make_join_rel.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  * make_join_rel.c
   4  *        Routines copied from PostgreSQL core distribution.
   5  *
   6  * src/backend/optimizer/path/joinrels.c
   7  *     make_join_rel()
   8  *
   9  * Portions Copyright (c) 2013, NIPPON TELEGRAPH AND TELEPHONE CORPORATION
  10  * Portions Copyright (c) 1996-2012, PostgreSQL Global Development Group
  11  * Portions Copyright (c) 1994, Regents of the University of California
  12  *
  13  *-------------------------------------------------------------------------
  14  */
  15
  16 /*
  17  * make_join_rel
  18  *         Find or create a join RelOptInfo that represents the join of
  19  *         the two given rels, and add to it path information for paths
  20  *         created with the two rels as outer and inner rel.
  21  *         (The join rel may already contain paths generated from other
  22  *         pairs of rels that add up to the same set of base rels.)
  23  *
  24  * NB: will return NULL if attempted join is not valid.  This can happen
  25  * when working with outer joins, or with IN or EXISTS clauses that have been
  26  * turned into joins.
  27  */
  28 static double
  29 adjust_rows(double rows, RowsHint *hint)
  30 {
  31         double          result = 0.0;   /* keep compiler quiet */
  32
  33         if (hint->value_type == RVT_ABSOLUTE)
  34                 result = hint->rows;
  35         else if (hint->value_type == RVT_ADD)
  36                 result = rows + hint->rows;
  37         else if (hint->value_type == RVT_SUB)
  38                 result =  rows - hint->rows;
  39         else if (hint->value_type == RVT_MULTI)
  40                 result = rows * hint->rows;
  41         else
  42                 Assert(false);  /* unrecognized rows value type */
  43
  44         hint->base.state = HINT_STATE_USED;
  45         result = clamp_row_est(result);
  46         elog(DEBUG1, "adjusted rows %d to %d", (int) rows, (int) result);
  47
  48         return result;
  49 }
  50
  51 RelOptInfo *
  52 make_join_rel(PlannerInfo *root, RelOptInfo *rel1, RelOptInfo *rel2)
  53 {
  54         Relids          joinrelids;
  55         SpecialJoinInfo *sjinfo;
  56         bool            reversed;
  57         SpecialJoinInfo sjinfo_data;
  58         RelOptInfo *joinrel;
  59         List       *restrictlist;
  60
  61         RowsHint   *rows_hint = NULL;
  62         int                     i;
  63
  64         /* We should never try to join two overlapping sets of rels. */
  65         Assert(!bms_overlap(rel1->relids, rel2->relids));
  66
  67         /* Construct Relids set that identifies the joinrel. */
  68         joinrelids = bms_union(rel1->relids, rel2->relids);
  69
  70         /* Check validity and determine join type. */
  71         if (!join_is_legal(root, rel1, rel2, joinrelids,
  72                                            &sjinfo, &reversed))
  73         {
  74                 /* invalid join path */
  75                 bms_free(joinrelids);
  76                 return NULL;
  77         }
  78
  79         /* Swap rels if needed to match the join info. */
  80         if (reversed)
  81         {
  82                 RelOptInfo *trel = rel1;
  83
  84                 rel1 = rel2;
  85                 rel2 = trel;
  86         }
  87
  88         /*
  89          * If it's a plain inner join, then we won't have found anything in
  90          * join_info_list.      Make up a SpecialJoinInfo so that selectivity
  91          * estimation functions will know what's being joined.
  92          */
  93         if (sjinfo == NULL)
  94         {
  95                 sjinfo = &sjinfo_data;
  96                 sjinfo->type = T_SpecialJoinInfo;
  97                 sjinfo->min_lefthand = rel1->relids;
  98                 sjinfo->min_righthand = rel2->relids;
  99                 sjinfo->syn_lefthand = rel1->relids;
 100                 sjinfo->syn_righthand = rel2->relids;
 101                 sjinfo->jointype = JOIN_INNER;
 102                 /* we don't bother trying to make the remaining fields valid */
 103                 sjinfo->lhs_strict = false;
 104                 sjinfo->delay_upper_joins = false;
 105                 sjinfo->join_quals = NIL;
 106         }
 107
 108         /*
 109          * Find or build the join RelOptInfo, and compute the restrictlist that
 110          * goes with this particular joining.
 111          */
 112         joinrel = build_join_rel(root, joinrelids, rel1, rel2, sjinfo,
 113                                                          &restrictlist);
 114
 115         /* Apply appropriate Rows hint to the join node, if any. */
 116         for (i = 0; i < current_hint->num_hints[HINT_TYPE_ROWS]; i++)
 117         {
 118                 rows_hint = current_hint->rows_hints[i];
 119
 120                 if (bms_equal(joinrelids, rows_hint->joinrelids))
 121                 {
 122                         /*
 123                          * This join RelOptInfo is exactly a Rows hint specifies, so adjust
 124                          * rows estimateion with the hint's content.  Here we never have
 125                          * another hint which has same relation combination, so we can skip
 126                          * rest of hints.
 127                          */
 128                         if (rows_hint->base.state == HINT_STATE_NOTUSED)
 129                                 joinrel->rows = adjust_rows(joinrel->rows, rows_hint);
 130                 }
 131                 else if (bms_is_subset(rows_hint->joinrelids, rel1->relids) ||
 132                                  bms_is_subset(rows_hint->joinrelids, rel2->relids))
 133                 {
 134                         /*
 135                          * Otherwise if the relation combination specified in thee Rows
 136                          * hint is subset of the set of join elements, re-estimate rows and
 137                          * costs again to reflect the adjustment done in down.  This is
 138                          * necessary for the first permutation of the combination the
 139                          * relations, but it's difficult to determine that this is the
 140                          * first, so do this everytime.
 141                          */
 142                         set_joinrel_size_estimates(root, joinrel, rel1, rel2, sjinfo,
 143                                                                            restrictlist);
 144                 }
 145                 else if (bms_is_subset(rows_hint->joinrelids, joinrelids))
 146                 {
 147                         /*
 148                          * If the combination specifed in the Rows hints is subset of the
 149                          * join relation and spreads over both children,
 150                          *
 151                          * We do adjust rows estimation only when the value type was
 152                          * multiplication, because other value types are meanless.
 153                          */
 154                         if (rows_hint->value_type == RVT_MULTI)
 155                         {
 156                                 set_joinrel_size_estimates(root, joinrel, rel1, rel2, sjinfo,
 157                                                                                    restrictlist);
 158                                 joinrel->rows = adjust_rows(joinrel->rows, rows_hint);
 159                         }
 160                 }
 161         }
 162
 163         /*
 164          * If we've already proven this join is empty, we needn't consider any
 165          * more paths for it.
 166          */
 167         if (is_dummy_rel(joinrel))
 168         {
 169                 bms_free(joinrelids);
 170                 return joinrel;
 171         }
 172
 173         /*
 174          * Consider paths using each rel as both outer and inner.  Depending on
 175          * the join type, a provably empty outer or inner rel might mean the join
 176          * is provably empty too; in which case throw away any previously computed
 177          * paths and mark the join as dummy.  (We do it this way since it's
 178          * conceivable that dummy-ness of a multi-element join might only be
 179          * noticeable for certain construction paths.)
 180          *
 181          * Also, a provably constant-false join restriction typically means that
 182          * we can skip evaluating one or both sides of the join.  We do this by
 183          * marking the appropriate rel as dummy.  For outer joins, a
 184          * constant-false restriction that is pushed down still means the whole
 185          * join is dummy, while a non-pushed-down one means that no inner rows
 186          * will join so we can treat the inner rel as dummy.
 187          *
 188          * We need only consider the jointypes that appear in join_info_list, plus
 189          * JOIN_INNER.
 190          */
 191         switch (sjinfo->jointype)
 192         {
 193                 case JOIN_INNER:
 194                         if (is_dummy_rel(rel1) || is_dummy_rel(rel2) ||
 195                                 restriction_is_constant_false(restrictlist, false))
 196                         {
 197                                 mark_dummy_rel(joinrel);
 198                                 break;
 199                         }
 200                         add_paths_to_joinrel(root, joinrel, rel1, rel2,
 201                                                                  JOIN_INNER, sjinfo,
 202                                                                  restrictlist);
 203                         add_paths_to_joinrel(root, joinrel, rel2, rel1,
 204                                                                  JOIN_INNER, sjinfo,
 205                                                                  restrictlist);
 206                         break;
 207                 case JOIN_LEFT:
 208                         if (is_dummy_rel(rel1) ||
 209                                 restriction_is_constant_false(restrictlist, true))
 210                         {
 211                                 mark_dummy_rel(joinrel);
 212                                 break;
 213                         }
 214                         if (restriction_is_constant_false(restrictlist, false) &&
 215                                 bms_is_subset(rel2->relids, sjinfo->syn_righthand))
 216                                 mark_dummy_rel(rel2);
 217                         add_paths_to_joinrel(root, joinrel, rel1, rel2,
 218                                                                  JOIN_LEFT, sjinfo,
 219                                                                  restrictlist);
 220                         add_paths_to_joinrel(root, joinrel, rel2, rel1,
 221                                                                  JOIN_RIGHT, sjinfo,
 222                                                                  restrictlist);
 223                         break;
 224                 case JOIN_FULL:
 225                         if ((is_dummy_rel(rel1) && is_dummy_rel(rel2)) ||
 226                                 restriction_is_constant_false(restrictlist, true))
 227                         {
 228                                 mark_dummy_rel(joinrel);
 229                                 break;
 230                         }
 231                         add_paths_to_joinrel(root, joinrel, rel1, rel2,
 232                                                                  JOIN_FULL, sjinfo,
 233                                                                  restrictlist);
 234                         add_paths_to_joinrel(root, joinrel, rel2, rel1,
 235                                                                  JOIN_FULL, sjinfo,
 236                                                                  restrictlist);
 237
 238                         /*
 239                          * If there are join quals that aren't mergeable or hashable, we
 240                          * may not be able to build any valid plan.  Complain here so that
 241                          * we can give a somewhat-useful error message.  (Since we have no
 242                          * flexibility of planning for a full join, there's no chance of
 243                          * succeeding later with another pair of input rels.)
 244                          */
 245                         if (joinrel->pathlist == NIL)
 246                                 ereport(ERROR,
 247                                                 (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
 248                                                  errmsg("FULL JOIN is only supported with merge-joinable or hash-joinable join conditions")));
 249                         break;
 250                 case JOIN_SEMI:
 251
 252                         /*
 253                          * We might have a normal semijoin, or a case where we don't have
 254                          * enough rels to do the semijoin but can unique-ify the RHS and
 255                          * then do an innerjoin (see comments in join_is_legal).  In the
 256                          * latter case we can't apply JOIN_SEMI joining.
 257                          */
 258                         if (bms_is_subset(sjinfo->min_lefthand, rel1->relids) &&
 259                                 bms_is_subset(sjinfo->min_righthand, rel2->relids))
 260                         {
 261                                 if (is_dummy_rel(rel1) || is_dummy_rel(rel2) ||
 262                                         restriction_is_constant_false(restrictlist, false))
 263                                 {
 264                                         mark_dummy_rel(joinrel);
 265                                         break;
 266                                 }
 267                                 add_paths_to_joinrel(root, joinrel, rel1, rel2,
 268                                                                          JOIN_SEMI, sjinfo,
 269                                                                          restrictlist);
 270                         }
 271
 272                         /*
 273                          * If we know how to unique-ify the RHS and one input rel is
 274                          * exactly the RHS (not a superset) we can consider unique-ifying
 275                          * it and then doing a regular join.  (The create_unique_path
 276                          * check here is probably redundant with what join_is_legal did,
 277                          * but if so the check is cheap because it's cached.  So test
 278                          * anyway to be sure.)
 279                          */
 280                         if (bms_equal(sjinfo->syn_righthand, rel2->relids) &&
 281                                 create_unique_path(root, rel2, rel2->cheapest_total_path,
 282                                                                    sjinfo) != NULL)
 283                         {
 284                                 if (is_dummy_rel(rel1) || is_dummy_rel(rel2) ||
 285                                         restriction_is_constant_false(restrictlist, false))
 286                                 {
 287                                         mark_dummy_rel(joinrel);
 288                                         break;
 289                                 }
 290                                 add_paths_to_joinrel(root, joinrel, rel1, rel2,
 291                                                                          JOIN_UNIQUE_INNER, sjinfo,
 292                                                                          restrictlist);
 293                                 add_paths_to_joinrel(root, joinrel, rel2, rel1,
 294                                                                          JOIN_UNIQUE_OUTER, sjinfo,
 295                                                                          restrictlist);
 296                         }
 297                         break;
 298                 case JOIN_ANTI:
 299                         if (is_dummy_rel(rel1) ||
 300                                 restriction_is_constant_false(restrictlist, true))
 301                         {
 302                                 mark_dummy_rel(joinrel);
 303                                 break;
 304                         }
 305                         if (restriction_is_constant_false(restrictlist, false) &&
 306                                 bms_is_subset(rel2->relids, sjinfo->syn_righthand))
 307                                 mark_dummy_rel(rel2);
 308                         add_paths_to_joinrel(root, joinrel, rel1, rel2,
 309                                                                  JOIN_ANTI, sjinfo,
 310                                                                  restrictlist);
 311                         break;
 312                 default:
 313                         /* other values not expected here */
 314                         elog(ERROR, "unrecognized join type: %d", (int) sjinfo->jointype);
 315                         break;
 316         }
 317
 318         bms_free(joinrelids);
 319
 320         return joinrel;
 321 }