Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * array_selfuncs.c
4 : * Functions for selectivity estimation of array operators
5 : *
6 : * Portions Copyright (c) 1996-2026, PostgreSQL Global Development Group
7 : * Portions Copyright (c) 1994, Regents of the University of California
8 : *
9 : *
10 : * IDENTIFICATION
11 : * src/backend/utils/adt/array_selfuncs.c
12 : *
13 : *-------------------------------------------------------------------------
14 : */
15 : #include "postgres.h"
16 :
17 : #include <math.h>
18 :
19 : #include "access/htup_details.h"
20 : #include "catalog/pg_operator.h"
21 : #include "catalog/pg_statistic.h"
22 : #include "utils/array.h"
23 : #include "utils/fmgrprotos.h"
24 : #include "utils/lsyscache.h"
25 : #include "utils/selfuncs.h"
26 : #include "utils/typcache.h"
27 :
28 :
29 : /* Default selectivity constant for "@>" and "<@" operators */
30 : #define DEFAULT_CONTAIN_SEL 0.005
31 :
32 : /* Default selectivity constant for "&&" operator */
33 : #define DEFAULT_OVERLAP_SEL 0.01
34 :
35 : /* Default selectivity for given operator */
36 : #define DEFAULT_SEL(operator) \
37 : ((operator) == OID_ARRAY_OVERLAP_OP ? \
38 : DEFAULT_OVERLAP_SEL : DEFAULT_CONTAIN_SEL)
39 :
40 : static Selectivity calc_arraycontsel(VariableStatData *vardata, Datum constval,
41 : Oid elemtype, Oid operator);
42 : static Selectivity mcelem_array_selec(const ArrayType *array,
43 : TypeCacheEntry *typentry,
44 : const Datum *mcelem, int nmcelem,
45 : const float4 *numbers, int nnumbers,
46 : const float4 *hist, int nhist,
47 : Oid operator);
48 : static Selectivity mcelem_array_contain_overlap_selec(const Datum *mcelem, int nmcelem,
49 : const float4 *numbers, int nnumbers,
50 : const Datum *array_data, int nitems,
51 : Oid operator, TypeCacheEntry *typentry);
52 : static Selectivity mcelem_array_contained_selec(const Datum *mcelem, int nmcelem,
53 : const float4 *numbers, int nnumbers,
54 : const Datum *array_data, int nitems,
55 : const float4 *hist, int nhist,
56 : Oid operator, TypeCacheEntry *typentry);
57 : static float *calc_hist(const float4 *hist, int nhist, int n);
58 : static float *calc_distr(const float *p, int n, int m, float rest);
59 : static int floor_log2(uint32 n);
60 : static bool find_next_mcelem(const Datum *mcelem, int nmcelem, Datum value,
61 : int *index, TypeCacheEntry *typentry);
62 : static int element_compare(const void *key1, const void *key2, void *arg);
63 : static int float_compare_desc(const void *key1, const void *key2);
64 :
65 :
66 : /*
67 : * scalararraysel_containment
68 : * Estimate selectivity of ScalarArrayOpExpr via array containment.
69 : *
70 : * If we have const =/<> ANY/ALL (array_var) then we can estimate the
71 : * selectivity as though this were an array containment operator,
72 : * array_var op ARRAY[const].
73 : *
74 : * scalararraysel() has already verified that the ScalarArrayOpExpr's operator
75 : * is the array element type's default equality or inequality operator, and
76 : * has aggressively simplified both inputs to constants.
77 : *
78 : * Returns selectivity (0..1), or -1 if we fail to estimate selectivity.
79 : */
80 : Selectivity
81 11808 : scalararraysel_containment(PlannerInfo *root,
82 : Node *leftop, Node *rightop,
83 : Oid elemtype, bool isEquality, bool useOr,
84 : int varRelid)
85 : {
86 : Selectivity selec;
87 : VariableStatData vardata;
88 : Datum constval;
89 : TypeCacheEntry *typentry;
90 : FmgrInfo *cmpfunc;
91 :
92 : /*
93 : * rightop must be a variable, else punt.
94 : */
95 11808 : examine_variable(root, rightop, varRelid, &vardata);
96 11808 : if (!vardata.rel)
97 : {
98 11536 : ReleaseVariableStats(vardata);
99 11536 : return -1.0;
100 : }
101 :
102 : /*
103 : * leftop must be a constant, else punt.
104 : */
105 272 : if (!IsA(leftop, Const))
106 : {
107 214 : ReleaseVariableStats(vardata);
108 214 : return -1.0;
109 : }
110 58 : if (((Const *) leftop)->constisnull)
111 : {
112 : /* qual can't succeed if null on left */
113 0 : ReleaseVariableStats(vardata);
114 0 : return (Selectivity) 0.0;
115 : }
116 58 : constval = ((Const *) leftop)->constvalue;
117 :
118 : /* Get element type's default comparison function */
119 58 : typentry = lookup_type_cache(elemtype, TYPECACHE_CMP_PROC_FINFO);
120 58 : if (!OidIsValid(typentry->cmp_proc_finfo.fn_oid))
121 : {
122 0 : ReleaseVariableStats(vardata);
123 0 : return -1.0;
124 : }
125 58 : cmpfunc = &typentry->cmp_proc_finfo;
126 :
127 : /*
128 : * If the operator is <>, swap ANY/ALL, then invert the result later.
129 : */
130 58 : if (!isEquality)
131 43 : useOr = !useOr;
132 :
133 : /* Get array element stats for var, if available */
134 64 : if (HeapTupleIsValid(vardata.statsTuple) &&
135 6 : statistic_proc_security_check(&vardata, cmpfunc->fn_oid))
136 6 : {
137 : Form_pg_statistic stats;
138 : AttStatsSlot sslot;
139 : AttStatsSlot hslot;
140 :
141 6 : stats = (Form_pg_statistic) GETSTRUCT(vardata.statsTuple);
142 :
143 : /* MCELEM will be an array of same type as element */
144 6 : if (get_attstatsslot(&sslot, vardata.statsTuple,
145 : STATISTIC_KIND_MCELEM, InvalidOid,
146 : ATTSTATSSLOT_VALUES | ATTSTATSSLOT_NUMBERS))
147 : {
148 : /* For ALL case, also get histogram of distinct-element counts */
149 6 : if (useOr ||
150 0 : !get_attstatsslot(&hslot, vardata.statsTuple,
151 : STATISTIC_KIND_DECHIST, InvalidOid,
152 : ATTSTATSSLOT_NUMBERS))
153 6 : memset(&hslot, 0, sizeof(hslot));
154 :
155 : /*
156 : * For = ANY, estimate as var @> ARRAY[const].
157 : *
158 : * For = ALL, estimate as var <@ ARRAY[const].
159 : */
160 6 : if (useOr)
161 6 : selec = mcelem_array_contain_overlap_selec(sslot.values,
162 : sslot.nvalues,
163 6 : sslot.numbers,
164 : sslot.nnumbers,
165 : &constval, 1,
166 : OID_ARRAY_CONTAINS_OP,
167 : typentry);
168 : else
169 0 : selec = mcelem_array_contained_selec(sslot.values,
170 : sslot.nvalues,
171 0 : sslot.numbers,
172 : sslot.nnumbers,
173 : &constval, 1,
174 0 : hslot.numbers,
175 : hslot.nnumbers,
176 : OID_ARRAY_CONTAINED_OP,
177 : typentry);
178 :
179 6 : free_attstatsslot(&hslot);
180 6 : free_attstatsslot(&sslot);
181 : }
182 : else
183 : {
184 : /* No most-common-elements info, so do without */
185 0 : if (useOr)
186 0 : selec = mcelem_array_contain_overlap_selec(NULL, 0,
187 : NULL, 0,
188 : &constval, 1,
189 : OID_ARRAY_CONTAINS_OP,
190 : typentry);
191 : else
192 0 : selec = mcelem_array_contained_selec(NULL, 0,
193 : NULL, 0,
194 : &constval, 1,
195 : NULL, 0,
196 : OID_ARRAY_CONTAINED_OP,
197 : typentry);
198 : }
199 :
200 : /*
201 : * MCE stats count only non-null rows, so adjust for null rows.
202 : */
203 6 : selec *= (1.0 - stats->stanullfrac);
204 : }
205 : else
206 : {
207 : /* No stats at all, so do without */
208 52 : if (useOr)
209 52 : selec = mcelem_array_contain_overlap_selec(NULL, 0,
210 : NULL, 0,
211 : &constval, 1,
212 : OID_ARRAY_CONTAINS_OP,
213 : typentry);
214 : else
215 0 : selec = mcelem_array_contained_selec(NULL, 0,
216 : NULL, 0,
217 : &constval, 1,
218 : NULL, 0,
219 : OID_ARRAY_CONTAINED_OP,
220 : typentry);
221 : /* we assume no nulls here, so no stanullfrac correction */
222 : }
223 :
224 58 : ReleaseVariableStats(vardata);
225 :
226 : /*
227 : * If the operator is <>, invert the results.
228 : */
229 58 : if (!isEquality)
230 43 : selec = 1.0 - selec;
231 :
232 58 : CLAMP_PROBABILITY(selec);
233 :
234 58 : return selec;
235 : }
236 :
237 : /*
238 : * arraycontsel -- restriction selectivity for array @>, &&, <@ operators
239 : */
240 : Datum
241 492 : arraycontsel(PG_FUNCTION_ARGS)
242 : {
243 492 : PlannerInfo *root = (PlannerInfo *) PG_GETARG_POINTER(0);
244 492 : Oid operator = PG_GETARG_OID(1);
245 492 : List *args = (List *) PG_GETARG_POINTER(2);
246 492 : int varRelid = PG_GETARG_INT32(3);
247 : VariableStatData vardata;
248 : Node *other;
249 : bool varonleft;
250 : Selectivity selec;
251 : Oid element_typeid;
252 :
253 : /*
254 : * If expression is not (variable op something) or (something op
255 : * variable), then punt and return a default estimate.
256 : */
257 492 : if (!get_restriction_variable(root, args, varRelid,
258 : &vardata, &other, &varonleft))
259 0 : PG_RETURN_FLOAT8(DEFAULT_SEL(operator));
260 :
261 : /*
262 : * Can't do anything useful if the something is not a constant, either.
263 : */
264 492 : if (!IsA(other, Const))
265 : {
266 0 : ReleaseVariableStats(vardata);
267 0 : PG_RETURN_FLOAT8(DEFAULT_SEL(operator));
268 : }
269 :
270 : /*
271 : * The "&&", "@>" and "<@" operators are strict, so we can cope with a
272 : * NULL constant right away.
273 : */
274 492 : if (((Const *) other)->constisnull)
275 : {
276 0 : ReleaseVariableStats(vardata);
277 0 : PG_RETURN_FLOAT8(0.0);
278 : }
279 :
280 : /*
281 : * If var is on the right, commute the operator, so that we can assume the
282 : * var is on the left in what follows.
283 : */
284 492 : if (!varonleft)
285 : {
286 12 : if (operator == OID_ARRAY_CONTAINS_OP)
287 0 : operator = OID_ARRAY_CONTAINED_OP;
288 12 : else if (operator == OID_ARRAY_CONTAINED_OP)
289 12 : operator = OID_ARRAY_CONTAINS_OP;
290 : }
291 :
292 : /*
293 : * OK, there's a Var and a Const we're dealing with here. We need the
294 : * Const to be an array with same element type as column, else we can't do
295 : * anything useful. (Such cases will likely fail at runtime, but here
296 : * we'd rather just return a default estimate.)
297 : */
298 492 : element_typeid = get_base_element_type(((Const *) other)->consttype);
299 984 : if (element_typeid != InvalidOid &&
300 492 : element_typeid == get_base_element_type(vardata.vartype))
301 : {
302 492 : selec = calc_arraycontsel(&vardata, ((Const *) other)->constvalue,
303 : element_typeid, operator);
304 : }
305 : else
306 : {
307 0 : selec = DEFAULT_SEL(operator);
308 : }
309 :
310 492 : ReleaseVariableStats(vardata);
311 :
312 492 : CLAMP_PROBABILITY(selec);
313 :
314 492 : PG_RETURN_FLOAT8((float8) selec);
315 : }
316 :
317 : /*
318 : * arraycontjoinsel -- join selectivity for array @>, &&, <@ operators
319 : */
320 : Datum
321 0 : arraycontjoinsel(PG_FUNCTION_ARGS)
322 : {
323 : /* For the moment this is just a stub */
324 0 : Oid operator = PG_GETARG_OID(1);
325 :
326 0 : PG_RETURN_FLOAT8(DEFAULT_SEL(operator));
327 : }
328 :
329 : /*
330 : * Calculate selectivity for "arraycolumn @> const", "arraycolumn && const"
331 : * or "arraycolumn <@ const" based on the statistics
332 : *
333 : * This function is mainly responsible for extracting the pg_statistic data
334 : * to be used; we then pass the problem on to mcelem_array_selec().
335 : */
336 : static Selectivity
337 492 : calc_arraycontsel(VariableStatData *vardata, Datum constval,
338 : Oid elemtype, Oid operator)
339 : {
340 : Selectivity selec;
341 : TypeCacheEntry *typentry;
342 : FmgrInfo *cmpfunc;
343 : ArrayType *array;
344 :
345 : /* Get element type's default comparison function */
346 492 : typentry = lookup_type_cache(elemtype, TYPECACHE_CMP_PROC_FINFO);
347 492 : if (!OidIsValid(typentry->cmp_proc_finfo.fn_oid))
348 0 : return DEFAULT_SEL(operator);
349 492 : cmpfunc = &typentry->cmp_proc_finfo;
350 :
351 : /*
352 : * The caller made sure the const is an array with same element type, so
353 : * get it now
354 : */
355 492 : array = DatumGetArrayTypeP(constval);
356 :
357 726 : if (HeapTupleIsValid(vardata->statsTuple) &&
358 234 : statistic_proc_security_check(vardata, cmpfunc->fn_oid))
359 234 : {
360 : Form_pg_statistic stats;
361 : AttStatsSlot sslot;
362 : AttStatsSlot hslot;
363 :
364 234 : stats = (Form_pg_statistic) GETSTRUCT(vardata->statsTuple);
365 :
366 : /* MCELEM will be an array of same type as column */
367 234 : if (get_attstatsslot(&sslot, vardata->statsTuple,
368 : STATISTIC_KIND_MCELEM, InvalidOid,
369 : ATTSTATSSLOT_VALUES | ATTSTATSSLOT_NUMBERS))
370 : {
371 : /*
372 : * For "array <@ const" case we also need histogram of distinct
373 : * element counts.
374 : */
375 234 : if (operator != OID_ARRAY_CONTAINED_OP ||
376 36 : !get_attstatsslot(&hslot, vardata->statsTuple,
377 : STATISTIC_KIND_DECHIST, InvalidOid,
378 : ATTSTATSSLOT_NUMBERS))
379 198 : memset(&hslot, 0, sizeof(hslot));
380 :
381 : /* Use the most-common-elements slot for the array Var. */
382 234 : selec = mcelem_array_selec(array, typentry,
383 234 : sslot.values, sslot.nvalues,
384 234 : sslot.numbers, sslot.nnumbers,
385 234 : hslot.numbers, hslot.nnumbers,
386 : operator);
387 :
388 234 : free_attstatsslot(&hslot);
389 234 : free_attstatsslot(&sslot);
390 : }
391 : else
392 : {
393 : /* No most-common-elements info, so do without */
394 0 : selec = mcelem_array_selec(array, typentry,
395 : NULL, 0, NULL, 0, NULL, 0,
396 : operator);
397 : }
398 :
399 : /*
400 : * MCE stats count only non-null rows, so adjust for null rows.
401 : */
402 234 : selec *= (1.0 - stats->stanullfrac);
403 : }
404 : else
405 : {
406 : /* No stats at all, so do without */
407 258 : selec = mcelem_array_selec(array, typentry,
408 : NULL, 0, NULL, 0, NULL, 0,
409 : operator);
410 : /* we assume no nulls here, so no stanullfrac correction */
411 : }
412 :
413 : /* If constant was toasted, release the copy we made */
414 492 : if (PointerGetDatum(array) != constval)
415 0 : pfree(array);
416 :
417 492 : return selec;
418 : }
419 :
420 : /*
421 : * Array selectivity estimation based on most common elements statistics
422 : *
423 : * This function just deconstructs and sorts the array constant's contents,
424 : * and then passes the problem on to mcelem_array_contain_overlap_selec or
425 : * mcelem_array_contained_selec depending on the operator.
426 : */
427 : static Selectivity
428 492 : mcelem_array_selec(const ArrayType *array, TypeCacheEntry *typentry,
429 : const Datum *mcelem, int nmcelem,
430 : const float4 *numbers, int nnumbers,
431 : const float4 *hist, int nhist,
432 : Oid operator)
433 : {
434 : Selectivity selec;
435 : int num_elems;
436 : Datum *elem_values;
437 : bool *elem_nulls;
438 : bool null_present;
439 : int nonnull_nitems;
440 : int i;
441 :
442 : /*
443 : * Prepare constant array data for sorting. Sorting lets us find unique
444 : * elements and efficiently merge with the MCELEM array.
445 : */
446 492 : deconstruct_array(array,
447 : typentry->type_id,
448 492 : typentry->typlen,
449 492 : typentry->typbyval,
450 492 : typentry->typalign,
451 : &elem_values, &elem_nulls, &num_elems);
452 :
453 : /* Collapse out any null elements */
454 492 : nonnull_nitems = 0;
455 492 : null_present = false;
456 2462 : for (i = 0; i < num_elems; i++)
457 : {
458 1970 : if (elem_nulls[i])
459 18 : null_present = true;
460 : else
461 1952 : elem_values[nonnull_nitems++] = elem_values[i];
462 : }
463 :
464 : /*
465 : * Query "column @> '{anything, null}'" matches nothing. For the other
466 : * two operators, presence of a null in the constant can be ignored.
467 : */
468 492 : if (null_present && operator == OID_ARRAY_CONTAINS_OP)
469 : {
470 6 : pfree(elem_values);
471 6 : pfree(elem_nulls);
472 6 : return (Selectivity) 0.0;
473 : }
474 :
475 : /* Sort extracted elements using their default comparison function. */
476 486 : qsort_arg(elem_values, nonnull_nitems, sizeof(Datum),
477 : element_compare, typentry);
478 :
479 : /* Separate cases according to operator */
480 486 : if (operator == OID_ARRAY_CONTAINS_OP || operator == OID_ARRAY_OVERLAP_OP)
481 449 : selec = mcelem_array_contain_overlap_selec(mcelem, nmcelem,
482 : numbers, nnumbers,
483 : elem_values, nonnull_nitems,
484 : operator, typentry);
485 37 : else if (operator == OID_ARRAY_CONTAINED_OP)
486 37 : selec = mcelem_array_contained_selec(mcelem, nmcelem,
487 : numbers, nnumbers,
488 : elem_values, nonnull_nitems,
489 : hist, nhist,
490 : operator, typentry);
491 : else
492 : {
493 0 : elog(ERROR, "arraycontsel called for unrecognized operator %u",
494 : operator);
495 : selec = 0.0; /* keep compiler quiet */
496 : }
497 :
498 486 : pfree(elem_values);
499 486 : pfree(elem_nulls);
500 486 : return selec;
501 : }
502 :
503 : /*
504 : * Estimate selectivity of "column @> const" and "column && const" based on
505 : * most common element statistics. This estimation assumes element
506 : * occurrences are independent.
507 : *
508 : * mcelem (of length nmcelem) and numbers (of length nnumbers) are from
509 : * the array column's MCELEM statistics slot, or are NULL/0 if stats are
510 : * not available. array_data (of length nitems) is the constant's elements.
511 : *
512 : * Both the mcelem and array_data arrays are assumed presorted according
513 : * to the element type's cmpfunc. Null elements are not present.
514 : *
515 : * TODO: this estimate probably could be improved by using the distinct
516 : * elements count histogram. For example, excepting the special case of
517 : * "column @> '{}'", we can multiply the calculated selectivity by the
518 : * fraction of nonempty arrays in the column.
519 : */
520 : static Selectivity
521 507 : mcelem_array_contain_overlap_selec(const Datum *mcelem, int nmcelem,
522 : const float4 *numbers, int nnumbers,
523 : const Datum *array_data, int nitems,
524 : Oid operator, TypeCacheEntry *typentry)
525 : {
526 : Selectivity selec,
527 : elem_selec;
528 : int mcelem_index,
529 : i;
530 : bool use_bsearch;
531 : float4 minfreq;
532 :
533 : /*
534 : * There should be three more Numbers than Values, because the last three
535 : * cells should hold minimal and maximal frequency among the non-null
536 : * elements, and then the frequency of null elements. Ignore the Numbers
537 : * if not right.
538 : */
539 507 : if (nnumbers != nmcelem + 3)
540 : {
541 309 : numbers = NULL;
542 309 : nnumbers = 0;
543 : }
544 :
545 507 : if (numbers)
546 : {
547 : /* Grab the minimal MCE frequency */
548 198 : minfreq = numbers[nmcelem];
549 : }
550 : else
551 : {
552 : /*
553 : * Without statistics, use DEFAULT_CONTAIN_SEL (the factor of 2 will
554 : * be removed again below).
555 : */
556 309 : minfreq = 2 * (float4) DEFAULT_CONTAIN_SEL;
557 : }
558 :
559 : /* Decide whether it is faster to use binary search or not. */
560 507 : if (nitems * floor_log2((uint32) nmcelem) < nmcelem + nitems)
561 408 : use_bsearch = true;
562 : else
563 99 : use_bsearch = false;
564 :
565 507 : if (operator == OID_ARRAY_CONTAINS_OP)
566 : {
567 : /*
568 : * Initial selectivity for "column @> const" query is 1.0, and it will
569 : * be decreased with each element of constant array.
570 : */
571 429 : selec = 1.0;
572 : }
573 : else
574 : {
575 : /*
576 : * Initial selectivity for "column && const" query is 0.0, and it will
577 : * be increased with each element of constant array.
578 : */
579 78 : selec = 0.0;
580 : }
581 :
582 : /* Scan mcelem and array in parallel. */
583 507 : mcelem_index = 0;
584 2436 : for (i = 0; i < nitems; i++)
585 : {
586 1929 : bool match = false;
587 :
588 : /* Ignore any duplicates in the array data. */
589 3489 : if (i > 0 &&
590 1560 : element_compare(&array_data[i - 1], &array_data[i], typentry) == 0)
591 0 : continue;
592 :
593 : /* Find the smallest MCELEM >= this array item. */
594 1929 : if (use_bsearch)
595 : {
596 1929 : match = find_next_mcelem(mcelem, nmcelem, array_data[i],
597 : &mcelem_index, typentry);
598 : }
599 : else
600 : {
601 0 : while (mcelem_index < nmcelem)
602 : {
603 0 : int cmp = element_compare(&mcelem[mcelem_index],
604 0 : &array_data[i],
605 : typentry);
606 :
607 0 : if (cmp < 0)
608 0 : mcelem_index++;
609 : else
610 : {
611 0 : if (cmp == 0)
612 0 : match = true; /* mcelem is found */
613 0 : break;
614 : }
615 : }
616 : }
617 :
618 1929 : if (match && numbers)
619 : {
620 : /* MCELEM matches the array item; use its frequency. */
621 204 : elem_selec = numbers[mcelem_index];
622 204 : mcelem_index++;
623 : }
624 : else
625 : {
626 : /*
627 : * The element is not in MCELEM. Estimate its frequency as half
628 : * that of the least-frequent MCE. (We know it cannot be more
629 : * than minfreq, and it could be a great deal less. Half seems
630 : * like a good compromise.) For probably-historical reasons,
631 : * clamp to not more than DEFAULT_CONTAIN_SEL.
632 : */
633 1725 : elem_selec = Min(DEFAULT_CONTAIN_SEL, minfreq / 2);
634 : }
635 :
636 : /*
637 : * Update overall selectivity using the current element's selectivity
638 : * and an assumption of element occurrence independence.
639 : */
640 1929 : if (operator == OID_ARRAY_CONTAINS_OP)
641 355 : selec *= elem_selec;
642 : else
643 1574 : selec = selec + elem_selec - selec * elem_selec;
644 :
645 : /* Clamp intermediate results to stay sane despite roundoff error */
646 1929 : CLAMP_PROBABILITY(selec);
647 : }
648 :
649 507 : return selec;
650 : }
651 :
652 : /*
653 : * Estimate selectivity of "column <@ const" based on most common element
654 : * statistics.
655 : *
656 : * mcelem (of length nmcelem) and numbers (of length nnumbers) are from
657 : * the array column's MCELEM statistics slot, or are NULL/0 if stats are
658 : * not available. array_data (of length nitems) is the constant's elements.
659 : * hist (of length nhist) is from the array column's DECHIST statistics slot,
660 : * or is NULL/0 if those stats are not available.
661 : *
662 : * Both the mcelem and array_data arrays are assumed presorted according
663 : * to the element type's cmpfunc. Null elements are not present.
664 : *
665 : * Independent element occurrence would imply a particular distribution of
666 : * distinct element counts among matching rows. Real data usually falsifies
667 : * that assumption. For example, in a set of 11-element integer arrays having
668 : * elements in the range [0..10], element occurrences are typically not
669 : * independent. If they were, a sufficiently-large set would include all
670 : * distinct element counts 0 through 11. We correct for this using the
671 : * histogram of distinct element counts.
672 : *
673 : * In the "column @> const" and "column && const" cases, we usually have a
674 : * "const" with low number of elements (otherwise we have selectivity close
675 : * to 0 or 1 respectively). That's why the effect of dependence related
676 : * to distinct element count distribution is negligible there. In the
677 : * "column <@ const" case, number of elements is usually high (otherwise we
678 : * have selectivity close to 0). That's why we should do a correction with
679 : * the array distinct element count distribution here.
680 : *
681 : * Using the histogram of distinct element counts produces a different
682 : * distribution law than independent occurrences of elements. This
683 : * distribution law can be described as follows:
684 : *
685 : * P(o1, o2, ..., on) = f1^o1 * (1 - f1)^(1 - o1) * f2^o2 *
686 : * (1 - f2)^(1 - o2) * ... * fn^on * (1 - fn)^(1 - on) * hist[m] / ind[m]
687 : *
688 : * where:
689 : * o1, o2, ..., on - occurrences of elements 1, 2, ..., n
690 : * (1 - occurrence, 0 - no occurrence) in row
691 : * f1, f2, ..., fn - frequencies of elements 1, 2, ..., n
692 : * (scalar values in [0..1]) according to collected statistics
693 : * m = o1 + o2 + ... + on = total number of distinct elements in row
694 : * hist[m] - histogram data for occurrence of m elements.
695 : * ind[m] - probability of m occurrences from n events assuming their
696 : * probabilities to be equal to frequencies of array elements.
697 : *
698 : * ind[m] = sum(f1^o1 * (1 - f1)^(1 - o1) * f2^o2 * (1 - f2)^(1 - o2) *
699 : * ... * fn^on * (1 - fn)^(1 - on), o1, o2, ..., on) | o1 + o2 + .. on = m
700 : */
701 : static Selectivity
702 37 : mcelem_array_contained_selec(const Datum *mcelem, int nmcelem,
703 : const float4 *numbers, int nnumbers,
704 : const Datum *array_data, int nitems,
705 : const float4 *hist, int nhist,
706 : Oid operator, TypeCacheEntry *typentry)
707 : {
708 : int mcelem_index,
709 : i,
710 37 : unique_nitems = 0;
711 : float selec,
712 : minfreq,
713 : nullelem_freq;
714 : float *dist,
715 : *mcelem_dist,
716 : *hist_part;
717 : float avg_count,
718 : mult,
719 : rest;
720 : float *elem_selec;
721 :
722 : /*
723 : * There should be three more Numbers than Values in the MCELEM slot,
724 : * because the last three cells should hold minimal and maximal frequency
725 : * among the non-null elements, and then the frequency of null elements.
726 : * Punt if not right, because we can't do much without the element freqs.
727 : */
728 37 : if (numbers == NULL || nnumbers != nmcelem + 3)
729 1 : return DEFAULT_CONTAIN_SEL;
730 :
731 : /* Can't do much without a count histogram, either */
732 36 : if (hist == NULL || nhist < 3)
733 0 : return DEFAULT_CONTAIN_SEL;
734 :
735 : /*
736 : * Grab some of the summary statistics that compute_array_stats() stores:
737 : * lowest MCE frequency, frequency of null elements, and average distinct
738 : * element count.
739 : */
740 36 : minfreq = numbers[nmcelem];
741 36 : nullelem_freq = numbers[nmcelem + 2];
742 36 : avg_count = hist[nhist - 1];
743 :
744 : /*
745 : * "rest" will be the sum of the frequencies of all elements not
746 : * represented in MCELEM. The average distinct element count is the sum
747 : * of the frequencies of *all* elements. Begin with that; we will proceed
748 : * to subtract the MCELEM frequencies.
749 : */
750 36 : rest = avg_count;
751 :
752 : /*
753 : * mult is a multiplier representing estimate of probability that each
754 : * mcelem that is not present in constant doesn't occur.
755 : */
756 36 : mult = 1.0f;
757 :
758 : /*
759 : * elem_selec is array of estimated frequencies for elements in the
760 : * constant.
761 : */
762 36 : elem_selec = palloc_array(float, nitems);
763 :
764 : /* Scan mcelem and array in parallel. */
765 36 : mcelem_index = 0;
766 114 : for (i = 0; i < nitems; i++)
767 : {
768 78 : bool match = false;
769 :
770 : /* Ignore any duplicates in the array data. */
771 138 : if (i > 0 &&
772 60 : element_compare(&array_data[i - 1], &array_data[i], typentry) == 0)
773 0 : continue;
774 :
775 : /*
776 : * Iterate over MCELEM until we find an entry greater than or equal to
777 : * this element of the constant. Update "rest" and "mult" for mcelem
778 : * entries skipped over.
779 : */
780 2064 : while (mcelem_index < nmcelem)
781 : {
782 2064 : int cmp = element_compare(&mcelem[mcelem_index],
783 2064 : &array_data[i],
784 : typentry);
785 :
786 2064 : if (cmp < 0)
787 : {
788 1986 : mult *= (1.0f - numbers[mcelem_index]);
789 1986 : rest -= numbers[mcelem_index];
790 1986 : mcelem_index++;
791 : }
792 : else
793 : {
794 78 : if (cmp == 0)
795 78 : match = true; /* mcelem is found */
796 78 : break;
797 : }
798 : }
799 :
800 78 : if (match)
801 : {
802 : /* MCELEM matches the array item. */
803 78 : elem_selec[unique_nitems] = numbers[mcelem_index];
804 : /* "rest" is decremented for all mcelems, matched or not */
805 78 : rest -= numbers[mcelem_index];
806 78 : mcelem_index++;
807 : }
808 : else
809 : {
810 : /*
811 : * The element is not in MCELEM. Estimate its frequency as half
812 : * that of the least-frequent MCE. (We know it cannot be more
813 : * than minfreq, and it could be a great deal less. Half seems
814 : * like a good compromise.) For probably-historical reasons,
815 : * clamp to not more than DEFAULT_CONTAIN_SEL.
816 : */
817 0 : elem_selec[unique_nitems] = Min(DEFAULT_CONTAIN_SEL,
818 : minfreq / 2);
819 : }
820 :
821 78 : unique_nitems++;
822 : }
823 :
824 : /*
825 : * If we handled all constant elements without exhausting the MCELEM
826 : * array, finish walking it to complete calculation of "rest" and "mult".
827 : */
828 3834 : while (mcelem_index < nmcelem)
829 : {
830 3798 : mult *= (1.0f - numbers[mcelem_index]);
831 3798 : rest -= numbers[mcelem_index];
832 3798 : mcelem_index++;
833 : }
834 :
835 : /*
836 : * The presence of many distinct rare elements materially decreases
837 : * selectivity. Use the Poisson distribution to estimate the probability
838 : * of a column value having zero occurrences of such elements. See above
839 : * for the definition of "rest".
840 : */
841 36 : mult *= exp(-rest);
842 :
843 : /*----------
844 : * Using the distinct element count histogram requires
845 : * O(unique_nitems * (nmcelem + unique_nitems))
846 : * operations. Beyond a certain computational cost threshold, it's
847 : * reasonable to sacrifice accuracy for decreased planning time. We limit
848 : * the number of operations to EFFORT * nmcelem; since nmcelem is limited
849 : * by the column's statistics target, the work done is user-controllable.
850 : *
851 : * If the number of operations would be too large, we can reduce it
852 : * without losing all accuracy by reducing unique_nitems and considering
853 : * only the most-common elements of the constant array. To make the
854 : * results exactly match what we would have gotten with only those
855 : * elements to start with, we'd have to remove any discarded elements'
856 : * frequencies from "mult", but since this is only an approximation
857 : * anyway, we don't bother with that. Therefore it's sufficient to qsort
858 : * elem_selec[] and take the largest elements. (They will no longer match
859 : * up with the elements of array_data[], but we don't care.)
860 : *----------
861 : */
862 : #define EFFORT 100
863 :
864 36 : if ((nmcelem + unique_nitems) > 0 &&
865 36 : unique_nitems > EFFORT * nmcelem / (nmcelem + unique_nitems))
866 : {
867 : /*
868 : * Use the quadratic formula to solve for largest allowable N. We
869 : * have A = 1, B = nmcelem, C = - EFFORT * nmcelem.
870 : */
871 0 : double b = (double) nmcelem;
872 : int n;
873 :
874 0 : n = (int) ((sqrt(b * b + 4 * EFFORT * b) - b) / 2);
875 :
876 : /* Sort, then take just the first n elements */
877 0 : qsort(elem_selec, unique_nitems, sizeof(float),
878 : float_compare_desc);
879 0 : unique_nitems = n;
880 : }
881 :
882 : /*
883 : * Calculate probabilities of each distinct element count for both mcelems
884 : * and constant elements. At this point, assume independent element
885 : * occurrence.
886 : */
887 36 : dist = calc_distr(elem_selec, unique_nitems, unique_nitems, 0.0f);
888 36 : mcelem_dist = calc_distr(numbers, nmcelem, unique_nitems, rest);
889 :
890 : /* ignore hist[nhist-1], which is the average not a histogram member */
891 36 : hist_part = calc_hist(hist, nhist - 1, unique_nitems);
892 :
893 36 : selec = 0.0f;
894 150 : for (i = 0; i <= unique_nitems; i++)
895 : {
896 : /*
897 : * mult * dist[i] / mcelem_dist[i] gives us probability of qual
898 : * matching from assumption of independent element occurrence with the
899 : * condition that distinct element count = i.
900 : */
901 114 : if (mcelem_dist[i] > 0)
902 114 : selec += hist_part[i] * mult * dist[i] / mcelem_dist[i];
903 : }
904 :
905 36 : pfree(dist);
906 36 : pfree(mcelem_dist);
907 36 : pfree(hist_part);
908 36 : pfree(elem_selec);
909 :
910 : /* Take into account occurrence of NULL element. */
911 36 : selec *= (1.0f - nullelem_freq);
912 :
913 36 : CLAMP_PROBABILITY(selec);
914 :
915 36 : return selec;
916 : }
917 :
918 : /*
919 : * Calculate the first n distinct element count probabilities from a
920 : * histogram of distinct element counts.
921 : *
922 : * Returns a palloc'd array of n+1 entries, with array[k] being the
923 : * probability of element count k, k in [0..n].
924 : *
925 : * We assume that a histogram box with bounds a and b gives 1 / ((b - a + 1) *
926 : * (nhist - 1)) probability to each value in (a,b) and an additional half of
927 : * that to a and b themselves.
928 : */
929 : static float *
930 36 : calc_hist(const float4 *hist, int nhist, int n)
931 : {
932 : float *hist_part;
933 : int k,
934 36 : i = 0;
935 36 : float prev_interval = 0,
936 : next_interval;
937 : float frac;
938 :
939 36 : hist_part = palloc_array(float, n + 1);
940 :
941 : /*
942 : * frac is a probability contribution for each interval between histogram
943 : * values. We have nhist - 1 intervals, so contribution of each one will
944 : * be 1 / (nhist - 1).
945 : */
946 36 : frac = 1.0f / ((float) (nhist - 1));
947 :
948 150 : for (k = 0; k <= n; k++)
949 : {
950 114 : int count = 0;
951 :
952 : /*
953 : * Count the histogram boundaries equal to k. (Although the histogram
954 : * should theoretically contain only exact integers, entries are
955 : * floats so there could be roundoff error in large values. Treat any
956 : * fractional value as equal to the next larger k.)
957 : */
958 1008 : while (i < nhist && hist[i] <= k)
959 : {
960 894 : count++;
961 894 : i++;
962 : }
963 :
964 114 : if (count > 0)
965 : {
966 : /* k is an exact bound for at least one histogram box. */
967 : float val;
968 :
969 : /* Find length between current histogram value and the next one */
970 108 : if (i < nhist)
971 108 : next_interval = hist[i] - hist[i - 1];
972 : else
973 0 : next_interval = 0;
974 :
975 : /*
976 : * count - 1 histogram boxes contain k exclusively. They
977 : * contribute a total of (count - 1) * frac probability. Also
978 : * factor in the partial histogram boxes on either side.
979 : */
980 108 : val = (float) (count - 1);
981 108 : if (next_interval > 0)
982 108 : val += 0.5f / next_interval;
983 108 : if (prev_interval > 0)
984 72 : val += 0.5f / prev_interval;
985 108 : hist_part[k] = frac * val;
986 :
987 108 : prev_interval = next_interval;
988 : }
989 : else
990 : {
991 : /* k does not appear as an exact histogram bound. */
992 6 : if (prev_interval > 0)
993 6 : hist_part[k] = frac / prev_interval;
994 : else
995 0 : hist_part[k] = 0.0f;
996 : }
997 : }
998 :
999 36 : return hist_part;
1000 : }
1001 :
1002 : /*
1003 : * Consider n independent events with probabilities p[]. This function
1004 : * calculates probabilities of exact k of events occurrence for k in [0..m].
1005 : * Returns a palloc'd array of size m+1.
1006 : *
1007 : * "rest" is the sum of the probabilities of all low-probability events not
1008 : * included in p.
1009 : *
1010 : * Imagine matrix M of size (n + 1) x (m + 1). Element M[i,j] denotes the
1011 : * probability that exactly j of first i events occur. Obviously M[0,0] = 1.
1012 : * For any constant j, each increment of i increases the probability iff the
1013 : * event occurs. So, by the law of total probability:
1014 : * M[i,j] = M[i - 1, j] * (1 - p[i]) + M[i - 1, j - 1] * p[i]
1015 : * for i > 0, j > 0.
1016 : * M[i,0] = M[i - 1, 0] * (1 - p[i]) for i > 0.
1017 : */
1018 : static float *
1019 72 : calc_distr(const float *p, int n, int m, float rest)
1020 : {
1021 : float *row,
1022 : *prev_row,
1023 : *tmp;
1024 : int i,
1025 : j;
1026 :
1027 : /*
1028 : * Since we return only the last row of the matrix and need only the
1029 : * current and previous row for calculations, allocate two rows.
1030 : */
1031 72 : row = palloc_array(float, m + 1);
1032 72 : prev_row = palloc_array(float, m + 1);
1033 :
1034 : /* M[0,0] = 1 */
1035 72 : row[0] = 1.0f;
1036 6012 : for (i = 1; i <= n; i++)
1037 : {
1038 5940 : float t = p[i - 1];
1039 :
1040 : /* Swap rows */
1041 5940 : tmp = row;
1042 5940 : row = prev_row;
1043 5940 : prev_row = tmp;
1044 :
1045 : /* Calculate next row */
1046 26574 : for (j = 0; j <= i && j <= m; j++)
1047 : {
1048 20634 : float val = 0.0f;
1049 :
1050 20634 : if (j < i)
1051 20478 : val += prev_row[j] * (1.0f - t);
1052 20634 : if (j > 0)
1053 14694 : val += prev_row[j - 1] * t;
1054 20634 : row[j] = val;
1055 : }
1056 : }
1057 :
1058 : /*
1059 : * The presence of many distinct rare (not in "p") elements materially
1060 : * decreases selectivity. Model their collective occurrence with the
1061 : * Poisson distribution.
1062 : */
1063 72 : if (rest > DEFAULT_CONTAIN_SEL)
1064 : {
1065 : float t;
1066 :
1067 : /* Swap rows */
1068 0 : tmp = row;
1069 0 : row = prev_row;
1070 0 : prev_row = tmp;
1071 :
1072 0 : for (i = 0; i <= m; i++)
1073 0 : row[i] = 0.0f;
1074 :
1075 : /* Value of Poisson distribution for 0 occurrences */
1076 0 : t = exp(-rest);
1077 :
1078 : /*
1079 : * Calculate convolution of previously computed distribution and the
1080 : * Poisson distribution.
1081 : */
1082 0 : for (i = 0; i <= m; i++)
1083 : {
1084 0 : for (j = 0; j <= m - i; j++)
1085 0 : row[j + i] += prev_row[j] * t;
1086 :
1087 : /* Get Poisson distribution value for (i + 1) occurrences */
1088 0 : t *= rest / (float) (i + 1);
1089 : }
1090 : }
1091 :
1092 72 : pfree(prev_row);
1093 72 : return row;
1094 : }
1095 :
1096 : /* Fast function for floor value of 2 based logarithm calculation. */
1097 : static int
1098 507 : floor_log2(uint32 n)
1099 : {
1100 507 : int logval = 0;
1101 :
1102 507 : if (n == 0)
1103 309 : return -1;
1104 198 : if (n >= (1 << 16))
1105 : {
1106 0 : n >>= 16;
1107 0 : logval += 16;
1108 : }
1109 198 : if (n >= (1 << 8))
1110 : {
1111 60 : n >>= 8;
1112 60 : logval += 8;
1113 : }
1114 198 : if (n >= (1 << 4))
1115 : {
1116 138 : n >>= 4;
1117 138 : logval += 4;
1118 : }
1119 198 : if (n >= (1 << 2))
1120 : {
1121 132 : n >>= 2;
1122 132 : logval += 2;
1123 : }
1124 198 : if (n >= (1 << 1))
1125 : {
1126 93 : logval += 1;
1127 : }
1128 198 : return logval;
1129 : }
1130 :
1131 : /*
1132 : * find_next_mcelem binary-searches a most common elements array, starting
1133 : * from *index, for the first member >= value. It saves the position of the
1134 : * match into *index and returns true if it's an exact match. (Note: we
1135 : * assume the mcelem elements are distinct so there can't be more than one
1136 : * exact match.)
1137 : */
1138 : static bool
1139 1929 : find_next_mcelem(const Datum *mcelem, int nmcelem, Datum value, int *index,
1140 : TypeCacheEntry *typentry)
1141 : {
1142 1929 : int l = *index,
1143 1929 : r = nmcelem - 1,
1144 : i,
1145 : res;
1146 :
1147 3171 : while (l <= r)
1148 : {
1149 1446 : i = (l + r) / 2;
1150 1446 : res = element_compare(&mcelem[i], &value, typentry);
1151 1446 : if (res == 0)
1152 : {
1153 204 : *index = i;
1154 204 : return true;
1155 : }
1156 1242 : else if (res < 0)
1157 441 : l = i + 1;
1158 : else
1159 801 : r = i - 1;
1160 : }
1161 1725 : *index = l;
1162 1725 : return false;
1163 : }
1164 :
1165 : /*
1166 : * Comparison function for elements.
1167 : *
1168 : * We use the element type's default btree opclass, and its default collation
1169 : * if the type is collation-sensitive.
1170 : *
1171 : * XXX consider using SortSupport infrastructure
1172 : */
1173 : static int
1174 6807 : element_compare(const void *key1, const void *key2, void *arg)
1175 : {
1176 6807 : Datum d1 = *((const Datum *) key1);
1177 6807 : Datum d2 = *((const Datum *) key2);
1178 6807 : TypeCacheEntry *typentry = (TypeCacheEntry *) arg;
1179 6807 : FmgrInfo *cmpfunc = &typentry->cmp_proc_finfo;
1180 : Datum c;
1181 :
1182 6807 : c = FunctionCall2Coll(cmpfunc, typentry->typcollation, d1, d2);
1183 6807 : return DatumGetInt32(c);
1184 : }
1185 :
1186 : /*
1187 : * Comparison function for sorting floats into descending order.
1188 : */
1189 : static int
1190 0 : float_compare_desc(const void *key1, const void *key2)
1191 : {
1192 0 : float d1 = *((const float *) key1);
1193 0 : float d2 = *((const float *) key2);
1194 :
1195 0 : if (d1 > d2)
1196 0 : return -1;
1197 0 : else if (d1 < d2)
1198 0 : return 1;
1199 : else
1200 0 : return 0;
1201 : }
|
