projects / p5sagit/p5-mst-13.2.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | history | HEAD
clarify what a "line" is
[p5sagit/p5-mst-13.2.git] / regexec.c
CommitLineData
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
b9d5759e 22#ifdef PERL_EXT_RE_BUILD
23/* need to replace pregcomp et al, so enable that */
24# ifndef PERL_IN_XSUB_RE
25# define PERL_IN_XSUB_RE
26# endif
27/* need access to debugger hooks */
28# ifndef DEBUGGING
29# define DEBUGGING
30# endif
31#endif
32
33#ifdef PERL_IN_XSUB_RE
d06ea78c 34/* We *really* need to overwrite these symbols: */
56953603 35# define Perl_regexec_flags my_regexec
36# define Perl_regdump my_regdump
37# define Perl_regprop my_regprop
d06ea78c 38/* *These* symbols are masked to allow static link. */
39# define Perl_pregexec my_pregexec
d88dccdf 40# define Perl_reginitcolors my_reginitcolors
56953603 41#endif
42
f0fcb552 43/*SUPPRESS 112*/
a687059c 44/*
e50aee73 45 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c 46 *
47 * Copyright (c) 1986 by University of Toronto.
48 * Written by Henry Spencer. Not derived from licensed software.
49 *
50 * Permission is granted to anyone to use this software for any
51 * purpose on any computer system, and to redistribute it freely,
52 * subject to the following restrictions:
53 *
54 * 1. The author is not responsible for the consequences of use of
55 * this software, no matter how awful, even if they arise
56 * from defects in it.
57 *
58 * 2. The origin of this software must not be misrepresented, either
59 * by explicit claim or by omission.
60 *
61 * 3. Altered versions must be plainly marked as such, and must not
62 * be misrepresented as being the original software.
63 *
64 **** Alterations to Henry's code are...
65 ****
a0ed51b3 66 **** Copyright (c) 1991-1998, Larry Wall
a687059c 67 ****
9ef589d8 68 **** You may distribute under the terms of either the GNU General Public
69 **** License or the Artistic License, as specified in the README file.
a687059c 70 *
71 * Beware that some of this code is subtly aware of the way operator
72 * precedence is structured in regular expressions. Serious changes in
73 * regular-expression syntax might require a total rethink.
74 */
75#include "EXTERN.h"
76#include "perl.h"
0f5d15d6 77
a687059c 78#include "regcomp.h"
79
c277df42 80#define RF_tainted 1 /* tainted information used? */
81#define RF_warned 2 /* warned about big count? */
ce862d02 82#define RF_evaled 4 /* Did an EVAL with setting? */
a0ed51b3 83#define RF_utf8 8 /* String contains multibyte chars? */
84
85#define UTF (PL_reg_flags & RF_utf8)
ce862d02 86
87#define RS_init 1 /* eval environment created */
88#define RS_set 2 /* replsv value is set */
c277df42 89
a687059c 90#ifndef STATIC
91#define STATIC static
92#endif
93
76e3520e 94#ifndef PERL_OBJECT
a0d0e21e 95typedef I32 CHECKPOINT;
96
c277df42 97/*
98 * Forwards.
99 */
100
101static I32 regmatch _((regnode *prog));
102static I32 regrepeat _((regnode *p, I32 max));
103static I32 regrepeat_hard _((regnode *p, I32 max, I32 *lp));
104static I32 regtry _((regexp *prog, char *startpos));
ae5c130c 105
c277df42 106static bool reginclass _((char *p, I32 c));
a0ed51b3 107static bool reginclassutf8 _((regnode *f, U8* p));
55497cff 108static CHECKPOINT regcppush _((I32 parenfloor));
109static char * regcppop _((void));
942e002e 110static char * regcp_set_to _((I32 ss));
111static void cache_re _((regexp *prog));
9661b544 112static void restore_pos _((void *arg));
76e3520e 113#endif
a0ed51b3 114
ae5c130c 115#define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c))
a0ed51b3 116#define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p))
117
118#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
119#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
120
dfe13c55 121#ifndef PERL_OBJECT
122static U8 * reghop _((U8 *pos, I32 off));
123static U8 * reghopmaybe _((U8 *pos, I32 off));
124#endif
125#define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
126#define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
127#define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off))
128#define HOPMAYBE(pos,off) (UTF ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
129#define HOPc(pos,off) ((char*)HOP(pos,off))
130#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
a0d0e21e 131
76e3520e 132STATIC CHECKPOINT
8ac85365 133regcppush(I32 parenfloor)
a0d0e21e 134{
11343788 135 dTHR;
3280af22 136 int retval = PL_savestack_ix;
137 int i = (PL_regsize - parenfloor) * 4;
a0d0e21e 138 int p;
139
140 SSCHECK(i + 5);
3280af22 141 for (p = PL_regsize; p > parenfloor; p--) {
142 SSPUSHPTR(PL_regendp[p]);
143 SSPUSHPTR(PL_regstartp[p]);
144 SSPUSHPTR(PL_reg_start_tmp[p]);
a0d0e21e 145 SSPUSHINT(p);
146 }
3280af22 147 SSPUSHINT(PL_regsize);
148 SSPUSHINT(*PL_reglastparen);
149 SSPUSHPTR(PL_reginput);
a0d0e21e 150 SSPUSHINT(i + 3);
151 SSPUSHINT(SAVEt_REGCONTEXT);
152 return retval;
153}
154
c277df42 155/* These are needed since we do not localize EVAL nodes: */
c3464db5 156# define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, \
157 " Setting an EVAL scope, savestack=%i\n", \
3280af22 158 PL_savestack_ix)); lastcp = PL_savestack_ix
c3464db5 159
3280af22 160# define REGCP_UNWIND DEBUG_r(lastcp != PL_savestack_ix ? \
c3464db5 161 PerlIO_printf(Perl_debug_log, \
162 " Clearing an EVAL scope, savestack=%i..%i\n", \
3280af22 163 lastcp, PL_savestack_ix) : 0); regcpblow(lastcp)
c277df42 164
76e3520e 165STATIC char *
8ac85365 166regcppop(void)
a0d0e21e 167{
11343788 168 dTHR;
a0d0e21e 169 I32 i = SSPOPINT;
170 U32 paren = 0;
171 char *input;
172 char *tmps;
173 assert(i == SAVEt_REGCONTEXT);
174 i = SSPOPINT;
175 input = (char *) SSPOPPTR;
3280af22 176 *PL_reglastparen = SSPOPINT;
177 PL_regsize = SSPOPINT;
c277df42 178 for (i -= 3; i > 0; i -= 4) {
a0d0e21e 179 paren = (U32)SSPOPINT;
3280af22 180 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
181 PL_regstartp[paren] = (char *) SSPOPPTR;
a0d0e21e 182 tmps = (char*)SSPOPPTR;
3280af22 183 if (paren <= *PL_reglastparen)
184 PL_regendp[paren] = tmps;
c277df42 185 DEBUG_r(
c3464db5 186 PerlIO_printf(Perl_debug_log,
187 " restoring \\%d to %d(%d)..%d%s\n",
3280af22 188 paren, PL_regstartp[paren] - PL_regbol,
189 PL_reg_start_tmp[paren] - PL_regbol,
190 PL_regendp[paren] - PL_regbol,
191 (paren > *PL_reglastparen ? "(no)" : ""));
c277df42 192 );
a0d0e21e 193 }
c277df42 194 DEBUG_r(
3280af22 195 if (*PL_reglastparen + 1 <= PL_regnpar) {
c3464db5 196 PerlIO_printf(Perl_debug_log,
197 " restoring \\%d..\\%d to undef\n",
3280af22 198 *PL_reglastparen + 1, PL_regnpar);
c277df42 199 }
200 );
3280af22 201 for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) {
202 if (paren > PL_regsize)
203 PL_regstartp[paren] = Nullch;
204 PL_regendp[paren] = Nullch;
a0d0e21e 205 }
206 return input;
207}
208
0f5d15d6 209STATIC char *
210regcp_set_to(I32 ss)
211{
46124e9e 212 dTHR;
0f5d15d6 213 I32 tmp = PL_savestack_ix;
214
215 PL_savestack_ix = ss;
216 regcppop();
217 PL_savestack_ix = tmp;
942e002e 218 return Nullch;
0f5d15d6 219}
220
221typedef struct re_cc_state
222{
223 I32 ss;
224 regnode *node;
225 struct re_cc_state *prev;
226 CURCUR *cc;
227 regexp *re;
228} re_cc_state;
229
c277df42 230#define regcpblow(cp) LEAVE_SCOPE(cp)
a0d0e21e 231
a687059c 232/*
e50aee73 233 * pregexec and friends
a687059c 234 */
235
236/*
c277df42 237 - pregexec - match a regexp against a string
a687059c 238 */
c277df42 239I32
c3464db5 240pregexec(register regexp *prog, char *stringarg, register char *strend,
241 char *strbeg, I32 minend, SV *screamer, U32 nosave)
c277df42 242/* strend: pointer to null at end of string */
243/* strbeg: real beginning of string */
244/* minend: end of match must be >=minend after stringarg. */
245/* nosave: For optimizations. */
246{
247 return
248 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
249 nosave ? 0 : REXEC_COPY_STR);
250}
0f5d15d6 251
252STATIC void
253cache_re(regexp *prog)
254{
46124e9e 255 dTHR;
0f5d15d6 256 PL_regprecomp = prog->precomp; /* Needed for FAIL. */
257#ifdef DEBUGGING
258 PL_regprogram = prog->program;
259#endif
260 PL_regnpar = prog->nparens;
261 PL_regdata = prog->data;
262 PL_reg_re = prog;
263}
22e551b9 264
9661b544 265STATIC void
266restore_pos(void *arg)
35ef4773 267{
268 dTHR;
9661b544 269 if (PL_reg_eval_set) {
270 PL_reg_magic->mg_len = PL_reg_oldpos;
271 PL_reg_eval_set = 0;
5c5e4c24 272 PL_curpm = PL_reg_oldcurpm;
9661b544 273 }
274}
275
276
a687059c 277/*
c277df42 278 - regexec_flags - match a regexp against a string
a687059c 279 */
79072805 280I32
c3464db5 281regexec_flags(register regexp *prog, char *stringarg, register char *strend,
22e551b9 282 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
c277df42 283/* strend: pointer to null at end of string */
284/* strbeg: real beginning of string */
285/* minend: end of match must be >=minend after stringarg. */
286/* data: May be used for some additional optimizations. */
287/* nosave: For optimizations. */
a687059c 288{
5c0ca799 289 dTHR;
a0d0e21e 290 register char *s;
c277df42 291 register regnode *c;
a0d0e21e 292 register char *startpos = stringarg;
293 register I32 tmp;
c277df42 294 I32 minlen; /* must match at least this many chars */
a0d0e21e 295 I32 dontbother = 0; /* how many characters not to try at end */
296 CURCUR cc;
c277df42 297 I32 start_shift = 0; /* Offset of the start to find
a0ed51b3 298 constant substr. */ /* CC */
299 I32 end_shift = 0; /* Same for the end. */ /* CC */
c277df42 300 I32 scream_pos = -1; /* Internal iterator of scream. */
301 char *scream_olds;
3280af22 302 SV* oreplsv = GvSV(PL_replgv);
a687059c 303
a0d0e21e 304 cc.cur = 0;
4633a7c4 305 cc.oldcc = 0;
3280af22 306 PL_regcc = &cc;
a0d0e21e 307
0f5d15d6 308 cache_re(prog);
a0d0e21e 309#ifdef DEBUGGING
3280af22 310 PL_regnarrate = PL_debug & 512;
a0d0e21e 311#endif
312
313 /* Be paranoid... */
314 if (prog == NULL || startpos == NULL) {
315 croak("NULL regexp parameter");
316 return 0;
317 }
318
c277df42 319 minlen = prog->minlen;
320 if (strend - startpos < minlen) goto phooey;
321
a0d0e21e 322 if (startpos == strbeg) /* is ^ valid at stringarg? */
3280af22 323 PL_regprev = '\n';
a0d0e21e 324 else {
a0ed51b3 325 PL_regprev = (U32)stringarg[-1];
3280af22 326 if (!PL_multiline && PL_regprev == '\n')
327 PL_regprev = '\0'; /* force ^ to NOT match */
a0d0e21e 328 }
bbce6d69 329
a0d0e21e 330 /* Check validity of program. */
22c35a8c 331 if (UCHARAT(prog->program) != REG_MAGIC) {
4327152a 332 croak("corrupted regexp program");
a0d0e21e 333 }
334
3280af22 335 PL_reg_flags = 0;
336 PL_reg_eval_set = 0;
a0d0e21e 337
a0ed51b3 338 if (prog->reganch & ROPT_UTF8)
339 PL_reg_flags |= RF_utf8;
340
341 /* Mark beginning of line for ^ and lookbehind. */
342 PL_regbol = startpos;
343 PL_bostr = strbeg;
9661b544 344 PL_reg_sv = sv;
a0ed51b3 345
346 /* Mark end of line for $ (and such) */
347 PL_regeol = strend;
348
349 /* see how far we have to get to not match where we matched before */
350 PL_regtill = startpos+minend;
351
0f5d15d6 352 /* We start without call_cc context. */
353 PL_reg_call_cc = 0;
354
a0d0e21e 355 /* If there is a "must appear" string, look for it. */
356 s = startpos;
c277df42 357 if (!(flags & REXEC_CHECKED)
358 && prog->check_substr != Nullsv &&
774d564b 359 !(prog->reganch & ROPT_ANCH_GPOS) &&
c277df42 360 (!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL))
3280af22 361 || (PL_multiline && prog->check_substr == prog->anchored_substr)) )
a0d0e21e 362 {
a0ed51b3 363 char *t;
364 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
c277df42 365 /* Should be nonnegative! */
a0ed51b3 366 end_shift = minlen - start_shift - CHR_SVLEN(prog->check_substr);
22e551b9 367 if (flags & REXEC_SCREAM) {
3280af22 368 if (PL_screamfirst[BmRARE(prog->check_substr)] >= 0)
22e551b9 369 s = screaminstr(sv, prog->check_substr,
c277df42 370 start_shift + (stringarg - strbeg),
371 end_shift, &scream_pos, 0);
a0d0e21e 372 else
373 s = Nullch;
c277df42 374 scream_olds = s;
0a12ae7d 375 }
a0d0e21e 376 else
c277df42 377 s = fbm_instr((unsigned char*)s + start_shift,
378 (unsigned char*)strend - end_shift,
411d5715 379 prog->check_substr, 0);
a0d0e21e 380 if (!s) {
c277df42 381 ++BmUSEFUL(prog->check_substr); /* hooray */
a0d0e21e 382 goto phooey; /* not present */
a0ed51b3 383 }
384 else if (s - stringarg > prog->check_offset_max &&
385 (UTF
dfe13c55 386 ? ((t = reghopmaybe_c(s, -(prog->check_offset_max))) && t >= stringarg)
a0ed51b3 387 : (t = s - prog->check_offset_max) != 0
388 )
389 )
390 {
c277df42 391 ++BmUSEFUL(prog->check_substr); /* hooray/2 */
a0ed51b3 392 s = t;
393 }
394 else if (!(prog->reganch & ROPT_NAUGHTY)
c277df42 395 && --BmUSEFUL(prog->check_substr) < 0
396 && prog->check_substr == prog->float_substr) { /* boo */
397 SvREFCNT_dec(prog->check_substr);
398 prog->check_substr = Nullsv; /* disable */
399 prog->float_substr = Nullsv; /* clear */
a0d0e21e 400 s = startpos;
a0ed51b3 401 }
402 else
403 s = startpos;
a0d0e21e 404 }
a687059c 405
35ef4773 406 DEBUG_r(if (!PL_colorset) reginitcolors());
407 DEBUG_r(PerlIO_printf(Perl_debug_log,
8d300b32 408 "%sMatching%s `%s%.60s%s%s' against `%s%.*s%s%s'\n",
409 PL_colors[4],PL_colors[5],PL_colors[0],
410 prog->precomp,
411 PL_colors[1],
c277df42 412 (strlen(prog->precomp) > 60 ? "..." : ""),
8d300b32 413 PL_colors[0],
c277df42 414 (strend - startpos > 60 ? 60 : strend - startpos),
8d300b32 415 startpos, PL_colors[1],
c277df42 416 (strend - startpos > 60 ? "..." : ""))
417 );
418
22e551b9 419 if (prog->reganch & ROPT_GPOS_SEEN) {
420 MAGIC *mg;
22e551b9 421
ad94a511 422 if (!(flags & REXEC_IGNOREPOS) && sv && SvTYPE(sv) >= SVt_PVMG
423 && SvMAGIC(sv) && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0)
424 PL_reg_ganch = strbeg + mg->mg_len;
425 else
426 PL_reg_ganch = startpos;
22e551b9 427 }
428
a0d0e21e 429 /* Simplest case: anchored match need be tried only once. */
774d564b 430 /* [unless only anchor is BOL and multiline is set] */
22e551b9 431 if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
a0d0e21e 432 if (regtry(prog, startpos))
433 goto got_it;
22e551b9 434 else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
435 || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
774d564b 436 {
a0d0e21e 437 if (minlen)
438 dontbother = minlen - 1;
dfe13c55 439 strend = HOPc(strend, -dontbother);
a0d0e21e 440 /* for multiline we only have to try after newlines */
441 if (s > startpos)
442 s--;
443 while (s < strend) {
6f06b55f 444 if (*s++ == '\n') { /* don't need PL_utf8skip here */
a0d0e21e 445 if (s < strend && regtry(prog, s))
446 goto got_it;
447 }
35c8bce7 448 }
35c8bce7 449 }
a0d0e21e 450 goto phooey;
22e551b9 451 } else if (prog->reganch & ROPT_ANCH_GPOS) {
452 if (regtry(prog, PL_reg_ganch))
453 goto got_it;
454 goto phooey;
a0d0e21e 455 }
35c8bce7 456
a0d0e21e 457 /* Messy cases: unanchored match. */
c277df42 458 if (prog->anchored_substr && prog->reganch & ROPT_SKIP) {
459 /* we have /x+whatever/ */
460 /* it must be a one character string */
461 char ch = SvPVX(prog->anchored_substr)[0];
a0ed51b3 462 if (UTF) {
463 while (s < strend) {
464 if (*s == ch) {
465 if (regtry(prog, s)) goto got_it;
466 s += UTF8SKIP(s);
467 while (s < strend && *s == ch)
468 s += UTF8SKIP(s);
469 }
470 s += UTF8SKIP(s);
471 }
472 }
473 else {
474 while (s < strend) {
475 if (*s == ch) {
476 if (regtry(prog, s)) goto got_it;
c277df42 477 s++;
a0ed51b3 478 while (s < strend && *s == ch)
479 s++;
480 }
481 s++;
a0d0e21e 482 }
a687059c 483 }
c277df42 484 }
485 /*SUPPRESS 560*/
486 else if (prog->anchored_substr != Nullsv
487 || (prog->float_substr != Nullsv
488 && prog->float_max_offset < strend - s)) {
489 SV *must = prog->anchored_substr
490 ? prog->anchored_substr : prog->float_substr;
491 I32 back_max =
492 prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset;
493 I32 back_min =
494 prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset;
495 I32 delta = back_max - back_min;
dfe13c55 496 char *last = HOPc(strend, 0-(CHR_SVLEN(must) + back_min)); /* Cannot start after this */
a0ed51b3 497 char *last1; /* Last position checked before */
498
499 if (s > PL_bostr)
dfe13c55 500 last1 = HOPc(s, -1);
a0ed51b3 501 else
502 last1 = s - 1; /* bogus */
c277df42 503
504 /* XXXX check_substr already used to find `s', can optimize if
505 check_substr==must. */
506 scream_pos = -1;
507 dontbother = end_shift;
dfe13c55 508 strend = HOPc(strend, -dontbother);
c277df42 509 while ( (s <= last) &&
22e551b9 510 ((flags & REXEC_SCREAM)
511 ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg,
c277df42 512 end_shift, &scream_pos, 0))
a0ed51b3 513 : (s = fbm_instr((unsigned char*)HOP(s, back_min),
411d5715 514 (unsigned char*)strend, must, 0))) ) {
dfe13c55 515 if (HOPc(s, -back_max) > last1) {
516 last1 = HOPc(s, -back_min);
517 s = HOPc(s, -back_max);
a0ed51b3 518 }
519 else {
dfe13c55 520 char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
c277df42 521
dfe13c55 522 last1 = HOPc(s, -back_min);
c277df42 523 s = t;
a0d0e21e 524 }
a0ed51b3 525 if (UTF) {
526 while (s <= last1) {
527 if (regtry(prog, s))
528 goto got_it;
529 s += UTF8SKIP(s);
530 }
531 }
532 else {
533 while (s <= last1) {
534 if (regtry(prog, s))
535 goto got_it;
536 s++;
537 }
a0d0e21e 538 }
539 }
540 goto phooey;
a0ed51b3 541 }
542 else if (c = prog->regstclass) {
a0d0e21e 543 I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
a0ed51b3 544 char *cc;
a687059c 545
a0d0e21e 546 if (minlen)
547 dontbother = minlen - 1;
dfe13c55 548 strend = HOPc(strend, -dontbother); /* don't bother with what can't match */
a0d0e21e 549 tmp = 1;
550 /* We know what class it must start with. */
551 switch (OP(c)) {
a0ed51b3 552 case ANYOFUTF8:
553 cc = (char *) OPERAND(c);
554 while (s < strend) {
555 if (REGINCLASSUTF8(c, (U8*)s)) {
556 if (tmp && regtry(prog, s))
557 goto got_it;
558 else
559 tmp = doevery;
560 }
561 else
562 tmp = 1;
563 s += UTF8SKIP(s);
564 }
565 break;
a0d0e21e 566 case ANYOF:
a0ed51b3 567 cc = (char *) OPERAND(c);
a0d0e21e 568 while (s < strend) {
a0ed51b3 569 if (REGINCLASS(cc, *s)) {
a0d0e21e 570 if (tmp && regtry(prog, s))
571 goto got_it;
572 else
573 tmp = doevery;
a687059c 574 }
a0d0e21e 575 else
576 tmp = 1;
577 s++;
578 }
579 break;
bbce6d69 580 case BOUNDL:
3280af22 581 PL_reg_flags |= RF_tainted;
bbce6d69 582 /* FALL THROUGH */
a0d0e21e 583 case BOUND:
a0ed51b3 584 if (minlen) {
585 dontbother++;
586 strend -= 1;
587 }
3280af22 588 tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev;
95bac841 589 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
a0d0e21e 590 while (s < strend) {
95bac841 591 if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
a0d0e21e 592 tmp = !tmp;
593 if (regtry(prog, s))
594 goto got_it;
a687059c 595 }
a0d0e21e 596 s++;
597 }
598 if ((minlen || tmp) && regtry(prog,s))
599 goto got_it;
600 break;
a0ed51b3 601 case BOUNDLUTF8:
602 PL_reg_flags |= RF_tainted;
603 /* FALL THROUGH */
604 case BOUNDUTF8:
605 if (minlen) {
606 dontbother++;
dfe13c55 607 strend = reghop_c(strend, -1);
a0ed51b3 608 }
dfe13c55 609 tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev;
a0ed51b3 610 tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
611 while (s < strend) {
dfe13c55 612 if (tmp == !(OP(c) == BOUND ?
613 swash_fetch(PL_utf8_alnum, (U8*)s) :
614 isALNUM_LC_utf8((U8*)s)))
615 {
a0ed51b3 616 tmp = !tmp;
617 if (regtry(prog, s))
618 goto got_it;
619 }
620 s += UTF8SKIP(s);
621 }
622 if ((minlen || tmp) && regtry(prog,s))
623 goto got_it;
624 break;
bbce6d69 625 case NBOUNDL:
3280af22 626 PL_reg_flags |= RF_tainted;
bbce6d69 627 /* FALL THROUGH */
a0d0e21e 628 case NBOUND:
a0ed51b3 629 if (minlen) {
630 dontbother++;
631 strend -= 1;
632 }
3280af22 633 tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev;
95bac841 634 tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
a0d0e21e 635 while (s < strend) {
95bac841 636 if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
a0d0e21e 637 tmp = !tmp;
638 else if (regtry(prog, s))
639 goto got_it;
640 s++;
641 }
642 if ((minlen || !tmp) && regtry(prog,s))
643 goto got_it;
644 break;
a0ed51b3 645 case NBOUNDLUTF8:
646 PL_reg_flags |= RF_tainted;
647 /* FALL THROUGH */
648 case NBOUNDUTF8:
649 if (minlen) {
650 dontbother++;
dfe13c55 651 strend = reghop_c(strend, -1);
a0ed51b3 652 }
dfe13c55 653 tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev;
a0ed51b3 654 tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
655 while (s < strend) {
dfe13c55 656 if (tmp == !(OP(c) == NBOUND ?
657 swash_fetch(PL_utf8_alnum, (U8*)s) :
658 isALNUM_LC_utf8((U8*)s)))
a0ed51b3 659 tmp = !tmp;
660 else if (regtry(prog, s))
661 goto got_it;
662 s += UTF8SKIP(s);
663 }
664 if ((minlen || !tmp) && regtry(prog,s))
665 goto got_it;
666 break;
a0d0e21e 667 case ALNUM:
668 while (s < strend) {
bbce6d69 669 if (isALNUM(*s)) {
670 if (tmp && regtry(prog, s))
671 goto got_it;
672 else
673 tmp = doevery;
674 }
675 else
676 tmp = 1;
677 s++;
678 }
679 break;
a0ed51b3 680 case ALNUMUTF8:
681 while (s < strend) {
dfe13c55 682 if (swash_fetch(PL_utf8_alnum, (U8*)s)) {
a0ed51b3 683 if (tmp && regtry(prog, s))
684 goto got_it;
685 else
686 tmp = doevery;
687 }
688 else
689 tmp = 1;
690 s += UTF8SKIP(s);
691 }
692 break;
bbce6d69 693 case ALNUML:
3280af22 694 PL_reg_flags |= RF_tainted;
bbce6d69 695 while (s < strend) {
696 if (isALNUM_LC(*s)) {
a0d0e21e 697 if (tmp && regtry(prog, s))
698 goto got_it;
a687059c 699 else
a0d0e21e 700 tmp = doevery;
701 }
702 else
703 tmp = 1;
704 s++;
705 }
706 break;
a0ed51b3 707 case ALNUMLUTF8:
708 PL_reg_flags |= RF_tainted;
709 while (s < strend) {
dfe13c55 710 if (isALNUM_LC_utf8((U8*)s)) {
a0ed51b3 711 if (tmp && regtry(prog, s))
712 goto got_it;
713 else
714 tmp = doevery;
715 }
716 else
717 tmp = 1;
718 s += UTF8SKIP(s);
719 }
720 break;
a0d0e21e 721 case NALNUM:
722 while (s < strend) {
bbce6d69 723 if (!isALNUM(*s)) {
724 if (tmp && regtry(prog, s))
725 goto got_it;
726 else
727 tmp = doevery;
728 }
729 else
730 tmp = 1;
731 s++;
732 }
733 break;
a0ed51b3 734 case NALNUMUTF8:
735 while (s < strend) {
dfe13c55 736 if (!swash_fetch(PL_utf8_alnum, (U8*)s)) {
a0ed51b3 737 if (tmp && regtry(prog, s))
738 goto got_it;
739 else
740 tmp = doevery;
741 }
742 else
743 tmp = 1;
744 s += UTF8SKIP(s);
745 }
746 break;
bbce6d69 747 case NALNUML:
3280af22 748 PL_reg_flags |= RF_tainted;
bbce6d69 749 while (s < strend) {
750 if (!isALNUM_LC(*s)) {
a0d0e21e 751 if (tmp && regtry(prog, s))
752 goto got_it;
a687059c 753 else
a0d0e21e 754 tmp = doevery;
a687059c 755 }
a0d0e21e 756 else
757 tmp = 1;
758 s++;
759 }
760 break;
a0ed51b3 761 case NALNUMLUTF8:
762 PL_reg_flags |= RF_tainted;
763 while (s < strend) {
dfe13c55 764 if (!isALNUM_LC_utf8((U8*)s)) {
a0ed51b3 765 if (tmp && regtry(prog, s))
766 goto got_it;
767 else
768 tmp = doevery;
769 }
770 else
771 tmp = 1;
772 s += UTF8SKIP(s);
773 }
774 break;
a0d0e21e 775 case SPACE:
776 while (s < strend) {
777 if (isSPACE(*s)) {
778 if (tmp && regtry(prog, s))
779 goto got_it;
780 else
781 tmp = doevery;
2304df62 782 }
a0d0e21e 783 else
784 tmp = 1;
785 s++;
786 }
787 break;
a0ed51b3 788 case SPACEUTF8:
789 while (s < strend) {
dfe13c55 790 if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) {
a0ed51b3 791 if (tmp && regtry(prog, s))
792 goto got_it;
793 else
794 tmp = doevery;
795 }
796 else
797 tmp = 1;
798 s += UTF8SKIP(s);
799 }
800 break;
bbce6d69 801 case SPACEL:
3280af22 802 PL_reg_flags |= RF_tainted;
bbce6d69 803 while (s < strend) {
804 if (isSPACE_LC(*s)) {
805 if (tmp && regtry(prog, s))
806 goto got_it;
807 else
808 tmp = doevery;
809 }
810 else
811 tmp = 1;
812 s++;
813 }
814 break;
a0ed51b3 815 case SPACELUTF8:
816 PL_reg_flags |= RF_tainted;
817 while (s < strend) {
dfe13c55 818 if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
a0ed51b3 819 if (tmp && regtry(prog, s))
820 goto got_it;
821 else
822 tmp = doevery;
823 }
824 else
825 tmp = 1;
826 s += UTF8SKIP(s);
827 }
828 break;
a0d0e21e 829 case NSPACE:
830 while (s < strend) {
831 if (!isSPACE(*s)) {
832 if (tmp && regtry(prog, s))
833 goto got_it;
834 else
835 tmp = doevery;
a687059c 836 }
a0d0e21e 837 else
838 tmp = 1;
839 s++;
840 }
841 break;
a0ed51b3 842 case NSPACEUTF8:
843 while (s < strend) {
dfe13c55 844 if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) {
a0ed51b3 845 if (tmp && regtry(prog, s))
846 goto got_it;
847 else
848 tmp = doevery;
849 }
850 else
851 tmp = 1;
852 s += UTF8SKIP(s);
853 }
854 break;
bbce6d69 855 case NSPACEL:
3280af22 856 PL_reg_flags |= RF_tainted;
bbce6d69 857 while (s < strend) {
858 if (!isSPACE_LC(*s)) {
859 if (tmp && regtry(prog, s))
860 goto got_it;
861 else
862 tmp = doevery;
863 }
864 else
865 tmp = 1;
866 s++;
867 }
868 break;
a0ed51b3 869 case NSPACELUTF8:
870 PL_reg_flags |= RF_tainted;
871 while (s < strend) {
dfe13c55 872 if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
a0ed51b3 873 if (tmp && regtry(prog, s))
874 goto got_it;
875 else
876 tmp = doevery;
877 }
878 else
879 tmp = 1;
880 s += UTF8SKIP(s);
881 }
882 break;
a0d0e21e 883 case DIGIT:
884 while (s < strend) {
885 if (isDIGIT(*s)) {
886 if (tmp && regtry(prog, s))
887 goto got_it;
888 else
889 tmp = doevery;
2b69d0c2 890 }
a0d0e21e 891 else
892 tmp = 1;
893 s++;
894 }
895 break;
a0ed51b3 896 case DIGITUTF8:
897 while (s < strend) {
dfe13c55 898 if (swash_fetch(PL_utf8_digit,(U8*)s)) {
a0ed51b3 899 if (tmp && regtry(prog, s))
900 goto got_it;
901 else
902 tmp = doevery;
903 }
904 else
905 tmp = 1;
906 s += UTF8SKIP(s);
907 }
908 break;
a0d0e21e 909 case NDIGIT:
910 while (s < strend) {
911 if (!isDIGIT(*s)) {
912 if (tmp && regtry(prog, s))
913 goto got_it;
914 else
915 tmp = doevery;
a687059c 916 }
a0d0e21e 917 else
918 tmp = 1;
919 s++;
920 }
921 break;
a0ed51b3 922 case NDIGITUTF8:
923 while (s < strend) {
dfe13c55 924 if (!swash_fetch(PL_utf8_digit,(U8*)s)) {
a0ed51b3 925 if (tmp && regtry(prog, s))
926 goto got_it;
927 else
928 tmp = doevery;
929 }
930 else
931 tmp = 1;
932 s += UTF8SKIP(s);
933 }
934 break;
a687059c 935 }
a0d0e21e 936 }
937 else {
c277df42 938 dontbother = 0;
939 if (prog->float_substr != Nullsv) { /* Trim the end. */
940 char *last;
941 I32 oldpos = scream_pos;
942
22e551b9 943 if (flags & REXEC_SCREAM) {
944 last = screaminstr(sv, prog->float_substr, s - strbeg,
c277df42 945 end_shift, &scream_pos, 1); /* last one */
946 if (!last) {
947 last = scream_olds; /* Only one occurence. */
948 }
a0ed51b3 949 }
950 else {
c277df42 951 STRLEN len;
952 char *little = SvPV(prog->float_substr, len);
19b4f81a 953 if (len)
954 last = rninstr(s, strend, little, little + len);
955 else
956 last = strend; /* matching `$' */
c277df42 957 }
958 if (last == NULL) goto phooey; /* Should not happen! */
19b4f81a 959 dontbother = strend - last + prog->float_min_offset;
c277df42 960 }
961 if (minlen && (dontbother < minlen))
a0d0e21e 962 dontbother = minlen - 1;
a0ed51b3 963 strend -= dontbother; /* this one's always in bytes! */
a0d0e21e 964 /* We don't know much -- general case. */
a0ed51b3 965 if (UTF) {
966 for (;;) {
84df6dba 967 if (regtry(prog, s))
a0ed51b3 968 goto got_it;
a0ed51b3 969 if (s >= strend)
970 break;
971 s += UTF8SKIP(s);
972 };
973 }
974 else {
975 do {
976 if (regtry(prog, s))
977 goto got_it;
978 } while (s++ < strend);
979 }
a0d0e21e 980 }
981
982 /* Failure. */
983 goto phooey;
a687059c 984
a0d0e21e 985got_it:
986 prog->subbeg = strbeg;
19b4f81a 987 prog->subend = PL_regeol; /* strend may have been modified */
3280af22 988 RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
5f05dabc 989
990 /* make sure $`, $&, $', and $digit will work later */
c277df42 991 if (strbeg != prog->subbase) { /* second+ //g match. */
992 if (!(flags & REXEC_COPY_STR)) {
137443ea 993 if (prog->subbase) {
994 Safefree(prog->subbase);
995 prog->subbase = Nullch;
996 }
997 }
998 else {
19b4f81a 999 I32 i = PL_regeol - startpos + (stringarg - strbeg);
137443ea 1000 s = savepvn(strbeg, i);
1001 Safefree(prog->subbase);
1002 prog->subbase = s;
1003 prog->subbeg = prog->subbase;
1004 prog->subend = prog->subbase + i;
1005 s = prog->subbase + (stringarg - strbeg);
1006 for (i = 0; i <= prog->nparens; i++) {
1007 if (prog->endp[i]) {
1008 prog->startp[i] = s + (prog->startp[i] - startpos);
1009 prog->endp[i] = s + (prog->endp[i] - startpos);
1010 }
a0d0e21e 1011 }
1012 }
a0d0e21e 1013 }
5c5e4c24 1014 if (PL_reg_eval_set) {
1015 /* Preserve the current value of $^R */
1016 if (oreplsv != GvSV(PL_replgv))
1017 sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
1018 restored, the value remains
1019 the same. */
9661b544 1020 restore_pos(0);
5c5e4c24 1021 }
1022
a0d0e21e 1023 return 1;
1024
1025phooey:
9661b544 1026 if (PL_reg_eval_set)
1027 restore_pos(0);
a0d0e21e 1028 return 0;
a687059c 1029}
1030
1031/*
1032 - regtry - try match at specific point
1033 */
76e3520e 1034STATIC I32 /* 0 failure, 1 success */
8ac85365 1035regtry(regexp *prog, char *startpos)
a687059c 1036{
c277df42 1037 dTHR;
a0d0e21e 1038 register I32 i;
1039 register char **sp;
1040 register char **ep;
c277df42 1041 CHECKPOINT lastcp;
a0d0e21e 1042
3280af22 1043 if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
9661b544 1044 MAGIC *mg;
1045
3280af22 1046 PL_reg_eval_set = RS_init;
ce862d02 1047 DEBUG_r(DEBUG_s(
c3464db5 1048 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n",
3280af22 1049 PL_stack_sp - PL_stack_base);
ce862d02 1050 ));
1051 SAVEINT(cxstack[cxstack_ix].blk_oldsp);
3280af22 1052 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
ce862d02 1053 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
1054 SAVETMPS;
1055 /* Apparently this is not needed, judging by wantarray. */
1056 /* SAVEINT(cxstack[cxstack_ix].blk_gimme);
1057 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
9661b544 1058
1059 if (PL_reg_sv) {
1060 /* Make $_ available to executed code. */
127ad2b7 1061 if (PL_reg_sv != DEFSV) {
1062 /* SAVE_DEFSV does *not* suffice here for USE_THREADS */
1063 SAVESPTR(DEFSV);
1064 DEFSV = PL_reg_sv;
9661b544 1065 }
1066
1067 if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
1068 && (mg = mg_find(PL_reg_sv, 'g')))) {
1069 /* prepare for quick setting of pos */
1070 sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0);
1071 mg = mg_find(PL_reg_sv, 'g');
1072 mg->mg_len = -1;
1073 }
1074 PL_reg_magic = mg;
1075 PL_reg_oldpos = mg->mg_len;
1076 SAVEDESTRUCTOR(restore_pos, 0);
1077 }
5c5e4c24 1078 if (!PL_reg_curpm)
1079 New(22,PL_reg_curpm, 1, PMOP);
1080 PL_reg_curpm->op_pmregexp = prog;
1081 PL_reg_oldcurpm = PL_curpm;
1082 PL_curpm = PL_reg_curpm;
1083 prog->subbeg = PL_bostr;
1084 prog->subend = PL_regeol; /* strend may have been modified */
ce862d02 1085 }
5c5e4c24 1086 prog->startp[0] = startpos;
3280af22 1087 PL_reginput = startpos;
1088 PL_regstartp = prog->startp;
1089 PL_regendp = prog->endp;
1090 PL_reglastparen = &prog->lastparen;
a0d0e21e 1091 prog->lastparen = 0;
3280af22 1092 PL_regsize = 0;
364723c2 1093 DEBUG_r(PL_reg_starttry = startpos);
3280af22 1094 if (PL_reg_start_tmpl <= prog->nparens) {
1095 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
1096 if(PL_reg_start_tmp)
1097 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
c277df42 1098 else
3280af22 1099 New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*);
c277df42 1100 }
a0d0e21e 1101
5c5e4c24 1102 /* XXXX What this code is doing here?!!! There should be no need
1103 to do this again and again, PL_reglastparen should take care of
1104 this! */
a0d0e21e 1105 sp = prog->startp;
1106 ep = prog->endp;
1107 if (prog->nparens) {
5c5e4c24 1108 for (i = prog->nparens; i >= 1; i--) {
1109 *++sp = NULL;
1110 *++ep = NULL;
a687059c 1111 }
a0d0e21e 1112 }
c277df42 1113 REGCP_SET;
7e5428c5 1114 if (regmatch(prog->program + 1)) {
3280af22 1115 prog->endp[0] = PL_reginput;
a0d0e21e 1116 return 1;
1117 }
c277df42 1118 REGCP_UNWIND;
1119 return 0;
a687059c 1120}
1121
1122/*
1123 - regmatch - main matching routine
1124 *
1125 * Conceptually the strategy is simple: check to see whether the current
1126 * node matches, call self recursively to see whether the rest matches,
1127 * and then act accordingly. In practice we make some effort to avoid
1128 * recursion, in particular by going through "ordinary" nodes (that don't
1129 * need to know whether the rest of the match failed) by a loop instead of
1130 * by recursion.
1131 */
1132/* [lwall] I've hoisted the register declarations to the outer block in order to
1133 * maybe save a little bit of pushing and popping on the stack. It also takes
1134 * advantage of machines that use a register save mask on subroutine entry.
1135 */
76e3520e 1136STATIC I32 /* 0 failure, 1 success */
c277df42 1137regmatch(regnode *prog)
a687059c 1138{
c277df42 1139 dTHR;
1140 register regnode *scan; /* Current node. */
1141 regnode *next; /* Next node. */
1142 regnode *inner; /* Next node in internal branch. */
c3464db5 1143 register I32 nextchr; /* renamed nextchr - nextchar colides with
1144 function of same name */
a0d0e21e 1145 register I32 n; /* no or next */
1146 register I32 ln; /* len or last */
1147 register char *s; /* operand or save */
3280af22 1148 register char *locinput = PL_reginput;
c277df42 1149 register I32 c1, c2, paren; /* case fold search, parenth */
1150 int minmod = 0, sw = 0, logical = 0;
4633a7c4 1151#ifdef DEBUGGING
3280af22 1152 PL_regindent++;
4633a7c4 1153#endif
a0d0e21e 1154
a0ed51b3 1155 /* Note that nextchr is a byte even in UTF */
76e3520e 1156 nextchr = UCHARAT(locinput);
a0d0e21e 1157 scan = prog;
1158 while (scan != NULL) {
c277df42 1159#define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO)
a687059c 1160#ifdef DEBUGGING
c277df42 1161# define sayYES goto yes
1162# define sayNO goto no
1163# define saySAME(x) if (x) goto yes; else goto no
1164# define REPORT_CODE_OFF 24
4633a7c4 1165#else
c277df42 1166# define sayYES return 1
1167# define sayNO return 0
1168# define saySAME(x) return x
a687059c 1169#endif
c277df42 1170 DEBUG_r( {
1171 SV *prop = sv_newmortal();
3280af22 1172 int docolor = *PL_colors[0];
c277df42 1173 int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
3280af22 1174 int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput);
beed8111 1175 /* The part of the string before starttry has one color
1176 (pref0_len chars), between starttry and current
1177 position another one (pref_len - pref0_len chars),
1178 after the current position the third one.
1179 We assume that pref0_len <= pref_len, otherwise we
1180 decrease pref0_len. */
3280af22 1181 int pref_len = (locinput - PL_bostr > (5 + taill) - l
1182 ? (5 + taill) - l : locinput - PL_bostr);
364723c2 1183 int pref0_len = pref_len - (locinput - PL_reg_starttry);
c277df42 1184
3280af22 1185 if (l + pref_len < (5 + taill) && l < PL_regeol - locinput)
1186 l = ( PL_regeol - locinput > (5 + taill) - pref_len
1187 ? (5 + taill) - pref_len : PL_regeol - locinput);
8d300b32 1188 if (pref0_len < 0)
1189 pref0_len = 0;
beed8111 1190 if (pref0_len > pref_len)
1191 pref0_len = pref_len;
c277df42 1192 regprop(prop, scan);
1193 PerlIO_printf(Perl_debug_log,
8d300b32 1194 "%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n",
3280af22 1195 locinput - PL_bostr,
8d300b32 1196 PL_colors[4], pref0_len,
1197 locinput - pref_len, PL_colors[5],
1198 PL_colors[2], pref_len - pref0_len,
1199 locinput - pref_len + pref0_len, PL_colors[3],
c277df42 1200 (docolor ? "" : "> <"),
3280af22 1201 PL_colors[0], l, locinput, PL_colors[1],
c277df42 1202 15 - l - pref_len + 1,
1203 "",
3280af22 1204 scan - PL_regprogram, PL_regindent*2, "",
c277df42 1205 SvPVX(prop));
1206 } );
a687059c 1207
c277df42 1208 next = scan + NEXT_OFF(scan);
a0d0e21e 1209 if (next == scan)
1210 next = NULL;
a687059c 1211
a0d0e21e 1212 switch (OP(scan)) {
1213 case BOL:
3280af22 1214 if (locinput == PL_bostr
1215 ? PL_regprev == '\n'
1216 : (PL_multiline &&
1217 (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
a0d0e21e 1218 {
a0ed51b3 1219 /* regtill = regbol; */
a0d0e21e 1220 break;
1221 }
4633a7c4 1222 sayNO;
a0d0e21e 1223 case MBOL:
3280af22 1224 if (locinput == PL_bostr
1225 ? PL_regprev == '\n'
1226 : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
a0d0e21e 1227 {
1228 break;
1229 }
4633a7c4 1230 sayNO;
a0d0e21e 1231 case SBOL:
3280af22 1232 if (locinput == PL_regbol && PL_regprev == '\n')
a0d0e21e 1233 break;
4633a7c4 1234 sayNO;
774d564b 1235 case GPOS:
22e551b9 1236 if (locinput == PL_reg_ganch)
a0d0e21e 1237 break;
4633a7c4 1238 sayNO;
a0d0e21e 1239 case EOL:
3280af22 1240 if (PL_multiline)
a0d0e21e 1241 goto meol;
1242 else
1243 goto seol;
1244 case MEOL:
1245 meol:
3280af22 1246 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
4633a7c4 1247 sayNO;
a0d0e21e 1248 break;
1249 case SEOL:
1250 seol:
3280af22 1251 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
4633a7c4 1252 sayNO;
3280af22 1253 if (PL_regeol - locinput > 1)
4633a7c4 1254 sayNO;
a0d0e21e 1255 break;
b85d18e9 1256 case EOS:
3280af22 1257 if (PL_regeol != locinput)
b85d18e9 1258 sayNO;
1259 break;
a0ed51b3 1260 case SANYUTF8:
1261 if (nextchr & 0x80) {
6f06b55f 1262 locinput += PL_utf8skip[nextchr];
a0ed51b3 1263 if (locinput > PL_regeol)
1264 sayNO;
1265 nextchr = UCHARAT(locinput);
1266 break;
1267 }
1268 if (!nextchr && locinput >= PL_regeol)
1269 sayNO;
1270 nextchr = UCHARAT(++locinput);
1271 break;
a0d0e21e 1272 case SANY:
3280af22 1273 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1274 sayNO;
76e3520e 1275 nextchr = UCHARAT(++locinput);
a0d0e21e 1276 break;
a0ed51b3 1277 case ANYUTF8:
1278 if (nextchr & 0x80) {
6f06b55f 1279 locinput += PL_utf8skip[nextchr];
a0ed51b3 1280 if (locinput > PL_regeol)
1281 sayNO;
1282 nextchr = UCHARAT(locinput);
1283 break;
1284 }
1285 if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
1286 sayNO;
1287 nextchr = UCHARAT(++locinput);
1288 break;
22c35a8c 1289 case REG_ANY:
3280af22 1290 if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
4633a7c4 1291 sayNO;
76e3520e 1292 nextchr = UCHARAT(++locinput);
a0d0e21e 1293 break;
bbce6d69 1294 case EXACT:
161b471a 1295 s = (char *) OPERAND(scan);
c277df42 1296 ln = UCHARAT(s++);
a0d0e21e 1297 /* Inline the first character, for speed. */
76e3520e 1298 if (UCHARAT(s) != nextchr)
4633a7c4 1299 sayNO;
3280af22 1300 if (PL_regeol - locinput < ln)
4633a7c4 1301 sayNO;
36477c24 1302 if (ln > 1 && memNE(s, locinput, ln))
4633a7c4 1303 sayNO;
a0d0e21e 1304 locinput += ln;
76e3520e 1305 nextchr = UCHARAT(locinput);
bbce6d69 1306 break;
1307 case EXACTFL:
3280af22 1308 PL_reg_flags |= RF_tainted;
bbce6d69 1309 /* FALL THROUGH */
1310 case EXACTF:
161b471a 1311 s = (char *) OPERAND(scan);
c277df42 1312 ln = UCHARAT(s++);
a0ed51b3 1313
1314 if (UTF) {
1315 char *l = locinput;
1316 char *e = s + ln;
1317 c1 = OP(scan) == EXACTF;
1318 while (s < e) {
1319 if (l >= PL_regeol)
1320 sayNO;
dfe13c55 1321 if (utf8_to_uv((U8*)s, 0) != (c1 ?
1322 toLOWER_utf8((U8*)l) :
1323 toLOWER_LC_utf8((U8*)l)))
1324 {
a0ed51b3 1325 sayNO;
dfe13c55 1326 }
a0ed51b3 1327 s += UTF8SKIP(s);
1328 l += UTF8SKIP(l);
1329 }
1330 locinput = l;
1331 nextchr = UCHARAT(locinput);
1332 break;
1333 }
1334
bbce6d69 1335 /* Inline the first character, for speed. */
76e3520e 1336 if (UCHARAT(s) != nextchr &&
bbce6d69 1337 UCHARAT(s) != ((OP(scan) == EXACTF)
22c35a8c 1338 ? PL_fold : PL_fold_locale)[nextchr])
bbce6d69 1339 sayNO;
3280af22 1340 if (PL_regeol - locinput < ln)
bbce6d69 1341 sayNO;
5f05dabc 1342 if (ln > 1 && (OP(scan) == EXACTF
1343 ? ibcmp(s, locinput, ln)
1344 : ibcmp_locale(s, locinput, ln)))
bbce6d69 1345 sayNO;
1346 locinput += ln;
76e3520e 1347 nextchr = UCHARAT(locinput);
a0d0e21e 1348 break;
a0ed51b3 1349 case ANYOFUTF8:
1350 s = (char *) OPERAND(scan);
1351 if (!REGINCLASSUTF8(scan, (U8*)locinput))
1352 sayNO;
1353 if (locinput >= PL_regeol)
1354 sayNO;
6f06b55f 1355 locinput += PL_utf8skip[nextchr];
a0ed51b3 1356 nextchr = UCHARAT(locinput);
1357 break;
a0d0e21e 1358 case ANYOF:
161b471a 1359 s = (char *) OPERAND(scan);
76e3520e 1360 if (nextchr < 0)
1361 nextchr = UCHARAT(locinput);
873ef191 1362 if (!REGINCLASS(s, nextchr))
4633a7c4 1363 sayNO;
3280af22 1364 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1365 sayNO;
76e3520e 1366 nextchr = UCHARAT(++locinput);
a0d0e21e 1367 break;
bbce6d69 1368 case ALNUML:
3280af22 1369 PL_reg_flags |= RF_tainted;
bbce6d69 1370 /* FALL THROUGH */
a0d0e21e 1371 case ALNUM:
76e3520e 1372 if (!nextchr)
4633a7c4 1373 sayNO;
bbce6d69 1374 if (!(OP(scan) == ALNUM
76e3520e 1375 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
4633a7c4 1376 sayNO;
76e3520e 1377 nextchr = UCHARAT(++locinput);
a0d0e21e 1378 break;
a0ed51b3 1379 case ALNUMLUTF8:
1380 PL_reg_flags |= RF_tainted;
1381 /* FALL THROUGH */
1382 case ALNUMUTF8:
1383 if (!nextchr)
1384 sayNO;
1385 if (nextchr & 0x80) {
1386 if (!(OP(scan) == ALNUMUTF8
dfe13c55 1387 ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
1388 : isALNUM_LC_utf8((U8*)locinput)))
1389 {
a0ed51b3 1390 sayNO;
dfe13c55 1391 }
6f06b55f 1392 locinput += PL_utf8skip[nextchr];
a0ed51b3 1393 nextchr = UCHARAT(locinput);
1394 break;
1395 }
1396 if (!(OP(scan) == ALNUMUTF8
1397 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
1398 sayNO;
1399 nextchr = UCHARAT(++locinput);
1400 break;
bbce6d69 1401 case NALNUML:
3280af22 1402 PL_reg_flags |= RF_tainted;
bbce6d69 1403 /* FALL THROUGH */
a0d0e21e 1404 case NALNUM:
3280af22 1405 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1406 sayNO;
bbce6d69 1407 if (OP(scan) == NALNUM
76e3520e 1408 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
4633a7c4 1409 sayNO;
76e3520e 1410 nextchr = UCHARAT(++locinput);
a0d0e21e 1411 break;
a0ed51b3 1412 case NALNUMLUTF8:
1413 PL_reg_flags |= RF_tainted;
1414 /* FALL THROUGH */
1415 case NALNUMUTF8:
1416 if (!nextchr && locinput >= PL_regeol)
1417 sayNO;
1418 if (nextchr & 0x80) {
1419 if (OP(scan) == NALNUMUTF8
dfe13c55 1420 ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
1421 : isALNUM_LC_utf8((U8*)locinput))
1422 {
a0ed51b3 1423 sayNO;
dfe13c55 1424 }
6f06b55f 1425 locinput += PL_utf8skip[nextchr];
a0ed51b3 1426 nextchr = UCHARAT(locinput);
1427 break;
1428 }
1429 if (OP(scan) == NALNUMUTF8
1430 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
1431 sayNO;
1432 nextchr = UCHARAT(++locinput);
1433 break;
bbce6d69 1434 case BOUNDL:
1435 case NBOUNDL:
3280af22 1436 PL_reg_flags |= RF_tainted;
bbce6d69 1437 /* FALL THROUGH */
a0d0e21e 1438 case BOUND:
bbce6d69 1439 case NBOUND:
1440 /* was last char in word? */
3280af22 1441 ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev;
bbce6d69 1442 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
1443 ln = isALNUM(ln);
76e3520e 1444 n = isALNUM(nextchr);
bbce6d69 1445 }
1446 else {
1447 ln = isALNUM_LC(ln);
76e3520e 1448 n = isALNUM_LC(nextchr);
bbce6d69 1449 }
95bac841 1450 if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL))
4633a7c4 1451 sayNO;
a0d0e21e 1452 break;
a0ed51b3 1453 case BOUNDLUTF8:
1454 case NBOUNDLUTF8:
1455 PL_reg_flags |= RF_tainted;
1456 /* FALL THROUGH */
1457 case BOUNDUTF8:
1458 case NBOUNDUTF8:
1459 /* was last char in word? */
dfe13c55 1460 ln = (locinput != PL_regbol)
1461 ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
a0ed51b3 1462 if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
1463 ln = isALNUM_uni(ln);
dfe13c55 1464 n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
a0ed51b3 1465 }
1466 else {
1467 ln = isALNUM_LC_uni(ln);
dfe13c55 1468 n = isALNUM_LC_utf8((U8*)locinput);
a0ed51b3 1469 }
1470 if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8))
1471 sayNO;
1472 break;
bbce6d69 1473 case SPACEL:
3280af22 1474 PL_reg_flags |= RF_tainted;
bbce6d69 1475 /* FALL THROUGH */
a0d0e21e 1476 case SPACE:
3280af22 1477 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1478 sayNO;
bbce6d69 1479 if (!(OP(scan) == SPACE
76e3520e 1480 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
4633a7c4 1481 sayNO;
76e3520e 1482 nextchr = UCHARAT(++locinput);
a0d0e21e 1483 break;
a0ed51b3 1484 case SPACELUTF8:
1485 PL_reg_flags |= RF_tainted;
1486 /* FALL THROUGH */
1487 case SPACEUTF8:
1488 if (!nextchr && locinput >= PL_regeol)
1489 sayNO;
1490 if (nextchr & 0x80) {
1491 if (!(OP(scan) == SPACEUTF8
dfe13c55 1492 ? swash_fetch(PL_utf8_space,(U8*)locinput)
1493 : isSPACE_LC_utf8((U8*)locinput)))
1494 {
a0ed51b3 1495 sayNO;
dfe13c55 1496 }
6f06b55f 1497 locinput += PL_utf8skip[nextchr];
a0ed51b3 1498 nextchr = UCHARAT(locinput);
1499 break;
1500 }
1501 if (!(OP(scan) == SPACEUTF8
1502 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
1503 sayNO;
1504 nextchr = UCHARAT(++locinput);
1505 break;
bbce6d69 1506 case NSPACEL:
3280af22 1507 PL_reg_flags |= RF_tainted;
bbce6d69 1508 /* FALL THROUGH */
a0d0e21e 1509 case NSPACE:
76e3520e 1510 if (!nextchr)
4633a7c4 1511 sayNO;
bbce6d69 1512 if (OP(scan) == SPACE
76e3520e 1513 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
4633a7c4 1514 sayNO;
76e3520e 1515 nextchr = UCHARAT(++locinput);
a0d0e21e 1516 break;
a0ed51b3 1517 case NSPACELUTF8:
1518 PL_reg_flags |= RF_tainted;
1519 /* FALL THROUGH */
1520 case NSPACEUTF8:
1521 if (!nextchr)
1522 sayNO;
1523 if (nextchr & 0x80) {
1524 if (OP(scan) == NSPACEUTF8
dfe13c55 1525 ? swash_fetch(PL_utf8_space,(U8*)locinput)
1526 : isSPACE_LC_utf8((U8*)locinput))
1527 {
a0ed51b3 1528 sayNO;
dfe13c55 1529 }
6f06b55f 1530 locinput += PL_utf8skip[nextchr];
a0ed51b3 1531 nextchr = UCHARAT(locinput);
1532 break;
1533 }
1534 if (OP(scan) == NSPACEUTF8
1535 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
1536 sayNO;
1537 nextchr = UCHARAT(++locinput);
1538 break;
a0d0e21e 1539 case DIGIT:
76e3520e 1540 if (!isDIGIT(nextchr))
4633a7c4 1541 sayNO;
76e3520e 1542 nextchr = UCHARAT(++locinput);
a0d0e21e 1543 break;
a0ed51b3 1544 case DIGITUTF8:
1545 if (nextchr & 0x80) {
dfe13c55 1546 if (!(swash_fetch(PL_utf8_digit,(U8*)locinput)))
a0ed51b3 1547 sayNO;
6f06b55f 1548 locinput += PL_utf8skip[nextchr];
a0ed51b3 1549 nextchr = UCHARAT(locinput);
1550 break;
1551 }
1552 if (!isDIGIT(nextchr))
1553 sayNO;
1554 nextchr = UCHARAT(++locinput);
1555 break;
a0d0e21e 1556 case NDIGIT:
3280af22 1557 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1558 sayNO;
76e3520e 1559 if (isDIGIT(nextchr))
4633a7c4 1560 sayNO;
76e3520e 1561 nextchr = UCHARAT(++locinput);
a0d0e21e 1562 break;
a0ed51b3 1563 case NDIGITUTF8:
1564 if (!nextchr && locinput >= PL_regeol)
1565 sayNO;
1566 if (nextchr & 0x80) {
dfe13c55 1567 if (swash_fetch(PL_utf8_digit,(U8*)locinput))
a0ed51b3 1568 sayNO;
6f06b55f 1569 locinput += PL_utf8skip[nextchr];
a0ed51b3 1570 nextchr = UCHARAT(locinput);
1571 break;
1572 }
1573 if (isDIGIT(nextchr))
1574 sayNO;
1575 nextchr = UCHARAT(++locinput);
1576 break;
1577 case CLUMP:
dfe13c55 1578 if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput))
a0ed51b3 1579 sayNO;
6f06b55f 1580 locinput += PL_utf8skip[nextchr];
dfe13c55 1581 while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput))
a0ed51b3 1582 locinput += UTF8SKIP(locinput);
1583 if (locinput > PL_regeol)
1584 sayNO;
1585 nextchr = UCHARAT(locinput);
1586 break;
c8756f30 1587 case REFFL:
3280af22 1588 PL_reg_flags |= RF_tainted;
c8756f30 1589 /* FALL THROUGH */
c277df42 1590 case REF:
c8756f30 1591 case REFF:
c277df42 1592 n = ARG(scan); /* which paren pair */
3280af22 1593 s = PL_regstartp[n];
1594 if (*PL_reglastparen < n || !s)
af3f8c16 1595 sayNO; /* Do not match unless seen CLOSEn. */
3280af22 1596 if (s == PL_regendp[n])
a0d0e21e 1597 break;
a0ed51b3 1598
1599 if (UTF && OP(scan) != REF) { /* REF can do byte comparison */
1600 char *l = locinput;
1601 char *e = PL_regendp[n];
1602 /*
1603 * Note that we can't do the "other character" lookup trick as
1604 * in the 8-bit case (no pun intended) because in Unicode we
1605 * have to map both upper and title case to lower case.
1606 */
1607 if (OP(scan) == REFF) {
1608 while (s < e) {
1609 if (l >= PL_regeol)
1610 sayNO;
dfe13c55 1611 if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l))
a0ed51b3 1612 sayNO;
1613 s += UTF8SKIP(s);
1614 l += UTF8SKIP(l);
1615 }
1616 }
1617 else {
1618 while (s < e) {
1619 if (l >= PL_regeol)
1620 sayNO;
dfe13c55 1621 if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l))
a0ed51b3 1622 sayNO;
1623 s += UTF8SKIP(s);
1624 l += UTF8SKIP(l);
1625 }
1626 }
1627 locinput = l;
1628 nextchr = UCHARAT(locinput);
1629 break;
1630 }
1631
a0d0e21e 1632 /* Inline the first character, for speed. */
76e3520e 1633 if (UCHARAT(s) != nextchr &&
c8756f30 1634 (OP(scan) == REF ||
1635 (UCHARAT(s) != ((OP(scan) == REFF
22c35a8c 1636 ? PL_fold : PL_fold_locale)[nextchr]))))
4633a7c4 1637 sayNO;
3280af22 1638 ln = PL_regendp[n] - s;
1639 if (locinput + ln > PL_regeol)
4633a7c4 1640 sayNO;
c8756f30 1641 if (ln > 1 && (OP(scan) == REF
1642 ? memNE(s, locinput, ln)
1643 : (OP(scan) == REFF
1644 ? ibcmp(s, locinput, ln)
1645 : ibcmp_locale(s, locinput, ln))))
4633a7c4 1646 sayNO;
a0d0e21e 1647 locinput += ln;
76e3520e 1648 nextchr = UCHARAT(locinput);
a0d0e21e 1649 break;
1650
1651 case NOTHING:
c277df42 1652 case TAIL:
a0d0e21e 1653 break;
1654 case BACK:
1655 break;
c277df42 1656 case EVAL:
1657 {
1658 dSP;
533c011a 1659 OP_4tree *oop = PL_op;
3280af22 1660 COP *ocurcop = PL_curcop;
1661 SV **ocurpad = PL_curpad;
c277df42 1662 SV *ret;
1663
1664 n = ARG(scan);
533c011a 1665 PL_op = (OP_4tree*)PL_regdata->data[n];
1666 DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) );
3280af22 1667 PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 1]);
9661b544 1668 PL_reg_magic->mg_len = locinput - PL_bostr;
5c5e4c24 1669 PL_regendp[0] = locinput;
c277df42 1670
76e3520e 1671 CALLRUNOPS(); /* Scalar context. */
c277df42 1672 SPAGAIN;
1673 ret = POPs;
1674 PUTBACK;
1675
0f5d15d6 1676 PL_op = oop;
1677 PL_curpad = ocurpad;
1678 PL_curcop = ocurcop;
c277df42 1679 if (logical) {
0f5d15d6 1680 if (logical == 2) { /* Postponed subexpression. */
1681 regexp *re;
22c35a8c 1682 MAGIC *mg = Null(MAGIC*);
0f5d15d6 1683 re_cc_state state;
1684 CURCUR cctmp;
1685 CHECKPOINT cp, lastcp;
1686
1687 if(SvROK(ret) || SvRMAGICAL(ret)) {
1688 SV *sv = SvROK(ret) ? SvRV(ret) : ret;
1689
1690 if(SvMAGICAL(sv))
1691 mg = mg_find(sv, 'r');
1692 }
1693 if (mg) {
1694 re = (regexp *)mg->mg_obj;
df0003d4 1695 (void)ReREFCNT_inc(re);
0f5d15d6 1696 }
1697 else {
1698 STRLEN len;
1699 char *t = SvPV(ret, len);
1700 PMOP pm;
1701 char *oprecomp = PL_regprecomp;
1702 I32 osize = PL_regsize;
1703 I32 onpar = PL_regnpar;
1704
1705 pm.op_pmflags = 0;
1706 re = CALLREGCOMP(t, t + len, &pm);
1707 if (!(SvFLAGS(ret)
1708 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY)))
1709 sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0);
1710 PL_regprecomp = oprecomp;
1711 PL_regsize = osize;
1712 PL_regnpar = onpar;
1713 }
1714 DEBUG_r(
1715 PerlIO_printf(Perl_debug_log,
1716 "Entering embedded `%s%.60s%s%s'\n",
1717 PL_colors[0],
1718 re->precomp,
1719 PL_colors[1],
1720 (strlen(re->precomp) > 60 ? "..." : ""))
1721 );
1722 state.node = next;
1723 state.prev = PL_reg_call_cc;
1724 state.cc = PL_regcc;
1725 state.re = PL_reg_re;
1726
1727 cctmp.cur = 0;
1728 cctmp.oldcc = 0;
1729 PL_regcc = &cctmp;
1730
1731 cp = regcppush(0); /* Save *all* the positions. */
1732 REGCP_SET;
1733 cache_re(re);
1734 state.ss = PL_savestack_ix;
1735 *PL_reglastparen = 0;
1736 PL_reg_call_cc = &state;
1737 PL_reginput = locinput;
1738 if (regmatch(re->program + 1)) {
1739 ReREFCNT_dec(re);
1740 regcpblow(cp);
1741 sayYES;
1742 }
1743 DEBUG_r(
1744 PerlIO_printf(Perl_debug_log,
1745 "%*s failed...\n",
1746 REPORT_CODE_OFF+PL_regindent*2, "")
1747 );
1748 ReREFCNT_dec(re);
1749 REGCP_UNWIND;
1750 regcppop();
1751 PL_reg_call_cc = state.prev;
1752 PL_regcc = state.cc;
1753 PL_reg_re = state.re;
d3790889 1754 cache_re(PL_reg_re);
0f5d15d6 1755 sayNO;
1756 }
c277df42 1757 sw = SvTRUE(ret);
0f5d15d6 1758 logical = 0;
a0ed51b3 1759 }
1760 else
3280af22 1761 sv_setsv(save_scalar(PL_replgv), ret);
c277df42 1762 break;
1763 }
a0d0e21e 1764 case OPEN:
c277df42 1765 n = ARG(scan); /* which paren pair */
3280af22 1766 PL_reg_start_tmp[n] = locinput;
1767 if (n > PL_regsize)
1768 PL_regsize = n;
a0d0e21e 1769 break;
1770 case CLOSE:
c277df42 1771 n = ARG(scan); /* which paren pair */
3280af22 1772 PL_regstartp[n] = PL_reg_start_tmp[n];
1773 PL_regendp[n] = locinput;
1774 if (n > *PL_reglastparen)
1775 *PL_reglastparen = n;
a0d0e21e 1776 break;
c277df42 1777 case GROUPP:
1778 n = ARG(scan); /* which paren pair */
3280af22 1779 sw = (*PL_reglastparen >= n && PL_regendp[n] != NULL);
c277df42 1780 break;
1781 case IFTHEN:
1782 if (sw)
1783 next = NEXTOPER(NEXTOPER(scan));
1784 else {
1785 next = scan + ARG(scan);
1786 if (OP(next) == IFTHEN) /* Fake one. */
1787 next = NEXTOPER(NEXTOPER(next));
1788 }
1789 break;
1790 case LOGICAL:
0f5d15d6 1791 logical = scan->flags;
c277df42 1792 break;
a0d0e21e 1793 case CURLYX: {
1794 CURCUR cc;
3280af22 1795 CHECKPOINT cp = PL_savestack_ix;
c277df42 1796
1797 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
1798 next += ARG(next);
3280af22 1799 cc.oldcc = PL_regcc;
1800 PL_regcc = &cc;
1801 cc.parenfloor = *PL_reglastparen;
a0d0e21e 1802 cc.cur = -1;
1803 cc.min = ARG1(scan);
1804 cc.max = ARG2(scan);
c277df42 1805 cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
a0d0e21e 1806 cc.next = next;
1807 cc.minmod = minmod;
1808 cc.lastloc = 0;
3280af22 1809 PL_reginput = locinput;
a0d0e21e 1810 n = regmatch(PREVOPER(next)); /* start on the WHILEM */
1811 regcpblow(cp);
3280af22 1812 PL_regcc = cc.oldcc;
4633a7c4 1813 saySAME(n);
a0d0e21e 1814 }
1815 /* NOT REACHED */
1816 case WHILEM: {
1817 /*
1818 * This is really hard to understand, because after we match
1819 * what we're trying to match, we must make sure the rest of
1820 * the RE is going to match for sure, and to do that we have
1821 * to go back UP the parse tree by recursing ever deeper. And
1822 * if it fails, we have to reset our parent's current state
1823 * that we can try again after backing off.
1824 */
1825
c277df42 1826 CHECKPOINT cp, lastcp;
3280af22 1827 CURCUR* cc = PL_regcc;
c277df42 1828 char *lastloc = cc->lastloc; /* Detection of 0-len. */
1829
4633a7c4 1830 n = cc->cur + 1; /* how many we know we matched */
3280af22 1831 PL_reginput = locinput;
a0d0e21e 1832
c277df42 1833 DEBUG_r(
1834 PerlIO_printf(Perl_debug_log,
1835 "%*s %ld out of %ld..%ld cc=%lx\n",
3280af22 1836 REPORT_CODE_OFF+PL_regindent*2, "",
c277df42 1837 (long)n, (long)cc->min,
1838 (long)cc->max, (long)cc)
1839 );
4633a7c4 1840
a0d0e21e 1841 /* If degenerate scan matches "", assume scan done. */
1842
579cf2c3 1843 if (locinput == cc->lastloc && n >= cc->min) {
3280af22 1844 PL_regcc = cc->oldcc;
1845 ln = PL_regcc->cur;
c277df42 1846 DEBUG_r(
c3464db5 1847 PerlIO_printf(Perl_debug_log,
1848 "%*s empty match detected, try continuation...\n",
3280af22 1849 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1850 );
a0d0e21e 1851 if (regmatch(cc->next))
4633a7c4 1852 sayYES;
c277df42 1853 DEBUG_r(
c3464db5 1854 PerlIO_printf(Perl_debug_log,
1855 "%*s failed...\n",
3280af22 1856 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1857 );
3280af22 1858 PL_regcc->cur = ln;
1859 PL_regcc = cc;
4633a7c4 1860 sayNO;
a0d0e21e 1861 }
1862
1863 /* First just match a string of min scans. */
1864
1865 if (n < cc->min) {
1866 cc->cur = n;
1867 cc->lastloc = locinput;
4633a7c4 1868 if (regmatch(cc->scan))
1869 sayYES;
1870 cc->cur = n - 1;
c277df42 1871 cc->lastloc = lastloc;
1872 DEBUG_r(
c3464db5 1873 PerlIO_printf(Perl_debug_log,
1874 "%*s failed...\n",
3280af22 1875 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1876 );
4633a7c4 1877 sayNO;
a0d0e21e 1878 }
1879
1880 /* Prefer next over scan for minimal matching. */
1881
1882 if (cc->minmod) {
3280af22 1883 PL_regcc = cc->oldcc;
1884 ln = PL_regcc->cur;
5f05dabc 1885 cp = regcppush(cc->parenfloor);
c277df42 1886 REGCP_SET;
5f05dabc 1887 if (regmatch(cc->next)) {
c277df42 1888 regcpblow(cp);
4633a7c4 1889 sayYES; /* All done. */
5f05dabc 1890 }
c277df42 1891 REGCP_UNWIND;
5f05dabc 1892 regcppop();
3280af22 1893 PL_regcc->cur = ln;
1894 PL_regcc = cc;
a0d0e21e 1895
c277df42 1896 if (n >= cc->max) { /* Maximum greed exceeded? */
599cee73 1897 if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
3280af22 1898 && !(PL_reg_flags & RF_warned)) {
1899 PL_reg_flags |= RF_warned;
599cee73 1900 warner(WARN_UNSAFE, "%s limit (%d) exceeded",
2f3ca594 1901 "Complex regular subexpression recursion",
1902 REG_INFTY - 1);
c277df42 1903 }
4633a7c4 1904 sayNO;
c277df42 1905 }
a687059c 1906
c277df42 1907 DEBUG_r(
c3464db5 1908 PerlIO_printf(Perl_debug_log,
1909 "%*s trying longer...\n",
3280af22 1910 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1911 );
a0d0e21e 1912 /* Try scanning more and see if it helps. */
3280af22 1913 PL_reginput = locinput;
a0d0e21e 1914 cc->cur = n;
1915 cc->lastloc = locinput;
5f05dabc 1916 cp = regcppush(cc->parenfloor);
c277df42 1917 REGCP_SET;
5f05dabc 1918 if (regmatch(cc->scan)) {
c277df42 1919 regcpblow(cp);
4633a7c4 1920 sayYES;
5f05dabc 1921 }
c277df42 1922 DEBUG_r(
c3464db5 1923 PerlIO_printf(Perl_debug_log,
1924 "%*s failed...\n",
3280af22 1925 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1926 );
1927 REGCP_UNWIND;
5f05dabc 1928 regcppop();
4633a7c4 1929 cc->cur = n - 1;
c277df42 1930 cc->lastloc = lastloc;
4633a7c4 1931 sayNO;
a0d0e21e 1932 }
1933
1934 /* Prefer scan over next for maximal matching. */
1935
1936 if (n < cc->max) { /* More greed allowed? */
5f05dabc 1937 cp = regcppush(cc->parenfloor);
a0d0e21e 1938 cc->cur = n;
1939 cc->lastloc = locinput;
c277df42 1940 REGCP_SET;
5f05dabc 1941 if (regmatch(cc->scan)) {
c277df42 1942 regcpblow(cp);
4633a7c4 1943 sayYES;
5f05dabc 1944 }
c277df42 1945 REGCP_UNWIND;
a0d0e21e 1946 regcppop(); /* Restore some previous $<digit>s? */
3280af22 1947 PL_reginput = locinput;
c277df42 1948 DEBUG_r(
c3464db5 1949 PerlIO_printf(Perl_debug_log,
1950 "%*s failed, try continuation...\n",
3280af22 1951 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1952 );
1953 }
599cee73 1954 if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
1955 && !(PL_reg_flags & RF_warned)) {
3280af22 1956 PL_reg_flags |= RF_warned;
599cee73 1957 warner(WARN_UNSAFE, "%s limit (%d) exceeded",
cb5d145d 1958 "Complex regular subexpression recursion",
1959 REG_INFTY - 1);
a0d0e21e 1960 }
1961
1962 /* Failed deeper matches of scan, so see if this one works. */
3280af22 1963 PL_regcc = cc->oldcc;
1964 ln = PL_regcc->cur;
a0d0e21e 1965 if (regmatch(cc->next))
4633a7c4 1966 sayYES;
c277df42 1967 DEBUG_r(
c3464db5 1968 PerlIO_printf(Perl_debug_log, "%*s failed...\n",
3280af22 1969 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1970 );
3280af22 1971 PL_regcc->cur = ln;
1972 PL_regcc = cc;
4633a7c4 1973 cc->cur = n - 1;
c277df42 1974 cc->lastloc = lastloc;
4633a7c4 1975 sayNO;
a0d0e21e 1976 }
1977 /* NOT REACHED */
c277df42 1978 case BRANCHJ:
1979 next = scan + ARG(scan);
1980 if (next == scan)
1981 next = NULL;
1982 inner = NEXTOPER(NEXTOPER(scan));
1983 goto do_branch;
1984 case BRANCH:
1985 inner = NEXTOPER(scan);
1986 do_branch:
1987 {
1988 CHECKPOINT lastcp;
1989 c1 = OP(scan);
1990 if (OP(next) != c1) /* No choice. */
1991 next = inner; /* Avoid recursion. */
a0d0e21e 1992 else {
3280af22 1993 int lastparen = *PL_reglastparen;
c277df42 1994
1995 REGCP_SET;
a0d0e21e 1996 do {
3280af22 1997 PL_reginput = locinput;
c277df42 1998 if (regmatch(inner))
4633a7c4 1999 sayYES;
c277df42 2000 REGCP_UNWIND;
3280af22 2001 for (n = *PL_reglastparen; n > lastparen; n--)
2002 PL_regendp[n] = 0;
2003 *PL_reglastparen = n;
c277df42 2004 scan = next;
a0d0e21e 2005 /*SUPPRESS 560*/
c277df42 2006 if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next)))
2007 next += n;
a0d0e21e 2008 else
c277df42 2009 next = NULL;
c277df42 2010 inner = NEXTOPER(scan);
2011 if (c1 == BRANCHJ) {
2012 inner = NEXTOPER(inner);
2013 }
2014 } while (scan != NULL && OP(scan) == c1);
4633a7c4 2015 sayNO;
a0d0e21e 2016 /* NOTREACHED */
a687059c 2017 }
a0d0e21e 2018 }
2019 break;
2020 case MINMOD:
2021 minmod = 1;
2022 break;
c277df42 2023 case CURLYM:
2024 {
00db4c45 2025 I32 l = 0;
c277df42 2026 CHECKPOINT lastcp;
2027
2028 /* We suppose that the next guy does not need
2029 backtracking: in particular, it is of constant length,
2030 and has no parenths to influence future backrefs. */
2031 ln = ARG1(scan); /* min to match */
2032 n = ARG2(scan); /* max to match */
c277df42 2033 paren = scan->flags;
2034 if (paren) {
3280af22 2035 if (paren > PL_regsize)
2036 PL_regsize = paren;
2037 if (paren > *PL_reglastparen)
2038 *PL_reglastparen = paren;
c277df42 2039 }
dc45a647 2040 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
c277df42 2041 if (paren)
2042 scan += NEXT_OFF(scan); /* Skip former OPEN. */
3280af22 2043 PL_reginput = locinput;
c277df42 2044 if (minmod) {
2045 minmod = 0;
2046 if (ln && regrepeat_hard(scan, ln, &l) < ln)
2047 sayNO;
5f4b28b2 2048 if (ln && l == 0 && n >= ln
c277df42 2049 /* In fact, this is tricky. If paren, then the
2050 fact that we did/didnot match may influence
2051 future execution. */
2052 && !(paren && ln == 0))
2053 ln = n;
3280af22 2054 locinput = PL_reginput;
22c35a8c 2055 if (PL_regkind[(U8)OP(next)] == EXACT) {
c277df42 2056 c1 = UCHARAT(OPERAND(next) + 1);
2057 if (OP(next) == EXACTF)
22c35a8c 2058 c2 = PL_fold[c1];
c277df42 2059 else if (OP(next) == EXACTFL)
22c35a8c 2060 c2 = PL_fold_locale[c1];
c277df42 2061 else
2062 c2 = c1;
a0ed51b3 2063 }
2064 else
c277df42 2065 c1 = c2 = -1000;
2066 REGCP_SET;
5f4b28b2 2067 /* This may be improved if l == 0. */
c277df42 2068 while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */
2069 /* If it could work, try it. */
2070 if (c1 == -1000 ||
3280af22 2071 UCHARAT(PL_reginput) == c1 ||
2072 UCHARAT(PL_reginput) == c2)
c277df42 2073 {
2074 if (paren) {
2075 if (n) {
dfe13c55 2076 PL_regstartp[paren] = HOPc(PL_reginput, -l);
3280af22 2077 PL_regendp[paren] = PL_reginput;
a0ed51b3 2078 }
2079 else
3280af22 2080 PL_regendp[paren] = NULL;
c277df42 2081 }
2082 if (regmatch(next))
2083 sayYES;
2084 REGCP_UNWIND;
2085 }
2086 /* Couldn't or didn't -- move forward. */
3280af22 2087 PL_reginput = locinput;
c277df42 2088 if (regrepeat_hard(scan, 1, &l)) {
2089 ln++;
3280af22 2090 locinput = PL_reginput;
c277df42 2091 }
2092 else
2093 sayNO;
2094 }
a0ed51b3 2095 }
2096 else {
c277df42 2097 n = regrepeat_hard(scan, n, &l);
2098 if (n != 0 && l == 0
2099 /* In fact, this is tricky. If paren, then the
2100 fact that we did/didnot match may influence
2101 future execution. */
2102 && !(paren && ln == 0))
2103 ln = n;
3280af22 2104 locinput = PL_reginput;
c277df42 2105 DEBUG_r(
5c0ca799 2106 PerlIO_printf(Perl_debug_log,
2107 "%*s matched %ld times, len=%ld...\n",
3280af22 2108 REPORT_CODE_OFF+PL_regindent*2, "", n, l)
c277df42 2109 );
2110 if (n >= ln) {
22c35a8c 2111 if (PL_regkind[(U8)OP(next)] == EXACT) {
c277df42 2112 c1 = UCHARAT(OPERAND(next) + 1);
2113 if (OP(next) == EXACTF)
22c35a8c 2114 c2 = PL_fold[c1];
c277df42 2115 else if (OP(next) == EXACTFL)
22c35a8c 2116 c2 = PL_fold_locale[c1];
c277df42 2117 else
2118 c2 = c1;
a0ed51b3 2119 }
2120 else
c277df42 2121 c1 = c2 = -1000;
2122 }
2123 REGCP_SET;
2124 while (n >= ln) {
2125 /* If it could work, try it. */
2126 if (c1 == -1000 ||
3280af22 2127 UCHARAT(PL_reginput) == c1 ||
2128 UCHARAT(PL_reginput) == c2)
a0ed51b3 2129 {
2130 DEBUG_r(
c3464db5 2131 PerlIO_printf(Perl_debug_log,
2132 "%*s trying tail with n=%ld...\n",
3280af22 2133 REPORT_CODE_OFF+PL_regindent*2, "", n)
a0ed51b3 2134 );
2135 if (paren) {
2136 if (n) {
dfe13c55 2137 PL_regstartp[paren] = HOPc(PL_reginput, -l);
a0ed51b3 2138 PL_regendp[paren] = PL_reginput;
c277df42 2139 }
a0ed51b3 2140 else
2141 PL_regendp[paren] = NULL;
c277df42 2142 }
a0ed51b3 2143 if (regmatch(next))
2144 sayYES;
2145 REGCP_UNWIND;
2146 }
c277df42 2147 /* Couldn't or didn't -- back up. */
2148 n--;
dfe13c55 2149 locinput = HOPc(locinput, -l);
3280af22 2150 PL_reginput = locinput;
c277df42 2151 }
2152 }
2153 sayNO;
2154 break;
2155 }
2156 case CURLYN:
2157 paren = scan->flags; /* Which paren to set */
3280af22 2158 if (paren > PL_regsize)
2159 PL_regsize = paren;
2160 if (paren > *PL_reglastparen)
2161 *PL_reglastparen = paren;
c277df42 2162 ln = ARG1(scan); /* min to match */
2163 n = ARG2(scan); /* max to match */
dc45a647 2164 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
c277df42 2165 goto repeat;
a0d0e21e 2166 case CURLY:
c277df42 2167 paren = 0;
a0d0e21e 2168 ln = ARG1(scan); /* min to match */
2169 n = ARG2(scan); /* max to match */
dc45a647 2170 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
a0d0e21e 2171 goto repeat;
2172 case STAR:
2173 ln = 0;
c277df42 2174 n = REG_INFTY;
a0d0e21e 2175 scan = NEXTOPER(scan);
c277df42 2176 paren = 0;
a0d0e21e 2177 goto repeat;
2178 case PLUS:
c277df42 2179 ln = 1;
2180 n = REG_INFTY;
2181 scan = NEXTOPER(scan);
2182 paren = 0;
2183 repeat:
a0d0e21e 2184 /*
2185 * Lookahead to avoid useless match attempts
2186 * when we know what character comes next.
2187 */
22c35a8c 2188 if (PL_regkind[(U8)OP(next)] == EXACT) {
bbce6d69 2189 c1 = UCHARAT(OPERAND(next) + 1);
2190 if (OP(next) == EXACTF)
22c35a8c 2191 c2 = PL_fold[c1];
bbce6d69 2192 else if (OP(next) == EXACTFL)
22c35a8c 2193 c2 = PL_fold_locale[c1];
bbce6d69 2194 else
2195 c2 = c1;
2196 }
a0d0e21e 2197 else
bbce6d69 2198 c1 = c2 = -1000;
3280af22 2199 PL_reginput = locinput;
a0d0e21e 2200 if (minmod) {
c277df42 2201 CHECKPOINT lastcp;
a0d0e21e 2202 minmod = 0;
2203 if (ln && regrepeat(scan, ln) < ln)
4633a7c4 2204 sayNO;
a0ed51b3 2205 locinput = PL_reginput;
c277df42 2206 REGCP_SET;
0fe9bf95 2207 if (c1 != -1000) {
2208 char *e = locinput + n - ln; /* Should not check after this */
2209 char *old = locinput;
2210
2211 if (e >= PL_regeol || (n == REG_INFTY))
2212 e = PL_regeol - 1;
2213 while (1) {
2214 /* Find place 'next' could work */
2215 if (c1 == c2) {
2216 while (locinput <= e && *locinput != c1)
2217 locinput++;
2218 } else {
2219 while (locinput <= e
2220 && *locinput != c1
2221 && *locinput != c2)
2222 locinput++;
2223 }
2224 if (locinput > e)
2225 sayNO;
2226 /* PL_reginput == old now */
2227 if (locinput != old) {
2228 ln = 1; /* Did some */
2229 if (regrepeat(scan, locinput - old) <
2230 locinput - old)
2231 sayNO;
2232 }
2233 /* PL_reginput == locinput now */
2234 if (paren) {
2235 if (ln) {
2236 PL_regstartp[paren] = HOPc(locinput, -1);
2237 PL_regendp[paren] = locinput;
2238 }
2239 else
2240 PL_regendp[paren] = NULL;
2241 }
2242 if (regmatch(next))
2243 sayYES;
2244 PL_reginput = locinput; /* Could be reset... */
2245 REGCP_UNWIND;
2246 /* Couldn't or didn't -- move forward. */
2247 old = locinput++;
2248 }
2249 }
2250 else
c277df42 2251 while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
a0d0e21e 2252 /* If it could work, try it. */
bbce6d69 2253 if (c1 == -1000 ||
3280af22 2254 UCHARAT(PL_reginput) == c1 ||
2255 UCHARAT(PL_reginput) == c2)
bbce6d69 2256 {
c277df42 2257 if (paren) {
2258 if (n) {
dfe13c55 2259 PL_regstartp[paren] = HOPc(PL_reginput, -1);
3280af22 2260 PL_regendp[paren] = PL_reginput;
a0ed51b3 2261 }
2262 else
3280af22 2263 PL_regendp[paren] = NULL;
c277df42 2264 }
a0d0e21e 2265 if (regmatch(next))
4633a7c4 2266 sayYES;
c277df42 2267 REGCP_UNWIND;
bbce6d69 2268 }
c277df42 2269 /* Couldn't or didn't -- move forward. */
a0ed51b3 2270 PL_reginput = locinput;
a0d0e21e 2271 if (regrepeat(scan, 1)) {
2272 ln++;
a0ed51b3 2273 locinput = PL_reginput;
2274 }
2275 else
4633a7c4 2276 sayNO;
a0d0e21e 2277 }
2278 }
2279 else {
c277df42 2280 CHECKPOINT lastcp;
a0d0e21e 2281 n = regrepeat(scan, n);
a0ed51b3 2282 locinput = PL_reginput;
22c35a8c 2283 if (ln < n && PL_regkind[(U8)OP(next)] == EOL &&
3280af22 2284 (!PL_multiline || OP(next) == SEOL))
a0d0e21e 2285 ln = n; /* why back off? */
c277df42 2286 REGCP_SET;
2287 if (paren) {
2288 while (n >= ln) {
2289 /* If it could work, try it. */
2290 if (c1 == -1000 ||
3280af22 2291 UCHARAT(PL_reginput) == c1 ||
2292 UCHARAT(PL_reginput) == c2)
c277df42 2293 {
2294 if (paren && n) {
2295 if (n) {
dfe13c55 2296 PL_regstartp[paren] = HOPc(PL_reginput, -1);
3280af22 2297 PL_regendp[paren] = PL_reginput;
a0ed51b3 2298 }
2299 else
3280af22 2300 PL_regendp[paren] = NULL;
c277df42 2301 }
2302 if (regmatch(next))
2303 sayYES;
2304 REGCP_UNWIND;
2305 }
2306 /* Couldn't or didn't -- back up. */
2307 n--;
dfe13c55 2308 PL_reginput = locinput = HOPc(locinput, -1);
c277df42 2309 }
a0ed51b3 2310 }
2311 else {
c277df42 2312 while (n >= ln) {
2313 /* If it could work, try it. */
2314 if (c1 == -1000 ||
3280af22 2315 UCHARAT(PL_reginput) == c1 ||
2316 UCHARAT(PL_reginput) == c2)
c277df42 2317 {
2318 if (regmatch(next))
2319 sayYES;
2320 REGCP_UNWIND;
2321 }
2322 /* Couldn't or didn't -- back up. */
2323 n--;
dfe13c55 2324 PL_reginput = locinput = HOPc(locinput, -1);
bbce6d69 2325 }
a0d0e21e 2326 }
2327 }
4633a7c4 2328 sayNO;
c277df42 2329 break;
a0d0e21e 2330 case END:
0f5d15d6 2331 if (PL_reg_call_cc) {
2332 re_cc_state *cur_call_cc = PL_reg_call_cc;
2333 CURCUR *cctmp = PL_regcc;
2334 regexp *re = PL_reg_re;
2335 CHECKPOINT cp, lastcp;
2336
2337 cp = regcppush(0); /* Save *all* the positions. */
2338 REGCP_SET;
2339 regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of
2340 the caller. */
2341 PL_reginput = locinput; /* Make position available to
2342 the callcc. */
2343 cache_re(PL_reg_call_cc->re);
2344 PL_regcc = PL_reg_call_cc->cc;
2345 PL_reg_call_cc = PL_reg_call_cc->prev;
2346 if (regmatch(cur_call_cc->node)) {
2347 PL_reg_call_cc = cur_call_cc;
2348 regcpblow(cp);
2349 sayYES;
2350 }
2351 REGCP_UNWIND;
2352 regcppop();
2353 PL_reg_call_cc = cur_call_cc;
2354 PL_regcc = cctmp;
2355 PL_reg_re = re;
2356 cache_re(re);
2357
2358 DEBUG_r(
2359 PerlIO_printf(Perl_debug_log,
2360 "%*s continuation failed...\n",
2361 REPORT_CODE_OFF+PL_regindent*2, "")
2362 );
2363 sayNO;
2364 }
3280af22 2365 if (locinput < PL_regtill)
7e5428c5 2366 sayNO; /* Cannot match: too short. */
2367 /* Fall through */
2368 case SUCCEED:
3280af22 2369 PL_reginput = locinput; /* put where regtry can find it */
4633a7c4 2370 sayYES; /* Success! */
c277df42 2371 case SUSPEND:
2372 n = 1;
9fe1d20c 2373 PL_reginput = locinput;
c277df42 2374 goto do_ifmatch;
a0d0e21e 2375 case UNLESSM:
c277df42 2376 n = 0;
a0ed51b3 2377 if (scan->flags) {
0fe9bf95 2378 if (UTF) { /* XXXX This is absolutely
2379 broken, we read before
2380 start of string. */
2381 s = HOPMAYBEc(locinput, -scan->flags);
2382 if (!s)
2383 goto say_yes;
2384 PL_reginput = s;
2385 }
2386 else {
2387 if (locinput < PL_bostr + scan->flags)
2388 goto say_yes;
2389 PL_reginput = locinput - scan->flags;
2390 goto do_ifmatch;
2391 }
a0ed51b3 2392 }
2393 else
2394 PL_reginput = locinput;
c277df42 2395 goto do_ifmatch;
2396 case IFMATCH:
2397 n = 1;
a0ed51b3 2398 if (scan->flags) {
0fe9bf95 2399 if (UTF) { /* XXXX This is absolutely
2400 broken, we read before
2401 start of string. */
2402 s = HOPMAYBEc(locinput, -scan->flags);
2403 if (!s || s < PL_bostr)
2404 goto say_no;
2405 PL_reginput = s;
2406 }
2407 else {
2408 if (locinput < PL_bostr + scan->flags)
2409 goto say_no;
2410 PL_reginput = locinput - scan->flags;
2411 goto do_ifmatch;
2412 }
a0ed51b3 2413 }
2414 else
2415 PL_reginput = locinput;
2416
c277df42 2417 do_ifmatch:
c277df42 2418 inner = NEXTOPER(NEXTOPER(scan));
2419 if (regmatch(inner) != n) {
2420 say_no:
2421 if (logical) {
2422 logical = 0;
2423 sw = 0;
2424 goto do_longjump;
a0ed51b3 2425 }
2426 else
c277df42 2427 sayNO;
2428 }
2429 say_yes:
2430 if (logical) {
2431 logical = 0;
2432 sw = 1;
2433 }
fe44a5e8 2434 if (OP(scan) == SUSPEND) {
3280af22 2435 locinput = PL_reginput;
565764a8 2436 nextchr = UCHARAT(locinput);
fe44a5e8 2437 }
c277df42 2438 /* FALL THROUGH. */
2439 case LONGJMP:
2440 do_longjump:
2441 next = scan + ARG(scan);
2442 if (next == scan)
2443 next = NULL;
a0d0e21e 2444 break;
2445 default:
c030ccd9 2446 PerlIO_printf(PerlIO_stderr(), "%lx %d\n",
c277df42 2447 (unsigned long)scan, OP(scan));
4327152a 2448 croak("regexp memory corruption");
a687059c 2449 }
a0d0e21e 2450 scan = next;
2451 }
a687059c 2452
a0d0e21e 2453 /*
2454 * We get here only if there's trouble -- normally "case END" is
2455 * the terminating point.
2456 */
4327152a 2457 croak("corrupted regexp pointers");
a0d0e21e 2458 /*NOTREACHED*/
4633a7c4 2459 sayNO;
2460
2461yes:
2462#ifdef DEBUGGING
3280af22 2463 PL_regindent--;
4633a7c4 2464#endif
2465 return 1;
2466
2467no:
2468#ifdef DEBUGGING
3280af22 2469 PL_regindent--;
4633a7c4 2470#endif
a0d0e21e 2471 return 0;
a687059c 2472}
2473
2474/*
2475 - regrepeat - repeatedly match something simple, report how many
2476 */
2477/*
2478 * [This routine now assumes that it will only match on things of length 1.
2479 * That was true before, but now we assume scan - reginput is the count,
a0ed51b3 2480 * rather than incrementing count on every character. [Er, except utf8.]]
a687059c 2481 */
76e3520e 2482STATIC I32
c277df42 2483regrepeat(regnode *p, I32 max)
a687059c 2484{
5c0ca799 2485 dTHR;
a0d0e21e 2486 register char *scan;
2487 register char *opnd;
2488 register I32 c;
3280af22 2489 register char *loceol = PL_regeol;
a0ed51b3 2490 register I32 hardcount = 0;
a0d0e21e 2491
3280af22 2492 scan = PL_reginput;
c277df42 2493 if (max != REG_INFTY && max < loceol - scan)
a0d0e21e 2494 loceol = scan + max;
161b471a 2495 opnd = (char *) OPERAND(p);
a0d0e21e 2496 switch (OP(p)) {
22c35a8c 2497 case REG_ANY:
a0d0e21e 2498 while (scan < loceol && *scan != '\n')
2499 scan++;
2500 break;
2501 case SANY:
2502 scan = loceol;
2503 break;
a0ed51b3 2504 case ANYUTF8:
2505 loceol = PL_regeol;
2506 while (scan < loceol && *scan != '\n') {
2507 scan += UTF8SKIP(scan);
2508 hardcount++;
2509 }
2510 break;
2511 case SANYUTF8:
2512 loceol = PL_regeol;
2513 while (scan < loceol) {
2514 scan += UTF8SKIP(scan);
2515 hardcount++;
2516 }
2517 break;
bbce6d69 2518 case EXACT: /* length of string is 1 */
2519 c = UCHARAT(++opnd);
2520 while (scan < loceol && UCHARAT(scan) == c)
2521 scan++;
2522 break;
2523 case EXACTF: /* length of string is 1 */
2524 c = UCHARAT(++opnd);
2525 while (scan < loceol &&
22c35a8c 2526 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c]))
bbce6d69 2527 scan++;
2528 break;
2529 case EXACTFL: /* length of string is 1 */
3280af22 2530 PL_reg_flags |= RF_tainted;
bbce6d69 2531 c = UCHARAT(++opnd);
2532 while (scan < loceol &&
22c35a8c 2533 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c]))
a0d0e21e 2534 scan++;
2535 break;
a0ed51b3 2536 case ANYOFUTF8:
2537 loceol = PL_regeol;
2538 while (scan < loceol && REGINCLASSUTF8(p, (U8*)scan)) {
2539 scan += UTF8SKIP(scan);
2540 hardcount++;
2541 }
2542 break;
a0d0e21e 2543 case ANYOF:
ae5c130c 2544 while (scan < loceol && REGINCLASS(opnd, *scan))
a0d0e21e 2545 scan++;
a0d0e21e 2546 break;
2547 case ALNUM:
2548 while (scan < loceol && isALNUM(*scan))
2549 scan++;
2550 break;
a0ed51b3 2551 case ALNUMUTF8:
2552 loceol = PL_regeol;
dfe13c55 2553 while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) {
a0ed51b3 2554 scan += UTF8SKIP(scan);
2555 hardcount++;
2556 }
2557 break;
bbce6d69 2558 case ALNUML:
3280af22 2559 PL_reg_flags |= RF_tainted;
bbce6d69 2560 while (scan < loceol && isALNUM_LC(*scan))
2561 scan++;
2562 break;
a0ed51b3 2563 case ALNUMLUTF8:
2564 PL_reg_flags |= RF_tainted;
2565 loceol = PL_regeol;
dfe13c55 2566 while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) {
a0ed51b3 2567 scan += UTF8SKIP(scan);
2568 hardcount++;
2569 }
2570 break;
2571 break;
a0d0e21e 2572 case NALNUM:
2573 while (scan < loceol && !isALNUM(*scan))
2574 scan++;
2575 break;
a0ed51b3 2576 case NALNUMUTF8:
2577 loceol = PL_regeol;
dfe13c55 2578 while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) {
a0ed51b3 2579 scan += UTF8SKIP(scan);
2580 hardcount++;
2581 }
2582 break;
bbce6d69 2583 case NALNUML:
3280af22 2584 PL_reg_flags |= RF_tainted;
bbce6d69 2585 while (scan < loceol && !isALNUM_LC(*scan))
2586 scan++;
2587 break;
a0ed51b3 2588 case NALNUMLUTF8:
2589 PL_reg_flags |= RF_tainted;
2590 loceol = PL_regeol;
dfe13c55 2591 while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) {
a0ed51b3 2592 scan += UTF8SKIP(scan);
2593 hardcount++;
2594 }
2595 break;
a0d0e21e 2596 case SPACE:
2597 while (scan < loceol && isSPACE(*scan))
2598 scan++;
2599 break;
a0ed51b3 2600 case SPACEUTF8:
2601 loceol = PL_regeol;
dfe13c55 2602 while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) {
a0ed51b3 2603 scan += UTF8SKIP(scan);
2604 hardcount++;
2605 }
2606 break;
bbce6d69 2607 case SPACEL:
3280af22 2608 PL_reg_flags |= RF_tainted;
bbce6d69 2609 while (scan < loceol && isSPACE_LC(*scan))
2610 scan++;
2611 break;
a0ed51b3 2612 case SPACELUTF8:
2613 PL_reg_flags |= RF_tainted;
2614 loceol = PL_regeol;
dfe13c55 2615 while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
a0ed51b3 2616 scan += UTF8SKIP(scan);
2617 hardcount++;
2618 }
2619 break;
a0d0e21e 2620 case NSPACE:
2621 while (scan < loceol && !isSPACE(*scan))
2622 scan++;
2623 break;
a0ed51b3 2624 case NSPACEUTF8:
2625 loceol = PL_regeol;
dfe13c55 2626 while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) {
a0ed51b3 2627 scan += UTF8SKIP(scan);
2628 hardcount++;
2629 }
2630 break;
bbce6d69 2631 case NSPACEL:
3280af22 2632 PL_reg_flags |= RF_tainted;
bbce6d69 2633 while (scan < loceol && !isSPACE_LC(*scan))
2634 scan++;
2635 break;
a0ed51b3 2636 case NSPACELUTF8:
2637 PL_reg_flags |= RF_tainted;
2638 loceol = PL_regeol;
dfe13c55 2639 while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
a0ed51b3 2640 scan += UTF8SKIP(scan);
2641 hardcount++;
2642 }
2643 break;
a0d0e21e 2644 case DIGIT:
2645 while (scan < loceol && isDIGIT(*scan))
2646 scan++;
2647 break;
a0ed51b3 2648 case DIGITUTF8:
2649 loceol = PL_regeol;
dfe13c55 2650 while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) {
a0ed51b3 2651 scan += UTF8SKIP(scan);
2652 hardcount++;
2653 }
2654 break;
2655 break;
a0d0e21e 2656 case NDIGIT:
2657 while (scan < loceol && !isDIGIT(*scan))
2658 scan++;
2659 break;
a0ed51b3 2660 case NDIGITUTF8:
2661 loceol = PL_regeol;
dfe13c55 2662 while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) {
a0ed51b3 2663 scan += UTF8SKIP(scan);
2664 hardcount++;
2665 }
2666 break;
a0d0e21e 2667 default: /* Called on something of 0 width. */
2668 break; /* So match right here or not at all. */
2669 }
a687059c 2670
a0ed51b3 2671 if (hardcount)
2672 c = hardcount;
2673 else
2674 c = scan - PL_reginput;
3280af22 2675 PL_reginput = scan;
a687059c 2676
c277df42 2677 DEBUG_r(
2678 {
2679 SV *prop = sv_newmortal();
2680
2681 regprop(prop, p);
2682 PerlIO_printf(Perl_debug_log,
2683 "%*s %s can match %ld times out of %ld...\n",
2684 REPORT_CODE_OFF+1, "", SvPVX(prop),c,max);
2685 });
2686
a0d0e21e 2687 return(c);
a687059c 2688}
2689
2690/*
c277df42 2691 - regrepeat_hard - repeatedly match something, report total lenth and length
2692 *
2693 * The repeater is supposed to have constant length.
2694 */
2695
76e3520e 2696STATIC I32
c277df42 2697regrepeat_hard(regnode *p, I32 max, I32 *lp)
2698{
5c0ca799 2699 dTHR;
c277df42 2700 register char *scan;
2701 register char *start;
3280af22 2702 register char *loceol = PL_regeol;
a0ed51b3 2703 I32 l = 0;
708e3b05 2704 I32 count = 0, res = 1;
a0ed51b3 2705
2706 if (!max)
2707 return 0;
c277df42 2708
3280af22 2709 start = PL_reginput;
a0ed51b3 2710 if (UTF) {
708e3b05 2711 while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
a0ed51b3 2712 if (!count++) {
2713 l = 0;
2714 while (start < PL_reginput) {
2715 l++;
2716 start += UTF8SKIP(start);
2717 }
2718 *lp = l;
2719 if (l == 0)
2720 return max;
2721 }
2722 if (count == max)
2723 return count;
2724 }
2725 }
2726 else {
708e3b05 2727 while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
a0ed51b3 2728 if (!count++) {
2729 *lp = l = PL_reginput - start;
2730 if (max != REG_INFTY && l*max < loceol - scan)
2731 loceol = scan + l*max;
2732 if (l == 0)
2733 return max;
c277df42 2734 }
2735 }
2736 }
708e3b05 2737 if (!res)
3280af22 2738 PL_reginput = scan;
c277df42 2739
a0ed51b3 2740 return count;
c277df42 2741}
2742
2743/*
cb8d8820 2744 - reginclass - determine if a character falls into a character class
bbce6d69 2745 */
2746
76e3520e 2747STATIC bool
8ac85365 2748reginclass(register char *p, register I32 c)
bbce6d69 2749{
5c0ca799 2750 dTHR;
bbce6d69 2751 char flags = *p;
2752 bool match = FALSE;
2753
2754 c &= 0xFF;
ae5c130c 2755 if (ANYOF_TEST(p, c))
bbce6d69 2756 match = TRUE;
2757 else if (flags & ANYOF_FOLD) {
2758 I32 cf;
2759 if (flags & ANYOF_LOCALE) {
3280af22 2760 PL_reg_flags |= RF_tainted;
22c35a8c 2761 cf = PL_fold_locale[c];
bbce6d69 2762 }
2763 else
22c35a8c 2764 cf = PL_fold[c];
ae5c130c 2765 if (ANYOF_TEST(p, cf))
bbce6d69 2766 match = TRUE;
2767 }
2768
2769 if (!match && (flags & ANYOF_ISA)) {
3280af22 2770 PL_reg_flags |= RF_tainted;
bbce6d69 2771
2772 if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) ||
2773 ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) ||
2774 ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) ||
2775 ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c)))
2776 {
2777 match = TRUE;
2778 }
2779 }
2780
ae5c130c 2781 return (flags & ANYOF_INVERT) ? !match : match;
bbce6d69 2782}
2783
a0ed51b3 2784STATIC bool
2785reginclassutf8(regnode *f, U8 *p)
c485e607 2786{
2787 dTHR;
a0ed51b3 2788 char flags = ARG1(f);
2789 bool match = FALSE;
2790 SV *sv = (SV*)PL_regdata->data[ARG2(f)];
2791
2792 if (swash_fetch(sv, p))
2793 match = TRUE;
2794 else if (flags & ANYOF_FOLD) {
2795 I32 cf;
dfe13c55 2796 U8 tmpbuf[10];
a0ed51b3 2797 if (flags & ANYOF_LOCALE) {
2798 PL_reg_flags |= RF_tainted;
2799 uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p));
2800 }
2801 else
2802 uv_to_utf8(tmpbuf, toLOWER_utf8(p));
2803 if (swash_fetch(sv, tmpbuf))
2804 match = TRUE;
2805 }
2806
2807 if (!match && (flags & ANYOF_ISA)) {
2808 PL_reg_flags |= RF_tainted;
2809
2810 if (((flags & ANYOF_ALNUML) && isALNUM_LC_utf8(p)) ||
2811 ((flags & ANYOF_NALNUML) && !isALNUM_LC_utf8(p)) ||
2812 ((flags & ANYOF_SPACEL) && isSPACE_LC_utf8(p)) ||
2813 ((flags & ANYOF_NSPACEL) && !isSPACE_LC_utf8(p)))
2814 {
2815 match = TRUE;
2816 }
2817 }
2818
2819 return (flags & ANYOF_INVERT) ? !match : match;
2820}
161b471a 2821
dfe13c55 2822STATIC U8 *
2823reghop(U8 *s, I32 off)
c485e607 2824{
2825 dTHR;
a0ed51b3 2826 if (off >= 0) {
2827 while (off-- && s < (U8*)PL_regeol)
2828 s += UTF8SKIP(s);
2829 }
2830 else {
2831 while (off++) {
2832 if (s > (U8*)PL_bostr) {
2833 s--;
2834 if (*s & 0x80) {
2835 while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80)
2836 s--;
2837 } /* XXX could check well-formedness here */
2838 }
2839 }
2840 }
2841 return s;
2842}
161b471a 2843
dfe13c55 2844STATIC U8 *
2845reghopmaybe(U8* s, I32 off)
a0ed51b3 2846{
c485e607 2847 dTHR;
a0ed51b3 2848 if (off >= 0) {
2849 while (off-- && s < (U8*)PL_regeol)
2850 s += UTF8SKIP(s);
2851 if (off >= 0)
2852 return 0;
2853 }
2854 else {
2855 while (off++) {
2856 if (s > (U8*)PL_bostr) {
2857 s--;
2858 if (*s & 0x80) {
2859 while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80)
2860 s--;
2861 } /* XXX could check well-formedness here */
2862 }
2863 else
2864 break;
2865 }
2866 if (off <= 0)
2867 return 0;
2868 }
2869 return s;
2870}
Unnamed repository; edit this file to name it for gitweb.