public inbox for [email protected]  

help / color / mirror / Atom feed

[PATCH v1] Fix usage of unified logging pg_log_* in pg_rewind and initdb
124+ messages / 5 participants
[nested] [flat]

* [PATCH v1] Fix usage of unified logging pg_log_* in pg_rewind and initdb
@ 2019-07-01 15:11 Alexey Kondratov <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Alexey Kondratov @ 2019-07-01 15:11 UTC (permalink / raw)

---
 src/bin/initdb/initdb.c       | 2 +-
 src/bin/pg_rewind/pg_rewind.c | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/bin/initdb/initdb.c b/src/bin/initdb/initdb.c
index 2ef179165b..70273be783 100644
--- a/src/bin/initdb/initdb.c
+++ b/src/bin/initdb/initdb.c
@@ -2497,7 +2497,7 @@ setup_bin_paths(const char *argv0)
 			pg_log_error("The program \"postgres\" is needed by %s but was not found in the\n"
 						 "same directory as \"%s\".\n"
 						 "Check your installation.",
-						 full_path, progname);
+						 progname, full_path);
 		else
 			pg_log_error("The program \"postgres\" was found by \"%s\"\n"
 						 "but was not the same version as %s.\n"
diff --git a/src/bin/pg_rewind/pg_rewind.c b/src/bin/pg_rewind/pg_rewind.c
index 6e77201be6..d378053de4 100644
--- a/src/bin/pg_rewind/pg_rewind.c
+++ b/src/bin/pg_rewind/pg_rewind.c
@@ -555,7 +555,7 @@ getTimelineHistory(ControlFileData *controlFile, int *nentries)
 		else if (controlFile == &ControlFile_target)
 			histfile = slurpFile(datadir_target, path, NULL);
 		else
-			pg_fatal("invalid control file\n");
+			pg_fatal("invalid control file");
 
 		history = rewind_parseTimeLineHistory(histfile, tli, nentries);
 		pg_free(histfile);

base-commit: 95bbe5d82e428db342fa3ec60b95f1b9873741e5
-- 
2.17.1


--------------0DB63CE8F653DAD250E3D5A4--





^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH 12/17] Add tests for buffile.c.
@ 2019-07-05 14:24 Antonin Houska <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Antonin Houska @ 2019-07-05 14:24 UTC (permalink / raw)

Since the encryption adds complexity to the buffile.c, regression tests seem
like a good idea. In order to be able to test corner cases that deal with
segment boudaries, this patch makes the segment size configurable - the tests
would be too expensive with 1GB segments.

The new GUC is marked with DEVELOPER_OPTIONS, DBA is not supposed to use it
during normal operation.
---
 src/backend/storage/file/buffile.c            |  73 ++---
 src/backend/utils/misc/guc.c                  |  34 +-
 src/include/storage/buffile.h                 |  19 ++
 src/test/modules/buffile/Makefile             |  21 ++
 src/test/modules/buffile/README               |  11 +
 src/test/modules/buffile/buffile--1.0.sql     |  54 ++++
 src/test/modules/buffile/buffile.c            | 429 ++++++++++++++++++++++++++
 src/test/modules/buffile/buffile.control      |   5 +
 src/test/modules/buffile/expected/test_01.out |  61 ++++
 src/test/modules/buffile/expected/test_02.out |  48 +++
 src/test/modules/buffile/expected/test_03.out |  27 ++
 src/test/modules/buffile/expected/test_04.out |  84 +++++
 src/test/modules/buffile/expected/test_05.out |  33 ++
 src/test/modules/buffile/expected/test_06.out |  43 +++
 src/test/modules/buffile/expected/test_07.out |  41 +++
 src/test/modules/buffile/expected/test_08.out |  41 +++
 src/test/modules/buffile/expected/test_09.out |  39 +++
 src/test/modules/buffile/expected/test_10.out |  76 +++++
 src/test/modules/buffile/expected/test_11.out |  34 ++
 src/test/modules/buffile/expected/test_12.out |  16 +
 src/test/modules/buffile/expected/test_13.out |  86 ++++++
 src/test/modules/buffile/sql/test_01.sql      |  17 +
 src/test/modules/buffile/sql/test_02.sql      |  13 +
 src/test/modules/buffile/sql/test_03.sql      |   7 +
 src/test/modules/buffile/sql/test_04.sql      |  25 ++
 src/test/modules/buffile/sql/test_05.sql      |   8 +
 src/test/modules/buffile/sql/test_06.sql      |  15 +
 src/test/modules/buffile/sql/test_07.sql      |  13 +
 src/test/modules/buffile/sql/test_08.sql      |  13 +
 src/test/modules/buffile/sql/test_09.sql      |   9 +
 src/test/modules/buffile/sql/test_10.sql      |  25 ++
 src/test/modules/buffile/sql/test_11.sql      |   9 +
 src/test/modules/buffile/sql/test_12.sql      |   7 +
 src/test/modules/buffile/sql/test_13.sql      |  25 ++
 34 files changed, 1420 insertions(+), 41 deletions(-)
 create mode 100644 src/test/modules/buffile/Makefile
 create mode 100644 src/test/modules/buffile/README
 create mode 100644 src/test/modules/buffile/buffile--1.0.sql
 create mode 100644 src/test/modules/buffile/buffile.c
 create mode 100644 src/test/modules/buffile/buffile.control
 create mode 100644 src/test/modules/buffile/expected/test_01.out
 create mode 100644 src/test/modules/buffile/expected/test_02.out
 create mode 100644 src/test/modules/buffile/expected/test_03.out
 create mode 100644 src/test/modules/buffile/expected/test_04.out
 create mode 100644 src/test/modules/buffile/expected/test_05.out
 create mode 100644 src/test/modules/buffile/expected/test_06.out
 create mode 100644 src/test/modules/buffile/expected/test_07.out
 create mode 100644 src/test/modules/buffile/expected/test_08.out
 create mode 100644 src/test/modules/buffile/expected/test_09.out
 create mode 100644 src/test/modules/buffile/expected/test_10.out
 create mode 100644 src/test/modules/buffile/expected/test_11.out
 create mode 100644 src/test/modules/buffile/expected/test_12.out
 create mode 100644 src/test/modules/buffile/expected/test_13.out
 create mode 100644 src/test/modules/buffile/sql/test_01.sql
 create mode 100644 src/test/modules/buffile/sql/test_02.sql
 create mode 100644 src/test/modules/buffile/sql/test_03.sql
 create mode 100644 src/test/modules/buffile/sql/test_04.sql
 create mode 100644 src/test/modules/buffile/sql/test_05.sql
 create mode 100644 src/test/modules/buffile/sql/test_06.sql
 create mode 100644 src/test/modules/buffile/sql/test_07.sql
 create mode 100644 src/test/modules/buffile/sql/test_08.sql
 create mode 100644 src/test/modules/buffile/sql/test_09.sql
 create mode 100644 src/test/modules/buffile/sql/test_10.sql
 create mode 100644 src/test/modules/buffile/sql/test_11.sql
 create mode 100644 src/test/modules/buffile/sql/test_12.sql
 create mode 100644 src/test/modules/buffile/sql/test_13.sql

diff --git a/src/backend/storage/file/buffile.c b/src/backend/storage/file/buffile.c
index 308ceb01b7..998dabd134 100644
--- a/src/backend/storage/file/buffile.c
+++ b/src/backend/storage/file/buffile.c
@@ -56,16 +56,11 @@
 #include "utils/resowner.h"
 
 /*
- * We break BufFiles into gigabyte-sized segments, regardless of RELSEG_SIZE.
- * The reason is that we'd like large BufFiles to be spread across multiple
- * tablespaces when available.
- *
- * An integer value indicating the number of useful bytes in the segment is
- * appended to each segment of if the file is both shared and encrypted, see
- * BufFile.useful.
+ * The functions bellow actually use integer constants so that the size can be
+ * controlled by GUC. This is useful for development and regression tests.
  */
-#define MAX_PHYSICAL_FILESIZE	0x40000000
-#define BUFFILE_SEG_SIZE		(MAX_PHYSICAL_FILESIZE / BLCKSZ)
+int buffile_max_filesize	=  MAX_PHYSICAL_FILESIZE;
+int buffile_seg_blocks	=	BUFFILE_SEG_BLOCKS(MAX_PHYSICAL_FILESIZE);
 
 /*
  * Fields that both BufFile and TransientBufFile structures need. It must be
@@ -123,7 +118,7 @@ struct BufFile
 	BufFileCommon common;		/* Common fields, see above. */
 
 	int			numFiles;		/* number of physical files in set */
-	/* all files except the last have length exactly MAX_PHYSICAL_FILESIZE */
+	/* all files except the last have length exactly buffile_max_filesize */
 	File	   *files;			/* palloc'd array with numFiles entries */
 
 	/*
@@ -291,7 +286,7 @@ extendBufFile(BufFile *file)
 
 /*
  * Create a BufFile for a new temporary file (which will expand to become
- * multiple temporary files if more than MAX_PHYSICAL_FILESIZE bytes are
+ * multiple temporary files if more than buffile_max_filesize bytes are
  * written to it).
  *
  * If interXact is true, the temp file will not be automatically deleted
@@ -595,7 +590,7 @@ BufFileLoadBuffer(BufFile *file)
 	File		thisfile;
 
 	/*
-	 * Only whole multiple of ENCRYPTION_BLOCK can be encrypted / decrypted,
+	 * Only whole multiple of ENCRYPTION_BLOCK can be encrypted / decrypted.
 	 */
 	Assert((file->common.curOffset % BLCKSZ == 0 &&
 			file->common.curOffset % ENCRYPTION_BLOCK == 0) ||
@@ -604,7 +599,7 @@ BufFileLoadBuffer(BufFile *file)
 	/*
 	 * Advance to next component file if necessary and possible.
 	 */
-	if (file->common.curOffset >= MAX_PHYSICAL_FILESIZE &&
+	if (file->common.curOffset >= buffile_max_filesize &&
 		file->common.curFile + 1 < file->numFiles)
 	{
 		file->common.curFile++;
@@ -715,7 +710,7 @@ BufFileDumpBuffer(BufFile *file)
 		/*
 		 * Advance to next component file if necessary and possible.
 		 */
-		if (file->common.curOffset >= MAX_PHYSICAL_FILESIZE)
+		if (file->common.curOffset >= buffile_max_filesize)
 		{
 			while (file->common.curFile + 1 >= file->numFiles)
 				extendBufFile(file);
@@ -727,7 +722,7 @@ BufFileDumpBuffer(BufFile *file)
 		 * Determine how much we need to write into this file.
 		 */
 		bytestowrite = file->common.nbytes - wpos;
-		availbytes = MAX_PHYSICAL_FILESIZE - file->common.curOffset;
+		availbytes = buffile_max_filesize - file->common.curOffset;
 
 		if ((off_t) bytestowrite > availbytes)
 			bytestowrite = (int) availbytes;
@@ -758,7 +753,7 @@ BufFileDumpBuffer(BufFile *file)
 	{
 		file->common.curFile--;
 		Assert(file->common.curFile >= 0);
-		file->common.curOffset += MAX_PHYSICAL_FILESIZE;
+		file->common.curOffset += buffile_max_filesize;
 	}
 
 	/*
@@ -796,7 +791,7 @@ BufFileDumpBufferEncrypted(BufFile *file)
 	/*
 	 * Advance to next component file if necessary and possible.
 	 */
-	if (file->common.curOffset >= MAX_PHYSICAL_FILESIZE)
+	if (file->common.curOffset >= buffile_max_filesize)
 	{
 		while (file->common.curFile + 1 >= file->numFiles)
 			extendBufFile(file);
@@ -809,15 +804,14 @@ BufFileDumpBufferEncrypted(BufFile *file)
 	 * above) ensure that the encrypted buffer never crosses component file
 	 * boundary.
 	 */
-	StaticAssertStmt((MAX_PHYSICAL_FILESIZE % BLCKSZ) == 0,
-					 "BLCKSZ is not whole multiple of MAX_PHYSICAL_FILESIZE");
+	Assert((buffile_max_filesize % BLCKSZ) == 0);
 
 	/*
 	 * Encrypted data is dumped all at once.
 	 *
 	 * Unlike BufFileDumpBuffer(), we don't have to check here how much bytes
 	 * is available in the segment. According to the assertions above,
-	 * currOffset should be lower than MAX_PHYSICAL_FILESIZE by non-zero
+	 * currOffset should be lower than buffile_max_filesize by non-zero
 	 * multiple of BLCKSZ.
 	 */
 	bytestowrite = BLCKSZ;
@@ -1021,7 +1015,7 @@ BufFileSeek(BufFile *file, int fileno, off_t offset, int whence)
 	{
 		if (--newFile < 0)
 			return EOF;
-		newOffset += MAX_PHYSICAL_FILESIZE;
+		newOffset += buffile_max_filesize;
 	}
 	if (newFile == file->common.curFile &&
 		newOffset >= file->common.curOffset &&
@@ -1050,13 +1044,13 @@ BufFileSeek(BufFile *file, int fileno, off_t offset, int whence)
 	if (newFile == file->numFiles && newOffset == 0)
 	{
 		newFile--;
-		newOffset = MAX_PHYSICAL_FILESIZE;
+		newOffset = buffile_max_filesize;
 	}
-	while (newOffset > MAX_PHYSICAL_FILESIZE)
+	while (newOffset > buffile_max_filesize)
 	{
 		if (++newFile >= file->numFiles)
 			return EOF;
-		newOffset -= MAX_PHYSICAL_FILESIZE;
+		newOffset -= buffile_max_filesize;
 	}
 	if (newFile >= file->numFiles)
 		return EOF;
@@ -1124,8 +1118,8 @@ int
 BufFileSeekBlock(BufFile *file, long blknum)
 {
 	return BufFileSeek(file,
-					   (int) (blknum / BUFFILE_SEG_SIZE),
-					   (off_t) (blknum % BUFFILE_SEG_SIZE) * BLCKSZ,
+					   (int) (blknum / buffile_seg_blocks),
+					   (off_t) (blknum % buffile_seg_blocks) * BLCKSZ,
 					   SEEK_SET);
 }
 
@@ -1166,7 +1160,7 @@ BufFileTweak(char *tweak, BufFileCommon *file, bool is_transient)
 		if (tmpfile->segnos)
 			curFile = tmpfile->segnos[curFile];
 
-		block = curFile * BUFFILE_SEG_SIZE + file->curOffset / BLCKSZ;
+		block = curFile * buffile_seg_blocks + file->curOffset / BLCKSZ;
 
 		StaticAssertStmt(sizeof(pid) + sizeof(number) + sizeof(block) <=
 						 TWEAK_SIZE,
@@ -1258,7 +1252,7 @@ BufFileTellBlock(BufFile *file)
 	long		blknum;
 
 	blknum = (file->common.curOffset + file->common.pos) / BLCKSZ;
-	blknum += file->common.curFile * BUFFILE_SEG_SIZE;
+	blknum += file->common.curFile * buffile_seg_blocks;
 	return blknum;
 }
 
@@ -1304,7 +1298,7 @@ BufFileSize(BufFile *file)
 		lastFileSize = file->common.useful[file->common.nuseful - 1];
 	}
 
-	return ((file->numFiles - 1) * (int64) MAX_PHYSICAL_FILESIZE) +
+	return ((file->numFiles - 1) * (int64) buffile_max_filesize) +
 		lastFileSize;
 }
 
@@ -1317,11 +1311,10 @@ BufFileSize(BufFile *file)
  * called here first.  Resource owners for source and target must match,
  * too.
  *
- * This operation works by manipulating lists of segment files, so the
- * file content is always appended at a MAX_PHYSICAL_FILESIZE-aligned
- * boundary, typically creating empty holes before the boundary.  These
- * areas do not contain any interesting data, and cannot be read from by
- * caller.
+ * This operation works by manipulating lists of segment files, so the file
+ * content is always appended at a buffile_max_filesize-aligned boundary,
+ * typically creating empty holes before the boundary.  These areas do not
+ * contain any interesting data, and cannot be read from by caller.
  *
  * Returns the block number within target where the contents of source
  * begins.  Caller should apply this as an offset when working off block
@@ -1330,7 +1323,7 @@ BufFileSize(BufFile *file)
 long
 BufFileAppend(BufFile *target, BufFile *source)
 {
-	long		startBlock = target->numFiles * BUFFILE_SEG_SIZE;
+	long		startBlock = target->numFiles * buffile_seg_blocks;
 	int			newNumFiles = target->numFiles + source->numFiles;
 	int			i;
 
@@ -2084,14 +2077,14 @@ BufFileUpdateUsefulLength(BufFileCommon *file, bool is_transient)
 		 * w/o dumping anything of it. While curFile will be fixed during the
 		 * next dump, we need valid fileno now.
 		 */
-		if (file->curOffset >= MAX_PHYSICAL_FILESIZE)
+		if (file->curOffset >= buffile_max_filesize)
 		{
 			/*
 			 * Even BufFileSeek() should not allow curOffset to become more
-			 * than MAX_PHYSICAL_FILESIZE (if caller passes higher offset,
+			 * than buffile_max_filesize (if caller passes higher offset,
 			 * curFile gets increased instead).
 			 */
-			Assert(file->curOffset == MAX_PHYSICAL_FILESIZE);
+			Assert(file->curOffset == buffile_max_filesize);
 
 			fileno++;
 		}
@@ -2114,8 +2107,8 @@ BufFileUpdateUsefulLength(BufFileCommon *file, bool is_transient)
 		 * new_useful is also relative to the start of that previous
 		 * segment. Make sure it's relative to the current (fileno) segment.
 		 */
-		if (file->curOffset % MAX_PHYSICAL_FILESIZE == 0)
-			new_useful %= MAX_PHYSICAL_FILESIZE;
+		if (file->curOffset % buffile_max_filesize == 0)
+			new_useful %= buffile_max_filesize;
 
 		/* Finalize the offset. */
 		new_useful += file->pos;
diff --git a/src/backend/utils/misc/guc.c b/src/backend/utils/misc/guc.c
index 440120635e..879c8fd816 100644
--- a/src/backend/utils/misc/guc.c
+++ b/src/backend/utils/misc/guc.c
@@ -70,9 +70,10 @@
 #include "replication/syncrep.h"
 #include "replication/walreceiver.h"
 #include "replication/walsender.h"
+#include "storage/buffile.h"
 #include "storage/bufmgr.h"
-#include "storage/encryption.h"
 #include "storage/dsm_impl.h"
+#include "storage/encryption.h"
 #include "storage/standby.h"
 #include "storage/fd.h"
 #include "storage/large_object.h"
@@ -189,6 +190,8 @@ static const char *show_tcp_keepalives_idle(void);
 static const char *show_tcp_keepalives_interval(void);
 static const char *show_tcp_keepalives_count(void);
 static const char *show_tcp_user_timeout(void);
+static bool check_buffile_max_filesize(int *newval, void **extra, GucSource source);
+static void assign_buffile_max_filesize(int newval, void *extra);
 static bool check_maxconnections(int *newval, void **extra, GucSource source);
 static bool check_max_worker_processes(int *newval, void **extra, GucSource source);
 static bool check_autovacuum_max_workers(int *newval, void **extra, GucSource source);
@@ -3209,6 +3212,18 @@ static struct config_int ConfigureNamesInt[] =
 		NULL, assign_tcp_user_timeout, show_tcp_user_timeout
 	},
 
+	{
+		/* Not for general use */
+		{"buffile_max_filesize", PGC_SUSET, DEVELOPER_OPTIONS,
+			gettext_noop("Maximum size of BufFile segment."),
+			gettext_noop("This makes testing of some corner cases easier, especially when read or write crosses segment boundary."),
+			GUC_NOT_IN_SAMPLE | GUC_UNIT_BYTE
+		},
+		&buffile_max_filesize,
+		MAX_PHYSICAL_FILESIZE, BLCKSZ, MAX_PHYSICAL_FILESIZE,
+		check_buffile_max_filesize, assign_buffile_max_filesize, NULL
+	},
+
 	/* End-of-list marker */
 	{
 		{NULL, 0, 0, NULL, NULL}, NULL, 0, 0, 0, NULL, NULL, NULL
@@ -11301,6 +11316,23 @@ show_tcp_user_timeout(void)
 }
 
 static bool
+check_buffile_max_filesize(int *newval, void **extra, GucSource source)
+{
+	if (*newval % BLCKSZ != 0)
+	{
+		GUC_check_errdetail("The value must be whole multiple of BLCKSZ.");
+		return false;
+	}
+	return true;
+}
+
+static void
+assign_buffile_max_filesize(int newval, void *extra)
+{
+	buffile_seg_blocks = BUFFILE_SEG_BLOCKS(newval);
+}
+
+static bool
 check_maxconnections(int *newval, void **extra, GucSource source)
 {
 	if (*newval + autovacuum_max_workers + 1 +
diff --git a/src/include/storage/buffile.h b/src/include/storage/buffile.h
index 625fa0014c..977942f4b3 100644
--- a/src/include/storage/buffile.h
+++ b/src/include/storage/buffile.h
@@ -37,6 +37,25 @@ typedef struct BufFile BufFile;
 typedef struct TransientBufFile TransientBufFile;
 
 /*
+ * We break BufFiles into gigabyte-sized segments, regardless of RELSEG_SIZE.
+ * The reason is that we'd like large BufFiles to be spread across multiple
+ * tablespaces when available.
+ *
+ * An integer value indicating the number of useful bytes in the segment is
+ * appended to each segment of if the file is both shared and encrypted, see
+ * BufFile.useful.
+ */
+#define MAX_PHYSICAL_FILESIZE		0x40000000
+
+/* Express segment size in the number of blocks. */
+#define BUFFILE_SEG_BLOCKS(phys)	((phys) / BLCKSZ)
+
+/* GUC to control size of the file segment. */
+extern int buffile_max_filesize;
+/* Segment size in blocks, derived from the above. */
+extern int buffile_seg_blocks;
+
+/*
  * prototypes for functions in buffile.c
  */
 
diff --git a/src/test/modules/buffile/Makefile b/src/test/modules/buffile/Makefile
new file mode 100644
index 0000000000..f0f163c45e
--- /dev/null
+++ b/src/test/modules/buffile/Makefile
@@ -0,0 +1,21 @@
+PG_CONFIG ?= pg_config
+MODULE_big = buffile_test
+OBJS = buffile.o $(WIN32RES)
+PGFILEDESC = "buffile_test"
+
+EXTENSION = buffile
+DATA = buffile--1.0.sql
+
+REGRESS = test_01 test_02 test_03 test_04 test_05 test_06 test_07 test_08 \
+test_09 test_10 test_11 test_12 test_13
+
+ifdef USE_PGXS
+PG_CONFIG = pg_config
+PGXS := $(shell $(PG_CONFIG) --pgxs)
+include $(PGXS)
+else
+subdir = src/test/modules/buffile
+top_builddir = ../../../..
+include $(top_builddir)/src/Makefile.global
+include $(top_srcdir)/contrib/contrib-global.mk
+endif
diff --git a/src/test/modules/buffile/README b/src/test/modules/buffile/README
new file mode 100644
index 0000000000..9c80787797
--- /dev/null
+++ b/src/test/modules/buffile/README
@@ -0,0 +1,11 @@
+This extension was written to check if changes introduced due to cluster
+encryption do not break buffile.c.
+
+Caution: To make the test cheaper, it was decided to adjust the segment sizeq,
+see
+
+#define MAX_PHYSICAL_FILESIZE	(4 * BLCKSZ)
+
+in buffile.c. BLCKSZ is 8192 (the default). All the tests rely on this
+value. So if you haven't compiled Postgres with this value, the tests will
+create 1 GB files and they will fail.
diff --git a/src/test/modules/buffile/buffile--1.0.sql b/src/test/modules/buffile/buffile--1.0.sql
new file mode 100644
index 0000000000..7c168f3514
--- /dev/null
+++ b/src/test/modules/buffile/buffile--1.0.sql
@@ -0,0 +1,54 @@
+CREATE FUNCTION buffile_create()
+RETURNS void
+AS 'MODULE_PATHNAME', 'buffile_create'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_close()
+RETURNS void
+AS 'MODULE_PATHNAME', 'buffile_close'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_write(text)
+RETURNS bigint
+AS 'MODULE_PATHNAME', 'buffile_write'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_read(bigint)
+RETURNS bytea
+AS 'MODULE_PATHNAME', 'buffile_read'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_seek(int, bigint)
+RETURNS int
+AS 'MODULE_PATHNAME', 'buffile_seek'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_assert_fileno(int)
+RETURNS void
+AS 'MODULE_PATHNAME', 'buffile_assert_fileno'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_test_shared()
+RETURNS void
+AS 'MODULE_PATHNAME', 'buffile_test_shared'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_test_shared_append()
+RETURNS void
+AS 'MODULE_PATHNAME', 'buffile_test_shared_append'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_open_transient(text, bool, bool)
+RETURNS void
+AS 'MODULE_PATHNAME', 'buffile_open_transient'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_close_transient()
+RETURNS void
+AS 'MODULE_PATHNAME', 'buffile_close_transient'
+LANGUAGE C;
+
+CREATE FUNCTION buffile_delete_file(text)
+RETURNS void
+AS 'MODULE_PATHNAME', 'buffile_delete_file'
+LANGUAGE C;
diff --git a/src/test/modules/buffile/buffile.c b/src/test/modules/buffile/buffile.c
new file mode 100644
index 0000000000..fe0680771a
--- /dev/null
+++ b/src/test/modules/buffile/buffile.c
@@ -0,0 +1,429 @@
+#include <fcntl.h>
+
+#include "postgres.h"
+#include "fmgr.h"
+#include "lib/stringinfo.h"
+#include "storage/buffile.h"
+#include "utils/array.h"
+#include "utils/builtins.h"
+#include "utils/lsyscache.h"
+#include "utils/memutils.h"
+#include "utils/resowner.h"
+
+PG_MODULE_MAGIC;
+
+/*
+ * To cope with files that span multiple segments w/o wasting resources, use
+ * the smallest possible segment size. The test scripts need to set
+ * buffile_max_filesize (GUC) accordingly.
+ */
+#define MAX_PHYSICAL_FILESIZE_TEST	(4 * BLCKSZ)
+
+static BufFile *bf = NULL;
+static TransientBufFile *bft = NULL;
+
+static void check_file(void);
+
+extern Datum buffile_create_transient(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_create);
+Datum
+buffile_create(PG_FUNCTION_ARGS)
+{
+	MemoryContext old_cxt;
+	ResourceOwner old_ro;
+
+	if (bf != NULL)
+		elog(ERROR, "file already exists");
+
+	old_cxt = MemoryContextSwitchTo(TopMemoryContext);
+
+	/*
+	 * Make sure the file is not deleted across function calls.
+	 */
+	old_ro = CurrentResourceOwner;
+	CurrentResourceOwner = TopTransactionResourceOwner;
+
+	bf = BufFileCreateTemp(false);
+
+	CurrentResourceOwner = old_ro;
+	MemoryContextSwitchTo(old_cxt);
+
+	PG_RETURN_VOID();
+}
+
+extern Datum buffile_close(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_close);
+Datum
+buffile_close(PG_FUNCTION_ARGS)
+{
+	if (bf == NULL)
+		elog(ERROR, "there's no file to close");
+
+	BufFileClose(bf);
+	bf = NULL;
+
+	PG_RETURN_VOID();
+}
+
+extern Datum buffile_write(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_write);
+Datum
+buffile_write(PG_FUNCTION_ARGS)
+{
+	Datum		d = PG_GETARG_DATUM(0);
+	char	   *s = TextDatumGetCString(d);
+	size_t		res;
+
+	if (bf)
+		res = BufFileWrite(bf, s, strlen(s));
+	else if (bft)
+		res = BufFileWriteTransient(bft, s, strlen(s));
+	else
+		elog(ERROR, "No file is open");
+
+	PG_RETURN_INT64(res);
+}
+
+extern Datum buffile_read(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_read);
+Datum
+buffile_read(PG_FUNCTION_ARGS)
+{
+	int64		size = PG_GETARG_INT64(0);
+	StringInfo	buf = makeStringInfo();
+	size_t		res_size;
+	bytea	   *result;
+
+	enlargeStringInfo(buf, size);
+
+	if (bf)
+		res_size = BufFileRead(bf, buf->data, size);
+	else if (bft)
+		res_size = BufFileReadTransient(bft, buf->data, size);
+	else
+		elog(ERROR, "No file is open");
+
+	buf->len = res_size;
+
+	result = DatumGetByteaPP(DirectFunctionCall1(bytearecv,
+												 PointerGetDatum(buf)));
+	PG_RETURN_BYTEA_P(result);
+}
+
+extern Datum buffile_seek(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_seek);
+Datum
+buffile_seek(PG_FUNCTION_ARGS)
+{
+	int32		fileno = PG_GETARG_INT32(0);
+	int64		offset = PG_GETARG_INT64(1);
+	int32		res;
+
+	check_file();
+	res = BufFileSeek(bf, fileno, offset, SEEK_SET);
+
+	PG_RETURN_INT32(res);
+}
+
+extern Datum buffile_assert_fileno(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_assert_fileno);
+Datum
+buffile_assert_fileno(PG_FUNCTION_ARGS)
+{
+	int32		fileno_expected = PG_GETARG_INT32(0);
+	int32		fileno;
+	off_t		offset;
+
+	check_file();
+	BufFileTell(bf, &fileno, &offset);
+
+	if (fileno != fileno_expected)
+	{
+		/*
+		 * Bring the backend down so that the following tests have no chance
+		 * to create the 1GB files.
+		 */
+		elog(FATAL, "file number does not match");
+	}
+
+	PG_RETURN_VOID();
+}
+
+static void
+check_file(void)
+{
+	if (bf == NULL)
+		elog(ERROR, "the file is not opened");
+}
+
+/*
+ * This test is especially important for shared encrypted files, see the
+ * comments below.
+ */
+extern Datum buffile_test_shared(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_test_shared);
+Datum
+buffile_test_shared(PG_FUNCTION_ARGS)
+{
+	dsm_segment *seg;
+	SharedFileSet *fileset;
+	BufFile    *bf_1,
+			   *bf_2;
+	char	   *data_1,
+			   *data_2,
+			   *data;
+	Size		chunk_size_1,
+				chunk_size_2;
+	int			fileno,
+				i;
+	off_t		offset,
+				res,
+				total_size;
+
+	/*
+	 * The size is not important, we actually do not need the shared memory.
+	 * The segment is only needed to initialize the fileset.
+	 */
+	seg = dsm_create(1024, 0);
+
+	/*
+	 * The fileset must survive error handling, so that dsm_detach works fine.
+	 * (The typical use case is that the fileset is in shared memory.)
+	 */
+	fileset = (SharedFileSet *) MemoryContextAlloc(TopTransactionContext,
+												   sizeof(SharedFileSet));
+	SharedFileSetInit(fileset, seg);
+
+	bf_1 = BufFileCreateShared(fileset, "file_1");
+
+	/*
+	 * Write more data than the buffer size, so that we can check that the
+	 * number of "useful bytes" word is only appended at the end of the
+	 * segment, not after each buffer.
+	 */
+	chunk_size_1 = BLCKSZ + 256;
+	data_1 = (char *) palloc(chunk_size_1);
+	memset(data_1, 1, chunk_size_1);
+	if (BufFileWrite(bf_1, data_1, chunk_size_1) != chunk_size_1)
+		elog(ERROR, "Failed to write data");
+	pfree(data_1);
+
+	/*
+	 * Enforce buffer flush (The BufFileFlush() function is not exported).
+	 * Thus the "useful bytes" metadata should appear at the current end the
+	 * first file segment. The next write will have to seek back to overwrite
+	 * the metadata.
+	 */
+	BufFileTell(bf_1, &fileno, &offset);
+	if (BufFileSeek(bf_1, 0, 0, SEEK_SET) != 0)
+		elog(ERROR, "seek failed");
+	if (BufFileSeek(bf_1, fileno, offset, SEEK_SET) != 0)
+		elog(ERROR, "seek failed");
+
+	/*
+	 * Write another chunk that does not fit into the first segment file. Thus
+	 * the "useful bytes" metadata should appear at the end of both segments.
+	 */
+	chunk_size_2 = 3 * BLCKSZ;
+	data_2 = (char *) palloc(chunk_size_2);
+	memset(data_2, 1, chunk_size_2);
+	if (BufFileWrite(bf_1, data_2, chunk_size_2) != chunk_size_2)
+		elog(ERROR, "Failed to write data");
+	pfree(data_2);
+	BufFileClose(bf_1);
+
+	/*
+	 * The word indicating the number of "useful bytes" (i.e. the actual data
+	 * w/o padding to buffer size) is stored at the end of each segment file.
+	 * Check that this metadata is read correctly.
+	 */
+	bf_2 = BufFileOpenShared(fileset, "file_1");
+	total_size = BufFileSize(bf_2);
+	if (total_size != (chunk_size_1 + chunk_size_2))
+		elog(ERROR, "Incorrect file size: %zu", total_size);
+
+	data = (char *) palloc(total_size);
+	res = BufFileRead(bf_2, data, total_size);
+	if (res != total_size)
+		elog(ERROR, "Incorrect chunk size read: %zu", res);
+	for (i = 0; i < total_size; i++)
+		if (data[i] != 1)
+			elog(ERROR, "Unexpected data read from the file");
+	pfree(data);
+	BufFileClose(bf_2);
+
+	dsm_detach(seg);
+
+	PG_RETURN_VOID();
+}
+
+
+/*
+ * Test BufFileAppend().
+ */
+extern Datum buffile_test_shared_append(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_test_shared_append);
+Datum
+buffile_test_shared_append(PG_FUNCTION_ARGS)
+{
+	dsm_segment *seg;
+	SharedFileSet *fileset;
+	BufFile    *bf_1,
+			   *bf_2,
+			   *bf_3;
+	char	   *data;
+	Size		chunk_size;
+	int			i;
+	off_t		res,
+				total_size;
+
+	seg = dsm_create(1024, 0);
+
+	fileset = (SharedFileSet *) MemoryContextAlloc(TopTransactionContext,
+												   sizeof(SharedFileSet));
+	SharedFileSetInit(fileset, seg);
+
+	/*
+	 * XXX Does the chunk size matter much?
+	 */
+	chunk_size = 8;
+	data = (char *) palloc(chunk_size);
+	memset(data, 1, chunk_size);
+
+	bf_1 = BufFileCreateShared(fileset, "file_1");
+	if (BufFileWrite(bf_1, data, chunk_size) != chunk_size)
+		elog(ERROR, "Failed to write data");
+
+	bf_2 = BufFileCreateShared(fileset, "file_2");
+	if (BufFileWrite(bf_2, data, chunk_size) != chunk_size)
+		elog(ERROR, "Failed to write data");
+
+	/*
+	 * Make sure it's read-only so that BufFileAppend() can accept it as
+	 * source.
+	 */
+	BufFileClose(bf_2);
+	bf_2 = BufFileOpenShared(fileset, "file_2");
+
+	bf_3 = BufFileCreateShared(fileset, "file_3");
+	if (BufFileWrite(bf_3, data, chunk_size) != chunk_size)
+		elog(ERROR, "Failed to write data");
+	BufFileClose(bf_3);
+	bf_3 = BufFileOpenShared(fileset, "file_3");
+
+	BufFileAppend(bf_1, bf_2);
+	BufFileAppend(bf_1, bf_3);
+
+	total_size = BufFileSize(bf_1);
+
+	/*
+	 * The result should contain complete segments of bf_1 and bf_2 and the
+	 * valid part of bf_3.
+	 */
+	if (total_size != (2 * MAX_PHYSICAL_FILESIZE_TEST + chunk_size))
+		elog(ERROR, "Incorrect total size of the appended data: %zu",
+			 total_size);
+
+	/*
+	 * Check that data of the 2nd segment was decrypted correctly.
+	 */
+	if (BufFileSeek(bf_1, 1, 0, SEEK_SET) != 0)
+		elog(ERROR, "seek failed");
+	res = BufFileRead(bf_1, data, chunk_size);
+	if (res != chunk_size)
+		elog(ERROR, "Incorrect chunk size read: %zu", res);
+	for (i = 0; i < chunk_size; i++)
+		if (data[i] != 1)
+			elog(ERROR, "Unexpected data read from the file");
+
+	/*
+	 * And the same for the 3rd segment.
+	 *
+	 * TODO Reuse the code above by putting it into a function.
+	 */
+	if (BufFileSeek(bf_1, 2, 0, SEEK_SET) != 0)
+		elog(ERROR, "seek failed");
+	res = BufFileRead(bf_1, data, chunk_size);
+	if (res != chunk_size)
+		elog(ERROR, "Incorrect chunk size read: %zu", res);
+	for (i = 0; i < chunk_size; i++)
+		if (data[i] != 1)
+			elog(ERROR, "Unexpected data read from the file");
+
+	BufFileClose(bf_1);
+	dsm_detach(seg);
+	PG_RETURN_VOID();
+}
+
+extern Datum buffile_open_transient(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_open_transient);
+Datum
+buffile_open_transient(PG_FUNCTION_ARGS)
+{
+	MemoryContext old_cxt;
+	Datum		d = PG_GETARG_DATUM(0);
+	char	   *path = TextDatumGetCString(d);
+	bool		write_only = PG_GETARG_BOOL(1);
+	bool		append = PG_GETARG_BOOL(2);
+	int			flags = O_CREAT | PG_BINARY;
+
+	if (bft != NULL)
+		elog(ERROR, "file already exists");
+
+	if (write_only)
+		flags |= O_WRONLY;
+	if (append)
+		flags |= O_APPEND;
+
+	old_cxt = MemoryContextSwitchTo(TopMemoryContext);
+
+	/*
+	 * Make sure the file is not deleted across function calls.
+	 */
+	bft = BufFileOpenTransient(path, flags);
+
+	MemoryContextSwitchTo(old_cxt);
+
+	PG_RETURN_VOID();
+}
+
+extern Datum buffile_close_transient(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_close_transient);
+Datum
+buffile_close_transient(PG_FUNCTION_ARGS)
+{
+	if (bft == NULL)
+		elog(ERROR, "there's no file to close");
+
+	BufFileCloseTransient(bft);
+	bft = NULL;
+
+	PG_RETURN_VOID();
+}
+
+extern Datum buffile_delete_file(PG_FUNCTION_ARGS);
+
+PG_FUNCTION_INFO_V1(buffile_delete_file);
+Datum
+buffile_delete_file(PG_FUNCTION_ARGS)
+{
+	Datum		d = PG_GETARG_DATUM(0);
+	char	   *path = TextDatumGetCString(d);
+
+	if (bft != NULL)
+		elog(ERROR, "the file is still open");
+
+	PathNameDeleteTemporaryFile(path, true);
+
+	PG_RETURN_VOID();
+}
diff --git a/src/test/modules/buffile/buffile.control b/src/test/modules/buffile/buffile.control
new file mode 100644
index 0000000000..8472c5a348
--- /dev/null
+++ b/src/test/modules/buffile/buffile.control
@@ -0,0 +1,5 @@
+# buffile_test extension
+comment = 'buffile_test'
+default_version = '1.0'
+module_pathname = '$libdir/buffile_test'
+relocatable = true
diff --git a/src/test/modules/buffile/expected/test_01.out b/src/test/modules/buffile/expected/test_01.out
new file mode 100644
index 0000000000..c1a4f5ac7b
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_01.out
@@ -0,0 +1,61 @@
+-- This is the first test, so make sure the test extension is there.
+CREATE EXTENSION IF NOT EXISTS buffile;
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+SELECT buffile_seek(0, 1);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_write('abc');
+ buffile_write 
+---------------
+             3
+(1 row)
+
+SELECT buffile_seek(0, 0);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+-- Check that the trailing zeroes are not fetched.
+SELECT buffile_read(16);
+ buffile_read 
+--------------
+ \x00616263
+(1 row)
+
+-- Adjust the number of useful bytes.
+SELECT buffile_write('abc');
+ buffile_write 
+---------------
+             3
+(1 row)
+
+-- ... and check again.
+SELECT buffile_seek(0, 0);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(16);
+   buffile_read   
+------------------
+ \x00616263616263
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_02.out b/src/test/modules/buffile/expected/test_02.out
new file mode 100644
index 0000000000..f783d0cb24
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_02.out
@@ -0,0 +1,48 @@
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+SELECT buffile_seek(0, 8189);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+-- Initialize the last 3 positions of the first buffer and the initial 3
+-- positions of the 2nd buffer.
+SELECT buffile_write('abcdef');
+ buffile_write 
+---------------
+             6
+(1 row)
+
+SELECT buffile_seek(0, 0);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+-- Read the first buffer.
+SELECT length(buffile_read(8192));
+ length 
+--------
+   8192
+(1 row)
+
+-- Only 3 bytes of the 2nd buffer should be fetched.
+SELECT length(buffile_read(8192));
+ length 
+--------
+      3
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_03.out b/src/test/modules/buffile/expected/test_03.out
new file mode 100644
index 0000000000..e899fa3b38
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_03.out
@@ -0,0 +1,27 @@
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+-- Read from an empty file.
+SELECT buffile_seek(0, 8);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(16);
+ buffile_read 
+--------------
+ \x
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_04.out b/src/test/modules/buffile/expected/test_04.out
new file mode 100644
index 0000000000..1f8eeabe48
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_04.out
@@ -0,0 +1,84 @@
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+-- Write something near the end of the first buffer, but leave some trailing
+-- space.
+SELECT buffile_seek(0, 8184);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_write('abcd');
+ buffile_write 
+---------------
+             4
+(1 row)
+
+-- Leave the 2nd buffer empty, as well as a few leading bytes. Thus we should
+-- get a hole that spans the whole 2nd buffer as well as a few adjacent bytes
+-- on each side.
+SELECT buffile_seek(0, 2 * 8192 + 4);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_write('efgh');
+ buffile_write 
+---------------
+             4
+(1 row)
+
+-- Check the initial part of the hole, which crosses the boundary of the 1st
+-- and the 2nd buffer.
+SELECT buffile_seek(0, 8184);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(16);
+            buffile_read            
+------------------------------------
+ \x61626364000000000000000000000000
+(1 row)
+
+-- Check the trailing part of the whole, which crosses the boundary of the 2nd
+-- and the 3rd buffer.
+SELECT buffile_seek(0, 2 * 8192 - 8);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(16);
+            buffile_read            
+------------------------------------
+ \x00000000000000000000000065666768
+(1 row)
+
+-- Check that the hole contains nothing but zeroes.
+SELECT buffile_seek(0, 8192 - 4);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT btrim(buffile_read(8192 + 8), '\x00');
+ btrim 
+-------
+ \x
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_05.out b/src/test/modules/buffile/expected/test_05.out
new file mode 100644
index 0000000000..6a73147711
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_05.out
@@ -0,0 +1,33 @@
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+-- Seek does not extend the file if it's not followed by write.
+SELECT buffile_seek(0, 1);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_seek(0, 0);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(2);
+ buffile_read 
+--------------
+ \x
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_06.out b/src/test/modules/buffile/expected/test_06.out
new file mode 100644
index 0000000000..d406bda9e6
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_06.out
@@ -0,0 +1,43 @@
+-- This test shows that the if first component file (segment) stays empty,
+-- read stops prematurely even if it starts within that segment, although it'd
+-- otherwise receive some data from the following one.
+BEGIN;
+-- Neither disk space nor time needs to be wasted.
+SET buffile_max_filesize TO 8192;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+SELECT buffile_seek(0, 8192);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_write('a');
+ buffile_write 
+---------------
+             1
+(1 row)
+
+SELECT buffile_seek(0, 8191);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(2);
+ buffile_read 
+--------------
+ \x
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_07.out b/src/test/modules/buffile/expected/test_07.out
new file mode 100644
index 0000000000..6f2ee3d2d9
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_07.out
@@ -0,0 +1,41 @@
+BEGIN;
+-- Use a small segment, not to waste disk space and time.
+SET buffile_max_filesize TO 8192;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+-- Write data at component file boundary and try to read it.
+SELECT buffile_seek(0, 8192);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_write('abcd');
+ buffile_write 
+---------------
+             4
+(1 row)
+
+SELECT buffile_seek(0, 8192);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(8);
+ buffile_read 
+--------------
+ \x61626364
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_08.out b/src/test/modules/buffile/expected/test_08.out
new file mode 100644
index 0000000000..6ffcb9dcd6
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_08.out
@@ -0,0 +1,41 @@
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+-- Neither disk space nor time needs to be wasted.
+SET buffile_max_filesize TO 8192;
+-- Write data across component file boundary and try to read it.
+SELECT buffile_seek(0, 8190);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_write('abcd');
+ buffile_write 
+---------------
+             4
+(1 row)
+
+SELECT buffile_seek(0, 8190);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(8);
+ buffile_read 
+--------------
+ \x61626364
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_09.out b/src/test/modules/buffile/expected/test_09.out
new file mode 100644
index 0000000000..b9d325b676
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_09.out
@@ -0,0 +1,39 @@
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+-- Write data across buffer boundary and try to read it.
+SELECT buffile_seek(0, 8190);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_write('abcd');
+ buffile_write 
+---------------
+             4
+(1 row)
+
+SELECT buffile_seek(0, 8190);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(8);
+ buffile_read 
+--------------
+ \x61626364
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_10.out b/src/test/modules/buffile/expected/test_10.out
new file mode 100644
index 0000000000..8e457fdcda
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_10.out
@@ -0,0 +1,76 @@
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+-- Write some data at the end of the buffer.
+SELECT buffile_seek(0, 8188);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_write('abcd');
+ buffile_write 
+---------------
+             4
+(1 row)
+
+SELECT buffile_seek(0, 8189);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+-- Enforce flush with the write position not at the end of the buffer. This is
+-- special by not moving curOffset to the next buffer.
+SELECT buffile_read(1);
+ buffile_read 
+--------------
+ \x62
+(1 row)
+
+-- Therefore the next writes should eventually affect the original data. (Here
+-- we also test going directly from read to write and vice versa.)
+SELECT buffile_write('x');
+ buffile_write 
+---------------
+             1
+(1 row)
+
+SELECT buffile_read(1);
+ buffile_read 
+--------------
+ \x64
+(1 row)
+
+-- Start a new buffer, i.e. force flushing of the previous one.
+SELECT buffile_write('z');
+ buffile_write 
+---------------
+             1
+(1 row)
+
+-- Check that the 'x' and 'y' letters are in the first buffer, not in the
+-- 2nd. (We read enough data to find any non-zero bytes in the 2nd buffer.)
+SELECT buffile_seek(0, 8188);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+SELECT buffile_read(4 + 8192);
+ buffile_read 
+--------------
+ \x616278647a
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_11.out b/src/test/modules/buffile/expected/test_11.out
new file mode 100644
index 0000000000..c6804d08af
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_11.out
@@ -0,0 +1,34 @@
+BEGIN;
+SELECT buffile_create();
+ buffile_create 
+----------------
+ 
+(1 row)
+
+SELECT buffile_write('abcd');
+ buffile_write 
+---------------
+             4
+(1 row)
+
+-- Seek beyond EOF not followed by write.
+SELECT buffile_seek(0, 5);
+ buffile_seek 
+--------------
+            0
+(1 row)
+
+-- Nothing should be fetched.
+SELECT buffile_read(8);
+ buffile_read 
+--------------
+ \x
+(1 row)
+
+SELECT buffile_close();
+ buffile_close 
+---------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/expected/test_12.out b/src/test/modules/buffile/expected/test_12.out
new file mode 100644
index 0000000000..fe507a8e58
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_12.out
@@ -0,0 +1,16 @@
+-- Neither disk space nor time needs to be wasted.
+--
+-- Here the tests are designed for segments of multiple buffers.
+SET buffile_max_filesize TO 32768;
+SELECT buffile_test_shared();
+ buffile_test_shared 
+---------------------
+ 
+(1 row)
+
+SELECT buffile_test_shared_append();
+ buffile_test_shared_append 
+----------------------------
+ 
+(1 row)
+
diff --git a/src/test/modules/buffile/expected/test_13.out b/src/test/modules/buffile/expected/test_13.out
new file mode 100644
index 0000000000..7ba50a9c9d
--- /dev/null
+++ b/src/test/modules/buffile/expected/test_13.out
@@ -0,0 +1,86 @@
+-- Use transaction block so that the file does not closed automatically at
+-- command boundary.
+BEGIN;
+SELECT buffile_open_transient('trans1', true, false);
+ buffile_open_transient 
+------------------------
+ 
+(1 row)
+
+SELECT buffile_write('01234567');
+ buffile_write 
+---------------
+             8
+(1 row)
+
+SELECT buffile_close_transient();
+ buffile_close_transient 
+-------------------------
+ 
+(1 row)
+
+-- Open for reading.
+SELECT buffile_open_transient('trans1', false, false);
+ buffile_open_transient 
+------------------------
+ 
+(1 row)
+
+SELECT length(buffile_read(65536));
+ length 
+--------
+      8
+(1 row)
+
+SELECT buffile_close_transient();
+ buffile_close_transient 
+-------------------------
+ 
+(1 row)
+
+-- Open for writing in append mode.
+SELECT buffile_open_transient('trans1', true, true);
+ buffile_open_transient 
+------------------------
+ 
+(1 row)
+
+-- Add BLCKSZ bytes, so that buffer boundary is crossed.
+SELECT buffile_write(repeat('x', 8192));
+ buffile_write 
+---------------
+          8192
+(1 row)
+
+SELECT buffile_close_transient();
+ buffile_close_transient 
+-------------------------
+ 
+(1 row)
+
+-- Open for reading and verify the valid part.
+SELECT buffile_open_transient('trans1', false, false);
+ buffile_open_transient 
+------------------------
+ 
+(1 row)
+
+SELECT length(buffile_read(65536));
+ length 
+--------
+   8200
+(1 row)
+
+SELECT buffile_close_transient();
+ buffile_close_transient 
+-------------------------
+ 
+(1 row)
+
+SELECT buffile_delete_file('trans1');
+ buffile_delete_file 
+---------------------
+ 
+(1 row)
+
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_01.sql b/src/test/modules/buffile/sql/test_01.sql
new file mode 100644
index 0000000000..92331dd0de
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_01.sql
@@ -0,0 +1,17 @@
+-- This is the first test, so make sure the test extension is there.
+CREATE EXTENSION IF NOT EXISTS buffile;
+
+BEGIN;
+SELECT buffile_create();
+SELECT buffile_seek(0, 1);
+SELECT buffile_write('abc');
+SELECT buffile_seek(0, 0);
+-- Check that the trailing zeroes are not fetched.
+SELECT buffile_read(16);
+-- Adjust the number of useful bytes.
+SELECT buffile_write('abc');
+-- ... and check again.
+SELECT buffile_seek(0, 0);
+SELECT buffile_read(16);
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_02.sql b/src/test/modules/buffile/sql/test_02.sql
new file mode 100644
index 0000000000..35d34722d7
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_02.sql
@@ -0,0 +1,13 @@
+BEGIN;
+SELECT buffile_create();
+SELECT buffile_seek(0, 8189);
+-- Initialize the last 3 positions of the first buffer and the initial 3
+-- positions of the 2nd buffer.
+SELECT buffile_write('abcdef');
+SELECT buffile_seek(0, 0);
+-- Read the first buffer.
+SELECT length(buffile_read(8192));
+-- Only 3 bytes of the 2nd buffer should be fetched.
+SELECT length(buffile_read(8192));
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_03.sql b/src/test/modules/buffile/sql/test_03.sql
new file mode 100644
index 0000000000..a95391f7c3
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_03.sql
@@ -0,0 +1,7 @@
+BEGIN;
+SELECT buffile_create();
+-- Read from an empty file.
+SELECT buffile_seek(0, 8);
+SELECT buffile_read(16);
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_04.sql b/src/test/modules/buffile/sql/test_04.sql
new file mode 100644
index 0000000000..64e8d39f94
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_04.sql
@@ -0,0 +1,25 @@
+BEGIN;
+SELECT buffile_create();
+-- Write something near the end of the first buffer, but leave some trailing
+-- space.
+SELECT buffile_seek(0, 8184);
+SELECT buffile_write('abcd');
+-- Leave the 2nd buffer empty, as well as a few leading bytes. Thus we should
+-- get a hole that spans the whole 2nd buffer as well as a few adjacent bytes
+-- on each side.
+SELECT buffile_seek(0, 2 * 8192 + 4);
+SELECT buffile_write('efgh');
+-- Check the initial part of the hole, which crosses the boundary of the 1st
+-- and the 2nd buffer.
+SELECT buffile_seek(0, 8184);
+SELECT buffile_read(16);
+-- Check the trailing part of the whole, which crosses the boundary of the 2nd
+-- and the 3rd buffer.
+SELECT buffile_seek(0, 2 * 8192 - 8);
+SELECT buffile_read(16);
+-- Check that the hole contains nothing but zeroes.
+SELECT buffile_seek(0, 8192 - 4);
+SELECT btrim(buffile_read(8192 + 8), '\x00');
+
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_05.sql b/src/test/modules/buffile/sql/test_05.sql
new file mode 100644
index 0000000000..5fd642e558
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_05.sql
@@ -0,0 +1,8 @@
+BEGIN;
+SELECT buffile_create();
+-- Seek does not extend the file if it's not followed by write.
+SELECT buffile_seek(0, 1);
+SELECT buffile_seek(0, 0);
+SELECT buffile_read(2);
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_06.sql b/src/test/modules/buffile/sql/test_06.sql
new file mode 100644
index 0000000000..35aa48857c
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_06.sql
@@ -0,0 +1,15 @@
+-- This test shows that the if first component file (segment) stays empty,
+-- read stops prematurely even if it starts within that segment, although it'd
+-- otherwise receive some data from the following one.
+BEGIN;
+
+-- Neither disk space nor time needs to be wasted.
+SET buffile_max_filesize TO 8192;
+
+SELECT buffile_create();
+SELECT buffile_seek(0, 8192);
+SELECT buffile_write('a');
+SELECT buffile_seek(0, 8191);
+SELECT buffile_read(2);
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_07.sql b/src/test/modules/buffile/sql/test_07.sql
new file mode 100644
index 0000000000..06057c5a1b
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_07.sql
@@ -0,0 +1,13 @@
+BEGIN;
+
+-- Use a small segment, not to waste disk space and time.
+SET buffile_max_filesize TO 8192;
+
+SELECT buffile_create();
+-- Write data at component file boundary and try to read it.
+SELECT buffile_seek(0, 8192);
+SELECT buffile_write('abcd');
+SELECT buffile_seek(0, 8192);
+SELECT buffile_read(8);
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_08.sql b/src/test/modules/buffile/sql/test_08.sql
new file mode 100644
index 0000000000..2736dc22df
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_08.sql
@@ -0,0 +1,13 @@
+BEGIN;
+SELECT buffile_create();
+
+-- Neither disk space nor time needs to be wasted.
+SET buffile_max_filesize TO 8192;
+
+-- Write data across component file boundary and try to read it.
+SELECT buffile_seek(0, 8190);
+SELECT buffile_write('abcd');
+SELECT buffile_seek(0, 8190);
+SELECT buffile_read(8);
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_09.sql b/src/test/modules/buffile/sql/test_09.sql
new file mode 100644
index 0000000000..cc7060932e
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_09.sql
@@ -0,0 +1,9 @@
+BEGIN;
+SELECT buffile_create();
+-- Write data across buffer boundary and try to read it.
+SELECT buffile_seek(0, 8190);
+SELECT buffile_write('abcd');
+SELECT buffile_seek(0, 8190);
+SELECT buffile_read(8);
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_10.sql b/src/test/modules/buffile/sql/test_10.sql
new file mode 100644
index 0000000000..63af760d9f
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_10.sql
@@ -0,0 +1,25 @@
+BEGIN;
+SELECT buffile_create();
+-- Write some data at the end of the buffer.
+SELECT buffile_seek(0, 8188);
+SELECT buffile_write('abcd');
+SELECT buffile_seek(0, 8189);
+-- Enforce flush with the write position not at the end of the buffer. This is
+-- special by not moving curOffset to the next buffer.
+SELECT buffile_read(1);
+
+-- Therefore the next writes should eventually affect the original data. (Here
+-- we also test going directly from read to write and vice versa.)
+SELECT buffile_write('x');
+SELECT buffile_read(1);
+
+-- Start a new buffer, i.e. force flushing of the previous one.
+SELECT buffile_write('z');
+
+-- Check that the 'x' and 'y' letters are in the first buffer, not in the
+-- 2nd. (We read enough data to find any non-zero bytes in the 2nd buffer.)
+SELECT buffile_seek(0, 8188);
+SELECT buffile_read(4 + 8192);
+
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_11.sql b/src/test/modules/buffile/sql/test_11.sql
new file mode 100644
index 0000000000..94300d253f
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_11.sql
@@ -0,0 +1,9 @@
+BEGIN;
+SELECT buffile_create();
+SELECT buffile_write('abcd');
+-- Seek beyond EOF not followed by write.
+SELECT buffile_seek(0, 5);
+-- Nothing should be fetched.
+SELECT buffile_read(8);
+SELECT buffile_close();
+COMMIT;
diff --git a/src/test/modules/buffile/sql/test_12.sql b/src/test/modules/buffile/sql/test_12.sql
new file mode 100644
index 0000000000..017050ef5b
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_12.sql
@@ -0,0 +1,7 @@
+-- Neither disk space nor time needs to be wasted.
+--
+-- Here the tests are designed for segments of multiple buffers.
+SET buffile_max_filesize TO 32768;
+
+SELECT buffile_test_shared();
+SELECT buffile_test_shared_append();
diff --git a/src/test/modules/buffile/sql/test_13.sql b/src/test/modules/buffile/sql/test_13.sql
new file mode 100644
index 0000000000..bcabf7bfe6
--- /dev/null
+++ b/src/test/modules/buffile/sql/test_13.sql
@@ -0,0 +1,25 @@
+-- Use transaction block so that the file does not closed automatically at
+-- command boundary.
+BEGIN;
+SELECT buffile_open_transient('trans1', true, false);
+SELECT buffile_write('01234567');
+SELECT buffile_close_transient();
+
+-- Open for reading.
+SELECT buffile_open_transient('trans1', false, false);
+SELECT length(buffile_read(65536));
+SELECT buffile_close_transient();
+
+-- Open for writing in append mode.
+SELECT buffile_open_transient('trans1', true, true);
+-- Add BLCKSZ bytes, so that buffer boundary is crossed.
+SELECT buffile_write(repeat('x', 8192));
+SELECT buffile_close_transient();
+
+-- Open for reading and verify the valid part.
+SELECT buffile_open_transient('trans1', false, false);
+SELECT length(buffile_read(65536));
+SELECT buffile_close_transient();
+
+SELECT buffile_delete_file('trans1');
+COMMIT;
-- 
2.13.7


--=-=-=
Content-Type: text/x-diff
Content-Disposition: attachment;
 filename=v04-0013-Make-database-OID-constant-for-postgres-database.patch



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v1 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 196 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 209 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..5bb11add3c 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,190 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+	if (wc->rpSkipTo != ST_NEXT_ROW)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("only AFTER MATCH SKIP TO NEXT_ROW is supported")));
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v1-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v1 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 196 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 209 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..5bb11add3c 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,190 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+	if (wc->rpSkipTo != ST_NEXT_ROW)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("only AFTER MATCH SKIP TO NEXT_ROW is supported")));
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v1-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v1 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 196 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 209 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..5bb11add3c 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,190 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+	if (wc->rpSkipTo != ST_NEXT_ROW)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("only AFTER MATCH SKIP TO NEXT_ROW is supported")));
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v1-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v1 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 196 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 209 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..5bb11add3c 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,190 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+	if (wc->rpSkipTo != ST_NEXT_ROW)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("only AFTER MATCH SKIP TO NEXT_ROW is supported")));
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v1-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v1 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-25 11:48 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-25 11:48 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 196 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 209 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..5bb11add3c 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,190 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+	if (wc->rpSkipTo != ST_NEXT_ROW)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("only AFTER MATCH SKIP TO NEXT_ROW is supported")));
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v1-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v2 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 192 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 205 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..ccf3332bef 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,186 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v2-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v2 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 192 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 205 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..ccf3332bef 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,186 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v2-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v2 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 192 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 205 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..ccf3332bef 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,186 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v2-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v2 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 192 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 205 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..ccf3332bef 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,186 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v2-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v2 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-06-26 08:05 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-06-26 08:05 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 192 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 205 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index f61f794755..ccf3332bef 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2949,6 +2952,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3814,3 +3821,186 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static List *
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	List		*patterns;
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return NULL;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+	return patterns;
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 346fd272b6..20231d9ec0 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -541,6 +541,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1754,6 +1755,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3133,6 +3135,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v2-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v3 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 190 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 203 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..ea2decc579 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,184 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v3-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v3 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 190 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 203 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..ea2decc579 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,184 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v3-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v3 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 190 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 203 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..ea2decc579 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,184 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v3-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v3 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-07-26 10:49 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-07-26 10:49 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 ++
 src/backend/parser/parse_clause.c | 190 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 203 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..ea2decc579 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void  transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,184 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	/* Check Frame option. Frame must start at current row */
+
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc;
+	ResTarget	*restarget, *r;
+	List		*restargets;
+
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	*name;
+		ListCell	*l;
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Make sure that row pattern definition search condition is a boolean
+		 * expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v3-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v4 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v4-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v4 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v4-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v4 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v4-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v4 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v4-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v4 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v4-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v4 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-08-09 07:56 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-08-09 07:56 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index fed8e4d089..8921b7ae01 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v4-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v5 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 64c582c344..18b58ac263 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v5-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v5 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 64c582c344..18b58ac263 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v5-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread

* [PATCH v5 2/7] Row pattern recognition patch (parse/analysis).
@ 2023-09-02 06:32 Tatsuo Ishii <[email protected]>
  0 siblings, 0 replies; 124+ messages in thread

From: Tatsuo Ishii @ 2023-09-02 06:32 UTC (permalink / raw)

---
 src/backend/parser/parse_agg.c    |   7 +
 src/backend/parser/parse_clause.c | 292 +++++++++++++++++++++++++++++-
 src/backend/parser/parse_expr.c   |   4 +
 src/backend/parser/parse_func.c   |   3 +
 4 files changed, 305 insertions(+), 1 deletion(-)

diff --git a/src/backend/parser/parse_agg.c b/src/backend/parser/parse_agg.c
index 85cd47b7ae..aa7a1cee80 100644
--- a/src/backend/parser/parse_agg.c
+++ b/src/backend/parser/parse_agg.c
@@ -564,6 +564,10 @@ check_agglevels_and_constraints(ParseState *pstate, Node *expr)
 			errkind = true;
 			break;
 
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
+
 			/*
 			 * There is intentionally no default: case here, so that the
 			 * compiler will warn if we add a new ParseExprKind without
@@ -953,6 +957,9 @@ transformWindowFuncCall(ParseState *pstate, WindowFunc *wfunc,
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_clause.c b/src/backend/parser/parse_clause.c
index 334b9b42bd..60020a7025 100644
--- a/src/backend/parser/parse_clause.c
+++ b/src/backend/parser/parse_clause.c
@@ -100,7 +100,10 @@ static WindowClause *findWindowClause(List *wclist, const char *name);
 static Node *transformFrameOffset(ParseState *pstate, int frameOptions,
 								  Oid rangeopfamily, Oid rangeopcintype, Oid *inRangeFunc,
 								  Node *clause);
-
+static void transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static List *transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist);
+static void transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
+static List *transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef);
 
 /*
  * transformFromClause -
@@ -2950,6 +2953,10 @@ transformWindowDefinitions(ParseState *pstate,
 											 rangeopfamily, rangeopcintype,
 											 &wc->endInRangeFunc,
 											 windef->endOffset);
+
+		/* Process Row Pattern Recognition related clauses */
+		transformRPR(pstate, wc, windef, targetlist);
+
 		wc->runCondition = NIL;
 		wc->winref = winref;
 
@@ -3815,3 +3822,286 @@ transformFrameOffset(ParseState *pstate, int frameOptions,
 
 	return node;
 }
+
+/*
+ * transformRPR
+ *		Process Row Pattern Recognition related clauses
+ */
+static void
+transformRPR(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/*
+	 * Window definition exists?
+	 */
+	if (windef == NULL)
+		return;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/* Check Frame option. Frame must start at current row */
+	if ((wc->frameOptions & FRAMEOPTION_START_CURRENT_ROW) == 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_SYNTAX_ERROR),
+				 errmsg("FRAME must start at current row when row patttern recognition is used")));
+
+	/* Transform AFTER MACH SKIP TO clause */
+	wc->rpSkipTo = windef->rpCommonSyntax->rpSkipTo;
+
+	/* Transform SEEK or INITIAL clause */
+	wc->initial = windef->rpCommonSyntax->initial;
+
+	/* Transform DEFINE clause into list of TargetEntry's */
+	wc->defineClause = transformDefineClause(pstate, wc, windef, targetlist);
+
+	/* Check PATTERN clause and copy to patternClause */
+	transformPatternClause(pstate, wc, windef);
+
+	/* Transform MEASURE clause */
+	transformMeasureClause(pstate, wc, windef);
+}
+
+/*
+ * transformDefineClause Process DEFINE clause and transform ResTarget into
+ *		list of TargetEntry.
+ *
+ * XXX we only support column reference in row pattern definition search
+ * condition, e.g. "price". <row pattern definition variable name>.<column
+ * reference> is not supported, e.g. "A.price".
+ */
+static List *
+transformDefineClause(ParseState *pstate, WindowClause *wc, WindowDef *windef, List **targetlist)
+{
+	/* DEFINE variable name initials */
+	static	char	*defineVariableInitials = "abcdefghijklmnopqrstuvwxyz";
+
+	ListCell		*lc, *l;
+	ResTarget		*restarget, *r;
+	List			*restargets;
+	char			*name;
+	int				initialLen;
+	int				i;
+
+	/*
+	 * If Row Definition Common Syntax exists, DEFINE clause must exist.
+	 * (the raw parser should have already checked it.)
+	 */
+	Assert(windef->rpCommonSyntax->rpDefs != NULL);
+
+	/*
+	 * Check and add "A AS A IS TRUE" if pattern variable is missing in DEFINE
+	 * per the SQL standard.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			/*
+			 * "name" is missing. So create "name AS name IS TRUE" ResTarget
+			 * node and add it to the temporary list.
+			 */
+			A_Const	   *n;
+
+			restarget = makeNode(ResTarget);
+			n = makeNode(A_Const);
+			n->val.boolval.type = T_Boolean;
+			n->val.boolval.boolval = true;
+			n->location = -1;
+			restarget->name = pstrdup(name);
+			restarget->indirection = NIL;
+			restarget->val = (Node *)n;
+			restarget->location = -1;
+			restargets = lappend((List *)restargets, restarget);
+		}
+	}
+
+	if (list_length(restargets) >= 1)
+	{
+		/* add missing DEFINEs */
+		windef->rpCommonSyntax->rpDefs = list_concat(windef->rpCommonSyntax->rpDefs,
+													 restargets);
+		list_free(restargets);
+	}
+
+	/*
+	 * Check for duplicate row pattern definition variables.  The standard
+	 * requires that no two row pattern definition variable names shall be
+	 * equivalent.
+	 */
+	restargets = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		/*
+		 * Add DEFINE expression (Restarget->val) to the targetlist as a
+		 * TargetEntry if it does not exist yet. Planner will add the column
+		 * ref var node to the outer plan's target list later on. This makes
+		 * DEFINE expression could access the outer tuple while evaluating
+		 * PATTERN.
+		 *
+		 * XXX: adding whole expressions of DEFINE to the plan.targetlist is
+		 * not so good, because it's not necessary to evalute the expression
+		 * in the target list while running the plan. We should extract the
+		 * var nodes only then add them to the plan.targetlist.
+		 */
+		findTargetlistEntrySQL99(pstate, (Node *)restarget->val, targetlist, EXPR_KIND_RPR_DEFINE);
+
+		/*
+		 * Make sure that the row pattern definition search condition is a
+		 * boolean expression.
+		 */
+		transformWhereClause(pstate, restarget->val,
+							 EXPR_KIND_RPR_DEFINE, "DEFINE");
+
+		foreach(l, restargets)
+		{
+			char		*n;
+
+			r = (ResTarget *) lfirst(l);
+			n = r->name;
+
+			if (!strcmp(n, name))
+				ereport(ERROR,
+						(errcode(ERRCODE_SYNTAX_ERROR),
+						 errmsg("row pattern definition variable name \"%s\" appears more than once in DEFINE clause",
+								name),
+						 parser_errposition(pstate, exprLocation((Node *)r))));
+		}
+		restargets = lappend(restargets, restarget);
+	}
+	list_free(restargets);
+
+	/*
+	 * Create list of row pattern DEFINE variable name's initial.
+	 * We assign [a-z] to them (up to 26 variable names are allowed).
+	 */
+	restargets = NIL;
+	i = 0;
+	initialLen = strlen(defineVariableInitials);
+
+	foreach(lc, windef->rpCommonSyntax->rpDefs)
+	{
+		char	initial[2];
+
+		restarget = (ResTarget *)lfirst(lc);
+		name = restarget->name;
+
+		if (i >= initialLen)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("number of row pattern definition variable names exceeds %d", initialLen),
+					 parser_errposition(pstate, exprLocation((Node *)restarget))));
+		}
+		initial[0] = defineVariableInitials[i++];
+		initial[1] = '\0';
+		wc->defineInitial = lappend(wc->defineInitial, makeString(pstrdup(initial)));
+	}
+
+	return transformTargetList(pstate, windef->rpCommonSyntax->rpDefs,
+							   EXPR_KIND_RPR_DEFINE);
+}
+
+/*
+ * transformPatternClause
+ *		Process PATTERN clause and return PATTERN clause in the raw parse tree
+ */
+static void
+transformPatternClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	ListCell	*lc, *l;
+
+	/*
+	 * Row Pattern Common Syntax clause exists?
+	 */
+	if (windef->rpCommonSyntax == NULL)
+		return;
+
+	/*
+	 * Primary row pattern variable names in PATTERN clause must appear in
+	 * DEFINE clause as row pattern definition variable names.
+	 */
+	wc->patternVariable = NIL;
+	wc->patternRegexp = NIL;
+	foreach(lc, windef->rpCommonSyntax->rpPatterns)
+	{
+		A_Expr	*a;
+		char	*name;
+		char	*regexp;
+		bool	found = false;
+
+		if (!IsA(lfirst(lc), A_Expr))
+			ereport(ERROR,
+					errmsg("node type is not A_Expr"));
+
+		a = (A_Expr *)lfirst(lc);
+		name = strVal(a->lexpr);
+
+		foreach(l, windef->rpCommonSyntax->rpDefs)
+		{
+			ResTarget	*restarget = (ResTarget *)lfirst(l);
+
+			if (!strcmp(restarget->name, name))
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+		{
+			ereport(ERROR,
+					(errcode(ERRCODE_SYNTAX_ERROR),
+					 errmsg("primary row pattern variable name \"%s\" does not appear in DEFINE clause",
+							name),
+					 parser_errposition(pstate, exprLocation((Node *)a))));
+		}
+		wc->patternVariable = lappend(wc->patternVariable, makeString(pstrdup(name)));
+		regexp = strVal(lfirst(list_head(a->name)));
+		wc->patternRegexp = lappend(wc->patternRegexp, makeString(pstrdup(regexp)));
+	}
+}
+
+/*
+ * transformMeasureClause
+ *		Process MEASURE clause
+ *	XXX MEASURE clause is not supported yet
+ */
+static List *
+transformMeasureClause(ParseState *pstate, WindowClause *wc, WindowDef *windef)
+{
+	if (windef->rowPatternMeasures == NIL)
+		return NIL;
+
+	ereport(ERROR,
+			(errcode(ERRCODE_SYNTAX_ERROR),
+			 errmsg("%s","MEASURE clause is not supported yet"),
+			 parser_errposition(pstate, exprLocation((Node *)windef->rowPatternMeasures))));
+}
diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c
index 64c582c344..18b58ac263 100644
--- a/src/backend/parser/parse_expr.c
+++ b/src/backend/parser/parse_expr.c
@@ -557,6 +557,7 @@ transformColumnRef(ParseState *pstate, ColumnRef *cref)
 		case EXPR_KIND_COPY_WHERE:
 		case EXPR_KIND_GENERATED_COLUMN:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 
@@ -1770,6 +1771,7 @@ transformSubLink(ParseState *pstate, SubLink *sublink)
 		case EXPR_KIND_VALUES:
 		case EXPR_KIND_VALUES_SINGLE:
 		case EXPR_KIND_CYCLE_MARK:
+		case EXPR_KIND_RPR_DEFINE:
 			/* okay */
 			break;
 		case EXPR_KIND_CHECK_CONSTRAINT:
@@ -3149,6 +3151,8 @@ ParseExprKindName(ParseExprKind exprKind)
 			return "GENERATED AS";
 		case EXPR_KIND_CYCLE_MARK:
 			return "CYCLE";
+		case EXPR_KIND_RPR_DEFINE:
+			return "DEFINE";
 
 			/*
 			 * There is intentionally no default: case here, so that the
diff --git a/src/backend/parser/parse_func.c b/src/backend/parser/parse_func.c
index b3f0b6a137..2ff3699538 100644
--- a/src/backend/parser/parse_func.c
+++ b/src/backend/parser/parse_func.c
@@ -2656,6 +2656,9 @@ check_srf_call_placement(ParseState *pstate, Node *last_srf, int location)
 		case EXPR_KIND_CYCLE_MARK:
 			errkind = true;
 			break;
+		case EXPR_KIND_RPR_DEFINE:
+			errkind = true;
+			break;
 
 			/*
 			 * There is intentionally no default: case here, so that the
-- 
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="v5-0003-Row-pattern-recognition-patch-planner.patch"



^ permalink  raw  reply  [nested|flat] 124+ messages in thread