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 1.14 2000-01-31 13:15:21 adam
49 * Removed uses of assert(3). Cleanup of ODR. CCL parser update so
50 * that some characters are not surrounded by spaces in resulting term.
53 * Revision 1.13 1999/12/22 13:13:32 adam
54 * Search terms may include "operators" without causing error.
56 * Revision 1.12 1999/11/30 13:47:11 adam
57 * Improved installation. Moved header files to include/yaz.
59 * Revision 1.11 1999/03/31 11:15:37 adam
60 * Fixed memory leaks in ccl_find_str and ccl_qual_rm.
62 * Revision 1.10 1998/02/11 11:53:33 adam
63 * Changed code so that it compiles as C++.
65 * Revision 1.9 1997/09/29 08:56:37 adam
66 * Changed CCL parser to be thread safe. New type, CCL_parser, declared
67 * and a create/destructers ccl_parser_create/ccl_parser/destory has
70 * Revision 1.8 1997/09/01 08:48:11 adam
71 * New windows NT/95 port using MSV5.0. Only a few changes made
74 * Revision 1.7 1997/05/14 06:53:26 adam
77 * Revision 1.6 1997/04/30 08:52:06 quinn
80 * Revision 1.5 1996/10/11 15:00:24 adam
81 * CCL parser from Europagate Email gateway 1.0.
83 * Revision 1.16 1996/01/08 08:41:13 adam
84 * Removed unused function.
86 * Revision 1.15 1995/07/20 08:14:34 adam
87 * Qualifiers were observed too often. Instead tokens are treated as
88 * qualifiers only when separated by comma.
90 * Revision 1.14 1995/05/16 09:39:26 adam
93 * Revision 1.13 1995/04/17 09:31:42 adam
94 * Improved handling of qualifiers. Aliases or reserved words.
96 * Revision 1.12 1995/03/20 15:27:43 adam
99 * Revision 1.11 1995/02/23 08:31:59 adam
102 * Revision 1.9 1995/02/16 13:20:06 adam
105 * Revision 1.8 1995/02/14 19:59:42 adam
106 * Removed a syntax error.
108 * Revision 1.7 1995/02/14 19:55:10 adam
109 * Header files ccl.h/cclp.h are gone! They have been merged an
110 * moved to ../include/ccl.h.
111 * Node kind(s) in ccl_rpn_node have changed names.
113 * Revision 1.6 1995/02/14 16:20:55 adam
114 * Qualifiers are read from a file now.
116 * Revision 1.5 1995/02/14 14:12:41 adam
117 * Ranges for ordered qualfiers implemented (e.g. pd=1980-1990).
119 * Revision 1.4 1995/02/14 13:16:29 adam
120 * Left and/or right truncation implemented.
122 * Revision 1.3 1995/02/14 10:25:56 adam
123 * The constructions 'qualifier rel term ...' implemented.
125 * Revision 1.2 1995/02/13 15:15:07 adam
126 * Added handling of qualifiers. Not finished yet.
128 * Revision 1.1 1995/02/13 12:35:20 adam
129 * First version of CCL. Qualifiers aren't handled yet.
139 /* returns type of current lookahead */
140 #define KIND (cclp->look_token->kind)
142 /* move one token forward */
143 #define ADVANCE cclp->look_token = cclp->look_token->next
146 * qual_val_type: test for existance of attribute type/value pair.
147 * qa: Attribute array
148 * type: Type of attribute to search for
149 * value: Value of attribute to seach for
150 * return: 1 if found; 0 otherwise.
152 static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value)
155 struct ccl_rpn_attr *q;
159 for (i = 0; (q=qa[i]); i++)
162 if (q->type == type && q->value == value)
170 * strxcat: concatenate strings.
171 * n: Null-terminated Destination string
172 * src: Source string to be appended (not null-terminated)
173 * len: Length of source string.
175 static void strxcat (char *n, const char *src, int len)
185 * copy_token_name: Return copy of CCL token name
186 * tp: Pointer to token info.
187 * return: malloc(3) allocated copy of token name.
189 static char *copy_token_name (struct ccl_token *tp)
191 char *str = (char *)malloc (tp->len + 1);
193 memcpy (str, tp->name, tp->len);
199 * mk_node: Create RPN node.
200 * kind: Type of node.
201 * return: pointer to allocated node.
203 static struct ccl_rpn_node *mk_node (int kind)
205 struct ccl_rpn_node *p;
206 p = (struct ccl_rpn_node *)malloc (sizeof(*p));
213 * ccl_rpn_delete: Delete RPN tree.
214 * rpn: Pointer to tree.
216 void ccl_rpn_delete (struct ccl_rpn_node *rpn)
218 struct ccl_rpn_attr *attr, *attr1;
226 ccl_rpn_delete (rpn->u.p[0]);
227 ccl_rpn_delete (rpn->u.p[1]);
230 free (rpn->u.t.term);
231 for (attr = rpn->u.t.attr_list; attr; attr = attr1)
238 free (rpn->u.setname);
241 ccl_rpn_delete (rpn->u.p[0]);
242 ccl_rpn_delete (rpn->u.p[1]);
248 static struct ccl_rpn_node *find_spec (CCL_parser cclp,
249 struct ccl_rpn_attr **qa);
251 static int is_term_ok (int look, int *list)
253 for (;*list >= 0; list++)
259 static struct ccl_rpn_node *search_terms (CCL_parser cclp,
260 struct ccl_rpn_attr **qa);
263 * add_attr: Add attribute (type/value) to RPN term node.
264 * p: RPN node of type term.
265 * type: Type of attribute
266 * value: Value of attribute
268 static void add_attr (struct ccl_rpn_node *p, int type, int value)
270 struct ccl_rpn_attr *n;
272 n = (struct ccl_rpn_attr *)malloc (sizeof(*n));
276 n->next = p->u.t.attr_list;
277 p->u.t.attr_list = n;
281 * search_term: Parse CCL search term.
283 * qa: Qualifier attributes already applied.
284 * return: pointer to node(s); NULL on error.
286 static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
287 struct ccl_rpn_attr **qa,
290 struct ccl_rpn_node *p;
291 struct ccl_token *lookahead = cclp->look_token;
297 int relation_value = -1;
298 int position_value = -1;
299 int structure_value = -1;
300 int truncation_value = -1;
301 int completeness_value = -1;
303 if (!is_term_ok(KIND, term_list))
305 cclp->error_code = CCL_ERR_TERM_EXPECTED;
308 /* create the term node, but wait a moment before adding the term */
309 p = mk_node (CCL_RPN_TERM);
310 p->u.t.attr_list = NULL;
315 /* no qualifier(s) applied. Use 'term' if it is defined */
317 qa = (struct ccl_rpn_attr **)malloc (2*sizeof(*qa));
319 qa[0] = ccl_qual_search (cclp, "term", 4);
323 /* go through all attributes and add them to the attribute list */
324 for (i=0; qa && qa[i]; i++)
326 struct ccl_rpn_attr *attr;
328 for (attr = qa[i]; attr; attr = attr->next)
330 { /* deal only with REAL attributes (positive) */
334 if (relation_value != -1)
336 relation_value = attr->value;
339 if (position_value != -1)
341 position_value = attr->value;
344 if (structure_value != -1)
346 structure_value = attr->value;
349 if (truncation_value != -1)
351 truncation_value = attr->value;
354 if (completeness_value != -1)
356 completeness_value = attr->value;
359 add_attr (p, attr->type, attr->value);
362 /* go through each TERM token. If no truncation attribute is yet
363 met, then look for left/right truncation markers (?) and
364 set left_trunc/right_trunc/mid_trunc accordingly */
365 for (no = 0; is_term_ok(lookahead->kind, term_list); no++)
367 for (i = 0; i<lookahead->len; i++)
368 if (truncation_value == -1 && lookahead->name[i] == '?')
370 if (no == 0 && i == 0 && lookahead->len >= 1)
372 else if (!is_term_ok(lookahead->next->kind, term_list) &&
373 i == lookahead->len-1 && i >= 1)
378 len += 1+lookahead->len;
379 lookahead = lookahead->next;
381 /* len now holds the number of characters in the RPN term */
382 /* no holds the number of CCL tokens (1 or more) */
384 if (structure_value == -1 &&
385 qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_WP))
386 { /* no structure attribute met. Apply either structure attribute
387 WORD or PHRASE depending on number of CCL tokens */
389 add_attr (p, CCL_BIB1_STR, 2);
391 add_attr (p, CCL_BIB1_STR, 1);
394 /* make the RPN token */
395 p->u.t.term = (char *)malloc (len);
396 ccl_assert (p->u.t.term);
397 p->u.t.term[0] = '\0';
398 for (i = 0; i<no; i++)
400 const char *src_str = cclp->look_token->name;
401 int src_len = cclp->look_token->len;
403 if (i == 0 && left_trunc)
408 else if (i == no-1 && right_trunc)
412 int len = strlen(p->u.t.term);
414 !strchr("-+", *src_str) &&
415 !strchr("-+", p->u.t.term[len-1]))
417 strcat (p->u.t.term, " ");
420 strxcat (p->u.t.term, src_str, src_len);
423 if (left_trunc && right_trunc)
425 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH))
427 cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
432 add_attr (p, CCL_BIB1_TRU, 3);
434 else if (right_trunc)
436 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT))
438 cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
443 add_attr (p, CCL_BIB1_TRU, 1);
447 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT))
449 cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
454 add_attr (p, CCL_BIB1_TRU, 2);
458 if (qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE))
459 add_attr (p, CCL_BIB1_TRU, 100);
464 static struct ccl_rpn_node *search_term (CCL_parser cclp,
465 struct ccl_rpn_attr **qa)
467 static int list[] = {CCL_TOK_TERM, CCL_TOK_COMMA, -1};
468 return search_term_x(cclp, qa, list);
472 * qualifiers: Parse CCL qualifiers and search terms.
474 * la: Token pointer to RELATION token.
475 * qa: Qualifier attributes already applied.
476 * return: pointer to node(s); NULL on error.
478 static struct ccl_rpn_node *qualifiers (CCL_parser cclp, struct ccl_token *la,
479 struct ccl_rpn_attr **qa)
481 struct ccl_token *lookahead = cclp->look_token;
482 struct ccl_rpn_attr **ap;
488 cclp->error_code = CCL_ERR_DOUBLE_QUAL;
492 for (lookahead = cclp->look_token; lookahead != la;
493 lookahead=lookahead->next)
496 for (i=0; qa[i]; i++)
498 ap = (struct ccl_rpn_attr **)malloc ((no+1) * sizeof(*ap));
500 for (i = 0; cclp->look_token != la; i++)
502 ap[i] = ccl_qual_search (cclp, cclp->look_token->name,
503 cclp->look_token->len);
506 cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
511 if (KIND == CCL_TOK_COMMA)
518 if (!qual_val_type (ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER))
520 /* unordered relation */
521 struct ccl_rpn_node *p;
522 if (KIND != CCL_TOK_EQ)
524 cclp->error_code = CCL_ERR_EQ_EXPECTED;
529 if (KIND == CCL_TOK_LP)
532 if (!(p = find_spec (cclp, ap)))
537 if (KIND != CCL_TOK_RP)
539 cclp->error_code = CCL_ERR_RP_EXPECTED;
547 p = search_terms (cclp, ap);
551 /* ordered relation ... */
553 if (cclp->look_token->len == 1)
555 if (cclp->look_token->name[0] == '<')
557 else if (cclp->look_token->name[0] == '=')
559 else if (cclp->look_token->name[0] == '>')
562 else if (cclp->look_token->len == 2)
564 if (!memcmp (cclp->look_token->name, "<=", 2))
566 else if (!memcmp (cclp->look_token->name, ">=", 2))
568 else if (!memcmp (cclp->look_token->name, "<>", 2))
572 cclp->error_code = CCL_ERR_BAD_RELATION;
575 struct ccl_rpn_node *p;
577 ADVANCE; /* skip relation */
578 if (KIND == CCL_TOK_TERM &&
579 cclp->look_token->next->kind == CCL_TOK_MINUS)
581 struct ccl_rpn_node *p1;
582 if (!(p1 = search_term (cclp, ap)))
587 ADVANCE; /* skip '-' */
588 if (KIND == CCL_TOK_TERM) /* = term - term ? */
590 struct ccl_rpn_node *p2;
592 if (!(p2 = search_term (cclp, ap)))
598 p = mk_node (CCL_RPN_AND);
600 add_attr (p1, CCL_BIB1_REL, 4);
602 add_attr (p2, CCL_BIB1_REL, 2);
608 add_attr (p1, CCL_BIB1_REL, 4);
613 else if (KIND == CCL_TOK_MINUS) /* = - term ? */
616 if (!(p = search_term (cclp, ap)))
621 add_attr (p, CCL_BIB1_REL, 2);
625 else if (KIND == CCL_TOK_LP)
628 if (!(p = find_spec (cclp, ap)))
633 if (KIND != CCL_TOK_RP)
635 cclp->error_code = CCL_ERR_RP_EXPECTED;
646 if (!(p = search_terms (cclp, ap)))
651 add_attr (p, CCL_BIB1_REL, rel);
655 cclp->error_code = CCL_ERR_TERM_EXPECTED;
662 * search_terms: Parse CCL search terms - including proximity.
664 * qa: Qualifier attributes already applied.
665 * return: pointer to node(s); NULL on error.
667 static struct ccl_rpn_node *search_terms (CCL_parser cclp,
668 struct ccl_rpn_attr **qa)
670 static int list[] = {
671 CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
672 CCL_TOK_REL, CCL_TOK_MINUS, -1};
673 struct ccl_rpn_node *p1, *p2, *pn;
674 p1 = search_term_x (cclp, qa, list);
679 if (KIND == CCL_TOK_PROX)
682 p2 = search_term_x (cclp, qa, list);
688 pn = mk_node (CCL_RPN_PROX);
693 else if (is_term_ok(KIND, list))
695 p2 = search_term_x (cclp, qa, list);
701 pn = mk_node (CCL_RPN_PROX);
713 * search_elements: Parse CCL search elements
715 * qa: Qualifier attributes already applied.
716 * return: pointer to node(s); NULL on error.
718 static struct ccl_rpn_node *search_elements (CCL_parser cclp,
719 struct ccl_rpn_attr **qa)
721 struct ccl_rpn_node *p1;
722 struct ccl_token *lookahead;
723 if (KIND == CCL_TOK_LP)
726 p1 = find_spec (cclp, qa);
729 if (KIND != CCL_TOK_RP)
731 cclp->error_code = CCL_ERR_RP_EXPECTED;
738 else if (KIND == CCL_TOK_SET)
741 if (KIND == CCL_TOK_EQ)
743 if (KIND != CCL_TOK_TERM)
745 cclp->error_code = CCL_ERR_SETNAME_EXPECTED;
748 p1 = mk_node (CCL_RPN_SET);
749 p1->u.setname = copy_token_name (cclp->look_token);
753 lookahead = cclp->look_token;
755 while (lookahead->kind==CCL_TOK_TERM)
757 lookahead = lookahead->next;
758 if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
759 return qualifiers (cclp, lookahead, qa);
760 if (lookahead->kind != CCL_TOK_COMMA)
762 lookahead = lookahead->next;
764 return search_terms (cclp, qa);
768 * find_spec: Parse CCL find specification
770 * qa: Qualifier attributes already applied.
771 * return: pointer to node(s); NULL on error.
773 static struct ccl_rpn_node *find_spec (CCL_parser cclp,
774 struct ccl_rpn_attr **qa)
776 struct ccl_rpn_node *p1, *p2, *pn;
777 if (!(p1 = search_elements (cclp, qa)))
785 p2 = search_elements (cclp, qa);
791 pn = mk_node (CCL_RPN_AND);
798 p2 = search_elements (cclp, qa);
804 pn = mk_node (CCL_RPN_OR);
811 p2 = search_elements (cclp, qa);
817 pn = mk_node (CCL_RPN_NOT);
828 struct ccl_rpn_node *ccl_parser_find (CCL_parser cclp, struct ccl_token *list)
830 struct ccl_rpn_node *p;
832 cclp->look_token = list;
833 p = find_spec (cclp, NULL);
834 if (p && KIND != CCL_TOK_EOL)
836 if (KIND == CCL_TOK_RP)
837 cclp->error_code = CCL_ERR_BAD_RP;
839 cclp->error_code = CCL_ERR_OP_EXPECTED;
843 cclp->error_pos = cclp->look_token->name;
845 cclp->error_code = CCL_ERR_OK;
847 cclp->error_code = cclp->error_code;
852 * ccl_find: Parse CCL find - token representation
853 * bibset: Bibset to be used for the parsing
854 * list: List of tokens
855 * error: Pointer to integer. Holds error no. on completion.
856 * pos: Pointer to char position. Holds approximate error position.
857 * return: RPN tree on successful completion; NULL otherwise.
859 struct ccl_rpn_node *ccl_find (CCL_bibset bibset, struct ccl_token *list,
860 int *error, const char **pos)
862 struct ccl_rpn_node *p;
863 CCL_parser cclp = ccl_parser_create ();
865 cclp->bibset = bibset;
867 p = ccl_parser_find (cclp, list);
869 *error = cclp->error_code;
870 *pos = cclp->error_pos;
872 ccl_parser_destroy (cclp);
878 * ccl_find_str: Parse CCL find - string representation
879 * bibset: Bibset to be used for the parsing
880 * str: String to be parsed
881 * error: Pointer to integer. Holds error no. on completion.
882 * pos: Pointer to char position. Holds approximate error position.
883 * return: RPN tree on successful completion; NULL otherwise.
885 struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset, const char *str,
886 int *error, int *pos)
888 CCL_parser cclp = ccl_parser_create ();
889 struct ccl_token *list;
890 struct ccl_rpn_node *p;
892 cclp->bibset = bibset;
894 list = ccl_parser_tokenize (cclp, str);
895 p = ccl_parser_find (cclp, list);
897 *error = cclp->error_code;
899 *pos = cclp->error_pos - str;
900 ccl_parser_destroy (cclp);
901 ccl_token_del (list);