]>
Commit | Line | Data |
---|---|---|
05f3dbba FK |
1 | /* kwset.c - search for any of a set of keywords. |
2 | Copyright 1989, 1998, 2000, 2005 Free Software Foundation, Inc. | |
3 | ||
4 | This program is free software; you can redistribute it and/or modify | |
5 | it under the terms of the GNU General Public License as published by | |
6 | the Free Software Foundation; either version 2, or (at your option) | |
7 | any later version. | |
8 | ||
9 | This program is distributed in the hope that it will be useful, | |
10 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
11 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
12 | GNU General Public License for more details. | |
13 | ||
14 | You should have received a copy of the GNU General Public License | |
15 | along with this program; if not, write to the Free Software | |
16 | Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston, MA | |
17 | 02110-1301, USA. */ | |
18 | ||
19 | /* Written August 1989 by Mike Haertel. | |
20 | The author may be reached (Email) at the address mike@ai.mit.edu, | |
21 | or (US mail) as Mike Haertel c/o Free Software Foundation. */ | |
22 | ||
23 | /* The algorithm implemented by these routines bears a startling resemblence | |
24 | to one discovered by Beate Commentz-Walter, although it is not identical. | |
25 | See "A String Matching Algorithm Fast on the Average," Technical Report, | |
26 | IBM-Germany, Scientific Center Heidelberg, Tiergartenstrasse 15, D-6900 | |
27 | Heidelberg, Germany. See also Aho, A.V., and M. Corasick, "Efficient | |
28 | String Matching: An Aid to Bibliographic Search," CACM June 1975, | |
29 | Vol. 18, No. 6, which describes the failure function used below. */ | |
30 | ||
31 | #ifdef HAVE_CONFIG_H | |
32 | # include <config.h> | |
33 | #endif | |
34 | #include <sys/types.h> | |
35 | #include "system.h" | |
36 | #include "kwset.h" | |
37 | #include "obstack.h" | |
38 | ||
39 | #ifdef GREP | |
40 | extern char *xmalloc(); | |
41 | # undef malloc | |
42 | # define malloc xmalloc | |
43 | #endif | |
44 | ||
45 | #define NCHAR (UCHAR_MAX + 1) | |
46 | #define obstack_chunk_alloc malloc | |
47 | #define obstack_chunk_free free | |
48 | ||
49 | #define U(c) ((unsigned char) (c)) | |
50 | ||
51 | /* Balanced tree of edges and labels leaving a given trie node. */ | |
52 | struct tree | |
53 | { | |
54 | struct tree *llink; /* Left link; MUST be first field. */ | |
55 | struct tree *rlink; /* Right link (to larger labels). */ | |
56 | struct trie *trie; /* Trie node pointed to by this edge. */ | |
57 | unsigned char label; /* Label on this edge. */ | |
58 | char balance; /* Difference in depths of subtrees. */ | |
59 | }; | |
60 | ||
61 | /* Node of a trie representing a set of reversed keywords. */ | |
62 | struct trie | |
63 | { | |
64 | unsigned int accepting; /* Word index of accepted word, or zero. */ | |
65 | struct tree *links; /* Tree of edges leaving this node. */ | |
66 | struct trie *parent; /* Parent of this node. */ | |
67 | struct trie *next; /* List of all trie nodes in level order. */ | |
68 | struct trie *fail; /* Aho-Corasick failure function. */ | |
69 | int depth; /* Depth of this node from the root. */ | |
70 | int shift; /* Shift function for search failures. */ | |
71 | int maxshift; /* Max shift of self and descendents. */ | |
72 | }; | |
73 | ||
74 | /* Structure returned opaquely to the caller, containing everything. */ | |
75 | struct kwset | |
76 | { | |
77 | struct obstack obstack; /* Obstack for node allocation. */ | |
78 | int words; /* Number of words in the trie. */ | |
79 | struct trie *trie; /* The trie itself. */ | |
80 | int mind; /* Minimum depth of an accepting node. */ | |
81 | int maxd; /* Maximum depth of any node. */ | |
82 | unsigned char delta[NCHAR]; /* Delta table for rapid search. */ | |
83 | struct trie *next[NCHAR]; /* Table of children of the root. */ | |
84 | char *target; /* Target string if there's only one. */ | |
85 | int mind2; /* Used in Boyer-Moore search for one string. */ | |
86 | char const *trans; /* Character translation table. */ | |
87 | }; | |
88 | ||
89 | /* Allocate and initialize a keyword set object, returning an opaque | |
90 | pointer to it. Return NULL if memory is not available. */ | |
91 | kwset_t | |
92 | kwsalloc (char const *trans) | |
93 | { | |
94 | struct kwset *kwset; | |
95 | ||
96 | kwset = (struct kwset *) malloc(sizeof (struct kwset)); | |
97 | if (!kwset) | |
98 | return NULL; | |
99 | ||
100 | obstack_init(&kwset->obstack); | |
101 | kwset->words = 0; | |
102 | kwset->trie | |
103 | = (struct trie *) obstack_alloc(&kwset->obstack, sizeof (struct trie)); | |
104 | if (!kwset->trie) | |
105 | { | |
106 | kwsfree((kwset_t) kwset); | |
107 | return NULL; | |
108 | } | |
109 | kwset->trie->accepting = 0; | |
110 | kwset->trie->links = NULL; | |
111 | kwset->trie->parent = NULL; | |
112 | kwset->trie->next = NULL; | |
113 | kwset->trie->fail = NULL; | |
114 | kwset->trie->depth = 0; | |
115 | kwset->trie->shift = 0; | |
116 | kwset->mind = INT_MAX; | |
117 | kwset->maxd = -1; | |
118 | kwset->target = NULL; | |
119 | kwset->trans = trans; | |
120 | ||
121 | return (kwset_t) kwset; | |
122 | } | |
123 | ||
124 | /* This upper bound is valid for CHAR_BIT >= 4 and | |
125 | exact for CHAR_BIT in { 4..11, 13, 15, 17, 19 }. */ | |
126 | #define DEPTH_SIZE (CHAR_BIT + CHAR_BIT/2) | |
127 | ||
128 | /* Add the given string to the contents of the keyword set. Return NULL | |
129 | for success, an error message otherwise. */ | |
130 | const char * | |
131 | kwsincr (kwset_t kws, char const *text, size_t len) | |
132 | { | |
133 | struct kwset *kwset; | |
134 | register struct trie *trie; | |
135 | register unsigned char label; | |
136 | register struct tree *link; | |
137 | register int depth; | |
138 | struct tree *links[DEPTH_SIZE]; | |
139 | enum { L, R } dirs[DEPTH_SIZE]; | |
140 | struct tree *t, *r, *l, *rl, *lr; | |
141 | ||
142 | kwset = (struct kwset *) kws; | |
143 | trie = kwset->trie; | |
144 | text += len; | |
145 | ||
146 | /* Descend the trie (built of reversed keywords) character-by-character, | |
147 | installing new nodes when necessary. */ | |
148 | while (len--) | |
149 | { | |
150 | label = kwset->trans ? kwset->trans[U(*--text)] : *--text; | |
151 | ||
152 | /* Descend the tree of outgoing links for this trie node, | |
153 | looking for the current character and keeping track | |
154 | of the path followed. */ | |
155 | link = trie->links; | |
156 | links[0] = (struct tree *) &trie->links; | |
157 | dirs[0] = L; | |
158 | depth = 1; | |
159 | ||
160 | while (link && label != link->label) | |
161 | { | |
162 | links[depth] = link; | |
163 | if (label < link->label) | |
164 | dirs[depth++] = L, link = link->llink; | |
165 | else | |
166 | dirs[depth++] = R, link = link->rlink; | |
167 | } | |
168 | ||
169 | /* The current character doesn't have an outgoing link at | |
170 | this trie node, so build a new trie node and install | |
171 | a link in the current trie node's tree. */ | |
172 | if (!link) | |
173 | { | |
174 | link = (struct tree *) obstack_alloc(&kwset->obstack, | |
175 | sizeof (struct tree)); | |
176 | if (!link) | |
177 | return _("memory exhausted"); | |
178 | link->llink = NULL; | |
179 | link->rlink = NULL; | |
180 | link->trie = (struct trie *) obstack_alloc(&kwset->obstack, | |
181 | sizeof (struct trie)); | |
182 | if (!link->trie) | |
183 | { | |
184 | obstack_free(&kwset->obstack, link); | |
185 | return _("memory exhausted"); | |
186 | } | |
187 | link->trie->accepting = 0; | |
188 | link->trie->links = NULL; | |
189 | link->trie->parent = trie; | |
190 | link->trie->next = NULL; | |
191 | link->trie->fail = NULL; | |
192 | link->trie->depth = trie->depth + 1; | |
193 | link->trie->shift = 0; | |
194 | link->label = label; | |
195 | link->balance = 0; | |
196 | ||
197 | /* Install the new tree node in its parent. */ | |
198 | if (dirs[--depth] == L) | |
199 | links[depth]->llink = link; | |
200 | else | |
201 | links[depth]->rlink = link; | |
202 | ||
203 | /* Back up the tree fixing the balance flags. */ | |
204 | while (depth && !links[depth]->balance) | |
205 | { | |
206 | if (dirs[depth] == L) | |
207 | --links[depth]->balance; | |
208 | else | |
209 | ++links[depth]->balance; | |
210 | --depth; | |
211 | } | |
212 | ||
213 | /* Rebalance the tree by pointer rotations if necessary. */ | |
214 | if (depth && ((dirs[depth] == L && --links[depth]->balance) | |
215 | || (dirs[depth] == R && ++links[depth]->balance))) | |
216 | { | |
217 | switch (links[depth]->balance) | |
218 | { | |
219 | case (char) -2: | |
220 | switch (dirs[depth + 1]) | |
221 | { | |
222 | case L: | |
223 | r = links[depth], t = r->llink, rl = t->rlink; | |
224 | t->rlink = r, r->llink = rl; | |
225 | t->balance = r->balance = 0; | |
226 | break; | |
227 | case R: | |
228 | r = links[depth], l = r->llink, t = l->rlink; | |
229 | rl = t->rlink, lr = t->llink; | |
230 | t->llink = l, l->rlink = lr, t->rlink = r, r->llink = rl; | |
231 | l->balance = t->balance != 1 ? 0 : -1; | |
232 | r->balance = t->balance != (char) -1 ? 0 : 1; | |
233 | t->balance = 0; | |
234 | break; | |
235 | default: | |
236 | abort (); | |
237 | } | |
238 | break; | |
239 | case 2: | |
240 | switch (dirs[depth + 1]) | |
241 | { | |
242 | case R: | |
243 | l = links[depth], t = l->rlink, lr = t->llink; | |
244 | t->llink = l, l->rlink = lr; | |
245 | t->balance = l->balance = 0; | |
246 | break; | |
247 | case L: | |
248 | l = links[depth], r = l->rlink, t = r->llink; | |
249 | lr = t->llink, rl = t->rlink; | |
250 | t->llink = l, l->rlink = lr, t->rlink = r, r->llink = rl; | |
251 | l->balance = t->balance != 1 ? 0 : -1; | |
252 | r->balance = t->balance != (char) -1 ? 0 : 1; | |
253 | t->balance = 0; | |
254 | break; | |
255 | default: | |
256 | abort (); | |
257 | } | |
258 | break; | |
259 | default: | |
260 | abort (); | |
261 | } | |
262 | ||
263 | if (dirs[depth - 1] == L) | |
264 | links[depth - 1]->llink = t; | |
265 | else | |
266 | links[depth - 1]->rlink = t; | |
267 | } | |
268 | } | |
269 | ||
270 | trie = link->trie; | |
271 | } | |
272 | ||
273 | /* Mark the node we finally reached as accepting, encoding the | |
274 | index number of this word in the keyword set so far. */ | |
275 | if (!trie->accepting) | |
276 | trie->accepting = 1 + 2 * kwset->words; | |
277 | ++kwset->words; | |
278 | ||
279 | /* Keep track of the longest and shortest string of the keyword set. */ | |
280 | if (trie->depth < kwset->mind) | |
281 | kwset->mind = trie->depth; | |
282 | if (trie->depth > kwset->maxd) | |
283 | kwset->maxd = trie->depth; | |
284 | ||
285 | return NULL; | |
286 | } | |
287 | ||
288 | /* Enqueue the trie nodes referenced from the given tree in the | |
289 | given queue. */ | |
290 | static void | |
291 | enqueue (struct tree *tree, struct trie **last) | |
292 | { | |
293 | if (!tree) | |
294 | return; | |
295 | enqueue(tree->llink, last); | |
296 | enqueue(tree->rlink, last); | |
297 | (*last) = (*last)->next = tree->trie; | |
298 | } | |
299 | ||
300 | /* Compute the Aho-Corasick failure function for the trie nodes referenced | |
301 | from the given tree, given the failure function for their parent as | |
302 | well as a last resort failure node. */ | |
303 | static void | |
304 | treefails (register struct tree const *tree, struct trie const *fail, | |
305 | struct trie *recourse) | |
306 | { | |
307 | register struct tree *link; | |
308 | ||
309 | if (!tree) | |
310 | return; | |
311 | ||
312 | treefails(tree->llink, fail, recourse); | |
313 | treefails(tree->rlink, fail, recourse); | |
314 | ||
315 | /* Find, in the chain of fails going back to the root, the first | |
316 | node that has a descendent on the current label. */ | |
317 | while (fail) | |
318 | { | |
319 | link = fail->links; | |
320 | while (link && tree->label != link->label) | |
321 | if (tree->label < link->label) | |
322 | link = link->llink; | |
323 | else | |
324 | link = link->rlink; | |
325 | if (link) | |
326 | { | |
327 | tree->trie->fail = link->trie; | |
328 | return; | |
329 | } | |
330 | fail = fail->fail; | |
331 | } | |
332 | ||
333 | tree->trie->fail = recourse; | |
334 | } | |
335 | ||
336 | /* Set delta entries for the links of the given tree such that | |
337 | the preexisting delta value is larger than the current depth. */ | |
338 | static void | |
339 | treedelta (register struct tree const *tree, | |
340 | register unsigned int depth, | |
341 | unsigned char delta[]) | |
342 | { | |
343 | if (!tree) | |
344 | return; | |
345 | treedelta(tree->llink, depth, delta); | |
346 | treedelta(tree->rlink, depth, delta); | |
347 | if (depth < delta[tree->label]) | |
348 | delta[tree->label] = depth; | |
349 | } | |
350 | ||
351 | /* Return true if A has every label in B. */ | |
352 | static int | |
353 | hasevery (register struct tree const *a, register struct tree const *b) | |
354 | { | |
355 | if (!b) | |
356 | return 1; | |
357 | if (!hasevery(a, b->llink)) | |
358 | return 0; | |
359 | if (!hasevery(a, b->rlink)) | |
360 | return 0; | |
361 | while (a && b->label != a->label) | |
362 | if (b->label < a->label) | |
363 | a = a->llink; | |
364 | else | |
365 | a = a->rlink; | |
366 | return !!a; | |
367 | } | |
368 | ||
369 | /* Compute a vector, indexed by character code, of the trie nodes | |
370 | referenced from the given tree. */ | |
371 | static void | |
372 | treenext (struct tree const *tree, struct trie *next[]) | |
373 | { | |
374 | if (!tree) | |
375 | return; | |
376 | treenext(tree->llink, next); | |
377 | treenext(tree->rlink, next); | |
378 | next[tree->label] = tree->trie; | |
379 | } | |
380 | ||
381 | /* Compute the shift for each trie node, as well as the delta | |
382 | table and next cache for the given keyword set. */ | |
383 | const char * | |
384 | kwsprep (kwset_t kws) | |
385 | { | |
386 | register struct kwset *kwset; | |
387 | register int i; | |
388 | register struct trie *curr; | |
389 | register char const *trans; | |
390 | unsigned char delta[NCHAR]; | |
391 | ||
392 | kwset = (struct kwset *) kws; | |
393 | ||
394 | /* Initial values for the delta table; will be changed later. The | |
395 | delta entry for a given character is the smallest depth of any | |
396 | node at which an outgoing edge is labeled by that character. */ | |
397 | memset(delta, kwset->mind < UCHAR_MAX ? kwset->mind : UCHAR_MAX, NCHAR); | |
398 | ||
399 | /* Check if we can use the simple boyer-moore algorithm, instead | |
400 | of the hairy commentz-walter algorithm. */ | |
401 | if (kwset->words == 1 && kwset->trans == NULL) | |
402 | { | |
403 | char c; | |
404 | ||
405 | /* Looking for just one string. Extract it from the trie. */ | |
406 | kwset->target = obstack_alloc(&kwset->obstack, kwset->mind); | |
407 | if (!kwset->target) | |
408 | return _("memory exhausted"); | |
409 | for (i = kwset->mind - 1, curr = kwset->trie; i >= 0; --i) | |
410 | { | |
411 | kwset->target[i] = curr->links->label; | |
412 | curr = curr->links->trie; | |
413 | } | |
414 | /* Build the Boyer Moore delta. Boy that's easy compared to CW. */ | |
415 | for (i = 0; i < kwset->mind; ++i) | |
416 | delta[U(kwset->target[i])] = kwset->mind - (i + 1); | |
417 | /* Find the minimal delta2 shift that we might make after | |
418 | a backwards match has failed. */ | |
419 | c = kwset->target[kwset->mind - 1]; | |
420 | for (i = kwset->mind - 2; i >= 0; --i) | |
421 | if (kwset->target[i] == c) | |
422 | break; | |
423 | kwset->mind2 = kwset->mind - (i + 1); | |
424 | } | |
425 | else | |
426 | { | |
427 | register struct trie *fail; | |
428 | struct trie *last, *next[NCHAR]; | |
429 | ||
430 | /* Traverse the nodes of the trie in level order, simultaneously | |
431 | computing the delta table, failure function, and shift function. */ | |
432 | for (curr = last = kwset->trie; curr; curr = curr->next) | |
433 | { | |
434 | /* Enqueue the immediate descendents in the level order queue. */ | |
435 | enqueue(curr->links, &last); | |
436 | ||
437 | curr->shift = kwset->mind; | |
438 | curr->maxshift = kwset->mind; | |
439 | ||
440 | /* Update the delta table for the descendents of this node. */ | |
441 | treedelta(curr->links, curr->depth, delta); | |
442 | ||
443 | /* Compute the failure function for the decendents of this node. */ | |
444 | treefails(curr->links, curr->fail, kwset->trie); | |
445 | ||
446 | /* Update the shifts at each node in the current node's chain | |
447 | of fails back to the root. */ | |
448 | for (fail = curr->fail; fail; fail = fail->fail) | |
449 | { | |
450 | /* If the current node has some outgoing edge that the fail | |
451 | doesn't, then the shift at the fail should be no larger | |
452 | than the difference of their depths. */ | |
453 | if (!hasevery(fail->links, curr->links)) | |
454 | if (curr->depth - fail->depth < fail->shift) | |
455 | fail->shift = curr->depth - fail->depth; | |
456 | ||
457 | /* If the current node is accepting then the shift at the | |
458 | fail and its descendents should be no larger than the | |
459 | difference of their depths. */ | |
460 | if (curr->accepting && fail->maxshift > curr->depth - fail->depth) | |
461 | fail->maxshift = curr->depth - fail->depth; | |
462 | } | |
463 | } | |
464 | ||
465 | /* Traverse the trie in level order again, fixing up all nodes whose | |
466 | shift exceeds their inherited maxshift. */ | |
467 | for (curr = kwset->trie->next; curr; curr = curr->next) | |
468 | { | |
469 | if (curr->maxshift > curr->parent->maxshift) | |
470 | curr->maxshift = curr->parent->maxshift; | |
471 | if (curr->shift > curr->maxshift) | |
472 | curr->shift = curr->maxshift; | |
473 | } | |
474 | ||
475 | /* Create a vector, indexed by character code, of the outgoing links | |
476 | from the root node. */ | |
477 | for (i = 0; i < NCHAR; ++i) | |
478 | next[i] = NULL; | |
479 | treenext(kwset->trie->links, next); | |
480 | ||
481 | if ((trans = kwset->trans) != NULL) | |
482 | for (i = 0; i < NCHAR; ++i) | |
483 | kwset->next[i] = next[U(trans[i])]; | |
484 | else | |
485 | memcpy(kwset->next, next, NCHAR * sizeof(struct trie *)); | |
486 | } | |
487 | ||
488 | /* Fix things up for any translation table. */ | |
489 | if ((trans = kwset->trans) != NULL) | |
490 | for (i = 0; i < NCHAR; ++i) | |
491 | kwset->delta[i] = delta[U(trans[i])]; | |
492 | else | |
493 | memcpy(kwset->delta, delta, NCHAR); | |
494 | ||
495 | return NULL; | |
496 | } | |
497 | ||
498 | /* Fast boyer-moore search. */ | |
499 | static size_t | |
500 | bmexec (kwset_t kws, char const *text, size_t size) | |
501 | { | |
502 | struct kwset const *kwset; | |
503 | register unsigned char const *d1; | |
504 | register char const *ep, *sp, *tp; | |
505 | register int d, gc, i, len, md2; | |
506 | ||
507 | kwset = (struct kwset const *) kws; | |
508 | len = kwset->mind; | |
509 | ||
510 | if (len == 0) | |
511 | return 0; | |
512 | if (len > size) | |
513 | return -1; | |
514 | if (len == 1) | |
515 | { | |
516 | tp = memchr (text, kwset->target[0], size); | |
517 | return tp ? tp - text : -1; | |
518 | } | |
519 | ||
520 | d1 = kwset->delta; | |
521 | sp = kwset->target + len; | |
522 | gc = U(sp[-2]); | |
523 | md2 = kwset->mind2; | |
524 | tp = text + len; | |
525 | ||
526 | /* Significance of 12: 1 (initial offset) + 10 (skip loop) + 1 (md2). */ | |
527 | if (size > 12 * len) | |
528 | /* 11 is not a bug, the initial offset happens only once. */ | |
529 | for (ep = text + size - 11 * len;;) | |
530 | { | |
531 | while (tp <= ep) | |
532 | { | |
533 | d = d1[U(tp[-1])], tp += d; | |
534 | d = d1[U(tp[-1])], tp += d; | |
535 | if (d == 0) | |
536 | goto found; | |
537 | d = d1[U(tp[-1])], tp += d; | |
538 | d = d1[U(tp[-1])], tp += d; | |
539 | d = d1[U(tp[-1])], tp += d; | |
540 | if (d == 0) | |
541 | goto found; | |
542 | d = d1[U(tp[-1])], tp += d; | |
543 | d = d1[U(tp[-1])], tp += d; | |
544 | d = d1[U(tp[-1])], tp += d; | |
545 | if (d == 0) | |
546 | goto found; | |
547 | d = d1[U(tp[-1])], tp += d; | |
548 | d = d1[U(tp[-1])], tp += d; | |
549 | } | |
550 | break; | |
551 | found: | |
552 | if (U(tp[-2]) == gc) | |
553 | { | |
554 | for (i = 3; i <= len && U(tp[-i]) == U(sp[-i]); ++i) | |
555 | ; | |
556 | if (i > len) | |
557 | return tp - len - text; | |
558 | } | |
559 | tp += md2; | |
560 | } | |
561 | ||
562 | /* Now we have only a few characters left to search. We | |
563 | carefully avoid ever producing an out-of-bounds pointer. */ | |
564 | ep = text + size; | |
565 | d = d1[U(tp[-1])]; | |
566 | while (d <= ep - tp) | |
567 | { | |
568 | d = d1[U((tp += d)[-1])]; | |
569 | if (d != 0) | |
570 | continue; | |
571 | if (U(tp[-2]) == gc) | |
572 | { | |
573 | for (i = 3; i <= len && U(tp[-i]) == U(sp[-i]); ++i) | |
574 | ; | |
575 | if (i > len) | |
576 | return tp - len - text; | |
577 | } | |
578 | d = md2; | |
579 | } | |
580 | ||
581 | return -1; | |
582 | } | |
583 | ||
584 | /* Hairy multiple string search. */ | |
585 | static size_t | |
586 | cwexec (kwset_t kws, char const *text, size_t len, struct kwsmatch *kwsmatch) | |
587 | { | |
588 | struct kwset const *kwset; | |
589 | struct trie * const *next; | |
590 | struct trie const *trie; | |
591 | struct trie const *accept; | |
592 | char const *beg, *lim, *mch, *lmch; | |
593 | register unsigned char c; | |
594 | register unsigned char const *delta; | |
595 | register int d; | |
596 | register char const *end, *qlim; | |
597 | register struct tree const *tree; | |
598 | register char const *trans; | |
599 | ||
600 | #ifdef lint | |
601 | accept = NULL; | |
602 | #endif | |
603 | ||
604 | /* Initialize register copies and look for easy ways out. */ | |
605 | kwset = (struct kwset *) kws; | |
606 | if (len < kwset->mind) | |
607 | return -1; | |
608 | next = kwset->next; | |
609 | delta = kwset->delta; | |
610 | trans = kwset->trans; | |
611 | lim = text + len; | |
612 | end = text; | |
613 | if ((d = kwset->mind) != 0) | |
614 | mch = NULL; | |
615 | else | |
616 | { | |
617 | mch = text, accept = kwset->trie; | |
618 | goto match; | |
619 | } | |
620 | ||
621 | if (len >= 4 * kwset->mind) | |
622 | qlim = lim - 4 * kwset->mind; | |
623 | else | |
624 | qlim = NULL; | |
625 | ||
626 | while (lim - end >= d) | |
627 | { | |
628 | if (qlim && end <= qlim) | |
629 | { | |
630 | end += d - 1; | |
631 | while ((d = delta[c = *end]) && end < qlim) | |
632 | { | |
633 | end += d; | |
634 | end += delta[U(*end)]; | |
635 | end += delta[U(*end)]; | |
636 | } | |
637 | ++end; | |
638 | } | |
639 | else | |
640 | d = delta[c = (end += d)[-1]]; | |
641 | if (d) | |
642 | continue; | |
643 | beg = end - 1; | |
644 | trie = next[c]; | |
645 | if (trie->accepting) | |
646 | { | |
647 | mch = beg; | |
648 | accept = trie; | |
649 | } | |
650 | d = trie->shift; | |
651 | while (beg > text) | |
652 | { | |
653 | c = trans ? trans[U(*--beg)] : *--beg; | |
654 | tree = trie->links; | |
655 | while (tree && c != tree->label) | |
656 | if (c < tree->label) | |
657 | tree = tree->llink; | |
658 | else | |
659 | tree = tree->rlink; | |
660 | if (tree) | |
661 | { | |
662 | trie = tree->trie; | |
663 | if (trie->accepting) | |
664 | { | |
665 | mch = beg; | |
666 | accept = trie; | |
667 | } | |
668 | } | |
669 | else | |
670 | break; | |
671 | d = trie->shift; | |
672 | } | |
673 | if (mch) | |
674 | goto match; | |
675 | } | |
676 | return -1; | |
677 | ||
678 | match: | |
679 | /* Given a known match, find the longest possible match anchored | |
680 | at or before its starting point. This is nearly a verbatim | |
681 | copy of the preceding main search loops. */ | |
682 | if (lim - mch > kwset->maxd) | |
683 | lim = mch + kwset->maxd; | |
684 | lmch = 0; | |
685 | d = 1; | |
686 | while (lim - end >= d) | |
687 | { | |
688 | if ((d = delta[c = (end += d)[-1]]) != 0) | |
689 | continue; | |
690 | beg = end - 1; | |
691 | if (!(trie = next[c])) | |
692 | { | |
693 | d = 1; | |
694 | continue; | |
695 | } | |
696 | if (trie->accepting && beg <= mch) | |
697 | { | |
698 | lmch = beg; | |
699 | accept = trie; | |
700 | } | |
701 | d = trie->shift; | |
702 | while (beg > text) | |
703 | { | |
704 | c = trans ? trans[U(*--beg)] : *--beg; | |
705 | tree = trie->links; | |
706 | while (tree && c != tree->label) | |
707 | if (c < tree->label) | |
708 | tree = tree->llink; | |
709 | else | |
710 | tree = tree->rlink; | |
711 | if (tree) | |
712 | { | |
713 | trie = tree->trie; | |
714 | if (trie->accepting && beg <= mch) | |
715 | { | |
716 | lmch = beg; | |
717 | accept = trie; | |
718 | } | |
719 | } | |
720 | else | |
721 | break; | |
722 | d = trie->shift; | |
723 | } | |
724 | if (lmch) | |
725 | { | |
726 | mch = lmch; | |
727 | goto match; | |
728 | } | |
729 | if (!d) | |
730 | d = 1; | |
731 | } | |
732 | ||
733 | if (kwsmatch) | |
734 | { | |
735 | kwsmatch->index = accept->accepting / 2; | |
736 | kwsmatch->offset[0] = mch - text; | |
737 | kwsmatch->size[0] = accept->depth; | |
738 | } | |
739 | return mch - text; | |
740 | } | |
741 | ||
742 | /* Search through the given text for a match of any member of the | |
743 | given keyword set. Return a pointer to the first character of | |
744 | the matching substring, or NULL if no match is found. If FOUNDLEN | |
745 | is non-NULL store in the referenced location the length of the | |
746 | matching substring. Similarly, if FOUNDIDX is non-NULL, store | |
747 | in the referenced location the index number of the particular | |
748 | keyword matched. */ | |
749 | size_t | |
750 | kwsexec (kwset_t kws, char const *text, size_t size, | |
751 | struct kwsmatch *kwsmatch) | |
752 | { | |
753 | struct kwset const *kwset = (struct kwset *) kws; | |
754 | if (kwset->words == 1 && kwset->trans == NULL) | |
755 | { | |
756 | size_t ret = bmexec (kws, text, size); | |
757 | if (kwsmatch != NULL && ret != (size_t) -1) | |
758 | { | |
759 | kwsmatch->index = 0; | |
760 | kwsmatch->offset[0] = ret; | |
761 | kwsmatch->size[0] = kwset->mind; | |
762 | } | |
763 | return ret; | |
764 | } | |
765 | else | |
766 | return cwexec(kws, text, size, kwsmatch); | |
767 | } | |
768 | ||
769 | /* Free the components of the given keyword set. */ | |
770 | void | |
771 | kwsfree (kwset_t kws) | |
772 | { | |
773 | struct kwset *kwset; | |
774 | ||
775 | kwset = (struct kwset *) kws; | |
776 | obstack_free(&kwset->obstack, NULL); | |
777 | free(kws); | |
778 | } |