Merge pull request #603 from citusdata/fix/outer-join-filter

Do not copy outer join clauses into WHERE

src/backend/distributed/planner/multi_logical_planner.c

@@ -59,7 +59,6 @@ static bool HasComplexJoinOrder(Query *queryTree);
 static bool HasComplexRangeTableType(Query *queryTree);
 static void ValidateClauseList(List *clauseList);
 static bool ExtractFromExpressionWalker(Node *node, List **qualifierList);
-static bool IsJoinClause(Node *clause);
 static List * MultiTableNodeList(List *tableEntryList, List *rangeTableList);
 static List * AddMultiCollectNodes(List *tableNodeList);
 static MultiNode * MultiJoinTree(List *joinOrderList, List *collectTableList,
@@ -834,8 +833,13 @@ ExtractFromExpressionWalker(Node *node, List **qualifierList)
 		List *joinQualifierList = NIL;
 		JoinExpr *joinExpression = (JoinExpr *) node;
 		Node *joinQualifiersNode = joinExpression->quals;
+		JoinType joinType = joinExpression->jointype;
 
-		if (joinQualifiersNode != NULL)
+		/*
+		 * We only extract qualifiers from inner join clauses, which can be
+		 * treated as WHERE clauses.
+		 */
+		if (joinQualifiersNode != NULL && joinType == JOIN_INNER)
 		{
 			if (IsA(joinQualifiersNode, List))
 			{
@@ -888,7 +892,7 @@ ExtractFromExpressionWalker(Node *node, List **qualifierList)
  * criteria. Our criteria defines a join clause as an equi join operator between
  * two columns that belong to two different tables.
  */
-static bool
+bool
 IsJoinClause(Node *clause)
 {
 	bool isJoinClause = false;

src/backend/distributed/planner/multi_physical_planner.c

@@ -3286,13 +3286,23 @@ JoinSequenceArray(List *rangeTableFragmentsList, Query *jobQuery, List *depended
 		foreach(nextJoinClauseCell, nextJoinClauseList)
 		{
 			OpExpr *nextJoinClause = (OpExpr *) lfirst(nextJoinClauseCell);
-			Var *leftColumn = LeftColumn(nextJoinClause);
+			Var *leftColumn = NULL;
-			Var *rightColumn = RightColumn(nextJoinClause);
+			Var *rightColumn = NULL;
-			Index leftRangeTableId = leftColumn->varno;
+			Index leftRangeTableId = 0;
-			Index rightRangeTableId = rightColumn->varno;
+			Index rightRangeTableId = 0;
 			bool leftPartitioned = false;
 			bool rightPartitioned = false;
 
+			if (!IsJoinClause((Node *) nextJoinClause))
+			{
+				continue;
+			}
+
+			leftColumn = LeftColumn(nextJoinClause);
+			rightColumn = RightColumn(nextJoinClause);
+			leftRangeTableId = leftColumn->varno;
+			rightRangeTableId = rightColumn->varno;
+
 			/*
 			 * We have a table from the existing join list joining with the next
 			 * table. First resolve the existing table's range table id.

src/include/distributed/multi_logical_planner.h

@@ -192,6 +192,7 @@ extern bool BinaryOperator(MultiNode *node);
 extern List * OutputTableIdList(MultiNode *multiNode);
 extern List * FindNodesOfType(MultiNode *node, int type);
 extern List * JoinClauseList(List *whereClauseList);
+extern bool IsJoinClause(Node *clause);
 extern List * SubqueryEntryList(Query *queryTree);
 extern bool ExtractRangeTableIndexWalker(Node *node, List **rangeTableIndexList);
 extern List * WhereClauseList(FromExpr *fromExpr);

src/test/regress/input/multi_outer_join.source

@@ -113,6 +113,27 @@ WHERE
 	r_custkey = 5;
 
 
+-- Apply a filter before the join
+SELECT
+	count(l_custkey), count(r_custkey)
+FROM
+	multi_outer_join_left a LEFT JOIN multi_outer_join_right b
+	ON (l_custkey = r_custkey AND r_custkey = 5);
+
+-- Apply a filter before the join (no matches right)
+SELECT
+	count(l_custkey), count(r_custkey)
+FROM
+	multi_outer_join_left a LEFT JOIN multi_outer_join_right b
+	ON (l_custkey = r_custkey AND r_custkey = -1 /* nonexistant */);
+
+-- Apply a filter before the join (no matches left)
+SELECT
+	count(l_custkey), count(r_custkey)
+FROM
+	multi_outer_join_left a LEFT JOIN multi_outer_join_right b
+	ON (l_custkey = r_custkey AND l_custkey = -1 /* nonexistant */);
+
 -- Right join should be disallowed in this case
 SELECT
 	min(r_custkey), max(r_custkey)
@@ -191,6 +212,14 @@ WHERE
 	r_custkey = 21;
 
 
+-- Apply a filter before the join
+SELECT
+	count(l_custkey), count(r_custkey)
+FROM
+	multi_outer_join_left a LEFT JOIN multi_outer_join_right b
+	ON (l_custkey = r_custkey AND r_custkey = 21);
+
+
 -- Right join should be allowed in this case
 SELECT
 	min(r_custkey), max(r_custkey)

src/test/regress/output/multi_outer_join.source

@@ -146,6 +146,42 @@ LOG:  join order: [ "multi_outer_join_left" ][ broadcast join "multi_outer_join_
    5 |   5
 (1 row)
 
+-- Apply a filter before the join
+SELECT
+	count(l_custkey), count(r_custkey)
+FROM
+	multi_outer_join_left a LEFT JOIN multi_outer_join_right b
+	ON (l_custkey = r_custkey AND r_custkey = 5);
+LOG:  join order: [ "multi_outer_join_left" ][ broadcast join "multi_outer_join_right" ]
+ count | count 
+-------+-------
+    20 |     1
+(1 row)
+
+-- Apply a filter before the join (no matches right)
+SELECT
+	count(l_custkey), count(r_custkey)
+FROM
+	multi_outer_join_left a LEFT JOIN multi_outer_join_right b
+	ON (l_custkey = r_custkey AND r_custkey = -1 /* nonexistant */);
+LOG:  join order: [ "multi_outer_join_left" ][ broadcast join "multi_outer_join_right" ]
+ count | count 
+-------+-------
+    20 |     0
+(1 row)
+
+-- Apply a filter before the join (no matches left)
+SELECT
+	count(l_custkey), count(r_custkey)
+FROM
+	multi_outer_join_left a LEFT JOIN multi_outer_join_right b
+	ON (l_custkey = r_custkey AND l_custkey = -1 /* nonexistant */);
+LOG:  join order: [ "multi_outer_join_left" ][ broadcast join "multi_outer_join_right" ]
+ count | count 
+-------+-------
+    20 |     0
+(1 row)
+
 -- Right join should be disallowed in this case
 SELECT
 	min(r_custkey), max(r_custkey)
@@ -249,6 +285,18 @@ LOG:  join order: [ "multi_outer_join_left" ][ local partition join "multi_outer
   21 |  21
 (1 row)
 
+-- Apply a filter before the join
+SELECT
+	count(l_custkey), count(r_custkey)
+FROM
+	multi_outer_join_left a LEFT JOIN multi_outer_join_right b
+	ON (l_custkey = r_custkey AND r_custkey = 21);
+LOG:  join order: [ "multi_outer_join_left" ][ local partition join "multi_outer_join_right" ]
+ count | count 
+-------+-------
+    25 |     1
+(1 row)
+
 -- Right join should be allowed in this case
 SELECT
 	min(r_custkey), max(r_custkey)