public inbox for [email protected]
help / color / mirror / Atom feed[PATCH 04/13] Jsonpath support for json
111+ messages / 5 participants
[nested] [flat]
* [PATCH 04/13] Jsonpath support for json
@ 2019-03-01 00:15 Nikita Glukhov <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Nikita Glukhov @ 2019-03-01 00:15 UTC (permalink / raw)
---
src/backend/catalog/system_views.sql | 58 +
src/backend/utils/adt/json.c | 855 ++++++++++-
src/backend/utils/adt/jsonb.c | 18 -
src/backend/utils/adt/jsonb_util.c | 33 +-
src/backend/utils/adt/jsonpath_exec.c | 1516 +++++++++++++------
src/include/catalog/pg_operator.dat | 8 +
src/include/catalog/pg_proc.dat | 34 +
src/include/utils/jsonapi.h | 69 +
src/include/utils/jsonb.h | 33 +-
src/test/regress/expected/json_jsonpath.out | 2120 +++++++++++++++++++++++++++
src/test/regress/parallel_schedule | 2 +-
src/test/regress/serial_schedule | 1 +
src/test/regress/sql/json_jsonpath.sql | 477 ++++++
src/tools/pgindent/typedefs.list | 2 +
14 files changed, 4728 insertions(+), 498 deletions(-)
create mode 100644 src/test/regress/expected/json_jsonpath.out
create mode 100644 src/test/regress/sql/json_jsonpath.sql
diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql
index f5d11d5..5ac1fab 100644
--- a/src/backend/catalog/system_views.sql
+++ b/src/backend/catalog/system_views.sql
@@ -1167,6 +1167,64 @@ LANGUAGE INTERNAL
STRICT IMMUTABLE PARALLEL SAFE
AS 'jsonb_path_query_first';
+CREATE OR REPLACE FUNCTION
+ jsonb_path_query_first_text(target jsonb, path jsonpath, vars jsonb DEFAULT '{}',
+ silent boolean DEFAULT false)
+RETURNS text
+LANGUAGE INTERNAL
+STRICT IMMUTABLE PARALLEL SAFE
+AS 'jsonb_path_query_first_text';
+
+
+
+CREATE OR REPLACE FUNCTION
+ json_path_exists(target json, path jsonpath, vars json DEFAULT '{}',
+ silent boolean DEFAULT false)
+RETURNS boolean
+LANGUAGE INTERNAL
+STRICT IMMUTABLE PARALLEL SAFE
+AS 'json_path_exists';
+
+CREATE OR REPLACE FUNCTION
+ json_path_match(target json, path jsonpath, vars json DEFAULT '{}',
+ silent boolean DEFAULT false)
+RETURNS boolean
+LANGUAGE INTERNAL
+STRICT IMMUTABLE PARALLEL SAFE
+AS 'json_path_match';
+
+CREATE OR REPLACE FUNCTION
+ json_path_query(target json, path jsonpath, vars json DEFAULT '{}',
+ silent boolean DEFAULT false)
+RETURNS SETOF json
+LANGUAGE INTERNAL
+STRICT IMMUTABLE PARALLEL SAFE
+AS 'json_path_query';
+
+CREATE OR REPLACE FUNCTION
+ json_path_query_array(target json, path jsonpath, vars json DEFAULT '{}',
+ silent boolean DEFAULT false)
+RETURNS json
+LANGUAGE INTERNAL
+STRICT IMMUTABLE PARALLEL SAFE
+AS 'json_path_query_array';
+
+CREATE OR REPLACE FUNCTION
+ json_path_query_first(target json, path jsonpath, vars json DEFAULT '{}',
+ silent boolean DEFAULT false)
+RETURNS json
+LANGUAGE INTERNAL
+STRICT IMMUTABLE PARALLEL SAFE
+AS 'json_path_query_first';
+
+CREATE OR REPLACE FUNCTION
+ json_path_query_first_text(target json, path jsonpath, vars json DEFAULT '{}',
+ silent boolean DEFAULT false)
+RETURNS text
+LANGUAGE INTERNAL
+STRICT IMMUTABLE PARALLEL SAFE
+AS 'json_path_query_first_text';
+
--
-- The default permissions for functions mean that anyone can execute them.
-- A number of functions shouldn't be executable by just anyone, but rather
diff --git a/src/backend/utils/adt/json.c b/src/backend/utils/adt/json.c
index 5239903..803d478 100644
--- a/src/backend/utils/adt/json.c
+++ b/src/backend/utils/adt/json.c
@@ -106,6 +106,9 @@ static void add_json(Datum val, bool is_null, StringInfo result,
Oid val_type, bool key_scalar);
static text *catenate_stringinfo_string(StringInfo buffer, const char *addon);
+static JsonIterator *JsonIteratorInitFromLex(JsonContainer *jc,
+ JsonLexContext *lex, JsonIterator *parent);
+
/* the null action object used for pure validation */
static JsonSemAction nullSemAction =
{
@@ -127,6 +130,27 @@ lex_peek(JsonLexContext *lex)
}
/*
+ * lex_peek_value
+ *
+ * get the current look_ahead de-escaped lexeme.
+*/
+static inline char *
+lex_peek_value(JsonLexContext *lex)
+{
+ if (lex->token_type == JSON_TOKEN_STRING)
+ return lex->strval ? pstrdup(lex->strval->data) : NULL;
+ else
+ {
+ int len = lex->token_terminator - lex->token_start;
+ char *tokstr = palloc(len + 1);
+
+ memcpy(tokstr, lex->token_start, len);
+ tokstr[len] = '\0';
+ return tokstr;
+ }
+}
+
+/*
* lex_accept
*
* accept the look_ahead token and move the lexer to the next token if the
@@ -141,22 +165,8 @@ lex_accept(JsonLexContext *lex, JsonTokenType token, char **lexeme)
if (lex->token_type == token)
{
if (lexeme != NULL)
- {
- if (lex->token_type == JSON_TOKEN_STRING)
- {
- if (lex->strval != NULL)
- *lexeme = pstrdup(lex->strval->data);
- }
- else
- {
- int len = (lex->token_terminator - lex->token_start);
- char *tokstr = palloc(len + 1);
+ *lexeme = lex_peek_value(lex);
- memcpy(tokstr, lex->token_start, len);
- tokstr[len] = '\0';
- *lexeme = tokstr;
- }
- }
json_lex(lex);
return true;
}
@@ -2573,3 +2583,818 @@ json_typeof(PG_FUNCTION_ARGS)
PG_RETURN_TEXT_P(cstring_to_text(type));
}
+
+/*
+ * Initialize a JsonContainer from a json text, its type and size.
+ * 'type' can be JB_FOBJECT, JB_FARRAY, (JB_FARRAY | JB_FSCALAR).
+ * 'size' is a number of elements/pairs in array/object, or -1 if unknown.
+ */
+static void
+jsonInitContainer(JsonContainerData *jc, char *json, int len, int type,
+ int size)
+{
+ if (size < 0 || size > JB_CMASK)
+ size = JB_CMASK; /* unknown size */
+
+ jc->data = json;
+ jc->len = len;
+ jc->header = type | size;
+}
+
+/*
+ * Initialize a JsonContainer from a text datum.
+ */
+static void
+jsonInit(JsonContainerData *jc, Datum value)
+{
+ text *json = DatumGetTextP(value);
+ JsonLexContext *lex = makeJsonLexContext(json, false);
+ JsonTokenType tok;
+ int type;
+ int size = -1;
+
+ /* Lex exactly one token from the input and check its type. */
+ json_lex(lex);
+ tok = lex_peek(lex);
+
+ switch (tok)
+ {
+ case JSON_TOKEN_OBJECT_START:
+ type = JB_FOBJECT;
+ lex_accept(lex, tok, NULL);
+ if (lex_peek(lex) == JSON_TOKEN_OBJECT_END)
+ size = 0;
+ break;
+ case JSON_TOKEN_ARRAY_START:
+ type = JB_FARRAY;
+ lex_accept(lex, tok, NULL);
+ if (lex_peek(lex) == JSON_TOKEN_ARRAY_END)
+ size = 0;
+ break;
+ case JSON_TOKEN_STRING:
+ case JSON_TOKEN_NUMBER:
+ case JSON_TOKEN_TRUE:
+ case JSON_TOKEN_FALSE:
+ case JSON_TOKEN_NULL:
+ type = JB_FARRAY | JB_FSCALAR;
+ size = 1;
+ break;
+ default:
+ elog(ERROR, "unexpected json token: %d", tok);
+ type = jbvNull;
+ break;
+ }
+
+ pfree(lex);
+
+ jsonInitContainer(jc, VARDATA(json), VARSIZE(json) - VARHDRSZ, type, size);
+}
+
+/*
+ * Wrap JSON text into a palloc()'d Json structure.
+ */
+Json *
+JsonCreate(text *json)
+{
+ Json *res = palloc0(sizeof(*res));
+
+ jsonInit((JsonContainerData *) &res->root, PointerGetDatum(json));
+
+ return res;
+}
+
+/*
+ * Fill JsonbValue from the current iterator token.
+ * Returns true if recursion into nested object or array is needed (in this case
+ * child iterator is created and put into *pit).
+ */
+static bool
+jsonFillValue(JsonIterator **pit, JsonbValue *res, bool skipNested,
+ JsontIterState nextState)
+{
+ JsonIterator *it = *pit;
+ JsonLexContext *lex = it->lex;
+ JsonTokenType tok = lex_peek(lex);
+
+ switch (tok)
+ {
+ case JSON_TOKEN_NULL:
+ res->type = jbvNull;
+ break;
+
+ case JSON_TOKEN_TRUE:
+ res->type = jbvBool;
+ res->val.boolean = true;
+ break;
+
+ case JSON_TOKEN_FALSE:
+ res->type = jbvBool;
+ res->val.boolean = false;
+ break;
+
+ case JSON_TOKEN_STRING:
+ {
+ char *token = lex_peek_value(lex);
+ res->type = jbvString;
+ res->val.string.val = token;
+ res->val.string.len = strlen(token);
+ break;
+ }
+
+ case JSON_TOKEN_NUMBER:
+ {
+ char *token = lex_peek_value(lex);
+ res->type = jbvNumeric;
+ res->val.numeric = DatumGetNumeric(DirectFunctionCall3(
+ numeric_in, CStringGetDatum(token), 0, -1));
+ break;
+ }
+
+ case JSON_TOKEN_OBJECT_START:
+ case JSON_TOKEN_ARRAY_START:
+ {
+ JsonContainerData *cont = palloc(sizeof(*cont));
+ char *token_start = lex->token_start;
+ int len;
+
+ if (skipNested)
+ {
+ /* find the end of a container for its length calculation */
+ if (tok == JSON_TOKEN_OBJECT_START)
+ parse_object(lex, &nullSemAction);
+ else
+ parse_array(lex, &nullSemAction);
+
+ len = lex->token_start - token_start;
+ }
+ else
+ len = lex->input_length - (lex->token_start - lex->input);
+
+ jsonInitContainer(cont,
+ token_start, len,
+ tok == JSON_TOKEN_OBJECT_START ?
+ JB_FOBJECT : JB_FARRAY,
+ -1);
+
+ res->type = jbvBinary;
+ res->val.binary.data = (JsonbContainer *) cont;
+ res->val.binary.len = len;
+
+ if (skipNested)
+ return false;
+
+ /* recurse into container */
+ it->state = nextState;
+ *pit = JsonIteratorInitFromLex(cont, lex, *pit);
+ return true;
+ }
+
+ default:
+ report_parse_error(JSON_PARSE_VALUE, lex);
+ }
+
+ lex_accept(lex, tok, NULL);
+
+ return false;
+}
+
+/*
+ * Free the topmost entry in the stack of JsonIterators.
+ */
+static inline JsonIterator *
+JsonIteratorFreeAndGetParent(JsonIterator *it)
+{
+ JsonIterator *parent = it->parent;
+
+ pfree(it);
+
+ return parent;
+}
+
+/*
+ * Free the entire stack of JsonIterators.
+ */
+void
+JsonIteratorFree(JsonIterator *it)
+{
+ while (it)
+ it = JsonIteratorFreeAndGetParent(it);
+}
+
+/*
+ * Get next JsonbValue while iterating through JsonContainer.
+ *
+ * For more details, see JsonbIteratorNext().
+ */
+JsonbIteratorToken
+JsonIteratorNext(JsonIterator **pit, JsonbValue *val, bool skipNested)
+{
+ JsonIterator *it;
+
+ if (*pit == NULL)
+ return WJB_DONE;
+
+recurse:
+ it = *pit;
+
+ /* parse by recursive descent */
+ switch (it->state)
+ {
+ case JTI_ARRAY_START:
+ val->type = jbvArray;
+ val->val.array.nElems = it->isScalar ? 1 : -1;
+ val->val.array.rawScalar = it->isScalar;
+ val->val.array.elems = NULL;
+ it->state = it->isScalar ? JTI_ARRAY_ELEM_SCALAR : JTI_ARRAY_ELEM;
+ return WJB_BEGIN_ARRAY;
+
+ case JTI_ARRAY_ELEM_SCALAR:
+ {
+ (void) jsonFillValue(pit, val, skipNested, JTI_ARRAY_END);
+ it->state = JTI_ARRAY_END;
+ return WJB_ELEM;
+ }
+
+ case JTI_ARRAY_END:
+ if (!it->parent && lex_peek(it->lex) != JSON_TOKEN_END)
+ report_parse_error(JSON_PARSE_END, it->lex);
+ *pit = JsonIteratorFreeAndGetParent(*pit);
+ return WJB_END_ARRAY;
+
+ case JTI_ARRAY_ELEM:
+ if (lex_accept(it->lex, JSON_TOKEN_ARRAY_END, NULL))
+ {
+ it->state = JTI_ARRAY_END;
+ goto recurse;
+ }
+
+ if (jsonFillValue(pit, val, skipNested, JTI_ARRAY_ELEM_AFTER))
+ goto recurse;
+
+ /* fall through */
+
+ case JTI_ARRAY_ELEM_AFTER:
+ if (!lex_accept(it->lex, JSON_TOKEN_COMMA, NULL))
+ {
+ if (lex_peek(it->lex) != JSON_TOKEN_ARRAY_END)
+ report_parse_error(JSON_PARSE_ARRAY_NEXT, it->lex);
+ }
+
+ if (it->state == JTI_ARRAY_ELEM_AFTER)
+ {
+ it->state = JTI_ARRAY_ELEM;
+ goto recurse;
+ }
+
+ return WJB_ELEM;
+
+ case JTI_OBJECT_START:
+ val->type = jbvObject;
+ val->val.object.nPairs = -1;
+ val->val.object.pairs = NULL;
+ val->val.object.uniquify = false;
+ it->state = JTI_OBJECT_KEY;
+ return WJB_BEGIN_OBJECT;
+
+ case JTI_OBJECT_KEY:
+ if (lex_accept(it->lex, JSON_TOKEN_OBJECT_END, NULL))
+ {
+ if (!it->parent && lex_peek(it->lex) != JSON_TOKEN_END)
+ report_parse_error(JSON_PARSE_END, it->lex);
+ *pit = JsonIteratorFreeAndGetParent(*pit);
+ return WJB_END_OBJECT;
+ }
+
+ if (lex_peek(it->lex) != JSON_TOKEN_STRING)
+ report_parse_error(JSON_PARSE_OBJECT_START, it->lex);
+
+ (void) jsonFillValue(pit, val, true, JTI_OBJECT_VALUE);
+
+ if (!lex_accept(it->lex, JSON_TOKEN_COLON, NULL))
+ report_parse_error(JSON_PARSE_OBJECT_LABEL, it->lex);
+
+ it->state = JTI_OBJECT_VALUE;
+ return WJB_KEY;
+
+ case JTI_OBJECT_VALUE:
+ if (jsonFillValue(pit, val, skipNested, JTI_OBJECT_VALUE_AFTER))
+ goto recurse;
+
+ /* fall through */
+
+ case JTI_OBJECT_VALUE_AFTER:
+ if (!lex_accept(it->lex, JSON_TOKEN_COMMA, NULL))
+ {
+ if (lex_peek(it->lex) != JSON_TOKEN_OBJECT_END)
+ report_parse_error(JSON_PARSE_OBJECT_NEXT, it->lex);
+ }
+
+ if (it->state == JTI_OBJECT_VALUE_AFTER)
+ {
+ it->state = JTI_OBJECT_KEY;
+ goto recurse;
+ }
+
+ it->state = JTI_OBJECT_KEY;
+ return WJB_VALUE;
+
+ default:
+ break;
+ }
+
+ return WJB_DONE;
+}
+
+/* Initialize JsonIterator from json lexer which onto the first token. */
+static JsonIterator *
+JsonIteratorInitFromLex(JsonContainer *jc, JsonLexContext *lex,
+ JsonIterator *parent)
+{
+ JsonIterator *it = palloc(sizeof(JsonIterator));
+ JsonTokenType tok;
+
+ it->container = jc;
+ it->parent = parent;
+ it->lex = lex;
+
+ tok = lex_peek(it->lex);
+
+ switch (tok)
+ {
+ case JSON_TOKEN_OBJECT_START:
+ it->isScalar = false;
+ it->state = JTI_OBJECT_START;
+ lex_accept(it->lex, tok, NULL);
+ break;
+ case JSON_TOKEN_ARRAY_START:
+ it->isScalar = false;
+ it->state = JTI_ARRAY_START;
+ lex_accept(it->lex, tok, NULL);
+ break;
+ case JSON_TOKEN_STRING:
+ case JSON_TOKEN_NUMBER:
+ case JSON_TOKEN_TRUE:
+ case JSON_TOKEN_FALSE:
+ case JSON_TOKEN_NULL:
+ it->isScalar = true;
+ it->state = JTI_ARRAY_START;
+ break;
+ default:
+ report_parse_error(JSON_PARSE_VALUE, it->lex);
+ }
+
+ return it;
+}
+
+/*
+ * Given a JsonContainer, expand to JsonIterator to iterate over items
+ * fully expanded to in-memory representation for manipulation.
+ *
+ * See JsonbIteratorNext() for notes on memory management.
+ */
+JsonIterator *
+JsonIteratorInit(JsonContainer *jc)
+{
+ JsonLexContext *lex = makeJsonLexContextCstringLen(jc->data, jc->len, true);
+ json_lex(lex);
+ return JsonIteratorInitFromLex(jc, lex, NULL);
+}
+
+/*
+ * Serialize a single JsonbValue into text buffer.
+ */
+static void
+JsonEncodeJsonbValue(StringInfo buf, JsonbValue *jbv)
+{
+ check_stack_depth();
+
+ switch (jbv->type)
+ {
+ case jbvNull:
+ appendBinaryStringInfo(buf, "null", 4);
+ break;
+
+ case jbvBool:
+ if (jbv->val.boolean)
+ appendBinaryStringInfo(buf, "true", 4);
+ else
+ appendBinaryStringInfo(buf, "false", 5);
+ break;
+
+ case jbvNumeric:
+ /* replace numeric NaN with string "NaN" */
+ if (numeric_is_nan(jbv->val.numeric))
+ appendBinaryStringInfo(buf, "\"NaN\"", 5);
+ else
+ {
+ Datum str = DirectFunctionCall1(numeric_out,
+ NumericGetDatum(jbv->val.numeric));
+
+ appendStringInfoString(buf, DatumGetCString(str));
+ }
+ break;
+
+ case jbvString:
+ {
+ char *str = jbv->val.string.len < 0 ? jbv->val.string.val :
+ pnstrdup(jbv->val.string.val, jbv->val.string.len);
+
+ escape_json(buf, str);
+
+ if (jbv->val.string.len >= 0)
+ pfree(str);
+
+ break;
+ }
+
+ case jbvArray:
+ {
+ int i;
+
+ if (!jbv->val.array.rawScalar)
+ appendStringInfoChar(buf, '[');
+
+ for (i = 0; i < jbv->val.array.nElems; i++)
+ {
+ if (i > 0)
+ appendBinaryStringInfo(buf, ", ", 2);
+
+ JsonEncodeJsonbValue(buf, &jbv->val.array.elems[i]);
+ }
+
+ if (!jbv->val.array.rawScalar)
+ appendStringInfoChar(buf, ']');
+
+ break;
+ }
+
+ case jbvObject:
+ {
+ int i;
+
+ appendStringInfoChar(buf, '{');
+
+ for (i = 0; i < jbv->val.object.nPairs; i++)
+ {
+ if (i > 0)
+ appendBinaryStringInfo(buf, ", ", 2);
+
+ JsonEncodeJsonbValue(buf, &jbv->val.object.pairs[i].key);
+ appendBinaryStringInfo(buf, ": ", 2);
+ JsonEncodeJsonbValue(buf, &jbv->val.object.pairs[i].value);
+ }
+
+ appendStringInfoChar(buf, '}');
+ break;
+ }
+
+ case jbvBinary:
+ {
+ JsonContainer *json = (JsonContainer *) jbv->val.binary.data;
+
+ appendBinaryStringInfo(buf, json->data, json->len);
+ break;
+ }
+
+ default:
+ elog(ERROR, "unknown jsonb value type: %d", jbv->type);
+ break;
+ }
+}
+
+/*
+ * Turn an in-memory JsonbValue into a json for on-disk storage.
+ */
+Json *
+JsonbValueToJson(JsonbValue *jbv)
+{
+ StringInfoData buf;
+ Json *json = palloc0(sizeof(*json));
+ int type;
+ int size;
+
+ if (jbv->type == jbvBinary)
+ {
+ /* simply copy the whole container and its data */
+ JsonContainer *src = (JsonContainer *) jbv->val.binary.data;
+ JsonContainerData *dst = (JsonContainerData *) &json->root;
+
+ *dst = *src;
+ dst->data = memcpy(palloc(src->len), src->data, src->len);
+
+ return json;
+ }
+
+ initStringInfo(&buf);
+
+ JsonEncodeJsonbValue(&buf, jbv);
+
+ switch (jbv->type)
+ {
+ case jbvArray:
+ type = JB_FARRAY;
+ size = jbv->val.array.nElems;
+ break;
+
+ case jbvObject:
+ type = JB_FOBJECT;
+ size = jbv->val.object.nPairs;
+ break;
+
+ default: /* scalar */
+ type = JB_FARRAY | JB_FSCALAR;
+ size = 1;
+ break;
+ }
+
+ jsonInitContainer((JsonContainerData *) &json->root,
+ buf.data, buf.len, type, size);
+
+ return json;
+}
+
+/* Context and semantic actions for JsonGetArraySize() */
+typedef struct JsonGetArraySizeState
+{
+ int level;
+ uint32 size;
+} JsonGetArraySizeState;
+
+static void
+JsonGetArraySize_array_start(void *state)
+{
+ ((JsonGetArraySizeState *) state)->level++;
+}
+
+static void
+JsonGetArraySize_array_end(void *state)
+{
+ ((JsonGetArraySizeState *) state)->level--;
+}
+
+static void
+JsonGetArraySize_array_element_start(void *state, bool isnull)
+{
+ JsonGetArraySizeState *s = state;
+ if (s->level == 1)
+ s->size++;
+}
+
+/*
+ * Calculate the size of a json array by iterating through its elements.
+ */
+uint32
+JsonGetArraySize(JsonContainer *jc)
+{
+ JsonLexContext *lex = makeJsonLexContextCstringLen(jc->data, jc->len, false);
+ JsonSemAction sem;
+ JsonGetArraySizeState state;
+
+ state.level = 0;
+ state.size = 0;
+
+ memset(&sem, 0, sizeof(sem));
+ sem.semstate = &state;
+ sem.array_start = JsonGetArraySize_array_start;
+ sem.array_end = JsonGetArraySize_array_end;
+ sem.array_element_end = JsonGetArraySize_array_element_start;
+
+ json_lex(lex);
+ parse_array(lex, &sem);
+
+ return state.size;
+}
+
+/*
+ * Find last key in a json object by name. Returns palloc()'d copy of the
+ * corresponding value, or NULL if is not found.
+ */
+static inline JsonbValue *
+jsonFindLastKeyInObject(JsonContainer *obj, const JsonbValue *key)
+{
+ JsonbValue *res = NULL;
+ JsonbValue jbv;
+ JsonIterator *it;
+ JsonbIteratorToken tok;
+
+ Assert(JsonContainerIsObject(obj));
+ Assert(key->type == jbvString);
+
+ it = JsonIteratorInit(obj);
+
+ while ((tok = JsonIteratorNext(&it, &jbv, true)) != WJB_DONE)
+ {
+ if (tok == WJB_KEY && !lengthCompareJsonbStringValue(key, &jbv))
+ {
+ if (!res)
+ res = palloc(sizeof(*res));
+
+ tok = JsonIteratorNext(&it, res, true);
+ Assert(tok == WJB_VALUE);
+ }
+ }
+
+ return res;
+}
+
+/*
+ * Find scalar element in a array. Returns palloc()'d copy of value or NULL.
+ */
+static JsonbValue *
+jsonFindValueInArray(JsonContainer *array, const JsonbValue *elem)
+{
+ JsonbValue *val = palloc(sizeof(*val));
+ JsonIterator *it;
+ JsonbIteratorToken tok;
+
+ Assert(JsonContainerIsArray(array));
+ Assert(IsAJsonbScalar(elem));
+
+ it = JsonIteratorInit(array);
+
+ while ((tok = JsonIteratorNext(&it, val, true)) != WJB_DONE)
+ {
+ if (tok == WJB_ELEM && val->type == elem->type &&
+ equalsJsonbScalarValue(val, (JsonbValue *) elem))
+ {
+ JsonIteratorFree(it);
+ return val;
+ }
+ }
+
+ pfree(val);
+ return NULL;
+}
+
+/*
+ * Find value in object (i.e. the "value" part of some key/value pair in an
+ * object), or find a matching element if we're looking through an array.
+ * The "flags" argument allows the caller to specify which container types are
+ * of interest. If we cannot find the value, return NULL. Otherwise, return
+ * palloc()'d copy of value.
+ *
+ * For more details, see findJsonbValueFromContainer().
+ */
+JsonbValue *
+findJsonValueFromContainer(JsonContainer *jc, uint32 flags, JsonbValue *key)
+{
+ Assert((flags & ~(JB_FARRAY | JB_FOBJECT)) == 0);
+
+ if (!JsonContainerSize(jc))
+ return NULL;
+
+ if ((flags & JB_FARRAY) && JsonContainerIsArray(jc))
+ return jsonFindValueInArray(jc, key);
+
+ if ((flags & JB_FOBJECT) && JsonContainerIsObject(jc))
+ return jsonFindLastKeyInObject(jc, key);
+
+ /* Not found */
+ return NULL;
+}
+
+/*
+ * Get i-th element of a json array.
+ *
+ * Returns palloc()'d copy of the value, or NULL if it does not exist.
+ */
+JsonbValue *
+getIthJsonValueFromContainer(JsonContainer *array, uint32 index)
+{
+ JsonbValue *val = palloc(sizeof(JsonbValue));
+ JsonIterator *it;
+ JsonbIteratorToken tok;
+
+ Assert(JsonContainerIsArray(array));
+
+ it = JsonIteratorInit(array);
+
+ while ((tok = JsonIteratorNext(&it, val, true)) != WJB_DONE)
+ {
+ if (tok == WJB_ELEM)
+ {
+ if (index-- == 0)
+ {
+ JsonIteratorFree(it);
+ return val;
+ }
+ }
+ }
+
+ pfree(val);
+
+ return NULL;
+}
+
+/*
+ * Push json JsonbValue into JsonbParseState.
+ *
+ * Used for converting an in-memory JsonbValue to a json. For more details,
+ * see pushJsonbValue(). This function differs from pushJsonbValue() only by
+ * resetting "uniquify" flag in objects.
+ */
+JsonbValue *
+pushJsonValue(JsonbParseState **pstate, JsonbIteratorToken seq,
+ JsonbValue *jbval)
+{
+ JsonIterator *it;
+ JsonbValue *res = NULL;
+ JsonbValue v;
+ JsonbIteratorToken tok;
+
+ if (!jbval || (seq != WJB_ELEM && seq != WJB_VALUE) ||
+ jbval->type != jbvBinary)
+ {
+ /* drop through */
+ res = pushJsonbValueScalar(pstate, seq, jbval);
+
+ /* reset "uniquify" flag of objects */
+ if (seq == WJB_BEGIN_OBJECT)
+ (*pstate)->contVal.val.object.uniquify = false;
+
+ return res;
+ }
+
+ /* unpack the binary and add each piece to the pstate */
+ it = JsonIteratorInit((JsonContainer *) jbval->val.binary.data);
+ while ((tok = JsonIteratorNext(&it, &v, false)) != WJB_DONE)
+ {
+ res = pushJsonbValueScalar(pstate, tok,
+ tok < WJB_BEGIN_ARRAY ? &v : NULL);
+
+ /* reset "uniquify" flag of objects */
+ if (tok == WJB_BEGIN_OBJECT)
+ (*pstate)->contVal.val.object.uniquify = false;
+ }
+
+ return res;
+}
+
+/*
+ * Extract scalar JsonbValue from a scalar json.
+ */
+bool
+JsonExtractScalar(JsonContainer *jbc, JsonbValue *res)
+{
+ JsonIterator *it = JsonIteratorInit(jbc);
+ JsonbIteratorToken tok PG_USED_FOR_ASSERTS_ONLY;
+ JsonbValue tmp;
+
+ if (!JsonContainerIsScalar(jbc))
+ return false;
+
+ tok = JsonIteratorNext(&it, &tmp, true);
+ Assert(tok == WJB_BEGIN_ARRAY);
+ Assert(tmp.val.array.nElems == 1 && tmp.val.array.rawScalar);
+
+ tok = JsonIteratorNext(&it, res, true);
+ Assert(tok == WJB_ELEM);
+ Assert(IsAJsonbScalar(res));
+
+ tok = JsonIteratorNext(&it, &tmp, true);
+ Assert(tok == WJB_END_ARRAY);
+
+ return true;
+}
+
+/*
+ * Turn a Json into its C-string representation with stripping quotes from
+ * scalar strings.
+ */
+char *
+JsonUnquote(Json *jb)
+{
+ if (JsonContainerIsScalar(&jb->root))
+ {
+ JsonbValue v;
+
+ JsonExtractScalar(&jb->root, &v);
+
+ if (v.type == jbvString)
+ return pnstrdup(v.val.string.val, v.val.string.len);
+ }
+
+ return JsonToCString(NULL, &jb->root, 0);
+}
+
+/*
+ * Turn a JsonContainer into its C-string representation.
+ */
+char *
+JsonToCString(StringInfo out, JsonContainer *jc, int estimated_len)
+{
+ if (out)
+ {
+ appendBinaryStringInfo(out, jc->data, jc->len);
+ return out->data;
+ }
+ else
+ {
+ char *str = palloc(jc->len + 1);
+
+ memcpy(str, jc->data, jc->len);
+ str[jc->len] = 0;
+
+ return str;
+ }
+}
diff --git a/src/backend/utils/adt/jsonb.c b/src/backend/utils/adt/jsonb.c
index 87d3a8d..71f0679 100644
--- a/src/backend/utils/adt/jsonb.c
+++ b/src/backend/utils/adt/jsonb.c
@@ -206,24 +206,6 @@ JsonbTypeName(JsonbValue *jbv)
return "boolean";
case jbvNull:
return "null";
- case jbvDatetime:
- switch (jbv->val.datetime.typid)
- {
- case DATEOID:
- return "date";
- case TIMEOID:
- return "time without time zone";
- case TIMETZOID:
- return "time with time zone";
- case TIMESTAMPOID:
- return "timestamp without time zone";
- case TIMESTAMPTZOID:
- return "timestamp with time zone";
- default:
- elog(ERROR, "unrecognized jsonb value datetime type: %d",
- jbv->val.datetime.typid);
- }
- return "unknown";
default:
elog(ERROR, "unrecognized jsonb value type: %d", jbv->type);
return "unknown";
diff --git a/src/backend/utils/adt/jsonb_util.c b/src/backend/utils/adt/jsonb_util.c
index 5618d39..def7c8f 100644
--- a/src/backend/utils/adt/jsonb_util.c
+++ b/src/backend/utils/adt/jsonb_util.c
@@ -39,7 +39,6 @@
static void fillJsonbValue(JsonbContainer *container, int index,
char *base_addr, uint32 offset,
JsonbValue *result);
-static bool equalsJsonbScalarValue(JsonbValue *a, JsonbValue *b);
static int compareJsonbScalarValue(JsonbValue *a, JsonbValue *b);
static Jsonb *convertToJsonb(JsonbValue *val);
static void convertJsonbValue(StringInfo buffer, JEntry *header, JsonbValue *val, int level);
@@ -58,12 +57,8 @@ static JsonbParseState *pushState(JsonbParseState **pstate);
static void appendKey(JsonbParseState *pstate, JsonbValue *scalarVal);
static void appendValue(JsonbParseState *pstate, JsonbValue *scalarVal);
static void appendElement(JsonbParseState *pstate, JsonbValue *scalarVal);
-static int lengthCompareJsonbStringValue(const void *a, const void *b);
static int lengthCompareJsonbPair(const void *a, const void *b, void *arg);
static void uniqueifyJsonbObject(JsonbValue *object);
-static JsonbValue *pushJsonbValueScalar(JsonbParseState **pstate,
- JsonbIteratorToken seq,
- JsonbValue *scalarVal);
/*
* Turn an in-memory JsonbValue into a Jsonb for on-disk storage.
@@ -244,7 +239,6 @@ compareJsonbContainers(JsonbContainer *a, JsonbContainer *b)
res = (va.val.object.nPairs > vb.val.object.nPairs) ? 1 : -1;
break;
case jbvBinary:
- case jbvDatetime:
elog(ERROR, "unexpected jbvBinary value");
}
}
@@ -546,7 +540,7 @@ pushJsonbValue(JsonbParseState **pstate, JsonbIteratorToken seq,
* Do the actual pushing, with only scalar or pseudo-scalar-array values
* accepted.
*/
-static JsonbValue *
+JsonbValue *
pushJsonbValueScalar(JsonbParseState **pstate, JsonbIteratorToken seq,
JsonbValue *scalarVal)
{
@@ -584,6 +578,7 @@ pushJsonbValueScalar(JsonbParseState **pstate, JsonbIteratorToken seq,
(*pstate)->size = 4;
(*pstate)->contVal.val.object.pairs = palloc(sizeof(JsonbPair) *
(*pstate)->size);
+ (*pstate)->contVal.val.object.uniquify = true;
break;
case WJB_KEY:
Assert(scalarVal->type == jbvString);
@@ -826,6 +821,7 @@ recurse:
/* Set v to object on first object call */
val->type = jbvObject;
val->val.object.nPairs = (*it)->nElems;
+ val->val.object.uniquify = true;
/*
* v->val.object.pairs is not actually set, because we aren't
@@ -1299,7 +1295,7 @@ JsonbHashScalarValueExtended(const JsonbValue *scalarVal, uint64 *hash,
/*
* Are two scalar JsonbValues of the same type a and b equal?
*/
-static bool
+bool
equalsJsonbScalarValue(JsonbValue *aScalar, JsonbValue *bScalar)
{
if (aScalar->type == bScalar->type)
@@ -1753,22 +1749,6 @@ convertJsonbScalar(StringInfo buffer, JEntry *jentry, JsonbValue *scalarVal)
JENTRY_ISBOOL_TRUE : JENTRY_ISBOOL_FALSE;
break;
- case jbvDatetime:
- {
- char buf[MAXDATELEN + 1];
- size_t len;
-
- JsonEncodeDateTime(buf,
- scalarVal->val.datetime.value,
- scalarVal->val.datetime.typid,
- &scalarVal->val.datetime.tz);
- len = strlen(buf);
- appendToBuffer(buffer, buf, len);
-
- *jentry = JENTRY_ISSTRING | len;
- }
- break;
-
default:
elog(ERROR, "invalid jsonb scalar type");
}
@@ -1786,7 +1766,7 @@ convertJsonbScalar(StringInfo buffer, JEntry *jentry, JsonbValue *scalarVal)
* a and b are first sorted based on their length. If a tie-breaker is
* required, only then do we consider string binary equality.
*/
-static int
+int
lengthCompareJsonbStringValue(const void *a, const void *b)
{
const JsonbValue *va = (const JsonbValue *) a;
@@ -1850,6 +1830,9 @@ uniqueifyJsonbObject(JsonbValue *object)
Assert(object->type == jbvObject);
+ if (!object->val.object.uniquify)
+ return;
+
if (object->val.object.nPairs > 1)
qsort_arg(object->val.object.pairs, object->val.object.nPairs, sizeof(JsonbPair),
lengthCompareJsonbPair, &hasNonUniq);
diff --git a/src/backend/utils/adt/jsonpath_exec.c b/src/backend/utils/adt/jsonpath_exec.c
index 7d84ef9..68668ea 100644
--- a/src/backend/utils/adt/jsonpath_exec.c
+++ b/src/backend/utils/adt/jsonpath_exec.c
@@ -72,12 +72,12 @@
#include "utils/float.h"
#include "utils/guc.h"
#include "utils/json.h"
+#include "utils/jsonapi.h"
#include "utils/jsonpath.h"
#include "utils/date.h"
#include "utils/timestamp.h"
#include "utils/varlena.h"
-
/* Standard error message for SQL/JSON errors */
#define ERRMSG_JSON_ARRAY_NOT_FOUND "SQL/JSON array not found"
#define ERRMSG_JSON_OBJECT_NOT_FOUND "SQL/JSON object not found"
@@ -90,23 +90,128 @@
#define ERRMSG_INVALID_ARGUMENT_FOR_JSON_DATETIME_FUNCTION \
"invalid argument for SQL/JSON datetime function"
+typedef enum JsonItemType
+{
+ /* Scalar types */
+ jsiNull = jbvNull,
+ jsiString = jbvString,
+ jsiNumeric = jbvNumeric,
+ jsiBool = jbvBool,
+ /* Composite types */
+ jsiArray = jbvArray,
+ jsiObject = jbvObject,
+ /* Binary (i.e. struct Jsonb) jbvArray/jbvObject */
+ jsiBinary = jbvBinary,
+
+ /*
+ * Virtual types.
+ *
+ * These types are used only for in-memory JSON processing and serialized
+ * into JSON strings when outputted to json/jsonb.
+ */
+ jsiDatetime = 0x20
+} JsonItemType;
+
+/* SQL/JSON item */
+typedef struct JsonItem
+{
+ struct JsonItem *next;
+
+ union
+ {
+ int type; /* XXX JsonItemType */
+
+ JsonbValue jbv;
+
+ struct
+ {
+ int type;
+ Datum value;
+ Oid typid;
+ int32 typmod;
+ int tz;
+ } datetime;
+ } val;
+} JsonItem;
+
+#define JsonItemJbv(jsi) (&(jsi)->val.jbv)
+#define JsonItemBool(jsi) (JsonItemJbv(jsi)->val.boolean)
+#define JsonItemNumeric(jsi) (JsonItemJbv(jsi)->val.numeric)
+#define JsonItemNumericDatum(jsi) NumericGetDatum(JsonItemNumeric(jsi))
+#define JsonItemString(jsi) (JsonItemJbv(jsi)->val.string)
+#define JsonItemBinary(jsi) (JsonItemJbv(jsi)->val.binary)
+#define JsonItemArray(jsi) (JsonItemJbv(jsi)->val.array)
+#define JsonItemObject(jsi) (JsonItemJbv(jsi)->val.object)
+#define JsonItemDatetime(jsi) ((jsi)->val.datetime)
+
+#define JsonItemGetType(jsi) ((jsi)->val.type)
+#define JsonItemIsNull(jsi) (JsonItemGetType(jsi) == jsiNull)
+#define JsonItemIsBool(jsi) (JsonItemGetType(jsi) == jsiBool)
+#define JsonItemIsNumeric(jsi) (JsonItemGetType(jsi) == jsiNumeric)
+#define JsonItemIsString(jsi) (JsonItemGetType(jsi) == jsiString)
+#define JsonItemIsBinary(jsi) (JsonItemGetType(jsi) == jsiBinary)
+#define JsonItemIsArray(jsi) (JsonItemGetType(jsi) == jsiArray)
+#define JsonItemIsObject(jsi) (JsonItemGetType(jsi) == jsiObject)
+#define JsonItemIsDatetime(jsi) (JsonItemGetType(jsi) == jsiDatetime)
+#define JsonItemIsScalar(jsi) (IsAJsonbScalar(JsonItemJbv(jsi)) || \
+ JsonItemIsDatetime(jsi))
+
+typedef union Jsonx
+{
+ Jsonb jb;
+ Json js;
+} Jsonx;
+
+#define DatumGetJsonxP(datum, isJsonb) \
+ ((isJsonb) ? (Jsonx *) DatumGetJsonbP(datum) : (Jsonx *) DatumGetJsonP(datum))
+
+typedef JsonbContainer JsonxContainer;
+
+typedef struct JsonxIterator
+{
+ bool isJsonb;
+ union
+ {
+ JsonbIterator *jb;
+ JsonIterator *js;
+ } it;
+} JsonxIterator;
+
/*
* Represents "base object" and it's "id" for .keyvalue() evaluation.
*/
typedef struct JsonBaseObjectInfo
{
- JsonbContainer *jbc;
+ JsonxContainer *jbc;
int id;
} JsonBaseObjectInfo;
/*
+ * Special data structure representing stack of current items. We use it
+ * instead of regular list in order to evade extra memory allocation. These
+ * items are always allocated in local variables.
+ */
+typedef struct JsonItemStackEntry
+{
+ JsonItem *item;
+ struct JsonItemStackEntry *parent;
+} JsonItemStackEntry;
+
+typedef JsonItemStackEntry *JsonItemStack;
+
+typedef int (*JsonPathVarCallback) (void *vars, bool isJsonb,
+ char *varName, int varNameLen,
+ JsonItem *val, JsonbValue *baseObject);
+
+/*
* Context of jsonpath execution.
*/
typedef struct JsonPathExecContext
{
- Jsonb *vars; /* variables to substitute into jsonpath */
- JsonbValue *root; /* for $ evaluation */
- JsonbValue *current; /* for @ evaluation */
+ void *vars; /* variables to substitute into jsonpath */
+ JsonPathVarCallback getVar;
+ JsonItem *root; /* for $ evaluation */
+ JsonItemStack stack; /* for @ evaluation */
JsonBaseObjectInfo baseObject; /* "base object" for .keyvalue()
* evaluation */
int lastGeneratedObjectId; /* "id" counter for .keyvalue()
@@ -120,6 +225,7 @@ typedef struct JsonPathExecContext
* ignored */
bool throwErrors; /* with "false" all suppressible errors are
* suppressed */
+ bool isJsonb;
} JsonPathExecContext;
/* Context for LIKE_REGEX execution. */
@@ -148,20 +254,34 @@ typedef enum JsonPathExecResult
#define jperIsError(jper) ((jper) == jperError)
/*
- * List of jsonb values with shortcut for single-value list.
+ * List of SQL/JSON items with shortcut for single-value list.
*/
typedef struct JsonValueList
{
- JsonbValue *singleton;
- List *list;
+ JsonItem *head;
+ JsonItem *tail;
+ int length;
} JsonValueList;
typedef struct JsonValueListIterator
{
- JsonbValue *value;
- ListCell *next;
+ JsonItem *next;
} JsonValueListIterator;
+/*
+ * Context for execution of jsonb_path_*(jsonb, jsonpath [, vars jsonb]) user
+ * functions.
+ */
+typedef struct JsonPathUserFuncContext
+{
+ FunctionCallInfo fcinfo;
+ void *js; /* first jsonb function argument */
+ Json *json;
+ JsonPath *jp; /* second jsonpath function argument */
+ void *vars; /* third vars function argument */
+ JsonValueList found; /* resulting item list */
+} JsonPathUserFuncContext;
+
/* strict/lax flags is decomposed into four [un]wrap/error flags */
#define jspStrictAbsenseOfErrors(cxt) (!(cxt)->laxMode)
#define jspAutoUnwrap(cxt) ((cxt)->laxMode)
@@ -178,86 +298,131 @@ do { \
} while (0)
typedef JsonPathBool (*JsonPathPredicateCallback) (JsonPathItem *jsp,
- JsonbValue *larg,
- JsonbValue *rarg,
+ JsonItem *larg,
+ JsonItem *rarg,
void *param);
typedef Numeric (*BinaryArithmFunc) (Numeric num1, Numeric num2, bool *error);
-static JsonPathExecResult executeJsonPath(JsonPath *path, Jsonb *vars,
- Jsonb *json, bool throwErrors, JsonValueList *result);
+typedef JsonbValue *(*JsonBuilderFunc) (JsonbParseState **,
+ JsonbIteratorToken,
+ JsonbValue *);
+
+static void freeUserFuncContext(JsonPathUserFuncContext *cxt);
+static JsonPathExecResult executeUserFunc(FunctionCallInfo fcinfo,
+ JsonPathUserFuncContext *cxt, bool isJsonb, bool invertSilent,
+ bool copy);
+
+static JsonPathExecResult executeJsonPath(JsonPath *path, void *vars,
+ JsonPathVarCallback getVar, Jsonx *json, bool isJsonb,
+ bool throwErrors, JsonValueList *result);
static JsonPathExecResult recursiveExecute(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, JsonValueList *found);
+ JsonPathItem *jsp, JsonItem *jb, JsonValueList *found);
static JsonPathExecResult recursiveExecuteUnwrap(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, bool unwrap, JsonValueList *found);
+ JsonPathItem *jsp, JsonItem *jb, bool unwrap,
+ JsonValueList *found);
+
static JsonPathExecResult recursiveExecuteUnwrapArray(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb,
+ JsonPathItem *jsp, JsonItem *jb,
JsonValueList *found, bool unwrapElements);
static JsonPathExecResult recursiveExecuteNext(JsonPathExecContext *cxt,
JsonPathItem *cur, JsonPathItem *next,
- JsonbValue *v, JsonValueList *found, bool copy);
+ JsonItem *v, JsonValueList *found, bool copy);
static JsonPathExecResult recursiveExecuteAndUnwrap(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, bool unwrap, JsonValueList *found);
+ JsonPathItem *jsp, JsonItem *jb, bool unwrap, JsonValueList *found);
static JsonPathExecResult recursiveExecuteAndUnwrapNoThrow(
JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, bool unwrap, JsonValueList *found);
+ JsonItem *jb, bool unwrap, JsonValueList *found);
static JsonPathBool recursiveExecuteBool(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, bool canHaveNext);
+ JsonPathItem *jsp, JsonItem *jb, bool canHaveNext);
static JsonPathBool recursiveExecuteBoolNested(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb);
+ JsonPathItem *jsp, JsonItem *jb);
static JsonPathExecResult recursiveAny(JsonPathExecContext *cxt,
JsonPathItem *jsp, JsonbContainer *jbc, JsonValueList *found,
uint32 level, uint32 first, uint32 last,
bool ignoreStructuralErrors, bool unwrapNext);
static JsonPathBool executePredicate(JsonPathExecContext *cxt,
JsonPathItem *pred, JsonPathItem *larg, JsonPathItem *rarg,
- JsonbValue *jb, bool unwrapRightArg,
+ JsonItem *jb, bool unwrapRightArg,
JsonPathPredicateCallback exec, void *param);
static JsonPathExecResult executeBinaryArithmExpr(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, BinaryArithmFunc func,
+ JsonPathItem *jsp, JsonItem *jb, BinaryArithmFunc func,
JsonValueList *found);
static JsonPathExecResult executeUnaryArithmExpr(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, PGFunction func,
+ JsonPathItem *jsp, JsonItem *jb, PGFunction func,
JsonValueList *found);
static JsonPathBool executeStartsWith(JsonPathItem *jsp,
- JsonbValue *whole, JsonbValue *initial, void *param);
-static JsonPathBool executeLikeRegex(JsonPathItem *jsp, JsonbValue *str,
- JsonbValue *rarg, void *param);
+ JsonItem *whole, JsonItem *initial, void *param);
+static JsonPathBool executeLikeRegex(JsonPathItem *jsp, JsonItem *str,
+ JsonItem *rarg, void *param);
static JsonPathExecResult executeNumericItemMethod(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, bool unwrap, PGFunction func,
+ JsonPathItem *jsp, JsonItem *jb, bool unwrap, PGFunction func,
JsonValueList *found);
static JsonPathExecResult executeKeyValueMethod(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, JsonValueList *found);
+ JsonPathItem *jsp, JsonItem *jb, JsonValueList *found);
static JsonPathExecResult appendBoolResult(JsonPathExecContext *cxt,
JsonPathItem *jsp, JsonValueList *found, JsonPathBool res);
static void getJsonPathItem(JsonPathExecContext *cxt, JsonPathItem *item,
- JsonbValue *value);
+ JsonItem *value);
static void getJsonPathVariable(JsonPathExecContext *cxt,
- JsonPathItem *variable, Jsonb *vars, JsonbValue *value);
-static int JsonbArraySize(JsonbValue *jb);
-static JsonPathBool executeComparison(JsonPathItem *cmp, JsonbValue *lv,
- JsonbValue *rv, void *p);
-static JsonPathBool compareItems(int32 op, JsonbValue *jb1, JsonbValue *jb2);
+ JsonPathItem *variable, JsonItem *value);
+static int getJsonPathVariableFromJsonx(void *varsJsonb, bool isJsonb,
+ char *varName, int varNameLen, JsonItem *val,
+ JsonbValue *baseObject);
+static int JsonxArraySize(JsonItem *jb, bool isJsonb);
+static JsonPathBool executeComparison(JsonPathItem *cmp, JsonItem *lv,
+ JsonItem *rv, void *p);
+static JsonPathBool compareItems(int32 op, JsonItem *jb1, JsonItem *jb2);
static int compareNumeric(Numeric a, Numeric b);
-static JsonbValue *copyJsonbValue(JsonbValue *src);
+
+static void JsonItemInitNull(JsonItem *item);
+static void JsonItemInitBool(JsonItem *item, bool val);
+static void JsonItemInitNumeric(JsonItem *item, Numeric val);
+#define JsonItemInitNumericDatum(item, val) \
+ JsonItemInitNumeric(item, DatumGetNumeric(val))
+static void JsonItemInitString(JsonItem *item, char *str, int len);
+static void JsonItemInitDatetime(JsonItem *item, Datum val, Oid typid,
+ int32 typmod, int tz);
+
+static JsonItem *copyJsonItem(JsonItem *src);
+static JsonItem *JsonbValueToJsonItem(JsonbValue *jbv, JsonItem *jsi);
+static JsonbValue *JsonItemToJsonbValue(JsonItem *jsi, JsonbValue *jbv);
+static Jsonb *JsonItemToJsonb(JsonItem *jsi);
+static const char *JsonItemTypeName(JsonItem *jsi);
static JsonPathExecResult getArrayIndex(JsonPathExecContext *cxt,
- JsonPathItem *jsp, JsonbValue *jb, int32 *index);
+ JsonPathItem *jsp, JsonItem *jb, int32 *index);
static JsonBaseObjectInfo setBaseObject(JsonPathExecContext *cxt,
- JsonbValue *jbv, int32 id);
-static void JsonValueListAppend(JsonValueList *jvl, JsonbValue *jbv);
+ JsonItem *jsi, int32 id);
+static void JsonValueListAppend(JsonValueList *jvl, JsonItem *jbv);
static int JsonValueListLength(const JsonValueList *jvl);
static bool JsonValueListIsEmpty(JsonValueList *jvl);
-static JsonbValue *JsonValueListHead(JsonValueList *jvl);
+static JsonItem *JsonValueListHead(JsonValueList *jvl);
static List *JsonValueListGetList(JsonValueList *jvl);
static void JsonValueListInitIterator(const JsonValueList *jvl,
JsonValueListIterator *it);
-static JsonbValue *JsonValueListNext(const JsonValueList *jvl,
+static JsonItem *JsonValueListNext(const JsonValueList *jvl,
JsonValueListIterator *it);
-static int JsonbType(JsonbValue *jb);
+static int JsonbType(JsonItem *jb);
static JsonbValue *JsonbInitBinary(JsonbValue *jbv, Jsonb *jb);
-static int JsonbType(JsonbValue *jb);
-static JsonbValue *getScalar(JsonbValue *scalar, enum jbvType type);
-static JsonbValue *wrapItemsInArray(const JsonValueList *items);
+static inline JsonbValue *JsonInitBinary(JsonbValue *jbv, Json *js);
+static JsonItem *getScalar(JsonItem *scalar, enum jbvType type);
+static JsonbValue *wrapItemsInArray(const JsonValueList *items, bool isJsonb);
+static text *JsonItemUnquoteText(JsonItem *jsi, bool isJsonb);
+
+static JsonItem *getJsonObjectKey(JsonItem *jb, char *keystr, int keylen,
+ bool isJsonb, JsonItem *val);
+static JsonItem *getJsonArrayElement(JsonItem *jb, uint32 index, bool isJsonb,
+ JsonItem *elem);
+
+static void JsonxIteratorInit(JsonxIterator *it, JsonxContainer *jxc,
+ bool isJsonb);
+static JsonbIteratorToken JsonxIteratorNext(JsonxIterator *it, JsonbValue *jbv,
+ bool skipNested);
+static JsonbValue *JsonItemToJsonbValue(JsonItem *jsi, JsonbValue *jbv);
+static Json *JsonItemToJson(JsonItem *jsi);
+static Jsonx *JsonbValueToJsonx(JsonbValue *jbv, bool isJsonb);
+static Datum JsonbValueToJsonxDatum(JsonbValue *jbv, bool isJsonb);
+static Datum JsonItemToJsonxDatum(JsonItem *jsi, bool isJsonb);
static bool tryToParseDatetime(text *fmt, text *datetime, char *tzname,
bool strict, Datum *value, Oid *typid,
@@ -266,10 +431,14 @@ static int compareDatetime(Datum val1, Oid typid1, int tz1,
Datum val2, Oid typid2, int tz2,
bool *error);
+static void pushJsonItem(JsonItemStack *stack,
+ JsonItemStackEntry *entry, JsonItem *item);
+static void popJsonItem(JsonItemStack *stack);
+
/****************** User interface to JsonPath executor ********************/
/*
- * jsonb_path_exists
+ * json[b]_path_exists
* Returns true if jsonpath returns at least one item for the specified
* jsonb value. This function and jsonb_path_match() are used to
* implement @? and @@ operators, which in turn are intended to have an
@@ -280,25 +449,10 @@ static int compareDatetime(Datum val1, Oid typid1, int tz1,
* SQL/JSON. Regarding jsonb_path_match(), this function doesn't have
* an analogy in SQL/JSON, so we define its behavior on our own.
*/
-Datum
-jsonb_path_exists(PG_FUNCTION_ARGS)
+static Datum
+jsonx_path_exists(PG_FUNCTION_ARGS, bool isJsonb)
{
- Jsonb *jb = PG_GETARG_JSONB_P(0);
- JsonPath *jp = PG_GETARG_JSONPATH_P(1);
- JsonPathExecResult res;
- Jsonb *vars = NULL;
- bool silent = true;
-
- if (PG_NARGS() == 4)
- {
- vars = PG_GETARG_JSONB_P(2);
- silent = !PG_GETARG_BOOL(3);
- }
-
- res = executeJsonPath(jp, vars, jb, !silent, NULL);
-
- PG_FREE_IF_COPY(jb, 0);
- PG_FREE_IF_COPY(jp, 1);
+ JsonPathExecResult res = executeUserFunc(fcinfo, NULL, isJsonb, true, false);
if (jperIsError(res))
PG_RETURN_NULL();
@@ -306,99 +460,121 @@ jsonb_path_exists(PG_FUNCTION_ARGS)
PG_RETURN_BOOL(res == jperOk);
}
+Datum
+jsonb_path_exists(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_exists(fcinfo, true);
+}
+
+Datum
+json_path_exists(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_exists(fcinfo, false);
+}
+
/*
- * jsonb_path_exists_novars
- * Implements the 2-argument version of jsonb_path_exists
+ * json[b]_path_exists_opr
+ * Implements the 2-argument version of json[b]_path_exists
*/
Datum
jsonb_path_exists_opr(PG_FUNCTION_ARGS)
{
- /* just call the other one -- it can handle both cases */
- return jsonb_path_exists(fcinfo);
+ return jsonx_path_exists(fcinfo, true);
+}
+
+Datum
+json_path_exists_opr(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_exists(fcinfo, false);
}
/*
- * jsonb_path_match
+ * json[b]_path_match
* Returns jsonpath predicate result item for the specified jsonb value.
* See jsonb_path_exists() comment for details regarding error handling.
*/
-Datum
-jsonb_path_match(PG_FUNCTION_ARGS)
+static Datum
+jsonx_path_match(PG_FUNCTION_ARGS, bool isJsonb)
{
- Jsonb *jb = PG_GETARG_JSONB_P(0);
- JsonPath *jp = PG_GETARG_JSONPATH_P(1);
- JsonbValue *jbv;
- JsonValueList found = {0};
- Jsonb *vars = NULL;
- bool silent = true;
+ JsonPathUserFuncContext cxt;
+ JsonItem *res;
- if (PG_NARGS() == 4)
- {
- vars = PG_GETARG_JSONB_P(2);
- silent = !PG_GETARG_BOOL(3);
- }
-
- (void) executeJsonPath(jp, vars, jb, !silent, &found);
+ (void) executeUserFunc(fcinfo, &cxt, isJsonb, true, false);
- if (JsonValueListLength(&found) < 1)
+ if (JsonValueListLength(&cxt.found) < 1)
PG_RETURN_NULL();
- jbv = JsonValueListHead(&found);
+ res = JsonValueListHead(&cxt.found);
- PG_FREE_IF_COPY(jb, 0);
- PG_FREE_IF_COPY(jp, 1);
+ freeUserFuncContext(&cxt);
- if (jbv->type != jbvBool)
+ if (!JsonItemIsBool(res))
PG_RETURN_NULL();
- PG_RETURN_BOOL(jbv->val.boolean);
+ PG_RETURN_BOOL(JsonItemBool(res));
+}
+
+Datum
+jsonb_path_match(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_match(fcinfo, true);
+}
+
+Datum
+json_path_match(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_match(fcinfo, false);
}
/*
- * jsonb_path_match_novars
- * Implements the 2-argument version of jsonb_path_match
+ * json[b]_path_match_opr
+ * Implements the 2-argument version of json[b]_path_match
*/
Datum
jsonb_path_match_opr(PG_FUNCTION_ARGS)
{
/* just call the other one -- it can handle both cases */
- return jsonb_path_match(fcinfo);
+ return jsonx_path_match(fcinfo, true);
+}
+
+Datum
+json_path_match_opr(PG_FUNCTION_ARGS)
+{
+ /* just call the other one -- it can handle both cases */
+ return jsonx_path_match(fcinfo, false);
}
/*
- * jsonb_path_query
+ * json[b]_path_query
* Executes jsonpath for given jsonb document and returns result as
* rowset.
*/
-Datum
-jsonb_path_query(PG_FUNCTION_ARGS)
+static Datum
+jsonx_path_query(PG_FUNCTION_ARGS, bool isJsonb)
{
FuncCallContext *funcctx;
List *found;
- JsonbValue *v;
+ JsonItem *v;
ListCell *c;
+ Datum res;
if (SRF_IS_FIRSTCALL())
{
- JsonPath *jp;
- Jsonb *jb;
+ JsonPathUserFuncContext jspcxt;
MemoryContext oldcontext;
- Jsonb *vars;
- bool silent;
- JsonValueList found = {0};
funcctx = SRF_FIRSTCALL_INIT();
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
- jb = PG_GETARG_JSONB_P_COPY(0);
- jp = PG_GETARG_JSONPATH_P_COPY(1);
- vars = PG_GETARG_JSONB_P_COPY(2);
- silent = PG_GETARG_BOOL(3);
+ /* jsonb and jsonpath arguments are copied into SRF context. */
+ (void) executeUserFunc(fcinfo, &jspcxt, isJsonb, false, true);
- (void) executeJsonPath(jp, vars, jb, !silent, &found);
-
- funcctx->user_fctx = JsonValueListGetList(&found);
+ /*
+ * Don't free jspcxt because items in jspcxt.found can reference
+ * untoasted copies of jsonb and jsonpath arguments.
+ */
+ funcctx->user_fctx = JsonValueListGetList(&jspcxt.found);
MemoryContextSwitchTo(oldcontext);
}
@@ -413,50 +589,214 @@ jsonb_path_query(PG_FUNCTION_ARGS)
v = lfirst(c);
funcctx->user_fctx = list_delete_first(found);
- SRF_RETURN_NEXT(funcctx, JsonbPGetDatum(JsonbValueToJsonb(v)));
+ res = isJsonb ?
+ JsonbPGetDatum(JsonItemToJsonb(v)) :
+ JsonPGetDatum(JsonItemToJson(v));
+
+ SRF_RETURN_NEXT(funcctx, res);
+}
+
+Datum
+jsonb_path_query(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_query(fcinfo, true);
+}
+
+Datum
+json_path_query(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_query(fcinfo, false);
}
/*
- * jsonb_path_query_array
+ * json[b]_path_query_array
* Executes jsonpath for given jsonb document and returns result as
* jsonb array.
*/
-Datum
-jsonb_path_query_array(FunctionCallInfo fcinfo)
+static Datum
+jsonx_path_query_array(PG_FUNCTION_ARGS, bool isJsonb)
{
- Jsonb *jb = PG_GETARG_JSONB_P(0);
- JsonPath *jp = PG_GETARG_JSONPATH_P(1);
- JsonValueList found = {0};
- Jsonb *vars = PG_GETARG_JSONB_P(2);
- bool silent = PG_GETARG_BOOL(3);
+ JsonPathUserFuncContext cxt;
+ Datum res;
- (void) executeJsonPath(jp, vars, jb, !silent, &found);
+ (void) executeUserFunc(fcinfo, &cxt, isJsonb, false, false);
- PG_RETURN_JSONB_P(JsonbValueToJsonb(wrapItemsInArray(&found)));
+ res = JsonbValueToJsonxDatum(wrapItemsInArray(&cxt.found, isJsonb), isJsonb);
+
+ freeUserFuncContext(&cxt);
+
+ PG_RETURN_DATUM(res);
+}
+
+Datum
+jsonb_path_query_array(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_query_array(fcinfo, true);
+}
+
+Datum
+json_path_query_array(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_query_array(fcinfo, false);
}
/*
- * jsonb_path_query_first
+ * json[b]_path_query_first
* Executes jsonpath for given jsonb document and returns first result
* item. If there are no items, NULL returned.
*/
+static Datum
+jsonx_path_query_first(PG_FUNCTION_ARGS, bool isJsonb)
+{
+ JsonPathUserFuncContext cxt;
+ Datum res;
+
+ (void) executeUserFunc(fcinfo, &cxt, isJsonb, false, false);
+
+ if (JsonValueListLength(&cxt.found) >= 1)
+ res = JsonItemToJsonxDatum(JsonValueListHead(&cxt.found), isJsonb);
+ else
+ res = (Datum) 0;
+
+ freeUserFuncContext(&cxt);
+
+ if (res)
+ PG_RETURN_DATUM(res);
+ else
+ PG_RETURN_NULL();
+}
+
Datum
-jsonb_path_query_first(FunctionCallInfo fcinfo)
+jsonb_path_query_first(PG_FUNCTION_ARGS)
{
- Jsonb *jb = PG_GETARG_JSONB_P(0);
- JsonPath *jp = PG_GETARG_JSONPATH_P(1);
- JsonValueList found = {0};
- Jsonb *vars = PG_GETARG_JSONB_P(2);
- bool silent = PG_GETARG_BOOL(3);
+ return jsonx_path_query_first(fcinfo, true);
+}
- (void) executeJsonPath(jp, vars, jb, !silent, &found);
+Datum
+json_path_query_first(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_query_first(fcinfo, false);
+}
- if (JsonValueListLength(&found) >= 1)
- PG_RETURN_JSONB_P(JsonbValueToJsonb(JsonValueListHead(&found)));
+/*
+ * json[b]_path_query_first_text
+ * Executes jsonpath for given jsonb document and returns first result
+ * item as text. If there are no items, NULL returned.
+ */
+static Datum
+jsonx_path_query_first_text(PG_FUNCTION_ARGS, bool isJsonb)
+{
+ JsonPathUserFuncContext cxt;
+ text *txt;
+
+ (void) executeUserFunc(fcinfo, &cxt, isJsonb, false, false);
+
+ if (JsonValueListLength(&cxt.found) >= 1)
+ txt = JsonItemUnquoteText(JsonValueListHead(&cxt.found), isJsonb);
+ else
+ txt = NULL;
+
+ freeUserFuncContext(&cxt);
+
+ if (txt)
+ PG_RETURN_TEXT_P(txt);
else
PG_RETURN_NULL();
}
+Datum
+jsonb_path_query_first_text(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_query_first_text(fcinfo, true);
+}
+
+Datum
+json_path_query_first_text(PG_FUNCTION_ARGS)
+{
+ return jsonx_path_query_first_text(fcinfo, false);
+}
+
+/* Free untoasted copies of jsonb and jsonpath arguments. */
+static void
+freeUserFuncContext(JsonPathUserFuncContext *cxt)
+{
+ FunctionCallInfo fcinfo = cxt->fcinfo;
+
+ PG_FREE_IF_COPY(cxt->js, 0);
+ PG_FREE_IF_COPY(cxt->jp, 1);
+ if (cxt->vars)
+ PG_FREE_IF_COPY(cxt->vars, 2);
+ if (cxt->json)
+ pfree(cxt->json);
+}
+
+/*
+ * Common code for jsonb_path_*(jsonb, jsonpath [, vars jsonb, silent bool])
+ * user functions.
+ *
+ * 'copy' flag enables copying of first three arguments into the current memory
+ * context.
+ */
+static JsonPathExecResult
+executeUserFunc(FunctionCallInfo fcinfo, JsonPathUserFuncContext *cxt,
+ bool isJsonb, bool invertSilent, bool copy)
+{
+ Datum js_toasted = PG_GETARG_DATUM(0);
+ struct varlena *js_detoasted = copy ?
+ PG_DETOAST_DATUM(js_toasted) :
+ PG_DETOAST_DATUM_COPY(js_toasted);
+ Jsonx *js = DatumGetJsonxP(js_detoasted, isJsonb);
+ JsonPath *jp = copy ? PG_GETARG_JSONPATH_P_COPY(1) : PG_GETARG_JSONPATH_P(1);
+ struct varlena *vars_detoasted = NULL;
+ Jsonx *vars = NULL;
+ bool silent = true;
+ JsonPathExecResult res;
+
+ if (PG_NARGS() == 4)
+ {
+ Datum vars_toasted = PG_GETARG_DATUM(2);
+
+ vars_detoasted = copy ?
+ PG_DETOAST_DATUM(vars_toasted) :
+ PG_DETOAST_DATUM_COPY(vars_toasted);
+
+ vars = DatumGetJsonxP(vars_detoasted, isJsonb);
+
+ silent = PG_GETARG_BOOL(3) ^ invertSilent;
+ }
+
+ if (cxt)
+ {
+ cxt->fcinfo = fcinfo;
+ cxt->js = js_detoasted;
+ cxt->jp = jp;
+ cxt->vars = vars_detoasted;
+ cxt->json = isJsonb ? NULL : &js->js;
+ memset(&cxt->found, 0, sizeof(cxt->found));
+ }
+
+ res = executeJsonPath(jp, vars, getJsonPathVariableFromJsonx,
+ js, isJsonb, !silent, cxt ? &cxt->found : NULL);
+
+ if (!cxt && !copy)
+ {
+ PG_FREE_IF_COPY(js_detoasted, 0);
+ PG_FREE_IF_COPY(jp, 1);
+
+ if (vars_detoasted)
+ PG_FREE_IF_COPY(vars_detoasted, 2);
+
+ if (!isJsonb)
+ {
+ pfree(js);
+ if (vars)
+ pfree(vars);
+ }
+ }
+
+ return res;
+}
+
/********************Execute functions for JsonPath**************************/
/*
@@ -478,37 +818,45 @@ jsonb_path_query_first(FunctionCallInfo fcinfo)
* In other case it tries to find all the satisfied result items.
*/
static JsonPathExecResult
-executeJsonPath(JsonPath *path, Jsonb *vars, Jsonb *json, bool throwErrors,
+executeJsonPath(JsonPath *path, void *vars, JsonPathVarCallback getVar,
+ Jsonx *json, bool isJsonb, bool throwErrors,
JsonValueList *result)
{
JsonPathExecContext cxt;
JsonPathExecResult res;
JsonPathItem jsp;
- JsonbValue jbv;
+ JsonItem jsi;
+ JsonbValue *jbv = JsonItemJbv(&jsi);
+ JsonItemStackEntry root;
jspInit(&jsp, path);
- if (!JsonbExtractScalar(&json->root, &jbv))
- JsonbInitBinary(&jbv, json);
-
- if (vars && !JsonContainerIsObject(&vars->root))
+ if (isJsonb)
{
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
- errmsg("jsonb containing jsonpath variables "
- "is not an object")));
+ if (!JsonbExtractScalar(&json->jb.root, jbv))
+ JsonbInitBinary(jbv, &json->jb);
+ }
+ else
+ {
+ if (!JsonExtractScalar(&json->js.root, jbv))
+ JsonInitBinary(jbv, &json->js);
}
cxt.vars = vars;
+ cxt.getVar = getVar;
cxt.laxMode = (path->header & JSONPATH_LAX) != 0;
cxt.ignoreStructuralErrors = cxt.laxMode;
- cxt.root = &jbv;
- cxt.current = &jbv;
+ cxt.root = &jsi;
+ cxt.stack = NULL;
cxt.baseObject.jbc = NULL;
cxt.baseObject.id = 0;
- cxt.lastGeneratedObjectId = vars ? 2 : 1;
+ /* 1 + number of base objects in vars */
+ cxt.lastGeneratedObjectId = 1 + getVar(vars, isJsonb, NULL, 0, NULL, NULL);
cxt.innermostArraySize = -1;
cxt.throwErrors = throwErrors;
+ cxt.isJsonb = isJsonb;
+
+ pushJsonItem(&cxt.stack, &root, cxt.root);
if (jspStrictAbsenseOfErrors(&cxt) && !result)
{
@@ -520,7 +868,7 @@ executeJsonPath(JsonPath *path, Jsonb *vars, Jsonb *json, bool throwErrors,
Assert(!throwErrors);
- res = recursiveExecute(&cxt, &jsp, &jbv, &vals);
+ res = recursiveExecute(&cxt, &jsp, &jsi, &vals);
if (jperIsError(res))
return res;
@@ -528,7 +876,7 @@ executeJsonPath(JsonPath *path, Jsonb *vars, Jsonb *json, bool throwErrors,
return JsonValueListIsEmpty(&vals) ? jperNotFound : jperOk;
}
- res = recursiveExecute(&cxt, &jsp, &jbv, result);
+ res = recursiveExecute(&cxt, &jsp, &jsi, result);
Assert(!throwErrors || !jperIsError(res));
@@ -540,7 +888,7 @@ executeJsonPath(JsonPath *path, Jsonb *vars, Jsonb *json, bool throwErrors,
*/
static JsonPathExecResult
recursiveExecute(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, JsonValueList *found)
+ JsonItem *jb, JsonValueList *found)
{
return recursiveExecuteUnwrap(cxt, jsp, jb, jspAutoUnwrap(cxt), found);
}
@@ -552,7 +900,7 @@ recursiveExecute(JsonPathExecContext *cxt, JsonPathItem *jsp,
*/
static JsonPathExecResult
recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, bool unwrap, JsonValueList *found)
+ JsonItem *jb, bool unwrap, JsonValueList *found)
{
JsonPathItem elem;
JsonPathExecResult res = jperNotFound;
@@ -587,23 +935,16 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
case jpiKey:
if (JsonbType(jb) == jbvObject)
{
- JsonbValue *v;
- JsonbValue key;
-
- key.type = jbvString;
- key.val.string.val = jspGetString(jsp, &key.val.string.len);
+ JsonItem val;
+ int keylen;
+ char *key = jspGetString(jsp, &keylen);
- v = findJsonbValueFromContainer(jb->val.binary.data,
- JB_FOBJECT, &key);
+ jb = getJsonObjectKey(jb, key, keylen, cxt->isJsonb, &val);
- if (v != NULL)
+ if (jb != NULL)
{
res = recursiveExecuteNext(cxt, jsp, NULL,
- v, found, false);
-
- /* free value if it was not added to found list */
- if (jspHasNext(jsp) || !found)
- pfree(v);
+ jb, found, true);
}
else if (!jspIgnoreStructuralErrors(cxt))
{
@@ -617,7 +958,7 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
initStringInfo(&keybuf);
- keystr = pnstrdup(key.val.string.val, key.val.string.len);
+ keystr = pnstrdup(key, keylen);
escape_json(&keybuf, keystr);
ereport(ERROR,
@@ -648,7 +989,7 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
break;
case jpiCurrent:
- res = recursiveExecuteNext(cxt, jsp, NULL, cxt->current,
+ res = recursiveExecuteNext(cxt, jsp, NULL, cxt->stack->item,
found, true);
break;
@@ -675,7 +1016,7 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
{
int innermostArraySize = cxt->innermostArraySize;
int i;
- int size = JsonbArraySize(jb);
+ int size = JsonxArraySize(jb, cxt->isJsonb);
bool singleton = size < 0;
bool hasNext = jspGetNext(jsp, &elem);
@@ -729,30 +1070,27 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
for (index = index_from; index <= index_to; index++)
{
- JsonbValue *v;
- bool copy;
+ JsonItem jsibuf;
+ JsonItem *jsi;
if (singleton)
{
- v = jb;
- copy = true;
+ jsi = jb;
}
else
{
- v = getIthJsonbValueFromContainer(jb->val.binary.data,
- (uint32) index);
+ jsi = getJsonArrayElement(jb, (uint32) index,
+ cxt->isJsonb, &jsibuf);
- if (v == NULL)
+ if (jsi == NULL)
continue;
-
- copy = false;
}
if (!hasNext && !found)
return jperOk;
- res = recursiveExecuteNext(cxt, jsp, &elem, v, found,
- copy);
+ res = recursiveExecuteNext(cxt, jsp, &elem, jsi, found,
+ true);
if (jperIsError(res))
break;
@@ -782,8 +1120,8 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
case jpiLast:
{
- JsonbValue tmpjbv;
- JsonbValue *lastjbv;
+ JsonItem tmpjsi;
+ JsonItem *lastjsi;
int last;
bool hasNext = jspGetNext(jsp, &elem);
@@ -799,15 +1137,14 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
last = cxt->innermostArraySize - 1;
- lastjbv = hasNext ? &tmpjbv : palloc(sizeof(*lastjbv));
+ lastjsi = hasNext ? &tmpjsi : palloc(sizeof(*lastjsi));
- lastjbv->type = jbvNumeric;
- lastjbv->val.numeric =
- DatumGetNumeric(DirectFunctionCall1(int4_numeric,
- Int32GetDatum(last)));
+ JsonItemInitNumericDatum(lastjsi,
+ DirectFunctionCall1(int4_numeric,
+ Int32GetDatum(last)));
res = recursiveExecuteNext(cxt, jsp, &elem,
- lastjbv, found, hasNext);
+ lastjsi, found, hasNext);
}
break;
@@ -816,11 +1153,12 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
{
bool hasNext = jspGetNext(jsp, &elem);
- if (jb->type != jbvBinary)
- elog(ERROR, "invalid jsonb object type: %d", jb->type);
+ if (!JsonItemIsBinary(jb))
+ elog(ERROR, "invalid jsonb object type: %d",
+ JsonItemGetType(jb));
return recursiveAny(cxt, hasNext ? &elem : NULL,
- jb->val.binary.data, found, 1, 1, 1,
+ JsonItemBinary(jb).data, found, 1, 1, 1,
false, jspAutoUnwrap(cxt));
}
else if (unwrap && JsonbType(jb) == jbvArray)
@@ -900,9 +1238,9 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
break;
}
- if (jb->type == jbvBinary)
+ if (JsonItemIsBinary(jb))
res = recursiveAny(cxt, hasNext ? &elem : NULL,
- jb->val.binary.data, found,
+ JsonItemBinary(jb).data, found,
1,
jsp->content.anybounds.first,
jsp->content.anybounds.last,
@@ -916,8 +1254,8 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
case jpiString:
case jpiVariable:
{
- JsonbValue vbuf;
- JsonbValue *v;
+ JsonItem vbuf;
+ JsonItem *v;
bool hasNext = jspGetNext(jsp, &elem);
if (!hasNext && !found)
@@ -925,7 +1263,6 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
res = jperOk; /* skip evaluation */
break;
}
-
v = hasNext ? &vbuf : palloc(sizeof(*v));
baseObject = cxt->baseObject;
@@ -939,20 +1276,18 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
case jpiType:
{
- JsonbValue *jbv = palloc(sizeof(*jbv));
+ JsonItem jsi;
+ const char *typname = JsonItemTypeName(jb);
- jbv->type = jbvString;
- jbv->val.string.val = pstrdup(JsonbTypeName(jb));
- jbv->val.string.len = strlen(jbv->val.string.val);
+ JsonItemInitString(&jsi, pstrdup(typname), strlen(typname));
- res = recursiveExecuteNext(cxt, jsp, NULL, jbv,
- found, false);
+ res = recursiveExecuteNext(cxt, jsp, NULL, &jsi, found, true);
}
break;
case jpiSize:
{
- int size = JsonbArraySize(jb);
+ int size = JsonxArraySize(jb, cxt->isJsonb);
if (size < 0)
{
@@ -973,10 +1308,8 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
jb = palloc(sizeof(*jb));
- jb->type = jbvNumeric;
- jb->val.numeric =
- DatumGetNumeric(DirectFunctionCall1(int4_numeric,
- Int32GetDatum(size)));
+ JsonItemInitNumericDatum(jb, DirectFunctionCall1(int4_numeric,
+ Int32GetDatum(size)));
res = recursiveExecuteNext(cxt, jsp, NULL, jb, found, false);
}
@@ -996,16 +1329,16 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
case jpiDouble:
{
- JsonbValue jbv;
+ JsonItem jsi;
if (unwrap && JsonbType(jb) == jbvArray)
return recursiveExecuteUnwrapArray(cxt, jsp, jb, found,
false);
- if (jb->type == jbvNumeric)
+ if (JsonItemIsNumeric(jb))
{
char *tmp = DatumGetCString(DirectFunctionCall1(numeric_out,
- NumericGetDatum(jb->val.numeric)));
+ NumericGetDatum(JsonItemNumeric(jb))));
bool error = false;
(void) float8in_internal_error(tmp,
@@ -1024,13 +1357,13 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
res = jperOk;
}
- else if (jb->type == jbvString)
+ else if (JsonItemIsString(jb))
{
/* cast string as double */
bool error = false;
double val;
- char *tmp = pnstrdup(jb->val.string.val,
- jb->val.string.len);
+ char *tmp = pnstrdup(JsonItemString(jb).val,
+ JsonItemString(jb).len);
val = float8in_internal_error(tmp,
NULL,
@@ -1046,10 +1379,9 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
"applied to not a numeric value",
jspOperationName(jsp->type)))));
- jb = &jbv;
- jb->type = jbvNumeric;
- jb->val.numeric = DatumGetNumeric(DirectFunctionCall1(float8_numeric,
- Float8GetDatum(val)));
+ jb = &jsi;
+ JsonItemInitNumericDatum(jb, DirectFunctionCall1(float8_numeric,
+ Float8GetDatum(val)));
res = jperOk;
}
@@ -1089,8 +1421,8 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
"applied to not a string",
jspOperationName(jsp->type)))));
- datetime = cstring_to_text_with_len(jb->val.string.val,
- jb->val.string.len);
+ datetime = cstring_to_text_with_len(JsonItemString(jb).val,
+ JsonItemString(jb).len);
if (jsp->content.args.left)
{
@@ -1110,7 +1442,7 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
{
JsonValueList tzlist = {0};
JsonPathExecResult tzres;
- JsonbValue *tzjbv;
+ JsonItem *tzjsi;
jspGetRightArg(jsp, &elem);
tzres = recursiveExecute(cxt, &elem, jb, &tzlist);
@@ -1118,8 +1450,8 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
return tzres;
if (JsonValueListLength(&tzlist) != 1 ||
- ((tzjbv = JsonValueListHead(&tzlist))->type != jbvString &&
- tzjbv->type != jbvNumeric))
+ (!JsonItemIsString((tzjsi = JsonValueListHead(&tzlist))) &&
+ !JsonItemIsNumeric(tzjsi)))
RETURN_ERROR(ereport(ERROR,
(errcode(ERRCODE_INVALID_ARGUMENT_FOR_JSON_DATETIME_FUNCTION),
errmsg(ERRMSG_INVALID_ARGUMENT_FOR_JSON_DATETIME_FUNCTION),
@@ -1128,14 +1460,14 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
"is not a singleton string or number",
jspOperationName(jsp->type)))));
- if (tzjbv->type == jbvString)
- tzname = pnstrdup(tzjbv->val.string.val,
- tzjbv->val.string.len);
+ if (JsonItemIsString(tzjsi))
+ tzname = pnstrdup(JsonItemString(tzjsi).val,
+ JsonItemString(tzjsi).len);
else
{
bool error = false;
- tz = numeric_int4_error(tzjbv->val.numeric, &error);
+ tz = numeric_int4_error(JsonItemNumeric(tzjsi), &error);
if (error || tz == PG_INT32_MIN)
RETURN_ERROR(ereport(ERROR,
@@ -1227,11 +1559,7 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
jb = hasNext ? &jbvbuf : palloc(sizeof(*jb));
- jb->type = jbvDatetime;
- jb->val.datetime.value = value;
- jb->val.datetime.typid = typid;
- jb->val.datetime.typmod = typmod;
- jb->val.datetime.tz = tz;
+ JsonItemInitDatetime(jb, value, typid, typmod, tz);
res = recursiveExecuteNext(cxt, jsp, &elem, jb, found, hasNext);
}
@@ -1255,16 +1583,16 @@ recursiveExecuteUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
*/
static JsonPathExecResult
recursiveExecuteUnwrapArray(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, JsonValueList *found,
+ JsonItem *jb, JsonValueList *found,
bool unwrapElements)
{
- if (jb->type != jbvBinary)
+ if (!JsonItemIsBinary(jb))
{
- Assert(jb->type != jbvArray);
- elog(ERROR, "invalid jsonb array value type: %d", jb->type);
+ Assert(!JsonItemIsArray(jb));
+ elog(ERROR, "invalid jsonb array value type: %d", JsonItemGetType(jb));
}
- return recursiveAny(cxt, jsp, jb->val.binary.data, found, 1, 1, 1,
+ return recursiveAny(cxt, jsp, JsonItemBinary(jb).data, found, 1, 1, 1,
false, unwrapElements);
}
@@ -1274,7 +1602,7 @@ recursiveExecuteUnwrapArray(JsonPathExecContext *cxt, JsonPathItem *jsp,
static JsonPathExecResult
recursiveExecuteNext(JsonPathExecContext *cxt,
JsonPathItem *cur, JsonPathItem *next,
- JsonbValue *v, JsonValueList *found, bool copy)
+ JsonItem *v, JsonValueList *found, bool copy)
{
JsonPathItem elem;
bool hasNext;
@@ -1293,7 +1621,7 @@ recursiveExecuteNext(JsonPathExecContext *cxt,
return recursiveExecute(cxt, next, v, found);
if (found)
- JsonValueListAppend(found, copy ? copyJsonbValue(v) : v);
+ JsonValueListAppend(found, copy ? copyJsonItem(v) : v);
return jperOk;
}
@@ -1304,22 +1632,40 @@ recursiveExecuteNext(JsonPathExecContext *cxt,
*/
static JsonPathExecResult
recursiveExecuteAndUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, bool unwrap, JsonValueList *found)
+ JsonItem *jb, bool unwrap, JsonValueList *found)
{
if (unwrap && jspAutoUnwrap(cxt))
{
JsonValueList seq = {0};
JsonValueListIterator it;
JsonPathExecResult res = recursiveExecute(cxt, jsp, jb, &seq);
- JsonbValue *item;
+ JsonItem *item;
+ int count;
if (jperIsError(res))
return res;
+ count = JsonValueListLength(&seq);
+
+ if (!count)
+ return jperNotFound;
+
+ /* Optimize copying of singleton item into empty list */
+ if (count == 1 &&
+ JsonbType((item = JsonValueListHead(&seq))) != jbvArray)
+ {
+ if (JsonValueListIsEmpty(found))
+ *found = seq;
+ else
+ JsonValueListAppend(found, item);
+
+ return jperOk;
+ }
+
JsonValueListInitIterator(&seq, &it);
while ((item = JsonValueListNext(&seq, &it)))
{
- Assert(item->type != jbvArray);
+ Assert(!JsonItemIsArray(item));
if (JsonbType(item) == jbvArray)
recursiveExecuteUnwrapArray(cxt, NULL, item, found, false);
@@ -1335,7 +1681,7 @@ recursiveExecuteAndUnwrap(JsonPathExecContext *cxt, JsonPathItem *jsp,
static JsonPathExecResult
recursiveExecuteAndUnwrapNoThrow(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, bool unwrap,
+ JsonItem *jb, bool unwrap,
JsonValueList *found)
{
JsonPathExecResult res;
@@ -1351,7 +1697,7 @@ recursiveExecuteAndUnwrapNoThrow(JsonPathExecContext *cxt, JsonPathItem *jsp,
/* Execute boolean-valued jsonpath expression. */
static JsonPathBool
recursiveExecuteBool(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, bool canHaveNext)
+ JsonItem *jb, bool canHaveNext)
{
JsonPathItem larg;
JsonPathItem rarg;
@@ -1482,15 +1828,14 @@ recursiveExecuteBool(JsonPathExecContext *cxt, JsonPathItem *jsp,
*/
static JsonPathBool
recursiveExecuteBoolNested(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb)
+ JsonItem *jb)
{
- JsonbValue *prev;
+ JsonItemStackEntry current;
JsonPathBool res;
- prev = cxt->current;
- cxt->current = jb;
+ pushJsonItem(&cxt->stack, ¤t, jb);
res = recursiveExecuteBool(cxt, jsp, jb, false);
- cxt->current = prev;
+ popJsonItem(&cxt->stack);
return res;
}
@@ -1507,25 +1852,26 @@ recursiveAny(JsonPathExecContext *cxt, JsonPathItem *jsp, JsonbContainer *jbc,
bool ignoreStructuralErrors, bool unwrapNext)
{
JsonPathExecResult res = jperNotFound;
- JsonbIterator *it;
+ JsonxIterator it;
int32 r;
- JsonbValue v;
+ JsonItem v;
check_stack_depth();
if (level > last)
return res;
- it = JsonbIteratorInit(jbc);
+
+ JsonxIteratorInit(&it, jbc, cxt->isJsonb);
/*
* Recursively iterate over jsonb objects/arrays
*/
- while ((r = JsonbIteratorNext(&it, &v, true)) != WJB_DONE)
+ while ((r = JsonxIteratorNext(&it, JsonItemJbv(&v), true)) != WJB_DONE)
{
if (r == WJB_KEY)
{
- r = JsonbIteratorNext(&it, &v, true);
+ r = JsonxIteratorNext(&it, JsonItemJbv(&v), true);
Assert(r == WJB_VALUE);
}
@@ -1534,7 +1880,7 @@ recursiveAny(JsonPathExecContext *cxt, JsonPathItem *jsp, JsonbContainer *jbc,
if (level >= first ||
(first == PG_UINT32_MAX && last == PG_UINT32_MAX &&
- v.type != jbvBinary)) /* leaves only requested */
+ !JsonItemIsBinary(&v))) /* leaves only requested */
{
/* check expression */
if (jsp)
@@ -1558,14 +1904,14 @@ recursiveAny(JsonPathExecContext *cxt, JsonPathItem *jsp, JsonbContainer *jbc,
break;
}
else if (found)
- JsonValueListAppend(found, copyJsonbValue(&v));
+ JsonValueListAppend(found, copyJsonItem(&v));
else
return jperOk;
}
- if (level < last && v.type == jbvBinary)
+ if (level < last && JsonItemIsBinary(&v))
{
- res = recursiveAny(cxt, jsp, v.val.binary.data, found,
+ res = recursiveAny(cxt, jsp, JsonItemBinary(&v).data, found,
level + 1, first, last,
ignoreStructuralErrors, unwrapNext);
@@ -1593,7 +1939,7 @@ recursiveAny(JsonPathExecContext *cxt, JsonPathItem *jsp, JsonbContainer *jbc,
*/
static JsonPathBool
executePredicate(JsonPathExecContext *cxt, JsonPathItem *pred,
- JsonPathItem *larg, JsonPathItem *rarg, JsonbValue *jb,
+ JsonPathItem *larg, JsonPathItem *rarg, JsonItem *jb,
bool unwrapRightArg, JsonPathPredicateCallback exec,
void *param)
{
@@ -1601,7 +1947,7 @@ executePredicate(JsonPathExecContext *cxt, JsonPathItem *pred,
JsonValueListIterator lseqit;
JsonValueList lseq = {0};
JsonValueList rseq = {0};
- JsonbValue *lval;
+ JsonItem *lval;
bool error = false;
bool found = false;
@@ -1623,7 +1969,7 @@ executePredicate(JsonPathExecContext *cxt, JsonPathItem *pred,
while ((lval = JsonValueListNext(&lseq, &lseqit)))
{
JsonValueListIterator rseqit;
- JsonbValue *rval;
+ JsonItem *rval;
bool first = true;
if (rarg)
@@ -1633,6 +1979,7 @@ executePredicate(JsonPathExecContext *cxt, JsonPathItem *pred,
}
else
{
+ rseqit.next = NULL;
rval = NULL;
}
@@ -1677,15 +2024,15 @@ executePredicate(JsonPathExecContext *cxt, JsonPathItem *pred,
*/
static JsonPathExecResult
executeBinaryArithmExpr(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, BinaryArithmFunc func,
+ JsonItem *jb, BinaryArithmFunc func,
JsonValueList *found)
{
JsonPathExecResult jper;
JsonPathItem elem;
JsonValueList lseq = {0};
JsonValueList rseq = {0};
- JsonbValue *lval;
- JsonbValue *rval;
+ JsonItem *lval;
+ JsonItem *rval;
Numeric res;
jspGetLeftArg(jsp, &elem);
@@ -1724,13 +2071,13 @@ executeBinaryArithmExpr(JsonPathExecContext *cxt, JsonPathItem *jsp,
if (jspThrowErrors(cxt))
{
- res = func(lval->val.numeric, rval->val.numeric, NULL);
+ res = func(JsonItemNumeric(lval), JsonItemNumeric(rval), NULL);
}
else
{
bool error = false;
- res = func(lval->val.numeric, rval->val.numeric, &error);
+ res = func(JsonItemNumeric(lval), JsonItemNumeric(rval), &error);
if (error)
return jperError;
@@ -1740,8 +2087,7 @@ executeBinaryArithmExpr(JsonPathExecContext *cxt, JsonPathItem *jsp,
return jperOk;
lval = palloc(sizeof(*lval));
- lval->type = jbvNumeric;
- lval->val.numeric = res;
+ JsonItemInitNumeric(lval, res);
return recursiveExecuteNext(cxt, jsp, &elem, lval, found, false);
}
@@ -1752,14 +2098,14 @@ executeBinaryArithmExpr(JsonPathExecContext *cxt, JsonPathItem *jsp,
*/
static JsonPathExecResult
executeUnaryArithmExpr(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, PGFunction func, JsonValueList *found)
+ JsonItem *jb, PGFunction func, JsonValueList *found)
{
JsonPathExecResult jper;
JsonPathExecResult jper2;
JsonPathItem elem;
JsonValueList seq = {0};
JsonValueListIterator it;
- JsonbValue *val;
+ JsonItem *val;
bool hasNext;
jspGetArg(jsp, &elem);
@@ -1794,9 +2140,9 @@ executeUnaryArithmExpr(JsonPathExecContext *cxt, JsonPathItem *jsp,
}
if (func)
- val->val.numeric =
+ JsonItemNumeric(val) =
DatumGetNumeric(DirectFunctionCall1(func,
- NumericGetDatum(val->val.numeric)));
+ NumericGetDatum(JsonItemNumeric(val))));
jper2 = recursiveExecuteNext(cxt, jsp, &elem, val, found, false);
@@ -1820,7 +2166,7 @@ executeUnaryArithmExpr(JsonPathExecContext *cxt, JsonPathItem *jsp,
* Check if the 'whole' string starts from 'initial' string.
*/
static JsonPathBool
-executeStartsWith(JsonPathItem *jsp, JsonbValue *whole, JsonbValue *initial,
+executeStartsWith(JsonPathItem *jsp, JsonItem *whole, JsonItem *initial,
void *param)
{
if (!(whole = getScalar(whole, jbvString)))
@@ -1829,10 +2175,10 @@ executeStartsWith(JsonPathItem *jsp, JsonbValue *whole, JsonbValue *initial,
if (!(initial = getScalar(initial, jbvString)))
return jpbUnknown; /* error */
- if (whole->val.string.len >= initial->val.string.len &&
- !memcmp(whole->val.string.val,
- initial->val.string.val,
- initial->val.string.len))
+ if (JsonItemString(whole).len >= JsonItemString(initial).len &&
+ !memcmp(JsonItemString(whole).val,
+ JsonItemString(initial).val,
+ JsonItemString(initial).len))
return jpbTrue;
return jpbFalse;
@@ -1844,7 +2190,7 @@ executeStartsWith(JsonPathItem *jsp, JsonbValue *whole, JsonbValue *initial,
* Check if the string matches regex pattern.
*/
static JsonPathBool
-executeLikeRegex(JsonPathItem *jsp, JsonbValue *str, JsonbValue *rarg,
+executeLikeRegex(JsonPathItem *jsp, JsonItem *str, JsonItem *rarg,
void *param)
{
JsonLikeRegexContext *cxt = param;
@@ -1874,8 +2220,8 @@ executeLikeRegex(JsonPathItem *jsp, JsonbValue *str, JsonbValue *rarg,
cxt->cflags |= REG_EXPANDED;
}
- if (RE_compile_and_execute(cxt->regex, str->val.string.val,
- str->val.string.len,
+ if (RE_compile_and_execute(cxt->regex, JsonItemString(str).val,
+ JsonItemString(str).len,
cxt->cflags, DEFAULT_COLLATION_OID, 0, NULL))
return jpbTrue;
@@ -1884,7 +2230,7 @@ executeLikeRegex(JsonPathItem *jsp, JsonbValue *str, JsonbValue *rarg,
static JsonPathExecResult
executeNumericItemMethod(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, bool unwrap, PGFunction func,
+ JsonItem *jb, bool unwrap, PGFunction func,
JsonValueList *found)
{
JsonPathItem next;
@@ -1901,15 +2247,14 @@ executeNumericItemMethod(JsonPathExecContext *cxt, JsonPathItem *jsp,
"not a numeric value",
jspOperationName(jsp->type)))));
- datum = NumericGetDatum(jb->val.numeric);
+ datum = NumericGetDatum(JsonItemNumeric(jb));
datum = DirectFunctionCall1(func, datum);
if (!jspGetNext(jsp, &next) && !found)
return jperOk;
jb = palloc(sizeof(*jb));
- jb->type = jbvNumeric;
- jb->val.numeric = DatumGetNumeric(datum);
+ JsonItemInitNumericDatum(jb, datum);
return recursiveExecuteNext(cxt, jsp, &next, jb, found, false);
}
@@ -1939,7 +2284,7 @@ executeNumericItemMethod(JsonPathExecContext *cxt, JsonPathItem *jsp,
*/
static JsonPathExecResult
executeKeyValueMethod(JsonPathExecContext *cxt, JsonPathItem *jsp,
- JsonbValue *jb, JsonValueList *found)
+ JsonItem *jb, JsonValueList *found)
{
JsonPathExecResult res = jperNotFound;
JsonPathItem next;
@@ -1950,12 +2295,13 @@ executeKeyValueMethod(JsonPathExecContext *cxt, JsonPathItem *jsp,
JsonbValue keystr;
JsonbValue valstr;
JsonbValue idstr;
- JsonbIterator *it;
+ JsonxIterator it;
JsonbIteratorToken tok;
+ JsonBuilderFunc push;
int64 id;
bool hasNext;
- if (JsonbType(jb) != jbvObject || jb->type != jbvBinary)
+ if (JsonbType(jb) != jbvObject || !JsonItemIsBinary(jb))
RETURN_ERROR(ereport(ERROR,
(errcode(ERRCODE_JSON_OBJECT_NOT_FOUND),
errmsg(ERRMSG_JSON_OBJECT_NOT_FOUND),
@@ -1963,7 +2309,7 @@ executeKeyValueMethod(JsonPathExecContext *cxt, JsonPathItem *jsp,
"to not an object",
jspOperationName(jsp->type)))));
- jbc = jb->val.binary.data;
+ jbc = JsonItemBinary(jb).data;
if (!JsonContainerSize(jbc))
return jperNotFound; /* no key-value pairs */
@@ -1983,23 +2329,29 @@ executeKeyValueMethod(JsonPathExecContext *cxt, JsonPathItem *jsp,
idstr.val.string.len = 2;
/* construct object id from its base object and offset inside that */
- id = jb->type != jbvBinary ? 0 :
- (int64) ((char *) jbc - (char *) cxt->baseObject.jbc);
+ id = cxt->isJsonb ?
+ (int64) ((char *)(JsonContainer *) jbc -
+ (char *)(JsonContainer *) cxt->baseObject.jbc) :
+ (int64) (((JsonContainer *) jbc)->data -
+ ((JsonContainer *) cxt->baseObject.jbc)->data);
+
id += (int64) cxt->baseObject.id * INT64CONST(10000000000);
idval.type = jbvNumeric;
idval.val.numeric = DatumGetNumeric(DirectFunctionCall1(int8_numeric,
Int64GetDatum(id)));
- it = JsonbIteratorInit(jbc);
+ push = cxt->isJsonb ? pushJsonbValue : pushJsonValue;
- while ((tok = JsonbIteratorNext(&it, &key, true)) != WJB_DONE)
+ JsonxIteratorInit(&it, jbc, cxt->isJsonb);
+
+ while ((tok = JsonxIteratorNext(&it, &key, true)) != WJB_DONE)
{
JsonBaseObjectInfo baseObject;
- JsonbValue obj;
+ JsonItem obj;
JsonbParseState *ps;
JsonbValue *keyval;
- Jsonb *jsonb;
+ Jsonx *jsonx;
if (tok != WJB_KEY)
continue;
@@ -2009,26 +2361,29 @@ executeKeyValueMethod(JsonPathExecContext *cxt, JsonPathItem *jsp,
if (!hasNext && !found)
break;
- tok = JsonbIteratorNext(&it, &val, true);
+ tok = JsonxIteratorNext(&it, &val, true);
Assert(tok == WJB_VALUE);
ps = NULL;
- pushJsonbValue(&ps, WJB_BEGIN_OBJECT, NULL);
+ push(&ps, WJB_BEGIN_OBJECT, NULL);
pushJsonbValue(&ps, WJB_KEY, &keystr);
pushJsonbValue(&ps, WJB_VALUE, &key);
pushJsonbValue(&ps, WJB_KEY, &valstr);
- pushJsonbValue(&ps, WJB_VALUE, &val);
+ push(&ps, WJB_VALUE, &val);
pushJsonbValue(&ps, WJB_KEY, &idstr);
pushJsonbValue(&ps, WJB_VALUE, &idval);
keyval = pushJsonbValue(&ps, WJB_END_OBJECT, NULL);
- jsonb = JsonbValueToJsonb(keyval);
+ jsonx = JsonbValueToJsonx(keyval, cxt->isJsonb);
- JsonbInitBinary(&obj, jsonb);
+ if (cxt->isJsonb)
+ JsonbInitBinary(JsonItemJbv(&obj), &jsonx->jb);
+ else
+ JsonInitBinary(JsonItemJbv(&obj), &jsonx->js);
baseObject = setBaseObject(cxt, &obj, cxt->lastGeneratedObjectId++);
@@ -2055,22 +2410,17 @@ appendBoolResult(JsonPathExecContext *cxt, JsonPathItem *jsp,
JsonValueList *found, JsonPathBool res)
{
JsonPathItem next;
- JsonbValue jbv;
+ JsonItem jsi;
if (!jspGetNext(jsp, &next) && !found)
return jperOk; /* found singleton boolean value */
if (res == jpbUnknown)
- {
- jbv.type = jbvNull;
- }
+ JsonItemInitNull(&jsi);
else
- {
- jbv.type = jbvBool;
- jbv.val.boolean = res == jpbTrue;
- }
+ JsonItemInitBool(&jsi, res == jpbTrue);
- return recursiveExecuteNext(cxt, jsp, &next, &jbv, found, true);
+ return recursiveExecuteNext(cxt, jsp, &next, &jsi, found, true);
}
/*
@@ -2080,28 +2430,29 @@ appendBoolResult(JsonPathExecContext *cxt, JsonPathItem *jsp,
*/
static void
getJsonPathItem(JsonPathExecContext *cxt, JsonPathItem *item,
- JsonbValue *value)
+ JsonItem *value)
{
switch (item->type)
{
case jpiNull:
- value->type = jbvNull;
+ JsonItemInitNull(value);
break;
case jpiBool:
- value->type = jbvBool;
- value->val.boolean = jspGetBool(item);
+ JsonItemInitBool(value, jspGetBool(item));
break;
case jpiNumeric:
- value->type = jbvNumeric;
- value->val.numeric = jspGetNumeric(item);
+ JsonItemInitNumeric(value, jspGetNumeric(item));
break;
case jpiString:
- value->type = jbvString;
- value->val.string.val = jspGetString(item,
- &value->val.string.len);
- break;
+ {
+ int len;
+ char *str = jspGetString(item, &len);
+
+ JsonItemInitString(value, str, len);
+ break;
+ }
case jpiVariable:
- getJsonPathVariable(cxt, item, cxt->vars, value);
+ getJsonPathVariable(cxt, item, value);
return;
default:
elog(ERROR, "unexpected jsonpath item type");
@@ -2113,42 +2464,86 @@ getJsonPathItem(JsonPathExecContext *cxt, JsonPathItem *item,
*/
static void
getJsonPathVariable(JsonPathExecContext *cxt, JsonPathItem *variable,
- Jsonb *vars, JsonbValue *value)
+ JsonItem *value)
{
char *varName;
int varNameLength;
- JsonbValue tmp;
- JsonbValue *v;
+ JsonItem baseObject;
+ int baseObjectId;
- if (!vars)
+ Assert(variable->type == jpiVariable);
+ varName = jspGetString(variable, &varNameLength);
+
+ if (!cxt->vars ||
+ (baseObjectId = cxt->getVar(cxt->vars, cxt->isJsonb,
+ varName, varNameLength,
+ value, JsonItemJbv(&baseObject))) < 0)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_OBJECT),
+ errmsg("cannot find jsonpath variable '%s'",
+ pnstrdup(varName, varNameLength))));
+
+ if (baseObjectId > 0)
+ setBaseObject(cxt, &baseObject, baseObjectId);
+}
+
+static int
+getJsonPathVariableFromJsonx(void *varsJsonx, bool isJsonb,
+ char *varName, int varNameLength,
+ JsonItem *value, JsonbValue *baseObject)
+{
+ Jsonx *vars = varsJsonx;
+ JsonbValue *val;
+ JsonbValue key;
+
+ if (!varName)
{
- value->type = jbvNull;
- return;
+ if (vars &&
+ !(isJsonb ?
+ JsonContainerIsObject(&vars->jb.root) :
+ JsonContainerIsObject(&vars->js.root)))
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+ errmsg("%s containing jsonpath variables is not an object",
+ isJsonb ? "jsonb" : "json")));
+
+ return vars ? 1 : 0; /* count of base objects */
}
- Assert(variable->type == jpiVariable);
- varName = jspGetString(variable, &varNameLength);
- tmp.type = jbvString;
- tmp.val.string.val = varName;
- tmp.val.string.len = varNameLength;
+ if (!vars)
+ return -1;
- v = findJsonbValueFromContainer(&vars->root, JB_FOBJECT, &tmp);
+ key.type = jbvString;
+ key.val.string.val = varName;
+ key.val.string.len = varNameLength;
- if (v)
+ if (isJsonb)
{
- *value = *v;
- pfree(v);
+ Jsonb *jb = &vars->jb;
+
+ val = findJsonbValueFromContainer(&jb->root, JB_FOBJECT, &key);
+
+ if (!val)
+ return -1;
+
+ JsonbInitBinary(baseObject, jb);
}
else
{
- ereport(ERROR,
- (errcode(ERRCODE_UNDEFINED_OBJECT),
- errmsg("cannot find jsonpath variable '%s'",
- pnstrdup(varName, varNameLength))));
+ Json *js = &vars->js;
+
+ val = findJsonValueFromContainer(&js->root, JB_FOBJECT, &key);
+
+ if (!val)
+ return -1;
+
+ JsonInitBinary(baseObject, js);
}
- JsonbInitBinary(&tmp, vars);
- setBaseObject(cxt, &tmp, 1);
+ *JsonItemJbv(value) = *val;
+ pfree(val);
+
+ return 1;
}
/**************** Support functions for JsonPath execution *****************/
@@ -2157,16 +2552,26 @@ getJsonPathVariable(JsonPathExecContext *cxt, JsonPathItem *variable,
* Returns the size of an array item, or -1 if item is not an array.
*/
static int
-JsonbArraySize(JsonbValue *jb)
+JsonxArraySize(JsonItem *jb, bool isJsonb)
{
- Assert(jb->type != jbvArray);
+ Assert(!JsonItemIsArray(jb));
- if (jb->type == jbvBinary)
+ if (JsonItemIsBinary(jb))
{
- JsonbContainer *jbc = jb->val.binary.data;
+ if (isJsonb)
+ {
+ JsonbContainer *jbc = JsonItemBinary(jb).data;
- if (JsonContainerIsArray(jbc) && !JsonContainerIsScalar(jbc))
- return JsonContainerSize(jbc);
+ if (JsonContainerIsArray(jbc) && !JsonContainerIsScalar(jbc))
+ return JsonContainerSize(jbc);
+ }
+ else
+ {
+ JsonContainer *jc = (JsonContainer *) JsonItemBinary(jb).data;
+
+ if (JsonContainerIsArray(jc) && !JsonContainerIsScalar(jc))
+ return JsonTextContainerSize(jc);
+ }
}
return -1;
@@ -2174,7 +2579,7 @@ JsonbArraySize(JsonbValue *jb)
/* Comparison predicate callback. */
static JsonPathBool
-executeComparison(JsonPathItem *cmp, JsonbValue *lv, JsonbValue *rv, void *p)
+executeComparison(JsonPathItem *cmp, JsonItem *lv, JsonItem *rv, void *p)
{
return compareItems(cmp->type, lv, rv);
}
@@ -2183,14 +2588,16 @@ executeComparison(JsonPathItem *cmp, JsonbValue *lv, JsonbValue *rv, void *p)
* Compare two SQL/JSON items using comparison operation 'op'.
*/
static JsonPathBool
-compareItems(int32 op, JsonbValue *jb1, JsonbValue *jb2)
+compareItems(int32 op, JsonItem *jsi1, JsonItem *jsi2)
{
+ JsonbValue *jb1 = JsonItemJbv(jsi1);
+ JsonbValue *jb2 = JsonItemJbv(jsi2);
int cmp;
bool res;
- if (jb1->type != jb2->type)
+ if (JsonItemGetType(jsi1) != JsonItemGetType(jsi2))
{
- if (jb1->type == jbvNull || jb2->type == jbvNull)
+ if (JsonItemIsNull(jsi1) || JsonItemIsNull(jsi2))
/*
* Equality and order comparison of nulls to non-nulls returns
@@ -2202,7 +2609,7 @@ compareItems(int32 op, JsonbValue *jb1, JsonbValue *jb2)
return jpbUnknown;
}
- switch (jb1->type)
+ switch (JsonItemGetType(jsi1))
{
case jbvNull:
cmp = 0;
@@ -2225,16 +2632,16 @@ compareItems(int32 op, JsonbValue *jb1, JsonbValue *jb2)
jb2->val.string.val, jb2->val.string.len,
DEFAULT_COLLATION_OID);
break;
- case jbvDatetime:
+ case jsiDatetime:
{
bool error = false;
- cmp = compareDatetime(jb1->val.datetime.value,
- jb1->val.datetime.typid,
- jb1->val.datetime.tz,
- jb2->val.datetime.value,
- jb2->val.datetime.typid,
- jb2->val.datetime.tz,
+ cmp = compareDatetime(JsonItemDatetime(jsi1).value,
+ JsonItemDatetime(jsi1).typid,
+ JsonItemDatetime(jsi1).tz,
+ JsonItemDatetime(jsi2).value,
+ JsonItemDatetime(jsi2).typid,
+ JsonItemDatetime(jsi2).tz,
&error);
if (error)
@@ -2248,7 +2655,7 @@ compareItems(int32 op, JsonbValue *jb1, JsonbValue *jb2)
return jpbUnknown; /* non-scalars are not comparable */
default:
- elog(ERROR, "invalid jsonb value type %d", jb1->type);
+ elog(ERROR, "invalid jsonb value type %d", JsonItemGetType(jsi1));
}
switch (op)
@@ -2288,25 +2695,88 @@ compareNumeric(Numeric a, Numeric b)
PointerGetDatum(b)));
}
-static JsonbValue *
-copyJsonbValue(JsonbValue *src)
+static JsonItem *
+copyJsonItem(JsonItem *src)
{
- JsonbValue *dst = palloc(sizeof(*dst));
+ JsonItem *dst = palloc(sizeof(*dst));
*dst = *src;
return dst;
}
+static JsonItem *
+JsonbValueToJsonItem(JsonbValue *jbv, JsonItem *jsi)
+{
+ *JsonItemJbv(jsi) = *jbv;
+ return jsi;
+}
+
+static JsonbValue *
+JsonItemToJsonbValue(JsonItem *jsi, JsonbValue *jbv)
+{
+ switch (JsonItemGetType(jsi))
+ {
+ case jsiDatetime:
+ jbv->type = jbvString;
+ jbv->val.string.val = JsonEncodeDateTime(NULL,
+ JsonItemDatetime(jsi).value,
+ JsonItemDatetime(jsi).typid,
+ &JsonItemDatetime(jsi).tz);
+ jbv->val.string.len = strlen(jbv->val.string.val);
+ return jbv;
+
+ default:
+ return JsonItemJbv(jsi);
+ }
+}
+
+static Jsonb *
+JsonItemToJsonb(JsonItem *jsi)
+{
+ JsonbValue jbv;
+
+ return JsonbValueToJsonb(JsonItemToJsonbValue(jsi, &jbv));
+}
+
+static const char *
+JsonItemTypeName(JsonItem *jsi)
+{
+ switch (JsonItemGetType(jsi))
+ {
+ case jsiDatetime:
+ switch (JsonItemDatetime(jsi).typid)
+ {
+ case DATEOID:
+ return "date";
+ case TIMEOID:
+ return "time without time zone";
+ case TIMETZOID:
+ return "time with time zone";
+ case TIMESTAMPOID:
+ return "timestamp without time zone";
+ case TIMESTAMPTZOID:
+ return "timestamp with time zone";
+ default:
+ elog(ERROR, "unrecognized jsonb value datetime type: %d",
+ JsonItemDatetime(jsi).typid);
+ return "unknown";
+ }
+
+ default:
+ return JsonbTypeName(JsonItemJbv(jsi));
+ }
+}
+
/*
* Execute array subscript expression and convert resulting numeric item to
* the integer type with truncation.
*/
static JsonPathExecResult
-getArrayIndex(JsonPathExecContext *cxt, JsonPathItem *jsp, JsonbValue *jb,
+getArrayIndex(JsonPathExecContext *cxt, JsonPathItem *jsp, JsonItem *jb,
int32 *index)
{
- JsonbValue *jbv;
+ JsonItem *jbv;
JsonValueList found = {0};
JsonPathExecResult res = recursiveExecute(cxt, jsp, jb, &found);
Datum numeric_index;
@@ -2324,7 +2794,7 @@ getArrayIndex(JsonPathExecContext *cxt, JsonPathItem *jsp, JsonbValue *jb,
"singleton numeric value"))));
numeric_index = DirectFunctionCall2(numeric_trunc,
- NumericGetDatum(jbv->val.numeric),
+ NumericGetDatum(JsonItemNumeric(jbv)),
Int32GetDatum(0));
*index = numeric_int4_error(DatumGetNumeric(numeric_index), &error);
@@ -2341,95 +2811,82 @@ getArrayIndex(JsonPathExecContext *cxt, JsonPathItem *jsp, JsonbValue *jb,
/* Save base object and its id needed for the execution of .keyvalue(). */
static JsonBaseObjectInfo
-setBaseObject(JsonPathExecContext *cxt, JsonbValue *jbv, int32 id)
+setBaseObject(JsonPathExecContext *cxt, JsonItem *jbv, int32 id)
{
JsonBaseObjectInfo baseObject = cxt->baseObject;
- cxt->baseObject.jbc = jbv->type != jbvBinary ? NULL :
- (JsonbContainer *) jbv->val.binary.data;
+ cxt->baseObject.jbc = !JsonItemIsBinary(jbv) ? NULL :
+ (JsonbContainer *) JsonItemBinary(jbv).data;
cxt->baseObject.id = id;
return baseObject;
}
static void
-JsonValueListAppend(JsonValueList *jvl, JsonbValue *jbv)
+JsonValueListAppend(JsonValueList *jvl, JsonItem *jsi)
{
- if (jvl->singleton)
+ jsi->next = NULL;
+
+ if (jvl->tail)
{
- jvl->list = list_make2(jvl->singleton, jbv);
- jvl->singleton = NULL;
+ jvl->tail->next = jsi;
+ jvl->tail = jsi;
}
- else if (!jvl->list)
- jvl->singleton = jbv;
else
- jvl->list = lappend(jvl->list, jbv);
+ {
+ Assert(!jvl->head);
+ jvl->head = jvl->tail = jsi;
+ }
+
+ jvl->length++;
}
static int
JsonValueListLength(const JsonValueList *jvl)
{
- return jvl->singleton ? 1 : list_length(jvl->list);
+ return jvl->length;
}
static bool
JsonValueListIsEmpty(JsonValueList *jvl)
{
- return !jvl->singleton && list_length(jvl->list) <= 0;
+ return !jvl->length;
}
-static JsonbValue *
+static JsonItem *
JsonValueListHead(JsonValueList *jvl)
{
- return jvl->singleton ? jvl->singleton : linitial(jvl->list);
+ return jvl->head;
}
static List *
JsonValueListGetList(JsonValueList *jvl)
{
- if (jvl->singleton)
- return list_make1(jvl->singleton);
+ List *list = NIL;
+ JsonItem *jsi;
- return jvl->list;
+ for (jsi = jvl->head; jsi; jsi = jsi->next)
+ list = lappend(list, jsi);
+
+ return list;
}
static void
JsonValueListInitIterator(const JsonValueList *jvl, JsonValueListIterator *it)
{
- if (jvl->singleton)
- {
- it->value = jvl->singleton;
- it->next = NULL;
- }
- else if (list_head(jvl->list) != NULL)
- {
- it->value = (JsonbValue *) linitial(jvl->list);
- it->next = lnext(list_head(jvl->list));
- }
- else
- {
- it->value = NULL;
- it->next = NULL;
- }
+ it->next = jvl->head;
}
/*
* Get the next item from the sequence advancing iterator.
*/
-static JsonbValue *
+static JsonItem *
JsonValueListNext(const JsonValueList *jvl, JsonValueListIterator *it)
{
- JsonbValue *result = it->value;
+ JsonItem *result = it->next;
- if (it->next)
- {
- it->value = lfirst(it->next);
- it->next = lnext(it->next);
- }
- else
- {
- it->value = NULL;
- }
+ if (result)
+ it->next = result->next;
return result;
}
@@ -2448,16 +2905,29 @@ JsonbInitBinary(JsonbValue *jbv, Jsonb *jb)
}
/*
+ * Initialize a binary JsonbValue with the given json container.
+ */
+static inline JsonbValue *
+JsonInitBinary(JsonbValue *jbv, Json *js)
+{
+ jbv->type = jbvBinary;
+ jbv->val.binary.data = (void *) &js->root;
+ jbv->val.binary.len = js->root.len;
+
+ return jbv;
+}
+
+/*
* Returns jbv* type of of JsonbValue. Note, it never returns jbvBinary as is.
*/
static int
-JsonbType(JsonbValue *jb)
+JsonbType(JsonItem *jb)
{
- int type = jb->type;
+ int type = JsonItemGetType(jb);
- if (jb->type == jbvBinary)
+ if (type == jbvBinary)
{
- JsonbContainer *jbc = (void *) jb->val.binary.data;
+ JsonbContainer *jbc = (void *) JsonItemBinary(jb).data;
/* Scalars should be always extracted during jsonpath execution. */
Assert(!JsonContainerIsScalar(jbc));
@@ -2473,32 +2943,204 @@ JsonbType(JsonbValue *jb)
return type;
}
+/*
+ * Convert jsonb to a C-string stripping quotes from scalar strings.
+ */
+static char *
+JsonbValueUnquote(JsonbValue *jbv, int *len, bool isJsonb)
+{
+ switch (jbv->type)
+ {
+ case jbvString:
+ *len = jbv->val.string.len;
+ return jbv->val.string.val;
+
+ case jbvBool:
+ *len = jbv->val.boolean ? 4 : 5;
+ return jbv->val.boolean ? "true" : "false";
+
+ case jbvNumeric:
+ *len = -1;
+ return DatumGetCString(DirectFunctionCall1(numeric_out,
+ PointerGetDatum(jbv->val.numeric)));
+
+ case jbvNull:
+ *len = 4;
+ return "null";
+
+ case jbvBinary:
+ {
+ JsonbValue jbvbuf;
+
+ if (isJsonb ?
+ JsonbExtractScalar(jbv->val.binary.data, &jbvbuf) :
+ JsonExtractScalar((JsonContainer *) jbv->val.binary.data, &jbvbuf))
+ return JsonbValueUnquote(&jbvbuf, len, isJsonb);
+
+ *len = -1;
+ return isJsonb ?
+ JsonbToCString(NULL, jbv->val.binary.data, jbv->val.binary.len) :
+ JsonToCString(NULL, (JsonContainer *) jbv->val.binary.data, jbv->val.binary.len);
+ }
+
+ default:
+ elog(ERROR, "unexpected jsonb value type: %d", jbv->type);
+ return NULL;
+ }
+}
+
+static char *
+JsonItemUnquote(JsonItem *jsi, int *len, bool isJsonb)
+{
+ switch (JsonItemGetType(jsi))
+ {
+ case jsiDatetime:
+ *len = -1;
+ return JsonEncodeDateTime(NULL,
+ JsonItemDatetime(jsi).value,
+ JsonItemDatetime(jsi).typid,
+ &JsonItemDatetime(jsi).tz);
+
+ default:
+ return JsonbValueUnquote(JsonItemJbv(jsi), len, isJsonb);
+ }
+}
+
+static text *
+JsonItemUnquoteText(JsonItem *jsi, bool isJsonb)
+{
+ int len;
+ char *str = JsonItemUnquote(jsi, &len, isJsonb);
+
+ if (len < 0)
+ return cstring_to_text(str);
+ else
+ return cstring_to_text_with_len(str, len);
+}
+
+static JsonItem *
+getJsonObjectKey(JsonItem *jsi, char *keystr, int keylen, bool isJsonb,
+ JsonItem *res)
+{
+ JsonbContainer *jbc = JsonItemBinary(jsi).data;
+ JsonbValue *val;
+ JsonbValue key;
+
+ key.type = jbvString;
+ key.val.string.val = keystr;
+ key.val.string.len = keylen;
+
+ val = isJsonb ?
+ findJsonbValueFromContainer(jbc, JB_FOBJECT, &key) :
+ findJsonValueFromContainer((JsonContainer *) jbc, JB_FOBJECT, &key);
+
+ return val ? JsonbValueToJsonItem(val, res) : NULL;
+}
+
+static JsonItem *
+getJsonArrayElement(JsonItem *jb, uint32 index, bool isJsonb, JsonItem *res)
+{
+ JsonbContainer *jbc = JsonItemBinary(jb).data;
+ JsonbValue *elem = isJsonb ?
+ getIthJsonbValueFromContainer(jbc, index) :
+ getIthJsonValueFromContainer((JsonContainer *) jbc, index);
+
+ return elem ? JsonbValueToJsonItem(elem, res) : NULL;
+}
+
+static inline void
+JsonxIteratorInit(JsonxIterator *it, JsonxContainer *jxc, bool isJsonb)
+{
+ it->isJsonb = isJsonb;
+ if (isJsonb)
+ it->it.jb = JsonbIteratorInit((JsonbContainer *) jxc);
+ else
+ it->it.js = JsonIteratorInit((JsonContainer *) jxc);
+}
+
+static JsonbIteratorToken
+JsonxIteratorNext(JsonxIterator *it, JsonbValue *jbv, bool skipNested)
+{
+ return it->isJsonb ?
+ JsonbIteratorNext(&it->it.jb, jbv, skipNested) :
+ JsonIteratorNext(&it->it.js, jbv, skipNested);
+}
+
+static Json *
+JsonItemToJson(JsonItem *jsi)
+{
+ JsonbValue jbv;
+
+ return JsonbValueToJson(JsonItemToJsonbValue(jsi, &jbv));
+}
+
+static Jsonx *
+JsonbValueToJsonx(JsonbValue *jbv, bool isJsonb)
+{
+ return isJsonb ?
+ (Jsonx *) JsonbValueToJsonb(jbv) :
+ (Jsonx *) JsonbValueToJson(jbv);
+}
+
+static Datum
+JsonbValueToJsonxDatum(JsonbValue *jbv, bool isJsonb)
+{
+ return isJsonb ?
+ JsonbPGetDatum(JsonbValueToJsonb(jbv)) :
+ JsonPGetDatum(JsonbValueToJson(jbv));
+}
+
+static Datum
+JsonItemToJsonxDatum(JsonItem *jsi, bool isJsonb)
+{
+ JsonbValue jbv;
+
+ return JsonbValueToJsonxDatum(JsonItemToJsonbValue(jsi, &jbv), isJsonb);
+}
+
/* Get scalar of given type or NULL on type mismatch */
-static JsonbValue *
-getScalar(JsonbValue *scalar, enum jbvType type)
+static JsonItem *
+getScalar(JsonItem *scalar, enum jbvType type)
{
/* Scalars should be always extracted during jsonpath execution. */
- Assert(scalar->type != jbvBinary ||
- !JsonContainerIsScalar(scalar->val.binary.data));
+ Assert(!JsonItemIsBinary(scalar) ||
+ !JsonContainerIsScalar(JsonItemBinary(scalar).data));
- return scalar->type == type ? scalar : NULL;
+ return JsonItemGetType(scalar) == type ? scalar : NULL;
}
/* Construct a JSON array from the item list */
static JsonbValue *
-wrapItemsInArray(const JsonValueList *items)
+wrapItemsInArray(const JsonValueList *items, bool isJsonb)
{
JsonbParseState *ps = NULL;
JsonValueListIterator it;
- JsonbValue *jbv;
+ JsonItem *jsi;
+ JsonbValue jbv;
+ JsonBuilderFunc push = isJsonb ? pushJsonbValue : pushJsonValue;
- pushJsonbValue(&ps, WJB_BEGIN_ARRAY, NULL);
+ push(&ps, WJB_BEGIN_ARRAY, NULL);
JsonValueListInitIterator(items, &it);
- while ((jbv = JsonValueListNext(items, &it)))
- pushJsonbValue(&ps, WJB_ELEM, jbv);
- return pushJsonbValue(&ps, WJB_END_ARRAY, NULL);
+ while ((jsi = JsonValueListNext(items, &it)))
+ push(&ps, WJB_ELEM, JsonItemToJsonbValue(jsi, &jbv));
+
+ return push(&ps, WJB_END_ARRAY, NULL);
+}
+
+static void
+pushJsonItem(JsonItemStack *stack, JsonItemStackEntry *entry, JsonItem *item)
+{
+ entry->item = item;
+ entry->parent = *stack;
+ *stack = entry;
+}
+
+static void
+popJsonItem(JsonItemStack *stack)
+{
+ *stack = (*stack)->parent;
}
static inline Datum
@@ -2739,3 +3381,43 @@ tryToParseDatetime(text *fmt, text *datetime, char *tzname, bool strict,
return !error;
}
+
+static void
+JsonItemInitNull(JsonItem *item)
+{
+ item->val.type = jbvNull;
+}
+
+static void
+JsonItemInitBool(JsonItem *item, bool val)
+{
+ item->val.type = jbvBool;
+ JsonItemBool(item) = val;
+}
+
+static void
+JsonItemInitNumeric(JsonItem *item, Numeric val)
+{
+ item->val.type = jbvNumeric;
+ JsonItemNumeric(item) = val;
+}
+
+#define JsonItemInitNumericDatum(item, val) JsonItemInitNumeric(item, DatumGetNumeric(val))
+
+static void
+JsonItemInitString(JsonItem *item, char *str, int len)
+{
+ item->val.type = jbvString;
+ JsonItemString(item).val = str;
+ JsonItemString(item).len = len;
+}
+
+static void
+JsonItemInitDatetime(JsonItem *item, Datum val, Oid typid, int32 typmod, int tz)
+{
+ item->val.type = jsiDatetime;
+ JsonItemDatetime(item).value = val;
+ JsonItemDatetime(item).typid = typid;
+ JsonItemDatetime(item).typmod = typmod;
+ JsonItemDatetime(item).tz = tz;
+}
diff --git a/src/include/catalog/pg_operator.dat b/src/include/catalog/pg_operator.dat
index 1c7af92..5e249dd 100644
--- a/src/include/catalog/pg_operator.dat
+++ b/src/include/catalog/pg_operator.dat
@@ -3263,5 +3263,13 @@
oprname => '@@', oprleft => 'jsonb', oprright => 'jsonpath',
oprresult => 'bool', oprcode => 'jsonb_path_match(jsonb,jsonpath)',
oprrest => 'contsel', oprjoin => 'contjoinsel' },
+{ oid => '6071', descr => 'jsonpath exists',
+ oprname => '@?', oprleft => 'json', oprright => 'jsonpath',
+ oprresult => 'bool', oprcode => 'json_path_exists(json,jsonpath)',
+ oprrest => 'contsel', oprjoin => 'contjoinsel' },
+{ oid => '6108', descr => 'jsonpath predicate',
+ oprname => '@@', oprleft => 'json', oprright => 'jsonpath',
+ oprresult => 'bool', oprcode => 'json_path_match(json,jsonpath)',
+ oprrest => 'contsel', oprjoin => 'contjoinsel' },
]
diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat
index 9a7a5db..4f80b95 100644
--- a/src/include/catalog/pg_proc.dat
+++ b/src/include/catalog/pg_proc.dat
@@ -9213,6 +9213,10 @@
{ oid => '6057', descr => 'jsonpath query first item',
proname => 'jsonb_path_query_first', prorettype => 'jsonb',
proargtypes => 'jsonb jsonpath jsonb bool', prosrc => 'jsonb_path_query_first' },
+{ oid => '6127', descr => 'jsonpath query first item text',
+ proname => 'jsonb_path_query_first_text', prorettype => 'text',
+ proargtypes => 'jsonb jsonpath jsonb bool',
+ prosrc => 'jsonb_path_query_first_text' },
{ oid => '6058', descr => 'jsonpath match', proname => 'jsonb_path_match',
prorettype => 'bool', proargtypes => 'jsonb jsonpath jsonb bool',
prosrc => 'jsonb_path_match' },
@@ -9224,6 +9228,36 @@
proname => 'jsonb_path_match', prorettype => 'bool',
proargtypes => 'jsonb jsonpath', prosrc => 'jsonb_path_match_opr' },
+{ oid => '6043', descr => 'implementation of @? operator',
+ proname => 'json_path_exists', prorettype => 'bool',
+ proargtypes => 'json jsonpath', prosrc => 'json_path_exists_opr' },
+{ oid => '6047', descr => 'implementation of @@ operator',
+ proname => 'json_path_match', prorettype => 'bool',
+ proargtypes => 'json jsonpath', prosrc => 'json_path_match_opr' },
+
+{ oid => '6045', descr => 'jsonpath exists test',
+ proname => 'json_path_exists', prorettype => 'bool',
+ proargtypes => 'json jsonpath json bool', prosrc => 'json_path_exists' },
+{ oid => '6046', descr => 'jsonpath query',
+ proname => 'json_path_query', prorows => '1000', proretset => 't',
+ prorettype => 'json', proargtypes => 'json jsonpath json bool',
+ prosrc => 'json_path_query' },
+{ oid => '6129', descr => 'jsonpath query with conditional wrapper',
+ proname => 'json_path_query_array', prorettype => 'json',
+ proargtypes => 'json jsonpath json bool',
+ prosrc => 'json_path_query_array' },
+{ oid => '6069', descr => 'jsonpath match test',
+ proname => 'json_path_match', prorettype => 'bool',
+ proargtypes => 'json jsonpath json bool', prosrc => 'json_path_match' },
+{ oid => '6109', descr => 'jsonpath query first item',
+ proname => 'json_path_query_first', prorettype => 'json',
+ proargtypes => 'json jsonpath json bool',
+ prosrc => 'json_path_query_first' },
+{ oid => '6044', descr => 'jsonpath query first item text',
+ proname => 'json_path_query_first_text', prorettype => 'text',
+ proargtypes => 'json jsonpath json bool',
+ prosrc => 'json_path_query_first_text' },
+
# txid
{ oid => '2939', descr => 'I/O',
proname => 'txid_snapshot_in', prorettype => 'txid_snapshot',
diff --git a/src/include/utils/jsonapi.h b/src/include/utils/jsonapi.h
index 9eda9a3..5535aba2 100644
--- a/src/include/utils/jsonapi.h
+++ b/src/include/utils/jsonapi.h
@@ -15,6 +15,7 @@
#define JSONAPI_H
#include "jsonb.h"
+#include "access/htup.h"
#include "lib/stringinfo.h"
typedef enum
@@ -93,6 +94,56 @@ typedef struct JsonSemAction
json_scalar_action scalar;
} JsonSemAction;
+typedef enum
+{
+ JTI_ARRAY_START,
+ JTI_ARRAY_ELEM,
+ JTI_ARRAY_ELEM_SCALAR,
+ JTI_ARRAY_ELEM_AFTER,
+ JTI_ARRAY_END,
+ JTI_OBJECT_START,
+ JTI_OBJECT_KEY,
+ JTI_OBJECT_VALUE,
+ JTI_OBJECT_VALUE_AFTER,
+} JsontIterState;
+
+typedef struct JsonContainerData
+{
+ uint32 header;
+ int len;
+ char *data;
+} JsonContainerData;
+
+typedef const JsonContainerData JsonContainer;
+
+#define JsonTextContainerSize(jc) \
+ (((jc)->header & JB_CMASK) == JB_CMASK && JsonContainerIsArray(jc) \
+ ? JsonGetArraySize(jc) : (jc)->header & JB_CMASK)
+
+typedef struct Json
+{
+ JsonContainer root;
+} Json;
+
+typedef struct JsonIterator
+{
+ struct JsonIterator *parent;
+ JsonContainer *container;
+ JsonLexContext *lex;
+ JsontIterState state;
+ bool isScalar;
+} JsonIterator;
+
+#define DatumGetJsonP(datum) JsonCreate(DatumGetTextP(datum))
+#define DatumGetJsonPCopy(datum) JsonCreate(DatumGetTextPCopy(datum))
+
+#define JsonPGetDatum(json) \
+ PointerGetDatum(cstring_to_text_with_len((json)->root.data, (json)->root.len))
+
+#define PG_GETARG_JSON_P(n) DatumGetJsonP(PG_GETARG_DATUM(n))
+#define PG_GETARG_JSON_P_COPY(n) DatumGetJsonPCopy(PG_GETARG_DATUM(n))
+#define PG_RETURN_JSON_P(json) PG_RETURN_DATUM(JsonPGetDatum(json))
+
/*
* parse_json will parse the string in the lex calling the
* action functions in sem at the appropriate points. It is
@@ -164,4 +215,22 @@ extern text *transform_json_string_values(text *json, void *action_state,
extern char *JsonEncodeDateTime(char *buf, Datum value, Oid typid,
const int *tzp);
+extern Json *JsonCreate(text *json);
+extern JsonbIteratorToken JsonIteratorNext(JsonIterator **pit, JsonbValue *val,
+ bool skipNested);
+extern JsonIterator *JsonIteratorInit(JsonContainer *jc);
+extern void JsonIteratorFree(JsonIterator *it);
+extern uint32 JsonGetArraySize(JsonContainer *jc);
+extern Json *JsonbValueToJson(JsonbValue *jbv);
+extern bool JsonExtractScalar(JsonContainer *jbc, JsonbValue *res);
+extern char *JsonUnquote(Json *jb);
+extern char *JsonToCString(StringInfo out, JsonContainer *jc,
+ int estimated_len);
+extern JsonbValue *pushJsonValue(JsonbParseState **pstate,
+ JsonbIteratorToken tok, JsonbValue *jbv);
+extern JsonbValue *findJsonValueFromContainer(JsonContainer *jc, uint32 flags,
+ JsonbValue *key);
+extern JsonbValue *getIthJsonValueFromContainer(JsonContainer *array,
+ uint32 index);
+
#endif /* JSONAPI_H */
diff --git a/src/include/utils/jsonb.h b/src/include/utils/jsonb.h
index 9698c46..a5d1df4 100644
--- a/src/include/utils/jsonb.h
+++ b/src/include/utils/jsonb.h
@@ -221,10 +221,10 @@ typedef struct
} Jsonb;
/* convenience macros for accessing the root container in a Jsonb datum */
-#define JB_ROOT_COUNT(jbp_) (*(uint32 *) VARDATA(jbp_) & JB_CMASK)
-#define JB_ROOT_IS_SCALAR(jbp_) ((*(uint32 *) VARDATA(jbp_) & JB_FSCALAR) != 0)
-#define JB_ROOT_IS_OBJECT(jbp_) ((*(uint32 *) VARDATA(jbp_) & JB_FOBJECT) != 0)
-#define JB_ROOT_IS_ARRAY(jbp_) ((*(uint32 *) VARDATA(jbp_) & JB_FARRAY) != 0)
+#define JB_ROOT_COUNT(jbp_) JsonContainerSize(&(jbp_)->root)
+#define JB_ROOT_IS_SCALAR(jbp_) JsonContainerIsScalar(&(jbp_)->root)
+#define JB_ROOT_IS_OBJECT(jbp_) JsonContainerIsObject(&(jbp_)->root)
+#define JB_ROOT_IS_ARRAY(jbp_) JsonContainerIsArray(&(jbp_)->root)
enum jbvType
@@ -239,14 +239,6 @@ enum jbvType
jbvObject,
/* Binary (i.e. struct Jsonb) jbvArray/jbvObject */
jbvBinary,
-
- /*
- * Virtual types.
- *
- * These types are used only for in-memory JSON processing and serialized
- * into JSON strings when outputted to json/jsonb.
- */
- jbvDatetime = 0x20,
};
/*
@@ -280,6 +272,8 @@ struct JsonbValue
{
int nPairs; /* 1 pair, 2 elements */
JsonbPair *pairs;
+ bool uniquify; /* Should we sort pairs by key name and
+ * remove duplicate keys? */
} object; /* Associative container type */
struct
@@ -287,20 +281,11 @@ struct JsonbValue
int len;
JsonbContainer *data;
} binary; /* Array or object, in on-disk format */
-
- struct
- {
- Datum value;
- Oid typid;
- int32 typmod;
- int tz;
- } datetime;
} val;
};
#define IsAJsonbScalar(jsonbval) (((jsonbval)->type >= jbvNull && \
- (jsonbval)->type <= jbvBool) || \
- (jsonbval)->type == jbvDatetime)
+ (jsonbval)->type <= jbvBool))
/*
* Key/value pair within an Object.
@@ -374,6 +359,8 @@ typedef struct JsonbIterator
/* Support functions */
extern uint32 getJsonbOffset(const JsonbContainer *jc, int index);
extern uint32 getJsonbLength(const JsonbContainer *jc, int index);
+extern int lengthCompareJsonbStringValue(const void *a, const void *b);
+extern bool equalsJsonbScalarValue(JsonbValue *a, JsonbValue *b);
extern int compareJsonbContainers(JsonbContainer *a, JsonbContainer *b);
extern JsonbValue *findJsonbValueFromContainer(JsonbContainer *sheader,
uint32 flags,
@@ -382,6 +369,8 @@ extern JsonbValue *getIthJsonbValueFromContainer(JsonbContainer *sheader,
uint32 i);
extern JsonbValue *pushJsonbValue(JsonbParseState **pstate,
JsonbIteratorToken seq, JsonbValue *jbVal);
+extern JsonbValue *pushJsonbValueScalar(JsonbParseState **pstate,
+ JsonbIteratorToken seq,JsonbValue *scalarVal);
extern JsonbIterator *JsonbIteratorInit(JsonbContainer *container);
extern JsonbIteratorToken JsonbIteratorNext(JsonbIterator **it, JsonbValue *val,
bool skipNested);
diff --git a/src/test/regress/expected/json_jsonpath.out b/src/test/regress/expected/json_jsonpath.out
new file mode 100644
index 0000000..220025d
--- /dev/null
+++ b/src/test/regress/expected/json_jsonpath.out
@@ -0,0 +1,2120 @@
+select json '{"a": 12}' @? '$';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": 12}' @? '1';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": 12}' @? '$.a.b';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"a": 12}' @? '$.b';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"a": 12}' @? '$.a + 2';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": 12}' @? '$.b + 2';
+ ?column?
+----------
+
+(1 row)
+
+select json '{"a": {"a": 12}}' @? '$.a.a';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"a": 12}}' @? '$.*.a';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"b": {"a": 12}}' @? '$.*.a';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"b": {"a": 12}}' @? '$.*.b';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"b": {"a": 12}}' @? 'strict $.*.b';
+ ?column?
+----------
+
+(1 row)
+
+select json '{}' @? '$.*';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"a": 1}' @? '$.*';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"b": 1}}' @? 'lax $.**{1}';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"b": 1}}' @? 'lax $.**{2}';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"b": 1}}' @? 'lax $.**{3}';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '[]' @? '$[*]';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '[1]' @? '$[*]';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1]' @? '$[1]';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '[1]' @? 'strict $[1]';
+ ?column?
+----------
+
+(1 row)
+
+select json_path_query('[1]', 'strict $[1]');
+ERROR: invalid SQL/JSON subscript
+DETAIL: jsonpath array subscript is out of bounds
+select json '[1]' @? 'lax $[10000000000000000]';
+ ?column?
+----------
+
+(1 row)
+
+select json '[1]' @? 'strict $[10000000000000000]';
+ ?column?
+----------
+
+(1 row)
+
+select json_path_query('[1]', 'lax $[10000000000000000]');
+ERROR: invalid SQL/JSON subscript
+DETAIL: jsonpath array subscript is out of integer range
+select json_path_query('[1]', 'strict $[10000000000000000]');
+ERROR: invalid SQL/JSON subscript
+DETAIL: jsonpath array subscript is out of integer range
+select json '[1]' @? '$[0]';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1]' @? '$[0.3]';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1]' @? '$[0.5]';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1]' @? '$[0.9]';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1]' @? '$[1.2]';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '[1]' @? 'strict $[1.2]';
+ ?column?
+----------
+
+(1 row)
+
+select json '{"a": [1,2,3], "b": [3,4,5]}' @? '$ ? (@.a[*] > @.b[*])';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"a": [1,2,3], "b": [3,4,5]}' @? '$ ? (@.a[*] >= @.b[*])';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": [1,2,3], "b": [3,4,"5"]}' @? '$ ? (@.a[*] >= @.b[*])';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": [1,2,3], "b": [3,4,"5"]}' @? 'strict $ ? (@.a[*] >= @.b[*])';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"a": [1,2,3], "b": [3,4,null]}' @? '$ ? (@.a[*] >= @.b[*])';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '1' @? '$ ? ((@ == "1") is unknown)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '1' @? '$ ? ((@ == 1) is unknown)';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '[{"a": 1}, {"a": 2}]' @? '$[0 to 1] ? (@.a > 1)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json_path_query('1', 'lax $.a');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('1', 'strict $.a');
+ERROR: SQL/JSON member not found
+DETAIL: jsonpath member accessor is applied to not an object
+select json_path_query('1', 'strict $.*');
+ERROR: SQL/JSON object not found
+DETAIL: jsonpath wildcard member accessor is applied to not an object
+select json_path_query('[]', 'lax $.a');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('[]', 'strict $.a');
+ERROR: SQL/JSON member not found
+DETAIL: jsonpath member accessor is applied to not an object
+select json_path_query('{}', 'lax $.a');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{}', 'strict $.a');
+ERROR: SQL/JSON member not found
+DETAIL: JSON object does not contain key "a"
+select json_path_query('1', 'strict $[1]');
+ERROR: SQL/JSON array not found
+DETAIL: jsonpath array accessor is applied to not an array
+select json_path_query('1', 'strict $[*]');
+ERROR: SQL/JSON array not found
+DETAIL: jsonpath wildcard array accessor is applied to not an array
+select json_path_query('[]', 'strict $[1]');
+ERROR: invalid SQL/JSON subscript
+DETAIL: jsonpath array subscript is out of bounds
+select json_path_query('[]', 'strict $["a"]');
+ERROR: invalid SQL/JSON subscript
+DETAIL: jsonpath array subscript is not a singleton numeric value
+select json_path_query('{"a": 12, "b": {"a": 13}}', '$.a');
+ json_path_query
+-----------------
+ 12
+(1 row)
+
+select json_path_query('{"a": 12, "b": {"a": 13}}', '$.b');
+ json_path_query
+-----------------
+ {"a": 13}
+(1 row)
+
+select json_path_query('{"a": 12, "b": {"a": 13}}', '$.*');
+ json_path_query
+-----------------
+ 12
+ {"a": 13}
+(2 rows)
+
+select json_path_query('{"a": 12, "b": {"a": 13}}', 'lax $.*.a');
+ json_path_query
+-----------------
+ 13
+(1 row)
+
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[*].a');
+ json_path_query
+-----------------
+ 13
+(1 row)
+
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[*].*');
+ json_path_query
+-----------------
+ 13
+ 14
+(2 rows)
+
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[0].a');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[1].a');
+ json_path_query
+-----------------
+ 13
+(1 row)
+
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[2].a');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[0,1].a');
+ json_path_query
+-----------------
+ 13
+(1 row)
+
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[0 to 10].a');
+ json_path_query
+-----------------
+ 13
+(1 row)
+
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[0 to 10 / 0].a');
+ERROR: division by zero
+select json_path_query('[12, {"a": 13}, {"b": 14}, "ccc", true]', '$[2.5 - 1 to $.size() - 2]');
+ json_path_query
+-----------------
+ {"a": 13}
+ {"b": 14}
+ "ccc"
+(3 rows)
+
+select json_path_query('1', 'lax $[0]');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('1', 'lax $[*]');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('[1]', 'lax $[0]');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('[1]', 'lax $[*]');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('[1,2,3]', 'lax $[*]');
+ json_path_query
+-----------------
+ 1
+ 2
+ 3
+(3 rows)
+
+select json_path_query('[1,2,3]', 'strict $[*].a');
+ERROR: SQL/JSON member not found
+DETAIL: jsonpath member accessor is applied to not an object
+select json_path_query('[]', '$[last]');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('[]', '$[last ? (exists(last))]');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('[]', 'strict $[last]');
+ERROR: invalid SQL/JSON subscript
+DETAIL: jsonpath array subscript is out of bounds
+select json_path_query('[1]', '$[last]');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('[1,2,3]', '$[last]');
+ json_path_query
+-----------------
+ 3
+(1 row)
+
+select json_path_query('[1,2,3]', '$[last - 1]');
+ json_path_query
+-----------------
+ 2
+(1 row)
+
+select json_path_query('[1,2,3]', '$[last ? (@.type() == "number")]');
+ json_path_query
+-----------------
+ 3
+(1 row)
+
+select json_path_query('[1,2,3]', '$[last ? (@.type() == "string")]');
+ERROR: invalid SQL/JSON subscript
+DETAIL: jsonpath array subscript is not a singleton numeric value
+select * from json_path_query('{"a": 10}', '$');
+ json_path_query
+-----------------
+ {"a": 10}
+(1 row)
+
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)');
+ERROR: cannot find jsonpath variable 'value'
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)', '1');
+ERROR: json containing jsonpath variables is not an object
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)', '[{"value" : 13}]');
+ERROR: json containing jsonpath variables is not an object
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)', '{"value" : 13}');
+ json_path_query
+-----------------
+ {"a": 10}
+(1 row)
+
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)', '{"value" : 8}');
+ json_path_query
+-----------------
+(0 rows)
+
+select * from json_path_query('{"a": 10}', '$.a ? (@ < $value)', '{"value" : 13}');
+ json_path_query
+-----------------
+ 10
+(1 row)
+
+select * from json_path_query('[10,11,12,13,14,15]', '$[*] ? (@ < $value)', '{"value" : 13}');
+ json_path_query
+-----------------
+ 10
+ 11
+ 12
+(3 rows)
+
+select * from json_path_query('[10,11,12,13,14,15]', '$[0,1] ? (@ < $x.value)', '{"x": {"value" : 13}}');
+ json_path_query
+-----------------
+ 10
+ 11
+(2 rows)
+
+select * from json_path_query('[10,11,12,13,14,15]', '$[0 to 2] ? (@ < $value)', '{"value" : 15}');
+ json_path_query
+-----------------
+ 10
+ 11
+ 12
+(3 rows)
+
+select * from json_path_query('[1,"1",2,"2",null]', '$[*] ? (@ == "1")');
+ json_path_query
+-----------------
+ "1"
+(1 row)
+
+select * from json_path_query('[1,"1",2,"2",null]', '$[*] ? (@ == $value)', '{"value" : "1"}');
+ json_path_query
+-----------------
+ "1"
+(1 row)
+
+select * from json_path_query('[1,"1",2,"2",null]', '$[*] ? (@ == $value)', '{"value" : null}');
+ json_path_query
+-----------------
+ null
+(1 row)
+
+select * from json_path_query('[1, "2", null]', '$[*] ? (@ != null)');
+ json_path_query
+-----------------
+ 1
+ "2"
+(2 rows)
+
+select * from json_path_query('[1, "2", null]', '$[*] ? (@ == null)');
+ json_path_query
+-----------------
+ null
+(1 row)
+
+select * from json_path_query('{}', '$ ? (@ == @)');
+ json_path_query
+-----------------
+(0 rows)
+
+select * from json_path_query('[]', 'strict $ ? (@ == @)');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**');
+ json_path_query
+-----------------
+ {"a": {"b": 1}}
+ {"b": 1}
+ 1
+(3 rows)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{0}');
+ json_path_query
+-----------------
+ {"a": {"b": 1}}
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{0 to last}');
+ json_path_query
+-----------------
+ {"a": {"b": 1}}
+ {"b": 1}
+ 1
+(3 rows)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1}');
+ json_path_query
+-----------------
+ {"b": 1}
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1 to last}');
+ json_path_query
+-----------------
+ {"b": 1}
+ 1
+(2 rows)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{2}');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{2 to last}');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{3 to last}');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{last}');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{0}.b ? (@ > 0)');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1}.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{0 to last}.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1 to last}.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1 to 2}.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{0}.b ? (@ > 0)');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{1}.b ? (@ > 0)');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{0 to last}.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{1 to last}.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{1 to 2}.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{2 to 3}.b ? (@ > 0)');
+ json_path_query
+-----------------
+ 1
+(1 row)
+
+select json '{"a": {"b": 1}}' @? '$.**.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"b": 1}}' @? '$.**{0}.b ? ( @ > 0)';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"a": {"b": 1}}' @? '$.**{1}.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"b": 1}}' @? '$.**{0 to last}.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"b": 1}}' @? '$.**{1 to last}.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"b": 1}}' @? '$.**{1 to 2}.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"c": {"b": 1}}}' @? '$.**.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{0}.b ? ( @ > 0)';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{1}.b ? ( @ > 0)';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{0 to last}.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{1 to last}.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{1 to 2}.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{2 to 3}.b ? ( @ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json_path_query('{"g": {"x": 2}}', '$.g ? (exists (@.x))');
+ json_path_query
+-----------------
+ {"x": 2}
+(1 row)
+
+select json_path_query('{"g": {"x": 2}}', '$.g ? (exists (@.y))');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"g": {"x": 2}}', '$.g ? (exists (@.x ? (@ >= 2) ))');
+ json_path_query
+-----------------
+ {"x": 2}
+(1 row)
+
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'lax $.g ? (exists (@.x))');
+ json_path_query
+-----------------
+ {"x": 2}
+(1 row)
+
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'lax $.g ? (exists (@.x + "3"))');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'lax $.g ? ((exists (@.x + "3")) is unknown)');
+ json_path_query
+-----------------
+ {"x": 2}
+ {"y": 3}
+(2 rows)
+
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'strict $.g[*] ? (exists (@.x))');
+ json_path_query
+-----------------
+ {"x": 2}
+(1 row)
+
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'strict $.g[*] ? ((exists (@.x)) is unknown)');
+ json_path_query
+-----------------
+ {"y": 3}
+(1 row)
+
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'strict $.g ? (exists (@[*].x))');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'strict $.g ? ((exists (@[*].x)) is unknown)');
+ json_path_query
+----------------------
+ [{"x": 2}, {"y": 3}]
+(1 row)
+
+--test ternary logic
+select
+ x, y,
+ json_path_query(
+ '[true, false, null]',
+ '$[*] ? (@ == true && ($x == true && $y == true) ||
+ @ == false && !($x == true && $y == true) ||
+ @ == null && ($x == true && $y == true) is unknown)',
+ json_build_object('x', x, 'y', y)
+ ) as "x && y"
+from
+ (values (json 'true'), ('false'), ('"null"')) x(x),
+ (values (json 'true'), ('false'), ('"null"')) y(y);
+ x | y | x && y
+--------+--------+--------
+ true | true | true
+ true | false | false
+ true | "null" | null
+ false | true | false
+ false | false | false
+ false | "null" | false
+ "null" | true | null
+ "null" | false | false
+ "null" | "null" | null
+(9 rows)
+
+select
+ x, y,
+ json_path_query(
+ '[true, false, null]',
+ '$[*] ? (@ == true && ($x == true || $y == true) ||
+ @ == false && !($x == true || $y == true) ||
+ @ == null && ($x == true || $y == true) is unknown)',
+ json_build_object('x', x, 'y', y)
+ ) as "x || y"
+from
+ (values (json 'true'), ('false'), ('"null"')) x(x),
+ (values (json 'true'), ('false'), ('"null"')) y(y);
+ x | y | x || y
+--------+--------+--------
+ true | true | true
+ true | false | true
+ true | "null" | true
+ false | true | true
+ false | false | false
+ false | "null" | null
+ "null" | true | true
+ "null" | false | null
+ "null" | "null" | null
+(9 rows)
+
+select json '{"a": 1, "b":1}' @? '$ ? (@.a == @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": 1, "b":1}}' @? '$ ? (@.a == @.b)';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '{"c": {"a": 1, "b":1}}' @? '$.c ? (@.a == @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": 1, "b":1}}' @? '$.c ? ($.c.a == @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": 1, "b":1}}' @? '$.* ? (@.a == @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": 1, "b":1}' @? '$.** ? (@.a == @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": 1, "b":1}}' @? '$.** ? (@.a == @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json_path_query('{"c": {"a": 2, "b":1}}', '$.** ? (@.a == 1 + 1)');
+ json_path_query
+-----------------
+ {"a": 2, "b":1}
+(1 row)
+
+select json_path_query('{"c": {"a": 2, "b":1}}', '$.** ? (@.a == (1 + 1))');
+ json_path_query
+-----------------
+ {"a": 2, "b":1}
+(1 row)
+
+select json_path_query('{"c": {"a": 2, "b":1}}', '$.** ? (@.a == @.b + 1)');
+ json_path_query
+-----------------
+ {"a": 2, "b":1}
+(1 row)
+
+select json_path_query('{"c": {"a": 2, "b":1}}', '$.** ? (@.a == (@.b + 1))');
+ json_path_query
+-----------------
+ {"a": 2, "b":1}
+(1 row)
+
+select json '{"c": {"a": -1, "b":1}}' @? '$.** ? (@.a == - 1)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": -1, "b":1}}' @? '$.** ? (@.a == -1)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": -1, "b":1}}' @? '$.** ? (@.a == [email protected])';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": -1, "b":1}}' @? '$.** ? (@.a == - @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": 0, "b":1}}' @? '$.** ? (@.a == 1 - @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": 2, "b":1}}' @? '$.** ? (@.a == 1 - - @.b)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"c": {"a": 0, "b":1}}' @? '$.** ? (@.a == 1 - [email protected])';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1,2,3]' @? '$ ? (+@[*] > +2)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1,2,3]' @? '$ ? (+@[*] > +3)';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '[1,2,3]' @? '$ ? (-@[*] < -2)';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1,2,3]' @? '$ ? (-@[*] < -3)';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '1' @? '$ ? ($ > 0)';
+ ?column?
+----------
+ t
+(1 row)
+
+-- arithmetic errors
+select json_path_query('[1,2,0,3]', '$[*] ? (2 / @ > 0)');
+ json_path_query
+-----------------
+ 1
+ 2
+ 3
+(3 rows)
+
+select json_path_query('[1,2,0,3]', '$[*] ? ((2 / @ > 0) is unknown)');
+ json_path_query
+-----------------
+ 0
+(1 row)
+
+select json_path_query('0', '1 / $');
+ERROR: division by zero
+select json_path_query('0', '1 / $ + 2');
+ERROR: division by zero
+select json_path_query('0', '-(3 + 1 % $)');
+ERROR: division by zero
+select json_path_query('1', '$ + "2"');
+ERROR: singleton SQL/JSON item required
+DETAIL: right operand of binary jsonpath operator + is not a singleton numeric value
+select json_path_query('[1, 2]', '3 * $');
+ERROR: singleton SQL/JSON item required
+DETAIL: right operand of binary jsonpath operator * is not a singleton numeric value
+select json_path_query('"a"', '-$');
+ERROR: SQL/JSON number not found
+DETAIL: operand of unary jsonpath operator - is not a numeric value
+select json_path_query('[1,"2",3]', '+$');
+ERROR: SQL/JSON number not found
+DETAIL: operand of unary jsonpath operator + is not a numeric value
+select json '["1",2,0,3]' @? '-$[*]';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '[1,"2",0,3]' @? '-$[*]';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '["1",2,0,3]' @? 'strict -$[*]';
+ ?column?
+----------
+
+(1 row)
+
+select json '[1,"2",0,3]' @? 'strict -$[*]';
+ ?column?
+----------
+
+(1 row)
+
+-- unwrapping of operator arguments in lax mode
+select json_path_query('{"a": [2]}', 'lax $.a * 3');
+ json_path_query
+-----------------
+ 6
+(1 row)
+
+select json_path_query('{"a": [2]}', 'lax $.a + 3');
+ json_path_query
+-----------------
+ 5
+(1 row)
+
+select json_path_query('{"a": [2, 3, 4]}', 'lax -$.a');
+ json_path_query
+-----------------
+ -2
+ -3
+ -4
+(3 rows)
+
+-- should fail
+select json_path_query('{"a": [1, 2]}', 'lax $.a * 3');
+ERROR: singleton SQL/JSON item required
+DETAIL: left operand of binary jsonpath operator * is not a singleton numeric value
+-- extension: boolean expressions
+select json_path_query('2', '$ > 1');
+ json_path_query
+-----------------
+ true
+(1 row)
+
+select json_path_query('2', '$ <= 1');
+ json_path_query
+-----------------
+ false
+(1 row)
+
+select json_path_query('2', '$ == "2"');
+ json_path_query
+-----------------
+ null
+(1 row)
+
+select json '2' @? '$ == "2"';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '2' @@ '$ > 1';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '2' @@ '$ <= 1';
+ ?column?
+----------
+ f
+(1 row)
+
+select json '2' @@ '$ == "2"';
+ ?column?
+----------
+
+(1 row)
+
+select json '2' @@ '1';
+ ?column?
+----------
+
+(1 row)
+
+select json '{}' @@ '$';
+ ?column?
+----------
+
+(1 row)
+
+select json '[]' @@ '$';
+ ?column?
+----------
+
+(1 row)
+
+select json '[1,2,3]' @@ '$[*]';
+ ?column?
+----------
+
+(1 row)
+
+select json '[]' @@ '$[*]';
+ ?column?
+----------
+
+(1 row)
+
+select json_path_match('[[1, true], [2, false]]', 'strict $[*] ? (@[0] > $x) [1]', '{"x": 1}');
+ json_path_match
+-----------------
+ f
+(1 row)
+
+select json_path_match('[[1, true], [2, false]]', 'strict $[*] ? (@[0] < $x) [1]', '{"x": 2}');
+ json_path_match
+-----------------
+ t
+(1 row)
+
+select json_path_query('[null,1,true,"a",[],{}]', '$.type()');
+ json_path_query
+-----------------
+ "array"
+(1 row)
+
+select json_path_query('[null,1,true,"a",[],{}]', 'lax $.type()');
+ json_path_query
+-----------------
+ "array"
+(1 row)
+
+select json_path_query('[null,1,true,"a",[],{}]', '$[*].type()');
+ json_path_query
+-----------------
+ "null"
+ "number"
+ "boolean"
+ "string"
+ "array"
+ "object"
+(6 rows)
+
+select json_path_query('null', 'null.type()');
+ json_path_query
+-----------------
+ "null"
+(1 row)
+
+select json_path_query('null', 'true.type()');
+ json_path_query
+-----------------
+ "boolean"
+(1 row)
+
+select json_path_query('null', '123.type()');
+ json_path_query
+-----------------
+ "number"
+(1 row)
+
+select json_path_query('null', '"123".type()');
+ json_path_query
+-----------------
+ "string"
+(1 row)
+
+select json_path_query('{"a": 2}', '($.a - 5).abs() + 10');
+ json_path_query
+-----------------
+ 13
+(1 row)
+
+select json_path_query('{"a": 2.5}', '-($.a * $.a).floor() % 4.3');
+ json_path_query
+-----------------
+ -1.7
+(1 row)
+
+select json_path_query('[1, 2, 3]', '($[*] > 2) ? (@ == true)');
+ json_path_query
+-----------------
+ true
+(1 row)
+
+select json_path_query('[1, 2, 3]', '($[*] > 3).type()');
+ json_path_query
+-----------------
+ "boolean"
+(1 row)
+
+select json_path_query('[1, 2, 3]', '($[*].a > 3).type()');
+ json_path_query
+-----------------
+ "boolean"
+(1 row)
+
+select json_path_query('[1, 2, 3]', 'strict ($[*].a > 3).type()');
+ json_path_query
+-----------------
+ "null"
+(1 row)
+
+select json_path_query('[1,null,true,"11",[],[1],[1,2,3],{},{"a":1,"b":2}]', 'strict $[*].size()');
+ERROR: SQL/JSON array not found
+DETAIL: jsonpath item method .size() is applied to not an array
+select json_path_query('[1,null,true,"11",[],[1],[1,2,3],{},{"a":1,"b":2}]', 'lax $[*].size()');
+ json_path_query
+-----------------
+ 1
+ 1
+ 1
+ 1
+ 0
+ 1
+ 3
+ 1
+ 1
+(9 rows)
+
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].abs()');
+ json_path_query
+-----------------
+ 0
+ 1
+ 2
+ 3.4
+ 5.6
+(5 rows)
+
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].floor()');
+ json_path_query
+-----------------
+ 0
+ 1
+ -2
+ -4
+ 5
+(5 rows)
+
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].ceiling()');
+ json_path_query
+-----------------
+ 0
+ 1
+ -2
+ -3
+ 6
+(5 rows)
+
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].ceiling().abs()');
+ json_path_query
+-----------------
+ 0
+ 1
+ 2
+ 3
+ 6
+(5 rows)
+
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].ceiling().abs().type()');
+ json_path_query
+-----------------
+ "number"
+ "number"
+ "number"
+ "number"
+ "number"
+(5 rows)
+
+select json_path_query('[{},1]', '$[*].keyvalue()');
+ERROR: SQL/JSON object not found
+DETAIL: jsonpath item method .keyvalue() is applied to not an object
+select json_path_query('{}', '$.keyvalue()');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('{"a": 1, "b": [1, 2], "c": {"a": "bbb"}}', '$.keyvalue()');
+ json_path_query
+----------------------------------------------
+ {"key": "a", "value": 1, "id": 0}
+ {"key": "b", "value": [1, 2], "id": 0}
+ {"key": "c", "value": {"a": "bbb"}, "id": 0}
+(3 rows)
+
+select json_path_query('[{"a": 1, "b": [1, 2]}, {"c": {"a": "bbb"}}]', '$[*].keyvalue()');
+ json_path_query
+-----------------------------------------------
+ {"key": "a", "value": 1, "id": 1}
+ {"key": "b", "value": [1, 2], "id": 1}
+ {"key": "c", "value": {"a": "bbb"}, "id": 24}
+(3 rows)
+
+select json_path_query('[{"a": 1, "b": [1, 2]}, {"c": {"a": "bbb"}}]', 'strict $.keyvalue()');
+ERROR: SQL/JSON object not found
+DETAIL: jsonpath item method .keyvalue() is applied to not an object
+select json_path_query('[{"a": 1, "b": [1, 2]}, {"c": {"a": "bbb"}}]', 'lax $.keyvalue()');
+ json_path_query
+-----------------------------------------------
+ {"key": "a", "value": 1, "id": 1}
+ {"key": "b", "value": [1, 2], "id": 1}
+ {"key": "c", "value": {"a": "bbb"}, "id": 24}
+(3 rows)
+
+select json_path_query('[{"a": 1, "b": [1, 2]}, {"c": {"a": "bbb"}}]', 'strict $.keyvalue().a');
+ERROR: SQL/JSON object not found
+DETAIL: jsonpath item method .keyvalue() is applied to not an object
+select json '{"a": 1, "b": [1, 2]}' @? 'lax $.keyvalue()';
+ ?column?
+----------
+ t
+(1 row)
+
+select json '{"a": 1, "b": [1, 2]}' @? 'lax $.keyvalue().key';
+ ?column?
+----------
+ t
+(1 row)
+
+select json_path_query('null', '$.double()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .double() is applied to neither a string nor numeric value
+select json_path_query('true', '$.double()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .double() is applied to neither a string nor numeric value
+select json_path_query('[]', '$.double()');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('[]', 'strict $.double()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .double() is applied to neither a string nor numeric value
+select json_path_query('{}', '$.double()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .double() is applied to neither a string nor numeric value
+select json_path_query('1.23', '$.double()');
+ json_path_query
+-----------------
+ 1.23
+(1 row)
+
+select json_path_query('"1.23"', '$.double()');
+ json_path_query
+-----------------
+ 1.23
+(1 row)
+
+select json_path_query('"1.23aaa"', '$.double()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .double() is applied to not a numeric value
+select json_path_query('"nan"', '$.double()');
+ json_path_query
+-----------------
+ "NaN"
+(1 row)
+
+select json_path_query('"NaN"', '$.double()');
+ json_path_query
+-----------------
+ "NaN"
+(1 row)
+
+select json_path_query('"inf"', '$.double()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .double() is applied to not a numeric value
+select json_path_query('"-inf"', '$.double()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .double() is applied to not a numeric value
+select json_path_query('{}', '$.abs()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .abs() is applied to not a numeric value
+select json_path_query('true', '$.floor()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .floor() is applied to not a numeric value
+select json_path_query('"1.2"', '$.ceiling()');
+ERROR: non-numeric SQL/JSON item
+DETAIL: jsonpath item method .ceiling() is applied to not a numeric value
+select json_path_query('["", "a", "abc", "abcabc"]', '$[*] ? (@ starts with "abc")');
+ json_path_query
+-----------------
+ "abc"
+ "abcabc"
+(2 rows)
+
+select json_path_query('["", "a", "abc", "abcabc"]', 'strict $ ? (@[*] starts with "abc")');
+ json_path_query
+----------------------------
+ ["", "a", "abc", "abcabc"]
+(1 row)
+
+select json_path_query('["", "a", "abd", "abdabc"]', 'strict $ ? (@[*] starts with "abc")');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('["abc", "abcabc", null, 1]', 'strict $ ? (@[*] starts with "abc")');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('["abc", "abcabc", null, 1]', 'strict $ ? ((@[*] starts with "abc") is unknown)');
+ json_path_query
+----------------------------
+ ["abc", "abcabc", null, 1]
+(1 row)
+
+select json_path_query('[[null, 1, "abc", "abcabc"]]', 'lax $ ? (@[*] starts with "abc")');
+ json_path_query
+----------------------------
+ [null, 1, "abc", "abcabc"]
+(1 row)
+
+select json_path_query('[[null, 1, "abd", "abdabc"]]', 'lax $ ? ((@[*] starts with "abc") is unknown)');
+ json_path_query
+----------------------------
+ [null, 1, "abd", "abdabc"]
+(1 row)
+
+select json_path_query('[null, 1, "abd", "abdabc"]', 'lax $[*] ? ((@ starts with "abc") is unknown)');
+ json_path_query
+-----------------
+ null
+ 1
+(2 rows)
+
+select json_path_query('[null, 1, "abc", "abd", "aBdC", "abdacb", "adc\nabc", "babc"]', 'lax $[*] ? (@ like_regex "^ab.*c")');
+ json_path_query
+-----------------
+ "abc"
+ "abdacb"
+(2 rows)
+
+select json_path_query('[null, 1, "abc", "abd", "aBdC", "abdacb", "adc\nabc", "babc"]', 'lax $[*] ? (@ like_regex "^a b.* c " flag "ix")');
+ json_path_query
+-----------------
+ "abc"
+ "aBdC"
+ "abdacb"
+(3 rows)
+
+select json_path_query('[null, 1, "abc", "abd", "aBdC", "abdacb", "adc\nabc", "babc"]', 'lax $[*] ? (@ like_regex "^ab.*c" flag "m")');
+ json_path_query
+-----------------
+ "abc"
+ "abdacb"
+ "adc\nabc"
+(3 rows)
+
+select json_path_query('[null, 1, "abc", "abd", "aBdC", "abdacb", "adc\nabc", "babc"]', 'lax $[*] ? (@ like_regex "^ab.*c" flag "s")');
+ json_path_query
+-----------------
+ "abc"
+ "abdacb"
+(2 rows)
+
+select json_path_query('null', '$.datetime()');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: jsonpath item method .datetime() is applied to not a string
+select json_path_query('true', '$.datetime()');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: jsonpath item method .datetime() is applied to not a string
+select json_path_query('1', '$.datetime()');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: jsonpath item method .datetime() is applied to not a string
+select json_path_query('[]', '$.datetime()');
+ json_path_query
+-----------------
+(0 rows)
+
+select json_path_query('[]', 'strict $.datetime()');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: jsonpath item method .datetime() is applied to not a string
+select json_path_query('{}', '$.datetime()');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: jsonpath item method .datetime() is applied to not a string
+select json_path_query('""', '$.datetime()');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: unrecognized datetime format
+HINT: use datetime template argument for explicit format specification
+select json_path_query('"12:34"', '$.datetime("aaa")');
+ERROR: datetime format is not dated and not timed
+select json_path_query('"12:34"', '$.datetime("aaa", 1)');
+ERROR: datetime format is not dated and not timed
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 1)');
+ json_path_query
+---------------------
+ "12:34:00+00:00:01"
+(1 row)
+
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 10000000000000)');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: timezone argument of jsonpath item method .datetime() is out of integer range
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 10000000000000)');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: timezone argument of jsonpath item method .datetime() is out of integer range
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 2147483647)');
+ json_path_query
+-------------------------
+ "12:34:00+596523:14:07"
+(1 row)
+
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 2147483648)');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: timezone argument of jsonpath item method .datetime() is out of integer range
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", -2147483647)');
+ json_path_query
+-------------------------
+ "12:34:00-596523:14:07"
+(1 row)
+
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", -2147483648)');
+ERROR: invalid argument for SQL/JSON datetime function
+DETAIL: timezone argument of jsonpath item method .datetime() is out of integer range
+select json_path_query('"aaaa"', '$.datetime("HH24")');
+ERROR: invalid value "aa" for "HH24"
+DETAIL: Value must be an integer.
+select json '"10-03-2017"' @? '$.datetime("dd-mm-yyyy")';
+ ?column?
+----------
+ t
+(1 row)
+
+select json_path_query('"10-03-2017"', '$.datetime("dd-mm-yyyy")');
+ json_path_query
+-----------------
+ "2017-03-10"
+(1 row)
+
+select json_path_query('"10-03-2017"', '$.datetime("dd-mm-yyyy").type()');
+ json_path_query
+-----------------
+ "date"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy")');
+ json_path_query
+-----------------
+ "2017-03-10"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy").type()');
+ json_path_query
+-----------------
+ "date"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', ' $.datetime("dd-mm-yyyy HH24:MI").type()');
+ json_path_query
+-------------------------------
+ "timestamp without time zone"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 +05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM").type()');
+ json_path_query
+----------------------------
+ "timestamp with time zone"
+(1 row)
+
+select json_path_query('"12:34:56"', '$.datetime("HH24:MI:SS").type()');
+ json_path_query
+--------------------------
+ "time without time zone"
+(1 row)
+
+select json_path_query('"12:34:56 +05:20"', '$.datetime("HH24:MI:SS TZH:TZM").type()');
+ json_path_query
+-----------------------
+ "time with time zone"
+(1 row)
+
+set time zone '+00';
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI")');
+ json_path_query
+-----------------------
+ "2017-03-10T12:34:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+ERROR: missing time-zone in input string for type timestamptz
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "+00")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+00:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "+00:12")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+00:12"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "-00:12:34")');
+ json_path_query
+--------------------------------
+ "2017-03-10T12:34:00-00:12:34"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "UTC")');
+ERROR: invalid input syntax for type timestamptz: "UTC"
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", 12 * 60)');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+00:12"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", -(5 * 3600 + 12 * 60 + 34))');
+ json_path_query
+--------------------------------
+ "2017-03-10T12:34:00-05:12:34"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", $tz)',
+ json_build_object('tz', extract(timezone from now())));
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+00:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 +05"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+05:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 -05"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00-05:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 +05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+05:20"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 -05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00-05:20"
+(1 row)
+
+select json_path_query('"12:34"', '$.datetime("HH24:MI")');
+ json_path_query
+-----------------
+ "12:34:00"
+(1 row)
+
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH")');
+ERROR: missing time-zone in input string for type timetz
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", "+00")');
+ json_path_query
+------------------
+ "12:34:00+00:00"
+(1 row)
+
+select json_path_query('"12:34 +05"', '$.datetime("HH24:MI TZH")');
+ json_path_query
+------------------
+ "12:34:00+05:00"
+(1 row)
+
+select json_path_query('"12:34 -05"', '$.datetime("HH24:MI TZH")');
+ json_path_query
+------------------
+ "12:34:00-05:00"
+(1 row)
+
+select json_path_query('"12:34 +05:20"', '$.datetime("HH24:MI TZH:TZM")');
+ json_path_query
+------------------
+ "12:34:00+05:20"
+(1 row)
+
+select json_path_query('"12:34 -05:20"', '$.datetime("HH24:MI TZH:TZM")');
+ json_path_query
+------------------
+ "12:34:00-05:20"
+(1 row)
+
+set time zone '+10';
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI")');
+ json_path_query
+-----------------------
+ "2017-03-10T12:34:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+ERROR: missing time-zone in input string for type timestamptz
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "+10")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+10:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", $tz)',
+ json_build_object('tz', extract(timezone from now())));
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+10:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 +05"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+05:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 -05"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00-05:00"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 +05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+05:20"
+(1 row)
+
+select json_path_query('"10-03-2017 12:34 -05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM")');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00-05:20"
+(1 row)
+
+select json_path_query('"12:34"', '$.datetime("HH24:MI")');
+ json_path_query
+-----------------
+ "12:34:00"
+(1 row)
+
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH")');
+ERROR: missing time-zone in input string for type timetz
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", "+10")');
+ json_path_query
+------------------
+ "12:34:00+10:00"
+(1 row)
+
+select json_path_query('"12:34 +05"', '$.datetime("HH24:MI TZH")');
+ json_path_query
+------------------
+ "12:34:00+05:00"
+(1 row)
+
+select json_path_query('"12:34 -05"', '$.datetime("HH24:MI TZH")');
+ json_path_query
+------------------
+ "12:34:00-05:00"
+(1 row)
+
+select json_path_query('"12:34 +05:20"', '$.datetime("HH24:MI TZH:TZM")');
+ json_path_query
+------------------
+ "12:34:00+05:20"
+(1 row)
+
+select json_path_query('"12:34 -05:20"', '$.datetime("HH24:MI TZH:TZM")');
+ json_path_query
+------------------
+ "12:34:00-05:20"
+(1 row)
+
+set time zone default;
+select json_path_query('"2017-03-10"', '$.datetime().type()');
+ json_path_query
+-----------------
+ "date"
+(1 row)
+
+select json_path_query('"2017-03-10"', '$.datetime()');
+ json_path_query
+-----------------
+ "2017-03-10"
+(1 row)
+
+select json_path_query('"2017-03-10 12:34:56"', '$.datetime().type()');
+ json_path_query
+-------------------------------
+ "timestamp without time zone"
+(1 row)
+
+select json_path_query('"2017-03-10 12:34:56"', '$.datetime()');
+ json_path_query
+-----------------------
+ "2017-03-10T12:34:56"
+(1 row)
+
+select json_path_query('"2017-03-10 12:34:56 +3"', '$.datetime().type()');
+ json_path_query
+----------------------------
+ "timestamp with time zone"
+(1 row)
+
+select json_path_query('"2017-03-10 12:34:56 +3"', '$.datetime()');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:56+03:00"
+(1 row)
+
+select json_path_query('"2017-03-10 12:34:56 +3:10"', '$.datetime().type()');
+ json_path_query
+----------------------------
+ "timestamp with time zone"
+(1 row)
+
+select json_path_query('"2017-03-10 12:34:56 +3:10"', '$.datetime()');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:56+03:10"
+(1 row)
+
+select json_path_query('"12:34:56"', '$.datetime().type()');
+ json_path_query
+--------------------------
+ "time without time zone"
+(1 row)
+
+select json_path_query('"12:34:56"', '$.datetime()');
+ json_path_query
+-----------------
+ "12:34:56"
+(1 row)
+
+select json_path_query('"12:34:56 +3"', '$.datetime().type()');
+ json_path_query
+-----------------------
+ "time with time zone"
+(1 row)
+
+select json_path_query('"12:34:56 +3"', '$.datetime()');
+ json_path_query
+------------------
+ "12:34:56+03:00"
+(1 row)
+
+select json_path_query('"12:34:56 +3:10"', '$.datetime().type()');
+ json_path_query
+-----------------------
+ "time with time zone"
+(1 row)
+
+select json_path_query('"12:34:56 +3:10"', '$.datetime()');
+ json_path_query
+------------------
+ "12:34:56+03:10"
+(1 row)
+
+set time zone '+00';
+-- date comparison
+select json_path_query(
+ '["2017-03-10", "2017-03-11", "2017-03-09", "12:34:56", "01:02:03 +04", "2017-03-10 00:00:00", "2017-03-10 12:34:56", "2017-03-10 01:02:03 +04", "2017-03-10 03:00:00 +03"]',
+ '$[*].datetime() ? (@ == "10.03.2017".datetime("dd.mm.yyyy"))');
+ json_path_query
+-----------------------
+ "2017-03-10"
+ "2017-03-10T00:00:00"
+(2 rows)
+
+select json_path_query(
+ '["2017-03-10", "2017-03-11", "2017-03-09", "12:34:56", "01:02:03 +04", "2017-03-10 00:00:00", "2017-03-10 12:34:56", "2017-03-10 01:02:03 +04", "2017-03-10 03:00:00 +03"]',
+ '$[*].datetime() ? (@ >= "10.03.2017".datetime("dd.mm.yyyy"))');
+ json_path_query
+-----------------------
+ "2017-03-10"
+ "2017-03-11"
+ "2017-03-10T00:00:00"
+ "2017-03-10T12:34:56"
+(4 rows)
+
+select json_path_query(
+ '["2017-03-10", "2017-03-11", "2017-03-09", "12:34:56", "01:02:03 +04", "2017-03-10 00:00:00", "2017-03-10 12:34:56", "2017-03-10 01:02:03 +04", "2017-03-10 03:00:00 +03"]',
+ '$[*].datetime() ? (@ < "10.03.2017".datetime("dd.mm.yyyy"))');
+ json_path_query
+-----------------
+ "2017-03-09"
+(1 row)
+
+-- time comparison
+select json_path_query(
+ '["12:34:00", "12:35:00", "12:36:00", "12:35:00 +00", "12:35:00 +01", "13:35:00 +01", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +01"]',
+ '$[*].datetime() ? (@ == "12:35".datetime("HH24:MI"))');
+ json_path_query
+-----------------
+ "12:35:00"
+(1 row)
+
+select json_path_query(
+ '["12:34:00", "12:35:00", "12:36:00", "12:35:00 +00", "12:35:00 +01", "13:35:00 +01", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +01"]',
+ '$[*].datetime() ? (@ >= "12:35".datetime("HH24:MI"))');
+ json_path_query
+-----------------
+ "12:35:00"
+ "12:36:00"
+(2 rows)
+
+select json_path_query(
+ '["12:34:00", "12:35:00", "12:36:00", "12:35:00 +00", "12:35:00 +01", "13:35:00 +01", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +01"]',
+ '$[*].datetime() ? (@ < "12:35".datetime("HH24:MI"))');
+ json_path_query
+-----------------
+ "12:34:00"
+(1 row)
+
+-- timetz comparison
+select json_path_query(
+ '["12:34:00 +01", "12:35:00 +01", "12:36:00 +01", "12:35:00 +02", "12:35:00 -02", "10:35:00", "11:35:00", "12:35:00", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +1"]',
+ '$[*].datetime() ? (@ == "12:35 +1".datetime("HH24:MI TZH"))');
+ json_path_query
+------------------
+ "12:35:00+01:00"
+(1 row)
+
+select json_path_query(
+ '["12:34:00 +01", "12:35:00 +01", "12:36:00 +01", "12:35:00 +02", "12:35:00 -02", "10:35:00", "11:35:00", "12:35:00", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +1"]',
+ '$[*].datetime() ? (@ >= "12:35 +1".datetime("HH24:MI TZH"))');
+ json_path_query
+------------------
+ "12:35:00+01:00"
+ "12:36:00+01:00"
+ "12:35:00-02:00"
+(3 rows)
+
+select json_path_query(
+ '["12:34:00 +01", "12:35:00 +01", "12:36:00 +01", "12:35:00 +02", "12:35:00 -02", "10:35:00", "11:35:00", "12:35:00", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +1"]',
+ '$[*].datetime() ? (@ < "12:35 +1".datetime("HH24:MI TZH"))');
+ json_path_query
+------------------
+ "12:34:00+01:00"
+ "12:35:00+02:00"
+(2 rows)
+
+-- timestamp comparison
+select json_path_query(
+ '["2017-03-10 12:34:00", "2017-03-10 12:35:00", "2017-03-10 12:36:00", "2017-03-10 12:35:00 +01", "2017-03-10 13:35:00 +01", "2017-03-10 12:35:00 -01", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ == "10.03.2017 12:35".datetime("dd.mm.yyyy HH24:MI"))');
+ json_path_query
+-----------------------
+ "2017-03-10T12:35:00"
+(1 row)
+
+select json_path_query(
+ '["2017-03-10 12:34:00", "2017-03-10 12:35:00", "2017-03-10 12:36:00", "2017-03-10 12:35:00 +01", "2017-03-10 13:35:00 +01", "2017-03-10 12:35:00 -01", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ >= "10.03.2017 12:35".datetime("dd.mm.yyyy HH24:MI"))');
+ json_path_query
+-----------------------
+ "2017-03-10T12:35:00"
+ "2017-03-10T12:36:00"
+ "2017-03-11"
+(3 rows)
+
+select json_path_query(
+ '["2017-03-10 12:34:00", "2017-03-10 12:35:00", "2017-03-10 12:36:00", "2017-03-10 12:35:00 +01", "2017-03-10 13:35:00 +01", "2017-03-10 12:35:00 -01", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ < "10.03.2017 12:35".datetime("dd.mm.yyyy HH24:MI"))');
+ json_path_query
+-----------------------
+ "2017-03-10T12:34:00"
+ "2017-03-10"
+(2 rows)
+
+-- timestamptz comparison
+select json_path_query(
+ '["2017-03-10 12:34:00 +01", "2017-03-10 12:35:00 +01", "2017-03-10 12:36:00 +01", "2017-03-10 12:35:00 +02", "2017-03-10 12:35:00 -02", "2017-03-10 10:35:00", "2017-03-10 11:35:00", "2017-03-10 12:35:00", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ == "10.03.2017 12:35 +1".datetime("dd.mm.yyyy HH24:MI TZH"))');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:35:00+01:00"
+(1 row)
+
+select json_path_query(
+ '["2017-03-10 12:34:00 +01", "2017-03-10 12:35:00 +01", "2017-03-10 12:36:00 +01", "2017-03-10 12:35:00 +02", "2017-03-10 12:35:00 -02", "2017-03-10 10:35:00", "2017-03-10 11:35:00", "2017-03-10 12:35:00", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ >= "10.03.2017 12:35 +1".datetime("dd.mm.yyyy HH24:MI TZH"))');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:35:00+01:00"
+ "2017-03-10T12:36:00+01:00"
+ "2017-03-10T12:35:00-02:00"
+(3 rows)
+
+select json_path_query(
+ '["2017-03-10 12:34:00 +01", "2017-03-10 12:35:00 +01", "2017-03-10 12:36:00 +01", "2017-03-10 12:35:00 +02", "2017-03-10 12:35:00 -02", "2017-03-10 10:35:00", "2017-03-10 11:35:00", "2017-03-10 12:35:00", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ < "10.03.2017 12:35 +1".datetime("dd.mm.yyyy HH24:MI TZH"))');
+ json_path_query
+-----------------------------
+ "2017-03-10T12:34:00+01:00"
+ "2017-03-10T12:35:00+02:00"
+(2 rows)
+
+set time zone default;
+-- jsonpath operators
+SELECT json_path_query('[{"a": 1}, {"a": 2}]', '$[*]');
+ json_path_query
+-----------------
+ {"a": 1}
+ {"a": 2}
+(2 rows)
+
+SELECT json_path_query('[{"a": 1}, {"a": 2}]', '$[*] ? (@.a > 10)');
+ json_path_query
+-----------------
+(0 rows)
+
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a');
+ERROR: SQL/JSON member not found
+DETAIL: JSON object does not contain key "a"
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a');
+ json_path_query_array
+-----------------------
+ [1, 2]
+(1 row)
+
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ == 1)');
+ json_path_query_array
+-----------------------
+ [1]
+(1 row)
+
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ > 10)');
+ json_path_query_array
+-----------------------
+ []
+(1 row)
+
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', '{"min": 1, "max": 4}');
+ json_path_query_array
+-----------------------
+ [2, 3]
+(1 row)
+
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', '{"min": 3, "max": 4}');
+ json_path_query_array
+-----------------------
+ []
+(1 row)
+
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a');
+ERROR: SQL/JSON member not found
+DETAIL: JSON object does not contain key "a"
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}]', '$[*].a');
+ json_path_query_first
+-----------------------
+ 1
+(1 row)
+
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ == 1)');
+ json_path_query_first
+-----------------------
+ 1
+(1 row)
+
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ > 10)');
+ json_path_query_first
+-----------------------
+
+(1 row)
+
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', '{"min": 1, "max": 4}');
+ json_path_query_first
+-----------------------
+ 2
+(1 row)
+
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', '{"min": 3, "max": 4}');
+ json_path_query_first
+-----------------------
+
+(1 row)
+
+SELECT json '[{"a": 1}, {"a": 2}]' @? '$[*].a ? (@ > 1)';
+ ?column?
+----------
+ t
+(1 row)
+
+SELECT json '[{"a": 1}, {"a": 2}]' @? '$[*] ? (@.a > 2)';
+ ?column?
+----------
+ f
+(1 row)
+
+SELECT json_path_exists('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*] ? (@.a > $min && @.a < $max)', '{"min": 1, "max": 4}');
+ json_path_exists
+------------------
+ t
+(1 row)
+
+SELECT json_path_exists('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*] ? (@.a > $min && @.a < $max)', '{"min": 3, "max": 4}');
+ json_path_exists
+------------------
+ f
+(1 row)
+
+SELECT json '[{"a": 1}, {"a": 2}]' @@ '$[*].a > 1';
+ ?column?
+----------
+ t
+(1 row)
+
+SELECT json '[{"a": 1}, {"a": 2}]' @@ '$[*].a > 2';
+ ?column?
+----------
+ f
+(1 row)
+
diff --git a/src/test/regress/parallel_schedule b/src/test/regress/parallel_schedule
index c10d909..2f1dd8d 100644
--- a/src/test/regress/parallel_schedule
+++ b/src/test/regress/parallel_schedule
@@ -115,7 +115,7 @@ test: select_views portals_p2 foreign_key cluster dependency guc bitmapops combo
# ----------
# Another group of parallel tests (JSON related)
# ----------
-test: json jsonb json_encoding jsonpath jsonb_jsonpath
+test: json jsonb json_encoding jsonpath json_jsonpath jsonb_jsonpath
# ----------
# Another group of parallel tests
diff --git a/src/test/regress/serial_schedule b/src/test/regress/serial_schedule
index d2bf88b..f1fb336 100644
--- a/src/test/regress/serial_schedule
+++ b/src/test/regress/serial_schedule
@@ -158,6 +158,7 @@ test: json
test: jsonb
test: json_encoding
test: jsonpath
+test: json_jsonpath
test: jsonb_jsonpath
test: indirect_toast
test: equivclass
diff --git a/src/test/regress/sql/json_jsonpath.sql b/src/test/regress/sql/json_jsonpath.sql
new file mode 100644
index 0000000..73781b0
--- /dev/null
+++ b/src/test/regress/sql/json_jsonpath.sql
@@ -0,0 +1,477 @@
+select json '{"a": 12}' @? '$';
+select json '{"a": 12}' @? '1';
+select json '{"a": 12}' @? '$.a.b';
+select json '{"a": 12}' @? '$.b';
+select json '{"a": 12}' @? '$.a + 2';
+select json '{"a": 12}' @? '$.b + 2';
+select json '{"a": {"a": 12}}' @? '$.a.a';
+select json '{"a": {"a": 12}}' @? '$.*.a';
+select json '{"b": {"a": 12}}' @? '$.*.a';
+select json '{"b": {"a": 12}}' @? '$.*.b';
+select json '{"b": {"a": 12}}' @? 'strict $.*.b';
+select json '{}' @? '$.*';
+select json '{"a": 1}' @? '$.*';
+select json '{"a": {"b": 1}}' @? 'lax $.**{1}';
+select json '{"a": {"b": 1}}' @? 'lax $.**{2}';
+select json '{"a": {"b": 1}}' @? 'lax $.**{3}';
+select json '[]' @? '$[*]';
+select json '[1]' @? '$[*]';
+select json '[1]' @? '$[1]';
+select json '[1]' @? 'strict $[1]';
+select json_path_query('[1]', 'strict $[1]');
+select json '[1]' @? 'lax $[10000000000000000]';
+select json '[1]' @? 'strict $[10000000000000000]';
+select json_path_query('[1]', 'lax $[10000000000000000]');
+select json_path_query('[1]', 'strict $[10000000000000000]');
+select json '[1]' @? '$[0]';
+select json '[1]' @? '$[0.3]';
+select json '[1]' @? '$[0.5]';
+select json '[1]' @? '$[0.9]';
+select json '[1]' @? '$[1.2]';
+select json '[1]' @? 'strict $[1.2]';
+select json '{"a": [1,2,3], "b": [3,4,5]}' @? '$ ? (@.a[*] > @.b[*])';
+select json '{"a": [1,2,3], "b": [3,4,5]}' @? '$ ? (@.a[*] >= @.b[*])';
+select json '{"a": [1,2,3], "b": [3,4,"5"]}' @? '$ ? (@.a[*] >= @.b[*])';
+select json '{"a": [1,2,3], "b": [3,4,"5"]}' @? 'strict $ ? (@.a[*] >= @.b[*])';
+select json '{"a": [1,2,3], "b": [3,4,null]}' @? '$ ? (@.a[*] >= @.b[*])';
+select json '1' @? '$ ? ((@ == "1") is unknown)';
+select json '1' @? '$ ? ((@ == 1) is unknown)';
+select json '[{"a": 1}, {"a": 2}]' @? '$[0 to 1] ? (@.a > 1)';
+
+select json_path_query('1', 'lax $.a');
+select json_path_query('1', 'strict $.a');
+select json_path_query('1', 'strict $.*');
+select json_path_query('[]', 'lax $.a');
+select json_path_query('[]', 'strict $.a');
+select json_path_query('{}', 'lax $.a');
+select json_path_query('{}', 'strict $.a');
+
+select json_path_query('1', 'strict $[1]');
+select json_path_query('1', 'strict $[*]');
+select json_path_query('[]', 'strict $[1]');
+select json_path_query('[]', 'strict $["a"]');
+
+select json_path_query('{"a": 12, "b": {"a": 13}}', '$.a');
+select json_path_query('{"a": 12, "b": {"a": 13}}', '$.b');
+select json_path_query('{"a": 12, "b": {"a": 13}}', '$.*');
+select json_path_query('{"a": 12, "b": {"a": 13}}', 'lax $.*.a');
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[*].a');
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[*].*');
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[0].a');
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[1].a');
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[2].a');
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[0,1].a');
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[0 to 10].a');
+select json_path_query('[12, {"a": 13}, {"b": 14}]', 'lax $[0 to 10 / 0].a');
+select json_path_query('[12, {"a": 13}, {"b": 14}, "ccc", true]', '$[2.5 - 1 to $.size() - 2]');
+select json_path_query('1', 'lax $[0]');
+select json_path_query('1', 'lax $[*]');
+select json_path_query('[1]', 'lax $[0]');
+select json_path_query('[1]', 'lax $[*]');
+select json_path_query('[1,2,3]', 'lax $[*]');
+select json_path_query('[1,2,3]', 'strict $[*].a');
+select json_path_query('[]', '$[last]');
+select json_path_query('[]', '$[last ? (exists(last))]');
+select json_path_query('[]', 'strict $[last]');
+select json_path_query('[1]', '$[last]');
+select json_path_query('[1,2,3]', '$[last]');
+select json_path_query('[1,2,3]', '$[last - 1]');
+select json_path_query('[1,2,3]', '$[last ? (@.type() == "number")]');
+select json_path_query('[1,2,3]', '$[last ? (@.type() == "string")]');
+
+select * from json_path_query('{"a": 10}', '$');
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)');
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)', '1');
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)', '[{"value" : 13}]');
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)', '{"value" : 13}');
+select * from json_path_query('{"a": 10}', '$ ? (@.a < $value)', '{"value" : 8}');
+select * from json_path_query('{"a": 10}', '$.a ? (@ < $value)', '{"value" : 13}');
+select * from json_path_query('[10,11,12,13,14,15]', '$[*] ? (@ < $value)', '{"value" : 13}');
+select * from json_path_query('[10,11,12,13,14,15]', '$[0,1] ? (@ < $x.value)', '{"x": {"value" : 13}}');
+select * from json_path_query('[10,11,12,13,14,15]', '$[0 to 2] ? (@ < $value)', '{"value" : 15}');
+select * from json_path_query('[1,"1",2,"2",null]', '$[*] ? (@ == "1")');
+select * from json_path_query('[1,"1",2,"2",null]', '$[*] ? (@ == $value)', '{"value" : "1"}');
+select * from json_path_query('[1,"1",2,"2",null]', '$[*] ? (@ == $value)', '{"value" : null}');
+select * from json_path_query('[1, "2", null]', '$[*] ? (@ != null)');
+select * from json_path_query('[1, "2", null]', '$[*] ? (@ == null)');
+select * from json_path_query('{}', '$ ? (@ == @)');
+select * from json_path_query('[]', 'strict $ ? (@ == @)');
+
+select json_path_query('{"a": {"b": 1}}', 'lax $.**');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{0}');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{0 to last}');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1}');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1 to last}');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{2}');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{2 to last}');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{3 to last}');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{last}');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**.b ? (@ > 0)');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{0}.b ? (@ > 0)');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1}.b ? (@ > 0)');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{0 to last}.b ? (@ > 0)');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1 to last}.b ? (@ > 0)');
+select json_path_query('{"a": {"b": 1}}', 'lax $.**{1 to 2}.b ? (@ > 0)');
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**.b ? (@ > 0)');
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{0}.b ? (@ > 0)');
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{1}.b ? (@ > 0)');
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{0 to last}.b ? (@ > 0)');
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{1 to last}.b ? (@ > 0)');
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{1 to 2}.b ? (@ > 0)');
+select json_path_query('{"a": {"c": {"b": 1}}}', 'lax $.**{2 to 3}.b ? (@ > 0)');
+
+select json '{"a": {"b": 1}}' @? '$.**.b ? ( @ > 0)';
+select json '{"a": {"b": 1}}' @? '$.**{0}.b ? ( @ > 0)';
+select json '{"a": {"b": 1}}' @? '$.**{1}.b ? ( @ > 0)';
+select json '{"a": {"b": 1}}' @? '$.**{0 to last}.b ? ( @ > 0)';
+select json '{"a": {"b": 1}}' @? '$.**{1 to last}.b ? ( @ > 0)';
+select json '{"a": {"b": 1}}' @? '$.**{1 to 2}.b ? ( @ > 0)';
+select json '{"a": {"c": {"b": 1}}}' @? '$.**.b ? ( @ > 0)';
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{0}.b ? ( @ > 0)';
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{1}.b ? ( @ > 0)';
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{0 to last}.b ? ( @ > 0)';
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{1 to last}.b ? ( @ > 0)';
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{1 to 2}.b ? ( @ > 0)';
+select json '{"a": {"c": {"b": 1}}}' @? '$.**{2 to 3}.b ? ( @ > 0)';
+
+select json_path_query('{"g": {"x": 2}}', '$.g ? (exists (@.x))');
+select json_path_query('{"g": {"x": 2}}', '$.g ? (exists (@.y))');
+select json_path_query('{"g": {"x": 2}}', '$.g ? (exists (@.x ? (@ >= 2) ))');
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'lax $.g ? (exists (@.x))');
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'lax $.g ? (exists (@.x + "3"))');
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'lax $.g ? ((exists (@.x + "3")) is unknown)');
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'strict $.g[*] ? (exists (@.x))');
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'strict $.g[*] ? ((exists (@.x)) is unknown)');
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'strict $.g ? (exists (@[*].x))');
+select json_path_query('{"g": [{"x": 2}, {"y": 3}]}', 'strict $.g ? ((exists (@[*].x)) is unknown)');
+
+--test ternary logic
+select
+ x, y,
+ json_path_query(
+ '[true, false, null]',
+ '$[*] ? (@ == true && ($x == true && $y == true) ||
+ @ == false && !($x == true && $y == true) ||
+ @ == null && ($x == true && $y == true) is unknown)',
+ json_build_object('x', x, 'y', y)
+ ) as "x && y"
+from
+ (values (json 'true'), ('false'), ('"null"')) x(x),
+ (values (json 'true'), ('false'), ('"null"')) y(y);
+
+select
+ x, y,
+ json_path_query(
+ '[true, false, null]',
+ '$[*] ? (@ == true && ($x == true || $y == true) ||
+ @ == false && !($x == true || $y == true) ||
+ @ == null && ($x == true || $y == true) is unknown)',
+ json_build_object('x', x, 'y', y)
+ ) as "x || y"
+from
+ (values (json 'true'), ('false'), ('"null"')) x(x),
+ (values (json 'true'), ('false'), ('"null"')) y(y);
+
+select json '{"a": 1, "b":1}' @? '$ ? (@.a == @.b)';
+select json '{"c": {"a": 1, "b":1}}' @? '$ ? (@.a == @.b)';
+select json '{"c": {"a": 1, "b":1}}' @? '$.c ? (@.a == @.b)';
+select json '{"c": {"a": 1, "b":1}}' @? '$.c ? ($.c.a == @.b)';
+select json '{"c": {"a": 1, "b":1}}' @? '$.* ? (@.a == @.b)';
+select json '{"a": 1, "b":1}' @? '$.** ? (@.a == @.b)';
+select json '{"c": {"a": 1, "b":1}}' @? '$.** ? (@.a == @.b)';
+
+select json_path_query('{"c": {"a": 2, "b":1}}', '$.** ? (@.a == 1 + 1)');
+select json_path_query('{"c": {"a": 2, "b":1}}', '$.** ? (@.a == (1 + 1))');
+select json_path_query('{"c": {"a": 2, "b":1}}', '$.** ? (@.a == @.b + 1)');
+select json_path_query('{"c": {"a": 2, "b":1}}', '$.** ? (@.a == (@.b + 1))');
+select json '{"c": {"a": -1, "b":1}}' @? '$.** ? (@.a == - 1)';
+select json '{"c": {"a": -1, "b":1}}' @? '$.** ? (@.a == -1)';
+select json '{"c": {"a": -1, "b":1}}' @? '$.** ? (@.a == [email protected])';
+select json '{"c": {"a": -1, "b":1}}' @? '$.** ? (@.a == - @.b)';
+select json '{"c": {"a": 0, "b":1}}' @? '$.** ? (@.a == 1 - @.b)';
+select json '{"c": {"a": 2, "b":1}}' @? '$.** ? (@.a == 1 - - @.b)';
+select json '{"c": {"a": 0, "b":1}}' @? '$.** ? (@.a == 1 - [email protected])';
+select json '[1,2,3]' @? '$ ? (+@[*] > +2)';
+select json '[1,2,3]' @? '$ ? (+@[*] > +3)';
+select json '[1,2,3]' @? '$ ? (-@[*] < -2)';
+select json '[1,2,3]' @? '$ ? (-@[*] < -3)';
+select json '1' @? '$ ? ($ > 0)';
+
+-- arithmetic errors
+select json_path_query('[1,2,0,3]', '$[*] ? (2 / @ > 0)');
+select json_path_query('[1,2,0,3]', '$[*] ? ((2 / @ > 0) is unknown)');
+select json_path_query('0', '1 / $');
+select json_path_query('0', '1 / $ + 2');
+select json_path_query('0', '-(3 + 1 % $)');
+select json_path_query('1', '$ + "2"');
+select json_path_query('[1, 2]', '3 * $');
+select json_path_query('"a"', '-$');
+select json_path_query('[1,"2",3]', '+$');
+select json '["1",2,0,3]' @? '-$[*]';
+select json '[1,"2",0,3]' @? '-$[*]';
+select json '["1",2,0,3]' @? 'strict -$[*]';
+select json '[1,"2",0,3]' @? 'strict -$[*]';
+
+-- unwrapping of operator arguments in lax mode
+select json_path_query('{"a": [2]}', 'lax $.a * 3');
+select json_path_query('{"a": [2]}', 'lax $.a + 3');
+select json_path_query('{"a": [2, 3, 4]}', 'lax -$.a');
+-- should fail
+select json_path_query('{"a": [1, 2]}', 'lax $.a * 3');
+
+-- extension: boolean expressions
+select json_path_query('2', '$ > 1');
+select json_path_query('2', '$ <= 1');
+select json_path_query('2', '$ == "2"');
+select json '2' @? '$ == "2"';
+
+select json '2' @@ '$ > 1';
+select json '2' @@ '$ <= 1';
+select json '2' @@ '$ == "2"';
+select json '2' @@ '1';
+select json '{}' @@ '$';
+select json '[]' @@ '$';
+select json '[1,2,3]' @@ '$[*]';
+select json '[]' @@ '$[*]';
+select json_path_match('[[1, true], [2, false]]', 'strict $[*] ? (@[0] > $x) [1]', '{"x": 1}');
+select json_path_match('[[1, true], [2, false]]', 'strict $[*] ? (@[0] < $x) [1]', '{"x": 2}');
+
+select json_path_query('[null,1,true,"a",[],{}]', '$.type()');
+select json_path_query('[null,1,true,"a",[],{}]', 'lax $.type()');
+select json_path_query('[null,1,true,"a",[],{}]', '$[*].type()');
+select json_path_query('null', 'null.type()');
+select json_path_query('null', 'true.type()');
+select json_path_query('null', '123.type()');
+select json_path_query('null', '"123".type()');
+
+select json_path_query('{"a": 2}', '($.a - 5).abs() + 10');
+select json_path_query('{"a": 2.5}', '-($.a * $.a).floor() % 4.3');
+select json_path_query('[1, 2, 3]', '($[*] > 2) ? (@ == true)');
+select json_path_query('[1, 2, 3]', '($[*] > 3).type()');
+select json_path_query('[1, 2, 3]', '($[*].a > 3).type()');
+select json_path_query('[1, 2, 3]', 'strict ($[*].a > 3).type()');
+
+select json_path_query('[1,null,true,"11",[],[1],[1,2,3],{},{"a":1,"b":2}]', 'strict $[*].size()');
+select json_path_query('[1,null,true,"11",[],[1],[1,2,3],{},{"a":1,"b":2}]', 'lax $[*].size()');
+
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].abs()');
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].floor()');
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].ceiling()');
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].ceiling().abs()');
+select json_path_query('[0, 1, -2, -3.4, 5.6]', '$[*].ceiling().abs().type()');
+
+select json_path_query('[{},1]', '$[*].keyvalue()');
+select json_path_query('{}', '$.keyvalue()');
+select json_path_query('{"a": 1, "b": [1, 2], "c": {"a": "bbb"}}', '$.keyvalue()');
+select json_path_query('[{"a": 1, "b": [1, 2]}, {"c": {"a": "bbb"}}]', '$[*].keyvalue()');
+select json_path_query('[{"a": 1, "b": [1, 2]}, {"c": {"a": "bbb"}}]', 'strict $.keyvalue()');
+select json_path_query('[{"a": 1, "b": [1, 2]}, {"c": {"a": "bbb"}}]', 'lax $.keyvalue()');
+select json_path_query('[{"a": 1, "b": [1, 2]}, {"c": {"a": "bbb"}}]', 'strict $.keyvalue().a');
+select json '{"a": 1, "b": [1, 2]}' @? 'lax $.keyvalue()';
+select json '{"a": 1, "b": [1, 2]}' @? 'lax $.keyvalue().key';
+
+select json_path_query('null', '$.double()');
+select json_path_query('true', '$.double()');
+select json_path_query('[]', '$.double()');
+select json_path_query('[]', 'strict $.double()');
+select json_path_query('{}', '$.double()');
+select json_path_query('1.23', '$.double()');
+select json_path_query('"1.23"', '$.double()');
+select json_path_query('"1.23aaa"', '$.double()');
+select json_path_query('"nan"', '$.double()');
+select json_path_query('"NaN"', '$.double()');
+select json_path_query('"inf"', '$.double()');
+select json_path_query('"-inf"', '$.double()');
+
+select json_path_query('{}', '$.abs()');
+select json_path_query('true', '$.floor()');
+select json_path_query('"1.2"', '$.ceiling()');
+
+select json_path_query('["", "a", "abc", "abcabc"]', '$[*] ? (@ starts with "abc")');
+select json_path_query('["", "a", "abc", "abcabc"]', 'strict $ ? (@[*] starts with "abc")');
+select json_path_query('["", "a", "abd", "abdabc"]', 'strict $ ? (@[*] starts with "abc")');
+select json_path_query('["abc", "abcabc", null, 1]', 'strict $ ? (@[*] starts with "abc")');
+select json_path_query('["abc", "abcabc", null, 1]', 'strict $ ? ((@[*] starts with "abc") is unknown)');
+select json_path_query('[[null, 1, "abc", "abcabc"]]', 'lax $ ? (@[*] starts with "abc")');
+select json_path_query('[[null, 1, "abd", "abdabc"]]', 'lax $ ? ((@[*] starts with "abc") is unknown)');
+select json_path_query('[null, 1, "abd", "abdabc"]', 'lax $[*] ? ((@ starts with "abc") is unknown)');
+
+select json_path_query('[null, 1, "abc", "abd", "aBdC", "abdacb", "adc\nabc", "babc"]', 'lax $[*] ? (@ like_regex "^ab.*c")');
+select json_path_query('[null, 1, "abc", "abd", "aBdC", "abdacb", "adc\nabc", "babc"]', 'lax $[*] ? (@ like_regex "^a b.* c " flag "ix")');
+select json_path_query('[null, 1, "abc", "abd", "aBdC", "abdacb", "adc\nabc", "babc"]', 'lax $[*] ? (@ like_regex "^ab.*c" flag "m")');
+select json_path_query('[null, 1, "abc", "abd", "aBdC", "abdacb", "adc\nabc", "babc"]', 'lax $[*] ? (@ like_regex "^ab.*c" flag "s")');
+
+select json_path_query('null', '$.datetime()');
+select json_path_query('true', '$.datetime()');
+select json_path_query('1', '$.datetime()');
+select json_path_query('[]', '$.datetime()');
+select json_path_query('[]', 'strict $.datetime()');
+select json_path_query('{}', '$.datetime()');
+select json_path_query('""', '$.datetime()');
+select json_path_query('"12:34"', '$.datetime("aaa")');
+select json_path_query('"12:34"', '$.datetime("aaa", 1)');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 1)');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 10000000000000)');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 10000000000000)');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 2147483647)');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", 2147483648)');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", -2147483647)');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", -2147483648)');
+select json_path_query('"aaaa"', '$.datetime("HH24")');
+
+select json '"10-03-2017"' @? '$.datetime("dd-mm-yyyy")';
+select json_path_query('"10-03-2017"', '$.datetime("dd-mm-yyyy")');
+select json_path_query('"10-03-2017"', '$.datetime("dd-mm-yyyy").type()');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy").type()');
+
+select json_path_query('"10-03-2017 12:34"', ' $.datetime("dd-mm-yyyy HH24:MI").type()');
+select json_path_query('"10-03-2017 12:34 +05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM").type()');
+select json_path_query('"12:34:56"', '$.datetime("HH24:MI:SS").type()');
+select json_path_query('"12:34:56 +05:20"', '$.datetime("HH24:MI:SS TZH:TZM").type()');
+
+set time zone '+00';
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "+00")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "+00:12")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "-00:12:34")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "UTC")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", 12 * 60)');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", -(5 * 3600 + 12 * 60 + 34))');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", $tz)',
+ json_build_object('tz', extract(timezone from now())));
+select json_path_query('"10-03-2017 12:34 +05"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+select json_path_query('"10-03-2017 12:34 -05"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+select json_path_query('"10-03-2017 12:34 +05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM")');
+select json_path_query('"10-03-2017 12:34 -05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM")');
+select json_path_query('"12:34"', '$.datetime("HH24:MI")');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH")');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", "+00")');
+select json_path_query('"12:34 +05"', '$.datetime("HH24:MI TZH")');
+select json_path_query('"12:34 -05"', '$.datetime("HH24:MI TZH")');
+select json_path_query('"12:34 +05:20"', '$.datetime("HH24:MI TZH:TZM")');
+select json_path_query('"12:34 -05:20"', '$.datetime("HH24:MI TZH:TZM")');
+
+set time zone '+10';
+
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", "+10")');
+select json_path_query('"10-03-2017 12:34"', '$.datetime("dd-mm-yyyy HH24:MI TZH", $tz)',
+ json_build_object('tz', extract(timezone from now())));
+select json_path_query('"10-03-2017 12:34 +05"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+select json_path_query('"10-03-2017 12:34 -05"', '$.datetime("dd-mm-yyyy HH24:MI TZH")');
+select json_path_query('"10-03-2017 12:34 +05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM")');
+select json_path_query('"10-03-2017 12:34 -05:20"', '$.datetime("dd-mm-yyyy HH24:MI TZH:TZM")');
+select json_path_query('"12:34"', '$.datetime("HH24:MI")');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH")');
+select json_path_query('"12:34"', '$.datetime("HH24:MI TZH", "+10")');
+select json_path_query('"12:34 +05"', '$.datetime("HH24:MI TZH")');
+select json_path_query('"12:34 -05"', '$.datetime("HH24:MI TZH")');
+select json_path_query('"12:34 +05:20"', '$.datetime("HH24:MI TZH:TZM")');
+select json_path_query('"12:34 -05:20"', '$.datetime("HH24:MI TZH:TZM")');
+
+set time zone default;
+
+select json_path_query('"2017-03-10"', '$.datetime().type()');
+select json_path_query('"2017-03-10"', '$.datetime()');
+select json_path_query('"2017-03-10 12:34:56"', '$.datetime().type()');
+select json_path_query('"2017-03-10 12:34:56"', '$.datetime()');
+select json_path_query('"2017-03-10 12:34:56 +3"', '$.datetime().type()');
+select json_path_query('"2017-03-10 12:34:56 +3"', '$.datetime()');
+select json_path_query('"2017-03-10 12:34:56 +3:10"', '$.datetime().type()');
+select json_path_query('"2017-03-10 12:34:56 +3:10"', '$.datetime()');
+select json_path_query('"12:34:56"', '$.datetime().type()');
+select json_path_query('"12:34:56"', '$.datetime()');
+select json_path_query('"12:34:56 +3"', '$.datetime().type()');
+select json_path_query('"12:34:56 +3"', '$.datetime()');
+select json_path_query('"12:34:56 +3:10"', '$.datetime().type()');
+select json_path_query('"12:34:56 +3:10"', '$.datetime()');
+
+set time zone '+00';
+
+-- date comparison
+select json_path_query(
+ '["2017-03-10", "2017-03-11", "2017-03-09", "12:34:56", "01:02:03 +04", "2017-03-10 00:00:00", "2017-03-10 12:34:56", "2017-03-10 01:02:03 +04", "2017-03-10 03:00:00 +03"]',
+ '$[*].datetime() ? (@ == "10.03.2017".datetime("dd.mm.yyyy"))');
+select json_path_query(
+ '["2017-03-10", "2017-03-11", "2017-03-09", "12:34:56", "01:02:03 +04", "2017-03-10 00:00:00", "2017-03-10 12:34:56", "2017-03-10 01:02:03 +04", "2017-03-10 03:00:00 +03"]',
+ '$[*].datetime() ? (@ >= "10.03.2017".datetime("dd.mm.yyyy"))');
+select json_path_query(
+ '["2017-03-10", "2017-03-11", "2017-03-09", "12:34:56", "01:02:03 +04", "2017-03-10 00:00:00", "2017-03-10 12:34:56", "2017-03-10 01:02:03 +04", "2017-03-10 03:00:00 +03"]',
+ '$[*].datetime() ? (@ < "10.03.2017".datetime("dd.mm.yyyy"))');
+
+-- time comparison
+select json_path_query(
+ '["12:34:00", "12:35:00", "12:36:00", "12:35:00 +00", "12:35:00 +01", "13:35:00 +01", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +01"]',
+ '$[*].datetime() ? (@ == "12:35".datetime("HH24:MI"))');
+select json_path_query(
+ '["12:34:00", "12:35:00", "12:36:00", "12:35:00 +00", "12:35:00 +01", "13:35:00 +01", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +01"]',
+ '$[*].datetime() ? (@ >= "12:35".datetime("HH24:MI"))');
+select json_path_query(
+ '["12:34:00", "12:35:00", "12:36:00", "12:35:00 +00", "12:35:00 +01", "13:35:00 +01", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +01"]',
+ '$[*].datetime() ? (@ < "12:35".datetime("HH24:MI"))');
+
+-- timetz comparison
+select json_path_query(
+ '["12:34:00 +01", "12:35:00 +01", "12:36:00 +01", "12:35:00 +02", "12:35:00 -02", "10:35:00", "11:35:00", "12:35:00", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +1"]',
+ '$[*].datetime() ? (@ == "12:35 +1".datetime("HH24:MI TZH"))');
+select json_path_query(
+ '["12:34:00 +01", "12:35:00 +01", "12:36:00 +01", "12:35:00 +02", "12:35:00 -02", "10:35:00", "11:35:00", "12:35:00", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +1"]',
+ '$[*].datetime() ? (@ >= "12:35 +1".datetime("HH24:MI TZH"))');
+select json_path_query(
+ '["12:34:00 +01", "12:35:00 +01", "12:36:00 +01", "12:35:00 +02", "12:35:00 -02", "10:35:00", "11:35:00", "12:35:00", "2017-03-10", "2017-03-10 12:35:00", "2017-03-10 12:35:00 +1"]',
+ '$[*].datetime() ? (@ < "12:35 +1".datetime("HH24:MI TZH"))');
+
+-- timestamp comparison
+select json_path_query(
+ '["2017-03-10 12:34:00", "2017-03-10 12:35:00", "2017-03-10 12:36:00", "2017-03-10 12:35:00 +01", "2017-03-10 13:35:00 +01", "2017-03-10 12:35:00 -01", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ == "10.03.2017 12:35".datetime("dd.mm.yyyy HH24:MI"))');
+select json_path_query(
+ '["2017-03-10 12:34:00", "2017-03-10 12:35:00", "2017-03-10 12:36:00", "2017-03-10 12:35:00 +01", "2017-03-10 13:35:00 +01", "2017-03-10 12:35:00 -01", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ >= "10.03.2017 12:35".datetime("dd.mm.yyyy HH24:MI"))');
+select json_path_query(
+ '["2017-03-10 12:34:00", "2017-03-10 12:35:00", "2017-03-10 12:36:00", "2017-03-10 12:35:00 +01", "2017-03-10 13:35:00 +01", "2017-03-10 12:35:00 -01", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ < "10.03.2017 12:35".datetime("dd.mm.yyyy HH24:MI"))');
+
+-- timestamptz comparison
+select json_path_query(
+ '["2017-03-10 12:34:00 +01", "2017-03-10 12:35:00 +01", "2017-03-10 12:36:00 +01", "2017-03-10 12:35:00 +02", "2017-03-10 12:35:00 -02", "2017-03-10 10:35:00", "2017-03-10 11:35:00", "2017-03-10 12:35:00", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ == "10.03.2017 12:35 +1".datetime("dd.mm.yyyy HH24:MI TZH"))');
+select json_path_query(
+ '["2017-03-10 12:34:00 +01", "2017-03-10 12:35:00 +01", "2017-03-10 12:36:00 +01", "2017-03-10 12:35:00 +02", "2017-03-10 12:35:00 -02", "2017-03-10 10:35:00", "2017-03-10 11:35:00", "2017-03-10 12:35:00", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ >= "10.03.2017 12:35 +1".datetime("dd.mm.yyyy HH24:MI TZH"))');
+select json_path_query(
+ '["2017-03-10 12:34:00 +01", "2017-03-10 12:35:00 +01", "2017-03-10 12:36:00 +01", "2017-03-10 12:35:00 +02", "2017-03-10 12:35:00 -02", "2017-03-10 10:35:00", "2017-03-10 11:35:00", "2017-03-10 12:35:00", "2017-03-10", "2017-03-11", "12:34:56", "12:34:56 +01"]',
+ '$[*].datetime() ? (@ < "10.03.2017 12:35 +1".datetime("dd.mm.yyyy HH24:MI TZH"))');
+
+set time zone default;
+
+-- jsonpath operators
+
+SELECT json_path_query('[{"a": 1}, {"a": 2}]', '$[*]');
+SELECT json_path_query('[{"a": 1}, {"a": 2}]', '$[*] ? (@.a > 10)');
+
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a');
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a');
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ == 1)');
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ > 10)');
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', '{"min": 1, "max": 4}');
+SELECT json_path_query_array('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', '{"min": 3, "max": 4}');
+
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a');
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}]', '$[*].a');
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ == 1)');
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ > 10)');
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', '{"min": 1, "max": 4}');
+SELECT json_path_query_first('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', '{"min": 3, "max": 4}');
+
+SELECT json '[{"a": 1}, {"a": 2}]' @? '$[*].a ? (@ > 1)';
+SELECT json '[{"a": 1}, {"a": 2}]' @? '$[*] ? (@.a > 2)';
+SELECT json_path_exists('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*] ? (@.a > $min && @.a < $max)', '{"min": 1, "max": 4}');
+SELECT json_path_exists('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*] ? (@.a > $min && @.a < $max)', '{"min": 3, "max": 4}');
+
+SELECT json '[{"a": 1}, {"a": 2}]' @@ '$[*].a > 1';
+SELECT json '[{"a": 1}, {"a": 2}]' @@ '$[*].a > 2';
diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list
index daaafa3..5a1518e 100644
--- a/src/tools/pgindent/typedefs.list
+++ b/src/tools/pgindent/typedefs.list
@@ -1097,6 +1097,7 @@ JsValue
JsonAggState
JsonBaseObjectInfo
JsonHashEntry
+JsonItem
JsonItemStack
JsonItemStackEntry
JsonIterateStringValuesAction
@@ -1112,6 +1113,7 @@ JsonPathItemType
JsonPathParseItem
JsonPathParseResult
JsonPathPredicateCallback
+JsonPathUserFuncContext
JsonPathVariable
JsonPathVariable_cb
JsonSemAction
--
2.7.4
--------------FD6F0D8EDD0CFB174B99F31A
Content-Type: text/x-patch;
name="0005-Jsonpath-GIN-support-v34.patch"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
filename="0005-Jsonpath-GIN-support-v34.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH 3/3] Add operator <->(box, point) to SP-GiST box_ops
@ 2019-03-07 20:49 Nikita Glukhov <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Nikita Glukhov @ 2019-03-07 20:49 UTC (permalink / raw)
---
doc/src/sgml/spgist.sgml | 1 +
src/include/catalog/pg_amop.dat | 4 ++
src/test/regress/expected/box.out | 82 +++++++++++++++++++++++++-----
src/test/regress/expected/sanity_check.out | 2 +
src/test/regress/sql/box.sql | 70 ++++++++++++++++++++-----
5 files changed, 135 insertions(+), 24 deletions(-)
diff --git a/doc/src/sgml/spgist.sgml b/doc/src/sgml/spgist.sgml
index a816856..81ddf5d 100644
--- a/doc/src/sgml/spgist.sgml
+++ b/doc/src/sgml/spgist.sgml
@@ -139,6 +139,7 @@
<literal>|&></literal>
</entry>
<entry>
+ <literal><-></literal>
</entry>
</row>
<row>
diff --git a/src/include/catalog/pg_amop.dat b/src/include/catalog/pg_amop.dat
index ebc38ae..232557e 100644
--- a/src/include/catalog/pg_amop.dat
+++ b/src/include/catalog/pg_amop.dat
@@ -1546,6 +1546,10 @@
amopstrategy => '11', amopopr => '|>>(box,box)', amopmethod => 'spgist' },
{ amopfamily => 'spgist/box_ops', amoplefttype => 'box', amoprighttype => 'box',
amopstrategy => '12', amopopr => '|&>(box,box)', amopmethod => 'spgist' },
+{ amopfamily => 'spgist/box_ops', amoplefttype => 'box',
+ amoprighttype => 'point', amopstrategy => '15', amoppurpose => 'o',
+ amopopr => '<->(box,point)', amopmethod => 'spgist',
+ amopsortfamily => 'btree/float_ops' },
# SP-GiST poly_ops (supports polygons)
{ amopfamily => 'spgist/poly_ops', amoplefttype => 'polygon',
diff --git a/src/test/regress/expected/box.out b/src/test/regress/expected/box.out
index 998b522..4d0f169 100644
--- a/src/test/regress/expected/box.out
+++ b/src/test/regress/expected/box.out
@@ -480,23 +480,33 @@ DROP INDEX box_spgist;
--
-- Test the SP-GiST index on the larger volume of data
--
-CREATE TABLE quad_box_tbl (b box);
+CREATE TABLE quad_box_tbl (id int, b box);
INSERT INTO quad_box_tbl
- SELECT box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
- FROM generate_series(1, 100) x,
- generate_series(1, 100) y;
+ SELECT (x - 1) * 100 + y, box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
+ FROM generate_series(1, 100) x,
+ generate_series(1, 100) y;
-- insert repeating data to test allTheSame
INSERT INTO quad_box_tbl
- SELECT '((200, 300),(210, 310))'
- FROM generate_series(1, 1000);
+ SELECT i, '((200, 300),(210, 310))'
+ FROM generate_series(10001, 11000) AS i;
INSERT INTO quad_box_tbl
- VALUES
- (NULL),
- (NULL),
- ('((-infinity,-infinity),(infinity,infinity))'),
- ('((-infinity,100),(-infinity,500))'),
- ('((-infinity,-infinity),(700,infinity))');
+VALUES
+ (11001, NULL),
+ (11002, NULL),
+ (11003, '((-infinity,-infinity),(infinity,infinity))'),
+ (11004, '((-infinity,100),(-infinity,500))'),
+ (11005, '((-infinity,-infinity),(700,infinity))');
CREATE INDEX quad_box_tbl_idx ON quad_box_tbl USING spgist(b);
+-- get reference results for ORDER BY distance from seq scan
+SET enable_seqscan = ON;
+SET enable_indexscan = OFF;
+SET enable_bitmapscan = OFF;
+CREATE TABLE quad_box_tbl_ord_seq1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+CREATE TABLE quad_box_tbl_ord_seq2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
SET enable_seqscan = OFF;
SET enable_indexscan = ON;
SET enable_bitmapscan = ON;
@@ -578,6 +588,54 @@ SELECT count(*) FROM quad_box_tbl WHERE b ~= box '((200,300),(205,305))';
1
(1 row)
+-- test ORDER BY distance
+SET enable_indexscan = ON;
+SET enable_bitmapscan = OFF;
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+ QUERY PLAN
+---------------------------------------------------------
+ WindowAgg
+ -> Index Scan using quad_box_tbl_idx on quad_box_tbl
+ Order By: (b <-> '(123,456)'::point)
+(3 rows)
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+SELECT *
+FROM quad_box_tbl_ord_seq1 seq FULL JOIN quad_box_tbl_ord_idx1 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+ n | dist | id | n | dist | id
+---+------+----+---+------+----
+(0 rows)
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+ QUERY PLAN
+---------------------------------------------------------
+ WindowAgg
+ -> Index Scan using quad_box_tbl_idx on quad_box_tbl
+ Index Cond: (b <@ '(500,600),(200,300)'::box)
+ Order By: (b <-> '(123,456)'::point)
+(4 rows)
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+SELECT *
+FROM quad_box_tbl_ord_seq2 seq FULL JOIN quad_box_tbl_ord_idx2 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+ n | dist | id | n | dist | id
+---+------+----+---+------+----
+(0 rows)
+
RESET enable_seqscan;
RESET enable_indexscan;
RESET enable_bitmapscan;
diff --git a/src/test/regress/expected/sanity_check.out b/src/test/regress/expected/sanity_check.out
index 8ff0da1..d6e75ff 100644
--- a/src/test/regress/expected/sanity_check.out
+++ b/src/test/regress/expected/sanity_check.out
@@ -165,6 +165,8 @@ pg_user_mapping|t
point_tbl|t
polygon_tbl|t
quad_box_tbl|t
+quad_box_tbl_ord_seq1|f
+quad_box_tbl_ord_seq2|f
quad_point_tbl|t
quad_poly_tbl|t
radix_text_tbl|t
diff --git a/src/test/regress/sql/box.sql b/src/test/regress/sql/box.sql
index 6710fc9..cd3e002 100644
--- a/src/test/regress/sql/box.sql
+++ b/src/test/regress/sql/box.sql
@@ -192,28 +192,41 @@ DROP INDEX box_spgist;
--
-- Test the SP-GiST index on the larger volume of data
--
-CREATE TABLE quad_box_tbl (b box);
+CREATE TABLE quad_box_tbl (id int, b box);
INSERT INTO quad_box_tbl
- SELECT box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
- FROM generate_series(1, 100) x,
- generate_series(1, 100) y;
+ SELECT (x - 1) * 100 + y, box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
+ FROM generate_series(1, 100) x,
+ generate_series(1, 100) y;
-- insert repeating data to test allTheSame
INSERT INTO quad_box_tbl
- SELECT '((200, 300),(210, 310))'
- FROM generate_series(1, 1000);
+ SELECT i, '((200, 300),(210, 310))'
+ FROM generate_series(10001, 11000) AS i;
INSERT INTO quad_box_tbl
- VALUES
- (NULL),
- (NULL),
- ('((-infinity,-infinity),(infinity,infinity))'),
- ('((-infinity,100),(-infinity,500))'),
- ('((-infinity,-infinity),(700,infinity))');
+VALUES
+ (11001, NULL),
+ (11002, NULL),
+ (11003, '((-infinity,-infinity),(infinity,infinity))'),
+ (11004, '((-infinity,100),(-infinity,500))'),
+ (11005, '((-infinity,-infinity),(700,infinity))');
CREATE INDEX quad_box_tbl_idx ON quad_box_tbl USING spgist(b);
+-- get reference results for ORDER BY distance from seq scan
+SET enable_seqscan = ON;
+SET enable_indexscan = OFF;
+SET enable_bitmapscan = OFF;
+
+CREATE TABLE quad_box_tbl_ord_seq1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+CREATE TABLE quad_box_tbl_ord_seq2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
SET enable_seqscan = OFF;
SET enable_indexscan = ON;
SET enable_bitmapscan = ON;
@@ -232,6 +245,39 @@ SELECT count(*) FROM quad_box_tbl WHERE b @> box '((201,301),(202,303))';
SELECT count(*) FROM quad_box_tbl WHERE b <@ box '((100,200),(300,500))';
SELECT count(*) FROM quad_box_tbl WHERE b ~= box '((200,300),(205,305))';
+-- test ORDER BY distance
+SET enable_indexscan = ON;
+SET enable_bitmapscan = OFF;
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+SELECT *
+FROM quad_box_tbl_ord_seq1 seq FULL JOIN quad_box_tbl_ord_idx1 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
+SELECT *
+FROM quad_box_tbl_ord_seq2 seq FULL JOIN quad_box_tbl_ord_idx2 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+
RESET enable_seqscan;
RESET enable_indexscan;
RESET enable_bitmapscan;
--
2.7.4
--------------C466D744B1EAB9FA85D22553--
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH 3/3] Add operator <->(box, point) to SP-GiST box_ops
@ 2019-03-07 20:49 Nikita Glukhov <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Nikita Glukhov @ 2019-03-07 20:49 UTC (permalink / raw)
---
doc/src/sgml/spgist.sgml | 1 +
src/include/catalog/pg_amop.dat | 4 ++
src/test/regress/expected/box.out | 82 +++++++++++++++++++++++++-----
src/test/regress/expected/sanity_check.out | 2 +
src/test/regress/sql/box.sql | 70 ++++++++++++++++++++-----
5 files changed, 135 insertions(+), 24 deletions(-)
diff --git a/doc/src/sgml/spgist.sgml b/doc/src/sgml/spgist.sgml
index 126d1f6..8bc8ad7 100644
--- a/doc/src/sgml/spgist.sgml
+++ b/doc/src/sgml/spgist.sgml
@@ -139,6 +139,7 @@
<literal>|&></literal>
</entry>
<entry>
+ <literal><-></literal>
</entry>
</row>
<row>
diff --git a/src/include/catalog/pg_amop.dat b/src/include/catalog/pg_amop.dat
index 8e74f86..797223d 100644
--- a/src/include/catalog/pg_amop.dat
+++ b/src/include/catalog/pg_amop.dat
@@ -1534,6 +1534,10 @@
amopstrategy => '11', amopopr => '|>>(box,box)', amopmethod => 'spgist' },
{ amopfamily => 'spgist/box_ops', amoplefttype => 'box', amoprighttype => 'box',
amopstrategy => '12', amopopr => '|&>(box,box)', amopmethod => 'spgist' },
+{ amopfamily => 'spgist/box_ops', amoplefttype => 'box',
+ amoprighttype => 'point', amopstrategy => '15', amoppurpose => 'o',
+ amopopr => '<->(box,point)', amopmethod => 'spgist',
+ amopsortfamily => 'btree/float_ops' },
# SP-GiST poly_ops (supports polygons)
{ amopfamily => 'spgist/poly_ops', amoplefttype => 'polygon',
diff --git a/src/test/regress/expected/box.out b/src/test/regress/expected/box.out
index 998b522..4d0f169 100644
--- a/src/test/regress/expected/box.out
+++ b/src/test/regress/expected/box.out
@@ -480,23 +480,33 @@ DROP INDEX box_spgist;
--
-- Test the SP-GiST index on the larger volume of data
--
-CREATE TABLE quad_box_tbl (b box);
+CREATE TABLE quad_box_tbl (id int, b box);
INSERT INTO quad_box_tbl
- SELECT box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
- FROM generate_series(1, 100) x,
- generate_series(1, 100) y;
+ SELECT (x - 1) * 100 + y, box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
+ FROM generate_series(1, 100) x,
+ generate_series(1, 100) y;
-- insert repeating data to test allTheSame
INSERT INTO quad_box_tbl
- SELECT '((200, 300),(210, 310))'
- FROM generate_series(1, 1000);
+ SELECT i, '((200, 300),(210, 310))'
+ FROM generate_series(10001, 11000) AS i;
INSERT INTO quad_box_tbl
- VALUES
- (NULL),
- (NULL),
- ('((-infinity,-infinity),(infinity,infinity))'),
- ('((-infinity,100),(-infinity,500))'),
- ('((-infinity,-infinity),(700,infinity))');
+VALUES
+ (11001, NULL),
+ (11002, NULL),
+ (11003, '((-infinity,-infinity),(infinity,infinity))'),
+ (11004, '((-infinity,100),(-infinity,500))'),
+ (11005, '((-infinity,-infinity),(700,infinity))');
CREATE INDEX quad_box_tbl_idx ON quad_box_tbl USING spgist(b);
+-- get reference results for ORDER BY distance from seq scan
+SET enable_seqscan = ON;
+SET enable_indexscan = OFF;
+SET enable_bitmapscan = OFF;
+CREATE TABLE quad_box_tbl_ord_seq1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+CREATE TABLE quad_box_tbl_ord_seq2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
SET enable_seqscan = OFF;
SET enable_indexscan = ON;
SET enable_bitmapscan = ON;
@@ -578,6 +588,54 @@ SELECT count(*) FROM quad_box_tbl WHERE b ~= box '((200,300),(205,305))';
1
(1 row)
+-- test ORDER BY distance
+SET enable_indexscan = ON;
+SET enable_bitmapscan = OFF;
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+ QUERY PLAN
+---------------------------------------------------------
+ WindowAgg
+ -> Index Scan using quad_box_tbl_idx on quad_box_tbl
+ Order By: (b <-> '(123,456)'::point)
+(3 rows)
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+SELECT *
+FROM quad_box_tbl_ord_seq1 seq FULL JOIN quad_box_tbl_ord_idx1 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+ n | dist | id | n | dist | id
+---+------+----+---+------+----
+(0 rows)
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+ QUERY PLAN
+---------------------------------------------------------
+ WindowAgg
+ -> Index Scan using quad_box_tbl_idx on quad_box_tbl
+ Index Cond: (b <@ '(500,600),(200,300)'::box)
+ Order By: (b <-> '(123,456)'::point)
+(4 rows)
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+SELECT *
+FROM quad_box_tbl_ord_seq2 seq FULL JOIN quad_box_tbl_ord_idx2 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+ n | dist | id | n | dist | id
+---+------+----+---+------+----
+(0 rows)
+
RESET enable_seqscan;
RESET enable_indexscan;
RESET enable_bitmapscan;
diff --git a/src/test/regress/expected/sanity_check.out b/src/test/regress/expected/sanity_check.out
index aaaa488..850de0c 100644
--- a/src/test/regress/expected/sanity_check.out
+++ b/src/test/regress/expected/sanity_check.out
@@ -167,6 +167,8 @@ pg_user_mapping|t
point_tbl|t
polygon_tbl|t
quad_box_tbl|t
+quad_box_tbl_ord_seq1|f
+quad_box_tbl_ord_seq2|f
quad_point_tbl|t
quad_poly_tbl|t
quad_poly_tbl_ord_seq1|f
diff --git a/src/test/regress/sql/box.sql b/src/test/regress/sql/box.sql
index 6710fc9..cd3e002 100644
--- a/src/test/regress/sql/box.sql
+++ b/src/test/regress/sql/box.sql
@@ -192,28 +192,41 @@ DROP INDEX box_spgist;
--
-- Test the SP-GiST index on the larger volume of data
--
-CREATE TABLE quad_box_tbl (b box);
+CREATE TABLE quad_box_tbl (id int, b box);
INSERT INTO quad_box_tbl
- SELECT box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
- FROM generate_series(1, 100) x,
- generate_series(1, 100) y;
+ SELECT (x - 1) * 100 + y, box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
+ FROM generate_series(1, 100) x,
+ generate_series(1, 100) y;
-- insert repeating data to test allTheSame
INSERT INTO quad_box_tbl
- SELECT '((200, 300),(210, 310))'
- FROM generate_series(1, 1000);
+ SELECT i, '((200, 300),(210, 310))'
+ FROM generate_series(10001, 11000) AS i;
INSERT INTO quad_box_tbl
- VALUES
- (NULL),
- (NULL),
- ('((-infinity,-infinity),(infinity,infinity))'),
- ('((-infinity,100),(-infinity,500))'),
- ('((-infinity,-infinity),(700,infinity))');
+VALUES
+ (11001, NULL),
+ (11002, NULL),
+ (11003, '((-infinity,-infinity),(infinity,infinity))'),
+ (11004, '((-infinity,100),(-infinity,500))'),
+ (11005, '((-infinity,-infinity),(700,infinity))');
CREATE INDEX quad_box_tbl_idx ON quad_box_tbl USING spgist(b);
+-- get reference results for ORDER BY distance from seq scan
+SET enable_seqscan = ON;
+SET enable_indexscan = OFF;
+SET enable_bitmapscan = OFF;
+
+CREATE TABLE quad_box_tbl_ord_seq1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+CREATE TABLE quad_box_tbl_ord_seq2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
SET enable_seqscan = OFF;
SET enable_indexscan = ON;
SET enable_bitmapscan = ON;
@@ -232,6 +245,39 @@ SELECT count(*) FROM quad_box_tbl WHERE b @> box '((201,301),(202,303))';
SELECT count(*) FROM quad_box_tbl WHERE b <@ box '((100,200),(300,500))';
SELECT count(*) FROM quad_box_tbl WHERE b ~= box '((200,300),(205,305))';
+-- test ORDER BY distance
+SET enable_indexscan = ON;
+SET enable_bitmapscan = OFF;
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+SELECT *
+FROM quad_box_tbl_ord_seq1 seq FULL JOIN quad_box_tbl_ord_idx1 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
+SELECT *
+FROM quad_box_tbl_ord_seq2 seq FULL JOIN quad_box_tbl_ord_idx2 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+
RESET enable_seqscan;
RESET enable_indexscan;
RESET enable_bitmapscan;
--
2.7.4
--------------B01ADF66983FA504E371CA13--
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH 3/3] Add operator <->(box, point) to SP-GiST box_ops
@ 2019-03-07 20:49 Nikita Glukhov <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Nikita Glukhov @ 2019-03-07 20:49 UTC (permalink / raw)
---
doc/src/sgml/spgist.sgml | 1 +
src/include/catalog/pg_amop.dat | 4 ++
src/test/regress/expected/box.out | 82 +++++++++++++++++++++++++-----
src/test/regress/expected/sanity_check.out | 2 +
src/test/regress/sql/box.sql | 70 ++++++++++++++++++++-----
5 files changed, 135 insertions(+), 24 deletions(-)
diff --git a/doc/src/sgml/spgist.sgml b/doc/src/sgml/spgist.sgml
index 7cf9d0e..9070914 100644
--- a/doc/src/sgml/spgist.sgml
+++ b/doc/src/sgml/spgist.sgml
@@ -139,6 +139,7 @@
<literal>|&></literal>
</entry>
<entry>
+ <literal><-></literal>
</entry>
</row>
<row>
diff --git a/src/include/catalog/pg_amop.dat b/src/include/catalog/pg_amop.dat
index ebc38ae..232557e 100644
--- a/src/include/catalog/pg_amop.dat
+++ b/src/include/catalog/pg_amop.dat
@@ -1546,6 +1546,10 @@
amopstrategy => '11', amopopr => '|>>(box,box)', amopmethod => 'spgist' },
{ amopfamily => 'spgist/box_ops', amoplefttype => 'box', amoprighttype => 'box',
amopstrategy => '12', amopopr => '|&>(box,box)', amopmethod => 'spgist' },
+{ amopfamily => 'spgist/box_ops', amoplefttype => 'box',
+ amoprighttype => 'point', amopstrategy => '15', amoppurpose => 'o',
+ amopopr => '<->(box,point)', amopmethod => 'spgist',
+ amopsortfamily => 'btree/float_ops' },
# SP-GiST poly_ops (supports polygons)
{ amopfamily => 'spgist/poly_ops', amoplefttype => 'polygon',
diff --git a/src/test/regress/expected/box.out b/src/test/regress/expected/box.out
index 998b522..4d0f169 100644
--- a/src/test/regress/expected/box.out
+++ b/src/test/regress/expected/box.out
@@ -480,23 +480,33 @@ DROP INDEX box_spgist;
--
-- Test the SP-GiST index on the larger volume of data
--
-CREATE TABLE quad_box_tbl (b box);
+CREATE TABLE quad_box_tbl (id int, b box);
INSERT INTO quad_box_tbl
- SELECT box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
- FROM generate_series(1, 100) x,
- generate_series(1, 100) y;
+ SELECT (x - 1) * 100 + y, box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
+ FROM generate_series(1, 100) x,
+ generate_series(1, 100) y;
-- insert repeating data to test allTheSame
INSERT INTO quad_box_tbl
- SELECT '((200, 300),(210, 310))'
- FROM generate_series(1, 1000);
+ SELECT i, '((200, 300),(210, 310))'
+ FROM generate_series(10001, 11000) AS i;
INSERT INTO quad_box_tbl
- VALUES
- (NULL),
- (NULL),
- ('((-infinity,-infinity),(infinity,infinity))'),
- ('((-infinity,100),(-infinity,500))'),
- ('((-infinity,-infinity),(700,infinity))');
+VALUES
+ (11001, NULL),
+ (11002, NULL),
+ (11003, '((-infinity,-infinity),(infinity,infinity))'),
+ (11004, '((-infinity,100),(-infinity,500))'),
+ (11005, '((-infinity,-infinity),(700,infinity))');
CREATE INDEX quad_box_tbl_idx ON quad_box_tbl USING spgist(b);
+-- get reference results for ORDER BY distance from seq scan
+SET enable_seqscan = ON;
+SET enable_indexscan = OFF;
+SET enable_bitmapscan = OFF;
+CREATE TABLE quad_box_tbl_ord_seq1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+CREATE TABLE quad_box_tbl_ord_seq2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
SET enable_seqscan = OFF;
SET enable_indexscan = ON;
SET enable_bitmapscan = ON;
@@ -578,6 +588,54 @@ SELECT count(*) FROM quad_box_tbl WHERE b ~= box '((200,300),(205,305))';
1
(1 row)
+-- test ORDER BY distance
+SET enable_indexscan = ON;
+SET enable_bitmapscan = OFF;
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+ QUERY PLAN
+---------------------------------------------------------
+ WindowAgg
+ -> Index Scan using quad_box_tbl_idx on quad_box_tbl
+ Order By: (b <-> '(123,456)'::point)
+(3 rows)
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+SELECT *
+FROM quad_box_tbl_ord_seq1 seq FULL JOIN quad_box_tbl_ord_idx1 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+ n | dist | id | n | dist | id
+---+------+----+---+------+----
+(0 rows)
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+ QUERY PLAN
+---------------------------------------------------------
+ WindowAgg
+ -> Index Scan using quad_box_tbl_idx on quad_box_tbl
+ Index Cond: (b <@ '(500,600),(200,300)'::box)
+ Order By: (b <-> '(123,456)'::point)
+(4 rows)
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+SELECT *
+FROM quad_box_tbl_ord_seq2 seq FULL JOIN quad_box_tbl_ord_idx2 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+ n | dist | id | n | dist | id
+---+------+----+---+------+----
+(0 rows)
+
RESET enable_seqscan;
RESET enable_indexscan;
RESET enable_bitmapscan;
diff --git a/src/test/regress/expected/sanity_check.out b/src/test/regress/expected/sanity_check.out
index 8ff0da1..d6e75ff 100644
--- a/src/test/regress/expected/sanity_check.out
+++ b/src/test/regress/expected/sanity_check.out
@@ -165,6 +165,8 @@ pg_user_mapping|t
point_tbl|t
polygon_tbl|t
quad_box_tbl|t
+quad_box_tbl_ord_seq1|f
+quad_box_tbl_ord_seq2|f
quad_point_tbl|t
quad_poly_tbl|t
radix_text_tbl|t
diff --git a/src/test/regress/sql/box.sql b/src/test/regress/sql/box.sql
index 6710fc9..cd3e002 100644
--- a/src/test/regress/sql/box.sql
+++ b/src/test/regress/sql/box.sql
@@ -192,28 +192,41 @@ DROP INDEX box_spgist;
--
-- Test the SP-GiST index on the larger volume of data
--
-CREATE TABLE quad_box_tbl (b box);
+CREATE TABLE quad_box_tbl (id int, b box);
INSERT INTO quad_box_tbl
- SELECT box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
- FROM generate_series(1, 100) x,
- generate_series(1, 100) y;
+ SELECT (x - 1) * 100 + y, box(point(x * 10, y * 10), point(x * 10 + 5, y * 10 + 5))
+ FROM generate_series(1, 100) x,
+ generate_series(1, 100) y;
-- insert repeating data to test allTheSame
INSERT INTO quad_box_tbl
- SELECT '((200, 300),(210, 310))'
- FROM generate_series(1, 1000);
+ SELECT i, '((200, 300),(210, 310))'
+ FROM generate_series(10001, 11000) AS i;
INSERT INTO quad_box_tbl
- VALUES
- (NULL),
- (NULL),
- ('((-infinity,-infinity),(infinity,infinity))'),
- ('((-infinity,100),(-infinity,500))'),
- ('((-infinity,-infinity),(700,infinity))');
+VALUES
+ (11001, NULL),
+ (11002, NULL),
+ (11003, '((-infinity,-infinity),(infinity,infinity))'),
+ (11004, '((-infinity,100),(-infinity,500))'),
+ (11005, '((-infinity,-infinity),(700,infinity))');
CREATE INDEX quad_box_tbl_idx ON quad_box_tbl USING spgist(b);
+-- get reference results for ORDER BY distance from seq scan
+SET enable_seqscan = ON;
+SET enable_indexscan = OFF;
+SET enable_bitmapscan = OFF;
+
+CREATE TABLE quad_box_tbl_ord_seq1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+CREATE TABLE quad_box_tbl_ord_seq2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
SET enable_seqscan = OFF;
SET enable_indexscan = ON;
SET enable_bitmapscan = ON;
@@ -232,6 +245,39 @@ SELECT count(*) FROM quad_box_tbl WHERE b @> box '((201,301),(202,303))';
SELECT count(*) FROM quad_box_tbl WHERE b <@ box '((100,200),(300,500))';
SELECT count(*) FROM quad_box_tbl WHERE b ~= box '((200,300),(205,305))';
+-- test ORDER BY distance
+SET enable_indexscan = ON;
+SET enable_bitmapscan = OFF;
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx1 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl;
+
+SELECT *
+FROM quad_box_tbl_ord_seq1 seq FULL JOIN quad_box_tbl_ord_idx1 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+
+
+EXPLAIN (COSTS OFF)
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
+CREATE TEMP TABLE quad_box_tbl_ord_idx2 AS
+SELECT rank() OVER (ORDER BY b <-> point '123,456') n, b <-> point '123,456' dist, id
+FROM quad_box_tbl WHERE b <@ box '((200,300),(500,600))';
+
+SELECT *
+FROM quad_box_tbl_ord_seq2 seq FULL JOIN quad_box_tbl_ord_idx2 idx
+ ON seq.n = idx.n AND seq.id = idx.id AND
+ (seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
+WHERE seq.id IS NULL OR idx.id IS NULL;
+
RESET enable_seqscan;
RESET enable_indexscan;
RESET enable_bitmapscan;
--
2.7.4
--------------07EE9595454874974EAD75E7--
^ permalink raw reply [nested|flat] 111+ messages in thread
* Which hook to use when overriding utility commands (COPY ...)
@ 2022-03-19 11:28 Hannu Krosing <[email protected]>
2022-03-19 12:44 ` Re: Which hook to use when overriding utility commands (COPY ...) Michael Paquier <[email protected]>
0 siblings, 1 reply; 111+ messages in thread
From: Hannu Krosing @ 2022-03-19 11:28 UTC (permalink / raw)
To: pgsql-hackers
Hi Pgsql-Hackers
Which hook should I use when overriding the COPY command in an extension?
I am working on adding new functionalities to COPY (compression, index
management, various other transports in addition to stdin and file, other
data formats, etc...) and while the aim is to contribute this to v15 I
would also like to have much of it in earlier versions.
As the current policy is to back-port only bugfixes and not "features" ,
the only way I can see to get it in earlier versions is to provide an
extension which intercepts the COPY command and replaces it with my own
implementation.
So my question is, which of the hooks would be easiest to use for this ?
At the syntax level it would still look the same COPY ... FROM/TO ... WITH
( options) and the extensibility will be in file names (using a URI scheme
mapped to transports) and in options part.
so I hope to fully reuse the parsing part and get in before the existence
checks.
Does anyone have experience in this and can [point to samples?
Cheers
Hannu
^ permalink raw reply [nested|flat] 111+ messages in thread
* Re: Which hook to use when overriding utility commands (COPY ...)
2022-03-19 11:28 Which hook to use when overriding utility commands (COPY ...) Hannu Krosing <[email protected]>
@ 2022-03-19 12:44 ` Michael Paquier <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Michael Paquier @ 2022-03-19 12:44 UTC (permalink / raw)
To: Hannu Krosing <[email protected]>; +Cc: pgsql-hackers
On Sat, Mar 19, 2022 at 12:28:46PM +0100, Hannu Krosing wrote:
> Which hook should I use when overriding the COPY command in an extension?
CopyStmt goes through ProcessUtility(), so you can use the hook called
ProcessUtility_hook to override what you want.
--
Michael
Attachments:
[application/pgp-signature] signature.asc (833B, ../../[email protected]/2-signature.asc)
download
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v1 5/7] Row pattern recognition patch (docs).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 51 ++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 18 ++++++++--
3 files changed, 136 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..23ee285b40 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,57 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..c0fc16d773 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,20 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses.
+
+<synopsis>
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Jun_25_21_05_09_2023_126)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v1-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v1 5/7] Row pattern recognition patch (docs).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 51 ++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 18 ++++++++--
3 files changed, 136 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..23ee285b40 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,57 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..c0fc16d773 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,20 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses.
+
+<synopsis>
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Jun_25_21_05_09_2023_126)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v1-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v1 5/7] Row pattern recognition patch (docs).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 51 ++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 18 ++++++++--
3 files changed, 136 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..23ee285b40 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,57 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..c0fc16d773 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,20 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses.
+
+<synopsis>
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Jun_25_21_05_09_2023_126)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v1-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v1 5/7] Row pattern recognition patch (docs).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 51 ++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 18 ++++++++--
3 files changed, 136 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..23ee285b40 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,57 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..c0fc16d773 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,20 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses.
+
+<synopsis>
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Jun_25_21_05_09_2023_126)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v1-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v2 5/7] Row pattern recognition patch (docs).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 +++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++--
3 files changed, 148 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..e9bbd5bc7c 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Jun_26_17_45_07_2023_724)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v2-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v2 5/7] Row pattern recognition patch (docs).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 +++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++--
3 files changed, 148 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..e9bbd5bc7c 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Jun_26_17_45_07_2023_724)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v2-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v2 5/7] Row pattern recognition patch (docs).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 +++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++--
3 files changed, 148 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..e9bbd5bc7c 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Jun_26_17_45_07_2023_724)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v2-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v2 5/7] Row pattern recognition patch (docs).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 +++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++--
3 files changed, 148 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..e9bbd5bc7c 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Jun_26_17_45_07_2023_724)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v2-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v2 5/7] Row pattern recognition patch (docs).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 +++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 69 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++--
3 files changed, 148 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..e9bbd5bc7c 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Window function <function>rpr</function> can be used with row pattern
+ common syntax to perform row pattern recognition in a query. Row pattern
+ common syntax includes two sub clauses. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An
+ example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, rpr(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5a47ce4343..8069c58ca5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21648,6 +21648,22 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>rpr</primary>
+ </indexterm>
+ <function>rpr</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Perform row pattern recognition using column specified
+ by <parameter>value</parameter> and returns the value of the column if
+ current row is the first matching row;
+ returns <literal>NULL</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
@@ -21687,6 +21703,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Jun_26_17_45_07_2023_724)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v2-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v3 5/7] Row pattern recognition patch (docs).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++++++--
3 files changed, 133 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index dcc9d6f59d..b333d62410 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21714,6 +21714,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21753,6 +21754,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Jul_26_21_21_34_2023_317)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v3-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v3 5/7] Row pattern recognition patch (docs).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++++++--
3 files changed, 133 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index dcc9d6f59d..b333d62410 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21714,6 +21714,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21753,6 +21754,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Jul_26_21_21_34_2023_317)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v3-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v3 5/7] Row pattern recognition patch (docs).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++++++--
3 files changed, 133 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index dcc9d6f59d..b333d62410 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21714,6 +21714,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21753,6 +21754,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Jul_26_21_21_34_2023_317)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v3-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v3 5/7] Row pattern recognition patch (docs).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++++++--
3 files changed, 133 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index dcc9d6f59d..b333d62410 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21714,6 +21714,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21753,6 +21754,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Jul_26_21_21_34_2023_317)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v3-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v3 5/7] Row pattern recognition patch (docs).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 29 +++++++++++++++++--
3 files changed, 133 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index dcc9d6f59d..b333d62410 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21714,6 +21714,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21753,6 +21754,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..16478a3950 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,31 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Jul_26_21_21_34_2023_317)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v3-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v4 5/7] Row pattern recognition patch (docs).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index be2f54c914..6cda164522 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21715,6 +21715,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21754,6 +21755,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Aug__9_17_41_12_2023_134)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v4-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v4 5/7] Row pattern recognition patch (docs).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index be2f54c914..6cda164522 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21715,6 +21715,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21754,6 +21755,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Aug__9_17_41_12_2023_134)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v4-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v4 5/7] Row pattern recognition patch (docs).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index be2f54c914..6cda164522 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21715,6 +21715,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21754,6 +21755,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Aug__9_17_41_12_2023_134)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v4-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v4 5/7] Row pattern recognition patch (docs).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index be2f54c914..6cda164522 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21715,6 +21715,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21754,6 +21755,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Aug__9_17_41_12_2023_134)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v4-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v4 5/7] Row pattern recognition patch (docs).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index be2f54c914..6cda164522 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21715,6 +21715,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21754,6 +21755,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Aug__9_17_41_12_2023_134)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v4-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v4 5/7] Row pattern recognition patch (docs).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index be2f54c914..6cda164522 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21715,6 +21715,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21754,6 +21755,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Aug__9_17_41_12_2023_134)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v4-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v5 5/7] Row pattern recognition patch (docs).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7a0d4b9134..b7bfc9271e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21772,6 +21772,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21811,6 +21812,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Sep__2_15_52_35_2023_273)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v5-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v5 5/7] Row pattern recognition patch (docs).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7a0d4b9134..b7bfc9271e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21772,6 +21772,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21811,6 +21812,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Sep__2_15_52_35_2023_273)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v5-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v5 5/7] Row pattern recognition patch (docs).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7a0d4b9134..b7bfc9271e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21772,6 +21772,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21811,6 +21812,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Sep__2_15_52_35_2023_273)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v5-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v5 5/7] Row pattern recognition patch (docs).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7a0d4b9134..b7bfc9271e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21772,6 +21772,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21811,6 +21812,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Sep__2_15_52_35_2023_273)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v5-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v5 5/7] Row pattern recognition patch (docs).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7a0d4b9134..b7bfc9271e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21772,6 +21772,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21811,6 +21812,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Sep__2_15_52_35_2023_273)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v5-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v5 5/7] Row pattern recognition patch (docs).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7a0d4b9134..b7bfc9271e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21772,6 +21772,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21811,6 +21812,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Sep__2_15_52_35_2023_273)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v5-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v6 5/7] Row pattern recognition patch (docs).
@ 2023-09-12 05:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-12 05:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Sep_12_15_18_43_2023_359)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v6-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v6 5/7] Row pattern recognition patch (docs).
@ 2023-09-12 05:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-12 05:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Sep_12_15_18_43_2023_359)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v6-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v6 5/7] Row pattern recognition patch (docs).
@ 2023-09-12 05:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-12 05:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Sep_12_15_18_43_2023_359)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v6-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v6 5/7] Row pattern recognition patch (docs).
@ 2023-09-12 05:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-12 05:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Sep_12_15_18_43_2023_359)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v6-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v6 5/7] Row pattern recognition patch (docs).
@ 2023-09-12 05:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-12 05:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Sep_12_15_18_43_2023_359)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v6-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v6 5/7] Row pattern recognition patch (docs).
@ 2023-09-12 05:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-12 05:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..eda3612822 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..8d3becd57a 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Sep_12_15_18_43_2023_359)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v6-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v7 5/7] Row pattern recognition patch (docs).
@ 2023-09-22 04:53 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-22 04:53 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Sep_22_14_16_40_2023_530)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v7-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v7 5/7] Row pattern recognition patch (docs).
@ 2023-09-22 04:53 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-22 04:53 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Sep_22_14_16_40_2023_530)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v7-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v7 5/7] Row pattern recognition patch (docs).
@ 2023-09-22 04:53 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-22 04:53 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Sep_22_14_16_40_2023_530)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v7-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v7 5/7] Row pattern recognition patch (docs).
@ 2023-09-22 04:53 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-22 04:53 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Sep_22_14_16_40_2023_530)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v7-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v7 5/7] Row pattern recognition patch (docs).
@ 2023-09-22 04:53 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-22 04:53 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Sep_22_14_16_40_2023_530)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v7-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v8 5/7] Row pattern recognition patch (docs).
@ 2023-09-25 05:01 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-25 05:01 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Sep_25_14_26_30_2023_752)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v8-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v8 5/7] Row pattern recognition patch (docs).
@ 2023-09-25 05:01 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-25 05:01 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Sep_25_14_26_30_2023_752)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v8-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v8 5/7] Row pattern recognition patch (docs).
@ 2023-09-25 05:01 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-25 05:01 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Sep_25_14_26_30_2023_752)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v8-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v8 5/7] Row pattern recognition patch (docs).
@ 2023-09-25 05:01 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-25 05:01 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Sep_25_14_26_30_2023_752)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v8-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v8 5/7] Row pattern recognition patch (docs).
@ 2023-09-25 05:01 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-09-25 05:01 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 24ad87f910..9c99dda4ae 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 0ee0cc7e64..056768b330 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -966,8 +966,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1074,6 +1074,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Sep_25_14_26_30_2023_752)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v8-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v9 5/7] Row pattern recognition patch (docs).
@ 2023-10-04 05:51 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-10-04 05:51 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index f1ad64c3d6..6c46ea4355 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Oct__4_15_03_28_2023_821)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v9-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v9 5/7] Row pattern recognition patch (docs).
@ 2023-10-04 05:51 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-10-04 05:51 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index f1ad64c3d6..6c46ea4355 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Oct__4_15_03_28_2023_821)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v9-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v9 5/7] Row pattern recognition patch (docs).
@ 2023-10-04 05:51 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-10-04 05:51 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index f1ad64c3d6..6c46ea4355 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Oct__4_15_03_28_2023_821)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v9-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v9 5/7] Row pattern recognition patch (docs).
@ 2023-10-04 05:51 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-10-04 05:51 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 52 ++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 +++++++++++++++++++++++--
3 files changed, 142 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..f39ec8f2d5 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,58 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used with row pattern common syntax to
+ perform row pattern recognition in a query. Row pattern common syntax
+ includes two sub clauses. <literal>DEFINE</literal> defines definition
+ variables along with an expression. The expression must be a logical
+ expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. Moreover if the expression comprises a column
+ reference, it must be the argument of <function>rpr</function>. An example
+ of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns price column in the previous
+ row if it's called in a context of row pattern recognition. So in the
+ second line means the definition variable "UP" is <literal>TRUE</literal>
+ when price column in the current row is greater than the price column in
+ the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". If a
+ sequence of rows found, rpr returns the column at the starting row.
+ Example of a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price, max(price) OVER w FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index f1ad64c3d6..6c46ea4355 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21780,6 +21780,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21819,6 +21820,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Oct__4_15_03_28_2023_821)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v9-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v10 5/7] Row pattern recognition patch (docs).
@ 2023-10-22 02:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-10-22 02:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7c3e940afe..1d835af15a 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21878,6 +21878,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21917,6 +21918,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Oct_22_11_39_20_2023_140)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v10-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v10 5/7] Row pattern recognition patch (docs).
@ 2023-10-22 02:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-10-22 02:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7c3e940afe..1d835af15a 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21878,6 +21878,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21917,6 +21918,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Oct_22_11_39_20_2023_140)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v10-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v10 5/7] Row pattern recognition patch (docs).
@ 2023-10-22 02:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-10-22 02:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7c3e940afe..1d835af15a 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21878,6 +21878,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21917,6 +21918,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Oct_22_11_39_20_2023_140)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v10-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v10 5/7] Row pattern recognition patch (docs).
@ 2023-10-22 02:22 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-10-22 02:22 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 7c3e940afe..1d835af15a 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21878,6 +21878,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21917,6 +21918,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Oct_22_11_39_20_2023_140)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v10-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v11 5/7] Row pattern recognition patch (docs).
@ 2023-11-08 06:57 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-11-08 06:57 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index d963f0a0a0..c3a8167c8e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21933,6 +21933,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21972,6 +21973,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Nov__8_16_37_05_2023_872)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v11-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v11 5/7] Row pattern recognition patch (docs).
@ 2023-11-08 06:57 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-11-08 06:57 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index d963f0a0a0..c3a8167c8e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21933,6 +21933,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21972,6 +21973,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Nov__8_16_37_05_2023_872)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v11-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v11 5/7] Row pattern recognition patch (docs).
@ 2023-11-08 06:57 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-11-08 06:57 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index d963f0a0a0..c3a8167c8e 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21933,6 +21933,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21972,6 +21973,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 42d78913cf..522ad9dd70 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_Nov__8_16_37_05_2023_872)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v11-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v12 5/7] Row pattern recognition patch (docs).
@ 2023-12-04 11:23 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-12-04 11:23 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 20da3ed033..8a18e0ee23 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21935,6 +21935,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21974,6 +21975,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 227ba1993b..dabaca9127 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Dec__8_10_16_13_2023_489)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v12-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v12 5/7] Row pattern recognition patch (docs).
@ 2023-12-04 11:23 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-12-04 11:23 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 20da3ed033..8a18e0ee23 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21935,6 +21935,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21974,6 +21975,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 227ba1993b..dabaca9127 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Dec__8_10_16_13_2023_489)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v12-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v12 5/7] Row pattern recognition patch (docs).
@ 2023-12-04 11:23 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2023-12-04 11:23 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 20da3ed033..8a18e0ee23 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21935,6 +21935,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21974,6 +21975,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 227ba1993b..dabaca9127 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Dec__8_10_16_13_2023_489)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v12-0006-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v13 6/8] Row pattern recognition patch (docs).
@ 2024-01-22 09:45 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-01-22 09:45 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 210c7c0b02..8422aa6b93 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21941,6 +21941,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21980,6 +21981,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 9917df7839..1575fc2167 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Jan_22_19_26_18_2024_011)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v13-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v13 6/8] Row pattern recognition patch (docs).
@ 2024-01-22 09:45 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-01-22 09:45 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 210c7c0b02..8422aa6b93 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21941,6 +21941,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21980,6 +21981,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 9917df7839..1575fc2167 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Jan_22_19_26_18_2024_011)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v13-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v13 6/8] Row pattern recognition patch (docs).
@ 2024-01-22 09:45 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-01-22 09:45 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 210c7c0b02..8422aa6b93 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -21941,6 +21941,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -21980,6 +21981,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 9917df7839..1575fc2167 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Jan_22_19_26_18_2024_011)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v13-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v14 6/8] Row pattern recognition patch (docs).
@ 2024-02-28 13:59 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-02-28 13:59 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index e5fa82c161..8dd97501a7 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -22364,6 +22364,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -22403,6 +22404,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 9917df7839..1575fc2167 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Feb_29_09_19_54_2024_640)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v14-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v14 6/8] Row pattern recognition patch (docs).
@ 2024-02-28 13:59 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-02-28 13:59 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index e5fa82c161..8dd97501a7 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -22364,6 +22364,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -22403,6 +22404,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 9917df7839..1575fc2167 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Feb_29_09_19_54_2024_640)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v14-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v14 6/8] Row pattern recognition patch (docs).
@ 2024-02-28 13:59 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-02-28 13:59 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index e5fa82c161..8dd97501a7 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -22364,6 +22364,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -22403,6 +22404,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 9917df7839..1575fc2167 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Feb_29_09_19_54_2024_640)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v14-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v15 6/8] Row pattern recognition patch (docs).
@ 2024-03-28 10:30 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-03-28 10:30 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 93b0bc2bc6..d25eeb3327 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -22637,6 +22637,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -22676,6 +22677,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Mar_28_19_59_25_2024_076)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v15-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v15 6/8] Row pattern recognition patch (docs).
@ 2024-03-28 10:30 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-03-28 10:30 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 93b0bc2bc6..d25eeb3327 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -22637,6 +22637,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -22676,6 +22677,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Mar_28_19_59_25_2024_076)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v15-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v15 6/8] Row pattern recognition patch (docs).
@ 2024-03-28 10:30 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-03-28 10:30 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 93b0bc2bc6..d25eeb3327 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -22637,6 +22637,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -22676,6 +22677,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Mar_28_19_59_25_2024_076)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v15-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v15 6/8] Row pattern recognition patch (docs).
@ 2024-03-28 10:30 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-03-28 10:30 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 93b0bc2bc6..d25eeb3327 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -22637,6 +22637,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -22676,6 +22677,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Mar_28_19_59_25_2024_076)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v15-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v15 6/8] Row pattern recognition patch (docs).
@ 2024-03-28 10:30 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-03-28 10:30 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 93b0bc2bc6..d25eeb3327 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -22637,6 +22637,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -22676,6 +22677,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Mar_28_19_59_25_2024_076)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v15-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v16 6/8] Row pattern recognition patch (docs).
@ 2024-04-12 06:49 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-04-12 06:49 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 8dfb42ad4d..f6ee99fe19 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23123,6 +23123,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23162,6 +23163,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Apr_12_16_09_08_2024_262)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v16-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v16 6/8] Row pattern recognition patch (docs).
@ 2024-04-12 06:49 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-04-12 06:49 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 8dfb42ad4d..f6ee99fe19 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23123,6 +23123,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23162,6 +23163,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Apr_12_16_09_08_2024_262)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v16-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v16 6/8] Row pattern recognition patch (docs).
@ 2024-04-12 06:49 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-04-12 06:49 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 80 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 170 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..cf18dd887e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,86 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. Row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. So in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. In the second or
+ subsequent rows all window functions and aggregates are NULL. For rows
+ that do not match the PATTERN, all window functions and aggregates are
+ shown AS NULL too, except count which shows 0. This is because the
+ unmatched rows are in an empty frame. Example of
+ a <literal>SELECT</literal> using the <literal>DEFINE</literal>
+ and <literal>PATTERN</literal> clause is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock,
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | |
+ company1 | 2023-07-03 | 150 | | |
+ company1 | 2023-07-04 | 140 | | |
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | |
+ company1 | 2023-07-08 | 130 | | |
+ company1 | 2023-07-09 | 120 | | |
+ company1 | 2023-07-10 | 130 | | | 0
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 8dfb42ad4d..f6ee99fe19 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23123,6 +23123,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23162,6 +23163,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Apr_12_16_09_08_2024_262)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v16-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v17 6/8] Row pattern recognition patch (docs).
@ 2024-04-28 11:00 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-04-28 11:00 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 1928de5762..adbcb1f279 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Apr_28_20_28_26_2024_444)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v17-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v17 6/8] Row pattern recognition patch (docs).
@ 2024-04-28 11:00 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-04-28 11:00 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 1928de5762..adbcb1f279 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Apr_28_20_28_26_2024_444)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v17-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v17 6/8] Row pattern recognition patch (docs).
@ 2024-04-28 11:00 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-04-28 11:00 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 1928de5762..adbcb1f279 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sun_Apr_28_20_28_26_2024_444)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v17-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v18 6/8] Row pattern recognition patch (docs).
@ 2024-05-11 07:11 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-11 07:11 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_May_11_16_23_07_2024_789)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v18-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v18 6/8] Row pattern recognition patch (docs).
@ 2024-05-11 07:11 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-11 07:11 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_May_11_16_23_07_2024_789)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v18-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v18 6/8] Row pattern recognition patch (docs).
@ 2024-05-11 07:11 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-11 07:11 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_May_11_16_23_07_2024_789)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v18-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v19 6/8] Row pattern recognition patch (docs).
@ 2024-05-14 23:26 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-14 23:26 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_May_15_09_02_03_2024_008)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v19-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v19 6/8] Row pattern recognition patch (docs).
@ 2024-05-14 23:26 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-14 23:26 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_May_15_09_02_03_2024_008)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v19-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v19 6/8] Row pattern recognition patch (docs).
@ 2024-05-14 23:26 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-14 23:26 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Wed_May_15_09_02_03_2024_008)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v19-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v20 6/8] Row pattern recognition patch (docs).
@ 2024-05-24 02:26 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-24 02:26 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_May_24_11_39_19_2024_763)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v20-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v20 6/8] Row pattern recognition patch (docs).
@ 2024-05-24 02:26 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-24 02:26 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_May_24_11_39_19_2024_763)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v20-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v20 6/8] Row pattern recognition patch (docs).
@ 2024-05-24 02:26 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-05-24 02:26 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 17c44bc338..8dbab31300 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23124,6 +23124,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23163,6 +23164,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index 066aed44e6..8f18718d58 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_May_24_11_39_19_2024_763)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v20-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v21 6/8] Row pattern recognition patch (docs).
@ 2024-08-26 04:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-08-26 04:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 461fc3f437..02ad2b0195 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23258,6 +23258,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23297,6 +23298,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Aug_26_13_39_47_2024_878)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v21-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v21 6/8] Row pattern recognition patch (docs).
@ 2024-08-26 04:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-08-26 04:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 461fc3f437..02ad2b0195 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23258,6 +23258,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23297,6 +23298,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Aug_26_13_39_47_2024_878)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v21-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v21 6/8] Row pattern recognition patch (docs).
@ 2024-08-26 04:32 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-08-26 04:32 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 461fc3f437..02ad2b0195 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23258,6 +23258,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23297,6 +23298,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Aug_26_13_39_47_2024_878)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v21-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v22 6/8] Row pattern recognition patch (docs).
@ 2024-09-19 04:48 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-09-19 04:48 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 6f75bd0c7d..4482c80a70 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23261,6 +23261,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23300,6 +23301,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Sep_19_13_59_47_2024_608)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v22-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v22 6/8] Row pattern recognition patch (docs).
@ 2024-09-19 04:48 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-09-19 04:48 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 6f75bd0c7d..4482c80a70 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23261,6 +23261,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23300,6 +23301,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Sep_19_13_59_47_2024_608)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v22-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v22 6/8] Row pattern recognition patch (docs).
@ 2024-09-19 04:48 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-09-19 04:48 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 6f75bd0c7d..4482c80a70 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23261,6 +23261,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23300,6 +23301,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Sep_19_13_59_47_2024_608)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v22-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v23 6/8] Row pattern recognition patch (docs).
@ 2024-10-25 03:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-10-25 03:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 58dc06b68b..ef4fa7144f 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23263,6 +23263,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23302,6 +23303,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Oct_25_13_04_53_2024_648)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v23-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v23 6/8] Row pattern recognition patch (docs).
@ 2024-10-25 03:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-10-25 03:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 58dc06b68b..ef4fa7144f 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23263,6 +23263,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23302,6 +23303,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Oct_25_13_04_53_2024_648)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v23-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v23 6/8] Row pattern recognition patch (docs).
@ 2024-10-25 03:56 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-10-25 03:56 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 58dc06b68b..ef4fa7144f 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23263,6 +23263,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23302,6 +23303,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Fri_Oct_25_13_04_53_2024_648)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v23-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v24 6/8] Row pattern recognition patch (docs).
@ 2024-12-19 06:06 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-19 06:06 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Dec_19_15_19_50_2024_894)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v24-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v24 6/8] Row pattern recognition patch (docs).
@ 2024-12-19 06:06 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-19 06:06 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Dec_19_15_19_50_2024_894)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v24-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v24 6/8] Row pattern recognition patch (docs).
@ 2024-12-19 06:06 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-19 06:06 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Thu_Dec_19_15_19_50_2024_894)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v24-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v25 6/9] Row pattern recognition patch (docs).
@ 2024-12-21 06:19 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-21 06:19 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Dec_21_18_20_04_2024_526)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v25-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v25 6/9] Row pattern recognition patch (docs).
@ 2024-12-21 06:19 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-21 06:19 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Dec_21_18_20_04_2024_526)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v25-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v25 6/9] Row pattern recognition patch (docs).
@ 2024-12-21 06:19 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-21 06:19 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Sat_Dec_21_18_20_04_2024_526)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v25-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v26 6/9] Row pattern recognition patch (docs).
@ 2024-12-30 12:44 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-30 12:44 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Dec_30_22_37_18_2024_171)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v26-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v26 6/9] Row pattern recognition patch (docs).
@ 2024-12-30 12:44 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-30 12:44 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Dec_30_22_37_18_2024_171)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v26-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v26 6/9] Row pattern recognition patch (docs).
@ 2024-12-30 12:44 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-30 12:44 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Mon_Dec_30_22_37_18_2024_171)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v26-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v27 6/9] Row pattern recognition patch (docs).
@ 2024-12-30 23:53 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-30 23:53 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Dec_31_08_57_07_2024_963)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v27-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v27 6/9] Row pattern recognition patch (docs).
@ 2024-12-30 23:53 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-30 23:53 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Dec_31_08_57_07_2024_963)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v27-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v27 6/9] Row pattern recognition patch (docs).
@ 2024-12-30 23:53 Tatsuo Ishii <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Tatsuo Ishii @ 2024-12-30 23:53 UTC (permalink / raw)
---
doc/src/sgml/advanced.sgml | 82 ++++++++++++++++++++++++++++++++++++
doc/src/sgml/func.sgml | 54 ++++++++++++++++++++++++
doc/src/sgml/ref/select.sgml | 38 ++++++++++++++++-
3 files changed, 172 insertions(+), 2 deletions(-)
diff --git a/doc/src/sgml/advanced.sgml b/doc/src/sgml/advanced.sgml
index 755c9f1485..b0b1d1c51e 100644
--- a/doc/src/sgml/advanced.sgml
+++ b/doc/src/sgml/advanced.sgml
@@ -537,6 +537,88 @@ WHERE pos < 3;
<literal>rank</literal> less than 3.
</para>
+ <para>
+ Row pattern common syntax can be used to perform row pattern recognition
+ in a query. The row pattern common syntax includes two sub
+ clauses: <literal>DEFINE</literal>
+ and <literal>PATTERN</literal>. <literal>DEFINE</literal> defines
+ definition variables along with an expression. The expression must be a
+ logical expression, which means it must
+ return <literal>TRUE</literal>, <literal>FALSE</literal>
+ or <literal>NULL</literal>. The expression may comprise column references
+ and functions. Window functions, aggregate functions and subqueries are
+ not allowed. An example of <literal>DEFINE</literal> is as follows.
+
+<programlisting>
+DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+</programlisting>
+
+ Note that <function>PREV</function> returns the price column in the
+ previous row if it's called in a context of row pattern recognition. Thus in
+ the second line the definition variable "UP" is <literal>TRUE</literal>
+ when the price column in the current row is greater than the price column
+ in the previous row. Likewise, "DOWN" is <literal>TRUE</literal> when when
+ the price column in the current row is lower than the price column in the
+ previous row.
+ </para>
+ <para>
+ Once <literal>DEFINE</literal> exists, <literal>PATTERN</literal> can be
+ used. <literal>PATTERN</literal> defines a sequence of rows that satisfies
+ certain conditions. For example following <literal>PATTERN</literal>
+ defines that a row starts with the condition "LOWPRICE", then one or more
+ rows satisfy "UP" and finally one or more rows satisfy "DOWN". Note that
+ "+" means one or more matches. Also you can use "*", which means zero or
+ more matches. If a sequence of rows which satisfies the PATTERN is found,
+ in the starting row of the sequence of rows all window functions and
+ aggregates are shown in the target list. Note that aggregations only look
+ into the matched rows, rather than whole frame. On the second or
+ subsequent rows all window functions are NULL. Aggregates are NULL or 0
+ (count case) depending on its aggregation definition. For rows that do not
+ match on the PATTERN, all window functions and aggregates are shown AS
+ NULL too, except count showing 0. This is because the rows do not match,
+ thus they are in an empty frame. Example of a <literal>SELECT</literal>
+ using the <literal>DEFINE</literal> and <literal>PATTERN</literal> clause
+ is as follows.
+
+<programlisting>
+SELECT company, tdate, price,
+ first_value(price) OVER w,
+ max(price) OVER w,
+ count(price) OVER w
+FROM stock
+ WINDOW w AS (
+ PARTITION BY company
+ ORDER BY tdate
+ ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+ AFTER MATCH SKIP PAST LAST ROW
+ INITIAL
+ PATTERN (LOWPRICE UP+ DOWN+)
+ DEFINE
+ LOWPRICE AS price <= 100,
+ UP AS price > PREV(price),
+ DOWN AS price < PREV(price)
+);
+</programlisting>
+<screen>
+ company | tdate | price | first_value | max | count
+----------+------------+-------+-------------+-----+-------
+ company1 | 2023-07-01 | 100 | 100 | 200 | 4
+ company1 | 2023-07-02 | 200 | | | 0
+ company1 | 2023-07-03 | 150 | | | 0
+ company1 | 2023-07-04 | 140 | | | 0
+ company1 | 2023-07-05 | 150 | | | 0
+ company1 | 2023-07-06 | 90 | 90 | 130 | 4
+ company1 | 2023-07-07 | 110 | | | 0
+ company1 | 2023-07-08 | 130 | | | 0
+ company1 | 2023-07-09 | 120 | | | 0
+ company1 | 2023-07-10 | 130 | | | 0
+(10 rows)
+</screen>
+ </para>
+
<para>
When a query involves multiple window functions, it is possible to write
out each one with a separate <literal>OVER</literal> clause, but this is
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 47370e581a..17a38c4046 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -23338,6 +23338,7 @@ SELECT count(*) FROM sometable;
returns <literal>NULL</literal> if there is no such row.
</para></entry>
</row>
+
</tbody>
</tgroup>
</table>
@@ -23377,6 +23378,59 @@ SELECT count(*) FROM sometable;
Other frame specifications can be used to obtain other effects.
</para>
+ <para>
+ Row pattern recognition navigation functions are listed in
+ <xref linkend="functions-rpr-navigation-table"/>. These functions
+ can be used to describe DEFINE clause of Row pattern recognition.
+ </para>
+
+ <table id="functions-rpr-navigation-table">
+ <title>Row Pattern Navigation Functions</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ Function
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>prev</primary>
+ </indexterm>
+ <function>prev</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the previous row;
+ returns NULL if there is no previous row in the window frame.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>next</primary>
+ </indexterm>
+ <function>next</function> ( <parameter>value</parameter> <type>anyelement</type> )
+ <returnvalue>anyelement</returnvalue>
+ </para>
+ <para>
+ Returns the column value at the next row;
+ returns NULL if there is no next row in the window frame.
+ </para></entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
<note>
<para>
The SQL standard defines a <literal>RESPECT NULLS</literal> or
diff --git a/doc/src/sgml/ref/select.sgml b/doc/src/sgml/ref/select.sgml
index d7089eac0b..7e1c9989ba 100644
--- a/doc/src/sgml/ref/select.sgml
+++ b/doc/src/sgml/ref/select.sgml
@@ -969,8 +969,8 @@ WINDOW <replaceable class="parameter">window_name</replaceable> AS ( <replaceabl
The <replaceable class="parameter">frame_clause</replaceable> can be one of
<synopsis>
-{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
-{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ]
+{ RANGE | ROWS | GROUPS } <replaceable>frame_start</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
+{ RANGE | ROWS | GROUPS } BETWEEN <replaceable>frame_start</replaceable> AND <replaceable>frame_end</replaceable> [ <replaceable>frame_exclusion</replaceable> ] [row_pattern_common_syntax]
</synopsis>
where <replaceable>frame_start</replaceable>
@@ -1077,6 +1077,40 @@ EXCLUDE NO OTHERS
a given peer group will be in the frame or excluded from it.
</para>
+ <para>
+ The
+ optional <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ defines the <firstterm>row pattern recognition condition</firstterm> for
+ this
+ window. <replaceable class="parameter">row_pattern_common_syntax</replaceable>
+ includes following subclauses. <literal>AFTER MATCH SKIP PAST LAST
+ ROW</literal> or <literal>AFTER MATCH SKIP TO NEXT ROW</literal> controls
+ how to proceed to next row position after a match
+ found. With <literal>AFTER MATCH SKIP PAST LAST ROW</literal> (the
+ default) next row position is next to the last row of previous match. On
+ the other hand, with <literal>AFTER MATCH SKIP TO NEXT ROW</literal> next
+ row position is always next to the last row of previous
+ match. <literal>DEFINE</literal> defines definition variables along with a
+ boolean expression. <literal>PATTERN</literal> defines a sequence of rows
+ that satisfies certain conditions using variables defined
+ in <literal>DEFINE</literal> clause. If the variable is not defined in
+ the <literal>DEFINE</literal> clause, it is implicitly assumed
+ following is defined in the <literal>DEFINE</literal> clause.
+
+<synopsis>
+<literal>variable_name</literal> AS TRUE
+</synopsis>
+
+ Note that the maximu number of variables defined
+ in <literal>DEFINE</literal> clause is 26.
+
+<synopsis>
+[ AFTER MATCH SKIP PAST LAST ROW | AFTER MATCH SKIP TO NEXT ROW ]
+PATTERN <replaceable class="parameter">pattern_variable_name</replaceable>[+] [, ...]
+DEFINE <replaceable class="parameter">definition_varible_name</replaceable> AS <replaceable class="parameter">expression</replaceable> [, ...]
+</synopsis>
+ </para>
+
<para>
The purpose of a <literal>WINDOW</literal> clause is to specify the
behavior of <firstterm>window functions</firstterm> appearing in the query's
--
2.25.1
----Next_Part(Tue_Dec_31_08_57_07_2024_963)--
Content-Type: Text/X-Patch; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="v27-0007-Row-pattern-recognition-patch-tests.patch"
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v2] remove #include <math.h> where not needed
@ 2026-01-15 13:33 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Álvaro Herrera @ 2026-01-15 13:33 UTC (permalink / raw)
Discussion: https://postgr.es/m/[email protected]
---
contrib/btree_gist/btree_numeric.c | 1 -
contrib/btree_gist/btree_utils_var.c | 1 -
contrib/intarray/_intbig_gist.c | 2 --
contrib/ltree/_ltree_gist.c | 2 --
contrib/seg/segparse.y | 1 -
src/backend/access/heap/vacuumlazy.c | 2 --
src/backend/access/transam/xlogrecovery.c | 1 -
src/backend/access/transam/xlogwait.c | 1 -
src/backend/commands/define.c | 1 -
src/backend/executor/nodeBitmapHeapscan.c | 2 --
src/backend/executor/nodeSubplan.c | 2 --
src/backend/lib/knapsack.c | 1 -
src/backend/nodes/readfuncs.c | 2 --
src/backend/optimizer/geqo/geqo_eval.c | 1 -
src/backend/optimizer/geqo/geqo_pool.c | 1 -
src/backend/optimizer/path/indxpath.c | 2 --
src/backend/optimizer/path/joinpath.c | 2 --
src/backend/optimizer/plan/createplan.c | 2 --
src/backend/optimizer/util/pathnode.c | 2 --
src/backend/statistics/mcv.c | 2 --
src/backend/utils/adt/numutils.c | 1 -
src/backend/utils/adt/tid.c | 1 -
src/common/binaryheap.c | 2 --
src/include/utils/date.h | 2 --
24 files changed, 37 deletions(-)
diff --git a/contrib/btree_gist/btree_numeric.c b/contrib/btree_gist/btree_numeric.c
index 052f27b0794..7ca097abb74 100644
--- a/contrib/btree_gist/btree_numeric.c
+++ b/contrib/btree_gist/btree_numeric.c
@@ -3,7 +3,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <float.h>
#include "btree_gist.h"
diff --git a/contrib/btree_gist/btree_utils_var.c b/contrib/btree_gist/btree_utils_var.c
index 40e06ae4908..6847e4e54d5 100644
--- a/contrib/btree_gist/btree_utils_var.c
+++ b/contrib/btree_gist/btree_utils_var.c
@@ -3,7 +3,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <limits.h>
#include <float.h>
diff --git a/contrib/intarray/_intbig_gist.c b/contrib/intarray/_intbig_gist.c
index 0afa8a73b68..6d6ff7413f9 100644
--- a/contrib/intarray/_intbig_gist.c
+++ b/contrib/intarray/_intbig_gist.c
@@ -3,8 +3,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "_int.h"
#include "access/gist.h"
#include "access/reloptions.h"
diff --git a/contrib/ltree/_ltree_gist.c b/contrib/ltree/_ltree_gist.c
index ceb92a6304d..6383b37e53f 100644
--- a/contrib/ltree/_ltree_gist.c
+++ b/contrib/ltree/_ltree_gist.c
@@ -7,8 +7,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/gist.h"
#include "access/reloptions.h"
#include "access/stratnum.h"
diff --git a/contrib/seg/segparse.y b/contrib/seg/segparse.y
index 0358ddb182c..a3dbd0cdbd6 100644
--- a/contrib/seg/segparse.y
+++ b/contrib/seg/segparse.y
@@ -4,7 +4,6 @@
#include "postgres.h"
#include <float.h>
-#include <math.h>
#include "fmgr.h"
#include "nodes/miscnodes.h"
diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c
index 2086a577199..1fcb212ab3d 100644
--- a/src/backend/access/heap/vacuumlazy.c
+++ b/src/backend/access/heap/vacuumlazy.c
@@ -129,8 +129,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/genam.h"
#include "access/heapam.h"
#include "access/htup_details.h"
diff --git a/src/backend/access/transam/xlogrecovery.c b/src/backend/access/transam/xlogrecovery.c
index 0b5f871abe7..117d8d8bb6b 100644
--- a/src/backend/access/transam/xlogrecovery.c
+++ b/src/backend/access/transam/xlogrecovery.c
@@ -25,7 +25,6 @@
#include "postgres.h"
#include <ctype.h>
-#include <math.h>
#include <time.h>
#include <sys/stat.h>
#include <sys/time.h>
diff --git a/src/backend/access/transam/xlogwait.c b/src/backend/access/transam/xlogwait.c
index cab7e5082dc..d286ff63123 100644
--- a/src/backend/access/transam/xlogwait.c
+++ b/src/backend/access/transam/xlogwait.c
@@ -47,7 +47,6 @@
#include "postgres.h"
#include <float.h>
-#include <math.h>
#include "access/xlog.h"
#include "access/xlogrecovery.h"
diff --git a/src/backend/commands/define.c b/src/backend/commands/define.c
index 8313431397f..4172cc9bacb 100644
--- a/src/backend/commands/define.c
+++ b/src/backend/commands/define.c
@@ -20,7 +20,6 @@
#include "postgres.h"
#include <ctype.h>
-#include <math.h>
#include "catalog/namespace.h"
#include "commands/defrem.h"
diff --git a/src/backend/executor/nodeBitmapHeapscan.c b/src/backend/executor/nodeBitmapHeapscan.c
index 2c68327cb29..c68c26cbf38 100644
--- a/src/backend/executor/nodeBitmapHeapscan.c
+++ b/src/backend/executor/nodeBitmapHeapscan.c
@@ -35,8 +35,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/relscan.h"
#include "access/tableam.h"
#include "access/visibilitymap.h"
diff --git a/src/backend/executor/nodeSubplan.c b/src/backend/executor/nodeSubplan.c
index 5e510c73926..8285c7101c2 100644
--- a/src/backend/executor/nodeSubplan.c
+++ b/src/backend/executor/nodeSubplan.c
@@ -26,8 +26,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/htup_details.h"
#include "executor/executor.h"
#include "executor/nodeSubplan.h"
diff --git a/src/backend/lib/knapsack.c b/src/backend/lib/knapsack.c
index 1a770b907ec..586f1881fd5 100644
--- a/src/backend/lib/knapsack.c
+++ b/src/backend/lib/knapsack.c
@@ -24,7 +24,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <limits.h>
#include "lib/knapsack.h"
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index c11728c0f17..981ab9c34ef 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -26,8 +26,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "miscadmin.h"
#include "nodes/bitmapset.h"
#include "nodes/readfuncs.h"
diff --git a/src/backend/optimizer/geqo/geqo_eval.c b/src/backend/optimizer/geqo/geqo_eval.c
index db2332e2364..56ad3df98fa 100644
--- a/src/backend/optimizer/geqo/geqo_eval.c
+++ b/src/backend/optimizer/geqo/geqo_eval.c
@@ -23,7 +23,6 @@
#include <float.h>
#include <limits.h>
-#include <math.h>
#include "optimizer/geqo.h"
#include "optimizer/joininfo.h"
diff --git a/src/backend/optimizer/geqo/geqo_pool.c b/src/backend/optimizer/geqo/geqo_pool.c
index fc43406d0c1..f330c739d3d 100644
--- a/src/backend/optimizer/geqo/geqo_pool.c
+++ b/src/backend/optimizer/geqo/geqo_pool.c
@@ -25,7 +25,6 @@
#include <float.h>
#include <limits.h>
-#include <math.h>
#include "optimizer/geqo_copy.h"
#include "optimizer/geqo_pool.h"
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 51b9d6677d3..29cb60d6b35 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -15,8 +15,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/stratnum.h"
#include "access/sysattr.h"
#include "access/transam.h"
diff --git a/src/backend/optimizer/path/joinpath.c b/src/backend/optimizer/path/joinpath.c
index 7a247493a39..36973eab3be 100644
--- a/src/backend/optimizer/path/joinpath.c
+++ b/src/backend/optimizer/path/joinpath.c
@@ -14,8 +14,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "executor/executor.h"
#include "foreign/fdwapi.h"
#include "nodes/nodeFuncs.h"
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index af41ca69929..eb4806b084a 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -16,8 +16,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/sysattr.h"
#include "catalog/pg_class.h"
#include "foreign/fdwapi.h"
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index 2e9becf3116..0f12421bac0 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -14,8 +14,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/htup_details.h"
#include "executor/nodeSetOp.h"
#include "foreign/fdwapi.h"
diff --git a/src/backend/statistics/mcv.c b/src/backend/statistics/mcv.c
index 9749871b18e..e5ac422c1b4 100644
--- a/src/backend/statistics/mcv.c
+++ b/src/backend/statistics/mcv.c
@@ -14,8 +14,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/htup_details.h"
#include "catalog/pg_statistic_ext.h"
#include "catalog/pg_statistic_ext_data.h"
diff --git a/src/backend/utils/adt/numutils.c b/src/backend/utils/adt/numutils.c
index b35e381b0be..47c2e21e6b3 100644
--- a/src/backend/utils/adt/numutils.c
+++ b/src/backend/utils/adt/numutils.c
@@ -14,7 +14,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <limits.h>
#include <ctype.h>
diff --git a/src/backend/utils/adt/tid.c b/src/backend/utils/adt/tid.c
index c627b385eec..07248b69e57 100644
--- a/src/backend/utils/adt/tid.c
+++ b/src/backend/utils/adt/tid.c
@@ -17,7 +17,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <limits.h>
#include "access/sysattr.h"
diff --git a/src/common/binaryheap.c b/src/common/binaryheap.c
index d2be882e65c..78823ff00f8 100644
--- a/src/common/binaryheap.c
+++ b/src/common/binaryheap.c
@@ -17,8 +17,6 @@
#include "postgres.h"
#endif
-#include <math.h>
-
#ifdef FRONTEND
#include "common/logging.h"
#endif
diff --git a/src/include/utils/date.h b/src/include/utils/date.h
index 1c5dfc94ee6..6063810891e 100644
--- a/src/include/utils/date.h
+++ b/src/include/utils/date.h
@@ -14,8 +14,6 @@
#ifndef DATE_H
#define DATE_H
-#include <math.h>
-
#include "datatype/timestamp.h"
#include "fmgr.h"
#include "pgtime.h"
--
2.47.3
--uzfvjzgwtlw4oib4--
^ permalink raw reply [nested|flat] 111+ messages in thread
* [PATCH v2] remove #include <math.h> where not needed
@ 2026-01-15 13:33 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 111+ messages in thread
From: Álvaro Herrera @ 2026-01-15 13:33 UTC (permalink / raw)
Discussion: https://postgr.es/m/[email protected]
---
contrib/btree_gist/btree_numeric.c | 1 -
contrib/btree_gist/btree_utils_var.c | 1 -
contrib/intarray/_intbig_gist.c | 2 --
contrib/ltree/_ltree_gist.c | 2 --
contrib/seg/segparse.y | 1 -
src/backend/access/heap/vacuumlazy.c | 2 --
src/backend/access/transam/xlogrecovery.c | 1 -
src/backend/access/transam/xlogwait.c | 1 -
src/backend/commands/define.c | 1 -
src/backend/executor/nodeBitmapHeapscan.c | 2 --
src/backend/executor/nodeSubplan.c | 2 --
src/backend/lib/knapsack.c | 1 -
src/backend/nodes/readfuncs.c | 2 --
src/backend/optimizer/geqo/geqo_eval.c | 1 -
src/backend/optimizer/geqo/geqo_pool.c | 1 -
src/backend/optimizer/path/indxpath.c | 2 --
src/backend/optimizer/path/joinpath.c | 2 --
src/backend/optimizer/plan/createplan.c | 2 --
src/backend/optimizer/util/pathnode.c | 2 --
src/backend/statistics/mcv.c | 2 --
src/backend/utils/adt/numutils.c | 1 -
src/backend/utils/adt/tid.c | 1 -
src/common/binaryheap.c | 2 --
src/include/utils/date.h | 2 --
24 files changed, 37 deletions(-)
diff --git a/contrib/btree_gist/btree_numeric.c b/contrib/btree_gist/btree_numeric.c
index 052f27b0794..7ca097abb74 100644
--- a/contrib/btree_gist/btree_numeric.c
+++ b/contrib/btree_gist/btree_numeric.c
@@ -3,7 +3,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <float.h>
#include "btree_gist.h"
diff --git a/contrib/btree_gist/btree_utils_var.c b/contrib/btree_gist/btree_utils_var.c
index 40e06ae4908..6847e4e54d5 100644
--- a/contrib/btree_gist/btree_utils_var.c
+++ b/contrib/btree_gist/btree_utils_var.c
@@ -3,7 +3,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <limits.h>
#include <float.h>
diff --git a/contrib/intarray/_intbig_gist.c b/contrib/intarray/_intbig_gist.c
index 0afa8a73b68..6d6ff7413f9 100644
--- a/contrib/intarray/_intbig_gist.c
+++ b/contrib/intarray/_intbig_gist.c
@@ -3,8 +3,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "_int.h"
#include "access/gist.h"
#include "access/reloptions.h"
diff --git a/contrib/ltree/_ltree_gist.c b/contrib/ltree/_ltree_gist.c
index ceb92a6304d..6383b37e53f 100644
--- a/contrib/ltree/_ltree_gist.c
+++ b/contrib/ltree/_ltree_gist.c
@@ -7,8 +7,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/gist.h"
#include "access/reloptions.h"
#include "access/stratnum.h"
diff --git a/contrib/seg/segparse.y b/contrib/seg/segparse.y
index 0358ddb182c..a3dbd0cdbd6 100644
--- a/contrib/seg/segparse.y
+++ b/contrib/seg/segparse.y
@@ -4,7 +4,6 @@
#include "postgres.h"
#include <float.h>
-#include <math.h>
#include "fmgr.h"
#include "nodes/miscnodes.h"
diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c
index 2086a577199..1fcb212ab3d 100644
--- a/src/backend/access/heap/vacuumlazy.c
+++ b/src/backend/access/heap/vacuumlazy.c
@@ -129,8 +129,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/genam.h"
#include "access/heapam.h"
#include "access/htup_details.h"
diff --git a/src/backend/access/transam/xlogrecovery.c b/src/backend/access/transam/xlogrecovery.c
index 0b5f871abe7..117d8d8bb6b 100644
--- a/src/backend/access/transam/xlogrecovery.c
+++ b/src/backend/access/transam/xlogrecovery.c
@@ -25,7 +25,6 @@
#include "postgres.h"
#include <ctype.h>
-#include <math.h>
#include <time.h>
#include <sys/stat.h>
#include <sys/time.h>
diff --git a/src/backend/access/transam/xlogwait.c b/src/backend/access/transam/xlogwait.c
index cab7e5082dc..d286ff63123 100644
--- a/src/backend/access/transam/xlogwait.c
+++ b/src/backend/access/transam/xlogwait.c
@@ -47,7 +47,6 @@
#include "postgres.h"
#include <float.h>
-#include <math.h>
#include "access/xlog.h"
#include "access/xlogrecovery.h"
diff --git a/src/backend/commands/define.c b/src/backend/commands/define.c
index 8313431397f..4172cc9bacb 100644
--- a/src/backend/commands/define.c
+++ b/src/backend/commands/define.c
@@ -20,7 +20,6 @@
#include "postgres.h"
#include <ctype.h>
-#include <math.h>
#include "catalog/namespace.h"
#include "commands/defrem.h"
diff --git a/src/backend/executor/nodeBitmapHeapscan.c b/src/backend/executor/nodeBitmapHeapscan.c
index 2c68327cb29..c68c26cbf38 100644
--- a/src/backend/executor/nodeBitmapHeapscan.c
+++ b/src/backend/executor/nodeBitmapHeapscan.c
@@ -35,8 +35,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/relscan.h"
#include "access/tableam.h"
#include "access/visibilitymap.h"
diff --git a/src/backend/executor/nodeSubplan.c b/src/backend/executor/nodeSubplan.c
index 5e510c73926..8285c7101c2 100644
--- a/src/backend/executor/nodeSubplan.c
+++ b/src/backend/executor/nodeSubplan.c
@@ -26,8 +26,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/htup_details.h"
#include "executor/executor.h"
#include "executor/nodeSubplan.h"
diff --git a/src/backend/lib/knapsack.c b/src/backend/lib/knapsack.c
index 1a770b907ec..586f1881fd5 100644
--- a/src/backend/lib/knapsack.c
+++ b/src/backend/lib/knapsack.c
@@ -24,7 +24,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <limits.h>
#include "lib/knapsack.h"
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index c11728c0f17..981ab9c34ef 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -26,8 +26,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "miscadmin.h"
#include "nodes/bitmapset.h"
#include "nodes/readfuncs.h"
diff --git a/src/backend/optimizer/geqo/geqo_eval.c b/src/backend/optimizer/geqo/geqo_eval.c
index db2332e2364..56ad3df98fa 100644
--- a/src/backend/optimizer/geqo/geqo_eval.c
+++ b/src/backend/optimizer/geqo/geqo_eval.c
@@ -23,7 +23,6 @@
#include <float.h>
#include <limits.h>
-#include <math.h>
#include "optimizer/geqo.h"
#include "optimizer/joininfo.h"
diff --git a/src/backend/optimizer/geqo/geqo_pool.c b/src/backend/optimizer/geqo/geqo_pool.c
index fc43406d0c1..f330c739d3d 100644
--- a/src/backend/optimizer/geqo/geqo_pool.c
+++ b/src/backend/optimizer/geqo/geqo_pool.c
@@ -25,7 +25,6 @@
#include <float.h>
#include <limits.h>
-#include <math.h>
#include "optimizer/geqo_copy.h"
#include "optimizer/geqo_pool.h"
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 51b9d6677d3..29cb60d6b35 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -15,8 +15,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/stratnum.h"
#include "access/sysattr.h"
#include "access/transam.h"
diff --git a/src/backend/optimizer/path/joinpath.c b/src/backend/optimizer/path/joinpath.c
index 7a247493a39..36973eab3be 100644
--- a/src/backend/optimizer/path/joinpath.c
+++ b/src/backend/optimizer/path/joinpath.c
@@ -14,8 +14,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "executor/executor.h"
#include "foreign/fdwapi.h"
#include "nodes/nodeFuncs.h"
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index af41ca69929..eb4806b084a 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -16,8 +16,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/sysattr.h"
#include "catalog/pg_class.h"
#include "foreign/fdwapi.h"
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index 2e9becf3116..0f12421bac0 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -14,8 +14,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/htup_details.h"
#include "executor/nodeSetOp.h"
#include "foreign/fdwapi.h"
diff --git a/src/backend/statistics/mcv.c b/src/backend/statistics/mcv.c
index 9749871b18e..e5ac422c1b4 100644
--- a/src/backend/statistics/mcv.c
+++ b/src/backend/statistics/mcv.c
@@ -14,8 +14,6 @@
*/
#include "postgres.h"
-#include <math.h>
-
#include "access/htup_details.h"
#include "catalog/pg_statistic_ext.h"
#include "catalog/pg_statistic_ext_data.h"
diff --git a/src/backend/utils/adt/numutils.c b/src/backend/utils/adt/numutils.c
index b35e381b0be..47c2e21e6b3 100644
--- a/src/backend/utils/adt/numutils.c
+++ b/src/backend/utils/adt/numutils.c
@@ -14,7 +14,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <limits.h>
#include <ctype.h>
diff --git a/src/backend/utils/adt/tid.c b/src/backend/utils/adt/tid.c
index c627b385eec..07248b69e57 100644
--- a/src/backend/utils/adt/tid.c
+++ b/src/backend/utils/adt/tid.c
@@ -17,7 +17,6 @@
*/
#include "postgres.h"
-#include <math.h>
#include <limits.h>
#include "access/sysattr.h"
diff --git a/src/common/binaryheap.c b/src/common/binaryheap.c
index d2be882e65c..78823ff00f8 100644
--- a/src/common/binaryheap.c
+++ b/src/common/binaryheap.c
@@ -17,8 +17,6 @@
#include "postgres.h"
#endif
-#include <math.h>
-
#ifdef FRONTEND
#include "common/logging.h"
#endif
diff --git a/src/include/utils/date.h b/src/include/utils/date.h
index 1c5dfc94ee6..6063810891e 100644
--- a/src/include/utils/date.h
+++ b/src/include/utils/date.h
@@ -14,8 +14,6 @@
#ifndef DATE_H
#define DATE_H
-#include <math.h>
-
#include "datatype/timestamp.h"
#include "fmgr.h"
#include "pgtime.h"
--
2.47.3
--uzfvjzgwtlw4oib4--
^ permalink raw reply [nested|flat] 111+ messages in thread
end of thread, other threads:[~2026-01-15 13:33 UTC | newest]
Thread overview: 111+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2019-03-01 00:15 [PATCH 04/13] Jsonpath support for json Nikita Glukhov <[email protected]>
2019-03-07 20:49 [PATCH 3/3] Add operator <->(box, point) to SP-GiST box_ops Nikita Glukhov <[email protected]>
2019-03-07 20:49 [PATCH 3/3] Add operator <->(box, point) to SP-GiST box_ops Nikita Glukhov <[email protected]>
2019-03-07 20:49 [PATCH 3/3] Add operator <->(box, point) to SP-GiST box_ops Nikita Glukhov <[email protected]>
2022-03-19 11:28 Which hook to use when overriding utility commands (COPY ...) Hannu Krosing <[email protected]>
2022-03-19 12:44 ` Re: Which hook to use when overriding utility commands (COPY ...) Michael Paquier <[email protected]>
2023-06-25 11:48 [PATCH v1 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-06-25 11:48 [PATCH v1 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-06-25 11:48 [PATCH v1 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-06-25 11:48 [PATCH v1 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-06-26 08:05 [PATCH v2 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-06-26 08:05 [PATCH v2 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-06-26 08:05 [PATCH v2 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-06-26 08:05 [PATCH v2 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-06-26 08:05 [PATCH v2 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-07-26 10:49 [PATCH v3 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-07-26 10:49 [PATCH v3 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-07-26 10:49 [PATCH v3 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-07-26 10:49 [PATCH v3 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-07-26 10:49 [PATCH v3 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-08-09 07:56 [PATCH v4 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-08-09 07:56 [PATCH v4 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-08-09 07:56 [PATCH v4 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-08-09 07:56 [PATCH v4 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-08-09 07:56 [PATCH v4 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-08-09 07:56 [PATCH v4 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-02 06:32 [PATCH v5 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-02 06:32 [PATCH v5 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-02 06:32 [PATCH v5 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-02 06:32 [PATCH v5 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-02 06:32 [PATCH v5 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-02 06:32 [PATCH v5 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-12 05:22 [PATCH v6 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-12 05:22 [PATCH v6 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-12 05:22 [PATCH v6 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-12 05:22 [PATCH v6 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-12 05:22 [PATCH v6 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-12 05:22 [PATCH v6 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-22 04:53 [PATCH v7 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-22 04:53 [PATCH v7 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-22 04:53 [PATCH v7 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-22 04:53 [PATCH v7 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-22 04:53 [PATCH v7 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-25 05:01 [PATCH v8 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-25 05:01 [PATCH v8 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-25 05:01 [PATCH v8 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-25 05:01 [PATCH v8 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-09-25 05:01 [PATCH v8 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-10-04 05:51 [PATCH v9 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-10-04 05:51 [PATCH v9 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-10-04 05:51 [PATCH v9 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-10-04 05:51 [PATCH v9 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-10-22 02:22 [PATCH v10 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-10-22 02:22 [PATCH v10 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-10-22 02:22 [PATCH v10 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-10-22 02:22 [PATCH v10 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-11-08 06:57 [PATCH v11 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-11-08 06:57 [PATCH v11 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-11-08 06:57 [PATCH v11 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-12-04 11:23 [PATCH v12 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-12-04 11:23 [PATCH v12 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2023-12-04 11:23 [PATCH v12 5/7] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-01-22 09:45 [PATCH v13 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-01-22 09:45 [PATCH v13 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-01-22 09:45 [PATCH v13 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-02-28 13:59 [PATCH v14 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-02-28 13:59 [PATCH v14 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-02-28 13:59 [PATCH v14 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-03-28 10:30 [PATCH v15 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-03-28 10:30 [PATCH v15 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-03-28 10:30 [PATCH v15 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-03-28 10:30 [PATCH v15 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-03-28 10:30 [PATCH v15 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-04-12 06:49 [PATCH v16 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-04-12 06:49 [PATCH v16 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-04-12 06:49 [PATCH v16 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-04-28 11:00 [PATCH v17 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-04-28 11:00 [PATCH v17 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-04-28 11:00 [PATCH v17 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-11 07:11 [PATCH v18 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-11 07:11 [PATCH v18 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-11 07:11 [PATCH v18 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-14 23:26 [PATCH v19 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-14 23:26 [PATCH v19 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-14 23:26 [PATCH v19 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-24 02:26 [PATCH v20 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-24 02:26 [PATCH v20 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-05-24 02:26 [PATCH v20 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-08-26 04:32 [PATCH v21 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-08-26 04:32 [PATCH v21 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-08-26 04:32 [PATCH v21 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-09-19 04:48 [PATCH v22 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-09-19 04:48 [PATCH v22 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-09-19 04:48 [PATCH v22 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-10-25 03:56 [PATCH v23 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-10-25 03:56 [PATCH v23 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-10-25 03:56 [PATCH v23 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-19 06:06 [PATCH v24 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-19 06:06 [PATCH v24 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-19 06:06 [PATCH v24 6/8] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-21 06:19 [PATCH v25 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-21 06:19 [PATCH v25 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-21 06:19 [PATCH v25 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-30 12:44 [PATCH v26 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-30 12:44 [PATCH v26 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-30 12:44 [PATCH v26 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-30 23:53 [PATCH v27 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-30 23:53 [PATCH v27 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2024-12-30 23:53 [PATCH v27 6/9] Row pattern recognition patch (docs). Tatsuo Ishii <[email protected]>
2026-01-15 13:33 [PATCH v2] remove #include <math.h> where not needed Álvaro Herrera <[email protected]>
2026-01-15 13:33 [PATCH v2] remove #include <math.h> where not needed Álvaro Herrera <[email protected]>
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox