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                 size_t len = strlen(p->u.t.term);
 389                 memcpy(p->u.t.term + len, cclp->look_token->ws_prefix_buf,
 390                        cclp->look_token->ws_prefix_len);
 391                 p->u.t.term[len + cclp->look_token->ws_prefix_len] = '\0';
 392             }
 393             for (j = 0; j < src_len; j++)
 394             {
 395                 if (j > 0 && src_str[j-1] == '\\')
 396                 {
 397                     if (regex_trunc && strchr(REGEX_CHARS "\\", src_str[j]))
 398                     {
 399                         regex_trunc = 2;
 400                         strcat(p->u.t.term, "\\");
 401                     }
 402                     else if (z3958_trunc && strchr(CCL_CHARS "\\", src_str[j]))
 403                     {
 404                         z3958_trunc = 2;
 405                         strcat(p->u.t.term, "\\");
 406                     }
 407                     strxcat(p->u.t.term, src_str + j, 1);
 408                 }
 409                 else if (src_str[j] == '"')
 410                     quote_mode = !quote_mode;
 411                 else if (!quote_mode && src_str[j] == '?')
 412                 {
 413                     if (regex_trunc)
 414                     {
 415                         strcat(p->u.t.term, ".*");
 416                         regex_trunc = 2; /* regex trunc is really needed */
 417                     }
 418                     else if (z3958_trunc)
 419                     {
 420                         strcat(p->u.t.term, "?");
 421                         z3958_trunc = 2;
 422                     }
 423                     else if (i == 0 && j == 0)
 424                         left_trunc = 1;
 425                     else if (i == no - 1 && j == src_len - 1)
 426                         right_trunc = 1;
 427                     else
 428                     {
 429                         cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
 430                         ccl_rpn_delete(p);
 431                         return NULL;
 432                     }
 433                 }
 434                 else if (!quote_mode && src_str[j] == '#')
 435                 {
 436                     if (regex_trunc)
 437                     {
 438                         strcat(p->u.t.term, ".");
 439                         regex_trunc = 2; /* regex trunc is really needed */
 440                     }
 441                     else if (z3958_trunc)
 442                     {
 443                         strcat(p->u.t.term, "#");
 444                         z3958_trunc = 2;
 445                     }
 446                     else
 447                     {
 448                         cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
 449                         ccl_rpn_delete(p);
 450                         return NULL;
 451                     }
 452                 }
 453                 else if (src_str[j] != '\\')
 454                 {
 455                     if (regex_trunc && strchr(REGEX_CHARS, src_str[j]))
 456                     {
 457                         regex_trunc = 2;
 458                         strcat(p->u.t.term, "\\");
 459                     }
 460                     else if (z3958_trunc && strchr(CCL_CHARS, src_str[j]))
 461                     {
 462                         z3958_trunc = 2;
 463                         strcat(p->u.t.term, "\\");
 464                     }
 465                     strxcat(p->u.t.term, src_str + j, 1);
 466                 }
 467             }
 468             ADVANCE;
 469         }
 470
 471         /* make the top node point to us.. */
 472         if (p_top)
 473         {
 474             struct ccl_rpn_node *tmp;
 475
 476             if (or_list)
 477                 tmp = ccl_rpn_node_create(CCL_RPN_OR);
 478             else if (and_list)
 479                 tmp = ccl_rpn_node_create(CCL_RPN_AND);
 480             else
 481                 tmp = ccl_rpn_node_create(CCL_RPN_AND);
 482             tmp->u.p[0] = p_top;
 483             tmp->u.p[1] = p;
 484
 485             p_top = tmp;
 486         }
 487         else
 488             p_top = p;
 489
 490
 491         if (left_trunc && right_trunc)
 492         {
 493             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH,
 494                                 &attset))
 495             {
 496                 cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
 497                 ccl_rpn_delete(p);
 498                 return NULL;
 499             }
 500             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 3);
 501         }
 502         else if (right_trunc)
 503         {
 504             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT,
 505                                  &attset))
 506             {
 507                 cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
 508                 ccl_rpn_delete(p);
 509                 return NULL;
 510             }
 511             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 1);
 512         }
 513         else if (left_trunc)
 514         {
 515             if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT,
 516                                 &attset))
 517             {
 518                 cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
 519                 ccl_rpn_delete(p);
 520                 return NULL;
 521             }
 522             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 2);
 523         }
 524         else if (regex_trunc == 2)
 525         {
 526             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 102);
 527         }
 528         else if (z3958_trunc == 2)
 529         {
 530             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 104);
 531         }
 532         else
 533         {
 534             if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE,
 535                                &attset))
 536                 ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 100);
 537         }
 538         if (!multi)
 539             break;
 540     }
 541     if (!p_top)
 542         cclp->error_code = CCL_ERR_TERM_EXPECTED;
 543     return p_top;
 544 }
 545
 546 static struct ccl_rpn_node *search_term(CCL_parser cclp, ccl_qualifier_t *qa)
 547 {
 548     static int list[] = {CCL_TOK_TERM, CCL_TOK_COMMA, -1};
 549     return search_term_x(cclp, qa, list, 0);
 550 }
 551
 552 static
 553 struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
 554                                       ccl_qualifier_t *ap, char *attset)
 555 {
 556     int rel = 0;
 557     struct ccl_rpn_node *p;
 558
 559     if (cclp->look_token->len == 1)
 560     {
 561         if (cclp->look_token->name[0] == '<')
 562             rel = 1;
 563         else if (cclp->look_token->name[0] == '=')
 564             rel = 3;
 565         else if (cclp->look_token->name[0] == '>')
 566             rel = 5;
 567     }
 568     else if (cclp->look_token->len == 2)
 569     {
 570         if (!memcmp(cclp->look_token->name, "<=", 2))
 571             rel = 2;
 572         else if (!memcmp(cclp->look_token->name, ">=", 2))
 573             rel = 4;
 574         else if (!memcmp(cclp->look_token->name, "<>", 2))
 575             rel = 6;
 576     }
 577     if (!rel)
 578     {
 579         cclp->error_code = CCL_ERR_BAD_RELATION;
 580         return NULL;
 581     }
 582     ADVANCE;  /* skip relation */
 583     if (rel == 3 &&
 584         qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, 0))
 585     {
 586         /* allow - inside term and treat it as range _always_ */
 587         /* relation is =. Extract "embedded" - to separate terms */
 588         if (KIND == CCL_TOK_TERM)
 589         {
 590             size_t i;
 591             for (i = 0; i<cclp->look_token->len; i++)
 592             {
 593                 if (cclp->look_token->name[i] == '-')
 594                     break;
 595             }
 596
 597             if (cclp->look_token->len > 1 && i == 0)
 598             {   /*  -xx*/
 599                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 600
 601                 ntoken->kind = CCL_TOK_TERM;
 602                 ntoken->name = cclp->look_token->name + 1;
 603                 ntoken->len = cclp->look_token->len - 1;
 604
 605                 cclp->look_token->len = 1;
 606                 cclp->look_token->name = "-";
 607             }
 608             else if (cclp->look_token->len > 1 && i == cclp->look_token->len-1)
 609             {   /* xx- */
 610                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 611
 612                 ntoken->kind = CCL_TOK_TERM;
 613                 ntoken->name = "-";
 614                 ntoken->len = 1;
 615
 616                 (cclp->look_token->len)--;
 617             }
 618             else if (cclp->look_token->len > 2 && i < cclp->look_token->len)
 619             {   /* xx-yy */
 620                 struct ccl_token *ntoken1 = ccl_token_add(cclp->look_token);
 621                 struct ccl_token *ntoken2 = ccl_token_add(ntoken1);
 622
 623                 ntoken1->kind = CCL_TOK_TERM;  /* generate - */
 624                 ntoken1->name = "-";
 625                 ntoken1->len = 1;
 626
 627                 ntoken2->kind = CCL_TOK_TERM;  /* generate yy */
 628                 ntoken2->name = cclp->look_token->name + (i+1);
 629                 ntoken2->len = cclp->look_token->len - (i+1);
 630
 631                 cclp->look_token->len = i;     /* adjust xx */
 632             }
 633             else if (i == cclp->look_token->len &&
 634                      cclp->look_token->next &&
 635                      cclp->look_token->next->kind == CCL_TOK_TERM &&
 636                      cclp->look_token->next->len > 1 &&
 637                      cclp->look_token->next->name[0] == '-')
 638
 639             {   /* xx -yy */
 640                 /* we _know_ that xx does not have - in it */
 641                 struct ccl_token *ntoken = ccl_token_add(cclp->look_token);
 642
 643                 ntoken->kind = CCL_TOK_TERM;    /* generate - */
 644                 ntoken->name = "-";
 645                 ntoken->len = 1;
 646
 647                 (ntoken->next->name)++;        /* adjust yy */
 648                 (ntoken->next->len)--;
 649             }
 650         }
 651     }
 652
 653     if (rel == 3 &&
 654         KIND == CCL_TOK_TERM &&
 655         cclp->look_token->next && cclp->look_token->next->len == 1 &&
 656         cclp->look_token->next->name[0] == '-')
 657     {
 658         struct ccl_rpn_node *p1;
 659         if (!(p1 = search_term(cclp, ap)))
 660             return NULL;
 661         ADVANCE;                   /* skip '-' */
 662         if (KIND == CCL_TOK_TERM)  /* = term - term  ? */
 663         {
 664             struct ccl_rpn_node *p2;
 665
 666             if (!(p2 = search_term(cclp, ap)))
 667             {
 668                 ccl_rpn_delete(p1);
 669                 return NULL;
 670             }
 671             p = ccl_rpn_node_create(CCL_RPN_AND);
 672             p->u.p[0] = p1;
 673             ccl_add_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
 674             p->u.p[1] = p2;
 675             ccl_add_attr_numeric(p2, attset, CCL_BIB1_REL, 2);
 676             return p;
 677         }
 678         else                       /* = term -    */
 679         {
 680             ccl_add_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
 681             return p1;
 682         }
 683     }
 684     else if (rel == 3 &&
 685              cclp->look_token->len == 1 &&
 686              cclp->look_token->name[0] == '-')   /* = - term  ? */
 687     {
 688         ADVANCE;
 689         if (!(p = search_term(cclp, ap)))
 690             return NULL;
 691         ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, 2);
 692         return p;
 693     }
 694     else if (KIND == CCL_TOK_LP)
 695     {
 696         ADVANCE;
 697         if (!(p = find_spec(cclp, ap)))
 698             return NULL;
 699         if (KIND != CCL_TOK_RP)
 700         {
 701             cclp->error_code = CCL_ERR_RP_EXPECTED;
 702             ccl_rpn_delete(p);
 703             return NULL;
 704         }
 705         ADVANCE;
 706         return p;
 707     }
 708     else
 709     {
 710         if (!(p = search_terms(cclp, ap)))
 711             return NULL;
 712         ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, rel);
 713         return p;
 714     }
 715     cclp->error_code = CCL_ERR_TERM_EXPECTED;
 716     return NULL;
 717 }
 718
 719 static
 720 struct ccl_rpn_node *qualifier_relation(CCL_parser cclp, ccl_qualifier_t *ap)
 721 {
 722     char *attset;
 723     struct ccl_rpn_node *p;
 724
 725     if (qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset)
 726         || qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, &attset))
 727         return qualifiers_order(cclp, ap, attset);
 728
 729     /* unordered relation */
 730     if (KIND != CCL_TOK_EQ)
 731     {
 732         cclp->error_code = CCL_ERR_EQ_EXPECTED;
 733         return NULL;
 734     }
 735     ADVANCE;
 736     if (KIND == CCL_TOK_LP)
 737     {
 738         ADVANCE;
 739         if (!(p = find_spec(cclp, ap)))
 740         {
 741             return NULL;
 742         }
 743         if (KIND != CCL_TOK_RP)
 744         {
 745             cclp->error_code = CCL_ERR_RP_EXPECTED;
 746             ccl_rpn_delete(p);
 747             return NULL;
 748         }
 749         ADVANCE;
 750     }
 751     else
 752         p = search_terms(cclp, ap);
 753     return p;
 754 }
 755
 756 /**
 757  * qualifier_list: Parse CCL qualifiers and search terms.
 758  * cclp:   CCL Parser
 759  * la:     Token pointer to RELATION token.
 760  * qa:     Qualifier attributes already applied.
 761  * return: pointer to node(s); NULL on error.
 762  */
 763 static struct ccl_rpn_node *qualifier_list(CCL_parser cclp,
 764                                            struct ccl_token *la,
 765                                            ccl_qualifier_t *qa)
 766 {
 767     struct ccl_token *lookahead = cclp->look_token;
 768     struct ccl_token *look_start = cclp->look_token;
 769     ccl_qualifier_t *ap;
 770     struct ccl_rpn_node *node = 0;
 771     const char **field_str;
 772     int no = 0;
 773     int seq = 0;
 774     int i;
 775     int mode_merge = 1;
 776 #if 0
 777     if (qa)
 778     {
 779         cclp->error_code = CCL_ERR_DOUBLE_QUAL;
 780         return NULL;
 781     }
 782 #endif
 783     for (lookahead = cclp->look_token; lookahead != la;
 784          lookahead=lookahead->next)
 785         no++;
 786     if (qa)
 787         for (i=0; qa[i]; i++)
 788             no++;
 789     ap = (ccl_qualifier_t *)xmalloc((no ? (no+1) : 2) * sizeof(*ap));
 790     ccl_assert(ap);
 791
 792     field_str = ccl_qual_search_special(cclp->bibset, "field");
 793     if (field_str)
 794     {
 795         if (!strcmp(field_str[0], "or"))
 796             mode_merge = 0;
 797         else if (!strcmp(field_str[0], "merge"))
 798             mode_merge = 1;
 799     }
 800     if (!mode_merge)
 801     {
 802         /* consider each field separately and OR */
 803         lookahead = look_start;
 804         while (lookahead != la)
 805         {
 806             ap[1] = 0;
 807             seq = 0;
 808             while ((ap[0] = ccl_qual_search(cclp, lookahead->name,
 809                                             lookahead->len, seq)) != 0)
 810             {
 811                 struct ccl_rpn_node *node_sub;
 812                 cclp->look_token = la;
 813
 814                 node_sub = qualifier_relation(cclp, ap);
 815                 if (!node_sub)
 816                 {
 817                     ccl_rpn_delete(node);
 818                     xfree(ap);
 819                     return 0;
 820                 }
 821                 if (node)
 822                 {
 823                     struct ccl_rpn_node *node_this =
 824                         ccl_rpn_node_create(CCL_RPN_OR);
 825                     node_this->u.p[0] = node;
 826                     node_this->u.p[1] = node_sub;
 827                     node = node_this;
 828                 }
 829                 else
 830                     node = node_sub;
 831                 seq++;
 832             }
 833             if (seq == 0)
 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     }
 845     else
 846     {
 847         /* merge attributes from ALL fields - including inherited ones */
 848         while (1)
 849         {
 850             struct ccl_rpn_node *node_sub;
 851             int found = 0;
 852             lookahead = look_start;
 853             for (i = 0; lookahead != la; i++)
 854             {
 855                 ap[i] = ccl_qual_search(cclp, lookahead->name,
 856                                          lookahead->len, seq);
 857                 if (ap[i])
 858                     found++;
 859                 if (!ap[i] && seq > 0)
 860                     ap[i] = ccl_qual_search(cclp, lookahead->name,
 861                                              lookahead->len, 0);
 862                 if (!ap[i])
 863                 {
 864                     cclp->look_token = lookahead;
 865                     cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
 866                     xfree(ap);
 867                     return NULL;
 868                 }
 869                 lookahead = lookahead->next;
 870                 if (lookahead->kind == CCL_TOK_COMMA)
 871                     lookahead = lookahead->next;
 872             }
 873             if (qa)
 874             {
 875                 ccl_qualifier_t *qa0 = qa;
 876
 877                 while (*qa0)
 878                     ap[i++] = *qa0++;
 879             }
 880             ap[i] = NULL;
 881
 882             if (!found)
 883                 break;
 884
 885             cclp->look_token = lookahead;
 886
 887             node_sub = qualifier_relation(cclp, ap);
 888             if (!node_sub)
 889             {
 890                 ccl_rpn_delete(node);
 891                 break;
 892             }
 893             if (node)
 894             {
 895                 struct ccl_rpn_node *node_this =
 896                     ccl_rpn_node_create(CCL_RPN_OR);
 897                 node_this->u.p[0] = node;
 898                 node_this->u.p[1] = node_sub;
 899                 node = node_this;
 900             }
 901             else
 902                 node = node_sub;
 903             seq++;
 904         }
 905     }
 906     xfree(ap);
 907     return node;
 908 }
 909
 910
 911 /**
 912  * search_terms: Parse CCL search terms - including proximity.
 913  * cclp:   CCL Parser
 914  * qa:     Qualifier attributes already applied.
 915  * return: pointer to node(s); NULL on error.
 916  */
 917 static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa)
 918 {
 919     static int list[] = {
 920         CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ, CCL_TOK_REL, CCL_TOK_SET, -1};
 921     struct ccl_rpn_node *p1, *p2, *pn;
 922     p1 = search_term_x(cclp, qa, list, 1);
 923     if (!p1)
 924         return NULL;
 925     while (1)
 926     {
 927         if (KIND == CCL_TOK_PROX)
 928         {
 929             struct ccl_rpn_node *p_prox = 0;
 930             /* ! word order specified */
 931             /* % word order not specified */
 932             p_prox = ccl_rpn_node_create(CCL_RPN_TERM);
 933             p_prox->u.t.term = (char *) xmalloc(1 + cclp->look_token->len);
 934             memcpy(p_prox->u.t.term, cclp->look_token->name,
 935                    cclp->look_token->len);
 936             p_prox->u.t.term[cclp->look_token->len] = 0;
 937             p_prox->u.t.attr_list = 0;
 938
 939             ADVANCE;
 940             p2 = search_term_x(cclp, qa, list, 1);
 941             if (!p2)
 942             {
 943                 ccl_rpn_delete(p1);
 944                 return NULL;
 945             }
 946             pn = ccl_rpn_node_create(CCL_RPN_PROX);
 947             pn->u.p[0] = p1;
 948             pn->u.p[1] = p2;
 949             pn->u.p[2] = p_prox;
 950             p1 = pn;
 951         }
 952         else if (is_term_ok(KIND, list))
 953         {
 954             p2 = search_term_x(cclp, qa, list, 1);
 955             if (!p2)
 956             {
 957                 ccl_rpn_delete(p1);
 958                 return NULL;
 959             }
 960             pn = ccl_rpn_node_create(CCL_RPN_PROX);
 961             pn->u.p[0] = p1;
 962             pn->u.p[1] = p2;
 963             pn->u.p[2] = 0;
 964             p1 = pn;
 965         }
 966         else
 967             break;
 968     }
 969     return p1;
 970 }
 971
 972 /**
 973  * search_elements: Parse CCL search elements
 974  * cclp:   CCL Parser
 975  * qa:     Qualifier attributes already applied.
 976  * return: pointer to node(s); NULL on error.
 977  */
 978 static struct ccl_rpn_node *search_elements(CCL_parser cclp,
 979                                             ccl_qualifier_t *qa)
 980 {
 981     struct ccl_rpn_node *p1;
 982     struct ccl_token *lookahead;
 983     if (KIND == CCL_TOK_LP)
 984     {
 985         ADVANCE;
 986         p1 = find_spec(cclp, qa);
 987         if (!p1)
 988             return NULL;
 989         if (KIND != CCL_TOK_RP)
 990         {
 991             cclp->error_code = CCL_ERR_RP_EXPECTED;
 992             ccl_rpn_delete(p1);
 993             return NULL;
 994         }
 995         ADVANCE;
 996         return p1;
 997     }
 998     else if (KIND == CCL_TOK_SET)
 999     {
1000         ADVANCE;
1001         if (KIND == CCL_TOK_EQ)
1002             ADVANCE;
1003         if (KIND != CCL_TOK_TERM)
1004         {
1005             cclp->error_code = CCL_ERR_SETNAME_EXPECTED;
1006             return NULL;
1007         }
1008         p1 = ccl_rpn_node_create(CCL_RPN_SET);
1009         p1->u.setname = copy_token_name(cclp->look_token);
1010         ADVANCE;
1011         return p1;
1012     }
1013     lookahead = cclp->look_token;
1014
1015     while (lookahead->kind==CCL_TOK_TERM)
1016     {
1017         lookahead = lookahead->next;
1018         if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
1019             return qualifier_list(cclp, lookahead, qa);
1020         if (lookahead->kind != CCL_TOK_COMMA)
1021             break;
1022         lookahead = lookahead->next;
1023     }
1024     if (qa)
1025         return search_terms(cclp, qa);
1026     else
1027     {
1028         ccl_qualifier_t qa[2];
1029         struct ccl_rpn_node *node = 0;
1030         int seq;
1031         lookahead = cclp->look_token;
1032
1033         qa[1] = 0;
1034         for(seq = 0; ;seq++)
1035         {
1036             struct ccl_rpn_node *node_sub;
1037             qa[0] = ccl_qual_search(cclp, "term", 4, seq);
1038             if (!qa[0])
1039                 break;
1040
1041             cclp->look_token = lookahead;
1042
1043             node_sub = search_terms(cclp, qa);
1044             if (!node_sub)
1045             {
1046                 ccl_rpn_delete(node);
1047                 return 0;
1048             }
1049             if (node)
1050             {
1051                 struct ccl_rpn_node *node_this =
1052                     ccl_rpn_node_create(CCL_RPN_OR);
1053                 node_this->u.p[0] = node;
1054                 node_this->u.p[1] = node_sub;
1055                 node_this->u.p[2] = 0;
1056                 node = node_this;
1057             }
1058             else
1059                 node = node_sub;
1060         }
1061         if (!node)
1062             node = search_terms(cclp, 0);
1063         return node;
1064     }
1065 }
1066
1067 /**
1068  * find_spec: Parse CCL find specification
1069  * cclp:   CCL Parser
1070  * qa:     Qualifier attributes already applied.
1071  * return: pointer to node(s); NULL on error.
1072  */
1073 static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa)
1074 {
1075     struct ccl_rpn_node *p1, *p2, *pn;
1076     if (!(p1 = search_elements(cclp, qa)))
1077         return NULL;
1078     while (1)
1079     {
1080         switch (KIND)
1081         {
1082         case CCL_TOK_AND:
1083             ADVANCE;
1084             p2 = search_elements(cclp, qa);
1085             if (!p2)
1086             {
1087                 ccl_rpn_delete(p1);
1088                 return NULL;
1089             }
1090             pn = ccl_rpn_node_create(CCL_RPN_AND);
1091             pn->u.p[0] = p1;
1092             pn->u.p[1] = p2;
1093             pn->u.p[2] = 0;
1094             p1 = pn;
1095             continue;
1096         case CCL_TOK_OR:
1097             ADVANCE;
1098             p2 = search_elements(cclp, qa);
1099             if (!p2)
1100             {
1101                 ccl_rpn_delete(p1);
1102                 return NULL;
1103             }
1104             pn = ccl_rpn_node_create(CCL_RPN_OR);
1105             pn->u.p[0] = p1;
1106             pn->u.p[1] = p2;
1107             pn->u.p[2] = 0;
1108             p1 = pn;
1109             continue;
1110         case CCL_TOK_NOT:
1111             ADVANCE;
1112             p2 = search_elements(cclp, qa);
1113             if (!p2)
1114             {
1115                 ccl_rpn_delete(p1);
1116                 return NULL;
1117             }
1118             pn = ccl_rpn_node_create(CCL_RPN_NOT);
1119             pn->u.p[0] = p1;
1120             pn->u.p[1] = p2;
1121             pn->u.p[2] = 0;
1122             p1 = pn;
1123             continue;
1124         }
1125         break;
1126     }
1127     return p1;
1128 }
1129
1130 struct ccl_rpn_node *ccl_parser_find_str(CCL_parser cclp, const char *str)
1131 {
1132     struct ccl_rpn_node *p;
1133     struct ccl_token *list = ccl_parser_tokenize(cclp, str);
1134     p = ccl_parser_find_token(cclp, list);
1135     ccl_token_del(list);
1136     return p;
1137 }
1138
1139 struct ccl_rpn_node *ccl_parser_find_token(CCL_parser cclp,
1140                                            struct ccl_token *list)
1141 {
1142     struct ccl_rpn_node *p;
1143
1144     cclp->look_token = list;
1145     p = find_spec(cclp, NULL);
1146     if (p && KIND != CCL_TOK_EOL)
1147     {
1148         if (KIND == CCL_TOK_RP)
1149             cclp->error_code = CCL_ERR_BAD_RP;
1150         else
1151             cclp->error_code = CCL_ERR_OP_EXPECTED;
1152         ccl_rpn_delete(p);
1153         p = NULL;
1154     }
1155     cclp->error_pos = cclp->look_token->name;
1156     if (p)
1157         cclp->error_code = CCL_ERR_OK;
1158     else
1159         cclp->error_code = cclp->error_code;
1160     return p;
1161 }
1162
1163 /**
1164  * ccl_find_str: Parse CCL find - string representation
1165  * bibset:  Bibset to be used for the parsing
1166  * str:     String to be parsed
1167  * error:   Pointer to integer. Holds error no. on completion.
1168  * pos:     Pointer to char position. Holds approximate error position.
1169  * return:  RPN tree on successful completion; NULL otherwise.
1170  */
1171 struct ccl_rpn_node *ccl_find_str(CCL_bibset bibset, const char *str,
1172                                   int *error, int *pos)
1173 {
1174     CCL_parser cclp = ccl_parser_create(bibset);
1175     struct ccl_token *list;
1176     struct ccl_rpn_node *p;
1177
1178     list = ccl_parser_tokenize(cclp, str);
1179     p = ccl_parser_find_token(cclp, list);
1180
1181     *error = cclp->error_code;
1182     if (*error)
1183         *pos = cclp->error_pos - str;
1184     ccl_parser_destroy(cclp);
1185     ccl_token_del(list);
1186     return p;
1187 }
1188
1189 /*
1190  * Local variables:
1191  * c-basic-offset: 4
1192  * c-file-style: "Stroustrup"
1193  * indent-tabs-mode: nil
1194  * End:
1195  * vim: shiftwidth=4 tabstop=8 expandtab
1196  */
1197