This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Forgot the latest mktables.PL from #9899.
[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 */
cad2e5aa 28# if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
b9d5759e
AD
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
cad2e5aa 38# define Perl_re_intuit_start my_re_intuit_start
d06ea78c
GS
39/* *These* symbols are masked to allow static link. */
40# define Perl_pregexec my_pregexec
9041c2e3 41# define Perl_reginitcolors my_reginitcolors
490a3f88 42# define Perl_regclass_swash my_regclass_swash
c5be433b
GS
43
44# define PERL_NO_GET_CONTEXT
9041c2e3 45#endif
56953603 46
f0fcb552 47/*SUPPRESS 112*/
a687059c 48/*
e50aee73 49 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
50 *
51 * Copyright (c) 1986 by University of Toronto.
52 * Written by Henry Spencer. Not derived from licensed software.
53 *
54 * Permission is granted to anyone to use this software for any
55 * purpose on any computer system, and to redistribute it freely,
56 * subject to the following restrictions:
57 *
58 * 1. The author is not responsible for the consequences of use of
59 * this software, no matter how awful, even if they arise
60 * from defects in it.
61 *
62 * 2. The origin of this software must not be misrepresented, either
63 * by explicit claim or by omission.
64 *
65 * 3. Altered versions must be plainly marked as such, and must not
66 * be misrepresented as being the original software.
67 *
68 **** Alterations to Henry's code are...
69 ****
bc89e66f 70 **** Copyright (c) 1991-2001, Larry Wall
a687059c 71 ****
9ef589d8
LW
72 **** You may distribute under the terms of either the GNU General Public
73 **** License or the Artistic License, as specified in the README file.
a687059c
LW
74 *
75 * Beware that some of this code is subtly aware of the way operator
76 * precedence is structured in regular expressions. Serious changes in
77 * regular-expression syntax might require a total rethink.
78 */
79#include "EXTERN.h"
864dbfa3 80#define PERL_IN_REGEXEC_C
a687059c 81#include "perl.h"
0f5d15d6 82
c5be433b
GS
83#ifdef PERL_IN_XSUB_RE
84# if defined(PERL_CAPI) || defined(PERL_OBJECT)
85# include "XSUB.h"
86# endif
87#endif
88
a687059c
LW
89#include "regcomp.h"
90
c277df42
IZ
91#define RF_tainted 1 /* tainted information used? */
92#define RF_warned 2 /* warned about big count? */
ce862d02 93#define RF_evaled 4 /* Did an EVAL with setting? */
a0ed51b3
LW
94#define RF_utf8 8 /* String contains multibyte chars? */
95
96#define UTF (PL_reg_flags & RF_utf8)
ce862d02
IZ
97
98#define RS_init 1 /* eval environment created */
99#define RS_set 2 /* replsv value is set */
c277df42 100
a687059c
LW
101#ifndef STATIC
102#define STATIC static
103#endif
104
c277df42
IZ
105/*
106 * Forwards.
107 */
108
a0ed51b3 109#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
1aa99e6b 110#define CHR_DIST(a,b) (DO_UTF8(PL_reg_sv) ? utf8_distance(a,b) : a - b)
a0ed51b3 111
dfe13c55
GS
112#define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
113#define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
1aa99e6b
IH
114#define HOP(pos,off) (DO_UTF8(PL_reg_sv) ? reghop((U8*)pos, off) : (U8*)(pos + off))
115#define HOPMAYBE(pos,off) (DO_UTF8(PL_reg_sv) ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
dfe13c55
GS
116#define HOPc(pos,off) ((char*)HOP(pos,off))
117#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
a0d0e21e 118
1aa99e6b
IH
119#define reghop3_c(pos,off,lim) ((char*)reghop3((U8*)pos, off, (U8*)lim))
120#define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim))
121#define HOP3(pos,off,lim) (DO_UTF8(PL_reg_sv) ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
122#define HOPMAYBE3(pos,off,lim) (DO_UTF8(PL_reg_sv) ? reghopmaybe3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
123#define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
124#define HOPMAYBE3c(pos,off,lim) ((char*)HOPMAYBE3(pos,off,lim))
125
8269fa76 126#define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) (void)CAT2(is_utf8_, a)((U8*)b); } STMT_END
51371543 127
8269fa76 128static void restore_pos(pTHXo_ void *arg);
51371543 129
76e3520e 130STATIC CHECKPOINT
cea2e8a9 131S_regcppush(pTHX_ I32 parenfloor)
a0d0e21e 132{
3280af22 133 int retval = PL_savestack_ix;
b1ce53c5
JH
134#define REGCP_PAREN_ELEMS 4
135 int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
a0d0e21e
LW
136 int p;
137
b1ce53c5
JH
138#define REGCP_OTHER_ELEMS 5
139 SSCHECK(paren_elems_to_push + REGCP_OTHER_ELEMS);
3280af22 140 for (p = PL_regsize; p > parenfloor; p--) {
b1ce53c5 141/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
cf93c79d
IZ
142 SSPUSHINT(PL_regendp[p]);
143 SSPUSHINT(PL_regstartp[p]);
3280af22 144 SSPUSHPTR(PL_reg_start_tmp[p]);
a0d0e21e
LW
145 SSPUSHINT(p);
146 }
b1ce53c5 147/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
3280af22
NIS
148 SSPUSHINT(PL_regsize);
149 SSPUSHINT(*PL_reglastparen);
150 SSPUSHPTR(PL_reginput);
41123dfd
JH
151#define REGCP_FRAME_ELEMS 2
152/* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and
153 * are needed for the regexp context stack bookkeeping. */
154 SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
b1ce53c5 155 SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */
41123dfd 156
a0d0e21e
LW
157 return retval;
158}
159
c277df42 160/* These are needed since we do not localize EVAL nodes: */
02db2b7b 161# define REGCP_SET(cp) DEBUG_r(PerlIO_printf(Perl_debug_log, \
faccc32b 162 " Setting an EVAL scope, savestack=%"IVdf"\n", \
02db2b7b 163 (IV)PL_savestack_ix)); cp = PL_savestack_ix
c3464db5 164
02db2b7b 165# define REGCP_UNWIND(cp) DEBUG_r(cp != PL_savestack_ix ? \
c3464db5 166 PerlIO_printf(Perl_debug_log, \
faccc32b 167 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
02db2b7b 168 (IV)(cp), (IV)PL_savestack_ix) : 0); regcpblow(cp)
c277df42 169
76e3520e 170STATIC char *
cea2e8a9 171S_regcppop(pTHX)
a0d0e21e 172{
b1ce53c5 173 I32 i;
a0d0e21e
LW
174 U32 paren = 0;
175 char *input;
cf93c79d 176 I32 tmps;
b1ce53c5
JH
177
178 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
a0d0e21e 179 i = SSPOPINT;
b1ce53c5
JH
180 assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
181 i = SSPOPINT; /* Parentheses elements to pop. */
a0d0e21e 182 input = (char *) SSPOPPTR;
3280af22
NIS
183 *PL_reglastparen = SSPOPINT;
184 PL_regsize = SSPOPINT;
b1ce53c5
JH
185
186 /* Now restore the parentheses context. */
41123dfd
JH
187 for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
188 i > 0; i -= REGCP_PAREN_ELEMS) {
a0d0e21e 189 paren = (U32)SSPOPINT;
3280af22 190 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
cf93c79d
IZ
191 PL_regstartp[paren] = SSPOPINT;
192 tmps = SSPOPINT;
3280af22
NIS
193 if (paren <= *PL_reglastparen)
194 PL_regendp[paren] = tmps;
c277df42 195 DEBUG_r(
c3464db5 196 PerlIO_printf(Perl_debug_log,
b900a521 197 " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
9041c2e3 198 (UV)paren, (IV)PL_regstartp[paren],
b900a521 199 (IV)(PL_reg_start_tmp[paren] - PL_bostr),
9041c2e3 200 (IV)PL_regendp[paren],
3280af22 201 (paren > *PL_reglastparen ? "(no)" : ""));
c277df42 202 );
a0d0e21e 203 }
c277df42 204 DEBUG_r(
3280af22 205 if (*PL_reglastparen + 1 <= PL_regnpar) {
c3464db5 206 PerlIO_printf(Perl_debug_log,
faccc32b
JH
207 " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
208 (IV)(*PL_reglastparen + 1), (IV)PL_regnpar);
c277df42
IZ
209 }
210 );
daf18116 211#if 1
dafc8851
JH
212 /* It would seem that the similar code in regtry()
213 * already takes care of this, and in fact it is in
214 * a better location to since this code can #if 0-ed out
215 * but the code in regtry() is needed or otherwise tests
216 * requiring null fields (pat.t#187 and split.t#{13,14}
daf18116
JH
217 * (as of patchlevel 7877) will fail. Then again,
218 * this code seems to be necessary or otherwise
219 * building DynaLoader will fail:
220 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
221 * --jhi */
3280af22
NIS
222 for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) {
223 if (paren > PL_regsize)
cf93c79d
IZ
224 PL_regstartp[paren] = -1;
225 PL_regendp[paren] = -1;
a0d0e21e 226 }
dafc8851 227#endif
a0d0e21e
LW
228 return input;
229}
230
0f5d15d6 231STATIC char *
cea2e8a9 232S_regcp_set_to(pTHX_ I32 ss)
0f5d15d6
IZ
233{
234 I32 tmp = PL_savestack_ix;
235
236 PL_savestack_ix = ss;
237 regcppop();
238 PL_savestack_ix = tmp;
942e002e 239 return Nullch;
0f5d15d6
IZ
240}
241
242typedef struct re_cc_state
243{
244 I32 ss;
245 regnode *node;
246 struct re_cc_state *prev;
247 CURCUR *cc;
248 regexp *re;
249} re_cc_state;
250
02db2b7b 251#define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
a0d0e21e 252
29d1e993
HS
253#define TRYPAREN(paren, n, input) { \
254 if (paren) { \
255 if (n) { \
256 PL_regstartp[paren] = HOPc(input, -1) - PL_bostr; \
257 PL_regendp[paren] = input - PL_bostr; \
258 } \
259 else \
260 PL_regendp[paren] = -1; \
261 } \
262 if (regmatch(next)) \
263 sayYES; \
264 if (paren && n) \
265 PL_regendp[paren] = -1; \
266}
267
268
a687059c 269/*
e50aee73 270 * pregexec and friends
a687059c
LW
271 */
272
273/*
c277df42 274 - pregexec - match a regexp against a string
a687059c 275 */
c277df42 276I32
864dbfa3 277Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend,
c3464db5 278 char *strbeg, I32 minend, SV *screamer, U32 nosave)
c277df42
IZ
279/* strend: pointer to null at end of string */
280/* strbeg: real beginning of string */
281/* minend: end of match must be >=minend after stringarg. */
282/* nosave: For optimizations. */
283{
284 return
9041c2e3 285 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
c277df42
IZ
286 nosave ? 0 : REXEC_COPY_STR);
287}
0f5d15d6
IZ
288
289STATIC void
cea2e8a9 290S_cache_re(pTHX_ regexp *prog)
0f5d15d6
IZ
291{
292 PL_regprecomp = prog->precomp; /* Needed for FAIL. */
293#ifdef DEBUGGING
294 PL_regprogram = prog->program;
295#endif
296 PL_regnpar = prog->nparens;
9041c2e3
NIS
297 PL_regdata = prog->data;
298 PL_reg_re = prog;
0f5d15d6 299}
22e551b9 300
9041c2e3 301/*
cad2e5aa
JH
302 * Need to implement the following flags for reg_anch:
303 *
304 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
305 * USE_INTUIT_ML
306 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
307 * INTUIT_AUTORITATIVE_ML
308 * INTUIT_ONCE_NOML - Intuit can match in one location only.
309 * INTUIT_ONCE_ML
310 *
311 * Another flag for this function: SECOND_TIME (so that float substrs
312 * with giant delta may be not rechecked).
313 */
314
315/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
316
2c2d71f5 317/* If SCREAM, then SvPVX(sv) should be compatible with strpos and strend.
cad2e5aa
JH
318 Otherwise, only SvCUR(sv) is used to get strbeg. */
319
320/* XXXX We assume that strpos is strbeg unless sv. */
321
6eb5f6b9
JH
322/* XXXX Some places assume that there is a fixed substring.
323 An update may be needed if optimizer marks as "INTUITable"
324 RExen without fixed substrings. Similarly, it is assumed that
325 lengths of all the strings are no more than minlen, thus they
326 cannot come from lookahead.
327 (Or minlen should take into account lookahead.) */
328
2c2d71f5
JH
329/* A failure to find a constant substring means that there is no need to make
330 an expensive call to REx engine, thus we celebrate a failure. Similarly,
331 finding a substring too deep into the string means that less calls to
30944b6d
IZ
332 regtry() should be needed.
333
334 REx compiler's optimizer found 4 possible hints:
335 a) Anchored substring;
336 b) Fixed substring;
337 c) Whether we are anchored (beginning-of-line or \G);
338 d) First node (of those at offset 0) which may distingush positions;
6eb5f6b9 339 We use a)b)d) and multiline-part of c), and try to find a position in the
30944b6d
IZ
340 string which does not contradict any of them.
341 */
2c2d71f5 342
6eb5f6b9
JH
343/* Most of decisions we do here should have been done at compile time.
344 The nodes of the REx which we used for the search should have been
345 deleted from the finite automaton. */
346
cad2e5aa
JH
347char *
348Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
349 char *strend, U32 flags, re_scream_pos_data *data)
350{
2c2d71f5 351 register I32 start_shift;
cad2e5aa 352 /* Should be nonnegative! */
2c2d71f5
JH
353 register I32 end_shift;
354 register char *s;
355 register SV *check;
a1933d95 356 char *strbeg;
cad2e5aa
JH
357 char *t;
358 I32 ml_anch;
2c2d71f5 359 char *tmp;
6eb5f6b9
JH
360 register char *other_last = Nullch; /* other substr checked before this */
361 char *check_at; /* check substr found at this pos */
30944b6d
IZ
362#ifdef DEBUGGING
363 char *i_strpos = strpos;
364#endif
cad2e5aa
JH
365
366 DEBUG_r( if (!PL_colorset) reginitcolors() );
367 DEBUG_r(PerlIO_printf(Perl_debug_log,
2c2d71f5 368 "%sGuessing start of match, REx%s `%s%.60s%s%s' against `%s%.*s%s%s'...\n",
cad2e5aa
JH
369 PL_colors[4],PL_colors[5],PL_colors[0],
370 prog->precomp,
371 PL_colors[1],
372 (strlen(prog->precomp) > 60 ? "..." : ""),
373 PL_colors[0],
b900a521 374 (int)(strend - strpos > 60 ? 60 : strend - strpos),
cad2e5aa
JH
375 strpos, PL_colors[1],
376 (strend - strpos > 60 ? "..." : ""))
377 );
378
1aa99e6b
IH
379 if (prog->reganch & ROPT_UTF8)
380 PL_reg_flags |= RF_utf8;
381
382 if (prog->minlen > CHR_DIST((U8*)strend, (U8*)strpos)) {
2c2d71f5 383 DEBUG_r(PerlIO_printf(Perl_debug_log, "String too short...\n"));
cad2e5aa 384 goto fail;
2c2d71f5 385 }
a1933d95 386 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
1aa99e6b 387 PL_regeol = strend;
653099ff 388 check = prog->check_substr;
2c2d71f5 389 if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */
cad2e5aa
JH
390 ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE)
391 || ( (prog->reganch & ROPT_ANCH_BOL)
2c2d71f5 392 && !PL_multiline ) ); /* Check after \n? */
cad2e5aa 393
7e25d62c
JH
394 if (!ml_anch) {
395 if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */
396 | ROPT_IMPLICIT)) /* not a real BOL */
397 /* SvCUR is not set on references: SvRV and SvPVX overlap */
398 && sv && !SvROK(sv)
399 && (strpos != strbeg)) {
400 DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
401 goto fail;
402 }
403 if (prog->check_offset_min == prog->check_offset_max &&
404 !(prog->reganch & ROPT_SANY_SEEN)) {
2c2d71f5 405 /* Substring at constant offset from beg-of-str... */
cad2e5aa
JH
406 I32 slen;
407
1aa99e6b 408 s = HOP3c(strpos, prog->check_offset_min, strend);
653099ff
GS
409 if (SvTAIL(check)) {
410 slen = SvCUR(check); /* >= 1 */
cad2e5aa 411
9041c2e3 412 if ( strend - s > slen || strend - s < slen - 1
2c2d71f5
JH
413 || (strend - s == slen && strend[-1] != '\n')) {
414 DEBUG_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
415 goto fail_finish;
cad2e5aa
JH
416 }
417 /* Now should match s[0..slen-2] */
418 slen--;
653099ff 419 if (slen && (*SvPVX(check) != *s
cad2e5aa 420 || (slen > 1
653099ff 421 && memNE(SvPVX(check), s, slen)))) {
2c2d71f5
JH
422 report_neq:
423 DEBUG_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
424 goto fail_finish;
425 }
cad2e5aa 426 }
653099ff
GS
427 else if (*SvPVX(check) != *s
428 || ((slen = SvCUR(check)) > 1
429 && memNE(SvPVX(check), s, slen)))
2c2d71f5
JH
430 goto report_neq;
431 goto success_at_start;
7e25d62c 432 }
cad2e5aa 433 }
2c2d71f5 434 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
cad2e5aa 435 s = strpos;
2c2d71f5 436 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
2c2d71f5 437 end_shift = prog->minlen - start_shift -
653099ff 438 CHR_SVLEN(check) + (SvTAIL(check) != 0);
2c2d71f5 439 if (!ml_anch) {
653099ff
GS
440 I32 end = prog->check_offset_max + CHR_SVLEN(check)
441 - (SvTAIL(check) != 0);
1aa99e6b 442 I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
2c2d71f5
JH
443
444 if (end_shift < eshift)
445 end_shift = eshift;
446 }
cad2e5aa 447 }
2c2d71f5 448 else { /* Can match at random position */
cad2e5aa
JH
449 ml_anch = 0;
450 s = strpos;
2c2d71f5
JH
451 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
452 /* Should be nonnegative! */
453 end_shift = prog->minlen - start_shift -
653099ff 454 CHR_SVLEN(check) + (SvTAIL(check) != 0);
cad2e5aa
JH
455 }
456
2c2d71f5 457#ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */
0033605d 458 if (end_shift < 0)
6bbae5e6 459 Perl_croak(aTHX_ "panic: end_shift");
2c2d71f5
JH
460#endif
461
2c2d71f5
JH
462 restart:
463 /* Find a possible match in the region s..strend by looking for
464 the "check" substring in the region corrected by start/end_shift. */
cad2e5aa 465 if (flags & REXEC_SCREAM) {
cad2e5aa
JH
466 I32 p = -1; /* Internal iterator of scream. */
467 I32 *pp = data ? data->scream_pos : &p;
468
2c2d71f5
JH
469 if (PL_screamfirst[BmRARE(check)] >= 0
470 || ( BmRARE(check) == '\n'
471 && (BmPREVIOUS(check) == SvCUR(check) - 1)
472 && SvTAIL(check) ))
9041c2e3 473 s = screaminstr(sv, check,
2c2d71f5 474 start_shift + (s - strbeg), end_shift, pp, 0);
cad2e5aa 475 else
2c2d71f5 476 goto fail_finish;
cad2e5aa
JH
477 if (data)
478 *data->scream_olds = s;
479 }
3baa4c62
JH
480 else if (prog->reganch & ROPT_SANY_SEEN)
481 s = fbm_instr((U8*)(s + start_shift),
482 (U8*)(strend - end_shift),
483 check, PL_multiline ? FBMrf_MULTILINE : 0);
cad2e5aa 484 else
1aa99e6b
IH
485 s = fbm_instr(HOP3(s, start_shift, strend),
486 HOP3(strend, -end_shift, strbeg),
2c2d71f5 487 check, PL_multiline ? FBMrf_MULTILINE : 0);
cad2e5aa
JH
488
489 /* Update the count-of-usability, remove useless subpatterns,
490 unshift s. */
2c2d71f5
JH
491
492 DEBUG_r(PerlIO_printf(Perl_debug_log, "%s %s substr `%s%.*s%s'%s%s",
493 (s ? "Found" : "Did not find"),
494 ((check == prog->anchored_substr) ? "anchored" : "floating"),
495 PL_colors[0],
7b0972df
JH
496 (int)(SvCUR(check) - (SvTAIL(check)!=0)),
497 SvPVX(check),
2c2d71f5
JH
498 PL_colors[1], (SvTAIL(check) ? "$" : ""),
499 (s ? " at offset " : "...\n") ) );
500
501 if (!s)
502 goto fail_finish;
503
6eb5f6b9
JH
504 check_at = s;
505
2c2d71f5 506 /* Finish the diagnostic message */
30944b6d 507 DEBUG_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
2c2d71f5
JH
508
509 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
510 Start with the other substr.
511 XXXX no SCREAM optimization yet - and a very coarse implementation
512 XXXX /ttx+/ results in anchored=`ttx', floating=`x'. floating will
513 *always* match. Probably should be marked during compile...
514 Probably it is right to do no SCREAM here...
515 */
516
517 if (prog->float_substr && prog->anchored_substr) {
30944b6d 518 /* Take into account the "other" substring. */
2c2d71f5
JH
519 /* XXXX May be hopelessly wrong for UTF... */
520 if (!other_last)
6eb5f6b9 521 other_last = strpos;
2c2d71f5 522 if (check == prog->float_substr) {
30944b6d
IZ
523 do_other_anchored:
524 {
1aa99e6b 525 char *last = HOP3c(s, -start_shift, strbeg), *last1, *last2;
2c2d71f5
JH
526 char *s1 = s;
527
2c2d71f5
JH
528 t = s - prog->check_offset_max;
529 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
530 && (!(prog->reganch & ROPT_UTF8)
1aa99e6b 531 || ((t = reghopmaybe3_c(s, -(prog->check_offset_max), strpos))
2c2d71f5 532 && t > strpos)))
30944b6d 533 /* EMPTY */;
2c2d71f5
JH
534 else
535 t = strpos;
1aa99e6b 536 t = HOP3c(t, prog->anchored_offset, strend);
6eb5f6b9
JH
537 if (t < other_last) /* These positions already checked */
538 t = other_last;
1aa99e6b 539 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
2c2d71f5
JH
540 if (last < last1)
541 last1 = last;
542 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
543 /* On end-of-str: see comment below. */
544 s = fbm_instr((unsigned char*)t,
1aa99e6b
IH
545 HOP3(HOP3(last1, prog->anchored_offset, strend)
546 + SvCUR(prog->anchored_substr),
547 -(SvTAIL(prog->anchored_substr)!=0), strbeg),
548 prog->anchored_substr,
549 PL_multiline ? FBMrf_MULTILINE : 0);
550 DEBUG_r(PerlIO_printf(Perl_debug_log,
551 "%s anchored substr `%s%.*s%s'%s",
2c2d71f5
JH
552 (s ? "Found" : "Contradicts"),
553 PL_colors[0],
7b0972df
JH
554 (int)(SvCUR(prog->anchored_substr)
555 - (SvTAIL(prog->anchored_substr)!=0)),
2c2d71f5
JH
556 SvPVX(prog->anchored_substr),
557 PL_colors[1], (SvTAIL(prog->anchored_substr) ? "$" : "")));
558 if (!s) {
559 if (last1 >= last2) {
560 DEBUG_r(PerlIO_printf(Perl_debug_log,
561 ", giving up...\n"));
562 goto fail_finish;
563 }
564 DEBUG_r(PerlIO_printf(Perl_debug_log,
565 ", trying floating at offset %ld...\n",
1aa99e6b
IH
566 (long)(HOP3c(s1, 1, strend) - i_strpos)));
567 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
568 s = HOP3c(last, 1, strend);
2c2d71f5
JH
569 goto restart;
570 }
571 else {
572 DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
30944b6d 573 (long)(s - i_strpos)));
1aa99e6b
IH
574 t = HOP3c(s, -prog->anchored_offset, strbeg);
575 other_last = HOP3c(s, 1, strend);
30944b6d 576 s = s1;
2c2d71f5
JH
577 if (t == strpos)
578 goto try_at_start;
2c2d71f5
JH
579 goto try_at_offset;
580 }
30944b6d 581 }
2c2d71f5
JH
582 }
583 else { /* Take into account the floating substring. */
584 char *last, *last1;
585 char *s1 = s;
586
1aa99e6b
IH
587 t = HOP3c(s, -start_shift, strbeg);
588 last1 = last =
589 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
590 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
591 last = HOP3c(t, prog->float_max_offset, strend);
592 s = HOP3c(t, prog->float_min_offset, strend);
6eb5f6b9
JH
593 if (s < other_last)
594 s = other_last;
2c2d71f5
JH
595 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
596 /* fbm_instr() takes into account exact value of end-of-str
597 if the check is SvTAIL(ed). Since false positives are OK,
598 and end-of-str is not later than strend we are OK. */
599 s = fbm_instr((unsigned char*)s,
600 (unsigned char*)last + SvCUR(prog->float_substr)
601 - (SvTAIL(prog->float_substr)!=0),
602 prog->float_substr, PL_multiline ? FBMrf_MULTILINE : 0);
603 DEBUG_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s",
604 (s ? "Found" : "Contradicts"),
605 PL_colors[0],
7b0972df
JH
606 (int)(SvCUR(prog->float_substr)
607 - (SvTAIL(prog->float_substr)!=0)),
2c2d71f5
JH
608 SvPVX(prog->float_substr),
609 PL_colors[1], (SvTAIL(prog->float_substr) ? "$" : "")));
610 if (!s) {
611 if (last1 == last) {
612 DEBUG_r(PerlIO_printf(Perl_debug_log,
613 ", giving up...\n"));
614 goto fail_finish;
615 }
616 DEBUG_r(PerlIO_printf(Perl_debug_log,
617 ", trying anchored starting at offset %ld...\n",
30944b6d 618 (long)(s1 + 1 - i_strpos)));
803ff556 619 other_last = last;
1aa99e6b 620 s = HOP3c(t, 1, strend);
2c2d71f5
JH
621 goto restart;
622 }
623 else {
624 DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
30944b6d 625 (long)(s - i_strpos)));
803ff556 626 other_last = s; /* Fix this later. --Hugo */
30944b6d 627 s = s1;
2c2d71f5
JH
628 if (t == strpos)
629 goto try_at_start;
2c2d71f5
JH
630 goto try_at_offset;
631 }
632 }
cad2e5aa 633 }
2c2d71f5
JH
634
635 t = s - prog->check_offset_max;
2c2d71f5
JH
636 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
637 && (!(prog->reganch & ROPT_UTF8)
1aa99e6b
IH
638 || ((t = reghopmaybe3_c(s, -prog->check_offset_max, strpos))
639 && t > strpos))) {
2c2d71f5
JH
640 /* Fixed substring is found far enough so that the match
641 cannot start at strpos. */
642 try_at_offset:
cad2e5aa 643 if (ml_anch && t[-1] != '\n') {
30944b6d
IZ
644 /* Eventually fbm_*() should handle this, but often
645 anchored_offset is not 0, so this check will not be wasted. */
646 /* XXXX In the code below we prefer to look for "^" even in
647 presence of anchored substrings. And we search even
648 beyond the found float position. These pessimizations
649 are historical artefacts only. */
650 find_anchor:
2c2d71f5 651 while (t < strend - prog->minlen) {
cad2e5aa 652 if (*t == '\n') {
4ee3650e 653 if (t < check_at - prog->check_offset_min) {
30944b6d 654 if (prog->anchored_substr) {
4ee3650e
GS
655 /* Since we moved from the found position,
656 we definitely contradict the found anchored
30944b6d
IZ
657 substr. Due to the above check we do not
658 contradict "check" substr.
659 Thus we can arrive here only if check substr
660 is float. Redo checking for "other"=="fixed".
661 */
9041c2e3 662 strpos = t + 1;
30944b6d
IZ
663 DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
664 PL_colors[0],PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
665 goto do_other_anchored;
666 }
4ee3650e
GS
667 /* We don't contradict the found floating substring. */
668 /* XXXX Why not check for STCLASS? */
cad2e5aa 669 s = t + 1;
2c2d71f5 670 DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
30944b6d 671 PL_colors[0],PL_colors[1], (long)(s - i_strpos)));
cad2e5aa
JH
672 goto set_useful;
673 }
4ee3650e
GS
674 /* Position contradicts check-string */
675 /* XXXX probably better to look for check-string
676 than for "\n", so one should lower the limit for t? */
677 DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
30944b6d 678 PL_colors[0],PL_colors[1], (long)(t + 1 - i_strpos)));
0e41cd87 679 other_last = strpos = s = t + 1;
cad2e5aa
JH
680 goto restart;
681 }
682 t++;
683 }
2c2d71f5
JH
684 DEBUG_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
685 PL_colors[0],PL_colors[1]));
686 goto fail_finish;
cad2e5aa 687 }
f5952150
GS
688 else {
689 DEBUG_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
690 PL_colors[0],PL_colors[1]));
691 }
cad2e5aa
JH
692 s = t;
693 set_useful:
2c2d71f5 694 ++BmUSEFUL(prog->check_substr); /* hooray/5 */
cad2e5aa
JH
695 }
696 else {
f5952150 697 /* The found string does not prohibit matching at strpos,
2c2d71f5 698 - no optimization of calling REx engine can be performed,
f5952150
GS
699 unless it was an MBOL and we are not after MBOL,
700 or a future STCLASS check will fail this. */
2c2d71f5
JH
701 try_at_start:
702 /* Even in this situation we may use MBOL flag if strpos is offset
703 wrt the start of the string. */
05b4157f 704 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
a1933d95 705 && (strpos != strbeg) && strpos[-1] != '\n'
d506a20d
IZ
706 /* May be due to an implicit anchor of m{.*foo} */
707 && !(prog->reganch & ROPT_IMPLICIT))
708 {
cad2e5aa
JH
709 t = strpos;
710 goto find_anchor;
711 }
30944b6d 712 DEBUG_r( if (ml_anch)
f5952150
GS
713 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
714 (long)(strpos - i_strpos), PL_colors[0],PL_colors[1]);
30944b6d 715 );
2c2d71f5 716 success_at_start:
30944b6d 717 if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */
66e933ab 718 && prog->check_substr /* Could be deleted already */
cad2e5aa 719 && --BmUSEFUL(prog->check_substr) < 0
66e933ab
GS
720 && prog->check_substr == prog->float_substr)
721 {
cad2e5aa 722 /* If flags & SOMETHING - do not do it many times on the same match */
f5952150 723 DEBUG_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
cad2e5aa
JH
724 SvREFCNT_dec(prog->check_substr);
725 prog->check_substr = Nullsv; /* disable */
726 prog->float_substr = Nullsv; /* clear */
5e39e1e5 727 check = Nullsv; /* abort */
cad2e5aa 728 s = strpos;
3cf5c195
IZ
729 /* XXXX This is a remnant of the old implementation. It
730 looks wasteful, since now INTUIT can use many
6eb5f6b9 731 other heuristics. */
cad2e5aa
JH
732 prog->reganch &= ~RE_USE_INTUIT;
733 }
734 else
735 s = strpos;
736 }
737
6eb5f6b9
JH
738 /* Last resort... */
739 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
740 if (prog->regstclass) {
741 /* minlen == 0 is possible if regstclass is \b or \B,
742 and the fixed substr is ''$.
743 Since minlen is already taken into account, s+1 is before strend;
744 accidentally, minlen >= 1 guaranties no false positives at s + 1
745 even for \b or \B. But (minlen? 1 : 0) below assumes that
746 regstclass does not come from lookahead... */
747 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
748 This leaves EXACTF only, which is dealt with in find_byclass(). */
1aa99e6b 749 U8* str = (U8*)STRING(prog->regstclass);
66e933ab 750 int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT
1aa99e6b 751 ? CHR_DIST(str+STR_LEN(prog->regstclass), str)
66e933ab 752 : 1);
6eb5f6b9 753 char *endpos = (prog->anchored_substr || ml_anch)
1aa99e6b
IH
754 ? HOP3c(s, (prog->minlen ? cl_l : 0), strend)
755 : (prog->float_substr
756 ? HOP3c(HOP3c(check_at, -start_shift, strbeg),
757 cl_l, strend)
758 : strend);
a1933d95 759 char *startpos = strbeg;
6eb5f6b9
JH
760
761 t = s;
76384e4a 762 if (prog->reganch & ROPT_UTF8) {
ffc61ed2 763 PL_regdata = prog->data;
76384e4a
GS
764 PL_bostr = startpos;
765 }
6eb5f6b9
JH
766 s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1);
767 if (!s) {
768#ifdef DEBUGGING
769 char *what;
770#endif
771 if (endpos == strend) {
772 DEBUG_r( PerlIO_printf(Perl_debug_log,
773 "Could not match STCLASS...\n") );
774 goto fail;
775 }
66e933ab
GS
776 DEBUG_r( PerlIO_printf(Perl_debug_log,
777 "This position contradicts STCLASS...\n") );
653099ff
GS
778 if ((prog->reganch & ROPT_ANCH) && !ml_anch)
779 goto fail;
6eb5f6b9
JH
780 /* Contradict one of substrings */
781 if (prog->anchored_substr) {
6eb5f6b9
JH
782 if (prog->anchored_substr == check) {
783 DEBUG_r( what = "anchored" );
784 hop_and_restart:
1aa99e6b 785 s = HOP3c(t, 1, strend);
66e933ab
GS
786 if (s + start_shift + end_shift > strend) {
787 /* XXXX Should be taken into account earlier? */
788 DEBUG_r( PerlIO_printf(Perl_debug_log,
789 "Could not match STCLASS...\n") );
790 goto fail;
791 }
5e39e1e5
HS
792 if (!check)
793 goto giveup;
6eb5f6b9 794 DEBUG_r( PerlIO_printf(Perl_debug_log,
f5952150 795 "Looking for %s substr starting at offset %ld...\n",
6eb5f6b9
JH
796 what, (long)(s + start_shift - i_strpos)) );
797 goto restart;
798 }
66e933ab 799 /* Have both, check_string is floating */
6eb5f6b9
JH
800 if (t + start_shift >= check_at) /* Contradicts floating=check */
801 goto retry_floating_check;
802 /* Recheck anchored substring, but not floating... */
9041c2e3 803 s = check_at;
5e39e1e5
HS
804 if (!check)
805 goto giveup;
6eb5f6b9 806 DEBUG_r( PerlIO_printf(Perl_debug_log,
f5952150 807 "Looking for anchored substr starting at offset %ld...\n",
6eb5f6b9
JH
808 (long)(other_last - i_strpos)) );
809 goto do_other_anchored;
810 }
60e71179
GS
811 /* Another way we could have checked stclass at the
812 current position only: */
813 if (ml_anch) {
814 s = t = t + 1;
5e39e1e5
HS
815 if (!check)
816 goto giveup;
60e71179 817 DEBUG_r( PerlIO_printf(Perl_debug_log,
f5952150
GS
818 "Looking for /%s^%s/m starting at offset %ld...\n",
819 PL_colors[0],PL_colors[1], (long)(t - i_strpos)) );
60e71179 820 goto try_at_offset;
66e933ab 821 }
60e71179
GS
822 if (!prog->float_substr) /* Could have been deleted */
823 goto fail;
6eb5f6b9
JH
824 /* Check is floating subtring. */
825 retry_floating_check:
826 t = check_at - start_shift;
827 DEBUG_r( what = "floating" );
828 goto hop_and_restart;
829 }
830 DEBUG_r( if (t != s)
9041c2e3 831 PerlIO_printf(Perl_debug_log,
6eb5f6b9
JH
832 "By STCLASS: moving %ld --> %ld\n",
833 (long)(t - i_strpos), (long)(s - i_strpos));
834 else
9041c2e3 835 PerlIO_printf(Perl_debug_log,
6eb5f6b9
JH
836 "Does not contradict STCLASS...\n") );
837 }
5e39e1e5
HS
838 giveup:
839 DEBUG_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
840 PL_colors[4], (check ? "Guessed" : "Giving up"),
841 PL_colors[5], (long)(s - i_strpos)) );
cad2e5aa 842 return s;
2c2d71f5
JH
843
844 fail_finish: /* Substring not found */
66e933ab
GS
845 if (prog->check_substr) /* could be removed already */
846 BmUSEFUL(prog->check_substr) += 5; /* hooray */
cad2e5aa 847 fail:
2c2d71f5 848 DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
cad2e5aa
JH
849 PL_colors[4],PL_colors[5]));
850 return Nullch;
851}
9661b544 852
6eb5f6b9 853/* We know what class REx starts with. Try to find this position... */
3c3eec57
GS
854STATIC char *
855S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *startpos, I32 norun)
a687059c 856{
6eb5f6b9
JH
857 I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
858 char *m;
d8093b23 859 STRLEN ln;
860 unsigned int c1;
861 unsigned int c2;
6eb5f6b9
JH
862 char *e;
863 register I32 tmp = 1; /* Scratch variable? */
ffc61ed2 864 register bool do_utf8 = DO_UTF8(PL_reg_sv);
cad2e5aa 865
6eb5f6b9
JH
866 /* We know what class it must start with. */
867 switch (OP(c)) {
6eb5f6b9 868 case ANYOF:
a0ed51b3 869 while (s < strend) {
ffc61ed2 870 if (reginclass(c, (U8*)s, do_utf8)) {
6eb5f6b9
JH
871 if (tmp && (norun || regtry(prog, s)))
872 goto got_it;
873 else
874 tmp = doevery;
a0ed51b3 875 }
6eb5f6b9
JH
876 else
877 tmp = 1;
ffc61ed2 878 s += do_utf8 ? UTF8SKIP(s) : 1;
a0d0e21e 879 }
6eb5f6b9
JH
880 break;
881 case EXACTF:
882 m = STRING(c);
883 ln = STR_LEN(c);
1aa99e6b
IH
884 if (UTF) {
885 c1 = to_utf8_lower((U8*)m);
886 c2 = to_utf8_upper((U8*)m);
887 }
888 else {
889 c1 = *(U8*)m;
890 c2 = PL_fold[c1];
891 }
6eb5f6b9
JH
892 goto do_exactf;
893 case EXACTFL:
894 m = STRING(c);
895 ln = STR_LEN(c);
d8093b23 896 c1 = *(U8*)m;
6eb5f6b9
JH
897 c2 = PL_fold_locale[c1];
898 do_exactf:
899 e = strend - ln;
b3c9acc1 900
6eb5f6b9
JH
901 if (norun && e < s)
902 e = s; /* Due to minlen logic of intuit() */
1aa99e6b
IH
903
904 if (do_utf8) {
905 STRLEN len;
906 if (c1 == c2)
907 while (s <= e) {
9041c2e3 908 if ( utf8_to_uvchr((U8*)s, &len) == c1
1aa99e6b
IH
909 && regtry(prog, s) )
910 goto got_it;
911 s += len;
912 }
913 else
914 while (s <= e) {
9041c2e3 915 UV c = utf8_to_uvchr((U8*)s, &len);
1aa99e6b
IH
916 if ( (c == c1 || c == c2) && regtry(prog, s) )
917 goto got_it;
918 s += len;
919 }
920 }
921 else {
922 if (c1 == c2)
923 while (s <= e) {
924 if ( *(U8*)s == c1
925 && (ln == 1 || !(OP(c) == EXACTF
926 ? ibcmp(s, m, ln)
927 : ibcmp_locale(s, m, ln)))
928 && (norun || regtry(prog, s)) )
929 goto got_it;
930 s++;
931 }
932 else
933 while (s <= e) {
934 if ( (*(U8*)s == c1 || *(U8*)s == c2)
935 && (ln == 1 || !(OP(c) == EXACTF
936 ? ibcmp(s, m, ln)
937 : ibcmp_locale(s, m, ln)))
938 && (norun || regtry(prog, s)) )
939 goto got_it;
940 s++;
941 }
b3c9acc1
IZ
942 }
943 break;
bbce6d69 944 case BOUNDL:
3280af22 945 PL_reg_flags |= RF_tainted;
bbce6d69 946 /* FALL THROUGH */
a0d0e21e 947 case BOUND:
ffc61ed2
JH
948 if (do_utf8) {
949 if (s == startpos)
950 tmp = '\n';
951 else {
1aa99e6b 952 U8 *r = reghop3((U8*)s, -1, (U8*)startpos);
9041c2e3 953
2b9d42f0 954 tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0);
ffc61ed2
JH
955 }
956 tmp = ((OP(c) == BOUND ?
9041c2e3 957 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
8269fa76 958 LOAD_UTF8_CHARCLASS(alnum,"a");
ffc61ed2
JH
959 while (s < strend) {
960 if (tmp == !(OP(c) == BOUND ?
961 swash_fetch(PL_utf8_alnum, (U8*)s) :
962 isALNUM_LC_utf8((U8*)s)))
963 {
964 tmp = !tmp;
965 if ((norun || regtry(prog, s)))
966 goto got_it;
967 }
968 s += UTF8SKIP(s);
a687059c 969 }
a0d0e21e 970 }
667bb95a 971 else {
ffc61ed2
JH
972 tmp = (s != startpos) ? UCHARAT(s - 1) : '\n';
973 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
974 while (s < strend) {
975 if (tmp ==
976 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
977 tmp = !tmp;
978 if ((norun || regtry(prog, s)))
979 goto got_it;
980 }
981 s++;
a0ed51b3 982 }
a0ed51b3 983 }
6eb5f6b9 984 if ((!prog->minlen && tmp) && (norun || regtry(prog, s)))
a0ed51b3
LW
985 goto got_it;
986 break;
bbce6d69 987 case NBOUNDL:
3280af22 988 PL_reg_flags |= RF_tainted;
bbce6d69 989 /* FALL THROUGH */
a0d0e21e 990 case NBOUND:
ffc61ed2
JH
991 if (do_utf8) {
992 if (s == startpos)
993 tmp = '\n';
994 else {
1aa99e6b 995 U8 *r = reghop3((U8*)s, -1, (U8*)startpos);
9041c2e3 996
2b9d42f0 997 tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0);
ffc61ed2
JH
998 }
999 tmp = ((OP(c) == NBOUND ?
9041c2e3 1000 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
8269fa76 1001 LOAD_UTF8_CHARCLASS(alnum,"a");
ffc61ed2
JH
1002 while (s < strend) {
1003 if (tmp == !(OP(c) == NBOUND ?
1004 swash_fetch(PL_utf8_alnum, (U8*)s) :
1005 isALNUM_LC_utf8((U8*)s)))
1006 tmp = !tmp;
1007 else if ((norun || regtry(prog, s)))
1008 goto got_it;
1009 s += UTF8SKIP(s);
1010 }
a0d0e21e 1011 }
667bb95a 1012 else {
ffc61ed2
JH
1013 tmp = (s != startpos) ? UCHARAT(s - 1) : '\n';
1014 tmp = ((OP(c) == NBOUND ?
1015 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1016 while (s < strend) {
1017 if (tmp ==
1018 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1019 tmp = !tmp;
1020 else if ((norun || regtry(prog, s)))
1021 goto got_it;
1022 s++;
1023 }
a0ed51b3 1024 }
6eb5f6b9 1025 if ((!prog->minlen && !tmp) && (norun || regtry(prog, s)))
a0ed51b3
LW
1026 goto got_it;
1027 break;
a0d0e21e 1028 case ALNUM:
ffc61ed2 1029 if (do_utf8) {
8269fa76 1030 LOAD_UTF8_CHARCLASS(alnum,"a");
ffc61ed2
JH
1031 while (s < strend) {
1032 if (swash_fetch(PL_utf8_alnum, (U8*)s)) {
1033 if (tmp && (norun || regtry(prog, s)))
1034 goto got_it;
1035 else
1036 tmp = doevery;
1037 }
bbce6d69 1038 else
ffc61ed2
JH
1039 tmp = 1;
1040 s += UTF8SKIP(s);
bbce6d69 1041 }
bbce6d69 1042 }
ffc61ed2
JH
1043 else {
1044 while (s < strend) {
1045 if (isALNUM(*s)) {
1046 if (tmp && (norun || regtry(prog, s)))
1047 goto got_it;
1048 else
1049 tmp = doevery;
1050 }
a0ed51b3 1051 else
ffc61ed2
JH
1052 tmp = 1;
1053 s++;
a0ed51b3 1054 }
a0ed51b3
LW
1055 }
1056 break;
bbce6d69 1057 case ALNUML:
3280af22 1058 PL_reg_flags |= RF_tainted;
ffc61ed2
JH
1059 if (do_utf8) {
1060 while (s < strend) {
1061 if (isALNUM_LC_utf8((U8*)s)) {
1062 if (tmp && (norun || regtry(prog, s)))
1063 goto got_it;
1064 else
1065 tmp = doevery;
1066 }
a687059c 1067 else
ffc61ed2
JH
1068 tmp = 1;
1069 s += UTF8SKIP(s);
a0d0e21e 1070 }
a0d0e21e 1071 }
ffc61ed2
JH
1072 else {
1073 while (s < strend) {
1074 if (isALNUM_LC(*s)) {
1075 if (tmp && (norun || regtry(prog, s)))
1076 goto got_it;
1077 else
1078 tmp = doevery;
1079 }
a0ed51b3 1080 else
ffc61ed2
JH
1081 tmp = 1;
1082 s++;
a0ed51b3 1083 }
a0ed51b3
LW
1084 }
1085 break;
a0d0e21e 1086 case NALNUM:
ffc61ed2 1087 if (do_utf8) {
8269fa76 1088 LOAD_UTF8_CHARCLASS(alnum,"a");
ffc61ed2
JH
1089 while (s < strend) {
1090 if (!swash_fetch(PL_utf8_alnum, (U8*)s)) {
1091 if (tmp && (norun || regtry(prog, s)))
1092 goto got_it;
1093 else
1094 tmp = doevery;
1095 }
bbce6d69 1096 else
ffc61ed2
JH
1097 tmp = 1;
1098 s += UTF8SKIP(s);
bbce6d69 1099 }
bbce6d69 1100 }
ffc61ed2
JH
1101 else {
1102 while (s < strend) {
1103 if (!isALNUM(*s)) {
1104 if (tmp && (norun || regtry(prog, s)))
1105 goto got_it;
1106 else
1107 tmp = doevery;
1108 }
a0ed51b3 1109 else
ffc61ed2
JH
1110 tmp = 1;
1111 s++;
a0ed51b3 1112 }
a0ed51b3
LW
1113 }
1114 break;
bbce6d69 1115 case NALNUML:
3280af22 1116 PL_reg_flags |= RF_tainted;
ffc61ed2
JH
1117 if (do_utf8) {
1118 while (s < strend) {
1119 if (!isALNUM_LC_utf8((U8*)s)) {
1120 if (tmp && (norun || regtry(prog, s)))
1121 goto got_it;
1122 else
1123 tmp = doevery;
1124 }
a687059c 1125 else
ffc61ed2
JH
1126 tmp = 1;
1127 s += UTF8SKIP(s);
a687059c 1128 }
a0d0e21e 1129 }
ffc61ed2
JH
1130 else {
1131 while (s < strend) {
1132 if (!isALNUM_LC(*s)) {
1133 if (tmp && (norun || regtry(prog, s)))
1134 goto got_it;
1135 else
1136 tmp = doevery;
1137 }
a0ed51b3 1138 else
ffc61ed2
JH
1139 tmp = 1;
1140 s++;
a0ed51b3 1141 }
a0ed51b3
LW
1142 }
1143 break;
a0d0e21e 1144 case SPACE:
ffc61ed2 1145 if (do_utf8) {
8269fa76 1146 LOAD_UTF8_CHARCLASS(space," ");
ffc61ed2
JH
1147 while (s < strend) {
1148 if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) {
1149 if (tmp && (norun || regtry(prog, s)))
1150 goto got_it;
1151 else
1152 tmp = doevery;
1153 }
a0d0e21e 1154 else
ffc61ed2
JH
1155 tmp = 1;
1156 s += UTF8SKIP(s);
2304df62 1157 }
a0d0e21e 1158 }
ffc61ed2
JH
1159 else {
1160 while (s < strend) {
1161 if (isSPACE(*s)) {
1162 if (tmp && (norun || regtry(prog, s)))
1163 goto got_it;
1164 else
1165 tmp = doevery;
1166 }
a0ed51b3 1167 else
ffc61ed2
JH
1168 tmp = 1;
1169 s++;
a0ed51b3 1170 }
a0ed51b3
LW
1171 }
1172 break;
bbce6d69 1173 case SPACEL:
3280af22 1174 PL_reg_flags |= RF_tainted;
ffc61ed2
JH
1175 if (do_utf8) {
1176 while (s < strend) {
1177 if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
1178 if (tmp && (norun || regtry(prog, s)))
1179 goto got_it;
1180 else
1181 tmp = doevery;
1182 }
bbce6d69 1183 else
ffc61ed2
JH
1184 tmp = 1;
1185 s += UTF8SKIP(s);
bbce6d69 1186 }
bbce6d69 1187 }
ffc61ed2
JH
1188 else {
1189 while (s < strend) {
1190 if (isSPACE_LC(*s)) {
1191 if (tmp && (norun || regtry(prog, s)))
1192 goto got_it;
1193 else
1194 tmp = doevery;
1195 }
a0ed51b3 1196 else
ffc61ed2
JH
1197 tmp = 1;
1198 s++;
a0ed51b3 1199 }
a0ed51b3
LW
1200 }
1201 break;
a0d0e21e 1202 case NSPACE:
ffc61ed2 1203 if (do_utf8) {
8269fa76 1204 LOAD_UTF8_CHARCLASS(space," ");
ffc61ed2
JH
1205 while (s < strend) {
1206 if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) {
1207 if (tmp && (norun || regtry(prog, s)))
1208 goto got_it;
1209 else
1210 tmp = doevery;
1211 }
a0d0e21e 1212 else
ffc61ed2
JH
1213 tmp = 1;
1214 s += UTF8SKIP(s);
a687059c 1215 }
a0d0e21e 1216 }
ffc61ed2
JH
1217 else {
1218 while (s < strend) {
1219 if (!isSPACE(*s)) {
1220 if (tmp && (norun || regtry(prog, s)))
1221 goto got_it;
1222 else
1223 tmp = doevery;
1224 }
a0ed51b3 1225 else
ffc61ed2
JH
1226 tmp = 1;
1227 s++;
a0ed51b3 1228 }
a0ed51b3
LW
1229 }
1230 break;
bbce6d69 1231 case NSPACEL:
3280af22 1232 PL_reg_flags |= RF_tainted;
ffc61ed2
JH
1233 if (do_utf8) {
1234 while (s < strend) {
1235 if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
1236 if (tmp && (norun || regtry(prog, s)))
1237 goto got_it;
1238 else
1239 tmp = doevery;
1240 }
bbce6d69 1241 else
ffc61ed2
JH
1242 tmp = 1;
1243 s += UTF8SKIP(s);
bbce6d69 1244 }
bbce6d69 1245 }
ffc61ed2
JH
1246 else {
1247 while (s < strend) {
1248 if (!isSPACE_LC(*s)) {
1249 if (tmp && (norun || regtry(prog, s)))
1250 goto got_it;
1251 else
1252 tmp = doevery;
1253 }
a0ed51b3 1254 else
ffc61ed2
JH
1255 tmp = 1;
1256 s++;
a0ed51b3 1257 }
a0ed51b3
LW
1258 }
1259 break;
a0d0e21e 1260 case DIGIT:
ffc61ed2 1261 if (do_utf8) {
8269fa76 1262 LOAD_UTF8_CHARCLASS(digit,"0");
ffc61ed2
JH
1263 while (s < strend) {
1264 if (swash_fetch(PL_utf8_digit,(U8*)s)) {
1265 if (tmp && (norun || regtry(prog, s)))
1266 goto got_it;
1267 else
1268 tmp = doevery;
1269 }
a0d0e21e 1270 else
ffc61ed2
JH
1271 tmp = 1;
1272 s += UTF8SKIP(s);
2b69d0c2 1273 }
a0d0e21e 1274 }
ffc61ed2
JH
1275 else {
1276 while (s < strend) {
1277 if (isDIGIT(*s)) {
1278 if (tmp && (norun || regtry(prog, s)))
1279 goto got_it;
1280 else
1281 tmp = doevery;
1282 }
a0ed51b3 1283 else
ffc61ed2
JH
1284 tmp = 1;
1285 s++;
a0ed51b3 1286 }
a0ed51b3
LW
1287 }
1288 break;
b8c5462f
JH
1289 case DIGITL:
1290 PL_reg_flags |= RF_tainted;
ffc61ed2
JH
1291 if (do_utf8) {
1292 while (s < strend) {
1293 if (isDIGIT_LC_utf8((U8*)s)) {
1294 if (tmp && (norun || regtry(prog, s)))
1295 goto got_it;
1296 else
1297 tmp = doevery;
1298 }
b8c5462f 1299 else
ffc61ed2
JH
1300 tmp = 1;
1301 s += UTF8SKIP(s);
b8c5462f 1302 }
b8c5462f 1303 }
ffc61ed2
JH
1304 else {
1305 while (s < strend) {
1306 if (isDIGIT_LC(*s)) {
1307 if (tmp && (norun || regtry(prog, s)))
1308 goto got_it;
1309 else
1310 tmp = doevery;
1311 }
b8c5462f 1312 else
ffc61ed2
JH
1313 tmp = 1;
1314 s++;
b8c5462f 1315 }
b8c5462f
JH
1316 }
1317 break;
a0d0e21e 1318 case NDIGIT:
ffc61ed2 1319 if (do_utf8) {
8269fa76 1320 LOAD_UTF8_CHARCLASS(digit,"0");
ffc61ed2
JH
1321 while (s < strend) {
1322 if (!swash_fetch(PL_utf8_digit,(U8*)s)) {
1323 if (tmp && (norun || regtry(prog, s)))
1324 goto got_it;
1325 else
1326 tmp = doevery;
1327 }
a0d0e21e 1328 else
ffc61ed2
JH
1329 tmp = 1;
1330 s += UTF8SKIP(s);
a687059c 1331 }
a0d0e21e 1332 }
ffc61ed2
JH
1333 else {
1334 while (s < strend) {
1335 if (!isDIGIT(*s)) {
1336 if (tmp && (norun || regtry(prog, s)))
1337 goto got_it;
1338 else
1339 tmp = doevery;
1340 }
a0ed51b3 1341 else
ffc61ed2
JH
1342 tmp = 1;
1343 s++;
a0ed51b3 1344 }
a0ed51b3
LW
1345 }
1346 break;
b8c5462f
JH
1347 case NDIGITL:
1348 PL_reg_flags |= RF_tainted;
ffc61ed2
JH
1349 if (do_utf8) {
1350 while (s < strend) {
1351 if (!isDIGIT_LC_utf8((U8*)s)) {
1352 if (tmp && (norun || regtry(prog, s)))
1353 goto got_it;
1354 else
1355 tmp = doevery;
1356 }
b8c5462f 1357 else
ffc61ed2
JH
1358 tmp = 1;
1359 s += UTF8SKIP(s);
b8c5462f 1360 }
a0ed51b3 1361 }
ffc61ed2
JH
1362 else {
1363 while (s < strend) {
1364 if (!isDIGIT_LC(*s)) {
1365 if (tmp && (norun || regtry(prog, s)))
1366 goto got_it;
1367 else
1368 tmp = doevery;
1369 }
cf93c79d 1370 else
ffc61ed2
JH
1371 tmp = 1;
1372 s++;
b8c5462f 1373 }
b8c5462f
JH
1374 }
1375 break;
b3c9acc1 1376 default:
3c3eec57
GS
1377 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
1378 break;
d6a28714 1379 }
6eb5f6b9
JH
1380 return 0;
1381 got_it:
1382 return s;
1383}
1384
1385/*
1386 - regexec_flags - match a regexp against a string
1387 */
1388I32
1389Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend,
1390 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
1391/* strend: pointer to null at end of string */
1392/* strbeg: real beginning of string */
1393/* minend: end of match must be >=minend after stringarg. */
1394/* data: May be used for some additional optimizations. */
1395/* nosave: For optimizations. */
1396{
6eb5f6b9
JH
1397 register char *s;
1398 register regnode *c;
1399 register char *startpos = stringarg;
6eb5f6b9
JH
1400 I32 minlen; /* must match at least this many chars */
1401 I32 dontbother = 0; /* how many characters not to try at end */
155aba94 1402 /* I32 start_shift = 0; */ /* Offset of the start to find
6eb5f6b9
JH
1403 constant substr. */ /* CC */
1404 I32 end_shift = 0; /* Same for the end. */ /* CC */
1405 I32 scream_pos = -1; /* Internal iterator of scream. */
1406 char *scream_olds;
1407 SV* oreplsv = GvSV(PL_replgv);
1aa99e6b 1408 bool do_utf8 = DO_UTF8(sv);
6eb5f6b9
JH
1409
1410 PL_regcc = 0;
1411
1412 cache_re(prog);
1413#ifdef DEBUGGING
aea4f609 1414 PL_regnarrate = DEBUG_r_TEST;
6eb5f6b9
JH
1415#endif
1416
1417 /* Be paranoid... */
1418 if (prog == NULL || startpos == NULL) {
1419 Perl_croak(aTHX_ "NULL regexp parameter");
1420 return 0;
1421 }
1422
1423 minlen = prog->minlen;
1aa99e6b 1424 if (do_utf8) {
3baa4c62
JH
1425 if (!(prog->reganch & ROPT_SANY_SEEN))
1426 if (utf8_distance((U8*)strend, (U8*)startpos) < minlen) goto phooey;
1aa99e6b
IH
1427 }
1428 else {
1429 if (strend - startpos < minlen) goto phooey;
1430 }
6eb5f6b9
JH
1431
1432 if (startpos == strbeg) /* is ^ valid at stringarg? */
1433 PL_regprev = '\n';
1434 else {
1aa99e6b
IH
1435 if (prog->reganch & ROPT_UTF8 && do_utf8) {
1436 U8 *s = reghop3((U8*)stringarg, -1, (U8*)strbeg);
2b9d42f0 1437 PL_regprev = utf8n_to_uvchr(s, (U8*)stringarg - s, NULL, 0);
1aa99e6b
IH
1438 }
1439 else
1440 PL_regprev = (U32)stringarg[-1];
6eb5f6b9
JH
1441 if (!PL_multiline && PL_regprev == '\n')
1442 PL_regprev = '\0'; /* force ^ to NOT match */
1443 }
1444
1445 /* Check validity of program. */
1446 if (UCHARAT(prog->program) != REG_MAGIC) {
1447 Perl_croak(aTHX_ "corrupted regexp program");
1448 }
1449
1450 PL_reg_flags = 0;
1451 PL_reg_eval_set = 0;
1452 PL_reg_maxiter = 0;
1453
1454 if (prog->reganch & ROPT_UTF8)
1455 PL_reg_flags |= RF_utf8;
1456
1457 /* Mark beginning of line for ^ and lookbehind. */
1458 PL_regbol = startpos;
1459 PL_bostr = strbeg;
1460 PL_reg_sv = sv;
1461
1462 /* Mark end of line for $ (and such) */
1463 PL_regeol = strend;
1464
1465 /* see how far we have to get to not match where we matched before */
1466 PL_regtill = startpos+minend;
1467
1468 /* We start without call_cc context. */
1469 PL_reg_call_cc = 0;
1470
1471 /* If there is a "must appear" string, look for it. */
1472 s = startpos;
1473
1474 if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to have PL_reg_ganch */
1475 MAGIC *mg;
1476
1477 if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
1478 PL_reg_ganch = startpos;
1479 else if (sv && SvTYPE(sv) >= SVt_PVMG
1480 && SvMAGIC(sv)
1481 && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) {
1482 PL_reg_ganch = strbeg + mg->mg_len; /* Defined pos() */
1483 if (prog->reganch & ROPT_ANCH_GPOS) {
1484 if (s > PL_reg_ganch)
1485 goto phooey;
1486 s = PL_reg_ganch;
1487 }
1488 }
1489 else /* pos() not defined */
1490 PL_reg_ganch = strbeg;
1491 }
1492
699c3c34
JH
1493 if (do_utf8 == (UTF!=0) &&
1494 !(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) {
6eb5f6b9
JH
1495 re_scream_pos_data d;
1496
1497 d.scream_olds = &scream_olds;
1498 d.scream_pos = &scream_pos;
1499 s = re_intuit_start(prog, sv, s, strend, flags, &d);
1500 if (!s)
1501 goto phooey; /* not present */
1502 }
1503
1504 DEBUG_r( if (!PL_colorset) reginitcolors() );
1505 DEBUG_r(PerlIO_printf(Perl_debug_log,
1506 "%sMatching REx%s `%s%.60s%s%s' against `%s%.*s%s%s'\n",
1507 PL_colors[4],PL_colors[5],PL_colors[0],
1508 prog->precomp,
1509 PL_colors[1],
1510 (strlen(prog->precomp) > 60 ? "..." : ""),
1511 PL_colors[0],
1512 (int)(strend - startpos > 60 ? 60 : strend - startpos),
1513 startpos, PL_colors[1],
1514 (strend - startpos > 60 ? "..." : ""))
1515 );
1516
1517 /* Simplest case: anchored match need be tried only once. */
1518 /* [unless only anchor is BOL and multiline is set] */
1519 if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
1520 if (s == startpos && regtry(prog, startpos))
1521 goto got_it;
1522 else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
1523 || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
1524 {
1525 char *end;
1526
1527 if (minlen)
1528 dontbother = minlen - 1;
1aa99e6b 1529 end = HOP3c(strend, -dontbother, strbeg) - 1;
6eb5f6b9
JH
1530 /* for multiline we only have to try after newlines */
1531 if (prog->check_substr) {
1532 if (s == startpos)
1533 goto after_try;
1534 while (1) {
1535 if (regtry(prog, s))
1536 goto got_it;
1537 after_try:
1538 if (s >= end)
1539 goto phooey;
1540 if (prog->reganch & RE_USE_INTUIT) {
1541 s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
1542 if (!s)
1543 goto phooey;
1544 }
1545 else
1546 s++;
1547 }
1548 } else {
1549 if (s > startpos)
1550 s--;
1551 while (s < end) {
1552 if (*s++ == '\n') { /* don't need PL_utf8skip here */
1553 if (regtry(prog, s))
1554 goto got_it;
1555 }
1556 }
1557 }
1558 }
1559 goto phooey;
1560 } else if (prog->reganch & ROPT_ANCH_GPOS) {
1561 if (regtry(prog, PL_reg_ganch))
1562 goto got_it;
1563 goto phooey;
1564 }
1565
1566 /* Messy cases: unanchored match. */
9041c2e3 1567 if (prog->anchored_substr && prog->reganch & ROPT_SKIP) {
6eb5f6b9
JH
1568 /* we have /x+whatever/ */
1569 /* it must be a one character string (XXXX Except UTF?) */
1570 char ch = SvPVX(prog->anchored_substr)[0];
bf93d4cc
GS
1571#ifdef DEBUGGING
1572 int did_match = 0;
1573#endif
1574
1aa99e6b 1575 if (do_utf8) {
6eb5f6b9
JH
1576 while (s < strend) {
1577 if (*s == ch) {
bf93d4cc 1578 DEBUG_r( did_match = 1 );
6eb5f6b9
JH
1579 if (regtry(prog, s)) goto got_it;
1580 s += UTF8SKIP(s);
1581 while (s < strend && *s == ch)
1582 s += UTF8SKIP(s);
1583 }
1584 s += UTF8SKIP(s);
1585 }
1586 }
1587 else {
1588 while (s < strend) {
1589 if (*s == ch) {
bf93d4cc 1590 DEBUG_r( did_match = 1 );
6eb5f6b9
JH
1591 if (regtry(prog, s)) goto got_it;
1592 s++;
1593 while (s < strend && *s == ch)
1594 s++;
1595 }
1596 s++;
1597 }
1598 }
bf93d4cc
GS
1599 DEBUG_r(did_match ||
1600 PerlIO_printf(Perl_debug_log,
1601 "Did not find anchored character...\n"));
6eb5f6b9
JH
1602 }
1603 /*SUPPRESS 560*/
1aa99e6b
IH
1604 else if (do_utf8 == (UTF!=0) &&
1605 (prog->anchored_substr != Nullsv
9041c2e3 1606 || (prog->float_substr != Nullsv
1aa99e6b 1607 && prog->float_max_offset < strend - s))) {
9041c2e3 1608 SV *must = prog->anchored_substr
6eb5f6b9 1609 ? prog->anchored_substr : prog->float_substr;
9041c2e3 1610 I32 back_max =
6eb5f6b9 1611 prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset;
9041c2e3 1612 I32 back_min =
6eb5f6b9 1613 prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset;
1aa99e6b 1614 char *last = HOP3c(strend, /* Cannot start after this */
6eb5f6b9 1615 -(I32)(CHR_SVLEN(must)
1aa99e6b 1616 - (SvTAIL(must) != 0) + back_min), strbeg);
6eb5f6b9 1617 char *last1; /* Last position checked before */
bf93d4cc
GS
1618#ifdef DEBUGGING
1619 int did_match = 0;
1620#endif
6eb5f6b9
JH
1621
1622 if (s > PL_bostr)
1623 last1 = HOPc(s, -1);
1624 else
1625 last1 = s - 1; /* bogus */
1626
1627 /* XXXX check_substr already used to find `s', can optimize if
1628 check_substr==must. */
1629 scream_pos = -1;
1630 dontbother = end_shift;
1631 strend = HOPc(strend, -dontbother);
1632 while ( (s <= last) &&
9041c2e3 1633 ((flags & REXEC_SCREAM)
1aa99e6b 1634 ? (s = screaminstr(sv, must, HOP3c(s, back_min, strend) - strbeg,
6eb5f6b9 1635 end_shift, &scream_pos, 0))
1aa99e6b 1636 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend),
9041c2e3 1637 (unsigned char*)strend, must,
6eb5f6b9 1638 PL_multiline ? FBMrf_MULTILINE : 0))) ) {
bf93d4cc 1639 DEBUG_r( did_match = 1 );
6eb5f6b9
JH
1640 if (HOPc(s, -back_max) > last1) {
1641 last1 = HOPc(s, -back_min);
1642 s = HOPc(s, -back_max);
1643 }
1644 else {
1645 char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
1646
1647 last1 = HOPc(s, -back_min);
1648 s = t;
1649 }
1aa99e6b 1650 if (do_utf8) {
6eb5f6b9
JH
1651 while (s <= last1) {
1652 if (regtry(prog, s))
1653 goto got_it;
1654 s += UTF8SKIP(s);
1655 }
1656 }
1657 else {
1658 while (s <= last1) {
1659 if (regtry(prog, s))
1660 goto got_it;
1661 s++;
1662 }
1663 }
1664 }
bf93d4cc
GS
1665 DEBUG_r(did_match ||
1666 PerlIO_printf(Perl_debug_log, "Did not find %s substr `%s%.*s%s'%s...\n",
1667 ((must == prog->anchored_substr)
1668 ? "anchored" : "floating"),
1669 PL_colors[0],
1670 (int)(SvCUR(must) - (SvTAIL(must)!=0)),
1671 SvPVX(must),
1672 PL_colors[1], (SvTAIL(must) ? "$" : "")));
6eb5f6b9
JH
1673 goto phooey;
1674 }
155aba94 1675 else if ((c = prog->regstclass)) {
66e933ab
GS
1676 if (minlen && PL_regkind[(U8)OP(prog->regstclass)] != EXACT)
1677 /* don't bother with what can't match */
6eb5f6b9 1678 strend = HOPc(strend, -(minlen - 1));
ffc61ed2
JH
1679 DEBUG_r({
1680 SV *prop = sv_newmortal();
1681 regprop(prop, c);
1682 PerlIO_printf(Perl_debug_log, "Matching stclass `%s' against `%s'\n", SvPVX(prop), s);
1683 });
6eb5f6b9
JH
1684 if (find_byclass(prog, c, s, strend, startpos, 0))
1685 goto got_it;
bf93d4cc 1686 DEBUG_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n"));
d6a28714
JH
1687 }
1688 else {
1689 dontbother = 0;
1690 if (prog->float_substr != Nullsv) { /* Trim the end. */
1691 char *last;
d6a28714
JH
1692
1693 if (flags & REXEC_SCREAM) {
1694 last = screaminstr(sv, prog->float_substr, s - strbeg,
1695 end_shift, &scream_pos, 1); /* last one */
1696 if (!last)
ffc61ed2 1697 last = scream_olds; /* Only one occurrence. */
b8c5462f 1698 }
d6a28714
JH
1699 else {
1700 STRLEN len;
1701 char *little = SvPV(prog->float_substr, len);
1702
1703 if (SvTAIL(prog->float_substr)) {
1704 if (memEQ(strend - len + 1, little, len - 1))
1705 last = strend - len + 1;
1706 else if (!PL_multiline)
9041c2e3 1707 last = memEQ(strend - len, little, len)
d6a28714 1708 ? strend - len : Nullch;
b8c5462f 1709 else
d6a28714
JH
1710 goto find_last;
1711 } else {
1712 find_last:
9041c2e3 1713 if (len)
d6a28714 1714 last = rninstr(s, strend, little, little + len);
b8c5462f 1715 else
d6a28714 1716 last = strend; /* matching `$' */
b8c5462f 1717 }
b8c5462f 1718 }
bf93d4cc
GS
1719 if (last == NULL) {
1720 DEBUG_r(PerlIO_printf(Perl_debug_log,
1721 "%sCan't trim the tail, match fails (should not happen)%s\n",
1722 PL_colors[4],PL_colors[5]));
1723 goto phooey; /* Should not happen! */
1724 }
d6a28714
JH
1725 dontbother = strend - last + prog->float_min_offset;
1726 }
1727 if (minlen && (dontbother < minlen))
1728 dontbother = minlen - 1;
1729 strend -= dontbother; /* this one's always in bytes! */
1730 /* We don't know much -- general case. */
1aa99e6b 1731 if (do_utf8) {
d6a28714
JH
1732 for (;;) {
1733 if (regtry(prog, s))
1734 goto got_it;
1735 if (s >= strend)
1736 break;
b8c5462f 1737 s += UTF8SKIP(s);
d6a28714
JH
1738 };
1739 }
1740 else {
1741 do {
1742 if (regtry(prog, s))
1743 goto got_it;
1744 } while (s++ < strend);
1745 }
1746 }
1747
1748 /* Failure. */
1749 goto phooey;
1750
1751got_it:
1752 RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
1753
1754 if (PL_reg_eval_set) {
1755 /* Preserve the current value of $^R */
1756 if (oreplsv != GvSV(PL_replgv))
1757 sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
1758 restored, the value remains
1759 the same. */
1760 restore_pos(aTHXo_ 0);
1761 }
1762
1763 /* make sure $`, $&, $', and $digit will work later */
1764 if ( !(flags & REXEC_NOT_FIRST) ) {
1765 if (RX_MATCH_COPIED(prog)) {
1766 Safefree(prog->subbeg);
1767 RX_MATCH_COPIED_off(prog);
1768 }
1769 if (flags & REXEC_COPY_STR) {
1770 I32 i = PL_regeol - startpos + (stringarg - strbeg);
1771
1772 s = savepvn(strbeg, i);
1773 prog->subbeg = s;
1774 prog->sublen = i;
1775 RX_MATCH_COPIED_on(prog);
1776 }
1777 else {
1778 prog->subbeg = strbeg;
1779 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
1780 }
1781 }
9041c2e3 1782
d6a28714
JH
1783 return 1;
1784
1785phooey:
bf93d4cc
GS
1786 DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
1787 PL_colors[4],PL_colors[5]));
d6a28714
JH
1788 if (PL_reg_eval_set)
1789 restore_pos(aTHXo_ 0);
1790 return 0;
1791}
1792
1793/*
1794 - regtry - try match at specific point
1795 */
1796STATIC I32 /* 0 failure, 1 success */
1797S_regtry(pTHX_ regexp *prog, char *startpos)
1798{
d6a28714
JH
1799 register I32 i;
1800 register I32 *sp;
1801 register I32 *ep;
1802 CHECKPOINT lastcp;
1803
02db2b7b
IZ
1804#ifdef DEBUGGING
1805 PL_regindent = 0; /* XXXX Not good when matches are reenterable... */
1806#endif
d6a28714
JH
1807 if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
1808 MAGIC *mg;
1809
1810 PL_reg_eval_set = RS_init;
1811 DEBUG_r(DEBUG_s(
b900a521
JH
1812 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
1813 (IV)(PL_stack_sp - PL_stack_base));
d6a28714 1814 ));
e8347627 1815 SAVEI32(cxstack[cxstack_ix].blk_oldsp);
d6a28714
JH
1816 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
1817 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
1818 SAVETMPS;
1819 /* Apparently this is not needed, judging by wantarray. */
e8347627 1820 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
d6a28714
JH
1821 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
1822
1823 if (PL_reg_sv) {
1824 /* Make $_ available to executed code. */
1825 if (PL_reg_sv != DEFSV) {
1826 /* SAVE_DEFSV does *not* suffice here for USE_THREADS */
1827 SAVESPTR(DEFSV);
1828 DEFSV = PL_reg_sv;
b8c5462f 1829 }
d6a28714 1830
9041c2e3 1831 if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
d6a28714
JH
1832 && (mg = mg_find(PL_reg_sv, 'g')))) {
1833 /* prepare for quick setting of pos */
1834 sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0);
1835 mg = mg_find(PL_reg_sv, 'g');
1836 mg->mg_len = -1;
b8c5462f 1837 }
d6a28714
JH
1838 PL_reg_magic = mg;
1839 PL_reg_oldpos = mg->mg_len;
c76ac1ee 1840 SAVEDESTRUCTOR_X(restore_pos, 0);
d6a28714
JH
1841 }
1842 if (!PL_reg_curpm)
0f79a09d 1843 Newz(22,PL_reg_curpm, 1, PMOP);
d6a28714
JH
1844 PL_reg_curpm->op_pmregexp = prog;
1845 PL_reg_oldcurpm = PL_curpm;
1846 PL_curpm = PL_reg_curpm;
1847 if (RX_MATCH_COPIED(prog)) {
1848 /* Here is a serious problem: we cannot rewrite subbeg,
1849 since it may be needed if this match fails. Thus
1850 $` inside (?{}) could fail... */
1851 PL_reg_oldsaved = prog->subbeg;
1852 PL_reg_oldsavedlen = prog->sublen;
1853 RX_MATCH_COPIED_off(prog);
1854 }
1855 else
1856 PL_reg_oldsaved = Nullch;
1857 prog->subbeg = PL_bostr;
1858 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
1859 }
1860 prog->startp[0] = startpos - PL_bostr;
1861 PL_reginput = startpos;
1862 PL_regstartp = prog->startp;
1863 PL_regendp = prog->endp;
1864 PL_reglastparen = &prog->lastparen;
1865 prog->lastparen = 0;
1866 PL_regsize = 0;
1867 DEBUG_r(PL_reg_starttry = startpos);
1868 if (PL_reg_start_tmpl <= prog->nparens) {
1869 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
1870 if(PL_reg_start_tmp)
1871 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
1872 else
1873 New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*);
1874 }
1875
1876 /* XXXX What this code is doing here?!!! There should be no need
1877 to do this again and again, PL_reglastparen should take care of
3dd2943c 1878 this! --ilya*/
dafc8851
JH
1879
1880 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
1881 * Actually, the code in regcppop() (which Ilya may be meaning by
daf18116
JH
1882 * PL_reglastparen), is not needed at all by the test suite
1883 * (op/regexp, op/pat, op/split), but that code is needed, oddly
1884 * enough, for building DynaLoader, or otherwise this
1885 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
1886 * will happen. Meanwhile, this code *is* needed for the
1887 * above-mentioned test suite tests to succeed. The common theme
1888 * on those tests seems to be returning null fields from matches.
1889 * --jhi */
dafc8851 1890#if 1
d6a28714
JH
1891 sp = prog->startp;
1892 ep = prog->endp;
1893 if (prog->nparens) {
09e8ae3b 1894 for (i = prog->nparens; i > *PL_reglastparen; i--) {
d6a28714
JH
1895 *++sp = -1;
1896 *++ep = -1;
1897 }
1898 }
dafc8851 1899#endif
02db2b7b 1900 REGCP_SET(lastcp);
d6a28714
JH
1901 if (regmatch(prog->program + 1)) {
1902 prog->endp[0] = PL_reginput - PL_bostr;
1903 return 1;
1904 }
02db2b7b 1905 REGCP_UNWIND(lastcp);
d6a28714
JH
1906 return 0;
1907}
1908
02db2b7b
IZ
1909#define RE_UNWIND_BRANCH 1
1910#define RE_UNWIND_BRANCHJ 2
1911
1912union re_unwind_t;
1913
1914typedef struct { /* XX: makes sense to enlarge it... */
1915 I32 type;
1916 I32 prev;
1917 CHECKPOINT lastcp;
1918} re_unwind_generic_t;
1919
1920typedef struct {
1921 I32 type;
1922 I32 prev;
1923 CHECKPOINT lastcp;
1924 I32 lastparen;
1925 regnode *next;
1926 char *locinput;
1927 I32 nextchr;
1928#ifdef DEBUGGING
1929 int regindent;
1930#endif
1931} re_unwind_branch_t;
1932
1933typedef union re_unwind_t {
1934 I32 type;
1935 re_unwind_generic_t generic;
1936 re_unwind_branch_t branch;
1937} re_unwind_t;
1938
d6a28714
JH
1939/*
1940 - regmatch - main matching routine
1941 *
1942 * Conceptually the strategy is simple: check to see whether the current
1943 * node matches, call self recursively to see whether the rest matches,
1944 * and then act accordingly. In practice we make some effort to avoid
1945 * recursion, in particular by going through "ordinary" nodes (that don't
1946 * need to know whether the rest of the match failed) by a loop instead of
1947 * by recursion.
1948 */
1949/* [lwall] I've hoisted the register declarations to the outer block in order to
1950 * maybe save a little bit of pushing and popping on the stack. It also takes
1951 * advantage of machines that use a register save mask on subroutine entry.
1952 */
1953STATIC I32 /* 0 failure, 1 success */
1954S_regmatch(pTHX_ regnode *prog)
1955{
d6a28714
JH
1956 register regnode *scan; /* Current node. */
1957 regnode *next; /* Next node. */
1958 regnode *inner; /* Next node in internal branch. */
1959 register I32 nextchr; /* renamed nextchr - nextchar colides with
1960 function of same name */
1961 register I32 n; /* no or next */
1962 register I32 ln; /* len or last */
1963 register char *s; /* operand or save */
1964 register char *locinput = PL_reginput;
1965 register I32 c1, c2, paren; /* case fold search, parenth */
1966 int minmod = 0, sw = 0, logical = 0;
02db2b7b
IZ
1967 I32 unwind = 0;
1968 I32 firstcp = PL_savestack_ix;
ffc61ed2 1969 register bool do_utf8 = DO_UTF8(PL_reg_sv);
02db2b7b 1970
d6a28714
JH
1971#ifdef DEBUGGING
1972 PL_regindent++;
1973#endif
1974
1975 /* Note that nextchr is a byte even in UTF */
1976 nextchr = UCHARAT(locinput);
1977 scan = prog;
1978 while (scan != NULL) {
1979#define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO)
02db2b7b 1980#if 1
d6a28714
JH
1981# define sayYES goto yes
1982# define sayNO goto no
7821416a
IZ
1983# define sayYES_FINAL goto yes_final
1984# define sayYES_LOUD goto yes_loud
1985# define sayNO_FINAL goto no_final
1986# define sayNO_SILENT goto do_no
d6a28714
JH
1987# define saySAME(x) if (x) goto yes; else goto no
1988# define REPORT_CODE_OFF 24
1989#else
1990# define sayYES return 1
1991# define sayNO return 0
7821416a
IZ
1992# define sayYES_FINAL return 1
1993# define sayYES_LOUD return 1
1994# define sayNO_FINAL return 0
1995# define sayNO_SILENT return 0
d6a28714
JH
1996# define saySAME(x) return x
1997#endif
1998 DEBUG_r( {
1999 SV *prop = sv_newmortal();
2000 int docolor = *PL_colors[0];
2001 int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
1aa99e6b 2002 int l = (PL_regeol - locinput) > taill ? taill : (PL_regeol - locinput);
d6a28714
JH
2003 /* The part of the string before starttry has one color
2004 (pref0_len chars), between starttry and current
2005 position another one (pref_len - pref0_len chars),
2006 after the current position the third one.
2007 We assume that pref0_len <= pref_len, otherwise we
2008 decrease pref0_len. */
9041c2e3 2009 int pref_len = (locinput - PL_bostr) > (5 + taill) - l
1aa99e6b
IH
2010 ? (5 + taill) - l : locinput - PL_bostr;
2011 int pref0_len;
d6a28714 2012
1aa99e6b
IH
2013 while (UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
2014 pref_len++;
2015 pref0_len = pref_len - (locinput - PL_reg_starttry);
d6a28714 2016 if (l + pref_len < (5 + taill) && l < PL_regeol - locinput)
9041c2e3 2017 l = ( PL_regeol - locinput > (5 + taill) - pref_len
d6a28714 2018 ? (5 + taill) - pref_len : PL_regeol - locinput);
1aa99e6b
IH
2019 while (UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
2020 l--;
d6a28714
JH
2021 if (pref0_len < 0)
2022 pref0_len = 0;
2023 if (pref0_len > pref_len)
2024 pref0_len = pref_len;
2025 regprop(prop, scan);
9041c2e3 2026 PerlIO_printf(Perl_debug_log,
b900a521 2027 "%4"IVdf" <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3"IVdf":%*s%s\n",
9041c2e3
NIS
2028 (IV)(locinput - PL_bostr),
2029 PL_colors[4], pref0_len,
d6a28714 2030 locinput - pref_len, PL_colors[5],
9041c2e3 2031 PL_colors[2], pref_len - pref0_len,
d6a28714
JH
2032 locinput - pref_len + pref0_len, PL_colors[3],
2033 (docolor ? "" : "> <"),
2034 PL_colors[0], l, locinput, PL_colors[1],
2035 15 - l - pref_len + 1,
2036 "",
b900a521 2037 (IV)(scan - PL_regprogram), PL_regindent*2, "",
d6a28714
JH
2038 SvPVX(prop));
2039 } );
2040
2041 next = scan + NEXT_OFF(scan);
2042 if (next == scan)
2043 next = NULL;
2044
2045 switch (OP(scan)) {
2046 case BOL:
2047 if (locinput == PL_bostr
2048 ? PL_regprev == '\n'
9041c2e3 2049 : (PL_multiline &&
d6a28714
JH
2050 (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
2051 {
2052 /* regtill = regbol; */
b8c5462f
JH
2053 break;
2054 }
d6a28714
JH
2055 sayNO;
2056 case MBOL:
2057 if (locinput == PL_bostr
2058 ? PL_regprev == '\n'
2059 : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
2060 {
b8c5462f
JH
2061 break;
2062 }
d6a28714
JH
2063 sayNO;
2064 case SBOL:
c2a73568 2065 if (locinput == PL_bostr)
b8c5462f 2066 break;
d6a28714
JH
2067 sayNO;
2068 case GPOS:
2069 if (locinput == PL_reg_ganch)
2070 break;
2071 sayNO;
2072 case EOL:
2073 if (PL_multiline)
2074 goto meol;
2075 else
2076 goto seol;
2077 case MEOL:
2078 meol:
2079 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 2080 sayNO;
b8c5462f 2081 break;
d6a28714
JH
2082 case SEOL:
2083 seol:
2084 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 2085 sayNO;
d6a28714 2086 if (PL_regeol - locinput > 1)
b8c5462f 2087 sayNO;
b8c5462f 2088 break;
d6a28714
JH
2089 case EOS:
2090 if (PL_regeol != locinput)
b8c5462f 2091 sayNO;
d6a28714 2092 break;
ffc61ed2 2093 case SANY:
d6a28714 2094 if (!nextchr && locinput >= PL_regeol)
4633a7c4 2095 sayNO;
b8c5462f 2096 nextchr = UCHARAT(++locinput);
a0d0e21e 2097 break;
ffc61ed2 2098 case REG_ANY:
1aa99e6b
IH
2099 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
2100 sayNO;
2101 if (do_utf8) {
b8c5462f 2102 locinput += PL_utf8skip[nextchr];
d6a28714
JH
2103 if (locinput > PL_regeol)
2104 sayNO;
a0ed51b3 2105 nextchr = UCHARAT(locinput);
a0ed51b3 2106 }
1aa99e6b
IH
2107 else
2108 nextchr = UCHARAT(++locinput);
a0ed51b3 2109 break;
d6a28714 2110 case EXACT:
cd439c50
IZ
2111 s = STRING(scan);
2112 ln = STR_LEN(scan);
1aa99e6b
IH
2113 if (do_utf8 != (UTF!=0)) {
2114 char *l = locinput;
2115 char *e = s + ln;
2116 STRLEN len;
2117 if (do_utf8)
2118 while (s < e) {
2119 if (l >= PL_regeol)
2120 sayNO;
9041c2e3 2121 if (*((U8*)s) != utf8_to_uvchr((U8*)l, &len))
1aa99e6b
IH
2122 sayNO;
2123 s++;
2124 l += len;
2125 }
2126 else
2127 while (s < e) {
2128 if (l >= PL_regeol)
2129 sayNO;
9041c2e3 2130 if (*((U8*)l) != utf8_to_uvchr((U8*)s, &len))
1aa99e6b
IH
2131 sayNO;
2132 s += len;
2133 l++;
2134 }
2135 locinput = l;
2136 nextchr = UCHARAT(locinput);
2137 break;
2138 }
d6a28714
JH
2139 /* Inline the first character, for speed. */
2140 if (UCHARAT(s) != nextchr)
2141 sayNO;
2142 if (PL_regeol - locinput < ln)
2143 sayNO;
2144 if (ln > 1 && memNE(s, locinput, ln))
2145 sayNO;
2146 locinput += ln;
2147 nextchr = UCHARAT(locinput);
2148 break;
2149 case EXACTFL:
b8c5462f
JH
2150 PL_reg_flags |= RF_tainted;
2151 /* FALL THROUGH */
d6a28714 2152 case EXACTF:
cd439c50
IZ
2153 s = STRING(scan);
2154 ln = STR_LEN(scan);
d6a28714 2155
1aa99e6b 2156 if (do_utf8) {
d6a28714 2157 char *l = locinput;
1aa99e6b
IH
2158 char *e;
2159 e = s + ln;
d6a28714
JH
2160 c1 = OP(scan) == EXACTF;
2161 while (s < e) {
1aa99e6b 2162 if (l >= PL_regeol) {
d6a28714
JH
2163 sayNO;
2164 }
2b9d42f0 2165 if ((UTF ? utf8n_to_uvchr((U8*)s, e - s, 0, 0) : *((U8*)s)) !=
1aa99e6b
IH
2166 (c1 ? toLOWER_utf8((U8*)l) : toLOWER_LC_utf8((U8*)l)))
2167 sayNO;
2168 s += UTF ? UTF8SKIP(s) : 1;
d6a28714 2169 l += UTF8SKIP(l);
b8c5462f 2170 }
d6a28714 2171 locinput = l;
a0ed51b3
LW
2172 nextchr = UCHARAT(locinput);
2173 break;
2174 }
d6a28714
JH
2175
2176 /* Inline the first character, for speed. */
2177 if (UCHARAT(s) != nextchr &&
2178 UCHARAT(s) != ((OP(scan) == EXACTF)
2179 ? PL_fold : PL_fold_locale)[nextchr])
a0ed51b3 2180 sayNO;
d6a28714 2181 if (PL_regeol - locinput < ln)
b8c5462f 2182 sayNO;
d6a28714
JH
2183 if (ln > 1 && (OP(scan) == EXACTF
2184 ? ibcmp(s, locinput, ln)
2185 : ibcmp_locale(s, locinput, ln)))
4633a7c4 2186 sayNO;
d6a28714
JH
2187 locinput += ln;
2188 nextchr = UCHARAT(locinput);
a0d0e21e 2189 break;
d6a28714 2190 case ANYOF:
ffc61ed2
JH
2191 if (do_utf8) {
2192 if (!reginclass(scan, (U8*)locinput, do_utf8))
2193 sayNO;
2194 if (locinput >= PL_regeol)
2195 sayNO;
2196 locinput += PL_utf8skip[nextchr];
b8c5462f 2197 nextchr = UCHARAT(locinput);
ffc61ed2
JH
2198 }
2199 else {
2200 if (nextchr < 0)
2201 nextchr = UCHARAT(locinput);
2202 if (!reginclass(scan, (U8*)locinput, do_utf8))
2203 sayNO;
2204 if (!nextchr && locinput >= PL_regeol)
2205 sayNO;
2206 nextchr = UCHARAT(++locinput);
2207 }
b8c5462f 2208 break;
d6a28714 2209 case ALNUML:
b8c5462f
JH
2210 PL_reg_flags |= RF_tainted;
2211 /* FALL THROUGH */
d6a28714 2212 case ALNUM:
b8c5462f 2213 if (!nextchr)
4633a7c4 2214 sayNO;
ffc61ed2
JH
2215 if (do_utf8) {
2216 if (!(OP(scan) == ALNUM
d6a28714
JH
2217 ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
2218 : isALNUM_LC_utf8((U8*)locinput)))
b8c5462f
JH
2219 {
2220 sayNO;
a0ed51b3 2221 }
b8c5462f 2222 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
2223 nextchr = UCHARAT(locinput);
2224 break;
2225 }
ffc61ed2 2226 if (!(OP(scan) == ALNUM
d6a28714 2227 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
bbce6d69 2228 sayNO;
b8c5462f 2229 nextchr = UCHARAT(++locinput);
a0d0e21e 2230 break;
d6a28714 2231 case NALNUML:
b8c5462f
JH
2232 PL_reg_flags |= RF_tainted;
2233 /* FALL THROUGH */
d6a28714
JH
2234 case NALNUM:
2235 if (!nextchr && locinput >= PL_regeol)
a0ed51b3 2236 sayNO;
ffc61ed2 2237 if (do_utf8) {
8269fa76 2238 LOAD_UTF8_CHARCLASS(alnum,"a");
ffc61ed2 2239 if (OP(scan) == NALNUM
d6a28714
JH
2240 ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
2241 : isALNUM_LC_utf8((U8*)locinput))
2242 {
b8c5462f 2243 sayNO;
d6a28714 2244 }
b8c5462f
JH
2245 locinput += PL_utf8skip[nextchr];
2246 nextchr = UCHARAT(locinput);
2247 break;
2248 }
ffc61ed2 2249 if (OP(scan) == NALNUM
d6a28714 2250 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
b8c5462f 2251 sayNO;
76e3520e 2252 nextchr = UCHARAT(++locinput);
a0d0e21e 2253 break;
d6a28714
JH
2254 case BOUNDL:
2255 case NBOUNDL:
3280af22 2256 PL_reg_flags |= RF_tainted;
bbce6d69 2257 /* FALL THROUGH */
d6a28714
JH
2258 case BOUND:
2259 case NBOUND:
2260 /* was last char in word? */
ffc61ed2
JH
2261 if (do_utf8) {
2262 if (locinput == PL_regbol)
2263 ln = PL_regprev;
2264 else {
2265 U8 *r = reghop((U8*)locinput, -1);
9041c2e3 2266
2b9d42f0 2267 ln = utf8n_to_uvchr(r, s - (char*)r, 0, 0);
ffc61ed2
JH
2268 }
2269 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
2270 ln = isALNUM_uni(ln);
8269fa76 2271 LOAD_UTF8_CHARCLASS(alnum,"a");
ffc61ed2
JH
2272 n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
2273 }
2274 else {
9041c2e3 2275 ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln));
ffc61ed2
JH
2276 n = isALNUM_LC_utf8((U8*)locinput);
2277 }
a0ed51b3 2278 }
d6a28714 2279 else {
ffc61ed2
JH
2280 ln = (locinput != PL_regbol) ?
2281 UCHARAT(locinput - 1) : PL_regprev;
2282 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
2283 ln = isALNUM(ln);
2284 n = isALNUM(nextchr);
2285 }
2286 else {
2287 ln = isALNUM_LC(ln);
2288 n = isALNUM_LC(nextchr);
2289 }
d6a28714 2290 }
ffc61ed2
JH
2291 if (((!ln) == (!n)) == (OP(scan) == BOUND ||
2292 OP(scan) == BOUNDL))
2293 sayNO;
a0ed51b3 2294 break;
d6a28714 2295 case SPACEL:
3280af22 2296 PL_reg_flags |= RF_tainted;
bbce6d69 2297 /* FALL THROUGH */
d6a28714 2298 case SPACE:
9442cb0e 2299 if (!nextchr)
4633a7c4 2300 sayNO;
1aa99e6b 2301 if (do_utf8) {
fd400ab9 2302 if (UTF8_IS_CONTINUED(nextchr)) {
8269fa76 2303 LOAD_UTF8_CHARCLASS(space," ");
ffc61ed2
JH
2304 if (!(OP(scan) == SPACE
2305 ? swash_fetch(PL_utf8_space, (U8*)locinput)
2306 : isSPACE_LC_utf8((U8*)locinput)))
2307 {
2308 sayNO;
2309 }
2310 locinput += PL_utf8skip[nextchr];
2311 nextchr = UCHARAT(locinput);
2312 break;
d6a28714 2313 }
ffc61ed2
JH
2314 if (!(OP(scan) == SPACE
2315 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
2316 sayNO;
2317 nextchr = UCHARAT(++locinput);
2318 }
2319 else {
2320 if (!(OP(scan) == SPACE
2321 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
2322 sayNO;
2323 nextchr = UCHARAT(++locinput);
a0ed51b3 2324 }
a0ed51b3 2325 break;
d6a28714 2326 case NSPACEL:
3280af22 2327 PL_reg_flags |= RF_tainted;
bbce6d69 2328 /* FALL THROUGH */
d6a28714 2329 case NSPACE:
9442cb0e 2330 if (!nextchr && locinput >= PL_regeol)
b8c5462f 2331 sayNO;
1aa99e6b 2332 if (do_utf8) {
8269fa76 2333 LOAD_UTF8_CHARCLASS(space," ");
ffc61ed2 2334 if (OP(scan) == NSPACE
9442cb0e 2335 ? swash_fetch(PL_utf8_space, (U8*)locinput)
d6a28714 2336 : isSPACE_LC_utf8((U8*)locinput))
b8c5462f
JH
2337 {
2338 sayNO;
2339 }
2340 locinput += PL_utf8skip[nextchr];
2341 nextchr = UCHARAT(locinput);
2342 break;
a0ed51b3 2343 }
ffc61ed2 2344 if (OP(scan) == NSPACE
d6a28714 2345 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
4633a7c4 2346 sayNO;
76e3520e 2347 nextchr = UCHARAT(++locinput);
a0d0e21e 2348 break;
d6a28714 2349 case DIGITL:
a0ed51b3
LW
2350 PL_reg_flags |= RF_tainted;
2351 /* FALL THROUGH */
d6a28714 2352 case DIGIT:
9442cb0e 2353 if (!nextchr)
a0ed51b3 2354 sayNO;
1aa99e6b 2355 if (do_utf8) {
8269fa76 2356 LOAD_UTF8_CHARCLASS(digit,"0");
ffc61ed2 2357 if (!(OP(scan) == DIGIT
9442cb0e
GS
2358 ? swash_fetch(PL_utf8_digit, (U8*)locinput)
2359 : isDIGIT_LC_utf8((U8*)locinput)))
dfe13c55 2360 {
a0ed51b3 2361 sayNO;
dfe13c55 2362 }
6f06b55f 2363 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
2364 nextchr = UCHARAT(locinput);
2365 break;
2366 }
ffc61ed2 2367 if (!(OP(scan) == DIGIT
9442cb0e 2368 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
a0ed51b3
LW
2369 sayNO;
2370 nextchr = UCHARAT(++locinput);
2371 break;
d6a28714 2372 case NDIGITL:
b8c5462f
JH
2373 PL_reg_flags |= RF_tainted;
2374 /* FALL THROUGH */
d6a28714 2375 case NDIGIT:
9442cb0e 2376 if (!nextchr && locinput >= PL_regeol)
b8c5462f 2377 sayNO;
1aa99e6b 2378 if (do_utf8) {
8269fa76 2379 LOAD_UTF8_CHARCLASS(digit,"0");
ffc61ed2 2380 if (OP(scan) == NDIGIT
9442cb0e
GS
2381 ? swash_fetch(PL_utf8_digit, (U8*)locinput)
2382 : isDIGIT_LC_utf8((U8*)locinput))
2383 {
a0ed51b3 2384 sayNO;
9442cb0e 2385 }
6f06b55f 2386 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
2387 nextchr = UCHARAT(locinput);
2388 break;
2389 }
ffc61ed2 2390 if (OP(scan) == NDIGIT
9442cb0e 2391 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
a0ed51b3
LW
2392 sayNO;
2393 nextchr = UCHARAT(++locinput);
2394 break;
2395 case CLUMP:
8269fa76 2396 LOAD_UTF8_CHARCLASS(mark,"~");
dfe13c55 2397 if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput))
a0ed51b3 2398 sayNO;
6f06b55f 2399 locinput += PL_utf8skip[nextchr];
dfe13c55 2400 while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput))
a0ed51b3
LW
2401 locinput += UTF8SKIP(locinput);
2402 if (locinput > PL_regeol)
2403 sayNO;
2404 nextchr = UCHARAT(locinput);
2405 break;
c8756f30 2406 case REFFL:
3280af22 2407 PL_reg_flags |= RF_tainted;
c8756f30 2408 /* FALL THROUGH */
c277df42 2409 case REF:
c8756f30 2410 case REFF:
c277df42 2411 n = ARG(scan); /* which paren pair */
cf93c79d 2412 ln = PL_regstartp[n];
2c2d71f5 2413 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
cf93c79d 2414 if (*PL_reglastparen < n || ln == -1)
af3f8c16 2415 sayNO; /* Do not match unless seen CLOSEn. */
cf93c79d 2416 if (ln == PL_regendp[n])
a0d0e21e 2417 break;
a0ed51b3 2418
cf93c79d 2419 s = PL_bostr + ln;
1aa99e6b 2420 if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */
a0ed51b3 2421 char *l = locinput;
cf93c79d 2422 char *e = PL_bostr + PL_regendp[n];
a0ed51b3
LW
2423 /*
2424 * Note that we can't do the "other character" lookup trick as
2425 * in the 8-bit case (no pun intended) because in Unicode we
2426 * have to map both upper and title case to lower case.
2427 */
2428 if (OP(scan) == REFF) {
2429 while (s < e) {
2430 if (l >= PL_regeol)
2431 sayNO;
dfe13c55 2432 if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l))
a0ed51b3
LW
2433 sayNO;
2434 s += UTF8SKIP(s);
2435 l += UTF8SKIP(l);
2436 }
2437 }
2438 else {
2439 while (s < e) {
2440 if (l >= PL_regeol)
2441 sayNO;
dfe13c55 2442 if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l))
a0ed51b3
LW
2443 sayNO;
2444 s += UTF8SKIP(s);
2445 l += UTF8SKIP(l);
2446 }
2447 }
2448 locinput = l;
2449 nextchr = UCHARAT(locinput);
2450 break;
2451 }
2452
a0d0e21e 2453 /* Inline the first character, for speed. */
76e3520e 2454 if (UCHARAT(s) != nextchr &&
c8756f30
AK
2455 (OP(scan) == REF ||
2456 (UCHARAT(s) != ((OP(scan) == REFF
22c35a8c 2457 ? PL_fold : PL_fold_locale)[nextchr]))))
4633a7c4 2458 sayNO;
cf93c79d 2459 ln = PL_regendp[n] - ln;
3280af22 2460 if (locinput + ln > PL_regeol)
4633a7c4 2461 sayNO;
c8756f30
AK
2462 if (ln > 1 && (OP(scan) == REF
2463 ? memNE(s, locinput, ln)
2464 : (OP(scan) == REFF
2465 ? ibcmp(s, locinput, ln)
2466 : ibcmp_locale(s, locinput, ln))))
4633a7c4 2467 sayNO;
a0d0e21e 2468 locinput += ln;
76e3520e 2469 nextchr = UCHARAT(locinput);
a0d0e21e
LW
2470 break;
2471
2472 case NOTHING:
c277df42 2473 case TAIL:
a0d0e21e
LW
2474 break;
2475 case BACK:
2476 break;
c277df42
IZ
2477 case EVAL:
2478 {
2479 dSP;
533c011a 2480 OP_4tree *oop = PL_op;
3280af22
NIS
2481 COP *ocurcop = PL_curcop;
2482 SV **ocurpad = PL_curpad;
c277df42 2483 SV *ret;
9041c2e3 2484
c277df42 2485 n = ARG(scan);
533c011a 2486 PL_op = (OP_4tree*)PL_regdata->data[n];
d7d93a81 2487 DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
dfad63ad 2488 PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]);
cf93c79d 2489 PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
c277df42 2490
cea2e8a9 2491 CALLRUNOPS(aTHX); /* Scalar context. */
c277df42
IZ
2492 SPAGAIN;
2493 ret = POPs;
2494 PUTBACK;
9041c2e3 2495
0f5d15d6
IZ
2496 PL_op = oop;
2497 PL_curpad = ocurpad;
2498 PL_curcop = ocurcop;
c277df42 2499 if (logical) {
0f5d15d6
IZ
2500 if (logical == 2) { /* Postponed subexpression. */
2501 regexp *re;
22c35a8c 2502 MAGIC *mg = Null(MAGIC*);
0f5d15d6 2503 re_cc_state state;
0f5d15d6
IZ
2504 CHECKPOINT cp, lastcp;
2505
2506 if(SvROK(ret) || SvRMAGICAL(ret)) {
2507 SV *sv = SvROK(ret) ? SvRV(ret) : ret;
2508
2509 if(SvMAGICAL(sv))
2510 mg = mg_find(sv, 'r');
2511 }
2512 if (mg) {
2513 re = (regexp *)mg->mg_obj;
df0003d4 2514 (void)ReREFCNT_inc(re);
0f5d15d6
IZ
2515 }
2516 else {
2517 STRLEN len;
2518 char *t = SvPV(ret, len);
2519 PMOP pm;
2520 char *oprecomp = PL_regprecomp;
2521 I32 osize = PL_regsize;
2522 I32 onpar = PL_regnpar;
2523
5fcd1c1b 2524 Zero(&pm, 1, PMOP);
cea2e8a9 2525 re = CALLREGCOMP(aTHX_ t, t + len, &pm);
9041c2e3 2526 if (!(SvFLAGS(ret)
0f5d15d6
IZ
2527 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY)))
2528 sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0);
2529 PL_regprecomp = oprecomp;
2530 PL_regsize = osize;
2531 PL_regnpar = onpar;
2532 }
2533 DEBUG_r(
9041c2e3 2534 PerlIO_printf(Perl_debug_log,
0f5d15d6
IZ
2535 "Entering embedded `%s%.60s%s%s'\n",
2536 PL_colors[0],
2537 re->precomp,
2538 PL_colors[1],
2539 (strlen(re->precomp) > 60 ? "..." : ""))
2540 );
2541 state.node = next;
2542 state.prev = PL_reg_call_cc;
2543 state.cc = PL_regcc;
2544 state.re = PL_reg_re;
2545
2ab05381 2546 PL_regcc = 0;
9041c2e3 2547
0f5d15d6 2548 cp = regcppush(0); /* Save *all* the positions. */
02db2b7b 2549 REGCP_SET(lastcp);
0f5d15d6
IZ
2550 cache_re(re);
2551 state.ss = PL_savestack_ix;
2552 *PL_reglastparen = 0;
2553 PL_reg_call_cc = &state;
2554 PL_reginput = locinput;
2c2d71f5
JH
2555
2556 /* XXXX This is too dramatic a measure... */
2557 PL_reg_maxiter = 0;
2558
0f5d15d6 2559 if (regmatch(re->program + 1)) {
2c914db6
IZ
2560 /* Even though we succeeded, we need to restore
2561 global variables, since we may be wrapped inside
2562 SUSPEND, thus the match may be not finished yet. */
2563
2564 /* XXXX Do this only if SUSPENDed? */
2565 PL_reg_call_cc = state.prev;
2566 PL_regcc = state.cc;
2567 PL_reg_re = state.re;
2568 cache_re(PL_reg_re);
2569
2570 /* XXXX This is too dramatic a measure... */
2571 PL_reg_maxiter = 0;
2572
2573 /* These are needed even if not SUSPEND. */
0f5d15d6
IZ
2574 ReREFCNT_dec(re);
2575 regcpblow(cp);
2576 sayYES;
2577 }
0f5d15d6 2578 ReREFCNT_dec(re);
02db2b7b 2579 REGCP_UNWIND(lastcp);
0f5d15d6
IZ
2580 regcppop();
2581 PL_reg_call_cc = state.prev;
2582 PL_regcc = state.cc;
2583 PL_reg_re = state.re;
d3790889 2584 cache_re(PL_reg_re);
2c2d71f5
JH
2585
2586 /* XXXX This is too dramatic a measure... */
2587 PL_reg_maxiter = 0;
2588
0f5d15d6
IZ
2589 sayNO;
2590 }
c277df42 2591 sw = SvTRUE(ret);
0f5d15d6 2592 logical = 0;
a0ed51b3
LW
2593 }
2594 else
3280af22 2595 sv_setsv(save_scalar(PL_replgv), ret);
c277df42
IZ
2596 break;
2597 }
a0d0e21e 2598 case OPEN:
c277df42 2599 n = ARG(scan); /* which paren pair */
3280af22
NIS
2600 PL_reg_start_tmp[n] = locinput;
2601 if (n > PL_regsize)
2602 PL_regsize = n;
a0d0e21e
LW
2603 break;
2604 case CLOSE:
c277df42 2605 n = ARG(scan); /* which paren pair */
cf93c79d
IZ
2606 PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr;
2607 PL_regendp[n] = locinput - PL_bostr;
3280af22
NIS
2608 if (n > *PL_reglastparen)
2609 *PL_reglastparen = n;
a0d0e21e 2610 break;
c277df42
IZ
2611 case GROUPP:
2612 n = ARG(scan); /* which paren pair */
cf93c79d 2613 sw = (*PL_reglastparen >= n && PL_regendp[n] != -1);
c277df42
IZ
2614 break;
2615 case IFTHEN:
2c2d71f5 2616 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
c277df42
IZ
2617 if (sw)
2618 next = NEXTOPER(NEXTOPER(scan));
2619 else {
2620 next = scan + ARG(scan);
2621 if (OP(next) == IFTHEN) /* Fake one. */
2622 next = NEXTOPER(NEXTOPER(next));
2623 }
2624 break;
2625 case LOGICAL:
0f5d15d6 2626 logical = scan->flags;
c277df42 2627 break;
2ab05381
IZ
2628/*******************************************************************
2629 PL_regcc contains infoblock about the innermost (...)* loop, and
2630 a pointer to the next outer infoblock.
2631
2632 Here is how Y(A)*Z is processed (if it is compiled into CURLYX/WHILEM):
2633
2634 1) After matching X, regnode for CURLYX is processed;
2635
9041c2e3 2636 2) This regnode creates infoblock on the stack, and calls
2ab05381
IZ
2637 regmatch() recursively with the starting point at WHILEM node;
2638
2639 3) Each hit of WHILEM node tries to match A and Z (in the order
2640 depending on the current iteration, min/max of {min,max} and
2641 greediness). The information about where are nodes for "A"
2642 and "Z" is read from the infoblock, as is info on how many times "A"
2643 was already matched, and greediness.
2644
2645 4) After A matches, the same WHILEM node is hit again.
2646
2647 5) Each time WHILEM is hit, PL_regcc is the infoblock created by CURLYX
2648 of the same pair. Thus when WHILEM tries to match Z, it temporarily
2649 resets PL_regcc, since this Y(A)*Z can be a part of some other loop:
2650 as in (Y(A)*Z)*. If Z matches, the automaton will hit the WHILEM node
2651 of the external loop.
2652
2653 Currently present infoblocks form a tree with a stem formed by PL_curcc
2654 and whatever it mentions via ->next, and additional attached trees
2655 corresponding to temporarily unset infoblocks as in "5" above.
2656
9041c2e3 2657 In the following picture infoblocks for outer loop of
2ab05381
IZ
2658 (Y(A)*?Z)*?T are denoted O, for inner I. NULL starting block
2659 is denoted by x. The matched string is YAAZYAZT. Temporarily postponed
2660 infoblocks are drawn below the "reset" infoblock.
2661
2662 In fact in the picture below we do not show failed matches for Z and T
2663 by WHILEM blocks. [We illustrate minimal matches, since for them it is
2664 more obvious *why* one needs to *temporary* unset infoblocks.]
2665
2666 Matched REx position InfoBlocks Comment
2667 (Y(A)*?Z)*?T x
2668 Y(A)*?Z)*?T x <- O
2669 Y (A)*?Z)*?T x <- O
2670 Y A)*?Z)*?T x <- O <- I
2671 YA )*?Z)*?T x <- O <- I
2672 YA A)*?Z)*?T x <- O <- I
2673 YAA )*?Z)*?T x <- O <- I
2674 YAA Z)*?T x <- O # Temporary unset I
2675 I
2676
2677 YAAZ Y(A)*?Z)*?T x <- O
2678 I
2679
2680 YAAZY (A)*?Z)*?T x <- O
2681 I
2682
2683 YAAZY A)*?Z)*?T x <- O <- I
2684 I
2685
2686 YAAZYA )*?Z)*?T x <- O <- I
2687 I
2688
2689 YAAZYA Z)*?T x <- O # Temporary unset I
2690 I,I
2691
2692 YAAZYAZ )*?T x <- O
2693 I,I
2694
2695 YAAZYAZ T x # Temporary unset O
2696 O
2697 I,I
2698
2699 YAAZYAZT x
2700 O
2701 I,I
2702 *******************************************************************/
a0d0e21e
LW
2703 case CURLYX: {
2704 CURCUR cc;
3280af22 2705 CHECKPOINT cp = PL_savestack_ix;
cb434fcc
IZ
2706 /* No need to save/restore up to this paren */
2707 I32 parenfloor = scan->flags;
c277df42
IZ
2708
2709 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
2710 next += ARG(next);
3280af22
NIS
2711 cc.oldcc = PL_regcc;
2712 PL_regcc = &cc;
cb434fcc
IZ
2713 /* XXXX Probably it is better to teach regpush to support
2714 parenfloor > PL_regsize... */
2715 if (parenfloor > *PL_reglastparen)
2716 parenfloor = *PL_reglastparen; /* Pessimization... */
2717 cc.parenfloor = parenfloor;
a0d0e21e
LW
2718 cc.cur = -1;
2719 cc.min = ARG1(scan);
2720 cc.max = ARG2(scan);
c277df42 2721 cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
a0d0e21e
LW
2722 cc.next = next;
2723 cc.minmod = minmod;
2724 cc.lastloc = 0;
3280af22 2725 PL_reginput = locinput;
a0d0e21e
LW
2726 n = regmatch(PREVOPER(next)); /* start on the WHILEM */
2727 regcpblow(cp);
3280af22 2728 PL_regcc = cc.oldcc;
4633a7c4 2729 saySAME(n);
a0d0e21e
LW
2730 }
2731 /* NOT REACHED */
2732 case WHILEM: {
2733 /*
2734 * This is really hard to understand, because after we match
2735 * what we're trying to match, we must make sure the rest of
2c2d71f5 2736 * the REx is going to match for sure, and to do that we have
a0d0e21e
LW
2737 * to go back UP the parse tree by recursing ever deeper. And
2738 * if it fails, we have to reset our parent's current state
2739 * that we can try again after backing off.
2740 */
2741
c277df42 2742 CHECKPOINT cp, lastcp;
3280af22 2743 CURCUR* cc = PL_regcc;
c277df42
IZ
2744 char *lastloc = cc->lastloc; /* Detection of 0-len. */
2745
4633a7c4 2746 n = cc->cur + 1; /* how many we know we matched */
3280af22 2747 PL_reginput = locinput;
a0d0e21e 2748
c277df42 2749 DEBUG_r(
9041c2e3
NIS
2750 PerlIO_printf(Perl_debug_log,
2751 "%*s %ld out of %ld..%ld cc=%lx\n",
3280af22 2752 REPORT_CODE_OFF+PL_regindent*2, "",
9041c2e3 2753 (long)n, (long)cc->min,
c277df42
IZ
2754 (long)cc->max, (long)cc)
2755 );
4633a7c4 2756
a0d0e21e
LW
2757 /* If degenerate scan matches "", assume scan done. */
2758
579cf2c3 2759 if (locinput == cc->lastloc && n >= cc->min) {
3280af22 2760 PL_regcc = cc->oldcc;
2ab05381
IZ
2761 if (PL_regcc)
2762 ln = PL_regcc->cur;
c277df42 2763 DEBUG_r(
c3464db5
DD
2764 PerlIO_printf(Perl_debug_log,
2765 "%*s empty match detected, try continuation...\n",
3280af22 2766 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 2767 );
a0d0e21e 2768 if (regmatch(cc->next))
4633a7c4 2769 sayYES;
2ab05381
IZ
2770 if (PL_regcc)
2771 PL_regcc->cur = ln;
3280af22 2772 PL_regcc = cc;
4633a7c4 2773 sayNO;
a0d0e21e
LW
2774 }
2775
2776 /* First just match a string of min scans. */
2777
2778 if (n < cc->min) {
2779 cc->cur = n;
2780 cc->lastloc = locinput;
4633a7c4
LW
2781 if (regmatch(cc->scan))
2782 sayYES;
2783 cc->cur = n - 1;
c277df42 2784 cc->lastloc = lastloc;
4633a7c4 2785 sayNO;
a0d0e21e
LW
2786 }
2787
2c2d71f5
JH
2788 if (scan->flags) {
2789 /* Check whether we already were at this position.
2790 Postpone detection until we know the match is not
2791 *that* much linear. */
2792 if (!PL_reg_maxiter) {
2793 PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4);
2794 PL_reg_leftiter = PL_reg_maxiter;
2795 }
2796 if (PL_reg_leftiter-- == 0) {
2797 I32 size = (PL_reg_maxiter + 7)/8;
2798 if (PL_reg_poscache) {
2799 if (PL_reg_poscache_size < size) {
2800 Renew(PL_reg_poscache, size, char);
2801 PL_reg_poscache_size = size;
2802 }
2803 Zero(PL_reg_poscache, size, char);
2804 }
2805 else {
2806 PL_reg_poscache_size = size;
2807 Newz(29, PL_reg_poscache, size, char);
2808 }
2809 DEBUG_r(
2810 PerlIO_printf(Perl_debug_log,
2811 "%sDetected a super-linear match, switching on caching%s...\n",
2812 PL_colors[4], PL_colors[5])
2813 );
2814 }
2815 if (PL_reg_leftiter < 0) {
2816 I32 o = locinput - PL_bostr, b;
2817
2818 o = (scan->flags & 0xf) - 1 + o * (scan->flags>>4);
2819 b = o % 8;
2820 o /= 8;
2821 if (PL_reg_poscache[o] & (1<<b)) {
2822 DEBUG_r(
2823 PerlIO_printf(Perl_debug_log,
2824 "%*s already tried at this position...\n",
2825 REPORT_CODE_OFF+PL_regindent*2, "")
2826 );
7821416a 2827 sayNO_SILENT;
2c2d71f5
JH
2828 }
2829 PL_reg_poscache[o] |= (1<<b);
2830 }
2831 }
2832
a0d0e21e
LW
2833 /* Prefer next over scan for minimal matching. */
2834
2835 if (cc->minmod) {
3280af22 2836 PL_regcc = cc->oldcc;
2ab05381
IZ
2837 if (PL_regcc)
2838 ln = PL_regcc->cur;
5f05dabc 2839 cp = regcppush(cc->parenfloor);
02db2b7b 2840 REGCP_SET(lastcp);
5f05dabc 2841 if (regmatch(cc->next)) {
c277df42 2842 regcpblow(cp);
4633a7c4 2843 sayYES; /* All done. */
5f05dabc 2844 }
02db2b7b 2845 REGCP_UNWIND(lastcp);
5f05dabc 2846 regcppop();
2ab05381
IZ
2847 if (PL_regcc)
2848 PL_regcc->cur = ln;
3280af22 2849 PL_regcc = cc;
a0d0e21e 2850
c277df42 2851 if (n >= cc->max) { /* Maximum greed exceeded? */
9041c2e3 2852 if (ckWARN(WARN_REGEXP) && n >= REG_INFTY
3280af22
NIS
2853 && !(PL_reg_flags & RF_warned)) {
2854 PL_reg_flags |= RF_warned;
e476b1b5 2855 Perl_warner(aTHX_ WARN_REGEXP, "%s limit (%d) exceeded",
2f3ca594
GS
2856 "Complex regular subexpression recursion",
2857 REG_INFTY - 1);
c277df42 2858 }
4633a7c4 2859 sayNO;
c277df42 2860 }
a687059c 2861
c277df42 2862 DEBUG_r(
c3464db5
DD
2863 PerlIO_printf(Perl_debug_log,
2864 "%*s trying longer...\n",
3280af22 2865 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 2866 );
a0d0e21e 2867 /* Try scanning more and see if it helps. */
3280af22 2868 PL_reginput = locinput;
a0d0e21e
LW
2869 cc->cur = n;
2870 cc->lastloc = locinput;
5f05dabc 2871 cp = regcppush(cc->parenfloor);
02db2b7b 2872 REGCP_SET(lastcp);
5f05dabc 2873 if (regmatch(cc->scan)) {
c277df42 2874 regcpblow(cp);
4633a7c4 2875 sayYES;
5f05dabc 2876 }
02db2b7b 2877 REGCP_UNWIND(lastcp);
5f05dabc 2878 regcppop();
4633a7c4 2879 cc->cur = n - 1;
c277df42 2880 cc->lastloc = lastloc;
4633a7c4 2881 sayNO;
a0d0e21e
LW
2882 }
2883
2884 /* Prefer scan over next for maximal matching. */
2885
2886 if (n < cc->max) { /* More greed allowed? */
5f05dabc 2887 cp = regcppush(cc->parenfloor);
a0d0e21e
LW
2888 cc->cur = n;
2889 cc->lastloc = locinput;
02db2b7b 2890 REGCP_SET(lastcp);
5f05dabc 2891 if (regmatch(cc->scan)) {
c277df42 2892 regcpblow(cp);
4633a7c4 2893 sayYES;
5f05dabc 2894 }
02db2b7b 2895 REGCP_UNWIND(lastcp);
a0d0e21e 2896 regcppop(); /* Restore some previous $<digit>s? */
3280af22 2897 PL_reginput = locinput;
c277df42 2898 DEBUG_r(
c3464db5
DD
2899 PerlIO_printf(Perl_debug_log,
2900 "%*s failed, try continuation...\n",
3280af22 2901 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42
IZ
2902 );
2903 }
9041c2e3 2904 if (ckWARN(WARN_REGEXP) && n >= REG_INFTY
599cee73 2905 && !(PL_reg_flags & RF_warned)) {
3280af22 2906 PL_reg_flags |= RF_warned;
e476b1b5 2907 Perl_warner(aTHX_ WARN_REGEXP, "%s limit (%d) exceeded",
cb5d145d
GS
2908 "Complex regular subexpression recursion",
2909 REG_INFTY - 1);
a0d0e21e
LW
2910 }
2911
2912 /* Failed deeper matches of scan, so see if this one works. */
3280af22 2913 PL_regcc = cc->oldcc;
2ab05381
IZ
2914 if (PL_regcc)
2915 ln = PL_regcc->cur;
a0d0e21e 2916 if (regmatch(cc->next))
4633a7c4 2917 sayYES;
2ab05381
IZ
2918 if (PL_regcc)
2919 PL_regcc->cur = ln;
3280af22 2920 PL_regcc = cc;
4633a7c4 2921 cc->cur = n - 1;
c277df42 2922 cc->lastloc = lastloc;
4633a7c4 2923 sayNO;
a0d0e21e
LW
2924 }
2925 /* NOT REACHED */
9041c2e3 2926 case BRANCHJ:
c277df42
IZ
2927 next = scan + ARG(scan);
2928 if (next == scan)
2929 next = NULL;
2930 inner = NEXTOPER(NEXTOPER(scan));
2931 goto do_branch;
9041c2e3 2932 case BRANCH:
c277df42
IZ
2933 inner = NEXTOPER(scan);
2934 do_branch:
2935 {
2936 CHECKPOINT lastcp;
2937 c1 = OP(scan);
2938 if (OP(next) != c1) /* No choice. */
2939 next = inner; /* Avoid recursion. */
a0d0e21e 2940 else {
02db2b7b
IZ
2941 I32 lastparen = *PL_reglastparen;
2942 I32 unwind1;
2943 re_unwind_branch_t *uw;
2944
2945 /* Put unwinding data on stack */
2946 unwind1 = SSNEWt(1,re_unwind_branch_t);
2947 uw = SSPTRt(unwind1,re_unwind_branch_t);
2948 uw->prev = unwind;
2949 unwind = unwind1;
2950 uw->type = ((c1 == BRANCH)
2951 ? RE_UNWIND_BRANCH
2952 : RE_UNWIND_BRANCHJ);
2953 uw->lastparen = lastparen;
2954 uw->next = next;
2955 uw->locinput = locinput;
2956 uw->nextchr = nextchr;
2957#ifdef DEBUGGING
2958 uw->regindent = ++PL_regindent;
2959#endif
c277df42 2960
02db2b7b
IZ
2961 REGCP_SET(uw->lastcp);
2962
2963 /* Now go into the first branch */
2964 next = inner;
a687059c 2965 }
a0d0e21e
LW
2966 }
2967 break;
2968 case MINMOD:
2969 minmod = 1;
2970 break;
c277df42
IZ
2971 case CURLYM:
2972 {
00db4c45 2973 I32 l = 0;
c277df42 2974 CHECKPOINT lastcp;
9041c2e3 2975
c277df42
IZ
2976 /* We suppose that the next guy does not need
2977 backtracking: in particular, it is of constant length,
2978 and has no parenths to influence future backrefs. */
2979 ln = ARG1(scan); /* min to match */
2980 n = ARG2(scan); /* max to match */
c277df42
IZ
2981 paren = scan->flags;
2982 if (paren) {
3280af22
NIS
2983 if (paren > PL_regsize)
2984 PL_regsize = paren;
2985 if (paren > *PL_reglastparen)
2986 *PL_reglastparen = paren;
c277df42 2987 }
dc45a647 2988 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
c277df42
IZ
2989 if (paren)
2990 scan += NEXT_OFF(scan); /* Skip former OPEN. */
3280af22 2991 PL_reginput = locinput;
c277df42
IZ
2992 if (minmod) {
2993 minmod = 0;
2994 if (ln && regrepeat_hard(scan, ln, &l) < ln)
2995 sayNO;
5f4b28b2 2996 if (ln && l == 0 && n >= ln
c277df42
IZ
2997 /* In fact, this is tricky. If paren, then the
2998 fact that we did/didnot match may influence
2999 future execution. */
3000 && !(paren && ln == 0))
3001 ln = n;
3280af22 3002 locinput = PL_reginput;
22c35a8c 3003 if (PL_regkind[(U8)OP(next)] == EXACT) {
cd439c50 3004 c1 = (U8)*STRING(next);
c277df42 3005 if (OP(next) == EXACTF)
22c35a8c 3006 c2 = PL_fold[c1];
c277df42 3007 else if (OP(next) == EXACTFL)
22c35a8c 3008 c2 = PL_fold_locale[c1];
c277df42
IZ
3009 else
3010 c2 = c1;
a0ed51b3
LW
3011 }
3012 else
c277df42 3013 c1 = c2 = -1000;
02db2b7b 3014 REGCP_SET(lastcp);
5f4b28b2 3015 /* This may be improved if l == 0. */
c277df42
IZ
3016 while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */
3017 /* If it could work, try it. */
3018 if (c1 == -1000 ||
3280af22
NIS
3019 UCHARAT(PL_reginput) == c1 ||
3020 UCHARAT(PL_reginput) == c2)
c277df42
IZ
3021 {
3022 if (paren) {
3023 if (n) {
cf93c79d
IZ
3024 PL_regstartp[paren] =
3025 HOPc(PL_reginput, -l) - PL_bostr;
3026 PL_regendp[paren] = PL_reginput - PL_bostr;
a0ed51b3
LW
3027 }
3028 else
cf93c79d 3029 PL_regendp[paren] = -1;
c277df42
IZ
3030 }
3031 if (regmatch(next))
3032 sayYES;
02db2b7b 3033 REGCP_UNWIND(lastcp);
c277df42
IZ
3034 }
3035 /* Couldn't or didn't -- move forward. */
3280af22 3036 PL_reginput = locinput;
c277df42
IZ
3037 if (regrepeat_hard(scan, 1, &l)) {
3038 ln++;
3280af22 3039 locinput = PL_reginput;
c277df42
IZ
3040 }
3041 else
3042 sayNO;
3043 }
a0ed51b3
LW
3044 }
3045 else {
c277df42
IZ
3046 n = regrepeat_hard(scan, n, &l);
3047 if (n != 0 && l == 0
3048 /* In fact, this is tricky. If paren, then the
3049 fact that we did/didnot match may influence
3050 future execution. */
3051 && !(paren && ln == 0))
3052 ln = n;
3280af22 3053 locinput = PL_reginput;
c277df42 3054 DEBUG_r(
5c0ca799 3055 PerlIO_printf(Perl_debug_log,
faccc32b 3056 "%*s matched %"IVdf" times, len=%"IVdf"...\n",
7b0972df 3057 (int)(REPORT_CODE_OFF+PL_regindent*2), "",
faccc32b 3058 (IV) n, (IV)l)
c277df42
IZ
3059 );
3060 if (n >= ln) {
22c35a8c 3061 if (PL_regkind[(U8)OP(next)] == EXACT) {
cd439c50 3062 c1 = (U8)*STRING(next);
c277df42 3063 if (OP(next) == EXACTF)
22c35a8c 3064 c2 = PL_fold[c1];
c277df42 3065 else if (OP(next) == EXACTFL)
22c35a8c 3066 c2 = PL_fold_locale[c1];
c277df42
IZ
3067 else
3068 c2 = c1;
a0ed51b3
LW
3069 }
3070 else
c277df42
IZ
3071 c1 = c2 = -1000;
3072 }
02db2b7b 3073 REGCP_SET(lastcp);
c277df42
IZ
3074 while (n >= ln) {
3075 /* If it could work, try it. */
3076 if (c1 == -1000 ||
3280af22
NIS
3077 UCHARAT(PL_reginput) == c1 ||
3078 UCHARAT(PL_reginput) == c2)
a0ed51b3
LW
3079 {
3080 DEBUG_r(
c3464db5 3081 PerlIO_printf(Perl_debug_log,
7b0972df
JH
3082 "%*s trying tail with n=%"IVdf"...\n",
3083 (int)(REPORT_CODE_OFF+PL_regindent*2), "", (IV)n)
a0ed51b3
LW
3084 );
3085 if (paren) {
3086 if (n) {
cf93c79d
IZ
3087 PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr;
3088 PL_regendp[paren] = PL_reginput - PL_bostr;
c277df42 3089 }
a0ed51b3 3090 else
cf93c79d 3091 PL_regendp[paren] = -1;
c277df42 3092 }
a0ed51b3
LW
3093 if (regmatch(next))
3094 sayYES;
02db2b7b 3095 REGCP_UNWIND(lastcp);
a0ed51b3 3096 }
c277df42
IZ
3097 /* Couldn't or didn't -- back up. */
3098 n--;
dfe13c55 3099 locinput = HOPc(locinput, -l);
3280af22 3100 PL_reginput = locinput;
c277df42
IZ
3101 }
3102 }
3103 sayNO;
3104 break;
3105 }
3106 case CURLYN:
3107 paren = scan->flags; /* Which paren to set */
3280af22
NIS
3108 if (paren > PL_regsize)
3109 PL_regsize = paren;
3110 if (paren > *PL_reglastparen)
3111 *PL_reglastparen = paren;
c277df42
IZ
3112 ln = ARG1(scan); /* min to match */
3113 n = ARG2(scan); /* max to match */
dc45a647 3114 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
c277df42 3115 goto repeat;
a0d0e21e 3116 case CURLY:
c277df42 3117 paren = 0;
a0d0e21e
LW
3118 ln = ARG1(scan); /* min to match */
3119 n = ARG2(scan); /* max to match */
dc45a647 3120 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
a0d0e21e
LW
3121 goto repeat;
3122 case STAR:
3123 ln = 0;
c277df42 3124 n = REG_INFTY;
a0d0e21e 3125 scan = NEXTOPER(scan);
c277df42 3126 paren = 0;
a0d0e21e
LW
3127 goto repeat;
3128 case PLUS:
c277df42
IZ
3129 ln = 1;
3130 n = REG_INFTY;
3131 scan = NEXTOPER(scan);
3132 paren = 0;
3133 repeat:
a0d0e21e
LW
3134 /*
3135 * Lookahead to avoid useless match attempts
3136 * when we know what character comes next.
3137 */
22c35a8c 3138 if (PL_regkind[(U8)OP(next)] == EXACT) {
1aa99e6b
IH
3139 U8 *s = (U8*)STRING(next);
3140 if (!UTF) {
3141 c2 = c1 = *s;
3142 if (OP(next) == EXACTF)
3143 c2 = PL_fold[c1];
3144 else if (OP(next) == EXACTFL)
3145 c2 = PL_fold_locale[c1];
3146 }
3147 else { /* UTF */
3148 if (OP(next) == EXACTF) {
3149 c1 = to_utf8_lower(s);
3150 c2 = to_utf8_upper(s);
3151 }
3152 else {
9041c2e3 3153 c2 = c1 = utf8_to_uvchr(s, NULL);
1aa99e6b
IH
3154 }
3155 }
bbce6d69 3156 }
a0d0e21e 3157 else
bbce6d69 3158 c1 = c2 = -1000;
3280af22 3159 PL_reginput = locinput;
a0d0e21e 3160 if (minmod) {
c277df42 3161 CHECKPOINT lastcp;
a0d0e21e
LW
3162 minmod = 0;
3163 if (ln && regrepeat(scan, ln) < ln)
4633a7c4 3164 sayNO;
a0ed51b3 3165 locinput = PL_reginput;
02db2b7b 3166 REGCP_SET(lastcp);
0fe9bf95 3167 if (c1 != -1000) {
1aa99e6b 3168 char *e; /* Should not check after this */
0fe9bf95
IZ
3169 char *old = locinput;
3170
1aa99e6b 3171 if (n == REG_INFTY) {
0fe9bf95 3172 e = PL_regeol - 1;
1aa99e6b
IH
3173 if (do_utf8)
3174 while (UTF8_IS_CONTINUATION(*(U8*)e))
3175 e--;
3176 }
3177 else if (do_utf8) {
3178 int m = n - ln;
3179 for (e = locinput;
3180 m >0 && e + UTF8SKIP(e) <= PL_regeol; m--)
3181 e += UTF8SKIP(e);
3182 }
3183 else {
3184 e = locinput + n - ln;
3185 if (e >= PL_regeol)
3186 e = PL_regeol - 1;
3187 }
0fe9bf95 3188 while (1) {
1aa99e6b 3189 int count;
0fe9bf95 3190 /* Find place 'next' could work */
1aa99e6b
IH
3191 if (!do_utf8) {
3192 if (c1 == c2) {
3193 while (locinput <= e && *locinput != c1)
3194 locinput++;
3195 } else {
9041c2e3 3196 while (locinput <= e
1aa99e6b
IH
3197 && *locinput != c1
3198 && *locinput != c2)
3199 locinput++;
3200 }
3201 count = locinput - old;
3202 }
3203 else {
3204 STRLEN len;
3205 if (c1 == c2) {
3206 for (count = 0;
3207 locinput <= e &&
9041c2e3 3208 utf8_to_uvchr((U8*)locinput, &len) != c1;
1aa99e6b
IH
3209 count++)
3210 locinput += len;
3211
3212 } else {
3213 for (count = 0; locinput <= e; count++) {
9041c2e3 3214 UV c = utf8_to_uvchr((U8*)locinput, &len);
1aa99e6b
IH
3215 if (c == c1 || c == c2)
3216 break;
9041c2e3 3217 locinput += len;
1aa99e6b
IH
3218 }
3219 }
0fe9bf95 3220 }
9041c2e3 3221 if (locinput > e)
0fe9bf95
IZ
3222 sayNO;
3223 /* PL_reginput == old now */
3224 if (locinput != old) {
3225 ln = 1; /* Did some */
1aa99e6b 3226 if (regrepeat(scan, count) < count)
0fe9bf95
IZ
3227 sayNO;
3228 }
3229 /* PL_reginput == locinput now */
29d1e993 3230 TRYPAREN(paren, ln, locinput);
0fe9bf95 3231 PL_reginput = locinput; /* Could be reset... */
02db2b7b 3232 REGCP_UNWIND(lastcp);
0fe9bf95 3233 /* Couldn't or didn't -- move forward. */
1aa99e6b
IH
3234 old = locinput;
3235 if (do_utf8)
3236 locinput += UTF8SKIP(locinput);
3237 else
3238 locinput++;
0fe9bf95
IZ
3239 }
3240 }
3241 else
c277df42 3242 while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
1aa99e6b
IH
3243 UV c;
3244 if (c1 != -1000) {
3245 if (do_utf8)
9041c2e3 3246 c = utf8_to_uvchr((U8*)PL_reginput, NULL);
1aa99e6b 3247 else
9041c2e3 3248 c = UCHARAT(PL_reginput);
2390ecbc
PP
3249 /* If it could work, try it. */
3250 if (c == c1 || c == c2)
3251 {
3252 TRYPAREN(paren, n, PL_reginput);
3253 REGCP_UNWIND(lastcp);
3254 }
1aa99e6b 3255 }
a0d0e21e 3256 /* If it could work, try it. */
2390ecbc 3257 else if (c1 == -1000)
bbce6d69 3258 {
29d1e993 3259 TRYPAREN(paren, n, PL_reginput);
02db2b7b 3260 REGCP_UNWIND(lastcp);
bbce6d69 3261 }
c277df42 3262 /* Couldn't or didn't -- move forward. */
a0ed51b3 3263 PL_reginput = locinput;
a0d0e21e
LW
3264 if (regrepeat(scan, 1)) {
3265 ln++;
a0ed51b3
LW
3266 locinput = PL_reginput;
3267 }
3268 else
4633a7c4 3269 sayNO;
a0d0e21e
LW
3270 }
3271 }
3272 else {
c277df42 3273 CHECKPOINT lastcp;
a0d0e21e 3274 n = regrepeat(scan, n);
a0ed51b3 3275 locinput = PL_reginput;
22c35a8c 3276 if (ln < n && PL_regkind[(U8)OP(next)] == EOL &&
1aeab75a 3277 (!PL_multiline || OP(next) == SEOL || OP(next) == EOS)) {
a0d0e21e 3278 ln = n; /* why back off? */
1aeab75a
GS
3279 /* ...because $ and \Z can match before *and* after
3280 newline at the end. Consider "\n\n" =~ /\n+\Z\n/.
3281 We should back off by one in this case. */
3282 if (UCHARAT(PL_reginput - 1) == '\n' && OP(next) != EOS)
3283 ln--;
3284 }
02db2b7b 3285 REGCP_SET(lastcp);
c277df42 3286 if (paren) {
1aa99e6b 3287 UV c;
c277df42 3288 while (n >= ln) {
1aa99e6b
IH
3289 if (c1 != -1000) {
3290 if (do_utf8)
9041c2e3 3291 c = utf8_to_uvchr((U8*)PL_reginput, NULL);
1aa99e6b 3292 else
9041c2e3 3293 c = UCHARAT(PL_reginput);
1aa99e6b 3294 }
c277df42 3295 /* If it could work, try it. */
1aa99e6b 3296 if (c1 == -1000 || c == c1 || c == c2)
c277df42 3297 {
29d1e993 3298 TRYPAREN(paren, n, PL_reginput);
02db2b7b 3299 REGCP_UNWIND(lastcp);
c277df42
IZ
3300 }
3301 /* Couldn't or didn't -- back up. */
3302 n--;
dfe13c55 3303 PL_reginput = locinput = HOPc(locinput, -1);
c277df42 3304 }
a0ed51b3
LW
3305 }
3306 else {
1aa99e6b 3307 UV c;
c277df42 3308 while (n >= ln) {
1aa99e6b
IH
3309 if (c1 != -1000) {
3310 if (do_utf8)
9041c2e3 3311 c = utf8_to_uvchr((U8*)PL_reginput, NULL);
1aa99e6b 3312 else
9041c2e3 3313 c = UCHARAT(PL_reginput);
1aa99e6b 3314 }
c277df42 3315 /* If it could work, try it. */
1aa99e6b 3316 if (c1 == -1000 || c == c1 || c == c2)
c277df42 3317 {
29d1e993 3318 TRYPAREN(paren, n, PL_reginput);
02db2b7b 3319 REGCP_UNWIND(lastcp);
c277df42
IZ
3320 }
3321 /* Couldn't or didn't -- back up. */
3322 n--;
dfe13c55 3323 PL_reginput = locinput = HOPc(locinput, -1);
bbce6d69 3324 }
a0d0e21e
LW
3325 }
3326 }
4633a7c4 3327 sayNO;
c277df42 3328 break;
a0d0e21e 3329 case END:
0f5d15d6
IZ
3330 if (PL_reg_call_cc) {
3331 re_cc_state *cur_call_cc = PL_reg_call_cc;
3332 CURCUR *cctmp = PL_regcc;
3333 regexp *re = PL_reg_re;
3334 CHECKPOINT cp, lastcp;
3335
3336 cp = regcppush(0); /* Save *all* the positions. */
02db2b7b 3337 REGCP_SET(lastcp);
0f5d15d6
IZ
3338 regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of
3339 the caller. */
3340 PL_reginput = locinput; /* Make position available to
3341 the callcc. */
3342 cache_re(PL_reg_call_cc->re);
3343 PL_regcc = PL_reg_call_cc->cc;
3344 PL_reg_call_cc = PL_reg_call_cc->prev;
3345 if (regmatch(cur_call_cc->node)) {
3346 PL_reg_call_cc = cur_call_cc;
3347 regcpblow(cp);
3348 sayYES;
3349 }
02db2b7b 3350 REGCP_UNWIND(lastcp);
0f5d15d6
IZ
3351 regcppop();
3352 PL_reg_call_cc = cur_call_cc;
3353 PL_regcc = cctmp;
3354 PL_reg_re = re;
3355 cache_re(re);
3356
3357 DEBUG_r(
3358 PerlIO_printf(Perl_debug_log,
3359 "%*s continuation failed...\n",
3360 REPORT_CODE_OFF+PL_regindent*2, "")
3361 );
7821416a 3362 sayNO_SILENT;
0f5d15d6 3363 }
7821416a
IZ
3364 if (locinput < PL_regtill) {
3365 DEBUG_r(PerlIO_printf(Perl_debug_log,
3366 "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
3367 PL_colors[4],
3368 (long)(locinput - PL_reg_starttry),
3369 (long)(PL_regtill - PL_reg_starttry),
3370 PL_colors[5]));
3371 sayNO_FINAL; /* Cannot match: too short. */
3372 }
3373 PL_reginput = locinput; /* put where regtry can find it */
3374 sayYES_FINAL; /* Success! */
7e5428c5 3375 case SUCCEED:
3280af22 3376 PL_reginput = locinput; /* put where regtry can find it */
7821416a 3377 sayYES_LOUD; /* Success! */
c277df42
IZ
3378 case SUSPEND:
3379 n = 1;
9fe1d20c 3380 PL_reginput = locinput;
9041c2e3 3381 goto do_ifmatch;
a0d0e21e 3382 case UNLESSM:
c277df42 3383 n = 0;
a0ed51b3 3384 if (scan->flags) {
0fe9bf95
IZ
3385 if (UTF) { /* XXXX This is absolutely
3386 broken, we read before
3387 start of string. */
3388 s = HOPMAYBEc(locinput, -scan->flags);
3389 if (!s)
3390 goto say_yes;
3391 PL_reginput = s;
3392 }
3393 else {
9041c2e3 3394 if (locinput < PL_bostr + scan->flags)
0fe9bf95
IZ
3395 goto say_yes;
3396 PL_reginput = locinput - scan->flags;
3397 goto do_ifmatch;
3398 }
a0ed51b3
LW
3399 }
3400 else
3401 PL_reginput = locinput;
c277df42
IZ
3402 goto do_ifmatch;
3403 case IFMATCH:
3404 n = 1;
a0ed51b3 3405 if (scan->flags) {
0fe9bf95
IZ
3406 if (UTF) { /* XXXX This is absolutely
3407 broken, we read before
3408 start of string. */
3409 s = HOPMAYBEc(locinput, -scan->flags);
3410 if (!s || s < PL_bostr)
3411 goto say_no;
3412 PL_reginput = s;
3413 }
3414 else {
9041c2e3 3415 if (locinput < PL_bostr + scan->flags)
0fe9bf95
IZ
3416 goto say_no;
3417 PL_reginput = locinput - scan->flags;
3418 goto do_ifmatch;
3419 }
a0ed51b3
LW
3420 }
3421 else
3422 PL_reginput = locinput;
3423
c277df42 3424 do_ifmatch:
c277df42
IZ
3425 inner = NEXTOPER(NEXTOPER(scan));
3426 if (regmatch(inner) != n) {
3427 say_no:
3428 if (logical) {
3429 logical = 0;
3430 sw = 0;
3431 goto do_longjump;
a0ed51b3
LW
3432 }
3433 else
c277df42
IZ
3434 sayNO;
3435 }
3436 say_yes:
3437 if (logical) {
3438 logical = 0;
3439 sw = 1;
3440 }
fe44a5e8 3441 if (OP(scan) == SUSPEND) {
3280af22 3442 locinput = PL_reginput;
565764a8 3443 nextchr = UCHARAT(locinput);
fe44a5e8 3444 }
c277df42
IZ
3445 /* FALL THROUGH. */
3446 case LONGJMP:
3447 do_longjump:
3448 next = scan + ARG(scan);
3449 if (next == scan)
3450 next = NULL;
a0d0e21e
LW
3451 break;
3452 default:
b900a521 3453 PerlIO_printf(Perl_error_log, "%"UVxf" %d\n",
d7d93a81 3454 PTR2UV(scan), OP(scan));
cea2e8a9 3455 Perl_croak(aTHX_ "regexp memory corruption");
a687059c 3456 }
02db2b7b 3457 reenter:
a0d0e21e
LW
3458 scan = next;
3459 }
a687059c 3460
a0d0e21e
LW
3461 /*
3462 * We get here only if there's trouble -- normally "case END" is
3463 * the terminating point.
3464 */
cea2e8a9 3465 Perl_croak(aTHX_ "corrupted regexp pointers");
a0d0e21e 3466 /*NOTREACHED*/
4633a7c4
LW
3467 sayNO;
3468
7821416a
IZ
3469yes_loud:
3470 DEBUG_r(
3471 PerlIO_printf(Perl_debug_log,
3472 "%*s %scould match...%s\n",
3473 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],PL_colors[5])
3474 );
3475 goto yes;
3476yes_final:
3477 DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
3478 PL_colors[4],PL_colors[5]));
4633a7c4
LW
3479yes:
3480#ifdef DEBUGGING
3280af22 3481 PL_regindent--;
4633a7c4 3482#endif
02db2b7b
IZ
3483
3484#if 0 /* Breaks $^R */
3485 if (unwind)
3486 regcpblow(firstcp);
3487#endif
4633a7c4
LW
3488 return 1;
3489
3490no:
7821416a
IZ
3491 DEBUG_r(
3492 PerlIO_printf(Perl_debug_log,
3493 "%*s %sfailed...%s\n",
3494 REPORT_CODE_OFF+PL_regindent*2, "",PL_colors[4],PL_colors[5])
3495 );
3496 goto do_no;
3497no_final:
3498do_no:
02db2b7b
IZ
3499 if (unwind) {
3500 re_unwind_t *uw = SSPTRt(unwind,re_unwind_t);
3501
3502 switch (uw->type) {
3503 case RE_UNWIND_BRANCH:
3504 case RE_UNWIND_BRANCHJ:
3505 {
3506 re_unwind_branch_t *uwb = &(uw->branch);
3507 I32 lastparen = uwb->lastparen;
9041c2e3 3508
02db2b7b
IZ
3509 REGCP_UNWIND(uwb->lastcp);
3510 for (n = *PL_reglastparen; n > lastparen; n--)
3511 PL_regendp[n] = -1;
3512 *PL_reglastparen = n;
3513 scan = next = uwb->next;
9041c2e3
NIS
3514 if ( !scan ||
3515 OP(scan) != (uwb->type == RE_UNWIND_BRANCH
02db2b7b
IZ
3516 ? BRANCH : BRANCHJ) ) { /* Failure */
3517 unwind = uwb->prev;
3518#ifdef DEBUGGING
3519 PL_regindent--;
3520#endif
3521 goto do_no;
3522 }
3523 /* Have more choice yet. Reuse the same uwb. */
3524 /*SUPPRESS 560*/
3525 if ((n = (uwb->type == RE_UNWIND_BRANCH
3526 ? NEXT_OFF(next) : ARG(next))))
3527 next += n;
3528 else
3529 next = NULL; /* XXXX Needn't unwinding in this case... */
3530 uwb->next = next;
3531 next = NEXTOPER(scan);
3532 if (uwb->type == RE_UNWIND_BRANCHJ)
3533 next = NEXTOPER(next);
3534 locinput = uwb->locinput;
3535 nextchr = uwb->nextchr;
3536#ifdef DEBUGGING
3537 PL_regindent = uwb->regindent;
3538#endif
3539
3540 goto reenter;
3541 }
3542 /* NOT REACHED */
3543 default:
3544 Perl_croak(aTHX_ "regexp unwind memory corruption");
3545 }
3546 /* NOT REACHED */
3547 }
4633a7c4 3548#ifdef DEBUGGING
3280af22 3549 PL_regindent--;
4633a7c4 3550#endif
a0d0e21e 3551 return 0;
a687059c
LW
3552}
3553
3554/*
3555 - regrepeat - repeatedly match something simple, report how many
3556 */
3557/*
3558 * [This routine now assumes that it will only match on things of length 1.
3559 * That was true before, but now we assume scan - reginput is the count,
a0ed51b3 3560 * rather than incrementing count on every character. [Er, except utf8.]]
a687059c 3561 */
76e3520e 3562STATIC I32
cea2e8a9 3563S_regrepeat(pTHX_ regnode *p, I32 max)
a687059c 3564{
a0d0e21e 3565 register char *scan;
a0d0e21e 3566 register I32 c;
3280af22 3567 register char *loceol = PL_regeol;
a0ed51b3 3568 register I32 hardcount = 0;
ffc61ed2 3569 register bool do_utf8 = DO_UTF8(PL_reg_sv);
a0d0e21e 3570
3280af22 3571 scan = PL_reginput;
c277df42 3572 if (max != REG_INFTY && max < loceol - scan)
a0d0e21e 3573 loceol = scan + max;
a0d0e21e 3574 switch (OP(p)) {
22c35a8c 3575 case REG_ANY:
1aa99e6b 3576 if (do_utf8) {
ffc61ed2 3577 loceol = PL_regeol;
1aa99e6b 3578 while (scan < loceol && hardcount < max && *scan != '\n') {
ffc61ed2
JH
3579 scan += UTF8SKIP(scan);
3580 hardcount++;
3581 }
3582 } else {
3583 while (scan < loceol && *scan != '\n')
3584 scan++;
a0ed51b3
LW
3585 }
3586 break;
ffc61ed2 3587 case SANY:
3baa4c62 3588 scan = loceol;
a0ed51b3 3589 break;
bbce6d69 3590 case EXACT: /* length of string is 1 */
cd439c50 3591 c = (U8)*STRING(p);
bbce6d69
PP
3592 while (scan < loceol && UCHARAT(scan) == c)
3593 scan++;
3594 break;
3595 case EXACTF: /* length of string is 1 */
cd439c50 3596 c = (U8)*STRING(p);
bbce6d69 3597 while (scan < loceol &&
22c35a8c 3598 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c]))
bbce6d69
PP
3599 scan++;
3600 break;
3601 case EXACTFL: /* length of string is 1 */
3280af22 3602 PL_reg_flags |= RF_tainted;
cd439c50 3603 c = (U8)*STRING(p);
bbce6d69 3604 while (scan < loceol &&
22c35a8c 3605 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c]))
a0d0e21e
LW
3606 scan++;
3607 break;
3608 case ANYOF:
ffc61ed2
JH
3609 if (do_utf8) {
3610 loceol = PL_regeol;
1aa99e6b
IH
3611 while (hardcount < max && scan < loceol &&
3612 reginclass(p, (U8*)scan, do_utf8)) {
ffc61ed2
JH
3613 scan += UTF8SKIP(scan);
3614 hardcount++;
3615 }
3616 } else {
3617 while (scan < loceol && reginclass(p, (U8*)scan, do_utf8))
3618 scan++;
3619 }
a0d0e21e
LW
3620 break;
3621 case ALNUM:
1aa99e6b 3622 if (do_utf8) {
ffc61ed2 3623 loceol = PL_regeol;
8269fa76 3624 LOAD_UTF8_CHARCLASS(alnum,"a");
1aa99e6b
IH
3625 while (hardcount < max && scan < loceol &&
3626 swash_fetch(PL_utf8_alnum, (U8*)scan)) {
ffc61ed2
JH
3627 scan += UTF8SKIP(scan);
3628 hardcount++;
3629 }
3630 } else {
3631 while (scan < loceol && isALNUM(*scan))
3632 scan++;
a0ed51b3
LW
3633 }
3634 break;
bbce6d69 3635 case ALNUML: