Commit | Line | Data |
a0d0e21e |
1 | /* regexec.c |
2 | */ |
3 | |
4 | /* |
5 | * "One Ring to rule them all, One Ring to find them..." |
6 | */ |
7 | |
a687059c |
8 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
9 | * confused with the original package (see point 3 below). Thanks, Henry! |
10 | */ |
11 | |
12 | /* Additional note: this code is very heavily munged from Henry's version |
13 | * in places. In some spots I've traded clarity for efficiency, so don't |
14 | * blame Henry for some of the lack of readability. |
15 | */ |
16 | |
e50aee73 |
17 | /* The names of the functions have been changed from regcomp and |
18 | * regexec to pregcomp and pregexec in order to avoid conflicts |
19 | * with the POSIX routines of the same names. |
20 | */ |
21 | |
f0fcb552 |
22 | /*SUPPRESS 112*/ |
a687059c |
23 | /* |
e50aee73 |
24 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c |
25 | * |
26 | * Copyright (c) 1986 by University of Toronto. |
27 | * Written by Henry Spencer. Not derived from licensed software. |
28 | * |
29 | * Permission is granted to anyone to use this software for any |
30 | * purpose on any computer system, and to redistribute it freely, |
31 | * subject to the following restrictions: |
32 | * |
33 | * 1. The author is not responsible for the consequences of use of |
34 | * this software, no matter how awful, even if they arise |
35 | * from defects in it. |
36 | * |
37 | * 2. The origin of this software must not be misrepresented, either |
38 | * by explicit claim or by omission. |
39 | * |
40 | * 3. Altered versions must be plainly marked as such, and must not |
41 | * be misrepresented as being the original software. |
42 | * |
43 | **** Alterations to Henry's code are... |
44 | **** |
9607fc9c |
45 | **** Copyright (c) 1991-1997, Larry Wall |
a687059c |
46 | **** |
9ef589d8 |
47 | **** You may distribute under the terms of either the GNU General Public |
48 | **** License or the Artistic License, as specified in the README file. |
a687059c |
49 | * |
50 | * Beware that some of this code is subtly aware of the way operator |
51 | * precedence is structured in regular expressions. Serious changes in |
52 | * regular-expression syntax might require a total rethink. |
53 | */ |
54 | #include "EXTERN.h" |
55 | #include "perl.h" |
56 | #include "regcomp.h" |
57 | |
58 | #ifndef STATIC |
59 | #define STATIC static |
60 | #endif |
61 | |
62 | #ifdef DEBUGGING |
a0d0e21e |
63 | static I32 regnarrate = 0; |
64 | static char* regprogram = 0; |
a687059c |
65 | #endif |
66 | |
a0d0e21e |
67 | /* Current curly descriptor */ |
68 | typedef struct curcur CURCUR; |
69 | struct curcur { |
70 | int parenfloor; /* how far back to strip paren data */ |
71 | int cur; /* how many instances of scan we've matched */ |
72 | int min; /* the minimal number of scans to match */ |
73 | int max; /* the maximal number of scans to match */ |
74 | int minmod; /* whether to work our way up or down */ |
75 | char * scan; /* the thing to match */ |
76 | char * next; /* what has to match after it */ |
77 | char * lastloc; /* where we started matching this scan */ |
78 | CURCUR * oldcc; /* current curly before we started this one */ |
79 | }; |
80 | |
81 | static CURCUR* regcc; |
82 | |
83 | typedef I32 CHECKPOINT; |
84 | |
55497cff |
85 | static CHECKPOINT regcppush _((I32 parenfloor)); |
86 | static char * regcppop _((void)); |
a0d0e21e |
87 | |
55497cff |
88 | static CHECKPOINT |
a0d0e21e |
89 | regcppush(parenfloor) |
90 | I32 parenfloor; |
91 | { |
92 | int retval = savestack_ix; |
93 | int i = (regsize - parenfloor) * 3; |
94 | int p; |
95 | |
96 | SSCHECK(i + 5); |
97 | for (p = regsize; p > parenfloor; p--) { |
98 | SSPUSHPTR(regendp[p]); |
99 | SSPUSHPTR(regstartp[p]); |
100 | SSPUSHINT(p); |
101 | } |
102 | SSPUSHINT(regsize); |
103 | SSPUSHINT(*reglastparen); |
104 | SSPUSHPTR(reginput); |
105 | SSPUSHINT(i + 3); |
106 | SSPUSHINT(SAVEt_REGCONTEXT); |
107 | return retval; |
108 | } |
109 | |
55497cff |
110 | static char * |
a0d0e21e |
111 | regcppop() |
112 | { |
113 | I32 i = SSPOPINT; |
114 | U32 paren = 0; |
115 | char *input; |
116 | char *tmps; |
117 | assert(i == SAVEt_REGCONTEXT); |
118 | i = SSPOPINT; |
119 | input = (char *) SSPOPPTR; |
120 | *reglastparen = SSPOPINT; |
121 | regsize = SSPOPINT; |
122 | for (i -= 3; i > 0; i -= 3) { |
123 | paren = (U32)SSPOPINT; |
124 | regstartp[paren] = (char *) SSPOPPTR; |
125 | tmps = (char*)SSPOPPTR; |
126 | if (paren <= *reglastparen) |
127 | regendp[paren] = tmps; |
128 | } |
129 | for (paren = *reglastparen + 1; paren <= regnpar; paren++) { |
130 | if (paren > regsize) |
131 | regstartp[paren] = Nullch; |
132 | regendp[paren] = Nullch; |
133 | } |
134 | return input; |
135 | } |
136 | |
137 | #define regcpblow(cp) leave_scope(cp) |
138 | |
a687059c |
139 | /* |
e50aee73 |
140 | * pregexec and friends |
a687059c |
141 | */ |
142 | |
143 | /* |
a687059c |
144 | * Forwards. |
145 | */ |
a0d0e21e |
146 | |
147 | static I32 regmatch _((char *prog)); |
148 | static I32 regrepeat _((char *p, I32 max)); |
149 | static I32 regtry _((regexp *prog, char *startpos)); |
bbce6d69 |
150 | static bool reginclass _((char *p, I32 c)); |
151 | |
152 | static bool regtainted; /* tainted information used? */ |
a687059c |
153 | |
154 | /* |
e50aee73 |
155 | - pregexec - match a regexp against a string |
a687059c |
156 | */ |
79072805 |
157 | I32 |
e50aee73 |
158 | pregexec(prog, stringarg, strend, strbeg, minend, screamer, safebase) |
a687059c |
159 | register regexp *prog; |
160 | char *stringarg; |
161 | register char *strend; /* pointer to null at end of string */ |
162 | char *strbeg; /* real beginning of string */ |
79072805 |
163 | I32 minend; /* end of match must be at least minend after stringarg */ |
164 | SV *screamer; |
165 | I32 safebase; /* no need to remember string in subbase */ |
a687059c |
166 | { |
a0d0e21e |
167 | register char *s; |
a0d0e21e |
168 | register char *c; |
169 | register char *startpos = stringarg; |
170 | register I32 tmp; |
171 | I32 minlen = 0; /* must match at least this many chars */ |
172 | I32 dontbother = 0; /* how many characters not to try at end */ |
173 | CURCUR cc; |
a687059c |
174 | |
a0d0e21e |
175 | cc.cur = 0; |
4633a7c4 |
176 | cc.oldcc = 0; |
a0d0e21e |
177 | regcc = &cc; |
178 | |
179 | #ifdef DEBUGGING |
180 | regnarrate = debug & 512; |
181 | regprogram = prog->program; |
182 | #endif |
183 | |
184 | /* Be paranoid... */ |
185 | if (prog == NULL || startpos == NULL) { |
186 | croak("NULL regexp parameter"); |
187 | return 0; |
188 | } |
189 | |
190 | if (startpos == strbeg) /* is ^ valid at stringarg? */ |
191 | regprev = '\n'; |
192 | else { |
193 | regprev = stringarg[-1]; |
194 | if (!multiline && regprev == '\n') |
195 | regprev = '\0'; /* force ^ to NOT match */ |
196 | } |
bbce6d69 |
197 | |
a0d0e21e |
198 | regprecomp = prog->precomp; |
a0d0e21e |
199 | /* Check validity of program. */ |
200 | if (UCHARAT(prog->program) != MAGIC) { |
201 | FAIL("corrupted regexp program"); |
202 | } |
203 | |
bbce6d69 |
204 | regnpar = prog->nparens; |
205 | regtainted = FALSE; |
a0d0e21e |
206 | |
207 | /* If there is a "must appear" string, look for it. */ |
208 | s = startpos; |
209 | if (prog->regmust != Nullsv && |
774d564b |
210 | !(prog->reganch & ROPT_ANCH_GPOS) && |
211 | (!(prog->reganch & ROPT_ANCH_BOL) |
a0d0e21e |
212 | || (multiline && prog->regback >= 0)) ) |
213 | { |
214 | if (stringarg == strbeg && screamer) { |
215 | if (screamfirst[BmRARE(prog->regmust)] >= 0) |
216 | s = screaminstr(screamer,prog->regmust); |
217 | else |
218 | s = Nullch; |
0a12ae7d |
219 | } |
a0d0e21e |
220 | else |
221 | s = fbm_instr((unsigned char*)s, (unsigned char*)strend, |
222 | prog->regmust); |
223 | if (!s) { |
224 | ++BmUSEFUL(prog->regmust); /* hooray */ |
225 | goto phooey; /* not present */ |
a687059c |
226 | } |
a0d0e21e |
227 | else if (prog->regback >= 0) { |
228 | s -= prog->regback; |
229 | if (s < startpos) |
230 | s = startpos; |
231 | minlen = prog->regback + SvCUR(prog->regmust); |
a687059c |
232 | } |
a0d0e21e |
233 | else if (!prog->naughty && --BmUSEFUL(prog->regmust) < 0) { /* boo */ |
234 | SvREFCNT_dec(prog->regmust); |
235 | prog->regmust = Nullsv; /* disable regmust */ |
236 | s = startpos; |
237 | } |
238 | else { |
239 | s = startpos; |
240 | minlen = SvCUR(prog->regmust); |
a687059c |
241 | } |
a0d0e21e |
242 | } |
a687059c |
243 | |
a0d0e21e |
244 | /* Mark beginning of line for ^ . */ |
245 | regbol = startpos; |
a687059c |
246 | |
a0d0e21e |
247 | /* Mark end of line for $ (and such) */ |
248 | regeol = strend; |
a687059c |
249 | |
a0d0e21e |
250 | /* see how far we have to get to not match where we matched before */ |
251 | regtill = startpos+minend; |
a687059c |
252 | |
a0d0e21e |
253 | /* Simplest case: anchored match need be tried only once. */ |
774d564b |
254 | /* [unless only anchor is BOL and multiline is set] */ |
a0d0e21e |
255 | if (prog->reganch & ROPT_ANCH) { |
256 | if (regtry(prog, startpos)) |
257 | goto got_it; |
774d564b |
258 | else if (!(prog->reganch & ROPT_ANCH_GPOS) && |
259 | (multiline || (prog->reganch & ROPT_IMPLICIT))) |
260 | { |
a0d0e21e |
261 | if (minlen) |
262 | dontbother = minlen - 1; |
263 | strend -= dontbother; |
264 | /* for multiline we only have to try after newlines */ |
265 | if (s > startpos) |
266 | s--; |
267 | while (s < strend) { |
268 | if (*s++ == '\n') { |
269 | if (s < strend && regtry(prog, s)) |
270 | goto got_it; |
271 | } |
35c8bce7 |
272 | } |
35c8bce7 |
273 | } |
a0d0e21e |
274 | goto phooey; |
275 | } |
35c8bce7 |
276 | |
a0d0e21e |
277 | /* Messy cases: unanchored match. */ |
278 | if (prog->regstart) { |
279 | if (prog->reganch & ROPT_SKIP) { /* we have /x+whatever/ */ |
280 | /* it must be a one character string */ |
bbce6d69 |
281 | char ch = SvPVX(prog->regstart)[0]; |
a0d0e21e |
282 | while (s < strend) { |
bbce6d69 |
283 | if (*s == ch) { |
a0d0e21e |
284 | if (regtry(prog, s)) |
a687059c |
285 | goto got_it; |
a0d0e21e |
286 | s++; |
bbce6d69 |
287 | while (s < strend && *s == ch) |
a0d0e21e |
288 | s++; |
a687059c |
289 | } |
a0d0e21e |
290 | s++; |
291 | } |
a687059c |
292 | } |
47109bfb |
293 | else if (SvTYPE(prog->regstart) == SVt_PVBM) { |
a0d0e21e |
294 | /* We know what string it must start with. */ |
295 | while ((s = fbm_instr((unsigned char*)s, |
296 | (unsigned char*)strend, prog->regstart)) != NULL) |
297 | { |
298 | if (regtry(prog, s)) |
299 | goto got_it; |
300 | s++; |
301 | } |
302 | } |
47109bfb |
303 | else { /* Optimized fbm_instr: */ |
a0d0e21e |
304 | c = SvPVX(prog->regstart); |
305 | while ((s = ninstr(s, strend, c, c + SvCUR(prog->regstart))) != NULL) |
306 | { |
307 | if (regtry(prog, s)) |
308 | goto got_it; |
309 | s++; |
310 | } |
311 | } |
312 | goto phooey; |
313 | } |
314 | /*SUPPRESS 560*/ |
315 | if (c = prog->regstclass) { |
316 | I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
a687059c |
317 | |
a0d0e21e |
318 | if (minlen) |
319 | dontbother = minlen - 1; |
320 | strend -= dontbother; /* don't bother with what can't match */ |
321 | tmp = 1; |
322 | /* We know what class it must start with. */ |
323 | switch (OP(c)) { |
324 | case ANYOF: |
325 | c = OPERAND(c); |
326 | while (s < strend) { |
bbce6d69 |
327 | if (reginclass(c, *s)) { |
a0d0e21e |
328 | if (tmp && regtry(prog, s)) |
329 | goto got_it; |
330 | else |
331 | tmp = doevery; |
a687059c |
332 | } |
a0d0e21e |
333 | else |
334 | tmp = 1; |
335 | s++; |
336 | } |
337 | break; |
bbce6d69 |
338 | case BOUNDL: |
339 | regtainted = TRUE; |
340 | /* FALL THROUGH */ |
a0d0e21e |
341 | case BOUND: |
342 | if (minlen) |
343 | dontbother++,strend--; |
bbce6d69 |
344 | tmp = (s != startpos) ? UCHARAT(s - 1) : regprev; |
95bac841 |
345 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
346 | while (s < strend) { |
95bac841 |
347 | if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { |
a0d0e21e |
348 | tmp = !tmp; |
349 | if (regtry(prog, s)) |
350 | goto got_it; |
a687059c |
351 | } |
a0d0e21e |
352 | s++; |
353 | } |
354 | if ((minlen || tmp) && regtry(prog,s)) |
355 | goto got_it; |
356 | break; |
bbce6d69 |
357 | case NBOUNDL: |
358 | regtainted = TRUE; |
359 | /* FALL THROUGH */ |
a0d0e21e |
360 | case NBOUND: |
361 | if (minlen) |
362 | dontbother++,strend--; |
bbce6d69 |
363 | tmp = (s != startpos) ? UCHARAT(s - 1) : regprev; |
95bac841 |
364 | tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
365 | while (s < strend) { |
95bac841 |
366 | if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) |
a0d0e21e |
367 | tmp = !tmp; |
368 | else if (regtry(prog, s)) |
369 | goto got_it; |
370 | s++; |
371 | } |
372 | if ((minlen || !tmp) && regtry(prog,s)) |
373 | goto got_it; |
374 | break; |
375 | case ALNUM: |
376 | while (s < strend) { |
bbce6d69 |
377 | if (isALNUM(*s)) { |
378 | if (tmp && regtry(prog, s)) |
379 | goto got_it; |
380 | else |
381 | tmp = doevery; |
382 | } |
383 | else |
384 | tmp = 1; |
385 | s++; |
386 | } |
387 | break; |
388 | case ALNUML: |
389 | regtainted = TRUE; |
390 | while (s < strend) { |
391 | if (isALNUM_LC(*s)) { |
a0d0e21e |
392 | if (tmp && regtry(prog, s)) |
393 | goto got_it; |
a687059c |
394 | else |
a0d0e21e |
395 | tmp = doevery; |
396 | } |
397 | else |
398 | tmp = 1; |
399 | s++; |
400 | } |
401 | break; |
402 | case NALNUM: |
403 | while (s < strend) { |
bbce6d69 |
404 | if (!isALNUM(*s)) { |
405 | if (tmp && regtry(prog, s)) |
406 | goto got_it; |
407 | else |
408 | tmp = doevery; |
409 | } |
410 | else |
411 | tmp = 1; |
412 | s++; |
413 | } |
414 | break; |
415 | case NALNUML: |
416 | regtainted = TRUE; |
417 | while (s < strend) { |
418 | if (!isALNUM_LC(*s)) { |
a0d0e21e |
419 | if (tmp && regtry(prog, s)) |
420 | goto got_it; |
a687059c |
421 | else |
a0d0e21e |
422 | tmp = doevery; |
a687059c |
423 | } |
a0d0e21e |
424 | else |
425 | tmp = 1; |
426 | s++; |
427 | } |
428 | break; |
429 | case SPACE: |
430 | while (s < strend) { |
431 | if (isSPACE(*s)) { |
432 | if (tmp && regtry(prog, s)) |
433 | goto got_it; |
434 | else |
435 | tmp = doevery; |
2304df62 |
436 | } |
a0d0e21e |
437 | else |
438 | tmp = 1; |
439 | s++; |
440 | } |
441 | break; |
bbce6d69 |
442 | case SPACEL: |
443 | regtainted = TRUE; |
444 | while (s < strend) { |
445 | if (isSPACE_LC(*s)) { |
446 | if (tmp && regtry(prog, s)) |
447 | goto got_it; |
448 | else |
449 | tmp = doevery; |
450 | } |
451 | else |
452 | tmp = 1; |
453 | s++; |
454 | } |
455 | break; |
a0d0e21e |
456 | case NSPACE: |
457 | while (s < strend) { |
458 | if (!isSPACE(*s)) { |
459 | if (tmp && regtry(prog, s)) |
460 | goto got_it; |
461 | else |
462 | tmp = doevery; |
a687059c |
463 | } |
a0d0e21e |
464 | else |
465 | tmp = 1; |
466 | s++; |
467 | } |
468 | break; |
bbce6d69 |
469 | case NSPACEL: |
470 | regtainted = TRUE; |
471 | while (s < strend) { |
472 | if (!isSPACE_LC(*s)) { |
473 | if (tmp && regtry(prog, s)) |
474 | goto got_it; |
475 | else |
476 | tmp = doevery; |
477 | } |
478 | else |
479 | tmp = 1; |
480 | s++; |
481 | } |
482 | break; |
a0d0e21e |
483 | case DIGIT: |
484 | while (s < strend) { |
485 | if (isDIGIT(*s)) { |
486 | if (tmp && regtry(prog, s)) |
487 | goto got_it; |
488 | else |
489 | tmp = doevery; |
2b69d0c2 |
490 | } |
a0d0e21e |
491 | else |
492 | tmp = 1; |
493 | s++; |
494 | } |
495 | break; |
496 | case NDIGIT: |
497 | while (s < strend) { |
498 | if (!isDIGIT(*s)) { |
499 | if (tmp && regtry(prog, s)) |
500 | goto got_it; |
501 | else |
502 | tmp = doevery; |
a687059c |
503 | } |
a0d0e21e |
504 | else |
505 | tmp = 1; |
506 | s++; |
507 | } |
508 | break; |
a687059c |
509 | } |
a0d0e21e |
510 | } |
511 | else { |
512 | if (minlen) |
513 | dontbother = minlen - 1; |
514 | strend -= dontbother; |
515 | /* We don't know much -- general case. */ |
516 | do { |
517 | if (regtry(prog, s)) |
518 | goto got_it; |
519 | } while (s++ < strend); |
520 | } |
521 | |
522 | /* Failure. */ |
523 | goto phooey; |
a687059c |
524 | |
a0d0e21e |
525 | got_it: |
420218e7 |
526 | strend += dontbother; /* uncheat */ |
a0d0e21e |
527 | prog->subbeg = strbeg; |
528 | prog->subend = strend; |
bbce6d69 |
529 | prog->exec_tainted = regtainted; |
5f05dabc |
530 | |
531 | /* make sure $`, $&, $', and $digit will work later */ |
532 | if (!safebase && (strbeg != prog->subbase)) { |
bbce6d69 |
533 | I32 i = strend - startpos + (stringarg - strbeg); |
5f05dabc |
534 | s = savepvn(strbeg, i); |
535 | Safefree(prog->subbase); |
536 | prog->subbase = s; |
537 | prog->subbeg = prog->subbase; |
538 | prog->subend = prog->subbase + i; |
539 | s = prog->subbase + (stringarg - strbeg); |
a0d0e21e |
540 | for (i = 0; i <= prog->nparens; i++) { |
541 | if (prog->endp[i]) { |
542 | prog->startp[i] = s + (prog->startp[i] - startpos); |
543 | prog->endp[i] = s + (prog->endp[i] - startpos); |
544 | } |
545 | } |
a0d0e21e |
546 | } |
547 | return 1; |
548 | |
549 | phooey: |
a0d0e21e |
550 | return 0; |
a687059c |
551 | } |
552 | |
553 | /* |
554 | - regtry - try match at specific point |
555 | */ |
79072805 |
556 | static I32 /* 0 failure, 1 success */ |
a0d0e21e |
557 | regtry(prog, startpos) |
a687059c |
558 | regexp *prog; |
a0d0e21e |
559 | char *startpos; |
a687059c |
560 | { |
a0d0e21e |
561 | register I32 i; |
562 | register char **sp; |
563 | register char **ep; |
564 | |
565 | reginput = startpos; |
566 | regstartp = prog->startp; |
567 | regendp = prog->endp; |
568 | reglastparen = &prog->lastparen; |
569 | prog->lastparen = 0; |
570 | regsize = 0; |
571 | |
572 | sp = prog->startp; |
573 | ep = prog->endp; |
574 | if (prog->nparens) { |
575 | for (i = prog->nparens; i >= 0; i--) { |
576 | *sp++ = NULL; |
577 | *ep++ = NULL; |
a687059c |
578 | } |
a0d0e21e |
579 | } |
580 | if (regmatch(prog->program + 1) && reginput >= regtill) { |
581 | prog->startp[0] = startpos; |
582 | prog->endp[0] = reginput; |
583 | return 1; |
584 | } |
585 | else |
586 | return 0; |
a687059c |
587 | } |
588 | |
589 | /* |
590 | - regmatch - main matching routine |
591 | * |
592 | * Conceptually the strategy is simple: check to see whether the current |
593 | * node matches, call self recursively to see whether the rest matches, |
594 | * and then act accordingly. In practice we make some effort to avoid |
595 | * recursion, in particular by going through "ordinary" nodes (that don't |
596 | * need to know whether the rest of the match failed) by a loop instead of |
597 | * by recursion. |
598 | */ |
599 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
600 | * maybe save a little bit of pushing and popping on the stack. It also takes |
601 | * advantage of machines that use a register save mask on subroutine entry. |
602 | */ |
79072805 |
603 | static I32 /* 0 failure, 1 success */ |
a687059c |
604 | regmatch(prog) |
605 | char *prog; |
606 | { |
a0d0e21e |
607 | register char *scan; /* Current node. */ |
608 | char *next; /* Next node. */ |
609 | register I32 nextchar; |
610 | register I32 n; /* no or next */ |
611 | register I32 ln; /* len or last */ |
612 | register char *s; /* operand or save */ |
613 | register char *locinput = reginput; |
bbce6d69 |
614 | register I32 c1, c2; /* case fold search */ |
a0d0e21e |
615 | int minmod = 0; |
4633a7c4 |
616 | #ifdef DEBUGGING |
617 | static int regindent = 0; |
618 | regindent++; |
619 | #endif |
a0d0e21e |
620 | |
bbce6d69 |
621 | nextchar = UCHARAT(locinput); |
a0d0e21e |
622 | scan = prog; |
623 | while (scan != NULL) { |
a687059c |
624 | #ifdef DEBUGGING |
4633a7c4 |
625 | #define sayYES goto yes |
626 | #define sayNO goto no |
627 | #define saySAME(x) if (x) goto yes; else goto no |
628 | if (regnarrate) { |
760ac839 |
629 | PerlIO_printf(Perl_debug_log, "%*s%2d%-8.8s\t<%.10s>\n", regindent*2, "", |
a0d0e21e |
630 | scan - regprogram, regprop(scan), locinput); |
4633a7c4 |
631 | } |
632 | #else |
633 | #define sayYES return 1 |
634 | #define sayNO return 0 |
635 | #define saySAME(x) return x |
a687059c |
636 | #endif |
637 | |
638 | #ifdef REGALIGN |
a0d0e21e |
639 | next = scan + NEXT(scan); |
640 | if (next == scan) |
641 | next = NULL; |
a687059c |
642 | #else |
a0d0e21e |
643 | next = regnext(scan); |
a687059c |
644 | #endif |
645 | |
a0d0e21e |
646 | switch (OP(scan)) { |
647 | case BOL: |
648 | if (locinput == regbol |
649 | ? regprev == '\n' |
650 | : ((nextchar || locinput < regeol) && locinput[-1] == '\n') ) |
651 | { |
652 | /* regtill = regbol; */ |
653 | break; |
654 | } |
4633a7c4 |
655 | sayNO; |
a0d0e21e |
656 | case MBOL: |
657 | if (locinput == regbol |
658 | ? regprev == '\n' |
659 | : ((nextchar || locinput < regeol) && locinput[-1] == '\n') ) |
660 | { |
661 | break; |
662 | } |
4633a7c4 |
663 | sayNO; |
a0d0e21e |
664 | case SBOL: |
665 | if (locinput == regbol && regprev == '\n') |
666 | break; |
4633a7c4 |
667 | sayNO; |
774d564b |
668 | case GPOS: |
a0d0e21e |
669 | if (locinput == regbol) |
670 | break; |
4633a7c4 |
671 | sayNO; |
a0d0e21e |
672 | case EOL: |
673 | if (multiline) |
674 | goto meol; |
675 | else |
676 | goto seol; |
677 | case MEOL: |
678 | meol: |
679 | if ((nextchar || locinput < regeol) && nextchar != '\n') |
4633a7c4 |
680 | sayNO; |
a0d0e21e |
681 | break; |
682 | case SEOL: |
683 | seol: |
684 | if ((nextchar || locinput < regeol) && nextchar != '\n') |
4633a7c4 |
685 | sayNO; |
a0d0e21e |
686 | if (regeol - locinput > 1) |
4633a7c4 |
687 | sayNO; |
a0d0e21e |
688 | break; |
689 | case SANY: |
690 | if (!nextchar && locinput >= regeol) |
4633a7c4 |
691 | sayNO; |
bbce6d69 |
692 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
693 | break; |
694 | case ANY: |
695 | if (!nextchar && locinput >= regeol || nextchar == '\n') |
4633a7c4 |
696 | sayNO; |
bbce6d69 |
697 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
698 | break; |
bbce6d69 |
699 | case EXACT: |
a0d0e21e |
700 | s = OPERAND(scan); |
701 | ln = *s++; |
702 | /* Inline the first character, for speed. */ |
95bac841 |
703 | if (UCHARAT(s) != nextchar) |
4633a7c4 |
704 | sayNO; |
a0d0e21e |
705 | if (regeol - locinput < ln) |
4633a7c4 |
706 | sayNO; |
36477c24 |
707 | if (ln > 1 && memNE(s, locinput, ln)) |
4633a7c4 |
708 | sayNO; |
a0d0e21e |
709 | locinput += ln; |
bbce6d69 |
710 | nextchar = UCHARAT(locinput); |
711 | break; |
712 | case EXACTFL: |
713 | regtainted = TRUE; |
714 | /* FALL THROUGH */ |
715 | case EXACTF: |
716 | s = OPERAND(scan); |
717 | ln = *s++; |
718 | /* Inline the first character, for speed. */ |
719 | if (UCHARAT(s) != nextchar && |
720 | UCHARAT(s) != ((OP(scan) == EXACTF) |
721 | ? fold : fold_locale)[nextchar]) |
722 | sayNO; |
723 | if (regeol - locinput < ln) |
724 | sayNO; |
5f05dabc |
725 | if (ln > 1 && (OP(scan) == EXACTF |
726 | ? ibcmp(s, locinput, ln) |
727 | : ibcmp_locale(s, locinput, ln))) |
bbce6d69 |
728 | sayNO; |
729 | locinput += ln; |
730 | nextchar = UCHARAT(locinput); |
a0d0e21e |
731 | break; |
732 | case ANYOF: |
733 | s = OPERAND(scan); |
734 | if (nextchar < 0) |
735 | nextchar = UCHARAT(locinput); |
bbce6d69 |
736 | if (!reginclass(s, nextchar)) |
4633a7c4 |
737 | sayNO; |
a0d0e21e |
738 | if (!nextchar && locinput >= regeol) |
4633a7c4 |
739 | sayNO; |
bbce6d69 |
740 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
741 | break; |
bbce6d69 |
742 | case ALNUML: |
743 | regtainted = TRUE; |
744 | /* FALL THROUGH */ |
a0d0e21e |
745 | case ALNUM: |
746 | if (!nextchar) |
4633a7c4 |
747 | sayNO; |
bbce6d69 |
748 | if (!(OP(scan) == ALNUM |
749 | ? isALNUM(nextchar) : isALNUM_LC(nextchar))) |
4633a7c4 |
750 | sayNO; |
bbce6d69 |
751 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
752 | break; |
bbce6d69 |
753 | case NALNUML: |
754 | regtainted = TRUE; |
755 | /* FALL THROUGH */ |
a0d0e21e |
756 | case NALNUM: |
757 | if (!nextchar && locinput >= regeol) |
4633a7c4 |
758 | sayNO; |
bbce6d69 |
759 | if (OP(scan) == NALNUM |
760 | ? isALNUM(nextchar) : isALNUM_LC(nextchar)) |
4633a7c4 |
761 | sayNO; |
bbce6d69 |
762 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
763 | break; |
bbce6d69 |
764 | case BOUNDL: |
765 | case NBOUNDL: |
766 | regtainted = TRUE; |
767 | /* FALL THROUGH */ |
a0d0e21e |
768 | case BOUND: |
bbce6d69 |
769 | case NBOUND: |
770 | /* was last char in word? */ |
771 | ln = (locinput != regbol) ? UCHARAT(locinput - 1) : regprev; |
772 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
773 | ln = isALNUM(ln); |
774 | n = isALNUM(nextchar); |
775 | } |
776 | else { |
777 | ln = isALNUM_LC(ln); |
778 | n = isALNUM_LC(nextchar); |
779 | } |
95bac841 |
780 | if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL)) |
4633a7c4 |
781 | sayNO; |
a0d0e21e |
782 | break; |
bbce6d69 |
783 | case SPACEL: |
784 | regtainted = TRUE; |
785 | /* FALL THROUGH */ |
a0d0e21e |
786 | case SPACE: |
787 | if (!nextchar && locinput >= regeol) |
4633a7c4 |
788 | sayNO; |
bbce6d69 |
789 | if (!(OP(scan) == SPACE |
790 | ? isSPACE(nextchar) : isSPACE_LC(nextchar))) |
4633a7c4 |
791 | sayNO; |
bbce6d69 |
792 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
793 | break; |
bbce6d69 |
794 | case NSPACEL: |
795 | regtainted = TRUE; |
796 | /* FALL THROUGH */ |
a0d0e21e |
797 | case NSPACE: |
798 | if (!nextchar) |
4633a7c4 |
799 | sayNO; |
bbce6d69 |
800 | if (OP(scan) == SPACE |
801 | ? isSPACE(nextchar) : isSPACE_LC(nextchar)) |
4633a7c4 |
802 | sayNO; |
bbce6d69 |
803 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
804 | break; |
805 | case DIGIT: |
806 | if (!isDIGIT(nextchar)) |
4633a7c4 |
807 | sayNO; |
bbce6d69 |
808 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
809 | break; |
810 | case NDIGIT: |
811 | if (!nextchar && locinput >= regeol) |
4633a7c4 |
812 | sayNO; |
a0d0e21e |
813 | if (isDIGIT(nextchar)) |
4633a7c4 |
814 | sayNO; |
bbce6d69 |
815 | nextchar = UCHARAT(++locinput); |
a0d0e21e |
816 | break; |
817 | case REF: |
818 | n = ARG1(scan); /* which paren pair */ |
819 | s = regstartp[n]; |
820 | if (!s) |
4633a7c4 |
821 | sayNO; |
a0d0e21e |
822 | if (!regendp[n]) |
4633a7c4 |
823 | sayNO; |
a0d0e21e |
824 | if (s == regendp[n]) |
825 | break; |
826 | /* Inline the first character, for speed. */ |
95bac841 |
827 | if (UCHARAT(s) != nextchar) |
4633a7c4 |
828 | sayNO; |
a0d0e21e |
829 | ln = regendp[n] - s; |
830 | if (locinput + ln > regeol) |
4633a7c4 |
831 | sayNO; |
36477c24 |
832 | if (ln > 1 && memNE(s, locinput, ln)) |
4633a7c4 |
833 | sayNO; |
a0d0e21e |
834 | locinput += ln; |
bbce6d69 |
835 | nextchar = UCHARAT(locinput); |
a0d0e21e |
836 | break; |
837 | |
838 | case NOTHING: |
839 | break; |
840 | case BACK: |
841 | break; |
842 | case OPEN: |
843 | n = ARG1(scan); /* which paren pair */ |
844 | regstartp[n] = locinput; |
845 | if (n > regsize) |
846 | regsize = n; |
847 | break; |
848 | case CLOSE: |
849 | n = ARG1(scan); /* which paren pair */ |
850 | regendp[n] = locinput; |
851 | if (n > *reglastparen) |
852 | *reglastparen = n; |
853 | break; |
854 | case CURLYX: { |
855 | CURCUR cc; |
856 | CHECKPOINT cp = savestack_ix; |
857 | cc.oldcc = regcc; |
858 | regcc = &cc; |
859 | cc.parenfloor = *reglastparen; |
860 | cc.cur = -1; |
861 | cc.min = ARG1(scan); |
862 | cc.max = ARG2(scan); |
863 | cc.scan = NEXTOPER(scan) + 4; |
864 | cc.next = next; |
865 | cc.minmod = minmod; |
866 | cc.lastloc = 0; |
867 | reginput = locinput; |
868 | n = regmatch(PREVOPER(next)); /* start on the WHILEM */ |
869 | regcpblow(cp); |
870 | regcc = cc.oldcc; |
4633a7c4 |
871 | saySAME(n); |
a0d0e21e |
872 | } |
873 | /* NOT REACHED */ |
874 | case WHILEM: { |
875 | /* |
876 | * This is really hard to understand, because after we match |
877 | * what we're trying to match, we must make sure the rest of |
878 | * the RE is going to match for sure, and to do that we have |
879 | * to go back UP the parse tree by recursing ever deeper. And |
880 | * if it fails, we have to reset our parent's current state |
881 | * that we can try again after backing off. |
882 | */ |
883 | |
5f05dabc |
884 | CHECKPOINT cp; |
a0d0e21e |
885 | CURCUR* cc = regcc; |
4633a7c4 |
886 | n = cc->cur + 1; /* how many we know we matched */ |
a0d0e21e |
887 | reginput = locinput; |
888 | |
4633a7c4 |
889 | #ifdef DEBUGGING |
890 | if (regnarrate) |
68dc0745 |
891 | PerlIO_printf(Perl_debug_log, "%*s %ld %lx\n", regindent*2, "", |
892 | (long)n, (long)cc); |
4633a7c4 |
893 | #endif |
894 | |
a0d0e21e |
895 | /* If degenerate scan matches "", assume scan done. */ |
896 | |
579cf2c3 |
897 | if (locinput == cc->lastloc && n >= cc->min) { |
a0d0e21e |
898 | regcc = cc->oldcc; |
899 | ln = regcc->cur; |
900 | if (regmatch(cc->next)) |
4633a7c4 |
901 | sayYES; |
a0d0e21e |
902 | regcc->cur = ln; |
903 | regcc = cc; |
4633a7c4 |
904 | sayNO; |
a0d0e21e |
905 | } |
906 | |
907 | /* First just match a string of min scans. */ |
908 | |
909 | if (n < cc->min) { |
910 | cc->cur = n; |
911 | cc->lastloc = locinput; |
4633a7c4 |
912 | if (regmatch(cc->scan)) |
913 | sayYES; |
914 | cc->cur = n - 1; |
915 | sayNO; |
a0d0e21e |
916 | } |
917 | |
918 | /* Prefer next over scan for minimal matching. */ |
919 | |
920 | if (cc->minmod) { |
921 | regcc = cc->oldcc; |
922 | ln = regcc->cur; |
5f05dabc |
923 | cp = regcppush(cc->parenfloor); |
924 | if (regmatch(cc->next)) { |
925 | regcpblow(cp); |
4633a7c4 |
926 | sayYES; /* All done. */ |
5f05dabc |
927 | } |
928 | regcppop(); |
a0d0e21e |
929 | regcc->cur = ln; |
930 | regcc = cc; |
931 | |
932 | if (n >= cc->max) /* Maximum greed exceeded? */ |
4633a7c4 |
933 | sayNO; |
a687059c |
934 | |
a0d0e21e |
935 | /* Try scanning more and see if it helps. */ |
936 | reginput = locinput; |
937 | cc->cur = n; |
938 | cc->lastloc = locinput; |
5f05dabc |
939 | cp = regcppush(cc->parenfloor); |
940 | if (regmatch(cc->scan)) { |
941 | regcpblow(cp); |
4633a7c4 |
942 | sayYES; |
5f05dabc |
943 | } |
944 | regcppop(); |
4633a7c4 |
945 | cc->cur = n - 1; |
946 | sayNO; |
a0d0e21e |
947 | } |
948 | |
949 | /* Prefer scan over next for maximal matching. */ |
950 | |
951 | if (n < cc->max) { /* More greed allowed? */ |
5f05dabc |
952 | cp = regcppush(cc->parenfloor); |
a0d0e21e |
953 | cc->cur = n; |
954 | cc->lastloc = locinput; |
5f05dabc |
955 | if (regmatch(cc->scan)) { |
956 | regcpblow(cp); |
4633a7c4 |
957 | sayYES; |
5f05dabc |
958 | } |
a0d0e21e |
959 | regcppop(); /* Restore some previous $<digit>s? */ |
960 | reginput = locinput; |
961 | } |
962 | |
963 | /* Failed deeper matches of scan, so see if this one works. */ |
964 | regcc = cc->oldcc; |
965 | ln = regcc->cur; |
966 | if (regmatch(cc->next)) |
4633a7c4 |
967 | sayYES; |
a0d0e21e |
968 | regcc->cur = ln; |
969 | regcc = cc; |
4633a7c4 |
970 | cc->cur = n - 1; |
971 | sayNO; |
a0d0e21e |
972 | } |
973 | /* NOT REACHED */ |
974 | case BRANCH: { |
975 | if (OP(next) != BRANCH) /* No choice. */ |
976 | next = NEXTOPER(scan);/* Avoid recursion. */ |
977 | else { |
748a9306 |
978 | int lastparen = *reglastparen; |
a0d0e21e |
979 | do { |
980 | reginput = locinput; |
981 | if (regmatch(NEXTOPER(scan))) |
4633a7c4 |
982 | sayYES; |
748a9306 |
983 | for (n = *reglastparen; n > lastparen; n--) |
984 | regendp[n] = 0; |
985 | *reglastparen = n; |
986 | |
a687059c |
987 | #ifdef REGALIGN |
a0d0e21e |
988 | /*SUPPRESS 560*/ |
989 | if (n = NEXT(scan)) |
990 | scan += n; |
991 | else |
992 | scan = NULL; |
a687059c |
993 | #else |
a0d0e21e |
994 | scan = regnext(scan); |
79072805 |
995 | #endif |
a0d0e21e |
996 | } while (scan != NULL && OP(scan) == BRANCH); |
4633a7c4 |
997 | sayNO; |
a0d0e21e |
998 | /* NOTREACHED */ |
a687059c |
999 | } |
a0d0e21e |
1000 | } |
1001 | break; |
1002 | case MINMOD: |
1003 | minmod = 1; |
1004 | break; |
1005 | case CURLY: |
1006 | ln = ARG1(scan); /* min to match */ |
1007 | n = ARG2(scan); /* max to match */ |
1008 | scan = NEXTOPER(scan) + 4; |
1009 | goto repeat; |
1010 | case STAR: |
1011 | ln = 0; |
1012 | n = 32767; |
1013 | scan = NEXTOPER(scan); |
1014 | goto repeat; |
1015 | case PLUS: |
1016 | /* |
1017 | * Lookahead to avoid useless match attempts |
1018 | * when we know what character comes next. |
1019 | */ |
1020 | ln = 1; |
1021 | n = 32767; |
1022 | scan = NEXTOPER(scan); |
1023 | repeat: |
bbce6d69 |
1024 | if (regkind[(U8)OP(next)] == EXACT) { |
1025 | c1 = UCHARAT(OPERAND(next) + 1); |
1026 | if (OP(next) == EXACTF) |
1027 | c2 = fold[c1]; |
1028 | else if (OP(next) == EXACTFL) |
1029 | c2 = fold_locale[c1]; |
1030 | else |
1031 | c2 = c1; |
1032 | } |
a0d0e21e |
1033 | else |
bbce6d69 |
1034 | c1 = c2 = -1000; |
a0d0e21e |
1035 | reginput = locinput; |
1036 | if (minmod) { |
1037 | minmod = 0; |
1038 | if (ln && regrepeat(scan, ln) < ln) |
4633a7c4 |
1039 | sayNO; |
8e07c86e |
1040 | while (n >= ln || (n == 32767 && ln > 0)) { /* ln overflow ? */ |
a0d0e21e |
1041 | /* If it could work, try it. */ |
bbce6d69 |
1042 | if (c1 == -1000 || |
1043 | UCHARAT(reginput) == c1 || |
1044 | UCHARAT(reginput) == c2) |
1045 | { |
a0d0e21e |
1046 | if (regmatch(next)) |
4633a7c4 |
1047 | sayYES; |
bbce6d69 |
1048 | } |
a0d0e21e |
1049 | /* Couldn't or didn't -- back up. */ |
748a9306 |
1050 | reginput = locinput + ln; |
a0d0e21e |
1051 | if (regrepeat(scan, 1)) { |
1052 | ln++; |
1053 | reginput = locinput + ln; |
1054 | } |
1055 | else |
4633a7c4 |
1056 | sayNO; |
a0d0e21e |
1057 | } |
1058 | } |
1059 | else { |
1060 | n = regrepeat(scan, n); |
1061 | if (ln < n && regkind[(U8)OP(next)] == EOL && |
1062 | (!multiline || OP(next) == SEOL)) |
1063 | ln = n; /* why back off? */ |
1064 | while (n >= ln) { |
1065 | /* If it could work, try it. */ |
bbce6d69 |
1066 | if (c1 == -1000 || |
1067 | UCHARAT(reginput) == c1 || |
1068 | UCHARAT(reginput) == c2) |
1069 | { |
a0d0e21e |
1070 | if (regmatch(next)) |
4633a7c4 |
1071 | sayYES; |
bbce6d69 |
1072 | } |
a0d0e21e |
1073 | /* Couldn't or didn't -- back up. */ |
1074 | n--; |
1075 | reginput = locinput + n; |
1076 | } |
1077 | } |
4633a7c4 |
1078 | sayNO; |
a0d0e21e |
1079 | case SUCCEED: |
1080 | case END: |
1081 | reginput = locinput; /* put where regtry can find it */ |
4633a7c4 |
1082 | sayYES; /* Success! */ |
a0d0e21e |
1083 | case IFMATCH: |
1084 | reginput = locinput; |
1085 | scan = NEXTOPER(scan); |
1086 | if (!regmatch(scan)) |
4633a7c4 |
1087 | sayNO; |
a0d0e21e |
1088 | break; |
1089 | case UNLESSM: |
1090 | reginput = locinput; |
1091 | scan = NEXTOPER(scan); |
1092 | if (regmatch(scan)) |
4633a7c4 |
1093 | sayNO; |
a0d0e21e |
1094 | break; |
1095 | default: |
c030ccd9 |
1096 | PerlIO_printf(PerlIO_stderr(), "%lx %d\n", |
1097 | (unsigned long)scan, scan[1]); |
a0d0e21e |
1098 | FAIL("regexp memory corruption"); |
a687059c |
1099 | } |
a0d0e21e |
1100 | scan = next; |
1101 | } |
a687059c |
1102 | |
a0d0e21e |
1103 | /* |
1104 | * We get here only if there's trouble -- normally "case END" is |
1105 | * the terminating point. |
1106 | */ |
1107 | FAIL("corrupted regexp pointers"); |
1108 | /*NOTREACHED*/ |
4633a7c4 |
1109 | sayNO; |
1110 | |
1111 | yes: |
1112 | #ifdef DEBUGGING |
1113 | regindent--; |
1114 | #endif |
1115 | return 1; |
1116 | |
1117 | no: |
1118 | #ifdef DEBUGGING |
1119 | regindent--; |
1120 | #endif |
a0d0e21e |
1121 | return 0; |
a687059c |
1122 | } |
1123 | |
1124 | /* |
1125 | - regrepeat - repeatedly match something simple, report how many |
1126 | */ |
1127 | /* |
1128 | * [This routine now assumes that it will only match on things of length 1. |
1129 | * That was true before, but now we assume scan - reginput is the count, |
1130 | * rather than incrementing count on every character.] |
1131 | */ |
79072805 |
1132 | static I32 |
00bf170e |
1133 | regrepeat(p, max) |
a687059c |
1134 | char *p; |
79072805 |
1135 | I32 max; |
a687059c |
1136 | { |
a0d0e21e |
1137 | register char *scan; |
1138 | register char *opnd; |
1139 | register I32 c; |
1140 | register char *loceol = regeol; |
1141 | |
1142 | scan = reginput; |
1143 | if (max != 32767 && max < loceol - scan) |
1144 | loceol = scan + max; |
1145 | opnd = OPERAND(p); |
1146 | switch (OP(p)) { |
1147 | case ANY: |
1148 | while (scan < loceol && *scan != '\n') |
1149 | scan++; |
1150 | break; |
1151 | case SANY: |
1152 | scan = loceol; |
1153 | break; |
bbce6d69 |
1154 | case EXACT: /* length of string is 1 */ |
1155 | c = UCHARAT(++opnd); |
1156 | while (scan < loceol && UCHARAT(scan) == c) |
1157 | scan++; |
1158 | break; |
1159 | case EXACTF: /* length of string is 1 */ |
1160 | c = UCHARAT(++opnd); |
1161 | while (scan < loceol && |
1162 | (UCHARAT(scan) == c || UCHARAT(scan) == fold[c])) |
1163 | scan++; |
1164 | break; |
1165 | case EXACTFL: /* length of string is 1 */ |
1166 | regtainted = TRUE; |
1167 | c = UCHARAT(++opnd); |
1168 | while (scan < loceol && |
1169 | (UCHARAT(scan) == c || UCHARAT(scan) == fold_locale[c])) |
a0d0e21e |
1170 | scan++; |
1171 | break; |
1172 | case ANYOF: |
bbce6d69 |
1173 | while (scan < loceol && reginclass(opnd, *scan)) |
a0d0e21e |
1174 | scan++; |
a0d0e21e |
1175 | break; |
1176 | case ALNUM: |
1177 | while (scan < loceol && isALNUM(*scan)) |
1178 | scan++; |
1179 | break; |
bbce6d69 |
1180 | case ALNUML: |
1181 | regtainted = TRUE; |
1182 | while (scan < loceol && isALNUM_LC(*scan)) |
1183 | scan++; |
1184 | break; |
a0d0e21e |
1185 | case NALNUM: |
1186 | while (scan < loceol && !isALNUM(*scan)) |
1187 | scan++; |
1188 | break; |
bbce6d69 |
1189 | case NALNUML: |
1190 | regtainted = TRUE; |
1191 | while (scan < loceol && !isALNUM_LC(*scan)) |
1192 | scan++; |
1193 | break; |
a0d0e21e |
1194 | case SPACE: |
1195 | while (scan < loceol && isSPACE(*scan)) |
1196 | scan++; |
1197 | break; |
bbce6d69 |
1198 | case SPACEL: |
1199 | regtainted = TRUE; |
1200 | while (scan < loceol && isSPACE_LC(*scan)) |
1201 | scan++; |
1202 | break; |
a0d0e21e |
1203 | case NSPACE: |
1204 | while (scan < loceol && !isSPACE(*scan)) |
1205 | scan++; |
1206 | break; |
bbce6d69 |
1207 | case NSPACEL: |
1208 | regtainted = TRUE; |
1209 | while (scan < loceol && !isSPACE_LC(*scan)) |
1210 | scan++; |
1211 | break; |
a0d0e21e |
1212 | case DIGIT: |
1213 | while (scan < loceol && isDIGIT(*scan)) |
1214 | scan++; |
1215 | break; |
1216 | case NDIGIT: |
1217 | while (scan < loceol && !isDIGIT(*scan)) |
1218 | scan++; |
1219 | break; |
1220 | default: /* Called on something of 0 width. */ |
1221 | break; /* So match right here or not at all. */ |
1222 | } |
a687059c |
1223 | |
a0d0e21e |
1224 | c = scan - reginput; |
1225 | reginput = scan; |
a687059c |
1226 | |
a0d0e21e |
1227 | return(c); |
a687059c |
1228 | } |
1229 | |
1230 | /* |
bbce6d69 |
1231 | - regclass - determine if a character falls into a character class |
1232 | */ |
1233 | |
1234 | static bool |
1235 | reginclass(p, c) |
1236 | register char *p; |
1237 | register I32 c; |
1238 | { |
1239 | char flags = *p; |
1240 | bool match = FALSE; |
1241 | |
1242 | c &= 0xFF; |
1243 | if (p[1 + (c >> 3)] & (1 << (c & 7))) |
1244 | match = TRUE; |
1245 | else if (flags & ANYOF_FOLD) { |
1246 | I32 cf; |
1247 | if (flags & ANYOF_LOCALE) { |
1248 | regtainted = TRUE; |
1249 | cf = fold_locale[c]; |
1250 | } |
1251 | else |
1252 | cf = fold[c]; |
1253 | if (p[1 + (cf >> 3)] & (1 << (cf & 7))) |
1254 | match = TRUE; |
1255 | } |
1256 | |
1257 | if (!match && (flags & ANYOF_ISA)) { |
1258 | regtainted = TRUE; |
1259 | |
1260 | if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) || |
1261 | ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) || |
1262 | ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) || |
1263 | ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c))) |
1264 | { |
1265 | match = TRUE; |
1266 | } |
1267 | } |
1268 | |
1269 | return match ^ ((flags & ANYOF_INVERT) != 0); |
1270 | } |
1271 | |
1272 | /* |
a687059c |
1273 | - regnext - dig the "next" pointer out of a node |
1274 | * |
1275 | * [Note, when REGALIGN is defined there are two places in regmatch() |
1276 | * that bypass this code for speed.] |
1277 | */ |
1278 | char * |
1279 | regnext(p) |
1280 | register char *p; |
1281 | { |
a0d0e21e |
1282 | register I32 offset; |
a687059c |
1283 | |
a0d0e21e |
1284 | if (p == ®dummy) |
1285 | return(NULL); |
a687059c |
1286 | |
a0d0e21e |
1287 | offset = NEXT(p); |
1288 | if (offset == 0) |
1289 | return(NULL); |
a687059c |
1290 | |
1291 | #ifdef REGALIGN |
a0d0e21e |
1292 | return(p+offset); |
a687059c |
1293 | #else |
a0d0e21e |
1294 | if (OP(p) == BACK) |
1295 | return(p-offset); |
1296 | else |
1297 | return(p+offset); |
a687059c |
1298 | #endif |
1299 | } |