Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * ts_typanalyze.c
4 : * functions for gathering statistics from tsvector columns
5 : *
6 : * Portions Copyright (c) 1996-2026, PostgreSQL Global Development Group
7 : *
8 : *
9 : * IDENTIFICATION
10 : * src/backend/tsearch/ts_typanalyze.c
11 : *
12 : *-------------------------------------------------------------------------
13 : */
14 : #include "postgres.h"
15 :
16 : #include "catalog/pg_collation.h"
17 : #include "catalog/pg_operator.h"
18 : #include "commands/vacuum.h"
19 : #include "common/hashfn.h"
20 : #include "tsearch/ts_type.h"
21 : #include "utils/builtins.h"
22 : #include "varatt.h"
23 :
24 :
25 : /* A hash key for lexemes */
26 : typedef struct
27 : {
28 : char *lexeme; /* lexeme (not NULL terminated!) */
29 : int length; /* its length in bytes */
30 : } LexemeHashKey;
31 :
32 : /* A hash table entry for the Lossy Counting algorithm */
33 : typedef struct
34 : {
35 : LexemeHashKey key; /* This is 'e' from the LC algorithm. */
36 : int frequency; /* This is 'f'. */
37 : int delta; /* And this is 'delta'. */
38 : } TrackItem;
39 :
40 : static void compute_tsvector_stats(VacAttrStats *stats,
41 : AnalyzeAttrFetchFunc fetchfunc,
42 : int samplerows,
43 : double totalrows);
44 : static void prune_lexemes_hashtable(HTAB *lexemes_tab, int b_current);
45 : static uint32 lexeme_hash(const void *key, Size keysize);
46 : static int lexeme_match(const void *key1, const void *key2, Size keysize);
47 : static int lexeme_compare(const void *key1, const void *key2);
48 : static int trackitem_compare_frequencies_desc(const void *e1, const void *e2,
49 : void *arg);
50 : static int trackitem_compare_lexemes(const void *e1, const void *e2,
51 : void *arg);
52 :
53 :
54 : /*
55 : * ts_typanalyze -- a custom typanalyze function for tsvector columns
56 : */
57 : Datum
58 4 : ts_typanalyze(PG_FUNCTION_ARGS)
59 : {
60 4 : VacAttrStats *stats = (VacAttrStats *) PG_GETARG_POINTER(0);
61 :
62 : /* If the attstattarget column is negative, use the default value */
63 4 : if (stats->attstattarget < 0)
64 4 : stats->attstattarget = default_statistics_target;
65 :
66 4 : stats->compute_stats = compute_tsvector_stats;
67 : /* see comment about the choice of minrows in commands/analyze.c */
68 4 : stats->minrows = 300 * stats->attstattarget;
69 :
70 4 : PG_RETURN_BOOL(true);
71 : }
72 :
73 : /*
74 : * compute_tsvector_stats() -- compute statistics for a tsvector column
75 : *
76 : * This function computes statistics that are useful for determining @@
77 : * operations' selectivity, along with the fraction of non-null rows and
78 : * average width.
79 : *
80 : * Instead of finding the most common values, as we do for most datatypes,
81 : * we're looking for the most common lexemes. This is more useful, because
82 : * there most probably won't be any two rows with the same tsvector and thus
83 : * the notion of a MCV is a bit bogus with this datatype. With a list of the
84 : * most common lexemes we can do a better job at figuring out @@ selectivity.
85 : *
86 : * For the same reasons we assume that tsvector columns are unique when
87 : * determining the number of distinct values.
88 : *
89 : * The algorithm used is Lossy Counting, as proposed in the paper "Approximate
90 : * frequency counts over data streams" by G. S. Manku and R. Motwani, in
91 : * Proceedings of the 28th International Conference on Very Large Data Bases,
92 : * Hong Kong, China, August 2002, section 4.2. The paper is available at
93 : * http://www.vldb.org/conf/2002/S10P03.pdf
94 : *
95 : * The Lossy Counting (aka LC) algorithm goes like this:
96 : * Let s be the threshold frequency for an item (the minimum frequency we
97 : * are interested in) and epsilon the error margin for the frequency. Let D
98 : * be a set of triples (e, f, delta), where e is an element value, f is that
99 : * element's frequency (actually, its current occurrence count) and delta is
100 : * the maximum error in f. We start with D empty and process the elements in
101 : * batches of size w. (The batch size is also known as "bucket size" and is
102 : * equal to 1/epsilon.) Let the current batch number be b_current, starting
103 : * with 1. For each element e we either increment its f count, if it's
104 : * already in D, or insert a new triple into D with values (e, 1, b_current
105 : * - 1). After processing each batch we prune D, by removing from it all
106 : * elements with f + delta <= b_current. After the algorithm finishes we
107 : * suppress all elements from D that do not satisfy f >= (s - epsilon) * N,
108 : * where N is the total number of elements in the input. We emit the
109 : * remaining elements with estimated frequency f/N. The LC paper proves
110 : * that this algorithm finds all elements with true frequency at least s,
111 : * and that no frequency is overestimated or is underestimated by more than
112 : * epsilon. Furthermore, given reasonable assumptions about the input
113 : * distribution, the required table size is no more than about 7 times w.
114 : *
115 : * We set s to be the estimated frequency of the K'th word in a natural
116 : * language's frequency table, where K is the target number of entries in
117 : * the MCELEM array plus an arbitrary constant, meant to reflect the fact
118 : * that the most common words in any language would usually be stopwords
119 : * so we will not actually see them in the input. We assume that the
120 : * distribution of word frequencies (including the stopwords) follows Zipf's
121 : * law with an exponent of 1.
122 : *
123 : * Assuming Zipfian distribution, the frequency of the K'th word is equal
124 : * to 1/(K * H(W)) where H(n) is 1/2 + 1/3 + ... + 1/n and W is the number of
125 : * words in the language. Putting W as one million, we get roughly 0.07/K.
126 : * Assuming top 10 words are stopwords gives s = 0.07/(K + 10). We set
127 : * epsilon = s/10, which gives bucket width w = (K + 10)/0.007 and
128 : * maximum expected hashtable size of about 1000 * (K + 10).
129 : *
130 : * Note: in the above discussion, s, epsilon, and f/N are in terms of a
131 : * lexeme's frequency as a fraction of all lexemes seen in the input.
132 : * However, what we actually want to store in the finished pg_statistic
133 : * entry is each lexeme's frequency as a fraction of all rows that it occurs
134 : * in. Assuming that the input tsvectors are correctly constructed, no
135 : * lexeme occurs more than once per tsvector, so the final count f is a
136 : * correct estimate of the number of input tsvectors it occurs in, and we
137 : * need only change the divisor from N to nonnull_cnt to get the number we
138 : * want.
139 : */
140 : static void
141 4 : compute_tsvector_stats(VacAttrStats *stats,
142 : AnalyzeAttrFetchFunc fetchfunc,
143 : int samplerows,
144 : double totalrows)
145 : {
146 : int num_mcelem;
147 4 : int null_cnt = 0;
148 4 : double total_width = 0;
149 :
150 : /* This is D from the LC algorithm. */
151 : HTAB *lexemes_tab;
152 : HASHCTL hash_ctl;
153 : HASH_SEQ_STATUS scan_status;
154 :
155 : /* This is the current bucket number from the LC algorithm */
156 : int b_current;
157 :
158 : /* This is 'w' from the LC algorithm */
159 : int bucket_width;
160 : int vector_no,
161 : lexeme_no;
162 : LexemeHashKey hash_key;
163 :
164 : /*
165 : * We want statistics_target * 10 lexemes in the MCELEM array. This
166 : * multiplier is pretty arbitrary, but is meant to reflect the fact that
167 : * the number of individual lexeme values tracked in pg_statistic ought to
168 : * be more than the number of values for a simple scalar column.
169 : */
170 4 : num_mcelem = stats->attstattarget * 10;
171 :
172 : /*
173 : * We set bucket width equal to (num_mcelem + 10) / 0.007 as per the
174 : * comment above.
175 : */
176 4 : bucket_width = (num_mcelem + 10) * 1000 / 7;
177 :
178 : /*
179 : * Create the hashtable. It will be in local memory, so we don't need to
180 : * worry about overflowing the initial size. Also we don't need to pay any
181 : * attention to locking and memory management.
182 : */
183 4 : hash_ctl.keysize = sizeof(LexemeHashKey);
184 4 : hash_ctl.entrysize = sizeof(TrackItem);
185 4 : hash_ctl.hash = lexeme_hash;
186 4 : hash_ctl.match = lexeme_match;
187 4 : hash_ctl.hcxt = CurrentMemoryContext;
188 4 : lexemes_tab = hash_create("Analyzed lexemes table",
189 : num_mcelem,
190 : &hash_ctl,
191 : HASH_ELEM | HASH_FUNCTION | HASH_COMPARE | HASH_CONTEXT);
192 :
193 : /* Initialize counters. */
194 4 : b_current = 1;
195 4 : lexeme_no = 0;
196 :
197 : /* Loop over the tsvectors. */
198 2039 : for (vector_no = 0; vector_no < samplerows; vector_no++)
199 : {
200 : Datum value;
201 : bool isnull;
202 : TSVector vector;
203 : WordEntry *curentryptr;
204 : char *lexemesptr;
205 : int j;
206 :
207 2035 : vacuum_delay_point(true);
208 :
209 2035 : value = fetchfunc(stats, vector_no, &isnull);
210 :
211 : /*
212 : * Check for null/nonnull.
213 : */
214 2035 : if (isnull)
215 : {
216 0 : null_cnt++;
217 0 : continue;
218 : }
219 :
220 : /*
221 : * Add up widths for average-width calculation. Since it's a
222 : * tsvector, we know it's varlena. As in the regular
223 : * compute_minimal_stats function, we use the toasted width for this
224 : * calculation.
225 : */
226 2035 : total_width += VARSIZE_ANY(DatumGetPointer(value));
227 :
228 : /*
229 : * Now detoast the tsvector if needed.
230 : */
231 2035 : vector = DatumGetTSVector(value);
232 :
233 : /*
234 : * We loop through the lexemes in the tsvector and add them to our
235 : * tracking hashtable.
236 : */
237 2035 : lexemesptr = STRPTR(vector);
238 2035 : curentryptr = ARRPTR(vector);
239 117239 : for (j = 0; j < vector->size; j++)
240 : {
241 : TrackItem *item;
242 : bool found;
243 :
244 : /*
245 : * Construct a hash key. The key points into the (detoasted)
246 : * tsvector value at this point, but if a new entry is created, we
247 : * make a copy of it. This way we can free the tsvector value
248 : * once we've processed all its lexemes.
249 : */
250 115204 : hash_key.lexeme = lexemesptr + curentryptr->pos;
251 115204 : hash_key.length = curentryptr->len;
252 :
253 : /* Lookup current lexeme in hashtable, adding it if new */
254 115204 : item = (TrackItem *) hash_search(lexemes_tab,
255 : &hash_key,
256 : HASH_ENTER, &found);
257 :
258 115204 : if (found)
259 : {
260 : /* The lexeme is already on the tracking list */
261 110636 : item->frequency++;
262 : }
263 : else
264 : {
265 : /* Initialize new tracking list element */
266 4568 : item->frequency = 1;
267 4568 : item->delta = b_current - 1;
268 :
269 4568 : item->key.lexeme = palloc(hash_key.length);
270 4568 : memcpy(item->key.lexeme, hash_key.lexeme, hash_key.length);
271 : }
272 :
273 : /* lexeme_no is the number of elements processed (ie N) */
274 115204 : lexeme_no++;
275 :
276 : /* We prune the D structure after processing each bucket */
277 115204 : if (lexeme_no % bucket_width == 0)
278 : {
279 0 : prune_lexemes_hashtable(lexemes_tab, b_current);
280 0 : b_current++;
281 : }
282 :
283 : /* Advance to the next WordEntry in the tsvector */
284 115204 : curentryptr++;
285 : }
286 :
287 : /* If the vector was toasted, free the detoasted copy. */
288 2035 : if (TSVectorGetDatum(vector) != value)
289 259 : pfree(vector);
290 : }
291 :
292 : /* We can only compute real stats if we found some non-null values. */
293 4 : if (null_cnt < samplerows)
294 : {
295 4 : int nonnull_cnt = samplerows - null_cnt;
296 : int i;
297 : TrackItem **sort_table;
298 : TrackItem *item;
299 : int track_len;
300 : int cutoff_freq;
301 : int minfreq,
302 : maxfreq;
303 :
304 4 : stats->stats_valid = true;
305 : /* Do the simple null-frac and average width stats */
306 4 : stats->stanullfrac = (double) null_cnt / (double) samplerows;
307 4 : stats->stawidth = total_width / (double) nonnull_cnt;
308 :
309 : /* Assume it's a unique column (see notes above) */
310 4 : stats->stadistinct = -1.0 * (1.0 - stats->stanullfrac);
311 :
312 : /*
313 : * Construct an array of the interesting hashtable items, that is,
314 : * those meeting the cutoff frequency (s - epsilon)*N. Also identify
315 : * the maximum frequency among these items.
316 : *
317 : * Since epsilon = s/10 and bucket_width = 1/epsilon, the cutoff
318 : * frequency is 9*N / bucket_width.
319 : */
320 4 : cutoff_freq = 9 * lexeme_no / bucket_width;
321 :
322 4 : i = hash_get_num_entries(lexemes_tab); /* surely enough space */
323 4 : sort_table = palloc_array(TrackItem *, i);
324 :
325 4 : hash_seq_init(&scan_status, lexemes_tab);
326 4 : track_len = 0;
327 4 : maxfreq = 0;
328 4576 : while ((item = (TrackItem *) hash_seq_search(&scan_status)) != NULL)
329 : {
330 4568 : if (item->frequency > cutoff_freq)
331 : {
332 4212 : sort_table[track_len++] = item;
333 4212 : maxfreq = Max(maxfreq, item->frequency);
334 : }
335 : }
336 : Assert(track_len <= i);
337 :
338 : /* emit some statistics for debug purposes */
339 4 : elog(DEBUG3, "tsvector_stats: target # mces = %d, bucket width = %d, "
340 : "# lexemes = %d, hashtable size = %d, usable entries = %d",
341 : num_mcelem, bucket_width, lexeme_no, i, track_len);
342 :
343 : /*
344 : * If we obtained more lexemes than we really want, get rid of those
345 : * with least frequencies. The easiest way is to qsort the array into
346 : * descending frequency order and truncate the array.
347 : *
348 : * If we did not find more elements than we want, then it is safe to
349 : * assume that the stored MCE array will contain every element with
350 : * frequency above the cutoff. In that case, rather than storing the
351 : * smallest frequency we are keeping, we want to store the minimum
352 : * frequency that would have been accepted as a valid MCE. The
353 : * selectivity functions can assume that that is an upper bound on the
354 : * frequency of elements not present in the array.
355 : *
356 : * If we found no candidate MCEs at all, we still want to record the
357 : * cutoff frequency, since it's still valid to assume that no element
358 : * has frequency more than that.
359 : */
360 4 : if (num_mcelem < track_len)
361 : {
362 4 : qsort_interruptible(sort_table, track_len, sizeof(TrackItem *),
363 : trackitem_compare_frequencies_desc, NULL);
364 : /* set minfreq to the smallest frequency we're keeping */
365 4 : minfreq = sort_table[num_mcelem - 1]->frequency;
366 : }
367 : else
368 : {
369 0 : num_mcelem = track_len;
370 : /* set minfreq to the minimum frequency above the cutoff */
371 0 : minfreq = cutoff_freq + 1;
372 : /* ensure maxfreq is nonzero, too */
373 0 : if (track_len == 0)
374 0 : maxfreq = minfreq;
375 : }
376 :
377 : /* Generate MCELEM slot entry */
378 4 : if (num_mcelem >= 0)
379 : {
380 : MemoryContext old_context;
381 : Datum *mcelem_values;
382 : float4 *mcelem_freqs;
383 :
384 : /*
385 : * We want to store statistics sorted on the lexeme value using
386 : * first length, then byte-for-byte comparison. The reason for
387 : * doing length comparison first is that we don't care about the
388 : * ordering so long as it's consistent, and comparing lengths
389 : * first gives us a chance to avoid a strncmp() call.
390 : *
391 : * This is different from what we do with scalar statistics --
392 : * they get sorted on frequencies. The rationale is that we
393 : * usually search through most common elements looking for a
394 : * specific value, so we can grab its frequency. When values are
395 : * presorted we can employ binary search for that. See
396 : * ts_selfuncs.c for a real usage scenario.
397 : */
398 4 : qsort_interruptible(sort_table, num_mcelem, sizeof(TrackItem *),
399 : trackitem_compare_lexemes, NULL);
400 :
401 : /* Must copy the target values into anl_context */
402 4 : old_context = MemoryContextSwitchTo(stats->anl_context);
403 :
404 : /*
405 : * We sorted statistics on the lexeme value, but we want to be
406 : * able to find out the minimal and maximal frequency without
407 : * going through all the values. We keep those two extra
408 : * frequencies in two extra cells in mcelem_freqs.
409 : *
410 : * (Note: the MCELEM statistics slot definition allows for a third
411 : * extra number containing the frequency of nulls, but we don't
412 : * create that for a tsvector column, since null elements aren't
413 : * possible.)
414 : */
415 4 : mcelem_values = palloc_array(Datum, num_mcelem);
416 4 : mcelem_freqs = palloc_array(float4, num_mcelem + 2);
417 :
418 : /*
419 : * See comments above about use of nonnull_cnt as the divisor for
420 : * the final frequency estimates.
421 : */
422 4004 : for (i = 0; i < num_mcelem; i++)
423 : {
424 4000 : TrackItem *titem = sort_table[i];
425 :
426 8000 : mcelem_values[i] =
427 4000 : PointerGetDatum(cstring_to_text_with_len(titem->key.lexeme,
428 : titem->key.length));
429 4000 : mcelem_freqs[i] = (double) titem->frequency / (double) nonnull_cnt;
430 : }
431 4 : mcelem_freqs[i++] = (double) minfreq / (double) nonnull_cnt;
432 4 : mcelem_freqs[i] = (double) maxfreq / (double) nonnull_cnt;
433 4 : MemoryContextSwitchTo(old_context);
434 :
435 4 : stats->stakind[0] = STATISTIC_KIND_MCELEM;
436 4 : stats->staop[0] = TextEqualOperator;
437 4 : stats->stacoll[0] = DEFAULT_COLLATION_OID;
438 4 : stats->stanumbers[0] = mcelem_freqs;
439 : /* See above comment about two extra frequency fields */
440 4 : stats->numnumbers[0] = num_mcelem + 2;
441 4 : stats->stavalues[0] = mcelem_values;
442 4 : stats->numvalues[0] = num_mcelem;
443 : /* We are storing text values */
444 4 : stats->statypid[0] = TEXTOID;
445 4 : stats->statyplen[0] = -1; /* typlen, -1 for varlena */
446 4 : stats->statypbyval[0] = false;
447 4 : stats->statypalign[0] = TYPALIGN_INT;
448 : }
449 : }
450 : else
451 : {
452 : /* We found only nulls; assume the column is entirely null */
453 0 : stats->stats_valid = true;
454 0 : stats->stanullfrac = 1.0;
455 0 : stats->stawidth = 0; /* "unknown" */
456 0 : stats->stadistinct = 0.0; /* "unknown" */
457 : }
458 :
459 : /*
460 : * We don't need to bother cleaning up any of our temporary palloc's. The
461 : * hashtable should also go away, as it used a child memory context.
462 : */
463 4 : }
464 :
465 : /*
466 : * A function to prune the D structure from the Lossy Counting algorithm.
467 : * Consult compute_tsvector_stats() for wider explanation.
468 : */
469 : static void
470 0 : prune_lexemes_hashtable(HTAB *lexemes_tab, int b_current)
471 : {
472 : HASH_SEQ_STATUS scan_status;
473 : TrackItem *item;
474 :
475 0 : hash_seq_init(&scan_status, lexemes_tab);
476 0 : while ((item = (TrackItem *) hash_seq_search(&scan_status)) != NULL)
477 : {
478 0 : if (item->frequency + item->delta <= b_current)
479 : {
480 0 : char *lexeme = item->key.lexeme;
481 :
482 0 : if (hash_search(lexemes_tab, &item->key,
483 : HASH_REMOVE, NULL) == NULL)
484 0 : elog(ERROR, "hash table corrupted");
485 0 : pfree(lexeme);
486 : }
487 : }
488 0 : }
489 :
490 : /*
491 : * Hash functions for lexemes. They are strings, but not NULL terminated,
492 : * so we need a special hash function.
493 : */
494 : static uint32
495 115204 : lexeme_hash(const void *key, Size keysize)
496 : {
497 115204 : const LexemeHashKey *l = (const LexemeHashKey *) key;
498 :
499 115204 : return DatumGetUInt32(hash_any((const unsigned char *) l->lexeme,
500 115204 : l->length));
501 : }
502 :
503 : /*
504 : * Matching function for lexemes, to be used in hashtable lookups.
505 : */
506 : static int
507 110636 : lexeme_match(const void *key1, const void *key2, Size keysize)
508 : {
509 : /* The keysize parameter is superfluous, the keys store their lengths */
510 110636 : return lexeme_compare(key1, key2);
511 : }
512 :
513 : /*
514 : * Comparison function for lexemes.
515 : */
516 : static int
517 151272 : lexeme_compare(const void *key1, const void *key2)
518 : {
519 151272 : const LexemeHashKey *d1 = (const LexemeHashKey *) key1;
520 151272 : const LexemeHashKey *d2 = (const LexemeHashKey *) key2;
521 :
522 : /* First, compare by length */
523 151272 : if (d1->length > d2->length)
524 0 : return 1;
525 151272 : else if (d1->length < d2->length)
526 0 : return -1;
527 : /* Lengths are equal, do a byte-by-byte comparison */
528 151272 : return strncmp(d1->lexeme, d2->lexeme, d1->length);
529 : }
530 :
531 : /*
532 : * Comparator for sorting TrackItems on frequencies (descending sort)
533 : */
534 : static int
535 25241 : trackitem_compare_frequencies_desc(const void *e1, const void *e2, void *arg)
536 : {
537 25241 : const TrackItem *const *t1 = (const TrackItem *const *) e1;
538 25241 : const TrackItem *const *t2 = (const TrackItem *const *) e2;
539 :
540 25241 : return (*t2)->frequency - (*t1)->frequency;
541 : }
542 :
543 : /*
544 : * Comparator for sorting TrackItems on lexemes
545 : */
546 : static int
547 40636 : trackitem_compare_lexemes(const void *e1, const void *e2, void *arg)
548 : {
549 40636 : const TrackItem *const *t1 = (const TrackItem *const *) e1;
550 40636 : const TrackItem *const *t2 = (const TrackItem *const *) e2;
551 :
552 40636 : return lexeme_compare(&(*t1)->key, &(*t2)->key);
553 : }
|
