src/cclfind.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2012 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 static size_t cmp_operator(const char **aliases, const char *input)
 216 {
 217     for (; *aliases; aliases++)
 218     {
 219         const char *cp = *aliases;
 220         size_t i;
 221         for (i = 0; *cp && *cp == input[i]; i++, cp++)
 222             ;
 223         if (*cp == '\0')
 224             return i;
 225     }
 226     return 0;
 227 }
 228
 229 #define REGEX_CHARS "^[]{}()|.*+?!$"
 230 #define CCL_CHARS "#?\\"
 231 /**
 232  * search_term: Parse CCL search term.
 233  * cclp:   CCL Parser
 234  * qa:     Qualifier attributes already applied.
 235  * term_list: tokens we accept as terms in context
 236  * multi:  whether we accept "multiple" tokens
 237  * return: pointer to node(s); NULL on error.
 238  */
 239 static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
 240                                           ccl_qualifier_t *qa,
 241                                           int *term_list, int multi)
 242 {
 243     struct ccl_rpn_node *p_top = 0;
 244     struct ccl_token *lookahead = cclp->look_token;
 245     int and_list = 0;
 246     int or_list = 0;
 247     char *attset;
 248     const char **truncation_aliases;
 249     const char *t_default[2];
 250
 251     truncation_aliases =
 252         ccl_qual_search_special(cclp->bibset, "truncation");
 253     if (!truncation_aliases)
 254     {
 255         truncation_aliases = t_default;
 256         t_default[0] = "?";
 257         t_default[1] = 0;
 258     }
 259
 260     if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_AND_LIST, 0))
 261         and_list = 1;
 262     if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_OR_LIST, 0))
 263         or_list = 1;
 264     while (1)
 265     {
 266         struct ccl_rpn_node *p;
 267         size_t no, i;
 268         int no_spaces = 0;
 269         int relation_value = -1;
 270         int position_value = -1;
 271         int structure_value = -1;
 272         int truncation_value = -1;
 273         int completeness_value = -1;
 274         int len = 0;
 275         int left_trunc = 0;
 276         int right_trunc = 0;
 277         int regex_trunc = 0;
 278         int z3958_trunc = 0;
 279         size_t max = 200;
 280         if (and_list || or_list || !multi)
 281             max = 1;
 282
 283         /* ignore commas when dealing with and-lists .. */
 284         if (and_list && lookahead && lookahead->kind == CCL_TOK_COMMA)
 285         {
 286             lookahead = lookahead->next;
 287             ADVANCE;
 288             continue;
 289         }
 290         /* go through each TERM token. If no truncation attribute is yet
 291            met, then look for left/right truncation markers (?) and
 292            set left_trunc/right_trunc/mid_trunc accordingly */
 293         for (no = 0; no < max && is_term_ok(lookahead->kind, term_list); no++)
 294         {
 295             for (i = 0; i<lookahead->len; i++)
 296                 if (lookahead->name[i] == ' ')
 297                     no_spaces++;
 298             len += 1+lookahead->len+lookahead->ws_prefix_len;
 299             lookahead = lookahead->next;
 300         }
 301
 302         if (len == 0)
 303             break;      /* no more terms . stop . */
 304
 305         /* create the term node, but wait a moment before adding the term */
 306         p = ccl_rpn_node_create(CCL_RPN_TERM);
 307         p->u.t.attr_list = NULL;
 308         p->u.t.term = NULL;
 309         if (qa && qa[0])
 310         {
 311             const char *n = ccl_qual_get_name(qa[0]);
 312             if (n)
 313                 p->u.t.qual = xstrdup(n);
 314         }
 315
 316         /* go through all attributes and add them to the attribute list */
 317         for (i=0; qa && qa[i]; i++)
 318         {
 319             struct ccl_rpn_attr *attr;
 320
 321             for (attr = ccl_qual_get_attr(qa[i]); attr; attr = attr->next)
 322                 switch(attr->kind)
 323                 {
 324                 case CCL_RPN_ATTR_STRING:
 325                     ccl_add_attr_string(p, attr->set, attr->type,
 326                                         attr->value.str);
 327                     break;
 328                 case CCL_RPN_ATTR_NUMERIC:
 329                     if (attr->value.numeric > 0)
 330                     {   /* deal only with REAL attributes (positive) */
 331                         switch (attr->type)
 332                         {
 333                         case CCL_BIB1_REL:
 334                             if (relation_value != -1)
 335                                 continue;
 336                             relation_value = attr->value.numeric;
 337                             break;
 338                         case CCL_BIB1_POS:
 339                             if (position_value != -1)
 340                                 continue;
 341                             position_value = attr->value.numeric;
 342                             break;
 343                         case CCL_BIB1_STR:
 344                             if (structure_value != -1)
 345                                 continue;
 346                             structure_value = attr->value.numeric;
 347                             break;
 348                         case CCL_BIB1_TRU:
 349                             if (truncation_value != -1)
 350                                 continue;
 351                             truncation_value = attr->value.numeric;
 352                             break;
 353                         case CCL_BIB1_COM:
 354                             if (completeness_value != -1)
 355                                 continue;
 356                             completeness_value = attr->value.numeric;
 357                             break;
 358                         }
 359                         ccl_add_attr_numeric(p, attr->set, attr->type,
 360                                              attr->value.numeric);
 361                     }
 362                 }
 363         }
 364         /* len now holds the number of characters in the RPN term */
 365         /* no holds the number of CCL tokens (1 or more) */
 366
 367         if (structure_value == -1 &&
 368             qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_WP, &attset))
 369         {   /* no structure attribute met. Apply either structure attribute
 370                WORD or PHRASE depending on number of CCL tokens */
 371             if (no == 1 && no_spaces == 0)
 372                 ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 2);
 373             else
 374                 ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 1);
 375         }
 376
 377         if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_REGEX,
 378                           &attset))
 379         {
 380             regex_trunc = 1; /* regex trunc (102) allowed */
 381         }
 382         else if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_Z3958,
 383                           &attset))
 384         {
 385             z3958_trunc = 1; /* Z39.58 trunc (CCL) trunc allowed */
 386         }
 387
 388         /* make the RPN token */
 389         p->u.t.term = (char *)xmalloc(len * 2 + 2);
 390         ccl_assert(p->u.t.term);
 391         p->u.t.term[0] = '\0';
 392         for (i = 0; i<no; i++)
 393         {
 394             const char *src_str = cclp->look_token->name;
 395             size_t src_len = cclp->look_token->len;
 396             int j;
 397             int quote_mode = 0;
 398
 399             if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
 400             {
 401                 strxcat(p->u.t.term, cclp->look_token->ws_prefix_buf,
 402                         cclp->look_token->ws_prefix_len);
 403             }
 404             for (j = 0; j < src_len; j++)
 405             {
 406                 size_t op_size;
 407                 if (j > 0 && src_str[j-1] == '\\')
 408                 {
 409                     if (regex_trunc && strchr(REGEX_CHARS "\\", src_str[j]))
 410                     {
 411                         regex_trunc = 2;
 412                         strcat(p->u.t.term, "\\");
 413                     }
 414                     else if (z3958_trunc && strchr(CCL_CHARS "\\", src_str[j]))
 415                     {
 416                         z3958_trunc = 2;
 417                         strcat(p->u.t.term, "\\");
 418                     }
 419                     strxcat(p->u.t.term, src_str + j, 1);
 420                 }
 421                 else if (src_str[j] == '"')
 422                     quote_mode = !quote_mode;
 423                 else if (!quote_mode &&
 424                          (op_size = cmp_operator(truncation_aliases,
 425                                                  src_str + j))
 426                     )
 427                 {
 428                     j += (op_size - 1);  /* j++ in for loop */
 429                     if (regex_trunc)
 430                     {
 431                         strcat(p->u.t.term, ".*");
 432                         regex_trunc = 2; /* regex trunc is really needed */
 433                     }
 434                     else if (z3958_trunc)
 435                     {
 436                         strcat(p->u.t.term, "?");
 437                         z3958_trunc = 2;
 438                     }
 439                     else if (i == 0 && j == 0)
 440                         left_trunc = 1;
 441                     else if (i == no - 1 && j == src_len - 1)
 442                         right_trunc = 1;
 443                     else
 444                     {
 445                         cclp->error_code = CCL_ERR_TRUNC_NOT_EMBED;
 446                         ccl_rpn_delete(p);
 447                         return NULL;
 448                     }
 449                 }
 450                 else if (!quote_mode && src_str[j] == '#')
 451                 {
 452                     if (regex_trunc)
 453                     {
 454                         strcat(p->u.t.term, ".");
 455                         regex_trunc = 2; /* regex trunc is really needed */
 456                     }
 457                     else if (z3958_trunc)
 458                     {
 459                         strcat(p->u.t.term, "#");
 460                         z3958_trunc = 2;
 461                     }
 462                     else
 463                     {
 464                         cclp->error_code = CCL_ERR_TRUNC_NOT_SINGLE;
 465                         ccl_rpn_delete(p);
 466                         return NULL;
 467                     }
 468                 }
 469                 else if (src_str[j] != '\\')
 470                 {
 471                     if (regex_trunc && strchr(REGEX_CHARS, src_str[j]))
 472                     {
 473                         regex_trunc = 2;
 474                         strcat(p->u.t.term, "\\");
 475                     }
 476                     else if (z3958_trunc && strchr(CCL_CHARS, src_str[j]))
 477                     {
 478                         z3958_trunc = 2;
 479                         strcat(p->u.t.term, "\\");
 480                     }
 481                     strxcat(p->u.t.term, src_str + j, 1);
 482                 }
 483             }
 484             ADVANCE;
 485         }
 486
 487         /* make the top node point to us.. */
 488         if (p_top)
 489         {
 490             struct ccl_rpn_node *tmp;
 491
 492             if (or_list)
 493                 tmp = ccl_rpn_node_create(CCL_RPN_OR);
 494             else if (and_list)
 495                 tmp = ccl_rpn_node_create(CCL_RPN_AND);
 496             else
 497                 tmp = ccl_rpn_node_create(CCL_RPN_AND);
 498             tmp->u.p[0] = p_top;
 499             tmp->u.p[1] = p;
 500
 501             p_top = tmp;
 502         }
 503         else
 504             p_top = p;
 505
 506
 507         if (left_trunc && right_trunc)
 508         {
 509             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH,
 510                                 &attset))
 511             {
 512                 cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
 513                 ccl_rpn_delete(p);
 514                 return NULL;
 515             }
 516             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 3);
 517         }
 518         else if (right_trunc)
 519         {
 520             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT,
 521                                  &attset))
 522             {
 523                 cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
 524                 ccl_rpn_delete(p);
 525                 return NULL;
 526             }
 527             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 1);
 528         }
 529         else if (left_trunc)
 530         {
 531             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT,
 532                                 &attset))
 533             {
 534                 cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
 535                 ccl_rpn_delete(p);
 536                 return NULL;
 537             }
 538             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 2);
 539         }
 540         else if (regex_trunc == 2)
 541         {
 542             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 102);
 543         }
 544         else if (z3958_trunc == 2)
 545         {
 546             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 104);
 547         }
 548         else
 549         {
 550             if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE,
 551                                &attset))
 552                 ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 100);
 553         }
 554         if (!multi)
 555             break;
 556     }
 557     if (!p_top)
 558         cclp->error_code = CCL_ERR_TERM_EXPECTED;
 559     return p_top;
 560 }
 561
 562 static struct ccl_rpn_node *search_term(CCL_parser cclp, ccl_qualifier_t *qa)
 563 {
 564     static int list[] = {CCL_TOK_TERM, CCL_TOK_COMMA, -1};
 565     return search_term_x(cclp, qa, list, 0);
 566 }
 567
 568
 569 static struct ccl_rpn_node *search_terms2(CCL_parser cclp,
 570                                           ccl_qualifier_t *qa)
 571 {
 572     if (KIND == CCL_TOK_LP)
 573     {
 574         struct ccl_rpn_node *p;
 575         ADVANCE;
 576         if (!(p = find_spec(cclp, qa)))
 577             return NULL;
 578         if (KIND != CCL_TOK_RP)
 579         {
 580             cclp->error_code = CCL_ERR_RP_EXPECTED;
 581             ccl_rpn_delete(p);
 582             return NULL;
 583         }
 584         ADVANCE;
 585         return p;
 586     }
 587     else
 588     {
 589         static int list[] = {
 590             CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
 591             CCL_TOK_REL, CCL_TOK_SET, -1};
 592
 593         return search_term_x(cclp, qa, list, 1);
 594     }
 595 }
 596
 597
 598
 599 static
 600 struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
 601                                       ccl_qualifier_t *ap, char *attset)
 602 {
 603     int rel = 0;
 604     struct ccl_rpn_node *p;
 605
 606     if (cclp->look_token->len == 1)
 607     {
 608         if (cclp->look_token->name[0] == '<')
 609             rel = 1;
 610         else if (cclp->look_token->name[0] == '=')
 611             rel = 3;
 612         else if (cclp->look_token->name[0] == '>')
 613             rel = 5;
 614     }
 615     else if (cclp->look_token->len == 2)
 616     {
 617         if (!memcmp(cclp->look_token->name, "<=", 2))
 618             rel = 2;
 619         else if (!memcmp(cclp->look_token->name, ">=", 2))
 620             rel = 4;
 621         else if (!memcmp(cclp->look_token->name, "<>", 2))
 622             rel = 6;
 623     }
 624     if (!rel)
 625     {
 626         cclp->error_code = CCL_ERR_BAD_RELATION;
 627         return NULL;
 628     }
 629     ADVANCE;  /* skip relation */
 630     if (rel == 3 &&
 631         qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, 0))
 632     {
 633         /* allow - inside term and treat it as range _always_ */
 634         /* relation is =. Extract "embedded" - to separate terms */
 635         if (KIND == CCL_TOK_TERM)
 636         {
 637             size_t i;
 638             for (i = 0; i<cclp->look_token->len; i++)
 639             {
 640                 if (cclp->look_token->name[i] == '-')
 641                     break;
 642             }
 643
 644             if (cclp->look_token->len > 1 && i == 0)
 645             {   /*  -xx*/
 646                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 647
 648                 ntoken->kind = CCL_TOK_TERM;
 649                 ntoken->name = cclp->look_token->name + 1;
 650                 ntoken->len = cclp->look_token->len - 1;
 651
 652                 cclp->look_token->len = 1;
 653                 cclp->look_token->name = "-";
 654             }
 655             else if (cclp->look_token->len > 1 && i == cclp->look_token->len-1)
 656             {   /* xx- */
 657                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 658
 659                 ntoken->kind = CCL_TOK_TERM;
 660                 ntoken->name = "-";
 661                 ntoken->len = 1;
 662
 663                 (cclp->look_token->len)--;
 664             }
 665             else if (cclp->look_token->len > 2 && i < cclp->look_token->len)
 666             {   /* xx-yy */
 667                 struct ccl_token *ntoken1 = ccl_token_add(cclp->look_token);
 668                 struct ccl_token *ntoken2 = ccl_token_add(ntoken1);
 669
 670                 ntoken1->kind = CCL_TOK_TERM;  /* generate - */
 671                 ntoken1->name = "-";
 672                 ntoken1->len = 1;
 673
 674                 ntoken2->kind = CCL_TOK_TERM;  /* generate yy */
 675                 ntoken2->name = cclp->look_token->name + (i+1);
 676                 ntoken2->len = cclp->look_token->len - (i+1);
 677
 678                 cclp->look_token->len = i;     /* adjust xx */
 679             }
 680             else if (i == cclp->look_token->len &&
 681                      cclp->look_token->next &&
 682                      cclp->look_token->next->kind == CCL_TOK_TERM &&
 683                      cclp->look_token->next->len > 1 &&
 684                      cclp->look_token->next->name[0] == '-')
 685
 686             {   /* xx -yy */
 687                 /* we _know_ that xx does not have - in it */
 688                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 689
 690                 ntoken->kind = CCL_TOK_TERM;    /* generate - */
 691                 ntoken->name = "-";
 692                 ntoken->len = 1;
 693
 694                 (ntoken->next->name)++;        /* adjust yy */
 695                 (ntoken->next->len)--;
 696             }
 697         }
 698     }
 699
 700     if (rel == 3 &&
 701         KIND == CCL_TOK_TERM &&
 702         cclp->look_token->next && cclp->look_token->next->len == 1 &&
 703         cclp->look_token->next->name[0] == '-')
 704     {
 705         struct ccl_rpn_node *p1;
 706         if (!(p1 = search_term(cclp, ap)))
 707             return NULL;
 708         ADVANCE;                   /* skip '-' */
 709         if (KIND == CCL_TOK_TERM)  /* = term - term  ? */
 710         {
 711             struct ccl_rpn_node *p2;
 712
 713             if (!(p2 = search_term(cclp, ap)))
 714             {
 715                 ccl_rpn_delete(p1);
 716                 return NULL;
 717             }
 718             p = ccl_rpn_node_create(CCL_RPN_AND);
 719             p->u.p[0] = p1;
 720             ccl_add_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
 721             p->u.p[1] = p2;
 722             ccl_add_attr_numeric(p2, attset, CCL_BIB1_REL, 2);
 723             return p;
 724         }
 725         else                       /* = term -    */
 726         {
 727             ccl_add_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
 728             return p1;
 729         }
 730     }
 731     else if (rel == 3 &&
 732              cclp->look_token->len == 1 &&
 733              cclp->look_token->name[0] == '-')   /* = - term  ? */
 734     {
 735         ADVANCE;
 736         if (!(p = search_term(cclp, ap)))
 737             return NULL;
 738         ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, 2);
 739         return p;
 740     }
 741     else
 742     {
 743         if (!(p = search_terms(cclp, ap)))
 744             return NULL;
 745         ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, rel);
 746         return p;
 747     }
 748     cclp->error_code = CCL_ERR_TERM_EXPECTED;
 749     return NULL;
 750 }
 751
 752 static
 753 struct ccl_rpn_node *qualifier_relation(CCL_parser cclp, ccl_qualifier_t *ap)
 754 {
 755     char *attset;
 756
 757     if (qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset)
 758         || qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, &attset))
 759         return qualifiers_order(cclp, ap, attset);
 760
 761     /* unordered relation */
 762     if (KIND != CCL_TOK_EQ)
 763     {
 764         cclp->error_code = CCL_ERR_EQ_EXPECTED;
 765         return NULL;
 766     }
 767     ADVANCE;
 768     return search_terms(cclp, ap);
 769 }
 770
 771 /**
 772  * qualifier_list: Parse CCL qualifiers and search terms.
 773  * cclp:   CCL Parser
 774  * la:     Token pointer to RELATION token.
 775  * qa:     Qualifier attributes already applied.
 776  * return: pointer to node(s); NULL on error.
 777  */
 778 static struct ccl_rpn_node *qualifier_list(CCL_parser cclp,
 779                                            struct ccl_token *la,
 780                                            ccl_qualifier_t *qa)
 781 {
 782     struct ccl_token *lookahead = cclp->look_token;
 783     struct ccl_token *look_start = cclp->look_token;
 784     ccl_qualifier_t *ap;
 785     struct ccl_rpn_node *node = 0;
 786     const char **field_str;
 787     int no = 0;
 788     int seq = 0;
 789     int i;
 790     int mode_merge = 1;
 791 #if 0
 792     if (qa)
 793     {
 794         cclp->error_code = CCL_ERR_DOUBLE_QUAL;
 795         return NULL;
 796     }
 797 #endif
 798     for (lookahead = cclp->look_token; lookahead != la;
 799          lookahead=lookahead->next)
 800         no++;
 801     if (qa)
 802         for (i=0; qa[i]; i++)
 803             no++;
 804     ap = (ccl_qualifier_t *)xmalloc((no ? (no+1) : 2) * sizeof(*ap));
 805     ccl_assert(ap);
 806
 807     field_str = ccl_qual_search_special(cclp->bibset, "field");
 808     if (field_str)
 809     {
 810         if (!strcmp(field_str[0], "or"))
 811             mode_merge = 0;
 812         else if (!strcmp(field_str[0], "merge"))
 813             mode_merge = 1;
 814     }
 815     if (!mode_merge)
 816     {
 817         /* consider each field separately and OR */
 818         lookahead = look_start;
 819         while (lookahead != la)
 820         {
 821             ap[1] = 0;
 822             seq = 0;
 823             while ((ap[0] = ccl_qual_search(cclp, lookahead->name,
 824                                             lookahead->len, seq)) != 0)
 825             {
 826                 struct ccl_rpn_node *node_sub;
 827                 cclp->look_token = la;
 828
 829                 node_sub = qualifier_relation(cclp, ap);
 830                 if (!node_sub)
 831                 {
 832                     ccl_rpn_delete(node);
 833                     xfree(ap);
 834                     return 0;
 835                 }
 836                 if (node)
 837                 {
 838                     struct ccl_rpn_node *node_this =
 839                         ccl_rpn_node_create(CCL_RPN_OR);
 840                     node_this->u.p[0] = node;
 841                     node_this->u.p[1] = node_sub;
 842                     node = node_this;
 843                 }
 844                 else
 845                     node = node_sub;
 846                 seq++;
 847             }
 848             if (seq == 0)
 849             {
 850                 cclp->look_token = lookahead;
 851                 cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
 852                 xfree(ap);
 853                 return NULL;
 854             }
 855             lookahead = lookahead->next;
 856             if (lookahead->kind == CCL_TOK_COMMA)
 857                 lookahead = lookahead->next;
 858         }
 859     }
 860     else
 861     {
 862         /* merge attributes from ALL fields - including inherited ones */
 863         while (1)
 864         {
 865             struct ccl_rpn_node *node_sub;
 866             int found = 0;
 867             lookahead = look_start;
 868             for (i = 0; lookahead != la; i++)
 869             {
 870                 ap[i] = ccl_qual_search(cclp, lookahead->name,
 871                                          lookahead->len, seq);
 872                 if (ap[i])
 873                     found++;
 874                 if (!ap[i] && seq > 0)
 875                     ap[i] = ccl_qual_search(cclp, lookahead->name,
 876                                              lookahead->len, 0);
 877                 if (!ap[i])
 878                 {
 879                     cclp->look_token = lookahead;
 880                     cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
 881                     xfree(ap);
 882                     return NULL;
 883                 }
 884                 lookahead = lookahead->next;
 885                 if (lookahead->kind == CCL_TOK_COMMA)
 886                     lookahead = lookahead->next;
 887             }
 888             if (qa)
 889             {
 890                 ccl_qualifier_t *qa0 = qa;
 891
 892                 while (*qa0)
 893                     ap[i++] = *qa0++;
 894             }
 895             ap[i] = NULL;
 896
 897             if (!found)
 898                 break;
 899
 900             cclp->look_token = lookahead;
 901
 902             node_sub = qualifier_relation(cclp, ap);
 903             if (!node_sub)
 904             {
 905                 ccl_rpn_delete(node);
 906                 break;
 907             }
 908             if (node)
 909             {
 910                 struct ccl_rpn_node *node_this =
 911                     ccl_rpn_node_create(CCL_RPN_OR);
 912                 node_this->u.p[0] = node;
 913                 node_this->u.p[1] = node_sub;
 914                 node = node_this;
 915             }
 916             else
 917                 node = node_sub;
 918             seq++;
 919         }
 920     }
 921     xfree(ap);
 922     return node;
 923 }
 924
 925
 926 /**
 927  * search_terms: Parse CCL search terms - including proximity.
 928  * cclp:   CCL Parser
 929  * qa:     Qualifier attributes already applied.
 930  * return: pointer to node(s); NULL on error.
 931  */
 932 static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa)
 933 {
 934     static int list[] = {
 935         CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
 936         CCL_TOK_REL, CCL_TOK_SET, -1};
 937     struct ccl_rpn_node *p1, *p2, *pn;
 938     p1 = search_terms2(cclp, qa);
 939     if (!p1)
 940         return NULL;
 941     while (1)
 942     {
 943         if (KIND == CCL_TOK_PROX)
 944         {
 945             struct ccl_rpn_node *p_prox = 0;
 946             /* ! word order specified */
 947             /* % word order not specified */
 948             p_prox = ccl_rpn_node_create(CCL_RPN_TERM);
 949             p_prox->u.t.term = (char *) xmalloc(1 + cclp->look_token->len);
 950             memcpy(p_prox->u.t.term, cclp->look_token->name,
 951                    cclp->look_token->len);
 952             p_prox->u.t.term[cclp->look_token->len] = 0;
 953             p_prox->u.t.attr_list = 0;
 954
 955             ADVANCE;
 956             p2 = search_terms2(cclp, qa);
 957             if (!p2)
 958             {
 959                 ccl_rpn_delete(p1);
 960                 return NULL;
 961             }
 962             pn = ccl_rpn_node_create(CCL_RPN_PROX);
 963             pn->u.p[0] = p1;
 964             pn->u.p[1] = p2;
 965             pn->u.p[2] = p_prox;
 966             p1 = pn;
 967         }
 968         else if (is_term_ok(KIND, list))
 969         {
 970             p2 = search_terms2(cclp, qa);
 971             if (!p2)
 972             {
 973                 ccl_rpn_delete(p1);
 974                 return NULL;
 975             }
 976             pn = ccl_rpn_node_create(CCL_RPN_PROX);
 977             pn->u.p[0] = p1;
 978             pn->u.p[1] = p2;
 979             pn->u.p[2] = 0;
 980             p1 = pn;
 981         }
 982         else
 983             break;
 984     }
 985     return p1;
 986 }
 987
 988 /**
 989  * search_elements: Parse CCL search elements
 990  * cclp:   CCL Parser
 991  * qa:     Qualifier attributes already applied.
 992  * return: pointer to node(s); NULL on error.
 993  */
 994 static struct ccl_rpn_node *search_elements(CCL_parser cclp,
 995                                             ccl_qualifier_t *qa)
 996 {
 997     struct ccl_rpn_node *p1;
 998     struct ccl_token *lookahead;
 999     if (KIND == CCL_TOK_SET)
1000     {
1001         ADVANCE;
1002         if (KIND == CCL_TOK_EQ)
1003             ADVANCE;
1004         if (KIND != CCL_TOK_TERM)
1005         {
1006             cclp->error_code = CCL_ERR_SETNAME_EXPECTED;
1007             return NULL;
1008         }
1009         p1 = ccl_rpn_node_create(CCL_RPN_SET);
1010         p1->u.setname = copy_token_name(cclp->look_token);
1011         ADVANCE;
1012         return p1;
1013     }
1014     lookahead = cclp->look_token;
1015
1016     while (lookahead->kind==CCL_TOK_TERM)
1017     {
1018         lookahead = lookahead->next;
1019         if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
1020             return qualifier_list(cclp, lookahead, qa);
1021         if (lookahead->kind != CCL_TOK_COMMA)
1022             break;
1023         lookahead = lookahead->next;
1024     }
1025     if (qa || lookahead->kind == CCL_TOK_LP)
1026         return search_terms(cclp, qa);
1027     else
1028     {
1029         ccl_qualifier_t qa[2];
1030         struct ccl_rpn_node *node = 0;
1031         int seq;
1032         lookahead = cclp->look_token;
1033
1034         qa[1] = 0;
1035         for(seq = 0; ;seq++)
1036         {
1037             struct ccl_rpn_node *node_sub;
1038             qa[0] = ccl_qual_search(cclp, "term", 4, seq);
1039             if (!qa[0])
1040                 break;
1041
1042             cclp->look_token = lookahead;
1043
1044             node_sub = search_terms(cclp, qa);
1045             if (!node_sub)
1046             {
1047                 ccl_rpn_delete(node);
1048                 return 0;
1049             }
1050             if (node)
1051             {
1052                 struct ccl_rpn_node *node_this =
1053                     ccl_rpn_node_create(CCL_RPN_OR);
1054                 node_this->u.p[0] = node;
1055                 node_this->u.p[1] = node_sub;
1056                 node_this->u.p[2] = 0;
1057                 node = node_this;
1058             }
1059             else
1060                 node = node_sub;
1061         }
1062         if (!node)
1063             node = search_terms(cclp, 0);
1064         return node;
1065     }
1066 }
1067
1068 /**
1069  * find_spec: Parse CCL find specification
1070  * cclp:   CCL Parser
1071  * qa:     Qualifier attributes already applied.
1072  * return: pointer to node(s); NULL on error.
1073  */
1074 static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa)
1075 {
1076     struct ccl_rpn_node *p1, *p2, *pn;
1077     if (!(p1 = search_elements(cclp, qa)))
1078         return NULL;
1079     while (1)
1080     {
1081         switch (KIND)
1082         {
1083         case CCL_TOK_AND:
1084             ADVANCE;
1085             p2 = search_elements(cclp, qa);
1086             if (!p2)
1087             {
1088                 ccl_rpn_delete(p1);
1089                 return NULL;
1090             }
1091             pn = ccl_rpn_node_create(CCL_RPN_AND);
1092             pn->u.p[0] = p1;
1093             pn->u.p[1] = p2;
1094             pn->u.p[2] = 0;
1095             p1 = pn;
1096             continue;
1097         case CCL_TOK_OR:
1098             ADVANCE;
1099             p2 = search_elements(cclp, qa);
1100             if (!p2)
1101             {
1102                 ccl_rpn_delete(p1);
1103                 return NULL;
1104             }
1105             pn = ccl_rpn_node_create(CCL_RPN_OR);
1106             pn->u.p[0] = p1;
1107             pn->u.p[1] = p2;
1108             pn->u.p[2] = 0;
1109             p1 = pn;
1110             continue;
1111         case CCL_TOK_NOT:
1112             ADVANCE;
1113             p2 = search_elements(cclp, qa);
1114             if (!p2)
1115             {
1116                 ccl_rpn_delete(p1);
1117                 return NULL;
1118             }
1119             pn = ccl_rpn_node_create(CCL_RPN_NOT);
1120             pn->u.p[0] = p1;
1121             pn->u.p[1] = p2;
1122             pn->u.p[2] = 0;
1123             p1 = pn;
1124             continue;
1125         }
1126         break;
1127     }
1128     return p1;
1129 }
1130
1131 struct ccl_rpn_node *ccl_parser_find_str(CCL_parser cclp, const char *str)
1132 {
1133     struct ccl_rpn_node *p;
1134     struct ccl_token *list = ccl_parser_tokenize(cclp, str);
1135     p = ccl_parser_find_token(cclp, list);
1136     ccl_token_del(list);
1137     return p;
1138 }
1139
1140 struct ccl_rpn_node *ccl_parser_find_token(CCL_parser cclp,
1141                                            struct ccl_token *list)
1142 {
1143     struct ccl_rpn_node *p;
1144
1145     cclp->look_token = list;
1146     p = find_spec(cclp, NULL);
1147     if (p && KIND != CCL_TOK_EOL)
1148     {
1149         if (KIND == CCL_TOK_RP)
1150             cclp->error_code = CCL_ERR_BAD_RP;
1151         else
1152             cclp->error_code = CCL_ERR_OP_EXPECTED;
1153         ccl_rpn_delete(p);
1154         p = NULL;
1155     }
1156     cclp->error_pos = cclp->look_token->name;
1157     if (p)
1158         cclp->error_code = CCL_ERR_OK;
1159     else
1160         cclp->error_code = cclp->error_code;
1161     return p;
1162 }
1163
1164 /**
1165  * ccl_find_str: Parse CCL find - string representation
1166  * bibset:  Bibset to be used for the parsing
1167  * str:     String to be parsed
1168  * error:   Pointer to integer. Holds error no. on completion.
1169  * pos:     Pointer to char position. Holds approximate error position.
1170  * return:  RPN tree on successful completion; NULL otherwise.
1171  */
1172 struct ccl_rpn_node *ccl_find_str(CCL_bibset bibset, const char *str,
1173                                   int *error, int *pos)
1174 {
1175     CCL_parser cclp = ccl_parser_create(bibset);
1176     struct ccl_token *list;
1177     struct ccl_rpn_node *p;
1178
1179     list = ccl_parser_tokenize(cclp, str);
1180     p = ccl_parser_find_token(cclp, list);
1181
1182     *error = cclp->error_code;
1183     if (*error)
1184         *pos = cclp->error_pos - str;
1185     ccl_parser_destroy(cclp);
1186     ccl_token_del(list);
1187     return p;
1188 }
1189
1190 /*
1191  * Local variables:
1192  * c-basic-offset: 4
1193  * c-file-style: "Stroustrup"
1194  * indent-tabs-mode: nil
1195  * End:
1196  * vim: shiftwidth=4 tabstop=8 expandtab
1197  */
1198