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 1.16 2000-03-14 09:06:11 adam
49 * Added POSIX threads support for frontend server.
51 * Revision 1.15 2000/02/24 23:49:13 adam
52 * Fixed memory allocation problem.
54 * Revision 1.14 2000/01/31 13:15:21 adam
55 * Removed uses of assert(3). Cleanup of ODR. CCL parser update so
56 * that some characters are not surrounded by spaces in resulting term.
59 * Revision 1.13 1999/12/22 13:13:32 adam
60 * Search terms may include "operators" without causing error.
62 * Revision 1.12 1999/11/30 13:47:11 adam
63 * Improved installation. Moved header files to include/yaz.
65 * Revision 1.11 1999/03/31 11:15:37 adam
66 * Fixed memory leaks in ccl_find_str and ccl_qual_rm.
68 * Revision 1.10 1998/02/11 11:53:33 adam
69 * Changed code so that it compiles as C++.
71 * Revision 1.9 1997/09/29 08:56:37 adam
72 * Changed CCL parser to be thread safe. New type, CCL_parser, declared
73 * and a create/destructers ccl_parser_create/ccl_parser/destory has
76 * Revision 1.8 1997/09/01 08:48:11 adam
77 * New windows NT/95 port using MSV5.0. Only a few changes made
80 * Revision 1.7 1997/05/14 06:53:26 adam
83 * Revision 1.6 1997/04/30 08:52:06 quinn
86 * Revision 1.5 1996/10/11 15:00:24 adam
87 * CCL parser from Europagate Email gateway 1.0.
89 * Revision 1.16 1996/01/08 08:41:13 adam
90 * Removed unused function.
92 * Revision 1.15 1995/07/20 08:14:34 adam
93 * Qualifiers were observed too often. Instead tokens are treated as
94 * qualifiers only when separated by comma.
96 * Revision 1.14 1995/05/16 09:39:26 adam
99 * Revision 1.13 1995/04/17 09:31:42 adam
100 * Improved handling of qualifiers. Aliases or reserved words.
102 * Revision 1.12 1995/03/20 15:27:43 adam
105 * Revision 1.11 1995/02/23 08:31:59 adam
108 * Revision 1.9 1995/02/16 13:20:06 adam
111 * Revision 1.8 1995/02/14 19:59:42 adam
112 * Removed a syntax error.
114 * Revision 1.7 1995/02/14 19:55:10 adam
115 * Header files ccl.h/cclp.h are gone! They have been merged an
116 * moved to ../include/ccl.h.
117 * Node kind(s) in ccl_rpn_node have changed names.
119 * Revision 1.6 1995/02/14 16:20:55 adam
120 * Qualifiers are read from a file now.
122 * Revision 1.5 1995/02/14 14:12:41 adam
123 * Ranges for ordered qualfiers implemented (e.g. pd=1980-1990).
125 * Revision 1.4 1995/02/14 13:16:29 adam
126 * Left and/or right truncation implemented.
128 * Revision 1.3 1995/02/14 10:25:56 adam
129 * The constructions 'qualifier rel term ...' implemented.
131 * Revision 1.2 1995/02/13 15:15:07 adam
132 * Added handling of qualifiers. Not finished yet.
134 * Revision 1.1 1995/02/13 12:35:20 adam
135 * First version of CCL. Qualifiers aren't handled yet.
144 /* returns type of current lookahead */
145 #define KIND (cclp->look_token->kind)
147 /* move one token forward */
148 #define ADVANCE cclp->look_token = cclp->look_token->next
151 * qual_val_type: test for existance of attribute type/value pair.
152 * qa: Attribute array
153 * type: Type of attribute to search for
154 * value: Value of attribute to seach for
155 * return: 1 if found; 0 otherwise.
157 static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value)
160 struct ccl_rpn_attr *q;
164 for (i = 0; (q=qa[i]); i++)
167 if (q->type == type && q->value == value)
175 * strxcat: concatenate strings.
176 * n: Null-terminated Destination string
177 * src: Source string to be appended (not null-terminated)
178 * len: Length of source string.
180 static void strxcat (char *n, const char *src, int len)
190 * copy_token_name: Return copy of CCL token name
191 * tp: Pointer to token info.
192 * return: malloc(3) allocated copy of token name.
194 static char *copy_token_name (struct ccl_token *tp)
196 char *str = (char *)malloc (tp->len + 1);
198 memcpy (str, tp->name, tp->len);
204 * mk_node: Create RPN node.
205 * kind: Type of node.
206 * return: pointer to allocated node.
208 static struct ccl_rpn_node *mk_node (int kind)
210 struct ccl_rpn_node *p;
211 p = (struct ccl_rpn_node *)malloc (sizeof(*p));
218 * ccl_rpn_delete: Delete RPN tree.
219 * rpn: Pointer to tree.
221 void ccl_rpn_delete (struct ccl_rpn_node *rpn)
223 struct ccl_rpn_attr *attr, *attr1;
231 ccl_rpn_delete (rpn->u.p[0]);
232 ccl_rpn_delete (rpn->u.p[1]);
235 free (rpn->u.t.term);
236 for (attr = rpn->u.t.attr_list; attr; attr = attr1)
243 free (rpn->u.setname);
246 ccl_rpn_delete (rpn->u.p[0]);
247 ccl_rpn_delete (rpn->u.p[1]);
253 static struct ccl_rpn_node *find_spec (CCL_parser cclp,
254 struct ccl_rpn_attr **qa);
256 static int is_term_ok (int look, int *list)
258 for (;*list >= 0; list++)
264 static struct ccl_rpn_node *search_terms (CCL_parser cclp,
265 struct ccl_rpn_attr **qa);
268 * add_attr: Add attribute (type/value) to RPN term node.
269 * p: RPN node of type term.
270 * type: Type of attribute
271 * value: Value of attribute
273 static void add_attr (struct ccl_rpn_node *p, int type, int value)
275 struct ccl_rpn_attr *n;
277 n = (struct ccl_rpn_attr *)malloc (sizeof(*n));
281 n->next = p->u.t.attr_list;
282 p->u.t.attr_list = n;
286 * search_term: Parse CCL search term.
288 * qa: Qualifier attributes already applied.
289 * return: pointer to node(s); NULL on error.
291 static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
292 struct ccl_rpn_attr **qa,
295 struct ccl_rpn_attr *qa_tmp[2];
296 struct ccl_rpn_node *p;
297 struct ccl_token *lookahead = cclp->look_token;
303 int relation_value = -1;
304 int position_value = -1;
305 int structure_value = -1;
306 int truncation_value = -1;
307 int completeness_value = -1;
309 if (!is_term_ok(KIND, term_list))
311 cclp->error_code = CCL_ERR_TERM_EXPECTED;
314 /* create the term node, but wait a moment before adding the term */
315 p = mk_node (CCL_RPN_TERM);
316 p->u.t.attr_list = NULL;
321 /* no qualifier(s) applied. Use 'term' if it is defined */
325 qa[0] = ccl_qual_search (cclp, "term", 4);
329 /* go through all attributes and add them to the attribute list */
330 for (i=0; qa && qa[i]; i++)
332 struct ccl_rpn_attr *attr;
334 for (attr = qa[i]; attr; attr = attr->next)
336 { /* deal only with REAL attributes (positive) */
340 if (relation_value != -1)
342 relation_value = attr->value;
345 if (position_value != -1)
347 position_value = attr->value;
350 if (structure_value != -1)
352 structure_value = attr->value;
355 if (truncation_value != -1)
357 truncation_value = attr->value;
360 if (completeness_value != -1)
362 completeness_value = attr->value;
365 add_attr (p, attr->type, attr->value);
368 /* go through each TERM token. If no truncation attribute is yet
369 met, then look for left/right truncation markers (?) and
370 set left_trunc/right_trunc/mid_trunc accordingly */
371 for (no = 0; is_term_ok(lookahead->kind, term_list); no++)
373 for (i = 0; i<lookahead->len; i++)
374 if (truncation_value == -1 && lookahead->name[i] == '?')
376 if (no == 0 && i == 0 && lookahead->len >= 1)
378 else if (!is_term_ok(lookahead->next->kind, term_list) &&
379 i == lookahead->len-1 && i >= 1)
384 len += 1+lookahead->len;
385 lookahead = lookahead->next;
387 /* len now holds the number of characters in the RPN term */
388 /* no holds the number of CCL tokens (1 or more) */
390 if (structure_value == -1 &&
391 qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_WP))
392 { /* no structure attribute met. Apply either structure attribute
393 WORD or PHRASE depending on number of CCL tokens */
395 add_attr (p, CCL_BIB1_STR, 2);
397 add_attr (p, CCL_BIB1_STR, 1);
400 /* make the RPN token */
401 p->u.t.term = (char *)malloc (len);
402 ccl_assert (p->u.t.term);
403 p->u.t.term[0] = '\0';
404 for (i = 0; i<no; i++)
406 const char *src_str = cclp->look_token->name;
407 int src_len = cclp->look_token->len;
409 if (i == 0 && left_trunc)
414 else if (i == no-1 && right_trunc)
418 int len = strlen(p->u.t.term);
420 !strchr("-+", *src_str) &&
421 !strchr("-+", p->u.t.term[len-1]))
423 strcat (p->u.t.term, " ");
426 strxcat (p->u.t.term, src_str, src_len);
429 if (left_trunc && right_trunc)
431 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH))
433 cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
437 add_attr (p, CCL_BIB1_TRU, 3);
439 else if (right_trunc)
441 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT))
443 cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
447 add_attr (p, CCL_BIB1_TRU, 1);
451 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT))
453 cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
457 add_attr (p, CCL_BIB1_TRU, 2);
461 if (qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE))
462 add_attr (p, CCL_BIB1_TRU, 100);
467 static struct ccl_rpn_node *search_term (CCL_parser cclp,
468 struct ccl_rpn_attr **qa)
470 static int list[] = {CCL_TOK_TERM, CCL_TOK_COMMA, -1};
471 return search_term_x(cclp, qa, list);
475 * qualifiers: Parse CCL qualifiers and search terms.
477 * la: Token pointer to RELATION token.
478 * qa: Qualifier attributes already applied.
479 * return: pointer to node(s); NULL on error.
481 static struct ccl_rpn_node *qualifiers (CCL_parser cclp, struct ccl_token *la,
482 struct ccl_rpn_attr **qa)
484 struct ccl_token *lookahead = cclp->look_token;
485 struct ccl_rpn_attr **ap;
491 cclp->error_code = CCL_ERR_DOUBLE_QUAL;
495 for (lookahead = cclp->look_token; lookahead != la;
496 lookahead=lookahead->next)
499 for (i=0; qa[i]; i++)
501 ap = (struct ccl_rpn_attr **)malloc ((no+1) * sizeof(*ap));
503 for (i = 0; cclp->look_token != la; i++)
505 ap[i] = ccl_qual_search (cclp, cclp->look_token->name,
506 cclp->look_token->len);
509 cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
514 if (KIND == CCL_TOK_COMMA)
521 if (!qual_val_type (ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER))
523 /* unordered relation */
524 struct ccl_rpn_node *p;
525 if (KIND != CCL_TOK_EQ)
527 cclp->error_code = CCL_ERR_EQ_EXPECTED;
532 if (KIND == CCL_TOK_LP)
535 if (!(p = find_spec (cclp, ap)))
540 if (KIND != CCL_TOK_RP)
542 cclp->error_code = CCL_ERR_RP_EXPECTED;
550 p = search_terms (cclp, ap);
554 /* ordered relation ... */
556 if (cclp->look_token->len == 1)
558 if (cclp->look_token->name[0] == '<')
560 else if (cclp->look_token->name[0] == '=')
562 else if (cclp->look_token->name[0] == '>')
565 else if (cclp->look_token->len == 2)
567 if (!memcmp (cclp->look_token->name, "<=", 2))
569 else if (!memcmp (cclp->look_token->name, ">=", 2))
571 else if (!memcmp (cclp->look_token->name, "<>", 2))
575 cclp->error_code = CCL_ERR_BAD_RELATION;
578 struct ccl_rpn_node *p;
580 ADVANCE; /* skip relation */
581 if (KIND == CCL_TOK_TERM &&
582 cclp->look_token->next->kind == CCL_TOK_MINUS)
584 struct ccl_rpn_node *p1;
585 if (!(p1 = search_term (cclp, ap)))
590 ADVANCE; /* skip '-' */
591 if (KIND == CCL_TOK_TERM) /* = term - term ? */
593 struct ccl_rpn_node *p2;
595 if (!(p2 = search_term (cclp, ap)))
601 p = mk_node (CCL_RPN_AND);
603 add_attr (p1, CCL_BIB1_REL, 4);
605 add_attr (p2, CCL_BIB1_REL, 2);
611 add_attr (p1, CCL_BIB1_REL, 4);
616 else if (KIND == CCL_TOK_MINUS) /* = - term ? */
619 if (!(p = search_term (cclp, ap)))
624 add_attr (p, CCL_BIB1_REL, 2);
628 else if (KIND == CCL_TOK_LP)
631 if (!(p = find_spec (cclp, ap)))
636 if (KIND != CCL_TOK_RP)
638 cclp->error_code = CCL_ERR_RP_EXPECTED;
649 if (!(p = search_terms (cclp, ap)))
654 add_attr (p, CCL_BIB1_REL, rel);
658 cclp->error_code = CCL_ERR_TERM_EXPECTED;
665 * search_terms: Parse CCL search terms - including proximity.
667 * qa: Qualifier attributes already applied.
668 * return: pointer to node(s); NULL on error.
670 static struct ccl_rpn_node *search_terms (CCL_parser cclp,
671 struct ccl_rpn_attr **qa)
673 static int list[] = {
674 CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
675 CCL_TOK_REL, CCL_TOK_MINUS, -1};
676 struct ccl_rpn_node *p1, *p2, *pn;
677 p1 = search_term_x (cclp, qa, list);
682 if (KIND == CCL_TOK_PROX)
685 p2 = search_term_x (cclp, qa, list);
691 pn = mk_node (CCL_RPN_PROX);
696 else if (is_term_ok(KIND, list))
698 p2 = search_term_x (cclp, qa, list);
704 pn = mk_node (CCL_RPN_PROX);
716 * search_elements: Parse CCL search elements
718 * qa: Qualifier attributes already applied.
719 * return: pointer to node(s); NULL on error.
721 static struct ccl_rpn_node *search_elements (CCL_parser cclp,
722 struct ccl_rpn_attr **qa)
724 struct ccl_rpn_node *p1;
725 struct ccl_token *lookahead;
726 if (KIND == CCL_TOK_LP)
729 p1 = find_spec (cclp, qa);
732 if (KIND != CCL_TOK_RP)
734 cclp->error_code = CCL_ERR_RP_EXPECTED;
741 else if (KIND == CCL_TOK_SET)
744 if (KIND == CCL_TOK_EQ)
746 if (KIND != CCL_TOK_TERM)
748 cclp->error_code = CCL_ERR_SETNAME_EXPECTED;
751 p1 = mk_node (CCL_RPN_SET);
752 p1->u.setname = copy_token_name (cclp->look_token);
756 lookahead = cclp->look_token;
758 while (lookahead->kind==CCL_TOK_TERM)
760 lookahead = lookahead->next;
761 if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
762 return qualifiers (cclp, lookahead, qa);
763 if (lookahead->kind != CCL_TOK_COMMA)
765 lookahead = lookahead->next;
767 return search_terms (cclp, qa);
771 * find_spec: Parse CCL find specification
773 * qa: Qualifier attributes already applied.
774 * return: pointer to node(s); NULL on error.
776 static struct ccl_rpn_node *find_spec (CCL_parser cclp,
777 struct ccl_rpn_attr **qa)
779 struct ccl_rpn_node *p1, *p2, *pn;
780 if (!(p1 = search_elements (cclp, qa)))
788 p2 = search_elements (cclp, qa);
794 pn = mk_node (CCL_RPN_AND);
801 p2 = search_elements (cclp, qa);
807 pn = mk_node (CCL_RPN_OR);
814 p2 = search_elements (cclp, qa);
820 pn = mk_node (CCL_RPN_NOT);
831 struct ccl_rpn_node *ccl_parser_find (CCL_parser cclp, struct ccl_token *list)
833 struct ccl_rpn_node *p;
835 cclp->look_token = list;
836 p = find_spec (cclp, NULL);
837 if (p && KIND != CCL_TOK_EOL)
839 if (KIND == CCL_TOK_RP)
840 cclp->error_code = CCL_ERR_BAD_RP;
842 cclp->error_code = CCL_ERR_OP_EXPECTED;
846 cclp->error_pos = cclp->look_token->name;
848 cclp->error_code = CCL_ERR_OK;
850 cclp->error_code = cclp->error_code;
855 * ccl_find: Parse CCL find - token representation
856 * bibset: Bibset to be used for the parsing
857 * list: List of tokens
858 * error: Pointer to integer. Holds error no. on completion.
859 * pos: Pointer to char position. Holds approximate error position.
860 * return: RPN tree on successful completion; NULL otherwise.
862 struct ccl_rpn_node *ccl_find (CCL_bibset bibset, struct ccl_token *list,
863 int *error, const char **pos)
865 struct ccl_rpn_node *p;
866 CCL_parser cclp = ccl_parser_create ();
868 cclp->bibset = bibset;
870 p = ccl_parser_find (cclp, list);
872 *error = cclp->error_code;
873 *pos = cclp->error_pos;
875 ccl_parser_destroy (cclp);
881 * ccl_find_str: Parse CCL find - string representation
882 * bibset: Bibset to be used for the parsing
883 * str: String to be parsed
884 * error: Pointer to integer. Holds error no. on completion.
885 * pos: Pointer to char position. Holds approximate error position.
886 * return: RPN tree on successful completion; NULL otherwise.
888 struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset, const char *str,
889 int *error, int *pos)
891 CCL_parser cclp = ccl_parser_create ();
892 struct ccl_token *list;
893 struct ccl_rpn_node *p;
895 cclp->bibset = bibset;
897 list = ccl_parser_tokenize (cclp, str);
898 p = ccl_parser_find (cclp, list);
900 *error = cclp->error_code;
902 *pos = cclp->error_pos - str;
903 ccl_parser_destroy (cclp);
904 ccl_token_del (list);