src/backend/utils/adt/tsquery.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  * tsquery.c
   4  *        I/O functions for tsquery
   5  *
   6  * Portions Copyright (c) 1996-2008, PostgreSQL Global Development Group
   7  *
   8  *
   9  * IDENTIFICATION
  10  *        $PostgreSQL$
  11  *
  12  *-------------------------------------------------------------------------
  13  */
  14
  15 #include "postgres.h"
  16
  17 #include "libpq/pqformat.h"
  18 #include "miscadmin.h"
  19 #include "tsearch/ts_locale.h"
  20 #include "tsearch/ts_type.h"
  21 #include "tsearch/ts_utils.h"
  22 #include "utils/builtins.h"
  23 #include "utils/memutils.h"
  24 #include "utils/pg_crc.h"
  25
  26
  27 struct TSQueryParserStateData
  28 {
  29         /* State for gettoken_query */
  30         char       *buffer;                     /* entire string we are scanning */
  31         char       *buf;                        /* current scan point */
  32         int                     state;
  33         int                     count;                  /* nesting count, incremented by (,
  34                                                                  * decremented by ) */
  35
  36         /* polish (prefix) notation in list, filled in by push* functions */
  37         List       *polstr;
  38
  39         /*
  40          * Strings from operands are collected in op. curop is a pointer to the
  41          * end of used space of op.
  42          */
  43         char       *op;
  44         char       *curop;
  45         int                     lenop;                  /* allocated size of op */
  46         int                     sumlen;                 /* used size of op */
  47
  48         /* state for value's parser */
  49         TSVectorParseState valstate;
  50 };
  51
  52 /* parser's states */
  53 #define WAITOPERAND 1
  54 #define WAITOPERATOR    2
  55 #define WAITFIRSTOPERAND 3
  56 #define WAITSINGLEOPERAND 4
  57
  58 /*
  59  * subroutine to parse the modifiers (weight and prefix flag currently)
  60  * part, like ':1AB' of a query.
  61  */
  62 static char *
  63 get_modifiers(char *buf, int16 *weight, bool *prefix)
  64 {
  65         *weight = 0;
  66         *prefix = false;
  67
  68         if (!t_iseq(buf, ':'))
  69                 return buf;
  70
  71         buf++;
  72         while (*buf && pg_mblen(buf) == 1)
  73         {
  74                 switch (*buf)
  75                 {
  76                         case 'a':
  77                         case 'A':
  78                                 *weight |= 1 << 3;
  79                                 break;
  80                         case 'b':
  81                         case 'B':
  82                                 *weight |= 1 << 2;
  83                                 break;
  84                         case 'c':
  85                         case 'C':
  86                                 *weight |= 1 << 1;
  87                                 break;
  88                         case 'd':
  89                         case 'D':
  90                                 *weight |= 1;
  91                                 break;
  92                         case '*':
  93                                 *prefix = true;
  94                                 break;
  95                         default:
  96                                 return buf;
  97                 }
  98                 buf++;
  99         }
 100
 101         return buf;
 102 }
 103
 104 /*
 105  * token types for parsing
 106  */
 107 typedef enum
 108 {
 109         PT_END = 0,
 110         PT_ERR = 1,
 111         PT_VAL = 2,
 112         PT_OPR = 3,
 113         PT_OPEN = 4,
 114         PT_CLOSE = 5
 115 } ts_tokentype;
 116
 117 /*
 118  * get token from query string
 119  *
 120  * *operator is filled in with OP_* when return values is PT_OPR
 121  * *strval, *lenval and *weight are filled in when return value is PT_VAL
 122  */
 123 static ts_tokentype
 124 gettoken_query(TSQueryParserState state,
 125                            int8 *operator,
 126                            int *lenval, char **strval, int16 *weight, bool *prefix)
 127 {
 128         *weight = 0;
 129         *prefix = false;
 130
 131         while (1)
 132         {
 133                 switch (state->state)
 134                 {
 135                         case WAITFIRSTOPERAND:
 136                         case WAITOPERAND:
 137                                 if (t_iseq(state->buf, '!'))
 138                                 {
 139                                         (state->buf)++;         /* can safely ++, t_iseq guarantee
 140                                                                                  * that pg_mblen()==1 */
 141                                         *operator = OP_NOT;
 142                                         state->state = WAITOPERAND;
 143                                         return PT_OPR;
 144                                 }
 145                                 else if (t_iseq(state->buf, '('))
 146                                 {
 147                                         state->count++;
 148                                         (state->buf)++;
 149                                         state->state = WAITOPERAND;
 150                                         return PT_OPEN;
 151                                 }
 152                                 else if (t_iseq(state->buf, ':'))
 153                                 {
 154                                         ereport(ERROR,
 155                                                         (errcode(ERRCODE_SYNTAX_ERROR),
 156                                                          errmsg("syntax error in tsquery: \"%s\"",
 157                                                                         state->buffer)));
 158                                 }
 159                                 else if (!t_isspace(state->buf))
 160                                 {
 161                                         /*
 162                                          * We rely on the tsvector parser to parse the value for
 163                                          * us
 164                                          */
 165                                         reset_tsvector_parser(state->valstate, state->buf);
 166                                         if (gettoken_tsvector(state->valstate, strval, lenval, NULL, NULL, &state->buf))
 167                                         {
 168                                                 state->buf = get_modifiers(state->buf, weight, prefix);
 169                                                 state->state = WAITOPERATOR;
 170                                                 return PT_VAL;
 171                                         }
 172                                         else if (state->state == WAITFIRSTOPERAND)
 173                                                 return PT_END;
 174                                         else
 175                                                 ereport(ERROR,
 176                                                                 (errcode(ERRCODE_SYNTAX_ERROR),
 177                                                                  errmsg("no operand in tsquery: \"%s\"",
 178                                                                                 state->buffer)));
 179                                 }
 180                                 break;
 181                         case WAITOPERATOR:
 182                                 if (t_iseq(state->buf, '&'))
 183                                 {
 184                                         state->state = WAITOPERAND;
 185                                         *operator = OP_AND;
 186                                         (state->buf)++;
 187                                         return PT_OPR;
 188                                 }
 189                                 if (t_iseq(state->buf, '|'))
 190                                 {
 191                                         state->state = WAITOPERAND;
 192                                         *operator = OP_OR;
 193                                         (state->buf)++;
 194                                         return PT_OPR;
 195                                 }
 196                                 else if (t_iseq(state->buf, ')'))
 197                                 {
 198                                         (state->buf)++;
 199                                         state->count--;
 200                                         return (state->count < 0) ? PT_ERR : PT_CLOSE;
 201                                 }
 202                                 else if (*(state->buf) == '\0')
 203                                         return (state->count) ? PT_ERR : PT_END;
 204                                 else if (!t_isspace(state->buf))
 205                                         return PT_ERR;
 206                                 break;
 207                         case WAITSINGLEOPERAND:
 208                                 if (*(state->buf) == '\0')
 209                                         return PT_END;
 210                                 *strval = state->buf;
 211                                 *lenval = strlen(state->buf);
 212                                 state->buf += strlen(state->buf);
 213                                 state->count++;
 214                                 return PT_VAL;
 215                         default:
 216                                 return PT_ERR;
 217                                 break;
 218                 }
 219                 state->buf += pg_mblen(state->buf);
 220         }
 221         return PT_END;
 222 }
 223
 224 /*
 225  * Push an operator to state->polstr
 226  */
 227 void
 228 pushOperator(TSQueryParserState state, int8 oper)
 229 {
 230         QueryOperator *tmp;
 231
 232         Assert(oper == OP_NOT || oper == OP_AND || oper == OP_OR);
 233
 234         tmp = (QueryOperator *) palloc0(sizeof(QueryOperator));
 235         tmp->type = QI_OPR;
 236         tmp->oper = oper;
 237         /* left is filled in later with findoprnd */
 238
 239         state->polstr = lcons(tmp, state->polstr);
 240 }
 241
 242 static void
 243 pushValue_internal(TSQueryParserState state, pg_crc32 valcrc, int distance, int lenval, int weight, bool prefix)
 244 {
 245         QueryOperand *tmp;
 246
 247         if (distance >= MAXSTRPOS)
 248                 ereport(ERROR,
 249                                 (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
 250                                  errmsg("value is too big in tsquery: \"%s\"",
 251                                                 state->buffer)));
 252         if (lenval >= MAXSTRLEN)
 253                 ereport(ERROR,
 254                                 (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
 255                                  errmsg("operand is too long in tsquery: \"%s\"",
 256                                                 state->buffer)));
 257
 258         tmp = (QueryOperand *) palloc0(sizeof(QueryOperand));
 259         tmp->type = QI_VAL;
 260         tmp->weight = weight;
 261         tmp->prefix = prefix;
 262         tmp->valcrc = (int32) valcrc;
 263         tmp->length = lenval;
 264         tmp->distance = distance;
 265
 266         state->polstr = lcons(tmp, state->polstr);
 267 }
 268
 269 /*
 270  * Push an operand to state->polstr.
 271  *
 272  * strval must point to a string equal to state->curop. lenval is the length
 273  * of the string.
 274  */
 275 void
 276 pushValue(TSQueryParserState state, char *strval, int lenval, int2 weight, bool prefix)
 277 {
 278         pg_crc32        valcrc;
 279
 280         if (lenval >= MAXSTRLEN)
 281                 ereport(ERROR,
 282                                 (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
 283                                  errmsg("word is too long in tsquery: \"%s\"",
 284                                                 state->buffer)));
 285
 286         INIT_CRC32(valcrc);
 287         COMP_CRC32(valcrc, strval, lenval);
 288         FIN_CRC32(valcrc);
 289         pushValue_internal(state, valcrc, state->curop - state->op, lenval, weight, prefix);
 290
 291         /* append the value string to state.op, enlarging buffer if needed first */
 292         while (state->curop - state->op + lenval + 1 >= state->lenop)
 293         {
 294                 int                     used = state->curop - state->op;
 295
 296                 state->lenop *= 2;
 297                 state->op = (char *) repalloc((void *) state->op, state->lenop);
 298                 state->curop = state->op + used;
 299         }
 300         memcpy((void *) state->curop, (void *) strval, lenval);
 301         state->curop += lenval;
 302         *(state->curop) = '\0';
 303         state->curop++;
 304         state->sumlen += lenval + 1 /* \0 */ ;
 305 }
 306
 307
 308 /*
 309  * Push a stopword placeholder to state->polstr
 310  */
 311 void
 312 pushStop(TSQueryParserState state)
 313 {
 314         QueryOperand *tmp;
 315
 316         tmp = (QueryOperand *) palloc0(sizeof(QueryOperand));
 317         tmp->type = QI_VALSTOP;
 318
 319         state->polstr = lcons(tmp, state->polstr);
 320 }
 321
 322
 323 #define STACKDEPTH      32
 324
 325 /*
 326  * Make polish (prefix) notation of query.
 327  *
 328  * See parse_tsquery for explanation of pushval.
 329  */
 330 static void
 331 makepol(TSQueryParserState state,
 332                 PushFunction pushval,
 333                 Datum opaque)
 334 {
 335         int8            operator = 0;
 336         ts_tokentype type;
 337         int                     lenval = 0;
 338         char       *strval = NULL;
 339         int8            opstack[STACKDEPTH];
 340         int                     lenstack = 0;
 341         int16           weight = 0;
 342         bool            prefix;
 343
 344         /* since this function recurses, it could be driven to stack overflow */
 345         check_stack_depth();
 346
 347         while ((type = gettoken_query(state, &operator, &lenval, &strval, &weight, &prefix)) != PT_END)
 348         {
 349                 switch (type)
 350                 {
 351                         case PT_VAL:
 352                                 pushval(opaque, state, strval, lenval, weight, prefix);
 353                                 while (lenstack && (opstack[lenstack - 1] == OP_AND ||
 354                                                                         opstack[lenstack - 1] == OP_NOT))
 355                                 {
 356                                         lenstack--;
 357                                         pushOperator(state, opstack[lenstack]);
 358                                 }
 359                                 break;
 360                         case PT_OPR:
 361                                 if (lenstack && operator == OP_OR)
 362                                         pushOperator(state, OP_OR);
 363                                 else
 364                                 {
 365                                         if (lenstack == STACKDEPTH) /* internal error */
 366                                                 elog(ERROR, "tsquery stack too small");
 367                                         opstack[lenstack] = operator;
 368                                         lenstack++;
 369                                 }
 370                                 break;
 371                         case PT_OPEN:
 372                                 makepol(state, pushval, opaque);
 373
 374                                 if (lenstack && (opstack[lenstack - 1] == OP_AND ||
 375                                                                  opstack[lenstack - 1] == OP_NOT))
 376                                 {
 377                                         lenstack--;
 378                                         pushOperator(state, opstack[lenstack]);
 379                                 }
 380                                 break;
 381                         case PT_CLOSE:
 382                                 while (lenstack)
 383                                 {
 384                                         lenstack--;
 385                                         pushOperator(state, opstack[lenstack]);
 386                                 };
 387                                 return;
 388                         case PT_ERR:
 389                         default:
 390                                 ereport(ERROR,
 391                                                 (errcode(ERRCODE_SYNTAX_ERROR),
 392                                                  errmsg("syntax error in tsquery: \"%s\"",
 393                                                                 state->buffer)));
 394                 }
 395         }
 396         while (lenstack)
 397         {
 398                 lenstack--;
 399                 pushOperator(state, opstack[lenstack]);
 400         }
 401 }
 402
 403 static void
 404 findoprnd_recurse(QueryItem *ptr, uint32 *pos, int nnodes)
 405 {
 406         /* since this function recurses, it could be driven to stack overflow. */
 407         check_stack_depth();
 408
 409         if (*pos >= nnodes)
 410                 elog(ERROR, "malformed tsquery: operand not found");
 411
 412         if (ptr[*pos].type == QI_VAL ||
 413                 ptr[*pos].type == QI_VALSTOP)   /* need to handle VALSTOP here, they
 414                                                                                  * haven't been cleaned away yet. */
 415         {
 416                 (*pos)++;
 417         }
 418         else
 419         {
 420                 Assert(ptr[*pos].type == QI_OPR);
 421
 422                 if (ptr[*pos].operator.oper == OP_NOT)
 423                 {
 424                         ptr[*pos].operator.left = 1;
 425                         (*pos)++;
 426                         findoprnd_recurse(ptr, pos, nnodes);
 427                 }
 428                 else
 429                 {
 430                         QueryOperator *curitem = &ptr[*pos].operator;
 431                         int                     tmp = *pos;
 432
 433                         Assert(curitem->oper == OP_AND || curitem->oper == OP_OR);
 434
 435                         (*pos)++;
 436                         findoprnd_recurse(ptr, pos, nnodes);
 437                         curitem->left = *pos - tmp;
 438                         findoprnd_recurse(ptr, pos, nnodes);
 439                 }
 440         }
 441 }
 442
 443
 444 /*
 445  * Fills in the left-fields previously left unfilled. The input
 446  * QueryItems must be in polish (prefix) notation.
 447  */
 448 static void
 449 findoprnd(QueryItem *ptr, int size)
 450 {
 451         uint32          pos;
 452
 453         pos = 0;
 454         findoprnd_recurse(ptr, &pos, size);
 455
 456         if (pos != size)
 457                 elog(ERROR, "malformed tsquery: extra nodes");
 458 }
 459
 460
 461 /*
 462  * Each value (operand) in the query is be passed to pushval. pushval can
 463  * transform the simple value to an arbitrarily complex expression using
 464  * pushValue and pushOperator. It must push a single value with pushValue,
 465  * a complete expression with all operands, or a a stopword placeholder
 466  * with pushStop, otherwise the prefix notation representation will be broken,
 467  * having an operator with no operand.
 468  *
 469  * opaque is passed on to pushval as is, pushval can use it to store its
 470  * private state.
 471  *
 472  * The returned query might contain QI_STOPVAL nodes. The caller is responsible
 473  * for cleaning them up (with clean_fakeval)
 474  */
 475 TSQuery
 476 parse_tsquery(char *buf,
 477                           PushFunction pushval,
 478                           Datum opaque,
 479                           bool isplain)
 480 {
 481         struct TSQueryParserStateData state;
 482         int                     i;
 483         TSQuery         query;
 484         int                     commonlen;
 485         QueryItem  *ptr;
 486         ListCell   *cell;
 487
 488         /* init state */
 489         state.buffer = buf;
 490         state.buf = buf;
 491         state.state = (isplain) ? WAITSINGLEOPERAND : WAITFIRSTOPERAND;
 492         state.count = 0;
 493         state.polstr = NIL;
 494
 495         /* init value parser's state */
 496         state.valstate = init_tsvector_parser(state.buffer, true, true);
 497
 498         /* init list of operand */
 499         state.sumlen = 0;
 500         state.lenop = 64;
 501         state.curop = state.op = (char *) palloc(state.lenop);
 502         *(state.curop) = '\0';
 503
 504         /* parse query & make polish notation (postfix, but in reverse order) */
 505         makepol(&state, pushval, opaque);
 506
 507         close_tsvector_parser(state.valstate);
 508
 509         if (list_length(state.polstr) == 0)
 510         {
 511                 ereport(NOTICE,
 512                                 (errmsg("text-search query doesn't contain lexemes: \"%s\"",
 513                                                 state.buffer)));
 514                 query = (TSQuery) palloc(HDRSIZETQ);
 515                 SET_VARSIZE(query, HDRSIZETQ);
 516                 query->size = 0;
 517                 return query;
 518         }
 519
 520         /* Pack the QueryItems in the final TSQuery struct to return to caller */
 521         commonlen = COMPUTESIZE(list_length(state.polstr), state.sumlen);
 522         query = (TSQuery) palloc0(commonlen);
 523         SET_VARSIZE(query, commonlen);
 524         query->size = list_length(state.polstr);
 525         ptr = GETQUERY(query);
 526
 527         /* Copy QueryItems to TSQuery */
 528         i = 0;
 529         foreach(cell, state.polstr)
 530         {
 531                 QueryItem  *item = (QueryItem *) lfirst(cell);
 532
 533                 switch (item->type)
 534                 {
 535                         case QI_VAL:
 536                                 memcpy(&ptr[i], item, sizeof(QueryOperand));
 537                                 break;
 538                         case QI_VALSTOP:
 539                                 ptr[i].type = QI_VALSTOP;
 540                                 break;
 541                         case QI_OPR:
 542                                 memcpy(&ptr[i], item, sizeof(QueryOperator));
 543                                 break;
 544                         default:
 545                                 elog(ERROR, "unrecognized QueryItem type: %d", item->type);
 546                 }
 547                 i++;
 548         }
 549
 550         /* Copy all the operand strings to TSQuery */
 551         memcpy((void *) GETOPERAND(query), (void *) state.op, state.sumlen);
 552         pfree(state.op);
 553
 554         /* Set left operand pointers for every operator. */
 555         findoprnd(ptr, query->size);
 556
 557         return query;
 558 }
 559
 560 static void
 561 pushval_asis(Datum opaque, TSQueryParserState state, char *strval, int lenval,
 562                          int16 weight, bool prefix)
 563 {
 564         pushValue(state, strval, lenval, weight, prefix);
 565 }
 566
 567 /*
 568  * in without morphology
 569  */
 570 Datum
 571 tsqueryin(PG_FUNCTION_ARGS)
 572 {
 573         char       *in = PG_GETARG_CSTRING(0);
 574
 575         pg_verifymbstr(in, strlen(in), false);
 576
 577         PG_RETURN_TSQUERY(parse_tsquery(in, pushval_asis, PointerGetDatum(NULL), false));
 578 }
 579
 580 /*
 581  * out function
 582  */
 583 typedef struct
 584 {
 585         QueryItem  *curpol;
 586         char       *buf;
 587         char       *cur;
 588         char       *op;
 589         int                     buflen;
 590 } INFIX;
 591
 592 /* Makes sure inf->buf is large enough for adding 'addsize' bytes */
 593 #define RESIZEBUF(inf, addsize) \
 594 while( ( (inf)->cur - (inf)->buf ) + (addsize) + 1 >= (inf)->buflen ) \
 595 { \
 596         int len = (inf)->cur - (inf)->buf; \
 597         (inf)->buflen *= 2; \
 598         (inf)->buf = (char*) repalloc( (void*)(inf)->buf, (inf)->buflen ); \
 599         (inf)->cur = (inf)->buf + len; \
 600 }
 601
 602 /*
 603  * recursive walk on tree and print it in
 604  * infix (human-readable) view
 605  */
 606 static void
 607 infix(INFIX *in, bool first)
 608 {
 609         /* since this function recurses, it could be driven to stack overflow. */
 610         check_stack_depth();
 611
 612         if (in->curpol->type == QI_VAL)
 613         {
 614                 QueryOperand *curpol = &in->curpol->operand;
 615                 char       *op = in->op + curpol->distance;
 616                 int                     clen;
 617
 618                 RESIZEBUF(in, curpol->length * (pg_database_encoding_max_length() + 1) + 2 + 6);
 619                 *(in->cur) = '\'';
 620                 in->cur++;
 621                 while (*op)
 622                 {
 623                         if (t_iseq(op, '\''))
 624                         {
 625                                 *(in->cur) = '\'';
 626                                 in->cur++;
 627                         }
 628                         else if (t_iseq(op, '\\'))
 629                         {
 630                                 *(in->cur) = '\\';
 631                                 in->cur++;
 632                         }
 633                         COPYCHAR(in->cur, op);
 634
 635                         clen = pg_mblen(op);
 636                         op += clen;
 637                         in->cur += clen;
 638                 }
 639                 *(in->cur) = '\'';
 640                 in->cur++;
 641                 if (curpol->weight || curpol->prefix)
 642                 {
 643                         *(in->cur) = ':';
 644                         in->cur++;
 645                         if ( curpol->prefix )
 646                         {
 647                                 *(in->cur) = '*';
 648                                 in->cur++;
 649                         }
 650                         if (curpol->weight & (1 << 3))
 651                         {
 652                                 *(in->cur) = 'A';
 653                                 in->cur++;
 654                         }
 655                         if (curpol->weight & (1 << 2))
 656                         {
 657                                 *(in->cur) = 'B';
 658                                 in->cur++;
 659                         }
 660                         if (curpol->weight & (1 << 1))
 661                         {
 662                                 *(in->cur) = 'C';
 663                                 in->cur++;
 664                         }
 665                         if (curpol->weight & 1)
 666                         {
 667                                 *(in->cur) = 'D';
 668                                 in->cur++;
 669                         }
 670                 }
 671                 *(in->cur) = '\0';
 672                 in->curpol++;
 673         }
 674         else if (in->curpol->operator.oper == OP_NOT)
 675         {
 676                 bool            isopr = false;
 677
 678                 RESIZEBUF(in, 1);
 679                 *(in->cur) = '!';
 680                 in->cur++;
 681                 *(in->cur) = '\0';
 682                 in->curpol++;
 683
 684                 if (in->curpol->type == QI_OPR)
 685                 {
 686                         isopr = true;
 687                         RESIZEBUF(in, 2);
 688                         sprintf(in->cur, "( ");
 689                         in->cur = strchr(in->cur, '\0');
 690                 }
 691
 692                 infix(in, isopr);
 693                 if (isopr)
 694                 {
 695                         RESIZEBUF(in, 2);
 696                         sprintf(in->cur, " )");
 697                         in->cur = strchr(in->cur, '\0');
 698                 }
 699         }
 700         else
 701         {
 702                 int8            op = in->curpol->operator.oper;
 703                 INFIX           nrm;
 704
 705                 in->curpol++;
 706                 if (op == OP_OR && !first)
 707                 {
 708                         RESIZEBUF(in, 2);
 709                         sprintf(in->cur, "( ");
 710                         in->cur = strchr(in->cur, '\0');
 711                 }
 712
 713                 nrm.curpol = in->curpol;
 714                 nrm.op = in->op;
 715                 nrm.buflen = 16;
 716                 nrm.cur = nrm.buf = (char *) palloc(sizeof(char) * nrm.buflen);
 717
 718                 /* get right operand */
 719                 infix(&nrm, false);
 720
 721                 /* get & print left operand */
 722                 in->curpol = nrm.curpol;
 723                 infix(in, false);
 724
 725                 /* print operator & right operand */
 726                 RESIZEBUF(in, 3 + (nrm.cur - nrm.buf));
 727                 switch (op)
 728                 {
 729                         case OP_OR:
 730                                 sprintf(in->cur, " | %s", nrm.buf);
 731                                 break;
 732                         case OP_AND:
 733                                 sprintf(in->cur, " & %s", nrm.buf);
 734                                 break;
 735                         default:
 736                                 /* OP_NOT is handled in above if-branch */
 737                                 elog(ERROR, "unrecognized operator type: %d", op);
 738                 }
 739                 in->cur = strchr(in->cur, '\0');
 740                 pfree(nrm.buf);
 741
 742                 if (op == OP_OR && !first)
 743                 {
 744                         RESIZEBUF(in, 2);
 745                         sprintf(in->cur, " )");
 746                         in->cur = strchr(in->cur, '\0');
 747                 }
 748         }
 749 }
 750
 751
 752 Datum
 753 tsqueryout(PG_FUNCTION_ARGS)
 754 {
 755         TSQuery         query = PG_GETARG_TSQUERY(0);
 756         INFIX           nrm;
 757
 758         if (query->size == 0)
 759         {
 760                 char       *b = palloc(1);
 761
 762                 *b = '\0';
 763                 PG_RETURN_POINTER(b);
 764         }
 765         nrm.curpol = GETQUERY(query);
 766         nrm.buflen = 32;
 767         nrm.cur = nrm.buf = (char *) palloc(sizeof(char) * nrm.buflen);
 768         *(nrm.cur) = '\0';
 769         nrm.op = GETOPERAND(query);
 770         infix(&nrm, true);
 771
 772         PG_FREE_IF_COPY(query, 0);
 773         PG_RETURN_CSTRING(nrm.buf);
 774 }
 775
 776 /*
 777  * Binary Input / Output functions. The binary format is as follows:
 778  *
 779  * uint32        number of operators/operands in the query
 780  *
 781  * Followed by the operators and operands, in prefix notation. For each
 782  * operand:
 783  *
 784  * uint8        type, QI_VAL
 785  * uint8        weight
 786  *                      operand text in client encoding, null-terminated
 787  * uint8        prefix
 788  *
 789  * For each operator:
 790  * uint8        type, QI_OPR
 791  * uint8        operator, one of OP_AND, OP_OR, OP_NOT.
 792  */
 793 Datum
 794 tsquerysend(PG_FUNCTION_ARGS)
 795 {
 796         TSQuery         query = PG_GETARG_TSQUERY(0);
 797         StringInfoData buf;
 798         int                     i;
 799         QueryItem  *item = GETQUERY(query);
 800
 801         pq_begintypsend(&buf);
 802
 803         pq_sendint(&buf, query->size, sizeof(uint32));
 804         for (i = 0; i < query->size; i++)
 805         {
 806                 pq_sendint(&buf, item->type, sizeof(item->type));
 807
 808                 switch (item->type)
 809                 {
 810                         case QI_VAL:
 811                                 pq_sendint(&buf, item->operand.weight, sizeof(uint8));
 812                                 pq_sendint(&buf, item->operand.prefix, sizeof(uint8));
 813                                 pq_sendstring(&buf, GETOPERAND(query) + item->operand.distance);
 814                                 break;
 815                         case QI_OPR:
 816                                 pq_sendint(&buf, item->operator.oper, sizeof(item->operator.oper));
 817                                 break;
 818                         default:
 819                                 elog(ERROR, "unrecognized tsquery node type: %d", item->type);
 820                 }
 821                 item++;
 822         }
 823
 824         PG_FREE_IF_COPY(query, 0);
 825
 826         PG_RETURN_BYTEA_P(pq_endtypsend(&buf));
 827 }
 828
 829 Datum
 830 tsqueryrecv(PG_FUNCTION_ARGS)
 831 {
 832         StringInfo      buf = (StringInfo) PG_GETARG_POINTER(0);
 833         TSQuery         query;
 834         int                     i,
 835                                 len;
 836         QueryItem  *item;
 837         int                     datalen;
 838         char       *ptr;
 839         uint32          size;
 840         const char **operands;
 841
 842         size = pq_getmsgint(buf, sizeof(uint32));
 843         if (size > (MaxAllocSize / sizeof(QueryItem)))
 844                 elog(ERROR, "invalid size of tsquery");
 845
 846         /* Allocate space to temporarily hold operand strings */
 847         operands = palloc(size * sizeof(char *));
 848
 849         /* Allocate space for all the QueryItems. */
 850         len = HDRSIZETQ + sizeof(QueryItem) * size;
 851         query = (TSQuery) palloc0(len);
 852         query->size = size;
 853         item = GETQUERY(query);
 854
 855         datalen = 0;
 856         for (i = 0; i < size; i++)
 857         {
 858                 item->type = (int8) pq_getmsgint(buf, sizeof(int8));
 859
 860                 if (item->type == QI_VAL)
 861                 {
 862                         size_t          val_len;        /* length after recoding to server encoding */
 863                         uint8           weight;
 864                         uint8           prefix;
 865                         const char *val;
 866                         pg_crc32        valcrc;
 867
 868                         weight = (uint8) pq_getmsgint(buf, sizeof(uint8));
 869                         prefix = (uint8) pq_getmsgint(buf, sizeof(uint8));
 870                         val = pq_getmsgstring(buf);
 871                         val_len = strlen(val);
 872
 873                         /* Sanity checks */
 874
 875                         if (weight > 0xF)
 876                                 elog(ERROR, "invalid tsquery: invalid weight bitmap");
 877
 878                         if (val_len > MAXSTRLEN)
 879                                 elog(ERROR, "invalid tsquery: operand too long");
 880
 881                         if (datalen > MAXSTRPOS)
 882                                 elog(ERROR, "invalid tsquery: total operand length exceeded");
 883
 884                         /* Looks valid. */
 885
 886                         INIT_CRC32(valcrc);
 887                         COMP_CRC32(valcrc, val, val_len);
 888                         FIN_CRC32(valcrc);
 889
 890                         item->operand.weight = weight;
 891                         item->operand.prefix = (prefix) ? true : false;
 892                         item->operand.valcrc = (int32) valcrc;
 893                         item->operand.length = val_len;
 894                         item->operand.distance = datalen;
 895
 896                         /*
 897                          * Operand strings are copied to the final struct after this loop;
 898                          * here we just collect them to an array
 899                          */
 900                         operands[i] = val;
 901
 902                         datalen += val_len + 1;         /* + 1 for the '\0' terminator */
 903                 }
 904                 else if (item->type == QI_OPR)
 905                 {
 906                         int8            oper;
 907
 908                         oper = (int8) pq_getmsgint(buf, sizeof(int8));
 909                         if (oper != OP_NOT && oper != OP_OR && oper != OP_AND)
 910                                 elog(ERROR, "invalid tsquery: unrecognized operator type %d",
 911                                          (int) oper);
 912                         if (i == size - 1)
 913                                 elog(ERROR, "invalid pointer to right operand");
 914
 915                         item->operator.oper = oper;
 916                 }
 917                 else
 918                         elog(ERROR, "unrecognized tsquery node type: %d", item->type);
 919
 920                 item++;
 921         }
 922
 923         /* Enlarge buffer to make room for the operand values. */
 924         query = (TSQuery) repalloc(query, len + datalen);
 925         item = GETQUERY(query);
 926         ptr = GETOPERAND(query);
 927
 928         /*
 929          * Fill in the left-pointers. Checks that the tree is well-formed as a
 930          * side-effect.
 931          */
 932         findoprnd(item, size);
 933
 934         /* Copy operands to output struct */
 935         for (i = 0; i < size; i++)
 936         {
 937                 if (item->type == QI_VAL)
 938                 {
 939                         memcpy(ptr, operands[i], item->operand.length + 1);
 940                         ptr += item->operand.length + 1;
 941                 }
 942                 item++;
 943         }
 944
 945         pfree(operands);
 946
 947         Assert(ptr - GETOPERAND(query) == datalen);
 948
 949         SET_VARSIZE(query, len + datalen);
 950
 951         PG_RETURN_TSVECTOR(query);
 952 }
 953
 954 /*
 955  * debug function, used only for view query
 956  * which will be executed in non-leaf pages in index
 957  */
 958 Datum
 959 tsquerytree(PG_FUNCTION_ARGS)
 960 {
 961         TSQuery         query = PG_GETARG_TSQUERY(0);
 962         INFIX           nrm;
 963         text       *res;
 964         QueryItem  *q;
 965         int                     len;
 966
 967         if (query->size == 0)
 968         {
 969                 res = (text *) palloc(VARHDRSZ);
 970                 SET_VARSIZE(res, VARHDRSZ);
 971                 PG_RETURN_POINTER(res);
 972         }
 973
 974         q = clean_NOT(GETQUERY(query), &len);
 975
 976         if (!q)
 977         {
 978                 res = cstring_to_text("T");
 979         }
 980         else
 981         {
 982                 nrm.curpol = q;
 983                 nrm.buflen = 32;
 984                 nrm.cur = nrm.buf = (char *) palloc(sizeof(char) * nrm.buflen);
 985                 *(nrm.cur) = '\0';
 986                 nrm.op = GETOPERAND(query);
 987                 infix(&nrm, true);
 988                 res = cstring_to_text_with_len(nrm.buf, nrm.cur - nrm.buf);
 989                 pfree(q);
 990         }
 991
 992         PG_FREE_IF_COPY(query, 0);
 993
 994         PG_RETURN_TEXT_P(res);
 995 }