Fix grammar for subscripting or field selection from a sub-SELECT result.

Such cases should work, but the grammar failed to accept them because of
our ancient precedence hacks to convince bison that extra parentheses
around a sub-SELECT in an expression are unambiguous.  (Formally, they
*are* ambiguous, but we don't especially care whether they're treated as
part of the sub-SELECT or part of the expression.  Bison cares, though.)
Fix by adding a redundant-looking production for this case.

This is a fine example of why fixing shift/reduce conflicts via
precedence declarations is more dangerous than it looks: you can easily
cause the parser to reject cases that should work.

This has been wrong since commit 3db4056e22
or maybe before, and apparently some people have been working around it
by inserting no-op casts.  That method introduces a dump/reload hazard,
as illustrated in bug #7838 from Jan Mate.  Hence, back-patch to all
active branches.

src/backend/parser/gram.y

@@ -10839,6 +10839,29 @@ c_expr:		columnref								{ $$ = $1; }
 					n->location = @1;
 					$$ = (Node *)n;
 				}
+			| select_with_parens indirection
+				{
+					/*
+					 * Because the select_with_parens nonterminal is designed
+					 * to "eat" as many levels of parens as possible, the
+					 * '(' a_expr ')' opt_indirection production above will
+					 * fail to match a sub-SELECT with indirection decoration;
+					 * the sub-SELECT won't be regarded as an a_expr as long
+					 * as there are parens around it.  To support applying
+					 * subscripting or field selection to a sub-SELECT result,
+					 * we need this redundant-looking production.
+					 */
+					SubLink *n = makeNode(SubLink);
+					A_Indirection *a = makeNode(A_Indirection);
+					n->subLinkType = EXPR_SUBLINK;
+					n->testexpr = NULL;
+					n->operName = NIL;
+					n->subselect = $1;
+					n->location = @1;
+					a->arg = (Node *)n;
+					a->indirection = check_indirection($2, yyscanner);
+					$$ = (Node *)a;
+				}
 			| EXISTS select_with_parens
 				{
 					SubLink *n = makeNode(SubLink);

src/test/regress/expected/subselect.out

@@ -17,6 +17,61 @@ SELECT 1 AS zero WHERE 1 IN (SELECT 2);
 ------
 (0 rows)
 
+-- Check grammar's handling of extra parens in assorted contexts
+SELECT * FROM (SELECT 1 AS x) ss;
+ x 
+---
+ 1
+(1 row)
+
+SELECT * FROM ((SELECT 1 AS x)) ss;
+ x 
+---
+ 1
+(1 row)
+
+(SELECT 2) UNION SELECT 2;
+ ?column? 
+----------
+        2
+(1 row)
+
+((SELECT 2)) UNION SELECT 2;
+ ?column? 
+----------
+        2
+(1 row)
+
+SELECT ((SELECT 2) UNION SELECT 2);
+ ?column? 
+----------
+        2
+(1 row)
+
+SELECT (((SELECT 2)) UNION SELECT 2);
+ ?column? 
+----------
+        2
+(1 row)
+
+SELECT (SELECT ARRAY[1,2,3])[1];
+ array 
+-------
+     1
+(1 row)
+
+SELECT ((SELECT ARRAY[1,2,3]))[2];
+ array 
+-------
+     2
+(1 row)
+
+SELECT (((SELECT ARRAY[1,2,3])))[3];
+ array 
+-------
+     3
+(1 row)
+
 -- Set up some simple test tables
 CREATE TABLE SUBSELECT_TBL (
   f1 integer,

src/test/regress/sql/subselect.sql

@@ -8,6 +8,21 @@ SELECT 1 AS zero WHERE 1 NOT IN (SELECT 1);
 
 SELECT 1 AS zero WHERE 1 IN (SELECT 2);
 
+-- Check grammar's handling of extra parens in assorted contexts
+
+SELECT * FROM (SELECT 1 AS x) ss;
+SELECT * FROM ((SELECT 1 AS x)) ss;
+
+(SELECT 2) UNION SELECT 2;
+((SELECT 2)) UNION SELECT 2;
+
+SELECT ((SELECT 2) UNION SELECT 2);
+SELECT (((SELECT 2)) UNION SELECT 2);
+
+SELECT (SELECT ARRAY[1,2,3])[1];
+SELECT ((SELECT ARRAY[1,2,3]))[2];
+SELECT (((SELECT ARRAY[1,2,3])))[3];
+
 -- Set up some simple test tables
 
 CREATE TABLE SUBSELECT_TBL (