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 | |
c277df42 |
58 | #define RF_tainted 1 /* tainted information used? */ |
59 | #define RF_warned 2 /* warned about big count? */ |
60 | #define RF_evaled 4 /* Did an EVAL? */ |
61 | |
a687059c |
62 | #ifndef STATIC |
63 | #define STATIC static |
64 | #endif |
65 | |
76e3520e |
66 | #ifndef PERL_OBJECT |
a0d0e21e |
67 | typedef I32 CHECKPOINT; |
68 | |
c277df42 |
69 | /* |
70 | * Forwards. |
71 | */ |
72 | |
73 | static I32 regmatch _((regnode *prog)); |
74 | static I32 regrepeat _((regnode *p, I32 max)); |
75 | static I32 regrepeat_hard _((regnode *p, I32 max, I32 *lp)); |
76 | static I32 regtry _((regexp *prog, char *startpos)); |
77 | static bool reginclass _((char *p, I32 c)); |
55497cff |
78 | static CHECKPOINT regcppush _((I32 parenfloor)); |
79 | static char * regcppop _((void)); |
76e3520e |
80 | #endif |
a0d0e21e |
81 | |
76e3520e |
82 | STATIC CHECKPOINT |
8ac85365 |
83 | regcppush(I32 parenfloor) |
a0d0e21e |
84 | { |
11343788 |
85 | dTHR; |
a0d0e21e |
86 | int retval = savestack_ix; |
c277df42 |
87 | int i = (regsize - parenfloor) * 4; |
a0d0e21e |
88 | int p; |
89 | |
90 | SSCHECK(i + 5); |
91 | for (p = regsize; p > parenfloor; p--) { |
92 | SSPUSHPTR(regendp[p]); |
93 | SSPUSHPTR(regstartp[p]); |
c5254dd6 |
94 | SSPUSHPTR(reg_start_tmp[p]); |
a0d0e21e |
95 | SSPUSHINT(p); |
96 | } |
97 | SSPUSHINT(regsize); |
98 | SSPUSHINT(*reglastparen); |
99 | SSPUSHPTR(reginput); |
100 | SSPUSHINT(i + 3); |
101 | SSPUSHINT(SAVEt_REGCONTEXT); |
102 | return retval; |
103 | } |
104 | |
c277df42 |
105 | /* These are needed since we do not localize EVAL nodes: */ |
106 | # define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, " Setting an EVAL scope, savestack=%i\n", savestack_ix)); lastcp = savestack_ix |
107 | # define REGCP_UNWIND DEBUG_r(lastcp != savestack_ix ? PerlIO_printf(Perl_debug_log," Clearing an EVAL scope, savestack=%i..%i\n", lastcp, savestack_ix) : 0); regcpblow(lastcp) |
108 | |
76e3520e |
109 | STATIC char * |
8ac85365 |
110 | regcppop(void) |
a0d0e21e |
111 | { |
11343788 |
112 | dTHR; |
a0d0e21e |
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; |
c277df42 |
122 | for (i -= 3; i > 0; i -= 4) { |
a0d0e21e |
123 | paren = (U32)SSPOPINT; |
c277df42 |
124 | reg_start_tmp[paren] = (char *) SSPOPPTR; |
a0d0e21e |
125 | regstartp[paren] = (char *) SSPOPPTR; |
126 | tmps = (char*)SSPOPPTR; |
127 | if (paren <= *reglastparen) |
128 | regendp[paren] = tmps; |
c277df42 |
129 | DEBUG_r( |
130 | PerlIO_printf(Perl_debug_log, " restoring \\%d to %d(%d)..%d%s\n", |
131 | paren, regstartp[paren] - regbol, |
132 | reg_start_tmp[paren] - regbol, |
133 | regendp[paren] - regbol, |
134 | (paren > *reglastparen ? "(no)" : "")); |
135 | ); |
a0d0e21e |
136 | } |
c277df42 |
137 | DEBUG_r( |
138 | if (*reglastparen + 1 <= regnpar) { |
139 | PerlIO_printf(Perl_debug_log, " restoring \\%d..\\%d to undef\n", |
140 | *reglastparen + 1, regnpar); |
141 | } |
142 | ); |
a0d0e21e |
143 | for (paren = *reglastparen + 1; paren <= regnpar; paren++) { |
144 | if (paren > regsize) |
145 | regstartp[paren] = Nullch; |
146 | regendp[paren] = Nullch; |
147 | } |
148 | return input; |
149 | } |
150 | |
c277df42 |
151 | #define regcpblow(cp) LEAVE_SCOPE(cp) |
a0d0e21e |
152 | |
a687059c |
153 | /* |
e50aee73 |
154 | * pregexec and friends |
a687059c |
155 | */ |
156 | |
157 | /* |
c277df42 |
158 | - pregexec - match a regexp against a string |
a687059c |
159 | */ |
c277df42 |
160 | I32 |
161 | pregexec(register regexp *prog, char *stringarg, register char *strend, char *strbeg, I32 minend, SV *screamer, U32 nosave) |
162 | /* strend: pointer to null at end of string */ |
163 | /* strbeg: real beginning of string */ |
164 | /* minend: end of match must be >=minend after stringarg. */ |
165 | /* nosave: For optimizations. */ |
166 | { |
167 | return |
168 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
169 | nosave ? 0 : REXEC_COPY_STR); |
170 | } |
171 | |
a687059c |
172 | /* |
c277df42 |
173 | - regexec_flags - match a regexp against a string |
a687059c |
174 | */ |
79072805 |
175 | I32 |
c277df42 |
176 | regexec_flags(register regexp *prog, char *stringarg, register char *strend, char *strbeg, I32 minend, SV *screamer, void *data, U32 flags) |
177 | /* strend: pointer to null at end of string */ |
178 | /* strbeg: real beginning of string */ |
179 | /* minend: end of match must be >=minend after stringarg. */ |
180 | /* data: May be used for some additional optimizations. */ |
181 | /* nosave: For optimizations. */ |
a687059c |
182 | { |
a0d0e21e |
183 | register char *s; |
c277df42 |
184 | register regnode *c; |
a0d0e21e |
185 | register char *startpos = stringarg; |
186 | register I32 tmp; |
c277df42 |
187 | I32 minlen; /* must match at least this many chars */ |
a0d0e21e |
188 | I32 dontbother = 0; /* how many characters not to try at end */ |
189 | CURCUR cc; |
c277df42 |
190 | I32 start_shift = 0; /* Offset of the start to find |
191 | constant substr. */ |
192 | I32 end_shift = 0; /* Same for the end. */ |
193 | I32 scream_pos = -1; /* Internal iterator of scream. */ |
194 | char *scream_olds; |
a687059c |
195 | |
a0d0e21e |
196 | cc.cur = 0; |
4633a7c4 |
197 | cc.oldcc = 0; |
a0d0e21e |
198 | regcc = &cc; |
199 | |
c277df42 |
200 | regprecomp = prog->precomp; /* Needed for error messages. */ |
a0d0e21e |
201 | #ifdef DEBUGGING |
202 | regnarrate = debug & 512; |
203 | regprogram = prog->program; |
204 | #endif |
205 | |
206 | /* Be paranoid... */ |
207 | if (prog == NULL || startpos == NULL) { |
208 | croak("NULL regexp parameter"); |
209 | return 0; |
210 | } |
211 | |
c277df42 |
212 | minlen = prog->minlen; |
213 | if (strend - startpos < minlen) goto phooey; |
214 | |
a0d0e21e |
215 | if (startpos == strbeg) /* is ^ valid at stringarg? */ |
216 | regprev = '\n'; |
217 | else { |
218 | regprev = stringarg[-1]; |
219 | if (!multiline && regprev == '\n') |
220 | regprev = '\0'; /* force ^ to NOT match */ |
221 | } |
bbce6d69 |
222 | |
a0d0e21e |
223 | /* Check validity of program. */ |
224 | if (UCHARAT(prog->program) != MAGIC) { |
225 | FAIL("corrupted regexp program"); |
226 | } |
227 | |
bbce6d69 |
228 | regnpar = prog->nparens; |
c277df42 |
229 | reg_flags = 0; |
230 | reg_eval_set = 0; |
a0d0e21e |
231 | |
232 | /* If there is a "must appear" string, look for it. */ |
233 | s = startpos; |
c277df42 |
234 | if (!(flags & REXEC_CHECKED) |
235 | && prog->check_substr != Nullsv && |
774d564b |
236 | !(prog->reganch & ROPT_ANCH_GPOS) && |
c277df42 |
237 | (!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL)) |
238 | || (multiline && prog->check_substr == prog->anchored_substr)) ) |
a0d0e21e |
239 | { |
c277df42 |
240 | start_shift = prog->check_offset_min; |
241 | /* Should be nonnegative! */ |
242 | end_shift = minlen - start_shift - SvCUR(prog->check_substr); |
243 | if (screamer) { |
244 | if (screamfirst[BmRARE(prog->check_substr)] >= 0) |
245 | s = screaminstr(screamer, prog->check_substr, |
246 | start_shift + (stringarg - strbeg), |
247 | end_shift, &scream_pos, 0); |
a0d0e21e |
248 | else |
249 | s = Nullch; |
c277df42 |
250 | scream_olds = s; |
0a12ae7d |
251 | } |
a0d0e21e |
252 | else |
c277df42 |
253 | s = fbm_instr((unsigned char*)s + start_shift, |
254 | (unsigned char*)strend - end_shift, |
255 | prog->check_substr); |
a0d0e21e |
256 | if (!s) { |
c277df42 |
257 | ++BmUSEFUL(prog->check_substr); /* hooray */ |
a0d0e21e |
258 | goto phooey; /* not present */ |
c277df42 |
259 | } else if ((s - stringarg) > prog->check_offset_max) { |
260 | ++BmUSEFUL(prog->check_substr); /* hooray/2 */ |
261 | s -= prog->check_offset_max; |
262 | } else if (!prog->naughty |
263 | && --BmUSEFUL(prog->check_substr) < 0 |
264 | && prog->check_substr == prog->float_substr) { /* boo */ |
265 | SvREFCNT_dec(prog->check_substr); |
266 | prog->check_substr = Nullsv; /* disable */ |
267 | prog->float_substr = Nullsv; /* clear */ |
a0d0e21e |
268 | s = startpos; |
c277df42 |
269 | } else s = startpos; |
a0d0e21e |
270 | } |
a687059c |
271 | |
c277df42 |
272 | /* Mark beginning of line for ^ and lookbehind. */ |
a0d0e21e |
273 | regbol = startpos; |
c277df42 |
274 | bostr = strbeg; |
a687059c |
275 | |
a0d0e21e |
276 | /* Mark end of line for $ (and such) */ |
277 | regeol = strend; |
a687059c |
278 | |
a0d0e21e |
279 | /* see how far we have to get to not match where we matched before */ |
280 | regtill = startpos+minend; |
a687059c |
281 | |
c277df42 |
282 | DEBUG_r( |
283 | PerlIO_printf(Perl_debug_log, |
284 | "Matching `%.60s%s' against `%.*s%s'\n", |
285 | prog->precomp, |
286 | (strlen(prog->precomp) > 60 ? "..." : ""), |
287 | (strend - startpos > 60 ? 60 : strend - startpos), |
288 | startpos, |
289 | (strend - startpos > 60 ? "..." : "")) |
290 | ); |
291 | |
a0d0e21e |
292 | /* Simplest case: anchored match need be tried only once. */ |
774d564b |
293 | /* [unless only anchor is BOL and multiline is set] */ |
a0d0e21e |
294 | if (prog->reganch & ROPT_ANCH) { |
295 | if (regtry(prog, startpos)) |
296 | goto got_it; |
774d564b |
297 | else if (!(prog->reganch & ROPT_ANCH_GPOS) && |
c277df42 |
298 | (multiline || (prog->reganch & ROPT_IMPLICIT) |
299 | || (prog->reganch & ROPT_ANCH_MBOL))) |
774d564b |
300 | { |
a0d0e21e |
301 | if (minlen) |
302 | dontbother = minlen - 1; |
303 | strend -= dontbother; |
304 | /* for multiline we only have to try after newlines */ |
305 | if (s > startpos) |
306 | s--; |
307 | while (s < strend) { |
308 | if (*s++ == '\n') { |
309 | if (s < strend && regtry(prog, s)) |
310 | goto got_it; |
311 | } |
35c8bce7 |
312 | } |
35c8bce7 |
313 | } |
a0d0e21e |
314 | goto phooey; |
315 | } |
35c8bce7 |
316 | |
a0d0e21e |
317 | /* Messy cases: unanchored match. */ |
c277df42 |
318 | if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { |
319 | /* we have /x+whatever/ */ |
320 | /* it must be a one character string */ |
321 | char ch = SvPVX(prog->anchored_substr)[0]; |
322 | while (s < strend) { |
323 | if (*s == ch) { |
324 | if (regtry(prog, s)) goto got_it; |
a0d0e21e |
325 | s++; |
c277df42 |
326 | while (s < strend && *s == ch) |
327 | s++; |
a0d0e21e |
328 | } |
c277df42 |
329 | s++; |
a687059c |
330 | } |
c277df42 |
331 | } |
332 | /*SUPPRESS 560*/ |
333 | else if (prog->anchored_substr != Nullsv |
334 | || (prog->float_substr != Nullsv |
335 | && prog->float_max_offset < strend - s)) { |
336 | SV *must = prog->anchored_substr |
337 | ? prog->anchored_substr : prog->float_substr; |
338 | I32 back_max = |
339 | prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; |
340 | I32 back_min = |
341 | prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; |
342 | I32 delta = back_max - back_min; |
343 | char *last = strend - SvCUR(must) - back_min; /* Cannot start after this */ |
344 | char *last1 = s - 1; /* Last position checked before */ |
345 | |
346 | /* XXXX check_substr already used to find `s', can optimize if |
347 | check_substr==must. */ |
348 | scream_pos = -1; |
349 | dontbother = end_shift; |
350 | strend -= dontbother; |
351 | while ( (s <= last) && |
352 | (screamer |
353 | ? (s = screaminstr(screamer, must, s + back_min - strbeg, |
354 | end_shift, &scream_pos, 0)) |
355 | : (s = fbm_instr((unsigned char*)s + back_min, |
356 | (unsigned char*)strend, must))) ) { |
357 | if (s - back_max > last1) { |
358 | last1 = s - back_min; |
359 | s = s - back_max; |
360 | } else { |
361 | char *t = last1 + 1; |
362 | |
363 | last1 = s - back_min; |
364 | s = t; |
a0d0e21e |
365 | } |
c277df42 |
366 | while (s <= last1) { |
a0d0e21e |
367 | if (regtry(prog, s)) |
368 | goto got_it; |
369 | s++; |
370 | } |
371 | } |
372 | goto phooey; |
c277df42 |
373 | } else if (c = prog->regstclass) { |
a0d0e21e |
374 | I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
161b471a |
375 | char *Class; |
a687059c |
376 | |
a0d0e21e |
377 | if (minlen) |
378 | dontbother = minlen - 1; |
379 | strend -= dontbother; /* don't bother with what can't match */ |
380 | tmp = 1; |
381 | /* We know what class it must start with. */ |
382 | switch (OP(c)) { |
383 | case ANYOF: |
161b471a |
384 | Class = (char *) OPERAND(c); |
a0d0e21e |
385 | while (s < strend) { |
161b471a |
386 | if (reginclass(Class, *s)) { |
a0d0e21e |
387 | if (tmp && regtry(prog, s)) |
388 | goto got_it; |
389 | else |
390 | tmp = doevery; |
a687059c |
391 | } |
a0d0e21e |
392 | else |
393 | tmp = 1; |
394 | s++; |
395 | } |
396 | break; |
bbce6d69 |
397 | case BOUNDL: |
c277df42 |
398 | reg_flags |= RF_tainted; |
bbce6d69 |
399 | /* FALL THROUGH */ |
a0d0e21e |
400 | case BOUND: |
401 | if (minlen) |
402 | dontbother++,strend--; |
bbce6d69 |
403 | tmp = (s != startpos) ? UCHARAT(s - 1) : regprev; |
95bac841 |
404 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
405 | while (s < strend) { |
95bac841 |
406 | if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { |
a0d0e21e |
407 | tmp = !tmp; |
408 | if (regtry(prog, s)) |
409 | goto got_it; |
a687059c |
410 | } |
a0d0e21e |
411 | s++; |
412 | } |
413 | if ((minlen || tmp) && regtry(prog,s)) |
414 | goto got_it; |
415 | break; |
bbce6d69 |
416 | case NBOUNDL: |
c277df42 |
417 | reg_flags |= RF_tainted; |
bbce6d69 |
418 | /* FALL THROUGH */ |
a0d0e21e |
419 | case NBOUND: |
420 | if (minlen) |
421 | dontbother++,strend--; |
bbce6d69 |
422 | tmp = (s != startpos) ? UCHARAT(s - 1) : regprev; |
95bac841 |
423 | tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
424 | while (s < strend) { |
95bac841 |
425 | if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) |
a0d0e21e |
426 | tmp = !tmp; |
427 | else if (regtry(prog, s)) |
428 | goto got_it; |
429 | s++; |
430 | } |
431 | if ((minlen || !tmp) && regtry(prog,s)) |
432 | goto got_it; |
433 | break; |
434 | case ALNUM: |
435 | while (s < strend) { |
bbce6d69 |
436 | if (isALNUM(*s)) { |
437 | if (tmp && regtry(prog, s)) |
438 | goto got_it; |
439 | else |
440 | tmp = doevery; |
441 | } |
442 | else |
443 | tmp = 1; |
444 | s++; |
445 | } |
446 | break; |
447 | case ALNUML: |
c277df42 |
448 | reg_flags |= RF_tainted; |
bbce6d69 |
449 | while (s < strend) { |
450 | if (isALNUM_LC(*s)) { |
a0d0e21e |
451 | if (tmp && regtry(prog, s)) |
452 | goto got_it; |
a687059c |
453 | else |
a0d0e21e |
454 | tmp = doevery; |
455 | } |
456 | else |
457 | tmp = 1; |
458 | s++; |
459 | } |
460 | break; |
461 | case NALNUM: |
462 | while (s < strend) { |
bbce6d69 |
463 | if (!isALNUM(*s)) { |
464 | if (tmp && regtry(prog, s)) |
465 | goto got_it; |
466 | else |
467 | tmp = doevery; |
468 | } |
469 | else |
470 | tmp = 1; |
471 | s++; |
472 | } |
473 | break; |
474 | case NALNUML: |
c277df42 |
475 | reg_flags |= RF_tainted; |
bbce6d69 |
476 | while (s < strend) { |
477 | if (!isALNUM_LC(*s)) { |
a0d0e21e |
478 | if (tmp && regtry(prog, s)) |
479 | goto got_it; |
a687059c |
480 | else |
a0d0e21e |
481 | tmp = doevery; |
a687059c |
482 | } |
a0d0e21e |
483 | else |
484 | tmp = 1; |
485 | s++; |
486 | } |
487 | break; |
488 | case SPACE: |
489 | while (s < strend) { |
490 | if (isSPACE(*s)) { |
491 | if (tmp && regtry(prog, s)) |
492 | goto got_it; |
493 | else |
494 | tmp = doevery; |
2304df62 |
495 | } |
a0d0e21e |
496 | else |
497 | tmp = 1; |
498 | s++; |
499 | } |
500 | break; |
bbce6d69 |
501 | case SPACEL: |
c277df42 |
502 | reg_flags |= RF_tainted; |
bbce6d69 |
503 | while (s < strend) { |
504 | if (isSPACE_LC(*s)) { |
505 | if (tmp && regtry(prog, s)) |
506 | goto got_it; |
507 | else |
508 | tmp = doevery; |
509 | } |
510 | else |
511 | tmp = 1; |
512 | s++; |
513 | } |
514 | break; |
a0d0e21e |
515 | case NSPACE: |
516 | while (s < strend) { |
517 | if (!isSPACE(*s)) { |
518 | if (tmp && regtry(prog, s)) |
519 | goto got_it; |
520 | else |
521 | tmp = doevery; |
a687059c |
522 | } |
a0d0e21e |
523 | else |
524 | tmp = 1; |
525 | s++; |
526 | } |
527 | break; |
bbce6d69 |
528 | case NSPACEL: |
c277df42 |
529 | reg_flags |= RF_tainted; |
bbce6d69 |
530 | while (s < strend) { |
531 | if (!isSPACE_LC(*s)) { |
532 | if (tmp && regtry(prog, s)) |
533 | goto got_it; |
534 | else |
535 | tmp = doevery; |
536 | } |
537 | else |
538 | tmp = 1; |
539 | s++; |
540 | } |
541 | break; |
a0d0e21e |
542 | case DIGIT: |
543 | while (s < strend) { |
544 | if (isDIGIT(*s)) { |
545 | if (tmp && regtry(prog, s)) |
546 | goto got_it; |
547 | else |
548 | tmp = doevery; |
2b69d0c2 |
549 | } |
a0d0e21e |
550 | else |
551 | tmp = 1; |
552 | s++; |
553 | } |
554 | break; |
555 | case NDIGIT: |
556 | while (s < strend) { |
557 | if (!isDIGIT(*s)) { |
558 | if (tmp && regtry(prog, s)) |
559 | goto got_it; |
560 | else |
561 | tmp = doevery; |
a687059c |
562 | } |
a0d0e21e |
563 | else |
564 | tmp = 1; |
565 | s++; |
566 | } |
567 | break; |
a687059c |
568 | } |
a0d0e21e |
569 | } |
570 | else { |
c277df42 |
571 | dontbother = 0; |
572 | if (prog->float_substr != Nullsv) { /* Trim the end. */ |
573 | char *last; |
574 | I32 oldpos = scream_pos; |
575 | |
576 | if (screamer) { |
577 | last = screaminstr(screamer, prog->float_substr, s - strbeg, |
578 | end_shift, &scream_pos, 1); /* last one */ |
579 | if (!last) { |
580 | last = scream_olds; /* Only one occurence. */ |
581 | } |
582 | } else { |
583 | STRLEN len; |
584 | char *little = SvPV(prog->float_substr, len); |
585 | last = rninstr(s, strend, little, little + len); |
586 | } |
587 | if (last == NULL) goto phooey; /* Should not happen! */ |
588 | dontbother = strend - last - 1; |
589 | } |
590 | if (minlen && (dontbother < minlen)) |
a0d0e21e |
591 | dontbother = minlen - 1; |
592 | strend -= dontbother; |
593 | /* We don't know much -- general case. */ |
594 | do { |
595 | if (regtry(prog, s)) |
596 | goto got_it; |
597 | } while (s++ < strend); |
598 | } |
599 | |
600 | /* Failure. */ |
601 | goto phooey; |
a687059c |
602 | |
a0d0e21e |
603 | got_it: |
420218e7 |
604 | strend += dontbother; /* uncheat */ |
a0d0e21e |
605 | prog->subbeg = strbeg; |
606 | prog->subend = strend; |
c277df42 |
607 | RX_MATCH_TAINTED_SET(prog, reg_flags & RF_tainted); |
5f05dabc |
608 | |
609 | /* make sure $`, $&, $', and $digit will work later */ |
c277df42 |
610 | if (strbeg != prog->subbase) { /* second+ //g match. */ |
611 | if (!(flags & REXEC_COPY_STR)) { |
137443ea |
612 | if (prog->subbase) { |
613 | Safefree(prog->subbase); |
614 | prog->subbase = Nullch; |
615 | } |
616 | } |
617 | else { |
618 | I32 i = strend - startpos + (stringarg - strbeg); |
619 | s = savepvn(strbeg, i); |
620 | Safefree(prog->subbase); |
621 | prog->subbase = s; |
622 | prog->subbeg = prog->subbase; |
623 | prog->subend = prog->subbase + i; |
624 | s = prog->subbase + (stringarg - strbeg); |
625 | for (i = 0; i <= prog->nparens; i++) { |
626 | if (prog->endp[i]) { |
627 | prog->startp[i] = s + (prog->startp[i] - startpos); |
628 | prog->endp[i] = s + (prog->endp[i] - startpos); |
629 | } |
a0d0e21e |
630 | } |
631 | } |
a0d0e21e |
632 | } |
633 | return 1; |
634 | |
635 | phooey: |
a0d0e21e |
636 | return 0; |
a687059c |
637 | } |
638 | |
639 | /* |
640 | - regtry - try match at specific point |
641 | */ |
76e3520e |
642 | STATIC I32 /* 0 failure, 1 success */ |
8ac85365 |
643 | regtry(regexp *prog, char *startpos) |
a687059c |
644 | { |
c277df42 |
645 | dTHR; |
a0d0e21e |
646 | register I32 i; |
647 | register char **sp; |
648 | register char **ep; |
c277df42 |
649 | CHECKPOINT lastcp; |
a0d0e21e |
650 | |
651 | reginput = startpos; |
652 | regstartp = prog->startp; |
653 | regendp = prog->endp; |
654 | reglastparen = &prog->lastparen; |
655 | prog->lastparen = 0; |
656 | regsize = 0; |
c277df42 |
657 | if (reg_start_tmpl <= prog->nparens) { |
658 | reg_start_tmpl = prog->nparens*3/2 + 3; |
659 | if(reg_start_tmp) |
660 | Renew(reg_start_tmp, reg_start_tmpl, char*); |
661 | else |
662 | New(22,reg_start_tmp, reg_start_tmpl, char*); |
663 | } |
a0d0e21e |
664 | |
665 | sp = prog->startp; |
666 | ep = prog->endp; |
7fae4e64 |
667 | regdata = prog->data; |
a0d0e21e |
668 | if (prog->nparens) { |
669 | for (i = prog->nparens; i >= 0; i--) { |
670 | *sp++ = NULL; |
671 | *ep++ = NULL; |
a687059c |
672 | } |
a0d0e21e |
673 | } |
c277df42 |
674 | REGCP_SET; |
a0d0e21e |
675 | if (regmatch(prog->program + 1) && reginput >= regtill) { |
676 | prog->startp[0] = startpos; |
677 | prog->endp[0] = reginput; |
678 | return 1; |
679 | } |
c277df42 |
680 | REGCP_UNWIND; |
681 | return 0; |
a687059c |
682 | } |
683 | |
684 | /* |
685 | - regmatch - main matching routine |
686 | * |
687 | * Conceptually the strategy is simple: check to see whether the current |
688 | * node matches, call self recursively to see whether the rest matches, |
689 | * and then act accordingly. In practice we make some effort to avoid |
690 | * recursion, in particular by going through "ordinary" nodes (that don't |
691 | * need to know whether the rest of the match failed) by a loop instead of |
692 | * by recursion. |
693 | */ |
694 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
695 | * maybe save a little bit of pushing and popping on the stack. It also takes |
696 | * advantage of machines that use a register save mask on subroutine entry. |
697 | */ |
76e3520e |
698 | STATIC I32 /* 0 failure, 1 success */ |
c277df42 |
699 | regmatch(regnode *prog) |
a687059c |
700 | { |
c277df42 |
701 | dTHR; |
702 | register regnode *scan; /* Current node. */ |
703 | regnode *next; /* Next node. */ |
704 | regnode *inner; /* Next node in internal branch. */ |
76e3520e |
705 | register I32 nextchr; /* renamed nextchr - nextchar colides with function of same name */ |
a0d0e21e |
706 | register I32 n; /* no or next */ |
707 | register I32 ln; /* len or last */ |
708 | register char *s; /* operand or save */ |
709 | register char *locinput = reginput; |
c277df42 |
710 | register I32 c1, c2, paren; /* case fold search, parenth */ |
711 | int minmod = 0, sw = 0, logical = 0; |
4633a7c4 |
712 | #ifdef DEBUGGING |
4633a7c4 |
713 | regindent++; |
714 | #endif |
a0d0e21e |
715 | |
76e3520e |
716 | nextchr = UCHARAT(locinput); |
a0d0e21e |
717 | scan = prog; |
718 | while (scan != NULL) { |
c277df42 |
719 | #define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO) |
a687059c |
720 | #ifdef DEBUGGING |
c277df42 |
721 | # define sayYES goto yes |
722 | # define sayNO goto no |
723 | # define saySAME(x) if (x) goto yes; else goto no |
724 | # define REPORT_CODE_OFF 24 |
4633a7c4 |
725 | #else |
c277df42 |
726 | # define sayYES return 1 |
727 | # define sayNO return 0 |
728 | # define saySAME(x) return x |
a687059c |
729 | #endif |
c277df42 |
730 | DEBUG_r( { |
731 | SV *prop = sv_newmortal(); |
732 | int docolor = *colors[0]; |
733 | int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
734 | int l = (regeol - locinput > taill ? taill : regeol - locinput); |
735 | int pref_len = (locinput - bostr > (5 + taill) - l |
736 | ? (5 + taill) - l : locinput - bostr); |
737 | |
738 | if (l + pref_len < (5 + taill) && l < regeol - locinput) |
739 | l = ( regeol - locinput > (5 + taill) - pref_len |
740 | ? (5 + taill) - pref_len : regeol - locinput); |
741 | regprop(prop, scan); |
742 | PerlIO_printf(Perl_debug_log, |
743 | "%4i <%s%.*s%s%s%s%.*s%s>%*s|%*s%2d%s\n", |
744 | locinput - bostr, |
745 | colors[2], pref_len, locinput - pref_len, colors[3], |
746 | (docolor ? "" : "> <"), |
747 | colors[0], l, locinput, colors[1], |
748 | 15 - l - pref_len + 1, |
749 | "", |
750 | regindent*2, "", scan - regprogram, |
751 | SvPVX(prop)); |
752 | } ); |
a687059c |
753 | |
754 | #ifdef REGALIGN |
c277df42 |
755 | next = scan + NEXT_OFF(scan); |
a0d0e21e |
756 | if (next == scan) |
757 | next = NULL; |
a687059c |
758 | #else |
a0d0e21e |
759 | next = regnext(scan); |
a687059c |
760 | #endif |
761 | |
a0d0e21e |
762 | switch (OP(scan)) { |
763 | case BOL: |
764 | if (locinput == regbol |
765 | ? regprev == '\n' |
c277df42 |
766 | : (multiline && |
76e3520e |
767 | (nextchr || locinput < regeol) && locinput[-1] == '\n') ) |
a0d0e21e |
768 | { |
769 | /* regtill = regbol; */ |
770 | break; |
771 | } |
4633a7c4 |
772 | sayNO; |
a0d0e21e |
773 | case MBOL: |
774 | if (locinput == regbol |
775 | ? regprev == '\n' |
76e3520e |
776 | : ((nextchr || locinput < regeol) && locinput[-1] == '\n') ) |
a0d0e21e |
777 | { |
778 | break; |
779 | } |
4633a7c4 |
780 | sayNO; |
a0d0e21e |
781 | case SBOL: |
782 | if (locinput == regbol && regprev == '\n') |
783 | break; |
4633a7c4 |
784 | sayNO; |
774d564b |
785 | case GPOS: |
a0d0e21e |
786 | if (locinput == regbol) |
787 | break; |
4633a7c4 |
788 | sayNO; |
a0d0e21e |
789 | case EOL: |
790 | if (multiline) |
791 | goto meol; |
792 | else |
793 | goto seol; |
794 | case MEOL: |
795 | meol: |
76e3520e |
796 | if ((nextchr || locinput < regeol) && nextchr != '\n') |
4633a7c4 |
797 | sayNO; |
a0d0e21e |
798 | break; |
799 | case SEOL: |
800 | seol: |
76e3520e |
801 | if ((nextchr || locinput < regeol) && nextchr != '\n') |
4633a7c4 |
802 | sayNO; |
a0d0e21e |
803 | if (regeol - locinput > 1) |
4633a7c4 |
804 | sayNO; |
a0d0e21e |
805 | break; |
806 | case SANY: |
76e3520e |
807 | if (!nextchr && locinput >= regeol) |
4633a7c4 |
808 | sayNO; |
76e3520e |
809 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
810 | break; |
811 | case ANY: |
76e3520e |
812 | if (!nextchr && locinput >= regeol || nextchr == '\n') |
4633a7c4 |
813 | sayNO; |
76e3520e |
814 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
815 | break; |
bbce6d69 |
816 | case EXACT: |
161b471a |
817 | s = (char *) OPERAND(scan); |
c277df42 |
818 | ln = UCHARAT(s++); |
a0d0e21e |
819 | /* Inline the first character, for speed. */ |
76e3520e |
820 | if (UCHARAT(s) != nextchr) |
4633a7c4 |
821 | sayNO; |
a0d0e21e |
822 | if (regeol - locinput < ln) |
4633a7c4 |
823 | sayNO; |
36477c24 |
824 | if (ln > 1 && memNE(s, locinput, ln)) |
4633a7c4 |
825 | sayNO; |
a0d0e21e |
826 | locinput += ln; |
76e3520e |
827 | nextchr = UCHARAT(locinput); |
bbce6d69 |
828 | break; |
829 | case EXACTFL: |
c277df42 |
830 | reg_flags |= RF_tainted; |
bbce6d69 |
831 | /* FALL THROUGH */ |
832 | case EXACTF: |
161b471a |
833 | s = (char *) OPERAND(scan); |
c277df42 |
834 | ln = UCHARAT(s++); |
bbce6d69 |
835 | /* Inline the first character, for speed. */ |
76e3520e |
836 | if (UCHARAT(s) != nextchr && |
bbce6d69 |
837 | UCHARAT(s) != ((OP(scan) == EXACTF) |
76e3520e |
838 | ? fold : fold_locale)[nextchr]) |
bbce6d69 |
839 | sayNO; |
840 | if (regeol - locinput < ln) |
841 | sayNO; |
5f05dabc |
842 | if (ln > 1 && (OP(scan) == EXACTF |
843 | ? ibcmp(s, locinput, ln) |
844 | : ibcmp_locale(s, locinput, ln))) |
bbce6d69 |
845 | sayNO; |
846 | locinput += ln; |
76e3520e |
847 | nextchr = UCHARAT(locinput); |
a0d0e21e |
848 | break; |
849 | case ANYOF: |
161b471a |
850 | s = (char *) OPERAND(scan); |
76e3520e |
851 | if (nextchr < 0) |
852 | nextchr = UCHARAT(locinput); |
853 | if (!reginclass(s, nextchr)) |
4633a7c4 |
854 | sayNO; |
76e3520e |
855 | if (!nextchr && locinput >= regeol) |
4633a7c4 |
856 | sayNO; |
76e3520e |
857 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
858 | break; |
bbce6d69 |
859 | case ALNUML: |
c277df42 |
860 | reg_flags |= RF_tainted; |
bbce6d69 |
861 | /* FALL THROUGH */ |
a0d0e21e |
862 | case ALNUM: |
76e3520e |
863 | if (!nextchr) |
4633a7c4 |
864 | sayNO; |
bbce6d69 |
865 | if (!(OP(scan) == ALNUM |
76e3520e |
866 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
4633a7c4 |
867 | sayNO; |
76e3520e |
868 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
869 | break; |
bbce6d69 |
870 | case NALNUML: |
c277df42 |
871 | reg_flags |= RF_tainted; |
bbce6d69 |
872 | /* FALL THROUGH */ |
a0d0e21e |
873 | case NALNUM: |
76e3520e |
874 | if (!nextchr && locinput >= regeol) |
4633a7c4 |
875 | sayNO; |
bbce6d69 |
876 | if (OP(scan) == NALNUM |
76e3520e |
877 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
4633a7c4 |
878 | sayNO; |
76e3520e |
879 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
880 | break; |
bbce6d69 |
881 | case BOUNDL: |
882 | case NBOUNDL: |
c277df42 |
883 | reg_flags |= RF_tainted; |
bbce6d69 |
884 | /* FALL THROUGH */ |
a0d0e21e |
885 | case BOUND: |
bbce6d69 |
886 | case NBOUND: |
887 | /* was last char in word? */ |
888 | ln = (locinput != regbol) ? UCHARAT(locinput - 1) : regprev; |
889 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
890 | ln = isALNUM(ln); |
76e3520e |
891 | n = isALNUM(nextchr); |
bbce6d69 |
892 | } |
893 | else { |
894 | ln = isALNUM_LC(ln); |
76e3520e |
895 | n = isALNUM_LC(nextchr); |
bbce6d69 |
896 | } |
95bac841 |
897 | if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL)) |
4633a7c4 |
898 | sayNO; |
a0d0e21e |
899 | break; |
bbce6d69 |
900 | case SPACEL: |
c277df42 |
901 | reg_flags |= RF_tainted; |
bbce6d69 |
902 | /* FALL THROUGH */ |
a0d0e21e |
903 | case SPACE: |
76e3520e |
904 | if (!nextchr && locinput >= regeol) |
4633a7c4 |
905 | sayNO; |
bbce6d69 |
906 | if (!(OP(scan) == SPACE |
76e3520e |
907 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) |
4633a7c4 |
908 | sayNO; |
76e3520e |
909 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
910 | break; |
bbce6d69 |
911 | case NSPACEL: |
c277df42 |
912 | reg_flags |= RF_tainted; |
bbce6d69 |
913 | /* FALL THROUGH */ |
a0d0e21e |
914 | case NSPACE: |
76e3520e |
915 | if (!nextchr) |
4633a7c4 |
916 | sayNO; |
bbce6d69 |
917 | if (OP(scan) == SPACE |
76e3520e |
918 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 |
919 | sayNO; |
76e3520e |
920 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
921 | break; |
922 | case DIGIT: |
76e3520e |
923 | if (!isDIGIT(nextchr)) |
4633a7c4 |
924 | sayNO; |
76e3520e |
925 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
926 | break; |
927 | case NDIGIT: |
76e3520e |
928 | if (!nextchr && locinput >= regeol) |
4633a7c4 |
929 | sayNO; |
76e3520e |
930 | if (isDIGIT(nextchr)) |
4633a7c4 |
931 | sayNO; |
76e3520e |
932 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
933 | break; |
c8756f30 |
934 | case REFFL: |
c277df42 |
935 | reg_flags |= RF_tainted; |
c8756f30 |
936 | /* FALL THROUGH */ |
c277df42 |
937 | case REF: |
c8756f30 |
938 | case REFF: |
c277df42 |
939 | n = ARG(scan); /* which paren pair */ |
a0d0e21e |
940 | s = regstartp[n]; |
c277df42 |
941 | if (*reglastparen < n || !s) |
af3f8c16 |
942 | sayNO; /* Do not match unless seen CLOSEn. */ |
a0d0e21e |
943 | if (s == regendp[n]) |
944 | break; |
945 | /* Inline the first character, for speed. */ |
76e3520e |
946 | if (UCHARAT(s) != nextchr && |
c8756f30 |
947 | (OP(scan) == REF || |
948 | (UCHARAT(s) != ((OP(scan) == REFF |
76e3520e |
949 | ? fold : fold_locale)[nextchr])))) |
4633a7c4 |
950 | sayNO; |
a0d0e21e |
951 | ln = regendp[n] - s; |
952 | if (locinput + ln > regeol) |
4633a7c4 |
953 | sayNO; |
c8756f30 |
954 | if (ln > 1 && (OP(scan) == REF |
955 | ? memNE(s, locinput, ln) |
956 | : (OP(scan) == REFF |
957 | ? ibcmp(s, locinput, ln) |
958 | : ibcmp_locale(s, locinput, ln)))) |
4633a7c4 |
959 | sayNO; |
a0d0e21e |
960 | locinput += ln; |
76e3520e |
961 | nextchr = UCHARAT(locinput); |
a0d0e21e |
962 | break; |
963 | |
964 | case NOTHING: |
c277df42 |
965 | case TAIL: |
a0d0e21e |
966 | break; |
967 | case BACK: |
968 | break; |
c277df42 |
969 | case EVAL: |
970 | { |
971 | dSP; |
972 | OP_4tree *oop = op; |
973 | COP *ocurcop = curcop; |
974 | SV **ocurpad = curpad; |
975 | SV *ret; |
976 | |
977 | n = ARG(scan); |
7fae4e64 |
978 | op = (OP_4tree*)regdata->data[n]; |
c277df42 |
979 | DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", op) ); |
7fae4e64 |
980 | curpad = AvARRAY((AV*)regdata->data[n + 1]); |
c277df42 |
981 | if (!reg_eval_set) { |
982 | /* Preserve whatever is on stack now, otherwise |
983 | OP_NEXTSTATE will overwrite it. */ |
984 | SAVEINT(reg_eval_set); /* Protect against unwinding. */ |
985 | reg_eval_set = 1; |
986 | DEBUG_r(DEBUG_s( |
987 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n", stack_sp - stack_base); |
988 | )); |
989 | SAVEINT(cxstack[cxstack_ix].blk_oldsp); |
990 | cxstack[cxstack_ix].blk_oldsp = stack_sp - stack_base; |
991 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ |
992 | SAVETMPS; |
993 | /* Apparently this is not needed, judging by wantarray. */ |
994 | /* SAVEINT(cxstack[cxstack_ix].blk_gimme); |
995 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
996 | } |
997 | |
76e3520e |
998 | CALLRUNOPS(); /* Scalar context. */ |
c277df42 |
999 | SPAGAIN; |
1000 | ret = POPs; |
1001 | PUTBACK; |
1002 | |
1003 | if (logical) { |
1004 | logical = 0; |
1005 | sw = SvTRUE(ret); |
1006 | } |
1007 | op = oop; |
1008 | curpad = ocurpad; |
1009 | curcop = ocurcop; |
1010 | break; |
1011 | } |
a0d0e21e |
1012 | case OPEN: |
c277df42 |
1013 | n = ARG(scan); /* which paren pair */ |
1014 | reg_start_tmp[n] = locinput; |
a0d0e21e |
1015 | if (n > regsize) |
1016 | regsize = n; |
1017 | break; |
1018 | case CLOSE: |
c277df42 |
1019 | n = ARG(scan); /* which paren pair */ |
1020 | regstartp[n] = reg_start_tmp[n]; |
a0d0e21e |
1021 | regendp[n] = locinput; |
1022 | if (n > *reglastparen) |
1023 | *reglastparen = n; |
1024 | break; |
c277df42 |
1025 | case GROUPP: |
1026 | n = ARG(scan); /* which paren pair */ |
1027 | sw = (*reglastparen >= n && regendp[n] != NULL); |
1028 | break; |
1029 | case IFTHEN: |
1030 | if (sw) |
1031 | next = NEXTOPER(NEXTOPER(scan)); |
1032 | else { |
1033 | next = scan + ARG(scan); |
1034 | if (OP(next) == IFTHEN) /* Fake one. */ |
1035 | next = NEXTOPER(NEXTOPER(next)); |
1036 | } |
1037 | break; |
1038 | case LOGICAL: |
1039 | logical = 1; |
1040 | break; |
a0d0e21e |
1041 | case CURLYX: { |
1042 | CURCUR cc; |
1043 | CHECKPOINT cp = savestack_ix; |
c277df42 |
1044 | |
1045 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ |
1046 | next += ARG(next); |
a0d0e21e |
1047 | cc.oldcc = regcc; |
1048 | regcc = &cc; |
1049 | cc.parenfloor = *reglastparen; |
1050 | cc.cur = -1; |
1051 | cc.min = ARG1(scan); |
1052 | cc.max = ARG2(scan); |
c277df42 |
1053 | cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; |
a0d0e21e |
1054 | cc.next = next; |
1055 | cc.minmod = minmod; |
1056 | cc.lastloc = 0; |
1057 | reginput = locinput; |
1058 | n = regmatch(PREVOPER(next)); /* start on the WHILEM */ |
1059 | regcpblow(cp); |
1060 | regcc = cc.oldcc; |
4633a7c4 |
1061 | saySAME(n); |
a0d0e21e |
1062 | } |
1063 | /* NOT REACHED */ |
1064 | case WHILEM: { |
1065 | /* |
1066 | * This is really hard to understand, because after we match |
1067 | * what we're trying to match, we must make sure the rest of |
1068 | * the RE is going to match for sure, and to do that we have |
1069 | * to go back UP the parse tree by recursing ever deeper. And |
1070 | * if it fails, we have to reset our parent's current state |
1071 | * that we can try again after backing off. |
1072 | */ |
1073 | |
c277df42 |
1074 | CHECKPOINT cp, lastcp; |
a0d0e21e |
1075 | CURCUR* cc = regcc; |
c277df42 |
1076 | char *lastloc = cc->lastloc; /* Detection of 0-len. */ |
1077 | |
4633a7c4 |
1078 | n = cc->cur + 1; /* how many we know we matched */ |
a0d0e21e |
1079 | reginput = locinput; |
1080 | |
c277df42 |
1081 | DEBUG_r( |
1082 | PerlIO_printf(Perl_debug_log, |
1083 | "%*s %ld out of %ld..%ld cc=%lx\n", |
1084 | REPORT_CODE_OFF+regindent*2, "", |
1085 | (long)n, (long)cc->min, |
1086 | (long)cc->max, (long)cc) |
1087 | ); |
4633a7c4 |
1088 | |
a0d0e21e |
1089 | /* If degenerate scan matches "", assume scan done. */ |
1090 | |
579cf2c3 |
1091 | if (locinput == cc->lastloc && n >= cc->min) { |
a0d0e21e |
1092 | regcc = cc->oldcc; |
1093 | ln = regcc->cur; |
c277df42 |
1094 | DEBUG_r( |
1095 | PerlIO_printf(Perl_debug_log, "%*s empty match detected, try continuation...\n", REPORT_CODE_OFF+regindent*2, "") |
1096 | ); |
a0d0e21e |
1097 | if (regmatch(cc->next)) |
4633a7c4 |
1098 | sayYES; |
c277df42 |
1099 | DEBUG_r( |
1100 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") |
1101 | ); |
a0d0e21e |
1102 | regcc->cur = ln; |
1103 | regcc = cc; |
4633a7c4 |
1104 | sayNO; |
a0d0e21e |
1105 | } |
1106 | |
1107 | /* First just match a string of min scans. */ |
1108 | |
1109 | if (n < cc->min) { |
1110 | cc->cur = n; |
1111 | cc->lastloc = locinput; |
4633a7c4 |
1112 | if (regmatch(cc->scan)) |
1113 | sayYES; |
1114 | cc->cur = n - 1; |
c277df42 |
1115 | cc->lastloc = lastloc; |
1116 | DEBUG_r( |
1117 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") |
1118 | ); |
4633a7c4 |
1119 | sayNO; |
a0d0e21e |
1120 | } |
1121 | |
1122 | /* Prefer next over scan for minimal matching. */ |
1123 | |
1124 | if (cc->minmod) { |
1125 | regcc = cc->oldcc; |
1126 | ln = regcc->cur; |
5f05dabc |
1127 | cp = regcppush(cc->parenfloor); |
c277df42 |
1128 | REGCP_SET; |
5f05dabc |
1129 | if (regmatch(cc->next)) { |
c277df42 |
1130 | regcpblow(cp); |
4633a7c4 |
1131 | sayYES; /* All done. */ |
5f05dabc |
1132 | } |
c277df42 |
1133 | REGCP_UNWIND; |
5f05dabc |
1134 | regcppop(); |
a0d0e21e |
1135 | regcc->cur = ln; |
1136 | regcc = cc; |
1137 | |
c277df42 |
1138 | if (n >= cc->max) { /* Maximum greed exceeded? */ |
1139 | if (dowarn && n >= REG_INFTY |
1140 | && !(reg_flags & RF_warned)) { |
1141 | reg_flags |= RF_warned; |
1142 | warn("count exceeded %d", REG_INFTY - 1); |
1143 | } |
4633a7c4 |
1144 | sayNO; |
c277df42 |
1145 | } |
a687059c |
1146 | |
c277df42 |
1147 | DEBUG_r( |
1148 | PerlIO_printf(Perl_debug_log, "%*s trying longer...\n", REPORT_CODE_OFF+regindent*2, "") |
1149 | ); |
a0d0e21e |
1150 | /* Try scanning more and see if it helps. */ |
1151 | reginput = locinput; |
1152 | cc->cur = n; |
1153 | cc->lastloc = locinput; |
5f05dabc |
1154 | cp = regcppush(cc->parenfloor); |
c277df42 |
1155 | REGCP_SET; |
5f05dabc |
1156 | if (regmatch(cc->scan)) { |
c277df42 |
1157 | regcpblow(cp); |
4633a7c4 |
1158 | sayYES; |
5f05dabc |
1159 | } |
c277df42 |
1160 | DEBUG_r( |
1161 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") |
1162 | ); |
1163 | REGCP_UNWIND; |
5f05dabc |
1164 | regcppop(); |
4633a7c4 |
1165 | cc->cur = n - 1; |
c277df42 |
1166 | cc->lastloc = lastloc; |
4633a7c4 |
1167 | sayNO; |
a0d0e21e |
1168 | } |
1169 | |
1170 | /* Prefer scan over next for maximal matching. */ |
1171 | |
1172 | if (n < cc->max) { /* More greed allowed? */ |
5f05dabc |
1173 | cp = regcppush(cc->parenfloor); |
a0d0e21e |
1174 | cc->cur = n; |
1175 | cc->lastloc = locinput; |
c277df42 |
1176 | REGCP_SET; |
5f05dabc |
1177 | if (regmatch(cc->scan)) { |
c277df42 |
1178 | regcpblow(cp); |
4633a7c4 |
1179 | sayYES; |
5f05dabc |
1180 | } |
c277df42 |
1181 | REGCP_UNWIND; |
a0d0e21e |
1182 | regcppop(); /* Restore some previous $<digit>s? */ |
1183 | reginput = locinput; |
c277df42 |
1184 | DEBUG_r( |
1185 | PerlIO_printf(Perl_debug_log, "%*s failed, try continuation...\n", REPORT_CODE_OFF+regindent*2, "") |
1186 | ); |
1187 | } |
1188 | if (dowarn && n >= REG_INFTY && !(reg_flags & RF_warned)) { |
1189 | reg_flags |= RF_warned; |
1190 | warn("count exceeded %d", REG_INFTY - 1); |
a0d0e21e |
1191 | } |
1192 | |
1193 | /* Failed deeper matches of scan, so see if this one works. */ |
1194 | regcc = cc->oldcc; |
1195 | ln = regcc->cur; |
1196 | if (regmatch(cc->next)) |
4633a7c4 |
1197 | sayYES; |
c277df42 |
1198 | DEBUG_r( |
1199 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") |
1200 | ); |
a0d0e21e |
1201 | regcc->cur = ln; |
1202 | regcc = cc; |
4633a7c4 |
1203 | cc->cur = n - 1; |
c277df42 |
1204 | cc->lastloc = lastloc; |
4633a7c4 |
1205 | sayNO; |
a0d0e21e |
1206 | } |
1207 | /* NOT REACHED */ |
c277df42 |
1208 | case BRANCHJ: |
1209 | next = scan + ARG(scan); |
1210 | if (next == scan) |
1211 | next = NULL; |
1212 | inner = NEXTOPER(NEXTOPER(scan)); |
1213 | goto do_branch; |
1214 | case BRANCH: |
1215 | inner = NEXTOPER(scan); |
1216 | do_branch: |
1217 | { |
1218 | CHECKPOINT lastcp; |
1219 | c1 = OP(scan); |
1220 | if (OP(next) != c1) /* No choice. */ |
1221 | next = inner; /* Avoid recursion. */ |
a0d0e21e |
1222 | else { |
748a9306 |
1223 | int lastparen = *reglastparen; |
c277df42 |
1224 | |
1225 | REGCP_SET; |
a0d0e21e |
1226 | do { |
1227 | reginput = locinput; |
c277df42 |
1228 | if (regmatch(inner)) |
4633a7c4 |
1229 | sayYES; |
c277df42 |
1230 | REGCP_UNWIND; |
748a9306 |
1231 | for (n = *reglastparen; n > lastparen; n--) |
1232 | regendp[n] = 0; |
1233 | *reglastparen = n; |
c277df42 |
1234 | scan = next; |
a687059c |
1235 | #ifdef REGALIGN |
a0d0e21e |
1236 | /*SUPPRESS 560*/ |
c277df42 |
1237 | if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next))) |
1238 | next += n; |
a0d0e21e |
1239 | else |
c277df42 |
1240 | next = NULL; |
a687059c |
1241 | #else |
c277df42 |
1242 | next = regnext(next); |
79072805 |
1243 | #endif |
c277df42 |
1244 | inner = NEXTOPER(scan); |
1245 | if (c1 == BRANCHJ) { |
1246 | inner = NEXTOPER(inner); |
1247 | } |
1248 | } while (scan != NULL && OP(scan) == c1); |
4633a7c4 |
1249 | sayNO; |
a0d0e21e |
1250 | /* NOTREACHED */ |
a687059c |
1251 | } |
a0d0e21e |
1252 | } |
1253 | break; |
1254 | case MINMOD: |
1255 | minmod = 1; |
1256 | break; |
c277df42 |
1257 | case CURLYM: |
1258 | { |
1259 | I32 l; |
1260 | CHECKPOINT lastcp; |
1261 | |
1262 | /* We suppose that the next guy does not need |
1263 | backtracking: in particular, it is of constant length, |
1264 | and has no parenths to influence future backrefs. */ |
1265 | ln = ARG1(scan); /* min to match */ |
1266 | n = ARG2(scan); /* max to match */ |
1267 | #ifdef REGALIGN_STRUCT |
1268 | paren = scan->flags; |
1269 | if (paren) { |
1270 | if (paren > regsize) |
1271 | regsize = paren; |
1272 | if (paren > *reglastparen) |
1273 | *reglastparen = paren; |
1274 | } |
1275 | #endif |
1276 | scan = NEXTOPER(scan) + 4/sizeof(regnode); |
1277 | if (paren) |
1278 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
1279 | reginput = locinput; |
1280 | if (minmod) { |
1281 | minmod = 0; |
1282 | if (ln && regrepeat_hard(scan, ln, &l) < ln) |
1283 | sayNO; |
1284 | if (l == 0 && n >= ln |
1285 | /* In fact, this is tricky. If paren, then the |
1286 | fact that we did/didnot match may influence |
1287 | future execution. */ |
1288 | && !(paren && ln == 0)) |
1289 | ln = n; |
1290 | locinput = reginput; |
1291 | if (regkind[(U8)OP(next)] == EXACT) { |
1292 | c1 = UCHARAT(OPERAND(next) + 1); |
1293 | if (OP(next) == EXACTF) |
1294 | c2 = fold[c1]; |
1295 | else if (OP(next) == EXACTFL) |
1296 | c2 = fold_locale[c1]; |
1297 | else |
1298 | c2 = c1; |
1299 | } else |
1300 | c1 = c2 = -1000; |
1301 | REGCP_SET; |
1302 | while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */ |
1303 | /* If it could work, try it. */ |
1304 | if (c1 == -1000 || |
1305 | UCHARAT(reginput) == c1 || |
1306 | UCHARAT(reginput) == c2) |
1307 | { |
1308 | if (paren) { |
1309 | if (n) { |
1310 | regstartp[paren] = reginput - l; |
1311 | regendp[paren] = reginput; |
1312 | } else |
1313 | regendp[paren] = NULL; |
1314 | } |
1315 | if (regmatch(next)) |
1316 | sayYES; |
1317 | REGCP_UNWIND; |
1318 | } |
1319 | /* Couldn't or didn't -- move forward. */ |
1320 | reginput = locinput; |
1321 | if (regrepeat_hard(scan, 1, &l)) { |
1322 | ln++; |
1323 | locinput = reginput; |
1324 | } |
1325 | else |
1326 | sayNO; |
1327 | } |
1328 | } else { |
1329 | n = regrepeat_hard(scan, n, &l); |
1330 | if (n != 0 && l == 0 |
1331 | /* In fact, this is tricky. If paren, then the |
1332 | fact that we did/didnot match may influence |
1333 | future execution. */ |
1334 | && !(paren && ln == 0)) |
1335 | ln = n; |
1336 | locinput = reginput; |
1337 | DEBUG_r( |
1338 | PerlIO_printf(Perl_debug_log, "%*s matched %ld times, len=%ld...\n", REPORT_CODE_OFF+regindent*2, "", n, l) |
1339 | ); |
1340 | if (n >= ln) { |
1341 | if (regkind[(U8)OP(next)] == EXACT) { |
1342 | c1 = UCHARAT(OPERAND(next) + 1); |
1343 | if (OP(next) == EXACTF) |
1344 | c2 = fold[c1]; |
1345 | else if (OP(next) == EXACTFL) |
1346 | c2 = fold_locale[c1]; |
1347 | else |
1348 | c2 = c1; |
1349 | } else |
1350 | c1 = c2 = -1000; |
1351 | } |
1352 | REGCP_SET; |
1353 | while (n >= ln) { |
1354 | /* If it could work, try it. */ |
1355 | if (c1 == -1000 || |
1356 | UCHARAT(reginput) == c1 || |
1357 | UCHARAT(reginput) == c2) |
1358 | { |
1359 | DEBUG_r( |
1360 | PerlIO_printf(Perl_debug_log, "%*s trying tail with n=%ld...\n", REPORT_CODE_OFF+regindent*2, "", n) |
1361 | ); |
1362 | if (paren) { |
1363 | if (n) { |
1364 | regstartp[paren] = reginput - l; |
1365 | regendp[paren] = reginput; |
1366 | } else |
1367 | regendp[paren] = NULL; |
1368 | } |
1369 | if (regmatch(next)) |
1370 | sayYES; |
1371 | REGCP_UNWIND; |
1372 | } |
1373 | /* Couldn't or didn't -- back up. */ |
1374 | n--; |
1375 | locinput -= l; |
1376 | reginput = locinput; |
1377 | } |
1378 | } |
1379 | sayNO; |
1380 | break; |
1381 | } |
1382 | case CURLYN: |
1383 | paren = scan->flags; /* Which paren to set */ |
1384 | if (paren > regsize) |
1385 | regsize = paren; |
1386 | if (paren > *reglastparen) |
1387 | *reglastparen = paren; |
1388 | ln = ARG1(scan); /* min to match */ |
1389 | n = ARG2(scan); /* max to match */ |
1390 | scan = regnext(NEXTOPER(scan) + 4/sizeof(regnode)); |
1391 | goto repeat; |
a0d0e21e |
1392 | case CURLY: |
c277df42 |
1393 | paren = 0; |
a0d0e21e |
1394 | ln = ARG1(scan); /* min to match */ |
1395 | n = ARG2(scan); /* max to match */ |
c277df42 |
1396 | scan = NEXTOPER(scan) + 4/sizeof(regnode); |
a0d0e21e |
1397 | goto repeat; |
1398 | case STAR: |
1399 | ln = 0; |
c277df42 |
1400 | n = REG_INFTY; |
a0d0e21e |
1401 | scan = NEXTOPER(scan); |
c277df42 |
1402 | paren = 0; |
a0d0e21e |
1403 | goto repeat; |
1404 | case PLUS: |
c277df42 |
1405 | ln = 1; |
1406 | n = REG_INFTY; |
1407 | scan = NEXTOPER(scan); |
1408 | paren = 0; |
1409 | repeat: |
a0d0e21e |
1410 | /* |
1411 | * Lookahead to avoid useless match attempts |
1412 | * when we know what character comes next. |
1413 | */ |
bbce6d69 |
1414 | if (regkind[(U8)OP(next)] == EXACT) { |
1415 | c1 = UCHARAT(OPERAND(next) + 1); |
1416 | if (OP(next) == EXACTF) |
1417 | c2 = fold[c1]; |
1418 | else if (OP(next) == EXACTFL) |
1419 | c2 = fold_locale[c1]; |
1420 | else |
1421 | c2 = c1; |
1422 | } |
a0d0e21e |
1423 | else |
bbce6d69 |
1424 | c1 = c2 = -1000; |
a0d0e21e |
1425 | reginput = locinput; |
1426 | if (minmod) { |
c277df42 |
1427 | CHECKPOINT lastcp; |
a0d0e21e |
1428 | minmod = 0; |
1429 | if (ln && regrepeat(scan, ln) < ln) |
4633a7c4 |
1430 | sayNO; |
c277df42 |
1431 | REGCP_SET; |
1432 | while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ |
a0d0e21e |
1433 | /* If it could work, try it. */ |
bbce6d69 |
1434 | if (c1 == -1000 || |
1435 | UCHARAT(reginput) == c1 || |
1436 | UCHARAT(reginput) == c2) |
1437 | { |
c277df42 |
1438 | if (paren) { |
1439 | if (n) { |
1440 | regstartp[paren] = reginput - 1; |
1441 | regendp[paren] = reginput; |
1442 | } else |
1443 | regendp[paren] = NULL; |
1444 | } |
a0d0e21e |
1445 | if (regmatch(next)) |
4633a7c4 |
1446 | sayYES; |
c277df42 |
1447 | REGCP_UNWIND; |
bbce6d69 |
1448 | } |
c277df42 |
1449 | /* Couldn't or didn't -- move forward. */ |
748a9306 |
1450 | reginput = locinput + ln; |
a0d0e21e |
1451 | if (regrepeat(scan, 1)) { |
1452 | ln++; |
1453 | reginput = locinput + ln; |
c277df42 |
1454 | } else |
4633a7c4 |
1455 | sayNO; |
a0d0e21e |
1456 | } |
1457 | } |
1458 | else { |
c277df42 |
1459 | CHECKPOINT lastcp; |
a0d0e21e |
1460 | n = regrepeat(scan, n); |
1461 | if (ln < n && regkind[(U8)OP(next)] == EOL && |
c277df42 |
1462 | (!multiline || OP(next) == SEOL)) |
a0d0e21e |
1463 | ln = n; /* why back off? */ |
c277df42 |
1464 | REGCP_SET; |
1465 | if (paren) { |
1466 | while (n >= ln) { |
1467 | /* If it could work, try it. */ |
1468 | if (c1 == -1000 || |
1469 | UCHARAT(reginput) == c1 || |
1470 | UCHARAT(reginput) == c2) |
1471 | { |
1472 | if (paren && n) { |
1473 | if (n) { |
1474 | regstartp[paren] = reginput - 1; |
1475 | regendp[paren] = reginput; |
1476 | } else |
1477 | regendp[paren] = NULL; |
1478 | } |
1479 | if (regmatch(next)) |
1480 | sayYES; |
1481 | REGCP_UNWIND; |
1482 | } |
1483 | /* Couldn't or didn't -- back up. */ |
1484 | n--; |
1485 | reginput = locinput + n; |
1486 | } |
1487 | } else { |
1488 | while (n >= ln) { |
1489 | /* If it could work, try it. */ |
1490 | if (c1 == -1000 || |
1491 | UCHARAT(reginput) == c1 || |
1492 | UCHARAT(reginput) == c2) |
1493 | { |
1494 | if (regmatch(next)) |
1495 | sayYES; |
1496 | REGCP_UNWIND; |
1497 | } |
1498 | /* Couldn't or didn't -- back up. */ |
1499 | n--; |
1500 | reginput = locinput + n; |
bbce6d69 |
1501 | } |
a0d0e21e |
1502 | } |
1503 | } |
4633a7c4 |
1504 | sayNO; |
c277df42 |
1505 | break; |
a0d0e21e |
1506 | case SUCCEED: |
1507 | case END: |
1508 | reginput = locinput; /* put where regtry can find it */ |
4633a7c4 |
1509 | sayYES; /* Success! */ |
c277df42 |
1510 | case SUSPEND: |
1511 | n = 1; |
1512 | goto do_ifmatch; |
a0d0e21e |
1513 | case UNLESSM: |
c277df42 |
1514 | n = 0; |
1515 | if (locinput < bostr + scan->flags) |
1516 | goto say_yes; |
1517 | goto do_ifmatch; |
1518 | case IFMATCH: |
1519 | n = 1; |
1520 | if (locinput < bostr + scan->flags) |
1521 | goto say_no; |
1522 | do_ifmatch: |
1523 | reginput = locinput - scan->flags; |
1524 | inner = NEXTOPER(NEXTOPER(scan)); |
1525 | if (regmatch(inner) != n) { |
1526 | say_no: |
1527 | if (logical) { |
1528 | logical = 0; |
1529 | sw = 0; |
1530 | goto do_longjump; |
1531 | } else |
1532 | sayNO; |
1533 | } |
1534 | say_yes: |
1535 | if (logical) { |
1536 | logical = 0; |
1537 | sw = 1; |
1538 | } |
fe44a5e8 |
1539 | if (OP(scan) == SUSPEND) { |
c277df42 |
1540 | locinput = reginput; |
565764a8 |
1541 | nextchr = UCHARAT(locinput); |
fe44a5e8 |
1542 | } |
c277df42 |
1543 | /* FALL THROUGH. */ |
1544 | case LONGJMP: |
1545 | do_longjump: |
1546 | next = scan + ARG(scan); |
1547 | if (next == scan) |
1548 | next = NULL; |
a0d0e21e |
1549 | break; |
1550 | default: |
c030ccd9 |
1551 | PerlIO_printf(PerlIO_stderr(), "%lx %d\n", |
c277df42 |
1552 | (unsigned long)scan, OP(scan)); |
a0d0e21e |
1553 | FAIL("regexp memory corruption"); |
a687059c |
1554 | } |
a0d0e21e |
1555 | scan = next; |
1556 | } |
a687059c |
1557 | |
a0d0e21e |
1558 | /* |
1559 | * We get here only if there's trouble -- normally "case END" is |
1560 | * the terminating point. |
1561 | */ |
1562 | FAIL("corrupted regexp pointers"); |
1563 | /*NOTREACHED*/ |
4633a7c4 |
1564 | sayNO; |
1565 | |
1566 | yes: |
1567 | #ifdef DEBUGGING |
1568 | regindent--; |
1569 | #endif |
1570 | return 1; |
1571 | |
1572 | no: |
1573 | #ifdef DEBUGGING |
1574 | regindent--; |
1575 | #endif |
a0d0e21e |
1576 | return 0; |
a687059c |
1577 | } |
1578 | |
1579 | /* |
1580 | - regrepeat - repeatedly match something simple, report how many |
1581 | */ |
1582 | /* |
1583 | * [This routine now assumes that it will only match on things of length 1. |
1584 | * That was true before, but now we assume scan - reginput is the count, |
1585 | * rather than incrementing count on every character.] |
1586 | */ |
76e3520e |
1587 | STATIC I32 |
c277df42 |
1588 | regrepeat(regnode *p, I32 max) |
a687059c |
1589 | { |
a0d0e21e |
1590 | register char *scan; |
1591 | register char *opnd; |
1592 | register I32 c; |
1593 | register char *loceol = regeol; |
1594 | |
1595 | scan = reginput; |
c277df42 |
1596 | if (max != REG_INFTY && max < loceol - scan) |
a0d0e21e |
1597 | loceol = scan + max; |
161b471a |
1598 | opnd = (char *) OPERAND(p); |
a0d0e21e |
1599 | switch (OP(p)) { |
1600 | case ANY: |
1601 | while (scan < loceol && *scan != '\n') |
1602 | scan++; |
1603 | break; |
1604 | case SANY: |
1605 | scan = loceol; |
1606 | break; |
bbce6d69 |
1607 | case EXACT: /* length of string is 1 */ |
1608 | c = UCHARAT(++opnd); |
1609 | while (scan < loceol && UCHARAT(scan) == c) |
1610 | scan++; |
1611 | break; |
1612 | case EXACTF: /* length of string is 1 */ |
1613 | c = UCHARAT(++opnd); |
1614 | while (scan < loceol && |
1615 | (UCHARAT(scan) == c || UCHARAT(scan) == fold[c])) |
1616 | scan++; |
1617 | break; |
1618 | case EXACTFL: /* length of string is 1 */ |
c277df42 |
1619 | reg_flags |= RF_tainted; |
bbce6d69 |
1620 | c = UCHARAT(++opnd); |
1621 | while (scan < loceol && |
1622 | (UCHARAT(scan) == c || UCHARAT(scan) == fold_locale[c])) |
a0d0e21e |
1623 | scan++; |
1624 | break; |
1625 | case ANYOF: |
bbce6d69 |
1626 | while (scan < loceol && reginclass(opnd, *scan)) |
a0d0e21e |
1627 | scan++; |
a0d0e21e |
1628 | break; |
1629 | case ALNUM: |
1630 | while (scan < loceol && isALNUM(*scan)) |
1631 | scan++; |
1632 | break; |
bbce6d69 |
1633 | case ALNUML: |
c277df42 |
1634 | reg_flags |= RF_tainted; |
bbce6d69 |
1635 | while (scan < loceol && isALNUM_LC(*scan)) |
1636 | scan++; |
1637 | break; |
a0d0e21e |
1638 | case NALNUM: |
1639 | while (scan < loceol && !isALNUM(*scan)) |
1640 | scan++; |
1641 | break; |
bbce6d69 |
1642 | case NALNUML: |
c277df42 |
1643 | reg_flags |= RF_tainted; |
bbce6d69 |
1644 | while (scan < loceol && !isALNUM_LC(*scan)) |
1645 | scan++; |
1646 | break; |
a0d0e21e |
1647 | case SPACE: |
1648 | while (scan < loceol && isSPACE(*scan)) |
1649 | scan++; |
1650 | break; |
bbce6d69 |
1651 | case SPACEL: |
c277df42 |
1652 | reg_flags |= RF_tainted; |
bbce6d69 |
1653 | while (scan < loceol && isSPACE_LC(*scan)) |
1654 | scan++; |
1655 | break; |
a0d0e21e |
1656 | case NSPACE: |
1657 | while (scan < loceol && !isSPACE(*scan)) |
1658 | scan++; |
1659 | break; |
bbce6d69 |
1660 | case NSPACEL: |
c277df42 |
1661 | reg_flags |= RF_tainted; |
bbce6d69 |
1662 | while (scan < loceol && !isSPACE_LC(*scan)) |
1663 | scan++; |
1664 | break; |
a0d0e21e |
1665 | case DIGIT: |
1666 | while (scan < loceol && isDIGIT(*scan)) |
1667 | scan++; |
1668 | break; |
1669 | case NDIGIT: |
1670 | while (scan < loceol && !isDIGIT(*scan)) |
1671 | scan++; |
1672 | break; |
1673 | default: /* Called on something of 0 width. */ |
1674 | break; /* So match right here or not at all. */ |
1675 | } |
a687059c |
1676 | |
a0d0e21e |
1677 | c = scan - reginput; |
1678 | reginput = scan; |
a687059c |
1679 | |
c277df42 |
1680 | DEBUG_r( |
1681 | { |
1682 | SV *prop = sv_newmortal(); |
1683 | |
1684 | regprop(prop, p); |
1685 | PerlIO_printf(Perl_debug_log, |
1686 | "%*s %s can match %ld times out of %ld...\n", |
1687 | REPORT_CODE_OFF+1, "", SvPVX(prop),c,max); |
1688 | }); |
1689 | |
a0d0e21e |
1690 | return(c); |
a687059c |
1691 | } |
1692 | |
1693 | /* |
c277df42 |
1694 | - regrepeat_hard - repeatedly match something, report total lenth and length |
1695 | * |
1696 | * The repeater is supposed to have constant length. |
1697 | */ |
1698 | |
76e3520e |
1699 | STATIC I32 |
c277df42 |
1700 | regrepeat_hard(regnode *p, I32 max, I32 *lp) |
1701 | { |
1702 | register char *scan; |
1703 | register char *start; |
1704 | register char *loceol = regeol; |
1705 | I32 l = -1; |
1706 | |
1707 | start = reginput; |
1708 | while (reginput < loceol && (scan = reginput, regmatch(p))) { |
1709 | if (l == -1) { |
1710 | *lp = l = reginput - start; |
1711 | if (max != REG_INFTY && l*max < loceol - scan) |
1712 | loceol = scan + l*max; |
1713 | if (l == 0) { |
1714 | return max; |
1715 | } |
1716 | } |
1717 | } |
1718 | if (reginput < loceol) |
1719 | reginput = scan; |
1720 | else |
1721 | scan = reginput; |
1722 | |
1723 | return (scan - start)/l; |
1724 | } |
1725 | |
1726 | /* |
bbce6d69 |
1727 | - regclass - determine if a character falls into a character class |
1728 | */ |
1729 | |
76e3520e |
1730 | STATIC bool |
8ac85365 |
1731 | reginclass(register char *p, register I32 c) |
bbce6d69 |
1732 | { |
1733 | char flags = *p; |
1734 | bool match = FALSE; |
1735 | |
1736 | c &= 0xFF; |
1737 | if (p[1 + (c >> 3)] & (1 << (c & 7))) |
1738 | match = TRUE; |
1739 | else if (flags & ANYOF_FOLD) { |
1740 | I32 cf; |
1741 | if (flags & ANYOF_LOCALE) { |
c277df42 |
1742 | reg_flags |= RF_tainted; |
bbce6d69 |
1743 | cf = fold_locale[c]; |
1744 | } |
1745 | else |
1746 | cf = fold[c]; |
1747 | if (p[1 + (cf >> 3)] & (1 << (cf & 7))) |
1748 | match = TRUE; |
1749 | } |
1750 | |
1751 | if (!match && (flags & ANYOF_ISA)) { |
c277df42 |
1752 | reg_flags |= RF_tainted; |
bbce6d69 |
1753 | |
1754 | if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) || |
1755 | ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) || |
1756 | ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) || |
1757 | ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c))) |
1758 | { |
1759 | match = TRUE; |
1760 | } |
1761 | } |
1762 | |
1763 | return match ^ ((flags & ANYOF_INVERT) != 0); |
1764 | } |
1765 | |
161b471a |
1766 | |
1767 | |