granicus.if.org Git - postgresql/blob - src/backend/optimizer/path/allpaths.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  * allpaths.c
   4  *        Routines to find possible search paths for processing a query
   5  *
   6  * Portions Copyright (c) 1996-2004, PostgreSQL Global Development Group
   7  * Portions Copyright (c) 1994, Regents of the University of California
   8  *
   9  *
  10  * IDENTIFICATION
  11  *        $PostgreSQL: pgsql/src/backend/optimizer/path/allpaths.c,v 1.120 2004/08/29 04:12:33 momjian Exp $
  12  *
  13  *-------------------------------------------------------------------------
  14  */
  15
  16 #include "postgres.h"
  17
  18 #ifdef OPTIMIZER_DEBUG
  19 #include "nodes/print.h"
  20 #endif
  21 #include "optimizer/clauses.h"
  22 #include "optimizer/cost.h"
  23 #include "optimizer/geqo.h"
  24 #include "optimizer/pathnode.h"
  25 #include "optimizer/paths.h"
  26 #include "optimizer/plancat.h"
  27 #include "optimizer/planner.h"
  28 #include "optimizer/prep.h"
  29 #include "optimizer/var.h"
  30 #include "parser/parsetree.h"
  31 #include "parser/parse_clause.h"
  32 #include "rewrite/rewriteManip.h"
  33
  34
  35 /* These parameters are set by GUC */
  36 bool            enable_geqo = false;    /* just in case GUC doesn't set it */
  37 int                     geqo_threshold;
  38
  39
  40 static void set_base_rel_pathlists(Query *root);
  41 static void set_plain_rel_pathlist(Query *root, RelOptInfo *rel,
  42                                            RangeTblEntry *rte);
  43 static void set_inherited_rel_pathlist(Query *root, RelOptInfo *rel,
  44                                                    Index rti, RangeTblEntry *rte,
  45                                                    List *inheritlist);
  46 static void set_subquery_pathlist(Query *root, RelOptInfo *rel,
  47                                           Index rti, RangeTblEntry *rte);
  48 static void set_function_pathlist(Query *root, RelOptInfo *rel,
  49                                           RangeTblEntry *rte);
  50 static RelOptInfo *make_one_rel_by_joins(Query *root, int levels_needed,
  51                                           List *initial_rels);
  52 static bool subquery_is_pushdown_safe(Query *subquery, Query *topquery,
  53                                                   bool *differentTypes);
  54 static bool recurse_pushdown_safe(Node *setOp, Query *topquery,
  55                                           bool *differentTypes);
  56 static void compare_tlist_datatypes(List *tlist, List *colTypes,
  57                                                 bool *differentTypes);
  58 static bool qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
  59                                           bool *differentTypes);
  60 static void subquery_push_qual(Query *subquery, List *rtable,
  61                                                            Index rti, Node *qual);
  62 static void recurse_push_qual(Node *setOp, Query *topquery,
  63                                                           List *rtable, Index rti, Node *qual);
  64
  65
  66 /*
  67  * make_one_rel
  68  *        Finds all possible access paths for executing a query, returning a
  69  *        single rel that represents the join of all base rels in the query.
  70  */
  71 RelOptInfo *
  72 make_one_rel(Query *root)
  73 {
  74         RelOptInfo *rel;
  75
  76         /*
  77          * Generate access paths for the base rels.
  78          */
  79         set_base_rel_pathlists(root);
  80
  81         /*
  82          * Generate access paths for the entire join tree.
  83          */
  84         Assert(root->jointree != NULL && IsA(root->jointree, FromExpr));
  85
  86         rel = make_fromexpr_rel(root, root->jointree);
  87
  88         /*
  89          * The result should join all the query's base rels.
  90          */
  91         Assert(bms_num_members(rel->relids) == list_length(root->base_rel_list));
  92
  93         return rel;
  94 }
  95
  96 /*
  97  * set_base_rel_pathlists
  98  *        Finds all paths available for scanning each base-relation entry.
  99  *        Sequential scan and any available indices are considered.
 100  *        Each useful path is attached to its relation's 'pathlist' field.
 101  */
 102 static void
 103 set_base_rel_pathlists(Query *root)
 104 {
 105         ListCell           *l;
 106
 107         foreach(l, root->base_rel_list)
 108         {
 109                 RelOptInfo *rel = (RelOptInfo *) lfirst(l);
 110                 Index           rti = rel->relid;
 111                 RangeTblEntry *rte;
 112                 List       *inheritlist;
 113
 114                 Assert(rti > 0);                /* better be base rel */
 115                 rte = rt_fetch(rti, root->rtable);
 116
 117                 if (rel->rtekind == RTE_SUBQUERY)
 118                 {
 119                         /* Subquery --- generate a separate plan for it */
 120                         set_subquery_pathlist(root, rel, rti, rte);
 121                 }
 122                 else if (rel->rtekind == RTE_FUNCTION)
 123                 {
 124                         /* RangeFunction --- generate a separate plan for it */
 125                         set_function_pathlist(root, rel, rte);
 126                 }
 127                 else if ((inheritlist = expand_inherited_rtentry(root, rti, true))
 128                                  != NIL)
 129                 {
 130                         /* Relation is root of an inheritance tree, process specially */
 131                         set_inherited_rel_pathlist(root, rel, rti, rte, inheritlist);
 132                 }
 133                 else
 134                 {
 135                         /* Plain relation */
 136                         set_plain_rel_pathlist(root, rel, rte);
 137                 }
 138
 139 #ifdef OPTIMIZER_DEBUG
 140                 debug_print_rel(root, rel);
 141 #endif
 142         }
 143 }
 144
 145 /*
 146  * set_plain_rel_pathlist
 147  *        Build access paths for a plain relation (no subquery, no inheritance)
 148  */
 149 static void
 150 set_plain_rel_pathlist(Query *root, RelOptInfo *rel, RangeTblEntry *rte)
 151 {
 152         /* Mark rel with estimated output rows, width, etc */
 153         set_baserel_size_estimates(root, rel);
 154
 155         /* Test any partial indexes of rel for applicability */
 156         check_partial_indexes(root, rel);
 157
 158         /*
 159          * Check to see if we can extract any restriction conditions from
 160          * join quals that are OR-of-AND structures.  If so, add them to the
 161          * rel's restriction list, and recompute the size estimates.
 162          */
 163         if (create_or_index_quals(root, rel))
 164                 set_baserel_size_estimates(root, rel);
 165
 166         /*
 167          * Generate paths and add them to the rel's pathlist.
 168          *
 169          * Note: add_path() will discard any paths that are dominated by another
 170          * available path, keeping only those paths that are superior along at
 171          * least one dimension of cost or sortedness.
 172          */
 173
 174         /* Consider sequential scan */
 175         add_path(rel, create_seqscan_path(root, rel));
 176
 177         /* Consider TID scans */
 178         create_tidscan_paths(root, rel);
 179
 180         /* Consider index paths for both simple and OR index clauses */
 181         create_index_paths(root, rel);
 182         create_or_index_paths(root, rel);
 183
 184         /* Now find the cheapest of the paths for this rel */
 185         set_cheapest(rel);
 186 }
 187
 188 /*
 189  * set_inherited_rel_pathlist
 190  *        Build access paths for a inheritance tree rooted at rel
 191  *
 192  * inheritlist is a list of RT indexes of all tables in the inheritance tree,
 193  * including a duplicate of the parent itself.  Note we will not come here
 194  * unless there's at least one child in addition to the parent.
 195  *
 196  * NOTE: the passed-in rel and RTE will henceforth represent the appended
 197  * result of the whole inheritance tree.  The members of inheritlist represent
 198  * the individual tables --- in particular, the inheritlist member that is a
 199  * duplicate of the parent RTE represents the parent table alone.
 200  * We will generate plans to scan the individual tables that refer to
 201  * the inheritlist RTEs, whereas Vars elsewhere in the plan tree that
 202  * refer to the original RTE are taken to refer to the append output.
 203  * In particular, this means we have separate RelOptInfos for the parent
 204  * table and for the append output, which is a good thing because they're
 205  * not the same size.
 206  */
 207 static void
 208 set_inherited_rel_pathlist(Query *root, RelOptInfo *rel,
 209                                                    Index rti, RangeTblEntry *rte,
 210                                                    List *inheritlist)
 211 {
 212         int                     parentRTindex = rti;
 213         Oid                     parentOID = rte->relid;
 214         List       *subpaths = NIL;
 215         ListCell   *il;
 216
 217         /*
 218          * XXX for now, can't handle inherited expansion of FOR UPDATE; can we
 219          * do better?
 220          */
 221         if (list_member_int(root->rowMarks, parentRTindex))
 222                 ereport(ERROR,
 223                                 (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
 224                                  errmsg("SELECT FOR UPDATE is not supported for inheritance queries")));
 225
 226         /*
 227          * The executor will check the parent table's access permissions when
 228          * it examines the parent's inheritlist entry.  There's no need to
 229          * check twice, so turn off access check bits in the original RTE.
 230          */
 231         rte->requiredPerms = 0;
 232
 233         /*
 234          * Initialize to compute size estimates for whole inheritance tree
 235          */
 236         rel->rows = 0;
 237         rel->width = 0;
 238
 239         /*
 240          * Generate access paths for each table in the tree (parent AND
 241          * children), and pick the cheapest path for each table.
 242          */
 243         foreach(il, inheritlist)
 244         {
 245                 int                     childRTindex = lfirst_int(il);
 246                 RangeTblEntry *childrte;
 247                 Oid                     childOID;
 248                 RelOptInfo *childrel;
 249                 ListCell   *parentvars;
 250                 ListCell   *childvars;
 251
 252                 childrte = rt_fetch(childRTindex, root->rtable);
 253                 childOID = childrte->relid;
 254
 255                 /*
 256                  * Make a RelOptInfo for the child so we can do planning.  Do NOT
 257                  * attach the RelOptInfo to the query's base_rel_list, however,
 258                  * since the child is not part of the main join tree.  Instead,
 259                  * the child RelOptInfo is added to other_rel_list.
 260                  */
 261                 childrel = build_other_rel(root, childRTindex);
 262
 263                 /*
 264                  * Copy the parent's targetlist and restriction quals to the
 265                  * child, with attribute-number adjustment as needed.  We don't
 266                  * bother to copy the join quals, since we can't do any joining of
 267                  * the individual tables.  Also, we just zap attr_needed rather
 268                  * than trying to adjust it; it won't be looked at in the child.
 269                  */
 270                 childrel->reltargetlist = (List *)
 271                         adjust_inherited_attrs((Node *) rel->reltargetlist,
 272                                                                    parentRTindex,
 273                                                                    parentOID,
 274                                                                    childRTindex,
 275                                                                    childOID);
 276                 childrel->attr_needed = NULL;
 277                 childrel->baserestrictinfo = (List *)
 278                         adjust_inherited_attrs((Node *) rel->baserestrictinfo,
 279                                                                    parentRTindex,
 280                                                                    parentOID,
 281                                                                    childRTindex,
 282                                                                    childOID);
 283
 284                 /*
 285                  * Now compute child access paths, and save the cheapest.
 286                  */
 287                 set_plain_rel_pathlist(root, childrel, childrte);
 288
 289                 subpaths = lappend(subpaths, childrel->cheapest_total_path);
 290
 291                 /*
 292                  * Propagate size information from the child back to the parent.
 293                  * For simplicity, we use the largest widths from any child as the
 294                  * parent estimates.
 295                  */
 296                 rel->rows += childrel->rows;
 297                 if (childrel->width > rel->width)
 298                         rel->width = childrel->width;
 299
 300                 forboth(parentvars, rel->reltargetlist,
 301                                 childvars, childrel->reltargetlist)
 302                 {
 303                         Var                *parentvar = (Var *) lfirst(parentvars);
 304                         Var                *childvar = (Var *) lfirst(childvars);
 305
 306                         if (IsA(parentvar, Var) && IsA(childvar, Var))
 307                         {
 308                                 int                     pndx = parentvar->varattno - rel->min_attr;
 309                                 int                     cndx = childvar->varattno - childrel->min_attr;
 310
 311                                 if (childrel->attr_widths[cndx] > rel->attr_widths[pndx])
 312                                         rel->attr_widths[pndx] = childrel->attr_widths[cndx];
 313                         }
 314                 }
 315         }
 316
 317         /*
 318          * Finally, build Append path and install it as the only access path
 319          * for the parent rel.
 320          */
 321         add_path(rel, (Path *) create_append_path(rel, subpaths));
 322
 323         /* Select cheapest path (pretty easy in this case...) */
 324         set_cheapest(rel);
 325 }
 326
 327 /*
 328  * set_subquery_pathlist
 329  *              Build the (single) access path for a subquery RTE
 330  */
 331 static void
 332 set_subquery_pathlist(Query *root, RelOptInfo *rel,
 333                                           Index rti, RangeTblEntry *rte)
 334 {
 335         Query      *subquery = rte->subquery;
 336         bool       *differentTypes;
 337         List       *pathkeys;
 338
 339         /* We need a workspace for keeping track of set-op type coercions */
 340         differentTypes = (bool *)
 341                 palloc0((list_length(subquery->targetList) + 1) * sizeof(bool));
 342
 343         /*
 344          * If there are any restriction clauses that have been attached to the
 345          * subquery relation, consider pushing them down to become HAVING
 346          * quals of the subquery itself.  (Not WHERE clauses, since they may
 347          * refer to subquery outputs that are aggregate results.  But
 348          * planner.c will transfer them into the subquery's WHERE if they do
 349          * not.)  This transformation is useful because it may allow us to
 350          * generate a better plan for the subquery than evaluating all the
 351          * subquery output rows and then filtering them.
 352          *
 353          * There are several cases where we cannot push down clauses.
 354          * Restrictions involving the subquery are checked by
 355          * subquery_is_pushdown_safe().  Restrictions on individual clauses
 356          * are checked by qual_is_pushdown_safe().
 357          *
 358          * Non-pushed-down clauses will get evaluated as qpquals of the
 359          * SubqueryScan node.
 360          *
 361          * XXX Are there any cases where we want to make a policy decision not to
 362          * push down a pushable qual, because it'd result in a worse plan?
 363          */
 364         if (rel->baserestrictinfo != NIL &&
 365                 subquery_is_pushdown_safe(subquery, subquery, differentTypes))
 366         {
 367                 /* OK to consider pushing down individual quals */
 368                 List       *upperrestrictlist = NIL;
 369                 ListCell   *l;
 370
 371                 foreach(l, rel->baserestrictinfo)
 372                 {
 373                         RestrictInfo *rinfo = (RestrictInfo *) lfirst(l);
 374                         Node       *clause = (Node *) rinfo->clause;
 375
 376                         if (qual_is_pushdown_safe(subquery, rti, clause, differentTypes))
 377                         {
 378                                 /* Push it down */
 379                                 subquery_push_qual(subquery, root->rtable, rti, clause);
 380                         }
 381                         else
 382                         {
 383                                 /* Keep it in the upper query */
 384                                 upperrestrictlist = lappend(upperrestrictlist, rinfo);
 385                         }
 386                 }
 387                 rel->baserestrictinfo = upperrestrictlist;
 388         }
 389
 390         pfree(differentTypes);
 391
 392         /* Generate the plan for the subquery */
 393         rel->subplan = subquery_planner(subquery, 0.0 /* default case */ );
 394
 395         /* Copy number of output rows from subplan */
 396         rel->tuples = rel->subplan->plan_rows;
 397
 398         /* Mark rel with estimated output rows, width, etc */
 399         set_baserel_size_estimates(root, rel);
 400
 401         /* Convert subquery pathkeys to outer representation */
 402         pathkeys = build_subquery_pathkeys(root, rel, subquery);
 403
 404         /* Generate appropriate path */
 405         add_path(rel, create_subqueryscan_path(rel, pathkeys));
 406
 407         /* Select cheapest path (pretty easy in this case...) */
 408         set_cheapest(rel);
 409 }
 410
 411 /*
 412  * set_function_pathlist
 413  *              Build the (single) access path for a function RTE
 414  */
 415 static void
 416 set_function_pathlist(Query *root, RelOptInfo *rel, RangeTblEntry *rte)
 417 {
 418         /* Mark rel with estimated output rows, width, etc */
 419         set_function_size_estimates(root, rel);
 420
 421         /* Generate appropriate path */
 422         add_path(rel, create_functionscan_path(root, rel));
 423
 424         /* Select cheapest path (pretty easy in this case...) */
 425         set_cheapest(rel);
 426 }
 427
 428 /*
 429  * make_fromexpr_rel
 430  *        Build access paths for a FromExpr jointree node.
 431  */
 432 RelOptInfo *
 433 make_fromexpr_rel(Query *root, FromExpr *from)
 434 {
 435         int                     levels_needed;
 436         List       *initial_rels = NIL;
 437         ListCell   *jt;
 438
 439         /*
 440          * Count the number of child jointree nodes.  This is the depth of the
 441          * dynamic-programming algorithm we must employ to consider all ways
 442          * of joining the child nodes.
 443          */
 444         levels_needed = list_length(from->fromlist);
 445
 446         if (levels_needed <= 0)
 447                 return NULL;                    /* nothing to do? */
 448
 449         /*
 450          * Construct a list of rels corresponding to the child jointree nodes.
 451          * This may contain both base rels and rels constructed according to
 452          * explicit JOIN directives.
 453          */
 454         foreach(jt, from->fromlist)
 455         {
 456                 Node       *jtnode = (Node *) lfirst(jt);
 457
 458                 initial_rels = lappend(initial_rels,
 459                                                            make_jointree_rel(root, jtnode));
 460         }
 461
 462         if (levels_needed == 1)
 463         {
 464                 /*
 465                  * Single jointree node, so we're done.
 466                  */
 467                 return (RelOptInfo *) linitial(initial_rels);
 468         }
 469         else
 470         {
 471                 /*
 472                  * Consider the different orders in which we could join the rels,
 473                  * using either GEQO or regular optimizer.
 474                  */
 475                 if (enable_geqo && levels_needed >= geqo_threshold)
 476                         return geqo(root, levels_needed, initial_rels);
 477                 else
 478                         return make_one_rel_by_joins(root, levels_needed, initial_rels);
 479         }
 480 }
 481
 482 /*
 483  * make_one_rel_by_joins
 484  *        Find all possible joinpaths for a query by successively finding ways
 485  *        to join component relations into join relations.
 486  *
 487  * 'levels_needed' is the number of iterations needed, ie, the number of
 488  *              independent jointree items in the query.  This is > 1.
 489  *
 490  * 'initial_rels' is a list of RelOptInfo nodes for each independent
 491  *              jointree item.  These are the components to be joined together.
 492  *
 493  * Returns the final level of join relations, i.e., the relation that is
 494  * the result of joining all the original relations together.
 495  */
 496 static RelOptInfo *
 497 make_one_rel_by_joins(Query *root, int levels_needed, List *initial_rels)
 498 {
 499         List      **joinitems;
 500         int                     lev;
 501         RelOptInfo *rel;
 502
 503         /*
 504          * We employ a simple "dynamic programming" algorithm: we first find
 505          * all ways to build joins of two jointree items, then all ways to
 506          * build joins of three items (from two-item joins and single items),
 507          * then four-item joins, and so on until we have considered all ways
 508          * to join all the items into one rel.
 509          *
 510          * joinitems[j] is a list of all the j-item rels.  Initially we set
 511          * joinitems[1] to represent all the single-jointree-item relations.
 512          */
 513         joinitems = (List **) palloc0((levels_needed + 1) * sizeof(List *));
 514
 515         joinitems[1] = initial_rels;
 516
 517         for (lev = 2; lev <= levels_needed; lev++)
 518         {
 519                 ListCell   *x;
 520
 521                 /*
 522                  * Determine all possible pairs of relations to be joined at this
 523                  * level, and build paths for making each one from every available
 524                  * pair of lower-level relations.
 525                  */
 526                 joinitems[lev] = make_rels_by_joins(root, lev, joinitems);
 527
 528                 /*
 529                  * Do cleanup work on each just-processed rel.
 530                  */
 531                 foreach(x, joinitems[lev])
 532                 {
 533                         rel = (RelOptInfo *) lfirst(x);
 534
 535                         /* Find and save the cheapest paths for this rel */
 536                         set_cheapest(rel);
 537
 538 #ifdef OPTIMIZER_DEBUG
 539                         debug_print_rel(root, rel);
 540 #endif
 541                 }
 542         }
 543
 544         /*
 545          * We should have a single rel at the final level.
 546          */
 547         if (joinitems[levels_needed] == NIL)
 548                 elog(ERROR, "failed to build any %d-way joins", levels_needed);
 549         Assert(list_length(joinitems[levels_needed]) == 1);
 550
 551         rel = (RelOptInfo *) linitial(joinitems[levels_needed]);
 552
 553         return rel;
 554 }
 555
 556 /*****************************************************************************
 557  *                      PUSHING QUALS DOWN INTO SUBQUERIES
 558  *****************************************************************************/
 559
 560 /*
 561  * subquery_is_pushdown_safe - is a subquery safe for pushing down quals?
 562  *
 563  * subquery is the particular component query being checked.  topquery
 564  * is the top component of a set-operations tree (the same Query if no
 565  * set-op is involved).
 566  *
 567  * Conditions checked here:
 568  *
 569  * 1. If the subquery has a LIMIT clause, we must not push down any quals,
 570  * since that could change the set of rows returned.
 571  *
 572  * 2. If the subquery contains EXCEPT or EXCEPT ALL set ops we cannot push
 573  * quals into it, because that would change the results.
 574  *
 575  * 3. For subqueries using UNION/UNION ALL/INTERSECT/INTERSECT ALL, we can
 576  * push quals into each component query, but the quals can only reference
 577  * subquery columns that suffer no type coercions in the set operation.
 578  * Otherwise there are possible semantic gotchas.  So, we check the
 579  * component queries to see if any of them have different output types;
 580  * differentTypes[k] is set true if column k has different type in any
 581  * component.
 582  */
 583 static bool
 584 subquery_is_pushdown_safe(Query *subquery, Query *topquery,
 585                                                   bool *differentTypes)
 586 {
 587         SetOperationStmt *topop;
 588
 589         /* Check point 1 */
 590         if (subquery->limitOffset != NULL || subquery->limitCount != NULL)
 591                 return false;
 592
 593         /* Are we at top level, or looking at a setop component? */
 594         if (subquery == topquery)
 595         {
 596                 /* Top level, so check any component queries */
 597                 if (subquery->setOperations != NULL)
 598                         if (!recurse_pushdown_safe(subquery->setOperations, topquery,
 599                                                                            differentTypes))
 600                                 return false;
 601         }
 602         else
 603         {
 604                 /* Setop component must not have more components (too weird) */
 605                 if (subquery->setOperations != NULL)
 606                         return false;
 607                 /* Check whether setop component output types match top level */
 608                 topop = (SetOperationStmt *) topquery->setOperations;
 609                 Assert(topop && IsA(topop, SetOperationStmt));
 610                 compare_tlist_datatypes(subquery->targetList,
 611                                                                 topop->colTypes,
 612                                                                 differentTypes);
 613         }
 614         return true;
 615 }
 616
 617 /*
 618  * Helper routine to recurse through setOperations tree
 619  */
 620 static bool
 621 recurse_pushdown_safe(Node *setOp, Query *topquery,
 622                                           bool *differentTypes)
 623 {
 624         if (IsA(setOp, RangeTblRef))
 625         {
 626                 RangeTblRef *rtr = (RangeTblRef *) setOp;
 627                 RangeTblEntry *rte = rt_fetch(rtr->rtindex, topquery->rtable);
 628                 Query      *subquery = rte->subquery;
 629
 630                 Assert(subquery != NULL);
 631                 return subquery_is_pushdown_safe(subquery, topquery, differentTypes);
 632         }
 633         else if (IsA(setOp, SetOperationStmt))
 634         {
 635                 SetOperationStmt *op = (SetOperationStmt *) setOp;
 636
 637                 /* EXCEPT is no good */
 638                 if (op->op == SETOP_EXCEPT)
 639                         return false;
 640                 /* Else recurse */
 641                 if (!recurse_pushdown_safe(op->larg, topquery, differentTypes))
 642                         return false;
 643                 if (!recurse_pushdown_safe(op->rarg, topquery, differentTypes))
 644                         return false;
 645         }
 646         else
 647         {
 648                 elog(ERROR, "unrecognized node type: %d",
 649                          (int) nodeTag(setOp));
 650         }
 651         return true;
 652 }
 653
 654 /*
 655  * Compare tlist's datatypes against the list of set-operation result types.
 656  * For any items that are different, mark the appropriate element of
 657  * differentTypes[] to show that this column will have type conversions.
 658  */
 659 static void
 660 compare_tlist_datatypes(List *tlist, List *colTypes,
 661                                                 bool *differentTypes)
 662 {
 663         ListCell   *l;
 664         ListCell   *colType = list_head(colTypes);
 665
 666         foreach(l, tlist)
 667         {
 668                 TargetEntry *tle = (TargetEntry *) lfirst(l);
 669
 670                 if (tle->resdom->resjunk)
 671                         continue;                       /* ignore resjunk columns */
 672                 if (colType == NULL)
 673                         elog(ERROR, "wrong number of tlist entries");
 674                 if (tle->resdom->restype != lfirst_oid(colType))
 675                         differentTypes[tle->resdom->resno] = true;
 676                 colType = lnext(colType);
 677         }
 678         if (colType != NULL)
 679                 elog(ERROR, "wrong number of tlist entries");
 680 }
 681
 682 /*
 683  * qual_is_pushdown_safe - is a particular qual safe to push down?
 684  *
 685  * qual is a restriction clause applying to the given subquery (whose RTE
 686  * has index rti in the parent query).
 687  *
 688  * Conditions checked here:
 689  *
 690  * 1. The qual must not contain any subselects (mainly because I'm not sure
 691  * it will work correctly: sublinks will already have been transformed into
 692  * subplans in the qual, but not in the subquery).
 693  *
 694  * 2. The qual must not refer to any subquery output columns that were
 695  * found to have inconsistent types across a set operation tree by
 696  * subquery_is_pushdown_safe().
 697  *
 698  * 3. If the subquery uses DISTINCT ON, we must not push down any quals that
 699  * refer to non-DISTINCT output columns, because that could change the set
 700  * of rows returned.  This condition is vacuous for DISTINCT, because then
 701  * there are no non-DISTINCT output columns, but unfortunately it's fairly
 702  * expensive to tell the difference between DISTINCT and DISTINCT ON in the
 703  * parsetree representation.  It's cheaper to just make sure all the Vars
 704  * in the qual refer to DISTINCT columns.
 705  *
 706  * 4. We must not push down any quals that refer to subselect outputs that
 707  * return sets, else we'd introduce functions-returning-sets into the
 708  * subquery's WHERE/HAVING quals.
 709  */
 710 static bool
 711 qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
 712                                           bool *differentTypes)
 713 {
 714         bool            safe = true;
 715         List       *vars;
 716         ListCell   *vl;
 717         Bitmapset  *tested = NULL;
 718
 719         /* Refuse subselects (point 1) */
 720         if (contain_subplans(qual))
 721                 return false;
 722
 723         /*
 724          * Examine all Vars used in clause; since it's a restriction clause,
 725          * all such Vars must refer to subselect output columns.
 726          */
 727         vars = pull_var_clause(qual, false);
 728         foreach(vl, vars)
 729         {
 730                 Var                *var = (Var *) lfirst(vl);
 731                 TargetEntry *tle;
 732
 733                 Assert(var->varno == rti);
 734
 735                 /*
 736                  * We use a bitmapset to avoid testing the same attno more than
 737                  * once.  (NB: this only works because subquery outputs can't have
 738                  * negative attnos.)
 739                  */
 740                 if (bms_is_member(var->varattno, tested))
 741                         continue;
 742                 tested = bms_add_member(tested, var->varattno);
 743
 744                 /* Check point 2 */
 745                 if (differentTypes[var->varattno])
 746                 {
 747                         safe = false;
 748                         break;
 749                 }
 750
 751                 /* Must find the tlist element referenced by the Var */
 752                 tle = get_tle_by_resno(subquery->targetList, var->varattno);
 753                 Assert(tle != NULL);
 754                 Assert(!tle->resdom->resjunk);
 755
 756                 /* If subquery uses DISTINCT or DISTINCT ON, check point 3 */
 757                 if (subquery->distinctClause != NIL &&
 758                         !targetIsInSortList(tle, subquery->distinctClause))
 759                 {
 760                         /* non-DISTINCT column, so fail */
 761                         safe = false;
 762                         break;
 763                 }
 764
 765                 /* Refuse functions returning sets (point 4) */
 766                 if (expression_returns_set((Node *) tle->expr))
 767                 {
 768                         safe = false;
 769                         break;
 770                 }
 771         }
 772
 773         list_free(vars);
 774         bms_free(tested);
 775
 776         return safe;
 777 }
 778
 779 /*
 780  * subquery_push_qual - push down a qual that we have determined is safe
 781  */
 782 static void
 783 subquery_push_qual(Query *subquery, List *rtable, Index rti, Node *qual)
 784 {
 785         if (subquery->setOperations != NULL)
 786         {
 787                 /* Recurse to push it separately to each component query */
 788                 recurse_push_qual(subquery->setOperations, subquery,
 789                                                   rtable, rti, qual);
 790         }
 791         else
 792         {
 793                 /*
 794                  * We need to replace Vars in the qual (which must refer to
 795                  * outputs of the subquery) with copies of the subquery's
 796                  * targetlist expressions.      Note that at this point, any uplevel
 797                  * Vars in the qual should have been replaced with Params, so they
 798                  * need no work.
 799                  *
 800                  * This step also ensures that when we are pushing into a setop tree,
 801                  * each component query gets its own copy of the qual.
 802                  */
 803                 qual = ResolveNew(qual, rti, 0, rtable,
 804                                                   subquery->targetList,
 805                                                   CMD_SELECT, 0);
 806                 subquery->havingQual = make_and_qual(subquery->havingQual,
 807                                                                                          qual);
 808
 809                 /*
 810                  * We need not change the subquery's hasAggs or hasSublinks flags,
 811                  * since we can't be pushing down any aggregates that weren't
 812                  * there before, and we don't push down subselects at all.
 813                  */
 814         }
 815 }
 816
 817 /*
 818  * Helper routine to recurse through setOperations tree
 819  */
 820 static void
 821 recurse_push_qual(Node *setOp, Query *topquery,
 822                                   List *rtable, Index rti, Node *qual)
 823 {
 824         if (IsA(setOp, RangeTblRef))
 825         {
 826                 RangeTblRef *rtr = (RangeTblRef *) setOp;
 827                 RangeTblEntry *subrte = rt_fetch(rtr->rtindex, topquery->rtable);
 828                 Query      *subquery = subrte->subquery;
 829
 830                 Assert(subquery != NULL);
 831                 subquery_push_qual(subquery, rtable, rti, qual);
 832         }
 833         else if (IsA(setOp, SetOperationStmt))
 834         {
 835                 SetOperationStmt *op = (SetOperationStmt *) setOp;
 836
 837                 recurse_push_qual(op->larg, topquery, rtable, rti, qual);
 838                 recurse_push_qual(op->rarg, topquery, rtable, rti, qual);
 839         }
 840         else
 841         {
 842                 elog(ERROR, "unrecognized node type: %d",
 843                          (int) nodeTag(setOp));
 844         }
 845 }
 846
 847 /*****************************************************************************
 848  *                      DEBUG SUPPORT
 849  *****************************************************************************/
 850
 851 #ifdef OPTIMIZER_DEBUG
 852
 853 static void
 854 print_relids(Relids relids)
 855 {
 856         Relids          tmprelids;
 857         int                     x;
 858         bool            first = true;
 859
 860         tmprelids = bms_copy(relids);
 861         while ((x = bms_first_member(tmprelids)) >= 0)
 862         {
 863                 if (!first)
 864                         printf(" ");
 865                 printf("%d", x);
 866                 first = false;
 867         }
 868         bms_free(tmprelids);
 869 }
 870
 871 static void
 872 print_restrictclauses(Query *root, List *clauses)
 873 {
 874         ListCell   *l;
 875
 876         foreach(l, clauses)
 877         {
 878                 RestrictInfo *c = lfirst(l);
 879
 880                 print_expr((Node *) c->clause, root->rtable);
 881                 if (lnext(l))
 882                         printf(", ");
 883         }
 884 }
 885
 886 static void
 887 print_path(Query *root, Path *path, int indent)
 888 {
 889         const char *ptype;
 890         bool            join = false;
 891         Path       *subpath = NULL;
 892         int                     i;
 893
 894         switch (nodeTag(path))
 895         {
 896                 case T_Path:
 897                         ptype = "SeqScan";
 898                         break;
 899                 case T_IndexPath:
 900                         ptype = "IdxScan";
 901                         break;
 902                 case T_TidPath:
 903                         ptype = "TidScan";
 904                         break;
 905                 case T_AppendPath:
 906                         ptype = "Append";
 907                         break;
 908                 case T_ResultPath:
 909                         ptype = "Result";
 910                         subpath = ((ResultPath *) path)->subpath;
 911                         break;
 912                 case T_MaterialPath:
 913                         ptype = "Material";
 914                         subpath = ((MaterialPath *) path)->subpath;
 915                         break;
 916                 case T_UniquePath:
 917                         ptype = "Unique";
 918                         subpath = ((UniquePath *) path)->subpath;
 919                         break;
 920                 case T_NestPath:
 921                         ptype = "NestLoop";
 922                         join = true;
 923                         break;
 924                 case T_MergePath:
 925                         ptype = "MergeJoin";
 926                         join = true;
 927                         break;
 928                 case T_HashPath:
 929                         ptype = "HashJoin";
 930                         join = true;
 931                         break;
 932                 default:
 933                         ptype = "???Path";
 934                         break;
 935         }
 936
 937         for (i = 0; i < indent; i++)
 938                 printf("\t");
 939         printf("%s", ptype);
 940
 941         if (path->parent)
 942         {
 943                 printf("(");
 944                 print_relids(path->parent->relids);
 945                 printf(") rows=%.0f", path->parent->rows);
 946         }
 947         printf(" cost=%.2f..%.2f\n", path->startup_cost, path->total_cost);
 948
 949         if (path->pathkeys)
 950         {
 951                 for (i = 0; i < indent; i++)
 952                         printf("\t");
 953                 printf("  pathkeys: ");
 954                 print_pathkeys(path->pathkeys, root->rtable);
 955         }
 956
 957         if (join)
 958         {
 959                 JoinPath   *jp = (JoinPath *) path;
 960
 961                 for (i = 0; i < indent; i++)
 962                         printf("\t");
 963                 printf("  clauses: ");
 964                 print_restrictclauses(root, jp->joinrestrictinfo);
 965                 printf("\n");
 966
 967                 if (IsA(path, MergePath))
 968                 {
 969                         MergePath  *mp = (MergePath *) path;
 970
 971                         if (mp->outersortkeys || mp->innersortkeys)
 972                         {
 973                                 for (i = 0; i < indent; i++)
 974                                         printf("\t");
 975                                 printf("  sortouter=%d sortinner=%d\n",
 976                                            ((mp->outersortkeys) ? 1 : 0),
 977                                            ((mp->innersortkeys) ? 1 : 0));
 978                         }
 979                 }
 980
 981                 print_path(root, jp->outerjoinpath, indent + 1);
 982                 print_path(root, jp->innerjoinpath, indent + 1);
 983         }
 984
 985         if (subpath)
 986                 print_path(root, subpath, indent + 1);
 987 }
 988
 989 void
 990 debug_print_rel(Query *root, RelOptInfo *rel)
 991 {
 992         ListCell   *l;
 993
 994         printf("RELOPTINFO (");
 995         print_relids(rel->relids);
 996         printf("): rows=%.0f width=%d\n", rel->rows, rel->width);
 997
 998         if (rel->baserestrictinfo)
 999         {
1000                 printf("\tbaserestrictinfo: ");
1001                 print_restrictclauses(root, rel->baserestrictinfo);
1002                 printf("\n");
1003         }
1004
1005         foreach(l, rel->joininfo)
1006         {
1007                 JoinInfo   *j = (JoinInfo *) lfirst(l);
1008
1009                 printf("\tjoininfo (");
1010                 print_relids(j->unjoined_relids);
1011                 printf("): ");
1012                 print_restrictclauses(root, j->jinfo_restrictinfo);
1013                 printf("\n");
1014         }
1015
1016         printf("\tpath list:\n");
1017         foreach(l, rel->pathlist)
1018                 print_path(root, lfirst(l), 1);
1019         printf("\n\tcheapest startup path:\n");
1020         print_path(root, rel->cheapest_startup_path, 1);
1021         printf("\n\tcheapest total path:\n");
1022         print_path(root, rel->cheapest_total_path, 1);
1023         printf("\n");
1024         fflush(stdout);
1025 }
1026
1027 #endif   /* OPTIMIZER_DEBUG */