clauses.c 31.2 KB
Newer Older
1 2
/*-------------------------------------------------------------------------
 *
3
 * clauses.c
4
 *	  routines to manipulate qualification clauses
5 6 7 8 9
 *
 * Copyright (c) 1994, Regents of the University of California
 *
 *
 * IDENTIFICATION
10
 *	  $Header: /cvsroot/pgsql/src/backend/optimizer/util/clauses.c,v 1.51 1999/09/09 02:35:53 tgl Exp $
11 12
 *
 * HISTORY
13 14
 *	  AUTHOR			DATE			MAJOR EVENT
 *	  Andrew Yu			Nov 3, 1994		clause.c and clauses.c combined
15 16 17 18 19 20
 *
 *-------------------------------------------------------------------------
 */

#include "postgres.h"

21
#include "catalog/pg_operator.h"
22 23
#include "nodes/makefuncs.h"
#include "nodes/nodeFuncs.h"
B
Bruce Momjian 已提交
24
#include "nodes/plannodes.h"
25 26
#include "optimizer/clauses.h"
#include "optimizer/internal.h"
27
#include "optimizer/tlist.h"
28
#include "optimizer/var.h"
B
Bruce Momjian 已提交
29
#include "utils/lsyscache.h"
30

31

32 33 34 35 36 37 38 39
typedef struct {
	List	   *groupClause;
	List	   *targetList;
} check_subplans_for_ungrouped_vars_context;

static bool pull_agg_clause_walker(Node *node, List **listptr);
static bool check_subplans_for_ungrouped_vars_walker(Node *node,
					check_subplans_for_ungrouped_vars_context *context);
40
static int is_single_func(Node *node);
41

42

43
Expr *
44
make_clause(int type, Node *oper, List *args)
45
{
46 47 48
	if (type == AND_EXPR || type == OR_EXPR || type == NOT_EXPR ||
		type == OP_EXPR || type == FUNC_EXPR)
	{
49
		Expr	   *expr = makeNode(Expr);
50 51 52 53 54 55 56 57 58 59 60 61 62

		/*
		 * assume type checking already done and we don't need the type of
		 * the expr any more.
		 */
		expr->typeOid = InvalidOid;
		expr->opType = type;
		expr->oper = oper;		/* ignored for AND, OR, NOT */
		expr->args = args;
		return expr;
	}
	else
	{
63
		elog(ERROR, "make_clause: unsupported type %d", type);
64
		/* will this ever happen? translated from lispy C code - ay 10/94 */
65
		return (Expr *) args;
66
	}
67 68 69 70
}


/*****************************************************************************
71
 *		OPERATOR clause functions
72 73 74
 *****************************************************************************/


75
/*
76
 * is_opclause
77
 *
78
 * Returns t iff the clause is an operator clause:
79
 *				(op expr expr) or (op expr).
80 81
 *
 * [historical note: is_clause has the exact functionality and is used
82 83
 *		throughout the code. They're renamed to is_opclause for clarity.
 *												- ay 10/94.]
84 85
 */
bool
86
is_opclause(Node *clause)
87
{
88
	return (clause != NULL &&
89 90
			IsA(clause, Expr) &&
			((Expr *) clause)->opType == OP_EXPR);
91 92
}

93
/*
94
 * make_opclause
95 96 97
 *	  Creates a clause given its operator left operand and right
 *	  operand (if it is non-null).
 *
98
 */
99
Expr *
100
make_opclause(Oper *op, Var *leftop, Var *rightop)
101
{
102
	Expr	   *expr = makeNode(Expr);
103

104 105 106
	expr->typeOid = InvalidOid; /* assume type checking done */
	expr->opType = OP_EXPR;
	expr->oper = (Node *) op;
107 108 109 110
	if (rightop)
		expr->args = lcons(leftop, lcons(rightop, NIL));
	else
		expr->args = lcons(leftop, NIL);
111
	return expr;
112 113
}

114
/*
115
 * get_leftop
116
 *
117
 * Returns the left operand of a clause of the form (op expr expr)
118
 *		or (op expr)
119 120 121 122
 *
 * NB: for historical reasons, the result is declared Var *, even
 * though many callers can cope with results that are not Vars.
 * The result really ought to be declared Expr * or Node *.
123
 */
124
Var *
125
get_leftop(Expr *clause)
126
{
127
	if (clause->args != NULL)
128
		return lfirst(clause->args);
129 130
	else
		return NULL;
131 132
}

133
/*
134
 * get_rightop
135
 *
136
 * Returns the right operand in a clause of the form (op expr expr).
137
 * NB: result will be NULL if applied to a unary op clause.
138
 */
139
Var *
140
get_rightop(Expr *clause)
141
{
142
	if (clause->args != NULL && lnext(clause->args) != NULL)
143
		return lfirst(lnext(clause->args));
144 145
	else
		return NULL;
146 147 148
}

/*****************************************************************************
149
 *		FUNC clause functions
150 151
 *****************************************************************************/

152
/*
153
 * is_funcclause
154
 *
155
 * Returns t iff the clause is a function clause: (func { expr }).
156
 *
157 158
 */
bool
159
is_funcclause(Node *clause)
160
{
161
	return (clause != NULL &&
162
			IsA(clause, Expr) &&
163
			((Expr *) clause)->opType == FUNC_EXPR);
164 165
}

166
/*
167
 * make_funcclause
168
 *
169 170
 * Creates a function clause given the FUNC node and the functional
 * arguments.
171
 *
172
 */
173
Expr *
174
make_funcclause(Func *func, List *funcargs)
175
{
176
	Expr	   *expr = makeNode(Expr);
177

178
	expr->typeOid = func->functype;
179 180 181 182
	expr->opType = FUNC_EXPR;
	expr->oper = (Node *) func;
	expr->args = funcargs;
	return expr;
183 184 185
}

/*****************************************************************************
186
 *		OR clause functions
187 188
 *****************************************************************************/

189
/*
190
 * or_clause
191
 *
192
 * Returns t iff the clause is an 'or' clause: (OR { expr }).
193
 *
194 195
 */
bool
196
or_clause(Node *clause)
197
{
198 199 200
	return (clause != NULL &&
			IsA(clause, Expr) &&
			((Expr *) clause)->opType == OR_EXPR);
201 202
}

203
/*
204
 * make_orclause
205
 *
206
 * Creates an 'or' clause given a list of its subclauses.
207
 *
208
 */
209
Expr *
210
make_orclause(List *orclauses)
211
{
212
	Expr	   *expr = makeNode(Expr);
213

214 215 216 217 218
	expr->typeOid = InvalidOid; /* assume type checking done */
	expr->opType = OR_EXPR;
	expr->oper = NULL;
	expr->args = orclauses;
	return expr;
219 220 221
}

/*****************************************************************************
222
 *		NOT clause functions
223 224
 *****************************************************************************/

225
/*
226
 * not_clause
227
 *
228
 * Returns t iff this is a 'not' clause: (NOT expr).
229
 *
230 231
 */
bool
232
not_clause(Node *clause)
233
{
234
	return (clause != NULL &&
235
			IsA(clause, Expr) &&
236
			((Expr *) clause)->opType == NOT_EXPR);
237 238
}

239
/*
240
 * make_notclause
241
 *
242
 * Create a 'not' clause given the expression to be negated.
243
 *
244
 */
245
Expr *
246
make_notclause(Expr *notclause)
247
{
248
	Expr	   *expr = makeNode(Expr);
249

250 251 252 253 254
	expr->typeOid = InvalidOid; /* assume type checking done */
	expr->opType = NOT_EXPR;
	expr->oper = NULL;
	expr->args = lcons(notclause, NIL);
	return expr;
255 256
}

257
/*
258
 * get_notclausearg
259
 *
260
 * Retrieve the clause within a 'not' clause
261
 *
262
 */
263
Expr *
264
get_notclausearg(Expr *notclause)
265
{
266
	return lfirst(notclause->args);
267 268 269
}

/*****************************************************************************
270
 *		AND clause functions
271 272 273
 *****************************************************************************/


274
/*
275
 * and_clause
276
 *
277
 * Returns t iff its argument is an 'and' clause: (AND { expr }).
278
 *
279 280
 */
bool
281
and_clause(Node *clause)
282
{
283
	return (clause != NULL &&
284
			IsA(clause, Expr) &&
285
			((Expr *) clause)->opType == AND_EXPR);
286
}
287 288

/*
289
 * make_andclause
290
 *
291
 * Create an 'and' clause given its arguments in a list.
292
 *
293
 */
294
Expr *
295
make_andclause(List *andclauses)
296
{
297
	Expr	   *expr = makeNode(Expr);
298

299 300 301 302 303
	expr->typeOid = InvalidOid; /* assume type checking done */
	expr->opType = AND_EXPR;
	expr->oper = NULL;
	expr->args = andclauses;
	return expr;
304 305
}

306 307
/*
 * Sometimes (such as in the result of cnfify), we use lists of expression
308 309 310
 * nodes with implicit AND semantics.  These functions convert between an
 * AND-semantics expression list and the ordinary representation of a
 * boolean expression.
311 312 313 314 315 316 317 318 319 320 321
 */
Expr *
make_ands_explicit(List *andclauses)
{
	if (andclauses == NIL)
		return NULL;
	else if (length(andclauses) == 1)
		return (Expr *) lfirst(andclauses);
	else
		return make_andclause(andclauses);
}
T
Thomas G. Lockhart 已提交
322

323 324 325 326 327 328 329 330 331 332 333
List *
make_ands_implicit(Expr *clause)
{
	if (clause == NULL)
		return NIL;
	else if (and_clause((Node *) clause))
		return clause->args;
	else
		return lcons(clause, NIL);
}

T
Thomas G. Lockhart 已提交
334

335
/*****************************************************************************
336
 *																			 *
337
 *		General clause-manipulating routines								 *
338
 *																			 *
339 340 341
 *****************************************************************************/


342
/*
343
 * pull_constant_clauses
344
 *	  Scans through a list of qualifications and find those that
345
 *	  contain no variables (of the current query level).
346
 *
347 348
 * Returns a list of the constant clauses in constantQual and the remaining
 * quals as the return value.
349
 *
350
 */
351
List *
352
pull_constant_clauses(List *quals, List **constantQual)
353
{
354 355 356
	List	   *q;
	List	   *constqual = NIL;
	List	   *restqual = NIL;
357 358 359 360 361 362 363

	foreach(q, quals)
	{
		if (!contain_var_clause(lfirst(q)))
			constqual = lcons(lfirst(q), constqual);
		else
			restqual = lcons(lfirst(q), restqual);
364
	}
365 366
	*constantQual = constqual;
	return restqual;
367 368
}

369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475
/*
 * pull_agg_clause
 *	  Recursively pulls all Aggref nodes from an expression tree.
 *
 *	  Returns list of Aggref nodes found.  Note the nodes themselves are not
 *	  copied, only referenced.
 */
List *
pull_agg_clause(Node *clause)
{
	List	   *result = NIL;

	pull_agg_clause_walker(clause, &result);
	return result;
}

static bool
pull_agg_clause_walker(Node *node, List **listptr)
{
	if (node == NULL)
		return false;
	if (IsA(node, Aggref))
	{
		*listptr = lappend(*listptr, node);
		/* continue, to iterate over agg's arg as well (do nested aggregates
		 * actually work?)
		 */
	}
	return expression_tree_walker(node, pull_agg_clause_walker,
								  (void *) listptr);
}

/*
 * check_subplans_for_ungrouped_vars
 *		Check for subplans that are being passed ungrouped variables as
 *		parameters; return TRUE if any are found.
 *
 * In most contexts, ungrouped variables will be detected by the parser (see
 * parse_agg.c, exprIsAggOrGroupCol()). But that routine currently does not
 * check subplans, because the necessary info is not computed until the
 * planner runs.  So we do it here, after we have processed the subplan.
 * This ought to be cleaned up someday.
 *
 * 'clause' is the expression tree to be searched for subplans.
 * 'groupClause' is the GROUP BY list (a list of GroupClause nodes).
 * 'targetList' is the target list that the group clauses refer to.
 */
bool
check_subplans_for_ungrouped_vars(Node *clause,
								  List *groupClause,
								  List *targetList)
{
	check_subplans_for_ungrouped_vars_context context;

	context.groupClause = groupClause;
	context.targetList = targetList;
	return check_subplans_for_ungrouped_vars_walker(clause, &context);
}

static bool
check_subplans_for_ungrouped_vars_walker(Node *node,
					check_subplans_for_ungrouped_vars_context *context)
{
	if (node == NULL)
		return false;
	/*
	 * We can ignore Vars other than in subplan args lists,
	 * since the parser already checked 'em.
	 */
	if (is_subplan(node))
	{
		/*
		 * The args list of the subplan node represents attributes from
		 * outside passed into the sublink.
		 */
		List	*t;

		foreach(t, ((Expr *) node)->args)
		{
			Node	   *thisarg = lfirst(t);
			bool		contained_in_group_clause = false;
			List	   *gl;

			foreach(gl, context->groupClause)
			{
				GroupClause	   *gcl = lfirst(gl);
				Node		   *groupexpr;

				groupexpr = get_sortgroupclause_expr(gcl,
													 context->targetList);
				if (equal(thisarg, groupexpr))
				{
					contained_in_group_clause = true;
					break;
				}
			}

			if (!contained_in_group_clause)
				return true;	/* found an ungrouped argument */
		}
	}
	return expression_tree_walker(node,
								  check_subplans_for_ungrouped_vars_walker,
								  (void *) context);
}


476
/*
477
 * clause_relids_vars
478 479 480 481 482 483 484 485
 *	  Retrieves distinct relids and vars appearing within a clause.
 *
 * '*relids' is set to an integer list of all distinct "varno"s appearing
 *		in Vars within the clause.
 * '*vars' is set to a list of all distinct Vars appearing within the clause.
 *		Var nodes are considered distinct if they have different varno
 *		or varattno values.  If there are several occurrences of the same
 *		varno/varattno, you get a randomly chosen one...
486 487 488
 *
 * Note that upper-level vars are ignored, since they normally will
 * become Params with respect to this query level.
489 490
 */
void
491
clause_get_relids_vars(Node *clause, Relids *relids, List **vars)
492
{
493
	List	   *clvars = pull_var_clause(clause, false);
494
	List	   *varno_list = NIL;
495
	List	   *var_list = NIL;
496
	List	   *i;
497

498
	foreach(i, clvars)
M
Marc G. Fournier 已提交
499
	{
500 501
		Var		   *var = (Var *) lfirst(i);
		List	   *vi;
502 503

		if (!intMember(var->varno, varno_list))
504
			varno_list = lconsi(var->varno, varno_list);
505 506
		foreach(vi, var_list)
		{
507
			Var		   *in_list = (Var *) lfirst(vi);
508

509
			if (in_list->varno == var->varno &&
V
Vadim B. Mikheev 已提交
510
				in_list->varattno == var->varattno)
511 512 513
				break;
		}
		if (vi == NIL)
514
			var_list = lcons(var, var_list);
M
Marc G. Fournier 已提交
515
	}
516
	freeList(clvars);
517

518 519
	*relids = varno_list;
	*vars = var_list;
520 521
}

522
/*
523 524
 * NumRelids
 *		(formerly clause_relids)
525
 *
526 527 528
 * Returns the number of different relations referenced in 'clause'.
 */
int
529
NumRelids(Node *clause)
530
{
531 532
	List	   *varno_list = pull_varnos(clause);
	int			result = length(varno_list);
533

534
	freeList(varno_list);
535
	return result;
536 537
}

538
/*
539
 * get_relattval
540 541 542 543 544 545 546 547 548 549 550 551 552 553 554
 *		Extract information from a restriction or join clause for
 *		selectivity estimation.  The inputs are an expression
 *		and a relation number (which can be 0 if we don't care which
 *		relation is used; that'd normally be the case for restriction
 *		clauses, where the caller already knows that only one relation
 *		is referenced in the clause).  The routine checks that the
 *		expression is of the form (var op something) or (something op var)
 *		where the var is an attribute of the specified relation, or
 *		a function of a var of the specified relation.  If so, it
 *		returns the following info:
 *			the found relation number (same as targetrelid unless that is 0)
 *			the found var number (or InvalidAttrNumber if a function)
 *			if the "something" is a constant, the value of the constant
 *			flags indicating whether a constant was found, and on which side.
 *		Default values are returned if the expression is too complicated,
555 556 557 558 559
 *		specifically 0 for the relid and attno, 0 for the constant value.
 *
 *		Note that negative attno values are *not* invalid, but represent
 *		system attributes such as OID.  It's sufficient to check for relid=0
 *		to determine whether the routine succeeded.
560 561
 */
void
562
get_relattval(Node *clause,
563
			  int targetrelid,
564
			  int *relid,
B
Bruce Momjian 已提交
565
			  AttrNumber *attno,
566
			  Datum *constval,
567
			  int *flag)
568
{
569
	Var		   *left,
570 571 572
			   *right,
			   *other;
	int			funcvarno;
573

574
	/* Careful; the passed clause might not be a binary operator at all */
575

576 577
	if (!is_opclause(clause))
		goto default_results;
578

579 580
	left = get_leftop((Expr *) clause);
	right = get_rightop((Expr *) clause);
581

582 583
	if (!right)
		goto default_results;
584

585 586 587 588
	/* First look for the var or func */

	if (IsA(left, Var) &&
		(targetrelid == 0 || targetrelid == left->varno))
589 590 591
	{
		*relid = left->varno;
		*attno = left->varattno;
592
		*flag = SEL_RIGHT;
593
	}
594 595
	else if (IsA(right, Var) &&
			 (targetrelid == 0 || targetrelid == right->varno))
596
	{
597 598 599
		*relid = right->varno;
		*attno = right->varattno;
		*flag = 0;
600
	}
601 602
	else if ((funcvarno = is_single_func((Node *) left)) != 0 &&
			 (targetrelid == 0 || targetrelid == funcvarno))
603
	{
604
		*relid = funcvarno;
605
		*attno = InvalidAttrNumber;
606
		*flag = SEL_RIGHT;
607
	}
608 609
	else if ((funcvarno = is_single_func((Node *) right)) != 0 &&
			 (targetrelid == 0 || targetrelid == funcvarno))
610
	{
611 612 613
		*relid = funcvarno;
		*attno = InvalidAttrNumber;
		*flag = 0;
614
	}
615
	else
616
	{
617 618
		/* Duh, it's too complicated for me... */
default_results:
619 620
		*relid = 0;
		*attno = 0;
621
		*constval = 0;
622 623
		*flag = 0;
		return;
624 625
	}

626 627 628 629 630 631 632 633
	/* OK, we identified the var or func; now look at the other side */

	other = (*flag == 0) ? left : right;

	if (IsA(other, Const))
	{
		*constval = ((Const *) other)->constvalue;
		*flag |= SEL_CONSTANT;
634 635 636
	}
	else
	{
637 638
		*constval = 0;
	}
639 640
}

641
/*
642 643 644 645 646 647 648 649
 * is_single_func
 *   If the given expression is a function of a single relation,
 *   return the relation number; else return 0
 */
static int is_single_func(Node *node)
{
	if (is_funcclause(node))
	{
650
		List	   *varnos = pull_varnos(node);
651

652
		if (length(varnos) == 1)
653
		{
654 655 656
			int		funcvarno = lfirsti(varnos);

			freeList(varnos);
657 658
			return funcvarno;
		}
659
		freeList(varnos);
660 661 662 663 664 665
	}
	return 0;
}

/*
 * get_rels_atts
666
 *
667
 * Returns the info
668 669 670
 *				( relid1 attno1 relid2 attno2 )
 *		for a joinclause.
 *
671
 * If the clause is not of the form (var op var) or if any of the vars
672
 * refer to nested attributes, then zeroes are returned.
673
 *
674 675
 */
void
676
get_rels_atts(Node *clause,
677
			  int *relid1,
B
Bruce Momjian 已提交
678
			  AttrNumber *attno1,
679
			  int *relid2,
B
Bruce Momjian 已提交
680
			  AttrNumber *attno2)
681
{
682
	/* set default values */
683 684 685 686
	*relid1 = 0;
	*attno1 = 0;
	*relid2 = 0;
	*attno2 = 0;
687

688 689
	if (is_opclause(clause))
	{
B
Bruce Momjian 已提交
690 691
		Var		   *left = get_leftop((Expr *) clause);
		Var		   *right = get_rightop((Expr *) clause);
692

693
		if (left && right)
694
		{
695
			int			funcvarno;
696

697
			if (IsA(left, Var))
698 699
			{
				*relid1 = left->varno;
700
				*attno1 = left->varattno;
701
			}
702
			else if ((funcvarno = is_single_func((Node *) left)) != 0)
703
			{
704 705 706
				*relid1 = funcvarno;
				*attno1 = InvalidAttrNumber;
			}
707

708 709
			if (IsA(right, Var))
			{
710
				*relid2 = right->varno;
711 712 713 714 715 716
				*attno2 = right->varattno;
			}
			else if ((funcvarno = is_single_func((Node *) right)) != 0)
			{
				*relid2 = funcvarno;
				*attno2 = InvalidAttrNumber;
717
			}
718
		}
719 720 721
	}
}

722 723
/*--------------------
 * CommuteClause: commute a binary operator clause
724 725
 *
 * XXX the clause is destructively modified!
726 727
 *--------------------
 */
728
void
729
CommuteClause(Expr *clause)
730
{
731
	HeapTuple	heapTup;
732 733 734
	Form_pg_operator commuTup;
	Oper	   *commu;
	Node	   *temp;
735

736 737 738
	if (!is_opclause((Node *) clause) ||
		length(clause->args) != 2)
		elog(ERROR, "CommuteClause: applied to non-binary-operator clause");
739

740
	heapTup = (HeapTuple)
741
		get_operator_tuple(get_commutator(((Oper *) clause->oper)->opno));
742

743
	if (heapTup == (HeapTuple) NULL)
744
		elog(ERROR, "CommuteClause: no commutator for operator %u",
745
			 ((Oper *) clause->oper)->opno);
746

747
	commuTup = (Form_pg_operator) GETSTRUCT(heapTup);
748

749
	commu = makeOper(heapTup->t_data->t_oid,
750
					 commuTup->oprcode,
751
					 commuTup->oprresult,
752
					 ((Oper *) clause->oper)->opsize,
753
					 NULL);
754

755
	/*
756
	 * re-form the clause in-place!
757
	 */
758 759 760 761
	clause->oper = (Node *) commu;
	temp = lfirst(clause->args);
	lfirst(clause->args) = lsecond(clause->args);
	lsecond(clause->args) = temp;
762
}
763 764


765
/*
766 767 768 769 770 771 772 773 774
 * Standard expression-tree walking support
 *
 * We used to have near-duplicate code in many different routines that
 * understood how to recurse through an expression node tree.  That was
 * a pain to maintain, and we frequently had bugs due to some particular
 * routine neglecting to support a particular node type.  In most cases,
 * these routines only actually care about certain node types, and don't
 * care about other types except insofar as they have to recurse through
 * non-primitive node types.  Therefore, we now provide generic tree-walking
775 776 777 778 779
 * logic to consolidate the redundant "boilerplate" code.  There are
 * two versions: expression_tree_walker() and expression_tree_mutator().
 */

/*--------------------
780 781
 * expression_tree_walker() is designed to support routines that traverse
 * a tree in a read-only fashion (although it will also work for routines
782 783
 * that modify nodes in-place but never add/delete/replace nodes).
 * A walker routine should look like this:
784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802
 *
 * bool my_walker (Node *node, my_struct *context)
 * {
 *		if (node == NULL)
 *			return false;
 *		// check for nodes that special work is required for, eg:
 *		if (IsA(node, Var))
 *		{
 *			... do special actions for Var nodes
 *		}
 *		else if (IsA(node, ...))
 *		{
 *			... do special actions for other node types
 *		}
 *		// for any node type not specially processed, do:
 *		return expression_tree_walker(node, my_walker, (void *) context);
 * }
 *
 * The "context" argument points to a struct that holds whatever context
803 804
 * information the walker routine needs --- it can be used to return data
 * gathered by the walker, too.  This argument is not touched by
805 806 807 808 809 810 811 812
 * expression_tree_walker, but it is passed down to recursive sub-invocations
 * of my_walker.  The tree walk is started from a setup routine that
 * fills in the appropriate context struct, calls my_walker with the top-level
 * node of the tree, and then examines the results.
 *
 * The walker routine should return "false" to continue the tree walk, or
 * "true" to abort the walk and immediately return "true" to the top-level
 * caller.  This can be used to short-circuit the traversal if the walker
813 814
 * has found what it came for.  "false" is returned to the top-level caller
 * iff no invocation of the walker returned "true".
815 816 817 818 819 820 821 822 823 824 825 826 827
 *
 * The node types handled by expression_tree_walker include all those
 * normally found in target lists and qualifier clauses during the planning
 * stage.  In particular, it handles List nodes since a cnf-ified qual clause
 * will have List structure at the top level, and it handles TargetEntry nodes
 * so that a scan of a target list can be handled without additional code.
 * (But only the "expr" part of a TargetEntry is examined, unless the walker
 * chooses to process TargetEntry nodes specially.)
 *
 * expression_tree_walker will handle a SUBPLAN_EXPR node by recursing into
 * the args and slink->oper lists (which belong to the outer plan), but it
 * will *not* visit the inner plan, since that's typically what expression
 * tree walkers want.  A walker that wants to visit the subplan can force
828 829
 * appropriate behavior by recognizing subplan expression nodes and doing
 * the right thing.
830
 *
831 832 833 834 835
 * Bare SubLink nodes (without a SUBPLAN_EXPR) are handled by recursing into
 * the "lefthand" argument list only.  (A bare SubLink should be seen only if
 * the tree has not yet been processed by subselect.c.)  Again, this can be
 * overridden by the walker, but it seems to be the most useful default
 * behavior.
836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864
 *--------------------
 */

bool
expression_tree_walker(Node *node, bool (*walker) (), void *context)
{
	List	   *temp;

	/*
	 * The walker has already visited the current node,
	 * and so we need only recurse into any sub-nodes it has.
	 *
	 * We assume that the walker is not interested in List nodes per se,
	 * so when we expect a List we just recurse directly to self without
	 * bothering to call the walker.
	 */
	if (node == NULL)
		return false;
	switch (nodeTag(node))
	{
		case T_Ident:
		case T_Const:
		case T_Var:
		case T_Param:
			/* primitive node types with no subnodes */
			break;
		case T_Expr:
			{
				Expr   *expr = (Expr *) node;
865

866 867
				if (expr->opType == SUBPLAN_EXPR)
				{
868 869 870
					/* recurse to the SubLink node (skipping SubPlan!) */
					if (walker((Node *) ((SubPlan *) expr->oper)->sublink,
							   context))
871 872
						return true;
				}
873 874 875 876
				/* for all Expr node types, examine args list */
				if (expression_tree_walker((Node *) expr->args,
										   walker, context))
					return true;
877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919
			}
			break;
		case T_Aggref:
			return walker(((Aggref *) node)->target, context);
		case T_Iter:
			return walker(((Iter *) node)->iterexpr, context);
		case T_ArrayRef:
			{
				ArrayRef   *aref = (ArrayRef *) node;
				/* recurse directly for upper/lower array index lists */
				if (expression_tree_walker((Node *) aref->refupperindexpr,
										   walker, context))
					return true;
				if (expression_tree_walker((Node *) aref->reflowerindexpr,
										   walker, context))
					return true;
				/* walker must see the refexpr and refassgnexpr, however */
				if (walker(aref->refexpr, context))
					return true;
				if (walker(aref->refassgnexpr, context))
					return true;
			}
			break;
		case T_CaseExpr:
			{
				CaseExpr   *caseexpr = (CaseExpr *) node;
				/* we assume walker doesn't care about CaseWhens, either */
				foreach(temp, caseexpr->args)
				{
					CaseWhen   *when = (CaseWhen *) lfirst(temp);
					Assert(IsA(when, CaseWhen));
					if (walker(when->expr, context))
						return true;
					if (walker(when->result, context))
						return true;
				}
				/* caseexpr->arg should be null, but we'll check it anyway */
				if (walker(caseexpr->arg, context))
					return true;
				if (walker(caseexpr->defresult, context))
					return true;
			}
			break;
920
		case T_SubLink:
921 922 923 924 925 926 927 928 929 930 931 932 933 934
			{
				SubLink   *sublink = (SubLink *) node;

				/* If the SubLink has already been processed by subselect.c,
				 * it will have lefthand=NIL, and we only need to look at
				 * the oper list.  Otherwise we only need to look at lefthand
				 * (the Oper nodes in the oper list are deemed uninteresting).
				 */
				if (sublink->lefthand)
					return walker((Node *) sublink->lefthand, context);
				else
					return walker((Node *) sublink->oper, context);
			}
			break;
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950
		case T_List:
			foreach(temp, (List *) node)
			{
				if (walker((Node *) lfirst(temp), context))
					return true;
			}
			break;
		case T_TargetEntry:
			return walker(((TargetEntry *) node)->expr, context);
		default:
			elog(ERROR, "expression_tree_walker: Unexpected node type %d",
				 nodeTag(node));
			break;
	}
	return false;
}
951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999

/*--------------------
 * expression_tree_mutator() is designed to support routines that make a
 * modified copy of an expression tree, with some nodes being added,
 * removed, or replaced by new subtrees.  The original tree is (normally)
 * not changed.  Each recursion level is responsible for returning a copy of
 * (or appropriately modified substitute for) the subtree it is handed.
 * A mutator routine should look like this:
 *
 * Node * my_mutator (Node *node, my_struct *context)
 * {
 *		if (node == NULL)
 *			return NULL;
 *		// check for nodes that special work is required for, eg:
 *		if (IsA(node, Var))
 *		{
 *			... create and return modified copy of Var node
 *		}
 *		else if (IsA(node, ...))
 *		{
 *			... do special transformations of other node types
 *		}
 *		// for any node type not specially processed, do:
 *		return expression_tree_mutator(node, my_mutator, (void *) context);
 * }
 *
 * The "context" argument points to a struct that holds whatever context
 * information the mutator routine needs --- it can be used to return extra
 * data gathered by the mutator, too.  This argument is not touched by
 * expression_tree_mutator, but it is passed down to recursive sub-invocations
 * of my_mutator.  The tree walk is started from a setup routine that
 * fills in the appropriate context struct, calls my_mutator with the
 * top-level node of the tree, and does any required post-processing.
 *
 * Each level of recursion must return an appropriately modified Node.
 * If expression_tree_mutator() is called, it will make an exact copy
 * of the given Node, but invoke my_mutator() to copy the sub-node(s)
 * of that Node.  In this way, my_mutator() has full control over the
 * copying process but need not directly deal with expression trees
 * that it has no interest in.
 *
 * Just as for expression_tree_walker, the node types handled by
 * expression_tree_mutator include all those normally found in target lists
 * and qualifier clauses during the planning stage.
 *
 * expression_tree_mutator will handle a SUBPLAN_EXPR node by recursing into
 * the args and slink->oper lists (which belong to the outer plan), but it
 * will simply copy the link to the inner plan, since that's typically what
 * expression tree mutators want.  A mutator that wants to modify the subplan
1000 1001
 * can force appropriate behavior by recognizing subplan expression nodes
 * and doing the right thing.
1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185
 *
 * Bare SubLink nodes (without a SUBPLAN_EXPR) are handled by recursing into
 * the "lefthand" argument list only.  (A bare SubLink should be seen only if
 * the tree has not yet been processed by subselect.c.)  Again, this can be
 * overridden by the mutator, but it seems to be the most useful default
 * behavior.
 *--------------------
 */

Node *
expression_tree_mutator(Node *node, Node * (*mutator) (), void *context)
{
	/*
	 * The mutator has already decided not to modify the current node,
	 * but we must call the mutator for any sub-nodes.
	 */

#define FLATCOPY(newnode, node, nodetype)  \
	( (newnode) = makeNode(nodetype), \
	  memcpy((newnode), (node), sizeof(nodetype)) )

#define CHECKFLATCOPY(newnode, node, nodetype)  \
	( AssertMacro(IsA((node), nodetype)), \
	  (newnode) = makeNode(nodetype), \
	  memcpy((newnode), (node), sizeof(nodetype)) )

#define MUTATE(newfield, oldfield, fieldtype)  \
		( (newfield) = (fieldtype) mutator((Node *) (oldfield), context) )

	if (node == NULL)
		return NULL;
	switch (nodeTag(node))
	{
		case T_Ident:
		case T_Const:
		case T_Var:
		case T_Param:
			/* primitive node types with no subnodes */
			return (Node *) copyObject(node);
		case T_Expr:
			{
				Expr   *expr = (Expr *) node;
				Expr   *newnode;

				FLATCOPY(newnode, expr, Expr);

				if (expr->opType == SUBPLAN_EXPR)
				{
					SubLink	   *oldsublink = ((SubPlan *) expr->oper)->sublink;
					SubPlan	   *newsubplan;

					/* flat-copy the oper node, which is a SubPlan */
					CHECKFLATCOPY(newsubplan, expr->oper, SubPlan);
					newnode->oper = (Node *) newsubplan;
					/* likewise its SubLink node */
					CHECKFLATCOPY(newsubplan->sublink, oldsublink, SubLink);
					/* transform args list (params to be passed to subplan) */
					MUTATE(newnode->args, expr->args, List *);
					/* transform sublink's oper list as well */
					MUTATE(newsubplan->sublink->oper, oldsublink->oper, List*);
					/* but not the subplan itself, which is referenced as-is */
				}
				else
				{
					/* for other Expr node types, just transform args list,
					 * linking to original oper node (OK?)
					 */
					MUTATE(newnode->args, expr->args, List *);
				}
				return (Node *) newnode;
			}
			break;
		case T_Aggref:
			{
				Aggref   *aggref = (Aggref *) node;
				Aggref   *newnode;

				FLATCOPY(newnode, aggref, Aggref);
				MUTATE(newnode->target, aggref->target, Node *);
				return (Node *) newnode;
			}
			break;
		case T_Iter:
			{
				Iter   *iter = (Iter *) node;
				Iter   *newnode;

				FLATCOPY(newnode, iter, Iter);
				MUTATE(newnode->iterexpr, iter->iterexpr, Node *);
				return (Node *) newnode;
			}
			break;
		case T_ArrayRef:
			{
				ArrayRef   *arrayref = (ArrayRef *) node;
				ArrayRef   *newnode;

				FLATCOPY(newnode, arrayref, ArrayRef);
				MUTATE(newnode->refupperindexpr, arrayref->refupperindexpr,
					   List *);
				MUTATE(newnode->reflowerindexpr, arrayref->reflowerindexpr,
					   List *);
				MUTATE(newnode->refexpr, arrayref->refexpr,
					   Node *);
				MUTATE(newnode->refassgnexpr, arrayref->refassgnexpr,
					   Node *);
				return (Node *) newnode;
			}
			break;
		case T_CaseExpr:
			{
				CaseExpr   *caseexpr = (CaseExpr *) node;
				CaseExpr   *newnode;

				FLATCOPY(newnode, caseexpr, CaseExpr);
				MUTATE(newnode->args, caseexpr->args, List *);
				/* caseexpr->arg should be null, but we'll check it anyway */
				MUTATE(newnode->arg, caseexpr->arg, Node *);
				MUTATE(newnode->defresult, caseexpr->defresult, Node *);
				return (Node *) newnode;
			}
			break;
		case T_CaseWhen:
			{
				CaseWhen   *casewhen = (CaseWhen *) node;
				CaseWhen   *newnode;

				FLATCOPY(newnode, casewhen, CaseWhen);
				MUTATE(newnode->expr, casewhen->expr, Node *);
				MUTATE(newnode->result, casewhen->result, Node *);
				return (Node *) newnode;
			}
			break;
		case T_SubLink:
			{
				/* A "bare" SubLink (note we will not come here if we found
				 * a SUBPLAN_EXPR node above it).  Transform the lefthand side,
				 * but not the oper list nor the subquery.
				 */
				SubLink   *sublink = (SubLink *) node;
				SubLink   *newnode;

				FLATCOPY(newnode, sublink, SubLink);
				MUTATE(newnode->lefthand, sublink->lefthand, List *);
				return (Node *) newnode;
			}
			break;
		case T_List:
			{
				/* We assume the mutator isn't interested in the list nodes
				 * per se, so just invoke it on each list element.
				 * NOTE: this would fail badly on a list with integer elements!
				 */
				List	   *resultlist = NIL;
				List	   *temp;

				foreach(temp, (List *) node)
				{
					resultlist = lappend(resultlist,
										 mutator((Node *) lfirst(temp),
												 context));
				}
				return (Node *) resultlist;
			}
			break;
		case T_TargetEntry:
			{
				/* We mutate the expression, but not the resdom, by default. */
				TargetEntry   *targetentry = (TargetEntry *) node;
				TargetEntry   *newnode;

				FLATCOPY(newnode, targetentry, TargetEntry);
				MUTATE(newnode->expr, targetentry->expr, Node *);
				return (Node *) newnode;
			}
			break;
		default:
			elog(ERROR, "expression_tree_mutator: Unexpected node type %d",
				 nodeTag(node));
			break;
	}
	/* can't get here, but keep compiler happy */
	return NULL;
}