agora inbox for [email protected]  
help / color / mirror / Atom feed
[PATCH v33 1/2] Unique key
11+ messages / 3 participants
[nested] [flat]

* [PATCH v33 1/2] Unique key
@ 2019-07-09 10:44  jesperpedersen <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: jesperpedersen @ 2019-07-09 10:44 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   3 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/indxpath.c  |  41 +++++++
 src/backend/optimizer/path/pathkeys.c  |  71 ++++++++++--
 src/backend/optimizer/path/uniquekey.c | 147 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planagg.c   |   1 +
 src/backend/optimizer/plan/planmain.c  |   1 +
 src/backend/optimizer/plan/planner.c   |  17 ++-
 src/backend/optimizer/util/pathnode.c  |  12 ++
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  18 +++
 src/include/nodes/print.h              |   1 +
 src/include/optimizer/pathnode.h       |   1 +
 src/include/optimizer/paths.h          |  11 ++
 16 files changed, 373 insertions(+), 13 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index d76fae44b8..16083e7a7e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1723,6 +1723,7 @@ _outPathInfo(StringInfo str, const Path *node)
 	WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 	WRITE_FLOAT_FIELD(total_cost, "%.2f");
 	WRITE_NODE_FIELD(pathkeys);
+	WRITE_NODE_FIELD(uniquekeys);
 }
 
 /*
@@ -2205,6 +2206,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(eq_classes);
 	WRITE_BOOL_FIELD(ec_merging_done);
 	WRITE_NODE_FIELD(canon_pathkeys);
+	WRITE_NODE_FIELD(canon_uniquekeys);
 	WRITE_NODE_FIELD(left_join_clauses);
 	WRITE_NODE_FIELD(right_join_clauses);
 	WRITE_NODE_FIELD(full_join_clauses);
@@ -2214,6 +2216,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(placeholder_list);
 	WRITE_NODE_FIELD(fkey_list);
 	WRITE_NODE_FIELD(query_pathkeys);
+	WRITE_NODE_FIELD(query_uniquekeys);
 	WRITE_NODE_FIELD(group_pathkeys);
 	WRITE_NODE_FIELD(window_pathkeys);
 	WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2401,6 +2404,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 	WRITE_BOOL_FIELD(pk_nulls_first);
 }
 
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+	WRITE_NODE_TYPE("UNIQUEKEY");
+
+	WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4092,6 +4103,9 @@ outNode(StringInfo str, const void *obj)
 			case T_PathKey:
 				_outPathKey(str, obj);
 				break;
+			case T_UniqueKey:
+				_outUniqueKey(str, obj);
+				break;
 			case T_PathTarget:
 				_outPathTarget(str, obj);
 				break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 42476724d8..d286b34544 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtable)
 	printf(")\n");
 }
 
+/*
+ * print_uniquekeys -
+ *	  uniquekeys list of UniqueKeys
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+	ListCell   *l;
+
+	printf("(");
+	foreach(l, uniquekeys)
+	{
+		UniqueKey *unique_key = (UniqueKey *) lfirst(l);
+		EquivalenceClass *eclass = (EquivalenceClass *) unique_key->eq_clause;
+		ListCell   *k;
+		bool		first = true;
+
+		/* chase up */
+		while (eclass->ec_merged)
+			eclass = eclass->ec_merged;
+
+		printf("(");
+		foreach(k, eclass->ec_members)
+		{
+			EquivalenceMember *mem = (EquivalenceMember *) lfirst(k);
+
+			if (first)
+				first = false;
+			else
+				printf(", ");
+			print_expr((Node *) mem->em_expr, rtable);
+		}
+		printf(")");
+		if (lnext(uniquekeys, l))
+			printf(", ");
+	}
+	printf(")\n");
+}
+
 /*
  * print_tl
  *	  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/path/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS = \
 	joinpath.o \
 	joinrels.o \
 	pathkeys.o \
-	tidpath.o
+	tidpath.o \
+	uniquekey.o
 
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 8286d9cf34..bbc13e6141 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent)
 		print_pathkeys(path->pathkeys, root->parse->rtable);
 	}
 
+	if (path->uniquekeys)
+	{
+		for (i = 0; i < indent; i++)
+			printf("\t");
+		printf("  uniquekeys: ");
+		print_uniquekeys(path->uniquekeys, root->parse->rtable);
+	}
+
 	if (join)
 	{
 		JoinPath   *jp = (JoinPath *) path;
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 2a50272da6..bd1ea53e5c 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 									   EquivalenceClass *ec, EquivalenceMember *em,
 									   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
 
 
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	List	   *orderbyclausecols;
 	List	   *index_pathkeys;
 	List	   *useful_pathkeys;
+	List	   *useful_uniquekeys = NIL;
 	bool		found_lower_saop_clause;
 	bool		pathkeys_possibly_useful;
 	bool		index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	if (index_clauses != NIL || useful_pathkeys != NIL || useful_predicate ||
 		index_only_scan)
 	{
+		if (has_useful_uniquekeys(root))
+			useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 		ipath = create_index_path(root, index,
 								  index_clauses,
 								  orderbyclauses,
 								  orderbyclausecols,
 								  useful_pathkeys,
+								  useful_uniquekeys,
 								  index_is_ordered ?
 								  ForwardScanDirection :
 								  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 									  orderbyclauses,
 									  orderbyclausecols,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  index_is_ordered ?
 									  ForwardScanDirection :
 									  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 													index_pathkeys);
 		if (useful_pathkeys != NIL)
 		{
+			if (has_useful_uniquekeys(root))
+				useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 			ipath = create_index_path(root, index,
 									  index_clauses,
 									  NIL,
 									  NIL,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  BackwardScanDirection,
 									  index_only_scan,
 									  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 										  NIL,
 										  NIL,
 										  useful_pathkeys,
+										  useful_uniquekeys,
 										  BackwardScanDirection,
 										  index_only_scan,
 										  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 	return clause;
 }
 
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+	ListCell *lc;
+
+	if (pathkeys)
+	{
+		List *uniquekeys = NIL;
+		foreach(lc, pathkeys)
+		{
+			UniqueKey *unique_key;
+			PathKey *pk = (PathKey *) lfirst(lc);
+			EquivalenceClass *ec = (EquivalenceClass *) pk->pk_eclass;
+
+			unique_key = makeNode(UniqueKey);
+			unique_key->eq_clause = ec;
+
+			lappend(uniquekeys, unique_key);
+		}
+
+		if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+			return uniquekeys;
+	}
+
+	return NIL;
+}
 
 /****************************************************************************
  *				----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS	----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/path/pathkeys.c
index 71b9d42c99..054df9a617 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
 
 
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 											 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 	return pk;
 }
 
+/*
+ * pathkey_is_unique
+ *	   Checks if the new pathkey's equivalence class is the same as that of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
+	ListCell   *lc;
+
+	/* If same EC already is already in the list, then not unique */
+	foreach(lc, pathkeys)
+	{
+		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
+
+		if (new_ec == old_pathkey->pk_eclass)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * pathkey_is_redundant
  *	   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
-	ListCell   *lc;
 
 	/* Check for EC containing a constant --- unconditionally redundant */
 	if (EC_MUST_BE_REDUNDANT(new_ec))
 		return true;
 
-	/* If same EC already used in list, then redundant */
-	foreach(lc, pathkeys)
-	{
-		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
-
-		if (new_ec == old_pathkey->pk_eclass)
-			return true;
-	}
-
-	return false;
+	return !pathkey_is_unique(new_pathkey, pathkeys);
 }
 
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 	return pathkeys;
 }
 
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *		Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+							 List *sortclauses,
+							 List *tlist)
+{
+	List	   *pathkeys = NIL;
+	ListCell   *l;
+
+	foreach(l, sortclauses)
+	{
+		SortGroupClause *sortcl = (SortGroupClause *) lfirst(l);
+		Expr	   *sortkey;
+		PathKey    *pathkey;
+
+		sortkey = (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+		Assert(OidIsValid(sortcl->sortop));
+		pathkey = make_pathkey_from_sortop(root,
+										   sortkey,
+										   root->nullable_baserels,
+										   sortcl->sortop,
+										   sortcl->nulls_first,
+										   sortcl->tleSortGroupRef,
+										   true);
+
+		if (pathkey_is_unique(pathkey, pathkeys))
+			pathkeys = lappend(pathkeys, pathkey);
+	}
+
+	return pathkeys;
+}
+
 /****************************************************************************
  *		PATHKEYS AND MERGECLAUSES
  ****************************************************************************/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer/path/uniquekey.c
new file mode 100644
index 0000000000..13d4ebb98c
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,147 @@
+/*-------------------------------------------------------------------------
+ *
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  src/backend/optimizer/path/uniquekey.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceClass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+	List *result = NIL;
+	List *sortkeys;
+	ListCell *l;
+
+	sortkeys = make_pathkeys_for_uniquekeys(root,
+											sortclauses,
+											root->processed_tlist);
+
+	/* Create a uniquekey and add it to the list */
+	foreach(l, sortkeys)
+	{
+		PathKey    *pathkey = (PathKey *) lfirst(l);
+		EquivalenceClass *ec = pathkey->pk_eclass;
+		UniqueKey *unique_key = make_canonical_uniquekey(root, ec);
+
+		result = lappend(result, unique_key);
+	}
+
+	return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *	  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+	ListCell   *key1,
+			   *key2;
+
+	/*
+	 * Fall out quickly if we are passed two identical lists.  This mostly
+	 * catches the case where both are NIL, but that's common enough to
+	 * warrant the test.
+	 */
+	if (keys1 == keys2)
+		return true;
+
+	foreach(key2, keys2)
+	{
+		bool found = false;
+		UniqueKey  *uniquekey2 = (UniqueKey *) lfirst(key2);
+
+		foreach(key1, keys1)
+		{
+			UniqueKey  *uniquekey1 = (UniqueKey *) lfirst(key1);
+
+			if (uniquekey1->eq_clause == uniquekey2->eq_clause)
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *		Detect whether the planner could have any uniquekeys that are
+ *		useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+	if (root->query_uniquekeys != NIL)
+		return true;	/* there are some */
+	return false;		/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *	  Given the parameters for a UniqueKey, find any pre-existing matching
+ *	  uniquekey in the query's list of "canonical" uniquekeys.  Make a new
+ *	  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instead,
+ * equivclass.c will complain if a merge occurs after root->canon_uniquekeys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+						 EquivalenceClass *eclass)
+{
+	UniqueKey  *uk;
+	ListCell   *lc;
+	MemoryContext oldcontext;
+
+	/* The passed eclass might be non-canonical, so chase up to the top */
+	while (eclass->ec_merged)
+		eclass = eclass->ec_merged;
+
+	foreach(lc, root->canon_uniquekeys)
+	{
+		uk = (UniqueKey *) lfirst(lc);
+		if (eclass == uk->eq_clause)
+			return uk;
+	}
+
+	/*
+	 * Be sure canonical uniquekeys are allocated in the main planning context.
+	 * Not an issue in normal planning, but it is for GEQO.
+	 */
+	oldcontext = MemoryContextSwitchTo(root->planner_cxt);
+
+	uk = makeNode(UniqueKey);
+	uk->eq_clause = eclass;
+
+	root->canon_uniquekeys = lappend(root->canon_uniquekeys, uk);
+
+	MemoryContextSwitchTo(oldcontext);
+
+	return uk;
+}
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/plan/planagg.c
index 8634940efc..dd64775d8f 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -511,6 +511,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 									  root->parse->targetList);
 
 	root->query_pathkeys = root->sort_pathkeys;
+	root->query_uniquekeys = NIL;
 }
 
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/plan/planmain.c
index 62dfc6d44a..3a372af91b 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 	root->join_rel_level = NULL;
 	root->join_cur_level = 0;
 	root->canon_pathkeys = NIL;
+	root->canon_uniquekeys = NIL;
 	root->left_join_clauses = NIL;
 	root->right_join_clauses = NIL;
 	root->full_join_clauses = NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index d6f2153593..984fca0696 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3657,15 +3657,30 @@ standard_qp_callback(PlannerInfo *root, void *extra)
 	 * much easier, since we know that the parser ensured that one is a
 	 * superset of the other.
 	 */
+	root->query_uniquekeys = NIL;
+
 	if (root->group_pathkeys)
+	{
 		root->query_pathkeys = root->group_pathkeys;
+
+		if (!root->parse->hasAggs)
+			root->query_uniquekeys = build_uniquekeys(root, qp_extra->groupClause);
+	}
 	else if (root->window_pathkeys)
 		root->query_pathkeys = root->window_pathkeys;
 	else if (list_length(root->distinct_pathkeys) >
 			 list_length(root->sort_pathkeys))
+	{
 		root->query_pathkeys = root->distinct_pathkeys;
+		root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else if (root->sort_pathkeys)
+	{
 		root->query_pathkeys = root->sort_pathkeys;
+
+		if (root->distinct_pathkeys)
+			root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else
 		root->query_pathkeys = NIL;
 }
@@ -6222,7 +6237,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
 
 	/* Estimate the cost of index scan */
 	indexScanPath = create_index_path(root, indexInfo,
-									  NIL, NIL, NIL, NIL,
+									  NIL, NIL, NIL, NIL, NIL,
 									  ForwardScanDirection, false,
 									  NULL, 1.0, false);
 
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index e6d08aede5..a006dbbe9c 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -940,6 +940,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = parallel_workers;
 	pathnode->pathkeys = NIL;	/* seqscan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_seqscan(pathnode, root, rel, pathnode->param_info);
 
@@ -964,6 +965,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* samplescan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_samplescan(pathnode, root, rel, pathnode->param_info);
 
@@ -1000,6 +1002,7 @@ create_index_path(PlannerInfo *root,
 				  List *indexorderbys,
 				  List *indexorderbycols,
 				  List *pathkeys,
+				  List *uniquekeys,
 				  ScanDirection indexscandir,
 				  bool indexonly,
 				  Relids required_outer,
@@ -1018,6 +1021,7 @@ create_index_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = 0;
 	pathnode->path.pathkeys = pathkeys;
+	pathnode->path.uniquekeys = uniquekeys;
 
 	pathnode->indexinfo = index;
 	pathnode->indexclauses = indexclauses;
@@ -1061,6 +1065,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = parallel_degree;
 	pathnode->path.pathkeys = NIL;	/* always unordered */
+	pathnode->path.uniquekeys = NIL;
 
 	pathnode->bitmapqual = bitmapqual;
 
@@ -1922,6 +1927,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = pathkeys;
+	pathnode->uniquekeys = NIL;
 
 	cost_functionscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1948,6 +1954,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1974,6 +1981,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_valuesscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1999,6 +2007,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* XXX for now, result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2025,6 +2034,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2051,6 +2061,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_resultscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2077,6 +2088,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	/* Cost is the same as for a regular CTE scan */
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index baced7eec0..a1511b46ea 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 	T_EquivalenceMember,
 	T_PathKey,
 	T_PathTarget,
+	T_UniqueKey,
 	T_RestrictInfo,
 	T_IndexClause,
 	T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 3d3be197e0..4e329f0fb5 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -269,6 +269,8 @@ struct PlannerInfo
 
 	List	   *canon_pathkeys; /* list of "canonical" PathKeys */
 
+	List	   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 	List	   *left_join_clauses;	/* list of RestrictInfos for mergejoinable
 									 * outer join clauses w/nonnullable var on
 									 * left */
@@ -297,6 +299,8 @@ struct PlannerInfo
 
 	List	   *query_pathkeys; /* desired pathkeys for query_planner() */
 
+	List	   *query_uniquekeys; /* unique keys used for the query */
+
 	List	   *group_pathkeys; /* groupClause pathkeys, if any */
 	List	   *window_pathkeys;	/* pathkeys of bottom window, if any */
 	List	   *distinct_pathkeys;	/* distinctClause pathkeys, if any */
@@ -1077,6 +1081,15 @@ typedef struct ParamPathInfo
 	List	   *ppi_clauses;	/* join clauses available from outer rels */
 } ParamPathInfo;
 
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+	NodeTag		type;
+
+	EquivalenceClass *eq_clause;	/* equivalence class */
+} UniqueKey;
 
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some other
@@ -1106,6 +1119,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1129,6 +1145,8 @@ typedef struct Path
 
 	List	   *pathkeys;		/* sort ordering of path's output */
 	/* pathkeys is a List of PathKey nodes; see above */
+
+	List	   *uniquekeys;	/* the unique keys, or NIL if none */
 } Path;
 
 /* Macro for extracting a path's parameterization relids; beware double eval */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index 6126b491bf..006248bfb5 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
 
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index e450fe112a..f75ff6f323 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -44,6 +44,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 									List *indexorderbys,
 									List *indexorderbycols,
 									List *pathkeys,
+									List *uniquekeys,
 									ScanDirection indexscandir,
 									bool indexonly,
 									Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index 9ab73bd20c..5b6be383b3 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 										   List *sortclauses,
 										   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+										  List *sortclauses,
+										  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 											RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *root,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 									List *live_childrels);
 
+/*
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif							/* PATHS_H */
-- 
2.21.0


--ezqb2fwnenb5aigz
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v33-0002-Index-skip-scan.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH] Unique key
@ 2019-07-09 10:44  jesperpedersen <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: jesperpedersen @ 2019-07-09 10:44 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   3 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/indxpath.c  |  41 +++++++
 src/backend/optimizer/path/pathkeys.c  |  71 ++++++++++--
 src/backend/optimizer/path/uniquekey.c | 147 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planagg.c   |   1 +
 src/backend/optimizer/plan/planmain.c  |   1 +
 src/backend/optimizer/plan/planner.c   |  17 ++-
 src/backend/optimizer/util/pathnode.c  |  12 ++
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  18 +++
 src/include/nodes/print.h              |   1 +
 src/include/optimizer/pathnode.h       |   1 +
 src/include/optimizer/paths.h          |  11 ++
 16 files changed, 373 insertions(+), 13 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index b0dcd02ff6..1ccd68d3aa 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1720,6 +1720,7 @@ _outPathInfo(StringInfo str, const Path *node)
 =09WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 =09WRITE_FLOAT_FIELD(total_cost, "%.2f");
 =09WRITE_NODE_FIELD(pathkeys);
+=09WRITE_NODE_FIELD(uniquekeys);
 }
=20
 /*
@@ -2201,6 +2202,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *no=
de)
 =09WRITE_NODE_FIELD(eq_classes);
 =09WRITE_BOOL_FIELD(ec_merging_done);
 =09WRITE_NODE_FIELD(canon_pathkeys);
+=09WRITE_NODE_FIELD(canon_uniquekeys);
 =09WRITE_NODE_FIELD(left_join_clauses);
 =09WRITE_NODE_FIELD(right_join_clauses);
 =09WRITE_NODE_FIELD(full_join_clauses);
@@ -2210,6 +2212,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *no=
de)
 =09WRITE_NODE_FIELD(placeholder_list);
 =09WRITE_NODE_FIELD(fkey_list);
 =09WRITE_NODE_FIELD(query_pathkeys);
+=09WRITE_NODE_FIELD(query_uniquekeys);
 =09WRITE_NODE_FIELD(group_pathkeys);
 =09WRITE_NODE_FIELD(window_pathkeys);
 =09WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2397,6 +2400,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 =09WRITE_BOOL_FIELD(pk_nulls_first);
 }
=20
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+=09WRITE_NODE_TYPE("UNIQUEKEY");
+
+=09WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4083,6 +4094,9 @@ outNode(StringInfo str, const void *obj)
 =09=09=09case T_PathKey:
 =09=09=09=09_outPathKey(str, obj);
 =09=09=09=09break;
+=09=09=09case T_UniqueKey:
+=09=09=09=09_outUniqueKey(str, obj);
+=09=09=09=09break;
 =09=09=09case T_PathTarget:
 =09=09=09=09_outPathTarget(str, obj);
 =09=09=09=09break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 4ecde6b421..435b32063c 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtabl=
e)
 =09printf(")\n");
 }
=20
+/*
+ * print_uniquekeys -
+ *=09  uniquekeys list of UniqueKeys
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+=09ListCell   *l;
+
+=09printf("(");
+=09foreach(l, uniquekeys)
+=09{
+=09=09UniqueKey *unique_key =3D (UniqueKey *) lfirst(l);
+=09=09EquivalenceClass *eclass =3D (EquivalenceClass *) unique_key->eq_cla=
use;
+=09=09ListCell   *k;
+=09=09bool=09=09first =3D true;
+
+=09=09/* chase up */
+=09=09while (eclass->ec_merged)
+=09=09=09eclass =3D eclass->ec_merged;
+
+=09=09printf("(");
+=09=09foreach(k, eclass->ec_members)
+=09=09{
+=09=09=09EquivalenceMember *mem =3D (EquivalenceMember *) lfirst(k);
+
+=09=09=09if (first)
+=09=09=09=09first =3D false;
+=09=09=09else
+=09=09=09=09printf(", ");
+=09=09=09print_expr((Node *) mem->em_expr, rtable);
+=09=09}
+=09=09printf(")");
+=09=09if (lnext(uniquekeys, l))
+=09=09=09printf(", ");
+=09}
+=09printf(")\n");
+}
+
 /*
  * print_tl
  *=09  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/pa=
th/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS =3D \
 =09joinpath.o \
 =09joinrels.o \
 =09pathkeys.o \
-=09tidpath.o
+=09tidpath.o \
+=09uniquekey.o
=20
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/=
path/allpaths.c
index db3a68a51d..5fc9b81746 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent=
)
 =09=09print_pathkeys(path->pathkeys, root->parse->rtable);
 =09}
=20
+=09if (path->uniquekeys)
+=09{
+=09=09for (i =3D 0; i < indent; i++)
+=09=09=09printf("\t");
+=09=09printf("  uniquekeys: ");
+=09=09print_uniquekeys(path->uniquekeys, root->parse->rtable);
+=09}
+
 =09if (join)
 =09{
 =09=09JoinPath   *jp =3D (JoinPath *) path;
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/=
path/indxpath.c
index 37b257cd0e..88c1dd0f59 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *=
index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 =09=09=09=09=09=09=09=09=09   EquivalenceClass *ec, EquivalenceMember *em,
 =09=09=09=09=09=09=09=09=09   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
=20
=20
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 =09List=09   *orderbyclausecols;
 =09List=09   *index_pathkeys;
 =09List=09   *useful_pathkeys;
+=09List=09   *useful_uniquekeys =3D NIL;
 =09bool=09=09found_lower_saop_clause;
 =09bool=09=09pathkeys_possibly_useful;
 =09bool=09=09index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *re=
l,
 =09if (index_clauses !=3D NIL || useful_pathkeys !=3D NIL || useful_predic=
ate ||
 =09=09index_only_scan)
 =09{
+=09=09if (has_useful_uniquekeys(root))
+=09=09=09useful_uniquekeys =3D get_uniquekeys_for_index(root, useful_pathk=
eys);
+
 =09=09ipath =3D create_index_path(root, index,
 =09=09=09=09=09=09=09=09  index_clauses,
 =09=09=09=09=09=09=09=09  orderbyclauses,
 =09=09=09=09=09=09=09=09  orderbyclausecols,
 =09=09=09=09=09=09=09=09  useful_pathkeys,
+=09=09=09=09=09=09=09=09  useful_uniquekeys,
 =09=09=09=09=09=09=09=09  index_is_ordered ?
 =09=09=09=09=09=09=09=09  ForwardScanDirection :
 =09=09=09=09=09=09=09=09  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 =09=09=09=09=09=09=09=09=09  orderbyclauses,
 =09=09=09=09=09=09=09=09=09  orderbyclausecols,
 =09=09=09=09=09=09=09=09=09  useful_pathkeys,
+=09=09=09=09=09=09=09=09=09  useful_uniquekeys,
 =09=09=09=09=09=09=09=09=09  index_is_ordered ?
 =09=09=09=09=09=09=09=09=09  ForwardScanDirection :
 =09=09=09=09=09=09=09=09=09  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *re=
l,
 =09=09=09=09=09=09=09=09=09=09=09=09=09index_pathkeys);
 =09=09if (useful_pathkeys !=3D NIL)
 =09=09{
+=09=09=09if (has_useful_uniquekeys(root))
+=09=09=09=09useful_uniquekeys =3D get_uniquekeys_for_index(root, useful_pa=
thkeys);
+
 =09=09=09ipath =3D create_index_path(root, index,
 =09=09=09=09=09=09=09=09=09  index_clauses,
 =09=09=09=09=09=09=09=09=09  NIL,
 =09=09=09=09=09=09=09=09=09  NIL,
 =09=09=09=09=09=09=09=09=09  useful_pathkeys,
+=09=09=09=09=09=09=09=09=09  useful_uniquekeys,
 =09=09=09=09=09=09=09=09=09  BackwardScanDirection,
 =09=09=09=09=09=09=09=09=09  index_only_scan,
 =09=09=09=09=09=09=09=09=09  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 =09=09=09=09=09=09=09=09=09=09  NIL,
 =09=09=09=09=09=09=09=09=09=09  NIL,
 =09=09=09=09=09=09=09=09=09=09  useful_pathkeys,
+=09=09=09=09=09=09=09=09=09=09  useful_uniquekeys,
 =09=09=09=09=09=09=09=09=09=09  BackwardScanDirection,
 =09=09=09=09=09=09=09=09=09=09  index_only_scan,
 =09=09=09=09=09=09=09=09=09=09  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 =09return clause;
 }
=20
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+=09ListCell *lc;
+
+=09if (pathkeys)
+=09{
+=09=09List *uniquekeys =3D NIL;
+=09=09foreach(lc, pathkeys)
+=09=09{
+=09=09=09UniqueKey *unique_key;
+=09=09=09PathKey *pk =3D (PathKey *) lfirst(lc);
+=09=09=09EquivalenceClass *ec =3D (EquivalenceClass *) pk->pk_eclass;
+
+=09=09=09unique_key =3D makeNode(UniqueKey);
+=09=09=09unique_key->eq_clause =3D ec;
+
+=09=09=09lappend(uniquekeys, unique_key);
+=09=09}
+
+=09=09if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+=09=09=09return uniquekeys;
+=09}
+
+=09return NIL;
+}
=20
 /*************************************************************************=
***
  *=09=09=09=09----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS=09----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/=
path/pathkeys.c
index 2f4fea241a..b0dc1bc22a 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
=20
=20
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 =09=09=09=09=09=09=09=09=09=09=09 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 =09return pk;
 }
=20
+/*
+ * pathkey_is_unique
+ *=09   Checks if the new pathkey's equivalence class is the same as that =
of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+=09EquivalenceClass *new_ec =3D new_pathkey->pk_eclass;
+=09ListCell   *lc;
+
+=09/* If same EC already is already in the list, then not unique */
+=09foreach(lc, pathkeys)
+=09{
+=09=09PathKey    *old_pathkey =3D (PathKey *) lfirst(lc);
+
+=09=09if (new_ec =3D=3D old_pathkey->pk_eclass)
+=09=09=09return false;
+=09}
+
+=09return true;
+}
+
 /*
  * pathkey_is_redundant
  *=09   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 =09EquivalenceClass *new_ec =3D new_pathkey->pk_eclass;
-=09ListCell   *lc;
=20
 =09/* Check for EC containing a constant --- unconditionally redundant */
 =09if (EC_MUST_BE_REDUNDANT(new_ec))
 =09=09return true;
=20
-=09/* If same EC already used in list, then redundant */
-=09foreach(lc, pathkeys)
-=09{
-=09=09PathKey    *old_pathkey =3D (PathKey *) lfirst(lc);
-
-=09=09if (new_ec =3D=3D old_pathkey->pk_eclass)
-=09=09=09return true;
-=09}
-
-=09return false;
+=09return !pathkey_is_unique(new_pathkey, pathkeys);
 }
=20
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 =09return pathkeys;
 }
=20
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *=09=09Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+=09=09=09=09=09=09=09 List *sortclauses,
+=09=09=09=09=09=09=09 List *tlist)
+{
+=09List=09   *pathkeys =3D NIL;
+=09ListCell   *l;
+
+=09foreach(l, sortclauses)
+=09{
+=09=09SortGroupClause *sortcl =3D (SortGroupClause *) lfirst(l);
+=09=09Expr=09   *sortkey;
+=09=09PathKey    *pathkey;
+
+=09=09sortkey =3D (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+=09=09Assert(OidIsValid(sortcl->sortop));
+=09=09pathkey =3D make_pathkey_from_sortop(root,
+=09=09=09=09=09=09=09=09=09=09   sortkey,
+=09=09=09=09=09=09=09=09=09=09   root->nullable_baserels,
+=09=09=09=09=09=09=09=09=09=09   sortcl->sortop,
+=09=09=09=09=09=09=09=09=09=09   sortcl->nulls_first,
+=09=09=09=09=09=09=09=09=09=09   sortcl->tleSortGroupRef,
+=09=09=09=09=09=09=09=09=09=09   true);
+
+=09=09if (pathkey_is_unique(pathkey, pathkeys))
+=09=09=09pathkeys =3D lappend(pathkeys, pathkey);
+=09}
+
+=09return pathkeys;
+}
+
 /*************************************************************************=
***
  *=09=09PATHKEYS AND MERGECLAUSES
  *************************************************************************=
***/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer=
/path/uniquekey.c
new file mode 100644
index 0000000000..13d4ebb98c
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,147 @@
+/*------------------------------------------------------------------------=
-
+ *
+ * uniquekey.c
+ *=09  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *=09  src/backend/optimizer/path/uniquekey.c
+ *
+ *------------------------------------------------------------------------=
-
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceC=
lass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+=09List *result =3D NIL;
+=09List *sortkeys;
+=09ListCell *l;
+
+=09sortkeys =3D make_pathkeys_for_uniquekeys(root,
+=09=09=09=09=09=09=09=09=09=09=09sortclauses,
+=09=09=09=09=09=09=09=09=09=09=09root->processed_tlist);
+
+=09/* Create a uniquekey and add it to the list */
+=09foreach(l, sortkeys)
+=09{
+=09=09PathKey    *pathkey =3D (PathKey *) lfirst(l);
+=09=09EquivalenceClass *ec =3D pathkey->pk_eclass;
+=09=09UniqueKey *unique_key =3D make_canonical_uniquekey(root, ec);
+
+=09=09result =3D lappend(result, unique_key);
+=09}
+
+=09return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *=09  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+=09ListCell   *key1,
+=09=09=09   *key2;
+
+=09/*
+=09 * Fall out quickly if we are passed two identical lists.  This mostly
+=09 * catches the case where both are NIL, but that's common enough to
+=09 * warrant the test.
+=09 */
+=09if (keys1 =3D=3D keys2)
+=09=09return true;
+
+=09foreach(key2, keys2)
+=09{
+=09=09bool found =3D false;
+=09=09UniqueKey  *uniquekey2 =3D (UniqueKey *) lfirst(key2);
+
+=09=09foreach(key1, keys1)
+=09=09{
+=09=09=09UniqueKey  *uniquekey1 =3D (UniqueKey *) lfirst(key1);
+
+=09=09=09if (uniquekey1->eq_clause =3D=3D uniquekey2->eq_clause)
+=09=09=09{
+=09=09=09=09found =3D true;
+=09=09=09=09break;
+=09=09=09}
+=09=09}
+
+=09=09if (!found)
+=09=09=09return false;
+=09}
+
+=09return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *=09=09Detect whether the planner could have any uniquekeys that are
+ *=09=09useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+=09if (root->query_uniquekeys !=3D NIL)
+=09=09return true;=09/* there are some */
+=09return false;=09=09/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *=09  Given the parameters for a UniqueKey, find any pre-existing matchin=
g
+ *=09  uniquekey in the query's list of "canonical" uniquekeys.  Make a ne=
w
+ *=09  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instea=
d,
+ * equivclass.c will complain if a merge occurs after root->canon_uniqueke=
ys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+=09=09=09=09=09=09 EquivalenceClass *eclass)
+{
+=09UniqueKey  *uk;
+=09ListCell   *lc;
+=09MemoryContext oldcontext;
+
+=09/* The passed eclass might be non-canonical, so chase up to the top */
+=09while (eclass->ec_merged)
+=09=09eclass =3D eclass->ec_merged;
+
+=09foreach(lc, root->canon_uniquekeys)
+=09{
+=09=09uk =3D (UniqueKey *) lfirst(lc);
+=09=09if (eclass =3D=3D uk->eq_clause)
+=09=09=09return uk;
+=09}
+
+=09/*
+=09 * Be sure canonical uniquekeys are allocated in the main planning cont=
ext.
+=09 * Not an issue in normal planning, but it is for GEQO.
+=09 */
+=09oldcontext =3D MemoryContextSwitchTo(root->planner_cxt);
+
+=09uk =3D makeNode(UniqueKey);
+=09uk->eq_clause =3D eclass;
+
+=09root->canon_uniquekeys =3D lappend(root->canon_uniquekeys, uk);
+
+=09MemoryContextSwitchTo(oldcontext);
+
+=09return uk;
+}
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/p=
lan/planagg.c
index 974f6204ca..f6144b2267 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -511,6 +511,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 =09=09=09=09=09=09=09=09=09  root->parse->targetList);
=20
 =09root->query_pathkeys =3D root->sort_pathkeys;
+=09root->query_uniquekeys =3D NIL;
 }
=20
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/=
plan/planmain.c
index f0c1b52a2e..3ccde03ab7 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 =09root->join_rel_level =3D NULL;
 =09root->join_cur_level =3D 0;
 =09root->canon_pathkeys =3D NIL;
+=09root->canon_uniquekeys =3D NIL;
 =09root->left_join_clauses =3D NIL;
 =09root->right_join_clauses =3D NIL;
 =09root->full_join_clauses =3D NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/p=
lan/planner.c
index 7fe11b59a0..8f03a20825 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3651,15 +3651,30 @@ standard_qp_callback(PlannerInfo *root, void *extra=
)
 =09 * much easier, since we know that the parser ensured that one is a
 =09 * superset of the other.
 =09 */
+=09root->query_uniquekeys =3D NIL;
+
 =09if (root->group_pathkeys)
+=09{
 =09=09root->query_pathkeys =3D root->group_pathkeys;
+
+=09=09if (!root->parse->hasAggs)
+=09=09=09root->query_uniquekeys =3D build_uniquekeys(root, qp_extra->group=
Clause);
+=09}
 =09else if (root->window_pathkeys)
 =09=09root->query_pathkeys =3D root->window_pathkeys;
 =09else if (list_length(root->distinct_pathkeys) >
 =09=09=09 list_length(root->sort_pathkeys))
+=09{
 =09=09root->query_pathkeys =3D root->distinct_pathkeys;
+=09=09root->query_uniquekeys =3D build_uniquekeys(root, parse->distinctCla=
use);
+=09}
 =09else if (root->sort_pathkeys)
+=09{
 =09=09root->query_pathkeys =3D root->sort_pathkeys;
+
+=09=09if (root->distinct_pathkeys)
+=09=09=09root->query_uniquekeys =3D build_uniquekeys(root, parse->distinct=
Clause);
+=09}
 =09else
 =09=09root->query_pathkeys =3D NIL;
 }
@@ -6216,7 +6231,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
=20
 =09/* Estimate the cost of index scan */
 =09indexScanPath =3D create_index_path(root, indexInfo,
-=09=09=09=09=09=09=09=09=09  NIL, NIL, NIL, NIL,
+=09=09=09=09=09=09=09=09=09  NIL, NIL, NIL, NIL, NIL,
 =09=09=09=09=09=09=09=09=09  ForwardScanDirection, false,
 =09=09=09=09=09=09=09=09=09  NULL, 1.0, false);
=20
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/=
util/pathnode.c
index 60c93ee7c5..ec02c468d0 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -940,6 +940,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D parallel_workers;
 =09pathnode->pathkeys =3D NIL;=09/* seqscan has unordered result */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_seqscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -964,6 +965,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *r=
el, Relids required_outer
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* samplescan has unordered result */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_samplescan(pathnode, root, rel, pathnode->param_info);
=20
@@ -1000,6 +1002,7 @@ create_index_path(PlannerInfo *root,
 =09=09=09=09  List *indexorderbys,
 =09=09=09=09  List *indexorderbycols,
 =09=09=09=09  List *pathkeys,
+=09=09=09=09  List *uniquekeys,
 =09=09=09=09  ScanDirection indexscandir,
 =09=09=09=09  bool indexonly,
 =09=09=09=09  Relids required_outer,
@@ -1018,6 +1021,7 @@ create_index_path(PlannerInfo *root,
 =09pathnode->path.parallel_safe =3D rel->consider_parallel;
 =09pathnode->path.parallel_workers =3D 0;
 =09pathnode->path.pathkeys =3D pathkeys;
+=09pathnode->path.uniquekeys =3D uniquekeys;
=20
 =09pathnode->indexinfo =3D index;
 =09pathnode->indexclauses =3D indexclauses;
@@ -1061,6 +1065,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 =09pathnode->path.parallel_safe =3D rel->consider_parallel;
 =09pathnode->path.parallel_workers =3D parallel_degree;
 =09pathnode->path.pathkeys =3D NIL;=09/* always unordered */
+=09pathnode->path.uniquekeys =3D NIL;
=20
 =09pathnode->bitmapqual =3D bitmapqual;
=20
@@ -1922,6 +1927,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInf=
o *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D pathkeys;
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_functionscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -1948,6 +1954,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptIn=
fo *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -1974,6 +1981,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo =
*rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_valuesscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -1999,6 +2007,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *re=
l, Relids required_outer)
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* XXX for now, result is always unordere=
d */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_ctescan(pathnode, root, rel, pathnode->param_info);
=20
@@ -2025,6 +2034,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, Re=
lOptInfo *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
=20
@@ -2051,6 +2061,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo =
*rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_resultscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -2077,6 +2088,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptIn=
fo *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09/* Cost is the same as for a regular CTE scan */
 =09cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index bce2d59b0d..cbb6ba2586 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 =09T_EquivalenceMember,
 =09T_PathKey,
 =09T_PathTarget,
+=09T_UniqueKey,
 =09T_RestrictInfo,
 =09T_IndexClause,
 =09T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 23a06d718e..10ece6c875 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -267,6 +267,8 @@ struct PlannerInfo
=20
 =09List=09   *canon_pathkeys; /* list of "canonical" PathKeys */
=20
+=09List=09   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 =09List=09   *left_join_clauses;=09/* list of RestrictInfos for mergejoina=
ble
 =09=09=09=09=09=09=09=09=09 * outer join clauses w/nonnullable var on
 =09=09=09=09=09=09=09=09=09 * left */
@@ -295,6 +297,8 @@ struct PlannerInfo
=20
 =09List=09   *query_pathkeys; /* desired pathkeys for query_planner() */
=20
+=09List=09   *query_uniquekeys; /* unique keys used for the query */
+
 =09List=09   *group_pathkeys; /* groupClause pathkeys, if any */
 =09List=09   *window_pathkeys;=09/* pathkeys of bottom window, if any */
 =09List=09   *distinct_pathkeys;=09/* distinctClause pathkeys, if any */
@@ -1075,6 +1079,15 @@ typedef struct ParamPathInfo
 =09List=09   *ppi_clauses;=09/* join clauses available from outer rels */
 } ParamPathInfo;
=20
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+=09NodeTag=09=09type;
+
+=09EquivalenceClass *eq_clause;=09/* equivalence class */
+} UniqueKey;
=20
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some ot=
her
@@ -1104,6 +1117,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1127,6 +1143,8 @@ typedef struct Path
=20
 =09List=09   *pathkeys;=09=09/* sort ordering of path's output */
 =09/* pathkeys is a List of PathKey nodes; see above */
+
+=09List=09   *uniquekeys;=09/* the unique keys, or NIL if none */
 } Path;
=20
 /* Macro for extracting a path's parameterization relids; beware double ev=
al */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index cbff56a724..f1a7112877 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
=20
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathn=
ode.h
index a12af54971..37a946f857 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -44,6 +44,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 =09=09=09=09=09=09=09=09=09List *indexorderbys,
 =09=09=09=09=09=09=09=09=09List *indexorderbycols,
 =09=09=09=09=09=09=09=09=09List *pathkeys,
+=09=09=09=09=09=09=09=09=09List *uniquekeys,
 =09=09=09=09=09=09=09=09=09ScanDirection indexscandir,
 =09=09=09=09=09=09=09=09=09bool indexonly,
 =09=09=09=09=09=09=09=09=09Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index c6c34630c2..c79e47eeaf 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 =09=09=09=09=09=09=09=09=09=09   List *sortclauses,
 =09=09=09=09=09=09=09=09=09=09   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+=09=09=09=09=09=09=09=09=09=09  List *sortclauses,
+=09=09=09=09=09=09=09=09=09=09  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 =09=09=09=09=09=09=09=09=09=09=09RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *ro=
ot,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 =09=09=09=09=09=09=09=09=09List *live_childrels);
=20
+/*
+ * uniquekey.c
+ *=09  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif=09=09=09=09=09=09=09/* PATHS_H */
--=20
2.21.0


--------------51513615E9A6AD19C4BF8178
Content-Type: text/x-patch; charset=UTF-8;
 name="v30_0002-Index-skip-scan.patch"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
 filename="v30_0002-Index-skip-scan.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH v31 1/2] Unique key
@ 2019-07-09 10:44  jesperpedersen <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: jesperpedersen @ 2019-07-09 10:44 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   3 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/indxpath.c  |  41 +++++++
 src/backend/optimizer/path/pathkeys.c  |  71 ++++++++++--
 src/backend/optimizer/path/uniquekey.c | 147 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planagg.c   |   1 +
 src/backend/optimizer/plan/planmain.c  |   1 +
 src/backend/optimizer/plan/planner.c   |  17 ++-
 src/backend/optimizer/util/pathnode.c  |  12 ++
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  18 +++
 src/include/nodes/print.h              |   1 +
 src/include/optimizer/pathnode.h       |   1 +
 src/include/optimizer/paths.h          |  11 ++
 16 files changed, 373 insertions(+), 13 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index d76fae44b8..16083e7a7e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1723,6 +1723,7 @@ _outPathInfo(StringInfo str, const Path *node)
 	WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 	WRITE_FLOAT_FIELD(total_cost, "%.2f");
 	WRITE_NODE_FIELD(pathkeys);
+	WRITE_NODE_FIELD(uniquekeys);
 }
 
 /*
@@ -2205,6 +2206,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(eq_classes);
 	WRITE_BOOL_FIELD(ec_merging_done);
 	WRITE_NODE_FIELD(canon_pathkeys);
+	WRITE_NODE_FIELD(canon_uniquekeys);
 	WRITE_NODE_FIELD(left_join_clauses);
 	WRITE_NODE_FIELD(right_join_clauses);
 	WRITE_NODE_FIELD(full_join_clauses);
@@ -2214,6 +2216,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(placeholder_list);
 	WRITE_NODE_FIELD(fkey_list);
 	WRITE_NODE_FIELD(query_pathkeys);
+	WRITE_NODE_FIELD(query_uniquekeys);
 	WRITE_NODE_FIELD(group_pathkeys);
 	WRITE_NODE_FIELD(window_pathkeys);
 	WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2401,6 +2404,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 	WRITE_BOOL_FIELD(pk_nulls_first);
 }
 
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+	WRITE_NODE_TYPE("UNIQUEKEY");
+
+	WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4092,6 +4103,9 @@ outNode(StringInfo str, const void *obj)
 			case T_PathKey:
 				_outPathKey(str, obj);
 				break;
+			case T_UniqueKey:
+				_outUniqueKey(str, obj);
+				break;
 			case T_PathTarget:
 				_outPathTarget(str, obj);
 				break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 42476724d8..d286b34544 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtable)
 	printf(")\n");
 }
 
+/*
+ * print_uniquekeys -
+ *	  uniquekeys list of UniqueKeys
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+	ListCell   *l;
+
+	printf("(");
+	foreach(l, uniquekeys)
+	{
+		UniqueKey *unique_key = (UniqueKey *) lfirst(l);
+		EquivalenceClass *eclass = (EquivalenceClass *) unique_key->eq_clause;
+		ListCell   *k;
+		bool		first = true;
+
+		/* chase up */
+		while (eclass->ec_merged)
+			eclass = eclass->ec_merged;
+
+		printf("(");
+		foreach(k, eclass->ec_members)
+		{
+			EquivalenceMember *mem = (EquivalenceMember *) lfirst(k);
+
+			if (first)
+				first = false;
+			else
+				printf(", ");
+			print_expr((Node *) mem->em_expr, rtable);
+		}
+		printf(")");
+		if (lnext(uniquekeys, l))
+			printf(", ");
+	}
+	printf(")\n");
+}
+
 /*
  * print_tl
  *	  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/path/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS = \
 	joinpath.o \
 	joinrels.o \
 	pathkeys.o \
-	tidpath.o
+	tidpath.o \
+	uniquekey.o
 
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 8286d9cf34..bbc13e6141 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent)
 		print_pathkeys(path->pathkeys, root->parse->rtable);
 	}
 
+	if (path->uniquekeys)
+	{
+		for (i = 0; i < indent; i++)
+			printf("\t");
+		printf("  uniquekeys: ");
+		print_uniquekeys(path->uniquekeys, root->parse->rtable);
+	}
+
 	if (join)
 	{
 		JoinPath   *jp = (JoinPath *) path;
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 2a50272da6..bd1ea53e5c 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 									   EquivalenceClass *ec, EquivalenceMember *em,
 									   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
 
 
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	List	   *orderbyclausecols;
 	List	   *index_pathkeys;
 	List	   *useful_pathkeys;
+	List	   *useful_uniquekeys = NIL;
 	bool		found_lower_saop_clause;
 	bool		pathkeys_possibly_useful;
 	bool		index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	if (index_clauses != NIL || useful_pathkeys != NIL || useful_predicate ||
 		index_only_scan)
 	{
+		if (has_useful_uniquekeys(root))
+			useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 		ipath = create_index_path(root, index,
 								  index_clauses,
 								  orderbyclauses,
 								  orderbyclausecols,
 								  useful_pathkeys,
+								  useful_uniquekeys,
 								  index_is_ordered ?
 								  ForwardScanDirection :
 								  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 									  orderbyclauses,
 									  orderbyclausecols,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  index_is_ordered ?
 									  ForwardScanDirection :
 									  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 													index_pathkeys);
 		if (useful_pathkeys != NIL)
 		{
+			if (has_useful_uniquekeys(root))
+				useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 			ipath = create_index_path(root, index,
 									  index_clauses,
 									  NIL,
 									  NIL,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  BackwardScanDirection,
 									  index_only_scan,
 									  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 										  NIL,
 										  NIL,
 										  useful_pathkeys,
+										  useful_uniquekeys,
 										  BackwardScanDirection,
 										  index_only_scan,
 										  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 	return clause;
 }
 
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+	ListCell *lc;
+
+	if (pathkeys)
+	{
+		List *uniquekeys = NIL;
+		foreach(lc, pathkeys)
+		{
+			UniqueKey *unique_key;
+			PathKey *pk = (PathKey *) lfirst(lc);
+			EquivalenceClass *ec = (EquivalenceClass *) pk->pk_eclass;
+
+			unique_key = makeNode(UniqueKey);
+			unique_key->eq_clause = ec;
+
+			lappend(uniquekeys, unique_key);
+		}
+
+		if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+			return uniquekeys;
+	}
+
+	return NIL;
+}
 
 /****************************************************************************
  *				----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS	----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/path/pathkeys.c
index 71b9d42c99..054df9a617 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
 
 
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 											 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 	return pk;
 }
 
+/*
+ * pathkey_is_unique
+ *	   Checks if the new pathkey's equivalence class is the same as that of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
+	ListCell   *lc;
+
+	/* If same EC already is already in the list, then not unique */
+	foreach(lc, pathkeys)
+	{
+		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
+
+		if (new_ec == old_pathkey->pk_eclass)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * pathkey_is_redundant
  *	   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
-	ListCell   *lc;
 
 	/* Check for EC containing a constant --- unconditionally redundant */
 	if (EC_MUST_BE_REDUNDANT(new_ec))
 		return true;
 
-	/* If same EC already used in list, then redundant */
-	foreach(lc, pathkeys)
-	{
-		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
-
-		if (new_ec == old_pathkey->pk_eclass)
-			return true;
-	}
-
-	return false;
+	return !pathkey_is_unique(new_pathkey, pathkeys);
 }
 
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 	return pathkeys;
 }
 
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *		Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+							 List *sortclauses,
+							 List *tlist)
+{
+	List	   *pathkeys = NIL;
+	ListCell   *l;
+
+	foreach(l, sortclauses)
+	{
+		SortGroupClause *sortcl = (SortGroupClause *) lfirst(l);
+		Expr	   *sortkey;
+		PathKey    *pathkey;
+
+		sortkey = (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+		Assert(OidIsValid(sortcl->sortop));
+		pathkey = make_pathkey_from_sortop(root,
+										   sortkey,
+										   root->nullable_baserels,
+										   sortcl->sortop,
+										   sortcl->nulls_first,
+										   sortcl->tleSortGroupRef,
+										   true);
+
+		if (pathkey_is_unique(pathkey, pathkeys))
+			pathkeys = lappend(pathkeys, pathkey);
+	}
+
+	return pathkeys;
+}
+
 /****************************************************************************
  *		PATHKEYS AND MERGECLAUSES
  ****************************************************************************/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer/path/uniquekey.c
new file mode 100644
index 0000000000..13d4ebb98c
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,147 @@
+/*-------------------------------------------------------------------------
+ *
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  src/backend/optimizer/path/uniquekey.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceClass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+	List *result = NIL;
+	List *sortkeys;
+	ListCell *l;
+
+	sortkeys = make_pathkeys_for_uniquekeys(root,
+											sortclauses,
+											root->processed_tlist);
+
+	/* Create a uniquekey and add it to the list */
+	foreach(l, sortkeys)
+	{
+		PathKey    *pathkey = (PathKey *) lfirst(l);
+		EquivalenceClass *ec = pathkey->pk_eclass;
+		UniqueKey *unique_key = make_canonical_uniquekey(root, ec);
+
+		result = lappend(result, unique_key);
+	}
+
+	return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *	  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+	ListCell   *key1,
+			   *key2;
+
+	/*
+	 * Fall out quickly if we are passed two identical lists.  This mostly
+	 * catches the case where both are NIL, but that's common enough to
+	 * warrant the test.
+	 */
+	if (keys1 == keys2)
+		return true;
+
+	foreach(key2, keys2)
+	{
+		bool found = false;
+		UniqueKey  *uniquekey2 = (UniqueKey *) lfirst(key2);
+
+		foreach(key1, keys1)
+		{
+			UniqueKey  *uniquekey1 = (UniqueKey *) lfirst(key1);
+
+			if (uniquekey1->eq_clause == uniquekey2->eq_clause)
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *		Detect whether the planner could have any uniquekeys that are
+ *		useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+	if (root->query_uniquekeys != NIL)
+		return true;	/* there are some */
+	return false;		/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *	  Given the parameters for a UniqueKey, find any pre-existing matching
+ *	  uniquekey in the query's list of "canonical" uniquekeys.  Make a new
+ *	  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instead,
+ * equivclass.c will complain if a merge occurs after root->canon_uniquekeys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+						 EquivalenceClass *eclass)
+{
+	UniqueKey  *uk;
+	ListCell   *lc;
+	MemoryContext oldcontext;
+
+	/* The passed eclass might be non-canonical, so chase up to the top */
+	while (eclass->ec_merged)
+		eclass = eclass->ec_merged;
+
+	foreach(lc, root->canon_uniquekeys)
+	{
+		uk = (UniqueKey *) lfirst(lc);
+		if (eclass == uk->eq_clause)
+			return uk;
+	}
+
+	/*
+	 * Be sure canonical uniquekeys are allocated in the main planning context.
+	 * Not an issue in normal planning, but it is for GEQO.
+	 */
+	oldcontext = MemoryContextSwitchTo(root->planner_cxt);
+
+	uk = makeNode(UniqueKey);
+	uk->eq_clause = eclass;
+
+	root->canon_uniquekeys = lappend(root->canon_uniquekeys, uk);
+
+	MemoryContextSwitchTo(oldcontext);
+
+	return uk;
+}
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/plan/planagg.c
index 8634940efc..dd64775d8f 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -511,6 +511,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 									  root->parse->targetList);
 
 	root->query_pathkeys = root->sort_pathkeys;
+	root->query_uniquekeys = NIL;
 }
 
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/plan/planmain.c
index 62dfc6d44a..3a372af91b 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 	root->join_rel_level = NULL;
 	root->join_cur_level = 0;
 	root->canon_pathkeys = NIL;
+	root->canon_uniquekeys = NIL;
 	root->left_join_clauses = NIL;
 	root->right_join_clauses = NIL;
 	root->full_join_clauses = NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index d6f2153593..984fca0696 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3657,15 +3657,30 @@ standard_qp_callback(PlannerInfo *root, void *extra)
 	 * much easier, since we know that the parser ensured that one is a
 	 * superset of the other.
 	 */
+	root->query_uniquekeys = NIL;
+
 	if (root->group_pathkeys)
+	{
 		root->query_pathkeys = root->group_pathkeys;
+
+		if (!root->parse->hasAggs)
+			root->query_uniquekeys = build_uniquekeys(root, qp_extra->groupClause);
+	}
 	else if (root->window_pathkeys)
 		root->query_pathkeys = root->window_pathkeys;
 	else if (list_length(root->distinct_pathkeys) >
 			 list_length(root->sort_pathkeys))
+	{
 		root->query_pathkeys = root->distinct_pathkeys;
+		root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else if (root->sort_pathkeys)
+	{
 		root->query_pathkeys = root->sort_pathkeys;
+
+		if (root->distinct_pathkeys)
+			root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else
 		root->query_pathkeys = NIL;
 }
@@ -6222,7 +6237,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
 
 	/* Estimate the cost of index scan */
 	indexScanPath = create_index_path(root, indexInfo,
-									  NIL, NIL, NIL, NIL,
+									  NIL, NIL, NIL, NIL, NIL,
 									  ForwardScanDirection, false,
 									  NULL, 1.0, false);
 
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index e6d08aede5..a006dbbe9c 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -940,6 +940,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = parallel_workers;
 	pathnode->pathkeys = NIL;	/* seqscan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_seqscan(pathnode, root, rel, pathnode->param_info);
 
@@ -964,6 +965,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* samplescan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_samplescan(pathnode, root, rel, pathnode->param_info);
 
@@ -1000,6 +1002,7 @@ create_index_path(PlannerInfo *root,
 				  List *indexorderbys,
 				  List *indexorderbycols,
 				  List *pathkeys,
+				  List *uniquekeys,
 				  ScanDirection indexscandir,
 				  bool indexonly,
 				  Relids required_outer,
@@ -1018,6 +1021,7 @@ create_index_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = 0;
 	pathnode->path.pathkeys = pathkeys;
+	pathnode->path.uniquekeys = uniquekeys;
 
 	pathnode->indexinfo = index;
 	pathnode->indexclauses = indexclauses;
@@ -1061,6 +1065,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = parallel_degree;
 	pathnode->path.pathkeys = NIL;	/* always unordered */
+	pathnode->path.uniquekeys = NIL;
 
 	pathnode->bitmapqual = bitmapqual;
 
@@ -1922,6 +1927,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = pathkeys;
+	pathnode->uniquekeys = NIL;
 
 	cost_functionscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1948,6 +1954,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1974,6 +1981,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_valuesscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1999,6 +2007,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* XXX for now, result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2025,6 +2034,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2051,6 +2061,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_resultscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2077,6 +2088,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	/* Cost is the same as for a regular CTE scan */
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index baced7eec0..a1511b46ea 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 	T_EquivalenceMember,
 	T_PathKey,
 	T_PathTarget,
+	T_UniqueKey,
 	T_RestrictInfo,
 	T_IndexClause,
 	T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 3d3be197e0..4e329f0fb5 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -269,6 +269,8 @@ struct PlannerInfo
 
 	List	   *canon_pathkeys; /* list of "canonical" PathKeys */
 
+	List	   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 	List	   *left_join_clauses;	/* list of RestrictInfos for mergejoinable
 									 * outer join clauses w/nonnullable var on
 									 * left */
@@ -297,6 +299,8 @@ struct PlannerInfo
 
 	List	   *query_pathkeys; /* desired pathkeys for query_planner() */
 
+	List	   *query_uniquekeys; /* unique keys used for the query */
+
 	List	   *group_pathkeys; /* groupClause pathkeys, if any */
 	List	   *window_pathkeys;	/* pathkeys of bottom window, if any */
 	List	   *distinct_pathkeys;	/* distinctClause pathkeys, if any */
@@ -1077,6 +1081,15 @@ typedef struct ParamPathInfo
 	List	   *ppi_clauses;	/* join clauses available from outer rels */
 } ParamPathInfo;
 
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+	NodeTag		type;
+
+	EquivalenceClass *eq_clause;	/* equivalence class */
+} UniqueKey;
 
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some other
@@ -1106,6 +1119,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1129,6 +1145,8 @@ typedef struct Path
 
 	List	   *pathkeys;		/* sort ordering of path's output */
 	/* pathkeys is a List of PathKey nodes; see above */
+
+	List	   *uniquekeys;	/* the unique keys, or NIL if none */
 } Path;
 
 /* Macro for extracting a path's parameterization relids; beware double eval */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index 6126b491bf..006248bfb5 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
 
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index e450fe112a..f75ff6f323 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -44,6 +44,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 									List *indexorderbys,
 									List *indexorderbycols,
 									List *pathkeys,
+									List *uniquekeys,
 									ScanDirection indexscandir,
 									bool indexonly,
 									Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index 9ab73bd20c..5b6be383b3 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 										   List *sortclauses,
 										   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+										  List *sortclauses,
+										  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 											RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *root,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 									List *live_childrels);
 
+/*
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif							/* PATHS_H */
-- 
2.21.0


--cxnum6r77tfmznfe
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v31-0002-Index-skip-scan.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH v32 1/2] Unique key
@ 2019-07-09 10:44  jesperpedersen <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: jesperpedersen @ 2019-07-09 10:44 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   3 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/indxpath.c  |  41 +++++++
 src/backend/optimizer/path/pathkeys.c  |  71 ++++++++++--
 src/backend/optimizer/path/uniquekey.c | 147 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planagg.c   |   1 +
 src/backend/optimizer/plan/planmain.c  |   1 +
 src/backend/optimizer/plan/planner.c   |  17 ++-
 src/backend/optimizer/util/pathnode.c  |  12 ++
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  18 +++
 src/include/nodes/print.h              |   1 +
 src/include/optimizer/pathnode.h       |   1 +
 src/include/optimizer/paths.h          |  11 ++
 16 files changed, 373 insertions(+), 13 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index d76fae44b8..16083e7a7e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1723,6 +1723,7 @@ _outPathInfo(StringInfo str, const Path *node)
 	WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 	WRITE_FLOAT_FIELD(total_cost, "%.2f");
 	WRITE_NODE_FIELD(pathkeys);
+	WRITE_NODE_FIELD(uniquekeys);
 }
 
 /*
@@ -2205,6 +2206,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(eq_classes);
 	WRITE_BOOL_FIELD(ec_merging_done);
 	WRITE_NODE_FIELD(canon_pathkeys);
+	WRITE_NODE_FIELD(canon_uniquekeys);
 	WRITE_NODE_FIELD(left_join_clauses);
 	WRITE_NODE_FIELD(right_join_clauses);
 	WRITE_NODE_FIELD(full_join_clauses);
@@ -2214,6 +2216,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(placeholder_list);
 	WRITE_NODE_FIELD(fkey_list);
 	WRITE_NODE_FIELD(query_pathkeys);
+	WRITE_NODE_FIELD(query_uniquekeys);
 	WRITE_NODE_FIELD(group_pathkeys);
 	WRITE_NODE_FIELD(window_pathkeys);
 	WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2401,6 +2404,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 	WRITE_BOOL_FIELD(pk_nulls_first);
 }
 
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+	WRITE_NODE_TYPE("UNIQUEKEY");
+
+	WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4092,6 +4103,9 @@ outNode(StringInfo str, const void *obj)
 			case T_PathKey:
 				_outPathKey(str, obj);
 				break;
+			case T_UniqueKey:
+				_outUniqueKey(str, obj);
+				break;
 			case T_PathTarget:
 				_outPathTarget(str, obj);
 				break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 42476724d8..d286b34544 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtable)
 	printf(")\n");
 }
 
+/*
+ * print_uniquekeys -
+ *	  uniquekeys list of UniqueKeys
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+	ListCell   *l;
+
+	printf("(");
+	foreach(l, uniquekeys)
+	{
+		UniqueKey *unique_key = (UniqueKey *) lfirst(l);
+		EquivalenceClass *eclass = (EquivalenceClass *) unique_key->eq_clause;
+		ListCell   *k;
+		bool		first = true;
+
+		/* chase up */
+		while (eclass->ec_merged)
+			eclass = eclass->ec_merged;
+
+		printf("(");
+		foreach(k, eclass->ec_members)
+		{
+			EquivalenceMember *mem = (EquivalenceMember *) lfirst(k);
+
+			if (first)
+				first = false;
+			else
+				printf(", ");
+			print_expr((Node *) mem->em_expr, rtable);
+		}
+		printf(")");
+		if (lnext(uniquekeys, l))
+			printf(", ");
+	}
+	printf(")\n");
+}
+
 /*
  * print_tl
  *	  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/path/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS = \
 	joinpath.o \
 	joinrels.o \
 	pathkeys.o \
-	tidpath.o
+	tidpath.o \
+	uniquekey.o
 
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 8286d9cf34..bbc13e6141 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent)
 		print_pathkeys(path->pathkeys, root->parse->rtable);
 	}
 
+	if (path->uniquekeys)
+	{
+		for (i = 0; i < indent; i++)
+			printf("\t");
+		printf("  uniquekeys: ");
+		print_uniquekeys(path->uniquekeys, root->parse->rtable);
+	}
+
 	if (join)
 	{
 		JoinPath   *jp = (JoinPath *) path;
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 2a50272da6..bd1ea53e5c 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 									   EquivalenceClass *ec, EquivalenceMember *em,
 									   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
 
 
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	List	   *orderbyclausecols;
 	List	   *index_pathkeys;
 	List	   *useful_pathkeys;
+	List	   *useful_uniquekeys = NIL;
 	bool		found_lower_saop_clause;
 	bool		pathkeys_possibly_useful;
 	bool		index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	if (index_clauses != NIL || useful_pathkeys != NIL || useful_predicate ||
 		index_only_scan)
 	{
+		if (has_useful_uniquekeys(root))
+			useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 		ipath = create_index_path(root, index,
 								  index_clauses,
 								  orderbyclauses,
 								  orderbyclausecols,
 								  useful_pathkeys,
+								  useful_uniquekeys,
 								  index_is_ordered ?
 								  ForwardScanDirection :
 								  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 									  orderbyclauses,
 									  orderbyclausecols,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  index_is_ordered ?
 									  ForwardScanDirection :
 									  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 													index_pathkeys);
 		if (useful_pathkeys != NIL)
 		{
+			if (has_useful_uniquekeys(root))
+				useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 			ipath = create_index_path(root, index,
 									  index_clauses,
 									  NIL,
 									  NIL,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  BackwardScanDirection,
 									  index_only_scan,
 									  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 										  NIL,
 										  NIL,
 										  useful_pathkeys,
+										  useful_uniquekeys,
 										  BackwardScanDirection,
 										  index_only_scan,
 										  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 	return clause;
 }
 
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+	ListCell *lc;
+
+	if (pathkeys)
+	{
+		List *uniquekeys = NIL;
+		foreach(lc, pathkeys)
+		{
+			UniqueKey *unique_key;
+			PathKey *pk = (PathKey *) lfirst(lc);
+			EquivalenceClass *ec = (EquivalenceClass *) pk->pk_eclass;
+
+			unique_key = makeNode(UniqueKey);
+			unique_key->eq_clause = ec;
+
+			lappend(uniquekeys, unique_key);
+		}
+
+		if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+			return uniquekeys;
+	}
+
+	return NIL;
+}
 
 /****************************************************************************
  *				----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS	----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/path/pathkeys.c
index 71b9d42c99..054df9a617 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
 
 
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 											 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 	return pk;
 }
 
+/*
+ * pathkey_is_unique
+ *	   Checks if the new pathkey's equivalence class is the same as that of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
+	ListCell   *lc;
+
+	/* If same EC already is already in the list, then not unique */
+	foreach(lc, pathkeys)
+	{
+		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
+
+		if (new_ec == old_pathkey->pk_eclass)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * pathkey_is_redundant
  *	   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
-	ListCell   *lc;
 
 	/* Check for EC containing a constant --- unconditionally redundant */
 	if (EC_MUST_BE_REDUNDANT(new_ec))
 		return true;
 
-	/* If same EC already used in list, then redundant */
-	foreach(lc, pathkeys)
-	{
-		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
-
-		if (new_ec == old_pathkey->pk_eclass)
-			return true;
-	}
-
-	return false;
+	return !pathkey_is_unique(new_pathkey, pathkeys);
 }
 
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 	return pathkeys;
 }
 
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *		Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+							 List *sortclauses,
+							 List *tlist)
+{
+	List	   *pathkeys = NIL;
+	ListCell   *l;
+
+	foreach(l, sortclauses)
+	{
+		SortGroupClause *sortcl = (SortGroupClause *) lfirst(l);
+		Expr	   *sortkey;
+		PathKey    *pathkey;
+
+		sortkey = (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+		Assert(OidIsValid(sortcl->sortop));
+		pathkey = make_pathkey_from_sortop(root,
+										   sortkey,
+										   root->nullable_baserels,
+										   sortcl->sortop,
+										   sortcl->nulls_first,
+										   sortcl->tleSortGroupRef,
+										   true);
+
+		if (pathkey_is_unique(pathkey, pathkeys))
+			pathkeys = lappend(pathkeys, pathkey);
+	}
+
+	return pathkeys;
+}
+
 /****************************************************************************
  *		PATHKEYS AND MERGECLAUSES
  ****************************************************************************/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer/path/uniquekey.c
new file mode 100644
index 0000000000..13d4ebb98c
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,147 @@
+/*-------------------------------------------------------------------------
+ *
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  src/backend/optimizer/path/uniquekey.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceClass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+	List *result = NIL;
+	List *sortkeys;
+	ListCell *l;
+
+	sortkeys = make_pathkeys_for_uniquekeys(root,
+											sortclauses,
+											root->processed_tlist);
+
+	/* Create a uniquekey and add it to the list */
+	foreach(l, sortkeys)
+	{
+		PathKey    *pathkey = (PathKey *) lfirst(l);
+		EquivalenceClass *ec = pathkey->pk_eclass;
+		UniqueKey *unique_key = make_canonical_uniquekey(root, ec);
+
+		result = lappend(result, unique_key);
+	}
+
+	return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *	  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+	ListCell   *key1,
+			   *key2;
+
+	/*
+	 * Fall out quickly if we are passed two identical lists.  This mostly
+	 * catches the case where both are NIL, but that's common enough to
+	 * warrant the test.
+	 */
+	if (keys1 == keys2)
+		return true;
+
+	foreach(key2, keys2)
+	{
+		bool found = false;
+		UniqueKey  *uniquekey2 = (UniqueKey *) lfirst(key2);
+
+		foreach(key1, keys1)
+		{
+			UniqueKey  *uniquekey1 = (UniqueKey *) lfirst(key1);
+
+			if (uniquekey1->eq_clause == uniquekey2->eq_clause)
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *		Detect whether the planner could have any uniquekeys that are
+ *		useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+	if (root->query_uniquekeys != NIL)
+		return true;	/* there are some */
+	return false;		/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *	  Given the parameters for a UniqueKey, find any pre-existing matching
+ *	  uniquekey in the query's list of "canonical" uniquekeys.  Make a new
+ *	  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instead,
+ * equivclass.c will complain if a merge occurs after root->canon_uniquekeys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+						 EquivalenceClass *eclass)
+{
+	UniqueKey  *uk;
+	ListCell   *lc;
+	MemoryContext oldcontext;
+
+	/* The passed eclass might be non-canonical, so chase up to the top */
+	while (eclass->ec_merged)
+		eclass = eclass->ec_merged;
+
+	foreach(lc, root->canon_uniquekeys)
+	{
+		uk = (UniqueKey *) lfirst(lc);
+		if (eclass == uk->eq_clause)
+			return uk;
+	}
+
+	/*
+	 * Be sure canonical uniquekeys are allocated in the main planning context.
+	 * Not an issue in normal planning, but it is for GEQO.
+	 */
+	oldcontext = MemoryContextSwitchTo(root->planner_cxt);
+
+	uk = makeNode(UniqueKey);
+	uk->eq_clause = eclass;
+
+	root->canon_uniquekeys = lappend(root->canon_uniquekeys, uk);
+
+	MemoryContextSwitchTo(oldcontext);
+
+	return uk;
+}
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/plan/planagg.c
index 8634940efc..dd64775d8f 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -511,6 +511,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 									  root->parse->targetList);
 
 	root->query_pathkeys = root->sort_pathkeys;
+	root->query_uniquekeys = NIL;
 }
 
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/plan/planmain.c
index 62dfc6d44a..3a372af91b 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 	root->join_rel_level = NULL;
 	root->join_cur_level = 0;
 	root->canon_pathkeys = NIL;
+	root->canon_uniquekeys = NIL;
 	root->left_join_clauses = NIL;
 	root->right_join_clauses = NIL;
 	root->full_join_clauses = NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index d6f2153593..984fca0696 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3657,15 +3657,30 @@ standard_qp_callback(PlannerInfo *root, void *extra)
 	 * much easier, since we know that the parser ensured that one is a
 	 * superset of the other.
 	 */
+	root->query_uniquekeys = NIL;
+
 	if (root->group_pathkeys)
+	{
 		root->query_pathkeys = root->group_pathkeys;
+
+		if (!root->parse->hasAggs)
+			root->query_uniquekeys = build_uniquekeys(root, qp_extra->groupClause);
+	}
 	else if (root->window_pathkeys)
 		root->query_pathkeys = root->window_pathkeys;
 	else if (list_length(root->distinct_pathkeys) >
 			 list_length(root->sort_pathkeys))
+	{
 		root->query_pathkeys = root->distinct_pathkeys;
+		root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else if (root->sort_pathkeys)
+	{
 		root->query_pathkeys = root->sort_pathkeys;
+
+		if (root->distinct_pathkeys)
+			root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else
 		root->query_pathkeys = NIL;
 }
@@ -6222,7 +6237,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
 
 	/* Estimate the cost of index scan */
 	indexScanPath = create_index_path(root, indexInfo,
-									  NIL, NIL, NIL, NIL,
+									  NIL, NIL, NIL, NIL, NIL,
 									  ForwardScanDirection, false,
 									  NULL, 1.0, false);
 
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index e6d08aede5..a006dbbe9c 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -940,6 +940,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = parallel_workers;
 	pathnode->pathkeys = NIL;	/* seqscan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_seqscan(pathnode, root, rel, pathnode->param_info);
 
@@ -964,6 +965,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* samplescan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_samplescan(pathnode, root, rel, pathnode->param_info);
 
@@ -1000,6 +1002,7 @@ create_index_path(PlannerInfo *root,
 				  List *indexorderbys,
 				  List *indexorderbycols,
 				  List *pathkeys,
+				  List *uniquekeys,
 				  ScanDirection indexscandir,
 				  bool indexonly,
 				  Relids required_outer,
@@ -1018,6 +1021,7 @@ create_index_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = 0;
 	pathnode->path.pathkeys = pathkeys;
+	pathnode->path.uniquekeys = uniquekeys;
 
 	pathnode->indexinfo = index;
 	pathnode->indexclauses = indexclauses;
@@ -1061,6 +1065,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = parallel_degree;
 	pathnode->path.pathkeys = NIL;	/* always unordered */
+	pathnode->path.uniquekeys = NIL;
 
 	pathnode->bitmapqual = bitmapqual;
 
@@ -1922,6 +1927,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = pathkeys;
+	pathnode->uniquekeys = NIL;
 
 	cost_functionscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1948,6 +1954,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1974,6 +1981,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_valuesscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1999,6 +2007,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* XXX for now, result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2025,6 +2034,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2051,6 +2061,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_resultscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2077,6 +2088,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	/* Cost is the same as for a regular CTE scan */
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index baced7eec0..a1511b46ea 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 	T_EquivalenceMember,
 	T_PathKey,
 	T_PathTarget,
+	T_UniqueKey,
 	T_RestrictInfo,
 	T_IndexClause,
 	T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 3d3be197e0..4e329f0fb5 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -269,6 +269,8 @@ struct PlannerInfo
 
 	List	   *canon_pathkeys; /* list of "canonical" PathKeys */
 
+	List	   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 	List	   *left_join_clauses;	/* list of RestrictInfos for mergejoinable
 									 * outer join clauses w/nonnullable var on
 									 * left */
@@ -297,6 +299,8 @@ struct PlannerInfo
 
 	List	   *query_pathkeys; /* desired pathkeys for query_planner() */
 
+	List	   *query_uniquekeys; /* unique keys used for the query */
+
 	List	   *group_pathkeys; /* groupClause pathkeys, if any */
 	List	   *window_pathkeys;	/* pathkeys of bottom window, if any */
 	List	   *distinct_pathkeys;	/* distinctClause pathkeys, if any */
@@ -1077,6 +1081,15 @@ typedef struct ParamPathInfo
 	List	   *ppi_clauses;	/* join clauses available from outer rels */
 } ParamPathInfo;
 
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+	NodeTag		type;
+
+	EquivalenceClass *eq_clause;	/* equivalence class */
+} UniqueKey;
 
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some other
@@ -1106,6 +1119,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1129,6 +1145,8 @@ typedef struct Path
 
 	List	   *pathkeys;		/* sort ordering of path's output */
 	/* pathkeys is a List of PathKey nodes; see above */
+
+	List	   *uniquekeys;	/* the unique keys, or NIL if none */
 } Path;
 
 /* Macro for extracting a path's parameterization relids; beware double eval */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index 6126b491bf..006248bfb5 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
 
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index e450fe112a..f75ff6f323 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -44,6 +44,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 									List *indexorderbys,
 									List *indexorderbycols,
 									List *pathkeys,
+									List *uniquekeys,
 									ScanDirection indexscandir,
 									bool indexonly,
 									Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index 9ab73bd20c..5b6be383b3 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 										   List *sortclauses,
 										   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+										  List *sortclauses,
+										  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 											RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *root,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 									List *live_childrels);
 
+/*
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif							/* PATHS_H */
-- 
2.21.0


--f7z2si56qzkbcsr7
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
 filename="v32-0002-Index-skip-scan-visibility-check.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH 1/2] Unique key
@ 2019-08-02 11:52  jesperpedersen <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: jesperpedersen @ 2019-08-02 11:52 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   2 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/costsize.c  |   5 +
 src/backend/optimizer/path/indxpath.c  |  41 +++++++
 src/backend/optimizer/path/pathkeys.c  |  72 ++++++++++--
 src/backend/optimizer/path/uniquekey.c | 147 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planner.c   |  18 ++-
 src/backend/optimizer/util/pathnode.c  |  12 ++
 src/backend/optimizer/util/tlist.c     |   1 -
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  18 +++
 src/include/nodes/print.h              |   2 +-
 src/include/optimizer/pathnode.h       |   1 +
 src/include/optimizer/paths.h          |  11 ++
 16 files changed, 377 insertions(+), 15 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e6ce8e2110..9a4f3e8e4b 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1720,6 +1720,7 @@ _outPathInfo(StringInfo str, const Path *node)
 	WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 	WRITE_FLOAT_FIELD(total_cost, "%.2f");
 	WRITE_NODE_FIELD(pathkeys);
+	WRITE_NODE_FIELD(uniquekeys);
 }
 
 /*
@@ -2201,6 +2202,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(eq_classes);
 	WRITE_BOOL_FIELD(ec_merging_done);
 	WRITE_NODE_FIELD(canon_pathkeys);
+	WRITE_NODE_FIELD(canon_uniquekeys);
 	WRITE_NODE_FIELD(left_join_clauses);
 	WRITE_NODE_FIELD(right_join_clauses);
 	WRITE_NODE_FIELD(full_join_clauses);
@@ -2210,6 +2212,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(placeholder_list);
 	WRITE_NODE_FIELD(fkey_list);
 	WRITE_NODE_FIELD(query_pathkeys);
+	WRITE_NODE_FIELD(query_uniquekeys);
 	WRITE_NODE_FIELD(group_pathkeys);
 	WRITE_NODE_FIELD(window_pathkeys);
 	WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2397,6 +2400,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 	WRITE_BOOL_FIELD(pk_nulls_first);
 }
 
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+	WRITE_NODE_TYPE("UNIQUEKEY");
+
+	WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4073,6 +4084,9 @@ outNode(StringInfo str, const void *obj)
 			case T_PathKey:
 				_outPathKey(str, obj);
 				break;
+			case T_UniqueKey:
+				_outUniqueKey(str, obj);
+				break;
 			case T_PathTarget:
 				_outPathTarget(str, obj);
 				break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 4ecde6b421..62c9d4ef7a 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtable)
 	printf(")\n");
 }
 
+/*
+ * print_uniquekeys -
+ *	  unique_key an UniqueKey
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+	ListCell   *l;
+
+	printf("(");
+	foreach(l, uniquekeys)
+	{
+		UniqueKey *unique_key = (UniqueKey *) lfirst(l);
+		EquivalenceClass *eclass = (EquivalenceClass *) unique_key->eq_clause;
+		ListCell   *k;
+		bool		first = true;
+
+		/* chase up */
+		while (eclass->ec_merged)
+			eclass = eclass->ec_merged;
+
+		printf("(");
+		foreach(k, eclass->ec_members)
+		{
+			EquivalenceMember *mem = (EquivalenceMember *) lfirst(k);
+
+			if (first)
+				first = false;
+			else
+				printf(", ");
+			print_expr((Node *) mem->em_expr, rtable);
+		}
+		printf(")");
+		if (lnext(uniquekeys, l))
+			printf(", ");
+	}
+	printf(")\n");
+}
+
 /*
  * print_tl
  *	  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/path/Makefile
index 6864a62132..8249a6b6db 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -13,6 +13,6 @@ top_builddir = ../../../..
 include $(top_builddir)/src/Makefile.global
 
 OBJS = allpaths.o clausesel.o costsize.o equivclass.o indxpath.o \
-       joinpath.o joinrels.o pathkeys.o tidpath.o
+       joinpath.o joinrels.o pathkeys.o tidpath.o uniquekey.o
 
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index e9ee32b7f4..6f4c25f7dd 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3957,6 +3957,14 @@ print_path(PlannerInfo *root, Path *path, int indent)
 		print_pathkeys(path->pathkeys, root->parse->rtable);
 	}
 
+	if (path->uniquekeys)
+	{
+		for (i = 0; i < indent; i++)
+			printf("\t");
+		printf("  uniquekeys: ");
+		print_uniquekeys(path->uniquekeys, root->parse->rtable);
+	}
+
 	if (join)
 	{
 		JoinPath   *jp = (JoinPath *) path;
diff --git a/src/backend/optimizer/path/costsize.c b/src/backend/optimizer/path/costsize.c
index 3a9a994733..2565dcf296 100644
--- a/src/backend/optimizer/path/costsize.c
+++ b/src/backend/optimizer/path/costsize.c
@@ -705,6 +705,11 @@ cost_index(IndexPath *path, PlannerInfo *root, double loop_count,
 		path->path.parallel_aware = true;
 	}
 
+	/* Consider cost based on unique key */
+	if (path->path.uniquekeys)
+	{
+	}
+
 	/*
 	 * Now interpolate based on estimated index order correlation to get total
 	 * disk I/O cost for main table accesses.
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 5f339fdfde..4b90dd378a 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 									   EquivalenceClass *ec, EquivalenceMember *em,
 									   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
 
 
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	List	   *orderbyclausecols;
 	List	   *index_pathkeys;
 	List	   *useful_pathkeys;
+	List	   *useful_uniquekeys;
 	bool		found_lower_saop_clause;
 	bool		pathkeys_possibly_useful;
 	bool		index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	if (index_clauses != NIL || useful_pathkeys != NIL || useful_predicate ||
 		index_only_scan)
 	{
+		if (has_useful_uniquekeys(root))
+			useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 		ipath = create_index_path(root, index,
 								  index_clauses,
 								  orderbyclauses,
 								  orderbyclausecols,
 								  useful_pathkeys,
+								  useful_uniquekeys,
 								  index_is_ordered ?
 								  ForwardScanDirection :
 								  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 									  orderbyclauses,
 									  orderbyclausecols,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  index_is_ordered ?
 									  ForwardScanDirection :
 									  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 													index_pathkeys);
 		if (useful_pathkeys != NIL)
 		{
+			if (has_useful_uniquekeys(root))
+				useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 			ipath = create_index_path(root, index,
 									  index_clauses,
 									  NIL,
 									  NIL,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  BackwardScanDirection,
 									  index_only_scan,
 									  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 										  NIL,
 										  NIL,
 										  useful_pathkeys,
+										  useful_uniquekeys,
 										  BackwardScanDirection,
 										  index_only_scan,
 										  outer_relids,
@@ -3369,6 +3381,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 	return clause;
 }
 
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+	ListCell *lc;
+
+	if (pathkeys)
+	{
+		List *uniquekeys = NIL;
+		foreach(lc, pathkeys)
+		{
+			UniqueKey *unique_key;
+			PathKey *pk = (PathKey *) lfirst(lc);
+			EquivalenceClass *ec = (EquivalenceClass *) pk->pk_eclass;
+
+			unique_key = makeNode(UniqueKey);
+			unique_key->eq_clause = ec;
+			
+			lappend(uniquekeys, unique_key);
+		}
+
+		if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+			return uniquekeys;
+	}
+
+	return NIL;
+}
 
 /****************************************************************************
  *				----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS	----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/path/pathkeys.c
index 2f4fea241a..0cba366c06 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
 
 
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 											 RelOptInfo *partrel,
@@ -96,6 +97,30 @@ make_canonical_pathkey(PlannerInfo *root,
 	return pk;
 }
 
+/*
+ * pathkey_is_unique
+ *	   Part of pathkey_is_redundant, that is reponsible for the case, when the
+ *	   new pathkey's equivalence class is the same as that of any existing
+ *	   member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
+	ListCell   *lc;
+
+	/* If same EC already is already in the list, then not unique */
+	foreach(lc, pathkeys)
+	{
+		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
+
+		if (new_ec == old_pathkey->pk_eclass)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * pathkey_is_redundant
  *	   Is a pathkey redundant with one already in the given list?
@@ -135,22 +160,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
-	ListCell   *lc;
 
 	/* Check for EC containing a constant --- unconditionally redundant */
 	if (EC_MUST_BE_REDUNDANT(new_ec))
 		return true;
 
-	/* If same EC already used in list, then redundant */
-	foreach(lc, pathkeys)
-	{
-		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
-
-		if (new_ec == old_pathkey->pk_eclass)
-			return true;
-	}
-
-	return false;
+	return !pathkey_is_unique(new_pathkey, pathkeys);
 }
 
 /*
@@ -1098,6 +1113,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 	return pathkeys;
 }
 
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *		Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+							 List *sortclauses,
+							 List *tlist)
+{
+	List	   *pathkeys = NIL;
+	ListCell   *l;
+
+	foreach(l, sortclauses)
+	{
+		SortGroupClause *sortcl = (SortGroupClause *) lfirst(l);
+		Expr	   *sortkey;
+		PathKey    *pathkey;
+
+		sortkey = (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+		Assert(OidIsValid(sortcl->sortop));
+		pathkey = make_pathkey_from_sortop(root,
+										   sortkey,
+										   root->nullable_baserels,
+										   sortcl->sortop,
+										   sortcl->nulls_first,
+										   sortcl->tleSortGroupRef,
+										   true);
+
+		if (pathkey_is_unique(pathkey, pathkeys))
+			pathkeys = lappend(pathkeys, pathkey);
+	}
+
+	return pathkeys;
+}
+
 /****************************************************************************
  *		PATHKEYS AND MERGECLAUSES
  ****************************************************************************/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer/path/uniquekey.c
new file mode 100644
index 0000000000..13d4ebb98c
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,147 @@
+/*-------------------------------------------------------------------------
+ *
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  src/backend/optimizer/path/uniquekey.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceClass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+	List *result = NIL;
+	List *sortkeys;
+	ListCell *l;
+
+	sortkeys = make_pathkeys_for_uniquekeys(root,
+											sortclauses,
+											root->processed_tlist);
+
+	/* Create a uniquekey and add it to the list */
+	foreach(l, sortkeys)
+	{
+		PathKey    *pathkey = (PathKey *) lfirst(l);
+		EquivalenceClass *ec = pathkey->pk_eclass;
+		UniqueKey *unique_key = make_canonical_uniquekey(root, ec);
+
+		result = lappend(result, unique_key);
+	}
+
+	return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *	  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+	ListCell   *key1,
+			   *key2;
+
+	/*
+	 * Fall out quickly if we are passed two identical lists.  This mostly
+	 * catches the case where both are NIL, but that's common enough to
+	 * warrant the test.
+	 */
+	if (keys1 == keys2)
+		return true;
+
+	foreach(key2, keys2)
+	{
+		bool found = false;
+		UniqueKey  *uniquekey2 = (UniqueKey *) lfirst(key2);
+
+		foreach(key1, keys1)
+		{
+			UniqueKey  *uniquekey1 = (UniqueKey *) lfirst(key1);
+
+			if (uniquekey1->eq_clause == uniquekey2->eq_clause)
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *		Detect whether the planner could have any uniquekeys that are
+ *		useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+	if (root->query_uniquekeys != NIL)
+		return true;	/* there are some */
+	return false;		/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *	  Given the parameters for a UniqueKey, find any pre-existing matching
+ *	  uniquekey in the query's list of "canonical" uniquekeys.  Make a new
+ *	  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instead,
+ * equivclass.c will complain if a merge occurs after root->canon_uniquekeys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+						 EquivalenceClass *eclass)
+{
+	UniqueKey  *uk;
+	ListCell   *lc;
+	MemoryContext oldcontext;
+
+	/* The passed eclass might be non-canonical, so chase up to the top */
+	while (eclass->ec_merged)
+		eclass = eclass->ec_merged;
+
+	foreach(lc, root->canon_uniquekeys)
+	{
+		uk = (UniqueKey *) lfirst(lc);
+		if (eclass == uk->eq_clause)
+			return uk;
+	}
+
+	/*
+	 * Be sure canonical uniquekeys are allocated in the main planning context.
+	 * Not an issue in normal planning, but it is for GEQO.
+	 */
+	oldcontext = MemoryContextSwitchTo(root->planner_cxt);
+
+	uk = makeNode(UniqueKey);
+	uk->eq_clause = eclass;
+
+	root->canon_uniquekeys = lappend(root->canon_uniquekeys, uk);
+
+	MemoryContextSwitchTo(oldcontext);
+
+	return uk;
+}
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index 8f51f59f8a..5ee9ee6595 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3656,15 +3656,31 @@ standard_qp_callback(PlannerInfo *root, void *extra)
 	 * much easier, since we know that the parser ensured that one is a
 	 * superset of the other.
 	 */
+	root->canon_uniquekeys = NIL;
+	root->query_uniquekeys = NIL;
+
 	if (root->group_pathkeys)
+	{
 		root->query_pathkeys = root->group_pathkeys;
+
+		if (!root->parse->hasAggs)
+			root->query_uniquekeys = build_uniquekeys(root, qp_extra->groupClause);
+	}
 	else if (root->window_pathkeys)
 		root->query_pathkeys = root->window_pathkeys;
 	else if (list_length(root->distinct_pathkeys) >
 			 list_length(root->sort_pathkeys))
+	{
 		root->query_pathkeys = root->distinct_pathkeys;
+		root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else if (root->sort_pathkeys)
+	{
 		root->query_pathkeys = root->sort_pathkeys;
+
+		if (root->distinct_pathkeys)
+			root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else
 		root->query_pathkeys = NIL;
 }
@@ -6222,7 +6238,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
 
 	/* Estimate the cost of index scan */
 	indexScanPath = create_index_path(root, indexInfo,
-									  NIL, NIL, NIL, NIL,
+									  NIL, NIL, NIL, NIL, NIL,
 									  ForwardScanDirection, false,
 									  NULL, 1.0, false);
 
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index 0ac73984d2..ac0b937895 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -941,6 +941,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = parallel_workers;
 	pathnode->pathkeys = NIL;	/* seqscan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_seqscan(pathnode, root, rel, pathnode->param_info);
 
@@ -965,6 +966,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* samplescan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_samplescan(pathnode, root, rel, pathnode->param_info);
 
@@ -1001,6 +1003,7 @@ create_index_path(PlannerInfo *root,
 				  List *indexorderbys,
 				  List *indexorderbycols,
 				  List *pathkeys,
+				  List *uniquekeys,
 				  ScanDirection indexscandir,
 				  bool indexonly,
 				  Relids required_outer,
@@ -1019,6 +1022,7 @@ create_index_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = 0;
 	pathnode->path.pathkeys = pathkeys;
+	pathnode->path.uniquekeys = uniquekeys;
 
 	pathnode->indexinfo = index;
 	pathnode->indexclauses = indexclauses;
@@ -1062,6 +1066,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = parallel_degree;
 	pathnode->path.pathkeys = NIL;	/* always unordered */
+	pathnode->path.uniquekeys = NIL;
 
 	pathnode->bitmapqual = bitmapqual;
 
@@ -1923,6 +1928,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = pathkeys;
+	pathnode->uniquekeys = NIL;
 
 	cost_functionscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1949,6 +1955,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1975,6 +1982,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_valuesscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2000,6 +2008,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* XXX for now, result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2026,6 +2035,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2052,6 +2062,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_resultscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2078,6 +2089,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	/* Cost is the same as for a regular CTE scan */
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/backend/optimizer/util/tlist.c b/src/backend/optimizer/util/tlist.c
index 7ccb10e4e1..618032e82c 100644
--- a/src/backend/optimizer/util/tlist.c
+++ b/src/backend/optimizer/util/tlist.c
@@ -427,7 +427,6 @@ get_sortgrouplist_exprs(List *sgClauses, List *targetList)
 	return result;
 }
 
-
 /*****************************************************************************
  *		Functions to extract data from a list of SortGroupClauses
  *
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index 4e2fb39105..a9b67c64f8 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 	T_EquivalenceMember,
 	T_PathKey,
 	T_PathTarget,
+	T_UniqueKey,
 	T_RestrictInfo,
 	T_IndexClause,
 	T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index e3c579ee44..c1d6f33fc0 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -269,6 +269,8 @@ struct PlannerInfo
 
 	List	   *canon_pathkeys; /* list of "canonical" PathKeys */
 
+	List	   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 	List	   *left_join_clauses;	/* list of RestrictInfos for mergejoinable
 									 * outer join clauses w/nonnullable var on
 									 * left */
@@ -297,6 +299,8 @@ struct PlannerInfo
 
 	List	   *query_pathkeys; /* desired pathkeys for query_planner() */
 
+	List	   *query_uniquekeys; /* */
+
 	List	   *group_pathkeys; /* groupClause pathkeys, if any */
 	List	   *window_pathkeys;	/* pathkeys of bottom window, if any */
 	List	   *distinct_pathkeys;	/* distinctClause pathkeys, if any */
@@ -1077,6 +1081,15 @@ typedef struct ParamPathInfo
 	List	   *ppi_clauses;	/* join clauses available from outer rels */
 } ParamPathInfo;
 
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+	NodeTag		type;
+
+	EquivalenceClass *eq_clause;	/* equivalence class */
+} UniqueKey;
 
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some other
@@ -1106,6 +1119,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1129,6 +1145,8 @@ typedef struct Path
 
 	List	   *pathkeys;		/* sort ordering of path's output */
 	/* pathkeys is a List of PathKey nodes; see above */
+
+	List	   *uniquekeys;	/* the unique keys, or NIL if none */
 } Path;
 
 /* Macro for extracting a path's parameterization relids; beware double eval */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index cbff56a724..31e8f0686e 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -16,7 +16,6 @@
 
 #include "executor/tuptable.h"
 
-
 #define nodeDisplay(x)		pprint(x)
 
 extern void print(const void *obj);
@@ -28,6 +27,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
 
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index 182ffeef4b..374cac157b 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -44,6 +44,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 									List *indexorderbys,
 									List *indexorderbycols,
 									List *pathkeys,
+									List *uniquekeys,
 									ScanDirection indexscandir,
 									bool indexonly,
 									Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index 7345137d1d..c7976d4a90 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -209,6 +209,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 										   List *sortclauses,
 										   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+										  List *sortclauses,
+										  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 											RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -235,4 +238,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *root,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 									List *live_childrels);
 
+/*
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif							/* PATHS_H */
-- 
2.21.0


--------------A6AA3F98E061114DE0DF40DC
Content-Type: text/x-patch;
 name="v23_0002-Index-skip-scan.patch"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
 filename="v23_0002-Index-skip-scan.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH 1/2] Unique key
@ 2019-11-11 13:49  jesperpedersen <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: jesperpedersen @ 2019-11-11 13:49 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   3 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/costsize.c  |   5 +
 src/backend/optimizer/path/indxpath.c  |  41 +++++++
 src/backend/optimizer/path/pathkeys.c  |  71 ++++++++++--
 src/backend/optimizer/path/uniquekey.c | 147 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planagg.c   |   1 +
 src/backend/optimizer/plan/planmain.c  |   1 +
 src/backend/optimizer/plan/planner.c   |  17 ++-
 src/backend/optimizer/util/pathnode.c  |  12 ++
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  18 +++
 src/include/nodes/print.h              |   1 +
 src/include/optimizer/pathnode.h       |   1 +
 src/include/optimizer/paths.h          |  11 ++
 17 files changed, 378 insertions(+), 13 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index b0dcd02ff6..1ccd68d3aa 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1720,6 +1720,7 @@ _outPathInfo(StringInfo str, const Path *node)
 =09WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 =09WRITE_FLOAT_FIELD(total_cost, "%.2f");
 =09WRITE_NODE_FIELD(pathkeys);
+=09WRITE_NODE_FIELD(uniquekeys);
 }
=20
 /*
@@ -2201,6 +2202,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *no=
de)
 =09WRITE_NODE_FIELD(eq_classes);
 =09WRITE_BOOL_FIELD(ec_merging_done);
 =09WRITE_NODE_FIELD(canon_pathkeys);
+=09WRITE_NODE_FIELD(canon_uniquekeys);
 =09WRITE_NODE_FIELD(left_join_clauses);
 =09WRITE_NODE_FIELD(right_join_clauses);
 =09WRITE_NODE_FIELD(full_join_clauses);
@@ -2210,6 +2212,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *no=
de)
 =09WRITE_NODE_FIELD(placeholder_list);
 =09WRITE_NODE_FIELD(fkey_list);
 =09WRITE_NODE_FIELD(query_pathkeys);
+=09WRITE_NODE_FIELD(query_uniquekeys);
 =09WRITE_NODE_FIELD(group_pathkeys);
 =09WRITE_NODE_FIELD(window_pathkeys);
 =09WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2397,6 +2400,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 =09WRITE_BOOL_FIELD(pk_nulls_first);
 }
=20
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+=09WRITE_NODE_TYPE("UNIQUEKEY");
+
+=09WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4083,6 +4094,9 @@ outNode(StringInfo str, const void *obj)
 =09=09=09case T_PathKey:
 =09=09=09=09_outPathKey(str, obj);
 =09=09=09=09break;
+=09=09=09case T_UniqueKey:
+=09=09=09=09_outUniqueKey(str, obj);
+=09=09=09=09break;
 =09=09=09case T_PathTarget:
 =09=09=09=09_outPathTarget(str, obj);
 =09=09=09=09break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 4ecde6b421..62c9d4ef7a 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtabl=
e)
 =09printf(")\n");
 }
=20
+/*
+ * print_uniquekeys -
+ *=09  unique_key an UniqueKey
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+=09ListCell   *l;
+
+=09printf("(");
+=09foreach(l, uniquekeys)
+=09{
+=09=09UniqueKey *unique_key =3D (UniqueKey *) lfirst(l);
+=09=09EquivalenceClass *eclass =3D (EquivalenceClass *) unique_key->eq_cla=
use;
+=09=09ListCell   *k;
+=09=09bool=09=09first =3D true;
+
+=09=09/* chase up */
+=09=09while (eclass->ec_merged)
+=09=09=09eclass =3D eclass->ec_merged;
+
+=09=09printf("(");
+=09=09foreach(k, eclass->ec_members)
+=09=09{
+=09=09=09EquivalenceMember *mem =3D (EquivalenceMember *) lfirst(k);
+
+=09=09=09if (first)
+=09=09=09=09first =3D false;
+=09=09=09else
+=09=09=09=09printf(", ");
+=09=09=09print_expr((Node *) mem->em_expr, rtable);
+=09=09}
+=09=09printf(")");
+=09=09if (lnext(uniquekeys, l))
+=09=09=09printf(", ");
+=09}
+=09printf(")\n");
+}
+
 /*
  * print_tl
  *=09  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/pa=
th/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS =3D \
 =09joinpath.o \
 =09joinrels.o \
 =09pathkeys.o \
-=09tidpath.o
+=09tidpath.o \
+=09uniquekey.o
=20
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/=
path/allpaths.c
index db3a68a51d..5fc9b81746 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent=
)
 =09=09print_pathkeys(path->pathkeys, root->parse->rtable);
 =09}
=20
+=09if (path->uniquekeys)
+=09{
+=09=09for (i =3D 0; i < indent; i++)
+=09=09=09printf("\t");
+=09=09printf("  uniquekeys: ");
+=09=09print_uniquekeys(path->uniquekeys, root->parse->rtable);
+=09}
+
 =09if (join)
 =09{
 =09=09JoinPath   *jp =3D (JoinPath *) path;
diff --git a/src/backend/optimizer/path/costsize.c b/src/backend/optimizer/=
path/costsize.c
index c5f6593485..0ec9a6db76 100644
--- a/src/backend/optimizer/path/costsize.c
+++ b/src/backend/optimizer/path/costsize.c
@@ -705,6 +705,11 @@ cost_index(IndexPath *path, PlannerInfo *root, double =
loop_count,
 =09=09path->path.parallel_aware =3D true;
 =09}
=20
+=09/* Consider cost based on unique key */
+=09if (path->path.uniquekeys)
+=09{
+=09}
+
 =09/*
 =09 * Now interpolate based on estimated index order correlation to get to=
tal
 =09 * disk I/O cost for main table accesses.
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/=
path/indxpath.c
index 37b257cd0e..aa0da49119 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *=
index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 =09=09=09=09=09=09=09=09=09   EquivalenceClass *ec, EquivalenceMember *em,
 =09=09=09=09=09=09=09=09=09   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
=20
=20
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 =09List=09   *orderbyclausecols;
 =09List=09   *index_pathkeys;
 =09List=09   *useful_pathkeys;
+=09List=09   *useful_uniquekeys;
 =09bool=09=09found_lower_saop_clause;
 =09bool=09=09pathkeys_possibly_useful;
 =09bool=09=09index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *re=
l,
 =09if (index_clauses !=3D NIL || useful_pathkeys !=3D NIL || useful_predic=
ate ||
 =09=09index_only_scan)
 =09{
+=09=09if (has_useful_uniquekeys(root))
+=09=09=09useful_uniquekeys =3D get_uniquekeys_for_index(root, useful_pathk=
eys);
+
 =09=09ipath =3D create_index_path(root, index,
 =09=09=09=09=09=09=09=09  index_clauses,
 =09=09=09=09=09=09=09=09  orderbyclauses,
 =09=09=09=09=09=09=09=09  orderbyclausecols,
 =09=09=09=09=09=09=09=09  useful_pathkeys,
+=09=09=09=09=09=09=09=09  useful_uniquekeys,
 =09=09=09=09=09=09=09=09  index_is_ordered ?
 =09=09=09=09=09=09=09=09  ForwardScanDirection :
 =09=09=09=09=09=09=09=09  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 =09=09=09=09=09=09=09=09=09  orderbyclauses,
 =09=09=09=09=09=09=09=09=09  orderbyclausecols,
 =09=09=09=09=09=09=09=09=09  useful_pathkeys,
+=09=09=09=09=09=09=09=09=09  useful_uniquekeys,
 =09=09=09=09=09=09=09=09=09  index_is_ordered ?
 =09=09=09=09=09=09=09=09=09  ForwardScanDirection :
 =09=09=09=09=09=09=09=09=09  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *re=
l,
 =09=09=09=09=09=09=09=09=09=09=09=09=09index_pathkeys);
 =09=09if (useful_pathkeys !=3D NIL)
 =09=09{
+=09=09=09if (has_useful_uniquekeys(root))
+=09=09=09=09useful_uniquekeys =3D get_uniquekeys_for_index(root, useful_pa=
thkeys);
+
 =09=09=09ipath =3D create_index_path(root, index,
 =09=09=09=09=09=09=09=09=09  index_clauses,
 =09=09=09=09=09=09=09=09=09  NIL,
 =09=09=09=09=09=09=09=09=09  NIL,
 =09=09=09=09=09=09=09=09=09  useful_pathkeys,
+=09=09=09=09=09=09=09=09=09  useful_uniquekeys,
 =09=09=09=09=09=09=09=09=09  BackwardScanDirection,
 =09=09=09=09=09=09=09=09=09  index_only_scan,
 =09=09=09=09=09=09=09=09=09  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 =09=09=09=09=09=09=09=09=09=09  NIL,
 =09=09=09=09=09=09=09=09=09=09  NIL,
 =09=09=09=09=09=09=09=09=09=09  useful_pathkeys,
+=09=09=09=09=09=09=09=09=09=09  useful_uniquekeys,
 =09=09=09=09=09=09=09=09=09=09  BackwardScanDirection,
 =09=09=09=09=09=09=09=09=09=09  index_only_scan,
 =09=09=09=09=09=09=09=09=09=09  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 =09return clause;
 }
=20
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+=09ListCell *lc;
+
+=09if (pathkeys)
+=09{
+=09=09List *uniquekeys =3D NIL;
+=09=09foreach(lc, pathkeys)
+=09=09{
+=09=09=09UniqueKey *unique_key;
+=09=09=09PathKey *pk =3D (PathKey *) lfirst(lc);
+=09=09=09EquivalenceClass *ec =3D (EquivalenceClass *) pk->pk_eclass;
+
+=09=09=09unique_key =3D makeNode(UniqueKey);
+=09=09=09unique_key->eq_clause =3D ec;
+=09=09=09
+=09=09=09lappend(uniquekeys, unique_key);
+=09=09}
+
+=09=09if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+=09=09=09return uniquekeys;
+=09}
+
+=09return NIL;
+}
=20
 /*************************************************************************=
***
  *=09=09=09=09----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS=09----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/=
path/pathkeys.c
index 2f4fea241a..b0dc1bc22a 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
=20
=20
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 =09=09=09=09=09=09=09=09=09=09=09 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 =09return pk;
 }
=20
+/*
+ * pathkey_is_unique
+ *=09   Checks if the new pathkey's equivalence class is the same as that =
of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+=09EquivalenceClass *new_ec =3D new_pathkey->pk_eclass;
+=09ListCell   *lc;
+
+=09/* If same EC already is already in the list, then not unique */
+=09foreach(lc, pathkeys)
+=09{
+=09=09PathKey    *old_pathkey =3D (PathKey *) lfirst(lc);
+
+=09=09if (new_ec =3D=3D old_pathkey->pk_eclass)
+=09=09=09return false;
+=09}
+
+=09return true;
+}
+
 /*
  * pathkey_is_redundant
  *=09   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 =09EquivalenceClass *new_ec =3D new_pathkey->pk_eclass;
-=09ListCell   *lc;
=20
 =09/* Check for EC containing a constant --- unconditionally redundant */
 =09if (EC_MUST_BE_REDUNDANT(new_ec))
 =09=09return true;
=20
-=09/* If same EC already used in list, then redundant */
-=09foreach(lc, pathkeys)
-=09{
-=09=09PathKey    *old_pathkey =3D (PathKey *) lfirst(lc);
-
-=09=09if (new_ec =3D=3D old_pathkey->pk_eclass)
-=09=09=09return true;
-=09}
-
-=09return false;
+=09return !pathkey_is_unique(new_pathkey, pathkeys);
 }
=20
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 =09return pathkeys;
 }
=20
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *=09=09Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+=09=09=09=09=09=09=09 List *sortclauses,
+=09=09=09=09=09=09=09 List *tlist)
+{
+=09List=09   *pathkeys =3D NIL;
+=09ListCell   *l;
+
+=09foreach(l, sortclauses)
+=09{
+=09=09SortGroupClause *sortcl =3D (SortGroupClause *) lfirst(l);
+=09=09Expr=09   *sortkey;
+=09=09PathKey    *pathkey;
+
+=09=09sortkey =3D (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+=09=09Assert(OidIsValid(sortcl->sortop));
+=09=09pathkey =3D make_pathkey_from_sortop(root,
+=09=09=09=09=09=09=09=09=09=09   sortkey,
+=09=09=09=09=09=09=09=09=09=09   root->nullable_baserels,
+=09=09=09=09=09=09=09=09=09=09   sortcl->sortop,
+=09=09=09=09=09=09=09=09=09=09   sortcl->nulls_first,
+=09=09=09=09=09=09=09=09=09=09   sortcl->tleSortGroupRef,
+=09=09=09=09=09=09=09=09=09=09   true);
+
+=09=09if (pathkey_is_unique(pathkey, pathkeys))
+=09=09=09pathkeys =3D lappend(pathkeys, pathkey);
+=09}
+
+=09return pathkeys;
+}
+
 /*************************************************************************=
***
  *=09=09PATHKEYS AND MERGECLAUSES
  *************************************************************************=
***/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer=
/path/uniquekey.c
new file mode 100644
index 0000000000..13d4ebb98c
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,147 @@
+/*------------------------------------------------------------------------=
-
+ *
+ * uniquekey.c
+ *=09  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *=09  src/backend/optimizer/path/uniquekey.c
+ *
+ *------------------------------------------------------------------------=
-
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceC=
lass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+=09List *result =3D NIL;
+=09List *sortkeys;
+=09ListCell *l;
+
+=09sortkeys =3D make_pathkeys_for_uniquekeys(root,
+=09=09=09=09=09=09=09=09=09=09=09sortclauses,
+=09=09=09=09=09=09=09=09=09=09=09root->processed_tlist);
+
+=09/* Create a uniquekey and add it to the list */
+=09foreach(l, sortkeys)
+=09{
+=09=09PathKey    *pathkey =3D (PathKey *) lfirst(l);
+=09=09EquivalenceClass *ec =3D pathkey->pk_eclass;
+=09=09UniqueKey *unique_key =3D make_canonical_uniquekey(root, ec);
+
+=09=09result =3D lappend(result, unique_key);
+=09}
+
+=09return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *=09  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+=09ListCell   *key1,
+=09=09=09   *key2;
+
+=09/*
+=09 * Fall out quickly if we are passed two identical lists.  This mostly
+=09 * catches the case where both are NIL, but that's common enough to
+=09 * warrant the test.
+=09 */
+=09if (keys1 =3D=3D keys2)
+=09=09return true;
+
+=09foreach(key2, keys2)
+=09{
+=09=09bool found =3D false;
+=09=09UniqueKey  *uniquekey2 =3D (UniqueKey *) lfirst(key2);
+
+=09=09foreach(key1, keys1)
+=09=09{
+=09=09=09UniqueKey  *uniquekey1 =3D (UniqueKey *) lfirst(key1);
+
+=09=09=09if (uniquekey1->eq_clause =3D=3D uniquekey2->eq_clause)
+=09=09=09{
+=09=09=09=09found =3D true;
+=09=09=09=09break;
+=09=09=09}
+=09=09}
+
+=09=09if (!found)
+=09=09=09return false;
+=09}
+
+=09return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *=09=09Detect whether the planner could have any uniquekeys that are
+ *=09=09useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+=09if (root->query_uniquekeys !=3D NIL)
+=09=09return true;=09/* there are some */
+=09return false;=09=09/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *=09  Given the parameters for a UniqueKey, find any pre-existing matchin=
g
+ *=09  uniquekey in the query's list of "canonical" uniquekeys.  Make a ne=
w
+ *=09  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instea=
d,
+ * equivclass.c will complain if a merge occurs after root->canon_uniqueke=
ys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+=09=09=09=09=09=09 EquivalenceClass *eclass)
+{
+=09UniqueKey  *uk;
+=09ListCell   *lc;
+=09MemoryContext oldcontext;
+
+=09/* The passed eclass might be non-canonical, so chase up to the top */
+=09while (eclass->ec_merged)
+=09=09eclass =3D eclass->ec_merged;
+
+=09foreach(lc, root->canon_uniquekeys)
+=09{
+=09=09uk =3D (UniqueKey *) lfirst(lc);
+=09=09if (eclass =3D=3D uk->eq_clause)
+=09=09=09return uk;
+=09}
+
+=09/*
+=09 * Be sure canonical uniquekeys are allocated in the main planning cont=
ext.
+=09 * Not an issue in normal planning, but it is for GEQO.
+=09 */
+=09oldcontext =3D MemoryContextSwitchTo(root->planner_cxt);
+
+=09uk =3D makeNode(UniqueKey);
+=09uk->eq_clause =3D eclass;
+
+=09root->canon_uniquekeys =3D lappend(root->canon_uniquekeys, uk);
+
+=09MemoryContextSwitchTo(oldcontext);
+
+=09return uk;
+}
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/p=
lan/planagg.c
index 9381939c82..3d32f6dfd6 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -512,6 +512,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 =09=09=09=09=09=09=09=09=09  root->parse->targetList);
=20
 =09root->query_pathkeys =3D root->sort_pathkeys;
+=09root->query_uniquekeys =3D NIL;
 }
=20
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/=
plan/planmain.c
index f0c1b52a2e..3ccde03ab7 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 =09root->join_rel_level =3D NULL;
 =09root->join_cur_level =3D 0;
 =09root->canon_pathkeys =3D NIL;
+=09root->canon_uniquekeys =3D NIL;
 =09root->left_join_clauses =3D NIL;
 =09root->right_join_clauses =3D NIL;
 =09root->full_join_clauses =3D NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/p=
lan/planner.c
index 17c5f086fb..2507ec7d2a 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3652,15 +3652,30 @@ standard_qp_callback(PlannerInfo *root, void *extra=
)
 =09 * much easier, since we know that the parser ensured that one is a
 =09 * superset of the other.
 =09 */
+=09root->query_uniquekeys =3D NIL;
+
 =09if (root->group_pathkeys)
+=09{
 =09=09root->query_pathkeys =3D root->group_pathkeys;
+
+=09=09if (!root->parse->hasAggs)
+=09=09=09root->query_uniquekeys =3D build_uniquekeys(root, qp_extra->group=
Clause);
+=09}
 =09else if (root->window_pathkeys)
 =09=09root->query_pathkeys =3D root->window_pathkeys;
 =09else if (list_length(root->distinct_pathkeys) >
 =09=09=09 list_length(root->sort_pathkeys))
+=09{
 =09=09root->query_pathkeys =3D root->distinct_pathkeys;
+=09=09root->query_uniquekeys =3D build_uniquekeys(root, parse->distinctCla=
use);
+=09}
 =09else if (root->sort_pathkeys)
+=09{
 =09=09root->query_pathkeys =3D root->sort_pathkeys;
+
+=09=09if (root->distinct_pathkeys)
+=09=09=09root->query_uniquekeys =3D build_uniquekeys(root, parse->distinct=
Clause);
+=09}
 =09else
 =09=09root->query_pathkeys =3D NIL;
 }
@@ -6217,7 +6232,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
=20
 =09/* Estimate the cost of index scan */
 =09indexScanPath =3D create_index_path(root, indexInfo,
-=09=09=09=09=09=09=09=09=09  NIL, NIL, NIL, NIL,
+=09=09=09=09=09=09=09=09=09  NIL, NIL, NIL, NIL, NIL,
 =09=09=09=09=09=09=09=09=09  ForwardScanDirection, false,
 =09=09=09=09=09=09=09=09=09  NULL, 1.0, false);
=20
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/=
util/pathnode.c
index 34acb732ee..f268112386 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -941,6 +941,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D parallel_workers;
 =09pathnode->pathkeys =3D NIL;=09/* seqscan has unordered result */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_seqscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -965,6 +966,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *r=
el, Relids required_outer
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* samplescan has unordered result */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_samplescan(pathnode, root, rel, pathnode->param_info);
=20
@@ -1001,6 +1003,7 @@ create_index_path(PlannerInfo *root,
 =09=09=09=09  List *indexorderbys,
 =09=09=09=09  List *indexorderbycols,
 =09=09=09=09  List *pathkeys,
+=09=09=09=09  List *uniquekeys,
 =09=09=09=09  ScanDirection indexscandir,
 =09=09=09=09  bool indexonly,
 =09=09=09=09  Relids required_outer,
@@ -1019,6 +1022,7 @@ create_index_path(PlannerInfo *root,
 =09pathnode->path.parallel_safe =3D rel->consider_parallel;
 =09pathnode->path.parallel_workers =3D 0;
 =09pathnode->path.pathkeys =3D pathkeys;
+=09pathnode->path.uniquekeys =3D uniquekeys;
=20
 =09pathnode->indexinfo =3D index;
 =09pathnode->indexclauses =3D indexclauses;
@@ -1062,6 +1066,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 =09pathnode->path.parallel_safe =3D rel->consider_parallel;
 =09pathnode->path.parallel_workers =3D parallel_degree;
 =09pathnode->path.pathkeys =3D NIL;=09/* always unordered */
+=09pathnode->path.uniquekeys =3D NIL;
=20
 =09pathnode->bitmapqual =3D bitmapqual;
=20
@@ -1923,6 +1928,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInf=
o *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D pathkeys;
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_functionscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -1949,6 +1955,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptIn=
fo *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -1975,6 +1982,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo =
*rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_valuesscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -2000,6 +2008,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *re=
l, Relids required_outer)
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* XXX for now, result is always unordere=
d */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_ctescan(pathnode, root, rel, pathnode->param_info);
=20
@@ -2026,6 +2035,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, Re=
lOptInfo *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
=20
@@ -2052,6 +2062,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo =
*rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09cost_resultscan(pathnode, root, rel, pathnode->param_info);
=20
@@ -2078,6 +2089,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptIn=
fo *rel,
 =09pathnode->parallel_safe =3D rel->consider_parallel;
 =09pathnode->parallel_workers =3D 0;
 =09pathnode->pathkeys =3D NIL;=09/* result is always unordered */
+=09pathnode->uniquekeys =3D NIL;
=20
 =09/* Cost is the same as for a regular CTE scan */
 =09cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index bce2d59b0d..cbb6ba2586 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 =09T_EquivalenceMember,
 =09T_PathKey,
 =09T_PathTarget,
+=09T_UniqueKey,
 =09T_RestrictInfo,
 =09T_IndexClause,
 =09T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 23a06d718e..6198c31cd4 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -267,6 +267,8 @@ struct PlannerInfo
=20
 =09List=09   *canon_pathkeys; /* list of "canonical" PathKeys */
=20
+=09List=09   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 =09List=09   *left_join_clauses;=09/* list of RestrictInfos for mergejoina=
ble
 =09=09=09=09=09=09=09=09=09 * outer join clauses w/nonnullable var on
 =09=09=09=09=09=09=09=09=09 * left */
@@ -295,6 +297,8 @@ struct PlannerInfo
=20
 =09List=09   *query_pathkeys; /* desired pathkeys for query_planner() */
=20
+=09List=09   *query_uniquekeys; /* */
+
 =09List=09   *group_pathkeys; /* groupClause pathkeys, if any */
 =09List=09   *window_pathkeys;=09/* pathkeys of bottom window, if any */
 =09List=09   *distinct_pathkeys;=09/* distinctClause pathkeys, if any */
@@ -1075,6 +1079,15 @@ typedef struct ParamPathInfo
 =09List=09   *ppi_clauses;=09/* join clauses available from outer rels */
 } ParamPathInfo;
=20
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+=09NodeTag=09=09type;
+
+=09EquivalenceClass *eq_clause;=09/* equivalence class */
+} UniqueKey;
=20
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some ot=
her
@@ -1104,6 +1117,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1127,6 +1143,8 @@ typedef struct Path
=20
 =09List=09   *pathkeys;=09=09/* sort ordering of path's output */
 =09/* pathkeys is a List of PathKey nodes; see above */
+
+=09List=09   *uniquekeys;=09/* the unique keys, or NIL if none */
 } Path;
=20
 /* Macro for extracting a path's parameterization relids; beware double ev=
al */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index cbff56a724..f1a7112877 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
=20
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathn=
ode.h
index a12af54971..37a946f857 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -44,6 +44,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 =09=09=09=09=09=09=09=09=09List *indexorderbys,
 =09=09=09=09=09=09=09=09=09List *indexorderbycols,
 =09=09=09=09=09=09=09=09=09List *pathkeys,
+=09=09=09=09=09=09=09=09=09List *uniquekeys,
 =09=09=09=09=09=09=09=09=09ScanDirection indexscandir,
 =09=09=09=09=09=09=09=09=09bool indexonly,
 =09=09=09=09=09=09=09=09=09Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index c6c34630c2..c79e47eeaf 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 =09=09=09=09=09=09=09=09=09=09   List *sortclauses,
 =09=09=09=09=09=09=09=09=09=09   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+=09=09=09=09=09=09=09=09=09=09  List *sortclauses,
+=09=09=09=09=09=09=09=09=09=09  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 =09=09=09=09=09=09=09=09=09=09=09RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *ro=
ot,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 =09=09=09=09=09=09=09=09=09List *live_childrels);
=20
+/*
+ * uniquekey.c
+ *=09  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif=09=09=09=09=09=09=09/* PATHS_H */
--=20
2.21.0


--------------25BA7482CCCC3CA4CEC0D330
Content-Type: text/x-patch; charset=UTF-8;
 name="v29_0002-Index-skip-scan.patch"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
 filename="v29_0002-Index-skip-scan.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH 1/2] Unique key
@ 2019-11-15 14:46  jesperpedersen <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: jesperpedersen @ 2019-11-15 14:46 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   3 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/indxpath.c  |  41 +++++++
 src/backend/optimizer/path/pathkeys.c  |  71 ++++++++++--
 src/backend/optimizer/path/uniquekey.c | 147 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planagg.c   |   1 +
 src/backend/optimizer/plan/planmain.c  |   1 +
 src/backend/optimizer/plan/planner.c   |  17 ++-
 src/backend/optimizer/util/pathnode.c  |  12 ++
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  18 +++
 src/include/nodes/print.h              |   1 +
 src/include/optimizer/pathnode.h       |   1 +
 src/include/optimizer/paths.h          |  11 ++
 16 files changed, 373 insertions(+), 13 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index d76fae44b8..16083e7a7e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1723,6 +1723,7 @@ _outPathInfo(StringInfo str, const Path *node)
 	WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 	WRITE_FLOAT_FIELD(total_cost, "%.2f");
 	WRITE_NODE_FIELD(pathkeys);
+	WRITE_NODE_FIELD(uniquekeys);
 }
 
 /*
@@ -2205,6 +2206,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(eq_classes);
 	WRITE_BOOL_FIELD(ec_merging_done);
 	WRITE_NODE_FIELD(canon_pathkeys);
+	WRITE_NODE_FIELD(canon_uniquekeys);
 	WRITE_NODE_FIELD(left_join_clauses);
 	WRITE_NODE_FIELD(right_join_clauses);
 	WRITE_NODE_FIELD(full_join_clauses);
@@ -2214,6 +2216,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(placeholder_list);
 	WRITE_NODE_FIELD(fkey_list);
 	WRITE_NODE_FIELD(query_pathkeys);
+	WRITE_NODE_FIELD(query_uniquekeys);
 	WRITE_NODE_FIELD(group_pathkeys);
 	WRITE_NODE_FIELD(window_pathkeys);
 	WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2401,6 +2404,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 	WRITE_BOOL_FIELD(pk_nulls_first);
 }
 
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+	WRITE_NODE_TYPE("UNIQUEKEY");
+
+	WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4092,6 +4103,9 @@ outNode(StringInfo str, const void *obj)
 			case T_PathKey:
 				_outPathKey(str, obj);
 				break;
+			case T_UniqueKey:
+				_outUniqueKey(str, obj);
+				break;
 			case T_PathTarget:
 				_outPathTarget(str, obj);
 				break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 42476724d8..d286b34544 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtable)
 	printf(")\n");
 }
 
+/*
+ * print_uniquekeys -
+ *	  uniquekeys list of UniqueKeys
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+	ListCell   *l;
+
+	printf("(");
+	foreach(l, uniquekeys)
+	{
+		UniqueKey *unique_key = (UniqueKey *) lfirst(l);
+		EquivalenceClass *eclass = (EquivalenceClass *) unique_key->eq_clause;
+		ListCell   *k;
+		bool		first = true;
+
+		/* chase up */
+		while (eclass->ec_merged)
+			eclass = eclass->ec_merged;
+
+		printf("(");
+		foreach(k, eclass->ec_members)
+		{
+			EquivalenceMember *mem = (EquivalenceMember *) lfirst(k);
+
+			if (first)
+				first = false;
+			else
+				printf(", ");
+			print_expr((Node *) mem->em_expr, rtable);
+		}
+		printf(")");
+		if (lnext(uniquekeys, l))
+			printf(", ");
+	}
+	printf(")\n");
+}
+
 /*
  * print_tl
  *	  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/path/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS = \
 	joinpath.o \
 	joinrels.o \
 	pathkeys.o \
-	tidpath.o
+	tidpath.o \
+	uniquekey.o
 
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 8286d9cf34..bbc13e6141 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent)
 		print_pathkeys(path->pathkeys, root->parse->rtable);
 	}
 
+	if (path->uniquekeys)
+	{
+		for (i = 0; i < indent; i++)
+			printf("\t");
+		printf("  uniquekeys: ");
+		print_uniquekeys(path->uniquekeys, root->parse->rtable);
+	}
+
 	if (join)
 	{
 		JoinPath   *jp = (JoinPath *) path;
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 2a50272da6..34e7fafa84 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 									   EquivalenceClass *ec, EquivalenceMember *em,
 									   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
 
 
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	List	   *orderbyclausecols;
 	List	   *index_pathkeys;
 	List	   *useful_pathkeys;
+	List	   *useful_uniquekeys = NIL;
 	bool		found_lower_saop_clause;
 	bool		pathkeys_possibly_useful;
 	bool		index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	if (index_clauses != NIL || useful_pathkeys != NIL || useful_predicate ||
 		index_only_scan)
 	{
+		if (has_useful_uniquekeys(root))
+			useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 		ipath = create_index_path(root, index,
 								  index_clauses,
 								  orderbyclauses,
 								  orderbyclausecols,
 								  useful_pathkeys,
+								  useful_uniquekeys,
 								  index_is_ordered ?
 								  ForwardScanDirection :
 								  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 									  orderbyclauses,
 									  orderbyclausecols,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  index_is_ordered ?
 									  ForwardScanDirection :
 									  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 													index_pathkeys);
 		if (useful_pathkeys != NIL)
 		{
+			if (has_useful_uniquekeys(root))
+				useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 			ipath = create_index_path(root, index,
 									  index_clauses,
 									  NIL,
 									  NIL,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  BackwardScanDirection,
 									  index_only_scan,
 									  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 										  NIL,
 										  NIL,
 										  useful_pathkeys,
+										  useful_uniquekeys,
 										  BackwardScanDirection,
 										  index_only_scan,
 										  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 	return clause;
 }
 
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+	ListCell *lc;
+
+	if (pathkeys)
+	{
+		List *uniquekeys = NIL;
+		foreach(lc, pathkeys)
+		{
+			UniqueKey *unique_key;
+			PathKey *pk = (PathKey *) lfirst(lc);
+			EquivalenceClass *ec = (EquivalenceClass *) pk->pk_eclass;
+
+			unique_key = makeNode(UniqueKey);
+			unique_key->eq_clause = ec;
+			
+			lappend(uniquekeys, unique_key);
+		}
+
+		if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+			return uniquekeys;
+	}
+
+	return NIL;
+}
 
 /****************************************************************************
  *				----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS	----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/path/pathkeys.c
index 71b9d42c99..054df9a617 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
 
 
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 											 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 	return pk;
 }
 
+/*
+ * pathkey_is_unique
+ *	   Checks if the new pathkey's equivalence class is the same as that of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
+	ListCell   *lc;
+
+	/* If same EC already is already in the list, then not unique */
+	foreach(lc, pathkeys)
+	{
+		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
+
+		if (new_ec == old_pathkey->pk_eclass)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * pathkey_is_redundant
  *	   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
-	ListCell   *lc;
 
 	/* Check for EC containing a constant --- unconditionally redundant */
 	if (EC_MUST_BE_REDUNDANT(new_ec))
 		return true;
 
-	/* If same EC already used in list, then redundant */
-	foreach(lc, pathkeys)
-	{
-		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
-
-		if (new_ec == old_pathkey->pk_eclass)
-			return true;
-	}
-
-	return false;
+	return !pathkey_is_unique(new_pathkey, pathkeys);
 }
 
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 	return pathkeys;
 }
 
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *		Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+							 List *sortclauses,
+							 List *tlist)
+{
+	List	   *pathkeys = NIL;
+	ListCell   *l;
+
+	foreach(l, sortclauses)
+	{
+		SortGroupClause *sortcl = (SortGroupClause *) lfirst(l);
+		Expr	   *sortkey;
+		PathKey    *pathkey;
+
+		sortkey = (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+		Assert(OidIsValid(sortcl->sortop));
+		pathkey = make_pathkey_from_sortop(root,
+										   sortkey,
+										   root->nullable_baserels,
+										   sortcl->sortop,
+										   sortcl->nulls_first,
+										   sortcl->tleSortGroupRef,
+										   true);
+
+		if (pathkey_is_unique(pathkey, pathkeys))
+			pathkeys = lappend(pathkeys, pathkey);
+	}
+
+	return pathkeys;
+}
+
 /****************************************************************************
  *		PATHKEYS AND MERGECLAUSES
  ****************************************************************************/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer/path/uniquekey.c
new file mode 100644
index 0000000000..13d4ebb98c
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,147 @@
+/*-------------------------------------------------------------------------
+ *
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  src/backend/optimizer/path/uniquekey.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceClass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+	List *result = NIL;
+	List *sortkeys;
+	ListCell *l;
+
+	sortkeys = make_pathkeys_for_uniquekeys(root,
+											sortclauses,
+											root->processed_tlist);
+
+	/* Create a uniquekey and add it to the list */
+	foreach(l, sortkeys)
+	{
+		PathKey    *pathkey = (PathKey *) lfirst(l);
+		EquivalenceClass *ec = pathkey->pk_eclass;
+		UniqueKey *unique_key = make_canonical_uniquekey(root, ec);
+
+		result = lappend(result, unique_key);
+	}
+
+	return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *	  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+	ListCell   *key1,
+			   *key2;
+
+	/*
+	 * Fall out quickly if we are passed two identical lists.  This mostly
+	 * catches the case where both are NIL, but that's common enough to
+	 * warrant the test.
+	 */
+	if (keys1 == keys2)
+		return true;
+
+	foreach(key2, keys2)
+	{
+		bool found = false;
+		UniqueKey  *uniquekey2 = (UniqueKey *) lfirst(key2);
+
+		foreach(key1, keys1)
+		{
+			UniqueKey  *uniquekey1 = (UniqueKey *) lfirst(key1);
+
+			if (uniquekey1->eq_clause == uniquekey2->eq_clause)
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *		Detect whether the planner could have any uniquekeys that are
+ *		useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+	if (root->query_uniquekeys != NIL)
+		return true;	/* there are some */
+	return false;		/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *	  Given the parameters for a UniqueKey, find any pre-existing matching
+ *	  uniquekey in the query's list of "canonical" uniquekeys.  Make a new
+ *	  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instead,
+ * equivclass.c will complain if a merge occurs after root->canon_uniquekeys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+						 EquivalenceClass *eclass)
+{
+	UniqueKey  *uk;
+	ListCell   *lc;
+	MemoryContext oldcontext;
+
+	/* The passed eclass might be non-canonical, so chase up to the top */
+	while (eclass->ec_merged)
+		eclass = eclass->ec_merged;
+
+	foreach(lc, root->canon_uniquekeys)
+	{
+		uk = (UniqueKey *) lfirst(lc);
+		if (eclass == uk->eq_clause)
+			return uk;
+	}
+
+	/*
+	 * Be sure canonical uniquekeys are allocated in the main planning context.
+	 * Not an issue in normal planning, but it is for GEQO.
+	 */
+	oldcontext = MemoryContextSwitchTo(root->planner_cxt);
+
+	uk = makeNode(UniqueKey);
+	uk->eq_clause = eclass;
+
+	root->canon_uniquekeys = lappend(root->canon_uniquekeys, uk);
+
+	MemoryContextSwitchTo(oldcontext);
+
+	return uk;
+}
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/plan/planagg.c
index 8634940efc..dd64775d8f 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -511,6 +511,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 									  root->parse->targetList);
 
 	root->query_pathkeys = root->sort_pathkeys;
+	root->query_uniquekeys = NIL;
 }
 
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/plan/planmain.c
index 62dfc6d44a..3a372af91b 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 	root->join_rel_level = NULL;
 	root->join_cur_level = 0;
 	root->canon_pathkeys = NIL;
+	root->canon_uniquekeys = NIL;
 	root->left_join_clauses = NIL;
 	root->right_join_clauses = NIL;
 	root->full_join_clauses = NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index d6f2153593..984fca0696 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3657,15 +3657,30 @@ standard_qp_callback(PlannerInfo *root, void *extra)
 	 * much easier, since we know that the parser ensured that one is a
 	 * superset of the other.
 	 */
+	root->query_uniquekeys = NIL;
+
 	if (root->group_pathkeys)
+	{
 		root->query_pathkeys = root->group_pathkeys;
+
+		if (!root->parse->hasAggs)
+			root->query_uniquekeys = build_uniquekeys(root, qp_extra->groupClause);
+	}
 	else if (root->window_pathkeys)
 		root->query_pathkeys = root->window_pathkeys;
 	else if (list_length(root->distinct_pathkeys) >
 			 list_length(root->sort_pathkeys))
+	{
 		root->query_pathkeys = root->distinct_pathkeys;
+		root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else if (root->sort_pathkeys)
+	{
 		root->query_pathkeys = root->sort_pathkeys;
+
+		if (root->distinct_pathkeys)
+			root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else
 		root->query_pathkeys = NIL;
 }
@@ -6222,7 +6237,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
 
 	/* Estimate the cost of index scan */
 	indexScanPath = create_index_path(root, indexInfo,
-									  NIL, NIL, NIL, NIL,
+									  NIL, NIL, NIL, NIL, NIL,
 									  ForwardScanDirection, false,
 									  NULL, 1.0, false);
 
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index e6d08aede5..a006dbbe9c 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -940,6 +940,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = parallel_workers;
 	pathnode->pathkeys = NIL;	/* seqscan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_seqscan(pathnode, root, rel, pathnode->param_info);
 
@@ -964,6 +965,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* samplescan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_samplescan(pathnode, root, rel, pathnode->param_info);
 
@@ -1000,6 +1002,7 @@ create_index_path(PlannerInfo *root,
 				  List *indexorderbys,
 				  List *indexorderbycols,
 				  List *pathkeys,
+				  List *uniquekeys,
 				  ScanDirection indexscandir,
 				  bool indexonly,
 				  Relids required_outer,
@@ -1018,6 +1021,7 @@ create_index_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = 0;
 	pathnode->path.pathkeys = pathkeys;
+	pathnode->path.uniquekeys = uniquekeys;
 
 	pathnode->indexinfo = index;
 	pathnode->indexclauses = indexclauses;
@@ -1061,6 +1065,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = parallel_degree;
 	pathnode->path.pathkeys = NIL;	/* always unordered */
+	pathnode->path.uniquekeys = NIL;
 
 	pathnode->bitmapqual = bitmapqual;
 
@@ -1922,6 +1927,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = pathkeys;
+	pathnode->uniquekeys = NIL;
 
 	cost_functionscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1948,6 +1954,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1974,6 +1981,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_valuesscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1999,6 +2007,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* XXX for now, result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2025,6 +2034,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2051,6 +2061,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_resultscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2077,6 +2088,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	/* Cost is the same as for a regular CTE scan */
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index baced7eec0..a1511b46ea 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 	T_EquivalenceMember,
 	T_PathKey,
 	T_PathTarget,
+	T_UniqueKey,
 	T_RestrictInfo,
 	T_IndexClause,
 	T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 3d3be197e0..4e329f0fb5 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -269,6 +269,8 @@ struct PlannerInfo
 
 	List	   *canon_pathkeys; /* list of "canonical" PathKeys */
 
+	List	   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 	List	   *left_join_clauses;	/* list of RestrictInfos for mergejoinable
 									 * outer join clauses w/nonnullable var on
 									 * left */
@@ -297,6 +299,8 @@ struct PlannerInfo
 
 	List	   *query_pathkeys; /* desired pathkeys for query_planner() */
 
+	List	   *query_uniquekeys; /* unique keys used for the query */
+
 	List	   *group_pathkeys; /* groupClause pathkeys, if any */
 	List	   *window_pathkeys;	/* pathkeys of bottom window, if any */
 	List	   *distinct_pathkeys;	/* distinctClause pathkeys, if any */
@@ -1077,6 +1081,15 @@ typedef struct ParamPathInfo
 	List	   *ppi_clauses;	/* join clauses available from outer rels */
 } ParamPathInfo;
 
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+	NodeTag		type;
+
+	EquivalenceClass *eq_clause;	/* equivalence class */
+} UniqueKey;
 
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some other
@@ -1106,6 +1119,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1129,6 +1145,8 @@ typedef struct Path
 
 	List	   *pathkeys;		/* sort ordering of path's output */
 	/* pathkeys is a List of PathKey nodes; see above */
+
+	List	   *uniquekeys;	/* the unique keys, or NIL if none */
 } Path;
 
 /* Macro for extracting a path's parameterization relids; beware double eval */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index 6126b491bf..006248bfb5 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
 
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index e450fe112a..f75ff6f323 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -44,6 +44,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 									List *indexorderbys,
 									List *indexorderbycols,
 									List *pathkeys,
+									List *uniquekeys,
 									ScanDirection indexscandir,
 									bool indexonly,
 									Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index 9ab73bd20c..5b6be383b3 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 										   List *sortclauses,
 										   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+										  List *sortclauses,
+										  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 											RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *root,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 									List *live_childrels);
 
+/*
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif							/* PATHS_H */
-- 
2.21.1


--------------51B4E6A19513F7FD264CD0FF
Content-Type: text/x-patch; charset=UTF-8;
 name="v31-0002-Index-skip-scan.patch"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
 filename="v31-0002-Index-skip-scan.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH v33 1/2] Unique key
@ 2020-03-24 16:04  Dmitrii Dolgov <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: Dmitrii Dolgov @ 2020-03-24 16:04 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   3 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/indxpath.c  |  41 ++++++++
 src/backend/optimizer/path/pathkeys.c  |  71 +++++++++++--
 src/backend/optimizer/path/uniquekey.c | 136 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planagg.c   |   1 +
 src/backend/optimizer/plan/planmain.c  |   1 +
 src/backend/optimizer/plan/planner.c   |  37 ++++++-
 src/backend/optimizer/util/pathnode.c  |  46 +++++++--
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  19 ++++
 src/include/nodes/print.h              |   1 +
 src/include/optimizer/pathnode.h       |   2 +
 src/include/optimizer/paths.h          |  11 ++
 16 files changed, 408 insertions(+), 23 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index d76fae44b8..16083e7a7e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1723,6 +1723,7 @@ _outPathInfo(StringInfo str, const Path *node)
 	WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 	WRITE_FLOAT_FIELD(total_cost, "%.2f");
 	WRITE_NODE_FIELD(pathkeys);
+	WRITE_NODE_FIELD(uniquekeys);
 }
 
 /*
@@ -2205,6 +2206,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(eq_classes);
 	WRITE_BOOL_FIELD(ec_merging_done);
 	WRITE_NODE_FIELD(canon_pathkeys);
+	WRITE_NODE_FIELD(canon_uniquekeys);
 	WRITE_NODE_FIELD(left_join_clauses);
 	WRITE_NODE_FIELD(right_join_clauses);
 	WRITE_NODE_FIELD(full_join_clauses);
@@ -2214,6 +2216,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(placeholder_list);
 	WRITE_NODE_FIELD(fkey_list);
 	WRITE_NODE_FIELD(query_pathkeys);
+	WRITE_NODE_FIELD(query_uniquekeys);
 	WRITE_NODE_FIELD(group_pathkeys);
 	WRITE_NODE_FIELD(window_pathkeys);
 	WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2401,6 +2404,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 	WRITE_BOOL_FIELD(pk_nulls_first);
 }
 
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+	WRITE_NODE_TYPE("UNIQUEKEY");
+
+	WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4092,6 +4103,9 @@ outNode(StringInfo str, const void *obj)
 			case T_PathKey:
 				_outPathKey(str, obj);
 				break;
+			case T_UniqueKey:
+				_outUniqueKey(str, obj);
+				break;
 			case T_PathTarget:
 				_outPathTarget(str, obj);
 				break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 42476724d8..d286b34544 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtable)
 	printf(")\n");
 }
 
+/*
+ * print_uniquekeys -
+ *	  uniquekeys list of UniqueKeys
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+	ListCell   *l;
+
+	printf("(");
+	foreach(l, uniquekeys)
+	{
+		UniqueKey *unique_key = (UniqueKey *) lfirst(l);
+		EquivalenceClass *eclass = (EquivalenceClass *) unique_key->eq_clause;
+		ListCell   *k;
+		bool		first = true;
+
+		/* chase up */
+		while (eclass->ec_merged)
+			eclass = eclass->ec_merged;
+
+		printf("(");
+		foreach(k, eclass->ec_members)
+		{
+			EquivalenceMember *mem = (EquivalenceMember *) lfirst(k);
+
+			if (first)
+				first = false;
+			else
+				printf(", ");
+			print_expr((Node *) mem->em_expr, rtable);
+		}
+		printf(")");
+		if (lnext(uniquekeys, l))
+			printf(", ");
+	}
+	printf(")\n");
+}
+
 /*
  * print_tl
  *	  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/path/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS = \
 	joinpath.o \
 	joinrels.o \
 	pathkeys.o \
-	tidpath.o
+	tidpath.o \
+	uniquekey.o
 
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 8286d9cf34..bbc13e6141 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent)
 		print_pathkeys(path->pathkeys, root->parse->rtable);
 	}
 
+	if (path->uniquekeys)
+	{
+		for (i = 0; i < indent; i++)
+			printf("\t");
+		printf("  uniquekeys: ");
+		print_uniquekeys(path->uniquekeys, root->parse->rtable);
+	}
+
 	if (join)
 	{
 		JoinPath   *jp = (JoinPath *) path;
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 2a50272da6..363f5349f1 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 									   EquivalenceClass *ec, EquivalenceMember *em,
 									   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
 
 
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	List	   *orderbyclausecols;
 	List	   *index_pathkeys;
 	List	   *useful_pathkeys;
+	List	   *useful_uniquekeys = NIL;
 	bool		found_lower_saop_clause;
 	bool		pathkeys_possibly_useful;
 	bool		index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	if (index_clauses != NIL || useful_pathkeys != NIL || useful_predicate ||
 		index_only_scan)
 	{
+		if (has_useful_uniquekeys(root))
+			useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 		ipath = create_index_path(root, index,
 								  index_clauses,
 								  orderbyclauses,
 								  orderbyclausecols,
 								  useful_pathkeys,
+								  useful_uniquekeys,
 								  index_is_ordered ?
 								  ForwardScanDirection :
 								  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 									  orderbyclauses,
 									  orderbyclausecols,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  index_is_ordered ?
 									  ForwardScanDirection :
 									  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 													index_pathkeys);
 		if (useful_pathkeys != NIL)
 		{
+			if (has_useful_uniquekeys(root))
+				useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 			ipath = create_index_path(root, index,
 									  index_clauses,
 									  NIL,
 									  NIL,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  BackwardScanDirection,
 									  index_only_scan,
 									  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 										  NIL,
 										  NIL,
 										  useful_pathkeys,
+										  useful_uniquekeys,
 										  BackwardScanDirection,
 										  index_only_scan,
 										  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 	return clause;
 }
 
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+	ListCell *lc;
+
+	if (pathkeys)
+	{
+		List *uniquekeys = NIL;
+		foreach(lc, pathkeys)
+		{
+			UniqueKey *unique_key;
+			PathKey *pk = (PathKey *) lfirst(lc);
+			EquivalenceClass *ec = (EquivalenceClass *) pk->pk_eclass;
+
+			unique_key = makeNode(UniqueKey);
+			unique_key->eq_clause = ec;
+
+			uniquekeys = lappend(uniquekeys, unique_key);
+		}
+
+		if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+			return uniquekeys;
+	}
+
+	return NIL;
+}
 
 /****************************************************************************
  *				----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS	----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/path/pathkeys.c
index 71b9d42c99..054df9a617 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
 
 
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 											 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 	return pk;
 }
 
+/*
+ * pathkey_is_unique
+ *	   Checks if the new pathkey's equivalence class is the same as that of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
+	ListCell   *lc;
+
+	/* If same EC already is already in the list, then not unique */
+	foreach(lc, pathkeys)
+	{
+		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
+
+		if (new_ec == old_pathkey->pk_eclass)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * pathkey_is_redundant
  *	   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
-	ListCell   *lc;
 
 	/* Check for EC containing a constant --- unconditionally redundant */
 	if (EC_MUST_BE_REDUNDANT(new_ec))
 		return true;
 
-	/* If same EC already used in list, then redundant */
-	foreach(lc, pathkeys)
-	{
-		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
-
-		if (new_ec == old_pathkey->pk_eclass)
-			return true;
-	}
-
-	return false;
+	return !pathkey_is_unique(new_pathkey, pathkeys);
 }
 
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 	return pathkeys;
 }
 
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *		Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+							 List *sortclauses,
+							 List *tlist)
+{
+	List	   *pathkeys = NIL;
+	ListCell   *l;
+
+	foreach(l, sortclauses)
+	{
+		SortGroupClause *sortcl = (SortGroupClause *) lfirst(l);
+		Expr	   *sortkey;
+		PathKey    *pathkey;
+
+		sortkey = (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+		Assert(OidIsValid(sortcl->sortop));
+		pathkey = make_pathkey_from_sortop(root,
+										   sortkey,
+										   root->nullable_baserels,
+										   sortcl->sortop,
+										   sortcl->nulls_first,
+										   sortcl->tleSortGroupRef,
+										   true);
+
+		if (pathkey_is_unique(pathkey, pathkeys))
+			pathkeys = lappend(pathkeys, pathkey);
+	}
+
+	return pathkeys;
+}
+
 /****************************************************************************
  *		PATHKEYS AND MERGECLAUSES
  ****************************************************************************/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer/path/uniquekey.c
new file mode 100644
index 0000000000..c421401d0f
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,136 @@
+/*-------------------------------------------------------------------------
+ *
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  src/backend/optimizer/path/uniquekey.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceClass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+	List *result = NIL;
+	List *sortkeys;
+	ListCell *l;
+
+	sortkeys = make_pathkeys_for_uniquekeys(root,
+											sortclauses,
+											root->processed_tlist);
+
+	/* Create a uniquekey and add it to the list */
+	foreach(l, sortkeys)
+	{
+		PathKey    *pathkey = (PathKey *) lfirst(l);
+		EquivalenceClass *ec = pathkey->pk_eclass;
+		UniqueKey *unique_key = make_canonical_uniquekey(root, ec);
+
+		result = lappend(result, unique_key);
+	}
+
+	return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *	  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+	ListCell   *key1,
+			   *key2;
+
+	foreach(key2, keys2)
+	{
+		bool found = false;
+		UniqueKey  *uniquekey2 = (UniqueKey *) lfirst(key2);
+
+		foreach(key1, keys1)
+		{
+			UniqueKey  *uniquekey1 = (UniqueKey *) lfirst(key1);
+
+			if (uniquekey1->eq_clause == uniquekey2->eq_clause)
+				return true;
+		}
+
+		if (!found)
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *		Detect whether the planner could have any uniquekeys that are
+ *		useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+	if (root->query_uniquekeys != NIL)
+		return true;	/* there are some */
+	return false;		/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *	  Given the parameters for a UniqueKey, find any pre-existing matching
+ *	  uniquekey in the query's list of "canonical" uniquekeys.  Make a new
+ *	  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instead,
+ * equivclass.c will complain if a merge occurs after root->canon_uniquekeys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+						 EquivalenceClass *eclass)
+{
+	UniqueKey  *uk;
+	ListCell   *lc;
+	MemoryContext oldcontext;
+
+	/* The passed eclass might be non-canonical, so chase up to the top */
+	while (eclass->ec_merged)
+		eclass = eclass->ec_merged;
+
+	foreach(lc, root->canon_uniquekeys)
+	{
+		uk = (UniqueKey *) lfirst(lc);
+		if (eclass == uk->eq_clause)
+			return uk;
+	}
+
+	/*
+	 * Be sure canonical uniquekeys are allocated in the main planning context.
+	 * Not an issue in normal planning, but it is for GEQO.
+	 */
+	oldcontext = MemoryContextSwitchTo(root->planner_cxt);
+
+	uk = makeNode(UniqueKey);
+	uk->eq_clause = eclass;
+
+	root->canon_uniquekeys = lappend(root->canon_uniquekeys, uk);
+
+	MemoryContextSwitchTo(oldcontext);
+
+	return uk;
+}
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/plan/planagg.c
index 8634940efc..dd64775d8f 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -511,6 +511,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 									  root->parse->targetList);
 
 	root->query_pathkeys = root->sort_pathkeys;
+	root->query_uniquekeys = NIL;
 }
 
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/plan/planmain.c
index 62dfc6d44a..3a372af91b 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 	root->join_rel_level = NULL;
 	root->join_cur_level = 0;
 	root->canon_pathkeys = NIL;
+	root->canon_uniquekeys = NIL;
 	root->left_join_clauses = NIL;
 	root->right_join_clauses = NIL;
 	root->full_join_clauses = NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index d6f2153593..a7de8476d9 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3657,15 +3657,30 @@ standard_qp_callback(PlannerInfo *root, void *extra)
 	 * much easier, since we know that the parser ensured that one is a
 	 * superset of the other.
 	 */
+	root->query_uniquekeys = NIL;
+
 	if (root->group_pathkeys)
+	{
 		root->query_pathkeys = root->group_pathkeys;
+
+		if (!root->parse->hasAggs)
+			root->query_uniquekeys = build_uniquekeys(root, qp_extra->groupClause);
+	}
 	else if (root->window_pathkeys)
 		root->query_pathkeys = root->window_pathkeys;
 	else if (list_length(root->distinct_pathkeys) >
 			 list_length(root->sort_pathkeys))
+	{
 		root->query_pathkeys = root->distinct_pathkeys;
+		root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else if (root->sort_pathkeys)
+	{
 		root->query_pathkeys = root->sort_pathkeys;
+
+		if (root->distinct_pathkeys)
+			root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else
 		root->query_pathkeys = NIL;
 }
@@ -6222,7 +6237,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
 
 	/* Estimate the cost of index scan */
 	indexScanPath = create_index_path(root, indexInfo,
-									  NIL, NIL, NIL, NIL,
+									  NIL, NIL, NIL, NIL, NIL,
 									  ForwardScanDirection, false,
 									  NULL, 1.0, false);
 
@@ -7107,6 +7122,26 @@ apply_scanjoin_target_to_paths(PlannerInfo *root,
 		}
 	}
 
+	foreach(lc, rel->unique_pathlist)
+	{
+		Path	   *subpath = (Path *) lfirst(lc);
+
+		/* Shouldn't have any parameterized paths anymore */
+		Assert(subpath->param_info == NULL);
+
+		if (tlist_same_exprs)
+			subpath->pathtarget->sortgrouprefs =
+				scanjoin_target->sortgrouprefs;
+		else
+		{
+			Path	   *newpath;
+
+			newpath = (Path *) create_projection_path(root, rel, subpath,
+													  scanjoin_target);
+			lfirst(lc) = newpath;
+		}
+	}
+
 	/*
 	 * Now, if final scan/join target contains SRFs, insert ProjectSetPath(s)
 	 * atop each existing path.  (Note that this function doesn't look at the
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index e6d08aede5..a4dfafbb59 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -361,9 +361,9 @@ set_cheapest(RelOptInfo *parent_rel)
 }
 
 /*
- * add_path
+ * add_path_to
  *	  Consider a potential implementation path for the specified parent rel,
- *	  and add it to the rel's pathlist if it is worthy of consideration.
+ *	  and add it to the specified pathlist if it is worthy of consideration.
  *	  A path is worthy if it has a better sort order (better pathkeys) or
  *	  cheaper cost (on either dimension), or generates fewer rows, than any
  *	  existing path that has the same or superset parameterization rels.
@@ -416,10 +416,10 @@ set_cheapest(RelOptInfo *parent_rel)
  * 'parent_rel' is the relation entry to which the path corresponds.
  * 'new_path' is a potential path for parent_rel.
  *
- * Returns nothing, but modifies parent_rel->pathlist.
+ * Returns modified pathlist.
  */
-void
-add_path(RelOptInfo *parent_rel, Path *new_path)
+static List *
+add_path_to(RelOptInfo *parent_rel, List *pathlist, Path *new_path)
 {
 	bool		accept_new = true;	/* unless we find a superior old path */
 	int			insert_at = 0;	/* where to insert new item */
@@ -440,7 +440,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 	 * for more than one old path to be tossed out because new_path dominates
 	 * it.
 	 */
-	foreach(p1, parent_rel->pathlist)
+	foreach(p1, pathlist)
 	{
 		Path	   *old_path = (Path *) lfirst(p1);
 		bool		remove_old = false; /* unless new proves superior */
@@ -584,8 +584,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 		 */
 		if (remove_old)
 		{
-			parent_rel->pathlist = foreach_delete_current(parent_rel->pathlist,
-														  p1);
+			pathlist = foreach_delete_current(pathlist, p1);
 
 			/*
 			 * Delete the data pointed-to by the deleted cell, if possible
@@ -612,8 +611,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 	if (accept_new)
 	{
 		/* Accept the new path: insert it at proper place in pathlist */
-		parent_rel->pathlist =
-			list_insert_nth(parent_rel->pathlist, insert_at, new_path);
+		pathlist = list_insert_nth(pathlist, insert_at, new_path);
 	}
 	else
 	{
@@ -621,6 +619,15 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 		if (!IsA(new_path, IndexPath))
 			pfree(new_path);
 	}
+
+	return pathlist;
+}
+
+void
+add_path(RelOptInfo *parent_rel, Path *new_path)
+{
+	parent_rel->pathlist = add_path_to(parent_rel,
+									   parent_rel->pathlist, new_path);
 }
 
 /*
@@ -915,6 +922,13 @@ add_partial_path_precheck(RelOptInfo *parent_rel, Cost total_cost,
 	return true;
 }
 
+void
+add_unique_path(RelOptInfo *parent_rel, Path *new_path)
+{
+	parent_rel->unique_pathlist = add_path_to(parent_rel,
+											  parent_rel->unique_pathlist,
+											  new_path);
+}
 
 /*****************************************************************************
  *		PATH NODE CREATION ROUTINES
@@ -940,6 +954,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = parallel_workers;
 	pathnode->pathkeys = NIL;	/* seqscan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_seqscan(pathnode, root, rel, pathnode->param_info);
 
@@ -964,6 +979,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* samplescan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_samplescan(pathnode, root, rel, pathnode->param_info);
 
@@ -1000,6 +1016,7 @@ create_index_path(PlannerInfo *root,
 				  List *indexorderbys,
 				  List *indexorderbycols,
 				  List *pathkeys,
+				  List *uniquekeys,
 				  ScanDirection indexscandir,
 				  bool indexonly,
 				  Relids required_outer,
@@ -1018,6 +1035,7 @@ create_index_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = 0;
 	pathnode->path.pathkeys = pathkeys;
+	pathnode->path.uniquekeys = uniquekeys;
 
 	pathnode->indexinfo = index;
 	pathnode->indexclauses = indexclauses;
@@ -1061,6 +1079,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = parallel_degree;
 	pathnode->path.pathkeys = NIL;	/* always unordered */
+	pathnode->path.uniquekeys = NIL;
 
 	pathnode->bitmapqual = bitmapqual;
 
@@ -1922,6 +1941,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = pathkeys;
+	pathnode->uniquekeys = NIL;
 
 	cost_functionscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1948,6 +1968,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1974,6 +1995,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_valuesscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1999,6 +2021,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* XXX for now, result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2025,6 +2048,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2051,6 +2075,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_resultscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2077,6 +2102,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	/* Cost is the same as for a regular CTE scan */
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index baced7eec0..a1511b46ea 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 	T_EquivalenceMember,
 	T_PathKey,
 	T_PathTarget,
+	T_UniqueKey,
 	T_RestrictInfo,
 	T_IndexClause,
 	T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 3d3be197e0..0de27f0ef3 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -269,6 +269,8 @@ struct PlannerInfo
 
 	List	   *canon_pathkeys; /* list of "canonical" PathKeys */
 
+	List	   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 	List	   *left_join_clauses;	/* list of RestrictInfos for mergejoinable
 									 * outer join clauses w/nonnullable var on
 									 * left */
@@ -297,6 +299,8 @@ struct PlannerInfo
 
 	List	   *query_pathkeys; /* desired pathkeys for query_planner() */
 
+	List	   *query_uniquekeys; /* unique keys used for the query */
+
 	List	   *group_pathkeys; /* groupClause pathkeys, if any */
 	List	   *window_pathkeys;	/* pathkeys of bottom window, if any */
 	List	   *distinct_pathkeys;	/* distinctClause pathkeys, if any */
@@ -657,6 +661,7 @@ typedef struct RelOptInfo
 	List	   *pathlist;		/* Path structures */
 	List	   *ppilist;		/* ParamPathInfos used in pathlist */
 	List	   *partial_pathlist;	/* partial Paths */
+	List	   *unique_pathlist;	/* unique Paths */
 	struct Path *cheapest_startup_path;
 	struct Path *cheapest_total_path;
 	struct Path *cheapest_unique_path;
@@ -1077,6 +1082,15 @@ typedef struct ParamPathInfo
 	List	   *ppi_clauses;	/* join clauses available from outer rels */
 } ParamPathInfo;
 
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+	NodeTag		type;
+
+	EquivalenceClass *eq_clause;	/* equivalence class */
+} UniqueKey;
 
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some other
@@ -1106,6 +1120,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1129,6 +1146,8 @@ typedef struct Path
 
 	List	   *pathkeys;		/* sort ordering of path's output */
 	/* pathkeys is a List of PathKey nodes; see above */
+
+	List	   *uniquekeys;	/* the unique keys, or NIL if none */
 } Path;
 
 /* Macro for extracting a path's parameterization relids; beware double eval */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index 6126b491bf..006248bfb5 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
 
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index e450fe112a..fd25997af5 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -34,6 +34,7 @@ extern void add_partial_path(RelOptInfo *parent_rel, Path *new_path);
 extern bool add_partial_path_precheck(RelOptInfo *parent_rel,
 									  Cost total_cost, List *pathkeys);
 
+extern void add_unique_path(RelOptInfo *parent_rel, Path *new_path);
 extern Path *create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 								 Relids required_outer, int parallel_workers);
 extern Path *create_samplescan_path(PlannerInfo *root, RelOptInfo *rel,
@@ -44,6 +45,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 									List *indexorderbys,
 									List *indexorderbycols,
 									List *pathkeys,
+									List *uniquekeys,
 									ScanDirection indexscandir,
 									bool indexonly,
 									Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index 9ab73bd20c..5b6be383b3 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 										   List *sortclauses,
 										   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+										  List *sortclauses,
+										  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 											RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *root,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 									List *live_childrels);
 
+/*
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif							/* PATHS_H */
-- 
2.21.0


--y3rsfeqxxd2vddcp
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v33-0002-Index-skip-scan.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH v33 1/2] Unique key
@ 2020-03-24 16:04  Dmitrii Dolgov <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: Dmitrii Dolgov @ 2020-03-24 16:04 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c          | 14 ++++++
 src/backend/nodes/print.c             | 39 +++++++++++++++
 src/backend/optimizer/path/Makefile   |  3 +-
 src/backend/optimizer/path/allpaths.c |  8 +++
 src/backend/optimizer/path/indxpath.c | 41 ++++++++++++++++
 src/backend/optimizer/path/pathkeys.c | 71 ++++++++++++++++++++++-----
 src/backend/optimizer/plan/planagg.c  |  1 +
 src/backend/optimizer/plan/planmain.c |  1 +
 src/backend/optimizer/plan/planner.c  | 37 +++++++++++++-
 src/backend/optimizer/util/pathnode.c | 46 +++++++++++++----
 src/include/nodes/nodes.h             |  1 +
 src/include/nodes/pathnodes.h         | 19 +++++++
 src/include/nodes/print.h             |  1 +
 src/include/optimizer/pathnode.h      |  2 +
 src/include/optimizer/paths.h         | 11 +++++
 15 files changed, 272 insertions(+), 23 deletions(-)

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index d76fae44b8..16083e7a7e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1723,6 +1723,7 @@ _outPathInfo(StringInfo str, const Path *node)
 	WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 	WRITE_FLOAT_FIELD(total_cost, "%.2f");
 	WRITE_NODE_FIELD(pathkeys);
+	WRITE_NODE_FIELD(uniquekeys);
 }
 
 /*
@@ -2205,6 +2206,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(eq_classes);
 	WRITE_BOOL_FIELD(ec_merging_done);
 	WRITE_NODE_FIELD(canon_pathkeys);
+	WRITE_NODE_FIELD(canon_uniquekeys);
 	WRITE_NODE_FIELD(left_join_clauses);
 	WRITE_NODE_FIELD(right_join_clauses);
 	WRITE_NODE_FIELD(full_join_clauses);
@@ -2214,6 +2216,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(placeholder_list);
 	WRITE_NODE_FIELD(fkey_list);
 	WRITE_NODE_FIELD(query_pathkeys);
+	WRITE_NODE_FIELD(query_uniquekeys);
 	WRITE_NODE_FIELD(group_pathkeys);
 	WRITE_NODE_FIELD(window_pathkeys);
 	WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2401,6 +2404,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 	WRITE_BOOL_FIELD(pk_nulls_first);
 }
 
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+	WRITE_NODE_TYPE("UNIQUEKEY");
+
+	WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4092,6 +4103,9 @@ outNode(StringInfo str, const void *obj)
 			case T_PathKey:
 				_outPathKey(str, obj);
 				break;
+			case T_UniqueKey:
+				_outUniqueKey(str, obj);
+				break;
 			case T_PathTarget:
 				_outPathTarget(str, obj);
 				break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 42476724d8..d286b34544 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtable)
 	printf(")\n");
 }
 
+/*
+ * print_uniquekeys -
+ *	  uniquekeys list of UniqueKeys
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+	ListCell   *l;
+
+	printf("(");
+	foreach(l, uniquekeys)
+	{
+		UniqueKey *unique_key = (UniqueKey *) lfirst(l);
+		EquivalenceClass *eclass = (EquivalenceClass *) unique_key->eq_clause;
+		ListCell   *k;
+		bool		first = true;
+
+		/* chase up */
+		while (eclass->ec_merged)
+			eclass = eclass->ec_merged;
+
+		printf("(");
+		foreach(k, eclass->ec_members)
+		{
+			EquivalenceMember *mem = (EquivalenceMember *) lfirst(k);
+
+			if (first)
+				first = false;
+			else
+				printf(", ");
+			print_expr((Node *) mem->em_expr, rtable);
+		}
+		printf(")");
+		if (lnext(uniquekeys, l))
+			printf(", ");
+	}
+	printf(")\n");
+}
+
 /*
  * print_tl
  *	  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/path/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS = \
 	joinpath.o \
 	joinrels.o \
 	pathkeys.o \
-	tidpath.o
+	tidpath.o \
+	uniquekey.o
 
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 8286d9cf34..bbc13e6141 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent)
 		print_pathkeys(path->pathkeys, root->parse->rtable);
 	}
 
+	if (path->uniquekeys)
+	{
+		for (i = 0; i < indent; i++)
+			printf("\t");
+		printf("  uniquekeys: ");
+		print_uniquekeys(path->uniquekeys, root->parse->rtable);
+	}
+
 	if (join)
 	{
 		JoinPath   *jp = (JoinPath *) path;
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 2a50272da6..363f5349f1 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 									   EquivalenceClass *ec, EquivalenceMember *em,
 									   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
 
 
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	List	   *orderbyclausecols;
 	List	   *index_pathkeys;
 	List	   *useful_pathkeys;
+	List	   *useful_uniquekeys = NIL;
 	bool		found_lower_saop_clause;
 	bool		pathkeys_possibly_useful;
 	bool		index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	if (index_clauses != NIL || useful_pathkeys != NIL || useful_predicate ||
 		index_only_scan)
 	{
+		if (has_useful_uniquekeys(root))
+			useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 		ipath = create_index_path(root, index,
 								  index_clauses,
 								  orderbyclauses,
 								  orderbyclausecols,
 								  useful_pathkeys,
+								  useful_uniquekeys,
 								  index_is_ordered ?
 								  ForwardScanDirection :
 								  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 									  orderbyclauses,
 									  orderbyclausecols,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  index_is_ordered ?
 									  ForwardScanDirection :
 									  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 													index_pathkeys);
 		if (useful_pathkeys != NIL)
 		{
+			if (has_useful_uniquekeys(root))
+				useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 			ipath = create_index_path(root, index,
 									  index_clauses,
 									  NIL,
 									  NIL,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  BackwardScanDirection,
 									  index_only_scan,
 									  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 										  NIL,
 										  NIL,
 										  useful_pathkeys,
+										  useful_uniquekeys,
 										  BackwardScanDirection,
 										  index_only_scan,
 										  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 	return clause;
 }
 
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+	ListCell *lc;
+
+	if (pathkeys)
+	{
+		List *uniquekeys = NIL;
+		foreach(lc, pathkeys)
+		{
+			UniqueKey *unique_key;
+			PathKey *pk = (PathKey *) lfirst(lc);
+			EquivalenceClass *ec = (EquivalenceClass *) pk->pk_eclass;
+
+			unique_key = makeNode(UniqueKey);
+			unique_key->eq_clause = ec;
+
+			uniquekeys = lappend(uniquekeys, unique_key);
+		}
+
+		if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+			return uniquekeys;
+	}
+
+	return NIL;
+}
 
 /****************************************************************************
  *				----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS	----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/path/pathkeys.c
index 71b9d42c99..054df9a617 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
 
 
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 											 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 	return pk;
 }
 
+/*
+ * pathkey_is_unique
+ *	   Checks if the new pathkey's equivalence class is the same as that of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
+	ListCell   *lc;
+
+	/* If same EC already is already in the list, then not unique */
+	foreach(lc, pathkeys)
+	{
+		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
+
+		if (new_ec == old_pathkey->pk_eclass)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * pathkey_is_redundant
  *	   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
-	ListCell   *lc;
 
 	/* Check for EC containing a constant --- unconditionally redundant */
 	if (EC_MUST_BE_REDUNDANT(new_ec))
 		return true;
 
-	/* If same EC already used in list, then redundant */
-	foreach(lc, pathkeys)
-	{
-		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
-
-		if (new_ec == old_pathkey->pk_eclass)
-			return true;
-	}
-
-	return false;
+	return !pathkey_is_unique(new_pathkey, pathkeys);
 }
 
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 	return pathkeys;
 }
 
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *		Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+							 List *sortclauses,
+							 List *tlist)
+{
+	List	   *pathkeys = NIL;
+	ListCell   *l;
+
+	foreach(l, sortclauses)
+	{
+		SortGroupClause *sortcl = (SortGroupClause *) lfirst(l);
+		Expr	   *sortkey;
+		PathKey    *pathkey;
+
+		sortkey = (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+		Assert(OidIsValid(sortcl->sortop));
+		pathkey = make_pathkey_from_sortop(root,
+										   sortkey,
+										   root->nullable_baserels,
+										   sortcl->sortop,
+										   sortcl->nulls_first,
+										   sortcl->tleSortGroupRef,
+										   true);
+
+		if (pathkey_is_unique(pathkey, pathkeys))
+			pathkeys = lappend(pathkeys, pathkey);
+	}
+
+	return pathkeys;
+}
+
 /****************************************************************************
  *		PATHKEYS AND MERGECLAUSES
  ****************************************************************************/
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/plan/planagg.c
index 8634940efc..dd64775d8f 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -511,6 +511,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 									  root->parse->targetList);
 
 	root->query_pathkeys = root->sort_pathkeys;
+	root->query_uniquekeys = NIL;
 }
 
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/plan/planmain.c
index 62dfc6d44a..3a372af91b 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 	root->join_rel_level = NULL;
 	root->join_cur_level = 0;
 	root->canon_pathkeys = NIL;
+	root->canon_uniquekeys = NIL;
 	root->left_join_clauses = NIL;
 	root->right_join_clauses = NIL;
 	root->full_join_clauses = NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index d6f2153593..a7de8476d9 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3657,15 +3657,30 @@ standard_qp_callback(PlannerInfo *root, void *extra)
 	 * much easier, since we know that the parser ensured that one is a
 	 * superset of the other.
 	 */
+	root->query_uniquekeys = NIL;
+
 	if (root->group_pathkeys)
+	{
 		root->query_pathkeys = root->group_pathkeys;
+
+		if (!root->parse->hasAggs)
+			root->query_uniquekeys = build_uniquekeys(root, qp_extra->groupClause);
+	}
 	else if (root->window_pathkeys)
 		root->query_pathkeys = root->window_pathkeys;
 	else if (list_length(root->distinct_pathkeys) >
 			 list_length(root->sort_pathkeys))
+	{
 		root->query_pathkeys = root->distinct_pathkeys;
+		root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else if (root->sort_pathkeys)
+	{
 		root->query_pathkeys = root->sort_pathkeys;
+
+		if (root->distinct_pathkeys)
+			root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else
 		root->query_pathkeys = NIL;
 }
@@ -6222,7 +6237,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
 
 	/* Estimate the cost of index scan */
 	indexScanPath = create_index_path(root, indexInfo,
-									  NIL, NIL, NIL, NIL,
+									  NIL, NIL, NIL, NIL, NIL,
 									  ForwardScanDirection, false,
 									  NULL, 1.0, false);
 
@@ -7107,6 +7122,26 @@ apply_scanjoin_target_to_paths(PlannerInfo *root,
 		}
 	}
 
+	foreach(lc, rel->unique_pathlist)
+	{
+		Path	   *subpath = (Path *) lfirst(lc);
+
+		/* Shouldn't have any parameterized paths anymore */
+		Assert(subpath->param_info == NULL);
+
+		if (tlist_same_exprs)
+			subpath->pathtarget->sortgrouprefs =
+				scanjoin_target->sortgrouprefs;
+		else
+		{
+			Path	   *newpath;
+
+			newpath = (Path *) create_projection_path(root, rel, subpath,
+													  scanjoin_target);
+			lfirst(lc) = newpath;
+		}
+	}
+
 	/*
 	 * Now, if final scan/join target contains SRFs, insert ProjectSetPath(s)
 	 * atop each existing path.  (Note that this function doesn't look at the
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index e6d08aede5..a4dfafbb59 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -361,9 +361,9 @@ set_cheapest(RelOptInfo *parent_rel)
 }
 
 /*
- * add_path
+ * add_path_to
  *	  Consider a potential implementation path for the specified parent rel,
- *	  and add it to the rel's pathlist if it is worthy of consideration.
+ *	  and add it to the specified pathlist if it is worthy of consideration.
  *	  A path is worthy if it has a better sort order (better pathkeys) or
  *	  cheaper cost (on either dimension), or generates fewer rows, than any
  *	  existing path that has the same or superset parameterization rels.
@@ -416,10 +416,10 @@ set_cheapest(RelOptInfo *parent_rel)
  * 'parent_rel' is the relation entry to which the path corresponds.
  * 'new_path' is a potential path for parent_rel.
  *
- * Returns nothing, but modifies parent_rel->pathlist.
+ * Returns modified pathlist.
  */
-void
-add_path(RelOptInfo *parent_rel, Path *new_path)
+static List *
+add_path_to(RelOptInfo *parent_rel, List *pathlist, Path *new_path)
 {
 	bool		accept_new = true;	/* unless we find a superior old path */
 	int			insert_at = 0;	/* where to insert new item */
@@ -440,7 +440,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 	 * for more than one old path to be tossed out because new_path dominates
 	 * it.
 	 */
-	foreach(p1, parent_rel->pathlist)
+	foreach(p1, pathlist)
 	{
 		Path	   *old_path = (Path *) lfirst(p1);
 		bool		remove_old = false; /* unless new proves superior */
@@ -584,8 +584,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 		 */
 		if (remove_old)
 		{
-			parent_rel->pathlist = foreach_delete_current(parent_rel->pathlist,
-														  p1);
+			pathlist = foreach_delete_current(pathlist, p1);
 
 			/*
 			 * Delete the data pointed-to by the deleted cell, if possible
@@ -612,8 +611,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 	if (accept_new)
 	{
 		/* Accept the new path: insert it at proper place in pathlist */
-		parent_rel->pathlist =
-			list_insert_nth(parent_rel->pathlist, insert_at, new_path);
+		pathlist = list_insert_nth(pathlist, insert_at, new_path);
 	}
 	else
 	{
@@ -621,6 +619,15 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 		if (!IsA(new_path, IndexPath))
 			pfree(new_path);
 	}
+
+	return pathlist;
+}
+
+void
+add_path(RelOptInfo *parent_rel, Path *new_path)
+{
+	parent_rel->pathlist = add_path_to(parent_rel,
+									   parent_rel->pathlist, new_path);
 }
 
 /*
@@ -915,6 +922,13 @@ add_partial_path_precheck(RelOptInfo *parent_rel, Cost total_cost,
 	return true;
 }
 
+void
+add_unique_path(RelOptInfo *parent_rel, Path *new_path)
+{
+	parent_rel->unique_pathlist = add_path_to(parent_rel,
+											  parent_rel->unique_pathlist,
+											  new_path);
+}
 
 /*****************************************************************************
  *		PATH NODE CREATION ROUTINES
@@ -940,6 +954,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = parallel_workers;
 	pathnode->pathkeys = NIL;	/* seqscan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_seqscan(pathnode, root, rel, pathnode->param_info);
 
@@ -964,6 +979,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* samplescan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_samplescan(pathnode, root, rel, pathnode->param_info);
 
@@ -1000,6 +1016,7 @@ create_index_path(PlannerInfo *root,
 				  List *indexorderbys,
 				  List *indexorderbycols,
 				  List *pathkeys,
+				  List *uniquekeys,
 				  ScanDirection indexscandir,
 				  bool indexonly,
 				  Relids required_outer,
@@ -1018,6 +1035,7 @@ create_index_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = 0;
 	pathnode->path.pathkeys = pathkeys;
+	pathnode->path.uniquekeys = uniquekeys;
 
 	pathnode->indexinfo = index;
 	pathnode->indexclauses = indexclauses;
@@ -1061,6 +1079,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = parallel_degree;
 	pathnode->path.pathkeys = NIL;	/* always unordered */
+	pathnode->path.uniquekeys = NIL;
 
 	pathnode->bitmapqual = bitmapqual;
 
@@ -1922,6 +1941,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = pathkeys;
+	pathnode->uniquekeys = NIL;
 
 	cost_functionscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1948,6 +1968,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1974,6 +1995,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_valuesscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1999,6 +2021,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* XXX for now, result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2025,6 +2048,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2051,6 +2075,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_resultscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2077,6 +2102,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	/* Cost is the same as for a regular CTE scan */
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index baced7eec0..a1511b46ea 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 	T_EquivalenceMember,
 	T_PathKey,
 	T_PathTarget,
+	T_UniqueKey,
 	T_RestrictInfo,
 	T_IndexClause,
 	T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 3d3be197e0..0de27f0ef3 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -269,6 +269,8 @@ struct PlannerInfo
 
 	List	   *canon_pathkeys; /* list of "canonical" PathKeys */
 
+	List	   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 	List	   *left_join_clauses;	/* list of RestrictInfos for mergejoinable
 									 * outer join clauses w/nonnullable var on
 									 * left */
@@ -297,6 +299,8 @@ struct PlannerInfo
 
 	List	   *query_pathkeys; /* desired pathkeys for query_planner() */
 
+	List	   *query_uniquekeys; /* unique keys used for the query */
+
 	List	   *group_pathkeys; /* groupClause pathkeys, if any */
 	List	   *window_pathkeys;	/* pathkeys of bottom window, if any */
 	List	   *distinct_pathkeys;	/* distinctClause pathkeys, if any */
@@ -657,6 +661,7 @@ typedef struct RelOptInfo
 	List	   *pathlist;		/* Path structures */
 	List	   *ppilist;		/* ParamPathInfos used in pathlist */
 	List	   *partial_pathlist;	/* partial Paths */
+	List	   *unique_pathlist;	/* unique Paths */
 	struct Path *cheapest_startup_path;
 	struct Path *cheapest_total_path;
 	struct Path *cheapest_unique_path;
@@ -1077,6 +1082,15 @@ typedef struct ParamPathInfo
 	List	   *ppi_clauses;	/* join clauses available from outer rels */
 } ParamPathInfo;
 
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+	NodeTag		type;
+
+	EquivalenceClass *eq_clause;	/* equivalence class */
+} UniqueKey;
 
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some other
@@ -1106,6 +1120,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1129,6 +1146,8 @@ typedef struct Path
 
 	List	   *pathkeys;		/* sort ordering of path's output */
 	/* pathkeys is a List of PathKey nodes; see above */
+
+	List	   *uniquekeys;	/* the unique keys, or NIL if none */
 } Path;
 
 /* Macro for extracting a path's parameterization relids; beware double eval */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index 6126b491bf..006248bfb5 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
 
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index e450fe112a..fd25997af5 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -34,6 +34,7 @@ extern void add_partial_path(RelOptInfo *parent_rel, Path *new_path);
 extern bool add_partial_path_precheck(RelOptInfo *parent_rel,
 									  Cost total_cost, List *pathkeys);
 
+extern void add_unique_path(RelOptInfo *parent_rel, Path *new_path);
 extern Path *create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 								 Relids required_outer, int parallel_workers);
 extern Path *create_samplescan_path(PlannerInfo *root, RelOptInfo *rel,
@@ -44,6 +45,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 									List *indexorderbys,
 									List *indexorderbycols,
 									List *pathkeys,
+									List *uniquekeys,
 									ScanDirection indexscandir,
 									bool indexonly,
 									Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index 9ab73bd20c..5b6be383b3 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 										   List *sortclauses,
 										   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+										  List *sortclauses,
+										  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 											RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *root,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 									List *live_childrels);
 
+/*
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif							/* PATHS_H */
-- 
2.21.0


--qfje4fcsuvt2suqm
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v33-0002-Index-skip-scan.patch"



^ permalink  raw  reply  [nested|flat] 11+ messages in thread

* [PATCH v33 1/2] Unique key
@ 2020-03-24 16:04  Dmitrii Dolgov <[email protected]>
  0 siblings, 0 replies; 11+ messages in thread

From: Dmitrii Dolgov @ 2020-03-24 16:04 UTC (permalink / raw)

Design by David Rowley.

Author: Jesper Pedersen
---
 src/backend/nodes/outfuncs.c           |  14 +++
 src/backend/nodes/print.c              |  39 +++++++
 src/backend/optimizer/path/Makefile    |   3 +-
 src/backend/optimizer/path/allpaths.c  |   8 ++
 src/backend/optimizer/path/indxpath.c  |  41 ++++++++
 src/backend/optimizer/path/pathkeys.c  |  71 +++++++++++--
 src/backend/optimizer/path/uniquekey.c | 136 +++++++++++++++++++++++++
 src/backend/optimizer/plan/planagg.c   |   1 +
 src/backend/optimizer/plan/planmain.c  |   1 +
 src/backend/optimizer/plan/planner.c   |  37 ++++++-
 src/backend/optimizer/util/pathnode.c  |  46 +++++++--
 src/include/nodes/nodes.h              |   1 +
 src/include/nodes/pathnodes.h          |  19 ++++
 src/include/nodes/print.h              |   1 +
 src/include/optimizer/pathnode.h       |   2 +
 src/include/optimizer/paths.h          |  11 ++
 16 files changed, 408 insertions(+), 23 deletions(-)
 create mode 100644 src/backend/optimizer/path/uniquekey.c

diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index d76fae44b8..16083e7a7e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -1723,6 +1723,7 @@ _outPathInfo(StringInfo str, const Path *node)
 	WRITE_FLOAT_FIELD(startup_cost, "%.2f");
 	WRITE_FLOAT_FIELD(total_cost, "%.2f");
 	WRITE_NODE_FIELD(pathkeys);
+	WRITE_NODE_FIELD(uniquekeys);
 }
 
 /*
@@ -2205,6 +2206,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(eq_classes);
 	WRITE_BOOL_FIELD(ec_merging_done);
 	WRITE_NODE_FIELD(canon_pathkeys);
+	WRITE_NODE_FIELD(canon_uniquekeys);
 	WRITE_NODE_FIELD(left_join_clauses);
 	WRITE_NODE_FIELD(right_join_clauses);
 	WRITE_NODE_FIELD(full_join_clauses);
@@ -2214,6 +2216,7 @@ _outPlannerInfo(StringInfo str, const PlannerInfo *node)
 	WRITE_NODE_FIELD(placeholder_list);
 	WRITE_NODE_FIELD(fkey_list);
 	WRITE_NODE_FIELD(query_pathkeys);
+	WRITE_NODE_FIELD(query_uniquekeys);
 	WRITE_NODE_FIELD(group_pathkeys);
 	WRITE_NODE_FIELD(window_pathkeys);
 	WRITE_NODE_FIELD(distinct_pathkeys);
@@ -2401,6 +2404,14 @@ _outPathKey(StringInfo str, const PathKey *node)
 	WRITE_BOOL_FIELD(pk_nulls_first);
 }
 
+static void
+_outUniqueKey(StringInfo str, const UniqueKey *node)
+{
+	WRITE_NODE_TYPE("UNIQUEKEY");
+
+	WRITE_NODE_FIELD(eq_clause);
+}
+
 static void
 _outPathTarget(StringInfo str, const PathTarget *node)
 {
@@ -4092,6 +4103,9 @@ outNode(StringInfo str, const void *obj)
 			case T_PathKey:
 				_outPathKey(str, obj);
 				break;
+			case T_UniqueKey:
+				_outUniqueKey(str, obj);
+				break;
 			case T_PathTarget:
 				_outPathTarget(str, obj);
 				break;
diff --git a/src/backend/nodes/print.c b/src/backend/nodes/print.c
index 42476724d8..d286b34544 100644
--- a/src/backend/nodes/print.c
+++ b/src/backend/nodes/print.c
@@ -459,6 +459,45 @@ print_pathkeys(const List *pathkeys, const List *rtable)
 	printf(")\n");
 }
 
+/*
+ * print_uniquekeys -
+ *	  uniquekeys list of UniqueKeys
+ */
+void
+print_uniquekeys(const List *uniquekeys, const List *rtable)
+{
+	ListCell   *l;
+
+	printf("(");
+	foreach(l, uniquekeys)
+	{
+		UniqueKey *unique_key = (UniqueKey *) lfirst(l);
+		EquivalenceClass *eclass = (EquivalenceClass *) unique_key->eq_clause;
+		ListCell   *k;
+		bool		first = true;
+
+		/* chase up */
+		while (eclass->ec_merged)
+			eclass = eclass->ec_merged;
+
+		printf("(");
+		foreach(k, eclass->ec_members)
+		{
+			EquivalenceMember *mem = (EquivalenceMember *) lfirst(k);
+
+			if (first)
+				first = false;
+			else
+				printf(", ");
+			print_expr((Node *) mem->em_expr, rtable);
+		}
+		printf(")");
+		if (lnext(uniquekeys, l))
+			printf(", ");
+	}
+	printf(")\n");
+}
+
 /*
  * print_tl
  *	  print targetlist in a more legible way.
diff --git a/src/backend/optimizer/path/Makefile b/src/backend/optimizer/path/Makefile
index 1e199ff66f..63cc1505d9 100644
--- a/src/backend/optimizer/path/Makefile
+++ b/src/backend/optimizer/path/Makefile
@@ -21,6 +21,7 @@ OBJS = \
 	joinpath.o \
 	joinrels.o \
 	pathkeys.o \
-	tidpath.o
+	tidpath.o \
+	uniquekey.o
 
 include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 8286d9cf34..bbc13e6141 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -3954,6 +3954,14 @@ print_path(PlannerInfo *root, Path *path, int indent)
 		print_pathkeys(path->pathkeys, root->parse->rtable);
 	}
 
+	if (path->uniquekeys)
+	{
+		for (i = 0; i < indent; i++)
+			printf("\t");
+		printf("  uniquekeys: ");
+		print_uniquekeys(path->uniquekeys, root->parse->rtable);
+	}
+
 	if (join)
 	{
 		JoinPath   *jp = (JoinPath *) path;
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
index 2a50272da6..363f5349f1 100644
--- a/src/backend/optimizer/path/indxpath.c
+++ b/src/backend/optimizer/path/indxpath.c
@@ -189,6 +189,7 @@ static Expr *match_clause_to_ordering_op(IndexOptInfo *index,
 static bool ec_member_matches_indexcol(PlannerInfo *root, RelOptInfo *rel,
 									   EquivalenceClass *ec, EquivalenceMember *em,
 									   void *arg);
+static List *get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys);
 
 
 /*
@@ -874,6 +875,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	List	   *orderbyclausecols;
 	List	   *index_pathkeys;
 	List	   *useful_pathkeys;
+	List	   *useful_uniquekeys = NIL;
 	bool		found_lower_saop_clause;
 	bool		pathkeys_possibly_useful;
 	bool		index_is_ordered;
@@ -1036,11 +1038,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 	if (index_clauses != NIL || useful_pathkeys != NIL || useful_predicate ||
 		index_only_scan)
 	{
+		if (has_useful_uniquekeys(root))
+			useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 		ipath = create_index_path(root, index,
 								  index_clauses,
 								  orderbyclauses,
 								  orderbyclausecols,
 								  useful_pathkeys,
+								  useful_uniquekeys,
 								  index_is_ordered ?
 								  ForwardScanDirection :
 								  NoMovementScanDirection,
@@ -1063,6 +1069,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 									  orderbyclauses,
 									  orderbyclausecols,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  index_is_ordered ?
 									  ForwardScanDirection :
 									  NoMovementScanDirection,
@@ -1093,11 +1100,15 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 													index_pathkeys);
 		if (useful_pathkeys != NIL)
 		{
+			if (has_useful_uniquekeys(root))
+				useful_uniquekeys = get_uniquekeys_for_index(root, useful_pathkeys);
+
 			ipath = create_index_path(root, index,
 									  index_clauses,
 									  NIL,
 									  NIL,
 									  useful_pathkeys,
+									  useful_uniquekeys,
 									  BackwardScanDirection,
 									  index_only_scan,
 									  outer_relids,
@@ -1115,6 +1126,7 @@ build_index_paths(PlannerInfo *root, RelOptInfo *rel,
 										  NIL,
 										  NIL,
 										  useful_pathkeys,
+										  useful_uniquekeys,
 										  BackwardScanDirection,
 										  index_only_scan,
 										  outer_relids,
@@ -3365,6 +3377,35 @@ match_clause_to_ordering_op(IndexOptInfo *index,
 	return clause;
 }
 
+/*
+ * get_uniquekeys_for_index
+ */
+static List *
+get_uniquekeys_for_index(PlannerInfo *root, List *pathkeys)
+{
+	ListCell *lc;
+
+	if (pathkeys)
+	{
+		List *uniquekeys = NIL;
+		foreach(lc, pathkeys)
+		{
+			UniqueKey *unique_key;
+			PathKey *pk = (PathKey *) lfirst(lc);
+			EquivalenceClass *ec = (EquivalenceClass *) pk->pk_eclass;
+
+			unique_key = makeNode(UniqueKey);
+			unique_key->eq_clause = ec;
+
+			uniquekeys = lappend(uniquekeys, unique_key);
+		}
+
+		if (uniquekeys_contained_in(root->canon_uniquekeys, uniquekeys))
+			return uniquekeys;
+	}
+
+	return NIL;
+}
 
 /****************************************************************************
  *				----  ROUTINES TO DO PARTIAL INDEX PREDICATE TESTS	----
diff --git a/src/backend/optimizer/path/pathkeys.c b/src/backend/optimizer/path/pathkeys.c
index 71b9d42c99..054df9a617 100644
--- a/src/backend/optimizer/path/pathkeys.c
+++ b/src/backend/optimizer/path/pathkeys.c
@@ -29,6 +29,7 @@
 #include "utils/lsyscache.h"
 
 
+static bool pathkey_is_unique(PathKey *new_pathkey, List *pathkeys);
 static bool pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys);
 static bool matches_boolean_partition_clause(RestrictInfo *rinfo,
 											 RelOptInfo *partrel,
@@ -96,6 +97,29 @@ make_canonical_pathkey(PlannerInfo *root,
 	return pk;
 }
 
+/*
+ * pathkey_is_unique
+ *	   Checks if the new pathkey's equivalence class is the same as that of
+ *     any existing member of the pathkey list.
+ */
+static bool
+pathkey_is_unique(PathKey *new_pathkey, List *pathkeys)
+{
+	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
+	ListCell   *lc;
+
+	/* If same EC already is already in the list, then not unique */
+	foreach(lc, pathkeys)
+	{
+		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
+
+		if (new_ec == old_pathkey->pk_eclass)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * pathkey_is_redundant
  *	   Is a pathkey redundant with one already in the given list?
@@ -135,22 +159,12 @@ static bool
 pathkey_is_redundant(PathKey *new_pathkey, List *pathkeys)
 {
 	EquivalenceClass *new_ec = new_pathkey->pk_eclass;
-	ListCell   *lc;
 
 	/* Check for EC containing a constant --- unconditionally redundant */
 	if (EC_MUST_BE_REDUNDANT(new_ec))
 		return true;
 
-	/* If same EC already used in list, then redundant */
-	foreach(lc, pathkeys)
-	{
-		PathKey    *old_pathkey = (PathKey *) lfirst(lc);
-
-		if (new_ec == old_pathkey->pk_eclass)
-			return true;
-	}
-
-	return false;
+	return !pathkey_is_unique(new_pathkey, pathkeys);
 }
 
 /*
@@ -1098,6 +1112,41 @@ make_pathkeys_for_sortclauses(PlannerInfo *root,
 	return pathkeys;
 }
 
+/*
+ * make_pathkeys_for_uniquekeyclauses
+ *		Generate a pathkeys list to be used for uniquekey clauses
+ */
+List *
+make_pathkeys_for_uniquekeys(PlannerInfo *root,
+							 List *sortclauses,
+							 List *tlist)
+{
+	List	   *pathkeys = NIL;
+	ListCell   *l;
+
+	foreach(l, sortclauses)
+	{
+		SortGroupClause *sortcl = (SortGroupClause *) lfirst(l);
+		Expr	   *sortkey;
+		PathKey    *pathkey;
+
+		sortkey = (Expr *) get_sortgroupclause_expr(sortcl, tlist);
+		Assert(OidIsValid(sortcl->sortop));
+		pathkey = make_pathkey_from_sortop(root,
+										   sortkey,
+										   root->nullable_baserels,
+										   sortcl->sortop,
+										   sortcl->nulls_first,
+										   sortcl->tleSortGroupRef,
+										   true);
+
+		if (pathkey_is_unique(pathkey, pathkeys))
+			pathkeys = lappend(pathkeys, pathkey);
+	}
+
+	return pathkeys;
+}
+
 /****************************************************************************
  *		PATHKEYS AND MERGECLAUSES
  ****************************************************************************/
diff --git a/src/backend/optimizer/path/uniquekey.c b/src/backend/optimizer/path/uniquekey.c
new file mode 100644
index 0000000000..c421401d0f
--- /dev/null
+++ b/src/backend/optimizer/path/uniquekey.c
@@ -0,0 +1,136 @@
+/*-------------------------------------------------------------------------
+ *
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ *
+ * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  src/backend/optimizer/path/uniquekey.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
+#include "nodes/pg_list.h"
+
+static UniqueKey *make_canonical_uniquekey(PlannerInfo *root, EquivalenceClass *eclass);
+
+/*
+ * Build a list of unique keys
+ */
+List*
+build_uniquekeys(PlannerInfo *root, List *sortclauses)
+{
+	List *result = NIL;
+	List *sortkeys;
+	ListCell *l;
+
+	sortkeys = make_pathkeys_for_uniquekeys(root,
+											sortclauses,
+											root->processed_tlist);
+
+	/* Create a uniquekey and add it to the list */
+	foreach(l, sortkeys)
+	{
+		PathKey    *pathkey = (PathKey *) lfirst(l);
+		EquivalenceClass *ec = pathkey->pk_eclass;
+		UniqueKey *unique_key = make_canonical_uniquekey(root, ec);
+
+		result = lappend(result, unique_key);
+	}
+
+	return result;
+}
+
+/*
+ * uniquekeys_contained_in
+ *	  Are the keys2 included in the keys1 superset
+ */
+bool
+uniquekeys_contained_in(List *keys1, List *keys2)
+{
+	ListCell   *key1,
+			   *key2;
+
+	foreach(key2, keys2)
+	{
+		bool found = false;
+		UniqueKey  *uniquekey2 = (UniqueKey *) lfirst(key2);
+
+		foreach(key1, keys1)
+		{
+			UniqueKey  *uniquekey1 = (UniqueKey *) lfirst(key1);
+
+			if (uniquekey1->eq_clause == uniquekey2->eq_clause)
+				return true;
+		}
+
+		if (!found)
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * has_useful_uniquekeys
+ *		Detect whether the planner could have any uniquekeys that are
+ *		useful.
+ */
+bool
+has_useful_uniquekeys(PlannerInfo *root)
+{
+	if (root->query_uniquekeys != NIL)
+		return true;	/* there are some */
+	return false;		/* definitely useless */
+}
+
+/*
+ * make_canonical_uniquekey
+ *	  Given the parameters for a UniqueKey, find any pre-existing matching
+ *	  uniquekey in the query's list of "canonical" uniquekeys.  Make a new
+ *	  entry if there's not one already.
+ *
+ * Note that this function must not be used until after we have completed
+ * merging EquivalenceClasses.  (We don't try to enforce that here; instead,
+ * equivclass.c will complain if a merge occurs after root->canon_uniquekeys
+ * has become nonempty.)
+ */
+static UniqueKey *
+make_canonical_uniquekey(PlannerInfo *root,
+						 EquivalenceClass *eclass)
+{
+	UniqueKey  *uk;
+	ListCell   *lc;
+	MemoryContext oldcontext;
+
+	/* The passed eclass might be non-canonical, so chase up to the top */
+	while (eclass->ec_merged)
+		eclass = eclass->ec_merged;
+
+	foreach(lc, root->canon_uniquekeys)
+	{
+		uk = (UniqueKey *) lfirst(lc);
+		if (eclass == uk->eq_clause)
+			return uk;
+	}
+
+	/*
+	 * Be sure canonical uniquekeys are allocated in the main planning context.
+	 * Not an issue in normal planning, but it is for GEQO.
+	 */
+	oldcontext = MemoryContextSwitchTo(root->planner_cxt);
+
+	uk = makeNode(UniqueKey);
+	uk->eq_clause = eclass;
+
+	root->canon_uniquekeys = lappend(root->canon_uniquekeys, uk);
+
+	MemoryContextSwitchTo(oldcontext);
+
+	return uk;
+}
diff --git a/src/backend/optimizer/plan/planagg.c b/src/backend/optimizer/plan/planagg.c
index 8634940efc..dd64775d8f 100644
--- a/src/backend/optimizer/plan/planagg.c
+++ b/src/backend/optimizer/plan/planagg.c
@@ -511,6 +511,7 @@ minmax_qp_callback(PlannerInfo *root, void *extra)
 									  root->parse->targetList);
 
 	root->query_pathkeys = root->sort_pathkeys;
+	root->query_uniquekeys = NIL;
 }
 
 /*
diff --git a/src/backend/optimizer/plan/planmain.c b/src/backend/optimizer/plan/planmain.c
index 62dfc6d44a..3a372af91b 100644
--- a/src/backend/optimizer/plan/planmain.c
+++ b/src/backend/optimizer/plan/planmain.c
@@ -70,6 +70,7 @@ query_planner(PlannerInfo *root,
 	root->join_rel_level = NULL;
 	root->join_cur_level = 0;
 	root->canon_pathkeys = NIL;
+	root->canon_uniquekeys = NIL;
 	root->left_join_clauses = NIL;
 	root->right_join_clauses = NIL;
 	root->full_join_clauses = NIL;
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index d6f2153593..a7de8476d9 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -3657,15 +3657,30 @@ standard_qp_callback(PlannerInfo *root, void *extra)
 	 * much easier, since we know that the parser ensured that one is a
 	 * superset of the other.
 	 */
+	root->query_uniquekeys = NIL;
+
 	if (root->group_pathkeys)
+	{
 		root->query_pathkeys = root->group_pathkeys;
+
+		if (!root->parse->hasAggs)
+			root->query_uniquekeys = build_uniquekeys(root, qp_extra->groupClause);
+	}
 	else if (root->window_pathkeys)
 		root->query_pathkeys = root->window_pathkeys;
 	else if (list_length(root->distinct_pathkeys) >
 			 list_length(root->sort_pathkeys))
+	{
 		root->query_pathkeys = root->distinct_pathkeys;
+		root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else if (root->sort_pathkeys)
+	{
 		root->query_pathkeys = root->sort_pathkeys;
+
+		if (root->distinct_pathkeys)
+			root->query_uniquekeys = build_uniquekeys(root, parse->distinctClause);
+	}
 	else
 		root->query_pathkeys = NIL;
 }
@@ -6222,7 +6237,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid)
 
 	/* Estimate the cost of index scan */
 	indexScanPath = create_index_path(root, indexInfo,
-									  NIL, NIL, NIL, NIL,
+									  NIL, NIL, NIL, NIL, NIL,
 									  ForwardScanDirection, false,
 									  NULL, 1.0, false);
 
@@ -7107,6 +7122,26 @@ apply_scanjoin_target_to_paths(PlannerInfo *root,
 		}
 	}
 
+	foreach(lc, rel->unique_pathlist)
+	{
+		Path	   *subpath = (Path *) lfirst(lc);
+
+		/* Shouldn't have any parameterized paths anymore */
+		Assert(subpath->param_info == NULL);
+
+		if (tlist_same_exprs)
+			subpath->pathtarget->sortgrouprefs =
+				scanjoin_target->sortgrouprefs;
+		else
+		{
+			Path	   *newpath;
+
+			newpath = (Path *) create_projection_path(root, rel, subpath,
+													  scanjoin_target);
+			lfirst(lc) = newpath;
+		}
+	}
+
 	/*
 	 * Now, if final scan/join target contains SRFs, insert ProjectSetPath(s)
 	 * atop each existing path.  (Note that this function doesn't look at the
diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c
index e6d08aede5..a4dfafbb59 100644
--- a/src/backend/optimizer/util/pathnode.c
+++ b/src/backend/optimizer/util/pathnode.c
@@ -361,9 +361,9 @@ set_cheapest(RelOptInfo *parent_rel)
 }
 
 /*
- * add_path
+ * add_path_to
  *	  Consider a potential implementation path for the specified parent rel,
- *	  and add it to the rel's pathlist if it is worthy of consideration.
+ *	  and add it to the specified pathlist if it is worthy of consideration.
  *	  A path is worthy if it has a better sort order (better pathkeys) or
  *	  cheaper cost (on either dimension), or generates fewer rows, than any
  *	  existing path that has the same or superset parameterization rels.
@@ -416,10 +416,10 @@ set_cheapest(RelOptInfo *parent_rel)
  * 'parent_rel' is the relation entry to which the path corresponds.
  * 'new_path' is a potential path for parent_rel.
  *
- * Returns nothing, but modifies parent_rel->pathlist.
+ * Returns modified pathlist.
  */
-void
-add_path(RelOptInfo *parent_rel, Path *new_path)
+static List *
+add_path_to(RelOptInfo *parent_rel, List *pathlist, Path *new_path)
 {
 	bool		accept_new = true;	/* unless we find a superior old path */
 	int			insert_at = 0;	/* where to insert new item */
@@ -440,7 +440,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 	 * for more than one old path to be tossed out because new_path dominates
 	 * it.
 	 */
-	foreach(p1, parent_rel->pathlist)
+	foreach(p1, pathlist)
 	{
 		Path	   *old_path = (Path *) lfirst(p1);
 		bool		remove_old = false; /* unless new proves superior */
@@ -584,8 +584,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 		 */
 		if (remove_old)
 		{
-			parent_rel->pathlist = foreach_delete_current(parent_rel->pathlist,
-														  p1);
+			pathlist = foreach_delete_current(pathlist, p1);
 
 			/*
 			 * Delete the data pointed-to by the deleted cell, if possible
@@ -612,8 +611,7 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 	if (accept_new)
 	{
 		/* Accept the new path: insert it at proper place in pathlist */
-		parent_rel->pathlist =
-			list_insert_nth(parent_rel->pathlist, insert_at, new_path);
+		pathlist = list_insert_nth(pathlist, insert_at, new_path);
 	}
 	else
 	{
@@ -621,6 +619,15 @@ add_path(RelOptInfo *parent_rel, Path *new_path)
 		if (!IsA(new_path, IndexPath))
 			pfree(new_path);
 	}
+
+	return pathlist;
+}
+
+void
+add_path(RelOptInfo *parent_rel, Path *new_path)
+{
+	parent_rel->pathlist = add_path_to(parent_rel,
+									   parent_rel->pathlist, new_path);
 }
 
 /*
@@ -915,6 +922,13 @@ add_partial_path_precheck(RelOptInfo *parent_rel, Cost total_cost,
 	return true;
 }
 
+void
+add_unique_path(RelOptInfo *parent_rel, Path *new_path)
+{
+	parent_rel->unique_pathlist = add_path_to(parent_rel,
+											  parent_rel->unique_pathlist,
+											  new_path);
+}
 
 /*****************************************************************************
  *		PATH NODE CREATION ROUTINES
@@ -940,6 +954,7 @@ create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = parallel_workers;
 	pathnode->pathkeys = NIL;	/* seqscan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_seqscan(pathnode, root, rel, pathnode->param_info);
 
@@ -964,6 +979,7 @@ create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* samplescan has unordered result */
+	pathnode->uniquekeys = NIL;
 
 	cost_samplescan(pathnode, root, rel, pathnode->param_info);
 
@@ -1000,6 +1016,7 @@ create_index_path(PlannerInfo *root,
 				  List *indexorderbys,
 				  List *indexorderbycols,
 				  List *pathkeys,
+				  List *uniquekeys,
 				  ScanDirection indexscandir,
 				  bool indexonly,
 				  Relids required_outer,
@@ -1018,6 +1035,7 @@ create_index_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = 0;
 	pathnode->path.pathkeys = pathkeys;
+	pathnode->path.uniquekeys = uniquekeys;
 
 	pathnode->indexinfo = index;
 	pathnode->indexclauses = indexclauses;
@@ -1061,6 +1079,7 @@ create_bitmap_heap_path(PlannerInfo *root,
 	pathnode->path.parallel_safe = rel->consider_parallel;
 	pathnode->path.parallel_workers = parallel_degree;
 	pathnode->path.pathkeys = NIL;	/* always unordered */
+	pathnode->path.uniquekeys = NIL;
 
 	pathnode->bitmapqual = bitmapqual;
 
@@ -1922,6 +1941,7 @@ create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = pathkeys;
+	pathnode->uniquekeys = NIL;
 
 	cost_functionscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1948,6 +1968,7 @@ create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1974,6 +1995,7 @@ create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_valuesscan(pathnode, root, rel, pathnode->param_info);
 
@@ -1999,6 +2021,7 @@ create_ctescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* XXX for now, result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2025,6 +2048,7 @@ create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
 
@@ -2051,6 +2075,7 @@ create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	cost_resultscan(pathnode, root, rel, pathnode->param_info);
 
@@ -2077,6 +2102,7 @@ create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
 	pathnode->parallel_safe = rel->consider_parallel;
 	pathnode->parallel_workers = 0;
 	pathnode->pathkeys = NIL;	/* result is always unordered */
+	pathnode->uniquekeys = NIL;
 
 	/* Cost is the same as for a regular CTE scan */
 	cost_ctescan(pathnode, root, rel, pathnode->param_info);
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index baced7eec0..a1511b46ea 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -261,6 +261,7 @@ typedef enum NodeTag
 	T_EquivalenceMember,
 	T_PathKey,
 	T_PathTarget,
+	T_UniqueKey,
 	T_RestrictInfo,
 	T_IndexClause,
 	T_PlaceHolderVar,
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 3d3be197e0..0de27f0ef3 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -269,6 +269,8 @@ struct PlannerInfo
 
 	List	   *canon_pathkeys; /* list of "canonical" PathKeys */
 
+	List	   *canon_uniquekeys; /* list of "canonical" UniqueKeys */
+
 	List	   *left_join_clauses;	/* list of RestrictInfos for mergejoinable
 									 * outer join clauses w/nonnullable var on
 									 * left */
@@ -297,6 +299,8 @@ struct PlannerInfo
 
 	List	   *query_pathkeys; /* desired pathkeys for query_planner() */
 
+	List	   *query_uniquekeys; /* unique keys used for the query */
+
 	List	   *group_pathkeys; /* groupClause pathkeys, if any */
 	List	   *window_pathkeys;	/* pathkeys of bottom window, if any */
 	List	   *distinct_pathkeys;	/* distinctClause pathkeys, if any */
@@ -657,6 +661,7 @@ typedef struct RelOptInfo
 	List	   *pathlist;		/* Path structures */
 	List	   *ppilist;		/* ParamPathInfos used in pathlist */
 	List	   *partial_pathlist;	/* partial Paths */
+	List	   *unique_pathlist;	/* unique Paths */
 	struct Path *cheapest_startup_path;
 	struct Path *cheapest_total_path;
 	struct Path *cheapest_unique_path;
@@ -1077,6 +1082,15 @@ typedef struct ParamPathInfo
 	List	   *ppi_clauses;	/* join clauses available from outer rels */
 } ParamPathInfo;
 
+/*
+ * UniqueKey
+ */
+typedef struct UniqueKey
+{
+	NodeTag		type;
+
+	EquivalenceClass *eq_clause;	/* equivalence class */
+} UniqueKey;
 
 /*
  * Type "Path" is used as-is for sequential-scan paths, as well as some other
@@ -1106,6 +1120,9 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * "uniquekeys", if not NIL, is a list of UniqueKey nodes (see above),
+ * describing the XXX.
  */
 typedef struct Path
 {
@@ -1129,6 +1146,8 @@ typedef struct Path
 
 	List	   *pathkeys;		/* sort ordering of path's output */
 	/* pathkeys is a List of PathKey nodes; see above */
+
+	List	   *uniquekeys;	/* the unique keys, or NIL if none */
 } Path;
 
 /* Macro for extracting a path's parameterization relids; beware double eval */
diff --git a/src/include/nodes/print.h b/src/include/nodes/print.h
index 6126b491bf..006248bfb5 100644
--- a/src/include/nodes/print.h
+++ b/src/include/nodes/print.h
@@ -28,6 +28,7 @@ extern char *pretty_format_node_dump(const char *dump);
 extern void print_rt(const List *rtable);
 extern void print_expr(const Node *expr, const List *rtable);
 extern void print_pathkeys(const List *pathkeys, const List *rtable);
+extern void print_uniquekeys(const List *uniquekeys, const List *rtable);
 extern void print_tl(const List *tlist, const List *rtable);
 extern void print_slot(TupleTableSlot *slot);
 
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index e450fe112a..fd25997af5 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -34,6 +34,7 @@ extern void add_partial_path(RelOptInfo *parent_rel, Path *new_path);
 extern bool add_partial_path_precheck(RelOptInfo *parent_rel,
 									  Cost total_cost, List *pathkeys);
 
+extern void add_unique_path(RelOptInfo *parent_rel, Path *new_path);
 extern Path *create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
 								 Relids required_outer, int parallel_workers);
 extern Path *create_samplescan_path(PlannerInfo *root, RelOptInfo *rel,
@@ -44,6 +45,7 @@ extern IndexPath *create_index_path(PlannerInfo *root,
 									List *indexorderbys,
 									List *indexorderbycols,
 									List *pathkeys,
+									List *uniquekeys,
 									ScanDirection indexscandir,
 									bool indexonly,
 									Relids required_outer,
diff --git a/src/include/optimizer/paths.h b/src/include/optimizer/paths.h
index 9ab73bd20c..5b6be383b3 100644
--- a/src/include/optimizer/paths.h
+++ b/src/include/optimizer/paths.h
@@ -214,6 +214,9 @@ extern List *build_join_pathkeys(PlannerInfo *root,
 extern List *make_pathkeys_for_sortclauses(PlannerInfo *root,
 										   List *sortclauses,
 										   List *tlist);
+extern List *make_pathkeys_for_uniquekeys(PlannerInfo *root,
+										  List *sortclauses,
+										  List *tlist);
 extern void initialize_mergeclause_eclasses(PlannerInfo *root,
 											RestrictInfo *restrictinfo);
 extern void update_mergeclause_eclasses(PlannerInfo *root,
@@ -240,4 +243,12 @@ extern PathKey *make_canonical_pathkey(PlannerInfo *root,
 extern void add_paths_to_append_rel(PlannerInfo *root, RelOptInfo *rel,
 									List *live_childrels);
 
+/*
+ * uniquekey.c
+ *	  Utilities for matching and building unique keys
+ */
+extern List *build_uniquekeys(PlannerInfo *root, List *sortclauses);
+extern bool uniquekeys_contained_in(List *keys1, List *keys2);
+extern bool has_useful_uniquekeys(PlannerInfo *root);
+
 #endif							/* PATHS_H */
-- 
2.21.0


--t7fko7qta7zyrtxo
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
 filename="v33-0002-Index-skip-scan-with-filtering.patch"



^ permalink  raw  reply  [nested|flat] 11+ 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; 11+ 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] 11+ messages in thread


end of thread, other threads:[~2023-07-26 10:49 UTC | newest]

Thread overview: 11+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2019-07-09 10:44 [PATCH v31 1/2] Unique key jesperpedersen <[email protected]>
2019-07-09 10:44 [PATCH v32 1/2] Unique key jesperpedersen <[email protected]>
2019-07-09 10:44 [PATCH v33 1/2] Unique key jesperpedersen <[email protected]>
2019-07-09 10:44 [PATCH] Unique key jesperpedersen <[email protected]>
2019-08-02 11:52 [PATCH 1/2] Unique key jesperpedersen <[email protected]>
2019-11-11 13:49 [PATCH 1/2] Unique key jesperpedersen <[email protected]>
2019-11-15 14:46 [PATCH 1/2] Unique key jesperpedersen <[email protected]>
2020-03-24 16:04 [PATCH v33 1/2] Unique key Dmitrii Dolgov <[email protected]>
2020-03-24 16:04 [PATCH v33 1/2] Unique key Dmitrii Dolgov <[email protected]>
2020-03-24 16:04 [PATCH v33 1/2] Unique key Dmitrii Dolgov <[email protected]>
2023-07-26 10:49 [PATCH v3 2/7] Row pattern recognition patch (parse/analysis). Tatsuo Ishii <[email protected]>

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox