]>
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 | |
6c9638b4 | 16 | Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, 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 | 35 | #include "defs.h" |
2b576293 | 36 | #include "gdb_string.h" |
6405302d FF |
37 | #undef malloc |
38 | #define malloc xmalloc | |
dd3b648e RP |
39 | |
40 | /* | |
41 | * Define the syntax stuff, so we can do the \<...\> things. | |
42 | */ | |
43 | ||
44 | #ifndef Sword /* must be non-zero in some of the tests below... */ | |
45 | #define Sword 1 | |
46 | #endif | |
47 | ||
48 | #define SYNTAX(c) re_syntax_table[c] | |
49 | ||
50 | #ifdef SYNTAX_TABLE | |
51 | ||
52 | char *re_syntax_table; | |
53 | ||
54 | #else | |
55 | ||
56 | static char re_syntax_table[256]; | |
57 | ||
58 | static void | |
59 | init_syntax_once () | |
60 | { | |
61 | register int c; | |
62 | static int done = 0; | |
63 | ||
64 | if (done) | |
65 | return; | |
66 | ||
4ed97c9a | 67 | memset (re_syntax_table, '\0', sizeof re_syntax_table); |
dd3b648e RP |
68 | |
69 | for (c = 'a'; c <= 'z'; c++) | |
70 | re_syntax_table[c] = Sword; | |
71 | ||
72 | for (c = 'A'; c <= 'Z'; c++) | |
73 | re_syntax_table[c] = Sword; | |
74 | ||
75 | for (c = '0'; c <= '9'; c++) | |
76 | re_syntax_table[c] = Sword; | |
77 | ||
78 | done = 1; | |
79 | } | |
80 | ||
81 | #endif /* SYNTAX_TABLE */ | |
82 | #endif /* not emacs */ | |
83 | ||
811f1bdc | 84 | #include "gnu-regex.h" |
dd3b648e RP |
85 | |
86 | /* Number of failure points to allocate space for initially, | |
87 | when matching. If this number is exceeded, more space is allocated, | |
88 | so it is not a hard limit. */ | |
89 | ||
90 | #ifndef NFAILURES | |
91 | #define NFAILURES 80 | |
92 | #endif /* NFAILURES */ | |
93 | ||
94 | /* width of a byte in bits */ | |
95 | ||
96 | #define BYTEWIDTH 8 | |
97 | ||
0d98155c PS |
98 | /* We remove any previous definition of `SIGN_EXTEND_CHAR', |
99 | since ours (we hope) works properly with all combinations of | |
100 | machines, compilers, `char' and `unsigned char' argument types. | |
101 | (Per Bothner suggested the basic approach.) */ | |
102 | #undef SIGN_EXTEND_CHAR | |
103 | #if __STDC__ | |
104 | #define SIGN_EXTEND_CHAR(c) ((signed char) (c)) | |
105 | #else /* not __STDC__ */ | |
106 | /* As in Harbison and Steele. */ | |
107 | #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128) | |
dd3b648e RP |
108 | #endif |
109 | \f | |
110 | static int obscure_syntax = 0; | |
111 | ||
112 | /* Specify the precise syntax of regexp for compilation. | |
113 | This provides for compatibility for various utilities | |
114 | which historically have different, incompatible syntaxes. | |
115 | ||
116 | The argument SYNTAX is a bit-mask containing the two bits | |
117 | RE_NO_BK_PARENS and RE_NO_BK_VBAR. */ | |
118 | ||
119 | int | |
120 | re_set_syntax (syntax) | |
51b57ded | 121 | int syntax; |
dd3b648e RP |
122 | { |
123 | int ret; | |
124 | ||
125 | ret = obscure_syntax; | |
126 | obscure_syntax = syntax; | |
127 | return ret; | |
128 | } | |
129 | \f | |
130 | /* re_compile_pattern takes a regular-expression string | |
131 | and converts it into a buffer full of byte commands for matching. | |
132 | ||
133 | PATTERN is the address of the pattern string | |
134 | SIZE is the length of it. | |
135 | BUFP is a struct re_pattern_buffer * which points to the info | |
136 | on where to store the byte commands. | |
137 | This structure contains a char * which points to the | |
138 | actual space, which should have been obtained with malloc. | |
139 | re_compile_pattern may use realloc to grow the buffer space. | |
140 | ||
141 | The number of bytes of commands can be found out by looking in | |
142 | the struct re_pattern_buffer that bufp pointed to, | |
143 | after re_compile_pattern returns. | |
144 | */ | |
145 | ||
146 | #define PATPUSH(ch) (*b++ = (char) (ch)) | |
147 | ||
148 | #define PATFETCH(c) \ | |
149 | {if (p == pend) goto end_of_pattern; \ | |
150 | c = * (unsigned char *) p++; \ | |
151 | if (translate) c = translate[c]; } | |
152 | ||
153 | #define PATFETCH_RAW(c) \ | |
154 | {if (p == pend) goto end_of_pattern; \ | |
155 | c = * (unsigned char *) p++; } | |
156 | ||
157 | #define PATUNFETCH p-- | |
158 | ||
ee6d646a JK |
159 | /* This is not an arbitrary limit: the arguments which represent offsets |
160 | into the pattern are two bytes long. So if 2^16 bytes turns out to | |
161 | be too small, many things would have to change. */ | |
162 | #define MAX_BUF_SIZE (1 << 16) | |
163 | ||
164 | ||
165 | /* Extend the buffer by twice its current size via realloc and | |
166 | reset the pointers that pointed into the old block to point to the | |
167 | correct places in the new one. If extending the buffer results in it | |
168 | being larger than MAX_BUF_SIZE, then flag memory exhausted. */ | |
169 | #define EXTEND_BUFFER \ | |
170 | do { \ | |
171 | char *old_buffer = bufp->buffer; \ | |
172 | if (bufp->allocated == MAX_BUF_SIZE) \ | |
173 | goto too_big; \ | |
174 | bufp->allocated <<= 1; \ | |
175 | if (bufp->allocated > MAX_BUF_SIZE) \ | |
176 | bufp->allocated = MAX_BUF_SIZE; \ | |
177 | bufp->buffer = (char *) realloc (bufp->buffer, bufp->allocated);\ | |
178 | if (bufp->buffer == NULL) \ | |
179 | goto memory_exhausted; \ | |
180 | /* If the buffer moved, move all the pointers into it. */ \ | |
181 | if (old_buffer != bufp->buffer) \ | |
182 | { \ | |
183 | b = (b - old_buffer) + bufp->buffer; \ | |
184 | begalt = (begalt - old_buffer) + bufp->buffer; \ | |
185 | if (fixup_jump) \ | |
186 | fixup_jump = (fixup_jump - old_buffer) + bufp->buffer;\ | |
187 | if (laststart) \ | |
188 | laststart = (laststart - old_buffer) + bufp->buffer; \ | |
189 | if (pending_exact) \ | |
190 | pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ | |
191 | } \ | |
192 | } while (0) | |
dd3b648e | 193 | |
50e0dc41 | 194 | static void store_jump (), insert_jump (); |
dd3b648e RP |
195 | |
196 | char * | |
197 | re_compile_pattern (pattern, size, bufp) | |
198 | char *pattern; | |
199 | int size; | |
200 | struct re_pattern_buffer *bufp; | |
201 | { | |
202 | register char *b = bufp->buffer; | |
203 | register char *p = pattern; | |
204 | char *pend = pattern + size; | |
205 | register unsigned c, c1; | |
206 | char *p1; | |
207 | unsigned char *translate = (unsigned char *) bufp->translate; | |
208 | ||
209 | /* address of the count-byte of the most recently inserted "exactn" command. | |
210 | This makes it possible to tell whether a new exact-match character | |
211 | can be added to that command or requires a new "exactn" command. */ | |
212 | ||
213 | char *pending_exact = 0; | |
214 | ||
215 | /* address of the place where a forward-jump should go | |
216 | to the end of the containing expression. | |
217 | Each alternative of an "or", except the last, ends with a forward-jump | |
218 | of this sort. */ | |
219 | ||
220 | char *fixup_jump = 0; | |
221 | ||
222 | /* address of start of the most recently finished expression. | |
223 | This tells postfix * where to find the start of its operand. */ | |
224 | ||
225 | char *laststart = 0; | |
226 | ||
227 | /* In processing a repeat, 1 means zero matches is allowed */ | |
228 | ||
229 | char zero_times_ok; | |
230 | ||
231 | /* In processing a repeat, 1 means many matches is allowed */ | |
232 | ||
233 | char many_times_ok; | |
234 | ||
235 | /* address of beginning of regexp, or inside of last \( */ | |
236 | ||
237 | char *begalt = b; | |
238 | ||
239 | /* Stack of information saved by \( and restored by \). | |
240 | Four stack elements are pushed by each \(: | |
241 | First, the value of b. | |
242 | Second, the value of fixup_jump. | |
243 | Third, the value of regnum. | |
244 | Fourth, the value of begalt. */ | |
245 | ||
246 | int stackb[40]; | |
247 | int *stackp = stackb; | |
248 | int *stacke = stackb + 40; | |
249 | int *stackt; | |
250 | ||
251 | /* Counts \('s as they are encountered. Remembered for the matching \), | |
252 | where it becomes the "register number" to put in the stop_memory command */ | |
253 | ||
254 | int regnum = 1; | |
255 | ||
256 | bufp->fastmap_accurate = 0; | |
257 | ||
258 | #ifndef emacs | |
259 | #ifndef SYNTAX_TABLE | |
260 | /* | |
261 | * Initialize the syntax table. | |
262 | */ | |
263 | init_syntax_once(); | |
264 | #endif | |
265 | #endif | |
266 | ||
267 | if (bufp->allocated == 0) | |
268 | { | |
269 | bufp->allocated = 28; | |
270 | if (bufp->buffer) | |
271 | /* EXTEND_BUFFER loses when bufp->allocated is 0 */ | |
272 | bufp->buffer = (char *) realloc (bufp->buffer, 28); | |
273 | else | |
274 | /* Caller did not allocate a buffer. Do it for him */ | |
275 | bufp->buffer = (char *) malloc (28); | |
276 | if (!bufp->buffer) goto memory_exhausted; | |
277 | begalt = b = bufp->buffer; | |
278 | } | |
279 | ||
280 | while (p != pend) | |
281 | { | |
282 | if (b - bufp->buffer > bufp->allocated - 10) | |
283 | /* Note that EXTEND_BUFFER clobbers c */ | |
284 | EXTEND_BUFFER; | |
285 | ||
286 | PATFETCH (c); | |
287 | ||
288 | switch (c) | |
289 | { | |
290 | case '$': | |
291 | if (obscure_syntax & RE_TIGHT_VBAR) | |
292 | { | |
293 | if (! (obscure_syntax & RE_CONTEXT_INDEP_OPS) && p != pend) | |
294 | goto normal_char; | |
295 | /* Make operand of last vbar end before this `$'. */ | |
296 | if (fixup_jump) | |
297 | store_jump (fixup_jump, jump, b); | |
298 | fixup_jump = 0; | |
299 | PATPUSH (endline); | |
300 | break; | |
301 | } | |
302 | ||
303 | /* $ means succeed if at end of line, but only in special contexts. | |
304 | If randomly in the middle of a pattern, it is a normal character. */ | |
305 | if (p == pend || *p == '\n' | |
306 | || (obscure_syntax & RE_CONTEXT_INDEP_OPS) | |
307 | || (obscure_syntax & RE_NO_BK_PARENS | |
308 | ? *p == ')' | |
309 | : *p == '\\' && p[1] == ')') | |
310 | || (obscure_syntax & RE_NO_BK_VBAR | |
311 | ? *p == '|' | |
312 | : *p == '\\' && p[1] == '|')) | |
313 | { | |
314 | PATPUSH (endline); | |
315 | break; | |
316 | } | |
317 | goto normal_char; | |
318 | ||
319 | case '^': | |
320 | /* ^ means succeed if at beg of line, but only if no preceding pattern. */ | |
321 | ||
322 | if (laststart && p[-2] != '\n' | |
323 | && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) | |
324 | goto normal_char; | |
325 | if (obscure_syntax & RE_TIGHT_VBAR) | |
326 | { | |
327 | if (p != pattern + 1 | |
328 | && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) | |
329 | goto normal_char; | |
330 | PATPUSH (begline); | |
331 | begalt = b; | |
332 | } | |
333 | else | |
334 | PATPUSH (begline); | |
335 | break; | |
336 | ||
337 | case '+': | |
338 | case '?': | |
339 | if (obscure_syntax & RE_BK_PLUS_QM) | |
340 | goto normal_char; | |
341 | handle_plus: | |
342 | case '*': | |
343 | /* If there is no previous pattern, char not special. */ | |
344 | if (!laststart && ! (obscure_syntax & RE_CONTEXT_INDEP_OPS)) | |
345 | goto normal_char; | |
346 | /* If there is a sequence of repetition chars, | |
347 | collapse it down to equivalent to just one. */ | |
348 | zero_times_ok = 0; | |
349 | many_times_ok = 0; | |
350 | while (1) | |
351 | { | |
352 | zero_times_ok |= c != '+'; | |
353 | many_times_ok |= c != '?'; | |
354 | if (p == pend) | |
355 | break; | |
356 | PATFETCH (c); | |
357 | if (c == '*') | |
358 | ; | |
359 | else if (!(obscure_syntax & RE_BK_PLUS_QM) | |
360 | && (c == '+' || c == '?')) | |
361 | ; | |
362 | else if ((obscure_syntax & RE_BK_PLUS_QM) | |
363 | && c == '\\') | |
364 | { | |
365 | int c1; | |
366 | PATFETCH (c1); | |
367 | if (!(c1 == '+' || c1 == '?')) | |
368 | { | |
369 | PATUNFETCH; | |
370 | PATUNFETCH; | |
371 | break; | |
372 | } | |
373 | c = c1; | |
374 | } | |
375 | else | |
376 | { | |
377 | PATUNFETCH; | |
378 | break; | |
379 | } | |
380 | } | |
381 | ||
382 | /* Star, etc. applied to an empty pattern is equivalent | |
383 | to an empty pattern. */ | |
384 | if (!laststart) | |
385 | break; | |
386 | ||
387 | /* Now we know whether 0 matches is allowed, | |
388 | and whether 2 or more matches is allowed. */ | |
389 | if (many_times_ok) | |
390 | { | |
391 | /* If more than one repetition is allowed, | |
392 | put in a backward jump at the end. */ | |
393 | store_jump (b, maybe_finalize_jump, laststart - 3); | |
394 | b += 3; | |
395 | } | |
396 | insert_jump (on_failure_jump, laststart, b + 3, b); | |
397 | pending_exact = 0; | |
398 | b += 3; | |
399 | if (!zero_times_ok) | |
400 | { | |
401 | /* At least one repetition required: insert before the loop | |
402 | a skip over the initial on-failure-jump instruction */ | |
403 | insert_jump (dummy_failure_jump, laststart, laststart + 6, b); | |
404 | b += 3; | |
405 | } | |
406 | break; | |
407 | ||
408 | case '.': | |
409 | laststart = b; | |
410 | PATPUSH (anychar); | |
411 | break; | |
412 | ||
413 | case '[': | |
414 | while (b - bufp->buffer | |
415 | > bufp->allocated - 3 - (1 << BYTEWIDTH) / BYTEWIDTH) | |
416 | /* Note that EXTEND_BUFFER clobbers c */ | |
417 | EXTEND_BUFFER; | |
418 | ||
419 | laststart = b; | |
420 | if (*p == '^') | |
421 | PATPUSH (charset_not), p++; | |
422 | else | |
423 | PATPUSH (charset); | |
424 | p1 = p; | |
425 | ||
426 | PATPUSH ((1 << BYTEWIDTH) / BYTEWIDTH); | |
427 | /* Clear the whole map */ | |
4ed97c9a | 428 | memset (b, '\0', (1 << BYTEWIDTH) / BYTEWIDTH); |
dd3b648e RP |
429 | /* Read in characters and ranges, setting map bits */ |
430 | while (1) | |
431 | { | |
432 | PATFETCH (c); | |
433 | if (c == ']' && p != p1 + 1) break; | |
434 | if (*p == '-' && p[1] != ']') | |
435 | { | |
436 | PATFETCH (c1); | |
437 | PATFETCH (c1); | |
438 | while (c <= c1) | |
439 | b[c / BYTEWIDTH] |= 1 << (c % BYTEWIDTH), c++; | |
440 | } | |
441 | else | |
442 | { | |
443 | b[c / BYTEWIDTH] |= 1 << (c % BYTEWIDTH); | |
444 | } | |
445 | } | |
446 | /* Discard any bitmap bytes that are all 0 at the end of the map. | |
447 | Decrement the map-length byte too. */ | |
448 | while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) | |
449 | b[-1]--; | |
450 | b += b[-1]; | |
451 | break; | |
452 | ||
453 | case '(': | |
454 | if (! (obscure_syntax & RE_NO_BK_PARENS)) | |
455 | goto normal_char; | |
456 | else | |
457 | goto handle_open; | |
458 | ||
459 | case ')': | |
460 | if (! (obscure_syntax & RE_NO_BK_PARENS)) | |
461 | goto normal_char; | |
462 | else | |
463 | goto handle_close; | |
464 | ||
465 | case '\n': | |
466 | if (! (obscure_syntax & RE_NEWLINE_OR)) | |
467 | goto normal_char; | |
468 | else | |
469 | goto handle_bar; | |
470 | ||
471 | case '|': | |
472 | if (! (obscure_syntax & RE_NO_BK_VBAR)) | |
473 | goto normal_char; | |
474 | else | |
475 | goto handle_bar; | |
476 | ||
477 | case '\\': | |
478 | if (p == pend) goto invalid_pattern; | |
479 | PATFETCH_RAW (c); | |
480 | switch (c) | |
481 | { | |
482 | case '(': | |
483 | if (obscure_syntax & RE_NO_BK_PARENS) | |
484 | goto normal_backsl; | |
485 | handle_open: | |
486 | if (stackp == stacke) goto nesting_too_deep; | |
487 | if (regnum < RE_NREGS) | |
488 | { | |
489 | PATPUSH (start_memory); | |
490 | PATPUSH (regnum); | |
491 | } | |
492 | *stackp++ = b - bufp->buffer; | |
493 | *stackp++ = fixup_jump ? fixup_jump - bufp->buffer + 1 : 0; | |
494 | *stackp++ = regnum++; | |
495 | *stackp++ = begalt - bufp->buffer; | |
496 | fixup_jump = 0; | |
497 | laststart = 0; | |
498 | begalt = b; | |
499 | break; | |
500 | ||
501 | case ')': | |
502 | if (obscure_syntax & RE_NO_BK_PARENS) | |
503 | goto normal_backsl; | |
504 | handle_close: | |
505 | if (stackp == stackb) goto unmatched_close; | |
506 | begalt = *--stackp + bufp->buffer; | |
507 | if (fixup_jump) | |
508 | store_jump (fixup_jump, jump, b); | |
509 | if (stackp[-1] < RE_NREGS) | |
510 | { | |
511 | PATPUSH (stop_memory); | |
512 | PATPUSH (stackp[-1]); | |
513 | } | |
514 | stackp -= 2; | |
515 | fixup_jump = 0; | |
516 | if (*stackp) | |
517 | fixup_jump = *stackp + bufp->buffer - 1; | |
518 | laststart = *--stackp + bufp->buffer; | |
519 | break; | |
520 | ||
521 | case '|': | |
522 | if (obscure_syntax & RE_NO_BK_VBAR) | |
523 | goto normal_backsl; | |
524 | handle_bar: | |
525 | insert_jump (on_failure_jump, begalt, b + 6, b); | |
526 | pending_exact = 0; | |
527 | b += 3; | |
528 | if (fixup_jump) | |
529 | store_jump (fixup_jump, jump, b); | |
530 | fixup_jump = b; | |
531 | b += 3; | |
532 | laststart = 0; | |
533 | begalt = b; | |
534 | break; | |
535 | ||
536 | #ifdef emacs | |
537 | case '=': | |
538 | PATPUSH (at_dot); | |
539 | break; | |
540 | ||
541 | case 's': | |
542 | laststart = b; | |
543 | PATPUSH (syntaxspec); | |
544 | PATFETCH (c); | |
545 | PATPUSH (syntax_spec_code[c]); | |
546 | break; | |
547 | ||
548 | case 'S': | |
549 | laststart = b; | |
550 | PATPUSH (notsyntaxspec); | |
551 | PATFETCH (c); | |
552 | PATPUSH (syntax_spec_code[c]); | |
553 | break; | |
554 | #endif /* emacs */ | |
555 | ||
556 | case 'w': | |
557 | laststart = b; | |
558 | PATPUSH (wordchar); | |
559 | break; | |
560 | ||
561 | case 'W': | |
562 | laststart = b; | |
563 | PATPUSH (notwordchar); | |
564 | break; | |
565 | ||
566 | case '<': | |
567 | PATPUSH (wordbeg); | |
568 | break; | |
569 | ||
570 | case '>': | |
571 | PATPUSH (wordend); | |
572 | break; | |
573 | ||
574 | case 'b': | |
575 | PATPUSH (wordbound); | |
576 | break; | |
577 | ||
578 | case 'B': | |
579 | PATPUSH (notwordbound); | |
580 | break; | |
581 | ||
582 | case '`': | |
583 | PATPUSH (begbuf); | |
584 | break; | |
585 | ||
586 | case '\'': | |
587 | PATPUSH (endbuf); | |
588 | break; | |
589 | ||
590 | case '1': | |
591 | case '2': | |
592 | case '3': | |
593 | case '4': | |
594 | case '5': | |
595 | case '6': | |
596 | case '7': | |
597 | case '8': | |
598 | case '9': | |
599 | c1 = c - '0'; | |
600 | if (c1 >= regnum) | |
601 | goto normal_char; | |
602 | for (stackt = stackp - 2; stackt > stackb; stackt -= 4) | |
603 | if (*stackt == c1) | |
604 | goto normal_char; | |
605 | laststart = b; | |
606 | PATPUSH (duplicate); | |
607 | PATPUSH (c1); | |
608 | break; | |
609 | ||
610 | case '+': | |
611 | case '?': | |
612 | if (obscure_syntax & RE_BK_PLUS_QM) | |
613 | goto handle_plus; | |
614 | ||
615 | default: | |
616 | normal_backsl: | |
617 | /* You might think it would be useful for \ to mean | |
618 | not to translate; but if we don't translate it | |
619 | it will never match anything. */ | |
620 | if (translate) c = translate[c]; | |
621 | goto normal_char; | |
622 | } | |
623 | break; | |
624 | ||
625 | default: | |
626 | normal_char: | |
627 | if (!pending_exact || pending_exact + *pending_exact + 1 != b | |
628 | || *pending_exact == 0177 || *p == '*' || *p == '^' | |
629 | || ((obscure_syntax & RE_BK_PLUS_QM) | |
630 | ? *p == '\\' && (p[1] == '+' || p[1] == '?') | |
631 | : (*p == '+' || *p == '?'))) | |
632 | { | |
633 | laststart = b; | |
634 | PATPUSH (exactn); | |
635 | pending_exact = b; | |
636 | PATPUSH (0); | |
637 | } | |
638 | PATPUSH (c); | |
639 | (*pending_exact)++; | |
640 | } | |
641 | } | |
642 | ||
643 | if (fixup_jump) | |
644 | store_jump (fixup_jump, jump, b); | |
645 | ||
646 | if (stackp != stackb) goto unmatched_open; | |
647 | ||
648 | bufp->used = b - bufp->buffer; | |
649 | return 0; | |
650 | ||
651 | invalid_pattern: | |
652 | return "Invalid regular expression"; | |
653 | ||
654 | unmatched_open: | |
655 | return "Unmatched \\("; | |
656 | ||
657 | unmatched_close: | |
658 | return "Unmatched \\)"; | |
659 | ||
660 | end_of_pattern: | |
661 | return "Premature end of regular expression"; | |
662 | ||
663 | nesting_too_deep: | |
664 | return "Nesting too deep"; | |
665 | ||
666 | too_big: | |
667 | return "Regular expression too big"; | |
668 | ||
669 | memory_exhausted: | |
670 | return "Memory exhausted"; | |
671 | } | |
672 | ||
673 | /* Store where `from' points a jump operation to jump to where `to' points. | |
674 | `opcode' is the opcode to store. */ | |
675 | ||
50e0dc41 | 676 | static void |
dd3b648e RP |
677 | store_jump (from, opcode, to) |
678 | char *from, *to; | |
679 | char opcode; | |
680 | { | |
681 | from[0] = opcode; | |
682 | from[1] = (to - (from + 3)) & 0377; | |
683 | from[2] = (to - (from + 3)) >> 8; | |
684 | } | |
685 | ||
686 | /* Open up space at char FROM, and insert there a jump to TO. | |
687 | CURRENT_END gives te end of the storage no in use, | |
688 | so we know how much data to copy up. | |
689 | OP is the opcode of the jump to insert. | |
690 | ||
691 | If you call this function, you must zero out pending_exact. */ | |
692 | ||
50e0dc41 | 693 | static void |
dd3b648e RP |
694 | insert_jump (op, from, to, current_end) |
695 | char op; | |
696 | char *from, *to, *current_end; | |
697 | { | |
698 | register char *pto = current_end + 3; | |
699 | register char *pfrom = current_end; | |
700 | while (pfrom != from) | |
701 | *--pto = *--pfrom; | |
702 | store_jump (from, op, to); | |
703 | } | |
704 | \f | |
705 | /* Given a pattern, compute a fastmap from it. | |
706 | The fastmap records which of the (1 << BYTEWIDTH) possible characters | |
707 | can start a string that matches the pattern. | |
708 | This fastmap is used by re_search to skip quickly over totally implausible text. | |
709 | ||
710 | The caller must supply the address of a (1 << BYTEWIDTH)-byte data area | |
711 | as bufp->fastmap. | |
712 | The other components of bufp describe the pattern to be used. */ | |
713 | ||
714 | void | |
715 | re_compile_fastmap (bufp) | |
716 | struct re_pattern_buffer *bufp; | |
717 | { | |
718 | unsigned char *pattern = (unsigned char *) bufp->buffer; | |
719 | int size = bufp->used; | |
720 | register char *fastmap = bufp->fastmap; | |
721 | register unsigned char *p = pattern; | |
722 | register unsigned char *pend = pattern + size; | |
51b57ded | 723 | register int j; |
dd3b648e RP |
724 | unsigned char *translate = (unsigned char *) bufp->translate; |
725 | ||
726 | unsigned char *stackb[NFAILURES]; | |
727 | unsigned char **stackp = stackb; | |
728 | ||
4ed97c9a | 729 | memset (fastmap, '\0', (1 << BYTEWIDTH)); |
dd3b648e RP |
730 | bufp->fastmap_accurate = 1; |
731 | bufp->can_be_null = 0; | |
732 | ||
733 | while (p) | |
734 | { | |
735 | if (p == pend) | |
736 | { | |
737 | bufp->can_be_null = 1; | |
738 | break; | |
739 | } | |
740 | #ifdef SWITCH_ENUM_BUG | |
741 | switch ((int) ((enum regexpcode) *p++)) | |
742 | #else | |
743 | switch ((enum regexpcode) *p++) | |
744 | #endif | |
745 | { | |
746 | case exactn: | |
747 | if (translate) | |
748 | fastmap[translate[p[1]]] = 1; | |
749 | else | |
750 | fastmap[p[1]] = 1; | |
751 | break; | |
752 | ||
753 | case begline: | |
754 | case before_dot: | |
755 | case at_dot: | |
756 | case after_dot: | |
757 | case begbuf: | |
758 | case endbuf: | |
759 | case wordbound: | |
760 | case notwordbound: | |
761 | case wordbeg: | |
762 | case wordend: | |
763 | continue; | |
764 | ||
765 | case endline: | |
766 | if (translate) | |
767 | fastmap[translate['\n']] = 1; | |
768 | else | |
769 | fastmap['\n'] = 1; | |
770 | if (bufp->can_be_null != 1) | |
771 | bufp->can_be_null = 2; | |
772 | break; | |
773 | ||
774 | case finalize_jump: | |
775 | case maybe_finalize_jump: | |
776 | case jump: | |
777 | case dummy_failure_jump: | |
778 | bufp->can_be_null = 1; | |
779 | j = *p++ & 0377; | |
780 | j += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
781 | p += j + 1; /* The 1 compensates for missing ++ above */ | |
782 | if (j > 0) | |
783 | continue; | |
784 | /* Jump backward reached implies we just went through | |
785 | the body of a loop and matched nothing. | |
786 | Opcode jumped to should be an on_failure_jump. | |
787 | Just treat it like an ordinary jump. | |
788 | For a * loop, it has pushed its failure point already; | |
789 | if so, discard that as redundant. */ | |
790 | if ((enum regexpcode) *p != on_failure_jump) | |
791 | continue; | |
792 | p++; | |
793 | j = *p++ & 0377; | |
794 | j += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
795 | p += j + 1; /* The 1 compensates for missing ++ above */ | |
796 | if (stackp != stackb && *stackp == p) | |
797 | stackp--; | |
798 | continue; | |
799 | ||
800 | case on_failure_jump: | |
801 | j = *p++ & 0377; | |
802 | j += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
803 | p++; | |
804 | *++stackp = p + j; | |
805 | continue; | |
806 | ||
807 | case start_memory: | |
808 | case stop_memory: | |
809 | p++; | |
810 | continue; | |
811 | ||
812 | case duplicate: | |
813 | bufp->can_be_null = 1; | |
814 | fastmap['\n'] = 1; | |
815 | case anychar: | |
816 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
817 | if (j != '\n') | |
818 | fastmap[j] = 1; | |
819 | if (bufp->can_be_null) | |
820 | return; | |
821 | /* Don't return; check the alternative paths | |
822 | so we can set can_be_null if appropriate. */ | |
823 | break; | |
824 | ||
825 | case wordchar: | |
826 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
827 | if (SYNTAX (j) == Sword) | |
828 | fastmap[j] = 1; | |
829 | break; | |
830 | ||
831 | case notwordchar: | |
832 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
833 | if (SYNTAX (j) != Sword) | |
834 | fastmap[j] = 1; | |
835 | break; | |
836 | ||
837 | #ifdef emacs | |
838 | case syntaxspec: | |
839 | k = *p++; | |
840 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
841 | if (SYNTAX (j) == (enum syntaxcode) k) | |
842 | fastmap[j] = 1; | |
843 | break; | |
844 | ||
845 | case notsyntaxspec: | |
846 | k = *p++; | |
847 | for (j = 0; j < (1 << BYTEWIDTH); j++) | |
848 | if (SYNTAX (j) != (enum syntaxcode) k) | |
849 | fastmap[j] = 1; | |
850 | break; | |
851 | #endif /* emacs */ | |
852 | ||
853 | case charset: | |
854 | for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | |
855 | if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) | |
856 | { | |
857 | if (translate) | |
858 | fastmap[translate[j]] = 1; | |
859 | else | |
860 | fastmap[j] = 1; | |
861 | } | |
862 | break; | |
863 | ||
864 | case charset_not: | |
865 | /* Chars beyond end of map must be allowed */ | |
866 | for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) | |
867 | if (translate) | |
868 | fastmap[translate[j]] = 1; | |
869 | else | |
870 | fastmap[j] = 1; | |
871 | ||
872 | for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | |
873 | if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) | |
874 | { | |
875 | if (translate) | |
876 | fastmap[translate[j]] = 1; | |
877 | else | |
878 | fastmap[j] = 1; | |
879 | } | |
880 | break; | |
6b14af2b FF |
881 | case unused: |
882 | case syntaxspec: | |
883 | case notsyntaxspec: | |
884 | default: | |
885 | break; | |
dd3b648e RP |
886 | } |
887 | ||
888 | /* Get here means we have successfully found the possible starting characters | |
889 | of one path of the pattern. We need not follow this path any farther. | |
890 | Instead, look at the next alternative remembered in the stack. */ | |
891 | if (stackp != stackb) | |
892 | p = *stackp--; | |
893 | else | |
894 | break; | |
895 | } | |
896 | } | |
897 | \f | |
898 | /* Like re_search_2, below, but only one string is specified. */ | |
899 | ||
900 | int | |
901 | re_search (pbufp, string, size, startpos, range, regs) | |
902 | struct re_pattern_buffer *pbufp; | |
903 | char *string; | |
904 | int size, startpos, range; | |
905 | struct re_registers *regs; | |
906 | { | |
907 | return re_search_2 (pbufp, 0, 0, string, size, startpos, range, regs, size); | |
908 | } | |
909 | ||
910 | /* Like re_match_2 but tries first a match starting at index STARTPOS, | |
911 | then at STARTPOS + 1, and so on. | |
912 | RANGE is the number of places to try before giving up. | |
913 | If RANGE is negative, the starting positions tried are | |
914 | STARTPOS, STARTPOS - 1, etc. | |
915 | It is up to the caller to make sure that range is not so large | |
916 | as to take the starting position outside of the input strings. | |
917 | ||
918 | The value returned is the position at which the match was found, | |
919 | or -1 if no match was found, | |
920 | or -2 if error (such as failure stack overflow). */ | |
921 | ||
922 | int | |
923 | re_search_2 (pbufp, string1, size1, string2, size2, startpos, range, regs, mstop) | |
924 | struct re_pattern_buffer *pbufp; | |
925 | char *string1, *string2; | |
926 | int size1, size2; | |
927 | int startpos; | |
928 | register int range; | |
929 | struct re_registers *regs; | |
930 | int mstop; | |
931 | { | |
932 | register char *fastmap = pbufp->fastmap; | |
933 | register unsigned char *translate = (unsigned char *) pbufp->translate; | |
934 | int total = size1 + size2; | |
935 | int val; | |
936 | ||
937 | /* Update the fastmap now if not correct already */ | |
938 | if (fastmap && !pbufp->fastmap_accurate) | |
939 | re_compile_fastmap (pbufp); | |
940 | ||
941 | /* Don't waste time in a long search for a pattern | |
942 | that says it is anchored. */ | |
943 | if (pbufp->used > 0 && (enum regexpcode) pbufp->buffer[0] == begbuf | |
944 | && range > 0) | |
945 | { | |
946 | if (startpos > 0) | |
947 | return -1; | |
948 | else | |
949 | range = 1; | |
950 | } | |
951 | ||
952 | while (1) | |
953 | { | |
954 | /* If a fastmap is supplied, skip quickly over characters | |
955 | that cannot possibly be the start of a match. | |
956 | Note, however, that if the pattern can possibly match | |
957 | the null string, we must test it at each starting point | |
958 | so that we take the first null string we get. */ | |
959 | ||
960 | if (fastmap && startpos < total && pbufp->can_be_null != 1) | |
961 | { | |
962 | if (range > 0) | |
963 | { | |
964 | register int lim = 0; | |
965 | register unsigned char *p; | |
966 | int irange = range; | |
967 | if (startpos < size1 && startpos + range >= size1) | |
968 | lim = range - (size1 - startpos); | |
969 | ||
970 | p = ((unsigned char *) | |
971 | &(startpos >= size1 ? string2 - size1 : string1)[startpos]); | |
972 | ||
973 | if (translate) | |
974 | { | |
975 | while (range > lim && !fastmap[translate[*p++]]) | |
976 | range--; | |
977 | } | |
978 | else | |
979 | { | |
980 | while (range > lim && !fastmap[*p++]) | |
981 | range--; | |
982 | } | |
983 | startpos += irange - range; | |
984 | } | |
985 | else | |
986 | { | |
987 | register unsigned char c; | |
988 | if (startpos >= size1) | |
989 | c = string2[startpos - size1]; | |
990 | else | |
991 | c = string1[startpos]; | |
992 | c &= 0xff; | |
993 | if (translate ? !fastmap[translate[c]] : !fastmap[c]) | |
994 | goto advance; | |
995 | } | |
996 | } | |
997 | ||
998 | if (range >= 0 && startpos == total | |
999 | && fastmap && pbufp->can_be_null == 0) | |
1000 | return -1; | |
1001 | ||
1002 | val = re_match_2 (pbufp, string1, size1, string2, size2, startpos, regs, mstop); | |
1003 | if (0 <= val) | |
1004 | { | |
1005 | if (val == -2) | |
1006 | return -2; | |
1007 | return startpos; | |
1008 | } | |
1009 | ||
1010 | #ifdef C_ALLOCA | |
1011 | alloca (0); | |
1012 | #endif /* C_ALLOCA */ | |
1013 | ||
1014 | advance: | |
1015 | if (!range) break; | |
1016 | if (range > 0) range--, startpos++; else range++, startpos--; | |
1017 | } | |
1018 | return -1; | |
1019 | } | |
1020 | \f | |
1021 | #ifndef emacs /* emacs never uses this */ | |
1022 | int | |
1023 | re_match (pbufp, string, size, pos, regs) | |
1024 | struct re_pattern_buffer *pbufp; | |
1025 | char *string; | |
1026 | int size, pos; | |
1027 | struct re_registers *regs; | |
1028 | { | |
1029 | return re_match_2 (pbufp, 0, 0, string, size, pos, regs, size); | |
1030 | } | |
1031 | #endif /* emacs */ | |
1032 | ||
1033 | /* Maximum size of failure stack. Beyond this, overflow is an error. */ | |
1034 | ||
1035 | int re_max_failures = 2000; | |
1036 | ||
9823e3f4 | 1037 | static int memcmp_translate(); |
dd3b648e RP |
1038 | /* Match the pattern described by PBUFP |
1039 | against data which is the virtual concatenation of STRING1 and STRING2. | |
1040 | SIZE1 and SIZE2 are the sizes of the two data strings. | |
1041 | Start the match at position POS. | |
1042 | Do not consider matching past the position MSTOP. | |
1043 | ||
1044 | If pbufp->fastmap is nonzero, then it had better be up to date. | |
1045 | ||
1046 | The reason that the data to match are specified as two components | |
1047 | which are to be regarded as concatenated | |
1048 | is so this function can be used directly on the contents of an Emacs buffer. | |
1049 | ||
1050 | -1 is returned if there is no match. -2 is returned if there is | |
1051 | an error (such as match stack overflow). Otherwise the value is the length | |
1052 | of the substring which was matched. */ | |
1053 | ||
1054 | int | |
1055 | re_match_2 (pbufp, string1, size1, string2, size2, pos, regs, mstop) | |
1056 | struct re_pattern_buffer *pbufp; | |
1057 | unsigned char *string1, *string2; | |
1058 | int size1, size2; | |
1059 | int pos; | |
1060 | struct re_registers *regs; | |
1061 | int mstop; | |
1062 | { | |
1063 | register unsigned char *p = (unsigned char *) pbufp->buffer; | |
1064 | register unsigned char *pend = p + pbufp->used; | |
1065 | /* End of first string */ | |
1066 | unsigned char *end1; | |
1067 | /* End of second string */ | |
1068 | unsigned char *end2; | |
1069 | /* Pointer just past last char to consider matching */ | |
1070 | unsigned char *end_match_1, *end_match_2; | |
1071 | register unsigned char *d, *dend; | |
1072 | register int mcnt; | |
1073 | unsigned char *translate = (unsigned char *) pbufp->translate; | |
1074 | ||
1075 | /* Failure point stack. Each place that can handle a failure further down the line | |
1076 | pushes a failure point on this stack. It consists of two char *'s. | |
1077 | The first one pushed is where to resume scanning the pattern; | |
1078 | the second pushed is where to resume scanning the strings. | |
1079 | If the latter is zero, the failure point is a "dummy". | |
1080 | If a failure happens and the innermost failure point is dormant, | |
1081 | it discards that failure point and tries the next one. */ | |
1082 | ||
1083 | unsigned char *initial_stack[2 * NFAILURES]; | |
1084 | unsigned char **stackb = initial_stack; | |
1085 | unsigned char **stackp = stackb, **stacke = &stackb[2 * NFAILURES]; | |
1086 | ||
1087 | /* Information on the "contents" of registers. | |
1088 | These are pointers into the input strings; they record | |
1089 | just what was matched (on this attempt) by some part of the pattern. | |
1090 | The start_memory command stores the start of a register's contents | |
1091 | and the stop_memory command stores the end. | |
1092 | ||
1093 | At that point, regstart[regnum] points to the first character in the register, | |
1094 | regend[regnum] points to the first character beyond the end of the register, | |
1095 | regstart_seg1[regnum] is true iff regstart[regnum] points into string1, | |
1096 | and regend_seg1[regnum] is true iff regend[regnum] points into string1. */ | |
1097 | ||
1098 | unsigned char *regstart[RE_NREGS]; | |
1099 | unsigned char *regend[RE_NREGS]; | |
1100 | unsigned char regstart_seg1[RE_NREGS], regend_seg1[RE_NREGS]; | |
1101 | ||
1102 | /* Set up pointers to ends of strings. | |
1103 | Don't allow the second string to be empty unless both are empty. */ | |
1104 | if (!size2) | |
1105 | { | |
1106 | string2 = string1; | |
1107 | size2 = size1; | |
1108 | string1 = 0; | |
1109 | size1 = 0; | |
1110 | } | |
1111 | end1 = string1 + size1; | |
1112 | end2 = string2 + size2; | |
1113 | ||
1114 | /* Compute where to stop matching, within the two strings */ | |
1115 | if (mstop <= size1) | |
1116 | { | |
1117 | end_match_1 = string1 + mstop; | |
1118 | end_match_2 = string2; | |
1119 | } | |
1120 | else | |
1121 | { | |
1122 | end_match_1 = end1; | |
1123 | end_match_2 = string2 + mstop - size1; | |
1124 | } | |
1125 | ||
1126 | /* Initialize \) text positions to -1 | |
1127 | to mark ones that no \( or \) has been seen for. */ | |
1128 | ||
b52cac6b | 1129 | for (mcnt = 0; mcnt < (int) (sizeof (regend) / sizeof (*regend)); mcnt++) |
dd3b648e RP |
1130 | regend[mcnt] = (unsigned char *) -1; |
1131 | ||
1132 | /* `p' scans through the pattern as `d' scans through the data. | |
1133 | `dend' is the end of the input string that `d' points within. | |
1134 | `d' is advanced into the following input string whenever necessary, | |
1135 | but this happens before fetching; | |
1136 | therefore, at the beginning of the loop, | |
1137 | `d' can be pointing at the end of a string, | |
1138 | but it cannot equal string2. */ | |
1139 | ||
1140 | if (pos <= size1) | |
1141 | d = string1 + pos, dend = end_match_1; | |
1142 | else | |
1143 | d = string2 + pos - size1, dend = end_match_2; | |
1144 | ||
1145 | /* Write PREFETCH; just before fetching a character with *d. */ | |
1146 | #define PREFETCH \ | |
1147 | while (d == dend) \ | |
1148 | { if (dend == end_match_2) goto fail; /* end of string2 => failure */ \ | |
1149 | d = string2; /* end of string1 => advance to string2. */ \ | |
1150 | dend = end_match_2; } | |
1151 | ||
1152 | /* This loop loops over pattern commands. | |
1153 | It exits by returning from the function if match is complete, | |
1154 | or it drops through if match fails at this starting point in the input data. */ | |
1155 | ||
1156 | while (1) | |
1157 | { | |
1158 | if (p == pend) | |
1159 | /* End of pattern means we have succeeded! */ | |
1160 | { | |
1161 | /* If caller wants register contents data back, convert it to indices */ | |
1162 | if (regs) | |
1163 | { | |
1164 | regs->start[0] = pos; | |
1165 | if (dend == end_match_1) | |
1166 | regs->end[0] = d - string1; | |
1167 | else | |
1168 | regs->end[0] = d - string2 + size1; | |
1169 | for (mcnt = 1; mcnt < RE_NREGS; mcnt++) | |
1170 | { | |
1171 | if (regend[mcnt] == (unsigned char *) -1) | |
1172 | { | |
1173 | regs->start[mcnt] = -1; | |
1174 | regs->end[mcnt] = -1; | |
1175 | continue; | |
1176 | } | |
1177 | if (regstart_seg1[mcnt]) | |
1178 | regs->start[mcnt] = regstart[mcnt] - string1; | |
1179 | else | |
1180 | regs->start[mcnt] = regstart[mcnt] - string2 + size1; | |
1181 | if (regend_seg1[mcnt]) | |
1182 | regs->end[mcnt] = regend[mcnt] - string1; | |
1183 | else | |
1184 | regs->end[mcnt] = regend[mcnt] - string2 + size1; | |
1185 | } | |
1186 | } | |
1187 | if (dend == end_match_1) | |
1188 | return (d - string1 - pos); | |
1189 | else | |
1190 | return d - string2 + size1 - pos; | |
1191 | } | |
1192 | ||
1193 | /* Otherwise match next pattern command */ | |
1194 | #ifdef SWITCH_ENUM_BUG | |
1195 | switch ((int) ((enum regexpcode) *p++)) | |
1196 | #else | |
1197 | switch ((enum regexpcode) *p++) | |
1198 | #endif | |
1199 | { | |
1200 | ||
1201 | /* \( is represented by a start_memory, \) by a stop_memory. | |
1202 | Both of those commands contain a "register number" argument. | |
1203 | The text matched within the \( and \) is recorded under that number. | |
1204 | Then, \<digit> turns into a `duplicate' command which | |
1205 | is followed by the numeric value of <digit> as the register number. */ | |
1206 | ||
1207 | case start_memory: | |
1208 | regstart[*p] = d; | |
1209 | regstart_seg1[*p++] = (dend == end_match_1); | |
1210 | break; | |
1211 | ||
1212 | case stop_memory: | |
1213 | regend[*p] = d; | |
1214 | regend_seg1[*p++] = (dend == end_match_1); | |
1215 | break; | |
1216 | ||
1217 | case duplicate: | |
1218 | { | |
1219 | int regno = *p++; /* Get which register to match against */ | |
1220 | register unsigned char *d2, *dend2; | |
1221 | ||
1222 | d2 = regstart[regno]; | |
1223 | dend2 = ((regstart_seg1[regno] == regend_seg1[regno]) | |
1224 | ? regend[regno] : end_match_1); | |
1225 | while (1) | |
1226 | { | |
1227 | /* Advance to next segment in register contents, if necessary */ | |
1228 | while (d2 == dend2) | |
1229 | { | |
1230 | if (dend2 == end_match_2) break; | |
1231 | if (dend2 == regend[regno]) break; | |
1232 | d2 = string2, dend2 = regend[regno]; /* end of string1 => advance to string2. */ | |
1233 | } | |
1234 | /* At end of register contents => success */ | |
1235 | if (d2 == dend2) break; | |
1236 | ||
1237 | /* Advance to next segment in data being matched, if necessary */ | |
1238 | PREFETCH; | |
1239 | ||
1240 | /* mcnt gets # consecutive chars to compare */ | |
1241 | mcnt = dend - d; | |
1242 | if (mcnt > dend2 - d2) | |
1243 | mcnt = dend2 - d2; | |
1244 | /* Compare that many; failure if mismatch, else skip them. */ | |
9823e3f4 | 1245 | if (translate ? memcmp_translate (d, d2, mcnt, translate) : memcmp (d, d2, mcnt)) |
dd3b648e RP |
1246 | goto fail; |
1247 | d += mcnt, d2 += mcnt; | |
1248 | } | |
1249 | } | |
1250 | break; | |
1251 | ||
1252 | case anychar: | |
1253 | /* fetch a data character */ | |
1254 | PREFETCH; | |
1255 | /* Match anything but a newline. */ | |
1256 | if ((translate ? translate[*d++] : *d++) == '\n') | |
1257 | goto fail; | |
1258 | break; | |
1259 | ||
1260 | case charset: | |
1261 | case charset_not: | |
1262 | { | |
1263 | /* Nonzero for charset_not */ | |
1264 | int not = 0; | |
1265 | register int c; | |
1266 | if (*(p - 1) == (unsigned char) charset_not) | |
1267 | not = 1; | |
1268 | ||
1269 | /* fetch a data character */ | |
1270 | PREFETCH; | |
1271 | ||
1272 | if (translate) | |
1273 | c = translate [*d]; | |
1274 | else | |
1275 | c = *d; | |
1276 | ||
1277 | if (c < *p * BYTEWIDTH | |
1278 | && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) | |
1279 | not = !not; | |
1280 | ||
1281 | p += 1 + *p; | |
1282 | ||
1283 | if (!not) goto fail; | |
1284 | d++; | |
1285 | break; | |
1286 | } | |
1287 | ||
1288 | case begline: | |
1289 | if (d == string1 || d[-1] == '\n') | |
1290 | break; | |
1291 | goto fail; | |
1292 | ||
1293 | case endline: | |
1294 | if (d == end2 | |
1295 | || (d == end1 ? (size2 == 0 || *string2 == '\n') : *d == '\n')) | |
1296 | break; | |
1297 | goto fail; | |
1298 | ||
1299 | /* "or" constructs ("|") are handled by starting each alternative | |
1300 | with an on_failure_jump that points to the start of the next alternative. | |
1301 | Each alternative except the last ends with a jump to the joining point. | |
1302 | (Actually, each jump except for the last one really jumps | |
1303 | to the following jump, because tensioning the jumps is a hassle.) */ | |
1304 | ||
1305 | /* The start of a stupid repeat has an on_failure_jump that points | |
1306 | past the end of the repeat text. | |
1307 | This makes a failure point so that, on failure to match a repetition, | |
1308 | matching restarts past as many repetitions have been found | |
1309 | with no way to fail and look for another one. */ | |
1310 | ||
1311 | /* A smart repeat is similar but loops back to the on_failure_jump | |
1312 | so that each repetition makes another failure point. */ | |
1313 | ||
1314 | case on_failure_jump: | |
1315 | if (stackp == stacke) | |
1316 | { | |
1317 | unsigned char **stackx; | |
1318 | if (stacke - stackb > re_max_failures * 2) | |
1319 | return -2; | |
1320 | stackx = (unsigned char **) alloca (2 * (stacke - stackb) | |
1321 | * sizeof (char *)); | |
ade40d31 | 1322 | memcpy (stackx, stackb, (stacke - stackb) * sizeof (char *)); |
dd3b648e RP |
1323 | stackp = stackx + (stackp - stackb); |
1324 | stacke = stackx + 2 * (stacke - stackb); | |
1325 | stackb = stackx; | |
1326 | } | |
1327 | mcnt = *p++ & 0377; | |
1328 | mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
1329 | p++; | |
1330 | *stackp++ = mcnt + p; | |
1331 | *stackp++ = d; | |
1332 | break; | |
1333 | ||
1334 | /* The end of a smart repeat has an maybe_finalize_jump back. | |
1335 | Change it either to a finalize_jump or an ordinary jump. */ | |
1336 | ||
1337 | case maybe_finalize_jump: | |
1338 | mcnt = *p++ & 0377; | |
1339 | mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
1340 | p++; | |
1341 | { | |
1342 | register unsigned char *p2 = p; | |
1343 | /* Compare what follows with the begining of the repeat. | |
1344 | If we can establish that there is nothing that they would | |
1345 | both match, we can change to finalize_jump */ | |
1346 | while (p2 != pend | |
1347 | && (*p2 == (unsigned char) stop_memory | |
1348 | || *p2 == (unsigned char) start_memory)) | |
1349 | p2++; | |
1350 | if (p2 == pend) | |
1351 | p[-3] = (unsigned char) finalize_jump; | |
1352 | else if (*p2 == (unsigned char) exactn | |
1353 | || *p2 == (unsigned char) endline) | |
1354 | { | |
1355 | register int c = *p2 == (unsigned char) endline ? '\n' : p2[2]; | |
1356 | register unsigned char *p1 = p + mcnt; | |
1357 | /* p1[0] ... p1[2] are an on_failure_jump. | |
1358 | Examine what follows that */ | |
1359 | if (p1[3] == (unsigned char) exactn && p1[5] != c) | |
1360 | p[-3] = (unsigned char) finalize_jump; | |
1361 | else if (p1[3] == (unsigned char) charset | |
1362 | || p1[3] == (unsigned char) charset_not) | |
1363 | { | |
1364 | int not = p1[3] == (unsigned char) charset_not; | |
1365 | if (c < p1[4] * BYTEWIDTH | |
1366 | && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) | |
1367 | not = !not; | |
1368 | /* not is 1 if c would match */ | |
1369 | /* That means it is not safe to finalize */ | |
1370 | if (!not) | |
1371 | p[-3] = (unsigned char) finalize_jump; | |
1372 | } | |
1373 | } | |
1374 | } | |
1375 | p -= 2; | |
1376 | if (p[-1] != (unsigned char) finalize_jump) | |
1377 | { | |
1378 | p[-1] = (unsigned char) jump; | |
1379 | goto nofinalize; | |
1380 | } | |
1381 | ||
1382 | /* The end of a stupid repeat has a finalize-jump | |
1383 | back to the start, where another failure point will be made | |
1384 | which will point after all the repetitions found so far. */ | |
1385 | ||
1386 | case finalize_jump: | |
1387 | stackp -= 2; | |
1388 | ||
1389 | case jump: | |
1390 | nofinalize: | |
1391 | mcnt = *p++ & 0377; | |
1392 | mcnt += SIGN_EXTEND_CHAR (*(char *)p) << 8; | |
1393 | p += mcnt + 1; /* The 1 compensates for missing ++ above */ | |
1394 | break; | |
1395 | ||
1396 | case dummy_failure_jump: | |
1397 | if (stackp == stacke) | |
1398 | { | |
1399 | unsigned char **stackx | |
1400 | = (unsigned char **) alloca (2 * (stacke - stackb) | |
1401 | * sizeof (char *)); | |
ade40d31 | 1402 | memcpy (stackx, stackb, (stacke - stackb) * sizeof (char *)); |
dd3b648e RP |
1403 | stackp = stackx + (stackp - stackb); |
1404 | stacke = stackx + 2 * (stacke - stackb); | |
1405 | stackb = stackx; | |
1406 | } | |
1407 | *stackp++ = 0; | |
1408 | *stackp++ = 0; | |
1409 | goto nofinalize; | |
1410 | ||
1411 | case wordbound: | |
1412 | if (d == string1 /* Points to first char */ | |
1413 | || d == end2 /* Points to end */ | |
1414 | || (d == end1 && size2 == 0)) /* Points to end */ | |
1415 | break; | |
1416 | if ((SYNTAX (d[-1]) == Sword) | |
1417 | != (SYNTAX (d == end1 ? *string2 : *d) == Sword)) | |
1418 | break; | |
1419 | goto fail; | |
1420 | ||
1421 | case notwordbound: | |
1422 | if (d == string1 /* Points to first char */ | |
1423 | || d == end2 /* Points to end */ | |
1424 | || (d == end1 && size2 == 0)) /* Points to end */ | |
1425 | goto fail; | |
1426 | if ((SYNTAX (d[-1]) == Sword) | |
1427 | != (SYNTAX (d == end1 ? *string2 : *d) == Sword)) | |
1428 | goto fail; | |
1429 | break; | |
1430 | ||
1431 | case wordbeg: | |
1432 | if (d == end2 /* Points to end */ | |
1433 | || (d == end1 && size2 == 0) /* Points to end */ | |
1434 | || SYNTAX (* (d == end1 ? string2 : d)) != Sword) /* Next char not a letter */ | |
1435 | goto fail; | |
1436 | if (d == string1 /* Points to first char */ | |
1437 | || SYNTAX (d[-1]) != Sword) /* prev char not letter */ | |
1438 | break; | |
1439 | goto fail; | |
1440 | ||
1441 | case wordend: | |
1442 | if (d == string1 /* Points to first char */ | |
1443 | || SYNTAX (d[-1]) != Sword) /* prev char not letter */ | |
1444 | goto fail; | |
1445 | if (d == end2 /* Points to end */ | |
1446 | || (d == end1 && size2 == 0) /* Points to end */ | |
1447 | || SYNTAX (d == end1 ? *string2 : *d) != Sword) /* Next char not a letter */ | |
1448 | break; | |
1449 | goto fail; | |
1450 | ||
1451 | #ifdef emacs | |
1452 | case before_dot: | |
1453 | if (((d - string2 <= (unsigned) size2) | |
1454 | ? d - bf_p2 : d - bf_p1) | |
1455 | <= point) | |
1456 | goto fail; | |
1457 | break; | |
1458 | ||
1459 | case at_dot: | |
1460 | if (((d - string2 <= (unsigned) size2) | |
1461 | ? d - bf_p2 : d - bf_p1) | |
1462 | == point) | |
1463 | goto fail; | |
1464 | break; | |
1465 | ||
1466 | case after_dot: | |
1467 | if (((d - string2 <= (unsigned) size2) | |
1468 | ? d - bf_p2 : d - bf_p1) | |
1469 | >= point) | |
1470 | goto fail; | |
1471 | break; | |
1472 | ||
1473 | case wordchar: | |
1474 | mcnt = (int) Sword; | |
1475 | goto matchsyntax; | |
1476 | ||
1477 | case syntaxspec: | |
1478 | mcnt = *p++; | |
1479 | matchsyntax: | |
1480 | PREFETCH; | |
1481 | if (SYNTAX (*d++) != (enum syntaxcode) mcnt) goto fail; | |
1482 | break; | |
1483 | ||
1484 | case notwordchar: | |
1485 | mcnt = (int) Sword; | |
1486 | goto matchnotsyntax; | |
1487 | ||
1488 | case notsyntaxspec: | |
1489 | mcnt = *p++; | |
1490 | matchnotsyntax: | |
1491 | PREFETCH; | |
1492 | if (SYNTAX (*d++) == (enum syntaxcode) mcnt) goto fail; | |
1493 | break; | |
1494 | #else | |
1495 | case wordchar: | |
1496 | PREFETCH; | |
1497 | if (SYNTAX (*d++) == 0) goto fail; | |
1498 | break; | |
1499 | ||
1500 | case notwordchar: | |
1501 | PREFETCH; | |
1502 | if (SYNTAX (*d++) != 0) goto fail; | |
1503 | break; | |
1504 | #endif /* not emacs */ | |
1505 | ||
1506 | case begbuf: | |
1507 | if (d == string1) /* Note, d cannot equal string2 */ | |
1508 | break; /* unless string1 == string2. */ | |
1509 | goto fail; | |
1510 | ||
1511 | case endbuf: | |
1512 | if (d == end2 || (d == end1 && size2 == 0)) | |
1513 | break; | |
1514 | goto fail; | |
1515 | ||
1516 | case exactn: | |
1517 | /* Match the next few pattern characters exactly. | |
1518 | mcnt is how many characters to match. */ | |
1519 | mcnt = *p++; | |
1520 | if (translate) | |
1521 | { | |
1522 | do | |
1523 | { | |
1524 | PREFETCH; | |
1525 | if (translate[*d++] != *p++) goto fail; | |
1526 | } | |
1527 | while (--mcnt); | |
1528 | } | |
1529 | else | |
1530 | { | |
1531 | do | |
1532 | { | |
1533 | PREFETCH; | |
1534 | if (*d++ != *p++) goto fail; | |
1535 | } | |
1536 | while (--mcnt); | |
1537 | } | |
1538 | break; | |
6b14af2b FF |
1539 | case unused: |
1540 | case before_dot: | |
1541 | case at_dot: | |
1542 | case after_dot: | |
1543 | case syntaxspec: | |
1544 | case notsyntaxspec: | |
1545 | default: | |
1546 | break; | |
dd3b648e RP |
1547 | } |
1548 | continue; /* Successfully matched one pattern command; keep matching */ | |
1549 | ||
1550 | /* Jump here if any matching operation fails. */ | |
1551 | fail: | |
1552 | if (stackp != stackb) | |
1553 | /* A restart point is known. Restart there and pop it. */ | |
1554 | { | |
1555 | if (!stackp[-2]) | |
1556 | { /* If innermost failure point is dormant, flush it and keep looking */ | |
1557 | stackp -= 2; | |
1558 | goto fail; | |
1559 | } | |
1560 | d = *--stackp; | |
1561 | p = *--stackp; | |
1562 | if (d >= string1 && d <= end1) | |
1563 | dend = end_match_1; | |
1564 | } | |
1565 | else break; /* Matching at this starting point really fails! */ | |
1566 | } | |
1567 | return -1; /* Failure to match */ | |
1568 | } | |
1569 | ||
1570 | static int | |
9823e3f4 | 1571 | memcmp_translate (s1, s2, len, translate) |
dd3b648e RP |
1572 | unsigned char *s1, *s2; |
1573 | register int len; | |
1574 | unsigned char *translate; | |
1575 | { | |
1576 | register unsigned char *p1 = s1, *p2 = s2; | |
1577 | while (len) | |
1578 | { | |
1579 | if (translate [*p1++] != translate [*p2++]) return 1; | |
1580 | len--; | |
1581 | } | |
1582 | return 0; | |
1583 | } | |
1584 | \f | |
1585 | /* Entry points compatible with bsd4.2 regex library */ | |
1586 | ||
1587 | #ifndef emacs | |
1588 | ||
1589 | static struct re_pattern_buffer re_comp_buf; | |
1590 | ||
1591 | char * | |
1592 | re_comp (s) | |
ba47c66a | 1593 | const char *s; |
dd3b648e RP |
1594 | { |
1595 | if (!s) | |
1596 | { | |
1597 | if (!re_comp_buf.buffer) | |
1598 | return "No previous regular expression"; | |
1599 | return 0; | |
1600 | } | |
1601 | ||
1602 | if (!re_comp_buf.buffer) | |
1603 | { | |
1604 | if (!(re_comp_buf.buffer = (char *) malloc (200))) | |
1605 | return "Memory exhausted"; | |
1606 | re_comp_buf.allocated = 200; | |
1607 | if (!(re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH))) | |
1608 | return "Memory exhausted"; | |
1609 | } | |
1610 | return re_compile_pattern (s, strlen (s), &re_comp_buf); | |
1611 | } | |
1612 | ||
1613 | int | |
1614 | re_exec (s) | |
1615 | char *s; | |
1616 | { | |
1617 | int len = strlen (s); | |
1618 | return 0 <= re_search (&re_comp_buf, s, len, 0, len, 0); | |
1619 | } | |
1620 | ||
1621 | #endif /* emacs */ | |
1622 | \f | |
1623 | #ifdef test | |
1624 | ||
1625 | #include <stdio.h> | |
1626 | ||
1627 | /* Indexed by a character, gives the upper case equivalent of the character */ | |
1628 | ||
1629 | static char upcase[0400] = | |
1630 | { 000, 001, 002, 003, 004, 005, 006, 007, | |
1631 | 010, 011, 012, 013, 014, 015, 016, 017, | |
1632 | 020, 021, 022, 023, 024, 025, 026, 027, | |
1633 | 030, 031, 032, 033, 034, 035, 036, 037, | |
1634 | 040, 041, 042, 043, 044, 045, 046, 047, | |
1635 | 050, 051, 052, 053, 054, 055, 056, 057, | |
1636 | 060, 061, 062, 063, 064, 065, 066, 067, | |
1637 | 070, 071, 072, 073, 074, 075, 076, 077, | |
1638 | 0100, 0101, 0102, 0103, 0104, 0105, 0106, 0107, | |
1639 | 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117, | |
1640 | 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127, | |
1641 | 0130, 0131, 0132, 0133, 0134, 0135, 0136, 0137, | |
1642 | 0140, 0101, 0102, 0103, 0104, 0105, 0106, 0107, | |
1643 | 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117, | |
1644 | 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127, | |
1645 | 0130, 0131, 0132, 0173, 0174, 0175, 0176, 0177, | |
1646 | 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207, | |
1647 | 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217, | |
1648 | 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227, | |
1649 | 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237, | |
1650 | 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247, | |
1651 | 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257, | |
1652 | 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267, | |
1653 | 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277, | |
1654 | 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307, | |
1655 | 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317, | |
1656 | 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0327, | |
1657 | 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0337, | |
1658 | 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347, | |
1659 | 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357, | |
1660 | 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0367, | |
1661 | 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0377 | |
1662 | }; | |
1663 | ||
1664 | main (argc, argv) | |
1665 | int argc; | |
1666 | char **argv; | |
1667 | { | |
1668 | char pat[80]; | |
1669 | struct re_pattern_buffer buf; | |
1670 | int i; | |
1671 | char c; | |
1672 | char fastmap[(1 << BYTEWIDTH)]; | |
1673 | ||
1674 | /* Allow a command argument to specify the style of syntax. */ | |
1675 | if (argc > 1) | |
1676 | obscure_syntax = atoi (argv[1]); | |
1677 | ||
1678 | buf.allocated = 40; | |
1679 | buf.buffer = (char *) malloc (buf.allocated); | |
1680 | buf.fastmap = fastmap; | |
1681 | buf.translate = upcase; | |
1682 | ||
1683 | while (1) | |
1684 | { | |
1685 | gets (pat); | |
1686 | ||
1687 | if (*pat) | |
1688 | { | |
1689 | re_compile_pattern (pat, strlen(pat), &buf); | |
1690 | ||
1691 | for (i = 0; i < buf.used; i++) | |
1692 | printchar (buf.buffer[i]); | |
1693 | ||
199b2450 | 1694 | putchar_unfiltered ('\n'); |
dd3b648e | 1695 | |
199b2450 | 1696 | printf_unfiltered ("%d allocated, %d used.\n", buf.allocated, buf.used); |
dd3b648e RP |
1697 | |
1698 | re_compile_fastmap (&buf); | |
199b2450 | 1699 | printf_unfiltered ("Allowed by fastmap: "); |
dd3b648e RP |
1700 | for (i = 0; i < (1 << BYTEWIDTH); i++) |
1701 | if (fastmap[i]) printchar (i); | |
199b2450 | 1702 | putchar_unfiltered ('\n'); |
dd3b648e RP |
1703 | } |
1704 | ||
1705 | gets (pat); /* Now read the string to match against */ | |
1706 | ||
1707 | i = re_match (&buf, pat, strlen (pat), 0, 0); | |
199b2450 | 1708 | printf_unfiltered ("Match value %d.\n", i); |
dd3b648e RP |
1709 | } |
1710 | } | |
1711 | ||
1712 | #ifdef NOTDEF | |
1713 | print_buf (bufp) | |
1714 | struct re_pattern_buffer *bufp; | |
1715 | { | |
1716 | int i; | |
1717 | ||
199b2450 | 1718 | printf_unfiltered ("buf is :\n----------------\n"); |
dd3b648e RP |
1719 | for (i = 0; i < bufp->used; i++) |
1720 | printchar (bufp->buffer[i]); | |
1721 | ||
199b2450 | 1722 | printf_unfiltered ("\n%d allocated, %d used.\n", bufp->allocated, bufp->used); |
dd3b648e | 1723 | |
199b2450 | 1724 | printf_unfiltered ("Allowed by fastmap: "); |
dd3b648e RP |
1725 | for (i = 0; i < (1 << BYTEWIDTH); i++) |
1726 | if (bufp->fastmap[i]) | |
1727 | printchar (i); | |
199b2450 | 1728 | printf_unfiltered ("\nAllowed by translate: "); |
dd3b648e RP |
1729 | if (bufp->translate) |
1730 | for (i = 0; i < (1 << BYTEWIDTH); i++) | |
1731 | if (bufp->translate[i]) | |
1732 | printchar (i); | |
199b2450 TL |
1733 | printf_unfiltered ("\nfastmap is%s accurate\n", bufp->fastmap_accurate ? "" : "n't"); |
1734 | printf_unfiltered ("can %s be null\n----------", bufp->can_be_null ? "" : "not"); | |
dd3b648e RP |
1735 | } |
1736 | #endif | |
1737 | ||
1738 | printchar (c) | |
1739 | char c; | |
1740 | { | |
1741 | if (c < 041 || c >= 0177) | |
1742 | { | |
199b2450 TL |
1743 | putchar_unfiltered ('\\'); |
1744 | putchar_unfiltered (((c >> 6) & 3) + '0'); | |
1745 | putchar_unfiltered (((c >> 3) & 7) + '0'); | |
1746 | putchar_unfiltered ((c & 7) + '0'); | |
dd3b648e RP |
1747 | } |
1748 | else | |
199b2450 | 1749 | putchar_unfiltered (c); |
dd3b648e RP |
1750 | } |
1751 | ||
1752 | error (string) | |
1753 | char *string; | |
1754 | { | |
199b2450 | 1755 | puts_unfiltered (string); |
dd3b648e RP |
1756 | exit (1); |
1757 | } | |
1758 | ||
1759 | #endif /* test */ |