?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Fix xslt_process() to ensure that it inserts a NULL terminator after the
[PostgreSQL.git] / contrib / btree_gist / btree_utils_var.c
blob7329ebd538573463652a3b65520bd27c562de311
1 /*
2 * $PostgreSQL$
3 */
4 #include "btree_gist.h"
5
6 #include <math.h>
7 #include <limits.h>
8 #include <float.h>
9
10 #include "btree_utils_var.h"
11 #include "utils/pg_locale.h"
12 #include "utils/builtins.h"
13 #include "utils/rel.h"
14
15 PG_FUNCTION_INFO_V1(gbt_var_decompress);
16 Datum gbt_var_decompress(PG_FUNCTION_ARGS);
17
18
19 Datum
20 gbt_var_decompress(PG_FUNCTION_ARGS)
21 {
22 GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
23 GBT_VARKEY *key = (GBT_VARKEY *) DatumGetPointer(PG_DETOAST_DATUM(entry->key));
24
25 if (key != (GBT_VARKEY *) DatumGetPointer(entry->key))
26 {
27 GISTENTRY *retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
28
29 gistentryinit(*retval, PointerGetDatum(key),
30 entry->rel, entry->page,
31 entry->offset, FALSE);
32
33 PG_RETURN_POINTER(retval);
34 }
35
36 PG_RETURN_POINTER(entry);
37 }
38
39 /* Returns a better readable representaion of variable key ( sets pointer ) */
40 GBT_VARKEY_R
41 gbt_var_key_readable(const GBT_VARKEY *k)
42 {
43
44 GBT_VARKEY_R r;
45
46 r.lower = (bytea *) &(((char *) k)[VARHDRSZ]);
47 if (VARSIZE(k) > (VARHDRSZ + (VARSIZE(r.lower))))
48 r.upper = (bytea *) &(((char *) k)[VARHDRSZ + INTALIGN(VARSIZE(r.lower))]);
49 else
50 r.upper = r.lower;
51 return r;
52 }
53
54
55 GBT_VARKEY *
56 gbt_var_key_copy(const GBT_VARKEY_R *u, bool force_node)
57 {
58 GBT_VARKEY *r = NULL;
59
60 if (u->lower == u->upper && !force_node)
61 { /* leaf key mode */
62 r = (GBT_VARKEY *) palloc(VARSIZE(u->lower) + VARHDRSZ);
63 memcpy(VARDATA(r), u->lower, VARSIZE(u->lower));
64 SET_VARSIZE(r, VARSIZE(u->lower) + VARHDRSZ);
65 }
66 else
67 { /* node key mode */
68 r = (GBT_VARKEY *) palloc(INTALIGN(VARSIZE(u->lower)) + VARSIZE(u->upper) + VARHDRSZ);
69 memcpy(VARDATA(r), u->lower, VARSIZE(u->lower));
70 memcpy(VARDATA(r) + INTALIGN(VARSIZE(u->lower)), u->upper, VARSIZE(u->upper));
71 SET_VARSIZE(r, INTALIGN(VARSIZE(u->lower)) + VARSIZE(u->upper) + VARHDRSZ);
72 }
73 return r;
74 }
75
76
77 static GBT_VARKEY *
78 gbt_var_leaf2node(GBT_VARKEY *leaf, const gbtree_vinfo *tinfo)
79 {
80 GBT_VARKEY *out = leaf;
81
82 if (tinfo->f_l2n)
83 out = (*tinfo->f_l2n) (leaf);
84
85 return out;
86 }
87
88
89 /*
90 * returns the common prefix length of a node key
91 */
92 static int32
93 gbt_var_node_cp_len(const GBT_VARKEY *node, const gbtree_vinfo *tinfo)
94 {
95
96 GBT_VARKEY_R r = gbt_var_key_readable(node);
97 int32 i = 0;
98 int32 l = 0;
99 int32 t1len = VARSIZE(r.lower) - VARHDRSZ;
100 int32 t2len = VARSIZE(r.upper) - VARHDRSZ;
101 int32 ml = Min(t1len, t2len);
102
103 char *p1 = VARDATA(r.lower);
104 char *p2 = VARDATA(r.upper);
105
106 if (ml == 0)
107 return 0;
108
109 while (i < ml)
110 {
111 if (tinfo->eml > 1 && l == 0)
112 {
113
114 if ((l = pg_mblen(p1)) != pg_mblen(p2))
115 {
116 return i;
117 }
118 }
119 if (*p1 != *p2)
120 {
121 if (tinfo->eml > 1)
122 {
123 return (i - l + 1);
124 }
125 else
126 {
127 return i;
128 }
129 }
130
131 p1++;
132 p2++;
133 l--;
134 i++;
135 }
136 return (ml); /* lower == upper */
137 }
138
139
140 /*
141 * returns true, if query matches prefix ( common prefix )
142 */
143 static bool
144 gbt_bytea_pf_match(const bytea *pf, const bytea *query, const gbtree_vinfo *tinfo)
145 {
146
147 bool out = FALSE;
148 int32 k = 0;
149 int32 qlen = VARSIZE(query) - VARHDRSZ;
150 int32 nlen = VARSIZE(pf) - VARHDRSZ;
151
152 if (nlen <= qlen)
153 {
154 char *q = VARDATA(query);
155 char *n = VARDATA(pf);
156
157 if (tinfo->eml > 1)
158 {
159 out = (varstr_cmp(q, nlen, n, nlen) == 0);
160 }
161 else
162 {
163 out = TRUE;
164 for (k = 0; k < nlen; k++)
165 {
166 if (*n != *q)
167 {
168 out = FALSE;
169 break;
170 }
171 if (k < (nlen - 1))
172 {
173 q++;
174 n++;
175 }
176 }
177 }
178 }
179
180 return out;
181 }
182
183
184 /*
185 * returns true, if query matches node using common prefix
186 */
187
188 static bool
189 gbt_var_node_pf_match(const GBT_VARKEY_R *node, const bytea *query, const gbtree_vinfo *tinfo)
190 {
191
192 return (tinfo->trnc && (
193 gbt_bytea_pf_match(node->lower, query, tinfo) ||
194 gbt_bytea_pf_match(node->upper, query, tinfo)
195 ));
196
197 }
198
199
200 /*
201 * truncates / compresses the node key
202 * cpf_length .. common prefix length
203 */
204 static GBT_VARKEY *
205 gbt_var_node_truncate(const GBT_VARKEY *node, int32 cpf_length, const gbtree_vinfo *tinfo)
206 {
207 GBT_VARKEY *out = NULL;
208 GBT_VARKEY_R r = gbt_var_key_readable(node);
209 int32 len1 = VARSIZE(r.lower) - VARHDRSZ;
210 int32 len2 = VARSIZE(r.upper) - VARHDRSZ;
211 int32 si;
212 char *out2;
213
214 len1 = Min(len1, (cpf_length + 1));
215 len2 = Min(len2, (cpf_length + 1));
216
217 si = 2 * VARHDRSZ + INTALIGN(len1 + VARHDRSZ) + len2;
218 out = (GBT_VARKEY *) palloc(si);
219 SET_VARSIZE(out, si);
220
221 memcpy(VARDATA(out), r.lower, len1 + VARHDRSZ);
222 SET_VARSIZE(VARDATA(out), len1 + VARHDRSZ);
223
224 out2 = VARDATA(out) + INTALIGN(len1 + VARHDRSZ);
225 memcpy(out2, r.upper, len2 + VARHDRSZ);
226 SET_VARSIZE(out2, len2 + VARHDRSZ);
227
228 return out;
229 }
230
231
232
233 void
234 gbt_var_bin_union(Datum *u, GBT_VARKEY *e, const gbtree_vinfo *tinfo)
235 {
236
237 GBT_VARKEY *nk = NULL;
238 GBT_VARKEY *tmp = NULL;
239 GBT_VARKEY_R nr;
240 GBT_VARKEY_R eo = gbt_var_key_readable(e);
241
242 if (eo.lower == eo.upper) /* leaf */
243 {
244 tmp = gbt_var_leaf2node(e, tinfo);
245 if (tmp != e)
246 eo = gbt_var_key_readable(tmp);
247 }
248
249 if (DatumGetPointer(*u))
250 {
251
252 GBT_VARKEY_R ro = gbt_var_key_readable((GBT_VARKEY *) DatumGetPointer(*u));
253
254 if ((*tinfo->f_cmp) ((bytea *) ro.lower, (bytea *) eo.lower) > 0)
255 {
256 nr.lower = eo.lower;
257 nr.upper = ro.upper;
258 nk = gbt_var_key_copy(&nr, TRUE);
259 }
260
261 if ((*tinfo->f_cmp) ((bytea *) ro.upper, (bytea *) eo.upper) < 0)
262 {
263 nr.upper = eo.upper;
264 nr.lower = ro.lower;
265 nk = gbt_var_key_copy(&nr, TRUE);
266 }
267
268 if (nk)
269 *u = PointerGetDatum(nk);
270 }
271 else
272 {
273 nr.lower = eo.lower;
274 nr.upper = eo.upper;
275 *u = PointerGetDatum(gbt_var_key_copy(&nr, TRUE));
276 }
277 }
278
279
280
281 GISTENTRY *
282 gbt_var_compress(GISTENTRY *entry, const gbtree_vinfo *tinfo)
283 {
284
285 GISTENTRY *retval;
286
287 if (entry->leafkey)
288 {
289 GBT_VARKEY *r = NULL;
290 bytea *leaf = (bytea *) DatumGetPointer(PG_DETOAST_DATUM(entry->key));
291 GBT_VARKEY_R u;
292
293 u.lower = u.upper = leaf;
294 r = gbt_var_key_copy(&u, FALSE);
295
296 retval = palloc(sizeof(GISTENTRY));
297 gistentryinit(*retval, PointerGetDatum(r),
298 entry->rel, entry->page,
299 entry->offset, TRUE);
300 }
301 else
302 retval = entry;
303
304 return (retval);
305 }
306
307
308
309 GBT_VARKEY *
310 gbt_var_union(const GistEntryVector *entryvec, int32 *size, const gbtree_vinfo *tinfo)
311 {
312
313 int i = 0,
314 numranges = entryvec->n;
315 GBT_VARKEY *cur;
316 Datum out;
317 GBT_VARKEY_R rk;
318
319 *size = sizeof(GBT_VARKEY);
320
321 cur = (GBT_VARKEY *) DatumGetPointer(entryvec->vector[0].key);
322 rk = gbt_var_key_readable(cur);
323 out = PointerGetDatum(gbt_var_key_copy(&rk, TRUE));
324
325 for (i = 1; i < numranges; i++)
326 {
327 cur = (GBT_VARKEY *) DatumGetPointer(entryvec->vector[i].key);
328 gbt_var_bin_union(&out, cur, tinfo);
329 }
330
331
332 /* Truncate (=compress) key */
333 if (tinfo->trnc)
334 {
335 int32 plen;
336 GBT_VARKEY *trc = NULL;
337
338 plen = gbt_var_node_cp_len((GBT_VARKEY *) DatumGetPointer(out), tinfo);
339 trc = gbt_var_node_truncate((GBT_VARKEY *) DatumGetPointer(out), plen + 1, tinfo);
340
341 out = PointerGetDatum(trc);
342 }
343
344 return ((GBT_VARKEY *) DatumGetPointer(out));
345 }
346
347
348 bool
349 gbt_var_same(bool *result, const Datum d1, const Datum d2, const gbtree_vinfo *tinfo)
350 {
351
352 GBT_VARKEY *t1 = (GBT_VARKEY *) DatumGetPointer(d1);
353 GBT_VARKEY *t2 = (GBT_VARKEY *) DatumGetPointer(d2);
354 GBT_VARKEY_R r1,
355 r2;
356
357 r1 = gbt_var_key_readable(t1);
358 r2 = gbt_var_key_readable(t2);
359
360 if (t1 && t2)
361 {
362 *result = (((*tinfo->f_cmp) ((bytea *) r1.lower, (bytea *) r2.lower) == 0
363 && (*tinfo->f_cmp) ((bytea *) r1.upper, (bytea *) r2.upper) == 0) ? TRUE : FALSE);
364 }
365 else
366 *result = (t1 == NULL && t2 == NULL) ? TRUE : FALSE;
367
368 PG_RETURN_POINTER(result);
369 }
370
371
372
373 float *
374 gbt_var_penalty(float *res, const GISTENTRY *o, const GISTENTRY *n, const gbtree_vinfo *tinfo)
375 {
376
377 GBT_VARKEY *orge = (GBT_VARKEY *) DatumGetPointer(o->key);
378 GBT_VARKEY *newe = (GBT_VARKEY *) DatumGetPointer(n->key);
379 GBT_VARKEY_R ok,
380 nk;
381 GBT_VARKEY *tmp = NULL;
382
383 *res = 0.0;
384
385 nk = gbt_var_key_readable(newe);
386 if (nk.lower == nk.upper) /* leaf */
387 {
388 tmp = gbt_var_leaf2node(newe, tinfo);
389 if (tmp != newe)
390 nk = gbt_var_key_readable(tmp);
391 }
392 ok = gbt_var_key_readable(orge);
393
394 if ((VARSIZE(ok.lower) - VARHDRSZ) == 0 && (VARSIZE(ok.upper) - VARHDRSZ) == 0)
395 *res = 0.0;
396 else if (!(
397 (
398 ((*tinfo->f_cmp) (nk.lower, ok.lower) >= 0 || gbt_bytea_pf_match(ok.lower, nk.lower, tinfo)) &&
399 ((*tinfo->f_cmp) (nk.upper, ok.upper) <= 0 || gbt_bytea_pf_match(ok.upper, nk.upper, tinfo))
400 )
401 ))
402 {
403 Datum d = PointerGetDatum(0);
404 double dres = 0.0;
405 int32 ol,
406 ul;
407
408 gbt_var_bin_union(&d, orge, tinfo);
409 ol = gbt_var_node_cp_len((GBT_VARKEY *) DatumGetPointer(d), tinfo);
410 gbt_var_bin_union(&d, newe, tinfo);
411 ul = gbt_var_node_cp_len((GBT_VARKEY *) DatumGetPointer(d), tinfo);
412
413 if (ul < ol)
414 {
415 dres = (ol - ul); /* lost of common prefix len */
416 }
417 else
418 {
419 GBT_VARKEY_R uk = gbt_var_key_readable((GBT_VARKEY *) DatumGetPointer(d));
420
421 char tmp[4];
422
423 tmp[0] = ((VARSIZE(ok.lower) - VARHDRSZ) == ul) ? (CHAR_MIN) : (VARDATA(ok.lower)[ul]);
424 tmp[1] = ((VARSIZE(uk.lower) - VARHDRSZ) == ul) ? (CHAR_MIN) : (VARDATA(uk.lower)[ul]);
425 tmp[2] = ((VARSIZE(ok.upper) - VARHDRSZ) == ul) ? (CHAR_MIN) : (VARDATA(ok.upper)[ul]);
426 tmp[3] = ((VARSIZE(uk.upper) - VARHDRSZ) == ul) ? (CHAR_MIN) : (VARDATA(uk.upper)[ul]);
427 dres = (tmp[0] - tmp[1]) +
428 (tmp[3] - tmp[2]);
429 dres /= 256.0;
430 }
431
432 *res += FLT_MIN;
433 *res += (float) (dres / ((double) (ol + 1)));
434 *res *= (FLT_MAX / (o->rel->rd_att->natts + 1));
435 }
436
437 return res;
438 }
439
440
441 static int
442 gbt_vsrt_cmp(const void *a, const void *b, void *arg)
443 {
444 GBT_VARKEY_R ar = gbt_var_key_readable(((const Vsrt *) a)->t);
445 GBT_VARKEY_R br = gbt_var_key_readable(((const Vsrt *) b)->t);
446 const gbtree_vinfo *tinfo = (const gbtree_vinfo *) arg;
447
448 return (*tinfo->f_cmp) (ar.lower, br.lower);
449 }
450
451 GIST_SPLITVEC *
452 gbt_var_picksplit(const GistEntryVector *entryvec, GIST_SPLITVEC *v, const gbtree_vinfo *tinfo)
453 {
454 OffsetNumber i,
455 maxoff = entryvec->n - 1;
456 Vsrt *arr;
457 int svcntr = 0,
458 nbytes;
459 char *cur;
460 GBT_VARKEY **sv = NULL;
461
462 arr = (Vsrt *) palloc((maxoff + 1) * sizeof(Vsrt));
463 nbytes = (maxoff + 2) * sizeof(OffsetNumber);
464 v->spl_left = (OffsetNumber *) palloc(nbytes);
465 v->spl_right = (OffsetNumber *) palloc(nbytes);
466 v->spl_ldatum = PointerGetDatum(0);
467 v->spl_rdatum = PointerGetDatum(0);
468 v->spl_nleft = 0;
469 v->spl_nright = 0;
470
471 sv = palloc(sizeof(bytea *) * (maxoff + 1));
472
473 /* Sort entries */
474
475 for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
476 {
477 GBT_VARKEY_R ro;
478
479 cur = (char *) DatumGetPointer(entryvec->vector[i].key);
480 ro = gbt_var_key_readable((GBT_VARKEY *) cur);
481 if (ro.lower == ro.upper) /* leaf */
482 {
483 sv[svcntr] = gbt_var_leaf2node((GBT_VARKEY *) cur, tinfo);
484 arr[i].t = sv[svcntr];
485 if (sv[svcntr] != (GBT_VARKEY *) cur)
486 svcntr++;
487 }
488 else
489 arr[i].t = (GBT_VARKEY *) cur;
490 arr[i].i = i;
491 }
492
493 /* sort */
494 qsort_arg((void *) &arr[FirstOffsetNumber],
495 maxoff - FirstOffsetNumber + 1,
496 sizeof(Vsrt),
497 gbt_vsrt_cmp,
498 (void *) tinfo);
499
500 /* We do simply create two parts */
501
502 for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
503 {
504 if (i <= (maxoff - FirstOffsetNumber + 1) / 2)
505 {
506 gbt_var_bin_union(&v->spl_ldatum, arr[i].t, tinfo);
507 v->spl_left[v->spl_nleft] = arr[i].i;
508 v->spl_nleft++;
509 }
510 else
511 {
512 gbt_var_bin_union(&v->spl_rdatum, arr[i].t, tinfo);
513 v->spl_right[v->spl_nright] = arr[i].i;
514 v->spl_nright++;
515 }
516 }
517
518 /* Truncate (=compress) key */
519 if (tinfo->trnc)
520 {
521 int32 ll = gbt_var_node_cp_len((GBT_VARKEY *) DatumGetPointer(v->spl_ldatum), tinfo);
522 int32 lr = gbt_var_node_cp_len((GBT_VARKEY *) DatumGetPointer(v->spl_rdatum), tinfo);
523 GBT_VARKEY *dl;
524 GBT_VARKEY *dr;
525
526 ll = Max(ll, lr);
527 ll++;
528
529 dl = gbt_var_node_truncate((GBT_VARKEY *) DatumGetPointer(v->spl_ldatum), ll, tinfo);
530 dr = gbt_var_node_truncate((GBT_VARKEY *) DatumGetPointer(v->spl_rdatum), ll, tinfo);
531 v->spl_ldatum = PointerGetDatum(dl);
532 v->spl_rdatum = PointerGetDatum(dr);
533 }
534
535 return v;
536 }
537
538
539 /*
540 * The GiST consistent method
541 */
542 bool
543 gbt_var_consistent(
544 GBT_VARKEY_R *key,
545 const void *query,
546 const StrategyNumber *strategy,
547 bool is_leaf,
548 const gbtree_vinfo *tinfo
549 )
550 {
551 bool retval = FALSE;
552
553 switch (*strategy)
554 {
555 case BTLessEqualStrategyNumber:
556 if (is_leaf)
557 retval = (*tinfo->f_ge) (query, (void *) key->lower);
558 else
559 retval = (*tinfo->f_cmp) ((bytea *) query, key->lower) >= 0
560 || gbt_var_node_pf_match(key, query, tinfo);
561 break;
562 case BTLessStrategyNumber:
563 if (is_leaf)
564 retval = (*tinfo->f_gt) (query, (void *) key->lower);
565 else
566 retval = (*tinfo->f_cmp) ((bytea *) query, key->lower) >= 0
567 || gbt_var_node_pf_match(key, query, tinfo);
568 break;
569 case BTEqualStrategyNumber:
570 if (is_leaf)
571 retval = (*tinfo->f_eq) (query, (void *) key->lower);
572 else
573 retval = (
574 (
575 (*tinfo->f_cmp) (key->lower, (bytea *) query) <= 0 &&
576 (*tinfo->f_cmp) ((bytea *) query, (void *) key->upper) <= 0
577 ) || gbt_var_node_pf_match(key, query, tinfo)
578 );
579 break;
580 case BTGreaterStrategyNumber:
581 if (is_leaf)
582 retval = (*tinfo->f_lt) (query, (void *) key->upper);
583 else
584 retval = (*tinfo->f_cmp) ((bytea *) query, key->upper) <= 0
585 || gbt_var_node_pf_match(key, query, tinfo);
586 break;
587 case BTGreaterEqualStrategyNumber:
588 if (is_leaf)
589 retval = (*tinfo->f_le) (query, (void *) key->upper);
590 else
591 retval = (*tinfo->f_cmp) ((bytea *) query, key->upper) <= 0
592 || gbt_var_node_pf_match(key, query, tinfo);
593 break;
594 default:
595 retval = FALSE;
596 }
597
598 return (retval);
599 }
PostgreSQL from CVS