smlar.c

   1 #include "smlar.h"
   2
   3 #include "fmgr.h"
   4 #include "access/genam.h"
   5 #include "access/heapam.h"
   6 #include "access/htup_details.h"
   7 #include "access/nbtree.h"
   8 #include "catalog/indexing.h"
   9 #include "catalog/pg_am.h"
  10 #include "catalog/pg_amproc.h"
  11 #include "catalog/pg_cast.h"
  12 #include "catalog/pg_opclass.h"
  13 #include "catalog/pg_type.h"
  14 #include "executor/spi.h"
  15 #include "utils/catcache.h"
  16 #include "utils/fmgroids.h"
  17 #include "utils/lsyscache.h"
  18 #include "utils/memutils.h"
  19 #if (PG_VERSION_NUM < 120000)
  20 #include "utils/tqual.h"
  21 #endif
  22 #include "utils/syscache.h"
  23 #include "utils/typcache.h"
  24
  25 PG_MODULE_MAGIC;
  26
  27 #if (PG_VERSION_NUM >= 90400)
  28 #define SNAPSHOT NULL
  29 #else
  30 #define SNAPSHOT SnapshotNow
  31 #endif
  32
  33 #if PG_VERSION_NUM >= 130000
  34 #define heap_open(r, l)                 table_open((r), (l))
  35 #define heap_close(r, l)                table_close((r), (l))
  36 #endif
  37
  38 static Oid
  39 getDefaultOpclass(Oid amoid, Oid typid)
  40 {
  41         ScanKeyData     skey;
  42         SysScanDesc     scan;
  43         HeapTuple       tuple;
  44         Relation        heapRel;
  45         Oid                     opclassOid = InvalidOid;
  46
  47         heapRel = heap_open(OperatorClassRelationId, AccessShareLock);
  48
  49         ScanKeyInit(&skey,
  50                                 Anum_pg_opclass_opcmethod,
  51                                 BTEqualStrategyNumber,  F_OIDEQ,
  52                                 ObjectIdGetDatum(amoid));
  53
  54         scan = systable_beginscan(heapRel,
  55                                                                 OpclassAmNameNspIndexId, true,
  56                                                                 SNAPSHOT, 1, &skey);
  57
  58         while (HeapTupleIsValid((tuple = systable_getnext(scan))))
  59         {
  60                 Form_pg_opclass opclass = (Form_pg_opclass)GETSTRUCT(tuple);
  61
  62                 if ( opclass->opcintype == typid && opclass->opcdefault )
  63                 {
  64                         if ( OidIsValid(opclassOid) )
  65                                 elog(ERROR, "Ambiguous opclass for type %u (access method %u)", typid, amoid);
  66 #if (PG_VERSION_NUM >= 120000)
  67                         opclassOid = opclass->oid;
  68 #else
  69                         opclassOid = HeapTupleGetOid(tuple);
  70 #endif
  71                 }
  72         }
  73
  74         systable_endscan(scan);
  75         heap_close(heapRel, AccessShareLock);
  76
  77         return opclassOid;
  78 }
  79
  80 static Oid
  81 getAMProc(Oid amoid, Oid typid)
  82 {
  83         Oid             opclassOid = getDefaultOpclass(amoid, typid);
  84         Oid             procOid = InvalidOid;
  85         Oid             opfamilyOid;
  86         ScanKeyData     skey[4];
  87         SysScanDesc     scan;
  88         HeapTuple       tuple;
  89         Relation        heapRel;
  90
  91         if ( !OidIsValid(opclassOid) )
  92         {
  93                 typid = getBaseType(typid);
  94                 opclassOid = getDefaultOpclass(amoid, typid);
  95         }
  96
  97         if ( !OidIsValid(opclassOid) )
  98         {
  99                 CatCList        *catlist;
 100                 int                     i;
 101
 102                 /*
 103                  * Search binary-coercible type
 104                  */
 105 #ifdef SearchSysCacheList1
 106                 catlist = SearchSysCacheList1(CASTSOURCETARGET,
 107                                                                           ObjectIdGetDatum(typid));
 108 #else
 109                 catlist = SearchSysCacheList(CASTSOURCETARGET, 1,
 110                                                                                 ObjectIdGetDatum(typid),
 111                                                                                 0, 0, 0);
 112 #endif
 113
 114                 for (i = 0; i < catlist->n_members; i++)
 115                 {
 116                         HeapTuple               tuple = &catlist->members[i]->tuple;
 117                         Form_pg_cast    castForm = (Form_pg_cast)GETSTRUCT(tuple);
 118
 119                         if ( castForm->castfunc == InvalidOid && castForm->castcontext == COERCION_CODE_IMPLICIT )
 120                         {
 121                                 typid = castForm->casttarget;
 122                                 opclassOid = getDefaultOpclass(amoid, typid);
 123                                 if( OidIsValid(opclassOid) )
 124                                         break;
 125                         }
 126                 }
 127
 128                 ReleaseSysCacheList(catlist);
 129         }
 130
 131         if ( !OidIsValid(opclassOid) )
 132                 return InvalidOid;
 133
 134         opfamilyOid = get_opclass_family(opclassOid);
 135
 136         heapRel = heap_open(AccessMethodProcedureRelationId, AccessShareLock);
 137         ScanKeyInit(&skey[0],
 138                                 Anum_pg_amproc_amprocfamily,
 139                                 BTEqualStrategyNumber, F_OIDEQ,
 140                                 ObjectIdGetDatum(opfamilyOid));
 141         ScanKeyInit(&skey[1],
 142                                 Anum_pg_amproc_amproclefttype,
 143                                 BTEqualStrategyNumber, F_OIDEQ,
 144                                 ObjectIdGetDatum(typid));
 145         ScanKeyInit(&skey[2],
 146                                 Anum_pg_amproc_amprocrighttype,
 147                                 BTEqualStrategyNumber, F_OIDEQ,
 148                                 ObjectIdGetDatum(typid));
 149 #if PG_VERSION_NUM >= 90200
 150         ScanKeyInit(&skey[3],
 151                                 Anum_pg_amproc_amprocnum,
 152                                 BTEqualStrategyNumber, F_OIDEQ,
 153                                 Int32GetDatum(BTORDER_PROC));
 154 #endif
 155
 156         scan = systable_beginscan(heapRel, AccessMethodProcedureIndexId, true,
 157                                                                 SNAPSHOT,
 158 #if PG_VERSION_NUM >= 90200
 159                                                                 4,
 160 #else
 161                                                                 3,
 162 #endif
 163                                                                 skey);
 164         while (HeapTupleIsValid(tuple = systable_getnext(scan)))
 165         {
 166                 Form_pg_amproc amprocform = (Form_pg_amproc) GETSTRUCT(tuple);
 167
 168                 switch(amoid)
 169                 {
 170                         case BTREE_AM_OID:
 171                         case HASH_AM_OID:
 172                                 if ( OidIsValid(procOid) )
 173                                         elog(ERROR,"Ambiguous support function for type %u (opclass %u)", typid, opfamilyOid);
 174                                 procOid = amprocform->amproc;
 175                                 break;
 176                         default:
 177                                 elog(ERROR,"Unsupported access method");
 178                 }
 179         }
 180
 181         systable_endscan(scan);
 182         heap_close(heapRel, AccessShareLock);
 183
 184         return procOid;
 185 }
 186
 187 static ProcTypeInfo *cacheProcs = NULL;
 188 static int nCacheProcs = 0;
 189
 190 #ifndef TupleDescAttr
 191 #define TupleDescAttr(tupdesc, i)       ((tupdesc)->attrs[(i)])
 192 #endif
 193
 194 static ProcTypeInfo
 195 fillProcs(Oid typid)
 196 {
 197         ProcTypeInfo    info = malloc(sizeof(ProcTypeInfoData));
 198
 199         if (!info)
 200                 elog(ERROR, "Can't allocate %u memory", (uint32)sizeof(ProcTypeInfoData));
 201
 202         info->typid = typid;
 203         info->typtype = get_typtype(typid);
 204
 205         if (info->typtype == 'c')
 206         {
 207                 /* composite type */
 208                 TupleDesc               tupdesc;
 209                 MemoryContext   oldcontext;
 210
 211                 tupdesc = lookup_rowtype_tupdesc(typid, -1);
 212
 213                 if (tupdesc->natts != 2)
 214                         elog(ERROR,"Composite type has wrong number of fields");
 215                 if (TupleDescAttr(tupdesc, 1)->atttypid != FLOAT4OID)
 216                         elog(ERROR,"Second field of composite type is not float4");
 217
 218                 oldcontext = MemoryContextSwitchTo(TopMemoryContext);
 219                 info->tupDesc = CreateTupleDescCopyConstr(tupdesc);
 220                 MemoryContextSwitchTo(oldcontext);
 221
 222                 ReleaseTupleDesc(tupdesc);
 223
 224                 info->cmpFuncOid = getAMProc(BTREE_AM_OID,
 225                                                                          TupleDescAttr(info->tupDesc, 0)->atttypid);
 226                 info->hashFuncOid = getAMProc(HASH_AM_OID,
 227                                                                           TupleDescAttr(info->tupDesc, 0)->atttypid);
 228         }
 229         else
 230         {
 231                 info->tupDesc = NULL;
 232
 233                 /* plain type */
 234                 info->cmpFuncOid = getAMProc(BTREE_AM_OID, typid);
 235                 info->hashFuncOid = getAMProc(HASH_AM_OID, typid);
 236         }
 237
 238         get_typlenbyvalalign(typid, &info->typlen, &info->typbyval, &info->typalign);
 239         info->hashFuncInited = info->cmpFuncInited = false;
 240
 241
 242         return info;
 243 }
 244
 245 void
 246 getFmgrInfoCmp(ProcTypeInfo info)
 247 {
 248         if ( info->cmpFuncInited == false )
 249         {
 250                 if ( !OidIsValid(info->cmpFuncOid) )
 251                         elog(ERROR, "Could not find cmp function for type %u", info->typid);
 252
 253                 fmgr_info_cxt( info->cmpFuncOid, &info->cmpFunc, TopMemoryContext );
 254                 info->cmpFuncInited = true;
 255         }
 256 }
 257
 258 void
 259 getFmgrInfoHash(ProcTypeInfo info)
 260 {
 261         if ( info->hashFuncInited == false )
 262         {
 263                 if ( !OidIsValid(info->hashFuncOid) )
 264                         elog(ERROR, "Could not find hash function for type %u", info->typid);
 265
 266                 fmgr_info_cxt( info->hashFuncOid, &info->hashFunc, TopMemoryContext );
 267                 info->hashFuncInited = true;
 268         }
 269 }
 270
 271 static int
 272 cmpProcTypeInfo(const void *a, const void *b)
 273 {
 274         ProcTypeInfo av = *(ProcTypeInfo*)a;
 275         ProcTypeInfo bv = *(ProcTypeInfo*)b;
 276
 277         Assert( av->typid != bv->typid );
 278
 279         return ( av->typid > bv->typid ) ? 1 : -1;
 280 }
 281
 282 ProcTypeInfo
 283 findProcs(Oid typid)
 284 {
 285         ProcTypeInfo    info = NULL;
 286
 287         if ( nCacheProcs == 1 )
 288         {
 289                 if ( cacheProcs[0]->typid == typid )
 290                 {
 291                         /*cacheProcs[0]->hashFuncInited = cacheProcs[0]->cmpFuncInited = false;*/
 292                         return cacheProcs[0];
 293                 }
 294         }
 295         else if ( nCacheProcs > 1 )
 296         {
 297                 ProcTypeInfo    *StopMiddle;
 298                 ProcTypeInfo    *StopLow = cacheProcs,
 299                                                 *StopHigh = cacheProcs + nCacheProcs;
 300
 301                 while (StopLow < StopHigh) {
 302                         StopMiddle = StopLow + ((StopHigh - StopLow) >> 1);
 303                         info = *StopMiddle;
 304
 305                         if ( info->typid == typid )
 306                         {
 307                                 /* info->hashFuncInited = info->cmpFuncInited = false; */
 308                                 return info;
 309                         }
 310                         else if ( info->typid < typid )
 311                                 StopLow = StopMiddle + 1;
 312                         else
 313                                 StopHigh = StopMiddle;
 314                 }
 315
 316                 /* not found */
 317         }
 318
 319         info = fillProcs(typid);
 320         if ( nCacheProcs == 0 )
 321         {
 322                 cacheProcs = malloc(sizeof(ProcTypeInfo));
 323
 324                 if (!cacheProcs)
 325                         elog(ERROR, "Can't allocate %u memory", (uint32)sizeof(ProcTypeInfo));
 326                 else
 327                 {
 328                         nCacheProcs = 1;
 329                         cacheProcs[0] = info;
 330                 }
 331         }
 332         else
 333         {
 334                 ProcTypeInfo    *cacheProcsTmp = realloc(cacheProcs, (nCacheProcs+1) * sizeof(ProcTypeInfo));
 335
 336                 if (!cacheProcsTmp)
 337                         elog(ERROR, "Can't allocate %u memory", (uint32)sizeof(ProcTypeInfo) * (nCacheProcs+1));
 338                 else
 339                 {
 340                         cacheProcs = cacheProcsTmp;
 341                         cacheProcs[ nCacheProcs ] = info;
 342                         nCacheProcs++;
 343                         qsort(cacheProcs, nCacheProcs, sizeof(ProcTypeInfo), cmpProcTypeInfo);
 344                 }
 345         }
 346
 347         /* info->hashFuncInited = info->cmpFuncInited = false; */
 348
 349         return info;
 350 }
 351
 352 /*
 353  * WARNING. Array2SimpleArray* doesn't copy Datum!
 354  */
 355 SimpleArray *
 356 Array2SimpleArray(ProcTypeInfo info, ArrayType *a)
 357 {
 358         SimpleArray     *s = palloc(sizeof(SimpleArray));
 359
 360         CHECKARRVALID(a);
 361
 362         if ( info == NULL )
 363                 info = findProcs(ARR_ELEMTYPE(a));
 364
 365         s->info = info;
 366         s->df = NULL;
 367         s->hash = NULL;
 368
 369         deconstruct_array(a, info->typid,
 370                                                 info->typlen, info->typbyval, info->typalign,
 371                                                 &s->elems, NULL, &s->nelems);
 372
 373         return s;
 374 }
 375
 376 static Datum
 377 deconstructCompositeType(ProcTypeInfo info, Datum in, double *weight)
 378 {
 379         HeapTupleHeader rec = DatumGetHeapTupleHeader(in);
 380         HeapTupleData   tuple;
 381         Datum                   values[2];
 382         bool                    nulls[2];
 383
 384         /* Build a temporary HeapTuple control structure */
 385         tuple.t_len = HeapTupleHeaderGetDatumLength(rec);
 386         ItemPointerSetInvalid(&(tuple.t_self));
 387         tuple.t_tableOid = InvalidOid;
 388         tuple.t_data = rec;
 389
 390         heap_deform_tuple(&tuple, info->tupDesc, values, nulls);
 391         if (nulls[0] || nulls[1])
 392                 elog(ERROR, "Both fields in composite type could not be NULL");
 393
 394         if (weight)
 395                 *weight = DatumGetFloat4(values[1]);
 396         return values[0];
 397 }
 398
 399 static int
 400 cmpArrayElem(const void *a, const void *b, void *arg)
 401 {
 402         ProcTypeInfo    info = (ProcTypeInfo)arg;
 403
 404         if (info->tupDesc)
 405                 /* composite type */
 406                 return DatumGetInt32( FCall2( &info->cmpFunc,
 407                                                 deconstructCompositeType(info, *(Datum*)a, NULL),
 408                                                 deconstructCompositeType(info, *(Datum*)b, NULL) ) );
 409
 410         return DatumGetInt32( FCall2( &info->cmpFunc,
 411                                                         *(Datum*)a, *(Datum*)b ) );
 412 }
 413
 414 SimpleArray *
 415 Array2SimpleArrayS(ProcTypeInfo info, ArrayType *a)
 416 {
 417         SimpleArray     *s = Array2SimpleArray(info, a);
 418
 419         if ( s->nelems > 1 )
 420         {
 421                 getFmgrInfoCmp(s->info);
 422
 423                 qsort_arg(s->elems, s->nelems, sizeof(Datum), cmpArrayElem, s->info);
 424         }
 425
 426         return s;
 427 }
 428
 429 typedef struct cmpArrayElemData {
 430         ProcTypeInfo    info;
 431         bool                    hasDuplicate;
 432
 433 } cmpArrayElemData;
 434
 435 static int
 436 cmpArrayElemArg(const void *a, const void *b, void *arg)
 437 {
 438         cmpArrayElemData        *data = (cmpArrayElemData*)arg;
 439         int                                     res;
 440
 441         if (data->info->tupDesc)
 442                 res =  DatumGetInt32( FCall2( &data->info->cmpFunc,
 443                                         deconstructCompositeType(data->info, *(Datum*)a, NULL),
 444                                         deconstructCompositeType(data->info, *(Datum*)b, NULL) ) );
 445         else
 446                 res = DatumGetInt32( FCall2( &data->info->cmpFunc,
 447                                                                 *(Datum*)a, *(Datum*)b ) );
 448
 449         if ( res == 0 )
 450                 data->hasDuplicate = true;
 451
 452         return res;
 453 }
 454
 455 /*
 456  * Uniquefy array and calculate TF. Although
 457  * result doesn't depend on normalization, we
 458  * normalize TF by length array to have possiblity
 459  * to limit estimation for index support.
 460  *
 461  * Cache signals of needing of TF caclulation
 462  */
 463
 464 SimpleArray *
 465 Array2SimpleArrayU(ProcTypeInfo info, ArrayType *a, void *cache)
 466 {
 467         SimpleArray     *s = Array2SimpleArray(info, a);
 468         StatElem        *stat = NULL;
 469
 470         if ( s->nelems > 0 && cache )
 471         {
 472                 s->df = palloc(sizeof(double) * s->nelems);
 473                 s->df[0] = 1.0; /* init */
 474         }
 475
 476         if ( s->nelems > 1 )
 477         {
 478                 cmpArrayElemData        data;
 479                 int                                     i;
 480
 481                 getFmgrInfoCmp(s->info);
 482                 data.info = s->info;
 483                 data.hasDuplicate = false;
 484
 485                 qsort_arg(s->elems, s->nelems, sizeof(Datum), cmpArrayElemArg, &data);
 486
 487                 if ( data.hasDuplicate )
 488                 {
 489                         Datum   *tmp,
 490                                         *dr,
 491                                         *data;
 492                         int             num = s->nelems,
 493                                         cmp;
 494
 495                         data = tmp = dr = s->elems;
 496
 497                         while (tmp - data < num)
 498                         {
 499                                 cmp = (tmp == dr) ? 0 : cmpArrayElem(tmp, dr, s->info);
 500                                 if ( cmp != 0 )
 501                                 {
 502                                         *(++dr) = *tmp++;
 503                                         if ( cache )
 504                                                 s->df[ dr - data ] = 1.0;
 505                                 }
 506                                 else
 507                                 {
 508                                         if ( cache )
 509                                                 s->df[ dr - data ] += 1.0;
 510                                         tmp++;
 511                                 }
 512                         }
 513
 514                         s->nelems = dr + 1 - s->elems;
 515
 516                         if ( cache )
 517                         {
 518                                 int tfm = getTFMethod();
 519
 520                                 for(i=0;i<s->nelems;i++)
 521                                 {
 522                                         stat = fingArrayStat(cache, s->info->typid, s->elems[i], stat);
 523                                         if ( stat )
 524                                         {
 525                                                 switch(tfm)
 526                                                 {
 527                                                         case TF_LOG:
 528                                                                 s->df[i] = (1.0 + log( s->df[i] ));
 529                                                         case TF_N:
 530                                                                 s->df[i] *= stat->idf;
 531                                                                 break;
 532                                                         case TF_CONST:
 533                                                                 s->df[i] = stat->idf;
 534                                                                 break;
 535                                                         default:
 536                                                                 elog(ERROR,"Unknown TF method: %d", tfm);
 537                                                 }
 538                                         }
 539                                         else
 540                                         {
 541                                                 s->df[i] = 0.0; /* unknown word */
 542                                         }
 543                                 }
 544                         }
 545                 }
 546                 else if ( cache )
 547                 {
 548                         for(i=0;i<s->nelems;i++)
 549                         {
 550                                 stat = fingArrayStat(cache, s->info->typid, s->elems[i], stat);
 551                                 if ( stat )
 552                                         s->df[i] = stat->idf;
 553                                 else
 554                                         s->df[i] = 0.0;
 555                         }
 556                 }
 557         }
 558         else if (s->nelems > 0 && cache)
 559         {
 560                 stat = fingArrayStat(cache, s->info->typid, s->elems[0], stat);
 561                 if ( stat )
 562                         s->df[0] = stat->idf;
 563                 else
 564                         s->df[0] = 0.0;
 565         }
 566
 567         return s;
 568 }
 569
 570 static int
 571 numOfIntersect(SimpleArray *a, SimpleArray *b)
 572 {
 573         int                             cnt = 0,
 574                                         cmp;
 575         Datum                   *aptr = a->elems,
 576                                         *bptr = b->elems;
 577         ProcTypeInfo    info = a->info;
 578
 579         Assert( a->info->typid == b->info->typid );
 580
 581         getFmgrInfoCmp(info);
 582
 583         while( aptr - a->elems < a->nelems && bptr - b->elems < b->nelems )
 584         {
 585                 cmp = cmpArrayElem(aptr, bptr, info);
 586                 if ( cmp < 0 )
 587                         aptr++;
 588                 else if ( cmp > 0 )
 589                         bptr++;
 590                 else
 591                 {
 592                         cnt++;
 593                         aptr++;
 594                         bptr++;
 595                 }
 596         }
 597
 598         return cnt;
 599 }
 600
 601 static double
 602 TFIDFSml(SimpleArray *a, SimpleArray *b)
 603 {
 604         int                             cmp;
 605         Datum                   *aptr = a->elems,
 606                                         *bptr = b->elems;
 607         ProcTypeInfo    info = a->info;
 608         double                  res = 0.0;
 609         double                  suma = 0.0, sumb = 0.0;
 610
 611         Assert( a->info->typid == b->info->typid );
 612         Assert( a->df );
 613         Assert( b->df );
 614
 615         getFmgrInfoCmp(info);
 616
 617         while( aptr - a->elems < a->nelems && bptr - b->elems < b->nelems )
 618         {
 619                 cmp = cmpArrayElem(aptr, bptr, info);
 620                 if ( cmp < 0 )
 621                 {
 622                         suma += a->df[ aptr - a->elems ] * a->df[ aptr - a->elems ];
 623                         aptr++;
 624                 }
 625                 else if ( cmp > 0 )
 626                 {
 627                         sumb += b->df[ bptr - b->elems ] * b->df[ bptr - b->elems ];
 628                         bptr++;
 629                 }
 630                 else
 631                 {
 632                         res += a->df[ aptr - a->elems ] * b->df[ bptr - b->elems ];
 633                         suma += a->df[ aptr - a->elems ] * a->df[ aptr - a->elems ];
 634                         sumb += b->df[ bptr - b->elems ] * b->df[ bptr - b->elems ];
 635                         aptr++;
 636                         bptr++;
 637                 }
 638         }
 639
 640         /*
 641          * Compute last elements
 642          */
 643         while( aptr - a->elems < a->nelems )
 644         {
 645                 suma += a->df[ aptr - a->elems ] * a->df[ aptr - a->elems ];
 646                 aptr++;
 647         }
 648
 649         while( bptr - b->elems < b->nelems )
 650         {
 651                 sumb += b->df[ bptr - b->elems ] * b->df[ bptr - b->elems ];
 652                 bptr++;
 653         }
 654
 655         if ( suma > 0.0 && sumb > 0.0 )
 656                 res = res / sqrt( suma * sumb );
 657         else
 658                 res = 0.0;
 659
 660         return res;
 661 }
 662
 663
 664 PG_FUNCTION_INFO_V1(arraysml);
 665 Datum   arraysml(PG_FUNCTION_ARGS);
 666 Datum
 667 arraysml(PG_FUNCTION_ARGS)
 668 {
 669         ArrayType               *a, *b;
 670         SimpleArray             *sa, *sb;
 671
 672         fcinfo->flinfo->fn_extra = SearchArrayCache(
 673                                                         fcinfo->flinfo->fn_extra,
 674                                                         fcinfo->flinfo->fn_mcxt,
 675                                                         PG_GETARG_DATUM(0), &a, &sa, NULL);
 676         fcinfo->flinfo->fn_extra = SearchArrayCache(
 677                                                         fcinfo->flinfo->fn_extra,
 678                                                         fcinfo->flinfo->fn_mcxt,
 679                                                         PG_GETARG_DATUM(1), &b, &sb, NULL);
 680
 681         if ( ARR_ELEMTYPE(a) != ARR_ELEMTYPE(b) )
 682                 elog(ERROR,"Arguments array are not the same type!");
 683
 684         if (ARRISVOID(a) || ARRISVOID(b))
 685                  PG_RETURN_FLOAT4(0.0);
 686
 687         switch(getSmlType())
 688         {
 689                 case ST_TFIDF:
 690                         PG_RETURN_FLOAT4( TFIDFSml(sa, sb) );
 691                         break;
 692                 case ST_COSINE:
 693                         {
 694                                 int                             cnt;
 695                                 double                  power;
 696
 697                                 power = ((double)(sa->nelems)) * ((double)(sb->nelems));
 698                                 cnt = numOfIntersect(sa, sb);
 699
 700                                 PG_RETURN_FLOAT4(  ((double)cnt) / sqrt( power ) );
 701                         }
 702                         break;
 703                 case ST_OVERLAP:
 704                         {
 705                                 float4 res = (float4)numOfIntersect(sa, sb);
 706
 707                                 PG_RETURN_FLOAT4(res);
 708                         }
 709                         break;
 710                 default:
 711                         elog(ERROR,"Unsupported formula type of similarity");
 712         }
 713
 714         PG_RETURN_FLOAT4(0.0); /* keep compiler quiet */
 715 }
 716
 717 PG_FUNCTION_INFO_V1(arraysmlw);
 718 Datum   arraysmlw(PG_FUNCTION_ARGS);
 719 Datum
 720 arraysmlw(PG_FUNCTION_ARGS)
 721 {
 722         ArrayType               *a, *b;
 723         SimpleArray             *sa, *sb;
 724         bool                    useIntersect = PG_GETARG_BOOL(2);
 725         double                  numerator = 0.0;
 726         double                  denominatorA = 0.0,
 727                                         denominatorB = 0.0,
 728                                         tmpA, tmpB;
 729         int                             cmp;
 730         ProcTypeInfo    info;
 731         int                             ai = 0, bi = 0;
 732
 733         fcinfo->flinfo->fn_extra = SearchArrayCache(
 734                                                         fcinfo->flinfo->fn_extra,
 735                                                         fcinfo->flinfo->fn_mcxt,
 736                                                         PG_GETARG_DATUM(0), &a, &sa, NULL);
 737         fcinfo->flinfo->fn_extra = SearchArrayCache(
 738                                                         fcinfo->flinfo->fn_extra,
 739                                                         fcinfo->flinfo->fn_mcxt,
 740                                                         PG_GETARG_DATUM(1), &b, &sb, NULL);
 741
 742         if ( ARR_ELEMTYPE(a) != ARR_ELEMTYPE(b) )
 743                 elog(ERROR,"Arguments array are not the same type!");
 744
 745         if (ARRISVOID(a) || ARRISVOID(b))
 746                  PG_RETURN_FLOAT4(0.0);
 747
 748         info = sa->info;
 749         if (info->tupDesc == NULL)
 750                 elog(ERROR, "Only weigthed (composite) types should be used");
 751         getFmgrInfoCmp(info);
 752
 753         while(ai < sa->nelems && bi < sb->nelems)
 754         {
 755                 Datum   ad = deconstructCompositeType(info, sa->elems[ai], &tmpA),
 756                                 bd = deconstructCompositeType(info, sb->elems[bi], &tmpB);
 757
 758                 cmp = DatumGetInt32(FCall2(&info->cmpFunc, ad, bd));
 759
 760                 if ( cmp < 0 ) {
 761                         if (useIntersect == false)
 762                                 denominatorA += tmpA * tmpA;
 763                         ai++;
 764                 } else if ( cmp > 0 ) {
 765                         if (useIntersect == false)
 766                                 denominatorB += tmpB * tmpB;
 767                         bi++;
 768                 } else {
 769                         denominatorA += tmpA * tmpA;
 770                         denominatorB += tmpB * tmpB;
 771                         numerator += tmpA * tmpB;
 772                         ai++;
 773                         bi++;
 774                 }
 775         }
 776
 777         if (useIntersect == false) {
 778                 while(ai < sa->nelems) {
 779                         deconstructCompositeType(info, sa->elems[ai], &tmpA);
 780                         denominatorA += tmpA * tmpA;
 781                         ai++;
 782                 }
 783
 784                 while(bi < sb->nelems) {
 785                         deconstructCompositeType(info, sb->elems[bi], &tmpB);
 786                         denominatorB += tmpB * tmpB;
 787                         bi++;
 788                 }
 789         }
 790
 791         if (numerator != 0.0) {
 792                 numerator = numerator / sqrt( denominatorA * denominatorB );
 793         }
 794
 795         PG_RETURN_FLOAT4(numerator);
 796 }
 797
 798 PG_FUNCTION_INFO_V1(arraysml_op);
 799 Datum   arraysml_op(PG_FUNCTION_ARGS);
 800 Datum
 801 arraysml_op(PG_FUNCTION_ARGS)
 802 {
 803         ArrayType               *a, *b;
 804         SimpleArray             *sa, *sb;
 805         double                  power = 0.0;
 806
 807         fcinfo->flinfo->fn_extra = SearchArrayCache(
 808                                                         fcinfo->flinfo->fn_extra,
 809                                                         fcinfo->flinfo->fn_mcxt,
 810                                                         PG_GETARG_DATUM(0), &a, &sa, NULL);
 811         fcinfo->flinfo->fn_extra = SearchArrayCache(
 812                                                         fcinfo->flinfo->fn_extra,
 813                                                         fcinfo->flinfo->fn_mcxt,
 814                                                         PG_GETARG_DATUM(1), &b, &sb, NULL);
 815
 816         if ( ARR_ELEMTYPE(a) != ARR_ELEMTYPE(b) )
 817                 elog(ERROR,"Arguments array are not the same type!");
 818
 819         if (ARRISVOID(a) || ARRISVOID(b))
 820                  PG_RETURN_BOOL(false);
 821
 822         switch(getSmlType())
 823         {
 824                 case ST_TFIDF:
 825                         power = TFIDFSml(sa, sb);
 826                         break;
 827                 case ST_COSINE:
 828                         {
 829                                 int                             cnt;
 830
 831                                 power = sqrt( ((double)(sa->nelems)) * ((double)(sb->nelems)) );
 832
 833                                 if (  ((double)Min(sa->nelems, sb->nelems)) / power < GetSmlarLimit()  )
 834                                         PG_RETURN_BOOL(false);
 835
 836                                 cnt = numOfIntersect(sa, sb);
 837                                 power = ((double)cnt) / power;
 838                         }
 839                         break;
 840                 case ST_OVERLAP:
 841                         power = (double)numOfIntersect(sa, sb);
 842                         break;
 843                 default:
 844                         elog(ERROR,"Unsupported formula type of similarity");
 845         }
 846
 847         PG_RETURN_BOOL(power >= GetSmlarLimit());
 848 }
 849
 850 #define QBSIZE          8192
 851 static char cachedFormula[QBSIZE];
 852 static int      cachedLen  = 0;
 853 static void     *cachedPlan = NULL;
 854
 855 PG_FUNCTION_INFO_V1(arraysml_func);
 856 Datum   arraysml_func(PG_FUNCTION_ARGS);
 857 Datum
 858 arraysml_func(PG_FUNCTION_ARGS)
 859 {
 860         ArrayType               *a, *b;
 861         SimpleArray             *sa, *sb;
 862         int                             cnt;
 863         float4                  result = -1.0;
 864         Oid                             arg[] = {INT4OID, INT4OID, INT4OID};
 865         Datum                   pars[3];
 866         bool                    isnull;
 867         void                    *plan;
 868         int                             stat;
 869         text                    *formula = PG_GETARG_TEXT_P(2);
 870
 871         fcinfo->flinfo->fn_extra = SearchArrayCache(
 872                                                         fcinfo->flinfo->fn_extra,
 873                                                         fcinfo->flinfo->fn_mcxt,
 874                                                         PG_GETARG_DATUM(0), &a, &sa, NULL);
 875         fcinfo->flinfo->fn_extra = SearchArrayCache(
 876                                                         fcinfo->flinfo->fn_extra,
 877                                                         fcinfo->flinfo->fn_mcxt,
 878                                                         PG_GETARG_DATUM(1), &b, &sb, NULL);
 879
 880         if ( ARR_ELEMTYPE(a) != ARR_ELEMTYPE(b) )
 881                 elog(ERROR,"Arguments array are not the same type!");
 882
 883         if (ARRISVOID(a) || ARRISVOID(b))
 884                  PG_RETURN_BOOL(false);
 885
 886         cnt = numOfIntersect(sa, sb);
 887
 888         if ( VARSIZE(formula) - VARHDRSZ > QBSIZE - 1024 )
 889                 elog(ERROR,"Formula is too long");
 890
 891         SPI_connect();
 892
 893         if ( cachedPlan == NULL || cachedLen != VARSIZE(formula) - VARHDRSZ ||
 894                                 memcmp( cachedFormula, VARDATA(formula), VARSIZE(formula) - VARHDRSZ ) != 0 )
 895         {
 896                 char                    *ptr, buf[QBSIZE];
 897
 898                 *cachedFormula = '\0';
 899                 if ( cachedPlan )
 900                         SPI_freeplan(cachedPlan);
 901                 cachedPlan = NULL;
 902                 cachedLen = 0;
 903
 904                 ptr = stpcpy( buf, "SELECT (" );
 905                 memcpy( ptr, VARDATA(formula), VARSIZE(formula) - VARHDRSZ );
 906                 ptr += VARSIZE(formula) - VARHDRSZ;
 907                 ptr = stpcpy( ptr, ")::float4 FROM");
 908                 ptr = stpcpy( ptr, " (SELECT $1 ::float8 AS i, $2 ::float8 AS a, $3 ::float8 AS b) AS N;");
 909                 *ptr = '\0';
 910
 911                 plan = SPI_prepare(buf, 3, arg);
 912                 if (!plan)
 913                         elog(ERROR, "SPI_prepare() failed");
 914
 915                 cachedPlan = SPI_saveplan(plan);
 916                 if (!cachedPlan)
 917                         elog(ERROR, "SPI_saveplan() failed");
 918
 919                 SPI_freeplan(plan);
 920                 cachedLen = VARSIZE(formula) - VARHDRSZ;
 921                 memcpy( cachedFormula, VARDATA(formula), VARSIZE(formula) - VARHDRSZ );
 922         }
 923
 924         plan = cachedPlan;
 925
 926
 927         pars[0] = Int32GetDatum( cnt );
 928         pars[1] = Int32GetDatum( sa->nelems );
 929         pars[2] = Int32GetDatum( sb->nelems );
 930
 931         stat = SPI_execute_plan(plan, pars, NULL, true, 3);
 932         if (stat < 0)
 933                 elog(ERROR, "SPI_execute_plan() returns %d", stat);
 934
 935         if ( SPI_processed > 0)
 936                 result = DatumGetFloat4(SPI_getbinval(SPI_tuptable->vals[0], SPI_tuptable->tupdesc, 1, &isnull));
 937
 938         SPI_finish();
 939
 940         PG_RETURN_FLOAT4(result);
 941 }
 942
 943 PG_FUNCTION_INFO_V1(array_unique);
 944 Datum   array_unique(PG_FUNCTION_ARGS);
 945 Datum
 946 array_unique(PG_FUNCTION_ARGS)
 947 {
 948         ArrayType               *a = PG_GETARG_ARRAYTYPE_P(0);
 949         ArrayType               *res;
 950         SimpleArray             *sa;
 951
 952         sa = Array2SimpleArrayU(NULL, a, NULL);
 953
 954         res = construct_array(  sa->elems,
 955                                                         sa->nelems,
 956                                                         sa->info->typid,
 957                                                         sa->info->typlen,
 958                                                         sa->info->typbyval,
 959                                                         sa->info->typalign);
 960
 961         pfree(sa->elems);
 962         pfree(sa);
 963         PG_FREE_IF_COPY(a, 0);
 964
 965         PG_RETURN_ARRAYTYPE_P(res);
 966 }
 967
 968 PG_FUNCTION_INFO_V1(inarray);
 969 Datum   inarray(PG_FUNCTION_ARGS);
 970 Datum
 971 inarray(PG_FUNCTION_ARGS)
 972 {
 973         ArrayType               *a;
 974         SimpleArray             *sa;
 975         Datum                   query = PG_GETARG_DATUM(1);
 976         Oid                             queryTypeOid;
 977         Datum                   *StopLow,
 978                                         *StopHigh,
 979                                         *StopMiddle;
 980         int                             cmp;
 981
 982         fcinfo->flinfo->fn_extra = SearchArrayCache(
 983                                                         fcinfo->flinfo->fn_extra,
 984                                                         fcinfo->flinfo->fn_mcxt,
 985                                                         PG_GETARG_DATUM(0), &a, &sa, NULL);
 986
 987         queryTypeOid = get_fn_expr_argtype(fcinfo->flinfo, 1);
 988
 989         if ( queryTypeOid == InvalidOid )
 990                 elog(ERROR,"inarray: could not determine actual argument type");
 991
 992         if ( queryTypeOid != sa->info->typid )
 993                 elog(ERROR,"inarray: Type of array's element and type of argument are not the same");
 994
 995         getFmgrInfoCmp(sa->info);
 996         StopLow = sa->elems;
 997         StopHigh = sa->elems + sa->nelems;
 998
 999         while (StopLow < StopHigh)
1000         {
1001                 StopMiddle = StopLow + ((StopHigh - StopLow) >> 1);
1002                 cmp = cmpArrayElem(StopMiddle, &query, sa->info);
1003
1004                 if ( cmp == 0 )
1005                 {
1006                         /* found */
1007                         if ( PG_NARGS() >= 3 )
1008                                 PG_RETURN_DATUM(PG_GETARG_DATUM(2));
1009                         PG_RETURN_FLOAT4(1.0);
1010                 }
1011                 else if (cmp < 0)
1012                         StopLow = StopMiddle + 1;
1013                 else
1014                         StopHigh = StopMiddle;
1015         }
1016
1017         if ( PG_NARGS() >= 4 )
1018                 PG_RETURN_DATUM(PG_GETARG_DATUM(3));
1019         PG_RETURN_FLOAT4(0.0);
1020 }