Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | /* regexec.c |
2 | */ | |
3 | ||
4 | /* | |
5 | * "One Ring to rule them all, One Ring to find them..." | |
6 | */ | |
7 | ||
a687059c LW |
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 AD |
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 LW |
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 LW |
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 LW |
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 IZ |
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 LW |
62 | #ifndef STATIC |
63 | #define STATIC static | |
64 | #endif | |
65 | ||
76e3520e | 66 | #ifndef PERL_OBJECT |
a0d0e21e LW |
67 | typedef I32 CHECKPOINT; |
68 | ||
c277df42 IZ |
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)); | |
ae5c130c | 77 | |
c277df42 | 78 | static bool reginclass _((char *p, I32 c)); |
55497cff | 79 | static CHECKPOINT regcppush _((I32 parenfloor)); |
80 | static char * regcppop _((void)); | |
76e3520e | 81 | #endif |
ae5c130c | 82 | #define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c)) |
a0d0e21e | 83 | |
76e3520e | 84 | STATIC CHECKPOINT |
8ac85365 | 85 | regcppush(I32 parenfloor) |
a0d0e21e | 86 | { |
11343788 | 87 | dTHR; |
a0d0e21e | 88 | int retval = savestack_ix; |
c277df42 | 89 | int i = (regsize - parenfloor) * 4; |
a0d0e21e LW |
90 | int p; |
91 | ||
92 | SSCHECK(i + 5); | |
93 | for (p = regsize; p > parenfloor; p--) { | |
94 | SSPUSHPTR(regendp[p]); | |
95 | SSPUSHPTR(regstartp[p]); | |
c5254dd6 | 96 | SSPUSHPTR(reg_start_tmp[p]); |
a0d0e21e LW |
97 | SSPUSHINT(p); |
98 | } | |
99 | SSPUSHINT(regsize); | |
100 | SSPUSHINT(*reglastparen); | |
101 | SSPUSHPTR(reginput); | |
102 | SSPUSHINT(i + 3); | |
103 | SSPUSHINT(SAVEt_REGCONTEXT); | |
104 | return retval; | |
105 | } | |
106 | ||
c277df42 IZ |
107 | /* These are needed since we do not localize EVAL nodes: */ |
108 | # define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, " Setting an EVAL scope, savestack=%i\n", savestack_ix)); lastcp = savestack_ix | |
109 | # 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) | |
110 | ||
76e3520e | 111 | STATIC char * |
8ac85365 | 112 | regcppop(void) |
a0d0e21e | 113 | { |
11343788 | 114 | dTHR; |
a0d0e21e LW |
115 | I32 i = SSPOPINT; |
116 | U32 paren = 0; | |
117 | char *input; | |
118 | char *tmps; | |
119 | assert(i == SAVEt_REGCONTEXT); | |
120 | i = SSPOPINT; | |
121 | input = (char *) SSPOPPTR; | |
122 | *reglastparen = SSPOPINT; | |
123 | regsize = SSPOPINT; | |
c277df42 | 124 | for (i -= 3; i > 0; i -= 4) { |
a0d0e21e | 125 | paren = (U32)SSPOPINT; |
c277df42 | 126 | reg_start_tmp[paren] = (char *) SSPOPPTR; |
a0d0e21e LW |
127 | regstartp[paren] = (char *) SSPOPPTR; |
128 | tmps = (char*)SSPOPPTR; | |
129 | if (paren <= *reglastparen) | |
130 | regendp[paren] = tmps; | |
c277df42 IZ |
131 | DEBUG_r( |
132 | PerlIO_printf(Perl_debug_log, " restoring \\%d to %d(%d)..%d%s\n", | |
133 | paren, regstartp[paren] - regbol, | |
134 | reg_start_tmp[paren] - regbol, | |
135 | regendp[paren] - regbol, | |
136 | (paren > *reglastparen ? "(no)" : "")); | |
137 | ); | |
a0d0e21e | 138 | } |
c277df42 IZ |
139 | DEBUG_r( |
140 | if (*reglastparen + 1 <= regnpar) { | |
141 | PerlIO_printf(Perl_debug_log, " restoring \\%d..\\%d to undef\n", | |
142 | *reglastparen + 1, regnpar); | |
143 | } | |
144 | ); | |
a0d0e21e LW |
145 | for (paren = *reglastparen + 1; paren <= regnpar; paren++) { |
146 | if (paren > regsize) | |
147 | regstartp[paren] = Nullch; | |
148 | regendp[paren] = Nullch; | |
149 | } | |
150 | return input; | |
151 | } | |
152 | ||
c277df42 | 153 | #define regcpblow(cp) LEAVE_SCOPE(cp) |
a0d0e21e | 154 | |
a687059c | 155 | /* |
e50aee73 | 156 | * pregexec and friends |
a687059c LW |
157 | */ |
158 | ||
159 | /* | |
c277df42 | 160 | - pregexec - match a regexp against a string |
a687059c | 161 | */ |
c277df42 IZ |
162 | I32 |
163 | pregexec(register regexp *prog, char *stringarg, register char *strend, char *strbeg, I32 minend, SV *screamer, U32 nosave) | |
164 | /* strend: pointer to null at end of string */ | |
165 | /* strbeg: real beginning of string */ | |
166 | /* minend: end of match must be >=minend after stringarg. */ | |
167 | /* nosave: For optimizations. */ | |
168 | { | |
169 | return | |
170 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, | |
171 | nosave ? 0 : REXEC_COPY_STR); | |
172 | } | |
173 | ||
a687059c | 174 | /* |
c277df42 | 175 | - regexec_flags - match a regexp against a string |
a687059c | 176 | */ |
79072805 | 177 | I32 |
c277df42 IZ |
178 | regexec_flags(register regexp *prog, char *stringarg, register char *strend, char *strbeg, I32 minend, SV *screamer, void *data, U32 flags) |
179 | /* strend: pointer to null at end of string */ | |
180 | /* strbeg: real beginning of string */ | |
181 | /* minend: end of match must be >=minend after stringarg. */ | |
182 | /* data: May be used for some additional optimizations. */ | |
183 | /* nosave: For optimizations. */ | |
a687059c | 184 | { |
a0d0e21e | 185 | register char *s; |
c277df42 | 186 | register regnode *c; |
a0d0e21e LW |
187 | register char *startpos = stringarg; |
188 | register I32 tmp; | |
c277df42 | 189 | I32 minlen; /* must match at least this many chars */ |
a0d0e21e LW |
190 | I32 dontbother = 0; /* how many characters not to try at end */ |
191 | CURCUR cc; | |
c277df42 IZ |
192 | I32 start_shift = 0; /* Offset of the start to find |
193 | constant substr. */ | |
194 | I32 end_shift = 0; /* Same for the end. */ | |
195 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
196 | char *scream_olds; | |
a687059c | 197 | |
a0d0e21e | 198 | cc.cur = 0; |
4633a7c4 | 199 | cc.oldcc = 0; |
a0d0e21e LW |
200 | regcc = &cc; |
201 | ||
c277df42 | 202 | regprecomp = prog->precomp; /* Needed for error messages. */ |
a0d0e21e LW |
203 | #ifdef DEBUGGING |
204 | regnarrate = debug & 512; | |
205 | regprogram = prog->program; | |
206 | #endif | |
207 | ||
208 | /* Be paranoid... */ | |
209 | if (prog == NULL || startpos == NULL) { | |
210 | croak("NULL regexp parameter"); | |
211 | return 0; | |
212 | } | |
213 | ||
c277df42 IZ |
214 | minlen = prog->minlen; |
215 | if (strend - startpos < minlen) goto phooey; | |
216 | ||
a0d0e21e LW |
217 | if (startpos == strbeg) /* is ^ valid at stringarg? */ |
218 | regprev = '\n'; | |
219 | else { | |
220 | regprev = stringarg[-1]; | |
221 | if (!multiline && regprev == '\n') | |
222 | regprev = '\0'; /* force ^ to NOT match */ | |
223 | } | |
bbce6d69 | 224 | |
a0d0e21e LW |
225 | /* Check validity of program. */ |
226 | if (UCHARAT(prog->program) != MAGIC) { | |
227 | FAIL("corrupted regexp program"); | |
228 | } | |
229 | ||
bbce6d69 | 230 | regnpar = prog->nparens; |
c277df42 IZ |
231 | reg_flags = 0; |
232 | reg_eval_set = 0; | |
a0d0e21e LW |
233 | |
234 | /* If there is a "must appear" string, look for it. */ | |
235 | s = startpos; | |
c277df42 IZ |
236 | if (!(flags & REXEC_CHECKED) |
237 | && prog->check_substr != Nullsv && | |
774d564b | 238 | !(prog->reganch & ROPT_ANCH_GPOS) && |
c277df42 IZ |
239 | (!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL)) |
240 | || (multiline && prog->check_substr == prog->anchored_substr)) ) | |
a0d0e21e | 241 | { |
c277df42 IZ |
242 | start_shift = prog->check_offset_min; |
243 | /* Should be nonnegative! */ | |
244 | end_shift = minlen - start_shift - SvCUR(prog->check_substr); | |
245 | if (screamer) { | |
246 | if (screamfirst[BmRARE(prog->check_substr)] >= 0) | |
247 | s = screaminstr(screamer, prog->check_substr, | |
248 | start_shift + (stringarg - strbeg), | |
249 | end_shift, &scream_pos, 0); | |
a0d0e21e LW |
250 | else |
251 | s = Nullch; | |
c277df42 | 252 | scream_olds = s; |
0a12ae7d | 253 | } |
a0d0e21e | 254 | else |
c277df42 IZ |
255 | s = fbm_instr((unsigned char*)s + start_shift, |
256 | (unsigned char*)strend - end_shift, | |
257 | prog->check_substr); | |
a0d0e21e | 258 | if (!s) { |
c277df42 | 259 | ++BmUSEFUL(prog->check_substr); /* hooray */ |
a0d0e21e | 260 | goto phooey; /* not present */ |
c277df42 IZ |
261 | } else if ((s - stringarg) > prog->check_offset_max) { |
262 | ++BmUSEFUL(prog->check_substr); /* hooray/2 */ | |
263 | s -= prog->check_offset_max; | |
264 | } else if (!prog->naughty | |
265 | && --BmUSEFUL(prog->check_substr) < 0 | |
266 | && prog->check_substr == prog->float_substr) { /* boo */ | |
267 | SvREFCNT_dec(prog->check_substr); | |
268 | prog->check_substr = Nullsv; /* disable */ | |
269 | prog->float_substr = Nullsv; /* clear */ | |
a0d0e21e | 270 | s = startpos; |
c277df42 | 271 | } else s = startpos; |
a0d0e21e | 272 | } |
a687059c | 273 | |
c277df42 | 274 | /* Mark beginning of line for ^ and lookbehind. */ |
a0d0e21e | 275 | regbol = startpos; |
c277df42 | 276 | bostr = strbeg; |
a687059c | 277 | |
a0d0e21e LW |
278 | /* Mark end of line for $ (and such) */ |
279 | regeol = strend; | |
a687059c | 280 | |
a0d0e21e LW |
281 | /* see how far we have to get to not match where we matched before */ |
282 | regtill = startpos+minend; | |
a687059c | 283 | |
c277df42 IZ |
284 | DEBUG_r( |
285 | PerlIO_printf(Perl_debug_log, | |
286 | "Matching `%.60s%s' against `%.*s%s'\n", | |
287 | prog->precomp, | |
288 | (strlen(prog->precomp) > 60 ? "..." : ""), | |
289 | (strend - startpos > 60 ? 60 : strend - startpos), | |
290 | startpos, | |
291 | (strend - startpos > 60 ? "..." : "")) | |
292 | ); | |
293 | ||
a0d0e21e | 294 | /* Simplest case: anchored match need be tried only once. */ |
774d564b | 295 | /* [unless only anchor is BOL and multiline is set] */ |
a0d0e21e LW |
296 | if (prog->reganch & ROPT_ANCH) { |
297 | if (regtry(prog, startpos)) | |
298 | goto got_it; | |
774d564b | 299 | else if (!(prog->reganch & ROPT_ANCH_GPOS) && |
c277df42 IZ |
300 | (multiline || (prog->reganch & ROPT_IMPLICIT) |
301 | || (prog->reganch & ROPT_ANCH_MBOL))) | |
774d564b | 302 | { |
a0d0e21e LW |
303 | if (minlen) |
304 | dontbother = minlen - 1; | |
305 | strend -= dontbother; | |
306 | /* for multiline we only have to try after newlines */ | |
307 | if (s > startpos) | |
308 | s--; | |
309 | while (s < strend) { | |
310 | if (*s++ == '\n') { | |
311 | if (s < strend && regtry(prog, s)) | |
312 | goto got_it; | |
313 | } | |
35c8bce7 | 314 | } |
35c8bce7 | 315 | } |
a0d0e21e LW |
316 | goto phooey; |
317 | } | |
35c8bce7 | 318 | |
a0d0e21e | 319 | /* Messy cases: unanchored match. */ |
c277df42 IZ |
320 | if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { |
321 | /* we have /x+whatever/ */ | |
322 | /* it must be a one character string */ | |
323 | char ch = SvPVX(prog->anchored_substr)[0]; | |
324 | while (s < strend) { | |
325 | if (*s == ch) { | |
326 | if (regtry(prog, s)) goto got_it; | |
a0d0e21e | 327 | s++; |
c277df42 IZ |
328 | while (s < strend && *s == ch) |
329 | s++; | |
a0d0e21e | 330 | } |
c277df42 | 331 | s++; |
a687059c | 332 | } |
c277df42 IZ |
333 | } |
334 | /*SUPPRESS 560*/ | |
335 | else if (prog->anchored_substr != Nullsv | |
336 | || (prog->float_substr != Nullsv | |
337 | && prog->float_max_offset < strend - s)) { | |
338 | SV *must = prog->anchored_substr | |
339 | ? prog->anchored_substr : prog->float_substr; | |
340 | I32 back_max = | |
341 | prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; | |
342 | I32 back_min = | |
343 | prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; | |
344 | I32 delta = back_max - back_min; | |
345 | char *last = strend - SvCUR(must) - back_min; /* Cannot start after this */ | |
346 | char *last1 = s - 1; /* Last position checked before */ | |
347 | ||
348 | /* XXXX check_substr already used to find `s', can optimize if | |
349 | check_substr==must. */ | |
350 | scream_pos = -1; | |
351 | dontbother = end_shift; | |
352 | strend -= dontbother; | |
353 | while ( (s <= last) && | |
354 | (screamer | |
355 | ? (s = screaminstr(screamer, must, s + back_min - strbeg, | |
356 | end_shift, &scream_pos, 0)) | |
357 | : (s = fbm_instr((unsigned char*)s + back_min, | |
358 | (unsigned char*)strend, must))) ) { | |
359 | if (s - back_max > last1) { | |
360 | last1 = s - back_min; | |
361 | s = s - back_max; | |
362 | } else { | |
363 | char *t = last1 + 1; | |
364 | ||
365 | last1 = s - back_min; | |
366 | s = t; | |
a0d0e21e | 367 | } |
c277df42 | 368 | while (s <= last1) { |
a0d0e21e LW |
369 | if (regtry(prog, s)) |
370 | goto got_it; | |
371 | s++; | |
372 | } | |
373 | } | |
374 | goto phooey; | |
c277df42 | 375 | } else if (c = prog->regstclass) { |
a0d0e21e | 376 | I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
161b471a | 377 | char *Class; |
a687059c | 378 | |
a0d0e21e LW |
379 | if (minlen) |
380 | dontbother = minlen - 1; | |
381 | strend -= dontbother; /* don't bother with what can't match */ | |
382 | tmp = 1; | |
383 | /* We know what class it must start with. */ | |
384 | switch (OP(c)) { | |
385 | case ANYOF: | |
161b471a | 386 | Class = (char *) OPERAND(c); |
a0d0e21e | 387 | while (s < strend) { |
ae5c130c | 388 | if (REGINCLASS(Class, *s)) { |
a0d0e21e LW |
389 | if (tmp && regtry(prog, s)) |
390 | goto got_it; | |
391 | else | |
392 | tmp = doevery; | |
a687059c | 393 | } |
a0d0e21e LW |
394 | else |
395 | tmp = 1; | |
396 | s++; | |
397 | } | |
398 | break; | |
bbce6d69 | 399 | case BOUNDL: |
c277df42 | 400 | reg_flags |= RF_tainted; |
bbce6d69 | 401 | /* FALL THROUGH */ |
a0d0e21e LW |
402 | case BOUND: |
403 | if (minlen) | |
404 | dontbother++,strend--; | |
bbce6d69 | 405 | tmp = (s != startpos) ? UCHARAT(s - 1) : regprev; |
95bac841 | 406 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e | 407 | while (s < strend) { |
95bac841 | 408 | if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { |
a0d0e21e LW |
409 | tmp = !tmp; |
410 | if (regtry(prog, s)) | |
411 | goto got_it; | |
a687059c | 412 | } |
a0d0e21e LW |
413 | s++; |
414 | } | |
415 | if ((minlen || tmp) && regtry(prog,s)) | |
416 | goto got_it; | |
417 | break; | |
bbce6d69 | 418 | case NBOUNDL: |
c277df42 | 419 | reg_flags |= RF_tainted; |
bbce6d69 | 420 | /* FALL THROUGH */ |
a0d0e21e LW |
421 | case NBOUND: |
422 | if (minlen) | |
423 | dontbother++,strend--; | |
bbce6d69 | 424 | tmp = (s != startpos) ? UCHARAT(s - 1) : regprev; |
95bac841 | 425 | tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e | 426 | while (s < strend) { |
95bac841 | 427 | if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) |
a0d0e21e LW |
428 | tmp = !tmp; |
429 | else if (regtry(prog, s)) | |
430 | goto got_it; | |
431 | s++; | |
432 | } | |
433 | if ((minlen || !tmp) && regtry(prog,s)) | |
434 | goto got_it; | |
435 | break; | |
436 | case ALNUM: | |
437 | while (s < strend) { | |
bbce6d69 | 438 | if (isALNUM(*s)) { |
439 | if (tmp && regtry(prog, s)) | |
440 | goto got_it; | |
441 | else | |
442 | tmp = doevery; | |
443 | } | |
444 | else | |
445 | tmp = 1; | |
446 | s++; | |
447 | } | |
448 | break; | |
449 | case ALNUML: | |
c277df42 | 450 | reg_flags |= RF_tainted; |
bbce6d69 | 451 | while (s < strend) { |
452 | if (isALNUM_LC(*s)) { | |
a0d0e21e LW |
453 | if (tmp && regtry(prog, s)) |
454 | goto got_it; | |
a687059c | 455 | else |
a0d0e21e LW |
456 | tmp = doevery; |
457 | } | |
458 | else | |
459 | tmp = 1; | |
460 | s++; | |
461 | } | |
462 | break; | |
463 | case NALNUM: | |
464 | while (s < strend) { | |
bbce6d69 | 465 | if (!isALNUM(*s)) { |
466 | if (tmp && regtry(prog, s)) | |
467 | goto got_it; | |
468 | else | |
469 | tmp = doevery; | |
470 | } | |
471 | else | |
472 | tmp = 1; | |
473 | s++; | |
474 | } | |
475 | break; | |
476 | case NALNUML: | |
c277df42 | 477 | reg_flags |= RF_tainted; |
bbce6d69 | 478 | while (s < strend) { |
479 | if (!isALNUM_LC(*s)) { | |
a0d0e21e LW |
480 | if (tmp && regtry(prog, s)) |
481 | goto got_it; | |
a687059c | 482 | else |
a0d0e21e | 483 | tmp = doevery; |
a687059c | 484 | } |
a0d0e21e LW |
485 | else |
486 | tmp = 1; | |
487 | s++; | |
488 | } | |
489 | break; | |
490 | case SPACE: | |
491 | while (s < strend) { | |
492 | if (isSPACE(*s)) { | |
493 | if (tmp && regtry(prog, s)) | |
494 | goto got_it; | |
495 | else | |
496 | tmp = doevery; | |
2304df62 | 497 | } |
a0d0e21e LW |
498 | else |
499 | tmp = 1; | |
500 | s++; | |
501 | } | |
502 | break; | |
bbce6d69 | 503 | case SPACEL: |
c277df42 | 504 | reg_flags |= RF_tainted; |
bbce6d69 | 505 | while (s < strend) { |
506 | if (isSPACE_LC(*s)) { | |
507 | if (tmp && regtry(prog, s)) | |
508 | goto got_it; | |
509 | else | |
510 | tmp = doevery; | |
511 | } | |
512 | else | |
513 | tmp = 1; | |
514 | s++; | |
515 | } | |
516 | break; | |
a0d0e21e LW |
517 | case NSPACE: |
518 | while (s < strend) { | |
519 | if (!isSPACE(*s)) { | |
520 | if (tmp && regtry(prog, s)) | |
521 | goto got_it; | |
522 | else | |
523 | tmp = doevery; | |
a687059c | 524 | } |
a0d0e21e LW |
525 | else |
526 | tmp = 1; | |
527 | s++; | |
528 | } | |
529 | break; | |
bbce6d69 | 530 | case NSPACEL: |
c277df42 | 531 | reg_flags |= RF_tainted; |
bbce6d69 | 532 | while (s < strend) { |
533 | if (!isSPACE_LC(*s)) { | |
534 | if (tmp && regtry(prog, s)) | |
535 | goto got_it; | |
536 | else | |
537 | tmp = doevery; | |
538 | } | |
539 | else | |
540 | tmp = 1; | |
541 | s++; | |
542 | } | |
543 | break; | |
a0d0e21e LW |
544 | case DIGIT: |
545 | while (s < strend) { | |
546 | if (isDIGIT(*s)) { | |
547 | if (tmp && regtry(prog, s)) | |
548 | goto got_it; | |
549 | else | |
550 | tmp = doevery; | |
2b69d0c2 | 551 | } |
a0d0e21e LW |
552 | else |
553 | tmp = 1; | |
554 | s++; | |
555 | } | |
556 | break; | |
557 | case NDIGIT: | |
558 | while (s < strend) { | |
559 | if (!isDIGIT(*s)) { | |
560 | if (tmp && regtry(prog, s)) | |
561 | goto got_it; | |
562 | else | |
563 | tmp = doevery; | |
a687059c | 564 | } |
a0d0e21e LW |
565 | else |
566 | tmp = 1; | |
567 | s++; | |
568 | } | |
569 | break; | |
a687059c | 570 | } |
a0d0e21e LW |
571 | } |
572 | else { | |
c277df42 IZ |
573 | dontbother = 0; |
574 | if (prog->float_substr != Nullsv) { /* Trim the end. */ | |
575 | char *last; | |
576 | I32 oldpos = scream_pos; | |
577 | ||
578 | if (screamer) { | |
579 | last = screaminstr(screamer, prog->float_substr, s - strbeg, | |
580 | end_shift, &scream_pos, 1); /* last one */ | |
581 | if (!last) { | |
582 | last = scream_olds; /* Only one occurence. */ | |
583 | } | |
584 | } else { | |
585 | STRLEN len; | |
586 | char *little = SvPV(prog->float_substr, len); | |
587 | last = rninstr(s, strend, little, little + len); | |
588 | } | |
589 | if (last == NULL) goto phooey; /* Should not happen! */ | |
590 | dontbother = strend - last - 1; | |
591 | } | |
592 | if (minlen && (dontbother < minlen)) | |
a0d0e21e LW |
593 | dontbother = minlen - 1; |
594 | strend -= dontbother; | |
595 | /* We don't know much -- general case. */ | |
596 | do { | |
597 | if (regtry(prog, s)) | |
598 | goto got_it; | |
599 | } while (s++ < strend); | |
600 | } | |
601 | ||
602 | /* Failure. */ | |
603 | goto phooey; | |
a687059c | 604 | |
a0d0e21e | 605 | got_it: |
420218e7 | 606 | strend += dontbother; /* uncheat */ |
a0d0e21e LW |
607 | prog->subbeg = strbeg; |
608 | prog->subend = strend; | |
c277df42 | 609 | RX_MATCH_TAINTED_SET(prog, reg_flags & RF_tainted); |
5f05dabc | 610 | |
611 | /* make sure $`, $&, $', and $digit will work later */ | |
c277df42 IZ |
612 | if (strbeg != prog->subbase) { /* second+ //g match. */ |
613 | if (!(flags & REXEC_COPY_STR)) { | |
137443ea | 614 | if (prog->subbase) { |
615 | Safefree(prog->subbase); | |
616 | prog->subbase = Nullch; | |
617 | } | |
618 | } | |
619 | else { | |
620 | I32 i = strend - startpos + (stringarg - strbeg); | |
621 | s = savepvn(strbeg, i); | |
622 | Safefree(prog->subbase); | |
623 | prog->subbase = s; | |
624 | prog->subbeg = prog->subbase; | |
625 | prog->subend = prog->subbase + i; | |
626 | s = prog->subbase + (stringarg - strbeg); | |
627 | for (i = 0; i <= prog->nparens; i++) { | |
628 | if (prog->endp[i]) { | |
629 | prog->startp[i] = s + (prog->startp[i] - startpos); | |
630 | prog->endp[i] = s + (prog->endp[i] - startpos); | |
631 | } | |
a0d0e21e LW |
632 | } |
633 | } | |
a0d0e21e LW |
634 | } |
635 | return 1; | |
636 | ||
637 | phooey: | |
a0d0e21e | 638 | return 0; |
a687059c LW |
639 | } |
640 | ||
641 | /* | |
642 | - regtry - try match at specific point | |
643 | */ | |
76e3520e | 644 | STATIC I32 /* 0 failure, 1 success */ |
8ac85365 | 645 | regtry(regexp *prog, char *startpos) |
a687059c | 646 | { |
c277df42 | 647 | dTHR; |
a0d0e21e LW |
648 | register I32 i; |
649 | register char **sp; | |
650 | register char **ep; | |
c277df42 | 651 | CHECKPOINT lastcp; |
a0d0e21e LW |
652 | |
653 | reginput = startpos; | |
654 | regstartp = prog->startp; | |
655 | regendp = prog->endp; | |
656 | reglastparen = &prog->lastparen; | |
657 | prog->lastparen = 0; | |
658 | regsize = 0; | |
c277df42 IZ |
659 | if (reg_start_tmpl <= prog->nparens) { |
660 | reg_start_tmpl = prog->nparens*3/2 + 3; | |
661 | if(reg_start_tmp) | |
662 | Renew(reg_start_tmp, reg_start_tmpl, char*); | |
663 | else | |
664 | New(22,reg_start_tmp, reg_start_tmpl, char*); | |
665 | } | |
a0d0e21e LW |
666 | |
667 | sp = prog->startp; | |
668 | ep = prog->endp; | |
7fae4e64 | 669 | regdata = prog->data; |
a0d0e21e LW |
670 | if (prog->nparens) { |
671 | for (i = prog->nparens; i >= 0; i--) { | |
672 | *sp++ = NULL; | |
673 | *ep++ = NULL; | |
a687059c | 674 | } |
a0d0e21e | 675 | } |
c277df42 | 676 | REGCP_SET; |
a0d0e21e LW |
677 | if (regmatch(prog->program + 1) && reginput >= regtill) { |
678 | prog->startp[0] = startpos; | |
679 | prog->endp[0] = reginput; | |
680 | return 1; | |
681 | } | |
c277df42 IZ |
682 | REGCP_UNWIND; |
683 | return 0; | |
a687059c LW |
684 | } |
685 | ||
686 | /* | |
687 | - regmatch - main matching routine | |
688 | * | |
689 | * Conceptually the strategy is simple: check to see whether the current | |
690 | * node matches, call self recursively to see whether the rest matches, | |
691 | * and then act accordingly. In practice we make some effort to avoid | |
692 | * recursion, in particular by going through "ordinary" nodes (that don't | |
693 | * need to know whether the rest of the match failed) by a loop instead of | |
694 | * by recursion. | |
695 | */ | |
696 | /* [lwall] I've hoisted the register declarations to the outer block in order to | |
697 | * maybe save a little bit of pushing and popping on the stack. It also takes | |
698 | * advantage of machines that use a register save mask on subroutine entry. | |
699 | */ | |
76e3520e | 700 | STATIC I32 /* 0 failure, 1 success */ |
c277df42 | 701 | regmatch(regnode *prog) |
a687059c | 702 | { |
c277df42 IZ |
703 | dTHR; |
704 | register regnode *scan; /* Current node. */ | |
705 | regnode *next; /* Next node. */ | |
706 | regnode *inner; /* Next node in internal branch. */ | |
76e3520e | 707 | register I32 nextchr; /* renamed nextchr - nextchar colides with function of same name */ |
a0d0e21e LW |
708 | register I32 n; /* no or next */ |
709 | register I32 ln; /* len or last */ | |
710 | register char *s; /* operand or save */ | |
711 | register char *locinput = reginput; | |
c277df42 IZ |
712 | register I32 c1, c2, paren; /* case fold search, parenth */ |
713 | int minmod = 0, sw = 0, logical = 0; | |
4633a7c4 | 714 | #ifdef DEBUGGING |
4633a7c4 LW |
715 | regindent++; |
716 | #endif | |
a0d0e21e | 717 | |
76e3520e | 718 | nextchr = UCHARAT(locinput); |
a0d0e21e LW |
719 | scan = prog; |
720 | while (scan != NULL) { | |
c277df42 | 721 | #define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO) |
a687059c | 722 | #ifdef DEBUGGING |
c277df42 IZ |
723 | # define sayYES goto yes |
724 | # define sayNO goto no | |
725 | # define saySAME(x) if (x) goto yes; else goto no | |
726 | # define REPORT_CODE_OFF 24 | |
4633a7c4 | 727 | #else |
c277df42 IZ |
728 | # define sayYES return 1 |
729 | # define sayNO return 0 | |
730 | # define saySAME(x) return x | |
a687059c | 731 | #endif |
c277df42 IZ |
732 | DEBUG_r( { |
733 | SV *prop = sv_newmortal(); | |
734 | int docolor = *colors[0]; | |
735 | int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ | |
736 | int l = (regeol - locinput > taill ? taill : regeol - locinput); | |
737 | int pref_len = (locinput - bostr > (5 + taill) - l | |
738 | ? (5 + taill) - l : locinput - bostr); | |
739 | ||
740 | if (l + pref_len < (5 + taill) && l < regeol - locinput) | |
741 | l = ( regeol - locinput > (5 + taill) - pref_len | |
742 | ? (5 + taill) - pref_len : regeol - locinput); | |
743 | regprop(prop, scan); | |
744 | PerlIO_printf(Perl_debug_log, | |
745 | "%4i <%s%.*s%s%s%s%.*s%s>%*s|%*s%2d%s\n", | |
746 | locinput - bostr, | |
747 | colors[2], pref_len, locinput - pref_len, colors[3], | |
748 | (docolor ? "" : "> <"), | |
749 | colors[0], l, locinput, colors[1], | |
750 | 15 - l - pref_len + 1, | |
751 | "", | |
752 | regindent*2, "", scan - regprogram, | |
753 | SvPVX(prop)); | |
754 | } ); | |
a687059c LW |
755 | |
756 | #ifdef REGALIGN | |
c277df42 | 757 | next = scan + NEXT_OFF(scan); |
a0d0e21e LW |
758 | if (next == scan) |
759 | next = NULL; | |
a687059c | 760 | #else |
a0d0e21e | 761 | next = regnext(scan); |
a687059c LW |
762 | #endif |
763 | ||
a0d0e21e LW |
764 | switch (OP(scan)) { |
765 | case BOL: | |
766 | if (locinput == regbol | |
767 | ? regprev == '\n' | |
c277df42 | 768 | : (multiline && |
76e3520e | 769 | (nextchr || locinput < regeol) && locinput[-1] == '\n') ) |
a0d0e21e LW |
770 | { |
771 | /* regtill = regbol; */ | |
772 | break; | |
773 | } | |
4633a7c4 | 774 | sayNO; |
a0d0e21e LW |
775 | case MBOL: |
776 | if (locinput == regbol | |
777 | ? regprev == '\n' | |
76e3520e | 778 | : ((nextchr || locinput < regeol) && locinput[-1] == '\n') ) |
a0d0e21e LW |
779 | { |
780 | break; | |
781 | } | |
4633a7c4 | 782 | sayNO; |
a0d0e21e LW |
783 | case SBOL: |
784 | if (locinput == regbol && regprev == '\n') | |
785 | break; | |
4633a7c4 | 786 | sayNO; |
774d564b | 787 | case GPOS: |
a0d0e21e LW |
788 | if (locinput == regbol) |
789 | break; | |
4633a7c4 | 790 | sayNO; |
a0d0e21e LW |
791 | case EOL: |
792 | if (multiline) | |
793 | goto meol; | |
794 | else | |
795 | goto seol; | |
796 | case MEOL: | |
797 | meol: | |
76e3520e | 798 | if ((nextchr || locinput < regeol) && nextchr != '\n') |
4633a7c4 | 799 | sayNO; |
a0d0e21e LW |
800 | break; |
801 | case SEOL: | |
802 | seol: | |
76e3520e | 803 | if ((nextchr || locinput < regeol) && nextchr != '\n') |
4633a7c4 | 804 | sayNO; |
a0d0e21e | 805 | if (regeol - locinput > 1) |
4633a7c4 | 806 | sayNO; |
a0d0e21e LW |
807 | break; |
808 | case SANY: | |
76e3520e | 809 | if (!nextchr && locinput >= regeol) |
4633a7c4 | 810 | sayNO; |
76e3520e | 811 | nextchr = UCHARAT(++locinput); |
a0d0e21e LW |
812 | break; |
813 | case ANY: | |
76e3520e | 814 | if (!nextchr && locinput >= regeol || nextchr == '\n') |
4633a7c4 | 815 | sayNO; |
76e3520e | 816 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 817 | break; |
bbce6d69 | 818 | case EXACT: |
161b471a | 819 | s = (char *) OPERAND(scan); |
c277df42 | 820 | ln = UCHARAT(s++); |
a0d0e21e | 821 | /* Inline the first character, for speed. */ |
76e3520e | 822 | if (UCHARAT(s) != nextchr) |
4633a7c4 | 823 | sayNO; |
a0d0e21e | 824 | if (regeol - locinput < ln) |
4633a7c4 | 825 | sayNO; |
36477c24 | 826 | if (ln > 1 && memNE(s, locinput, ln)) |
4633a7c4 | 827 | sayNO; |
a0d0e21e | 828 | locinput += ln; |
76e3520e | 829 | nextchr = UCHARAT(locinput); |
bbce6d69 | 830 | break; |
831 | case EXACTFL: | |
c277df42 | 832 | reg_flags |= RF_tainted; |
bbce6d69 | 833 | /* FALL THROUGH */ |
834 | case EXACTF: | |
161b471a | 835 | s = (char *) OPERAND(scan); |
c277df42 | 836 | ln = UCHARAT(s++); |
bbce6d69 | 837 | /* Inline the first character, for speed. */ |
76e3520e | 838 | if (UCHARAT(s) != nextchr && |
bbce6d69 | 839 | UCHARAT(s) != ((OP(scan) == EXACTF) |
76e3520e | 840 | ? fold : fold_locale)[nextchr]) |
bbce6d69 | 841 | sayNO; |
842 | if (regeol - locinput < ln) | |
843 | sayNO; | |
5f05dabc | 844 | if (ln > 1 && (OP(scan) == EXACTF |
845 | ? ibcmp(s, locinput, ln) | |
846 | : ibcmp_locale(s, locinput, ln))) | |
bbce6d69 | 847 | sayNO; |
848 | locinput += ln; | |
76e3520e | 849 | nextchr = UCHARAT(locinput); |
a0d0e21e LW |
850 | break; |
851 | case ANYOF: | |
161b471a | 852 | s = (char *) OPERAND(scan); |
76e3520e GS |
853 | if (nextchr < 0) |
854 | nextchr = UCHARAT(locinput); | |
873ef191 | 855 | if (!REGINCLASS(s, nextchr)) |
4633a7c4 | 856 | sayNO; |
76e3520e | 857 | if (!nextchr && locinput >= regeol) |
4633a7c4 | 858 | sayNO; |
76e3520e | 859 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 860 | break; |
bbce6d69 | 861 | case ALNUML: |
c277df42 | 862 | reg_flags |= RF_tainted; |
bbce6d69 | 863 | /* FALL THROUGH */ |
a0d0e21e | 864 | case ALNUM: |
76e3520e | 865 | if (!nextchr) |
4633a7c4 | 866 | sayNO; |
bbce6d69 | 867 | if (!(OP(scan) == ALNUM |
76e3520e | 868 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
4633a7c4 | 869 | sayNO; |
76e3520e | 870 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 871 | break; |
bbce6d69 | 872 | case NALNUML: |
c277df42 | 873 | reg_flags |= RF_tainted; |
bbce6d69 | 874 | /* FALL THROUGH */ |
a0d0e21e | 875 | case NALNUM: |
76e3520e | 876 | if (!nextchr && locinput >= regeol) |
4633a7c4 | 877 | sayNO; |
bbce6d69 | 878 | if (OP(scan) == NALNUM |
76e3520e | 879 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
4633a7c4 | 880 | sayNO; |
76e3520e | 881 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 882 | break; |
bbce6d69 | 883 | case BOUNDL: |
884 | case NBOUNDL: | |
c277df42 | 885 | reg_flags |= RF_tainted; |
bbce6d69 | 886 | /* FALL THROUGH */ |
a0d0e21e | 887 | case BOUND: |
bbce6d69 | 888 | case NBOUND: |
889 | /* was last char in word? */ | |
890 | ln = (locinput != regbol) ? UCHARAT(locinput - 1) : regprev; | |
891 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
892 | ln = isALNUM(ln); | |
76e3520e | 893 | n = isALNUM(nextchr); |
bbce6d69 | 894 | } |
895 | else { | |
896 | ln = isALNUM_LC(ln); | |
76e3520e | 897 | n = isALNUM_LC(nextchr); |
bbce6d69 | 898 | } |
95bac841 | 899 | if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL)) |
4633a7c4 | 900 | sayNO; |
a0d0e21e | 901 | break; |
bbce6d69 | 902 | case SPACEL: |
c277df42 | 903 | reg_flags |= RF_tainted; |
bbce6d69 | 904 | /* FALL THROUGH */ |
a0d0e21e | 905 | case SPACE: |
76e3520e | 906 | if (!nextchr && locinput >= regeol) |
4633a7c4 | 907 | sayNO; |
bbce6d69 | 908 | if (!(OP(scan) == SPACE |
76e3520e | 909 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) |
4633a7c4 | 910 | sayNO; |
76e3520e | 911 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 912 | break; |
bbce6d69 | 913 | case NSPACEL: |
c277df42 | 914 | reg_flags |= RF_tainted; |
bbce6d69 | 915 | /* FALL THROUGH */ |
a0d0e21e | 916 | case NSPACE: |
76e3520e | 917 | if (!nextchr) |
4633a7c4 | 918 | sayNO; |
bbce6d69 | 919 | if (OP(scan) == SPACE |
76e3520e | 920 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 921 | sayNO; |
76e3520e | 922 | nextchr = UCHARAT(++locinput); |
a0d0e21e LW |
923 | break; |
924 | case DIGIT: | |
76e3520e | 925 | if (!isDIGIT(nextchr)) |
4633a7c4 | 926 | sayNO; |
76e3520e | 927 | nextchr = UCHARAT(++locinput); |
a0d0e21e LW |
928 | break; |
929 | case NDIGIT: | |
76e3520e | 930 | if (!nextchr && locinput >= regeol) |
4633a7c4 | 931 | sayNO; |
76e3520e | 932 | if (isDIGIT(nextchr)) |
4633a7c4 | 933 | sayNO; |
76e3520e | 934 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 935 | break; |
c8756f30 | 936 | case REFFL: |
c277df42 | 937 | reg_flags |= RF_tainted; |
c8756f30 | 938 | /* FALL THROUGH */ |
c277df42 | 939 | case REF: |
c8756f30 | 940 | case REFF: |
c277df42 | 941 | n = ARG(scan); /* which paren pair */ |
a0d0e21e | 942 | s = regstartp[n]; |
c277df42 | 943 | if (*reglastparen < n || !s) |
af3f8c16 | 944 | sayNO; /* Do not match unless seen CLOSEn. */ |
a0d0e21e LW |
945 | if (s == regendp[n]) |
946 | break; | |
947 | /* Inline the first character, for speed. */ | |
76e3520e | 948 | if (UCHARAT(s) != nextchr && |
c8756f30 AK |
949 | (OP(scan) == REF || |
950 | (UCHARAT(s) != ((OP(scan) == REFF | |
76e3520e | 951 | ? fold : fold_locale)[nextchr])))) |
4633a7c4 | 952 | sayNO; |
a0d0e21e LW |
953 | ln = regendp[n] - s; |
954 | if (locinput + ln > regeol) | |
4633a7c4 | 955 | sayNO; |
c8756f30 AK |
956 | if (ln > 1 && (OP(scan) == REF |
957 | ? memNE(s, locinput, ln) | |
958 | : (OP(scan) == REFF | |
959 | ? ibcmp(s, locinput, ln) | |
960 | : ibcmp_locale(s, locinput, ln)))) | |
4633a7c4 | 961 | sayNO; |
a0d0e21e | 962 | locinput += ln; |
76e3520e | 963 | nextchr = UCHARAT(locinput); |
a0d0e21e LW |
964 | break; |
965 | ||
966 | case NOTHING: | |
c277df42 | 967 | case TAIL: |
a0d0e21e LW |
968 | break; |
969 | case BACK: | |
970 | break; | |
c277df42 IZ |
971 | case EVAL: |
972 | { | |
973 | dSP; | |
974 | OP_4tree *oop = op; | |
975 | COP *ocurcop = curcop; | |
976 | SV **ocurpad = curpad; | |
977 | SV *ret; | |
978 | ||
979 | n = ARG(scan); | |
7fae4e64 | 980 | op = (OP_4tree*)regdata->data[n]; |
c277df42 | 981 | DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", op) ); |
7fae4e64 | 982 | curpad = AvARRAY((AV*)regdata->data[n + 1]); |
c277df42 IZ |
983 | if (!reg_eval_set) { |
984 | /* Preserve whatever is on stack now, otherwise | |
985 | OP_NEXTSTATE will overwrite it. */ | |
986 | SAVEINT(reg_eval_set); /* Protect against unwinding. */ | |
987 | reg_eval_set = 1; | |
988 | DEBUG_r(DEBUG_s( | |
989 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n", stack_sp - stack_base); | |
990 | )); | |
991 | SAVEINT(cxstack[cxstack_ix].blk_oldsp); | |
992 | cxstack[cxstack_ix].blk_oldsp = stack_sp - stack_base; | |
993 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
994 | SAVETMPS; | |
995 | /* Apparently this is not needed, judging by wantarray. */ | |
996 | /* SAVEINT(cxstack[cxstack_ix].blk_gimme); | |
997 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ | |
998 | } | |
999 | ||
76e3520e | 1000 | CALLRUNOPS(); /* Scalar context. */ |
c277df42 IZ |
1001 | SPAGAIN; |
1002 | ret = POPs; | |
1003 | PUTBACK; | |
1004 | ||
1005 | if (logical) { | |
1006 | logical = 0; | |
1007 | sw = SvTRUE(ret); | |
1008 | } | |
1009 | op = oop; | |
1010 | curpad = ocurpad; | |
1011 | curcop = ocurcop; | |
1012 | break; | |
1013 | } | |
a0d0e21e | 1014 | case OPEN: |
c277df42 IZ |
1015 | n = ARG(scan); /* which paren pair */ |
1016 | reg_start_tmp[n] = locinput; | |
a0d0e21e LW |
1017 | if (n > regsize) |
1018 | regsize = n; | |
1019 | break; | |
1020 | case CLOSE: | |
c277df42 IZ |
1021 | n = ARG(scan); /* which paren pair */ |
1022 | regstartp[n] = reg_start_tmp[n]; | |
a0d0e21e LW |
1023 | regendp[n] = locinput; |
1024 | if (n > *reglastparen) | |
1025 | *reglastparen = n; | |
1026 | break; | |
c277df42 IZ |
1027 | case GROUPP: |
1028 | n = ARG(scan); /* which paren pair */ | |
1029 | sw = (*reglastparen >= n && regendp[n] != NULL); | |
1030 | break; | |
1031 | case IFTHEN: | |
1032 | if (sw) | |
1033 | next = NEXTOPER(NEXTOPER(scan)); | |
1034 | else { | |
1035 | next = scan + ARG(scan); | |
1036 | if (OP(next) == IFTHEN) /* Fake one. */ | |
1037 | next = NEXTOPER(NEXTOPER(next)); | |
1038 | } | |
1039 | break; | |
1040 | case LOGICAL: | |
1041 | logical = 1; | |
1042 | break; | |
a0d0e21e LW |
1043 | case CURLYX: { |
1044 | CURCUR cc; | |
1045 | CHECKPOINT cp = savestack_ix; | |
c277df42 IZ |
1046 | |
1047 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ | |
1048 | next += ARG(next); | |
a0d0e21e LW |
1049 | cc.oldcc = regcc; |
1050 | regcc = &cc; | |
1051 | cc.parenfloor = *reglastparen; | |
1052 | cc.cur = -1; | |
1053 | cc.min = ARG1(scan); | |
1054 | cc.max = ARG2(scan); | |
c277df42 | 1055 | cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; |
a0d0e21e LW |
1056 | cc.next = next; |
1057 | cc.minmod = minmod; | |
1058 | cc.lastloc = 0; | |
1059 | reginput = locinput; | |
1060 | n = regmatch(PREVOPER(next)); /* start on the WHILEM */ | |
1061 | regcpblow(cp); | |
1062 | regcc = cc.oldcc; | |
4633a7c4 | 1063 | saySAME(n); |
a0d0e21e LW |
1064 | } |
1065 | /* NOT REACHED */ | |
1066 | case WHILEM: { | |
1067 | /* | |
1068 | * This is really hard to understand, because after we match | |
1069 | * what we're trying to match, we must make sure the rest of | |
1070 | * the RE is going to match for sure, and to do that we have | |
1071 | * to go back UP the parse tree by recursing ever deeper. And | |
1072 | * if it fails, we have to reset our parent's current state | |
1073 | * that we can try again after backing off. | |
1074 | */ | |
1075 | ||
c277df42 | 1076 | CHECKPOINT cp, lastcp; |
a0d0e21e | 1077 | CURCUR* cc = regcc; |
c277df42 IZ |
1078 | char *lastloc = cc->lastloc; /* Detection of 0-len. */ |
1079 | ||
4633a7c4 | 1080 | n = cc->cur + 1; /* how many we know we matched */ |
a0d0e21e LW |
1081 | reginput = locinput; |
1082 | ||
c277df42 IZ |
1083 | DEBUG_r( |
1084 | PerlIO_printf(Perl_debug_log, | |
1085 | "%*s %ld out of %ld..%ld cc=%lx\n", | |
1086 | REPORT_CODE_OFF+regindent*2, "", | |
1087 | (long)n, (long)cc->min, | |
1088 | (long)cc->max, (long)cc) | |
1089 | ); | |
4633a7c4 | 1090 | |
a0d0e21e LW |
1091 | /* If degenerate scan matches "", assume scan done. */ |
1092 | ||
579cf2c3 | 1093 | if (locinput == cc->lastloc && n >= cc->min) { |
a0d0e21e LW |
1094 | regcc = cc->oldcc; |
1095 | ln = regcc->cur; | |
c277df42 IZ |
1096 | DEBUG_r( |
1097 | PerlIO_printf(Perl_debug_log, "%*s empty match detected, try continuation...\n", REPORT_CODE_OFF+regindent*2, "") | |
1098 | ); | |
a0d0e21e | 1099 | if (regmatch(cc->next)) |
4633a7c4 | 1100 | sayYES; |
c277df42 IZ |
1101 | DEBUG_r( |
1102 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") | |
1103 | ); | |
a0d0e21e LW |
1104 | regcc->cur = ln; |
1105 | regcc = cc; | |
4633a7c4 | 1106 | sayNO; |
a0d0e21e LW |
1107 | } |
1108 | ||
1109 | /* First just match a string of min scans. */ | |
1110 | ||
1111 | if (n < cc->min) { | |
1112 | cc->cur = n; | |
1113 | cc->lastloc = locinput; | |
4633a7c4 LW |
1114 | if (regmatch(cc->scan)) |
1115 | sayYES; | |
1116 | cc->cur = n - 1; | |
c277df42 IZ |
1117 | cc->lastloc = lastloc; |
1118 | DEBUG_r( | |
1119 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") | |
1120 | ); | |
4633a7c4 | 1121 | sayNO; |
a0d0e21e LW |
1122 | } |
1123 | ||
1124 | /* Prefer next over scan for minimal matching. */ | |
1125 | ||
1126 | if (cc->minmod) { | |
1127 | regcc = cc->oldcc; | |
1128 | ln = regcc->cur; | |
5f05dabc | 1129 | cp = regcppush(cc->parenfloor); |
c277df42 | 1130 | REGCP_SET; |
5f05dabc | 1131 | if (regmatch(cc->next)) { |
c277df42 | 1132 | regcpblow(cp); |
4633a7c4 | 1133 | sayYES; /* All done. */ |
5f05dabc | 1134 | } |
c277df42 | 1135 | REGCP_UNWIND; |
5f05dabc | 1136 | regcppop(); |
a0d0e21e LW |
1137 | regcc->cur = ln; |
1138 | regcc = cc; | |
1139 | ||
c277df42 IZ |
1140 | if (n >= cc->max) { /* Maximum greed exceeded? */ |
1141 | if (dowarn && n >= REG_INFTY | |
1142 | && !(reg_flags & RF_warned)) { | |
1143 | reg_flags |= RF_warned; | |
1144 | warn("count exceeded %d", REG_INFTY - 1); | |
1145 | } | |
4633a7c4 | 1146 | sayNO; |
c277df42 | 1147 | } |
a687059c | 1148 | |
c277df42 IZ |
1149 | DEBUG_r( |
1150 | PerlIO_printf(Perl_debug_log, "%*s trying longer...\n", REPORT_CODE_OFF+regindent*2, "") | |
1151 | ); | |
a0d0e21e LW |
1152 | /* Try scanning more and see if it helps. */ |
1153 | reginput = locinput; | |
1154 | cc->cur = n; | |
1155 | cc->lastloc = locinput; | |
5f05dabc | 1156 | cp = regcppush(cc->parenfloor); |
c277df42 | 1157 | REGCP_SET; |
5f05dabc | 1158 | if (regmatch(cc->scan)) { |
c277df42 | 1159 | regcpblow(cp); |
4633a7c4 | 1160 | sayYES; |
5f05dabc | 1161 | } |
c277df42 IZ |
1162 | DEBUG_r( |
1163 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") | |
1164 | ); | |
1165 | REGCP_UNWIND; | |
5f05dabc | 1166 | regcppop(); |
4633a7c4 | 1167 | cc->cur = n - 1; |
c277df42 | 1168 | cc->lastloc = lastloc; |
4633a7c4 | 1169 | sayNO; |
a0d0e21e LW |
1170 | } |
1171 | ||
1172 | /* Prefer scan over next for maximal matching. */ | |
1173 | ||
1174 | if (n < cc->max) { /* More greed allowed? */ | |
5f05dabc | 1175 | cp = regcppush(cc->parenfloor); |
a0d0e21e LW |
1176 | cc->cur = n; |
1177 | cc->lastloc = locinput; | |
c277df42 | 1178 | REGCP_SET; |
5f05dabc | 1179 | if (regmatch(cc->scan)) { |
c277df42 | 1180 | regcpblow(cp); |
4633a7c4 | 1181 | sayYES; |
5f05dabc | 1182 | } |
c277df42 | 1183 | REGCP_UNWIND; |
a0d0e21e LW |
1184 | regcppop(); /* Restore some previous $<digit>s? */ |
1185 | reginput = locinput; | |
c277df42 IZ |
1186 | DEBUG_r( |
1187 | PerlIO_printf(Perl_debug_log, "%*s failed, try continuation...\n", REPORT_CODE_OFF+regindent*2, "") | |
1188 | ); | |
1189 | } | |
1190 | if (dowarn && n >= REG_INFTY && !(reg_flags & RF_warned)) { | |
1191 | reg_flags |= RF_warned; | |
1192 | warn("count exceeded %d", REG_INFTY - 1); | |
a0d0e21e LW |
1193 | } |
1194 | ||
1195 | /* Failed deeper matches of scan, so see if this one works. */ | |
1196 | regcc = cc->oldcc; | |
1197 | ln = regcc->cur; | |
1198 | if (regmatch(cc->next)) | |
4633a7c4 | 1199 | sayYES; |
c277df42 IZ |
1200 | DEBUG_r( |
1201 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") | |
1202 | ); | |
a0d0e21e LW |
1203 | regcc->cur = ln; |
1204 | regcc = cc; | |
4633a7c4 | 1205 | cc->cur = n - 1; |
c277df42 | 1206 | cc->lastloc = lastloc; |
4633a7c4 | 1207 | sayNO; |
a0d0e21e LW |
1208 | } |
1209 | /* NOT REACHED */ | |
c277df42 IZ |
1210 | case BRANCHJ: |
1211 | next = scan + ARG(scan); | |
1212 | if (next == scan) | |
1213 | next = NULL; | |
1214 | inner = NEXTOPER(NEXTOPER(scan)); | |
1215 | goto do_branch; | |
1216 | case BRANCH: | |
1217 | inner = NEXTOPER(scan); | |
1218 | do_branch: | |
1219 | { | |
1220 | CHECKPOINT lastcp; | |
1221 | c1 = OP(scan); | |
1222 | if (OP(next) != c1) /* No choice. */ | |
1223 | next = inner; /* Avoid recursion. */ | |
a0d0e21e | 1224 | else { |
748a9306 | 1225 | int lastparen = *reglastparen; |
c277df42 IZ |
1226 | |
1227 | REGCP_SET; | |
a0d0e21e LW |
1228 | do { |
1229 | reginput = locinput; | |
c277df42 | 1230 | if (regmatch(inner)) |
4633a7c4 | 1231 | sayYES; |
c277df42 | 1232 | REGCP_UNWIND; |
748a9306 LW |
1233 | for (n = *reglastparen; n > lastparen; n--) |
1234 | regendp[n] = 0; | |
1235 | *reglastparen = n; | |
c277df42 | 1236 | scan = next; |
a687059c | 1237 | #ifdef REGALIGN |
a0d0e21e | 1238 | /*SUPPRESS 560*/ |
c277df42 IZ |
1239 | if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next))) |
1240 | next += n; | |
a0d0e21e | 1241 | else |
c277df42 | 1242 | next = NULL; |
a687059c | 1243 | #else |
c277df42 | 1244 | next = regnext(next); |
79072805 | 1245 | #endif |
c277df42 IZ |
1246 | inner = NEXTOPER(scan); |
1247 | if (c1 == BRANCHJ) { | |
1248 | inner = NEXTOPER(inner); | |
1249 | } | |
1250 | } while (scan != NULL && OP(scan) == c1); | |
4633a7c4 | 1251 | sayNO; |
a0d0e21e | 1252 | /* NOTREACHED */ |
a687059c | 1253 | } |
a0d0e21e LW |
1254 | } |
1255 | break; | |
1256 | case MINMOD: | |
1257 | minmod = 1; | |
1258 | break; | |
c277df42 IZ |
1259 | case CURLYM: |
1260 | { | |
1261 | I32 l; | |
1262 | CHECKPOINT lastcp; | |
1263 | ||
1264 | /* We suppose that the next guy does not need | |
1265 | backtracking: in particular, it is of constant length, | |
1266 | and has no parenths to influence future backrefs. */ | |
1267 | ln = ARG1(scan); /* min to match */ | |
1268 | n = ARG2(scan); /* max to match */ | |
1269 | #ifdef REGALIGN_STRUCT | |
1270 | paren = scan->flags; | |
1271 | if (paren) { | |
1272 | if (paren > regsize) | |
1273 | regsize = paren; | |
1274 | if (paren > *reglastparen) | |
1275 | *reglastparen = paren; | |
1276 | } | |
1277 | #endif | |
dc45a647 | 1278 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
c277df42 IZ |
1279 | if (paren) |
1280 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ | |
1281 | reginput = locinput; | |
1282 | if (minmod) { | |
1283 | minmod = 0; | |
1284 | if (ln && regrepeat_hard(scan, ln, &l) < ln) | |
1285 | sayNO; | |
5f4b28b2 | 1286 | if (ln && l == 0 && n >= ln |
c277df42 IZ |
1287 | /* In fact, this is tricky. If paren, then the |
1288 | fact that we did/didnot match may influence | |
1289 | future execution. */ | |
1290 | && !(paren && ln == 0)) | |
1291 | ln = n; | |
1292 | locinput = reginput; | |
1293 | if (regkind[(U8)OP(next)] == EXACT) { | |
1294 | c1 = UCHARAT(OPERAND(next) + 1); | |
1295 | if (OP(next) == EXACTF) | |
1296 | c2 = fold[c1]; | |
1297 | else if (OP(next) == EXACTFL) | |
1298 | c2 = fold_locale[c1]; | |
1299 | else | |
1300 | c2 = c1; | |
1301 | } else | |
1302 | c1 = c2 = -1000; | |
1303 | REGCP_SET; | |
5f4b28b2 | 1304 | /* This may be improved if l == 0. */ |
c277df42 IZ |
1305 | while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */ |
1306 | /* If it could work, try it. */ | |
1307 | if (c1 == -1000 || | |
1308 | UCHARAT(reginput) == c1 || | |
1309 | UCHARAT(reginput) == c2) | |
1310 | { | |
1311 | if (paren) { | |
1312 | if (n) { | |
1313 | regstartp[paren] = reginput - l; | |
1314 | regendp[paren] = reginput; | |
1315 | } else | |
1316 | regendp[paren] = NULL; | |
1317 | } | |
1318 | if (regmatch(next)) | |
1319 | sayYES; | |
1320 | REGCP_UNWIND; | |
1321 | } | |
1322 | /* Couldn't or didn't -- move forward. */ | |
1323 | reginput = locinput; | |
1324 | if (regrepeat_hard(scan, 1, &l)) { | |
1325 | ln++; | |
1326 | locinput = reginput; | |
1327 | } | |
1328 | else | |
1329 | sayNO; | |
1330 | } | |
1331 | } else { | |
1332 | n = regrepeat_hard(scan, n, &l); | |
1333 | if (n != 0 && l == 0 | |
1334 | /* In fact, this is tricky. If paren, then the | |
1335 | fact that we did/didnot match may influence | |
1336 | future execution. */ | |
1337 | && !(paren && ln == 0)) | |
1338 | ln = n; | |
1339 | locinput = reginput; | |
1340 | DEBUG_r( | |
1341 | PerlIO_printf(Perl_debug_log, "%*s matched %ld times, len=%ld...\n", REPORT_CODE_OFF+regindent*2, "", n, l) | |
1342 | ); | |
1343 | if (n >= ln) { | |
1344 | if (regkind[(U8)OP(next)] == EXACT) { | |
1345 | c1 = UCHARAT(OPERAND(next) + 1); | |
1346 | if (OP(next) == EXACTF) | |
1347 | c2 = fold[c1]; | |
1348 | else if (OP(next) == EXACTFL) | |
1349 | c2 = fold_locale[c1]; | |
1350 | else | |
1351 | c2 = c1; | |
1352 | } else | |
1353 | c1 = c2 = -1000; | |
1354 | } | |
1355 | REGCP_SET; | |
1356 | while (n >= ln) { | |
1357 | /* If it could work, try it. */ | |
1358 | if (c1 == -1000 || | |
1359 | UCHARAT(reginput) == c1 || | |
1360 | UCHARAT(reginput) == c2) | |
1361 | { | |
1362 | DEBUG_r( | |
1363 | PerlIO_printf(Perl_debug_log, "%*s trying tail with n=%ld...\n", REPORT_CODE_OFF+regindent*2, "", n) | |
1364 | ); | |
1365 | if (paren) { | |
1366 | if (n) { | |
1367 | regstartp[paren] = reginput - l; | |
1368 | regendp[paren] = reginput; | |
1369 | } else | |
1370 | regendp[paren] = NULL; | |
1371 | } | |
1372 | if (regmatch(next)) | |
1373 | sayYES; | |
1374 | REGCP_UNWIND; | |
1375 | } | |
1376 | /* Couldn't or didn't -- back up. */ | |
1377 | n--; | |
1378 | locinput -= l; | |
1379 | reginput = locinput; | |
1380 | } | |
1381 | } | |
1382 | sayNO; | |
1383 | break; | |
1384 | } | |
1385 | case CURLYN: | |
1386 | paren = scan->flags; /* Which paren to set */ | |
1387 | if (paren > regsize) | |
1388 | regsize = paren; | |
1389 | if (paren > *reglastparen) | |
1390 | *reglastparen = paren; | |
1391 | ln = ARG1(scan); /* min to match */ | |
1392 | n = ARG2(scan); /* max to match */ | |
dc45a647 | 1393 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); |
c277df42 | 1394 | goto repeat; |
a0d0e21e | 1395 | case CURLY: |
c277df42 | 1396 | paren = 0; |
a0d0e21e LW |
1397 | ln = ARG1(scan); /* min to match */ |
1398 | n = ARG2(scan); /* max to match */ | |
dc45a647 | 1399 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
a0d0e21e LW |
1400 | goto repeat; |
1401 | case STAR: | |
1402 | ln = 0; | |
c277df42 | 1403 | n = REG_INFTY; |
a0d0e21e | 1404 | scan = NEXTOPER(scan); |
c277df42 | 1405 | paren = 0; |
a0d0e21e LW |
1406 | goto repeat; |
1407 | case PLUS: | |
c277df42 IZ |
1408 | ln = 1; |
1409 | n = REG_INFTY; | |
1410 | scan = NEXTOPER(scan); | |
1411 | paren = 0; | |
1412 | repeat: | |
a0d0e21e LW |
1413 | /* |
1414 | * Lookahead to avoid useless match attempts | |
1415 | * when we know what character comes next. | |
1416 | */ | |
bbce6d69 | 1417 | if (regkind[(U8)OP(next)] == EXACT) { |
1418 | c1 = UCHARAT(OPERAND(next) + 1); | |
1419 | if (OP(next) == EXACTF) | |
1420 | c2 = fold[c1]; | |
1421 | else if (OP(next) == EXACTFL) | |
1422 | c2 = fold_locale[c1]; | |
1423 | else | |
1424 | c2 = c1; | |
1425 | } | |
a0d0e21e | 1426 | else |
bbce6d69 | 1427 | c1 = c2 = -1000; |
a0d0e21e LW |
1428 | reginput = locinput; |
1429 | if (minmod) { | |
c277df42 | 1430 | CHECKPOINT lastcp; |
a0d0e21e LW |
1431 | minmod = 0; |
1432 | if (ln && regrepeat(scan, ln) < ln) | |
4633a7c4 | 1433 | sayNO; |
c277df42 IZ |
1434 | REGCP_SET; |
1435 | while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ | |
a0d0e21e | 1436 | /* If it could work, try it. */ |
bbce6d69 | 1437 | if (c1 == -1000 || |
1438 | UCHARAT(reginput) == c1 || | |
1439 | UCHARAT(reginput) == c2) | |
1440 | { | |
c277df42 IZ |
1441 | if (paren) { |
1442 | if (n) { | |
1443 | regstartp[paren] = reginput - 1; | |
1444 | regendp[paren] = reginput; | |
1445 | } else | |
1446 | regendp[paren] = NULL; | |
1447 | } | |
a0d0e21e | 1448 | if (regmatch(next)) |
4633a7c4 | 1449 | sayYES; |
c277df42 | 1450 | REGCP_UNWIND; |
bbce6d69 | 1451 | } |
c277df42 | 1452 | /* Couldn't or didn't -- move forward. */ |
748a9306 | 1453 | reginput = locinput + ln; |
a0d0e21e LW |
1454 | if (regrepeat(scan, 1)) { |
1455 | ln++; | |
1456 | reginput = locinput + ln; | |
c277df42 | 1457 | } else |
4633a7c4 | 1458 | sayNO; |
a0d0e21e LW |
1459 | } |
1460 | } | |
1461 | else { | |
c277df42 | 1462 | CHECKPOINT lastcp; |
a0d0e21e LW |
1463 | n = regrepeat(scan, n); |
1464 | if (ln < n && regkind[(U8)OP(next)] == EOL && | |
c277df42 | 1465 | (!multiline || OP(next) == SEOL)) |
a0d0e21e | 1466 | ln = n; /* why back off? */ |
c277df42 IZ |
1467 | REGCP_SET; |
1468 | if (paren) { | |
1469 | while (n >= ln) { | |
1470 | /* If it could work, try it. */ | |
1471 | if (c1 == -1000 || | |
1472 | UCHARAT(reginput) == c1 || | |
1473 | UCHARAT(reginput) == c2) | |
1474 | { | |
1475 | if (paren && n) { | |
1476 | if (n) { | |
1477 | regstartp[paren] = reginput - 1; | |
1478 | regendp[paren] = reginput; | |
1479 | } else | |
1480 | regendp[paren] = NULL; | |
1481 | } | |
1482 | if (regmatch(next)) | |
1483 | sayYES; | |
1484 | REGCP_UNWIND; | |
1485 | } | |
1486 | /* Couldn't or didn't -- back up. */ | |
1487 | n--; | |
1488 | reginput = locinput + n; | |
1489 | } | |
1490 | } else { | |
1491 | while (n >= ln) { | |
1492 | /* If it could work, try it. */ | |
1493 | if (c1 == -1000 || | |
1494 | UCHARAT(reginput) == c1 || | |
1495 | UCHARAT(reginput) == c2) | |
1496 | { | |
1497 | if (regmatch(next)) | |
1498 | sayYES; | |
1499 | REGCP_UNWIND; | |
1500 | } | |
1501 | /* Couldn't or didn't -- back up. */ | |
1502 | n--; | |
1503 | reginput = locinput + n; | |
bbce6d69 | 1504 | } |
a0d0e21e LW |
1505 | } |
1506 | } | |
4633a7c4 | 1507 | sayNO; |
c277df42 | 1508 | break; |
a0d0e21e LW |
1509 | case SUCCEED: |
1510 | case END: | |
1511 | reginput = locinput; /* put where regtry can find it */ | |
4633a7c4 | 1512 | sayYES; /* Success! */ |
c277df42 IZ |
1513 | case SUSPEND: |
1514 | n = 1; | |
1515 | goto do_ifmatch; | |
a0d0e21e | 1516 | case UNLESSM: |
c277df42 IZ |
1517 | n = 0; |
1518 | if (locinput < bostr + scan->flags) | |
1519 | goto say_yes; | |
1520 | goto do_ifmatch; | |
1521 | case IFMATCH: | |
1522 | n = 1; | |
1523 | if (locinput < bostr + scan->flags) | |
1524 | goto say_no; | |
1525 | do_ifmatch: | |
1526 | reginput = locinput - scan->flags; | |
1527 | inner = NEXTOPER(NEXTOPER(scan)); | |
1528 | if (regmatch(inner) != n) { | |
1529 | say_no: | |
1530 | if (logical) { | |
1531 | logical = 0; | |
1532 | sw = 0; | |
1533 | goto do_longjump; | |
1534 | } else | |
1535 | sayNO; | |
1536 | } | |
1537 | say_yes: | |
1538 | if (logical) { | |
1539 | logical = 0; | |
1540 | sw = 1; | |
1541 | } | |
fe44a5e8 | 1542 | if (OP(scan) == SUSPEND) { |
c277df42 | 1543 | locinput = reginput; |
565764a8 | 1544 | nextchr = UCHARAT(locinput); |
fe44a5e8 | 1545 | } |
c277df42 IZ |
1546 | /* FALL THROUGH. */ |
1547 | case LONGJMP: | |
1548 | do_longjump: | |
1549 | next = scan + ARG(scan); | |
1550 | if (next == scan) | |
1551 | next = NULL; | |
a0d0e21e LW |
1552 | break; |
1553 | default: | |
c030ccd9 | 1554 | PerlIO_printf(PerlIO_stderr(), "%lx %d\n", |
c277df42 | 1555 | (unsigned long)scan, OP(scan)); |
a0d0e21e | 1556 | FAIL("regexp memory corruption"); |
a687059c | 1557 | } |
a0d0e21e LW |
1558 | scan = next; |
1559 | } | |
a687059c | 1560 | |
a0d0e21e LW |
1561 | /* |
1562 | * We get here only if there's trouble -- normally "case END" is | |
1563 | * the terminating point. | |
1564 | */ | |
1565 | FAIL("corrupted regexp pointers"); | |
1566 | /*NOTREACHED*/ | |
4633a7c4 LW |
1567 | sayNO; |
1568 | ||
1569 | yes: | |
1570 | #ifdef DEBUGGING | |
1571 | regindent--; | |
1572 | #endif | |
1573 | return 1; | |
1574 | ||
1575 | no: | |
1576 | #ifdef DEBUGGING | |
1577 | regindent--; | |
1578 | #endif | |
a0d0e21e | 1579 | return 0; |
a687059c LW |
1580 | } |
1581 | ||
1582 | /* | |
1583 | - regrepeat - repeatedly match something simple, report how many | |
1584 | */ | |
1585 | /* | |
1586 | * [This routine now assumes that it will only match on things of length 1. | |
1587 | * That was true before, but now we assume scan - reginput is the count, | |
1588 | * rather than incrementing count on every character.] | |
1589 | */ | |
76e3520e | 1590 | STATIC I32 |
c277df42 | 1591 | regrepeat(regnode *p, I32 max) |
a687059c | 1592 | { |
a0d0e21e LW |
1593 | register char *scan; |
1594 | register char *opnd; | |
1595 | register I32 c; | |
1596 | register char *loceol = regeol; | |
1597 | ||
1598 | scan = reginput; | |
c277df42 | 1599 | if (max != REG_INFTY && max < loceol - scan) |
a0d0e21e | 1600 | loceol = scan + max; |
161b471a | 1601 | opnd = (char *) OPERAND(p); |
a0d0e21e LW |
1602 | switch (OP(p)) { |
1603 | case ANY: | |
1604 | while (scan < loceol && *scan != '\n') | |
1605 | scan++; | |
1606 | break; | |
1607 | case SANY: | |
1608 | scan = loceol; | |
1609 | break; | |
bbce6d69 | 1610 | case EXACT: /* length of string is 1 */ |
1611 | c = UCHARAT(++opnd); | |
1612 | while (scan < loceol && UCHARAT(scan) == c) | |
1613 | scan++; | |
1614 | break; | |
1615 | case EXACTF: /* length of string is 1 */ | |
1616 | c = UCHARAT(++opnd); | |
1617 | while (scan < loceol && | |
1618 | (UCHARAT(scan) == c || UCHARAT(scan) == fold[c])) | |
1619 | scan++; | |
1620 | break; | |
1621 | case EXACTFL: /* length of string is 1 */ | |
c277df42 | 1622 | reg_flags |= RF_tainted; |
bbce6d69 | 1623 | c = UCHARAT(++opnd); |
1624 | while (scan < loceol && | |
1625 | (UCHARAT(scan) == c || UCHARAT(scan) == fold_locale[c])) | |
a0d0e21e LW |
1626 | scan++; |
1627 | break; | |
1628 | case ANYOF: | |
ae5c130c | 1629 | while (scan < loceol && REGINCLASS(opnd, *scan)) |
a0d0e21e | 1630 | scan++; |
a0d0e21e LW |
1631 | break; |
1632 | case ALNUM: | |
1633 | while (scan < loceol && isALNUM(*scan)) | |
1634 | scan++; | |
1635 | break; | |
bbce6d69 | 1636 | case ALNUML: |
c277df42 | 1637 | reg_flags |= RF_tainted; |
bbce6d69 | 1638 | while (scan < loceol && isALNUM_LC(*scan)) |
1639 | scan++; | |
1640 | break; | |
a0d0e21e LW |
1641 | case NALNUM: |
1642 | while (scan < loceol && !isALNUM(*scan)) | |
1643 | scan++; | |
1644 | break; | |
bbce6d69 | 1645 | case NALNUML: |
c277df42 | 1646 | reg_flags |= RF_tainted; |
bbce6d69 | 1647 | while (scan < loceol && !isALNUM_LC(*scan)) |
1648 | scan++; | |
1649 | break; | |
a0d0e21e LW |
1650 | case SPACE: |
1651 | while (scan < loceol && isSPACE(*scan)) | |
1652 | scan++; | |
1653 | break; | |
bbce6d69 | 1654 | case SPACEL: |
c277df42 | 1655 | reg_flags |= RF_tainted; |
bbce6d69 | 1656 | while (scan < loceol && isSPACE_LC(*scan)) |
1657 | scan++; | |
1658 | break; | |
a0d0e21e LW |
1659 | case NSPACE: |
1660 | while (scan < loceol && !isSPACE(*scan)) | |
1661 | scan++; | |
1662 | break; | |
bbce6d69 | 1663 | case NSPACEL: |
c277df42 | 1664 | reg_flags |= RF_tainted; |
bbce6d69 | 1665 | while (scan < loceol && !isSPACE_LC(*scan)) |
1666 | scan++; | |
1667 | break; | |
a0d0e21e LW |
1668 | case DIGIT: |
1669 | while (scan < loceol && isDIGIT(*scan)) | |
1670 | scan++; | |
1671 | break; | |
1672 | case NDIGIT: | |
1673 | while (scan < loceol && !isDIGIT(*scan)) | |
1674 | scan++; | |
1675 | break; | |
1676 | default: /* Called on something of 0 width. */ | |
1677 | break; /* So match right here or not at all. */ | |
1678 | } | |
a687059c | 1679 | |
a0d0e21e LW |
1680 | c = scan - reginput; |
1681 | reginput = scan; | |
a687059c | 1682 | |
c277df42 IZ |
1683 | DEBUG_r( |
1684 | { | |
1685 | SV *prop = sv_newmortal(); | |
1686 | ||
1687 | regprop(prop, p); | |
1688 | PerlIO_printf(Perl_debug_log, | |
1689 | "%*s %s can match %ld times out of %ld...\n", | |
1690 | REPORT_CODE_OFF+1, "", SvPVX(prop),c,max); | |
1691 | }); | |
1692 | ||
a0d0e21e | 1693 | return(c); |
a687059c LW |
1694 | } |
1695 | ||
1696 | /* | |
c277df42 IZ |
1697 | - regrepeat_hard - repeatedly match something, report total lenth and length |
1698 | * | |
1699 | * The repeater is supposed to have constant length. | |
1700 | */ | |
1701 | ||
76e3520e | 1702 | STATIC I32 |
c277df42 IZ |
1703 | regrepeat_hard(regnode *p, I32 max, I32 *lp) |
1704 | { | |
1705 | register char *scan; | |
1706 | register char *start; | |
1707 | register char *loceol = regeol; | |
1708 | I32 l = -1; | |
1709 | ||
1710 | start = reginput; | |
1711 | while (reginput < loceol && (scan = reginput, regmatch(p))) { | |
1712 | if (l == -1) { | |
1713 | *lp = l = reginput - start; | |
1714 | if (max != REG_INFTY && l*max < loceol - scan) | |
1715 | loceol = scan + l*max; | |
1716 | if (l == 0) { | |
1717 | return max; | |
1718 | } | |
1719 | } | |
1720 | } | |
1721 | if (reginput < loceol) | |
1722 | reginput = scan; | |
1723 | else | |
1724 | scan = reginput; | |
1725 | ||
1726 | return (scan - start)/l; | |
1727 | } | |
1728 | ||
1729 | /* | |
bbce6d69 | 1730 | - regclass - determine if a character falls into a character class |
1731 | */ | |
1732 | ||
76e3520e | 1733 | STATIC bool |
8ac85365 | 1734 | reginclass(register char *p, register I32 c) |
bbce6d69 | 1735 | { |
1736 | char flags = *p; | |
1737 | bool match = FALSE; | |
1738 | ||
1739 | c &= 0xFF; | |
ae5c130c | 1740 | if (ANYOF_TEST(p, c)) |
bbce6d69 | 1741 | match = TRUE; |
1742 | else if (flags & ANYOF_FOLD) { | |
1743 | I32 cf; | |
1744 | if (flags & ANYOF_LOCALE) { | |
c277df42 | 1745 | reg_flags |= RF_tainted; |
bbce6d69 | 1746 | cf = fold_locale[c]; |
1747 | } | |
1748 | else | |
1749 | cf = fold[c]; | |
ae5c130c | 1750 | if (ANYOF_TEST(p, cf)) |
bbce6d69 | 1751 | match = TRUE; |
1752 | } | |
1753 | ||
1754 | if (!match && (flags & ANYOF_ISA)) { | |
c277df42 | 1755 | reg_flags |= RF_tainted; |
bbce6d69 | 1756 | |
1757 | if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) || | |
1758 | ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) || | |
1759 | ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) || | |
1760 | ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c))) | |
1761 | { | |
1762 | match = TRUE; | |
1763 | } | |
1764 | } | |
1765 | ||
ae5c130c | 1766 | return (flags & ANYOF_INVERT) ? !match : match; |
bbce6d69 | 1767 | } |
1768 | ||
161b471a NIS |
1769 | |
1770 |