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 1.8 1997-09-01 08:48:11 adam
49 * New windows NT/95 port using MSV5.0. Only a few changes made
52 * Revision 1.7 1997/05/14 06:53:26 adam
55 * Revision 1.6 1997/04/30 08:52:06 quinn
58 * Revision 1.5 1996/10/11 15:00:24 adam
59 * CCL parser from Europagate Email gateway 1.0.
61 * Revision 1.16 1996/01/08 08:41:13 adam
62 * Removed unused function.
64 * Revision 1.15 1995/07/20 08:14:34 adam
65 * Qualifiers were observed too often. Instead tokens are treated as
66 * qualifiers only when separated by comma.
68 * Revision 1.14 1995/05/16 09:39:26 adam
71 * Revision 1.13 1995/04/17 09:31:42 adam
72 * Improved handling of qualifiers. Aliases or reserved words.
74 * Revision 1.12 1995/03/20 15:27:43 adam
77 * Revision 1.11 1995/02/23 08:31:59 adam
80 * Revision 1.9 1995/02/16 13:20:06 adam
83 * Revision 1.8 1995/02/14 19:59:42 adam
84 * Removed a syntax error.
86 * Revision 1.7 1995/02/14 19:55:10 adam
87 * Header files ccl.h/cclp.h are gone! They have been merged an
88 * moved to ../include/ccl.h.
89 * Node kind(s) in ccl_rpn_node have changed names.
91 * Revision 1.6 1995/02/14 16:20:55 adam
92 * Qualifiers are read from a file now.
94 * Revision 1.5 1995/02/14 14:12:41 adam
95 * Ranges for ordered qualfiers implemented (e.g. pd=1980-1990).
97 * Revision 1.4 1995/02/14 13:16:29 adam
98 * Left and/or right truncation implemented.
100 * Revision 1.3 1995/02/14 10:25:56 adam
101 * The constructions 'qualifier rel term ...' implemented.
103 * Revision 1.2 1995/02/13 15:15:07 adam
104 * Added handling of qualifiers. Not finished yet.
106 * Revision 1.1 1995/02/13 12:35:20 adam
107 * First version of CCL. Qualifiers aren't handled yet.
118 /* current lookahead token */
119 static struct ccl_token *look_token;
121 /* holds error no if error occur */
122 static int ccl_error;
125 static CCL_bibset bibset;
127 /* returns type of current lookahead */
128 #define KIND (look_token->kind)
130 /* move one token forward */
131 #define ADVANCE look_token = look_token->next
134 * qual_val_type: test for existance of attribute type/value pair.
135 * qa: Attribute array
136 * type: Type of attribute to search for
137 * value: Value of attribute to seach for
138 * return: 1 if found; 0 otherwise.
140 static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value)
143 struct ccl_rpn_attr *q;
147 for (i = 0; (q=qa[i]); i++)
150 if (q->type == type && q->value == value)
158 * strxcat: concatenate strings.
159 * n: Null-terminated Destination string
160 * src: Source string to be appended (not null-terminated)
161 * len: Length of source string.
163 static void strxcat (char *n, const char *src, int len)
173 * copy_token_name: Return copy of CCL token name
174 * tp: Pointer to token info.
175 * return: malloc(3) allocated copy of token name.
177 static char *copy_token_name (struct ccl_token *tp)
179 char *str = malloc (tp->len + 1);
181 memcpy (str, tp->name, tp->len);
187 * mk_node: Create RPN node.
188 * kind: Type of node.
189 * return: pointer to allocated node.
191 static struct ccl_rpn_node *mk_node (int kind)
193 struct ccl_rpn_node *p;
194 p = malloc (sizeof(*p));
201 * ccl_rpn_delete: Delete RPN tree.
202 * rpn: Pointer to tree.
204 void ccl_rpn_delete (struct ccl_rpn_node *rpn)
206 struct ccl_rpn_attr *attr, *attr1;
214 ccl_rpn_delete (rpn->u.p[0]);
215 ccl_rpn_delete (rpn->u.p[1]);
218 free (rpn->u.t.term);
219 for (attr = rpn->u.t.attr_list; attr; attr = attr1)
226 free (rpn->u.setname);
229 ccl_rpn_delete (rpn->u.p[0]);
230 ccl_rpn_delete (rpn->u.p[1]);
236 static struct ccl_rpn_node *find_spec (struct ccl_rpn_attr **qa);
237 static struct ccl_rpn_node *search_terms (struct ccl_rpn_attr **qa);
240 * add_attr: Add attribute (type/value) to RPN term node.
241 * p: RPN node of type term.
242 * type: Type of attribute
243 * value: Value of attribute
245 static void add_attr (struct ccl_rpn_node *p, int type, int value)
247 struct ccl_rpn_attr *n;
249 n = malloc (sizeof(*n));
253 n->next = p->u.t.attr_list;
254 p->u.t.attr_list = n;
258 * search_term: Parse CCL search term.
259 * qa: Qualifier attributes already applied.
260 * return: pointer to node(s); NULL on error.
262 static struct ccl_rpn_node *search_term (struct ccl_rpn_attr **qa)
264 struct ccl_rpn_node *p;
265 struct ccl_token *lookahead = look_token;
271 int relation_value = -1;
272 int position_value = -1;
273 int structure_value = -1;
274 int truncation_value = -1;
275 int completeness_value = -1;
277 if (KIND != CCL_TOK_TERM)
279 ccl_error = CCL_ERR_TERM_EXPECTED;
282 /* create the term node, but wait a moment before adding the term */
283 p = mk_node (CCL_RPN_TERM);
284 p->u.t.attr_list = NULL;
289 /* no qualifier(s) applied. Use 'term' if it is defined */
291 qa = malloc (2*sizeof(*qa));
293 qa[0] = ccl_qual_search (bibset, "term", 4);
297 /* go through all attributes and add them to the attribute list */
298 for (i=0; qa && qa[i]; i++)
300 struct ccl_rpn_attr *attr;
302 for (attr = qa[i]; attr; attr = attr->next)
304 { /* deal only with REAL attributes (positive) */
308 if (relation_value != -1)
310 relation_value = attr->value;
313 if (position_value != -1)
315 position_value = attr->value;
318 if (structure_value != -1)
320 structure_value = attr->value;
323 if (truncation_value != -1)
325 truncation_value = attr->value;
328 if (completeness_value != -1)
330 completeness_value = attr->value;
333 add_attr (p, attr->type, attr->value);
336 /* go through each TERM token. If no truncation attribute is yet
337 met, then look for left/right truncation markers (?) and
338 set left_trunc/right_trunc/mid_trunc accordingly */
339 for (no = 0; lookahead->kind == CCL_TOK_TERM; no++)
341 for (i = 0; i<lookahead->len; i++)
342 if (truncation_value == -1 && lookahead->name[i] == '?')
344 if (no == 0 && i == 0 && lookahead->len >= 1)
346 else if (lookahead->next->kind != CCL_TOK_TERM &&
347 i == lookahead->len-1 && i >= 1)
352 len += 1+lookahead->len;
353 lookahead = lookahead->next;
355 /* len now holds the number of characters in the RPN term */
356 /* no holds the number of CCL tokens (1 or more) */
358 if (structure_value == -1 &&
359 qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_WP))
360 { /* no structure attribute met. Apply either structure attribute
361 WORD or PHRASE depending on number of CCL tokens */
363 add_attr (p, CCL_BIB1_STR, 2);
365 add_attr (p, CCL_BIB1_STR, 1);
368 /* make the RPN token */
369 p->u.t.term = malloc (len);
370 assert (p->u.t.term);
371 p->u.t.term[0] = '\0';
372 for (i = 0; i<no; i++)
374 const char *src_str = look_token->name;
375 int src_len = look_token->len;
377 if (i == 0 && left_trunc)
382 else if (i == no-1 && right_trunc)
385 strcat (p->u.t.term, " ");
386 strxcat (p->u.t.term, src_str, src_len);
389 if (left_trunc && right_trunc)
391 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH))
393 ccl_error = CCL_ERR_TRUNC_NOT_BOTH;
398 add_attr (p, CCL_BIB1_TRU, 3);
400 else if (right_trunc)
402 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT))
404 ccl_error = CCL_ERR_TRUNC_NOT_RIGHT;
409 add_attr (p, CCL_BIB1_TRU, 1);
413 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT))
415 ccl_error = CCL_ERR_TRUNC_NOT_LEFT;
420 add_attr (p, CCL_BIB1_TRU, 2);
424 if (qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE))
425 add_attr (p, CCL_BIB1_TRU, 100);
431 * qualifiers: Parse CCL qualifiers and search terms.
432 * la: Token pointer to RELATION token.
433 * qa: Qualifier attributes already applied.
434 * return: pointer to node(s); NULL on error.
436 static struct ccl_rpn_node *qualifiers (struct ccl_token *la,
437 struct ccl_rpn_attr **qa)
439 struct ccl_token *lookahead = look_token;
440 struct ccl_rpn_attr **ap;
446 ccl_error = CCL_ERR_DOUBLE_QUAL;
450 for (lookahead = look_token; lookahead != la; lookahead=lookahead->next)
453 for (i=0; qa[i]; i++)
455 ap = malloc ((no+1) * sizeof(*ap));
457 for (i = 0; look_token != la; i++)
459 ap[i] = ccl_qual_search (bibset, look_token->name, look_token->len);
462 ccl_error = CCL_ERR_UNKNOWN_QUAL;
467 if (KIND == CCL_TOK_COMMA)
474 if (!qual_val_type (ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER))
476 /* unordered relation */
477 struct ccl_rpn_node *p;
478 if (KIND != CCL_TOK_EQ)
480 ccl_error = CCL_ERR_EQ_EXPECTED;
485 if (KIND == CCL_TOK_LP)
488 if (!(p = find_spec (ap)))
493 if (KIND != CCL_TOK_RP)
495 ccl_error = CCL_ERR_RP_EXPECTED;
503 p = search_terms (ap);
508 if (look_token->len == 1)
510 if (look_token->name[0] == '<')
512 else if (look_token->name[0] == '=')
514 else if (look_token->name[0] == '>')
517 else if (look_token->len == 2)
519 if (!memcmp (look_token->name, "<=", 2))
521 else if (!memcmp (look_token->name, ">=", 2))
523 else if (!memcmp (look_token->name, "<>", 2))
527 ccl_error = CCL_ERR_BAD_RELATION;
530 struct ccl_rpn_node *p;
532 ADVANCE; /* skip relation */
533 if (KIND == CCL_TOK_TERM && look_token->next->kind == CCL_TOK_MINUS)
535 struct ccl_rpn_node *p1;
536 if (!(p1 = search_term (ap)))
541 ADVANCE; /* skip '-' */
542 if (KIND == CCL_TOK_TERM) /* = term - term ? */
544 struct ccl_rpn_node *p2;
546 if (!(p2 = search_term (ap)))
552 p = mk_node (CCL_RPN_AND);
554 add_attr (p1, CCL_BIB1_REL, 4);
556 add_attr (p2, CCL_BIB1_REL, 2);
562 add_attr (p1, CCL_BIB1_REL, 4);
567 else if (KIND == CCL_TOK_MINUS) /* = - term ? */
570 if (!(p = search_term (ap)))
575 add_attr (p, CCL_BIB1_REL, 2);
579 else if (KIND == CCL_TOK_LP)
582 if (!(p = find_spec (ap)))
587 if (KIND != CCL_TOK_RP)
589 ccl_error = CCL_ERR_RP_EXPECTED;
600 if (!(p = search_terms (ap)))
605 add_attr (p, CCL_BIB1_REL, rel);
609 ccl_error = CCL_ERR_TERM_EXPECTED;
616 * search_terms: Parse CCL search terms - including proximity.
617 * qa: Qualifier attributes already applied.
618 * return: pointer to node(s); NULL on error.
620 static struct ccl_rpn_node *search_terms (struct ccl_rpn_attr **qa)
622 struct ccl_rpn_node *p1, *p2, *pn;
623 p1 = search_term (qa);
628 if (KIND == CCL_TOK_PROX)
631 p2 = search_term (qa);
637 pn = mk_node (CCL_RPN_PROX);
642 else if (KIND == CCL_TOK_TERM)
644 p2 = search_term (qa);
650 pn = mk_node (CCL_RPN_PROX);
662 * search_elements: Parse CCL search elements
663 * qa: Qualifier attributes already applied.
664 * return: pointer to node(s); NULL on error.
666 static struct ccl_rpn_node *search_elements (struct ccl_rpn_attr **qa)
668 struct ccl_rpn_node *p1;
669 struct ccl_token *lookahead;
670 if (KIND == CCL_TOK_LP)
676 if (KIND != CCL_TOK_RP)
678 ccl_error = CCL_ERR_RP_EXPECTED;
685 else if (KIND == CCL_TOK_SET)
688 if (KIND == CCL_TOK_EQ)
690 if (KIND != CCL_TOK_TERM)
692 ccl_error = CCL_ERR_SETNAME_EXPECTED;
695 p1 = mk_node (CCL_RPN_SET);
696 p1->u.setname = copy_token_name (look_token);
700 lookahead = look_token;
702 while (lookahead->kind==CCL_TOK_TERM)
704 lookahead = lookahead->next;
705 if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
706 return qualifiers (lookahead, qa);
707 if (lookahead->kind != CCL_TOK_COMMA)
709 lookahead = lookahead->next;
711 return search_terms (qa);
715 * find_spec: Parse CCL find specification
716 * qa: Qualifier attributes already applied.
717 * return: pointer to node(s); NULL on error.
719 static struct ccl_rpn_node *find_spec (struct ccl_rpn_attr **qa)
721 struct ccl_rpn_node *p1, *p2, *pn;
722 if (!(p1 = search_elements (qa)))
730 p2 = search_elements (qa);
736 pn = mk_node (CCL_RPN_AND);
743 p2 = search_elements (qa);
749 pn = mk_node (CCL_RPN_OR);
756 p2 = search_elements (qa);
762 pn = mk_node (CCL_RPN_NOT);
774 * ccl_find: Parse CCL find - token representation
775 * abibset: Bibset to be used for the parsing
776 * list: List of tokens
777 * error: Pointer to integer. Holds error no. on completion.
778 * pos: Pointer to char position. Holds approximate error position.
779 * return: RPN tree on successful completion; NULL otherwise.
781 struct ccl_rpn_node *ccl_find (CCL_bibset abibset, struct ccl_token *list,
782 int *error, const char **pos)
784 struct ccl_rpn_node *p;
788 p = find_spec (NULL);
789 if (p && KIND != CCL_TOK_EOL)
791 if (KIND == CCL_TOK_RP)
792 ccl_error = CCL_ERR_BAD_RP;
794 ccl_error = CCL_ERR_OP_EXPECTED;
798 *pos = look_token->name;
807 * ccl_find_str: Parse CCL find - string representation
808 * bibset: Bibset to be used for the parsing
809 * str: String to be parsed
810 * error: Pointer to integer. Holds error no. on completion.
811 * pos: Pointer to char position. Holds approximate error position.
812 * return: RPN tree on successful completion; NULL otherwise.
814 struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset, const char *str,
815 int *error, int *pos)
817 struct ccl_token *list;
818 struct ccl_rpn_node *rpn;
819 const char *char_pos;
821 list = ccl_tokenize (str);
822 rpn = ccl_find (bibset, list, error, &char_pos);
824 *pos = char_pos - str;