3 * Teodor Sigaev <teodor@sigaev.ru>
9 #include "access/gist.h"
10 #include "access/itup.h"
11 #include "catalog/namespace.h"
12 #include "commands/trigger.h"
13 #include "executor/spi.h"
16 #include "nodes/pg_list.h"
17 #include "storage/bufpage.h"
18 #include "utils/array.h"
19 #include "utils/builtins.h"
25 PG_FUNCTION_INFO_V1(rank);
26 Datum rank(PG_FUNCTION_ARGS);
28 PG_FUNCTION_INFO_V1(rank_def);
29 Datum rank_def(PG_FUNCTION_ARGS);
31 PG_FUNCTION_INFO_V1(rank_cd);
32 Datum rank_cd(PG_FUNCTION_ARGS);
34 PG_FUNCTION_INFO_V1(rank_cd_def);
35 Datum rank_cd_def(PG_FUNCTION_ARGS);
37 PG_FUNCTION_INFO_V1(get_covers);
38 Datum get_covers(PG_FUNCTION_ARGS);
40 static float weights[] = {0.1, 0.2, 0.4, 1.0};
42 #define wpos(wep) ( w[ WEP_GETWEIGHT(wep) ] )
44 #define RANK_NO_NORM 0x00
45 #define RANK_NORM_LOGLENGTH 0x01
46 #define RANK_NORM_LENGTH 0x02
47 #define RANK_NORM_EXTDIST 0x04
48 #define RANK_NORM_UNIQ 0x08
49 #define RANK_NORM_LOGUNIQ 0x10
50 #define DEF_NORM_METHOD RANK_NO_NORM
52 static float calc_rank_or(float *w, tsvector * t, QUERYTYPE * q);
53 static float calc_rank_and(float *w, tsvector * t, QUERYTYPE * q);
56 * Returns a weight of a word collocation
64 return 1.0 / (1.005 + 0.05 * exp(((float4) w) / 1.5 - 2));
68 cnt_length(tsvector * t)
70 WordEntry *ptr = ARRPTR(t),
71 *end = (WordEntry *) STRPTR(t);
77 if ((clen = POSDATALEN(t, ptr)) == 0)
88 WordECompareITEM(char *eval, char *qval, WordEntry * ptr, ITEM * item)
90 if (ptr->len == item->length)
93 qval + item->distance,
96 return (ptr->len > item->length) ? 1 : -1;
100 find_wordentry(tsvector * t, QUERYTYPE * q, ITEM * item)
102 WordEntry *StopLow = ARRPTR(t);
103 WordEntry *StopHigh = (WordEntry *) STRPTR(t);
104 WordEntry *StopMiddle;
107 /* Loop invariant: StopLow <= item < StopHigh */
109 while (StopLow < StopHigh)
111 StopMiddle = StopLow + (StopHigh - StopLow) / 2;
112 difference = WordECompareITEM(STRPTR(t), GETOPERAND(q), StopMiddle, item);
115 else if (difference < 0)
116 StopLow = StopMiddle + 1;
118 StopHigh = StopMiddle;
125 static char *SortAndUniqOperand = NULL;
128 compareITEM(const void *a, const void *b)
130 if ((*(ITEM **) a)->length == (*(ITEM **) b)->length)
131 return strncmp(SortAndUniqOperand + (*(ITEM **) a)->distance,
132 SortAndUniqOperand + (*(ITEM **) b)->distance,
133 (*(ITEM **) b)->length);
135 return ((*(ITEM **) a)->length > (*(ITEM **) b)->length) ? 1 : -1;
139 SortAndUniqItems(char *operand, ITEM * item, int *size)
145 ptr = res = (ITEM **) palloc(sizeof(ITEM *) * *size);
149 if (item->type == VAL)
161 SortAndUniqOperand = operand;
162 qsort(res, *size, sizeof(ITEM **), compareITEM);
167 while (ptr - res < *size)
169 if (compareITEM((void *) ptr, (void *) prevptr) != 0)
177 *size = prevptr + 1 - res;
181 static WordEntryPos POSNULL[] = {
187 calc_rank_and(float *w, tsvector * t, QUERYTYPE * q)
204 item = SortAndUniqItems(GETOPERAND(q), GETQUERY(q), &size);
208 return calc_rank_or(w, t, q);
210 pos = (uint16 **) palloc(sizeof(uint16 *) * q->size);
211 memset(pos, 0, sizeof(uint16 *) * q->size);
212 *(uint16 *) POSNULL = lengthof(POSNULL) - 1;
213 WEP_SETPOS(POSNULL[1], MAXENTRYPOS - 1);
215 for (i = 0; i < size; i++)
217 entry = find_wordentry(t, q, item[i]);
222 pos[i] = (uint16 *) _POSDATAPTR(t, entry);
224 pos[i] = (uint16 *) POSNULL;
227 dimt = *(uint16 *) (pos[i]);
228 post = (WordEntryPos *) (pos[i] + 1);
229 for (k = 0; k < i; k++)
233 lenct = *(uint16 *) (pos[k]);
234 ct = (WordEntryPos *) (pos[k] + 1);
235 for (l = 0; l < dimt; l++)
237 for (p = 0; p < lenct; p++)
239 dist = Abs((int) WEP_GETPOS(post[l]) - (int) WEP_GETPOS(ct[p]));
240 if (dist || (dist == 0 && (pos[i] == (uint16 *) POSNULL || pos[k] == (uint16 *) POSNULL)))
246 curw = sqrt(wpos(post[l]) * wpos(ct[p]) * word_distance(dist));
247 res = (res < 0) ? curw : 1.0 - (1.0 - res) * (1.0 - curw);
259 calc_rank_or(float *w, tsvector * t, QUERYTYPE * q)
270 *(uint16 *) POSNULL = lengthof(POSNULL) - 1;
271 item = SortAndUniqItems(GETOPERAND(q), GETQUERY(q), &size);
273 for (i = 0; i < size; i++)
279 entry = find_wordentry(t, q, item[i]);
285 dimt = POSDATALEN(t, entry);
286 post = POSDATAPTR(t, entry);
290 dimt = *(uint16 *) POSNULL;
297 for (j = 0; j < dimt; j++)
299 resj = resj + wpos(post[j]) / ((j + 1) * (j + 1));
300 if (wpos(post[j]) > wjm)
307 limit (sum(i/i^2),i->inf) = pi^2/6
308 resj = sum(wi/i^2),i=1,noccurence,
309 wi - should be sorted desc,
310 don't sort for now, just choose maximum weight. This should be corrected
313 res = res + (wjm + resj - wjm / ((jm + 1) * (jm + 1))) / 1.64493406685;
322 calc_rank(float *w, tsvector * t, QUERYTYPE * q, int4 method)
324 ITEM *item = GETQUERY(q);
328 if (!t->size || !q->size)
331 res = (item->type != VAL && item->val == (int4) '&') ?
332 calc_rank_and(w, t, q) : calc_rank_or(w, t, q);
337 if ((method & RANK_NORM_LOGLENGTH) && t->size > 0)
338 res /= log((double) (cnt_length(t) + 1)) / log(2.0);
340 if (method & RANK_NORM_LENGTH)
347 if ((method & RANK_NORM_UNIQ) && t->size > 0)
348 res /= (float) (t->size);
350 if ((method & RANK_NORM_LOGUNIQ) && t->size > 0)
351 res /= log((double) (t->size + 1)) / log(2.0);
357 rank(PG_FUNCTION_ARGS)
359 ArrayType *win = (ArrayType *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
360 tsvector *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(1));
361 QUERYTYPE *query = (QUERYTYPE *) PG_DETOAST_DATUM(PG_GETARG_DATUM(2));
362 int method = DEF_NORM_METHOD;
364 float ws[lengthof(weights)];
368 if (ARR_NDIM(win) != 1)
370 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
371 errmsg("array of weight must be one-dimensional")));
373 if (ARRNELEMS(win) < lengthof(weights))
375 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
376 errmsg("array of weight is too short")));
378 if (ARR_HASNULL(win))
380 (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
381 errmsg("array of weight must not contain nulls")));
383 arrdata = (float4 *) ARR_DATA_PTR(win);
384 for (i = 0; i < lengthof(weights); i++)
386 ws[i] = (arrdata[i] >= 0) ? arrdata[i] : weights[i];
389 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
390 errmsg("weight out of range")));
394 method = PG_GETARG_INT32(3);
396 res = calc_rank(ws, txt, query, method);
398 PG_FREE_IF_COPY(win, 0);
399 PG_FREE_IF_COPY(txt, 1);
400 PG_FREE_IF_COPY(query, 2);
401 PG_RETURN_FLOAT4(res);
405 rank_def(PG_FUNCTION_ARGS)
407 tsvector *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
408 QUERYTYPE *query = (QUERYTYPE *) PG_DETOAST_DATUM(PG_GETARG_DATUM(1));
410 int method = DEF_NORM_METHOD;
413 method = PG_GETARG_INT32(2);
415 res = calc_rank(weights, txt, query, method);
417 PG_FREE_IF_COPY(txt, 0);
418 PG_FREE_IF_COPY(query, 1);
419 PG_RETURN_FLOAT4(res);
433 compareDocR(const void *a, const void *b)
435 if (((DocRepresentation *) a)->pos == ((DocRepresentation *) b)->pos)
437 return (((DocRepresentation *) a)->pos > ((DocRepresentation *) b)->pos) ? 1 : -1;
441 checkcondition_ITEM(void *checkval, ITEM * val)
443 return (bool) (val->istrue);
447 reset_istrue_flag(QUERYTYPE * query)
449 ITEM *item = GETQUERY(query);
452 /* reset istrue flag */
453 for (i = 0; i < query->size; i++)
455 if (item->type == VAL)
466 DocRepresentation *begin;
467 DocRepresentation *end;
472 Cover(DocRepresentation * doc, int len, QUERYTYPE * query, Extention * ext)
474 DocRepresentation *ptr;
475 int lastpos = ext->pos;
479 reset_istrue_flag(query);
483 ptr = doc + ext->pos;
485 /* find upper bound of cover from current position, move up */
486 while (ptr - doc < len)
488 for (i = 0; i < ptr->nitem; i++)
489 ptr->item[i]->istrue = 1;
490 if (TS_execute(GETQUERY(query), NULL, false, checkcondition_ITEM))
492 if (ptr->pos > ext->q)
507 reset_istrue_flag(query);
511 /* find lower bound of cover from founded upper bound, move down */
514 for (i = 0; i < ptr->nitem; i++)
515 ptr->item[i]->istrue = 1;
516 if (TS_execute(GETQUERY(query), NULL, true, checkcondition_ITEM))
518 if (ptr->pos < ext->p)
528 if (ext->p <= ext->q)
531 * set position for next try to next lexeme after begining of founded
534 ext->pos = (ptr - doc) + 1;
539 return Cover(doc, len, query, ext);
542 static DocRepresentation *
543 get_docrep(tsvector * txt, QUERYTYPE * query, int *doclen)
545 ITEM *item = GETQUERY(query);
551 int len = query->size * 4,
553 DocRepresentation *doc;
555 *(uint16 *) POSNULL = lengthof(POSNULL) - 1;
556 doc = (DocRepresentation *) palloc(sizeof(DocRepresentation) * len);
557 SortAndUniqOperand = GETOPERAND(query);
558 reset_istrue_flag(query);
560 for (i = 0; i < query->size; i++)
562 if (item[i].type != VAL || item[i].istrue)
565 entry = find_wordentry(txt, query, &(item[i]));
571 dimt = POSDATALEN(txt, entry);
572 post = POSDATAPTR(txt, entry);
576 dimt = *(uint16 *) POSNULL;
580 while (cur + dimt >= len)
583 doc = (DocRepresentation *) repalloc(doc, sizeof(DocRepresentation) * len);
586 for (j = 0; j < dimt; j++)
594 doc[cur].needfree = false;
596 doc[cur].item = (ITEM **) palloc(sizeof(ITEM *) * query->size);
598 for (k = 0; k < query->size; k++)
601 if (k == i || (item[k].type == VAL && compareITEM(&kptr, &iptr) == 0))
603 doc[cur].item[doc[cur].nitem] = item + k;
611 doc[cur].needfree = false;
612 doc[cur].nitem = doc[cur - 1].nitem;
613 doc[cur].item = doc[cur - 1].item;
615 doc[cur].pos = WEP_GETPOS(post[j]);
616 doc[cur].wclass = WEP_GETWEIGHT(post[j]);
626 qsort((void *) doc, cur, sizeof(DocRepresentation), compareDocR);
635 calc_rank_cd(float4 *arrdata, tsvector * txt, QUERYTYPE * query, int method)
637 DocRepresentation *doc;
643 double invws[lengthof(weights)];
644 double SumDist = 0.0,
649 for (i = 0; i < lengthof(weights); i++)
651 invws[i] = ((double) ((arrdata[i] >= 0) ? arrdata[i] : weights[i]));
654 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
655 errmsg("weight out of range")));
656 invws[i] = 1.0 / invws[i];
659 doc = get_docrep(txt, query, &doclen);
663 MemSet(&ext, 0, sizeof(Extention));
664 while (Cover(doc, doclen, query, &ext))
668 DocRepresentation *ptr = ext.begin;
670 while (ptr <= ext.end)
672 InvSum += invws[ptr->wclass];
676 Cpos = ((double) (ext.end - ext.begin + 1)) / InvSum;
677 Wdoc += Cpos / ((double) ((1 + (ext.q - ext.p) - (ext.end - ext.begin))));
679 CurExtPos = ((double) (ext.q + ext.p)) / 2.0;
680 if (NExtent > 0 && CurExtPos > PrevExtPos /* prevent devision by
683 SumDist += 1.0 / (CurExtPos - PrevExtPos);
685 PrevExtPos = CurExtPos;
689 if ((method & RANK_NORM_LOGLENGTH) && txt->size > 0)
690 Wdoc /= log((double) (cnt_length(txt) + 1));
692 if (method & RANK_NORM_LENGTH)
694 len = cnt_length(txt);
696 Wdoc /= (double) len;
699 if ((method & RANK_NORM_EXTDIST) && SumDist > 0)
700 Wdoc /= ((double) NExtent) / SumDist;
702 if ((method & RANK_NORM_UNIQ) && txt->size > 0)
703 Wdoc /= (double) (txt->size);
705 if ((method & RANK_NORM_LOGUNIQ) && txt->size > 0)
706 Wdoc /= log((double) (txt->size + 1)) / log(2.0);
708 for (i = 0; i < doclen; i++)
713 return (float4) Wdoc;
717 rank_cd(PG_FUNCTION_ARGS)
719 ArrayType *win = (ArrayType *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
720 tsvector *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(1));
721 QUERYTYPE *query = (QUERYTYPE *) PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(2));
722 int method = DEF_NORM_METHOD;
725 if (ARR_NDIM(win) != 1)
727 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
728 errmsg("array of weight must be one-dimensional")));
730 if (ARRNELEMS(win) < lengthof(weights))
732 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
733 errmsg("array of weight is too short")));
735 if (ARR_HASNULL(win))
737 (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
738 errmsg("array of weight must not contain nulls")));
741 method = PG_GETARG_INT32(3);
743 res = calc_rank_cd((float4 *) ARR_DATA_PTR(win), txt, query, method);
745 PG_FREE_IF_COPY(win, 0);
746 PG_FREE_IF_COPY(txt, 1);
747 PG_FREE_IF_COPY(query, 2);
749 PG_RETURN_FLOAT4(res);
754 rank_cd_def(PG_FUNCTION_ARGS)
756 tsvector *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
757 QUERYTYPE *query = (QUERYTYPE *) PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(1));
760 res = calc_rank_cd(weights, txt, query, (PG_NARGS() == 3) ? PG_GETARG_DATUM(2) : DEF_NORM_METHOD);
762 PG_FREE_IF_COPY(txt, 0);
763 PG_FREE_IF_COPY(query, 1);
765 PG_RETURN_FLOAT4(res);
768 /**************debug*************/
780 compareDocWord(const void *a, const void *b)
782 if (((DocWord *) a)->pos == ((DocWord *) b)->pos)
784 return (((DocWord *) a)->pos > ((DocWord *) b)->pos) ? 1 : -1;
789 get_covers(PG_FUNCTION_ARGS)
791 tsvector *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
792 QUERYTYPE *query = (QUERYTYPE *) PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(1));
793 WordEntry *pptr = ARRPTR(txt);
804 DocRepresentation *doc;
809 doc = get_docrep(txt, query, &rlen);
813 out = palloc(VARHDRSZ);
814 VARATT_SIZEP(out) = VARHDRSZ;
815 PG_FREE_IF_COPY(txt, 0);
816 PG_FREE_IF_COPY(query, 1);
817 PG_RETURN_POINTER(out);
820 for (i = 0; i < txt->size; i++)
824 (errcode(ERRCODE_SYNTAX_ERROR),
825 errmsg("no pos info")));
826 dlen += POSDATALEN(txt, &(pptr[i]));
829 dwptr = dw = palloc(sizeof(DocWord) * dlen);
830 memset(dw, 0, sizeof(DocWord) * dlen);
832 for (i = 0; i < txt->size; i++)
834 WordEntryPos *posdata = POSDATAPTR(txt, &(pptr[i]));
836 for (j = 0; j < POSDATALEN(txt, &(pptr[i])); j++)
838 dw[cur].w = STRPTR(txt) + pptr[i].pos;
839 dw[cur].len = pptr[i].len;
840 dw[cur].pos = WEP_GETPOS(posdata[j]);
843 len += (pptr[i].len + 1) * (int) POSDATALEN(txt, &(pptr[i]));
845 qsort((void *) dw, dlen, sizeof(DocWord), compareDocWord);
847 MemSet(&ext, 0, sizeof(Extention));
848 while (Cover(doc, rlen, query, &ext))
850 dwptr = dw + olddwpos;
851 while (dwptr->pos < ext.p && dwptr - dw < dlen)
853 olddwpos = dwptr - dw;
854 dwptr->start = ncover;
855 while (dwptr->pos < ext.q + 1 && dwptr - dw < dlen)
857 (dwptr - 1)->finish = ncover;
858 len += 4 /* {}+two spaces */ + 2 * 16 /* numbers */ ;
862 out = palloc(VARHDRSZ + len);
863 cptr = ((char *) out) + VARHDRSZ;
866 while (dwptr - dw < dlen)
870 sprintf(cptr, "{%d ", dwptr->start);
871 cptr = strchr(cptr, '\0');
873 memcpy(cptr, dwptr->w, dwptr->len);
879 sprintf(cptr, "}%d ", dwptr->finish);
880 cptr = strchr(cptr, '\0');
885 VARATT_SIZEP(out) = cptr - ((char *) out);
888 for (i = 0; i < rlen; i++)
893 PG_FREE_IF_COPY(txt, 0);
894 PG_FREE_IF_COPY(query, 1);
895 PG_RETURN_POINTER(out);