This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
sample checksum code in "perlfunc" man page is wrong
[perl5.git] / regexec.c
CommitLineData
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
b9d5759e
AD
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
IZ
35# define Perl_regexec_flags my_regexec
36# define Perl_regdump my_regdump
37# define Perl_regprop my_regprop
d06ea78c
GS
38/* *These* symbols are masked to allow static link. */
39# define Perl_pregexec my_pregexec
d88dccdf 40# define Perl_reginitcolors my_reginitcolors
56953603
IZ
41#endif
42
f0fcb552 43/*SUPPRESS 112*/
a687059c 44/*
e50aee73 45 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
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
LW
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
LW
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
LW
78#include "regcomp.h"
79
c277df42
IZ
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
LW
83#define RF_utf8 8 /* String contains multibyte chars? */
84
85#define UTF (PL_reg_flags & RF_utf8)
ce862d02
IZ
86
87#define RS_init 1 /* eval environment created */
88#define RS_set 2 /* replsv value is set */
c277df42 89
a687059c
LW
90#ifndef STATIC
91#define STATIC static
92#endif
93
76e3520e 94#ifndef PERL_OBJECT
a0d0e21e
LW
95typedef I32 CHECKPOINT;
96
c277df42
IZ
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
PP
108static CHECKPOINT regcppush _((I32 parenfloor));
109static char * regcppop _((void));
942e002e
GS
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
LW
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
GS
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
NIS
136 int retval = PL_savestack_ix;
137 int i = (PL_regsize - parenfloor) * 4;
a0d0e21e
LW
138 int p;
139
140 SSCHECK(i + 5);
3280af22
NIS
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
LW
145 SSPUSHINT(p);
146 }
3280af22
NIS
147 SSPUSHINT(PL_regsize);
148 SSPUSHINT(*PL_reglastparen);
149 SSPUSHPTR(PL_reginput);
a0d0e21e
LW
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
DD
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
DD
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
LW
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
NIS
176 *PL_reglastparen = SSPOPINT;
177 PL_regsize = SSPOPINT;
c277df42 178 for (i -= 3; i > 0; i -= 4) {
a0d0e21e 179 paren = (U32)SSPOPINT;
3280af22
NIS
180 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
181 PL_regstartp[paren] = (char *) SSPOPPTR;
a0d0e21e 182 tmps = (char*)SSPOPPTR;
3280af22
NIS
183 if (paren <= *PL_reglastparen)
184 PL_regendp[paren] = tmps;
c277df42 185 DEBUG_r(
c3464db5
DD
186 PerlIO_printf(Perl_debug_log,
187 " restoring \\%d to %d(%d)..%d%s\n",
3280af22
NIS
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
DD
196 PerlIO_printf(Perl_debug_log,
197 " restoring \\%d..\\%d to undef\n",
3280af22 198 *PL_reglastparen + 1, PL_regnpar);
c277df42
IZ
199 }
200 );
3280af22
NIS
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
LW
205 }
206 return input;
207}
208
0f5d15d6
IZ
209STATIC char *
210regcp_set_to(I32 ss)
211{
46124e9e 212 dTHR;
0f5d15d6
IZ
213 I32 tmp = PL_savestack_ix;
214
215 PL_savestack_ix = ss;
216 regcppop();
217 PL_savestack_ix = tmp;
942e002e 218 return Nullch;
0f5d15d6
IZ
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
LW
234 */
235
236/*
c277df42 237 - pregexec - match a regexp against a string
a687059c 238 */
c277df42 239I32
c3464db5
DD
240pregexec(register regexp *prog, char *stringarg, register char *strend,
241 char *strbeg, I32 minend, SV *screamer, U32 nosave)
c277df42
IZ
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
IZ
251
252STATIC void
253cache_re(regexp *prog)
254{
46124e9e 255 dTHR;
0f5d15d6
IZ
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
IZ
265STATIC void
266restore_pos(void *arg)
35ef4773
GS
267{
268 dTHR;
9661b544
IZ
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
IZ
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
IZ
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
LW
292 register char *startpos = stringarg;
293 register I32 tmp;
c277df42 294 I32 minlen; /* must match at least this many chars */
a0d0e21e
LW
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
LW
298 constant substr. */ /* CC */
299 I32 end_shift = 0; /* Same for the end. */ /* CC */
c277df42
IZ
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
LW
311#endif
312
313 /* Be paranoid... */
314 if (prog == NULL || startpos == NULL) {
315 croak("NULL regexp parameter");
316 return 0;
317 }
318
c277df42
IZ
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
NIS
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) {
a0d0e21e
LW
332 FAIL("corrupted regexp program");
333 }
334
3280af22
NIS
335 PL_reg_flags = 0;
336 PL_reg_eval_set = 0;
a0d0e21e 337
a0ed51b3
LW
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
LW
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
IZ
352 /* We start without call_cc context. */
353 PL_reg_call_cc = 0;
354
a0d0e21e
LW
355 /* If there is a "must appear" string, look for it. */
356 s = startpos;
c277df42
IZ
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
LW
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
IZ
370 start_shift + (stringarg - strbeg),
371 end_shift, &scream_pos, 0);
a0d0e21e
LW
372 else
373 s = Nullch;
c277df42 374 scream_olds = s;
0a12ae7d 375 }
a0d0e21e 376 else
c277df42
IZ
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
LW
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
LW
387 : (t = s - prog->check_offset_max) != 0
388 )
389 )
390 {
c277df42 391 ++BmUSEFUL(prog->check_substr); /* hooray/2 */
a0ed51b3
LW
392 s = t;
393 }
394 else if (!(prog->reganch & ROPT_NAUGHTY)
c277df42
IZ
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
LW
401 }
402 else
403 s = startpos;
a0d0e21e 404 }
a687059c 405
35ef4773
GS
406 DEBUG_r(if (!PL_colorset) reginitcolors());
407 DEBUG_r(PerlIO_printf(Perl_debug_log,
8d300b32
GS
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
IZ
416 (strend - startpos > 60 ? "..." : ""))
417 );
418
22e551b9
IZ
419 if (prog->reganch & ROPT_GPOS_SEEN) {
420 MAGIC *mg;
22e551b9 421
ad94a511
IZ
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
IZ
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
LW
432 if (regtry(prog, startpos))
433 goto got_it;
22e551b9
IZ
434 else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
435 || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
774d564b 436 {
a0d0e21e
LW
437 if (minlen)
438 dontbother = minlen - 1;
dfe13c55 439 strend = HOPc(strend, -dontbother);
a0d0e21e
LW
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
LW
445 if (s < strend && regtry(prog, s))
446 goto got_it;
447 }
35c8bce7 448 }
35c8bce7 449 }
a0d0e21e 450 goto phooey;
22e551b9
IZ
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
IZ
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
LW
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
LW
478 while (s < strend && *s == ch)
479 s++;
480 }
481 s++;
a0d0e21e 482 }
a687059c 483 }
c277df42
IZ
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
LW
497 char *last1; /* Last position checked before */
498
499 if (s > PL_bostr)
dfe13c55 500 last1 = HOPc(s, -1);
a0ed51b3
LW
501 else
502 last1 = s - 1; /* bogus */
c277df42
IZ
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
IZ
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
GS
515 if (HOPc(s, -back_max) > last1) {
516 last1 = HOPc(s, -back_min);
517 s = HOPc(s, -back_max);
a0ed51b3
LW
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
LW
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
LW
538 }
539 }
540 goto phooey;
a0ed51b3
LW
541 }
542 else if (c = prog->regstclass) {
a0d0e21e 543 I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
a0ed51b3 544 char *cc;
a687059c 545
a0d0e21e
LW
546 if (minlen)
547 dontbother = minlen - 1;
dfe13c55 548 strend = HOPc(strend, -dontbother); /* don't bother with what can't match */
a0d0e21e
LW
549 tmp = 1;
550 /* We know what class it must start with. */
551 switch (OP(c)) {
a0ed51b3
LW
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
LW
570 if (tmp && regtry(prog, s))
571 goto got_it;
572 else
573 tmp = doevery;
a687059c 574 }
a0d0e21e
LW
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
LW
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
LW
592 tmp = !tmp;
593 if (regtry(prog, s))
594 goto got_it;
a687059c 595 }
a0d0e21e
LW
596 s++;
597 }
598 if ((minlen || tmp) && regtry(prog,s))
599 goto got_it;
600 break;
a0ed51b3
LW
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
LW
610 tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
611 while (s < strend) {
dfe13c55
GS
612 if (tmp == !(OP(c) == BOUND ?
613 swash_fetch(PL_utf8_alnum, (U8*)s) :
614 isALNUM_LC_utf8((U8*)s)))
615 {
a0ed51b3
LW
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
LW
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
LW
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
LW
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
LW
654 tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
655 while (s < strend) {
dfe13c55
GS
656 if (tmp == !(OP(c) == NBOUND ?
657 swash_fetch(PL_utf8_alnum, (U8*)s) :
658 isALNUM_LC_utf8((U8*)s)))
a0ed51b3
LW
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
LW
667 case ALNUM:
668 while (s < strend) {
bbce6d69
PP
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
LW
680 case ALNUMUTF8:
681 while (s < strend) {
dfe13c55 682 if (swash_fetch(PL_utf8_alnum, (U8*)s)) {
a0ed51b3
LW
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
PP
695 while (s < strend) {
696 if (isALNUM_LC(*s)) {
a0d0e21e
LW
697 if (tmp && regtry(prog, s))
698 goto got_it;
a687059c 699 else
a0d0e21e
LW
700 tmp = doevery;
701 }
702 else
703 tmp = 1;
704 s++;
705 }
706 break;
a0ed51b3
LW
707 case ALNUMLUTF8:
708 PL_reg_flags |= RF_tainted;
709 while (s < strend) {
dfe13c55 710 if (isALNUM_LC_utf8((U8*)s)) {
a0ed51b3
LW
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
LW
721 case NALNUM:
722 while (s < strend) {
bbce6d69
PP
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
LW
734 case NALNUMUTF8:
735 while (s < strend) {
dfe13c55 736 if (!swash_fetch(PL_utf8_alnum, (U8*)s)) {
a0ed51b3
LW
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
PP
749 while (s < strend) {
750 if (!isALNUM_LC(*s)) {
a0d0e21e
LW
751 if (tmp && regtry(prog, s))
752 goto got_it;
a687059c 753 else
a0d0e21e 754 tmp = doevery;
a687059c 755 }
a0d0e21e
LW
756 else
757 tmp = 1;
758 s++;
759 }
760 break;
a0ed51b3
LW
761 case NALNUMLUTF8:
762 PL_reg_flags |= RF_tainted;
763 while (s < strend) {
dfe13c55 764 if (!isALNUM_LC_utf8((U8*)s)) {
a0ed51b3
LW
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
LW
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
LW
783 else
784 tmp = 1;
785 s++;
786 }
787 break;
a0ed51b3
LW
788 case SPACEUTF8:
789 while (s < strend) {
dfe13c55 790 if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) {
a0ed51b3
LW
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
PP
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
LW
815 case SPACELUTF8:
816 PL_reg_flags |= RF_tainted;
817 while (s < strend) {
dfe13c55 818 if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
a0ed51b3
LW
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
LW
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
LW
837 else
838 tmp = 1;
839 s++;
840 }
841 break;
a0ed51b3
LW
842 case NSPACEUTF8:
843 while (s < strend) {
dfe13c55 844 if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) {
a0ed51b3
LW
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
PP
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
LW
869 case NSPACELUTF8:
870 PL_reg_flags |= RF_tainted;
871 while (s < strend) {
dfe13c55 872 if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
a0ed51b3
LW
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
LW
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
LW
891 else
892 tmp = 1;
893 s++;
894 }
895 break;
a0ed51b3
LW
896 case DIGITUTF8:
897 while (s < strend) {
dfe13c55 898 if (swash_fetch(PL_utf8_digit,(U8*)s)) {
a0ed51b3
LW
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
LW
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
LW
917 else
918 tmp = 1;
919 s++;
920 }
921 break;
a0ed51b3
LW
922 case NDIGITUTF8:
923 while (s < strend) {
dfe13c55 924 if (!swash_fetch(PL_utf8_digit,(U8*)s)) {
a0ed51b3
LW
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
LW
936 }
937 else {
c277df42
IZ
938 dontbother = 0;
939 if (prog->float_substr != Nullsv) { /* Trim the end. */
940 char *last;
941 I32 oldpos = scream_pos;
942
22e551b9
IZ
943 if (flags & REXEC_SCREAM) {
944 last = screaminstr(sv, prog->float_substr, s - strbeg,
c277df42
IZ
945 end_shift, &scream_pos, 1); /* last one */
946 if (!last) {
947 last = scream_olds; /* Only one occurence. */
948 }
a0ed51b3
LW
949 }
950 else {
c277df42
IZ
951 STRLEN len;
952 char *little = SvPV(prog->float_substr, len);
19b4f81a
JPC
953 if (len)
954 last = rninstr(s, strend, little, little + len);
955 else
956 last = strend; /* matching `$' */
c277df42
IZ
957 }
958 if (last == NULL) goto phooey; /* Should not happen! */
19b4f81a 959 dontbother = strend - last + prog->float_min_offset;
c277df42
IZ
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
LW
965 if (UTF) {
966 for (;;) {
84df6dba 967 if (regtry(prog, s))
a0ed51b3 968 goto got_it;
a0ed51b3
LW
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
LW
980 }
981
982 /* Failure. */
983 goto phooey;
a687059c 984
a0d0e21e
LW
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
PP
989
990 /* make sure $`, $&, $', and $digit will work later */
c277df42
IZ
991 if (strbeg != prog->subbase) { /* second+ //g match. */
992 if (!(flags & REXEC_COPY_STR)) {
137443ea
PP
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
PP
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
LW
1011 }
1012 }
a0d0e21e 1013 }
5c5e4c24
IZ
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
IZ
1021 }
1022
a0d0e21e
LW
1023 return 1;
1024
1025phooey:
9661b544
IZ
1026 if (PL_reg_eval_set)
1027 restore_pos(0);
a0d0e21e 1028 return 0;
a687059c
LW
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
LW
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
IZ
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
IZ
1050 ));
1051 SAVEINT(cxstack[cxstack_ix].blk_oldsp);
3280af22 1052 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
ce862d02
IZ
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
IZ
1058
1059 if (PL_reg_sv) {
1060 /* Make $_ available to executed code. */
127ad2b7
GS
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
IZ
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
IZ
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
NIS
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
NIS
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
IZ
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
LW
1105 sp = prog->startp;
1106 ep = prog->endp;
1107 if (prog->nparens) {
5c5e4c24
IZ
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
LW
1116 return 1;
1117 }
c277df42
IZ
1118 REGCP_UNWIND;
1119 return 0;
a687059c
LW
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
IZ
1139 dTHR;
1140 register regnode *scan; /* Current node. */
1141 regnode *next; /* Next node. */
1142 regnode *inner; /* Next node in internal branch. */
c3464db5
DD
1143 register I32 nextchr; /* renamed nextchr - nextchar colides with
1144 function of same name */
a0d0e21e
LW
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
IZ
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
LW
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
IZ
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
IZ
1166# define sayYES return 1
1167# define sayNO return 0
1168# define saySAME(x) return x
a687059c 1169#endif
c277df42
IZ
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
NIS
1174 int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput);
1175 int pref_len = (locinput - PL_bostr > (5 + taill) - l
1176 ? (5 + taill) - l : locinput - PL_bostr);
364723c2 1177 int pref0_len = pref_len - (locinput - PL_reg_starttry);
c277df42 1178
3280af22
NIS
1179 if (l + pref_len < (5 + taill) && l < PL_regeol - locinput)
1180 l = ( PL_regeol - locinput > (5 + taill) - pref_len
1181 ? (5 + taill) - pref_len : PL_regeol - locinput);
8d300b32
GS
1182 if (pref0_len < 0)
1183 pref0_len = 0;
c277df42
IZ
1184 regprop(prop, scan);
1185 PerlIO_printf(Perl_debug_log,
8d300b32 1186 "%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n",
3280af22 1187 locinput - PL_bostr,
8d300b32
GS
1188 PL_colors[4], pref0_len,
1189 locinput - pref_len, PL_colors[5],
1190 PL_colors[2], pref_len - pref0_len,
1191 locinput - pref_len + pref0_len, PL_colors[3],
c277df42 1192 (docolor ? "" : "> <"),
3280af22 1193 PL_colors[0], l, locinput, PL_colors[1],
c277df42
IZ
1194 15 - l - pref_len + 1,
1195 "",
3280af22 1196 scan - PL_regprogram, PL_regindent*2, "",
c277df42
IZ
1197 SvPVX(prop));
1198 } );
a687059c 1199
c277df42 1200 next = scan + NEXT_OFF(scan);
a0d0e21e
LW
1201 if (next == scan)
1202 next = NULL;
a687059c 1203
a0d0e21e
LW
1204 switch (OP(scan)) {
1205 case BOL:
3280af22
NIS
1206 if (locinput == PL_bostr
1207 ? PL_regprev == '\n'
1208 : (PL_multiline &&
1209 (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
a0d0e21e 1210 {
a0ed51b3 1211 /* regtill = regbol; */
a0d0e21e
LW
1212 break;
1213 }
4633a7c4 1214 sayNO;
a0d0e21e 1215 case MBOL:
3280af22
NIS
1216 if (locinput == PL_bostr
1217 ? PL_regprev == '\n'
1218 : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
a0d0e21e
LW
1219 {
1220 break;
1221 }
4633a7c4 1222 sayNO;
a0d0e21e 1223 case SBOL:
3280af22 1224 if (locinput == PL_regbol && PL_regprev == '\n')
a0d0e21e 1225 break;
4633a7c4 1226 sayNO;
774d564b 1227 case GPOS:
22e551b9 1228 if (locinput == PL_reg_ganch)
a0d0e21e 1229 break;
4633a7c4 1230 sayNO;
a0d0e21e 1231 case EOL:
3280af22 1232 if (PL_multiline)
a0d0e21e
LW
1233 goto meol;
1234 else
1235 goto seol;
1236 case MEOL:
1237 meol:
3280af22 1238 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
4633a7c4 1239 sayNO;
a0d0e21e
LW
1240 break;
1241 case SEOL:
1242 seol:
3280af22 1243 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
4633a7c4 1244 sayNO;
3280af22 1245 if (PL_regeol - locinput > 1)
4633a7c4 1246 sayNO;
a0d0e21e 1247 break;
b85d18e9 1248 case EOS:
3280af22 1249 if (PL_regeol != locinput)
b85d18e9
IZ
1250 sayNO;
1251 break;
a0ed51b3
LW
1252 case SANYUTF8:
1253 if (nextchr & 0x80) {
6f06b55f 1254 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1255 if (locinput > PL_regeol)
1256 sayNO;
1257 nextchr = UCHARAT(locinput);
1258 break;
1259 }
1260 if (!nextchr && locinput >= PL_regeol)
1261 sayNO;
1262 nextchr = UCHARAT(++locinput);
1263 break;
a0d0e21e 1264 case SANY:
3280af22 1265 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1266 sayNO;
76e3520e 1267 nextchr = UCHARAT(++locinput);
a0d0e21e 1268 break;
a0ed51b3
LW
1269 case ANYUTF8:
1270 if (nextchr & 0x80) {
6f06b55f 1271 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1272 if (locinput > PL_regeol)
1273 sayNO;
1274 nextchr = UCHARAT(locinput);
1275 break;
1276 }
1277 if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
1278 sayNO;
1279 nextchr = UCHARAT(++locinput);
1280 break;
22c35a8c 1281 case REG_ANY:
3280af22 1282 if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
4633a7c4 1283 sayNO;
76e3520e 1284 nextchr = UCHARAT(++locinput);
a0d0e21e 1285 break;
bbce6d69 1286 case EXACT:
161b471a 1287 s = (char *) OPERAND(scan);
c277df42 1288 ln = UCHARAT(s++);
a0d0e21e 1289 /* Inline the first character, for speed. */
76e3520e 1290 if (UCHARAT(s) != nextchr)
4633a7c4 1291 sayNO;
3280af22 1292 if (PL_regeol - locinput < ln)
4633a7c4 1293 sayNO;
36477c24 1294 if (ln > 1 && memNE(s, locinput, ln))
4633a7c4 1295 sayNO;
a0d0e21e 1296 locinput += ln;
76e3520e 1297 nextchr = UCHARAT(locinput);
bbce6d69
PP
1298 break;
1299 case EXACTFL:
3280af22 1300 PL_reg_flags |= RF_tainted;
bbce6d69
PP
1301 /* FALL THROUGH */
1302 case EXACTF:
161b471a 1303 s = (char *) OPERAND(scan);
c277df42 1304 ln = UCHARAT(s++);
a0ed51b3
LW
1305
1306 if (UTF) {
1307 char *l = locinput;
1308 char *e = s + ln;
1309 c1 = OP(scan) == EXACTF;
1310 while (s < e) {
1311 if (l >= PL_regeol)
1312 sayNO;
dfe13c55
GS
1313 if (utf8_to_uv((U8*)s, 0) != (c1 ?
1314 toLOWER_utf8((U8*)l) :
1315 toLOWER_LC_utf8((U8*)l)))
1316 {
a0ed51b3 1317 sayNO;
dfe13c55 1318 }
a0ed51b3
LW
1319 s += UTF8SKIP(s);
1320 l += UTF8SKIP(l);
1321 }
1322 locinput = l;
1323 nextchr = UCHARAT(locinput);
1324 break;
1325 }
1326
bbce6d69 1327 /* Inline the first character, for speed. */
76e3520e 1328 if (UCHARAT(s) != nextchr &&
bbce6d69 1329 UCHARAT(s) != ((OP(scan) == EXACTF)
22c35a8c 1330 ? PL_fold : PL_fold_locale)[nextchr])
bbce6d69 1331 sayNO;
3280af22 1332 if (PL_regeol - locinput < ln)
bbce6d69 1333 sayNO;
5f05dabc
PP
1334 if (ln > 1 && (OP(scan) == EXACTF
1335 ? ibcmp(s, locinput, ln)
1336 : ibcmp_locale(s, locinput, ln)))
bbce6d69
PP
1337 sayNO;
1338 locinput += ln;
76e3520e 1339 nextchr = UCHARAT(locinput);
a0d0e21e 1340 break;
a0ed51b3
LW
1341 case ANYOFUTF8:
1342 s = (char *) OPERAND(scan);
1343 if (!REGINCLASSUTF8(scan, (U8*)locinput))
1344 sayNO;
1345 if (locinput >= PL_regeol)
1346 sayNO;
6f06b55f 1347 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1348 nextchr = UCHARAT(locinput);
1349 break;
a0d0e21e 1350 case ANYOF:
161b471a 1351 s = (char *) OPERAND(scan);
76e3520e
GS
1352 if (nextchr < 0)
1353 nextchr = UCHARAT(locinput);
873ef191 1354 if (!REGINCLASS(s, nextchr))
4633a7c4 1355 sayNO;
3280af22 1356 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1357 sayNO;
76e3520e 1358 nextchr = UCHARAT(++locinput);
a0d0e21e 1359 break;
bbce6d69 1360 case ALNUML:
3280af22 1361 PL_reg_flags |= RF_tainted;
bbce6d69 1362 /* FALL THROUGH */
a0d0e21e 1363 case ALNUM:
76e3520e 1364 if (!nextchr)
4633a7c4 1365 sayNO;
bbce6d69 1366 if (!(OP(scan) == ALNUM
76e3520e 1367 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
4633a7c4 1368 sayNO;
76e3520e 1369 nextchr = UCHARAT(++locinput);
a0d0e21e 1370 break;
a0ed51b3
LW
1371 case ALNUMLUTF8:
1372 PL_reg_flags |= RF_tainted;
1373 /* FALL THROUGH */
1374 case ALNUMUTF8:
1375 if (!nextchr)
1376 sayNO;
1377 if (nextchr & 0x80) {
1378 if (!(OP(scan) == ALNUMUTF8
dfe13c55
GS
1379 ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
1380 : isALNUM_LC_utf8((U8*)locinput)))
1381 {
a0ed51b3 1382 sayNO;
dfe13c55 1383 }
6f06b55f 1384 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1385 nextchr = UCHARAT(locinput);
1386 break;
1387 }
1388 if (!(OP(scan) == ALNUMUTF8
1389 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
1390 sayNO;
1391 nextchr = UCHARAT(++locinput);
1392 break;
bbce6d69 1393 case NALNUML:
3280af22 1394 PL_reg_flags |= RF_tainted;
bbce6d69 1395 /* FALL THROUGH */
a0d0e21e 1396 case NALNUM:
3280af22 1397 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1398 sayNO;
bbce6d69 1399 if (OP(scan) == NALNUM
76e3520e 1400 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
4633a7c4 1401 sayNO;
76e3520e 1402 nextchr = UCHARAT(++locinput);
a0d0e21e 1403 break;
a0ed51b3
LW
1404 case NALNUMLUTF8:
1405 PL_reg_flags |= RF_tainted;
1406 /* FALL THROUGH */
1407 case NALNUMUTF8:
1408 if (!nextchr && locinput >= PL_regeol)
1409 sayNO;
1410 if (nextchr & 0x80) {
1411 if (OP(scan) == NALNUMUTF8
dfe13c55
GS
1412 ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
1413 : isALNUM_LC_utf8((U8*)locinput))
1414 {
a0ed51b3 1415 sayNO;
dfe13c55 1416 }
6f06b55f 1417 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1418 nextchr = UCHARAT(locinput);
1419 break;
1420 }
1421 if (OP(scan) == NALNUMUTF8
1422 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
1423 sayNO;
1424 nextchr = UCHARAT(++locinput);
1425 break;
bbce6d69
PP
1426 case BOUNDL:
1427 case NBOUNDL:
3280af22 1428 PL_reg_flags |= RF_tainted;
bbce6d69 1429 /* FALL THROUGH */
a0d0e21e 1430 case BOUND:
bbce6d69
PP
1431 case NBOUND:
1432 /* was last char in word? */
3280af22 1433 ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev;
bbce6d69
PP
1434 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
1435 ln = isALNUM(ln);
76e3520e 1436 n = isALNUM(nextchr);
bbce6d69
PP
1437 }
1438 else {
1439 ln = isALNUM_LC(ln);
76e3520e 1440 n = isALNUM_LC(nextchr);
bbce6d69 1441 }
95bac841 1442 if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL))
4633a7c4 1443 sayNO;
a0d0e21e 1444 break;
a0ed51b3
LW
1445 case BOUNDLUTF8:
1446 case NBOUNDLUTF8:
1447 PL_reg_flags |= RF_tainted;
1448 /* FALL THROUGH */
1449 case BOUNDUTF8:
1450 case NBOUNDUTF8:
1451 /* was last char in word? */
dfe13c55
GS
1452 ln = (locinput != PL_regbol)
1453 ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
a0ed51b3
LW
1454 if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
1455 ln = isALNUM_uni(ln);
dfe13c55 1456 n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
a0ed51b3
LW
1457 }
1458 else {
1459 ln = isALNUM_LC_uni(ln);
dfe13c55 1460 n = isALNUM_LC_utf8((U8*)locinput);
a0ed51b3
LW
1461 }
1462 if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8))
1463 sayNO;
1464 break;
bbce6d69 1465 case SPACEL:
3280af22 1466 PL_reg_flags |= RF_tainted;
bbce6d69 1467 /* FALL THROUGH */
a0d0e21e 1468 case SPACE:
3280af22 1469 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1470 sayNO;
bbce6d69 1471 if (!(OP(scan) == SPACE
76e3520e 1472 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
4633a7c4 1473 sayNO;
76e3520e 1474 nextchr = UCHARAT(++locinput);
a0d0e21e 1475 break;
a0ed51b3
LW
1476 case SPACELUTF8:
1477 PL_reg_flags |= RF_tainted;
1478 /* FALL THROUGH */
1479 case SPACEUTF8:
1480 if (!nextchr && locinput >= PL_regeol)
1481 sayNO;
1482 if (nextchr & 0x80) {
1483 if (!(OP(scan) == SPACEUTF8
dfe13c55
GS
1484 ? swash_fetch(PL_utf8_space,(U8*)locinput)
1485 : isSPACE_LC_utf8((U8*)locinput)))
1486 {
a0ed51b3 1487 sayNO;
dfe13c55 1488 }
6f06b55f 1489 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1490 nextchr = UCHARAT(locinput);
1491 break;
1492 }
1493 if (!(OP(scan) == SPACEUTF8
1494 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
1495 sayNO;
1496 nextchr = UCHARAT(++locinput);
1497 break;
bbce6d69 1498 case NSPACEL:
3280af22 1499 PL_reg_flags |= RF_tainted;
bbce6d69 1500 /* FALL THROUGH */
a0d0e21e 1501 case NSPACE:
76e3520e 1502 if (!nextchr)
4633a7c4 1503 sayNO;
bbce6d69 1504 if (OP(scan) == SPACE
76e3520e 1505 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
4633a7c4 1506 sayNO;
76e3520e 1507 nextchr = UCHARAT(++locinput);
a0d0e21e 1508 break;
a0ed51b3
LW
1509 case NSPACELUTF8:
1510 PL_reg_flags |= RF_tainted;
1511 /* FALL THROUGH */
1512 case NSPACEUTF8:
1513 if (!nextchr)
1514 sayNO;
1515 if (nextchr & 0x80) {
1516 if (OP(scan) == NSPACEUTF8
dfe13c55
GS
1517 ? swash_fetch(PL_utf8_space,(U8*)locinput)
1518 : isSPACE_LC_utf8((U8*)locinput))
1519 {
a0ed51b3 1520 sayNO;
dfe13c55 1521 }
6f06b55f 1522 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1523 nextchr = UCHARAT(locinput);
1524 break;
1525 }
1526 if (OP(scan) == NSPACEUTF8
1527 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
1528 sayNO;
1529 nextchr = UCHARAT(++locinput);
1530 break;
a0d0e21e 1531 case DIGIT:
76e3520e 1532 if (!isDIGIT(nextchr))
4633a7c4 1533 sayNO;
76e3520e 1534 nextchr = UCHARAT(++locinput);
a0d0e21e 1535 break;
a0ed51b3
LW
1536 case DIGITUTF8:
1537 if (nextchr & 0x80) {
dfe13c55 1538 if (!(swash_fetch(PL_utf8_digit,(U8*)locinput)))
a0ed51b3 1539 sayNO;
6f06b55f 1540 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1541 nextchr = UCHARAT(locinput);
1542 break;
1543 }
1544 if (!isDIGIT(nextchr))
1545 sayNO;
1546 nextchr = UCHARAT(++locinput);
1547 break;
a0d0e21e 1548 case NDIGIT:
3280af22 1549 if (!nextchr && locinput >= PL_regeol)
4633a7c4 1550 sayNO;
76e3520e 1551 if (isDIGIT(nextchr))
4633a7c4 1552 sayNO;
76e3520e 1553 nextchr = UCHARAT(++locinput);
a0d0e21e 1554 break;
a0ed51b3
LW
1555 case NDIGITUTF8:
1556 if (!nextchr && locinput >= PL_regeol)
1557 sayNO;
1558 if (nextchr & 0x80) {
dfe13c55 1559 if (swash_fetch(PL_utf8_digit,(U8*)locinput))
a0ed51b3 1560 sayNO;
6f06b55f 1561 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
1562 nextchr = UCHARAT(locinput);
1563 break;
1564 }
1565 if (isDIGIT(nextchr))
1566 sayNO;
1567 nextchr = UCHARAT(++locinput);
1568 break;
1569 case CLUMP:
dfe13c55 1570 if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput))
a0ed51b3 1571 sayNO;
6f06b55f 1572 locinput += PL_utf8skip[nextchr];
dfe13c55 1573 while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput))
a0ed51b3
LW
1574 locinput += UTF8SKIP(locinput);
1575 if (locinput > PL_regeol)
1576 sayNO;
1577 nextchr = UCHARAT(locinput);
1578 break;
c8756f30 1579 case REFFL:
3280af22 1580 PL_reg_flags |= RF_tainted;
c8756f30 1581 /* FALL THROUGH */
c277df42 1582 case REF:
c8756f30 1583 case REFF:
c277df42 1584 n = ARG(scan); /* which paren pair */
3280af22
NIS
1585 s = PL_regstartp[n];
1586 if (*PL_reglastparen < n || !s)
af3f8c16 1587 sayNO; /* Do not match unless seen CLOSEn. */
3280af22 1588 if (s == PL_regendp[n])
a0d0e21e 1589 break;
a0ed51b3
LW
1590
1591 if (UTF && OP(scan) != REF) { /* REF can do byte comparison */
1592 char *l = locinput;
1593 char *e = PL_regendp[n];
1594 /*
1595 * Note that we can't do the "other character" lookup trick as
1596 * in the 8-bit case (no pun intended) because in Unicode we
1597 * have to map both upper and title case to lower case.
1598 */
1599 if (OP(scan) == REFF) {
1600 while (s < e) {
1601 if (l >= PL_regeol)
1602 sayNO;
dfe13c55 1603 if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l))
a0ed51b3
LW
1604 sayNO;
1605 s += UTF8SKIP(s);
1606 l += UTF8SKIP(l);
1607 }
1608 }
1609 else {
1610 while (s < e) {
1611 if (l >= PL_regeol)
1612 sayNO;
dfe13c55 1613 if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l))
a0ed51b3
LW
1614 sayNO;
1615 s += UTF8SKIP(s);
1616 l += UTF8SKIP(l);
1617 }
1618 }
1619 locinput = l;
1620 nextchr = UCHARAT(locinput);
1621 break;
1622 }
1623
a0d0e21e 1624 /* Inline the first character, for speed. */
76e3520e 1625 if (UCHARAT(s) != nextchr &&
c8756f30
AK
1626 (OP(scan) == REF ||
1627 (UCHARAT(s) != ((OP(scan) == REFF
22c35a8c 1628 ? PL_fold : PL_fold_locale)[nextchr]))))
4633a7c4 1629 sayNO;
3280af22
NIS
1630 ln = PL_regendp[n] - s;
1631 if (locinput + ln > PL_regeol)
4633a7c4 1632 sayNO;
c8756f30
AK
1633 if (ln > 1 && (OP(scan) == REF
1634 ? memNE(s, locinput, ln)
1635 : (OP(scan) == REFF
1636 ? ibcmp(s, locinput, ln)
1637 : ibcmp_locale(s, locinput, ln))))
4633a7c4 1638 sayNO;
a0d0e21e 1639 locinput += ln;
76e3520e 1640 nextchr = UCHARAT(locinput);
a0d0e21e
LW
1641 break;
1642
1643 case NOTHING:
c277df42 1644 case TAIL:
a0d0e21e
LW
1645 break;
1646 case BACK:
1647 break;
c277df42
IZ
1648 case EVAL:
1649 {
1650 dSP;
533c011a 1651 OP_4tree *oop = PL_op;
3280af22
NIS
1652 COP *ocurcop = PL_curcop;
1653 SV **ocurpad = PL_curpad;
c277df42
IZ
1654 SV *ret;
1655
1656 n = ARG(scan);
533c011a
NIS
1657 PL_op = (OP_4tree*)PL_regdata->data[n];
1658 DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) );
3280af22 1659 PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 1]);
9661b544 1660 PL_reg_magic->mg_len = locinput - PL_bostr;
5c5e4c24 1661 PL_regendp[0] = locinput;
c277df42 1662
76e3520e 1663 CALLRUNOPS(); /* Scalar context. */
c277df42
IZ
1664 SPAGAIN;
1665 ret = POPs;
1666 PUTBACK;
1667
0f5d15d6
IZ
1668 PL_op = oop;
1669 PL_curpad = ocurpad;
1670 PL_curcop = ocurcop;
c277df42 1671 if (logical) {
0f5d15d6
IZ
1672 if (logical == 2) { /* Postponed subexpression. */
1673 regexp *re;
22c35a8c 1674 MAGIC *mg = Null(MAGIC*);
0f5d15d6
IZ
1675 re_cc_state state;
1676 CURCUR cctmp;
1677 CHECKPOINT cp, lastcp;
1678
1679 if(SvROK(ret) || SvRMAGICAL(ret)) {
1680 SV *sv = SvROK(ret) ? SvRV(ret) : ret;
1681
1682 if(SvMAGICAL(sv))
1683 mg = mg_find(sv, 'r');
1684 }
1685 if (mg) {
1686 re = (regexp *)mg->mg_obj;
df0003d4 1687 (void)ReREFCNT_inc(re);
0f5d15d6
IZ
1688 }
1689 else {
1690 STRLEN len;
1691 char *t = SvPV(ret, len);
1692 PMOP pm;
1693 char *oprecomp = PL_regprecomp;
1694 I32 osize = PL_regsize;
1695 I32 onpar = PL_regnpar;
1696
1697 pm.op_pmflags = 0;
1698 re = CALLREGCOMP(t, t + len, &pm);
1699 if (!(SvFLAGS(ret)
1700 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY)))
1701 sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0);
1702 PL_regprecomp = oprecomp;
1703 PL_regsize = osize;
1704 PL_regnpar = onpar;
1705 }
1706 DEBUG_r(
1707 PerlIO_printf(Perl_debug_log,
1708 "Entering embedded `%s%.60s%s%s'\n",
1709 PL_colors[0],
1710 re->precomp,
1711 PL_colors[1],
1712 (strlen(re->precomp) > 60 ? "..." : ""))
1713 );
1714 state.node = next;
1715 state.prev = PL_reg_call_cc;
1716 state.cc = PL_regcc;
1717 state.re = PL_reg_re;
1718
1719 cctmp.cur = 0;
1720 cctmp.oldcc = 0;
1721 PL_regcc = &cctmp;
1722
1723 cp = regcppush(0); /* Save *all* the positions. */
1724 REGCP_SET;
1725 cache_re(re);
1726 state.ss = PL_savestack_ix;
1727 *PL_reglastparen = 0;
1728 PL_reg_call_cc = &state;
1729 PL_reginput = locinput;
1730 if (regmatch(re->program + 1)) {
1731 ReREFCNT_dec(re);
1732 regcpblow(cp);
1733 sayYES;
1734 }
1735 DEBUG_r(
1736 PerlIO_printf(Perl_debug_log,
1737 "%*s failed...\n",
1738 REPORT_CODE_OFF+PL_regindent*2, "")
1739 );
1740 ReREFCNT_dec(re);
1741 REGCP_UNWIND;
1742 regcppop();
1743 PL_reg_call_cc = state.prev;
1744 PL_regcc = state.cc;
1745 PL_reg_re = state.re;
d3790889 1746 cache_re(PL_reg_re);
0f5d15d6
IZ
1747 sayNO;
1748 }
c277df42 1749 sw = SvTRUE(ret);
0f5d15d6 1750 logical = 0;
a0ed51b3
LW
1751 }
1752 else
3280af22 1753 sv_setsv(save_scalar(PL_replgv), ret);
c277df42
IZ
1754 break;
1755 }
a0d0e21e 1756 case OPEN:
c277df42 1757 n = ARG(scan); /* which paren pair */
3280af22
NIS
1758 PL_reg_start_tmp[n] = locinput;
1759 if (n > PL_regsize)
1760 PL_regsize = n;
a0d0e21e
LW
1761 break;
1762 case CLOSE:
c277df42 1763 n = ARG(scan); /* which paren pair */
3280af22
NIS
1764 PL_regstartp[n] = PL_reg_start_tmp[n];
1765 PL_regendp[n] = locinput;
1766 if (n > *PL_reglastparen)
1767 *PL_reglastparen = n;
a0d0e21e 1768 break;
c277df42
IZ
1769 case GROUPP:
1770 n = ARG(scan); /* which paren pair */
3280af22 1771 sw = (*PL_reglastparen >= n && PL_regendp[n] != NULL);
c277df42
IZ
1772 break;
1773 case IFTHEN:
1774 if (sw)
1775 next = NEXTOPER(NEXTOPER(scan));
1776 else {
1777 next = scan + ARG(scan);
1778 if (OP(next) == IFTHEN) /* Fake one. */
1779 next = NEXTOPER(NEXTOPER(next));
1780 }
1781 break;
1782 case LOGICAL:
0f5d15d6 1783 logical = scan->flags;
c277df42 1784 break;
a0d0e21e
LW
1785 case CURLYX: {
1786 CURCUR cc;
3280af22 1787 CHECKPOINT cp = PL_savestack_ix;
c277df42
IZ
1788
1789 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
1790 next += ARG(next);
3280af22
NIS
1791 cc.oldcc = PL_regcc;
1792 PL_regcc = &cc;
1793 cc.parenfloor = *PL_reglastparen;
a0d0e21e
LW
1794 cc.cur = -1;
1795 cc.min = ARG1(scan);
1796 cc.max = ARG2(scan);
c277df42 1797 cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
a0d0e21e
LW
1798 cc.next = next;
1799 cc.minmod = minmod;
1800 cc.lastloc = 0;
3280af22 1801 PL_reginput = locinput;
a0d0e21e
LW
1802 n = regmatch(PREVOPER(next)); /* start on the WHILEM */
1803 regcpblow(cp);
3280af22 1804 PL_regcc = cc.oldcc;
4633a7c4 1805 saySAME(n);
a0d0e21e
LW
1806 }
1807 /* NOT REACHED */
1808 case WHILEM: {
1809 /*
1810 * This is really hard to understand, because after we match
1811 * what we're trying to match, we must make sure the rest of
1812 * the RE is going to match for sure, and to do that we have
1813 * to go back UP the parse tree by recursing ever deeper. And
1814 * if it fails, we have to reset our parent's current state
1815 * that we can try again after backing off.
1816 */
1817
c277df42 1818 CHECKPOINT cp, lastcp;
3280af22 1819 CURCUR* cc = PL_regcc;
c277df42
IZ
1820 char *lastloc = cc->lastloc; /* Detection of 0-len. */
1821
4633a7c4 1822 n = cc->cur + 1; /* how many we know we matched */
3280af22 1823 PL_reginput = locinput;
a0d0e21e 1824
c277df42
IZ
1825 DEBUG_r(
1826 PerlIO_printf(Perl_debug_log,
1827 "%*s %ld out of %ld..%ld cc=%lx\n",
3280af22 1828 REPORT_CODE_OFF+PL_regindent*2, "",
c277df42
IZ
1829 (long)n, (long)cc->min,
1830 (long)cc->max, (long)cc)
1831 );
4633a7c4 1832
a0d0e21e
LW
1833 /* If degenerate scan matches "", assume scan done. */
1834
579cf2c3 1835 if (locinput == cc->lastloc && n >= cc->min) {
3280af22
NIS
1836 PL_regcc = cc->oldcc;
1837 ln = PL_regcc->cur;
c277df42 1838 DEBUG_r(
c3464db5
DD
1839 PerlIO_printf(Perl_debug_log,
1840 "%*s empty match detected, try continuation...\n",
3280af22 1841 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1842 );
a0d0e21e 1843 if (regmatch(cc->next))
4633a7c4 1844 sayYES;
c277df42 1845 DEBUG_r(
c3464db5
DD
1846 PerlIO_printf(Perl_debug_log,
1847 "%*s failed...\n",
3280af22 1848 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1849 );
3280af22
NIS
1850 PL_regcc->cur = ln;
1851 PL_regcc = cc;
4633a7c4 1852 sayNO;
a0d0e21e
LW
1853 }
1854
1855 /* First just match a string of min scans. */
1856
1857 if (n < cc->min) {
1858 cc->cur = n;
1859 cc->lastloc = locinput;
4633a7c4
LW
1860 if (regmatch(cc->scan))
1861 sayYES;
1862 cc->cur = n - 1;
c277df42
IZ
1863 cc->lastloc = lastloc;
1864 DEBUG_r(
c3464db5
DD
1865 PerlIO_printf(Perl_debug_log,
1866 "%*s failed...\n",
3280af22 1867 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1868 );
4633a7c4 1869 sayNO;
a0d0e21e
LW
1870 }
1871
1872 /* Prefer next over scan for minimal matching. */
1873
1874 if (cc->minmod) {
3280af22
NIS
1875 PL_regcc = cc->oldcc;
1876 ln = PL_regcc->cur;
5f05dabc 1877 cp = regcppush(cc->parenfloor);
c277df42 1878 REGCP_SET;
5f05dabc 1879 if (regmatch(cc->next)) {
c277df42 1880 regcpblow(cp);
4633a7c4 1881 sayYES; /* All done. */
5f05dabc 1882 }
c277df42 1883 REGCP_UNWIND;
5f05dabc 1884 regcppop();
3280af22
NIS
1885 PL_regcc->cur = ln;
1886 PL_regcc = cc;
a0d0e21e 1887
c277df42 1888 if (n >= cc->max) { /* Maximum greed exceeded? */
599cee73 1889 if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
3280af22
NIS
1890 && !(PL_reg_flags & RF_warned)) {
1891 PL_reg_flags |= RF_warned;
599cee73 1892 warner(WARN_UNSAFE, "%s limit (%d) exceeded",
2f3ca594
GS
1893 "Complex regular subexpression recursion",
1894 REG_INFTY - 1);
c277df42 1895 }
4633a7c4 1896 sayNO;
c277df42 1897 }
a687059c 1898
c277df42 1899 DEBUG_r(
c3464db5
DD
1900 PerlIO_printf(Perl_debug_log,
1901 "%*s trying longer...\n",
3280af22 1902 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1903 );
a0d0e21e 1904 /* Try scanning more and see if it helps. */
3280af22 1905 PL_reginput = locinput;
a0d0e21e
LW
1906 cc->cur = n;
1907 cc->lastloc = locinput;
5f05dabc 1908 cp = regcppush(cc->parenfloor);
c277df42 1909 REGCP_SET;
5f05dabc 1910 if (regmatch(cc->scan)) {
c277df42 1911 regcpblow(cp);
4633a7c4 1912 sayYES;
5f05dabc 1913 }
c277df42 1914 DEBUG_r(
c3464db5
DD
1915 PerlIO_printf(Perl_debug_log,
1916 "%*s failed...\n",
3280af22 1917 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42
IZ
1918 );
1919 REGCP_UNWIND;
5f05dabc 1920 regcppop();
4633a7c4 1921 cc->cur = n - 1;
c277df42 1922 cc->lastloc = lastloc;
4633a7c4 1923 sayNO;
a0d0e21e
LW
1924 }
1925
1926 /* Prefer scan over next for maximal matching. */
1927
1928 if (n < cc->max) { /* More greed allowed? */
5f05dabc 1929 cp = regcppush(cc->parenfloor);
a0d0e21e
LW
1930 cc->cur = n;
1931 cc->lastloc = locinput;
c277df42 1932 REGCP_SET;
5f05dabc 1933 if (regmatch(cc->scan)) {
c277df42 1934 regcpblow(cp);
4633a7c4 1935 sayYES;
5f05dabc 1936 }
c277df42 1937 REGCP_UNWIND;
a0d0e21e 1938 regcppop(); /* Restore some previous $<digit>s? */
3280af22 1939 PL_reginput = locinput;
c277df42 1940 DEBUG_r(
c3464db5
DD
1941 PerlIO_printf(Perl_debug_log,
1942 "%*s failed, try continuation...\n",
3280af22 1943 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42
IZ
1944 );
1945 }
599cee73
PM
1946 if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
1947 && !(PL_reg_flags & RF_warned)) {
3280af22 1948 PL_reg_flags |= RF_warned;
599cee73 1949 warner(WARN_UNSAFE, "%s limit (%d) exceeded",
cb5d145d
GS
1950 "Complex regular subexpression recursion",
1951 REG_INFTY - 1);
a0d0e21e
LW
1952 }
1953
1954 /* Failed deeper matches of scan, so see if this one works. */
3280af22
NIS
1955 PL_regcc = cc->oldcc;
1956 ln = PL_regcc->cur;
a0d0e21e 1957 if (regmatch(cc->next))
4633a7c4 1958 sayYES;
c277df42 1959 DEBUG_r(
c3464db5 1960 PerlIO_printf(Perl_debug_log, "%*s failed...\n",
3280af22 1961 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 1962 );
3280af22
NIS
1963 PL_regcc->cur = ln;
1964 PL_regcc = cc;
4633a7c4 1965 cc->cur = n - 1;
c277df42 1966 cc->lastloc = lastloc;
4633a7c4 1967 sayNO;
a0d0e21e
LW
1968 }
1969 /* NOT REACHED */
c277df42
IZ
1970 case BRANCHJ:
1971 next = scan + ARG(scan);
1972 if (next == scan)
1973 next = NULL;
1974 inner = NEXTOPER(NEXTOPER(scan));
1975 goto do_branch;
1976 case BRANCH:
1977 inner = NEXTOPER(scan);
1978 do_branch:
1979 {
1980 CHECKPOINT lastcp;
1981 c1 = OP(scan);
1982 if (OP(next) != c1) /* No choice. */
1983 next = inner; /* Avoid recursion. */
a0d0e21e 1984 else {
3280af22 1985 int lastparen = *PL_reglastparen;
c277df42
IZ
1986
1987 REGCP_SET;
a0d0e21e 1988 do {
3280af22 1989 PL_reginput = locinput;
c277df42 1990 if (regmatch(inner))
4633a7c4 1991 sayYES;
c277df42 1992 REGCP_UNWIND;
3280af22
NIS
1993 for (n = *PL_reglastparen; n > lastparen; n--)
1994 PL_regendp[n] = 0;
1995 *PL_reglastparen = n;
c277df42 1996 scan = next;
a0d0e21e 1997 /*SUPPRESS 560*/
c277df42
IZ
1998 if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next)))
1999 next += n;
a0d0e21e 2000 else
c277df42 2001 next = NULL;
c277df42
IZ
2002 inner = NEXTOPER(scan);
2003 if (c1 == BRANCHJ) {
2004 inner = NEXTOPER(inner);
2005 }
2006 } while (scan != NULL && OP(scan) == c1);
4633a7c4 2007 sayNO;
a0d0e21e 2008 /* NOTREACHED */
a687059c 2009 }
a0d0e21e
LW
2010 }
2011 break;
2012 case MINMOD:
2013 minmod = 1;
2014 break;
c277df42
IZ
2015 case CURLYM:
2016 {
00db4c45 2017 I32 l = 0;
c277df42
IZ
2018 CHECKPOINT lastcp;
2019
2020 /* We suppose that the next guy does not need
2021 backtracking: in particular, it is of constant length,
2022 and has no parenths to influence future backrefs. */
2023 ln = ARG1(scan); /* min to match */
2024 n = ARG2(scan); /* max to match */
c277df42
IZ
2025 paren = scan->flags;
2026 if (paren) {
3280af22
NIS
2027 if (paren > PL_regsize)
2028 PL_regsize = paren;
2029 if (paren > *PL_reglastparen)
2030 *PL_reglastparen = paren;
c277df42 2031 }
dc45a647 2032 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
c277df42
IZ
2033 if (paren)
2034 scan += NEXT_OFF(scan); /* Skip former OPEN. */
3280af22 2035 PL_reginput = locinput;
c277df42
IZ
2036 if (minmod) {
2037 minmod = 0;
2038 if (ln && regrepeat_hard(scan, ln, &l) < ln)
2039 sayNO;
5f4b28b2 2040 if (ln && l == 0 && n >= ln
c277df42
IZ
2041 /* In fact, this is tricky. If paren, then the
2042 fact that we did/didnot match may influence
2043 future execution. */
2044 && !(paren && ln == 0))
2045 ln = n;
3280af22 2046 locinput = PL_reginput;
22c35a8c 2047 if (PL_regkind[(U8)OP(next)] == EXACT) {
c277df42
IZ
2048 c1 = UCHARAT(OPERAND(next) + 1);
2049 if (OP(next) == EXACTF)
22c35a8c 2050 c2 = PL_fold[c1];
c277df42 2051 else if (OP(next) == EXACTFL)
22c35a8c 2052 c2 = PL_fold_locale[c1];
c277df42
IZ
2053 else
2054 c2 = c1;
a0ed51b3
LW
2055 }
2056 else
c277df42
IZ
2057 c1 = c2 = -1000;
2058 REGCP_SET;
5f4b28b2 2059 /* This may be improved if l == 0. */
c277df42
IZ
2060 while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */
2061 /* If it could work, try it. */
2062 if (c1 == -1000 ||
3280af22
NIS
2063 UCHARAT(PL_reginput) == c1 ||
2064 UCHARAT(PL_reginput) == c2)
c277df42
IZ
2065 {
2066 if (paren) {
2067 if (n) {
dfe13c55 2068 PL_regstartp[paren] = HOPc(PL_reginput, -l);
3280af22 2069 PL_regendp[paren] = PL_reginput;
a0ed51b3
LW
2070 }
2071 else
3280af22 2072 PL_regendp[paren] = NULL;
c277df42
IZ
2073 }
2074 if (regmatch(next))
2075 sayYES;
2076 REGCP_UNWIND;
2077 }
2078 /* Couldn't or didn't -- move forward. */
3280af22 2079 PL_reginput = locinput;
c277df42
IZ
2080 if (regrepeat_hard(scan, 1, &l)) {
2081 ln++;
3280af22 2082 locinput = PL_reginput;
c277df42
IZ
2083 }
2084 else
2085 sayNO;
2086 }
a0ed51b3
LW
2087 }
2088 else {
c277df42
IZ
2089 n = regrepeat_hard(scan, n, &l);
2090 if (n != 0 && l == 0
2091 /* In fact, this is tricky. If paren, then the
2092 fact that we did/didnot match may influence
2093 future execution. */
2094 && !(paren && ln == 0))
2095 ln = n;
3280af22 2096 locinput = PL_reginput;
c277df42 2097 DEBUG_r(
5c0ca799
GS
2098 PerlIO_printf(Perl_debug_log,
2099 "%*s matched %ld times, len=%ld...\n",
3280af22 2100 REPORT_CODE_OFF+PL_regindent*2, "", n, l)
c277df42
IZ
2101 );
2102 if (n >= ln) {
22c35a8c 2103 if (PL_regkind[(U8)OP(next)] == EXACT) {
c277df42
IZ
2104 c1 = UCHARAT(OPERAND(next) + 1);
2105 if (OP(next) == EXACTF)
22c35a8c 2106 c2 = PL_fold[c1];
c277df42 2107 else if (OP(next) == EXACTFL)
22c35a8c 2108 c2 = PL_fold_locale[c1];
c277df42
IZ
2109 else
2110 c2 = c1;
a0ed51b3
LW
2111 }
2112 else
c277df42
IZ
2113 c1 = c2 = -1000;
2114 }
2115 REGCP_SET;
2116 while (n >= ln) {
2117 /* If it could work, try it. */
2118 if (c1 == -1000 ||
3280af22
NIS
2119 UCHARAT(PL_reginput) == c1 ||
2120 UCHARAT(PL_reginput) == c2)
a0ed51b3
LW
2121 {
2122 DEBUG_r(
c3464db5
DD
2123 PerlIO_printf(Perl_debug_log,
2124 "%*s trying tail with n=%ld...\n",
3280af22 2125 REPORT_CODE_OFF+PL_regindent*2, "", n)
a0ed51b3
LW
2126 );
2127 if (paren) {
2128 if (n) {
dfe13c55 2129 PL_regstartp[paren] = HOPc(PL_reginput, -l);
a0ed51b3 2130 PL_regendp[paren] = PL_reginput;
c277df42 2131 }
a0ed51b3
LW
2132 else
2133 PL_regendp[paren] = NULL;
c277df42 2134 }
a0ed51b3
LW
2135 if (regmatch(next))
2136 sayYES;
2137 REGCP_UNWIND;
2138 }
c277df42
IZ
2139 /* Couldn't or didn't -- back up. */
2140 n--;
dfe13c55 2141 locinput = HOPc(locinput, -l);
3280af22 2142 PL_reginput = locinput;
c277df42
IZ
2143 }
2144 }
2145 sayNO;
2146 break;
2147 }
2148 case CURLYN:
2149 paren = scan->flags; /* Which paren to set */
3280af22
NIS
2150 if (paren > PL_regsize)
2151 PL_regsize = paren;
2152 if (paren > *PL_reglastparen)
2153 *PL_reglastparen = paren;
c277df42
IZ
2154 ln = ARG1(scan); /* min to match */
2155 n = ARG2(scan); /* max to match */
dc45a647 2156 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
c277df42 2157 goto repeat;
a0d0e21e 2158 case CURLY:
c277df42 2159 paren = 0;
a0d0e21e
LW
2160 ln = ARG1(scan); /* min to match */
2161 n = ARG2(scan); /* max to match */
dc45a647 2162 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
a0d0e21e
LW
2163 goto repeat;
2164 case STAR:
2165 ln = 0;
c277df42 2166 n = REG_INFTY;
a0d0e21e 2167 scan = NEXTOPER(scan);
c277df42 2168 paren = 0;
a0d0e21e
LW
2169 goto repeat;
2170 case PLUS:
c277df42
IZ
2171 ln = 1;
2172 n = REG_INFTY;
2173 scan = NEXTOPER(scan);
2174 paren = 0;
2175 repeat:
a0d0e21e
LW
2176 /*
2177 * Lookahead to avoid useless match attempts
2178 * when we know what character comes next.
2179 */
22c35a8c 2180 if (PL_regkind[(U8)OP(next)] == EXACT) {
bbce6d69
PP
2181 c1 = UCHARAT(OPERAND(next) + 1);
2182 if (OP(next) == EXACTF)
22c35a8c 2183 c2 = PL_fold[c1];
bbce6d69 2184 else if (OP(next) == EXACTFL)
22c35a8c 2185 c2 = PL_fold_locale[c1];
bbce6d69
PP
2186 else
2187 c2 = c1;
2188 }
a0d0e21e 2189 else
bbce6d69 2190 c1 = c2 = -1000;
3280af22 2191 PL_reginput = locinput;
a0d0e21e 2192 if (minmod) {
c277df42 2193 CHECKPOINT lastcp;
a0d0e21e
LW
2194 minmod = 0;
2195 if (ln && regrepeat(scan, ln) < ln)
4633a7c4 2196 sayNO;
a0ed51b3 2197 locinput = PL_reginput;
c277df42
IZ
2198 REGCP_SET;
2199 while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
a0d0e21e 2200 /* If it could work, try it. */
bbce6d69 2201 if (c1 == -1000 ||
3280af22
NIS
2202 UCHARAT(PL_reginput) == c1 ||
2203 UCHARAT(PL_reginput) == c2)
bbce6d69 2204 {
c277df42
IZ
2205 if (paren) {
2206 if (n) {
dfe13c55 2207 PL_regstartp[paren] = HOPc(PL_reginput, -1);
3280af22 2208 PL_regendp[paren] = PL_reginput;
a0ed51b3
LW
2209 }
2210 else
3280af22 2211 PL_regendp[paren] = NULL;
c277df42 2212 }
a0d0e21e 2213 if (regmatch(next))
4633a7c4 2214 sayYES;
c277df42 2215 REGCP_UNWIND;
bbce6d69 2216 }
c277df42 2217 /* Couldn't or didn't -- move forward. */
a0ed51b3 2218 PL_reginput = locinput;
a0d0e21e
LW
2219 if (regrepeat(scan, 1)) {
2220 ln++;
a0ed51b3
LW
2221 locinput = PL_reginput;
2222 }
2223 else
4633a7c4 2224 sayNO;
a0d0e21e
LW
2225 }
2226 }
2227 else {
c277df42 2228 CHECKPOINT lastcp;
a0d0e21e 2229 n = regrepeat(scan, n);
a0ed51b3 2230 locinput = PL_reginput;
22c35a8c 2231 if (ln < n && PL_regkind[(U8)OP(next)] == EOL &&
3280af22 2232 (!PL_multiline || OP(next) == SEOL))
a0d0e21e 2233 ln = n; /* why back off? */
c277df42
IZ
2234 REGCP_SET;
2235 if (paren) {
2236 while (n >= ln) {
2237 /* If it could work, try it. */
2238 if (c1 == -1000 ||
3280af22
NIS
2239 UCHARAT(PL_reginput) == c1 ||
2240 UCHARAT(PL_reginput) == c2)
c277df42
IZ
2241 {
2242 if (paren && n) {
2243 if (n) {
dfe13c55 2244 PL_regstartp[paren] = HOPc(PL_reginput, -1);
3280af22 2245 PL_regendp[paren] = PL_reginput;
a0ed51b3
LW
2246 }
2247 else
3280af22 2248 PL_regendp[paren] = NULL;
c277df42
IZ
2249 }
2250 if (regmatch(next))
2251 sayYES;
2252 REGCP_UNWIND;
2253 }
2254 /* Couldn't or didn't -- back up. */
2255 n--;
dfe13c55 2256 PL_reginput = locinput = HOPc(locinput, -1);
c277df42 2257 }
a0ed51b3
LW
2258 }
2259 else {
c277df42
IZ
2260 while (n >= ln) {
2261 /* If it could work, try it. */
2262 if (c1 == -1000 ||
3280af22
NIS
2263 UCHARAT(PL_reginput) == c1 ||
2264 UCHARAT(PL_reginput) == c2)
c277df42
IZ
2265 {
2266 if (regmatch(next))
2267 sayYES;
2268 REGCP_UNWIND;
2269 }
2270 /* Couldn't or didn't -- back up. */
2271 n--;
dfe13c55 2272 PL_reginput = locinput = HOPc(locinput, -1);
bbce6d69 2273 }
a0d0e21e
LW
2274 }
2275 }
4633a7c4 2276 sayNO;
c277df42 2277 break;
a0d0e21e 2278 case END:
0f5d15d6
IZ
2279 if (PL_reg_call_cc) {
2280 re_cc_state *cur_call_cc = PL_reg_call_cc;
2281 CURCUR *cctmp = PL_regcc;
2282 regexp *re = PL_reg_re;
2283 CHECKPOINT cp, lastcp;
2284
2285 cp = regcppush(0); /* Save *all* the positions. */
2286 REGCP_SET;
2287 regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of
2288 the caller. */
2289 PL_reginput = locinput; /* Make position available to
2290 the callcc. */
2291 cache_re(PL_reg_call_cc->re);
2292 PL_regcc = PL_reg_call_cc->cc;
2293 PL_reg_call_cc = PL_reg_call_cc->prev;
2294 if (regmatch(cur_call_cc->node)) {
2295 PL_reg_call_cc = cur_call_cc;
2296 regcpblow(cp);
2297 sayYES;
2298 }
2299 REGCP_UNWIND;
2300 regcppop();
2301 PL_reg_call_cc = cur_call_cc;
2302 PL_regcc = cctmp;
2303 PL_reg_re = re;
2304 cache_re(re);
2305
2306 DEBUG_r(
2307 PerlIO_printf(Perl_debug_log,
2308 "%*s continuation failed...\n",
2309 REPORT_CODE_OFF+PL_regindent*2, "")
2310 );
2311 sayNO;
2312 }
3280af22 2313 if (locinput < PL_regtill)
7e5428c5
IZ
2314 sayNO; /* Cannot match: too short. */
2315 /* Fall through */
2316 case SUCCEED:
3280af22 2317 PL_reginput = locinput; /* put where regtry can find it */
4633a7c4 2318 sayYES; /* Success! */
c277df42
IZ
2319 case SUSPEND:
2320 n = 1;
9fe1d20c 2321 PL_reginput = locinput;
c277df42 2322 goto do_ifmatch;
a0d0e21e 2323 case UNLESSM:
c277df42 2324 n = 0;
a0ed51b3 2325 if (scan->flags) {
dfe13c55 2326 s = HOPMAYBEc(locinput, -scan->flags);
a0ed51b3
LW
2327 if (!s)
2328 goto say_yes;
2329 PL_reginput = s;
2330 }
2331 else
2332 PL_reginput = locinput;
c277df42
IZ
2333 goto do_ifmatch;
2334 case IFMATCH:
2335 n = 1;
a0ed51b3 2336 if (scan->flags) {
dfe13c55 2337 s = HOPMAYBEc(locinput, -scan->flags);
642653c7 2338 if (!s || s < PL_bostr)
a0ed51b3
LW
2339 goto say_no;
2340 PL_reginput = s;
2341 }
2342 else
2343 PL_reginput = locinput;
2344
c277df42 2345 do_ifmatch:
c277df42
IZ
2346 inner = NEXTOPER(NEXTOPER(scan));
2347 if (regmatch(inner) != n) {
2348 say_no:
2349 if (logical) {
2350 logical = 0;
2351 sw = 0;
2352 goto do_longjump;
a0ed51b3
LW
2353 }
2354 else
c277df42
IZ
2355 sayNO;
2356 }
2357 say_yes:
2358 if (logical) {
2359 logical = 0;
2360 sw = 1;
2361 }
fe44a5e8 2362 if (OP(scan) == SUSPEND) {
3280af22 2363 locinput = PL_reginput;
565764a8 2364 nextchr = UCHARAT(locinput);
fe44a5e8 2365 }
c277df42
IZ
2366 /* FALL THROUGH. */
2367 case LONGJMP:
2368 do_longjump:
2369 next = scan + ARG(scan);
2370 if (next == scan)
2371 next = NULL;
a0d0e21e
LW
2372 break;
2373 default:
c030ccd9 2374 PerlIO_printf(PerlIO_stderr(), "%lx %d\n",
c277df42 2375 (unsigned long)scan, OP(scan));
a0d0e21e 2376 FAIL("regexp memory corruption");
a687059c 2377 }
a0d0e21e
LW
2378 scan = next;
2379 }
a687059c 2380
a0d0e21e
LW
2381 /*
2382 * We get here only if there's trouble -- normally "case END" is
2383 * the terminating point.
2384 */
2385 FAIL("corrupted regexp pointers");
2386 /*NOTREACHED*/
4633a7c4
LW
2387 sayNO;
2388
2389yes:
2390#ifdef DEBUGGING
3280af22 2391 PL_regindent--;
4633a7c4
LW
2392#endif
2393 return 1;
2394
2395no:
2396#ifdef DEBUGGING
3280af22 2397 PL_regindent--;
4633a7c4 2398#endif
a0d0e21e 2399 return 0;
a687059c
LW
2400}
2401
2402/*
2403 - regrepeat - repeatedly match something simple, report how many
2404 */
2405/*
2406 * [This routine now assumes that it will only match on things of length 1.
2407 * That was true before, but now we assume scan - reginput is the count,
a0ed51b3 2408 * rather than incrementing count on every character. [Er, except utf8.]]
a687059c 2409 */
76e3520e 2410STATIC I32
c277df42 2411regrepeat(regnode *p, I32 max)
a687059c 2412{
5c0ca799 2413 dTHR;
a0d0e21e
LW
2414 register char *scan;
2415 register char *opnd;
2416 register I32 c;
3280af22 2417 register char *loceol = PL_regeol;
a0ed51b3 2418 register I32 hardcount = 0;
a0d0e21e 2419
3280af22 2420 scan = PL_reginput;
c277df42 2421 if (max != REG_INFTY && max < loceol - scan)
a0d0e21e 2422 loceol = scan + max;
161b471a 2423 opnd = (char *) OPERAND(p);
a0d0e21e 2424 switch (OP(p)) {
22c35a8c 2425 case REG_ANY:
a0d0e21e
LW
2426 while (scan < loceol && *scan != '\n')
2427 scan++;
2428 break;
2429 case SANY:
2430 scan = loceol;
2431 break;
a0ed51b3
LW
2432 case ANYUTF8:
2433 loceol = PL_regeol;
2434 while (scan < loceol && *scan != '\n') {
2435 scan += UTF8SKIP(scan);
2436 hardcount++;
2437 }
2438 break;
2439 case SANYUTF8:
2440 loceol = PL_regeol;
2441 while (scan < loceol) {
2442 scan += UTF8SKIP(scan);
2443 hardcount++;
2444 }
2445 break;
bbce6d69
PP
2446 case EXACT: /* length of string is 1 */
2447 c = UCHARAT(++opnd);
2448 while (scan < loceol && UCHARAT(scan) == c)
2449 scan++;
2450 break;
2451 case EXACTF: /* length of string is 1 */
2452 c = UCHARAT(++opnd);
2453 while (scan < loceol &&
22c35a8c 2454 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c]))
bbce6d69
PP
2455 scan++;
2456 break;
2457 case EXACTFL: /* length of string is 1 */
3280af22 2458 PL_reg_flags |= RF_tainted;
bbce6d69
PP
2459 c = UCHARAT(++opnd);
2460 while (scan < loceol &&
22c35a8c 2461 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c]))
a0d0e21e
LW
2462 scan++;
2463 break;
a0ed51b3
LW
2464 case ANYOFUTF8:
2465 loceol = PL_regeol;
2466 while (scan < loceol && REGINCLASSUTF8(p, (U8*)scan)) {
2467 scan += UTF8SKIP(scan);
2468 hardcount++;
2469 }
2470 break;
a0d0e21e 2471 case ANYOF:
ae5c130c 2472 while (scan < loceol && REGINCLASS(opnd, *scan))
a0d0e21e 2473 scan++;
a0d0e21e
LW
2474 break;
2475 case ALNUM:
2476 while (scan < loceol && isALNUM(*scan))
2477 scan++;
2478 break;
a0ed51b3
LW
2479 case ALNUMUTF8:
2480 loceol = PL_regeol;
dfe13c55 2481 while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) {
a0ed51b3
LW
2482 scan += UTF8SKIP(scan);
2483 hardcount++;
2484 }
2485 break;
bbce6d69 2486 case ALNUML:
3280af22 2487 PL_reg_flags |= RF_tainted;
bbce6d69
PP
2488 while (scan < loceol && isALNUM_LC(*scan))
2489 scan++;
2490 break;
a0ed51b3
LW
2491 case ALNUMLUTF8:
2492 PL_reg_flags |= RF_tainted;
2493 loceol = PL_regeol;
dfe13c55 2494 while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) {
a0ed51b3
LW
2495 scan += UTF8SKIP(scan);
2496 hardcount++;
2497 }
2498 break;
2499 break;
a0d0e21e
LW
2500 case NALNUM:
2501 while (scan < loceol && !isALNUM(*scan))
2502 scan++;
2503 break;
a0ed51b3
LW
2504 case NALNUMUTF8:
2505 loceol = PL_regeol;
dfe13c55 2506 while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) {
a0ed51b3
LW
2507 scan += UTF8SKIP(scan);
2508 hardcount++;
2509 }
2510 break;
bbce6d69 2511 case NALNUML:
3280af22 2512 PL_reg_flags |= RF_tainted;
bbce6d69
PP
2513 while (scan < loceol && !isALNUM_LC(*scan))
2514 scan++;
2515 break;
a0ed51b3
LW
2516 case NALNUMLUTF8:
2517 PL_reg_flags |= RF_tainted;
2518 loceol = PL_regeol;
dfe13c55 2519 while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) {
a0ed51b3
LW
2520 scan += UTF8SKIP(scan);
2521 hardcount++;
2522 }
2523 break;
a0d0e21e
LW
2524 case SPACE:
2525 while (scan < loceol && isSPACE(*scan))
2526 scan++;
2527 break;
a0ed51b3
LW
2528 case SPACEUTF8:
2529 loceol = PL_regeol;
dfe13c55 2530 while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) {
a0ed51b3
LW
2531 scan += UTF8SKIP(scan);
2532 hardcount++;
2533 }
2534 break;
bbce6d69 2535 case SPACEL:
3280af22 2536 PL_reg_flags |= RF_tainted;
bbce6d69
PP
2537 while (scan < loceol && isSPACE_LC(*scan))
2538 scan++;
2539 break;
a0ed51b3
LW
2540 case SPACELUTF8:
2541 PL_reg_flags |= RF_tainted;
2542 loceol = PL_regeol;
dfe13c55 2543 while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
a0ed51b3
LW
2544 scan += UTF8SKIP(scan);
2545 hardcount++;
2546 }
2547 break;
a0d0e21e
LW
2548 case NSPACE:
2549 while (scan < loceol && !isSPACE(*scan))
2550 scan++;
2551 break;
a0ed51b3
LW
2552 case NSPACEUTF8:
2553 loceol = PL_regeol;
dfe13c55 2554 while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) {
a0ed51b3
LW
2555 scan += UTF8SKIP(scan);
2556 hardcount++;
2557 }
2558 break;
bbce6d69 2559 case NSPACEL:
3280af22 2560 PL_reg_flags |= RF_tainted;
bbce6d69
PP
2561 while (scan < loceol && !isSPACE_LC(*scan))
2562 scan++;
2563 break;
a0ed51b3
LW
2564 case NSPACELUTF8:
2565 PL_reg_flags |= RF_tainted;
2566 loceol = PL_regeol;
dfe13c55 2567 while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
a0ed51b3
LW
2568 scan += UTF8SKIP(scan);
2569 hardcount++;
2570 }
2571 break;
a0d0e21e
LW
2572 case DIGIT:
2573 while (scan < loceol && isDIGIT(*scan))
2574 scan++;
2575 break;
a0ed51b3
LW
2576 case DIGITUTF8:
2577 loceol = PL_regeol;
dfe13c55 2578 while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) {
a0ed51b3
LW
2579 scan += UTF8SKIP(scan);
2580 hardcount++;
2581 }
2582 break;
2583 break;
a0d0e21e
LW
2584 case NDIGIT:
2585 while (scan < loceol && !isDIGIT(*scan))
2586 scan++;
2587 break;
a0ed51b3
LW
2588 case NDIGITUTF8:
2589 loceol = PL_regeol;
dfe13c55 2590 while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) {
a0ed51b3
LW
2591 scan += UTF8SKIP(scan);
2592 hardcount++;
2593 }
2594 break;
a0d0e21e
LW
2595 default: /* Called on something of 0 width. */
2596 break; /* So match right here or not at all. */
2597 }
a687059c 2598
a0ed51b3
LW
2599 if (hardcount)
2600 c = hardcount;
2601 else
2602 c = scan - PL_reginput;
3280af22 2603 PL_reginput = scan;
a687059c 2604
c277df42
IZ
2605 DEBUG_r(
2606 {
2607 SV *prop = sv_newmortal();
2608
2609 regprop(prop, p);
2610 PerlIO_printf(Perl_debug_log,
2611 "%*s %s can match %ld times out of %ld...\n",
2612 REPORT_CODE_OFF+1, "", SvPVX(prop),c,max);
2613 });
2614
a0d0e21e 2615 return(c);
a687059c
LW
2616}
2617
2618/*
c277df42
IZ
2619 - regrepeat_hard - repeatedly match something, report total lenth and length
2620 *
2621 * The repeater is supposed to have constant length.
2622 */
2623
76e3520e 2624STATIC I32
c277df42
IZ
2625regrepeat_hard(regnode *p, I32 max, I32 *lp)
2626{
5c0ca799 2627 dTHR;
c277df42
IZ
2628 register char *scan;
2629 register char *start;
3280af22 2630 register char *loceol = PL_regeol;
a0ed51b3 2631 I32 l = 0;
708e3b05 2632 I32 count = 0, res = 1;
a0ed51b3
LW
2633
2634 if (!max)
2635 return 0;
c277df42 2636
3280af22 2637 start = PL_reginput;
a0ed51b3 2638 if (UTF) {
708e3b05 2639 while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
a0ed51b3
LW
2640 if (!count++) {
2641 l = 0;
2642 while (start < PL_reginput) {
2643 l++;
2644 start += UTF8SKIP(start);
2645 }
2646 *lp = l;
2647 if (l == 0)
2648 return max;
2649 }
2650 if (count == max)
2651 return count;
2652 }
2653 }
2654 else {
708e3b05 2655 while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
a0ed51b3
LW
2656 if (!count++) {
2657 *lp = l = PL_reginput - start;
2658 if (max != REG_INFTY && l*max < loceol - scan)
2659 loceol = scan + l*max;
2660 if (l == 0)
2661 return max;
c277df42
IZ
2662 }
2663 }
2664 }
708e3b05 2665 if (!res)
3280af22 2666 PL_reginput = scan;
c277df42 2667
a0ed51b3 2668 return count;
c277df42
IZ
2669}
2670
2671/*
cb8d8820 2672 - reginclass - determine if a character falls into a character class
bbce6d69
PP
2673 */
2674
76e3520e 2675STATIC bool
8ac85365 2676reginclass(register char *p, register I32 c)
bbce6d69 2677{
5c0ca799 2678 dTHR;
bbce6d69
PP
2679 char flags = *p;
2680 bool match = FALSE;
2681
2682 c &= 0xFF;
ae5c130c 2683 if (ANYOF_TEST(p, c))
bbce6d69
PP
2684 match = TRUE;
2685 else if (flags & ANYOF_FOLD) {
2686 I32 cf;
2687 if (flags & ANYOF_LOCALE) {
3280af22 2688 PL_reg_flags |= RF_tainted;
22c35a8c 2689 cf = PL_fold_locale[c];
bbce6d69
PP
2690 }
2691 else
22c35a8c 2692 cf = PL_fold[c];
ae5c130c 2693 if (ANYOF_TEST(p, cf))
bbce6d69
PP
2694 match = TRUE;
2695 }
2696
2697 if (!match && (flags & ANYOF_ISA)) {
3280af22 2698 PL_reg_flags |= RF_tainted;
bbce6d69
PP
2699
2700 if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) ||
2701 ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) ||
2702 ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) ||
2703 ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c)))
2704 {
2705 match = TRUE;
2706 }
2707 }
2708
ae5c130c 2709 return (flags & ANYOF_INVERT) ? !match : match;
bbce6d69
PP
2710}
2711
a0ed51b3
LW
2712STATIC bool
2713reginclassutf8(regnode *f, U8 *p)
c485e607
NIS
2714{
2715 dTHR;
a0ed51b3
LW
2716 char flags = ARG1(f);
2717 bool match = FALSE;
2718 SV *sv = (SV*)PL_regdata->data[ARG2(f)];
2719
2720 if (swash_fetch(sv, p))
2721 match = TRUE;
2722 else if (flags & ANYOF_FOLD) {
2723 I32 cf;
dfe13c55 2724 U8 tmpbuf[10];
a0ed51b3
LW
2725 if (flags & ANYOF_LOCALE) {
2726 PL_reg_flags |= RF_tainted;
2727 uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p));
2728 }
2729 else
2730 uv_to_utf8(tmpbuf, toLOWER_utf8(p));
2731 if (swash_fetch(sv, tmpbuf))
2732 match = TRUE;
2733 }
2734
2735 if (!match && (flags & ANYOF_ISA)) {
2736 PL_reg_flags |= RF_tainted;
2737
2738 if (((flags & ANYOF_ALNUML) && isALNUM_LC_utf8(p)) ||
2739 ((flags & ANYOF_NALNUML) && !isALNUM_LC_utf8(p)) ||
2740 ((flags & ANYOF_SPACEL) && isSPACE_LC_utf8(p)) ||
2741 ((flags & ANYOF_NSPACEL) && !isSPACE_LC_utf8(p)))
2742 {
2743 match = TRUE;
2744 }
2745 }
2746
2747 return (flags & ANYOF_INVERT) ? !match : match;
2748}
161b471a 2749
dfe13c55
GS
2750STATIC U8 *
2751reghop(U8 *s, I32 off)
c485e607
NIS
2752{
2753 dTHR;
a0ed51b3
LW
2754 if (off >= 0) {
2755 while (off-- && s < (U8*)PL_regeol)
2756 s += UTF8SKIP(s);
2757 }
2758 else {
2759 while (off++) {
2760 if (s > (U8*)PL_bostr) {
2761 s--;
2762 if (*s & 0x80) {
2763 while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80)
2764 s--;
2765 } /* XXX could check well-formedness here */
2766 }
2767 }
2768 }
2769 return s;
2770}
161b471a 2771
dfe13c55
GS
2772STATIC U8 *
2773reghopmaybe(U8* s, I32 off)
a0ed51b3 2774{
c485e607 2775 dTHR;
a0ed51b3
LW
2776 if (off >= 0) {
2777 while (off-- && s < (U8*)PL_regeol)
2778 s += UTF8SKIP(s);
2779 if (off >= 0)
2780 return 0;
2781 }
2782 else {
2783 while (off++) {
2784 if (s > (U8*)PL_bostr) {
2785 s--;
2786 if (*s & 0x80) {
2787 while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80)
2788 s--;
2789 } /* XXX could check well-formedness here */
2790 }
2791 else
2792 break;
2793 }
2794 if (off <= 0)
2795 return 0;
2796 }
2797 return s;
2798}