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 1.19 2000-11-16 09:58:02 adam
49 * Implemented local AttributeSet setting for CCL field maps.
51 * Revision 1.18 2000/10/17 19:50:28 adam
52 * Implemented and-list and or-list for CCL module.
54 * Revision 1.17 2000/05/01 09:36:50 adam
55 * Range operator only treated in ordered ranges so that minus (-) can be
56 * used for, say, the and-not operator.
58 * Revision 1.16 2000/03/14 09:06:11 adam
59 * Added POSIX threads support for frontend server.
61 * Revision 1.15 2000/02/24 23:49:13 adam
62 * Fixed memory allocation problem.
64 * Revision 1.14 2000/01/31 13:15:21 adam
65 * Removed uses of assert(3). Cleanup of ODR. CCL parser update so
66 * that some characters are not surrounded by spaces in resulting term.
69 * Revision 1.13 1999/12/22 13:13:32 adam
70 * Search terms may include "operators" without causing error.
72 * Revision 1.12 1999/11/30 13:47:11 adam
73 * Improved installation. Moved header files to include/yaz.
75 * Revision 1.11 1999/03/31 11:15:37 adam
76 * Fixed memory leaks in ccl_find_str and ccl_qual_rm.
78 * Revision 1.10 1998/02/11 11:53:33 adam
79 * Changed code so that it compiles as C++.
81 * Revision 1.9 1997/09/29 08:56:37 adam
82 * Changed CCL parser to be thread safe. New type, CCL_parser, declared
83 * and a create/destructers ccl_parser_create/ccl_parser/destory has
86 * Revision 1.8 1997/09/01 08:48:11 adam
87 * New windows NT/95 port using MSV5.0. Only a few changes made
90 * Revision 1.7 1997/05/14 06:53:26 adam
93 * Revision 1.6 1997/04/30 08:52:06 quinn
96 * Revision 1.5 1996/10/11 15:00:24 adam
97 * CCL parser from Europagate Email gateway 1.0.
99 * Revision 1.16 1996/01/08 08:41:13 adam
100 * Removed unused function.
102 * Revision 1.15 1995/07/20 08:14:34 adam
103 * Qualifiers were observed too often. Instead tokens are treated as
104 * qualifiers only when separated by comma.
106 * Revision 1.14 1995/05/16 09:39:26 adam
109 * Revision 1.13 1995/04/17 09:31:42 adam
110 * Improved handling of qualifiers. Aliases or reserved words.
112 * Revision 1.12 1995/03/20 15:27:43 adam
115 * Revision 1.11 1995/02/23 08:31:59 adam
118 * Revision 1.9 1995/02/16 13:20:06 adam
121 * Revision 1.8 1995/02/14 19:59:42 adam
122 * Removed a syntax error.
124 * Revision 1.7 1995/02/14 19:55:10 adam
125 * Header files ccl.h/cclp.h are gone! They have been merged an
126 * moved to ../include/ccl.h.
127 * Node kind(s) in ccl_rpn_node have changed names.
129 * Revision 1.6 1995/02/14 16:20:55 adam
130 * Qualifiers are read from a file now.
132 * Revision 1.5 1995/02/14 14:12:41 adam
133 * Ranges for ordered qualfiers implemented (e.g. pd=1980-1990).
135 * Revision 1.4 1995/02/14 13:16:29 adam
136 * Left and/or right truncation implemented.
138 * Revision 1.3 1995/02/14 10:25:56 adam
139 * The constructions 'qualifier rel term ...' implemented.
141 * Revision 1.2 1995/02/13 15:15:07 adam
142 * Added handling of qualifiers. Not finished yet.
144 * Revision 1.1 1995/02/13 12:35:20 adam
145 * First version of CCL. Qualifiers aren't handled yet.
154 /* returns type of current lookahead */
155 #define KIND (cclp->look_token->kind)
157 /* move one token forward */
158 #define ADVANCE cclp->look_token = cclp->look_token->next
161 * qual_val_type: test for existance of attribute type/value pair.
162 * qa: Attribute array
163 * type: Type of attribute to search for
164 * value: Value of attribute to seach for
165 * return: 1 if found; 0 otherwise.
167 static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value,
171 struct ccl_rpn_attr *q;
175 for (i = 0; (q=qa[i]); i++)
178 if (q->type == type && q->value == value)
190 * strxcat: concatenate strings.
191 * n: Null-terminated Destination string
192 * src: Source string to be appended (not null-terminated)
193 * len: Length of source string.
195 static void strxcat (char *n, const char *src, int len)
205 * copy_token_name: Return copy of CCL token name
206 * tp: Pointer to token info.
207 * return: malloc(3) allocated copy of token name.
209 static char *copy_token_name (struct ccl_token *tp)
211 char *str = (char *)malloc (tp->len + 1);
213 memcpy (str, tp->name, tp->len);
219 * mk_node: Create RPN node.
220 * kind: Type of node.
221 * return: pointer to allocated node.
223 static struct ccl_rpn_node *mk_node (int kind)
225 struct ccl_rpn_node *p;
226 p = (struct ccl_rpn_node *)malloc (sizeof(*p));
233 * ccl_rpn_delete: Delete RPN tree.
234 * rpn: Pointer to tree.
236 void ccl_rpn_delete (struct ccl_rpn_node *rpn)
238 struct ccl_rpn_attr *attr, *attr1;
246 ccl_rpn_delete (rpn->u.p[0]);
247 ccl_rpn_delete (rpn->u.p[1]);
250 free (rpn->u.t.term);
251 for (attr = rpn->u.t.attr_list; attr; attr = attr1)
260 free (rpn->u.setname);
263 ccl_rpn_delete (rpn->u.p[0]);
264 ccl_rpn_delete (rpn->u.p[1]);
270 static struct ccl_rpn_node *find_spec (CCL_parser cclp,
271 struct ccl_rpn_attr **qa);
273 static int is_term_ok (int look, int *list)
275 for (;*list >= 0; list++)
281 static struct ccl_rpn_node *search_terms (CCL_parser cclp,
282 struct ccl_rpn_attr **qa);
285 * add_attr: Add attribute (type/value) to RPN term node.
286 * p: RPN node of type term.
287 * type: Type of attribute
288 * value: Value of attribute
289 * set: Attribute set name
291 static void add_attr (struct ccl_rpn_node *p, const char *set,
294 struct ccl_rpn_attr *n;
296 n = (struct ccl_rpn_attr *)malloc (sizeof(*n));
300 n->set = malloc (strlen(set)+1);
301 strcpy (n->set, set);
307 n->next = p->u.t.attr_list;
308 p->u.t.attr_list = n;
312 * search_term: Parse CCL search term.
314 * qa: Qualifier attributes already applied.
315 * return: pointer to node(s); NULL on error.
317 static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
318 struct ccl_rpn_attr **qa,
321 struct ccl_rpn_attr *qa_tmp[2];
322 struct ccl_rpn_node *p_top = 0;
323 struct ccl_token *lookahead = cclp->look_token;
330 /* no qualifier(s) applied. Use 'term' if it is defined */
334 qa[0] = ccl_qual_search (cclp, "term", 4);
337 if (qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_AND_LIST, 0))
339 if (qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_OR_LIST, 0))
343 struct ccl_rpn_node *p;
348 int relation_value = -1;
349 int position_value = -1;
350 int structure_value = -1;
351 int truncation_value = -1;
352 int completeness_value = -1;
355 if (and_list || or_list)
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; no < max && 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;
379 break; /* no more terms . stop . */
384 p = mk_node (CCL_RPN_OR);
386 p = mk_node (CCL_RPN_AND);
388 p = mk_node (CCL_RPN_AND);
393 /* create the term node, but wait a moment before adding the term */
394 p = mk_node (CCL_RPN_TERM);
395 p->u.t.attr_list = NULL;
398 /* make the top node point to us.. */
405 /* go through all attributes and add them to the attribute list */
406 for (i=0; qa && qa[i]; i++)
408 struct ccl_rpn_attr *attr;
410 for (attr = qa[i]; attr; attr = attr->next)
412 { /* deal only with REAL attributes (positive) */
416 if (relation_value != -1)
418 relation_value = attr->value;
421 if (position_value != -1)
423 position_value = attr->value;
426 if (structure_value != -1)
428 structure_value = attr->value;
431 if (truncation_value != -1)
433 truncation_value = attr->value;
434 left_trunc = right_trunc = mid_trunc = 0;
437 if (completeness_value != -1)
439 completeness_value = attr->value;
442 add_attr (p, attr->set, attr->type, attr->value);
445 /* len now holds the number of characters in the RPN term */
446 /* no holds the number of CCL tokens (1 or more) */
448 if (structure_value == -1 &&
449 qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_WP, &attset))
450 { /* no structure attribute met. Apply either structure attribute
451 WORD or PHRASE depending on number of CCL tokens */
453 add_attr (p, attset, CCL_BIB1_STR, 2);
455 add_attr (p, attset, CCL_BIB1_STR, 1);
458 /* make the RPN token */
459 p->u.t.term = (char *)malloc (len);
460 ccl_assert (p->u.t.term);
461 p->u.t.term[0] = '\0';
462 for (i = 0; i<no; i++)
464 const char *src_str = cclp->look_token->name;
465 int src_len = cclp->look_token->len;
467 if (i == 0 && left_trunc)
472 else if (i == no-1 && right_trunc)
476 int len = strlen(p->u.t.term);
478 !strchr("-+", *src_str) &&
479 !strchr("-+", p->u.t.term[len-1]))
481 strcat (p->u.t.term, " ");
484 strxcat (p->u.t.term, src_str, src_len);
487 if (left_trunc && right_trunc)
489 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH,
492 cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
496 add_attr (p, attset, CCL_BIB1_TRU, 3);
498 else if (right_trunc)
500 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT,
503 cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
507 add_attr (p, attset, CCL_BIB1_TRU, 1);
511 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT,
514 cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
518 add_attr (p, attset, CCL_BIB1_TRU, 2);
522 if (qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE,
524 add_attr (p, attset, CCL_BIB1_TRU, 100);
528 cclp->error_code = CCL_ERR_TERM_EXPECTED;
532 static struct ccl_rpn_node *search_term (CCL_parser cclp,
533 struct ccl_rpn_attr **qa)
535 static int list[] = {CCL_TOK_TERM, CCL_TOK_COMMA, -1};
536 return search_term_x(cclp, qa, list);
540 * qualifiers: Parse CCL qualifiers and search terms.
542 * la: Token pointer to RELATION token.
543 * qa: Qualifier attributes already applied.
544 * return: pointer to node(s); NULL on error.
546 static struct ccl_rpn_node *qualifiers (CCL_parser cclp, struct ccl_token *la,
547 struct ccl_rpn_attr **qa)
549 struct ccl_token *lookahead = cclp->look_token;
550 struct ccl_rpn_attr **ap;
557 cclp->error_code = CCL_ERR_DOUBLE_QUAL;
561 for (lookahead = cclp->look_token; lookahead != la;
562 lookahead=lookahead->next)
565 for (i=0; qa[i]; i++)
567 ap = (struct ccl_rpn_attr **)malloc ((no+1) * sizeof(*ap));
569 for (i = 0; cclp->look_token != la; i++)
571 ap[i] = ccl_qual_search (cclp, cclp->look_token->name,
572 cclp->look_token->len);
575 cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
580 if (KIND == CCL_TOK_COMMA)
587 if (!qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset))
589 /* unordered relation */
590 struct ccl_rpn_node *p;
591 if (KIND != CCL_TOK_EQ)
593 cclp->error_code = CCL_ERR_EQ_EXPECTED;
598 if (KIND == CCL_TOK_LP)
601 if (!(p = find_spec (cclp, ap)))
606 if (KIND != CCL_TOK_RP)
608 cclp->error_code = CCL_ERR_RP_EXPECTED;
616 p = search_terms (cclp, ap);
620 /* ordered relation ... */
622 if (cclp->look_token->len == 1)
624 if (cclp->look_token->name[0] == '<')
626 else if (cclp->look_token->name[0] == '=')
628 else if (cclp->look_token->name[0] == '>')
631 else if (cclp->look_token->len == 2)
633 if (!memcmp (cclp->look_token->name, "<=", 2))
635 else if (!memcmp (cclp->look_token->name, ">=", 2))
637 else if (!memcmp (cclp->look_token->name, "<>", 2))
641 cclp->error_code = CCL_ERR_BAD_RELATION;
644 struct ccl_rpn_node *p;
646 ADVANCE; /* skip relation */
647 if (KIND == CCL_TOK_TERM &&
648 cclp->look_token->next->len == 1 &&
649 cclp->look_token->next->name[0] == '-')
651 struct ccl_rpn_node *p1;
652 if (!(p1 = search_term (cclp, ap)))
657 ADVANCE; /* skip '-' */
658 if (KIND == CCL_TOK_TERM) /* = term - term ? */
660 struct ccl_rpn_node *p2;
662 if (!(p2 = search_term (cclp, ap)))
668 p = mk_node (CCL_RPN_AND);
670 add_attr (p1, attset, CCL_BIB1_REL, 4);
672 add_attr (p2, attset, CCL_BIB1_REL, 2);
678 add_attr (p1, attset, CCL_BIB1_REL, 4);
683 else if (cclp->look_token->len == 1 &&
684 cclp->look_token->name[0] == '"') /* = - term ? */
687 if (!(p = search_term (cclp, ap)))
692 add_attr (p, attset, CCL_BIB1_REL, 2);
696 else if (KIND == CCL_TOK_LP)
699 if (!(p = find_spec (cclp, ap)))
704 if (KIND != CCL_TOK_RP)
706 cclp->error_code = CCL_ERR_RP_EXPECTED;
717 if (!(p = search_terms (cclp, ap)))
722 add_attr (p, attset, CCL_BIB1_REL, rel);
726 cclp->error_code = CCL_ERR_TERM_EXPECTED;
733 * search_terms: Parse CCL search terms - including proximity.
735 * qa: Qualifier attributes already applied.
736 * return: pointer to node(s); NULL on error.
738 static struct ccl_rpn_node *search_terms (CCL_parser cclp,
739 struct ccl_rpn_attr **qa)
741 static int list[] = {
742 CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ, CCL_TOK_REL, -1};
743 struct ccl_rpn_node *p1, *p2, *pn;
744 p1 = search_term_x (cclp, qa, list);
749 if (KIND == CCL_TOK_PROX)
752 p2 = search_term_x (cclp, qa, list);
758 pn = mk_node (CCL_RPN_PROX);
763 else if (is_term_ok(KIND, list))
765 p2 = search_term_x (cclp, qa, list);
771 pn = mk_node (CCL_RPN_PROX);
783 * search_elements: Parse CCL search elements
785 * qa: Qualifier attributes already applied.
786 * return: pointer to node(s); NULL on error.
788 static struct ccl_rpn_node *search_elements (CCL_parser cclp,
789 struct ccl_rpn_attr **qa)
791 struct ccl_rpn_node *p1;
792 struct ccl_token *lookahead;
793 if (KIND == CCL_TOK_LP)
796 p1 = find_spec (cclp, qa);
799 if (KIND != CCL_TOK_RP)
801 cclp->error_code = CCL_ERR_RP_EXPECTED;
808 else if (KIND == CCL_TOK_SET)
811 if (KIND == CCL_TOK_EQ)
813 if (KIND != CCL_TOK_TERM)
815 cclp->error_code = CCL_ERR_SETNAME_EXPECTED;
818 p1 = mk_node (CCL_RPN_SET);
819 p1->u.setname = copy_token_name (cclp->look_token);
823 lookahead = cclp->look_token;
825 while (lookahead->kind==CCL_TOK_TERM)
827 lookahead = lookahead->next;
828 if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
829 return qualifiers (cclp, lookahead, qa);
830 if (lookahead->kind != CCL_TOK_COMMA)
832 lookahead = lookahead->next;
834 return search_terms (cclp, qa);
838 * find_spec: Parse CCL find specification
840 * qa: Qualifier attributes already applied.
841 * return: pointer to node(s); NULL on error.
843 static struct ccl_rpn_node *find_spec (CCL_parser cclp,
844 struct ccl_rpn_attr **qa)
846 struct ccl_rpn_node *p1, *p2, *pn;
847 if (!(p1 = search_elements (cclp, qa)))
855 p2 = search_elements (cclp, qa);
861 pn = mk_node (CCL_RPN_AND);
868 p2 = search_elements (cclp, qa);
874 pn = mk_node (CCL_RPN_OR);
881 p2 = search_elements (cclp, qa);
887 pn = mk_node (CCL_RPN_NOT);
898 struct ccl_rpn_node *ccl_parser_find (CCL_parser cclp, struct ccl_token *list)
900 struct ccl_rpn_node *p;
902 cclp->look_token = list;
903 p = find_spec (cclp, NULL);
904 if (p && KIND != CCL_TOK_EOL)
906 if (KIND == CCL_TOK_RP)
907 cclp->error_code = CCL_ERR_BAD_RP;
909 cclp->error_code = CCL_ERR_OP_EXPECTED;
913 cclp->error_pos = cclp->look_token->name;
915 cclp->error_code = CCL_ERR_OK;
917 cclp->error_code = cclp->error_code;
922 * ccl_find: Parse CCL find - token representation
923 * bibset: Bibset to be used for the parsing
924 * list: List of tokens
925 * error: Pointer to integer. Holds error no. on completion.
926 * pos: Pointer to char position. Holds approximate error position.
927 * return: RPN tree on successful completion; NULL otherwise.
929 struct ccl_rpn_node *ccl_find (CCL_bibset bibset, struct ccl_token *list,
930 int *error, const char **pos)
932 struct ccl_rpn_node *p;
933 CCL_parser cclp = ccl_parser_create ();
935 cclp->bibset = bibset;
937 p = ccl_parser_find (cclp, list);
939 *error = cclp->error_code;
940 *pos = cclp->error_pos;
942 ccl_parser_destroy (cclp);
948 * ccl_find_str: Parse CCL find - string representation
949 * bibset: Bibset to be used for the parsing
950 * str: String to be parsed
951 * error: Pointer to integer. Holds error no. on completion.
952 * pos: Pointer to char position. Holds approximate error position.
953 * return: RPN tree on successful completion; NULL otherwise.
955 struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset, const char *str,
956 int *error, int *pos)
958 CCL_parser cclp = ccl_parser_create ();
959 struct ccl_token *list;
960 struct ccl_rpn_node *p;
962 cclp->bibset = bibset;
964 list = ccl_parser_tokenize (cclp, str);
965 p = ccl_parser_find (cclp, list);
967 *error = cclp->error_code;
969 *pos = cclp->error_pos - str;
970 ccl_parser_destroy (cclp);
971 ccl_token_del (list);