granicus.if.org Git - postgresql/blob - contrib/tsearch2/rank.c

   1 /*
   2  * Relevation
   3  * Teodor Sigaev <teodor@sigaev.ru>
   4  */
   5 #include "postgres.h"
   6
   7 #include <math.h>
   8
   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"
  14 #include "fmgr.h"
  15 #include "funcapi.h"
  16 #include "nodes/pg_list.h"
  17 #include "storage/bufpage.h"
  18 #include "utils/array.h"
  19 #include "utils/builtins.h"
  20
  21 #include "tsvector.h"
  22 #include "query.h"
  23 #include "common.h"
  24
  25 PG_FUNCTION_INFO_V1(rank);
  26 Datum           rank(PG_FUNCTION_ARGS);
  27
  28 PG_FUNCTION_INFO_V1(rank_def);
  29 Datum           rank_def(PG_FUNCTION_ARGS);
  30
  31 PG_FUNCTION_INFO_V1(rank_cd);
  32 Datum           rank_cd(PG_FUNCTION_ARGS);
  33
  34 PG_FUNCTION_INFO_V1(rank_cd_def);
  35 Datum           rank_cd_def(PG_FUNCTION_ARGS);
  36
  37 PG_FUNCTION_INFO_V1(get_covers);
  38 Datum           get_covers(PG_FUNCTION_ARGS);
  39
  40 static float weights[] = {0.1, 0.2, 0.4, 1.0};
  41
  42 #define wpos(wep)       ( w[ WEP_GETWEIGHT(wep) ] )
  43
  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
  51
  52 static float calc_rank_or(float *w, tsvector * t, QUERYTYPE * q);
  53 static float calc_rank_and(float *w, tsvector * t, QUERYTYPE * q);
  54
  55 /*
  56  * Returns a weight of a word collocation
  57  */
  58 static float4
  59 word_distance(int4 w)
  60 {
  61         if (w > 100)
  62                 return 1e-30;
  63
  64         return 1.0 / (1.005 + 0.05 * exp(((float4) w) / 1.5 - 2));
  65 }
  66
  67 static int
  68 cnt_length(tsvector * t)
  69 {
  70         WordEntry  *ptr = ARRPTR(t),
  71                            *end = (WordEntry *) STRPTR(t);
  72         int                     len = 0,
  73                                 clen;
  74
  75         while (ptr < end)
  76         {
  77                 if ((clen = POSDATALEN(t, ptr)) == 0)
  78                         len += 1;
  79                 else
  80                         len += clen;
  81                 ptr++;
  82         }
  83
  84         return len;
  85 }
  86
  87 static int4
  88 WordECompareITEM(char *eval, char *qval, WordEntry * ptr, ITEM * item)
  89 {
  90         if (ptr->len == item->length)
  91                 return strncmp(
  92                                            eval + ptr->pos,
  93                                            qval + item->distance,
  94                                            item->length);
  95
  96         return (ptr->len > item->length) ? 1 : -1;
  97 }
  98
  99 static WordEntry *
 100 find_wordentry(tsvector * t, QUERYTYPE * q, ITEM * item)
 101 {
 102         WordEntry  *StopLow = ARRPTR(t);
 103         WordEntry  *StopHigh = (WordEntry *) STRPTR(t);
 104         WordEntry  *StopMiddle;
 105         int                     difference;
 106
 107         /* Loop invariant: StopLow <= item < StopHigh */
 108
 109         while (StopLow < StopHigh)
 110         {
 111                 StopMiddle = StopLow + (StopHigh - StopLow) / 2;
 112                 difference = WordECompareITEM(STRPTR(t), GETOPERAND(q), StopMiddle, item);
 113                 if (difference == 0)
 114                         return StopMiddle;
 115                 else if (difference < 0)
 116                         StopLow = StopMiddle + 1;
 117                 else
 118                         StopHigh = StopMiddle;
 119         }
 120
 121         return NULL;
 122 }
 123
 124
 125 static char *SortAndUniqOperand = NULL;
 126
 127 static int
 128 compareITEM(const void *a, const void *b)
 129 {
 130         if ((*(ITEM **) a)->length == (*(ITEM **) b)->length)
 131                 return strncmp(SortAndUniqOperand + (*(ITEM **) a)->distance,
 132                                            SortAndUniqOperand + (*(ITEM **) b)->distance,
 133                                            (*(ITEM **) b)->length);
 134
 135         return ((*(ITEM **) a)->length > (*(ITEM **) b)->length) ? 1 : -1;
 136 }
 137
 138 static ITEM **
 139 SortAndUniqItems(char *operand, ITEM * item, int *size)
 140 {
 141         ITEM      **res,
 142                           **ptr,
 143                           **prevptr;
 144
 145         ptr = res = (ITEM **) palloc(sizeof(ITEM *) * *size);
 146
 147         while ((*size)--)
 148         {
 149                 if (item->type == VAL)
 150                 {
 151                         *ptr = item;
 152                         ptr++;
 153                 }
 154                 item++;
 155         }
 156
 157         *size = ptr - res;
 158         if (*size < 2)
 159                 return res;
 160
 161         SortAndUniqOperand = operand;
 162         qsort(res, *size, sizeof(ITEM **), compareITEM);
 163
 164         ptr = res + 1;
 165         prevptr = res;
 166
 167         while (ptr - res < *size)
 168         {
 169                 if (compareITEM((void *) ptr, (void *) prevptr) != 0)
 170                 {
 171                         prevptr++;
 172                         *prevptr = *ptr;
 173                 }
 174                 ptr++;
 175         }
 176
 177         *size = prevptr + 1 - res;
 178         return res;
 179 }
 180
 181 static WordEntryPos POSNULL[] = {
 182         0,
 183         0
 184 };
 185
 186 static float
 187 calc_rank_and(float *w, tsvector * t, QUERYTYPE * q)
 188 {
 189         uint16    **pos;
 190         int                     i,
 191                                 k,
 192                                 l,
 193                                 p;
 194         WordEntry  *entry;
 195         WordEntryPos *post,
 196                            *ct;
 197         int4            dimt,
 198                                 lenct,
 199                                 dist;
 200         float           res = -1.0;
 201         ITEM      **item;
 202         int                     size = q->size;
 203
 204         item = SortAndUniqItems(GETOPERAND(q), GETQUERY(q), &size);
 205         if (size < 2)
 206         {
 207                 pfree(item);
 208                 return calc_rank_or(w, t, q);
 209         }
 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);
 214
 215         for (i = 0; i < size; i++)
 216         {
 217                 entry = find_wordentry(t, q, item[i]);
 218                 if (!entry)
 219                         continue;
 220
 221                 if (entry->haspos)
 222                         pos[i] = (uint16 *) _POSDATAPTR(t, entry);
 223                 else
 224                         pos[i] = (uint16 *) POSNULL;
 225
 226
 227                 dimt = *(uint16 *) (pos[i]);
 228                 post = (WordEntryPos *) (pos[i] + 1);
 229                 for (k = 0; k < i; k++)
 230                 {
 231                         if (!pos[k])
 232                                 continue;
 233                         lenct = *(uint16 *) (pos[k]);
 234                         ct = (WordEntryPos *) (pos[k] + 1);
 235                         for (l = 0; l < dimt; l++)
 236                         {
 237                                 for (p = 0; p < lenct; p++)
 238                                 {
 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)))
 241                                         {
 242                                                 float           curw;
 243
 244                                                 if (!dist)
 245                                                         dist = MAXENTRYPOS;
 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);
 248                                         }
 249                                 }
 250                         }
 251                 }
 252         }
 253         pfree(pos);
 254         pfree(item);
 255         return res;
 256 }
 257
 258 static float
 259 calc_rank_or(float *w, tsvector * t, QUERYTYPE * q)
 260 {
 261         WordEntry  *entry;
 262         WordEntryPos *post;
 263         int4            dimt,
 264                                 j,
 265                                 i;
 266         float           res = 0.0;
 267         ITEM      **item;
 268         int                     size = q->size;
 269
 270         *(uint16 *) POSNULL = lengthof(POSNULL) - 1;
 271         item = SortAndUniqItems(GETOPERAND(q), GETQUERY(q), &size);
 272
 273         for (i = 0; i < size; i++)
 274         {
 275                 float           resj,
 276                                         wjm;
 277                 int4            jm;
 278
 279                 entry = find_wordentry(t, q, item[i]);
 280                 if (!entry)
 281                         continue;
 282
 283                 if (entry->haspos)
 284                 {
 285                         dimt = POSDATALEN(t, entry);
 286                         post = POSDATAPTR(t, entry);
 287                 }
 288                 else
 289                 {
 290                         dimt = *(uint16 *) POSNULL;
 291                         post = POSNULL + 1;
 292                 }
 293
 294                 resj = 0.0;
 295                 wjm = -1.0;
 296                 jm = 0;
 297                 for (j = 0; j < dimt; j++)
 298                 {
 299                         resj = resj + wpos(post[j]) / ((j + 1) * (j + 1));
 300                         if (wpos(post[j]) > wjm)
 301                         {
 302                                 wjm = wpos(post[j]);
 303                                 jm = j;
 304                         }
 305                 }
 306 /*
 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
 311                 Oleg Bartunov
 312 */
 313                 res = res + (wjm + resj - wjm / ((jm + 1) * (jm + 1))) / 1.64493406685;
 314         }
 315         if (size > 0)
 316                 res = res / size;
 317         pfree(item);
 318         return res;
 319 }
 320
 321 static float
 322 calc_rank(float *w, tsvector * t, QUERYTYPE * q, int4 method)
 323 {
 324         ITEM       *item = GETQUERY(q);
 325         float           res = 0.0;
 326         int                     len;
 327
 328         if (!t->size || !q->size)
 329                 return 0.0;
 330
 331         res = (item->type != VAL && item->val == (int4) '&') ?
 332                 calc_rank_and(w, t, q) : calc_rank_or(w, t, q);
 333
 334         if (res < 0)
 335                 res = 1e-20;
 336
 337         if ( (method & RANK_NORM_LOGLENGTH) && t->size>0 )
 338                 res /= log((double) (cnt_length(t) + 1)) / log(2.0);
 339
 340         if ( method & RANK_NORM_LENGTH ) {
 341                 len = cnt_length(t);
 342                 if ( len>0 )
 343                         res /= (float) len;
 344         }
 345
 346         if ( (method & RANK_NORM_UNIQ) && t->size > 0 )
 347                 res /= (float)( t->size );
 348
 349         if ( (method & RANK_NORM_LOGUNIQ) && t->size > 0 )
 350                 res /= log((double) (t->size + 1)) / log(2.0);
 351
 352         return res;
 353 }
 354
 355 Datum
 356 rank(PG_FUNCTION_ARGS)
 357 {
 358         ArrayType  *win = (ArrayType *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
 359         tsvector   *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(1));
 360         QUERYTYPE  *query = (QUERYTYPE *) PG_DETOAST_DATUM(PG_GETARG_DATUM(2));
 361         int                     method = DEF_NORM_METHOD;
 362         float           res = 0.0;
 363         float           ws[lengthof(weights)];
 364         float4     *arrdata;
 365         int                     i;
 366
 367         if (ARR_NDIM(win) != 1)
 368                 ereport(ERROR,
 369                                 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
 370                                  errmsg("array of weight must be one-dimensional")));
 371
 372         if (ARRNELEMS(win) < lengthof(weights))
 373                 ereport(ERROR,
 374                                 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
 375                                  errmsg("array of weight is too short")));
 376
 377         if (ARR_HASNULL(win))
 378                 ereport(ERROR,
 379                                 (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
 380                                  errmsg("array of weight must not contain nulls")));
 381
 382         arrdata = (float4 *) ARR_DATA_PTR(win);
 383         for (i = 0; i < lengthof(weights); i++)
 384         {
 385                 ws[i] = (arrdata[i] >= 0) ? arrdata[i] : weights[i];
 386                 if (ws[i] > 1.0)
 387                         ereport(ERROR,
 388                                         (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
 389                                          errmsg("weight out of range")));
 390         }
 391
 392         if (PG_NARGS() == 4)
 393                 method = PG_GETARG_INT32(3);
 394
 395         res = calc_rank(ws, txt, query, method);
 396
 397         PG_FREE_IF_COPY(win, 0);
 398         PG_FREE_IF_COPY(txt, 1);
 399         PG_FREE_IF_COPY(query, 2);
 400         PG_RETURN_FLOAT4(res);
 401 }
 402
 403 Datum
 404 rank_def(PG_FUNCTION_ARGS)
 405 {
 406         tsvector   *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
 407         QUERYTYPE  *query = (QUERYTYPE *) PG_DETOAST_DATUM(PG_GETARG_DATUM(1));
 408         float           res = 0.0;
 409         int                     method = DEF_NORM_METHOD;
 410
 411         if (PG_NARGS() == 3)
 412                 method = PG_GETARG_INT32(2);
 413
 414         res = calc_rank(weights, txt, query, method);
 415
 416         PG_FREE_IF_COPY(txt, 0);
 417         PG_FREE_IF_COPY(query, 1);
 418         PG_RETURN_FLOAT4(res);
 419 }
 420
 421
 422 typedef struct
 423 {
 424         ITEM      **item;
 425         int16           nitem;
 426         bool            needfree;
 427         uint8           wclass;
 428         int32           pos;
 429 }       DocRepresentation;
 430
 431 static int
 432 compareDocR(const void *a, const void *b)
 433 {
 434         if (((DocRepresentation *) a)->pos == ((DocRepresentation *) b)->pos)
 435                 return 0;
 436         return (((DocRepresentation *) a)->pos > ((DocRepresentation *) b)->pos) ? 1 : -1;
 437 }
 438
 439 static bool
 440 checkcondition_ITEM(void *checkval, ITEM * val)
 441 {
 442         return (bool) (val->istrue);
 443 }
 444
 445 static void
 446 reset_istrue_flag(QUERYTYPE * query)
 447 {
 448         ITEM       *item = GETQUERY(query);
 449         int                     i;
 450
 451         /* reset istrue flag */
 452         for (i = 0; i < query->size; i++)
 453         {
 454                 if (item->type == VAL)
 455                         item->istrue = 0;
 456                 item++;
 457         }
 458 }
 459
 460 typedef struct {
 461         int     pos;
 462         int     p;
 463         int     q;
 464         DocRepresentation       *begin;
 465         DocRepresentation       *end;
 466 } Extention;
 467
 468
 469 static bool
 470 Cover(DocRepresentation * doc, int len, QUERYTYPE * query, Extention *ext)
 471 {
 472         DocRepresentation *ptr;
 473         int                     lastpos = ext->pos;
 474         int                     i;
 475         bool            found = false;
 476
 477         reset_istrue_flag(query);
 478
 479         ext->p = 0x7fffffff;
 480         ext->q = 0;
 481         ptr = doc + ext->pos;
 482
 483         /* find upper bound of cover from current position, move up */
 484         while (ptr - doc < len)
 485         {
 486                 for (i = 0; i < ptr->nitem; i++)
 487                         ptr->item[i]->istrue = 1;
 488                 if (TS_execute(GETQUERY(query), NULL, false, checkcondition_ITEM))
 489                 {
 490                         if (ptr->pos > ext->q)
 491                         {
 492                                 ext->q = ptr->pos;
 493                                 ext->end = ptr;
 494                                 lastpos = ptr - doc;
 495                                 found = true;
 496                         }
 497                         break;
 498                 }
 499                 ptr++;
 500         }
 501
 502         if (!found)
 503                 return false;
 504
 505         reset_istrue_flag(query);
 506
 507         ptr = doc + lastpos;
 508
 509         /* find lower bound of cover from founded upper bound, move down */
 510         while (ptr >= doc)
 511         {
 512                 for (i = 0; i < ptr->nitem; i++)
 513                         ptr->item[i]->istrue = 1;
 514                 if (TS_execute(GETQUERY(query), NULL, true, checkcondition_ITEM))
 515                 {
 516                         if (ptr->pos < ext->p) {
 517                                 ext->begin = ptr;
 518                                 ext->p = ptr->pos;
 519                         }
 520                         break;
 521                 }
 522                 ptr--;
 523         }
 524
 525         if (ext->p <= ext->q)
 526         {
 527                 /*
 528                  * set position for next try to next lexeme after begining of founded
 529                  * cover
 530                  */
 531                 ext->pos = (ptr - doc) + 1;
 532                 return true;
 533         }
 534
 535         ext->pos++;
 536         return Cover(doc, len, query, ext);
 537 }
 538
 539 static DocRepresentation *
 540 get_docrep(tsvector * txt, QUERYTYPE * query, int *doclen)
 541 {
 542         ITEM       *item = GETQUERY(query);
 543         WordEntry  *entry;
 544         WordEntryPos *post;
 545         int4            dimt,
 546                                 j,
 547                                 i;
 548         int                     len = query->size * 4,
 549                                 cur = 0;
 550         DocRepresentation *doc;
 551
 552         *(uint16 *) POSNULL = lengthof(POSNULL) - 1;
 553         doc = (DocRepresentation *) palloc(sizeof(DocRepresentation) * len);
 554         SortAndUniqOperand = GETOPERAND(query);
 555         reset_istrue_flag(query);
 556
 557         for (i = 0; i < query->size; i++)
 558         {
 559                 if (item[i].type != VAL || item[i].istrue)
 560                         continue;
 561
 562                 entry = find_wordentry(txt, query, &(item[i]));
 563                 if (!entry)
 564                         continue;
 565
 566                 if (entry->haspos)
 567                 {
 568                         dimt = POSDATALEN(txt, entry);
 569                         post = POSDATAPTR(txt, entry);
 570                 }
 571                 else
 572                 {
 573                         dimt = *(uint16 *) POSNULL;
 574                         post = POSNULL + 1;
 575                 }
 576
 577                 while (cur + dimt >= len)
 578                 {
 579                         len *= 2;
 580                         doc = (DocRepresentation *) repalloc(doc, sizeof(DocRepresentation) * len);
 581                 }
 582
 583                 for (j = 0; j < dimt; j++)
 584                 {
 585                         if (j == 0)
 586                         {
 587                                 ITEM       *kptr,
 588                                                    *iptr = item + i;
 589                                 int                     k;
 590
 591                                 doc[cur].needfree = false;
 592                                 doc[cur].nitem = 0;
 593                                 doc[cur].item = (ITEM **) palloc(sizeof(ITEM *) * query->size);
 594
 595                                 for (k = 0; k < query->size; k++)
 596                                 {
 597                                         kptr = item + k;
 598                                         if (k == i || (item[k].type == VAL && compareITEM(&kptr, &iptr) == 0))
 599                                         {
 600                                                 doc[cur].item[doc[cur].nitem] = item + k;
 601                                                 doc[cur].nitem++;
 602                                                 kptr->istrue = 1;
 603                                         }
 604                                 }
 605                         }
 606                         else
 607                         {
 608                                 doc[cur].needfree = false;
 609                                 doc[cur].nitem = doc[cur - 1].nitem;
 610                                 doc[cur].item = doc[cur - 1].item;
 611                         }
 612                         doc[cur].pos = WEP_GETPOS(post[j]);
 613                         doc[cur].wclass = WEP_GETWEIGHT(post[j]);
 614                         cur++;
 615                 }
 616         }
 617
 618         *doclen = cur;
 619
 620         if (cur > 0)
 621         {
 622                 if (cur > 1)
 623                         qsort((void *) doc, cur, sizeof(DocRepresentation), compareDocR);
 624                 return doc;
 625         }
 626
 627         pfree(doc);
 628         return NULL;
 629 }
 630
 631 static float4
 632 calc_rank_cd(float4 *arrdata, tsvector *txt, QUERYTYPE *query, int method) {
 633         DocRepresentation *doc;
 634         int                     len,
 635                                 i,
 636                                 doclen = 0;
 637         Extention       ext;
 638         double          Wdoc = 0.0;
 639         double          invws[lengthof(weights)];
 640         double          SumDist=0.0, PrevExtPos=0.0, CurExtPos=0.0;
 641         int             NExtent=0;
 642
 643         for (i = 0; i < lengthof(weights); i++)
 644         {
 645                 invws[i] = ((double)((arrdata[i] >= 0) ? arrdata[i] : weights[i]));
 646                 if (invws[i] > 1.0)
 647                         ereport(ERROR,
 648                                         (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
 649                                          errmsg("weight out of range")));
 650                 invws[i] = 1.0/invws[i];
 651         }
 652
 653         doc = get_docrep(txt, query, &doclen);
 654         if (!doc)
 655                 return 0.0;
 656
 657         MemSet( &ext, 0, sizeof(Extention) );
 658         while (Cover(doc, doclen, query, &ext)) {
 659                 double  Cpos = 0.0;
 660                 double  InvSum = 0.0;
 661                 DocRepresentation *ptr = ext.begin;
 662
 663                 while ( ptr<=ext.end ) {
 664                         InvSum += invws[ ptr->wclass ];
 665                         ptr++;
 666                 }
 667
 668                 Cpos = ((double)( ext.end-ext.begin+1 )) / InvSum;
 669                 Wdoc += Cpos / ( (double)(( 1 + (ext.q - ext.p) - (ext.end - ext.begin) )) );
 670
 671                 CurExtPos = ((double)(ext.q + ext.p))/2.0;
 672                 if ( NExtent>0 && CurExtPos > PrevExtPos /* prevent devision by zero in a case of multiple lexize */ )
 673                         SumDist += 1.0/( CurExtPos - PrevExtPos );
 674
 675                 PrevExtPos = CurExtPos;
 676                 NExtent++;
 677         }
 678
 679         if ( (method & RANK_NORM_LOGLENGTH) && txt->size > 0 )
 680                 Wdoc /= log((double) (cnt_length(txt) + 1));
 681
 682         if ( method & RANK_NORM_LENGTH ) {
 683                 len = cnt_length(txt);
 684                 if ( len>0 )
 685                         Wdoc /= (double) len;
 686         }
 687
 688         if ( (method & RANK_NORM_EXTDIST) && SumDist > 0 )
 689                 Wdoc /= ((double)NExtent) / SumDist;
 690
 691         if ( (method & RANK_NORM_UNIQ) && txt->size > 0 )
 692                 Wdoc /= (double)( txt->size );
 693
 694         if ( (method & RANK_NORM_LOGUNIQ) && txt->size > 0 )
 695                 Wdoc /= log((double) (txt->size + 1)) / log(2.0);
 696
 697         for (i = 0; i < doclen; i++)
 698                 if (doc[i].needfree)
 699                         pfree(doc[i].item);
 700         pfree(doc);
 701
 702         return (float4)Wdoc;
 703 }
 704
 705 Datum
 706 rank_cd(PG_FUNCTION_ARGS)
 707 {
 708         ArrayType *win =  (ArrayType *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
 709         tsvector   *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(1));
 710         QUERYTYPE  *query = (QUERYTYPE *) PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(2));
 711         int                     method = DEF_NORM_METHOD;
 712         float4          res;
 713
 714         if (ARR_NDIM(win) != 1)
 715                 ereport(ERROR,
 716                                 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
 717                                  errmsg("array of weight must be one-dimensional")));
 718
 719         if (ARRNELEMS(win) < lengthof(weights))
 720                 ereport(ERROR,
 721                                 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
 722                                  errmsg("array of weight is too short")));
 723
 724         if (ARR_HASNULL(win))
 725                 ereport(ERROR,
 726                                 (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
 727                                  errmsg("array of weight must not contain nulls")));
 728
 729         if (PG_NARGS() == 4)
 730                 method = PG_GETARG_INT32(3);
 731
 732         res = calc_rank_cd( (float4 *) ARR_DATA_PTR(win), txt, query, method);
 733
 734         PG_FREE_IF_COPY(win, 0);
 735         PG_FREE_IF_COPY(txt, 1);
 736         PG_FREE_IF_COPY(query, 2);
 737
 738         PG_RETURN_FLOAT4(res);
 739 }
 740
 741
 742 Datum
 743 rank_cd_def(PG_FUNCTION_ARGS)
 744 {
 745         tsvector   *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
 746         QUERYTYPE  *query = (QUERYTYPE *) PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(1));
 747         float4 res;
 748
 749         res = calc_rank_cd( weights, txt, query, (PG_NARGS() == 3) ? PG_GETARG_DATUM(2) : DEF_NORM_METHOD);
 750
 751         PG_FREE_IF_COPY(txt, 0);
 752         PG_FREE_IF_COPY(query, 1);
 753
 754         PG_RETURN_FLOAT4(res);
 755 }
 756
 757 /**************debug*************/
 758
 759 typedef struct
 760 {
 761         char       *w;
 762         int2            len;
 763         int2            pos;
 764         int2            start;
 765         int2            finish;
 766 }       DocWord;
 767
 768 static int
 769 compareDocWord(const void *a, const void *b)
 770 {
 771         if (((DocWord *) a)->pos == ((DocWord *) b)->pos)
 772                 return 0;
 773         return (((DocWord *) a)->pos > ((DocWord *) b)->pos) ? 1 : -1;
 774 }
 775
 776
 777 Datum
 778 get_covers(PG_FUNCTION_ARGS)
 779 {
 780         tsvector   *txt = (tsvector *) PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
 781         QUERYTYPE  *query = (QUERYTYPE *) PG_DETOAST_DATUM_COPY(PG_GETARG_DATUM(1));
 782         WordEntry  *pptr = ARRPTR(txt);
 783         int                     i,
 784                                 dlen = 0,
 785                                 j,
 786                                 cur = 0,
 787                                 len = 0,
 788                                 rlen;
 789         DocWord    *dw,
 790                            *dwptr;
 791         text       *out;
 792         char       *cptr;
 793         DocRepresentation *doc;
 794         int                     olddwpos = 0;
 795         int                     ncover = 1;
 796         Extention       ext;
 797
 798         doc = get_docrep(txt, query, &rlen);
 799
 800         if (!doc)
 801         {
 802                 out = palloc(VARHDRSZ);
 803                 VARATT_SIZEP(out) = VARHDRSZ;
 804                 PG_FREE_IF_COPY(txt, 0);
 805                 PG_FREE_IF_COPY(query, 1);
 806                 PG_RETURN_POINTER(out);
 807         }
 808
 809         for (i = 0; i < txt->size; i++)
 810         {
 811                 if (!pptr[i].haspos)
 812                         ereport(ERROR,
 813                                         (errcode(ERRCODE_SYNTAX_ERROR),
 814                                          errmsg("no pos info")));
 815                 dlen += POSDATALEN(txt, &(pptr[i]));
 816         }
 817
 818         dwptr = dw = palloc(sizeof(DocWord) * dlen);
 819         memset(dw, 0, sizeof(DocWord) * dlen);
 820
 821         for (i = 0; i < txt->size; i++)
 822         {
 823                 WordEntryPos *posdata = POSDATAPTR(txt, &(pptr[i]));
 824
 825                 for (j = 0; j < POSDATALEN(txt, &(pptr[i])); j++)
 826                 {
 827                         dw[cur].w = STRPTR(txt) + pptr[i].pos;
 828                         dw[cur].len = pptr[i].len;
 829                         dw[cur].pos = WEP_GETPOS(posdata[j]);
 830                         cur++;
 831                 }
 832                 len += (pptr[i].len + 1) * (int) POSDATALEN(txt, &(pptr[i]));
 833         }
 834         qsort((void *) dw, dlen, sizeof(DocWord), compareDocWord);
 835
 836         MemSet( &ext, 0, sizeof(Extention) );
 837         while (Cover(doc, rlen, query, &ext))
 838         {
 839                 dwptr = dw + olddwpos;
 840                 while (dwptr->pos < ext.p && dwptr - dw < dlen)
 841                         dwptr++;
 842                 olddwpos = dwptr - dw;
 843                 dwptr->start = ncover;
 844                 while (dwptr->pos < ext.q + 1 && dwptr - dw < dlen)
 845                         dwptr++;
 846                 (dwptr - 1)->finish = ncover;
 847                 len += 4 /* {}+two spaces */ + 2 * 16 /* numbers */ ;
 848                 ncover++;
 849         }
 850
 851         out = palloc(VARHDRSZ + len);
 852         cptr = ((char *) out) + VARHDRSZ;
 853         dwptr = dw;
 854
 855         while (dwptr - dw < dlen)
 856         {
 857                 if (dwptr->start)
 858                 {
 859                         sprintf(cptr, "{%d ", dwptr->start);
 860                         cptr = strchr(cptr, '\0');
 861                 }
 862                 memcpy(cptr, dwptr->w, dwptr->len);
 863                 cptr += dwptr->len;
 864                 *cptr = ' ';
 865                 cptr++;
 866                 if (dwptr->finish)
 867                 {
 868                         sprintf(cptr, "}%d ", dwptr->finish);
 869                         cptr = strchr(cptr, '\0');
 870                 }
 871                 dwptr++;
 872         }
 873
 874         VARATT_SIZEP(out) = cptr - ((char *) out);
 875
 876         pfree(dw);
 877         for (i = 0; i < rlen; i++)
 878                 if (doc[i].needfree)
 879                         pfree(doc[i].item);
 880         pfree(doc);
 881
 882         PG_FREE_IF_COPY(txt, 0);
 883         PG_FREE_IF_COPY(query, 1);
 884         PG_RETURN_POINTER(out);
 885 }