1 | /*
|
---|
2 | * Copyright (c) 1995, the EUROPAGATE consortium (see below).
|
---|
3 | *
|
---|
4 | * The EUROPAGATE consortium members are:
|
---|
5 | *
|
---|
6 | * University College Dublin
|
---|
7 | * Danmarks Teknologiske Videnscenter
|
---|
8 | * An Chomhairle Leabharlanna
|
---|
9 | * Consejo Superior de Investigaciones Cientificas
|
---|
10 | *
|
---|
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,
|
---|
13 | * provided that:
|
---|
14 | *
|
---|
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.
|
---|
18 | *
|
---|
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.
|
---|
22 | *
|
---|
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.
|
---|
28 | *
|
---|
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.
|
---|
32 | *
|
---|
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.
|
---|
42 | *
|
---|
43 | */
|
---|
44 | /* CCL find (to rpn conversion)
|
---|
45 | * Europagate, 1995
|
---|
46 | *
|
---|
47 | * $Log$
|
---|
48 | * Revision 1.1 2000/08/03 03:09:56 johnmcp
|
---|
49 | * Added the YAZ toolkit source to the packages directory (for z39.50 stuff)
|
---|
50 | *
|
---|
51 | * Revision 1.15 2000/02/24 23:49:13 adam
|
---|
52 | * Fixed memory allocation problem.
|
---|
53 | *
|
---|
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.
|
---|
57 | * ILL-code updates.
|
---|
58 | *
|
---|
59 | * Revision 1.13 1999/12/22 13:13:32 adam
|
---|
60 | * Search terms may include "operators" without causing error.
|
---|
61 | *
|
---|
62 | * Revision 1.12 1999/11/30 13:47:11 adam
|
---|
63 | * Improved installation. Moved header files to include/yaz.
|
---|
64 | *
|
---|
65 | * Revision 1.11 1999/03/31 11:15:37 adam
|
---|
66 | * Fixed memory leaks in ccl_find_str and ccl_qual_rm.
|
---|
67 | *
|
---|
68 | * Revision 1.10 1998/02/11 11:53:33 adam
|
---|
69 | * Changed code so that it compiles as C++.
|
---|
70 | *
|
---|
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
|
---|
74 | * been added.
|
---|
75 | *
|
---|
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
|
---|
78 | * to avoid warnings.
|
---|
79 | *
|
---|
80 | * Revision 1.7 1997/05/14 06:53:26 adam
|
---|
81 | * C++ support.
|
---|
82 | *
|
---|
83 | * Revision 1.6 1997/04/30 08:52:06 quinn
|
---|
84 | * Null
|
---|
85 | *
|
---|
86 | * Revision 1.5 1996/10/11 15:00:24 adam
|
---|
87 | * CCL parser from Europagate Email gateway 1.0.
|
---|
88 | *
|
---|
89 | * Revision 1.16 1996/01/08 08:41:13 adam
|
---|
90 | * Removed unused function.
|
---|
91 | *
|
---|
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.
|
---|
95 | *
|
---|
96 | * Revision 1.14 1995/05/16 09:39:26 adam
|
---|
97 | * LICENSE.
|
---|
98 | *
|
---|
99 | * Revision 1.13 1995/04/17 09:31:42 adam
|
---|
100 | * Improved handling of qualifiers. Aliases or reserved words.
|
---|
101 | *
|
---|
102 | * Revision 1.12 1995/03/20 15:27:43 adam
|
---|
103 | * Minor changes.
|
---|
104 | *
|
---|
105 | * Revision 1.11 1995/02/23 08:31:59 adam
|
---|
106 | * Changed header.
|
---|
107 | *
|
---|
108 | * Revision 1.9 1995/02/16 13:20:06 adam
|
---|
109 | * Spell fix.
|
---|
110 | *
|
---|
111 | * Revision 1.8 1995/02/14 19:59:42 adam
|
---|
112 | * Removed a syntax error.
|
---|
113 | *
|
---|
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.
|
---|
118 | *
|
---|
119 | * Revision 1.6 1995/02/14 16:20:55 adam
|
---|
120 | * Qualifiers are read from a file now.
|
---|
121 | *
|
---|
122 | * Revision 1.5 1995/02/14 14:12:41 adam
|
---|
123 | * Ranges for ordered qualfiers implemented (e.g. pd=1980-1990).
|
---|
124 | *
|
---|
125 | * Revision 1.4 1995/02/14 13:16:29 adam
|
---|
126 | * Left and/or right truncation implemented.
|
---|
127 | *
|
---|
128 | * Revision 1.3 1995/02/14 10:25:56 adam
|
---|
129 | * The constructions 'qualifier rel term ...' implemented.
|
---|
130 | *
|
---|
131 | * Revision 1.2 1995/02/13 15:15:07 adam
|
---|
132 | * Added handling of qualifiers. Not finished yet.
|
---|
133 | *
|
---|
134 | * Revision 1.1 1995/02/13 12:35:20 adam
|
---|
135 | * First version of CCL. Qualifiers aren't handled yet.
|
---|
136 | *
|
---|
137 | */
|
---|
138 |
|
---|
139 | #include <stdio.h>
|
---|
140 | #include <stdlib.h>
|
---|
141 | #include <string.h>
|
---|
142 |
|
---|
143 | #include <yaz/ccl.h>
|
---|
144 |
|
---|
145 | /* returns type of current lookahead */
|
---|
146 | #define KIND (cclp->look_token->kind)
|
---|
147 |
|
---|
148 | /* move one token forward */
|
---|
149 | #define ADVANCE cclp->look_token = cclp->look_token->next
|
---|
150 |
|
---|
151 | /*
|
---|
152 | * qual_val_type: test for existance of attribute type/value pair.
|
---|
153 | * qa: Attribute array
|
---|
154 | * type: Type of attribute to search for
|
---|
155 | * value: Value of attribute to seach for
|
---|
156 | * return: 1 if found; 0 otherwise.
|
---|
157 | */
|
---|
158 | static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value)
|
---|
159 | {
|
---|
160 | int i;
|
---|
161 | struct ccl_rpn_attr *q;
|
---|
162 |
|
---|
163 | if (!qa)
|
---|
164 | return 0;
|
---|
165 | for (i = 0; (q=qa[i]); i++)
|
---|
166 | while (q)
|
---|
167 | {
|
---|
168 | if (q->type == type && q->value == value)
|
---|
169 | return 1;
|
---|
170 | q = q->next;
|
---|
171 | }
|
---|
172 | return 0;
|
---|
173 | }
|
---|
174 |
|
---|
175 | /*
|
---|
176 | * strxcat: concatenate strings.
|
---|
177 | * n: Null-terminated Destination string
|
---|
178 | * src: Source string to be appended (not null-terminated)
|
---|
179 | * len: Length of source string.
|
---|
180 | */
|
---|
181 | static void strxcat (char *n, const char *src, int len)
|
---|
182 | {
|
---|
183 | while (*n)
|
---|
184 | n++;
|
---|
185 | while (--len >= 0)
|
---|
186 | *n++ = *src++;
|
---|
187 | *n = '\0';
|
---|
188 | }
|
---|
189 |
|
---|
190 | /*
|
---|
191 | * copy_token_name: Return copy of CCL token name
|
---|
192 | * tp: Pointer to token info.
|
---|
193 | * return: malloc(3) allocated copy of token name.
|
---|
194 | */
|
---|
195 | static char *copy_token_name (struct ccl_token *tp)
|
---|
196 | {
|
---|
197 | char *str = (char *)malloc (tp->len + 1);
|
---|
198 | ccl_assert (str);
|
---|
199 | memcpy (str, tp->name, tp->len);
|
---|
200 | str[tp->len] = '\0';
|
---|
201 | return str;
|
---|
202 | }
|
---|
203 |
|
---|
204 | /*
|
---|
205 | * mk_node: Create RPN node.
|
---|
206 | * kind: Type of node.
|
---|
207 | * return: pointer to allocated node.
|
---|
208 | */
|
---|
209 | static struct ccl_rpn_node *mk_node (int kind)
|
---|
210 | {
|
---|
211 | struct ccl_rpn_node *p;
|
---|
212 | p = (struct ccl_rpn_node *)malloc (sizeof(*p));
|
---|
213 | ccl_assert (p);
|
---|
214 | p->kind = kind;
|
---|
215 | return p;
|
---|
216 | }
|
---|
217 |
|
---|
218 | /*
|
---|
219 | * ccl_rpn_delete: Delete RPN tree.
|
---|
220 | * rpn: Pointer to tree.
|
---|
221 | */
|
---|
222 | void ccl_rpn_delete (struct ccl_rpn_node *rpn)
|
---|
223 | {
|
---|
224 | struct ccl_rpn_attr *attr, *attr1;
|
---|
225 | if (!rpn)
|
---|
226 | return;
|
---|
227 | switch (rpn->kind)
|
---|
228 | {
|
---|
229 | case CCL_RPN_AND:
|
---|
230 | case CCL_RPN_OR:
|
---|
231 | case CCL_RPN_NOT:
|
---|
232 | ccl_rpn_delete (rpn->u.p[0]);
|
---|
233 | ccl_rpn_delete (rpn->u.p[1]);
|
---|
234 | break;
|
---|
235 | case CCL_RPN_TERM:
|
---|
236 | free (rpn->u.t.term);
|
---|
237 | for (attr = rpn->u.t.attr_list; attr; attr = attr1)
|
---|
238 | {
|
---|
239 | attr1 = attr->next;
|
---|
240 | free (attr);
|
---|
241 | }
|
---|
242 | break;
|
---|
243 | case CCL_RPN_SET:
|
---|
244 | free (rpn->u.setname);
|
---|
245 | break;
|
---|
246 | case CCL_RPN_PROX:
|
---|
247 | ccl_rpn_delete (rpn->u.p[0]);
|
---|
248 | ccl_rpn_delete (rpn->u.p[1]);
|
---|
249 | break;
|
---|
250 | }
|
---|
251 | free (rpn);
|
---|
252 | }
|
---|
253 |
|
---|
254 | static struct ccl_rpn_node *find_spec (CCL_parser cclp,
|
---|
255 | struct ccl_rpn_attr **qa);
|
---|
256 |
|
---|
257 | static int is_term_ok (int look, int *list)
|
---|
258 | {
|
---|
259 | for (;*list >= 0; list++)
|
---|
260 | if (look == *list)
|
---|
261 | return 1;
|
---|
262 | return 0;
|
---|
263 | }
|
---|
264 |
|
---|
265 | static struct ccl_rpn_node *search_terms (CCL_parser cclp,
|
---|
266 | struct ccl_rpn_attr **qa);
|
---|
267 |
|
---|
268 | /*
|
---|
269 | * add_attr: Add attribute (type/value) to RPN term node.
|
---|
270 | * p: RPN node of type term.
|
---|
271 | * type: Type of attribute
|
---|
272 | * value: Value of attribute
|
---|
273 | */
|
---|
274 | static void add_attr (struct ccl_rpn_node *p, int type, int value)
|
---|
275 | {
|
---|
276 | struct ccl_rpn_attr *n;
|
---|
277 |
|
---|
278 | n = (struct ccl_rpn_attr *)malloc (sizeof(*n));
|
---|
279 | ccl_assert (n);
|
---|
280 | n->type = type;
|
---|
281 | n->value = value;
|
---|
282 | n->next = p->u.t.attr_list;
|
---|
283 | p->u.t.attr_list = n;
|
---|
284 | }
|
---|
285 |
|
---|
286 | /*
|
---|
287 | * search_term: Parse CCL search term.
|
---|
288 | * cclp: CCL Parser
|
---|
289 | * qa: Qualifier attributes already applied.
|
---|
290 | * return: pointer to node(s); NULL on error.
|
---|
291 | */
|
---|
292 | static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
|
---|
293 | struct ccl_rpn_attr **qa,
|
---|
294 | int *term_list)
|
---|
295 | {
|
---|
296 | struct ccl_rpn_attr *qa_tmp[2];
|
---|
297 | struct ccl_rpn_node *p;
|
---|
298 | struct ccl_token *lookahead = cclp->look_token;
|
---|
299 | int len = 0;
|
---|
300 | size_t no, i;
|
---|
301 | int left_trunc = 0;
|
---|
302 | int right_trunc = 0;
|
---|
303 | int mid_trunc = 0;
|
---|
304 | int relation_value = -1;
|
---|
305 | int position_value = -1;
|
---|
306 | int structure_value = -1;
|
---|
307 | int truncation_value = -1;
|
---|
308 | int completeness_value = -1;
|
---|
309 |
|
---|
310 | if (!is_term_ok(KIND, term_list))
|
---|
311 | {
|
---|
312 | cclp->error_code = CCL_ERR_TERM_EXPECTED;
|
---|
313 | return NULL;
|
---|
314 | }
|
---|
315 | /* create the term node, but wait a moment before adding the term */
|
---|
316 | p = mk_node (CCL_RPN_TERM);
|
---|
317 | p->u.t.attr_list = NULL;
|
---|
318 | p->u.t.term = NULL;
|
---|
319 |
|
---|
320 | if (!qa)
|
---|
321 | {
|
---|
322 | /* no qualifier(s) applied. Use 'term' if it is defined */
|
---|
323 |
|
---|
324 | qa = qa_tmp;
|
---|
325 | ccl_assert (qa);
|
---|
326 | qa[0] = ccl_qual_search (cclp, "term", 4);
|
---|
327 | qa[1] = NULL;
|
---|
328 | }
|
---|
329 |
|
---|
330 | /* go through all attributes and add them to the attribute list */
|
---|
331 | for (i=0; qa && qa[i]; i++)
|
---|
332 | {
|
---|
333 | struct ccl_rpn_attr *attr;
|
---|
334 |
|
---|
335 | for (attr = qa[i]; attr; attr = attr->next)
|
---|
336 | if (attr->value > 0)
|
---|
337 | { /* deal only with REAL attributes (positive) */
|
---|
338 | switch (attr->type)
|
---|
339 | {
|
---|
340 | case CCL_BIB1_REL:
|
---|
341 | if (relation_value != -1)
|
---|
342 | continue;
|
---|
343 | relation_value = attr->value;
|
---|
344 | break;
|
---|
345 | case CCL_BIB1_POS:
|
---|
346 | if (position_value != -1)
|
---|
347 | continue;
|
---|
348 | position_value = attr->value;
|
---|
349 | break;
|
---|
350 | case CCL_BIB1_STR:
|
---|
351 | if (structure_value != -1)
|
---|
352 | continue;
|
---|
353 | structure_value = attr->value;
|
---|
354 | break;
|
---|
355 | case CCL_BIB1_TRU:
|
---|
356 | if (truncation_value != -1)
|
---|
357 | continue;
|
---|
358 | truncation_value = attr->value;
|
---|
359 | break;
|
---|
360 | case CCL_BIB1_COM:
|
---|
361 | if (completeness_value != -1)
|
---|
362 | continue;
|
---|
363 | completeness_value = attr->value;
|
---|
364 | break;
|
---|
365 | }
|
---|
366 | add_attr (p, attr->type, attr->value);
|
---|
367 | }
|
---|
368 | }
|
---|
369 | /* go through each TERM token. If no truncation attribute is yet
|
---|
370 | met, then look for left/right truncation markers (?) and
|
---|
371 | set left_trunc/right_trunc/mid_trunc accordingly */
|
---|
372 | for (no = 0; is_term_ok(lookahead->kind, term_list); no++)
|
---|
373 | {
|
---|
374 | for (i = 0; i<lookahead->len; i++)
|
---|
375 | if (truncation_value == -1 && lookahead->name[i] == '?')
|
---|
376 | {
|
---|
377 | if (no == 0 && i == 0 && lookahead->len >= 1)
|
---|
378 | left_trunc = 1;
|
---|
379 | else if (!is_term_ok(lookahead->next->kind, term_list) &&
|
---|
380 | i == lookahead->len-1 && i >= 1)
|
---|
381 | right_trunc = 1;
|
---|
382 | else
|
---|
383 | mid_trunc = 1;
|
---|
384 | }
|
---|
385 | len += 1+lookahead->len;
|
---|
386 | lookahead = lookahead->next;
|
---|
387 | }
|
---|
388 | /* len now holds the number of characters in the RPN term */
|
---|
389 | /* no holds the number of CCL tokens (1 or more) */
|
---|
390 |
|
---|
391 | if (structure_value == -1 &&
|
---|
392 | qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_WP))
|
---|
393 | { /* no structure attribute met. Apply either structure attribute
|
---|
394 | WORD or PHRASE depending on number of CCL tokens */
|
---|
395 | if (no == 1)
|
---|
396 | add_attr (p, CCL_BIB1_STR, 2);
|
---|
397 | else
|
---|
398 | add_attr (p, CCL_BIB1_STR, 1);
|
---|
399 | }
|
---|
400 |
|
---|
401 | /* make the RPN token */
|
---|
402 | p->u.t.term = (char *)malloc (len);
|
---|
403 | ccl_assert (p->u.t.term);
|
---|
404 | p->u.t.term[0] = '\0';
|
---|
405 | for (i = 0; i<no; i++)
|
---|
406 | {
|
---|
407 | const char *src_str = cclp->look_token->name;
|
---|
408 | int src_len = cclp->look_token->len;
|
---|
409 |
|
---|
410 | if (i == 0 && left_trunc)
|
---|
411 | {
|
---|
412 | src_len--;
|
---|
413 | src_str++;
|
---|
414 | }
|
---|
415 | else if (i == no-1 && right_trunc)
|
---|
416 | src_len--;
|
---|
417 | if (src_len)
|
---|
418 | {
|
---|
419 | int len = strlen(p->u.t.term);
|
---|
420 | if (len &&
|
---|
421 | !strchr("-+", *src_str) &&
|
---|
422 | !strchr("-+", p->u.t.term[len-1]))
|
---|
423 | {
|
---|
424 | strcat (p->u.t.term, " ");
|
---|
425 | }
|
---|
426 | }
|
---|
427 | strxcat (p->u.t.term, src_str, src_len);
|
---|
428 | ADVANCE;
|
---|
429 | }
|
---|
430 | if (left_trunc && right_trunc)
|
---|
431 | {
|
---|
432 | if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH))
|
---|
433 | {
|
---|
434 | cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
|
---|
435 | ccl_rpn_delete (p);
|
---|
436 | return NULL;
|
---|
437 | }
|
---|
438 | add_attr (p, CCL_BIB1_TRU, 3);
|
---|
439 | }
|
---|
440 | else if (right_trunc)
|
---|
441 | {
|
---|
442 | if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT))
|
---|
443 | {
|
---|
444 | cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
|
---|
445 | ccl_rpn_delete (p);
|
---|
446 | return NULL;
|
---|
447 | }
|
---|
448 | add_attr (p, CCL_BIB1_TRU, 1);
|
---|
449 | }
|
---|
450 | else if (left_trunc)
|
---|
451 | {
|
---|
452 | if (!qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT))
|
---|
453 | {
|
---|
454 | cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
|
---|
455 | ccl_rpn_delete (p);
|
---|
456 | return NULL;
|
---|
457 | }
|
---|
458 | add_attr (p, CCL_BIB1_TRU, 2);
|
---|
459 | }
|
---|
460 | else
|
---|
461 | {
|
---|
462 | if (qual_val_type (qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE))
|
---|
463 | add_attr (p, CCL_BIB1_TRU, 100);
|
---|
464 | }
|
---|
465 | return p;
|
---|
466 | }
|
---|
467 |
|
---|
468 | static struct ccl_rpn_node *search_term (CCL_parser cclp,
|
---|
469 | struct ccl_rpn_attr **qa)
|
---|
470 | {
|
---|
471 | static int list[] = {CCL_TOK_TERM, CCL_TOK_COMMA, -1};
|
---|
472 | return search_term_x(cclp, qa, list);
|
---|
473 | }
|
---|
474 |
|
---|
475 | /*
|
---|
476 | * qualifiers: Parse CCL qualifiers and search terms.
|
---|
477 | * cclp: CCL Parser
|
---|
478 | * la: Token pointer to RELATION token.
|
---|
479 | * qa: Qualifier attributes already applied.
|
---|
480 | * return: pointer to node(s); NULL on error.
|
---|
481 | */
|
---|
482 | static struct ccl_rpn_node *qualifiers (CCL_parser cclp, struct ccl_token *la,
|
---|
483 | struct ccl_rpn_attr **qa)
|
---|
484 | {
|
---|
485 | struct ccl_token *lookahead = cclp->look_token;
|
---|
486 | struct ccl_rpn_attr **ap;
|
---|
487 | int no = 0;
|
---|
488 | int i, rel;
|
---|
489 | #if 0
|
---|
490 | if (qa)
|
---|
491 | {
|
---|
492 | cclp->error_code = CCL_ERR_DOUBLE_QUAL;
|
---|
493 | return NULL;
|
---|
494 | }
|
---|
495 | #endif
|
---|
496 | for (lookahead = cclp->look_token; lookahead != la;
|
---|
497 | lookahead=lookahead->next)
|
---|
498 | no++;
|
---|
499 | if (qa)
|
---|
500 | for (i=0; qa[i]; i++)
|
---|
501 | no++;
|
---|
502 | ap = (struct ccl_rpn_attr **)malloc ((no+1) * sizeof(*ap));
|
---|
503 | ccl_assert (ap);
|
---|
504 | for (i = 0; cclp->look_token != la; i++)
|
---|
505 | {
|
---|
506 | ap[i] = ccl_qual_search (cclp, cclp->look_token->name,
|
---|
507 | cclp->look_token->len);
|
---|
508 | if (!ap[i])
|
---|
509 | {
|
---|
510 | cclp->error_code = CCL_ERR_UNKNOWN_QUAL;
|
---|
511 | free (ap);
|
---|
512 | return NULL;
|
---|
513 | }
|
---|
514 | ADVANCE;
|
---|
515 | if (KIND == CCL_TOK_COMMA)
|
---|
516 | ADVANCE;
|
---|
517 | }
|
---|
518 | if (qa)
|
---|
519 | while (*qa)
|
---|
520 | ap[i++] = *qa++;
|
---|
521 | ap[i] = NULL;
|
---|
522 | if (!qual_val_type (ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER))
|
---|
523 | {
|
---|
524 | /* unordered relation */
|
---|
525 | struct ccl_rpn_node *p;
|
---|
526 | if (KIND != CCL_TOK_EQ)
|
---|
527 | {
|
---|
528 | cclp->error_code = CCL_ERR_EQ_EXPECTED;
|
---|
529 | free (ap);
|
---|
530 | return NULL;
|
---|
531 | }
|
---|
532 | ADVANCE;
|
---|
533 | if (KIND == CCL_TOK_LP)
|
---|
534 | {
|
---|
535 | ADVANCE;
|
---|
536 | if (!(p = find_spec (cclp, ap)))
|
---|
537 | {
|
---|
538 | free (ap);
|
---|
539 | return NULL;
|
---|
540 | }
|
---|
541 | if (KIND != CCL_TOK_RP)
|
---|
542 | {
|
---|
543 | cclp->error_code = CCL_ERR_RP_EXPECTED;
|
---|
544 | ccl_rpn_delete (p);
|
---|
545 | free (ap);
|
---|
546 | return NULL;
|
---|
547 | }
|
---|
548 | ADVANCE;
|
---|
549 | }
|
---|
550 | else
|
---|
551 | p = search_terms (cclp, ap);
|
---|
552 | free (ap);
|
---|
553 | return p;
|
---|
554 | }
|
---|
555 | /* ordered relation ... */
|
---|
556 | rel = 0;
|
---|
557 | if (cclp->look_token->len == 1)
|
---|
558 | {
|
---|
559 | if (cclp->look_token->name[0] == '<')
|
---|
560 | rel = 1;
|
---|
561 | else if (cclp->look_token->name[0] == '=')
|
---|
562 | rel = 3;
|
---|
563 | else if (cclp->look_token->name[0] == '>')
|
---|
564 | rel = 5;
|
---|
565 | }
|
---|
566 | else if (cclp->look_token->len == 2)
|
---|
567 | {
|
---|
568 | if (!memcmp (cclp->look_token->name, "<=", 2))
|
---|
569 | rel = 2;
|
---|
570 | else if (!memcmp (cclp->look_token->name, ">=", 2))
|
---|
571 | rel = 4;
|
---|
572 | else if (!memcmp (cclp->look_token->name, "<>", 2))
|
---|
573 | rel = 6;
|
---|
574 | }
|
---|
575 | if (!rel)
|
---|
576 | cclp->error_code = CCL_ERR_BAD_RELATION;
|
---|
577 | else
|
---|
578 | {
|
---|
579 | struct ccl_rpn_node *p;
|
---|
580 |
|
---|
581 | ADVANCE; /* skip relation */
|
---|
582 | if (KIND == CCL_TOK_TERM &&
|
---|
583 | cclp->look_token->next->kind == CCL_TOK_MINUS)
|
---|
584 | {
|
---|
585 | struct ccl_rpn_node *p1;
|
---|
586 | if (!(p1 = search_term (cclp, ap)))
|
---|
587 | {
|
---|
588 | free (ap);
|
---|
589 | return NULL;
|
---|
590 | }
|
---|
591 | ADVANCE; /* skip '-' */
|
---|
592 | if (KIND == CCL_TOK_TERM) /* = term - term ? */
|
---|
593 | {
|
---|
594 | struct ccl_rpn_node *p2;
|
---|
595 |
|
---|
596 | if (!(p2 = search_term (cclp, ap)))
|
---|
597 | {
|
---|
598 | ccl_rpn_delete (p1);
|
---|
599 | free (ap);
|
---|
600 | return NULL;
|
---|
601 | }
|
---|
602 | p = mk_node (CCL_RPN_AND);
|
---|
603 | p->u.p[0] = p1;
|
---|
604 | add_attr (p1, CCL_BIB1_REL, 4);
|
---|
605 | p->u.p[1] = p2;
|
---|
606 | add_attr (p2, CCL_BIB1_REL, 2);
|
---|
607 | free (ap);
|
---|
608 | return p;
|
---|
609 | }
|
---|
610 | else /* = term - */
|
---|
611 | {
|
---|
612 | add_attr (p1, CCL_BIB1_REL, 4);
|
---|
613 | free (ap);
|
---|
614 | return p1;
|
---|
615 | }
|
---|
616 | }
|
---|
617 | else if (KIND == CCL_TOK_MINUS) /* = - term ? */
|
---|
618 | {
|
---|
619 | ADVANCE;
|
---|
620 | if (!(p = search_term (cclp, ap)))
|
---|
621 | {
|
---|
622 | free (ap);
|
---|
623 | return NULL;
|
---|
624 | }
|
---|
625 | add_attr (p, CCL_BIB1_REL, 2);
|
---|
626 | free (ap);
|
---|
627 | return p;
|
---|
628 | }
|
---|
629 | else if (KIND == CCL_TOK_LP)
|
---|
630 | {
|
---|
631 | ADVANCE;
|
---|
632 | if (!(p = find_spec (cclp, ap)))
|
---|
633 | {
|
---|
634 | free (ap);
|
---|
635 | return NULL;
|
---|
636 | }
|
---|
637 | if (KIND != CCL_TOK_RP)
|
---|
638 | {
|
---|
639 | cclp->error_code = CCL_ERR_RP_EXPECTED;
|
---|
640 | ccl_rpn_delete (p);
|
---|
641 | free (ap);
|
---|
642 | return NULL;
|
---|
643 | }
|
---|
644 | ADVANCE;
|
---|
645 | free (ap);
|
---|
646 | return p;
|
---|
647 | }
|
---|
648 | else
|
---|
649 | {
|
---|
650 | if (!(p = search_terms (cclp, ap)))
|
---|
651 | {
|
---|
652 | free (ap);
|
---|
653 | return NULL;
|
---|
654 | }
|
---|
655 | add_attr (p, CCL_BIB1_REL, rel);
|
---|
656 | free (ap);
|
---|
657 | return p;
|
---|
658 | }
|
---|
659 | cclp->error_code = CCL_ERR_TERM_EXPECTED;
|
---|
660 | }
|
---|
661 | free (ap);
|
---|
662 | return NULL;
|
---|
663 | }
|
---|
664 |
|
---|
665 | /*
|
---|
666 | * search_terms: Parse CCL search terms - including proximity.
|
---|
667 | * cclp: CCL Parser
|
---|
668 | * qa: Qualifier attributes already applied.
|
---|
669 | * return: pointer to node(s); NULL on error.
|
---|
670 | */
|
---|
671 | static struct ccl_rpn_node *search_terms (CCL_parser cclp,
|
---|
672 | struct ccl_rpn_attr **qa)
|
---|
673 | {
|
---|
674 | static int list[] = {
|
---|
675 | CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
|
---|
676 | CCL_TOK_REL, CCL_TOK_MINUS, -1};
|
---|
677 | struct ccl_rpn_node *p1, *p2, *pn;
|
---|
678 | p1 = search_term_x (cclp, qa, list);
|
---|
679 | if (!p1)
|
---|
680 | return NULL;
|
---|
681 | while (1)
|
---|
682 | {
|
---|
683 | if (KIND == CCL_TOK_PROX)
|
---|
684 | {
|
---|
685 | ADVANCE;
|
---|
686 | p2 = search_term_x (cclp, qa, list);
|
---|
687 | if (!p2)
|
---|
688 | {
|
---|
689 | ccl_rpn_delete (p1);
|
---|
690 | return NULL;
|
---|
691 | }
|
---|
692 | pn = mk_node (CCL_RPN_PROX);
|
---|
693 | pn->u.p[0] = p1;
|
---|
694 | pn->u.p[1] = p2;
|
---|
695 | p1 = pn;
|
---|
696 | }
|
---|
697 | else if (is_term_ok(KIND, list))
|
---|
698 | {
|
---|
699 | p2 = search_term_x (cclp, qa, list);
|
---|
700 | if (!p2)
|
---|
701 | {
|
---|
702 | ccl_rpn_delete (p1);
|
---|
703 | return NULL;
|
---|
704 | }
|
---|
705 | pn = mk_node (CCL_RPN_PROX);
|
---|
706 | pn->u.p[0] = p1;
|
---|
707 | pn->u.p[1] = p2;
|
---|
708 | p1 = pn;
|
---|
709 | }
|
---|
710 | else
|
---|
711 | break;
|
---|
712 | }
|
---|
713 | return p1;
|
---|
714 | }
|
---|
715 |
|
---|
716 | /*
|
---|
717 | * search_elements: Parse CCL search elements
|
---|
718 | * cclp: CCL Parser
|
---|
719 | * qa: Qualifier attributes already applied.
|
---|
720 | * return: pointer to node(s); NULL on error.
|
---|
721 | */
|
---|
722 | static struct ccl_rpn_node *search_elements (CCL_parser cclp,
|
---|
723 | struct ccl_rpn_attr **qa)
|
---|
724 | {
|
---|
725 | struct ccl_rpn_node *p1;
|
---|
726 | struct ccl_token *lookahead;
|
---|
727 | if (KIND == CCL_TOK_LP)
|
---|
728 | {
|
---|
729 | ADVANCE;
|
---|
730 | p1 = find_spec (cclp, qa);
|
---|
731 | if (!p1)
|
---|
732 | return NULL;
|
---|
733 | if (KIND != CCL_TOK_RP)
|
---|
734 | {
|
---|
735 | cclp->error_code = CCL_ERR_RP_EXPECTED;
|
---|
736 | ccl_rpn_delete (p1);
|
---|
737 | return NULL;
|
---|
738 | }
|
---|
739 | ADVANCE;
|
---|
740 | return p1;
|
---|
741 | }
|
---|
742 | else if (KIND == CCL_TOK_SET)
|
---|
743 | {
|
---|
744 | ADVANCE;
|
---|
745 | if (KIND == CCL_TOK_EQ)
|
---|
746 | ADVANCE;
|
---|
747 | if (KIND != CCL_TOK_TERM)
|
---|
748 | {
|
---|
749 | cclp->error_code = CCL_ERR_SETNAME_EXPECTED;
|
---|
750 | return NULL;
|
---|
751 | }
|
---|
752 | p1 = mk_node (CCL_RPN_SET);
|
---|
753 | p1->u.setname = copy_token_name (cclp->look_token);
|
---|
754 | ADVANCE;
|
---|
755 | return p1;
|
---|
756 | }
|
---|
757 | lookahead = cclp->look_token;
|
---|
758 |
|
---|
759 | while (lookahead->kind==CCL_TOK_TERM)
|
---|
760 | {
|
---|
761 | lookahead = lookahead->next;
|
---|
762 | if (lookahead->kind == CCL_TOK_REL || lookahead->kind == CCL_TOK_EQ)
|
---|
763 | return qualifiers (cclp, lookahead, qa);
|
---|
764 | if (lookahead->kind != CCL_TOK_COMMA)
|
---|
765 | break;
|
---|
766 | lookahead = lookahead->next;
|
---|
767 | }
|
---|
768 | return search_terms (cclp, qa);
|
---|
769 | }
|
---|
770 |
|
---|
771 | /*
|
---|
772 | * find_spec: Parse CCL find specification
|
---|
773 | * cclp: CCL Parser
|
---|
774 | * qa: Qualifier attributes already applied.
|
---|
775 | * return: pointer to node(s); NULL on error.
|
---|
776 | */
|
---|
777 | static struct ccl_rpn_node *find_spec (CCL_parser cclp,
|
---|
778 | struct ccl_rpn_attr **qa)
|
---|
779 | {
|
---|
780 | struct ccl_rpn_node *p1, *p2, *pn;
|
---|
781 | if (!(p1 = search_elements (cclp, qa)))
|
---|
782 | return NULL;
|
---|
783 | while (1)
|
---|
784 | {
|
---|
785 | switch (KIND)
|
---|
786 | {
|
---|
787 | case CCL_TOK_AND:
|
---|
788 | ADVANCE;
|
---|
789 | p2 = search_elements (cclp, qa);
|
---|
790 | if (!p2)
|
---|
791 | {
|
---|
792 | ccl_rpn_delete (p1);
|
---|
793 | return NULL;
|
---|
794 | }
|
---|
795 | pn = mk_node (CCL_RPN_AND);
|
---|
796 | pn->u.p[0] = p1;
|
---|
797 | pn->u.p[1] = p2;
|
---|
798 | p1 = pn;
|
---|
799 | continue;
|
---|
800 | case CCL_TOK_OR:
|
---|
801 | ADVANCE;
|
---|
802 | p2 = search_elements (cclp, qa);
|
---|
803 | if (!p2)
|
---|
804 | {
|
---|
805 | ccl_rpn_delete (p1);
|
---|
806 | return NULL;
|
---|
807 | }
|
---|
808 | pn = mk_node (CCL_RPN_OR);
|
---|
809 | pn->u.p[0] = p1;
|
---|
810 | pn->u.p[1] = p2;
|
---|
811 | p1 = pn;
|
---|
812 | continue;
|
---|
813 | case CCL_TOK_NOT:
|
---|
814 | ADVANCE;
|
---|
815 | p2 = search_elements (cclp, qa);
|
---|
816 | if (!p2)
|
---|
817 | {
|
---|
818 | ccl_rpn_delete (p1);
|
---|
819 | return NULL;
|
---|
820 | }
|
---|
821 | pn = mk_node (CCL_RPN_NOT);
|
---|
822 | pn->u.p[0] = p1;
|
---|
823 | pn->u.p[1] = p2;
|
---|
824 | p1 = pn;
|
---|
825 | continue;
|
---|
826 | }
|
---|
827 | break;
|
---|
828 | }
|
---|
829 | return p1;
|
---|
830 | }
|
---|
831 |
|
---|
832 | struct ccl_rpn_node *ccl_parser_find (CCL_parser cclp, struct ccl_token *list)
|
---|
833 | {
|
---|
834 | struct ccl_rpn_node *p;
|
---|
835 |
|
---|
836 | cclp->look_token = list;
|
---|
837 | p = find_spec (cclp, NULL);
|
---|
838 | if (p && KIND != CCL_TOK_EOL)
|
---|
839 | {
|
---|
840 | if (KIND == CCL_TOK_RP)
|
---|
841 | cclp->error_code = CCL_ERR_BAD_RP;
|
---|
842 | else
|
---|
843 | cclp->error_code = CCL_ERR_OP_EXPECTED;
|
---|
844 | ccl_rpn_delete (p);
|
---|
845 | p = NULL;
|
---|
846 | }
|
---|
847 | cclp->error_pos = cclp->look_token->name;
|
---|
848 | if (p)
|
---|
849 | cclp->error_code = CCL_ERR_OK;
|
---|
850 | else
|
---|
851 | cclp->error_code = cclp->error_code;
|
---|
852 | return p;
|
---|
853 | }
|
---|
854 |
|
---|
855 | /*
|
---|
856 | * ccl_find: Parse CCL find - token representation
|
---|
857 | * bibset: Bibset to be used for the parsing
|
---|
858 | * list: List of tokens
|
---|
859 | * error: Pointer to integer. Holds error no. on completion.
|
---|
860 | * pos: Pointer to char position. Holds approximate error position.
|
---|
861 | * return: RPN tree on successful completion; NULL otherwise.
|
---|
862 | */
|
---|
863 | struct ccl_rpn_node *ccl_find (CCL_bibset bibset, struct ccl_token *list,
|
---|
864 | int *error, const char **pos)
|
---|
865 | {
|
---|
866 | struct ccl_rpn_node *p;
|
---|
867 | CCL_parser cclp = ccl_parser_create ();
|
---|
868 |
|
---|
869 | cclp->bibset = bibset;
|
---|
870 |
|
---|
871 | p = ccl_parser_find (cclp, list);
|
---|
872 |
|
---|
873 | *error = cclp->error_code;
|
---|
874 | *pos = cclp->error_pos;
|
---|
875 |
|
---|
876 | ccl_parser_destroy (cclp);
|
---|
877 |
|
---|
878 | return p;
|
---|
879 | }
|
---|
880 |
|
---|
881 | /*
|
---|
882 | * ccl_find_str: Parse CCL find - string representation
|
---|
883 | * bibset: Bibset to be used for the parsing
|
---|
884 | * str: String to be parsed
|
---|
885 | * error: Pointer to integer. Holds error no. on completion.
|
---|
886 | * pos: Pointer to char position. Holds approximate error position.
|
---|
887 | * return: RPN tree on successful completion; NULL otherwise.
|
---|
888 | */
|
---|
889 | struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset, const char *str,
|
---|
890 | int *error, int *pos)
|
---|
891 | {
|
---|
892 | CCL_parser cclp = ccl_parser_create ();
|
---|
893 | struct ccl_token *list;
|
---|
894 | struct ccl_rpn_node *p;
|
---|
895 |
|
---|
896 | cclp->bibset = bibset;
|
---|
897 |
|
---|
898 | list = ccl_parser_tokenize (cclp, str);
|
---|
899 | p = ccl_parser_find (cclp, list);
|
---|
900 |
|
---|
901 | *error = cclp->error_code;
|
---|
902 | if (*error)
|
---|
903 | *pos = cclp->error_pos - str;
|
---|
904 | ccl_parser_destroy (cclp);
|
---|
905 | ccl_token_del (list);
|
---|
906 | return p;
|
---|
907 | }
|
---|