Commit | Line | Data |
a687059c |
1 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
2 | * confused with the original package (see point 3 below). Thanks, Henry! |
3 | */ |
4 | |
5 | /* Additional note: this code is very heavily munged from Henry's version |
6 | * in places. In some spots I've traded clarity for efficiency, so don't |
7 | * blame Henry for some of the lack of readability. |
8 | */ |
9 | |
2b69d0c2 |
10 | /* $RCSfile: regexec.c,v $$Revision: 4.0.1.4 $$Date: 92/06/08 15:25:50 $ |
a687059c |
11 | * |
12 | * $Log: regexec.c,v $ |
2b69d0c2 |
13 | * Revision 4.0.1.4 92/06/08 15:25:50 lwall |
14 | * patch20: pattern modifiers i and g didn't interact right |
15 | * patch20: in some cases $` and $' didn't get set by match |
16 | * patch20: /x{0}/ was wrongly interpreted as /x{0,}/ |
17 | * |
f0fcb552 |
18 | * Revision 4.0.1.3 91/11/05 18:23:55 lwall |
19 | * patch11: prepared for ctype implementations that don't define isascii() |
20 | * patch11: initial .* in pattern had dependency on value of $* |
21 | * |
9ef589d8 |
22 | * Revision 4.0.1.2 91/06/07 11:50:33 lwall |
23 | * patch4: new copyright notice |
24 | * patch4: // wouldn't use previous pattern if it started with a null character |
25 | * |
35c8bce7 |
26 | * Revision 4.0.1.1 91/04/12 09:07:39 lwall |
27 | * patch1: regexec only allocated space for 9 subexpresssions |
28 | * |
fe14fcc3 |
29 | * Revision 4.0 91/03/20 01:39:16 lwall |
30 | * 4.0 baseline. |
a687059c |
31 | * |
32 | */ |
f0fcb552 |
33 | /*SUPPRESS 112*/ |
a687059c |
34 | /* |
35 | * regcomp and regexec -- regsub and regerror are not used in perl |
36 | * |
37 | * Copyright (c) 1986 by University of Toronto. |
38 | * Written by Henry Spencer. Not derived from licensed software. |
39 | * |
40 | * Permission is granted to anyone to use this software for any |
41 | * purpose on any computer system, and to redistribute it freely, |
42 | * subject to the following restrictions: |
43 | * |
44 | * 1. The author is not responsible for the consequences of use of |
45 | * this software, no matter how awful, even if they arise |
46 | * from defects in it. |
47 | * |
48 | * 2. The origin of this software must not be misrepresented, either |
49 | * by explicit claim or by omission. |
50 | * |
51 | * 3. Altered versions must be plainly marked as such, and must not |
52 | * be misrepresented as being the original software. |
53 | * |
54 | **** Alterations to Henry's code are... |
55 | **** |
9ef589d8 |
56 | **** Copyright (c) 1991, Larry Wall |
a687059c |
57 | **** |
9ef589d8 |
58 | **** You may distribute under the terms of either the GNU General Public |
59 | **** License or the Artistic License, as specified in the README file. |
a687059c |
60 | * |
61 | * Beware that some of this code is subtly aware of the way operator |
62 | * precedence is structured in regular expressions. Serious changes in |
63 | * regular-expression syntax might require a total rethink. |
64 | */ |
65 | #include "EXTERN.h" |
66 | #include "perl.h" |
67 | #include "regcomp.h" |
68 | |
69 | #ifndef STATIC |
70 | #define STATIC static |
71 | #endif |
72 | |
73 | #ifdef DEBUGGING |
74 | int regnarrate = 0; |
75 | #endif |
76 | |
77 | /* |
78 | * regexec and friends |
79 | */ |
80 | |
81 | /* |
82 | * Global work variables for regexec(). |
83 | */ |
84 | static char *regprecomp; |
85 | static char *reginput; /* String-input pointer. */ |
ac58e20f |
86 | static char regprev; /* char before regbol, \n if none */ |
a687059c |
87 | static char *regbol; /* Beginning of input, for ^ check. */ |
88 | static char *regeol; /* End of input, for $ check. */ |
89 | static char **regstartp; /* Pointer to startp array. */ |
90 | static char **regendp; /* Ditto for endp. */ |
91 | static char *reglastparen; /* Similarly for lastparen. */ |
92 | static char *regtill; |
93 | |
35c8bce7 |
94 | static int regmyp_size = 0; |
95 | static char **regmystartp = Null(char**); |
96 | static char **regmyendp = Null(char**); |
a687059c |
97 | |
98 | /* |
99 | * Forwards. |
100 | */ |
101 | STATIC int regtry(); |
102 | STATIC int regmatch(); |
103 | STATIC int regrepeat(); |
104 | |
105 | extern int multiline; |
106 | |
107 | /* |
108 | - regexec - match a regexp against a string |
109 | */ |
110 | int |
111 | regexec(prog, stringarg, strend, strbeg, minend, screamer, safebase) |
112 | register regexp *prog; |
113 | char *stringarg; |
114 | register char *strend; /* pointer to null at end of string */ |
115 | char *strbeg; /* real beginning of string */ |
116 | int minend; /* end of match must be at least minend after stringarg */ |
117 | STR *screamer; |
118 | int safebase; /* no need to remember string in subbase */ |
119 | { |
120 | register char *s; |
121 | register int i; |
122 | register char *c; |
123 | register char *string = stringarg; |
124 | register int tmp; |
125 | int minlen = 0; /* must match at least this many chars */ |
126 | int dontbother = 0; /* how many characters not to try at end */ |
a687059c |
127 | |
128 | /* Be paranoid... */ |
129 | if (prog == NULL || string == NULL) { |
130 | fatal("NULL regexp parameter"); |
131 | return(0); |
132 | } |
133 | |
ac58e20f |
134 | if (string == strbeg) /* is ^ valid at stringarg? */ |
135 | regprev = '\n'; |
0a12ae7d |
136 | else { |
ac58e20f |
137 | regprev = stringarg[-1]; |
0a12ae7d |
138 | if (!multiline && regprev == '\n') |
139 | regprev = '\0'; /* force ^ to NOT match */ |
140 | } |
a687059c |
141 | regprecomp = prog->precomp; |
142 | /* Check validity of program. */ |
143 | if (UCHARAT(prog->program) != MAGIC) { |
144 | FAIL("corrupted regexp program"); |
145 | } |
146 | |
147 | if (prog->do_folding) { |
a687059c |
148 | i = strend - string; |
149 | New(1101,c,i+1,char); |
2b69d0c2 |
150 | Copy(string, c, i+1, char); |
a687059c |
151 | string = c; |
152 | strend = string + i; |
153 | for (s = string; s < strend; s++) |
ac58e20f |
154 | if (isUPPER(*s)) |
a687059c |
155 | *s = tolower(*s); |
156 | } |
157 | |
158 | /* If there is a "must appear" string, look for it. */ |
159 | s = string; |
34de22dd |
160 | if (prog->regmust != Nullstr && |
9ef589d8 |
161 | (!(prog->reganch & ROPT_ANCH) |
162 | || (multiline && prog->regback >= 0)) ) { |
ac58e20f |
163 | if (stringarg == strbeg && screamer) { |
a687059c |
164 | if (screamfirst[prog->regmust->str_rare] >= 0) |
165 | s = screaminstr(screamer,prog->regmust); |
166 | else |
167 | s = Nullch; |
168 | } |
169 | #ifndef lint |
170 | else |
171 | s = fbminstr((unsigned char*)s, (unsigned char*)strend, |
172 | prog->regmust); |
173 | #endif |
174 | if (!s) { |
175 | ++prog->regmust->str_u.str_useful; /* hooray */ |
176 | goto phooey; /* not present */ |
177 | } |
178 | else if (prog->regback >= 0) { |
179 | s -= prog->regback; |
180 | if (s < string) |
181 | s = string; |
182 | minlen = prog->regback + prog->regmust->str_cur; |
183 | } |
184 | else if (--prog->regmust->str_u.str_useful < 0) { /* boo */ |
185 | str_free(prog->regmust); |
186 | prog->regmust = Nullstr; /* disable regmust */ |
187 | s = string; |
188 | } |
189 | else { |
190 | s = string; |
191 | minlen = prog->regmust->str_cur; |
192 | } |
193 | } |
194 | |
195 | /* Mark beginning of line for ^ . */ |
ac58e20f |
196 | regbol = string; |
a687059c |
197 | |
198 | /* Mark end of line for $ (and such) */ |
199 | regeol = strend; |
200 | |
201 | /* see how far we have to get to not match where we matched before */ |
202 | regtill = string+minend; |
203 | |
35c8bce7 |
204 | /* Allocate our backreference arrays */ |
205 | if ( regmyp_size < prog->nparens + 1 ) { |
206 | /* Allocate or enlarge the arrays */ |
207 | regmyp_size = prog->nparens + 1; |
208 | if ( regmyp_size < 10 ) regmyp_size = 10; /* minimum */ |
209 | if ( regmystartp ) { |
210 | /* reallocate larger */ |
211 | Renew(regmystartp,regmyp_size,char*); |
212 | Renew(regmyendp, regmyp_size,char*); |
213 | } |
214 | else { |
215 | /* Initial allocation */ |
216 | New(1102,regmystartp,regmyp_size,char*); |
217 | New(1102,regmyendp, regmyp_size,char*); |
218 | } |
219 | |
220 | } |
221 | |
a687059c |
222 | /* Simplest case: anchored match need be tried only once. */ |
223 | /* [unless multiline is set] */ |
9ef589d8 |
224 | if (prog->reganch & ROPT_ANCH) { |
a687059c |
225 | if (regtry(prog, string)) |
226 | goto got_it; |
f0fcb552 |
227 | else if (multiline || (prog->reganch & ROPT_IMPLICIT)) { |
a687059c |
228 | if (minlen) |
229 | dontbother = minlen - 1; |
230 | strend -= dontbother; |
231 | /* for multiline we only have to try after newlines */ |
232 | if (s > string) |
233 | s--; |
00bf170e |
234 | while (s < strend) { |
235 | if (*s++ == '\n') { |
236 | if (s < strend && regtry(prog, s)) |
a687059c |
237 | goto got_it; |
238 | } |
239 | } |
240 | } |
241 | goto phooey; |
242 | } |
243 | |
244 | /* Messy cases: unanchored match. */ |
245 | if (prog->regstart) { |
9ef589d8 |
246 | if (prog->reganch & ROPT_SKIP) { /* we have /x+whatever/ */ |
00bf170e |
247 | /* it must be a one character string */ |
248 | i = prog->regstart->str_ptr[0]; |
249 | while (s < strend) { |
250 | if (*s == i) { |
251 | if (regtry(prog, s)) |
252 | goto got_it; |
253 | s++; |
254 | while (s < strend && *s == i) |
255 | s++; |
256 | } |
257 | s++; |
258 | } |
259 | } |
260 | else if (prog->regstart->str_pok == 3) { |
261 | /* We know what string it must start with. */ |
a687059c |
262 | #ifndef lint |
263 | while ((s = fbminstr((unsigned char*)s, |
264 | (unsigned char*)strend, prog->regstart)) != NULL) |
265 | #else |
266 | while (s = Nullch) |
267 | #endif |
268 | { |
269 | if (regtry(prog, s)) |
270 | goto got_it; |
271 | s++; |
272 | } |
273 | } |
274 | else { |
275 | c = prog->regstart->str_ptr; |
276 | while ((s = ninstr(s, strend, |
277 | c, c + prog->regstart->str_cur )) != NULL) { |
278 | if (regtry(prog, s)) |
279 | goto got_it; |
280 | s++; |
281 | } |
282 | } |
283 | goto phooey; |
284 | } |
f0fcb552 |
285 | /*SUPPRESS 560*/ |
a687059c |
286 | if (c = prog->regstclass) { |
9ef589d8 |
287 | int doevery = (prog->reganch & ROPT_SKIP) == 0; |
00bf170e |
288 | |
a687059c |
289 | if (minlen) |
290 | dontbother = minlen - 1; |
291 | strend -= dontbother; /* don't bother with what can't match */ |
00bf170e |
292 | tmp = 1; |
a687059c |
293 | /* We know what class it must start with. */ |
294 | switch (OP(c)) { |
00bf170e |
295 | case ANYOF: |
a687059c |
296 | c = OPERAND(c); |
297 | while (s < strend) { |
ac58e20f |
298 | i = UCHARAT(s); |
00bf170e |
299 | if (!(c[i >> 3] & (1 << (i&7)))) { |
300 | if (tmp && regtry(prog, s)) |
a687059c |
301 | goto got_it; |
00bf170e |
302 | else |
303 | tmp = doevery; |
304 | } |
305 | else |
306 | tmp = 1; |
a687059c |
307 | s++; |
308 | } |
309 | break; |
310 | case BOUND: |
311 | if (minlen) |
312 | dontbother++,strend--; |
313 | if (s != string) { |
314 | i = s[-1]; |
ac58e20f |
315 | tmp = isALNUM(i); |
a687059c |
316 | } |
317 | else |
ac58e20f |
318 | tmp = isALNUM(regprev); /* assume not alphanumeric */ |
a687059c |
319 | while (s < strend) { |
320 | i = *s; |
ac58e20f |
321 | if (tmp != isALNUM(i)) { |
a687059c |
322 | tmp = !tmp; |
323 | if (regtry(prog, s)) |
324 | goto got_it; |
325 | } |
326 | s++; |
327 | } |
ae986130 |
328 | if ((minlen || tmp) && regtry(prog,s)) |
a687059c |
329 | goto got_it; |
330 | break; |
331 | case NBOUND: |
332 | if (minlen) |
333 | dontbother++,strend--; |
334 | if (s != string) { |
335 | i = s[-1]; |
ac58e20f |
336 | tmp = isALNUM(i); |
a687059c |
337 | } |
338 | else |
ac58e20f |
339 | tmp = isALNUM(regprev); /* assume not alphanumeric */ |
a687059c |
340 | while (s < strend) { |
341 | i = *s; |
ac58e20f |
342 | if (tmp != isALNUM(i)) |
a687059c |
343 | tmp = !tmp; |
344 | else if (regtry(prog, s)) |
345 | goto got_it; |
346 | s++; |
347 | } |
ae986130 |
348 | if ((minlen || !tmp) && regtry(prog,s)) |
a687059c |
349 | goto got_it; |
350 | break; |
351 | case ALNUM: |
352 | while (s < strend) { |
353 | i = *s; |
00bf170e |
354 | if (isALNUM(i)) { |
355 | if (tmp && regtry(prog, s)) |
a687059c |
356 | goto got_it; |
00bf170e |
357 | else |
358 | tmp = doevery; |
359 | } |
360 | else |
361 | tmp = 1; |
a687059c |
362 | s++; |
363 | } |
364 | break; |
365 | case NALNUM: |
366 | while (s < strend) { |
367 | i = *s; |
00bf170e |
368 | if (!isALNUM(i)) { |
369 | if (tmp && regtry(prog, s)) |
a687059c |
370 | goto got_it; |
00bf170e |
371 | else |
372 | tmp = doevery; |
373 | } |
374 | else |
375 | tmp = 1; |
a687059c |
376 | s++; |
377 | } |
378 | break; |
379 | case SPACE: |
380 | while (s < strend) { |
00bf170e |
381 | if (isSPACE(*s)) { |
382 | if (tmp && regtry(prog, s)) |
a687059c |
383 | goto got_it; |
00bf170e |
384 | else |
385 | tmp = doevery; |
386 | } |
387 | else |
388 | tmp = 1; |
a687059c |
389 | s++; |
390 | } |
391 | break; |
392 | case NSPACE: |
393 | while (s < strend) { |
00bf170e |
394 | if (!isSPACE(*s)) { |
395 | if (tmp && regtry(prog, s)) |
a687059c |
396 | goto got_it; |
00bf170e |
397 | else |
398 | tmp = doevery; |
399 | } |
400 | else |
401 | tmp = 1; |
a687059c |
402 | s++; |
403 | } |
404 | break; |
405 | case DIGIT: |
406 | while (s < strend) { |
00bf170e |
407 | if (isDIGIT(*s)) { |
408 | if (tmp && regtry(prog, s)) |
a687059c |
409 | goto got_it; |
00bf170e |
410 | else |
411 | tmp = doevery; |
412 | } |
413 | else |
414 | tmp = 1; |
a687059c |
415 | s++; |
416 | } |
417 | break; |
418 | case NDIGIT: |
419 | while (s < strend) { |
00bf170e |
420 | if (!isDIGIT(*s)) { |
421 | if (tmp && regtry(prog, s)) |
a687059c |
422 | goto got_it; |
00bf170e |
423 | else |
424 | tmp = doevery; |
425 | } |
426 | else |
427 | tmp = 1; |
a687059c |
428 | s++; |
429 | } |
430 | break; |
431 | } |
432 | } |
433 | else { |
663a0e37 |
434 | if (minlen) |
435 | dontbother = minlen - 1; |
a687059c |
436 | strend -= dontbother; |
437 | /* We don't know much -- general case. */ |
438 | do { |
439 | if (regtry(prog, s)) |
440 | goto got_it; |
441 | } while (s++ < strend); |
442 | } |
443 | |
444 | /* Failure. */ |
445 | goto phooey; |
446 | |
447 | got_it: |
2b69d0c2 |
448 | prog->subbeg = strbeg; |
449 | prog->subend = strend; |
a687059c |
450 | if ((!safebase && (prog->nparens || sawampersand)) || prog->do_folding){ |
451 | strend += dontbother; /* uncheat */ |
452 | if (safebase) /* no need for $digit later */ |
453 | s = strbeg; |
454 | else if (strbeg != prog->subbase) { |
455 | i = strend - string + (stringarg - strbeg); |
456 | s = nsavestr(strbeg,i); /* so $digit will work later */ |
457 | if (prog->subbase) |
458 | Safefree(prog->subbase); |
9ef589d8 |
459 | prog->subbeg = prog->subbase = s; |
00bf170e |
460 | prog->subend = s+i; |
a687059c |
461 | } |
2b69d0c2 |
462 | else { |
463 | i = strend - string + (stringarg - strbeg); |
464 | prog->subbeg = s = prog->subbase; |
465 | prog->subend = s+i; |
466 | } |
a687059c |
467 | s += (stringarg - strbeg); |
468 | for (i = 0; i <= prog->nparens; i++) { |
469 | if (prog->endp[i]) { |
470 | prog->startp[i] = s + (prog->startp[i] - string); |
471 | prog->endp[i] = s + (prog->endp[i] - string); |
472 | } |
473 | } |
474 | if (prog->do_folding) |
475 | Safefree(string); |
476 | } |
477 | return(1); |
478 | |
479 | phooey: |
480 | if (prog->do_folding) |
481 | Safefree(string); |
482 | return(0); |
483 | } |
484 | |
485 | /* |
486 | - regtry - try match at specific point |
487 | */ |
488 | static int /* 0 failure, 1 success */ |
489 | regtry(prog, string) |
490 | regexp *prog; |
491 | char *string; |
492 | { |
493 | register int i; |
494 | register char **sp; |
495 | register char **ep; |
496 | |
497 | reginput = string; |
498 | regstartp = prog->startp; |
499 | regendp = prog->endp; |
500 | reglastparen = &prog->lastparen; |
501 | prog->lastparen = 0; |
502 | |
503 | sp = prog->startp; |
504 | ep = prog->endp; |
505 | if (prog->nparens) { |
fe14fcc3 |
506 | for (i = prog->nparens; i >= 0; i--) { |
a687059c |
507 | *sp++ = NULL; |
508 | *ep++ = NULL; |
509 | } |
510 | } |
511 | if (regmatch(prog->program + 1) && reginput >= regtill) { |
512 | prog->startp[0] = string; |
513 | prog->endp[0] = reginput; |
514 | return(1); |
515 | } else |
516 | return(0); |
517 | } |
518 | |
519 | /* |
520 | - regmatch - main matching routine |
521 | * |
522 | * Conceptually the strategy is simple: check to see whether the current |
523 | * node matches, call self recursively to see whether the rest matches, |
524 | * and then act accordingly. In practice we make some effort to avoid |
525 | * recursion, in particular by going through "ordinary" nodes (that don't |
526 | * need to know whether the rest of the match failed) by a loop instead of |
527 | * by recursion. |
528 | */ |
529 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
530 | * maybe save a little bit of pushing and popping on the stack. It also takes |
531 | * advantage of machines that use a register save mask on subroutine entry. |
532 | */ |
533 | static int /* 0 failure, 1 success */ |
534 | regmatch(prog) |
535 | char *prog; |
536 | { |
537 | register char *scan; /* Current node. */ |
538 | char *next; /* Next node. */ |
539 | register int nextchar; |
540 | register int n; /* no or next */ |
541 | register int ln; /* len or last */ |
542 | register char *s; /* operand or save */ |
543 | register char *locinput = reginput; |
544 | |
545 | nextchar = *locinput; |
546 | scan = prog; |
547 | #ifdef DEBUGGING |
548 | if (scan != NULL && regnarrate) |
549 | fprintf(stderr, "%s(\n", regprop(scan)); |
550 | #endif |
551 | while (scan != NULL) { |
552 | #ifdef DEBUGGING |
553 | if (regnarrate) |
554 | fprintf(stderr, "%s...\n", regprop(scan)); |
555 | #endif |
556 | |
557 | #ifdef REGALIGN |
558 | next = scan + NEXT(scan); |
559 | if (next == scan) |
560 | next = NULL; |
561 | #else |
562 | next = regnext(scan); |
563 | #endif |
564 | |
565 | switch (OP(scan)) { |
566 | case BOL: |
ac58e20f |
567 | if (locinput == regbol ? regprev == '\n' : |
a687059c |
568 | ((nextchar || locinput < regeol) && |
569 | locinput[-1] == '\n') ) |
570 | { |
fe14fcc3 |
571 | /* regtill = regbol; */ |
a687059c |
572 | break; |
573 | } |
574 | return(0); |
575 | case EOL: |
576 | if ((nextchar || locinput < regeol) && nextchar != '\n') |
577 | return(0); |
00bf170e |
578 | if (!multiline && regeol - locinput > 1) |
579 | return 0; |
fe14fcc3 |
580 | /* regtill = regbol; */ |
a687059c |
581 | break; |
582 | case ANY: |
583 | if ((nextchar == '\0' && locinput >= regeol) || |
584 | nextchar == '\n') |
585 | return(0); |
586 | nextchar = *++locinput; |
587 | break; |
588 | case EXACTLY: |
589 | s = OPERAND(scan); |
590 | ln = *s++; |
591 | /* Inline the first character, for speed. */ |
592 | if (*s != nextchar) |
593 | return(0); |
00bf170e |
594 | if (regeol - locinput < ln) |
a687059c |
595 | return 0; |
596 | if (ln > 1 && bcmp(s, locinput, ln) != 0) |
597 | return(0); |
598 | locinput += ln; |
599 | nextchar = *locinput; |
600 | break; |
601 | case ANYOF: |
a687059c |
602 | s = OPERAND(scan); |
603 | if (nextchar < 0) |
604 | nextchar = UCHARAT(locinput); |
605 | if (s[nextchar >> 3] & (1 << (nextchar&7))) |
606 | return(0); |
34de22dd |
607 | if (!nextchar && locinput >= regeol) |
a687059c |
608 | return 0; |
34de22dd |
609 | nextchar = *++locinput; |
a687059c |
610 | break; |
611 | case ALNUM: |
612 | if (!nextchar) |
613 | return(0); |
ac58e20f |
614 | if (!isALNUM(nextchar)) |
a687059c |
615 | return(0); |
616 | nextchar = *++locinput; |
617 | break; |
618 | case NALNUM: |
619 | if (!nextchar && locinput >= regeol) |
620 | return(0); |
ac58e20f |
621 | if (isALNUM(nextchar)) |
a687059c |
622 | return(0); |
623 | nextchar = *++locinput; |
624 | break; |
625 | case NBOUND: |
626 | case BOUND: |
627 | if (locinput == regbol) /* was last char in word? */ |
ac58e20f |
628 | ln = isALNUM(regprev); |
a687059c |
629 | else |
ac58e20f |
630 | ln = isALNUM(locinput[-1]); |
631 | n = isALNUM(nextchar); /* is next char in word? */ |
a687059c |
632 | if ((ln == n) == (OP(scan) == BOUND)) |
633 | return(0); |
634 | break; |
635 | case SPACE: |
636 | if (!nextchar && locinput >= regeol) |
637 | return(0); |
ac58e20f |
638 | if (!isSPACE(nextchar)) |
a687059c |
639 | return(0); |
640 | nextchar = *++locinput; |
641 | break; |
642 | case NSPACE: |
643 | if (!nextchar) |
644 | return(0); |
ac58e20f |
645 | if (isSPACE(nextchar)) |
a687059c |
646 | return(0); |
647 | nextchar = *++locinput; |
648 | break; |
649 | case DIGIT: |
ac58e20f |
650 | if (!isDIGIT(nextchar)) |
a687059c |
651 | return(0); |
652 | nextchar = *++locinput; |
653 | break; |
654 | case NDIGIT: |
655 | if (!nextchar && locinput >= regeol) |
656 | return(0); |
ac58e20f |
657 | if (isDIGIT(nextchar)) |
a687059c |
658 | return(0); |
659 | nextchar = *++locinput; |
660 | break; |
661 | case REF: |
fe14fcc3 |
662 | n = ARG1(scan); /* which paren pair */ |
a687059c |
663 | s = regmystartp[n]; |
664 | if (!s) |
665 | return(0); |
666 | if (!regmyendp[n]) |
667 | return(0); |
668 | if (s == regmyendp[n]) |
669 | break; |
670 | /* Inline the first character, for speed. */ |
671 | if (*s != nextchar) |
672 | return(0); |
673 | ln = regmyendp[n] - s; |
674 | if (locinput + ln > regeol) |
675 | return 0; |
676 | if (ln > 1 && bcmp(s, locinput, ln) != 0) |
677 | return(0); |
678 | locinput += ln; |
679 | nextchar = *locinput; |
680 | break; |
681 | |
682 | case NOTHING: |
683 | break; |
684 | case BACK: |
685 | break; |
fe14fcc3 |
686 | case OPEN: |
687 | n = ARG1(scan); /* which paren pair */ |
a687059c |
688 | reginput = locinput; |
689 | |
690 | regmystartp[n] = locinput; /* for REF */ |
691 | if (regmatch(next)) { |
692 | /* |
693 | * Don't set startp if some later |
694 | * invocation of the same parentheses |
695 | * already has. |
696 | */ |
697 | if (regstartp[n] == NULL) |
698 | regstartp[n] = locinput; |
699 | return(1); |
700 | } else |
701 | return(0); |
702 | /* NOTREACHED */ |
fe14fcc3 |
703 | case CLOSE: { |
704 | n = ARG1(scan); /* which paren pair */ |
a687059c |
705 | reginput = locinput; |
706 | |
707 | regmyendp[n] = locinput; /* for REF */ |
708 | if (regmatch(next)) { |
709 | /* |
710 | * Don't set endp if some later |
711 | * invocation of the same parentheses |
712 | * already has. |
713 | */ |
714 | if (regendp[n] == NULL) { |
715 | regendp[n] = locinput; |
716 | if (n > *reglastparen) |
717 | *reglastparen = n; |
718 | } |
719 | return(1); |
720 | } else |
721 | return(0); |
722 | } |
723 | /*NOTREACHED*/ |
724 | case BRANCH: { |
725 | if (OP(next) != BRANCH) /* No choice. */ |
726 | next = NEXTOPER(scan); /* Avoid recursion. */ |
727 | else { |
728 | do { |
729 | reginput = locinput; |
730 | if (regmatch(NEXTOPER(scan))) |
731 | return(1); |
732 | #ifdef REGALIGN |
f0fcb552 |
733 | /*SUPPRESS 560*/ |
a687059c |
734 | if (n = NEXT(scan)) |
735 | scan += n; |
736 | else |
737 | scan = NULL; |
738 | #else |
739 | scan = regnext(scan); |
740 | #endif |
741 | } while (scan != NULL && OP(scan) == BRANCH); |
742 | return(0); |
743 | /* NOTREACHED */ |
744 | } |
745 | } |
746 | break; |
00bf170e |
747 | case CURLY: |
748 | ln = ARG1(scan); /* min to match */ |
749 | n = ARG2(scan); /* max to match */ |
750 | scan = NEXTOPER(scan) + 4; |
751 | goto repeat; |
a687059c |
752 | case STAR: |
00bf170e |
753 | ln = 0; |
2b69d0c2 |
754 | n = 32767; |
00bf170e |
755 | scan = NEXTOPER(scan); |
756 | goto repeat; |
a687059c |
757 | case PLUS: |
758 | /* |
759 | * Lookahead to avoid useless match attempts |
760 | * when we know what character comes next. |
761 | */ |
00bf170e |
762 | ln = 1; |
2b69d0c2 |
763 | n = 32767; |
00bf170e |
764 | scan = NEXTOPER(scan); |
765 | repeat: |
a687059c |
766 | if (OP(next) == EXACTLY) |
767 | nextchar = *(OPERAND(next)+1); |
768 | else |
769 | nextchar = -1000; |
a687059c |
770 | reginput = locinput; |
00bf170e |
771 | n = regrepeat(scan, n); |
0a12ae7d |
772 | if (!multiline && OP(next) == EOL && ln < n) |
00bf170e |
773 | ln = n; /* why back off? */ |
a687059c |
774 | while (n >= ln) { |
775 | /* If it could work, try it. */ |
776 | if (nextchar == -1000 || *reginput == nextchar) |
777 | if (regmatch(next)) |
778 | return(1); |
779 | /* Couldn't or didn't -- back up. */ |
780 | n--; |
781 | reginput = locinput + n; |
782 | } |
783 | return(0); |
784 | case END: |
785 | reginput = locinput; /* put where regtry can find it */ |
786 | return(1); /* Success! */ |
787 | default: |
788 | printf("%x %d\n",scan,scan[1]); |
789 | FAIL("regexp memory corruption"); |
790 | } |
791 | |
792 | scan = next; |
793 | } |
794 | |
795 | /* |
796 | * We get here only if there's trouble -- normally "case END" is |
797 | * the terminating point. |
798 | */ |
799 | FAIL("corrupted regexp pointers"); |
800 | /*NOTREACHED*/ |
801 | #ifdef lint |
802 | return 0; |
803 | #endif |
804 | } |
805 | |
806 | /* |
807 | - regrepeat - repeatedly match something simple, report how many |
808 | */ |
809 | /* |
810 | * [This routine now assumes that it will only match on things of length 1. |
811 | * That was true before, but now we assume scan - reginput is the count, |
812 | * rather than incrementing count on every character.] |
813 | */ |
814 | static int |
00bf170e |
815 | regrepeat(p, max) |
a687059c |
816 | char *p; |
00bf170e |
817 | int max; |
a687059c |
818 | { |
819 | register char *scan; |
820 | register char *opnd; |
821 | register int c; |
822 | register char *loceol = regeol; |
823 | |
824 | scan = reginput; |
2b69d0c2 |
825 | if (max != 32767 && max < loceol - scan) |
00bf170e |
826 | loceol = scan + max; |
a687059c |
827 | opnd = OPERAND(p); |
828 | switch (OP(p)) { |
829 | case ANY: |
830 | while (scan < loceol && *scan != '\n') |
831 | scan++; |
832 | break; |
833 | case EXACTLY: /* length of string is 1 */ |
834 | opnd++; |
835 | while (scan < loceol && *opnd == *scan) |
836 | scan++; |
837 | break; |
838 | case ANYOF: |
a687059c |
839 | c = UCHARAT(scan); |
840 | while (scan < loceol && !(opnd[c >> 3] & (1 << (c & 7)))) { |
841 | scan++; |
842 | c = UCHARAT(scan); |
843 | } |
844 | break; |
845 | case ALNUM: |
0a12ae7d |
846 | while (scan < loceol && isALNUM(*scan)) |
a687059c |
847 | scan++; |
848 | break; |
849 | case NALNUM: |
ac58e20f |
850 | while (scan < loceol && !isALNUM(*scan)) |
a687059c |
851 | scan++; |
852 | break; |
853 | case SPACE: |
ac58e20f |
854 | while (scan < loceol && isSPACE(*scan)) |
a687059c |
855 | scan++; |
856 | break; |
857 | case NSPACE: |
ac58e20f |
858 | while (scan < loceol && !isSPACE(*scan)) |
a687059c |
859 | scan++; |
860 | break; |
861 | case DIGIT: |
0a12ae7d |
862 | while (scan < loceol && isDIGIT(*scan)) |
a687059c |
863 | scan++; |
864 | break; |
865 | case NDIGIT: |
ac58e20f |
866 | while (scan < loceol && !isDIGIT(*scan)) |
a687059c |
867 | scan++; |
868 | break; |
869 | default: /* Oh dear. Called inappropriately. */ |
870 | FAIL("internal regexp foulup"); |
871 | /* NOTREACHED */ |
872 | } |
873 | |
874 | c = scan - reginput; |
875 | reginput = scan; |
876 | |
877 | return(c); |
878 | } |
879 | |
880 | /* |
881 | - regnext - dig the "next" pointer out of a node |
882 | * |
883 | * [Note, when REGALIGN is defined there are two places in regmatch() |
884 | * that bypass this code for speed.] |
885 | */ |
886 | char * |
887 | regnext(p) |
888 | register char *p; |
889 | { |
890 | register int offset; |
891 | |
892 | if (p == ®dummy) |
893 | return(NULL); |
894 | |
895 | offset = NEXT(p); |
896 | if (offset == 0) |
897 | return(NULL); |
898 | |
899 | #ifdef REGALIGN |
900 | return(p+offset); |
901 | #else |
902 | if (OP(p) == BACK) |
903 | return(p-offset); |
904 | else |
905 | return(p+offset); |
906 | #endif |
907 | } |