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 1.7 1997-05-14 06:53:26 adam
51 * Revision 1.6 1997/04/30 08:52:06 quinn
54 * Revision 1.5 1996/10/11 15:00:24 adam
55 * CCL parser from Europagate Email gateway 1.0.
57 * Revision 1.16 1996/01/08 08:41:13 adam
58 * Removed unused function.
60 * Revision 1.15 1995/07/20 08:14:34 adam
61 * Qualifiers were observed too often. Instead tokens are treated as
62 * qualifiers only when separated by comma.
64 * Revision 1.14 1995/05/16 09:39:26 adam
67 * Revision 1.13 1995/04/17 09:31:42 adam
68 * Improved handling of qualifiers. Aliases or reserved words.
70 * Revision 1.12 1995/03/20 15:27:43 adam
73 * Revision 1.11 1995/02/23 08:31:59 adam
76 * Revision 1.9 1995/02/16 13:20:06 adam
79 * Revision 1.8 1995/02/14 19:59:42 adam
80 * Removed a syntax error.
82 * Revision 1.7 1995/02/14 19:55:10 adam
83 * Header files ccl.h/cclp.h are gone! They have been merged an
84 * moved to ../include/ccl.h.
85 * Node kind(s) in ccl_rpn_node have changed names.
87 * Revision 1.6 1995/02/14 16:20:55 adam
88 * Qualifiers are read from a file now.
90 * Revision 1.5 1995/02/14 14:12:41 adam
91 * Ranges for ordered qualfiers implemented (e.g. pd=1980-1990).
93 * Revision 1.4 1995/02/14 13:16:29 adam
94 * Left and/or right truncation implemented.
96 * Revision 1.3 1995/02/14 10:25:56 adam
97 * The constructions 'qualifier rel term ...' implemented.
99 * Revision 1.2 1995/02/13 15:15:07 adam
100 * Added handling of qualifiers. Not finished yet.
102 * Revision 1.1 1995/02/13 12:35:20 adam
103 * First version of CCL. Qualifiers aren't handled yet.
114 /* current lookahead token */
115 static struct ccl_token *look_token;
117 /* holds error no if error occur */
118 static int ccl_error;
121 static CCL_bibset bibset;
123 /* returns type of current lookahead */
124 #define KIND (look_token->kind)
126 /* move one token forward */
127 #define ADVANCE look_token = look_token->next
130 * qual_val_type: test for existance of attribute type/value pair.
131 * qa: Attribute array
132 * type: Type of attribute to search for
133 * value: Value of attribute to seach for
134 * return: 1 if found; 0 otherwise.
136 static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value)
139 struct ccl_rpn_attr *q;
143 for (i = 0; (q=qa[i]); i++)
146 if (q->type == type && q->value == value)
154 * strxcat: concatenate strings.
155 * n: Null-terminated Destination string
156 * src: Source string to be appended (not null-terminated)
157 * len: Length of source string.
159 static void strxcat (char *n, const char *src, int len)
169 * copy_token_name: Return copy of CCL token name
170 * tp: Pointer to token info.
171 * return: malloc(3) allocated copy of token name.
173 static char *copy_token_name (struct ccl_token *tp)
175 char *str = malloc (tp->len + 1);
177 memcpy (str, tp->name, tp->len);
183 * mk_node: Create RPN node.
184 * kind: Type of node.
185 * return: pointer to allocated node.
187 static struct ccl_rpn_node *mk_node (int kind)
189 struct ccl_rpn_node *p;
190 p = malloc (sizeof(*p));
197 * ccl_rpn_delete: Delete RPN tree.
198 * rpn: Pointer to tree.
200 void ccl_rpn_delete (struct ccl_rpn_node *rpn)
202 struct ccl_rpn_attr *attr, *attr1;
210 ccl_rpn_delete (rpn->u.p[0]);
211 ccl_rpn_delete (rpn->u.p[1]);
214 free (rpn->u.t.term);
215 for (attr = rpn->u.t.attr_list; attr; attr = attr1)
222 free (rpn->u.setname);
225 ccl_rpn_delete (rpn->u.p[0]);
226 ccl_rpn_delete (rpn->u.p[1]);
232 static struct ccl_rpn_node *find_spec (struct ccl_rpn_attr **qa);
233 static struct ccl_rpn_node *search_terms (struct ccl_rpn_attr **qa);
236 * add_attr: Add attribute (type/value) to RPN term node.
237 * p: RPN node of type term.
238 * type: Type of attribute
239 * value: Value of attribute
241 static void add_attr (struct ccl_rpn_node *p, int type, int value)
243 struct ccl_rpn_attr *n;
245 n = malloc (sizeof(*n));
249 n->next = p->u.t.attr_list;
250 p->u.t.attr_list = n;
254 * search_term: Parse CCL search term.
255 * qa: Qualifier attributes already applied.
256 * return: pointer to node(s); NULL on error.
258 static struct ccl_rpn_node *search_term (struct ccl_rpn_attr **qa)
260 struct ccl_rpn_node *p;
261 struct ccl_token *lookahead = look_token;
267 int relation_value = -1;
268 int position_value = -1;
269 int structure_value = -1;
270 int truncation_value = -1;
271 int completeness_value = -1;
273 if (KIND != CCL_TOK_TERM)
275 ccl_error = CCL_ERR_TERM_EXPECTED;
278 /* create the term node, but wait a moment before adding the term */
279 p = mk_node (CCL_RPN_TERM);
280 p->u.t.attr_list = NULL;
285 /* no qualifier(s) applied. Use 'term' if it is defined */
287 qa = malloc (2*sizeof(*qa));
289 qa[0] = ccl_qual_search (bibset, "term", 4);
293 /* go through all attributes and add them to the attribute list */
294 for (i=0; qa && qa[i]; i++)
296 struct ccl_rpn_attr *attr;
298 for (attr = qa[i]; attr; attr = attr->next)
300 { /* deal only with REAL attributes (positive) */
304 if (relation_value != -1)
306 relation_value = attr->value;
309 if (position_value != -1)
311 position_value = attr->value;
314 if (structure_value != -1)
316 structure_value = attr->value;
319 if (truncation_value != -1)
321 truncation_value = attr->value;
324 if (completeness_value != -1)
326 completeness_value = attr->value;
329 add_attr (p, attr->type, attr->value);
332 /* go through each TERM token. If no truncation attribute is yet
333 met, then look for left/right truncation markers (?) and
334 set left_trunc/right_trunc/mid_trunc accordingly */
335 for (no = 0; lookahead->kind == CCL_TOK_TERM; no++)
337 for (i = 0; i<lookahead->len; i++)
338 if (truncation_value == -1 && lookahead->name[i] == '?')
340 if (no == 0 && i == 0 && lookahead->len >= 1)
342 else if (lookahead->next->kind != CCL_TOK_TERM &&
343 i == lookahead->len-1 && i >= 1)
348 len += 1+lookahead->len;
349 lookahead = lookahead->next;
351 /* len now holds the number of characters in the RPN term */
352 /* no holds the number of CCL tokens (1 or more) */
354 if (structure_value == -1 &&
355 qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_WP))
356 { /* no structure attribute met. Apply either structure attribute
357 WORD or PHRASE depending on number of CCL tokens */
359 add_attr (p, CCL_BIB1_STR, 2);
361 add_attr (p, CCL_BIB1_STR, 1);
364 /* make the RPN token */
365 p->u.t.term = malloc (len);
366 assert (p->u.t.term);
367 p->u.t.term[0] = '\0';
368 for (i = 0; i<no; i++)
370 const char *src_str = look_token->name;
371 int src_len = look_token->len;
373 if (i == 0 && left_trunc)
378 else if (i == no-1 && right_trunc)
381 strcat (p->u.t.term, " ");
382 strxcat (p->u.t.term, src_str, src_len);
385 if (left_trunc && right_trunc)
387 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH))
389 ccl_error = CCL_ERR_TRUNC_NOT_BOTH;
394 add_attr (p, CCL_BIB1_TRU, 3);
396 else if (right_trunc)
398 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT))
400 ccl_error = CCL_ERR_TRUNC_NOT_RIGHT;
405 add_attr (p, CCL_BIB1_TRU, 1);
409 if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT))
411 ccl_error = CCL_ERR_TRUNC_NOT_LEFT;
416 add_attr (p, CCL_BIB1_TRU, 2);
420 if (qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE))
421 add_attr (p, CCL_BIB1_TRU, 100);
427 * qualifiers: Parse CCL qualifiers and search terms.
428 * la: Token pointer to RELATION token.
429 * qa: Qualifier attributes already applied.
430 * return: pointer to node(s); NULL on error.
432 static struct ccl_rpn_node *qualifiers (struct ccl_token *la,
433 struct ccl_rpn_attr **qa)
435 struct ccl_token *lookahead = look_token;
436 struct ccl_rpn_attr **ap;
442 ccl_error = CCL_ERR_DOUBLE_QUAL;
446 for (lookahead = look_token; lookahead != la; lookahead=lookahead->next)
449 for (i=0; qa[i]; i++)
451 ap = malloc ((no+1) * sizeof(*ap));
453 for (i = 0; look_token != la; i++)
455 ap[i] = ccl_qual_search (bibset, look_token->name, look_token->len);
458 ccl_error = CCL_ERR_UNKNOWN_QUAL;
463 if (KIND == CCL_TOK_COMMA)
470 if (!qual_val_type (ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER))
472 /* unordered relation */
473 struct ccl_rpn_node *p;
474 if (KIND != CCL_TOK_EQ)
476 ccl_error = CCL_ERR_EQ_EXPECTED;
481 if (KIND == CCL_TOK_LP)
484 if (!(p = find_spec (ap)))
489 if (KIND != CCL_TOK_RP)
491 ccl_error = CCL_ERR_RP_EXPECTED;
499 p = search_terms (ap);
504 if (look_token->len == 1)
506 if (look_token->name[0] == '<')
508 else if (look_token->name[0] == '=')
510 else if (look_token->name[0] == '>')
513 else if (look_token->len == 2)
515 if (!memcmp (look_token->name, "<=", 2))
517 else if (!memcmp (look_token->name, ">=", 2))
519 else if (!memcmp (look_token->name, "<>", 2))
523 ccl_error = CCL_ERR_BAD_RELATION;
526 struct ccl_rpn_node *p;
528 ADVANCE; /* skip relation */
529 if (KIND == CCL_TOK_TERM && look_token->next->kind == CCL_TOK_MINUS)
531 struct ccl_rpn_node *p1;
532 if (!(p1 = search_term (ap)))
537 ADVANCE; /* skip '-' */
538 if (KIND == CCL_TOK_TERM) /* = term - term ? */
540 struct ccl_rpn_node *p2;
542 if (!(p2 = search_term (ap)))
548 p = mk_node (CCL_RPN_AND);
550 add_attr (p1, CCL_BIB1_REL, 4);
552 add_attr (p2, CCL_BIB1_REL, 2);
558 add_attr (p1, CCL_BIB1_REL, 4);
563 else if (KIND == CCL_TOK_MINUS) /* = - term ? */
566 if (!(p = search_term (ap)))
571 add_attr (p, CCL_BIB1_REL, 2);
575 else if (KIND == CCL_TOK_LP)
578 if (!(p = find_spec (ap)))
583 if (KIND != CCL_TOK_RP)
585 ccl_error = CCL_ERR_RP_EXPECTED;
596 if (!(p = search_terms (ap)))
601 add_attr (p, CCL_BIB1_REL, rel);
605 ccl_error = CCL_ERR_TERM_EXPECTED;
612 * search_terms: Parse CCL search terms - including proximity.
613 * qa: Qualifier attributes already applied.
614 * return: pointer to node(s); NULL on error.
616 static struct ccl_rpn_node *search_terms (struct ccl_rpn_attr **qa)
618 struct ccl_rpn_node *p1, *p2, *pn;
619 p1 = search_term (qa);
624 if (KIND == CCL_TOK_PROX)
627 p2 = search_term (qa);
633 pn = mk_node (CCL_RPN_PROX);
638 else if (KIND == CCL_TOK_TERM)
640 p2 = search_term (qa);
646 pn = mk_node (CCL_RPN_PROX);
658 * search_elements: Parse CCL search elements
659 * qa: Qualifier attributes already applied.
660 * return: pointer to node(s); NULL on error.
662 static struct ccl_rpn_node *search_elements (struct ccl_rpn_attr **qa)
664 struct ccl_rpn_node *p1;
665 struct ccl_token *lookahead;
666 if (KIND == CCL_TOK_LP)
672 if (KIND != CCL_TOK_RP)
674 ccl_error = CCL_ERR_RP_EXPECTED;
681 else if (KIND == CCL_TOK_SET)
684 if (KIND == CCL_TOK_EQ)
686 if (KIND != CCL_TOK_TERM)
688 ccl_error = CCL_ERR_SETNAME_EXPECTED;
691 p1 = mk_node (CCL_RPN_SET);
692 p1->u.setname = copy_token_name (look_token);
696 lookahead = look_token;
698 while (lookahead->kind==CCL_TOK_TERM)
700 lookahead = lookahead->next;
701 if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
702 return qualifiers (lookahead, qa);
703 if (lookahead->kind != CCL_TOK_COMMA)
705 lookahead = lookahead->next;
707 return search_terms (qa);
711 * find_spec: Parse CCL find specification
712 * qa: Qualifier attributes already applied.
713 * return: pointer to node(s); NULL on error.
715 static struct ccl_rpn_node *find_spec (struct ccl_rpn_attr **qa)
717 struct ccl_rpn_node *p1, *p2, *pn;
718 if (!(p1 = search_elements (qa)))
726 p2 = search_elements (qa);
732 pn = mk_node (CCL_RPN_AND);
739 p2 = search_elements (qa);
745 pn = mk_node (CCL_RPN_OR);
752 p2 = search_elements (qa);
758 pn = mk_node (CCL_RPN_NOT);
770 * ccl_find: Parse CCL find - token representation
771 * abibset: Bibset to be used for the parsing
772 * list: List of tokens
773 * error: Pointer to integer. Holds error no. on completion.
774 * pos: Pointer to char position. Holds approximate error position.
775 * return: RPN tree on successful completion; NULL otherwise.
777 struct ccl_rpn_node *ccl_find (CCL_bibset abibset, struct ccl_token *list,
778 int *error, const char **pos)
780 struct ccl_rpn_node *p;
784 p = find_spec (NULL);
785 if (p && KIND != CCL_TOK_EOL)
787 if (KIND == CCL_TOK_RP)
788 ccl_error = CCL_ERR_BAD_RP;
790 ccl_error = CCL_ERR_OP_EXPECTED;
794 *pos = look_token->name;
803 * ccl_find_str: Parse CCL find - string representation
804 * bibset: Bibset to be used for the parsing
805 * str: String to be parsed
806 * error: Pointer to integer. Holds error no. on completion.
807 * pos: Pointer to char position. Holds approximate error position.
808 * return: RPN tree on successful completion; NULL otherwise.
810 struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset, const char *str,
811 int *error, int *pos)
813 struct ccl_token *list;
814 struct ccl_rpn_node *rpn;
815 const char *char_pos;
817 list = ccl_tokenize (str);
818 rpn = ccl_find (bibset, list, error, &char_pos);
820 *pos = char_pos - str;