2 * Copyright (c) 1995, the EUROPAGATE consortium (see below).
4 * The EUROPAGATE consortium members are:
6 * University College Dublin
7 * Danmarks Teknologiske Videnscenter
8 * An Chomhairle Leabharlanna
9 * Consejo Superior de Investigaciones Cientificas
11 * Permission to use, copy, modify, distribute, and sell this software and
12 * its documentation, in whole or in part, for any purpose, is hereby granted,
15 * 1. This copyright and permission notice appear in all copies of the
16 * software and its documentation. Notices of copyright or attribution
17 * which appear at the beginning of any file must remain unchanged.
19 * 2. The names of EUROPAGATE or the project partners may not be used to
20 * endorse or promote products derived from this software without specific
21 * prior written permission.
23 * 3. Users of this software (implementors and gateway operators) agree to
24 * inform the EUROPAGATE consortium of their use of the software. This
25 * information will be used to evaluate the EUROPAGATE project and the
26 * software, and to plan further developments. The consortium may use
27 * the information in later publications.
29 * 4. Users of this software agree to make their best efforts, when
30 * documenting their use of the software, to acknowledge the EUROPAGATE
31 * consortium, and the role played by the software in their work.
33 * THIS SOFTWARE IS PROVIDED "AS IS" AND WITHOUT WARRANTY OF ANY KIND,
34 * EXPRESS, IMPLIED, OR OTHERWISE, INCLUDING WITHOUT LIMITATION, ANY
35 * WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
36 * IN NO EVENT SHALL THE EUROPAGATE CONSORTIUM OR ITS MEMBERS BE LIABLE
37 * FOR ANY SPECIAL, INCIDENTAL, INDIRECT OR CONSEQUENTIAL DAMAGES OF
38 * ANY KIND, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA
39 * OR PROFITS, WHETHER OR NOT ADVISED OF THE POSSIBILITY OF DAMAGE, AND
40 * ON ANY THEORY OF LIABILITY, ARISING OUT OF OR IN CONNECTION WITH THE
41 * USE OR PERFORMANCE OF THIS SOFTWARE.
44 /* CCL find (to rpn conversion)
48 * Revision 1.13 1999-12-22 13:13:32 adam
49 * Search terms may include "operators" without causing error.
51 * Revision 1.12 1999/11/30 13:47:11 adam
52 * Improved installation. Moved header files to include/yaz.
54 * Revision 1.11 1999/03/31 11:15:37 adam
55 * Fixed memory leaks in ccl_find_str and ccl_qual_rm.
57 * Revision 1.10 1998/02/11 11:53:33 adam
58 * Changed code so that it compiles as C++.
60 * Revision 1.9 1997/09/29 08:56:37 adam
61 * Changed CCL parser to be thread safe. New type, CCL_parser, declared
62 * and a create/destructers ccl_parser_create/ccl_parser/destory has
65 * Revision 1.8 1997/09/01 08:48:11 adam
66 * New windows NT/95 port using MSV5.0. Only a few changes made
69 * Revision 1.7 1997/05/14 06:53:26 adam
72 * Revision 1.6 1997/04/30 08:52:06 quinn
75 * Revision 1.5 1996/10/11 15:00:24 adam
76 * CCL parser from Europagate Email gateway 1.0.
78 * Revision 1.16 1996/01/08 08:41:13 adam
79 * Removed unused function.
81 * Revision 1.15 1995/07/20 08:14:34 adam
82 * Qualifiers were observed too often. Instead tokens are treated as
83 * qualifiers only when separated by comma.
85 * Revision 1.14 1995/05/16 09:39:26 adam
88 * Revision 1.13 1995/04/17 09:31:42 adam
89 * Improved handling of qualifiers. Aliases or reserved words.
91 * Revision 1.12 1995/03/20 15:27:43 adam
94 * Revision 1.11 1995/02/23 08:31:59 adam
97 * Revision 1.9 1995/02/16 13:20:06 adam
100 * Revision 1.8 1995/02/14 19:59:42 adam
101 * Removed a syntax error.
103 * Revision 1.7 1995/02/14 19:55:10 adam
104 * Header files ccl.h/cclp.h are gone! They have been merged an
105 * moved to ../include/ccl.h.
106 * Node kind(s) in ccl_rpn_node have changed names.
108 * Revision 1.6 1995/02/14 16:20:55 adam
109 * Qualifiers are read from a file now.
111 * Revision 1.5 1995/02/14 14:12:41 adam
112 * Ranges for ordered qualfiers implemented (e.g. pd=1980-1990).
114 * Revision 1.4 1995/02/14 13:16:29 adam
115 * Left and/or right truncation implemented.
117 * Revision 1.3 1995/02/14 10:25:56 adam
118 * The constructions 'qualifier rel term ...' implemented.
120 * Revision 1.2 1995/02/13 15:15:07 adam
121 * Added handling of qualifiers. Not finished yet.
123 * Revision 1.1 1995/02/13 12:35:20 adam
124 * First version of CCL. Qualifiers aren't handled yet.
135 /* returns type of current lookahead */
136 #define KIND (cclp->look_token->kind)
138 /* move one token forward */
139 #define ADVANCE cclp->look_token = cclp->look_token->next
142 * qual_val_type: test for existance of attribute type/value pair.
143 * qa: Attribute array
144 * type: Type of attribute to search for
145 * value: Value of attribute to seach for
146 * return: 1 if found; 0 otherwise.
148 static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value)
151 struct ccl_rpn_attr *q;
155 for (i = 0; (q=qa[i]); i++)
158 if (q->type == type && q->value == value)
166 * strxcat: concatenate strings.
167 * n: Null-terminated Destination string
168 * src: Source string to be appended (not null-terminated)
169 * len: Length of source string.
171 static void strxcat (char *n, const char *src, int len)
181 * copy_token_name: Return copy of CCL token name
182 * tp: Pointer to token info.
183 * return: malloc(3) allocated copy of token name.
185 static char *copy_token_name (struct ccl_token *tp)
187 char *str = (char *)malloc (tp->len + 1);
189 memcpy (str, tp->name, tp->len);
195 * mk_node: Create RPN node.
196 * kind: Type of node.
197 * return: pointer to allocated node.
199 static struct ccl_rpn_node *mk_node (int kind)
201 struct ccl_rpn_node *p;
202 p = (struct ccl_rpn_node *)malloc (sizeof(*p));
209 * ccl_rpn_delete: Delete RPN tree.
210 * rpn: Pointer to tree.
212 void ccl_rpn_delete (struct ccl_rpn_node *rpn)
214 struct ccl_rpn_attr *attr, *attr1;
222 ccl_rpn_delete (rpn->u.p[0]);
223 ccl_rpn_delete (rpn->u.p[1]);
226 free (rpn->u.t.term);
227 for (attr = rpn->u.t.attr_list; attr; attr = attr1)
234 free (rpn->u.setname);
237 ccl_rpn_delete (rpn->u.p[0]);
238 ccl_rpn_delete (rpn->u.p[1]);
244 static struct ccl_rpn_node *find_spec (CCL_parser cclp,
245 struct ccl_rpn_attr **qa);
247 static int is_term_ok (int look, int *list)
249 for (;*list >= 0; list++)
255 static struct ccl_rpn_node *search_terms (CCL_parser cclp,
256 struct ccl_rpn_attr **qa);
259 * add_attr: Add attribute (type/value) to RPN term node.
260 * p: RPN node of type term.
261 * type: Type of attribute
262 * value: Value of attribute
264 static void add_attr (struct ccl_rpn_node *p, int type, int value)
266 struct ccl_rpn_attr *n;
268 n = (struct ccl_rpn_attr *)malloc (sizeof(*n));
272 n->next = p->u.t.attr_list;
273 p->u.t.attr_list = n;
277 * search_term: Parse CCL search term.
279 * qa: Qualifier attributes already applied.
280 * return: pointer to node(s); NULL on error.
282 static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
283 struct ccl_rpn_attr **qa,
286 struct ccl_rpn_node *p;
287 struct ccl_token *lookahead = cclp->look_token;
293 int relation_value = -1;
294 int position_value = -1;
295 int structure_value = -1;
296 int truncation_value = -1;
297 int completeness_value = -1;
299 if (!is_term_ok(KIND, term_list))
301 cclp->error_code = CCL_ERR_TERM_EXPECTED;
304 /* create the term node, but wait a moment before adding the term */
305 p = mk_node (CCL_RPN_TERM);
306 p->u.t.attr_list = NULL;
311 /* no qualifier(s) applied. Use 'term' if it is defined */
313 qa = (struct ccl_rpn_attr **)malloc (2*sizeof(*qa));
315 qa[0] = ccl_qual_search (cclp, "term", 4);
319 /* go through all attributes and add them to the attribute list */
320 for (i=0; qa && qa[i]; i++)
322 struct ccl_rpn_attr *attr;
324 for (attr = qa[i]; attr; attr = attr->next)
326 { /* deal only with REAL attributes (positive) */
330 if (relation_value != -1)
332 relation_value = attr->value;
335 if (position_value != -1)
337 position_value = attr->value;
340 if (structure_value != -1)
342 structure_value = attr->value;
345 if (truncation_value != -1)
347 truncation_value = attr->value;
350 if (completeness_value != -1)
352 completeness_value = attr->value;
355 add_attr (p, attr->type, attr->value);
358 /* go through each TERM token. If no truncation attribute is yet
359 met, then look for left/right truncation markers (?) and
360 set left_trunc/right_trunc/mid_trunc accordingly */
361 for (no = 0; is_term_ok(lookahead->kind, term_list); no++)
363 for (i = 0; i<lookahead->len; i++)
364 if (truncation_value == -1 && lookahead->name[i] == '?')
366 if (no == 0 && i == 0 && lookahead->len >= 1)
368 else if (!is_term_ok(lookahead->next->kind, term_list) &&
369 i == lookahead->len-1 && i >= 1)
374 len += 1+lookahead->len;
375 lookahead = lookahead->next;
377 /* len now holds the number of characters in the RPN term */
378 /* no holds the number of CCL tokens (1 or more) */
380 if (structure_value == -1 &&
381 qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_WP))
382 { /* no structure attribute met. Apply either structure attribute
383 WORD or PHRASE depending on number of CCL tokens */
385 add_attr (p, CCL_BIB1_STR, 2);
387 add_attr (p, CCL_BIB1_STR, 1);
390 /* make the RPN token */
391 p->u.t.term = (char *)malloc (len);
392 assert (p->u.t.term);
393 p->u.t.term[0] = '\0';
394 for (i = 0; i<no; i++)
396 const char *src_str = cclp->look_token->name;
397 int src_len = cclp->look_token->len;
399 if (i == 0 && left_trunc)
404 else if (i == no-1 && right_trunc)
407 strcat (p->u.t.term, " ");
408 strxcat (p->u.t.term, src_str, src_len);
411 if (left_trunc && right_trunc)
413 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH))
415 cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
420 add_attr (p, CCL_BIB1_TRU, 3);
422 else if (right_trunc)
424 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT))
426 cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
431 add_attr (p, CCL_BIB1_TRU, 1);
435 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT))
437 cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
442 add_attr (p, CCL_BIB1_TRU, 2);
446 if (qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE))
447 add_attr (p, CCL_BIB1_TRU, 100);
452 static struct ccl_rpn_node *search_term (CCL_parser cclp,
453 struct ccl_rpn_attr **qa)
455 static int list[] = {CCL_TOK_TERM, CCL_TOK_COMMA, -1};
456 return search_term_x(cclp, qa, list);
460 * qualifiers: Parse CCL qualifiers and search terms.
462 * la: Token pointer to RELATION token.
463 * qa: Qualifier attributes already applied.
464 * return: pointer to node(s); NULL on error.
466 static struct ccl_rpn_node *qualifiers (CCL_parser cclp, struct ccl_token *la,
467 struct ccl_rpn_attr **qa)
469 struct ccl_token *lookahead = cclp->look_token;
470 struct ccl_rpn_attr **ap;
476 cclp->error_code = CCL_ERR_DOUBLE_QUAL;
480 for (lookahead = cclp->look_token; lookahead != la;
481 lookahead=lookahead->next)
484 for (i=0; qa[i]; i++)
486 ap = (struct ccl_rpn_attr **)malloc ((no+1) * sizeof(*ap));
488 for (i = 0; cclp->look_token != la; i++)
490 ap[i] = ccl_qual_search (cclp, cclp->look_token->name,
491 cclp->look_token->len);
494 cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
499 if (KIND == CCL_TOK_COMMA)
506 if (!qual_val_type (ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER))
508 /* unordered relation */
509 struct ccl_rpn_node *p;
510 if (KIND != CCL_TOK_EQ)
512 cclp->error_code = CCL_ERR_EQ_EXPECTED;
517 if (KIND == CCL_TOK_LP)
520 if (!(p = find_spec (cclp, ap)))
525 if (KIND != CCL_TOK_RP)
527 cclp->error_code = CCL_ERR_RP_EXPECTED;
535 p = search_terms (cclp, ap);
539 /* ordered relation ... */
541 if (cclp->look_token->len == 1)
543 if (cclp->look_token->name[0] == '<')
545 else if (cclp->look_token->name[0] == '=')
547 else if (cclp->look_token->name[0] == '>')
550 else if (cclp->look_token->len == 2)
552 if (!memcmp (cclp->look_token->name, "<=", 2))
554 else if (!memcmp (cclp->look_token->name, ">=", 2))
556 else if (!memcmp (cclp->look_token->name, "<>", 2))
560 cclp->error_code = CCL_ERR_BAD_RELATION;
563 struct ccl_rpn_node *p;
565 ADVANCE; /* skip relation */
566 if (KIND == CCL_TOK_TERM &&
567 cclp->look_token->next->kind == CCL_TOK_MINUS)
569 struct ccl_rpn_node *p1;
570 if (!(p1 = search_term (cclp, ap)))
575 ADVANCE; /* skip '-' */
576 if (KIND == CCL_TOK_TERM) /* = term - term ? */
578 struct ccl_rpn_node *p2;
580 if (!(p2 = search_term (cclp, ap)))
586 p = mk_node (CCL_RPN_AND);
588 add_attr (p1, CCL_BIB1_REL, 4);
590 add_attr (p2, CCL_BIB1_REL, 2);
596 add_attr (p1, CCL_BIB1_REL, 4);
601 else if (KIND == CCL_TOK_MINUS) /* = - term ? */
604 if (!(p = search_term (cclp, ap)))
609 add_attr (p, CCL_BIB1_REL, 2);
613 else if (KIND == CCL_TOK_LP)
616 if (!(p = find_spec (cclp, ap)))
621 if (KIND != CCL_TOK_RP)
623 cclp->error_code = CCL_ERR_RP_EXPECTED;
634 if (!(p = search_terms (cclp, ap)))
639 add_attr (p, CCL_BIB1_REL, rel);
643 cclp->error_code = CCL_ERR_TERM_EXPECTED;
650 * search_terms: Parse CCL search terms - including proximity.
652 * qa: Qualifier attributes already applied.
653 * return: pointer to node(s); NULL on error.
655 static struct ccl_rpn_node *search_terms (CCL_parser cclp,
656 struct ccl_rpn_attr **qa)
658 static int list[] = {
659 CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
660 CCL_TOK_REL, CCL_TOK_MINUS, -1};
661 struct ccl_rpn_node *p1, *p2, *pn;
662 p1 = search_term_x (cclp, qa, list);
667 if (KIND == CCL_TOK_PROX)
670 p2 = search_term_x (cclp, qa, list);
676 pn = mk_node (CCL_RPN_PROX);
681 else if (is_term_ok(KIND, list))
683 p2 = search_term_x (cclp, qa, list);
689 pn = mk_node (CCL_RPN_PROX);
701 * search_elements: Parse CCL search elements
703 * qa: Qualifier attributes already applied.
704 * return: pointer to node(s); NULL on error.
706 static struct ccl_rpn_node *search_elements (CCL_parser cclp,
707 struct ccl_rpn_attr **qa)
709 struct ccl_rpn_node *p1;
710 struct ccl_token *lookahead;
711 if (KIND == CCL_TOK_LP)
714 p1 = find_spec (cclp, qa);
717 if (KIND != CCL_TOK_RP)
719 cclp->error_code = CCL_ERR_RP_EXPECTED;
726 else if (KIND == CCL_TOK_SET)
729 if (KIND == CCL_TOK_EQ)
731 if (KIND != CCL_TOK_TERM)
733 cclp->error_code = CCL_ERR_SETNAME_EXPECTED;
736 p1 = mk_node (CCL_RPN_SET);
737 p1->u.setname = copy_token_name (cclp->look_token);
741 lookahead = cclp->look_token;
743 while (lookahead->kind==CCL_TOK_TERM)
745 lookahead = lookahead->next;
746 if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
747 return qualifiers (cclp, lookahead, qa);
748 if (lookahead->kind != CCL_TOK_COMMA)
750 lookahead = lookahead->next;
752 return search_terms (cclp, qa);
756 * find_spec: Parse CCL find specification
758 * qa: Qualifier attributes already applied.
759 * return: pointer to node(s); NULL on error.
761 static struct ccl_rpn_node *find_spec (CCL_parser cclp,
762 struct ccl_rpn_attr **qa)
764 struct ccl_rpn_node *p1, *p2, *pn;
765 if (!(p1 = search_elements (cclp, qa)))
773 p2 = search_elements (cclp, qa);
779 pn = mk_node (CCL_RPN_AND);
786 p2 = search_elements (cclp, qa);
792 pn = mk_node (CCL_RPN_OR);
799 p2 = search_elements (cclp, qa);
805 pn = mk_node (CCL_RPN_NOT);
816 struct ccl_rpn_node *ccl_parser_find (CCL_parser cclp, struct ccl_token *list)
818 struct ccl_rpn_node *p;
820 cclp->look_token = list;
821 p = find_spec (cclp, NULL);
822 if (p && KIND != CCL_TOK_EOL)
824 if (KIND == CCL_TOK_RP)
825 cclp->error_code = CCL_ERR_BAD_RP;
827 cclp->error_code = CCL_ERR_OP_EXPECTED;
831 cclp->error_pos = cclp->look_token->name;
833 cclp->error_code = CCL_ERR_OK;
835 cclp->error_code = cclp->error_code;
840 * ccl_find: Parse CCL find - token representation
841 * bibset: Bibset to be used for the parsing
842 * list: List of tokens
843 * error: Pointer to integer. Holds error no. on completion.
844 * pos: Pointer to char position. Holds approximate error position.
845 * return: RPN tree on successful completion; NULL otherwise.
847 struct ccl_rpn_node *ccl_find (CCL_bibset bibset, struct ccl_token *list,
848 int *error, const char **pos)
850 struct ccl_rpn_node *p;
851 CCL_parser cclp = ccl_parser_create ();
853 cclp->bibset = bibset;
855 p = ccl_parser_find (cclp, list);
857 *error = cclp->error_code;
858 *pos = cclp->error_pos;
860 ccl_parser_destroy (cclp);
866 * ccl_find_str: Parse CCL find - string representation
867 * bibset: Bibset to be used for the parsing
868 * str: String to be parsed
869 * error: Pointer to integer. Holds error no. on completion.
870 * pos: Pointer to char position. Holds approximate error position.
871 * return: RPN tree on successful completion; NULL otherwise.
873 struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset, const char *str,
874 int *error, int *pos)
876 CCL_parser cclp = ccl_parser_create ();
877 struct ccl_token *list;
878 struct ccl_rpn_node *p;
880 cclp->bibset = bibset;
882 list = ccl_parser_tokenize (cclp, str);
883 p = ccl_parser_find (cclp, list);
885 *error = cclp->error_code;
887 *pos = cclp->error_pos - str;
888 ccl_parser_destroy (cclp);
889 ccl_token_del (list);