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