src/cclfind.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2011 Index Data
   3  * See the file LICENSE for details.
   4  */
   5 /**
   6  * \file cclfind.c
   7  * \brief Implements parsing of a CCL FIND query.
   8  *
   9  * This source file implements parsing of a CCL Query (ISO8777).
  10  * The parser uses predictive parsing, but it does several tokens
  11  * of lookahead in the handling of relational operations.. So
  12  * it's not really pure.
  13  */
  14 #if HAVE_CONFIG_H
  15 #include <config.h>
  16 #endif
  17
  18 #include <stdlib.h>
  19 #include <string.h>
  20 #include <assert.h>
  21
  22 #include "cclp.h"
  23
  24 /* returns type of current lookahead */
  25 #define KIND (cclp->look_token->kind)
  26
  27 /* move one token forward */
  28 #define ADVANCE cclp->look_token = cclp->look_token->next
  29
  30 /**
  31  * qual_val_type: test for existance of attribute type/value pair.
  32  * qa:     Attribute array
  33  * type:   Type of attribute to search for
  34  * value:  Value of attribute to seach for
  35  * return: 1 if found; 0 otherwise.
  36  */
  37 static int qual_val_type(ccl_qualifier_t *qa, int type, int value,
  38                          char **attset)
  39 {
  40     int i;
  41
  42     if (!qa)
  43         return 0;
  44     for (i = 0; qa[i]; i++)
  45     {
  46         struct ccl_rpn_attr *q = ccl_qual_get_attr(qa[i]);
  47         while (q)
  48         {
  49             if (q->type == type && q->kind == CCL_RPN_ATTR_NUMERIC &&
  50                 q->value.numeric == value)
  51             {
  52                 if (attset)
  53                     *attset = q->set;
  54                 return 1;
  55             }
  56             q = q->next;
  57         }
  58     }
  59     return 0;
  60 }
  61
  62 /**
  63  * strxcat: concatenate strings.
  64  * n:      Null-terminated Destination string
  65  * src:    Source string to be appended (not null-terminated)
  66  * len:    Length of source string.
  67  */
  68 static void strxcat(char *n, const char *src, int len)
  69 {
  70     while (*n)
  71         n++;
  72     while (--len >= 0)
  73         *n++ = *src++;
  74     *n = '\0';
  75 }
  76
  77 /**
  78  * copy_token_name: Return copy of CCL token name
  79  * tp:      Pointer to token info.
  80  * return:  malloc(3) allocated copy of token name.
  81  */
  82 static char *copy_token_name(struct ccl_token *tp)
  83 {
  84     char *str = (char *)xmalloc(tp->len + 1);
  85     ccl_assert(str);
  86     memcpy(str, tp->name, tp->len);
  87     str[tp->len] = '\0';
  88     return str;
  89 }
  90
  91 /**
  92  * mk_node: Create RPN node.
  93  * kind:   Type of node.
  94  * return: pointer to allocated node.
  95  */
  96 struct ccl_rpn_node *ccl_rpn_node_create(enum ccl_rpn_kind kind)
  97 {
  98     struct ccl_rpn_node *p;
  99     p = (struct ccl_rpn_node *)xmalloc(sizeof(*p));
 100     ccl_assert(p);
 101     p->kind = kind;
 102
 103     switch(kind)
 104     {
 105     case CCL_RPN_TERM:
 106         p->u.t.attr_list = 0;
 107         p->u.t.term = 0;
 108         p->u.t.qual = 0;
 109         break;
 110     default:
 111         break;
 112     }
 113     return p;
 114 }
 115
 116 /**
 117  * ccl_rpn_delete: Delete RPN tree.
 118  * rpn:   Pointer to tree.
 119  */
 120 void ccl_rpn_delete(struct ccl_rpn_node *rpn)
 121 {
 122     struct ccl_rpn_attr *attr, *attr1;
 123     if (!rpn)
 124         return;
 125     switch (rpn->kind)
 126     {
 127     case CCL_RPN_AND:
 128     case CCL_RPN_OR:
 129     case CCL_RPN_NOT:
 130         ccl_rpn_delete(rpn->u.p[0]);
 131         ccl_rpn_delete(rpn->u.p[1]);
 132         break;
 133     case CCL_RPN_TERM:
 134         xfree(rpn->u.t.term);
 135         xfree(rpn->u.t.qual);
 136         for (attr = rpn->u.t.attr_list; attr; attr = attr1)
 137         {
 138             attr1 = attr->next;
 139             if (attr->kind == CCL_RPN_ATTR_STRING)
 140                 xfree(attr->value.str);
 141             if (attr->set)
 142                 xfree(attr->set);
 143             xfree(attr);
 144         }
 145         break;
 146     case CCL_RPN_SET:
 147         xfree(rpn->u.setname);
 148         break;
 149     case CCL_RPN_PROX:
 150         ccl_rpn_delete(rpn->u.p[0]);
 151         ccl_rpn_delete(rpn->u.p[1]);
 152         ccl_rpn_delete(rpn->u.p[2]);
 153         break;
 154     }
 155     xfree(rpn);
 156 }
 157
 158 static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa);
 159
 160 static int is_term_ok(int look, int *list)
 161 {
 162     for (;*list >= 0; list++)
 163         if (look == *list)
 164             return 1;
 165     return 0;
 166 }
 167
 168 static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa);
 169
 170 static struct ccl_rpn_attr *add_attr_node(struct ccl_rpn_node *p,
 171                                            const char *set, int type)
 172 {
 173     struct ccl_rpn_attr *n;
 174
 175     n = (struct ccl_rpn_attr *)xmalloc(sizeof(*n));
 176     ccl_assert(n);
 177     if (set)
 178         n->set = xstrdup(set);
 179     else
 180         n->set = 0;
 181     n->type = type;
 182     n->next = p->u.t.attr_list;
 183     p->u.t.attr_list = n;
 184
 185     return n;
 186 }
 187
 188 /**
 189  * add_attr_numeric: Add attribute (type/value) to RPN term node.
 190  * p:     RPN node of type term.
 191  * type:  Type of attribute
 192  * value: Value of attribute
 193  * set: Attribute set name
 194  */
 195 void ccl_add_attr_numeric(struct ccl_rpn_node *p, const char *set,
 196                           int type, int value)
 197 {
 198     struct ccl_rpn_attr *n;
 199
 200     n = add_attr_node(p, set, type);
 201     n->kind = CCL_RPN_ATTR_NUMERIC;
 202     n->value.numeric = value;
 203 }
 204
 205 void ccl_add_attr_string(struct ccl_rpn_node *p, const char *set,
 206                          int type, char *value)
 207 {
 208     struct ccl_rpn_attr *n;
 209
 210     n = add_attr_node(p, set, type);
 211     n->kind = CCL_RPN_ATTR_STRING;
 212     n->value.str = xstrdup(value);
 213 }
 214
 215
 216 #define REGEX_CHARS "^[]{}()|.*+?!$"
 217 #define CCL_CHARS "#?\\"
 218 /**
 219  * search_term: Parse CCL search term.
 220  * cclp:   CCL Parser
 221  * qa:     Qualifier attributes already applied.
 222  * term_list: tokens we accept as terms in context
 223  * multi:  whether we accept "multiple" tokens
 224  * return: pointer to node(s); NULL on error.
 225  */
 226 static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
 227                                           ccl_qualifier_t *qa,
 228                                           int *term_list, int multi)
 229 {
 230     struct ccl_rpn_node *p_top = 0;
 231     struct ccl_token *lookahead = cclp->look_token;
 232     int and_list = 0;
 233     int or_list = 0;
 234     char *attset;
 235     const char **truncation_aliases;
 236     const char *t_default[2];
 237
 238     truncation_aliases =
 239         ccl_qual_search_special(cclp->bibset, "truncation");
 240     if (!truncation_aliases)
 241     {
 242         truncation_aliases = t_default;
 243         t_default[0] = "?";
 244         t_default[1] = 0;
 245     }
 246
 247     if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_AND_LIST, 0))
 248         and_list = 1;
 249     if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_OR_LIST, 0))
 250         or_list = 1;
 251     while (1)
 252     {
 253         struct ccl_rpn_node *p;
 254         size_t no, i;
 255         int no_spaces = 0;
 256         int relation_value = -1;
 257         int position_value = -1;
 258         int structure_value = -1;
 259         int truncation_value = -1;
 260         int completeness_value = -1;
 261         int len = 0;
 262         int left_trunc = 0;
 263         int right_trunc = 0;
 264         int regex_trunc = 0;
 265         int z3958_trunc = 0;
 266         size_t max = 200;
 267         if (and_list || or_list || !multi)
 268             max = 1;
 269
 270         /* ignore commas when dealing with and-lists .. */
 271         if (and_list && lookahead && lookahead->kind == CCL_TOK_COMMA)
 272         {
 273             lookahead = lookahead->next;
 274             ADVANCE;
 275             continue;
 276         }
 277         /* go through each TERM token. If no truncation attribute is yet
 278            met, then look for left/right truncation markers (?) and
 279            set left_trunc/right_trunc/mid_trunc accordingly */
 280         for (no = 0; no < max && is_term_ok(lookahead->kind, term_list); no++)
 281         {
 282             for (i = 0; i<lookahead->len; i++)
 283                 if (lookahead->name[i] == ' ')
 284                     no_spaces++;
 285             len += 1+lookahead->len+lookahead->ws_prefix_len;
 286             lookahead = lookahead->next;
 287         }
 288
 289         if (len == 0)
 290             break;      /* no more terms . stop . */
 291
 292         /* create the term node, but wait a moment before adding the term */
 293         p = ccl_rpn_node_create(CCL_RPN_TERM);
 294         p->u.t.attr_list = NULL;
 295         p->u.t.term = NULL;
 296         if (qa && qa[0])
 297         {
 298             const char *n = ccl_qual_get_name(qa[0]);
 299             if (n)
 300                 p->u.t.qual = xstrdup(n);
 301         }
 302
 303         /* go through all attributes and add them to the attribute list */
 304         for (i=0; qa && qa[i]; i++)
 305         {
 306             struct ccl_rpn_attr *attr;
 307
 308             for (attr = ccl_qual_get_attr(qa[i]); attr; attr = attr->next)
 309                 switch(attr->kind)
 310                 {
 311                 case CCL_RPN_ATTR_STRING:
 312                     ccl_add_attr_string(p, attr->set, attr->type,
 313                                         attr->value.str);
 314                     break;
 315                 case CCL_RPN_ATTR_NUMERIC:
 316                     if (attr->value.numeric > 0)
 317                     {   /* deal only with REAL attributes (positive) */
 318                         switch (attr->type)
 319                         {
 320                         case CCL_BIB1_REL:
 321                             if (relation_value != -1)
 322                                 continue;
 323                             relation_value = attr->value.numeric;
 324                             break;
 325                         case CCL_BIB1_POS:
 326                             if (position_value != -1)
 327                                 continue;
 328                             position_value = attr->value.numeric;
 329                             break;
 330                         case CCL_BIB1_STR:
 331                             if (structure_value != -1)
 332                                 continue;
 333                             structure_value = attr->value.numeric;
 334                             break;
 335                         case CCL_BIB1_TRU:
 336                             if (truncation_value != -1)
 337                                 continue;
 338                             truncation_value = attr->value.numeric;
 339                             break;
 340                         case CCL_BIB1_COM:
 341                             if (completeness_value != -1)
 342                                 continue;
 343                             completeness_value = attr->value.numeric;
 344                             break;
 345                         }
 346                         ccl_add_attr_numeric(p, attr->set, attr->type,
 347                                              attr->value.numeric);
 348                     }
 349                 }
 350         }
 351         /* len now holds the number of characters in the RPN term */
 352         /* no holds the number of CCL tokens (1 or more) */
 353
 354         if (structure_value == -1 &&
 355             qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_WP, &attset))
 356         {   /* no structure attribute met. Apply either structure attribute
 357                WORD or PHRASE depending on number of CCL tokens */
 358             if (no == 1 && no_spaces == 0)
 359                 ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 2);
 360             else
 361                 ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 1);
 362         }
 363
 364         if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_REGEX,
 365                           &attset))
 366         {
 367             regex_trunc = 1; /* regex trunc (102) allowed */
 368         }
 369         else if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_Z3958,
 370                           &attset))
 371         {
 372             z3958_trunc = 1; /* Z39.58 trunc (CCL) trunc allowed */
 373         }
 374
 375         /* make the RPN token */
 376         p->u.t.term = (char *)xmalloc(len * 2 + 2);
 377         ccl_assert(p->u.t.term);
 378         p->u.t.term[0] = '\0';
 379         for (i = 0; i<no; i++)
 380         {
 381             const char *src_str = cclp->look_token->name;
 382             size_t src_len = cclp->look_token->len;
 383             int j;
 384             int quote_mode = 0;
 385
 386             if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
 387             {
 388                 strxcat(p->u.term, cclp->look_token->ws_prefix_buf,
 389                         cclp->look_token->ws_prefix_len);
 390             }
 391             for (j = 0; j < src_len; j++)
 392             {
 393                 if (j > 0 && src_str[j-1] == '\\')
 394                 {
 395                     if (regex_trunc && strchr(REGEX_CHARS "\\", src_str[j]))
 396                     {
 397                         regex_trunc = 2;
 398                         strcat(p->u.t.term, "\\");
 399                     }
 400                     else if (z3958_trunc && strchr(CCL_CHARS "\\", src_str[j]))
 401                     {
 402                         z3958_trunc = 2;
 403                         strcat(p->u.t.term, "\\");
 404                     }
 405                     strxcat(p->u.t.term, src_str + j, 1);
 406                 }
 407                 else if (src_str[j] == '"')
 408                     quote_mode = !quote_mode;
 409                 else if (!quote_mode && src_str[j] == '?')
 410                 {
 411                     if (regex_trunc)
 412                     {
 413                         strcat(p->u.t.term, ".*");
 414                         regex_trunc = 2; /* regex trunc is really needed */
 415                     }
 416                     else if (z3958_trunc)
 417                     {
 418                         strcat(p->u.t.term, "?");
 419                         z3958_trunc = 2;
 420                     }
 421                     else if (i == 0 && j == 0)
 422                         left_trunc = 1;
 423                     else if (i == no - 1 && j == src_len - 1)
 424                         right_trunc = 1;
 425                     else
 426                     {
 427                         cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
 428                         ccl_rpn_delete(p);
 429                         return NULL;
 430                     }
 431                 }
 432                 else if (!quote_mode && src_str[j] == '#')
 433                 {
 434                     if (regex_trunc)
 435                     {
 436                         strcat(p->u.t.term, ".");
 437                         regex_trunc = 2; /* regex trunc is really needed */
 438                     }
 439                     else if (z3958_trunc)
 440                     {
 441                         strcat(p->u.t.term, "#");
 442                         z3958_trunc = 2;
 443                     }
 444                     else
 445                     {
 446                         cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
 447                         ccl_rpn_delete(p);
 448                         return NULL;
 449                     }
 450                 }
 451                 else if (src_str[j] != '\\')
 452                 {
 453                     if (regex_trunc && strchr(REGEX_CHARS, src_str[j]))
 454                     {
 455                         regex_trunc = 2;
 456                         strcat(p->u.t.term, "\\");
 457                     }
 458                     else if (z3958_trunc && strchr(CCL_CHARS, src_str[j]))
 459                     {
 460                         z3958_trunc = 2;
 461                         strcat(p->u.t.term, "\\");
 462                     }
 463                     strxcat(p->u.t.term, src_str + j, 1);
 464                 }
 465             }
 466             ADVANCE;
 467         }
 468
 469         /* make the top node point to us.. */
 470         if (p_top)
 471         {
 472             struct ccl_rpn_node *tmp;
 473
 474             if (or_list)
 475                 tmp = ccl_rpn_node_create(CCL_RPN_OR);
 476             else if (and_list)
 477                 tmp = ccl_rpn_node_create(CCL_RPN_AND);
 478             else
 479                 tmp = ccl_rpn_node_create(CCL_RPN_AND);
 480             tmp->u.p[0] = p_top;
 481             tmp->u.p[1] = p;
 482
 483             p_top = tmp;
 484         }
 485         else
 486             p_top = p;
 487
 488
 489         if (left_trunc && right_trunc)
 490         {
 491             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH,
 492                                 &attset))
 493             {
 494                 cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
 495                 ccl_rpn_delete(p);
 496                 return NULL;
 497             }
 498             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 3);
 499         }
 500         else if (right_trunc)
 501         {
 502             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT,
 503                                  &attset))
 504             {
 505                 cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
 506                 ccl_rpn_delete(p);
 507                 return NULL;
 508             }
 509             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 1);
 510         }
 511         else if (left_trunc)
 512         {
 513             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT,
 514                                 &attset))
 515             {
 516                 cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
 517                 ccl_rpn_delete(p);
 518                 return NULL;
 519             }
 520             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 2);
 521         }
 522         else if (regex_trunc == 2)
 523         {
 524             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 102);
 525         }
 526         else if (z3958_trunc == 2)
 527         {
 528             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 104);
 529         }
 530         else
 531         {
 532             if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE,
 533                                &attset))
 534                 ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 100);
 535         }
 536         if (!multi)
 537             break;
 538     }
 539     if (!p_top)
 540         cclp->error_code = CCL_ERR_TERM_EXPECTED;
 541     return p_top;
 542 }
 543
 544 static struct ccl_rpn_node *search_term(CCL_parser cclp, ccl_qualifier_t *qa)
 545 {
 546     static int list[] = {CCL_TOK_TERM, CCL_TOK_COMMA, -1};
 547     return search_term_x(cclp, qa, list, 0);
 548 }
 549
 550 static
 551 struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
 552                                       ccl_qualifier_t *ap, char *attset)
 553 {
 554     int rel = 0;
 555     struct ccl_rpn_node *p;
 556
 557     if (cclp->look_token->len == 1)
 558     {
 559         if (cclp->look_token->name[0] == '<')
 560             rel = 1;
 561         else if (cclp->look_token->name[0] == '=')
 562             rel = 3;
 563         else if (cclp->look_token->name[0] == '>')
 564             rel = 5;
 565     }
 566     else if (cclp->look_token->len == 2)
 567     {
 568         if (!memcmp(cclp->look_token->name, "<=", 2))
 569             rel = 2;
 570         else if (!memcmp(cclp->look_token->name, ">=", 2))
 571             rel = 4;
 572         else if (!memcmp(cclp->look_token->name, "<>", 2))
 573             rel = 6;
 574     }
 575     if (!rel)
 576     {
 577         cclp->error_code = CCL_ERR_BAD_RELATION;
 578         return NULL;
 579     }
 580     ADVANCE;  /* skip relation */
 581     if (rel == 3 &&
 582         qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, 0))
 583     {
 584         /* allow - inside term and treat it as range _always_ */
 585         /* relation is =. Extract "embedded" - to separate terms */
 586         if (KIND == CCL_TOK_TERM)
 587         {
 588             size_t i;
 589             for (i = 0; i<cclp->look_token->len; i++)
 590             {
 591                 if (cclp->look_token->name[i] == '-')
 592                     break;
 593             }
 594
 595             if (cclp->look_token->len > 1 && i == 0)
 596             {   /*  -xx*/
 597                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 598
 599                 ntoken->kind = CCL_TOK_TERM;
 600                 ntoken->name = cclp->look_token->name + 1;
 601                 ntoken->len = cclp->look_token->len - 1;
 602
 603                 cclp->look_token->len = 1;
 604                 cclp->look_token->name = "-";
 605             }
 606             else if (cclp->look_token->len > 1 && i == cclp->look_token->len-1)
 607             {   /* xx- */
 608                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 609
 610                 ntoken->kind = CCL_TOK_TERM;
 611                 ntoken->name = "-";
 612                 ntoken->len = 1;
 613
 614                 (cclp->look_token->len)--;
 615             }
 616             else if (cclp->look_token->len > 2 && i < cclp->look_token->len)
 617             {   /* xx-yy */
 618                 struct ccl_token *ntoken1 = ccl_token_add(cclp->look_token);
 619                 struct ccl_token *ntoken2 = ccl_token_add(ntoken1);
 620
 621                 ntoken1->kind = CCL_TOK_TERM;  /* generate - */
 622                 ntoken1->name = "-";
 623                 ntoken1->len = 1;
 624
 625                 ntoken2->kind = CCL_TOK_TERM;  /* generate yy */
 626                 ntoken2->name = cclp->look_token->name + (i+1);
 627                 ntoken2->len = cclp->look_token->len - (i+1);
 628
 629                 cclp->look_token->len = i;     /* adjust xx */
 630             }
 631             else if (i == cclp->look_token->len &&
 632                      cclp->look_token->next &&
 633                      cclp->look_token->next->kind == CCL_TOK_TERM &&
 634                      cclp->look_token->next->len > 1 &&
 635                      cclp->look_token->next->name[0] == '-')
 636
 637             {   /* xx -yy */
 638                 /* we _know_ that xx does not have - in it */
 639                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 640
 641                 ntoken->kind = CCL_TOK_TERM;    /* generate - */
 642                 ntoken->name = "-";
 643                 ntoken->len = 1;
 644
 645                 (ntoken->next->name)++;        /* adjust yy */
 646                 (ntoken->next->len)--;
 647             }
 648         }
 649     }
 650
 651     if (rel == 3 &&
 652         KIND == CCL_TOK_TERM &&
 653         cclp->look_token->next && cclp->look_token->next->len == 1 &&
 654         cclp->look_token->next->name[0] == '-')
 655     {
 656         struct ccl_rpn_node *p1;
 657         if (!(p1 = search_term(cclp, ap)))
 658             return NULL;
 659         ADVANCE;                   /* skip '-' */
 660         if (KIND == CCL_TOK_TERM)  /* = term - term  ? */
 661         {
 662             struct ccl_rpn_node *p2;
 663
 664             if (!(p2 = search_term(cclp, ap)))
 665             {
 666                 ccl_rpn_delete(p1);
 667                 return NULL;
 668             }
 669             p = ccl_rpn_node_create(CCL_RPN_AND);
 670             p->u.p[0] = p1;
 671             ccl_add_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
 672             p->u.p[1] = p2;
 673             ccl_add_attr_numeric(p2, attset, CCL_BIB1_REL, 2);
 674             return p;
 675         }
 676         else                       /* = term -    */
 677         {
 678             ccl_add_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
 679             return p1;
 680         }
 681     }
 682     else if (rel == 3 &&
 683              cclp->look_token->len == 1 &&
 684              cclp->look_token->name[0] == '-')   /* = - term  ? */
 685     {
 686         ADVANCE;
 687         if (!(p = search_term(cclp, ap)))
 688             return NULL;
 689         ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, 2);
 690         return p;
 691     }
 692     else if (KIND == CCL_TOK_LP)
 693     {
 694         ADVANCE;
 695         if (!(p = find_spec(cclp, ap)))
 696             return NULL;
 697         if (KIND != CCL_TOK_RP)
 698         {
 699             cclp->error_code = CCL_ERR_RP_EXPECTED;
 700             ccl_rpn_delete(p);
 701             return NULL;
 702         }
 703         ADVANCE;
 704         return p;
 705     }
 706     else
 707     {
 708         if (!(p = search_terms(cclp, ap)))
 709             return NULL;
 710         ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, rel);
 711         return p;
 712     }
 713     cclp->error_code = CCL_ERR_TERM_EXPECTED;
 714     return NULL;
 715 }
 716
 717 static
 718 struct ccl_rpn_node *qualifier_relation(CCL_parser cclp, ccl_qualifier_t *ap)
 719 {
 720     char *attset;
 721     struct ccl_rpn_node *p;
 722
 723     if (qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset)
 724         || qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, &attset))
 725         return qualifiers_order(cclp, ap, attset);
 726
 727     /* unordered relation */
 728     if (KIND != CCL_TOK_EQ)
 729     {
 730         cclp->error_code = CCL_ERR_EQ_EXPECTED;
 731         return NULL;
 732     }
 733     ADVANCE;
 734     if (KIND == CCL_TOK_LP)
 735     {
 736         ADVANCE;
 737         if (!(p = find_spec(cclp, ap)))
 738         {
 739             return NULL;
 740         }
 741         if (KIND != CCL_TOK_RP)
 742         {
 743             cclp->error_code = CCL_ERR_RP_EXPECTED;
 744             ccl_rpn_delete(p);
 745             return NULL;
 746         }
 747         ADVANCE;
 748     }
 749     else
 750         p = search_terms(cclp, ap);
 751     return p;
 752 }
 753
 754 /**
 755  * qualifier_list: Parse CCL qualifiers and search terms.
 756  * cclp:   CCL Parser
 757  * la:     Token pointer to RELATION token.
 758  * qa:     Qualifier attributes already applied.
 759  * return: pointer to node(s); NULL on error.
 760  */
 761 static struct ccl_rpn_node *qualifier_list(CCL_parser cclp,
 762                                            struct ccl_token *la,
 763                                            ccl_qualifier_t *qa)
 764 {
 765     struct ccl_token *lookahead = cclp->look_token;
 766     struct ccl_token *look_start = cclp->look_token;
 767     ccl_qualifier_t *ap;
 768     struct ccl_rpn_node *node = 0;
 769     const char **field_str;
 770     int no = 0;
 771     int seq = 0;
 772     int i;
 773     int mode_merge = 1;
 774 #if 0
 775     if (qa)
 776     {
 777         cclp->error_code = CCL_ERR_DOUBLE_QUAL;
 778         return NULL;
 779     }
 780 #endif
 781     for (lookahead = cclp->look_token; lookahead != la;
 782          lookahead=lookahead->next)
 783         no++;
 784     if (qa)
 785         for (i=0; qa[i]; i++)
 786             no++;
 787     ap = (ccl_qualifier_t *)xmalloc((no ? (no+1) : 2) * sizeof(*ap));
 788     ccl_assert(ap);
 789
 790     field_str = ccl_qual_search_special(cclp->bibset, "field");
 791     if (field_str)
 792     {
 793         if (!strcmp(field_str[0], "or"))
 794             mode_merge = 0;
 795         else if (!strcmp(field_str[0], "merge"))
 796             mode_merge = 1;
 797     }
 798     if (!mode_merge)
 799     {
 800         /* consider each field separately and OR */
 801         lookahead = look_start;
 802         while (lookahead != la)
 803         {
 804             ap[1] = 0;
 805             seq = 0;
 806             while ((ap[0] = ccl_qual_search(cclp, lookahead->name,
 807                                             lookahead->len, seq)) != 0)
 808             {
 809                 struct ccl_rpn_node *node_sub;
 810                 cclp->look_token = la;
 811
 812                 node_sub = qualifier_relation(cclp, ap);
 813                 if (!node_sub)
 814                 {
 815                     ccl_rpn_delete(node);
 816                     xfree(ap);
 817                     return 0;
 818                 }
 819                 if (node)
 820                 {
 821                     struct ccl_rpn_node *node_this =
 822                         ccl_rpn_node_create(CCL_RPN_OR);
 823                     node_this->u.p[0] = node;
 824                     node_this->u.p[1] = node_sub;
 825                     node = node_this;
 826                 }
 827                 else
 828                     node = node_sub;
 829                 seq++;
 830             }
 831             if (seq == 0)
 832             {
 833                 cclp->look_token = lookahead;
 834                 cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
 835                 xfree(ap);
 836                 return NULL;
 837             }
 838             lookahead = lookahead->next;
 839             if (lookahead->kind == CCL_TOK_COMMA)
 840                 lookahead = lookahead->next;
 841         }
 842     }
 843     else
 844     {
 845         /* merge attributes from ALL fields - including inherited ones */
 846         while (1)
 847         {
 848             struct ccl_rpn_node *node_sub;
 849             int found = 0;
 850             lookahead = look_start;
 851             for (i = 0; lookahead != la; i++)
 852             {
 853                 ap[i] = ccl_qual_search(cclp, lookahead->name,
 854                                          lookahead->len, seq);
 855                 if (ap[i])
 856                     found++;
 857                 if (!ap[i] && seq > 0)
 858                     ap[i] = ccl_qual_search(cclp, lookahead->name,
 859                                              lookahead->len, 0);
 860                 if (!ap[i])
 861                 {
 862                     cclp->look_token = lookahead;
 863                     cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
 864                     xfree(ap);
 865                     return NULL;
 866                 }
 867                 lookahead = lookahead->next;
 868                 if (lookahead->kind == CCL_TOK_COMMA)
 869                     lookahead = lookahead->next;
 870             }
 871             if (qa)
 872             {
 873                 ccl_qualifier_t *qa0 = qa;
 874
 875                 while (*qa0)
 876                     ap[i++] = *qa0++;
 877             }
 878             ap[i] = NULL;
 879
 880             if (!found)
 881                 break;
 882
 883             cclp->look_token = lookahead;
 884
 885             node_sub = qualifier_relation(cclp, ap);
 886             if (!node_sub)
 887             {
 888                 ccl_rpn_delete(node);
 889                 break;
 890             }
 891             if (node)
 892             {
 893                 struct ccl_rpn_node *node_this =
 894                     ccl_rpn_node_create(CCL_RPN_OR);
 895                 node_this->u.p[0] = node;
 896                 node_this->u.p[1] = node_sub;
 897                 node = node_this;
 898             }
 899             else
 900                 node = node_sub;
 901             seq++;
 902         }
 903     }
 904     xfree(ap);
 905     return node;
 906 }
 907
 908
 909 /**
 910  * search_terms: Parse CCL search terms - including proximity.
 911  * cclp:   CCL Parser
 912  * qa:     Qualifier attributes already applied.
 913  * return: pointer to node(s); NULL on error.
 914  */
 915 static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa)
 916 {
 917     static int list[] = {
 918         CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ, CCL_TOK_REL, CCL_TOK_SET, -1};
 919     struct ccl_rpn_node *p1, *p2, *pn;
 920     p1 = search_term_x(cclp, qa, list, 1);
 921     if (!p1)
 922         return NULL;
 923     while (1)
 924     {
 925         if (KIND == CCL_TOK_PROX)
 926         {
 927             struct ccl_rpn_node *p_prox = 0;
 928             /* ! word order specified */
 929             /* % word order not specified */
 930             p_prox = ccl_rpn_node_create(CCL_RPN_TERM);
 931             p_prox->u.t.term = (char *) xmalloc(1 + cclp->look_token->len);
 932             memcpy(p_prox->u.t.term, cclp->look_token->name,
 933                    cclp->look_token->len);
 934             p_prox->u.t.term[cclp->look_token->len] = 0;
 935             p_prox->u.t.attr_list = 0;
 936
 937             ADVANCE;
 938             p2 = search_term_x(cclp, qa, list, 1);
 939             if (!p2)
 940             {
 941                 ccl_rpn_delete(p1);
 942                 return NULL;
 943             }
 944             pn = ccl_rpn_node_create(CCL_RPN_PROX);
 945             pn->u.p[0] = p1;
 946             pn->u.p[1] = p2;
 947             pn->u.p[2] = p_prox;
 948             p1 = pn;
 949         }
 950         else if (is_term_ok(KIND, list))
 951         {
 952             p2 = search_term_x(cclp, qa, list, 1);
 953             if (!p2)
 954             {
 955                 ccl_rpn_delete(p1);
 956                 return NULL;
 957             }
 958             pn = ccl_rpn_node_create(CCL_RPN_PROX);
 959             pn->u.p[0] = p1;
 960             pn->u.p[1] = p2;
 961             pn->u.p[2] = 0;
 962             p1 = pn;
 963         }
 964         else
 965             break;
 966     }
 967     return p1;
 968 }
 969
 970 /**
 971  * search_elements: Parse CCL search elements
 972  * cclp:   CCL Parser
 973  * qa:     Qualifier attributes already applied.
 974  * return: pointer to node(s); NULL on error.
 975  */
 976 static struct ccl_rpn_node *search_elements(CCL_parser cclp,
 977                                             ccl_qualifier_t *qa)
 978 {
 979     struct ccl_rpn_node *p1;
 980     struct ccl_token *lookahead;
 981     if (KIND == CCL_TOK_LP)
 982     {
 983         ADVANCE;
 984         p1 = find_spec(cclp, qa);
 985         if (!p1)
 986             return NULL;
 987         if (KIND != CCL_TOK_RP)
 988         {
 989             cclp->error_code = CCL_ERR_RP_EXPECTED;
 990             ccl_rpn_delete(p1);
 991             return NULL;
 992         }
 993         ADVANCE;
 994         return p1;
 995     }
 996     else if (KIND == CCL_TOK_SET)
 997     {
 998         ADVANCE;
 999         if (KIND == CCL_TOK_EQ)
1000             ADVANCE;
1001         if (KIND != CCL_TOK_TERM)
1002         {
1003             cclp->error_code = CCL_ERR_SETNAME_EXPECTED;
1004             return NULL;
1005         }
1006         p1 = ccl_rpn_node_create(CCL_RPN_SET);
1007         p1->u.setname = copy_token_name(cclp->look_token);
1008         ADVANCE;
1009         return p1;
1010     }
1011     lookahead = cclp->look_token;
1012
1013     while (lookahead->kind==CCL_TOK_TERM)
1014     {
1015         lookahead = lookahead->next;
1016         if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
1017             return qualifier_list(cclp, lookahead, qa);
1018         if (lookahead->kind != CCL_TOK_COMMA)
1019             break;
1020         lookahead = lookahead->next;
1021     }
1022     if (qa)
1023         return search_terms(cclp, qa);
1024     else
1025     {
1026         ccl_qualifier_t qa[2];
1027         struct ccl_rpn_node *node = 0;
1028         int seq;
1029         lookahead = cclp->look_token;
1030
1031         qa[1] = 0;
1032         for(seq = 0; ;seq++)
1033         {
1034             struct ccl_rpn_node *node_sub;
1035             qa[0] = ccl_qual_search(cclp, "term", 4, seq);
1036             if (!qa[0])
1037                 break;
1038
1039             cclp->look_token = lookahead;
1040
1041             node_sub = search_terms(cclp, qa);
1042             if (!node_sub)
1043             {
1044                 ccl_rpn_delete(node);
1045                 return 0;
1046             }
1047             if (node)
1048             {
1049                 struct ccl_rpn_node *node_this =
1050                     ccl_rpn_node_create(CCL_RPN_OR);
1051                 node_this->u.p[0] = node;
1052                 node_this->u.p[1] = node_sub;
1053                 node_this->u.p[2] = 0;
1054                 node = node_this;
1055             }
1056             else
1057                 node = node_sub;
1058         }
1059         if (!node)
1060             node = search_terms(cclp, 0);
1061         return node;
1062     }
1063 }
1064
1065 /**
1066  * find_spec: Parse CCL find specification
1067  * cclp:   CCL Parser
1068  * qa:     Qualifier attributes already applied.
1069  * return: pointer to node(s); NULL on error.
1070  */
1071 static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa)
1072 {
1073     struct ccl_rpn_node *p1, *p2, *pn;
1074     if (!(p1 = search_elements(cclp, qa)))
1075         return NULL;
1076     while (1)
1077     {
1078         switch (KIND)
1079         {
1080         case CCL_TOK_AND:
1081             ADVANCE;
1082             p2 = search_elements(cclp, qa);
1083             if (!p2)
1084             {
1085                 ccl_rpn_delete(p1);
1086                 return NULL;
1087             }
1088             pn = ccl_rpn_node_create(CCL_RPN_AND);
1089             pn->u.p[0] = p1;
1090             pn->u.p[1] = p2;
1091             pn->u.p[2] = 0;
1092             p1 = pn;
1093             continue;
1094         case CCL_TOK_OR:
1095             ADVANCE;
1096             p2 = search_elements(cclp, qa);
1097             if (!p2)
1098             {
1099                 ccl_rpn_delete(p1);
1100                 return NULL;
1101             }
1102             pn = ccl_rpn_node_create(CCL_RPN_OR);
1103             pn->u.p[0] = p1;
1104             pn->u.p[1] = p2;
1105             pn->u.p[2] = 0;
1106             p1 = pn;
1107             continue;
1108         case CCL_TOK_NOT:
1109             ADVANCE;
1110             p2 = search_elements(cclp, qa);
1111             if (!p2)
1112             {
1113                 ccl_rpn_delete(p1);
1114                 return NULL;
1115             }
1116             pn = ccl_rpn_node_create(CCL_RPN_NOT);
1117             pn->u.p[0] = p1;
1118             pn->u.p[1] = p2;
1119             pn->u.p[2] = 0;
1120             p1 = pn;
1121             continue;
1122         }
1123         break;
1124     }
1125     return p1;
1126 }
1127
1128 struct ccl_rpn_node *ccl_parser_find_str(CCL_parser cclp, const char *str)
1129 {
1130     struct ccl_rpn_node *p;
1131     struct ccl_token *list = ccl_parser_tokenize(cclp, str);
1132     p = ccl_parser_find_token(cclp, list);
1133     ccl_token_del(list);
1134     return p;
1135 }
1136
1137 struct ccl_rpn_node *ccl_parser_find_token(CCL_parser cclp,
1138                                            struct ccl_token *list)
1139 {
1140     struct ccl_rpn_node *p;
1141
1142     cclp->look_token = list;
1143     p = find_spec(cclp, NULL);
1144     if (p && KIND != CCL_TOK_EOL)
1145     {
1146         if (KIND == CCL_TOK_RP)
1147             cclp->error_code = CCL_ERR_BAD_RP;
1148         else
1149             cclp->error_code = CCL_ERR_OP_EXPECTED;
1150         ccl_rpn_delete(p);
1151         p = NULL;
1152     }
1153     cclp->error_pos = cclp->look_token->name;
1154     if (p)
1155         cclp->error_code = CCL_ERR_OK;
1156     else
1157         cclp->error_code = cclp->error_code;
1158     return p;
1159 }
1160
1161 /**
1162  * ccl_find_str: Parse CCL find - string representation
1163  * bibset:  Bibset to be used for the parsing
1164  * str:     String to be parsed
1165  * error:   Pointer to integer. Holds error no. on completion.
1166  * pos:     Pointer to char position. Holds approximate error position.
1167  * return:  RPN tree on successful completion; NULL otherwise.
1168  */
1169 struct ccl_rpn_node *ccl_find_str(CCL_bibset bibset, const char *str,
1170                                   int *error, int *pos)
1171 {
1172     CCL_parser cclp = ccl_parser_create(bibset);
1173     struct ccl_token *list;
1174     struct ccl_rpn_node *p;
1175
1176     list = ccl_parser_tokenize(cclp, str);
1177     p = ccl_parser_find_token(cclp, list);
1178
1179     *error = cclp->error_code;
1180     if (*error)
1181         *pos = cclp->error_pos - str;
1182     ccl_parser_destroy(cclp);
1183     ccl_token_del(list);
1184     return p;
1185 }
1186
1187 /*
1188  * Local variables:
1189  * c-basic-offset: 4
1190  * c-file-style: "Stroustrup"
1191  * indent-tabs-mode: nil
1192  * End:
1193  * vim: shiftwidth=4 tabstop=8 expandtab
1194  */
1195