]>
Commit | Line | Data |
---|---|---|
dd3b648e RP |
1 | /* Extended regular expression matching and search library. |
2 | Copyright (C) 1985, 1989 Free Software Foundation, Inc. | |
3 | ||
99a7de40 JG |
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 of the License, or | |
7 | (at your option) 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., 675 Mass Ave, Cambridge, MA 02139, USA. */ | |
dd3b648e RP |
17 | |
18 | /* To test, compile with -Dtest. | |
19 | This Dtestable feature turns this into a self-contained program | |
20 | which reads a pattern, describes how it compiles, | |
21 | then reads a string and searches for it. */ | |
22 | ||
23 | #ifdef emacs | |
24 | ||
25 | /* The `emacs' switch turns on certain special matching commands | |
26 | that make sense only in emacs. */ | |
27 | ||
28 | #include "config.h" | |
29 | #include "lisp.h" | |
30 | #include "buffer.h" | |
31 | #include "syntax.h" | |
32 | ||
33 | #else /* not emacs */ | |
34 | ||
ba47c66a PS |
35 | #include "defs.h" |
36 | #include <string.h> | |
dd3b648e RP |
37 | |
38 | /* | |
39 | * Define the syntax stuff, so we can do the \<...\> things. | |
40 | */ | |
41 | ||
42 | #ifndef Sword /* must be non-zero in some of the tests below... */ | |
43 | #define Sword 1 | |
44 | #endif | |
45 | ||
46 | #define SYNTAX(c) re_syntax_table[c] | |
47 | ||
48 | #ifdef SYNTAX_TABLE | |
49 | ||
50 | char *re_syntax_table; | |
51 | ||
52 | #else | |
53 | ||
54 | static char re_syntax_table[256]; | |
55 | ||
56 | static void | |
57 | init_syntax_once () | |
58 | { | |
59 | register int c; | |
60 | static int done = 0; | |
61 | ||
62 | if (done) | |
63 | return; | |
64 | ||
4ed97c9a | 65 | memset (re_syntax_table, '\0', sizeof re_syntax_table); |
dd3b648e RP |
66 | |
67 | for (c = 'a'; c <= 'z'; c++) | |
68 | re_syntax_table[c] = Sword; | |
69 | ||
70 | for (c = 'A'; c <= 'Z'; c++) | |
71 | re_syntax_table[c] = Sword; | |
72 | ||
73 | for (c = '0'; c <= '9'; c++) | |
74 | re_syntax_table[c] = Sword; | |
75 | ||
76 | done = 1; | |
77 | } | |
78 | ||
79 | #endif /* SYNTAX_TABLE */ | |
80 | #endif /* not emacs */ | |
81 | ||
82 | #include "regex.h" | |
83 | ||
84 | /* Number of failure points to allocate space for initially, | |
85 | when matching. If this number is exceeded, more space is allocated, | |
86 | so it is not a hard limit. */ | |
87 | ||
88 | #ifndef NFAILURES | |
89 | #define NFAILURES 80 | |
90 | #endif /* NFAILURES */ | |
91 | ||
92 | /* width of a byte in bits */ | |
93 | ||
94 | #define BYTEWIDTH 8 | |
95 | ||
96 | #ifndef SIGN_EXTEND_CHAR | |
97 | #define SIGN_EXTEND_CHAR(x) (x) | |
98 | #endif | |
99 | \f | |
100 | static int obscure_syntax = 0; | |
101 | ||
102 | /* Specify the precise syntax of regexp for compilation. | |
103 | This provides for compatibility for various utilities | |
104 | which historically have different, incompatible syntaxes. | |
105 | ||
106 | The argument SYNTAX is a bit-mask containing the two bits | |
107 | RE_NO_BK_PARENS and RE_NO_BK_VBAR. */ | |
108 | ||
109 | int | |
110 | re_set_syntax (syntax) | |
51b57ded | 111 | int syntax; |
dd3b648e RP |
112 | { |
113 | int ret; | |
114 | ||
115 | ret = obscure_syntax; | |
116 | obscure_syntax = syntax; | |
117 | return ret; | |
118 | } | |
119 | \f | |
120 | /* re_compile_pattern takes a regular-expression string | |
121 | and converts it into a buffer full of byte commands for matching. | |
122 | ||
123 | PATTERN is the address of the pattern string | |
124 | SIZE is the length of it. | |
125 | BUFP is a struct re_pattern_buffer * which points to the info | |
126 | on where to store the byte commands. | |
127 | This structure contains a char * which points to the | |
128 | actual space, which should have been obtained with malloc. | |
129 | re_compile_pattern may use realloc to grow the buffer space. | |
130 | ||
131 | The number of bytes of commands can be found out by looking in | |
132 | the struct re_pattern_buffer that bufp pointed to, | |
133 | after re_compile_pattern returns. | |
134 | */ | |
135 | ||
136 | #define PATPUSH(ch) (*b++ = (char) (ch)) | |
137 | ||
138 | #define PATFETCH(c) \ | |
139 | {if (p == pend) goto end_of_pattern; \ | |
140 | c = * (unsigned char *) p++; \ | |
141 | if (translate) c = translate[c]; } | |
142 | ||
143 | #define PATFETCH_RAW(c) \ | |
144 | {if (p == pend) goto end_of_pattern; \ | |
145 | c = * (unsigned char *) p++; } | |
146 | ||
147 | #define PATUNFETCH p-- | |
148 | ||
ee6d646a JK |
149 | /* This is not an arbitrary limit: the arguments which represent offsets |
150 | into the pattern are two bytes long. So if 2^16 bytes turns out to | |
151 | be too small, many things would have to change. */ | |
152 | #define MAX_BUF_SIZE (1 << 16) | |
153 | ||
154 | ||
155 | /* Extend the buffer by twice its current size via realloc and | |
156 | reset the pointers that pointed into the old block to point to the | |
157 | correct places in the new one. If extending the buffer results in it | |
158 | being larger than MAX_BUF_SIZE, then flag memory exhausted. */ | |
159 | #define EXTEND_BUFFER \ | |
160 | do { \ | |
161 | char *old_buffer = bufp->buffer; \ | |
162 | if (bufp->allocated == MAX_BUF_SIZE) \ | |
163 | goto too_big; \ | |
164 | bufp->allocated <<= 1; \ | |
165 | if (bufp->allocated > MAX_BUF_SIZE) \ | |
166 | bufp->allocated = MAX_BUF_SIZE; \ | |
167 | bufp->buffer = (char *) realloc (bufp->buffer, bufp->allocated);\ | |
168 | if (bufp->buffer == NULL) \ | |
169 | goto memory_exhausted; \ | |
170 | /* If the buffer moved, move all the pointers into it. */ \ | |
171 | if (old_buffer != bufp->buffer) \ | |
172 | { \ | |
173 | b = (b - old_buffer) + bufp->buffer; \ | |
174 | begalt = (begalt - old_buffer) + bufp->buffer; \ | |
175 | if (fixup_jump) \ | |
176 | fixup_jump = (fixup_jump - old_buffer) + bufp->buffer;\ | |
177 | if (laststart) \ | |
178 | laststart = (laststart - old_buffer) + bufp->buffer; \ | |
179 | if (pending_exact) \ | |
180 | pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ | |
181 | } \ | |
182 | } while (0) | |
dd3b648e | 183 | |
50e0dc41 | 184 | static void store_jump (), insert_jump (); |
dd3b648e RP |
185 | |
186 | char * | |
187 | re_compile_pattern (pattern, size, bufp) | |
188 | char *pattern; | |
189 | int size; | |
190 | struct re_pattern_buffer *bufp; | |
191 | { | |
192 | register char *b = bufp->buffer; | |
193 | register char *p = pattern; | |
194 | char *pend = pattern + size; | |
195 | register unsigned c, c1; | |
196 | char *p1; | |
197 | unsigned char *translate = (unsigned char *) bufp->translate; | |
198 | ||
199 | /* address of the count-byte of the most recently inserted "exactn" command. | |
200 | This makes it possible to tell whether a new exact-match character | |
201 | can be added to that command or requires a new "exactn" command. */ | |
202 | ||
203 | char *pending_exact = 0; | |
204 | ||
205 | /* address of the place where a forward-jump should go | |
206 | to the end of the containing expression. | |
207 | Each alternative of an "or", except the last, ends with a forward-jump | |
208 | of this sort. */ | |
209 | ||
210 | char *fixup_jump = 0; | |
211 | ||
212 | /* address of start of the most recently finished expression. | |
213 | This tells postfix * where to find the start of its operand. */ | |
214 | ||
215 | char *laststart = 0; | |
216 | ||
217 | /* In processing a repeat, 1 means zero matches is allowed */ | |
218 | ||
219 | char zero_times_ok; | |
220 | ||
221 | /* In processing a repeat, 1 means many matches is allowed */ | |
222 | ||
223 | char many_times_ok; | |
224 | ||
225 | /* address of beginning of regexp, or inside of last \( */ | |
226 | ||
227 | char *begalt = b; | |
228 | ||
229 | /* Stack of information saved by \( and restored by \). | |
230 | Four stack elements are pushed by each \(: | |
231 | First, the value of b. | |
232 | Second, the value of fixup_jump. | |
233 | Third, the value of regnum. | |
234 | Fourth, the value of begalt. */ | |
235 | ||
236 | int stackb[40]; | |
237 | int *stackp = stackb; | |
238 | int *stacke = stackb + 40; | |
239 | int *stackt; | |
240 | ||
241 | /* Counts \('s as they are encountered. Remembered for the matching \), | |
242 | where it becomes the "register number" to put in the stop_memory command */ | |
243 | ||
244 | int regnum = 1; | |
245 | ||
246 | bufp->fastmap_accurate = 0; | |
247 | ||
248 | #ifndef emacs | |
249 | #ifndef SYNTAX_TABLE | |
250 | /* | |
251 | * Initialize the syntax table. | |
252 | */ | |
253 | init_syntax_once(); | |
254 | #endif | |
255 | #endif | |
256 | ||
257 | if (bufp->allocated == 0) | |
258 | { | |
259 | bufp->allocated = 28; | |
260 | if (bufp->buffer) | |
261 | /* EXTEND_BUFFER loses when bufp->allocated is 0 */ | |
262 | bufp->buffer = (char *) realloc (bufp->buffer, 28); | |
263 | else | |
264 | /* Caller did not allocate a buffer. Do it for him */ | |
265 | bufp->buffer = (char *) malloc (28); | |
266 | if (!bufp->buffer) goto memory_exhausted; | |
267 | begalt = b = bufp->buffer; | |
268 | } | |
269 | ||
270 | while (p != pend) | |
271 | { | |
272 | if (b - bufp->buffer > bufp->allocated - 10) | |
273 | /* Note that EXTEND_BUFFER clobbers c */ | |
274 | EXTEND_BUFFER; | |
275 | ||
276 | PATFETCH (c); | |
277 | ||
278 | switch (c) | |
279 | { | |
280 | case '$': | |
281 | if (obscure_syntax & RE_TIGHT_VBAR) | |
282 | { | |
283 | if (! (obscure_syntax & RE_CONTEXT_INDEP_OPS) && p != pend) | |
284 | goto normal_char; | |
285 | /* Make operand of last vbar end before this `$'. */ | |
286 | if (fixup_jump) | |
287 | store_jump (fixup_jump, jump, b); | |
288 | fixup_jump = 0; | |
289 | PATPUSH (endline); | |
290 | break; | |
291 | } | |
292 | ||
293 | /* $ means succeed if at end of line, but only in special contexts. | |
294 | If randomly in the middle of a pattern, it is a normal character. */ | |
295 | if (p == pend || *p == '\n' | |
296 | || (obscure_syntax & RE_CONTEXT_INDEP_OPS) | |
297 | || (obscure_syntax & RE_NO_BK_PARENS | |
298 | ? *p == ')' | |
299 | : *p == '\\' && p[1] == ')') | |
300 | || (obscure_syntax & RE_NO_BK_VBAR | |
301 | ? *p == '|' | |
302 | : *p == '\\' && p[1] == '|')) | |
303 | { | |
304 | PATPUSH (endline); | |
305 | break; | |
306 | } | |
307 | goto normal_char; | |
308 | ||
309 | case '^': | |
310 | /* ^ means succeed if at beg of line, but only if no preceding pattern. */ | |
311 | ||
312 | if (laststart && p[-2] != '\n' | |
313 | && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) | |
314 | goto normal_char; | |
315 | if (obscure_syntax & RE_TIGHT_VBAR) | |
316 | { | |
317 | if (p != pattern + 1 | |
318 | && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) | |
319 | goto normal_char; | |
320 | PATPUSH (begline); | |
321 | begalt = b; | |
322 | } | |
323 | else | |
324 | PATPUSH (begline); | |
325 | break; | |
326 | ||
327 | case '+': | |
328 | case '?': | |
329 | if (obscure_syntax & RE_BK_PLUS_QM) | |
330 | goto normal_char; | |
331 | handle_plus: | |
332 | case '*': | |
333 | /* If there is no previous pattern, char not special. */ | |
334 | if (!laststart && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) | |
335 | goto normal_char; | |
336 | /* If there is a sequence of repetition chars, | |
337 | collapse it down to equivalent to just one. */ | |
338 | zero_times_ok = 0; | |
339 | many_times_ok = 0; | |
340 | while (1) | |
341 | { | |
342 | zero_times_ok |= c != '+'; | |
343 | many_times_ok |= c != '?'; | |
344 | if (p == pend) | |
345 | break; | |
346 | PATFETCH (c); | |
347 | if (c == '*') | |
348 | ; | |
349 | else if (!(obscure_syntax & RE_BK_PLUS_QM) | |
350 | && (c == '+' || c == '?')) | |
351 | ; | |
352 | else if ((obscure_syntax & RE_BK_PLUS_QM) | |
353 | && c == '\\') | |
354 | { | |
355 | int c1; | |
356 | PATFETCH (c1); | |
357 | if (!(c1 == '+' || c1 == '?')) | |
358 | { | |
359 | PATUNFETCH; | |
360 | PATUNFETCH; | |
361 | break; | |
362 | } | |
363 | c = c1; | |
364 | } | |
365 | else | |
366 | { | |
367 | PATUNFETCH; | |
368 | break; | |
369 | } | |
370 | } | |
371 | ||
372 | /* Star, etc. applied to an empty pattern is equivalent | |
373 | to an empty pattern. */ | |
374 | if (!laststart) | |
375 | break; | |
376 | ||
377 | /* Now we know whether 0 matches is allowed, | |
378 | and whether 2 or more matches is allowed. */ | |
379 | if (many_times_ok) | |
380 | { | |
381 | /* If more than one repetition is allowed, | |
382 | put in a backward jump at the end. */ | |
383 | store_jump (b, maybe_finalize_jump, laststart - 3); | |
384 | b += 3; | |
385 | } | |
386 | insert_jump (on_failure_jump, laststart, b + 3, b); | |
387 | pending_exact = 0; | |
388 | b += 3; | |
389 | if (!zero_times_ok) | |
390 | { | |
391 | /* At least one repetition required: insert before the loop | |
392 | a skip over the initial on-failure-jump instruction */ | |
393 | insert_jump (dummy_failure_jump, laststart, laststart + 6, b); | |
394 | b += 3; | |
395 | } | |
396 | break; | |
397 | ||
398 | case '.': | |
399 | laststart = b; | |
400 | PATPUSH (anychar); | |
401 | break; | |
402 | ||
403 | case '[': | |
404 | while (b - bufp->buffer | |
405 | > bufp->allocated - 3 - (1 << BYTEWIDTH) / BYTEWIDTH) | |
406 | /* Note that EXTEND_BUFFER clobbers c */ | |
407 | EXTEND_BUFFER; | |
408 | ||
409 | laststart = b; | |
410 | if (*p == '^') | |
411 | PATPUSH (charset_not), p++; | |
412 | else | |
413 | PATPUSH (charset); | |
414 | p1 = p; | |
415 | ||
416 | PATPUSH ((1 << BYTEWIDTH) / BYTEWIDTH); | |
417 | /* Clear the whole map */ | |
4ed97c9a | 418 | memset (b, '\0', (1 << BYTEWIDTH) / BYTEWIDTH); |
dd3b648e RP |
419 | /* Read in characters and ranges, setting map bits */ |
420 | while (1) | |
421 | { | |
422 | PATFETCH (c); | |
423 | if (c == ']' && p != p1 + 1) break; | |
424 | if (*p == '-' && p[1] != ']') | |
425 | { | |
426 | PATFETCH (c1); | |
427 | PATFETCH (c1); | |
428 | while (c <= c1) | |
429 | b[c / BYTEWIDTH] |= 1 << (c % BYTEWIDTH), c++; | |
430 | } | |
431 | else | |
432 | { | |
433 | b[c / BYTEWIDTH] |= 1 << (c % BYTEWIDTH); | |
434 | } | |
435 | } | |
436 | /* Discard any bitmap bytes that are all 0 at the end of the map. | |
437 | Decrement the map-length byte too. */ | |
438 | while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) | |
439 | b[-1]--; | |
440 | b += b[-1]; | |
441 | break; | |
442 | ||
443 | case '(': | |
444 | if (! (obscure_syntax & RE_NO_BK_PARENS)) | |
445 | goto normal_char; | |
446 | else | |
447 | goto handle_open; | |
448 | ||
449 | case ')': | |
450 | if (! (obscure_syntax & RE_NO_BK_PARENS)) | |
451 | goto normal_char; | |
452 | else | |
453 | goto handle_close; | |
454 | ||
455 | case '\n': | |
456 | if (! (obscure_syntax & RE_NEWLINE_OR)) | |
457 | goto normal_char; | |
458 | else | |
459 | goto handle_bar; | |
460 | ||
461 | case '|': | |
462 | if (! (obscure_syntax & RE_NO_BK_VBAR)) | |
463 | goto normal_char; | |
464 | else | |
465 | goto handle_bar; | |
466 | ||
467 | case '\\': | |
468 | if (p == pend) goto invalid_pattern; | |
469 | PATFETCH_RAW (c); | |
470 | switch (c) | |
471 | { | |
472 | case '(': | |
473 | if (obscure_syntax & RE_NO_BK_PARENS) | |
474 | goto normal_backsl; | |
475 | handle_open: | |
476 | if (stackp == stacke) goto nesting_too_deep; | |
477 | if (regnum < RE_NREGS) | |
478 | { | |
479 | PATPUSH (start_memory); | |
480 | PATPUSH (regnum); | |
481 | } | |
482 | *stackp++ = b - bufp->buffer; | |
483 | *stackp++ = fixup_jump ? fixup_jump - bufp->buffer + 1 : 0; | |
484 | *stackp++ = regnum++; | |
485 | *stackp++ = begalt - bufp->buffer; | |
486 | fixup_jump = 0; | |
487 | laststart = 0; | |
488 | begalt = b; | |
489 | break; | |
490 | ||
491 | case ')': | |
492 | if (obscure_syntax & RE_NO_BK_PARENS) | |
493 | goto normal_backsl; | |
494 | handle_close: | |
495 | if (stackp == stackb) goto unmatched_close; | |
496 | begalt = *--stackp + bufp->buffer; | |
497 | if (fixup_jump) | |
498 | store_jump (fixup_jump, jump, b); | |
499 | if (stackp[-1] < RE_NREGS) | |
500 | { | |
501 | PATPUSH (stop_memory); | |
502 | PATPUSH (stackp[-1]); | |
503 | } | |
504 | stackp -= 2; | |
505 | fixup_jump = 0; | |
506 | if (*stackp) | |
507 | fixup_jump = *stackp + bufp->buffer - 1; | |
508 | laststart = *--stackp + bufp->buffer; | |
509 | break; | |
510 | ||
511 | case '|': | |
512 | if (obscure_syntax & RE_NO_BK_VBAR) | |
513 | goto normal_backsl; | |
514 | handle_bar: | |
515 | insert_jump (on_failure_jump, begalt, b + 6, b); | |
516 | pending_exact = 0; | |
517 | b += 3; | |
518 | if (fixup_jump) | |
519 | store_jump (fixup_jump, jump, b); | |
520 | fixup_jump = b; | |
521 | b += 3; | |
522 | laststart = 0; | |
523 | begalt = b; | |
524 | break; | |
525 | ||
526 | #ifdef emacs | |
527 | case '=': | |
528 | PATPUSH (at_dot); | |
529 | break; | |
530 | ||
531 | case 's': | |
532 | laststart = b; | |
533 | PATPUSH (syntaxspec); | |
534 | PATFETCH (c); | |
535 | PATPUSH (syntax_spec_code[c]); | |
536 | break; | |
537 | ||
538 | case 'S': | |
539 | laststart = b; | |
540 | PATPUSH (notsyntaxspec); | |
541 | PATFETCH (c); | |
542 | PATPUSH (syntax_spec_code[c]); | |
543 | break; | |
544 | #endif /* emacs */ | |
545 | ||
546 | case 'w': | |
547 | laststart = b; | |
548 | PATPUSH (wordchar); | |
549 | break; | |
550 | ||
551 | case 'W': | |
552 | laststart = b; | |
553 | PATPUSH (notwordchar); | |
554 | break; | |
555 | ||
556 | case '<': | |
557 | PATPUSH (wordbeg); | |
558 | break; | |
559 | ||
560 | case '>': | |
561 | PATPUSH (wordend); | |
562 | break; | |
563 | ||
564 | case 'b': | |
565 | PATPUSH (wordbound); | |
566 | break; | |
567 | ||
568 | case 'B': | |
569 | PATPUSH (notwordbound); | |
570 | break; | |
571 | ||
572 | case '`': | |
573 | PATPUSH (begbuf); | |
574 | break; | |
575 | ||
576 | case '\'': | |
577 | PATPUSH (endbuf); | |
578 | break; | |
579 | ||
580 | case '1': | |
581 | case '2': | |
582 | case '3': | |
583 | case '4': | |
584 | case '5': | |
585 | case '6': | |
586 | case '7': | |
587 | case '8': | |
588 | case '9': | |
589 | c1 = c - '0'; | |
590 | if (c1 >= regnum) | |
591 | goto normal_char; | |
592 | for (stackt = stackp - 2; stackt > stackb; stackt -= 4) | |
593 | if (*stackt == c1) | |
594 | goto normal_char; | |
595 | laststart = b; | |
596 | PATPUSH (duplicate); | |
597 | PATPUSH (c1); | |
598 | break; | |
599 | ||
600 | case '+': | |
601 | case '?': | |
602 | if (obscure_syntax & RE_BK_PLUS_QM) | |
603 | goto handle_plus; | |
604 | ||
605 | default: | |
606 | normal_backsl: | |
607 | /* You might think it would be useful for \ to mean | |
608 | not to translate; but if we don't translate it | |
609 | it will never match anything. */ | |
610 | if (translate) c = translate[c]; | |
611 | goto normal_char; | |
612 | } | |
613 | break; | |
614 | ||
615 | default: | |
616 | normal_char: | |
617 | if (!pending_exact || pending_exact + *pending_exact + 1 != b | |
618 | || *pending_exact == 0177 || *p == '*' || *p == '^' | |
619 | || ((obscure_syntax & RE_BK_PLUS_QM) | |
620 | ? *p == '\\' && (p[1] == '+' || p[1] == '?') | |
621 | : (*p == '+' || *p == '?'))) | |
622 | { | |
623 | laststart = b; | |
624 | PATPUSH (exactn); | |
625 | pending_exact = b; | |
626 | PATPUSH (0); | |
627 | } | |
628 | PATPUSH (c); | |
629 | (*pending_exact)++; | |
630 | } | |
631 | } | |
632 | ||
633 | if (fixup_jump) | |
634 | store_jump (fixup_jump, jump, b); | |
635 | ||
636 | if (stackp != stackb) goto unmatched_open; | |
637 | ||
638 | bufp->used = b - bufp->buffer; | |
639 | return 0; | |
640 | ||
641 | invalid_pattern: | |
642 | return "Invalid regular expression"; | |
643 | ||
644 | unmatched_open: | |
645 | return "Unmatched \\("; | |
646 | ||
647 | unmatched_close: | |
648 | return "Unmatched \\)"; | |
649 | ||
650 | end_of_pattern: | |
651 | return "Premature end of regular expression"; | |
652 | ||
653 | nesting_too_deep: | |
654 | return "Nesting too deep"; | |
655 | ||
656 | too_big: | |
657 | return "Regular expression too big"; | |
658 | ||
659 | memory_exhausted: | |
660 | return "Memory exhausted"; | |
661 | } | |
662 | ||
663 | /* Store where `from' points a jump operation to jump to where `to' points. | |
664 | `opcode' is the opcode to store. */ | |
665 | ||
50e0dc41 | 666 | static void |
dd3b648e RP |
667 | store_jump (from, opcode, to) |
668 | char *from, *to; | |
669 | char opcode; | |
670 | { | |
671 | from[0] = opcode; | |
672 | from[1] = (to - (from + 3)) & 0377; | |
673 | from[2] = (to - (from + 3)) >> 8; | |
674 | } | |
675 | ||
676 | /* Open up space at char FROM, and insert there a jump to TO. | |
677 | CURRENT_END gives te end of the storage no in use, | |
678 | so we know how much data to copy up. | |
679 | OP is the opcode of the jump to insert. | |
680 | ||
681 | If you call this function, you must zero out pending_exact. */ | |
682 | ||
50e0dc41 | 683 | static void |
dd3b648e RP |
684 | insert_jump (op, from, to, current_end) |
685 | char op; | |
686 | char *from, *to, *current_end; | |
687 | { | |
688 | register char *pto = current_end + 3; | |
689 | register char *pfrom = current_end; | |
690 | while (pfrom != from) | |
691 | *--pto = *--pfrom; | |
692 | store_jump (from, op, to); | |
693 | } | |
694 | \f | |
695 | /* Given a pattern, compute a fastmap from it. | |
696 | The fastmap records which of the (1 << BYTEWIDTH) possible characters | |
697 | can start a string that matches the pattern. | |
698 | This fastmap is used by re_search to skip quickly over totally implausible text. | |
699 | ||
700 | The caller must supply the address of a (1 << BYTEWIDTH)-byte data area | |
701 | as bufp->fastmap. | |
702 | The other components of bufp describe the pattern to be used. */ | |
703 | ||
704 | void | |
705 | re_compile_fastmap (bufp) | |
706 | struct re_pattern_buffer *bufp; | |
707 | { | |
708 | unsigned char *pattern = (unsigned char *) bufp->buffer; | |
709 | int size = bufp->used; | |
710 | register char *fastmap = bufp->fastmap; | |
711 | register unsigned char *p = pattern; | |
712 | register unsigned char *pend = pattern + size; | |
51b57ded | 713 | register int j; |
dd3b648e RP |
714 | unsigned char *translate = (unsigned char *) bufp->translate; |
715 | ||
716 | unsigned char *stackb[NFAILURES]; | |
717 | unsigned char **stackp = stackb; | |
718 | ||
4ed97c9a | 719 | memset (fastmap, '\0', (1 << BYTEWIDTH)); |
dd3b648e RP |
720 | bufp->fastmap_accurate = 1; |
721 | bufp->can_be_null = 0; | |
722 | ||
723 | while (p) | |
724 | { | |
725 | if (p == pend) | |
726 | { | |
727 | bufp->can_be_null = 1; | |
728 | break; | |
729 | } | |
730 | #ifdef SWITCH_ENUM_BUG | |
731 | switch ((int) ((enum regexpcode) *p++)) | |
732 | #else | |
733 | switch ((enum regexpcode) *p++) | |
734 | #endif | |
735 | { | |
736 | case exactn: | |
737 | if (translate) | |
738 | fastmap[translate[p[1]]] = 1; | |
739 | else | |
740 | fastmap[p[1]] = 1; | |
741 | break; | |
742 | ||
743 | case begline: | |
744 | case before_dot: | |
745 | case at_dot: | |
746 | case after_dot: | |
747 | case begbuf: | |
748 | case endbuf: | |
749 | case wordbound: | |
750 | case notwordbound: | |
751 | case wordbeg: | |
752 | case wordend: | |
753 | continue; | |
754 | ||
755 | case endline: | |
756 | if (translate) | |
757 | fastmap[translate['\n']] = 1; | |
758 | else | |
759 | fastmap['\n'] = 1; | |
760 | if (bufp->can_be_null != 1) | |
761 | bufp->can_be_null = 2; | |
762 | break; | |
763 | ||
764 | case finalize_jump: | |
765 | case maybe_finalize_jump: | |
766 | case jump: | |
767 | case dummy_failure_jump: | |
768 | bufp->can_be_null = 1; | |
769 | j = *p++ & 0377; | |
770 | j += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
771 | p += j + 1; /* The 1 compensates for missing ++ above */ | |
772 | if (j > 0) | |
773 | continue; | |
774 | /* Jump backward reached implies we just went through | |
775 | the body of a loop and matched nothing. | |
776 | Opcode jumped to should be an on_failure_jump. | |
777 | Just treat it like an ordinary jump. | |
778 | For a * loop, it has pushed its failure point already; | |
779 | if so, discard that as redundant. */ | |
780 | if ((enum regexpcode) *p != on_failure_jump) | |
781 | continue; | |
782 | p++; | |
783 | j = *p++ & 0377; | |
784 | j += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
785 | p += j + 1; /* The 1 compensates for missing ++ above */ | |
786 | if (stackp != stackb && *stackp == p) | |
787 | stackp--; | |
788 | continue; | |
789 | ||
790 | case on_failure_jump: | |
791 | j = *p++ & 0377; | |
792 | j += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
793 | p++; | |
794 | *++stackp = p + j; | |
795 | continue; | |
796 | ||
797 | case start_memory: | |
798 | case stop_memory: | |
799 | p++; | |
800 | continue; | |
801 | ||
802 | case duplicate: | |
803 | bufp->can_be_null = 1; | |
804 | fastmap['\n'] = 1; | |
805 | case anychar: | |
806 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
807 | if (j != '\n') | |
808 | fastmap[j] = 1; | |
809 | if (bufp->can_be_null) | |
810 | return; | |
811 | /* Don't return; check the alternative paths | |
812 | so we can set can_be_null if appropriate. */ | |
813 | break; | |
814 | ||
815 | case wordchar: | |
816 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
817 | if (SYNTAX (j) == Sword) | |
818 | fastmap[j] = 1; | |
819 | break; | |
820 | ||
821 | case notwordchar: | |
822 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
823 | if (SYNTAX (j) != Sword) | |
824 | fastmap[j] = 1; | |
825 | break; | |
826 | ||
827 | #ifdef emacs | |
828 | case syntaxspec: | |
829 | k = *p++; | |
830 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
831 | if (SYNTAX (j) == (enum syntaxcode) k) | |
832 | fastmap[j] = 1; | |
833 | break; | |
834 | ||
835 | case notsyntaxspec: | |
836 | k = *p++; | |
837 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
838 | if (SYNTAX (j) != (enum syntaxcode) k) | |
839 | fastmap[j] = 1; | |
840 | break; | |
841 | #endif /* emacs */ | |
842 | ||
843 | case charset: | |
844 | for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | |
845 | if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) | |
846 | { | |
847 | if (translate) | |
848 | fastmap[translate[j]] = 1; | |
849 | else | |
850 | fastmap[j] = 1; | |
851 | } | |
852 | break; | |
853 | ||
854 | case charset_not: | |
855 | /* Chars beyond end of map must be allowed */ | |
856 | for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) | |
857 | if (translate) | |
858 | fastmap[translate[j]] = 1; | |
859 | else | |
860 | fastmap[j] = 1; | |
861 | ||
862 | for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | |
863 | if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) | |
864 | { | |
865 | if (translate) | |
866 | fastmap[translate[j]] = 1; | |
867 | else | |
868 | fastmap[j] = 1; | |
869 | } | |
870 | break; | |
871 | } | |
872 | ||
873 | /* Get here means we have successfully found the possible starting characters | |
874 | of one path of the pattern. We need not follow this path any farther. | |
875 | Instead, look at the next alternative remembered in the stack. */ | |
876 | if (stackp != stackb) | |
877 | p = *stackp--; | |
878 | else | |
879 | break; | |
880 | } | |
881 | } | |
882 | \f | |
883 | /* Like re_search_2, below, but only one string is specified. */ | |
884 | ||
885 | int | |
886 | re_search (pbufp, string, size, startpos, range, regs) | |
887 | struct re_pattern_buffer *pbufp; | |
888 | char *string; | |
889 | int size, startpos, range; | |
890 | struct re_registers *regs; | |
891 | { | |
892 | return re_search_2 (pbufp, 0, 0, string, size, startpos, range, regs, size); | |
893 | } | |
894 | ||
895 | /* Like re_match_2 but tries first a match starting at index STARTPOS, | |
896 | then at STARTPOS + 1, and so on. | |
897 | RANGE is the number of places to try before giving up. | |
898 | If RANGE is negative, the starting positions tried are | |
899 | STARTPOS, STARTPOS - 1, etc. | |
900 | It is up to the caller to make sure that range is not so large | |
901 | as to take the starting position outside of the input strings. | |
902 | ||
903 | The value returned is the position at which the match was found, | |
904 | or -1 if no match was found, | |
905 | or -2 if error (such as failure stack overflow). */ | |
906 | ||
907 | int | |
908 | re_search_2 (pbufp, string1, size1, string2, size2, startpos, range, regs, mstop) | |
909 | struct re_pattern_buffer *pbufp; | |
910 | char *string1, *string2; | |
911 | int size1, size2; | |
912 | int startpos; | |
913 | register int range; | |
914 | struct re_registers *regs; | |
915 | int mstop; | |
916 | { | |
917 | register char *fastmap = pbufp->fastmap; | |
918 | register unsigned char *translate = (unsigned char *) pbufp->translate; | |
919 | int total = size1 + size2; | |
920 | int val; | |
921 | ||
922 | /* Update the fastmap now if not correct already */ | |
923 | if (fastmap && !pbufp->fastmap_accurate) | |
924 | re_compile_fastmap (pbufp); | |
925 | ||
926 | /* Don't waste time in a long search for a pattern | |
927 | that says it is anchored. */ | |
928 | if (pbufp->used > 0 && (enum regexpcode) pbufp->buffer[0] == begbuf | |
929 | && range > 0) | |
930 | { | |
931 | if (startpos > 0) | |
932 | return -1; | |
933 | else | |
934 | range = 1; | |
935 | } | |
936 | ||
937 | while (1) | |
938 | { | |
939 | /* If a fastmap is supplied, skip quickly over characters | |
940 | that cannot possibly be the start of a match. | |
941 | Note, however, that if the pattern can possibly match | |
942 | the null string, we must test it at each starting point | |
943 | so that we take the first null string we get. */ | |
944 | ||
945 | if (fastmap && startpos < total && pbufp->can_be_null != 1) | |
946 | { | |
947 | if (range > 0) | |
948 | { | |
949 | register int lim = 0; | |
950 | register unsigned char *p; | |
951 | int irange = range; | |
952 | if (startpos < size1 && startpos + range >= size1) | |
953 | lim = range - (size1 - startpos); | |
954 | ||
955 | p = ((unsigned char *) | |
956 | &(startpos >= size1 ? string2 - size1 : string1)[startpos]); | |
957 | ||
958 | if (translate) | |
959 | { | |
960 | while (range > lim && !fastmap[translate[*p++]]) | |
961 | range--; | |
962 | } | |
963 | else | |
964 | { | |
965 | while (range > lim && !fastmap[*p++]) | |
966 | range--; | |
967 | } | |
968 | startpos += irange - range; | |
969 | } | |
970 | else | |
971 | { | |
972 | register unsigned char c; | |
973 | if (startpos >= size1) | |
974 | c = string2[startpos - size1]; | |
975 | else | |
976 | c = string1[startpos]; | |
977 | c &= 0xff; | |
978 | if (translate ? !fastmap[translate[c]] : !fastmap[c]) | |
979 | goto advance; | |
980 | } | |
981 | } | |
982 | ||
983 | if (range >= 0 && startpos == total | |
984 | && fastmap && pbufp->can_be_null == 0) | |
985 | return -1; | |
986 | ||
987 | val = re_match_2 (pbufp, string1, size1, string2, size2, startpos, regs, mstop); | |
988 | if (0 <= val) | |
989 | { | |
990 | if (val == -2) | |
991 | return -2; | |
992 | return startpos; | |
993 | } | |
994 | ||
995 | #ifdef C_ALLOCA | |
996 | alloca (0); | |
997 | #endif /* C_ALLOCA */ | |
998 | ||
999 | advance: | |
1000 | if (!range) break; | |
1001 | if (range > 0) range--, startpos++; else range++, startpos--; | |
1002 | } | |
1003 | return -1; | |
1004 | } | |
1005 | \f | |
1006 | #ifndef emacs /* emacs never uses this */ | |
1007 | int | |
1008 | re_match (pbufp, string, size, pos, regs) | |
1009 | struct re_pattern_buffer *pbufp; | |
1010 | char *string; | |
1011 | int size, pos; | |
1012 | struct re_registers *regs; | |
1013 | { | |
1014 | return re_match_2 (pbufp, 0, 0, string, size, pos, regs, size); | |
1015 | } | |
1016 | #endif /* emacs */ | |
1017 | ||
1018 | /* Maximum size of failure stack. Beyond this, overflow is an error. */ | |
1019 | ||
1020 | int re_max_failures = 2000; | |
1021 | ||
9823e3f4 | 1022 | static int memcmp_translate(); |
dd3b648e RP |
1023 | /* Match the pattern described by PBUFP |
1024 | against data which is the virtual concatenation of STRING1 and STRING2. | |
1025 | SIZE1 and SIZE2 are the sizes of the two data strings. | |
1026 | Start the match at position POS. | |
1027 | Do not consider matching past the position MSTOP. | |
1028 | ||
1029 | If pbufp->fastmap is nonzero, then it had better be up to date. | |
1030 | ||
1031 | The reason that the data to match are specified as two components | |
1032 | which are to be regarded as concatenated | |
1033 | is so this function can be used directly on the contents of an Emacs buffer. | |
1034 | ||
1035 | -1 is returned if there is no match. -2 is returned if there is | |
1036 | an error (such as match stack overflow). Otherwise the value is the length | |
1037 | of the substring which was matched. */ | |
1038 | ||
1039 | int | |
1040 | re_match_2 (pbufp, string1, size1, string2, size2, pos, regs, mstop) | |
1041 | struct re_pattern_buffer *pbufp; | |
1042 | unsigned char *string1, *string2; | |
1043 | int size1, size2; | |
1044 | int pos; | |
1045 | struct re_registers *regs; | |
1046 | int mstop; | |
1047 | { | |
1048 | register unsigned char *p = (unsigned char *) pbufp->buffer; | |
1049 | register unsigned char *pend = p + pbufp->used; | |
1050 | /* End of first string */ | |
1051 | unsigned char *end1; | |
1052 | /* End of second string */ | |
1053 | unsigned char *end2; | |
1054 | /* Pointer just past last char to consider matching */ | |
1055 | unsigned char *end_match_1, *end_match_2; | |
1056 | register unsigned char *d, *dend; | |
1057 | register int mcnt; | |
1058 | unsigned char *translate = (unsigned char *) pbufp->translate; | |
1059 | ||
1060 | /* Failure point stack. Each place that can handle a failure further down the line | |
1061 | pushes a failure point on this stack. It consists of two char *'s. | |
1062 | The first one pushed is where to resume scanning the pattern; | |
1063 | the second pushed is where to resume scanning the strings. | |
1064 | If the latter is zero, the failure point is a "dummy". | |
1065 | If a failure happens and the innermost failure point is dormant, | |
1066 | it discards that failure point and tries the next one. */ | |
1067 | ||
1068 | unsigned char *initial_stack[2 * NFAILURES]; | |
1069 | unsigned char **stackb = initial_stack; | |
1070 | unsigned char **stackp = stackb, **stacke = &stackb[2 * NFAILURES]; | |
1071 | ||
1072 | /* Information on the "contents" of registers. | |
1073 | These are pointers into the input strings; they record | |
1074 | just what was matched (on this attempt) by some part of the pattern. | |
1075 | The start_memory command stores the start of a register's contents | |
1076 | and the stop_memory command stores the end. | |
1077 | ||
1078 | At that point, regstart[regnum] points to the first character in the register, | |
1079 | regend[regnum] points to the first character beyond the end of the register, | |
1080 | regstart_seg1[regnum] is true iff regstart[regnum] points into string1, | |
1081 | and regend_seg1[regnum] is true iff regend[regnum] points into string1. */ | |
1082 | ||
1083 | unsigned char *regstart[RE_NREGS]; | |
1084 | unsigned char *regend[RE_NREGS]; | |
1085 | unsigned char regstart_seg1[RE_NREGS], regend_seg1[RE_NREGS]; | |
1086 | ||
1087 | /* Set up pointers to ends of strings. | |
1088 | Don't allow the second string to be empty unless both are empty. */ | |
1089 | if (!size2) | |
1090 | { | |
1091 | string2 = string1; | |
1092 | size2 = size1; | |
1093 | string1 = 0; | |
1094 | size1 = 0; | |
1095 | } | |
1096 | end1 = string1 + size1; | |
1097 | end2 = string2 + size2; | |
1098 | ||
1099 | /* Compute where to stop matching, within the two strings */ | |
1100 | if (mstop <= size1) | |
1101 | { | |
1102 | end_match_1 = string1 + mstop; | |
1103 | end_match_2 = string2; | |
1104 | } | |
1105 | else | |
1106 | { | |
1107 | end_match_1 = end1; | |
1108 | end_match_2 = string2 + mstop - size1; | |
1109 | } | |
1110 | ||
1111 | /* Initialize \) text positions to -1 | |
1112 | to mark ones that no \( or \) has been seen for. */ | |
1113 | ||
1114 | for (mcnt = 0; mcnt < sizeof (regend) / sizeof (*regend); mcnt++) | |
1115 | regend[mcnt] = (unsigned char *) -1; | |
1116 | ||
1117 | /* `p' scans through the pattern as `d' scans through the data. | |
1118 | `dend' is the end of the input string that `d' points within. | |
1119 | `d' is advanced into the following input string whenever necessary, | |
1120 | but this happens before fetching; | |
1121 | therefore, at the beginning of the loop, | |
1122 | `d' can be pointing at the end of a string, | |
1123 | but it cannot equal string2. */ | |
1124 | ||
1125 | if (pos <= size1) | |
1126 | d = string1 + pos, dend = end_match_1; | |
1127 | else | |
1128 | d = string2 + pos - size1, dend = end_match_2; | |
1129 | ||
1130 | /* Write PREFETCH; just before fetching a character with *d. */ | |
1131 | #define PREFETCH \ | |
1132 | while (d == dend) \ | |
1133 | { if (dend == end_match_2) goto fail; /* end of string2 => failure */ \ | |
1134 | d = string2; /* end of string1 => advance to string2. */ \ | |
1135 | dend = end_match_2; } | |
1136 | ||
1137 | /* This loop loops over pattern commands. | |
1138 | It exits by returning from the function if match is complete, | |
1139 | or it drops through if match fails at this starting point in the input data. */ | |
1140 | ||
1141 | while (1) | |
1142 | { | |
1143 | if (p == pend) | |
1144 | /* End of pattern means we have succeeded! */ | |
1145 | { | |
1146 | /* If caller wants register contents data back, convert it to indices */ | |
1147 | if (regs) | |
1148 | { | |
1149 | regs->start[0] = pos; | |
1150 | if (dend == end_match_1) | |
1151 | regs->end[0] = d - string1; | |
1152 | else | |
1153 | regs->end[0] = d - string2 + size1; | |
1154 | for (mcnt = 1; mcnt < RE_NREGS; mcnt++) | |
1155 | { | |
1156 | if (regend[mcnt] == (unsigned char *) -1) | |
1157 | { | |
1158 | regs->start[mcnt] = -1; | |
1159 | regs->end[mcnt] = -1; | |
1160 | continue; | |
1161 | } | |
1162 | if (regstart_seg1[mcnt]) | |
1163 | regs->start[mcnt] = regstart[mcnt] - string1; | |
1164 | else | |
1165 | regs->start[mcnt] = regstart[mcnt] - string2 + size1; | |
1166 | if (regend_seg1[mcnt]) | |
1167 | regs->end[mcnt] = regend[mcnt] - string1; | |
1168 | else | |
1169 | regs->end[mcnt] = regend[mcnt] - string2 + size1; | |
1170 | } | |
1171 | } | |
1172 | if (dend == end_match_1) | |
1173 | return (d - string1 - pos); | |
1174 | else | |
1175 | return d - string2 + size1 - pos; | |
1176 | } | |
1177 | ||
1178 | /* Otherwise match next pattern command */ | |
1179 | #ifdef SWITCH_ENUM_BUG | |
1180 | switch ((int) ((enum regexpcode) *p++)) | |
1181 | #else | |
1182 | switch ((enum regexpcode) *p++) | |
1183 | #endif | |
1184 | { | |
1185 | ||
1186 | /* \( is represented by a start_memory, \) by a stop_memory. | |
1187 | Both of those commands contain a "register number" argument. | |
1188 | The text matched within the \( and \) is recorded under that number. | |
1189 | Then, \<digit> turns into a `duplicate' command which | |
1190 | is followed by the numeric value of <digit> as the register number. */ | |
1191 | ||
1192 | case start_memory: | |
1193 | regstart[*p] = d; | |
1194 | regstart_seg1[*p++] = (dend == end_match_1); | |
1195 | break; | |
1196 | ||
1197 | case stop_memory: | |
1198 | regend[*p] = d; | |
1199 | regend_seg1[*p++] = (dend == end_match_1); | |
1200 | break; | |
1201 | ||
1202 | case duplicate: | |
1203 | { | |
1204 | int regno = *p++; /* Get which register to match against */ | |
1205 | register unsigned char *d2, *dend2; | |
1206 | ||
1207 | d2 = regstart[regno]; | |
1208 | dend2 = ((regstart_seg1[regno] == regend_seg1[regno]) | |
1209 | ? regend[regno] : end_match_1); | |
1210 | while (1) | |
1211 | { | |
1212 | /* Advance to next segment in register contents, if necessary */ | |
1213 | while (d2 == dend2) | |
1214 | { | |
1215 | if (dend2 == end_match_2) break; | |
1216 | if (dend2 == regend[regno]) break; | |
1217 | d2 = string2, dend2 = regend[regno]; /* end of string1 => advance to string2. */ | |
1218 | } | |
1219 | /* At end of register contents => success */ | |
1220 | if (d2 == dend2) break; | |
1221 | ||
1222 | /* Advance to next segment in data being matched, if necessary */ | |
1223 | PREFETCH; | |
1224 | ||
1225 | /* mcnt gets # consecutive chars to compare */ | |
1226 | mcnt = dend - d; | |
1227 | if (mcnt > dend2 - d2) | |
1228 | mcnt = dend2 - d2; | |
1229 | /* Compare that many; failure if mismatch, else skip them. */ | |
9823e3f4 | 1230 | if (translate ? memcmp_translate (d, d2, mcnt, translate) : memcmp (d, d2, mcnt)) |
dd3b648e RP |
1231 | goto fail; |
1232 | d += mcnt, d2 += mcnt; | |
1233 | } | |
1234 | } | |
1235 | break; | |
1236 | ||
1237 | case anychar: | |
1238 | /* fetch a data character */ | |
1239 | PREFETCH; | |
1240 | /* Match anything but a newline. */ | |
1241 | if ((translate ? translate[*d++] : *d++) == '\n') | |
1242 | goto fail; | |
1243 | break; | |
1244 | ||
1245 | case charset: | |
1246 | case charset_not: | |
1247 | { | |
1248 | /* Nonzero for charset_not */ | |
1249 | int not = 0; | |
1250 | register int c; | |
1251 | if (*(p - 1) == (unsigned char) charset_not) | |
1252 | not = 1; | |
1253 | ||
1254 | /* fetch a data character */ | |
1255 | PREFETCH; | |
1256 | ||
1257 | if (translate) | |
1258 | c = translate [*d]; | |
1259 | else | |
1260 | c = *d; | |
1261 | ||
1262 | if (c < *p * BYTEWIDTH | |
1263 | && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) | |
1264 | not = !not; | |
1265 | ||
1266 | p += 1 + *p; | |
1267 | ||
1268 | if (!not) goto fail; | |
1269 | d++; | |
1270 | break; | |
1271 | } | |
1272 | ||
1273 | case begline: | |
1274 | if (d == string1 || d[-1] == '\n') | |
1275 | break; | |
1276 | goto fail; | |
1277 | ||
1278 | case endline: | |
1279 | if (d == end2 | |
1280 | || (d == end1 ? (size2 == 0 || *string2 == '\n') : *d == '\n')) | |
1281 | break; | |
1282 | goto fail; | |
1283 | ||
1284 | /* "or" constructs ("|") are handled by starting each alternative | |
1285 | with an on_failure_jump that points to the start of the next alternative. | |
1286 | Each alternative except the last ends with a jump to the joining point. | |
1287 | (Actually, each jump except for the last one really jumps | |
1288 | to the following jump, because tensioning the jumps is a hassle.) */ | |
1289 | ||
1290 | /* The start of a stupid repeat has an on_failure_jump that points | |
1291 | past the end of the repeat text. | |
1292 | This makes a failure point so that, on failure to match a repetition, | |
1293 | matching restarts past as many repetitions have been found | |
1294 | with no way to fail and look for another one. */ | |
1295 | ||
1296 | /* A smart repeat is similar but loops back to the on_failure_jump | |
1297 | so that each repetition makes another failure point. */ | |
1298 | ||
1299 | case on_failure_jump: | |
1300 | if (stackp == stacke) | |
1301 | { | |
1302 | unsigned char **stackx; | |
1303 | if (stacke - stackb > re_max_failures * 2) | |
1304 | return -2; | |
1305 | stackx = (unsigned char **) alloca (2 * (stacke - stackb) | |
1306 | * sizeof (char *)); | |
ade40d31 | 1307 | memcpy (stackx, stackb, (stacke - stackb) * sizeof (char *)); |
dd3b648e RP |
1308 | stackp = stackx + (stackp - stackb); |
1309 | stacke = stackx + 2 * (stacke - stackb); | |
1310 | stackb = stackx; | |
1311 | } | |
1312 | mcnt = *p++ & 0377; | |
1313 | mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
1314 | p++; | |
1315 | *stackp++ = mcnt + p; | |
1316 | *stackp++ = d; | |
1317 | break; | |
1318 | ||
1319 | /* The end of a smart repeat has an maybe_finalize_jump back. | |
1320 | Change it either to a finalize_jump or an ordinary jump. */ | |
1321 | ||
1322 | case maybe_finalize_jump: | |
1323 | mcnt = *p++ & 0377; | |
1324 | mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
1325 | p++; | |
1326 | { | |
1327 | register unsigned char *p2 = p; | |
1328 | /* Compare what follows with the begining of the repeat. | |
1329 | If we can establish that there is nothing that they would | |
1330 | both match, we can change to finalize_jump */ | |
1331 | while (p2 != pend | |
1332 | && (*p2 == (unsigned char) stop_memory | |
1333 | || *p2 == (unsigned char) start_memory)) | |
1334 | p2++; | |
1335 | if (p2 == pend) | |
1336 | p[-3] = (unsigned char) finalize_jump; | |
1337 | else if (*p2 == (unsigned char) exactn | |
1338 | || *p2 == (unsigned char) endline) | |
1339 | { | |
1340 | register int c = *p2 == (unsigned char) endline ? '\n' : p2[2]; | |
1341 | register unsigned char *p1 = p + mcnt; | |
1342 | /* p1[0] ... p1[2] are an on_failure_jump. | |
1343 | Examine what follows that */ | |
1344 | if (p1[3] == (unsigned char) exactn && p1[5] != c) | |
1345 | p[-3] = (unsigned char) finalize_jump; | |
1346 | else if (p1[3] == (unsigned char) charset | |
1347 | || p1[3] == (unsigned char) charset_not) | |
1348 | { | |
1349 | int not = p1[3] == (unsigned char) charset_not; | |
1350 | if (c < p1[4] * BYTEWIDTH | |
1351 | && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) | |
1352 | not = !not; | |
1353 | /* not is 1 if c would match */ | |
1354 | /* That means it is not safe to finalize */ | |
1355 | if (!not) | |
1356 | p[-3] = (unsigned char) finalize_jump; | |
1357 | } | |
1358 | } | |
1359 | } | |
1360 | p -= 2; | |
1361 | if (p[-1] != (unsigned char) finalize_jump) | |
1362 | { | |
1363 | p[-1] = (unsigned char) jump; | |
1364 | goto nofinalize; | |
1365 | } | |
1366 | ||
1367 | /* The end of a stupid repeat has a finalize-jump | |
1368 | back to the start, where another failure point will be made | |
1369 | which will point after all the repetitions found so far. */ | |
1370 | ||
1371 | case finalize_jump: | |
1372 | stackp -= 2; | |
1373 | ||
1374 | case jump: | |
1375 | nofinalize: | |
1376 | mcnt = *p++ & 0377; | |
1377 | mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
1378 | p += mcnt + 1; /* The 1 compensates for missing ++ above */ | |
1379 | break; | |
1380 | ||
1381 | case dummy_failure_jump: | |
1382 | if (stackp == stacke) | |
1383 | { | |
1384 | unsigned char **stackx | |
1385 | = (unsigned char **) alloca (2 * (stacke - stackb) | |
1386 | * sizeof (char *)); | |
ade40d31 | 1387 | memcpy (stackx, stackb, (stacke - stackb) * sizeof (char *)); |
dd3b648e RP |
1388 | stackp = stackx + (stackp - stackb); |
1389 | stacke = stackx + 2 * (stacke - stackb); | |
1390 | stackb = stackx; | |
1391 | } | |
1392 | *stackp++ = 0; | |
1393 | *stackp++ = 0; | |
1394 | goto nofinalize; | |
1395 | ||
1396 | case wordbound: | |
1397 | if (d == string1 /* Points to first char */ | |
1398 | || d == end2 /* Points to end */ | |
1399 | || (d == end1 && size2 == 0)) /* Points to end */ | |
1400 | break; | |
1401 | if ((SYNTAX (d[-1]) == Sword) | |
1402 | != (SYNTAX (d == end1 ? *string2 : *d) == Sword)) | |
1403 | break; | |
1404 | goto fail; | |
1405 | ||
1406 | case notwordbound: | |
1407 | if (d == string1 /* Points to first char */ | |
1408 | || d == end2 /* Points to end */ | |
1409 | || (d == end1 && size2 == 0)) /* Points to end */ | |
1410 | goto fail; | |
1411 | if ((SYNTAX (d[-1]) == Sword) | |
1412 | != (SYNTAX (d == end1 ? *string2 : *d) == Sword)) | |
1413 | goto fail; | |
1414 | break; | |
1415 | ||
1416 | case wordbeg: | |
1417 | if (d == end2 /* Points to end */ | |
1418 | || (d == end1 && size2 == 0) /* Points to end */ | |
1419 | || SYNTAX (* (d == end1 ? string2 : d)) != Sword) /* Next char not a letter */ | |
1420 | goto fail; | |
1421 | if (d == string1 /* Points to first char */ | |
1422 | || SYNTAX (d[-1]) != Sword) /* prev char not letter */ | |
1423 | break; | |
1424 | goto fail; | |
1425 | ||
1426 | case wordend: | |
1427 | if (d == string1 /* Points to first char */ | |
1428 | || SYNTAX (d[-1]) != Sword) /* prev char not letter */ | |
1429 | goto fail; | |
1430 | if (d == end2 /* Points to end */ | |
1431 | || (d == end1 && size2 == 0) /* Points to end */ | |
1432 | || SYNTAX (d == end1 ? *string2 : *d) != Sword) /* Next char not a letter */ | |
1433 | break; | |
1434 | goto fail; | |
1435 | ||
1436 | #ifdef emacs | |
1437 | case before_dot: | |
1438 | if (((d - string2 <= (unsigned) size2) | |
1439 | ? d - bf_p2 : d - bf_p1) | |
1440 | <= point) | |
1441 | goto fail; | |
1442 | break; | |
1443 | ||
1444 | case at_dot: | |
1445 | if (((d - string2 <= (unsigned) size2) | |
1446 | ? d - bf_p2 : d - bf_p1) | |
1447 | == point) | |
1448 | goto fail; | |
1449 | break; | |
1450 | ||
1451 | case after_dot: | |
1452 | if (((d - string2 <= (unsigned) size2) | |
1453 | ? d - bf_p2 : d - bf_p1) | |
1454 | >= point) | |
1455 | goto fail; | |
1456 | break; | |
1457 | ||
1458 | case wordchar: | |
1459 | mcnt = (int) Sword; | |
1460 | goto matchsyntax; | |
1461 | ||
1462 | case syntaxspec: | |
1463 | mcnt = *p++; | |
1464 | matchsyntax: | |
1465 | PREFETCH; | |
1466 | if (SYNTAX (*d++) != (enum syntaxcode) mcnt) goto fail; | |
1467 | break; | |
1468 | ||
1469 | case notwordchar: | |
1470 | mcnt = (int) Sword; | |
1471 | goto matchnotsyntax; | |
1472 | ||
1473 | case notsyntaxspec: | |
1474 | mcnt = *p++; | |
1475 | matchnotsyntax: | |
1476 | PREFETCH; | |
1477 | if (SYNTAX (*d++) == (enum syntaxcode) mcnt) goto fail; | |
1478 | break; | |
1479 | #else | |
1480 | case wordchar: | |
1481 | PREFETCH; | |
1482 | if (SYNTAX (*d++) == 0) goto fail; | |
1483 | break; | |
1484 | ||
1485 | case notwordchar: | |
1486 | PREFETCH; | |
1487 | if (SYNTAX (*d++) != 0) goto fail; | |
1488 | break; | |
1489 | #endif /* not emacs */ | |
1490 | ||
1491 | case begbuf: | |
1492 | if (d == string1) /* Note, d cannot equal string2 */ | |
1493 | break; /* unless string1 == string2. */ | |
1494 | goto fail; | |
1495 | ||
1496 | case endbuf: | |
1497 | if (d == end2 || (d == end1 && size2 == 0)) | |
1498 | break; | |
1499 | goto fail; | |
1500 | ||
1501 | case exactn: | |
1502 | /* Match the next few pattern characters exactly. | |
1503 | mcnt is how many characters to match. */ | |
1504 | mcnt = *p++; | |
1505 | if (translate) | |
1506 | { | |
1507 | do | |
1508 | { | |
1509 | PREFETCH; | |
1510 | if (translate[*d++] != *p++) goto fail; | |
1511 | } | |
1512 | while (--mcnt); | |
1513 | } | |
1514 | else | |
1515 | { | |
1516 | do | |
1517 | { | |
1518 | PREFETCH; | |
1519 | if (*d++ != *p++) goto fail; | |
1520 | } | |
1521 | while (--mcnt); | |
1522 | } | |
1523 | break; | |
1524 | } | |
1525 | continue; /* Successfully matched one pattern command; keep matching */ | |
1526 | ||
1527 | /* Jump here if any matching operation fails. */ | |
1528 | fail: | |
1529 | if (stackp != stackb) | |
1530 | /* A restart point is known. Restart there and pop it. */ | |
1531 | { | |
1532 | if (!stackp[-2]) | |
1533 | { /* If innermost failure point is dormant, flush it and keep looking */ | |
1534 | stackp -= 2; | |
1535 | goto fail; | |
1536 | } | |
1537 | d = *--stackp; | |
1538 | p = *--stackp; | |
1539 | if (d >= string1 && d <= end1) | |
1540 | dend = end_match_1; | |
1541 | } | |
1542 | else break; /* Matching at this starting point really fails! */ | |
1543 | } | |
1544 | return -1; /* Failure to match */ | |
1545 | } | |
1546 | ||
1547 | static int | |
9823e3f4 | 1548 | memcmp_translate (s1, s2, len, translate) |
dd3b648e RP |
1549 | unsigned char *s1, *s2; |
1550 | register int len; | |
1551 | unsigned char *translate; | |
1552 | { | |
1553 | register unsigned char *p1 = s1, *p2 = s2; | |
1554 | while (len) | |
1555 | { | |
1556 | if (translate [*p1++] != translate [*p2++]) return 1; | |
1557 | len--; | |
1558 | } | |
1559 | return 0; | |
1560 | } | |
1561 | \f | |
1562 | /* Entry points compatible with bsd4.2 regex library */ | |
1563 | ||
1564 | #ifndef emacs | |
1565 | ||
1566 | static struct re_pattern_buffer re_comp_buf; | |
1567 | ||
1568 | char * | |
1569 | re_comp (s) | |
ba47c66a | 1570 | const char *s; |
dd3b648e RP |
1571 | { |
1572 | if (!s) | |
1573 | { | |
1574 | if (!re_comp_buf.buffer) | |
1575 | return "No previous regular expression"; | |
1576 | return 0; | |
1577 | } | |
1578 | ||
1579 | if (!re_comp_buf.buffer) | |
1580 | { | |
1581 | if (!(re_comp_buf.buffer = (char *) malloc (200))) | |
1582 | return "Memory exhausted"; | |
1583 | re_comp_buf.allocated = 200; | |
1584 | if (!(re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH))) | |
1585 | return "Memory exhausted"; | |
1586 | } | |
1587 | return re_compile_pattern (s, strlen (s), &re_comp_buf); | |
1588 | } | |
1589 | ||
1590 | int | |
1591 | re_exec (s) | |
1592 | char *s; | |
1593 | { | |
1594 | int len = strlen (s); | |
1595 | return 0 <= re_search (&re_comp_buf, s, len, 0, len, 0); | |
1596 | } | |
1597 | ||
1598 | #endif /* emacs */ | |
1599 | \f | |
1600 | #ifdef test | |
1601 | ||
1602 | #include <stdio.h> | |
1603 | ||
1604 | /* Indexed by a character, gives the upper case equivalent of the character */ | |
1605 | ||
1606 | static char upcase[0400] = | |
1607 | { 000, 001, 002, 003, 004, 005, 006, 007, | |
1608 | 010, 011, 012, 013, 014, 015, 016, 017, | |
1609 | 020, 021, 022, 023, 024, 025, 026, 027, | |
1610 | 030, 031, 032, 033, 034, 035, 036, 037, | |
1611 | 040, 041, 042, 043, 044, 045, 046, 047, | |
1612 | 050, 051, 052, 053, 054, 055, 056, 057, | |
1613 | 060, 061, 062, 063, 064, 065, 066, 067, | |
1614 | 070, 071, 072, 073, 074, 075, 076, 077, | |
1615 | 0100, 0101, 0102, 0103, 0104, 0105, 0106, 0107, | |
1616 | 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117, | |
1617 | 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127, | |
1618 | 0130, 0131, 0132, 0133, 0134, 0135, 0136, 0137, | |
1619 | 0140, 0101, 0102, 0103, 0104, 0105, 0106, 0107, | |
1620 | 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117, | |
1621 | 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127, | |
1622 | 0130, 0131, 0132, 0173, 0174, 0175, 0176, 0177, | |
1623 | 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207, | |
1624 | 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217, | |
1625 | 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227, | |
1626 | 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237, | |
1627 | 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247, | |
1628 | 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257, | |
1629 | 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267, | |
1630 | 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277, | |
1631 | 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307, | |
1632 | 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317, | |
1633 | 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0327, | |
1634 | 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0337, | |
1635 | 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347, | |
1636 | 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357, | |
1637 | 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0367, | |
1638 | 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0377 | |
1639 | }; | |
1640 | ||
1641 | main (argc, argv) | |
1642 | int argc; | |
1643 | char **argv; | |
1644 | { | |
1645 | char pat[80]; | |
1646 | struct re_pattern_buffer buf; | |
1647 | int i; | |
1648 | char c; | |
1649 | char fastmap[(1 << BYTEWIDTH)]; | |
1650 | ||
1651 | /* Allow a command argument to specify the style of syntax. */ | |
1652 | if (argc > 1) | |
1653 | obscure_syntax = atoi (argv[1]); | |
1654 | ||
1655 | buf.allocated = 40; | |
1656 | buf.buffer = (char *) malloc (buf.allocated); | |
1657 | buf.fastmap = fastmap; | |
1658 | buf.translate = upcase; | |
1659 | ||
1660 | while (1) | |
1661 | { | |
1662 | gets (pat); | |
1663 | ||
1664 | if (*pat) | |
1665 | { | |
1666 | re_compile_pattern (pat, strlen(pat), &buf); | |
1667 | ||
1668 | for (i = 0; i < buf.used; i++) | |
1669 | printchar (buf.buffer[i]); | |
1670 | ||
199b2450 | 1671 | putchar_unfiltered ('\n'); |
dd3b648e | 1672 | |
199b2450 | 1673 | printf_unfiltered ("%d allocated, %d used.\n", buf.allocated, buf.used); |
dd3b648e RP |
1674 | |
1675 | re_compile_fastmap (&buf); | |
199b2450 | 1676 | printf_unfiltered ("Allowed by fastmap: "); |
dd3b648e RP |
1677 | for (i = 0; i < (1 << BYTEWIDTH); i++) |
1678 | if (fastmap[i]) printchar (i); | |
199b2450 | 1679 | putchar_unfiltered ('\n'); |
dd3b648e RP |
1680 | } |
1681 | ||
1682 | gets (pat); /* Now read the string to match against */ | |
1683 | ||
1684 | i = re_match (&buf, pat, strlen (pat), 0, 0); | |
199b2450 | 1685 | printf_unfiltered ("Match value %d.\n", i); |
dd3b648e RP |
1686 | } |
1687 | } | |
1688 | ||
1689 | #ifdef NOTDEF | |
1690 | print_buf (bufp) | |
1691 | struct re_pattern_buffer *bufp; | |
1692 | { | |
1693 | int i; | |
1694 | ||
199b2450 | 1695 | printf_unfiltered ("buf is :\n----------------\n"); |
dd3b648e RP |
1696 | for (i = 0; i < bufp->used; i++) |
1697 | printchar (bufp->buffer[i]); | |
1698 | ||
199b2450 | 1699 | printf_unfiltered ("\n%d allocated, %d used.\n", bufp->allocated, bufp->used); |
dd3b648e | 1700 | |
199b2450 | 1701 | printf_unfiltered ("Allowed by fastmap: "); |
dd3b648e RP |
1702 | for (i = 0; i < (1 << BYTEWIDTH); i++) |
1703 | if (bufp->fastmap[i]) | |
1704 | printchar (i); | |
199b2450 | 1705 | printf_unfiltered ("\nAllowed by translate: "); |
dd3b648e RP |
1706 | if (bufp->translate) |
1707 | for (i = 0; i < (1 << BYTEWIDTH); i++) | |
1708 | if (bufp->translate[i]) | |
1709 | printchar (i); | |
199b2450 TL |
1710 | printf_unfiltered ("\nfastmap is%s accurate\n", bufp->fastmap_accurate ? "" : "n't"); |
1711 | printf_unfiltered ("can %s be null\n----------", bufp->can_be_null ? "" : "not"); | |
dd3b648e RP |
1712 | } |
1713 | #endif | |
1714 | ||
1715 | printchar (c) | |
1716 | char c; | |
1717 | { | |
1718 | if (c < 041 || c >= 0177) | |
1719 | { | |
199b2450 TL |
1720 | putchar_unfiltered ('\\'); |
1721 | putchar_unfiltered (((c >> 6) & 3) + '0'); | |
1722 | putchar_unfiltered (((c >> 3) & 7) + '0'); | |
1723 | putchar_unfiltered ((c & 7) + '0'); | |
dd3b648e RP |
1724 | } |
1725 | else | |
199b2450 | 1726 | putchar_unfiltered (c); |
dd3b648e RP |
1727 | } |
1728 | ||
1729 | error (string) | |
1730 | char *string; | |
1731 | { | |
199b2450 | 1732 | puts_unfiltered (string); |
dd3b648e RP |
1733 | exit (1); |
1734 | } | |
1735 | ||
1736 | #endif /* test */ |