This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Minor grammar fix by Uri Guttman
[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
61296642
DM
8/* This file contains functions for executing a regular expression. See
9 * also regcomp.c which funnily enough, contains functions for compiling
166f8a29 10 * a regular expression.
e4a054ea
DM
11 *
12 * This file is also copied at build time to ext/re/re_exec.c, where
13 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
14 * This causes the main functions to be compiled under new names and with
15 * debugging support added, which makes "use re 'debug'" work.
16
166f8a29
DM
17 */
18
a687059c
LW
19/* NOTE: this is derived from Henry Spencer's regexp code, and should not
20 * confused with the original package (see point 3 below). Thanks, Henry!
21 */
22
23/* Additional note: this code is very heavily munged from Henry's version
24 * in places. In some spots I've traded clarity for efficiency, so don't
25 * blame Henry for some of the lack of readability.
26 */
27
e50aee73
AD
28/* The names of the functions have been changed from regcomp and
29 * regexec to pregcomp and pregexec in order to avoid conflicts
30 * with the POSIX routines of the same names.
31*/
32
b9d5759e
AD
33#ifdef PERL_EXT_RE_BUILD
34/* need to replace pregcomp et al, so enable that */
35# ifndef PERL_IN_XSUB_RE
36# define PERL_IN_XSUB_RE
37# endif
38/* need access to debugger hooks */
cad2e5aa 39# if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
b9d5759e
AD
40# define DEBUGGING
41# endif
42#endif
43
44#ifdef PERL_IN_XSUB_RE
d06ea78c 45/* We *really* need to overwrite these symbols: */
56953603
IZ
46# define Perl_regexec_flags my_regexec
47# define Perl_regdump my_regdump
48# define Perl_regprop my_regprop
cad2e5aa 49# define Perl_re_intuit_start my_re_intuit_start
d06ea78c
GS
50/* *These* symbols are masked to allow static link. */
51# define Perl_pregexec my_pregexec
9041c2e3 52# define Perl_reginitcolors my_reginitcolors
490a3f88 53# define Perl_regclass_swash my_regclass_swash
c5be433b
GS
54
55# define PERL_NO_GET_CONTEXT
9041c2e3 56#endif
56953603 57
f0fcb552 58/*SUPPRESS 112*/
a687059c 59/*
e50aee73 60 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
61 *
62 * Copyright (c) 1986 by University of Toronto.
63 * Written by Henry Spencer. Not derived from licensed software.
64 *
65 * Permission is granted to anyone to use this software for any
66 * purpose on any computer system, and to redistribute it freely,
67 * subject to the following restrictions:
68 *
69 * 1. The author is not responsible for the consequences of use of
70 * this software, no matter how awful, even if they arise
71 * from defects in it.
72 *
73 * 2. The origin of this software must not be misrepresented, either
74 * by explicit claim or by omission.
75 *
76 * 3. Altered versions must be plainly marked as such, and must not
77 * be misrepresented as being the original software.
78 *
79 **** Alterations to Henry's code are...
80 ****
4bb101f2 81 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
f2b990bf 82 **** 2000, 2001, 2002, 2003, 2004, 2005, by Larry Wall and others
a687059c 83 ****
9ef589d8
LW
84 **** You may distribute under the terms of either the GNU General Public
85 **** License or the Artistic License, as specified in the README file.
a687059c
LW
86 *
87 * Beware that some of this code is subtly aware of the way operator
88 * precedence is structured in regular expressions. Serious changes in
89 * regular-expression syntax might require a total rethink.
90 */
91#include "EXTERN.h"
864dbfa3 92#define PERL_IN_REGEXEC_C
a687059c 93#include "perl.h"
0f5d15d6 94
a687059c
LW
95#include "regcomp.h"
96
c277df42
IZ
97#define RF_tainted 1 /* tainted information used? */
98#define RF_warned 2 /* warned about big count? */
ce862d02 99#define RF_evaled 4 /* Did an EVAL with setting? */
a0ed51b3
LW
100#define RF_utf8 8 /* String contains multibyte chars? */
101
eb160463 102#define UTF ((PL_reg_flags & RF_utf8) != 0)
ce862d02
IZ
103
104#define RS_init 1 /* eval environment created */
105#define RS_set 2 /* replsv value is set */
c277df42 106
a687059c
LW
107#ifndef STATIC
108#define STATIC static
109#endif
110
ba7b4546 111#define REGINCLASS(p,c) (ANYOF_FLAGS(p) ? reginclass(p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
7d3e948e 112
c277df42
IZ
113/*
114 * Forwards.
115 */
116
33b8afdf 117#define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv))
53c4c00c 118#define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)
a0ed51b3 119
dfe13c55
GS
120#define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
121#define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
53c4c00c
JH
122#define HOP(pos,off) (PL_reg_match_utf8 ? reghop((U8*)pos, off) : (U8*)(pos + off))
123#define HOPMAYBE(pos,off) (PL_reg_match_utf8 ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
dfe13c55
GS
124#define HOPc(pos,off) ((char*)HOP(pos,off))
125#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
a0d0e21e 126
efb30f32 127#define HOPBACK(pos, off) ( \
e54858b0 128 (PL_reg_match_utf8) \
efb30f32
HS
129 ? reghopmaybe((U8*)pos, -off) \
130 : (pos - off >= PL_bostr) \
131 ? (U8*)(pos - off) \
132 : (U8*)NULL \
133)
134#define HOPBACKc(pos, off) (char*)HOPBACK(pos, off)
135
1aa99e6b
IH
136#define reghop3_c(pos,off,lim) ((char*)reghop3((U8*)pos, off, (U8*)lim))
137#define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim))
53c4c00c
JH
138#define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
139#define HOPMAYBE3(pos,off,lim) (PL_reg_match_utf8 ? reghopmaybe3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
1aa99e6b
IH
140#define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
141#define HOPMAYBE3c(pos,off,lim) ((char*)HOPMAYBE3(pos,off,lim))
142
bfed75c6 143#define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) { ENTER; save_re_context(); (void)CAT2(is_utf8_, a)((const U8*)b); LEAVE; } } STMT_END
51371543 144
5f80c4cf 145/* for use after a quantifier and before an EXACT-like node -- japhy */
e2d8ce26
JP
146#define JUMPABLE(rn) ( \
147 OP(rn) == OPEN || OP(rn) == CLOSE || OP(rn) == EVAL || \
cca55fe3
JP
148 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
149 OP(rn) == PLUS || OP(rn) == MINMOD || \
150 (PL_regkind[(U8)OP(rn)] == CURLY && ARG1(rn) > 0) \
e2d8ce26
JP
151)
152
cca55fe3
JP
153#define HAS_TEXT(rn) ( \
154 PL_regkind[(U8)OP(rn)] == EXACT || PL_regkind[(U8)OP(rn)] == REF \
155)
e2d8ce26 156
a84d97b6
HS
157/*
158 Search for mandatory following text node; for lookahead, the text must
159 follow but for lookbehind (rn->flags != 0) we skip to the next step.
160*/
cca55fe3 161#define FIND_NEXT_IMPT(rn) STMT_START { \
e2d8ce26 162 while (JUMPABLE(rn)) \
a84d97b6 163 if (OP(rn) == SUSPEND || PL_regkind[(U8)OP(rn)] == CURLY) \
e2d8ce26 164 rn = NEXTOPER(NEXTOPER(rn)); \
cca55fe3
JP
165 else if (OP(rn) == PLUS) \
166 rn = NEXTOPER(rn); \
a84d97b6
HS
167 else if (OP(rn) == IFMATCH) \
168 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
e2d8ce26 169 else rn += NEXT_OFF(rn); \
5f80c4cf 170} STMT_END
74750237 171
acfe0abc 172static void restore_pos(pTHX_ void *arg);
51371543 173
76e3520e 174STATIC CHECKPOINT
cea2e8a9 175S_regcppush(pTHX_ I32 parenfloor)
a0d0e21e 176{
a3b680e6 177 const int retval = PL_savestack_ix;
b1ce53c5 178#define REGCP_PAREN_ELEMS 4
a3b680e6 179 const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
a0d0e21e
LW
180 int p;
181
e49a9654
IH
182 if (paren_elems_to_push < 0)
183 Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
184
a01268b5 185#define REGCP_OTHER_ELEMS 6
4b3c1a47 186 SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
3280af22 187 for (p = PL_regsize; p > parenfloor; p--) {
b1ce53c5 188/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
cf93c79d
IZ
189 SSPUSHINT(PL_regendp[p]);
190 SSPUSHINT(PL_regstartp[p]);
3280af22 191 SSPUSHPTR(PL_reg_start_tmp[p]);
a0d0e21e
LW
192 SSPUSHINT(p);
193 }
b1ce53c5 194/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
3280af22
NIS
195 SSPUSHINT(PL_regsize);
196 SSPUSHINT(*PL_reglastparen);
a01268b5 197 SSPUSHINT(*PL_reglastcloseparen);
3280af22 198 SSPUSHPTR(PL_reginput);
41123dfd
JH
199#define REGCP_FRAME_ELEMS 2
200/* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and
201 * are needed for the regexp context stack bookkeeping. */
202 SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
b1ce53c5 203 SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */
41123dfd 204
a0d0e21e
LW
205 return retval;
206}
207
c277df42 208/* These are needed since we do not localize EVAL nodes: */
a3621e74 209# define REGCP_SET(cp) DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, \
faccc32b 210 " Setting an EVAL scope, savestack=%"IVdf"\n", \
02db2b7b 211 (IV)PL_savestack_ix)); cp = PL_savestack_ix
c3464db5 212
a3621e74 213# define REGCP_UNWIND(cp) DEBUG_EXECUTE_r(cp != PL_savestack_ix ? \
c3464db5 214 PerlIO_printf(Perl_debug_log, \
faccc32b 215 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
02db2b7b 216 (IV)(cp), (IV)PL_savestack_ix) : 0); regcpblow(cp)
c277df42 217
76e3520e 218STATIC char *
cea2e8a9 219S_regcppop(pTHX)
a0d0e21e 220{
b1ce53c5 221 I32 i;
a0d0e21e
LW
222 U32 paren = 0;
223 char *input;
cf93c79d 224 I32 tmps;
b1ce53c5 225
a3621e74
YO
226 GET_RE_DEBUG_FLAGS_DECL;
227
b1ce53c5 228 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
a0d0e21e 229 i = SSPOPINT;
b1ce53c5
JH
230 assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
231 i = SSPOPINT; /* Parentheses elements to pop. */
a0d0e21e 232 input = (char *) SSPOPPTR;
a01268b5 233 *PL_reglastcloseparen = SSPOPINT;
3280af22
NIS
234 *PL_reglastparen = SSPOPINT;
235 PL_regsize = SSPOPINT;
b1ce53c5
JH
236
237 /* Now restore the parentheses context. */
41123dfd
JH
238 for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
239 i > 0; i -= REGCP_PAREN_ELEMS) {
a0d0e21e 240 paren = (U32)SSPOPINT;
3280af22 241 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
cf93c79d
IZ
242 PL_regstartp[paren] = SSPOPINT;
243 tmps = SSPOPINT;
3280af22
NIS
244 if (paren <= *PL_reglastparen)
245 PL_regendp[paren] = tmps;
a3621e74 246 DEBUG_EXECUTE_r(
c3464db5 247 PerlIO_printf(Perl_debug_log,
b900a521 248 " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
9041c2e3 249 (UV)paren, (IV)PL_regstartp[paren],
b900a521 250 (IV)(PL_reg_start_tmp[paren] - PL_bostr),
9041c2e3 251 (IV)PL_regendp[paren],
3280af22 252 (paren > *PL_reglastparen ? "(no)" : ""));
c277df42 253 );
a0d0e21e 254 }
a3621e74 255 DEBUG_EXECUTE_r(
eb160463 256 if ((I32)(*PL_reglastparen + 1) <= PL_regnpar) {
c3464db5 257 PerlIO_printf(Perl_debug_log,
faccc32b
JH
258 " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
259 (IV)(*PL_reglastparen + 1), (IV)PL_regnpar);
c277df42
IZ
260 }
261 );
daf18116 262#if 1
dafc8851
JH
263 /* It would seem that the similar code in regtry()
264 * already takes care of this, and in fact it is in
265 * a better location to since this code can #if 0-ed out
266 * but the code in regtry() is needed or otherwise tests
267 * requiring null fields (pat.t#187 and split.t#{13,14}
daf18116
JH
268 * (as of patchlevel 7877) will fail. Then again,
269 * this code seems to be necessary or otherwise
270 * building DynaLoader will fail:
271 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
272 * --jhi */
eb160463
GS
273 for (paren = *PL_reglastparen + 1; (I32)paren <= PL_regnpar; paren++) {
274 if ((I32)paren > PL_regsize)
cf93c79d
IZ
275 PL_regstartp[paren] = -1;
276 PL_regendp[paren] = -1;
a0d0e21e 277 }
dafc8851 278#endif
a0d0e21e
LW
279 return input;
280}
281
0f5d15d6 282STATIC char *
cea2e8a9 283S_regcp_set_to(pTHX_ I32 ss)
0f5d15d6 284{
a3b680e6 285 const I32 tmp = PL_savestack_ix;
0f5d15d6
IZ
286
287 PL_savestack_ix = ss;
288 regcppop();
289 PL_savestack_ix = tmp;
942e002e 290 return Nullch;
0f5d15d6
IZ
291}
292
293typedef struct re_cc_state
294{
295 I32 ss;
296 regnode *node;
297 struct re_cc_state *prev;
298 CURCUR *cc;
299 regexp *re;
300} re_cc_state;
301
02db2b7b 302#define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
a0d0e21e 303
29d1e993
HS
304#define TRYPAREN(paren, n, input) { \
305 if (paren) { \
306 if (n) { \
307 PL_regstartp[paren] = HOPc(input, -1) - PL_bostr; \
308 PL_regendp[paren] = input - PL_bostr; \
309 } \
310 else \
311 PL_regendp[paren] = -1; \
312 } \
313 if (regmatch(next)) \
314 sayYES; \
315 if (paren && n) \
316 PL_regendp[paren] = -1; \
317}
318
319
a687059c 320/*
e50aee73 321 * pregexec and friends
a687059c
LW
322 */
323
324/*
c277df42 325 - pregexec - match a regexp against a string
a687059c 326 */
c277df42 327I32
864dbfa3 328Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend,
c3464db5 329 char *strbeg, I32 minend, SV *screamer, U32 nosave)
c277df42
IZ
330/* strend: pointer to null at end of string */
331/* strbeg: real beginning of string */
332/* minend: end of match must be >=minend after stringarg. */
333/* nosave: For optimizations. */
334{
335 return
9041c2e3 336 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
c277df42
IZ
337 nosave ? 0 : REXEC_COPY_STR);
338}
0f5d15d6
IZ
339
340STATIC void
cea2e8a9 341S_cache_re(pTHX_ regexp *prog)
0f5d15d6
IZ
342{
343 PL_regprecomp = prog->precomp; /* Needed for FAIL. */
344#ifdef DEBUGGING
345 PL_regprogram = prog->program;
346#endif
347 PL_regnpar = prog->nparens;
9041c2e3
NIS
348 PL_regdata = prog->data;
349 PL_reg_re = prog;
0f5d15d6 350}
22e551b9 351
9041c2e3 352/*
cad2e5aa
JH
353 * Need to implement the following flags for reg_anch:
354 *
355 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
356 * USE_INTUIT_ML
357 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
358 * INTUIT_AUTORITATIVE_ML
359 * INTUIT_ONCE_NOML - Intuit can match in one location only.
360 * INTUIT_ONCE_ML
361 *
362 * Another flag for this function: SECOND_TIME (so that float substrs
363 * with giant delta may be not rechecked).
364 */
365
366/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
367
2c2d71f5 368/* If SCREAM, then SvPVX(sv) should be compatible with strpos and strend.
cad2e5aa
JH
369 Otherwise, only SvCUR(sv) is used to get strbeg. */
370
371/* XXXX We assume that strpos is strbeg unless sv. */
372
6eb5f6b9
JH
373/* XXXX Some places assume that there is a fixed substring.
374 An update may be needed if optimizer marks as "INTUITable"
375 RExen without fixed substrings. Similarly, it is assumed that
376 lengths of all the strings are no more than minlen, thus they
377 cannot come from lookahead.
378 (Or minlen should take into account lookahead.) */
379
2c2d71f5
JH
380/* A failure to find a constant substring means that there is no need to make
381 an expensive call to REx engine, thus we celebrate a failure. Similarly,
382 finding a substring too deep into the string means that less calls to
30944b6d
IZ
383 regtry() should be needed.
384
385 REx compiler's optimizer found 4 possible hints:
386 a) Anchored substring;
387 b) Fixed substring;
388 c) Whether we are anchored (beginning-of-line or \G);
389 d) First node (of those at offset 0) which may distingush positions;
6eb5f6b9 390 We use a)b)d) and multiline-part of c), and try to find a position in the
30944b6d
IZ
391 string which does not contradict any of them.
392 */
2c2d71f5 393
6eb5f6b9
JH
394/* Most of decisions we do here should have been done at compile time.
395 The nodes of the REx which we used for the search should have been
396 deleted from the finite automaton. */
397
cad2e5aa
JH
398char *
399Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
400 char *strend, U32 flags, re_scream_pos_data *data)
401{
b7953727 402 register I32 start_shift = 0;
cad2e5aa 403 /* Should be nonnegative! */
b7953727 404 register I32 end_shift = 0;
2c2d71f5
JH
405 register char *s;
406 register SV *check;
a1933d95 407 char *strbeg;
cad2e5aa 408 char *t;
a3b680e6 409 const int do_utf8 = sv ? SvUTF8(sv) : 0; /* if no sv we have to assume bytes */
cad2e5aa 410 I32 ml_anch;
6eb5f6b9 411 register char *other_last = Nullch; /* other substr checked before this */
b7953727 412 char *check_at = Nullch; /* check substr found at this pos */
7fba1cd6 413 I32 multiline = prog->reganch & PMf_MULTILINE;
30944b6d
IZ
414#ifdef DEBUGGING
415 char *i_strpos = strpos;
ce333219 416 SV *dsv = PERL_DEBUG_PAD_ZERO(0);
30944b6d 417#endif
a3621e74
YO
418
419 GET_RE_DEBUG_FLAGS_DECL;
420
a30b2f1f 421 RX_MATCH_UTF8_set(prog,do_utf8);
cad2e5aa 422
b8d68ded 423 if (prog->reganch & ROPT_UTF8) {
a3621e74 424 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
b8d68ded
JH
425 "UTF-8 regex...\n"));
426 PL_reg_flags |= RF_utf8;
427 }
428
a3621e74 429 DEBUG_EXECUTE_r({
b8d68ded 430 char *s = PL_reg_match_utf8 ?
c728cb41
JH
431 sv_uni_display(dsv, sv, 60, UNI_DISPLAY_REGEX) :
432 strpos;
b8d68ded
JH
433 int len = PL_reg_match_utf8 ?
434 strlen(s) : strend - strpos;
2a782b5b
JH
435 if (!PL_colorset)
436 reginitcolors();
b8d68ded 437 if (PL_reg_match_utf8)
a3621e74 438 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
b8d68ded 439 "UTF-8 target...\n"));
2a782b5b
JH
440 PerlIO_printf(Perl_debug_log,
441 "%sGuessing start of match, REx%s `%s%.60s%s%s' against `%s%.*s%s%s'...\n",
e4584336 442 PL_colors[4], PL_colors[5], PL_colors[0],
2a782b5b
JH
443 prog->precomp,
444 PL_colors[1],
445 (strlen(prog->precomp) > 60 ? "..." : ""),
446 PL_colors[0],
447 (int)(len > 60 ? 60 : len),
448 s, PL_colors[1],
449 (len > 60 ? "..." : "")
450 );
451 });
cad2e5aa 452
c344f387
JH
453 /* CHR_DIST() would be more correct here but it makes things slow. */
454 if (prog->minlen > strend - strpos) {
a3621e74 455 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
a72c7584 456 "String too short... [re_intuit_start]\n"));
cad2e5aa 457 goto fail;
2c2d71f5 458 }
a1933d95 459 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
1aa99e6b 460 PL_regeol = strend;
33b8afdf
JH
461 if (do_utf8) {
462 if (!prog->check_utf8 && prog->check_substr)
463 to_utf8_substr(prog);
464 check = prog->check_utf8;
465 } else {
466 if (!prog->check_substr && prog->check_utf8)
467 to_byte_substr(prog);
468 check = prog->check_substr;
469 }
470 if (check == &PL_sv_undef) {
a3621e74 471 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf
JH
472 "Non-utf string cannot match utf check string\n"));
473 goto fail;
474 }
2c2d71f5 475 if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */
cad2e5aa
JH
476 ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE)
477 || ( (prog->reganch & ROPT_ANCH_BOL)
7fba1cd6 478 && !multiline ) ); /* Check after \n? */
cad2e5aa 479
7e25d62c
JH
480 if (!ml_anch) {
481 if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */
482 | ROPT_IMPLICIT)) /* not a real BOL */
483 /* SvCUR is not set on references: SvRV and SvPVX overlap */
484 && sv && !SvROK(sv)
485 && (strpos != strbeg)) {
a3621e74 486 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
7e25d62c
JH
487 goto fail;
488 }
489 if (prog->check_offset_min == prog->check_offset_max &&
cce850e4 490 !(prog->reganch & ROPT_CANY_SEEN)) {
2c2d71f5 491 /* Substring at constant offset from beg-of-str... */
cad2e5aa
JH
492 I32 slen;
493
1aa99e6b 494 s = HOP3c(strpos, prog->check_offset_min, strend);
653099ff
GS
495 if (SvTAIL(check)) {
496 slen = SvCUR(check); /* >= 1 */
cad2e5aa 497
9041c2e3 498 if ( strend - s > slen || strend - s < slen - 1
2c2d71f5 499 || (strend - s == slen && strend[-1] != '\n')) {
a3621e74 500 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
2c2d71f5 501 goto fail_finish;
cad2e5aa
JH
502 }
503 /* Now should match s[0..slen-2] */
504 slen--;
653099ff 505 if (slen && (*SvPVX(check) != *s
cad2e5aa 506 || (slen > 1
653099ff 507 && memNE(SvPVX(check), s, slen)))) {
2c2d71f5 508 report_neq:
a3621e74 509 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
2c2d71f5
JH
510 goto fail_finish;
511 }
cad2e5aa 512 }
653099ff
GS
513 else if (*SvPVX(check) != *s
514 || ((slen = SvCUR(check)) > 1
515 && memNE(SvPVX(check), s, slen)))
2c2d71f5
JH
516 goto report_neq;
517 goto success_at_start;
7e25d62c 518 }
cad2e5aa 519 }
2c2d71f5 520 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
cad2e5aa 521 s = strpos;
2c2d71f5 522 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
2c2d71f5 523 end_shift = prog->minlen - start_shift -
653099ff 524 CHR_SVLEN(check) + (SvTAIL(check) != 0);
2c2d71f5 525 if (!ml_anch) {
a3b680e6 526 const I32 end = prog->check_offset_max + CHR_SVLEN(check)
653099ff 527 - (SvTAIL(check) != 0);
a3b680e6 528 const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
2c2d71f5
JH
529
530 if (end_shift < eshift)
531 end_shift = eshift;
532 }
cad2e5aa 533 }
2c2d71f5 534 else { /* Can match at random position */
cad2e5aa
JH
535 ml_anch = 0;
536 s = strpos;
2c2d71f5
JH
537 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
538 /* Should be nonnegative! */
539 end_shift = prog->minlen - start_shift -
653099ff 540 CHR_SVLEN(check) + (SvTAIL(check) != 0);
cad2e5aa
JH
541 }
542
2c2d71f5 543#ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */
0033605d 544 if (end_shift < 0)
6bbae5e6 545 Perl_croak(aTHX_ "panic: end_shift");
2c2d71f5
JH
546#endif
547
2c2d71f5
JH
548 restart:
549 /* Find a possible match in the region s..strend by looking for
550 the "check" substring in the region corrected by start/end_shift. */
cad2e5aa 551 if (flags & REXEC_SCREAM) {
cad2e5aa 552 I32 p = -1; /* Internal iterator of scream. */
a3b680e6 553 I32 * const pp = data ? data->scream_pos : &p;
cad2e5aa 554
2c2d71f5
JH
555 if (PL_screamfirst[BmRARE(check)] >= 0
556 || ( BmRARE(check) == '\n'
557 && (BmPREVIOUS(check) == SvCUR(check) - 1)
558 && SvTAIL(check) ))
9041c2e3 559 s = screaminstr(sv, check,
2c2d71f5 560 start_shift + (s - strbeg), end_shift, pp, 0);
cad2e5aa 561 else
2c2d71f5 562 goto fail_finish;
4addbd3b
HS
563 /* we may be pointing at the wrong string */
564 if (s && RX_MATCH_COPIED(prog))
7ef91622 565 s = strbeg + (s - SvPVX(sv));
cad2e5aa
JH
566 if (data)
567 *data->scream_olds = s;
568 }
f33976b4 569 else if (prog->reganch & ROPT_CANY_SEEN)
3baa4c62
JH
570 s = fbm_instr((U8*)(s + start_shift),
571 (U8*)(strend - end_shift),
7fba1cd6 572 check, multiline ? FBMrf_MULTILINE : 0);
cad2e5aa 573 else
1aa99e6b
IH
574 s = fbm_instr(HOP3(s, start_shift, strend),
575 HOP3(strend, -end_shift, strbeg),
7fba1cd6 576 check, multiline ? FBMrf_MULTILINE : 0);
cad2e5aa
JH
577
578 /* Update the count-of-usability, remove useless subpatterns,
579 unshift s. */
2c2d71f5 580
a3621e74 581 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s %s substr `%s%.*s%s'%s%s",
2c2d71f5 582 (s ? "Found" : "Did not find"),
33b8afdf 583 (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) ? "anchored" : "floating"),
2c2d71f5 584 PL_colors[0],
7b0972df
JH
585 (int)(SvCUR(check) - (SvTAIL(check)!=0)),
586 SvPVX(check),
2c2d71f5
JH
587 PL_colors[1], (SvTAIL(check) ? "$" : ""),
588 (s ? " at offset " : "...\n") ) );
589
590 if (!s)
591 goto fail_finish;
592
6eb5f6b9
JH
593 check_at = s;
594
2c2d71f5 595 /* Finish the diagnostic message */
a3621e74 596 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
2c2d71f5
JH
597
598 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
599 Start with the other substr.
600 XXXX no SCREAM optimization yet - and a very coarse implementation
601 XXXX /ttx+/ results in anchored=`ttx', floating=`x'. floating will
602 *always* match. Probably should be marked during compile...
603 Probably it is right to do no SCREAM here...
604 */
605
33b8afdf 606 if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) : (prog->float_substr && prog->anchored_substr)) {
30944b6d 607 /* Take into account the "other" substring. */
2c2d71f5
JH
608 /* XXXX May be hopelessly wrong for UTF... */
609 if (!other_last)
6eb5f6b9 610 other_last = strpos;
33b8afdf 611 if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) {
30944b6d
IZ
612 do_other_anchored:
613 {
1aa99e6b 614 char *last = HOP3c(s, -start_shift, strbeg), *last1, *last2;
2c2d71f5 615 char *s1 = s;
33b8afdf 616 SV* must;
2c2d71f5 617
2c2d71f5
JH
618 t = s - prog->check_offset_max;
619 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
1d86a7f9 620 && (!do_utf8
1aa99e6b 621 || ((t = reghopmaybe3_c(s, -(prog->check_offset_max), strpos))
2c2d71f5 622 && t > strpos)))
30944b6d 623 /* EMPTY */;
2c2d71f5
JH
624 else
625 t = strpos;
1aa99e6b 626 t = HOP3c(t, prog->anchored_offset, strend);
6eb5f6b9
JH
627 if (t < other_last) /* These positions already checked */
628 t = other_last;
1aa99e6b 629 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
2c2d71f5
JH
630 if (last < last1)
631 last1 = last;
632 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
633 /* On end-of-str: see comment below. */
33b8afdf
JH
634 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
635 if (must == &PL_sv_undef) {
636 s = (char*)NULL;
a3621e74 637 DEBUG_EXECUTE_r(must = prog->anchored_utf8); /* for debug */
33b8afdf
JH
638 }
639 else
640 s = fbm_instr(
641 (unsigned char*)t,
642 HOP3(HOP3(last1, prog->anchored_offset, strend)
643 + SvCUR(must), -(SvTAIL(must)!=0), strbeg),
644 must,
7fba1cd6 645 multiline ? FBMrf_MULTILINE : 0
33b8afdf 646 );
a3621e74 647 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1aa99e6b 648 "%s anchored substr `%s%.*s%s'%s",
2c2d71f5
JH
649 (s ? "Found" : "Contradicts"),
650 PL_colors[0],
33b8afdf
JH
651 (int)(SvCUR(must)
652 - (SvTAIL(must)!=0)),
653 SvPVX(must),
654 PL_colors[1], (SvTAIL(must) ? "$" : "")));
2c2d71f5
JH
655 if (!s) {
656 if (last1 >= last2) {
a3621e74 657 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2c2d71f5
JH
658 ", giving up...\n"));
659 goto fail_finish;
660 }
a3621e74 661 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2c2d71f5 662 ", trying floating at offset %ld...\n",
1aa99e6b
IH
663 (long)(HOP3c(s1, 1, strend) - i_strpos)));
664 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
665 s = HOP3c(last, 1, strend);
2c2d71f5
JH
666 goto restart;
667 }
668 else {
a3621e74 669 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
30944b6d 670 (long)(s - i_strpos)));
1aa99e6b
IH
671 t = HOP3c(s, -prog->anchored_offset, strbeg);
672 other_last = HOP3c(s, 1, strend);
30944b6d 673 s = s1;
2c2d71f5
JH
674 if (t == strpos)
675 goto try_at_start;
2c2d71f5
JH
676 goto try_at_offset;
677 }
30944b6d 678 }
2c2d71f5
JH
679 }
680 else { /* Take into account the floating substring. */
33b8afdf
JH
681 char *last, *last1;
682 char *s1 = s;
683 SV* must;
684
685 t = HOP3c(s, -start_shift, strbeg);
686 last1 = last =
687 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
688 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
689 last = HOP3c(t, prog->float_max_offset, strend);
690 s = HOP3c(t, prog->float_min_offset, strend);
691 if (s < other_last)
692 s = other_last;
2c2d71f5 693 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
33b8afdf
JH
694 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
695 /* fbm_instr() takes into account exact value of end-of-str
696 if the check is SvTAIL(ed). Since false positives are OK,
697 and end-of-str is not later than strend we are OK. */
698 if (must == &PL_sv_undef) {
699 s = (char*)NULL;
a3621e74 700 DEBUG_EXECUTE_r(must = prog->float_utf8); /* for debug message */
33b8afdf
JH
701 }
702 else
2c2d71f5 703 s = fbm_instr((unsigned char*)s,
33b8afdf
JH
704 (unsigned char*)last + SvCUR(must)
705 - (SvTAIL(must)!=0),
7fba1cd6 706 must, multiline ? FBMrf_MULTILINE : 0);
a3621e74 707 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s",
33b8afdf
JH
708 (s ? "Found" : "Contradicts"),
709 PL_colors[0],
710 (int)(SvCUR(must) - (SvTAIL(must)!=0)),
711 SvPVX(must),
712 PL_colors[1], (SvTAIL(must) ? "$" : "")));
713 if (!s) {
714 if (last1 == last) {
a3621e74 715 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf
JH
716 ", giving up...\n"));
717 goto fail_finish;
2c2d71f5 718 }
a3621e74 719 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf
JH
720 ", trying anchored starting at offset %ld...\n",
721 (long)(s1 + 1 - i_strpos)));
722 other_last = last;
723 s = HOP3c(t, 1, strend);
724 goto restart;
725 }
726 else {
a3621e74 727 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
33b8afdf
JH
728 (long)(s - i_strpos)));
729 other_last = s; /* Fix this later. --Hugo */
730 s = s1;
731 if (t == strpos)
732 goto try_at_start;
733 goto try_at_offset;
734 }
2c2d71f5 735 }
cad2e5aa 736 }
2c2d71f5
JH
737
738 t = s - prog->check_offset_max;
2c2d71f5 739 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
1d86a7f9 740 && (!do_utf8
1aa99e6b
IH
741 || ((t = reghopmaybe3_c(s, -prog->check_offset_max, strpos))
742 && t > strpos))) {
2c2d71f5
JH
743 /* Fixed substring is found far enough so that the match
744 cannot start at strpos. */
745 try_at_offset:
cad2e5aa 746 if (ml_anch && t[-1] != '\n') {
30944b6d
IZ
747 /* Eventually fbm_*() should handle this, but often
748 anchored_offset is not 0, so this check will not be wasted. */
749 /* XXXX In the code below we prefer to look for "^" even in
750 presence of anchored substrings. And we search even
751 beyond the found float position. These pessimizations
752 are historical artefacts only. */
753 find_anchor:
2c2d71f5 754 while (t < strend - prog->minlen) {
cad2e5aa 755 if (*t == '\n') {
4ee3650e 756 if (t < check_at - prog->check_offset_min) {
33b8afdf 757 if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) {
4ee3650e
GS
758 /* Since we moved from the found position,
759 we definitely contradict the found anchored
30944b6d
IZ
760 substr. Due to the above check we do not
761 contradict "check" substr.
762 Thus we can arrive here only if check substr
763 is float. Redo checking for "other"=="fixed".
764 */
9041c2e3 765 strpos = t + 1;
a3621e74 766 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
e4584336 767 PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
30944b6d
IZ
768 goto do_other_anchored;
769 }
4ee3650e
GS
770 /* We don't contradict the found floating substring. */
771 /* XXXX Why not check for STCLASS? */
cad2e5aa 772 s = t + 1;
a3621e74 773 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
e4584336 774 PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
cad2e5aa
JH
775 goto set_useful;
776 }
4ee3650e
GS
777 /* Position contradicts check-string */
778 /* XXXX probably better to look for check-string
779 than for "\n", so one should lower the limit for t? */
a3621e74 780 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
e4584336 781 PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
0e41cd87 782 other_last = strpos = s = t + 1;
cad2e5aa
JH
783 goto restart;
784 }
785 t++;
786 }
a3621e74 787 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
e4584336 788 PL_colors[0], PL_colors[1]));
2c2d71f5 789 goto fail_finish;
cad2e5aa 790 }
f5952150 791 else {
a3621e74 792 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
e4584336 793 PL_colors[0], PL_colors[1]));
f5952150 794 }
cad2e5aa
JH
795 s = t;
796 set_useful:
33b8afdf 797 ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
cad2e5aa
JH
798 }
799 else {
f5952150 800 /* The found string does not prohibit matching at strpos,
2c2d71f5 801 - no optimization of calling REx engine can be performed,
f5952150
GS
802 unless it was an MBOL and we are not after MBOL,
803 or a future STCLASS check will fail this. */
2c2d71f5
JH
804 try_at_start:
805 /* Even in this situation we may use MBOL flag if strpos is offset
806 wrt the start of the string. */
05b4157f 807 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
a1933d95 808 && (strpos != strbeg) && strpos[-1] != '\n'
d506a20d
IZ
809 /* May be due to an implicit anchor of m{.*foo} */
810 && !(prog->reganch & ROPT_IMPLICIT))
811 {
cad2e5aa
JH
812 t = strpos;
813 goto find_anchor;
814 }
a3621e74 815 DEBUG_EXECUTE_r( if (ml_anch)
f5952150 816 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
e4584336 817 (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
30944b6d 818 );
2c2d71f5 819 success_at_start:
30944b6d 820 if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */
33b8afdf
JH
821 && (do_utf8 ? (
822 prog->check_utf8 /* Could be deleted already */
823 && --BmUSEFUL(prog->check_utf8) < 0
824 && (prog->check_utf8 == prog->float_utf8)
825 ) : (
826 prog->check_substr /* Could be deleted already */
827 && --BmUSEFUL(prog->check_substr) < 0
828 && (prog->check_substr == prog->float_substr)
829 )))
66e933ab 830 {
cad2e5aa 831 /* If flags & SOMETHING - do not do it many times on the same match */
a3621e74 832 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
33b8afdf
JH
833 SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr);
834 if (do_utf8 ? prog->check_substr : prog->check_utf8)
835 SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8);
836 prog->check_substr = prog->check_utf8 = Nullsv; /* disable */
837 prog->float_substr = prog->float_utf8 = Nullsv; /* clear */
5e39e1e5 838 check = Nullsv; /* abort */
cad2e5aa 839 s = strpos;
3cf5c195
IZ
840 /* XXXX This is a remnant of the old implementation. It
841 looks wasteful, since now INTUIT can use many
6eb5f6b9 842 other heuristics. */
cad2e5aa
JH
843 prog->reganch &= ~RE_USE_INTUIT;
844 }
845 else
846 s = strpos;
847 }
848
6eb5f6b9
JH
849 /* Last resort... */
850 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
851 if (prog->regstclass) {
852 /* minlen == 0 is possible if regstclass is \b or \B,
853 and the fixed substr is ''$.
854 Since minlen is already taken into account, s+1 is before strend;
855 accidentally, minlen >= 1 guaranties no false positives at s + 1
856 even for \b or \B. But (minlen? 1 : 0) below assumes that
857 regstclass does not come from lookahead... */
858 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
859 This leaves EXACTF only, which is dealt with in find_byclass(). */
06b5626a
AL
860 const U8* str = (U8*)STRING(prog->regstclass);
861 const int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT
1aa99e6b 862 ? CHR_DIST(str+STR_LEN(prog->regstclass), str)
66e933ab 863 : 1);
a3b680e6 864 const char * const endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
1aa99e6b 865 ? HOP3c(s, (prog->minlen ? cl_l : 0), strend)
33b8afdf 866 : (prog->float_substr || prog->float_utf8
1aa99e6b
IH
867 ? HOP3c(HOP3c(check_at, -start_shift, strbeg),
868 cl_l, strend)
869 : strend);
6eb5f6b9
JH
870
871 t = s;
9926ca43 872 cache_re(prog);
06b5626a 873 s = find_byclass(prog, prog->regstclass, s, endpos, 1);
6eb5f6b9
JH
874 if (!s) {
875#ifdef DEBUGGING
e1ec3a88 876 const char *what = 0;
6eb5f6b9
JH
877#endif
878 if (endpos == strend) {
a3621e74 879 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6eb5f6b9
JH
880 "Could not match STCLASS...\n") );
881 goto fail;
882 }
a3621e74 883 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab 884 "This position contradicts STCLASS...\n") );
653099ff
GS
885 if ((prog->reganch & ROPT_ANCH) && !ml_anch)
886 goto fail;
6eb5f6b9 887 /* Contradict one of substrings */
33b8afdf
JH
888 if (prog->anchored_substr || prog->anchored_utf8) {
889 if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) {
a3621e74 890 DEBUG_EXECUTE_r( what = "anchored" );
6eb5f6b9 891 hop_and_restart:
1aa99e6b 892 s = HOP3c(t, 1, strend);
66e933ab
GS
893 if (s + start_shift + end_shift > strend) {
894 /* XXXX Should be taken into account earlier? */
a3621e74 895 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab
GS
896 "Could not match STCLASS...\n") );
897 goto fail;
898 }
5e39e1e5
HS
899 if (!check)
900 goto giveup;
a3621e74 901 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 902 "Looking for %s substr starting at offset %ld...\n",
6eb5f6b9
JH
903 what, (long)(s + start_shift - i_strpos)) );
904 goto restart;
905 }
66e933ab 906 /* Have both, check_string is floating */
6eb5f6b9
JH
907 if (t + start_shift >= check_at) /* Contradicts floating=check */
908 goto retry_floating_check;
909 /* Recheck anchored substring, but not floating... */
9041c2e3 910 s = check_at;
5e39e1e5
HS
911 if (!check)
912 goto giveup;
a3621e74 913 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 914 "Looking for anchored substr starting at offset %ld...\n",
6eb5f6b9
JH
915 (long)(other_last - i_strpos)) );
916 goto do_other_anchored;
917 }
60e71179
GS
918 /* Another way we could have checked stclass at the
919 current position only: */
920 if (ml_anch) {
921 s = t = t + 1;
5e39e1e5
HS
922 if (!check)
923 goto giveup;
a3621e74 924 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 925 "Looking for /%s^%s/m starting at offset %ld...\n",
e4584336 926 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
60e71179 927 goto try_at_offset;
66e933ab 928 }
33b8afdf 929 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
60e71179 930 goto fail;
6eb5f6b9
JH
931 /* Check is floating subtring. */
932 retry_floating_check:
933 t = check_at - start_shift;
a3621e74 934 DEBUG_EXECUTE_r( what = "floating" );
6eb5f6b9
JH
935 goto hop_and_restart;
936 }
b7953727 937 if (t != s) {
a3621e74 938 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
6eb5f6b9 939 "By STCLASS: moving %ld --> %ld\n",
b7953727
JH
940 (long)(t - i_strpos), (long)(s - i_strpos))
941 );
942 }
943 else {
a3621e74 944 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
b7953727
JH
945 "Does not contradict STCLASS...\n");
946 );
947 }
6eb5f6b9 948 }
5e39e1e5 949 giveup:
a3621e74 950 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
5e39e1e5
HS
951 PL_colors[4], (check ? "Guessed" : "Giving up"),
952 PL_colors[5], (long)(s - i_strpos)) );
cad2e5aa 953 return s;
2c2d71f5
JH
954
955 fail_finish: /* Substring not found */
33b8afdf
JH
956 if (prog->check_substr || prog->check_utf8) /* could be removed already */
957 BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
cad2e5aa 958 fail:
a3621e74 959 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
e4584336 960 PL_colors[4], PL_colors[5]));
cad2e5aa
JH
961 return Nullch;
962}
9661b544 963
6eb5f6b9 964/* We know what class REx starts with. Try to find this position... */
3c3eec57 965STATIC char *
a3b680e6 966S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, const char *strend, I32 norun)
a687059c 967{
27da23d5 968 dVAR;
6eb5f6b9
JH
969 I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
970 char *m;
d8093b23 971 STRLEN ln;
5dab1207 972 STRLEN lnc;
078c425b 973 register STRLEN uskip;
d8093b23
G
974 unsigned int c1;
975 unsigned int c2;
6eb5f6b9
JH
976 char *e;
977 register I32 tmp = 1; /* Scratch variable? */
a3b680e6 978 register const bool do_utf8 = PL_reg_match_utf8;
cad2e5aa 979
6eb5f6b9
JH
980 /* We know what class it must start with. */
981 switch (OP(c)) {
6eb5f6b9 982 case ANYOF:
388cc4de 983 if (do_utf8) {
078c425b 984 while (s + (uskip = UTF8SKIP(s)) <= strend) {
388cc4de
HS
985 if ((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
986 !UTF8_IS_INVARIANT((U8)s[0]) ?
987 reginclass(c, (U8*)s, 0, do_utf8) :
988 REGINCLASS(c, (U8*)s)) {
989 if (tmp && (norun || regtry(prog, s)))
990 goto got_it;
991 else
992 tmp = doevery;
993 }
994 else
995 tmp = 1;
078c425b 996 s += uskip;
388cc4de
HS
997 }
998 }
999 else {
1000 while (s < strend) {
1001 STRLEN skip = 1;
1002
1003 if (REGINCLASS(c, (U8*)s) ||
1004 (ANYOF_FOLD_SHARP_S(c, s, strend) &&
1005 /* The assignment of 2 is intentional:
1006 * for the folded sharp s, the skip is 2. */
1007 (skip = SHARP_S_SKIP))) {
1008 if (tmp && (norun || regtry(prog, s)))
1009 goto got_it;
1010 else
1011 tmp = doevery;
1012 }
1013 else
1014 tmp = 1;
1015 s += skip;
1016 }
a0d0e21e 1017 }
6eb5f6b9 1018 break;
f33976b4
DB
1019 case CANY:
1020 while (s < strend) {
1021 if (tmp && (norun || regtry(prog, s)))
1022 goto got_it;
1023 else
1024 tmp = doevery;
1025 s++;
1026 }
1027 break;
6eb5f6b9 1028 case EXACTF:
5dab1207
NIS
1029 m = STRING(c);
1030 ln = STR_LEN(c); /* length to match in octets/bytes */
1031 lnc = (I32) ln; /* length to match in characters */
1aa99e6b 1032 if (UTF) {
a2a2844f 1033 STRLEN ulen1, ulen2;
5dab1207 1034 U8 *sm = (U8 *) m;
89ebb4a3
JH
1035 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
1036 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
a2a2844f
JH
1037
1038 to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1039 to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1040
89ebb4a3 1041 c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
872c91ae 1042 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
89ebb4a3 1043 c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
872c91ae 1044 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
5dab1207
NIS
1045 lnc = 0;
1046 while (sm < ((U8 *) m + ln)) {
1047 lnc++;
1048 sm += UTF8SKIP(sm);
1049 }
1aa99e6b
IH
1050 }
1051 else {
1052 c1 = *(U8*)m;
1053 c2 = PL_fold[c1];
1054 }
6eb5f6b9
JH
1055 goto do_exactf;
1056 case EXACTFL:
5dab1207
NIS
1057 m = STRING(c);
1058 ln = STR_LEN(c);
1059 lnc = (I32) ln;
d8093b23 1060 c1 = *(U8*)m;
6eb5f6b9
JH
1061 c2 = PL_fold_locale[c1];
1062 do_exactf:
db12adc6 1063 e = HOP3c(strend, -((I32)lnc), s);
b3c9acc1 1064
6eb5f6b9
JH
1065 if (norun && e < s)
1066 e = s; /* Due to minlen logic of intuit() */
1aa99e6b 1067
60a8b682
JH
1068 /* The idea in the EXACTF* cases is to first find the
1069 * first character of the EXACTF* node and then, if
1070 * necessary, case-insensitively compare the full
1071 * text of the node. The c1 and c2 are the first
1072 * characters (though in Unicode it gets a bit
1073 * more complicated because there are more cases
7f16dd3d
JH
1074 * than just upper and lower: one needs to use
1075 * the so-called folding case for case-insensitive
1076 * matching (called "loose matching" in Unicode).
1077 * ibcmp_utf8() will do just that. */
60a8b682 1078
1aa99e6b 1079 if (do_utf8) {
575cac57 1080 UV c, f;
89ebb4a3
JH
1081 U8 tmpbuf [UTF8_MAXBYTES+1];
1082 U8 foldbuf[UTF8_MAXBYTES_CASE+1];
575cac57 1083 STRLEN len, foldlen;
d7f013c8 1084
09091399 1085 if (c1 == c2) {
5dab1207
NIS
1086 /* Upper and lower of 1st char are equal -
1087 * probably not a "letter". */
1aa99e6b 1088 while (s <= e) {
89ebb4a3 1089 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
872c91ae
JH
1090 ckWARN(WARN_UTF8) ?
1091 0 : UTF8_ALLOW_ANY);
80aecb99
JH
1092 if ( c == c1
1093 && (ln == len ||
66423254 1094 ibcmp_utf8(s, (char **)0, 0, do_utf8,
eb160463 1095 m, (char **)0, ln, (bool)UTF))
55da9344 1096 && (norun || regtry(prog, s)) )
1aa99e6b 1097 goto got_it;
80aecb99
JH
1098 else {
1099 uvchr_to_utf8(tmpbuf, c);
ac7e0132 1100 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen);
80aecb99
JH
1101 if ( f != c
1102 && (f == c1 || f == c2)
1103 && (ln == foldlen ||
66423254
JH
1104 !ibcmp_utf8((char *) foldbuf,
1105 (char **)0, foldlen, do_utf8,
d07ddd77 1106 m,
eb160463 1107 (char **)0, ln, (bool)UTF))
80aecb99
JH
1108 && (norun || regtry(prog, s)) )
1109 goto got_it;
1110 }
1aa99e6b
IH
1111 s += len;
1112 }
09091399
JH
1113 }
1114 else {
1aa99e6b 1115 while (s <= e) {
89ebb4a3 1116 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
872c91ae
JH
1117 ckWARN(WARN_UTF8) ?
1118 0 : UTF8_ALLOW_ANY);
80aecb99 1119
60a8b682 1120 /* Handle some of the three Greek sigmas cases.
8c01da3c
JH
1121 * Note that not all the possible combinations
1122 * are handled here: some of them are handled
1123 * by the standard folding rules, and some of
1124 * them (the character class or ANYOF cases)
1125 * are handled during compiletime in
1126 * regexec.c:S_regclass(). */
880bd946
JH
1127 if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
1128 c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
1129 c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
80aecb99
JH
1130
1131 if ( (c == c1 || c == c2)
1132 && (ln == len ||
66423254 1133 ibcmp_utf8(s, (char **)0, 0, do_utf8,
eb160463 1134 m, (char **)0, ln, (bool)UTF))
55da9344 1135 && (norun || regtry(prog, s)) )
1aa99e6b 1136 goto got_it;
80aecb99
JH
1137 else {
1138 uvchr_to_utf8(tmpbuf, c);
ac7e0132 1139 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen);
80aecb99
JH
1140 if ( f != c
1141 && (f == c1 || f == c2)
1142 && (ln == foldlen ||
a6872d42 1143 !ibcmp_utf8((char *) foldbuf,
66423254 1144 (char **)0, foldlen, do_utf8,
d07ddd77 1145 m,
eb160463 1146 (char **)0, ln, (bool)UTF))
80aecb99
JH
1147 && (norun || regtry(prog, s)) )
1148 goto got_it;
1149 }
1aa99e6b
IH
1150 s += len;
1151 }
09091399 1152 }
1aa99e6b
IH
1153 }
1154 else {
1155 if (c1 == c2)
1156 while (s <= e) {
1157 if ( *(U8*)s == c1
1158 && (ln == 1 || !(OP(c) == EXACTF
1159 ? ibcmp(s, m, ln)
1160 : ibcmp_locale(s, m, ln)))
1161 && (norun || regtry(prog, s)) )
1162 goto got_it;
1163 s++;
1164 }
1165 else
1166 while (s <= e) {
1167 if ( (*(U8*)s == c1 || *(U8*)s == c2)
1168 && (ln == 1 || !(OP(c) == EXACTF
1169 ? ibcmp(s, m, ln)
1170 : ibcmp_locale(s, m, ln)))
1171 && (norun || regtry(prog, s)) )
1172 goto got_it;
1173 s++;
1174 }
b3c9acc1
IZ
1175 }
1176 break;
bbce6d69 1177 case BOUNDL:
3280af22 1178 PL_reg_flags |= RF_tainted;
bbce6d69 1179 /* FALL THROUGH */
a0d0e21e 1180 case BOUND:
ffc61ed2 1181 if (do_utf8) {
12d33761 1182 if (s == PL_bostr)
ffc61ed2
JH
1183 tmp = '\n';
1184 else {
b4f7163a 1185 U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr);
9041c2e3 1186
b4f7163a 1187 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
ffc61ed2
JH
1188 }
1189 tmp = ((OP(c) == BOUND ?
9041c2e3 1190 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
8269fa76 1191 LOAD_UTF8_CHARCLASS(alnum,"a");
078c425b 1192 while (s + (uskip = UTF8SKIP(s)) <= strend) {
ffc61ed2 1193 if (tmp == !(OP(c) == BOUND ?
3568d838 1194 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
ffc61ed2
JH
1195 isALNUM_LC_utf8((U8*)s)))
1196 {
1197 tmp = !tmp;
1198 if ((norun || regtry(prog, s)))
1199 goto got_it;
1200 }
078c425b 1201 s += uskip;
a687059c 1202 }
a0d0e21e 1203 }
667bb95a 1204 else {
12d33761 1205 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2
JH
1206 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1207 while (s < strend) {
1208 if (tmp ==
1209 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
1210 tmp = !tmp;
1211 if ((norun || regtry(prog, s)))
1212 goto got_it;
1213 }
1214 s++;
a0ed51b3 1215 }
a0ed51b3 1216 }
6eb5f6b9 1217 if ((!prog->minlen && tmp) && (norun || regtry(prog, s)))
a0ed51b3
LW
1218 goto got_it;
1219 break;
bbce6d69 1220 case NBOUNDL:
3280af22 1221 PL_reg_flags |= RF_tainted;
bbce6d69 1222 /* FALL THROUGH */
a0d0e21e 1223 case NBOUND:
ffc61ed2 1224 if (do_utf8) {
12d33761 1225 if (s == PL_bostr)
ffc61ed2
JH
1226 tmp = '\n';
1227 else {
b4f7163a 1228 U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr);
9041c2e3 1229
b4f7163a 1230 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
ffc61ed2
JH
1231 }
1232 tmp = ((OP(c) == NBOUND ?
9041c2e3 1233 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
8269fa76 1234 LOAD_UTF8_CHARCLASS(alnum,"a");
078c425b 1235 while (s + (uskip = UTF8SKIP(s)) <= strend) {
ffc61ed2 1236 if (tmp == !(OP(c) == NBOUND ?
3568d838 1237 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
ffc61ed2
JH
1238 isALNUM_LC_utf8((U8*)s)))
1239 tmp = !tmp;
1240 else if ((norun || regtry(prog, s)))
1241 goto got_it;
078c425b 1242 s += uskip;
ffc61ed2 1243 }
a0d0e21e 1244 }
667bb95a 1245 else {
12d33761 1246 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2
JH
1247 tmp = ((OP(c) == NBOUND ?
1248 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1249 while (s < strend) {
1250 if (tmp ==
1251 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1252 tmp = !tmp;
1253 else if ((norun || regtry(prog, s)))
1254 goto got_it;
1255 s++;
1256 }
a0ed51b3 1257 }
6eb5f6b9 1258 if ((!prog->minlen && !tmp) && (norun || regtry(prog, s)))
a0ed51b3
LW
1259 goto got_it;
1260 break;
a0d0e21e 1261 case ALNUM:
ffc61ed2 1262 if (do_utf8) {
8269fa76 1263 LOAD_UTF8_CHARCLASS(alnum,"a");
078c425b 1264 while (s + (uskip = UTF8SKIP(s)) <= strend) {
3568d838 1265 if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
ffc61ed2
JH
1266 if (tmp && (norun || regtry(prog, s)))
1267 goto got_it;
1268 else
1269 tmp = doevery;
1270 }
bbce6d69 1271 else
ffc61ed2 1272 tmp = 1;
078c425b 1273 s += uskip;
bbce6d69 1274 }
bbce6d69 1275 }
ffc61ed2
JH
1276 else {
1277 while (s < strend) {
1278 if (isALNUM(*s)) {
1279 if (tmp && (norun || regtry(prog, s)))
1280 goto got_it;
1281 else
1282 tmp = doevery;
1283 }
a0ed51b3 1284 else
ffc61ed2
JH
1285 tmp = 1;
1286 s++;
a0ed51b3 1287 }
a0ed51b3
LW
1288 }
1289 break;
bbce6d69 1290 case ALNUML:
3280af22 1291 PL_reg_flags |= RF_tainted;
ffc61ed2 1292 if (do_utf8) {
078c425b 1293 while (s + (uskip = UTF8SKIP(s)) <= strend) {
ffc61ed2
JH
1294 if (isALNUM_LC_utf8((U8*)s)) {
1295 if (tmp && (norun || regtry(prog, s)))
1296 goto got_it;
1297 else
1298 tmp = doevery;
1299 }
a687059c 1300 else
ffc61ed2 1301 tmp = 1;
078c425b 1302 s += uskip;
a0d0e21e 1303 }
a0d0e21e 1304 }
ffc61ed2
JH
1305 else {
1306 while (s < strend) {
1307 if (isALNUM_LC(*s)) {
1308 if (tmp && (norun || regtry(prog, s)))
1309 goto got_it;
1310 else
1311 tmp = doevery;
1312 }
a0ed51b3 1313 else
ffc61ed2
JH
1314 tmp = 1;
1315 s++;
a0ed51b3 1316 }
a0ed51b3
LW
1317 }
1318 break;
a0d0e21e 1319 case NALNUM:
ffc61ed2 1320 if (do_utf8) {
8269fa76 1321 LOAD_UTF8_CHARCLASS(alnum,"a");
078c425b 1322 while (s + (uskip = UTF8SKIP(s)) <= strend) {
3568d838 1323 if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
ffc61ed2
JH
1324 if (tmp && (norun || regtry(prog, s)))
1325 goto got_it;
1326 else
1327 tmp = doevery;
1328 }
bbce6d69 1329 else
ffc61ed2 1330 tmp = 1;
078c425b 1331 s += uskip;
bbce6d69 1332 }
bbce6d69 1333 }
ffc61ed2
JH
1334 else {
1335 while (s < strend) {
1336 if (!isALNUM(*s)) {
1337 if (tmp && (norun || regtry(prog, s)))
1338 goto got_it;
1339 else
1340 tmp = doevery;
1341 }
a0ed51b3 1342 else
ffc61ed2
JH
1343 tmp = 1;
1344 s++;
a0ed51b3 1345 }
a0ed51b3
LW
1346 }
1347 break;
bbce6d69 1348 case NALNUML:
3280af22 1349 PL_reg_flags |= RF_tainted;
ffc61ed2 1350 if (do_utf8) {
078c425b 1351 while (s + (uskip = UTF8SKIP(s)) <= strend) {
ffc61ed2
JH
1352 if (!isALNUM_LC_utf8((U8*)s)) {
1353 if (tmp && (norun || regtry(prog, s)))
1354 goto got_it;
1355 else
1356 tmp = doevery;
1357 }
a687059c 1358 else
ffc61ed2 1359 tmp = 1;
078c425b 1360 s += uskip;
a687059c 1361 }
a0d0e21e 1362 }
ffc61ed2
JH
1363 else {
1364 while (s < strend) {
1365 if (!isALNUM_LC(*s)) {
1366 if (tmp && (norun || regtry(prog, s)))
1367 goto got_it;
1368 else
1369 tmp = doevery;
1370 }
a0ed51b3 1371 else
ffc61ed2
JH
1372 tmp = 1;
1373 s++;
a0ed51b3 1374 }
a0ed51b3
LW
1375 }
1376 break;
a0d0e21e 1377 case SPACE:
ffc61ed2 1378 if (do_utf8) {
8269fa76 1379 LOAD_UTF8_CHARCLASS(space," ");
078c425b 1380 while (s + (uskip = UTF8SKIP(s)) <= strend) {
3568d838 1381 if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) {
ffc61ed2
JH
1382 if (tmp && (norun || regtry(prog, s)))
1383 goto got_it;
1384 else
1385 tmp = doevery;
1386 }
a0d0e21e 1387 else
ffc61ed2 1388 tmp = 1;
078c425b 1389 s += uskip;
2304df62 1390 }
a0d0e21e 1391 }
ffc61ed2
JH
1392 else {
1393 while (s < strend) {
1394 if (isSPACE(*s)) {
1395 if (tmp && (norun || regtry(prog, s)))
1396 goto got_it;
1397 else
1398 tmp = doevery;
1399 }
a0ed51b3 1400 else
ffc61ed2
JH
1401 tmp = 1;
1402 s++;
a0ed51b3 1403 }
a0ed51b3
LW
1404 }
1405 break;
bbce6d69 1406 case SPACEL:
3280af22 1407 PL_reg_flags |= RF_tainted;
ffc61ed2 1408 if (do_utf8) {
078c425b 1409 while (s + (uskip = UTF8SKIP(s)) <= strend) {
ffc61ed2
JH
1410 if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
1411 if (tmp && (norun || regtry(prog, s)))
1412 goto got_it;
1413 else
1414 tmp = doevery;
1415 }
bbce6d69 1416 else
ffc61ed2 1417 tmp = 1;
078c425b 1418 s += uskip;
bbce6d69 1419 }
bbce6d69 1420 }
ffc61ed2
JH
1421 else {
1422 while (s < strend) {
1423 if (isSPACE_LC(*s)) {
1424 if (tmp && (norun || regtry(prog, s)))
1425 goto got_it;
1426 else
1427 tmp = doevery;
1428 }
a0ed51b3 1429 else
ffc61ed2
JH
1430 tmp = 1;
1431 s++;
a0ed51b3 1432 }
a0ed51b3
LW
1433 }
1434 break;
a0d0e21e 1435 case NSPACE:
ffc61ed2 1436 if (do_utf8) {
8269fa76 1437 LOAD_UTF8_CHARCLASS(space," ");
078c425b 1438 while (s + (uskip = UTF8SKIP(s)) <= strend) {
3568d838 1439 if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) {
ffc61ed2
JH
1440 if (tmp && (norun || regtry(prog, s)))
1441 goto got_it;
1442 else
1443 tmp = doevery;
1444 }
a0d0e21e 1445 else
ffc61ed2 1446 tmp = 1;
078c425b 1447 s += uskip;
a687059c 1448 }
a0d0e21e 1449 }
ffc61ed2
JH
1450 else {
1451 while (s < strend) {
1452 if (!isSPACE(*s)) {
1453 if (tmp && (norun || regtry(prog, s)))
1454 goto got_it;
1455 else
1456 tmp = doevery;
1457 }
a0ed51b3 1458 else
ffc61ed2
JH
1459 tmp = 1;
1460 s++;
a0ed51b3 1461 }
a0ed51b3
LW
1462 }
1463 break;
bbce6d69 1464 case NSPACEL:
3280af22 1465 PL_reg_flags |= RF_tainted;
ffc61ed2 1466 if (do_utf8) {
078c425b 1467 while (s + (uskip = UTF8SKIP(s)) <= strend) {
ffc61ed2
JH
1468 if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
1469 if (tmp && (norun || regtry(prog, s)))
1470 goto got_it;
1471 else
1472 tmp = doevery;
1473 }
bbce6d69 1474 else
ffc61ed2 1475 tmp = 1;
078c425b 1476 s += uskip;
bbce6d69 1477 }
bbce6d69 1478 }
ffc61ed2
JH
1479 else {
1480 while (s < strend) {
1481 if (!isSPACE_LC(*s)) {
1482 if (tmp && (norun || regtry(prog, s)))
1483 goto got_it;
1484 else
1485 tmp = doevery;
1486 }
a0ed51b3 1487 else
ffc61ed2
JH
1488 tmp = 1;
1489 s++;
a0ed51b3 1490 }
a0ed51b3
LW
1491 }
1492 break;
a0d0e21e 1493 case DIGIT:
ffc61ed2 1494 if (do_utf8) {
8269fa76 1495 LOAD_UTF8_CHARCLASS(digit,"0");
078c425b 1496 while (s + (uskip = UTF8SKIP(s)) <= strend) {
3568d838 1497 if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
ffc61ed2
JH
1498 if (tmp && (norun || regtry(prog, s)))
1499 goto got_it;
1500 else
1501 tmp = doevery;
1502 }
a0d0e21e 1503 else
ffc61ed2 1504 tmp = 1;
078c425b 1505 s += uskip;
2b69d0c2 1506 }
a0d0e21e 1507 }
ffc61ed2
JH
1508 else {
1509 while (s < strend) {
1510 if (isDIGIT(*s)) {
1511 if (tmp && (norun || regtry(prog, s)))
1512 goto got_it;
1513 else
1514 tmp = doevery;
1515 }
a0ed51b3 1516 else
ffc61ed2
JH
1517 tmp = 1;
1518 s++;
a0ed51b3 1519 }
a0ed51b3
LW
1520 }
1521 break;
b8c5462f
JH
1522 case DIGITL:
1523 PL_reg_flags |= RF_tainted;
ffc61ed2 1524 if (do_utf8) {
078c425b 1525 while (s + (uskip = UTF8SKIP(s)) <= strend) {
ffc61ed2
JH
1526 if (isDIGIT_LC_utf8((U8*)s)) {
1527 if (tmp && (norun || regtry(prog, s)))
1528 goto got_it;
1529 else
1530 tmp = doevery;
1531 }
b8c5462f 1532 else
ffc61ed2 1533 tmp = 1;
078c425b 1534 s += uskip;
b8c5462f 1535 }
b8c5462f 1536 }
ffc61ed2
JH
1537 else {
1538 while (s < strend) {
1539 if (isDIGIT_LC(*s)) {
1540 if (tmp && (norun || regtry(prog, s)))
1541 goto got_it;
1542 else
1543 tmp = doevery;
1544 }
b8c5462f 1545 else
ffc61ed2
JH
1546 tmp = 1;
1547 s++;
b8c5462f 1548 }
b8c5462f
JH
1549 }
1550 break;
a0d0e21e 1551 case NDIGIT:
ffc61ed2 1552 if (do_utf8) {
8269fa76 1553 LOAD_UTF8_CHARCLASS(digit,"0");
078c425b 1554 while (s + (uskip = UTF8SKIP(s)) <= strend) {
3568d838 1555 if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
ffc61ed2
JH
1556 if (tmp && (norun || regtry(prog, s)))
1557 goto got_it;
1558 else
1559 tmp = doevery;
1560 }
a0d0e21e 1561 else
ffc61ed2 1562 tmp = 1;
078c425b 1563 s += uskip;
a687059c 1564 }
a0d0e21e 1565 }
ffc61ed2
JH
1566 else {
1567 while (s < strend) {
1568 if (!isDIGIT(*s)) {
1569 if (tmp && (norun || regtry(prog, s)))
1570 goto got_it;
1571 else
1572 tmp = doevery;
1573 }
a0ed51b3 1574 else
ffc61ed2
JH
1575 tmp = 1;
1576 s++;
a0ed51b3 1577 }
a0ed51b3
LW
1578 }
1579 break;
b8c5462f
JH
1580 case NDIGITL:
1581 PL_reg_flags |= RF_tainted;
ffc61ed2 1582 if (do_utf8) {
078c425b 1583 while (s + (uskip = UTF8SKIP(s)) <= strend) {
ffc61ed2
JH
1584 if (!isDIGIT_LC_utf8((U8*)s)) {
1585 if (tmp && (norun || regtry(prog, s)))
1586 goto got_it;
1587 else
1588 tmp = doevery;
1589 }
b8c5462f 1590 else
ffc61ed2 1591 tmp = 1;
078c425b 1592 s += uskip;
b8c5462f 1593 }
a0ed51b3 1594 }
ffc61ed2
JH
1595 else {
1596 while (s < strend) {
1597 if (!isDIGIT_LC(*s)) {
1598 if (tmp && (norun || regtry(prog, s)))
1599 goto got_it;
1600 else
1601 tmp = doevery;
1602 }
cf93c79d 1603 else
ffc61ed2
JH
1604 tmp = 1;
1605 s++;
b8c5462f 1606 }
b8c5462f
JH
1607 }
1608 break;
b3c9acc1 1609 default:
3c3eec57
GS
1610 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
1611 break;
d6a28714 1612 }
6eb5f6b9
JH
1613 return 0;
1614 got_it:
1615 return s;
1616}
1617
1618/*
1619 - regexec_flags - match a regexp against a string
1620 */
1621I32
1622Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend,
1623 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
1624/* strend: pointer to null at end of string */
1625/* strbeg: real beginning of string */
1626/* minend: end of match must be >=minend after stringarg. */
1627/* data: May be used for some additional optimizations. */
1628/* nosave: For optimizations. */
1629{
6eb5f6b9
JH
1630 register char *s;
1631 register regnode *c;
1632 register char *startpos = stringarg;
6eb5f6b9
JH
1633 I32 minlen; /* must match at least this many chars */
1634 I32 dontbother = 0; /* how many characters not to try at end */
155aba94 1635 /* I32 start_shift = 0; */ /* Offset of the start to find
6eb5f6b9
JH
1636 constant substr. */ /* CC */
1637 I32 end_shift = 0; /* Same for the end. */ /* CC */
1638 I32 scream_pos = -1; /* Internal iterator of scream. */
1639 char *scream_olds;
1640 SV* oreplsv = GvSV(PL_replgv);
1aa99e6b 1641 bool do_utf8 = DO_UTF8(sv);
a3b680e6 1642 const I32 multiline = prog->reganch & PMf_MULTILINE;
2a782b5b 1643#ifdef DEBUGGING
9e55ce06
JH
1644 SV *dsv0 = PERL_DEBUG_PAD_ZERO(0);
1645 SV *dsv1 = PERL_DEBUG_PAD_ZERO(1);
2a782b5b 1646#endif
a3621e74
YO
1647
1648 GET_RE_DEBUG_FLAGS_DECL;
1649
a30b2f1f 1650 RX_MATCH_UTF8_set(prog,do_utf8);
6eb5f6b9
JH
1651
1652 PL_regcc = 0;
1653
1654 cache_re(prog);
1655#ifdef DEBUGGING
aea4f609 1656 PL_regnarrate = DEBUG_r_TEST;
6eb5f6b9
JH
1657#endif
1658
1659 /* Be paranoid... */
1660 if (prog == NULL || startpos == NULL) {
1661 Perl_croak(aTHX_ "NULL regexp parameter");
1662 return 0;
1663 }
1664
1665 minlen = prog->minlen;
61a36c01 1666 if (strend - startpos < minlen) {
a3621e74 1667 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
a72c7584
JH
1668 "String too short [regexec_flags]...\n"));
1669 goto phooey;
1aa99e6b 1670 }
6eb5f6b9 1671
6eb5f6b9
JH
1672 /* Check validity of program. */
1673 if (UCHARAT(prog->program) != REG_MAGIC) {
1674 Perl_croak(aTHX_ "corrupted regexp program");
1675 }
1676
1677 PL_reg_flags = 0;
1678 PL_reg_eval_set = 0;
1679 PL_reg_maxiter = 0;
1680
1681 if (prog->reganch & ROPT_UTF8)
1682 PL_reg_flags |= RF_utf8;
1683
1684 /* Mark beginning of line for ^ and lookbehind. */
1685 PL_regbol = startpos;
1686 PL_bostr = strbeg;
1687 PL_reg_sv = sv;
1688
1689 /* Mark end of line for $ (and such) */
1690 PL_regeol = strend;
1691
1692 /* see how far we have to get to not match where we matched before */
1693 PL_regtill = startpos+minend;
1694
1695 /* We start without call_cc context. */
1696 PL_reg_call_cc = 0;
1697
1698 /* If there is a "must appear" string, look for it. */
1699 s = startpos;
1700
1701 if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to have PL_reg_ganch */
1702 MAGIC *mg;
1703
1704 if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
1705 PL_reg_ganch = startpos;
1706 else if (sv && SvTYPE(sv) >= SVt_PVMG
1707 && SvMAGIC(sv)
14befaf4
DM
1708 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
1709 && mg->mg_len >= 0) {
6eb5f6b9
JH
1710 PL_reg_ganch = strbeg + mg->mg_len; /* Defined pos() */
1711 if (prog->reganch & ROPT_ANCH_GPOS) {
f33976b4 1712 if (s > PL_reg_ganch)
6eb5f6b9
JH
1713 goto phooey;
1714 s = PL_reg_ganch;
1715 }
1716 }
1717 else /* pos() not defined */
1718 PL_reg_ganch = strbeg;
1719 }
1720
33b8afdf 1721 if (!(flags & REXEC_CHECKED) && (prog->check_substr != Nullsv || prog->check_utf8 != Nullsv)) {
6eb5f6b9
JH
1722 re_scream_pos_data d;
1723
1724 d.scream_olds = &scream_olds;
1725 d.scream_pos = &scream_pos;
1726 s = re_intuit_start(prog, sv, s, strend, flags, &d);
3fa9c3d7 1727 if (!s) {
a3621e74 1728 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
6eb5f6b9 1729 goto phooey; /* not present */
3fa9c3d7 1730 }
6eb5f6b9
JH
1731 }
1732
a3621e74 1733 DEBUG_EXECUTE_r({
9e55ce06
JH
1734 char *s0 = UTF ?
1735 pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60,
c728cb41 1736 UNI_DISPLAY_REGEX) :
9e55ce06
JH
1737 prog->precomp;
1738 int len0 = UTF ? SvCUR(dsv0) : prog->prelen;
1739 char *s1 = do_utf8 ? sv_uni_display(dsv1, sv, 60,
c728cb41 1740 UNI_DISPLAY_REGEX) : startpos;
9e55ce06 1741 int len1 = do_utf8 ? SvCUR(dsv1) : strend - startpos;
2a782b5b
JH
1742 if (!PL_colorset)
1743 reginitcolors();
1744 PerlIO_printf(Perl_debug_log,
9e55ce06 1745 "%sMatching REx%s `%s%*.*s%s%s' against `%s%.*s%s%s'\n",
e4584336 1746 PL_colors[4], PL_colors[5], PL_colors[0],
9e55ce06 1747 len0, len0, s0,
2a782b5b 1748 PL_colors[1],
9e55ce06 1749 len0 > 60 ? "..." : "",
2a782b5b 1750 PL_colors[0],
9e55ce06
JH
1751 (int)(len1 > 60 ? 60 : len1),
1752 s1, PL_colors[1],
1753 (len1 > 60 ? "..." : "")
2a782b5b
JH
1754 );
1755 });
6eb5f6b9
JH
1756
1757 /* Simplest case: anchored match need be tried only once. */
1758 /* [unless only anchor is BOL and multiline is set] */
1759 if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
1760 if (s == startpos && regtry(prog, startpos))
1761 goto got_it;
7fba1cd6 1762 else if (multiline || (prog->reganch & ROPT_IMPLICIT)
6eb5f6b9
JH
1763 || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
1764 {
1765 char *end;
1766
1767 if (minlen)
1768 dontbother = minlen - 1;
1aa99e6b 1769 end = HOP3c(strend, -dontbother, strbeg) - 1;
6eb5f6b9 1770 /* for multiline we only have to try after newlines */
33b8afdf 1771 if (prog->check_substr || prog->check_utf8) {
6eb5f6b9
JH
1772 if (s == startpos)
1773 goto after_try;
1774 while (1) {
1775 if (regtry(prog, s))
1776 goto got_it;
1777 after_try:
1778 if (s >= end)
1779 goto phooey;
1780 if (prog->reganch & RE_USE_INTUIT) {
1781 s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
1782 if (!s)
1783 goto phooey;
1784 }
1785 else
1786 s++;
1787 }
1788 } else {
1789 if (s > startpos)
1790 s--;
1791 while (s < end) {
1792 if (*s++ == '\n') { /* don't need PL_utf8skip here */
1793 if (regtry(prog, s))
1794 goto got_it;
1795 }
1796 }
1797 }
1798 }
1799 goto phooey;
1800 } else if (prog->reganch & ROPT_ANCH_GPOS) {
1801 if (regtry(prog, PL_reg_ganch))
1802 goto got_it;
1803 goto phooey;
1804 }
1805
1806 /* Messy cases: unanchored match. */
33b8afdf 1807 if ((prog->anchored_substr || prog->anchored_utf8) && prog->reganch & ROPT_SKIP) {
6eb5f6b9
JH
1808 /* we have /x+whatever/ */
1809 /* it must be a one character string (XXXX Except UTF?) */
33b8afdf 1810 char ch;
bf93d4cc
GS
1811#ifdef DEBUGGING
1812 int did_match = 0;
1813#endif
33b8afdf
JH
1814 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1815 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1816 ch = SvPVX(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
bf93d4cc 1817
1aa99e6b 1818 if (do_utf8) {
6eb5f6b9
JH
1819 while (s < strend) {
1820 if (*s == ch) {
a3621e74 1821 DEBUG_EXECUTE_r( did_match = 1 );
6eb5f6b9
JH
1822 if (regtry(prog, s)) goto got_it;
1823 s += UTF8SKIP(s);
1824 while (s < strend && *s == ch)
1825 s += UTF8SKIP(s);
1826 }
1827 s += UTF8SKIP(s);
1828 }
1829 }
1830 else {
1831 while (s < strend) {
1832 if (*s == ch) {
a3621e74 1833 DEBUG_EXECUTE_r( did_match = 1 );
6eb5f6b9
JH
1834 if (regtry(prog, s)) goto got_it;
1835 s++;
1836 while (s < strend && *s == ch)
1837 s++;
1838 }
1839 s++;
1840 }
1841 }
a3621e74 1842 DEBUG_EXECUTE_r(if (!did_match)
bf93d4cc 1843 PerlIO_printf(Perl_debug_log,
b7953727
JH
1844 "Did not find anchored character...\n")
1845 );
6eb5f6b9
JH
1846 }
1847 /*SUPPRESS 560*/
33b8afdf
JH
1848 else if (prog->anchored_substr != Nullsv
1849 || prog->anchored_utf8 != Nullsv
1850 || ((prog->float_substr != Nullsv || prog->float_utf8 != Nullsv)
1851 && prog->float_max_offset < strend - s)) {
1852 SV *must;
1853 I32 back_max;
1854 I32 back_min;
1855 char *last;
6eb5f6b9 1856 char *last1; /* Last position checked before */
bf93d4cc
GS
1857#ifdef DEBUGGING
1858 int did_match = 0;
1859#endif
33b8afdf
JH
1860 if (prog->anchored_substr || prog->anchored_utf8) {
1861 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1862 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1863 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
1864 back_max = back_min = prog->anchored_offset;
1865 } else {
1866 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1867 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1868 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
1869 back_max = prog->float_max_offset;
1870 back_min = prog->float_min_offset;
1871 }
1872 if (must == &PL_sv_undef)
1873 /* could not downgrade utf8 check substring, so must fail */
1874 goto phooey;
1875
1876 last = HOP3c(strend, /* Cannot start after this */
1877 -(I32)(CHR_SVLEN(must)
1878 - (SvTAIL(must) != 0) + back_min), strbeg);
6eb5f6b9
JH
1879
1880 if (s > PL_bostr)
1881 last1 = HOPc(s, -1);
1882 else
1883 last1 = s - 1; /* bogus */
1884
1885 /* XXXX check_substr already used to find `s', can optimize if
1886 check_substr==must. */
1887 scream_pos = -1;
1888 dontbother = end_shift;
1889 strend = HOPc(strend, -dontbother);
1890 while ( (s <= last) &&
9041c2e3 1891 ((flags & REXEC_SCREAM)
1aa99e6b 1892 ? (s = screaminstr(sv, must, HOP3c(s, back_min, strend) - strbeg,
6eb5f6b9 1893 end_shift, &scream_pos, 0))
1aa99e6b 1894 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend),
9041c2e3 1895 (unsigned char*)strend, must,
7fba1cd6 1896 multiline ? FBMrf_MULTILINE : 0))) ) {
4addbd3b
HS
1897 /* we may be pointing at the wrong string */
1898 if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog))
7ef91622 1899 s = strbeg + (s - SvPVX(sv));
a3621e74 1900 DEBUG_EXECUTE_r( did_match = 1 );
6eb5f6b9
JH
1901 if (HOPc(s, -back_max) > last1) {
1902 last1 = HOPc(s, -back_min);
1903 s = HOPc(s, -back_max);
1904 }
1905 else {
1906 char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
1907
1908 last1 = HOPc(s, -back_min);
1909 s = t;
1910 }
1aa99e6b 1911 if (do_utf8) {
6eb5f6b9
JH
1912 while (s <= last1) {
1913 if (regtry(prog, s))
1914 goto got_it;
1915 s += UTF8SKIP(s);
1916 }
1917 }
1918 else {
1919 while (s <= last1) {
1920 if (regtry(prog, s))
1921 goto got_it;
1922 s++;
1923 }
1924 }
1925 }
a3621e74 1926 DEBUG_EXECUTE_r(if (!did_match)
b7953727
JH
1927 PerlIO_printf(Perl_debug_log,
1928 "Did not find %s substr `%s%.*s%s'%s...\n",
33b8afdf 1929 ((must == prog->anchored_substr || must == prog->anchored_utf8)
bf93d4cc
GS
1930 ? "anchored" : "floating"),
1931 PL_colors[0],
1932 (int)(SvCUR(must) - (SvTAIL(must)!=0)),
1933 SvPVX(must),
b7953727
JH
1934 PL_colors[1], (SvTAIL(must) ? "$" : ""))
1935 );
6eb5f6b9
JH
1936 goto phooey;
1937 }
155aba94 1938 else if ((c = prog->regstclass)) {
f14c76ed
RGS
1939 if (minlen) {
1940 I32 op = (U8)OP(prog->regstclass);
66e933ab 1941 /* don't bother with what can't match */
f14c76ed
RGS
1942 if (PL_regkind[op] != EXACT && op != CANY)
1943 strend = HOPc(strend, -(minlen - 1));
1944 }
a3621e74 1945 DEBUG_EXECUTE_r({
ffc61ed2 1946 SV *prop = sv_newmortal();
9e55ce06
JH
1947 char *s0;
1948 char *s1;
1949 int len0;
1950 int len1;
1951
ffc61ed2 1952 regprop(prop, c);
9e55ce06
JH
1953 s0 = UTF ?
1954 pv_uni_display(dsv0, (U8*)SvPVX(prop), SvCUR(prop), 60,
c728cb41 1955 UNI_DISPLAY_REGEX) :
9e55ce06
JH
1956 SvPVX(prop);
1957 len0 = UTF ? SvCUR(dsv0) : SvCUR(prop);
1958 s1 = UTF ?
c728cb41 1959 sv_uni_display(dsv1, sv, 60, UNI_DISPLAY_REGEX) : s;
9e55ce06
JH
1960 len1 = UTF ? SvCUR(dsv1) : strend - s;
1961 PerlIO_printf(Perl_debug_log,
1962 "Matching stclass `%*.*s' against `%*.*s'\n",
1963 len0, len0, s0,
1964 len1, len1, s1);
ffc61ed2 1965 });
06b5626a 1966 if (find_byclass(prog, c, s, strend, 0))
6eb5f6b9 1967 goto got_it;
a3621e74 1968 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n"));
d6a28714
JH
1969 }
1970 else {
1971 dontbother = 0;
33b8afdf
JH
1972 if (prog->float_substr != Nullsv || prog->float_utf8 != Nullsv) {
1973 /* Trim the end. */
d6a28714 1974 char *last;
33b8afdf
JH
1975 SV* float_real;
1976
1977 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1978 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1979 float_real = do_utf8 ? prog->float_utf8 : prog->float_substr;
d6a28714
JH
1980
1981 if (flags & REXEC_SCREAM) {
33b8afdf 1982 last = screaminstr(sv, float_real, s - strbeg,
d6a28714
JH
1983 end_shift, &scream_pos, 1); /* last one */
1984 if (!last)
ffc61ed2 1985 last = scream_olds; /* Only one occurrence. */
4addbd3b
HS
1986 /* we may be pointing at the wrong string */
1987 else if (RX_MATCH_COPIED(prog))
7ef91622 1988 s = strbeg + (s - SvPVX(sv));
b8c5462f 1989 }
d6a28714
JH
1990 else {
1991 STRLEN len;
06b5626a 1992 const char * const little = SvPV(float_real, len);
d6a28714 1993
33b8afdf 1994 if (SvTAIL(float_real)) {
d6a28714
JH
1995 if (memEQ(strend - len + 1, little, len - 1))
1996 last = strend - len + 1;
7fba1cd6 1997 else if (!multiline)
9041c2e3 1998 last = memEQ(strend - len, little, len)
d6a28714 1999 ? strend - len : Nullch;
b8c5462f 2000 else
d6a28714
JH
2001 goto find_last;
2002 } else {
2003 find_last:
9041c2e3 2004 if (len)
d6a28714 2005 last = rninstr(s, strend, little, little + len);
b8c5462f 2006 else
d6a28714 2007 last = strend; /* matching `$' */
b8c5462f 2008 }
b8c5462f 2009 }
bf93d4cc 2010 if (last == NULL) {
a3621e74 2011 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
bf93d4cc 2012 "%sCan't trim the tail, match fails (should not happen)%s\n",
e4584336 2013 PL_colors[4], PL_colors[5]));
bf93d4cc
GS
2014 goto phooey; /* Should not happen! */
2015 }
d6a28714
JH
2016 dontbother = strend - last + prog->float_min_offset;
2017 }
2018 if (minlen && (dontbother < minlen))
2019 dontbother = minlen - 1;
2020 strend -= dontbother; /* this one's always in bytes! */
2021 /* We don't know much -- general case. */
1aa99e6b 2022 if (do_utf8) {
d6a28714
JH
2023 for (;;) {
2024 if (regtry(prog, s))
2025 goto got_it;
2026 if (s >= strend)
2027 break;
b8c5462f 2028 s += UTF8SKIP(s);
d6a28714
JH
2029 };
2030 }
2031 else {
2032 do {
2033 if (regtry(prog, s))
2034 goto got_it;
2035 } while (s++ < strend);
2036 }
2037 }
2038
2039 /* Failure. */
2040 goto phooey;
2041
2042got_it:
2043 RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
2044
2045 if (PL_reg_eval_set) {
2046 /* Preserve the current value of $^R */
2047 if (oreplsv != GvSV(PL_replgv))
2048 sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
2049 restored, the value remains
2050 the same. */
acfe0abc 2051 restore_pos(aTHX_ 0);
d6a28714
JH
2052 }
2053
2054 /* make sure $`, $&, $', and $digit will work later */
2055 if ( !(flags & REXEC_NOT_FIRST) ) {
ed252734 2056 RX_MATCH_COPY_FREE(prog);
d6a28714
JH
2057 if (flags & REXEC_COPY_STR) {
2058 I32 i = PL_regeol - startpos + (stringarg - strbeg);
ed252734
NC
2059#ifdef PERL_COPY_ON_WRITE
2060 if ((SvIsCOW(sv)
2061 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2062 if (DEBUG_C_TEST) {
2063 PerlIO_printf(Perl_debug_log,
2064 "Copy on write: regexp capture, type %d\n",
2065 (int) SvTYPE(sv));
2066 }
2067 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
2068 prog->subbeg = SvPVX(prog->saved_copy);
2069 assert (SvPOKp(prog->saved_copy));
2070 } else
2071#endif
2072 {
2073 RX_MATCH_COPIED_on(prog);
2074 s = savepvn(strbeg, i);
2075 prog->subbeg = s;
2076 }
d6a28714 2077 prog->sublen = i;
d6a28714
JH
2078 }
2079 else {
2080 prog->subbeg = strbeg;
2081 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2082 }
2083 }
9041c2e3 2084
d6a28714
JH
2085 return 1;
2086
2087phooey:
a3621e74 2088 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
e4584336 2089 PL_colors[4], PL_colors[5]));
d6a28714 2090 if (PL_reg_eval_set)
acfe0abc 2091 restore_pos(aTHX_ 0);
d6a28714
JH
2092 return 0;
2093}
2094
2095/*
2096 - regtry - try match at specific point
2097 */
2098STATIC I32 /* 0 failure, 1 success */
2099S_regtry(pTHX_ regexp *prog, char *startpos)
2100{
d6a28714
JH
2101 register I32 i;
2102 register I32 *sp;
2103 register I32 *ep;
2104 CHECKPOINT lastcp;
a3621e74 2105 GET_RE_DEBUG_FLAGS_DECL;
d6a28714 2106
02db2b7b
IZ
2107#ifdef DEBUGGING
2108 PL_regindent = 0; /* XXXX Not good when matches are reenterable... */
2109#endif
d6a28714
JH
2110 if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
2111 MAGIC *mg;
2112
2113 PL_reg_eval_set = RS_init;
a3621e74 2114 DEBUG_EXECUTE_r(DEBUG_s(
b900a521
JH
2115 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
2116 (IV)(PL_stack_sp - PL_stack_base));
d6a28714 2117 ));
e8347627 2118 SAVEI32(cxstack[cxstack_ix].blk_oldsp);
d6a28714
JH
2119 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
2120 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
2121 SAVETMPS;
2122 /* Apparently this is not needed, judging by wantarray. */
e8347627 2123 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
d6a28714
JH
2124 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
2125
2126 if (PL_reg_sv) {
2127 /* Make $_ available to executed code. */
2128 if (PL_reg_sv != DEFSV) {
59f00321 2129 SAVE_DEFSV;
d6a28714 2130 DEFSV = PL_reg_sv;
b8c5462f 2131 }
d6a28714 2132
9041c2e3 2133 if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
14befaf4 2134 && (mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global)))) {
d6a28714 2135 /* prepare for quick setting of pos */
14befaf4
DM
2136 sv_magic(PL_reg_sv, (SV*)0,
2137 PERL_MAGIC_regex_global, Nullch, 0);
2138 mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global);
d6a28714 2139 mg->mg_len = -1;
b8c5462f 2140 }
d6a28714
JH
2141 PL_reg_magic = mg;
2142 PL_reg_oldpos = mg->mg_len;
c76ac1ee 2143 SAVEDESTRUCTOR_X(restore_pos, 0);
d6a28714 2144 }
09687e5a 2145 if (!PL_reg_curpm) {
e4584336 2146 Newz(22, PL_reg_curpm, 1, PMOP);
09687e5a
AB
2147#ifdef USE_ITHREADS
2148 {
2149 SV* repointer = newSViv(0);
577e12cc 2150 /* so we know which PL_regex_padav element is PL_reg_curpm */
35061a7e 2151 SvFLAGS(repointer) |= SVf_BREAK;
09687e5a
AB
2152 av_push(PL_regex_padav,repointer);
2153 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2154 PL_regex_pad = AvARRAY(PL_regex_padav);
2155 }
2156#endif
2157 }
aaa362c4 2158 PM_SETRE(PL_reg_curpm, prog);
d6a28714
JH
2159 PL_reg_oldcurpm = PL_curpm;
2160 PL_curpm = PL_reg_curpm;
2161 if (RX_MATCH_COPIED(prog)) {
2162 /* Here is a serious problem: we cannot rewrite subbeg,
2163 since it may be needed if this match fails. Thus
2164 $` inside (?{}) could fail... */
2165 PL_reg_oldsaved = prog->subbeg;
2166 PL_reg_oldsavedlen = prog->sublen;
ed252734
NC
2167#ifdef PERL_COPY_ON_WRITE
2168 PL_nrs = prog->saved_copy;
2169#endif
d6a28714
JH
2170 RX_MATCH_COPIED_off(prog);
2171 }
2172 else
2173 PL_reg_oldsaved = Nullch;
2174 prog->subbeg = PL_bostr;
2175 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2176 }
2177 prog->startp[0] = startpos - PL_bostr;
2178 PL_reginput = startpos;
2179 PL_regstartp = prog->startp;
2180 PL_regendp = prog->endp;
2181 PL_reglastparen = &prog->lastparen;
a01268b5 2182 PL_reglastcloseparen = &prog->lastcloseparen;
d6a28714 2183 prog->lastparen = 0;
03994de8 2184 prog->lastcloseparen = 0;
d6a28714 2185 PL_regsize = 0;
a3621e74 2186 DEBUG_EXECUTE_r(PL_reg_starttry = startpos);
d6a28714
JH
2187 if (PL_reg_start_tmpl <= prog->nparens) {
2188 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
2189 if(PL_reg_start_tmp)
2190 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2191 else
e4584336 2192 New(22, PL_reg_start_tmp, PL_reg_start_tmpl, char*);
d6a28714
JH
2193 }
2194
2195 /* XXXX What this code is doing here?!!! There should be no need
2196 to do this again and again, PL_reglastparen should take care of
3dd2943c 2197 this! --ilya*/
dafc8851
JH
2198
2199 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2200 * Actually, the code in regcppop() (which Ilya may be meaning by
daf18116
JH
2201 * PL_reglastparen), is not needed at all by the test suite
2202 * (op/regexp, op/pat, op/split), but that code is needed, oddly
2203 * enough, for building DynaLoader, or otherwise this
2204 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
2205 * will happen. Meanwhile, this code *is* needed for the
2206 * above-mentioned test suite tests to succeed. The common theme
2207 * on those tests seems to be returning null fields from matches.
2208 * --jhi */
dafc8851 2209#if 1
d6a28714
JH
2210 sp = prog->startp;
2211 ep = prog->endp;
2212 if (prog->nparens) {
eb160463 2213 for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
d6a28714
JH
2214 *++sp = -1;
2215 *++ep = -1;
2216 }
2217 }
dafc8851 2218#endif
02db2b7b 2219 REGCP_SET(lastcp);
d6a28714
JH
2220 if (regmatch(prog->program + 1)) {
2221 prog->endp[0] = PL_reginput - PL_bostr;
2222 return 1;
2223 }
02db2b7b 2224 REGCP_UNWIND(lastcp);
d6a28714
JH
2225 return 0;
2226}
2227
02db2b7b
IZ
2228#define RE_UNWIND_BRANCH 1
2229#define RE_UNWIND_BRANCHJ 2
2230
2231union re_unwind_t;
2232
2233typedef struct { /* XX: makes sense to enlarge it... */
2234 I32 type;
2235 I32 prev;
2236 CHECKPOINT lastcp;
2237} re_unwind_generic_t;
2238
2239typedef struct {
2240 I32 type;
2241 I32 prev;
2242 CHECKPOINT lastcp;
2243 I32 lastparen;
2244 regnode *next;
2245 char *locinput;
2246 I32 nextchr;
2247#ifdef DEBUGGING
2248 int regindent;
2249#endif
2250} re_unwind_branch_t;
2251
2252typedef union re_unwind_t {
2253 I32 type;
2254 re_unwind_generic_t generic;
2255 re_unwind_branch_t branch;
2256} re_unwind_t;
2257
8ba1375e
MJD
2258#define sayYES goto yes
2259#define sayNO goto no
e0f9d4a8 2260#define sayNO_ANYOF goto no_anyof
8ba1375e
MJD
2261#define sayYES_FINAL goto yes_final
2262#define sayYES_LOUD goto yes_loud
2263#define sayNO_FINAL goto no_final
2264#define sayNO_SILENT goto do_no
2265#define saySAME(x) if (x) goto yes; else goto no
2266
3ab3c9b4
HS
2267#define POSCACHE_SUCCESS 0 /* caching success rather than failure */
2268#define POSCACHE_SEEN 1 /* we know what we're caching */
2269#define POSCACHE_START 2 /* the real cache: this bit maps to pos 0 */
2270#define CACHEsayYES STMT_START { \
2271 if (cache_offset | cache_bit) { \
2272 if (!(PL_reg_poscache[0] & (1<<POSCACHE_SEEN))) \
2273 PL_reg_poscache[0] |= (1<<POSCACHE_SUCCESS) || (1<<POSCACHE_SEEN); \
2274 else if (!(PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))) { \
2275 /* cache records failure, but this is success */ \
2276 DEBUG_r( \
2277 PerlIO_printf(Perl_debug_log, \
2278 "%*s (remove success from failure cache)\n", \
2279 REPORT_CODE_OFF+PL_regindent*2, "") \
2280 ); \
2281 PL_reg_poscache[cache_offset] &= ~(1<<cache_bit); \
2282 } \
2283 } \
2284 sayYES; \
2285} STMT_END
2286#define CACHEsayNO STMT_START { \
2287 if (cache_offset | cache_bit) { \
2288 if (!(PL_reg_poscache[0] & (1<<POSCACHE_SEEN))) \
2289 PL_reg_poscache[0] |= (1<<POSCACHE_SEEN); \
2290 else if ((PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))) { \
2291 /* cache records success, but this is failure */ \
2292 DEBUG_r( \
2293 PerlIO_printf(Perl_debug_log, \
2294 "%*s (remove failure from success cache)\n", \
2295 REPORT_CODE_OFF+PL_regindent*2, "") \
2296 ); \
2297 PL_reg_poscache[cache_offset] &= ~(1<<cache_bit); \
2298 } \
2299 } \
2300 sayNO; \
2301} STMT_END
2302
a3621e74
YO
2303/* this is used to determine how far from the left messages like
2304 'failed...' are printed. Currently 29 makes these messages line
2305 up with the opcode they refer to. Earlier perls used 25 which
2306 left these messages outdented making reviewing a debug output
2307 quite difficult.
2308*/
2309#define REPORT_CODE_OFF 29
2310
2311
2312/* Make sure there is a test for this +1 options in re_tests */
2313#define TRIE_INITAL_ACCEPT_BUFFLEN 4;
2314
2315#define TRIE_CHECK_STATE_IS_ACCEPTING STMT_START { \
2316 if ( trie->states[ state ].wordnum ) { \
2317 if ( !accepted ) { \
2318 ENTER; \
2319 SAVETMPS; \
2320 bufflen = TRIE_INITAL_ACCEPT_BUFFLEN ; \
2321 sv_accept_buff=NEWSV( 1234, \
2322 bufflen * sizeof(reg_trie_accepted) - 1 ); \
2323 SvCUR_set( sv_accept_buff, sizeof(reg_trie_accepted) ); \
2324 SvPOK_on( sv_accept_buff ); \
2325 sv_2mortal( sv_accept_buff ); \
2326 accept_buff = (reg_trie_accepted*)SvPV_nolen( sv_accept_buff );\
2327 } else { \
2328 if ( accepted >= bufflen ) { \
2329 bufflen *= 2; \
2330 accept_buff =(reg_trie_accepted*)SvGROW( sv_accept_buff, \
2331 bufflen * sizeof(reg_trie_accepted) ); \
2332 } \
2333 SvCUR_set( sv_accept_buff,SvCUR( sv_accept_buff ) \
2334 + sizeof( reg_trie_accepted ) ); \
2335 } \
2336 accept_buff[ accepted ].wordnum = trie->states[ state ].wordnum; \
2337 accept_buff[ accepted ].endpos = uc; \
2338 ++accepted; \
2339 } } STMT_END
2340
2341#define TRIE_HANDLE_CHAR STMT_START { \
2342 if ( uvc < 256 ) { \
2343 charid = trie->charmap[ uvc ]; \
2344 } else { \
2345 charid = 0; \
2346 if( trie->widecharmap ) { \
2347 SV** svpp = (SV**)NULL; \
2348 svpp = hv_fetch( trie->widecharmap, (char*)&uvc, \
2349 sizeof( UV ), 0 ); \
2350 if ( svpp ) { \
2351 charid = (U16)SvIV( *svpp ); \
2352 } \
2353 } \
2354 } \
2355 if ( charid && \
cc601c31
YO
2356 ( base + charid > trie->uniquecharcount ) && \
2357 ( base + charid - 1 - trie->uniquecharcount < trie->lasttrans) && \
a3621e74
YO
2358 trie->trans[ base + charid - 1 - trie->uniquecharcount ].check == state ) \
2359 { \
2360 state = trie->trans[ base + charid - 1 - trie->uniquecharcount ].next; \
2361 } else { \
2362 state = 0; \
2363 } \
2364 uc += len; \
2365 } STMT_END
8ba1375e 2366
d6a28714
JH
2367/*
2368 - regmatch - main matching routine
2369 *
2370 * Conceptually the strategy is simple: check to see whether the current
2371 * node matches, call self recursively to see whether the rest matches,
2372 * and then act accordingly. In practice we make some effort to avoid
2373 * recursion, in particular by going through "ordinary" nodes (that don't
2374 * need to know whether the rest of the match failed) by a loop instead of
2375 * by recursion.
2376 */
2377/* [lwall] I've hoisted the register declarations to the outer block in order to
2378 * maybe save a little bit of pushing and popping on the stack. It also takes
2379 * advantage of machines that use a register save mask on subroutine entry.
2380 */
2381STATIC I32 /* 0 failure, 1 success */
2382S_regmatch(pTHX_ regnode *prog)
2383{
27da23d5 2384 dVAR;
d6a28714
JH
2385 register regnode *scan; /* Current node. */
2386 regnode *next; /* Next node. */
2387 regnode *inner; /* Next node in internal branch. */
2388 register I32 nextchr; /* renamed nextchr - nextchar colides with
2389 function of same name */
2390 register I32 n; /* no or next */
b7953727
JH
2391 register I32 ln = 0; /* len or last */
2392 register char *s = Nullch; /* operand or save */
d6a28714 2393 register char *locinput = PL_reginput;
b7953727 2394 register I32 c1 = 0, c2 = 0, paren; /* case fold search, parenth */
d6a28714 2395 int minmod = 0, sw = 0, logical = 0;
02db2b7b 2396 I32 unwind = 0;
a3621e74
YO
2397
2398 /* used by the trie code */
ab74612d
NC
2399 SV *sv_accept_buff = 0; /* accepting states we have traversed */
2400 reg_trie_accepted *accept_buff = 0; /* "" */
2401 reg_trie_data *trie; /* what trie are we using right now */
2402 U32 accepted = 0; /* how many accepting states we have seen*/
a3621e74 2403
b7953727 2404#if 0
02db2b7b 2405 I32 firstcp = PL_savestack_ix;
b7953727 2406#endif
a3b680e6 2407 const register bool do_utf8 = PL_reg_match_utf8;
2a782b5b 2408#ifdef DEBUGGING
ce333219
JH
2409 SV *dsv0 = PERL_DEBUG_PAD_ZERO(0);
2410 SV *dsv1 = PERL_DEBUG_PAD_ZERO(1);
2411 SV *dsv2 = PERL_DEBUG_PAD_ZERO(2);
a3621e74 2412
ab74612d 2413 SV *re_debug_flags = NULL;
2a782b5b 2414#endif
02db2b7b 2415
a3621e74
YO
2416 GET_RE_DEBUG_FLAGS;
2417
d6a28714
JH
2418#ifdef DEBUGGING
2419 PL_regindent++;
2420#endif
2421
a3621e74 2422
d6a28714
JH
2423 /* Note that nextchr is a byte even in UTF */
2424 nextchr = UCHARAT(locinput);
2425 scan = prog;
2426 while (scan != NULL) {
8ba1375e 2427
a3621e74 2428 DEBUG_EXECUTE_r( {
d6a28714
JH
2429 SV *prop = sv_newmortal();
2430 int docolor = *PL_colors[0];
2431 int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
1aa99e6b 2432 int l = (PL_regeol - locinput) > taill ? taill : (PL_regeol - locinput);
d6a28714
JH
2433 /* The part of the string before starttry has one color
2434 (pref0_len chars), between starttry and current
2435 position another one (pref_len - pref0_len chars),
2436 after the current position the third one.
2437 We assume that pref0_len <= pref_len, otherwise we
2438 decrease pref0_len. */
9041c2e3 2439 int pref_len = (locinput - PL_bostr) > (5 + taill) - l
1aa99e6b
IH
2440 ? (5 + taill) - l : locinput - PL_bostr;
2441 int pref0_len;
d6a28714 2442
df1ffd02 2443 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
1aa99e6b
IH
2444 pref_len++;
2445 pref0_len = pref_len - (locinput - PL_reg_starttry);
d6a28714 2446 if (l + pref_len < (5 + taill) && l < PL_regeol - locinput)
9041c2e3 2447 l = ( PL_regeol - locinput > (5 + taill) - pref_len
d6a28714 2448 ? (5 + taill) - pref_len : PL_regeol - locinput);
df1ffd02 2449 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
1aa99e6b 2450 l--;
d6a28714
JH
2451 if (pref0_len < 0)
2452 pref0_len = 0;
2453 if (pref0_len > pref_len)
2454 pref0_len = pref_len;
2455 regprop(prop, scan);
2a782b5b
JH
2456 {
2457 char *s0 =
f14c76ed 2458 do_utf8 && OP(scan) != CANY ?
2a782b5b 2459 pv_uni_display(dsv0, (U8*)(locinput - pref_len),
c728cb41 2460 pref0_len, 60, UNI_DISPLAY_REGEX) :
2a782b5b 2461 locinput - pref_len;
df1ffd02 2462 int len0 = do_utf8 ? strlen(s0) : pref0_len;
f14c76ed 2463 char *s1 = do_utf8 && OP(scan) != CANY ?
2a782b5b 2464 pv_uni_display(dsv1, (U8*)(locinput - pref_len + pref0_len),
c728cb41 2465 pref_len - pref0_len, 60, UNI_DISPLAY_REGEX) :
2a782b5b 2466 locinput - pref_len + pref0_len;
df1ffd02 2467 int len1 = do_utf8 ? strlen(s1) : pref_len - pref0_len;
f14c76ed 2468 char *s2 = do_utf8 && OP(scan) != CANY ?
2a782b5b 2469 pv_uni_display(dsv2, (U8*)locinput,
c728cb41 2470 PL_regeol - locinput, 60, UNI_DISPLAY_REGEX) :
2a782b5b 2471 locinput;
df1ffd02 2472 int len2 = do_utf8 ? strlen(s2) : l;
2a782b5b
JH
2473 PerlIO_printf(Perl_debug_log,
2474 "%4"IVdf" <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3"IVdf":%*s%s\n",
2475 (IV)(locinput - PL_bostr),
2476 PL_colors[4],
2477 len0, s0,
2478 PL_colors[5],
2479 PL_colors[2],
2480 len1, s1,
2481 PL_colors[3],
2482 (docolor ? "" : "> <"),
2483 PL_colors[0],
2484 len2, s2,
2485 PL_colors[1],
2486 15 - l - pref_len + 1,
2487 "",
2488 (IV)(scan - PL_regprogram), PL_regindent*2, "",
2489 SvPVX(prop));
2490 }
2491 });
d6a28714
JH
2492
2493 next = scan + NEXT_OFF(scan);
2494 if (next == scan)
2495 next = NULL;
2496
2497 switch (OP(scan)) {
2498 case BOL:
7fba1cd6 2499 if (locinput == PL_bostr)
d6a28714
JH
2500 {
2501 /* regtill = regbol; */
b8c5462f
JH
2502 break;
2503 }
d6a28714
JH
2504 sayNO;
2505 case MBOL:
12d33761
HS
2506 if (locinput == PL_bostr ||
2507 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
d6a28714 2508 {
b8c5462f
JH
2509 break;
2510 }
d6a28714
JH
2511 sayNO;
2512 case SBOL:
c2a73568 2513 if (locinput == PL_bostr)
b8c5462f 2514 break;
d6a28714
JH
2515 sayNO;
2516 case GPOS:
2517 if (locinput == PL_reg_ganch)
2518 break;
2519 sayNO;
2520 case EOL:
d6a28714
JH
2521 goto seol;
2522 case MEOL:
d6a28714 2523 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 2524 sayNO;
b8c5462f 2525 break;
d6a28714
JH
2526 case SEOL:
2527 seol:
2528 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 2529 sayNO;
d6a28714 2530 if (PL_regeol - locinput > 1)
b8c5462f 2531 sayNO;
b8c5462f 2532 break;
d6a28714
JH
2533 case EOS:
2534 if (PL_regeol != locinput)
b8c5462f 2535 sayNO;
d6a28714 2536 break;
ffc61ed2 2537 case SANY:
d6a28714 2538 if (!nextchr && locinput >= PL_regeol)
4633a7c4 2539 sayNO;
f33976b4
DB
2540 if (do_utf8) {
2541 locinput += PL_utf8skip[nextchr];
2542 if (locinput > PL_regeol)
2543 sayNO;
2544 nextchr = UCHARAT(locinput);
2545 }
2546 else
2547 nextchr = UCHARAT(++locinput);
2548 break;
2549 case CANY:
2550 if (!nextchr && locinput >= PL_regeol)
2551 sayNO;
b8c5462f 2552 nextchr = UCHARAT(++locinput);
a0d0e21e 2553 break;
ffc61ed2 2554 case REG_ANY:
1aa99e6b
IH
2555 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
2556 sayNO;
2557 if (do_utf8) {
b8c5462f 2558 locinput += PL_utf8skip[nextchr];
d6a28714
JH
2559 if (locinput > PL_regeol)
2560 sayNO;
a0ed51b3 2561 nextchr = UCHARAT(locinput);
a0ed51b3 2562 }
1aa99e6b
IH
2563 else
2564 nextchr = UCHARAT(++locinput);
a0ed51b3 2565 break;
a3621e74
YO
2566
2567
2568
2569 /*
2570 traverse the TRIE keeping track of all accepting states
2571 we transition through until we get to a failing node.
2572
2573 we use two slightly different pieces of code to handle
2574 the traversal depending on whether its case sensitive or
2575 not. we reuse the accept code however. (this should probably
2576 be turned into a macro.)
2577
2578 */
2579 case TRIEF:
2580 case TRIEFL:
2581 {
2582
a3b680e6 2583 const U32 uniflags = ckWARN( WARN_UTF8 ) ? 0 : UTF8_ALLOW_ANY;
a3621e74
YO
2584 U8 *uc = ( U8* )locinput;
2585 U32 state = 1;
2586 U16 charid = 0;
2587 U32 base = 0;
2588 UV uvc = 0;
2589 STRLEN len = 0;
2590 STRLEN foldlen = 0;
2591 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2592 U8 *uscan = (U8*)NULL;
2593 STRLEN bufflen=0;
2594 accepted = 0;
2595
2596 trie = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ];
2597
2598 while ( state && uc <= (U8*)PL_regeol ) {
2599
2600 TRIE_CHECK_STATE_IS_ACCEPTING;
2601
2602 base = trie->states[ state ].trans.base;
2603
2604 DEBUG_TRIE_EXECUTE_r(
2605 PerlIO_printf( Perl_debug_log,
e4584336 2606 "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ",
a3621e74 2607 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
e4584336 2608 (UV)state, (UV)base, (UV)accepted );
a3621e74
YO
2609 );
2610
2611 if ( base ) {
2612
2613 if ( do_utf8 || UTF ) {
2614 if ( foldlen>0 ) {
2615 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags );
2616 foldlen -= len;
2617 uscan += len;
2618 len=0;
2619 } else {
2620 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags );
2621 uvc = to_uni_fold( uvc, foldbuf, &foldlen );
2622 foldlen -= UNISKIP( uvc );
2623 uscan = foldbuf + UNISKIP( uvc );
2624 }
2625 } else {
e4584336 2626 uvc = (UV)*uc;
a3621e74
YO
2627 len = 1;
2628 }
2629
2630 TRIE_HANDLE_CHAR;
2631
2632 } else {
2633 state = 0;
2634 }
2635 DEBUG_TRIE_EXECUTE_r(
e4584336
RB
2636 PerlIO_printf( Perl_debug_log,
2637 "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n",
2638 charid, uvc, (UV)state, PL_colors[5] );
a3621e74
YO
2639 );
2640 }
2641 if ( !accepted ) {
2642 sayNO;
2643 } else {
2644 goto TrieAccept;
2645 }
2646 }
2647 /* unreached codepoint: we jump into the middle of the next case
2648 from previous if blocks */
2649 case TRIE:
2650 {
a3b680e6 2651 const U32 uniflags = ckWARN( WARN_UTF8 ) ? 0 : UTF8_ALLOW_ANY;
a3621e74
YO
2652 U8 *uc = (U8*)locinput;
2653 U32 state = 1;
2654 U16 charid = 0;
2655 U32 base = 0;
2656 UV uvc = 0;
2657 STRLEN len = 0;
2658 STRLEN bufflen = 0;
2659 accepted = 0;
2660
2661 trie = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ];
2662
2663 while ( state && uc <= (U8*)PL_regeol ) {
2664
2665 TRIE_CHECK_STATE_IS_ACCEPTING;
2666
2667 base = trie->states[ state ].trans.base;
2668
2669 DEBUG_TRIE_EXECUTE_r(
2670 PerlIO_printf( Perl_debug_log,
e4584336 2671 "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ",
a3621e74 2672 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
e4584336 2673 (UV)state, (UV)base, (UV)accepted );
a3621e74
YO
2674 );
2675
2676 if ( base ) {
2677
2678 if ( do_utf8 || UTF ) {
2679 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags );
2680 } else {
2681 uvc = (U32)*uc;
2682 len = 1;
2683 }
2684
2685 TRIE_HANDLE_CHAR;
2686
2687 } else {
2688 state = 0;
2689 }
2690 DEBUG_TRIE_EXECUTE_r(
2691 PerlIO_printf( Perl_debug_log,
e4584336
RB
2692 "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n",
2693 charid, uvc, (UV)state, PL_colors[5] );
a3621e74
YO
2694 );
2695 }
2696 if ( !accepted ) {
2697 sayNO;
2698 }
2699 }
2700
2701
2702 /*
2703 There was at least one accepting state that we
2704 transitioned through. Presumably the number of accepting
2705 states is going to be low, typically one or two. So we
2706 simply scan through to find the one with lowest wordnum.
2707 Once we find it, we swap the last state into its place
2708 and decrement the size. We then try to match the rest of
2709 the pattern at the point where the word ends, if we
2710 succeed then we end the loop, otherwise the loop
2711 eventually terminates once all of the accepting states
2712 have been tried.
2713 */
2714 TrieAccept:
2715 {
2716 int gotit = 0;
2717
2718 if ( accepted == 1 ) {
2719 DEBUG_EXECUTE_r({
2720 SV **tmp = av_fetch( trie->words, accept_buff[ 0 ].wordnum-1, 0 );
2721 PerlIO_printf( Perl_debug_log,
2722 "%*s %sonly one match : #%d <%s>%s\n",
e4584336 2723 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],
a3621e74
YO
2724 accept_buff[ 0 ].wordnum,
2725 tmp ? SvPV_nolen( *tmp ) : "not compiled under -Dr",
2726 PL_colors[5] );
2727 });
cc601c31 2728 PL_reginput = (char *)accept_buff[ 0 ].endpos;
a3621e74
YO
2729 /* in this case we free tmps/leave before we call regmatch
2730 as we wont be using accept_buff again. */
2731 FREETMPS;
2732 LEAVE;
2733 gotit = regmatch( scan + NEXT_OFF( scan ) );
2734 } else {
2735 DEBUG_EXECUTE_r(
e4584336
RB
2736 PerlIO_printf( Perl_debug_log,"%*s %sgot %"IVdf" possible matches%s\n",
2737 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], (IV)accepted,
a3621e74
YO
2738 PL_colors[5] );
2739 );
2740 while ( !gotit && accepted-- ) {
2741 U32 best = 0;
2742 U32 cur;
2743 for( cur = 1 ; cur <= accepted ; cur++ ) {
e4584336
RB
2744 DEBUG_TRIE_EXECUTE_r(
2745 PerlIO_printf( Perl_debug_log,
2746 "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n",
2747 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
2748 (IV)best, accept_buff[ best ].wordnum, (IV)cur,
2749 accept_buff[ cur ].wordnum, PL_colors[5] );
2750 );
a3621e74
YO
2751
2752 if ( accept_buff[ cur ].wordnum < accept_buff[ best ].wordnum )
2753 best = cur;
2754 }
2755 DEBUG_EXECUTE_r({
2756 SV **tmp = av_fetch( trie->words, accept_buff[ best ].wordnum - 1, 0 );
2757 PerlIO_printf( Perl_debug_log, "%*s %strying alternation #%d <%s> at 0x%p%s\n",
e4584336 2758 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],
a3621e74
YO
2759 accept_buff[best].wordnum,
2760 tmp ? SvPV_nolen( *tmp ) : "not compiled under -Dr",scan,
2761 PL_colors[5] );
2762 });
2763 if ( best<accepted ) {
2764 reg_trie_accepted tmp = accept_buff[ best ];
2765 accept_buff[ best ] = accept_buff[ accepted ];
2766 accept_buff[ accepted ] = tmp;
2767 best = accepted;
2768 }
cc601c31 2769 PL_reginput = (char *)accept_buff[ best ].endpos;
a3621e74
YO
2770
2771 /*
2772 as far as I can tell we only need the SAVETMPS/FREETMPS
2773 for re's with EVAL in them but I'm leaving them in for
2774 all until I can be sure.
2775 */
2776 SAVETMPS;
2777 gotit = regmatch( scan + NEXT_OFF( scan ) ) ;
2778 FREETMPS;
2779 }
2780 FREETMPS;
2781 LEAVE;
2782 }
2783
2784 if ( gotit ) {
2785 sayYES;
2786 } else {
2787 sayNO;
2788 }
2789 }
2790 /* unreached codepoint */
d6a28714 2791 case EXACT:
cd439c50
IZ
2792 s = STRING(scan);
2793 ln = STR_LEN(scan);
eb160463 2794 if (do_utf8 != UTF) {
bc517b45 2795 /* The target and the pattern have differing utf8ness. */
1aa99e6b 2796 char *l = locinput;
a3b680e6 2797 const char *e = s + ln;
a72c7584 2798
5ff6fc6d
JH
2799 if (do_utf8) {
2800 /* The target is utf8, the pattern is not utf8. */
1aa99e6b 2801 while (s < e) {
a3b680e6 2802 STRLEN ulen;
1aa99e6b 2803 if (l >= PL_regeol)
5ff6fc6d
JH
2804 sayNO;
2805 if (NATIVE_TO_UNI(*(U8*)s) !=
89ebb4a3 2806 utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
872c91ae
JH
2807 ckWARN(WARN_UTF8) ?
2808 0 : UTF8_ALLOW_ANY))
5ff6fc6d 2809 sayNO;
bc517b45 2810 l += ulen;
5ff6fc6d 2811 s ++;
1aa99e6b 2812 }
5ff6fc6d
JH
2813 }
2814 else {
2815 /* The target is not utf8, the pattern is utf8. */
1aa99e6b 2816 while (s < e) {
a3b680e6 2817 STRLEN ulen;
1aa99e6b
IH
2818 if (l >= PL_regeol)
2819 sayNO;
5ff6fc6d 2820 if (NATIVE_TO_UNI(*((U8*)l)) !=
89ebb4a3 2821 utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
872c91ae
JH
2822 ckWARN(WARN_UTF8) ?
2823 0 : UTF8_ALLOW_ANY))
1aa99e6b 2824 sayNO;
bc517b45 2825 s += ulen;
a72c7584 2826 l ++;
1aa99e6b 2827 }
5ff6fc6d 2828 }
1aa99e6b
IH
2829 locinput = l;
2830 nextchr = UCHARAT(locinput);
2831 break;
2832 }
bc517b45 2833 /* The target and the pattern have the same utf8ness. */
d6a28714
JH
2834 /* Inline the first character, for speed. */
2835 if (UCHARAT(s) != nextchr)
2836 sayNO;
2837 if (PL_regeol - locinput < ln)
2838 sayNO;
2839 if (ln > 1 && memNE(s, locinput, ln))
2840 sayNO;
2841 locinput += ln;
2842 nextchr = UCHARAT(locinput);
2843 break;
2844 case EXACTFL:
b8c5462f
JH
2845 PL_reg_flags |= RF_tainted;
2846 /* FALL THROUGH */
d6a28714 2847 case EXACTF:
cd439c50
IZ
2848 s = STRING(scan);
2849 ln = STR_LEN(scan);
d6a28714 2850
d07ddd77
JH
2851 if (do_utf8 || UTF) {
2852 /* Either target or the pattern are utf8. */
d6a28714 2853 char *l = locinput;
d07ddd77 2854 char *e = PL_regeol;
bc517b45 2855
eb160463 2856 if (ibcmp_utf8(s, 0, ln, (bool)UTF,
1feea2c7 2857 l, &e, 0, do_utf8)) {
5486206c
JH
2858 /* One more case for the sharp s:
2859 * pack("U0U*", 0xDF) =~ /ss/i,
2860 * the 0xC3 0x9F are the UTF-8
2861 * byte sequence for the U+00DF. */
2862 if (!(do_utf8 &&
2863 toLOWER(s[0]) == 's' &&
2864 ln >= 2 &&
2865 toLOWER(s[1]) == 's' &&
2866 (U8)l[0] == 0xC3 &&
2867 e - l >= 2 &&
2868 (U8)l[1] == 0x9F))
2869 sayNO;
2870 }
d07ddd77
JH
2871 locinput = e;
2872 nextchr = UCHARAT(locinput);
2873 break;
a0ed51b3 2874 }
d6a28714 2875
bc517b45
JH
2876 /* Neither the target and the pattern are utf8. */
2877
d6a28714
JH
2878 /* Inline the first character, for speed. */
2879 if (UCHARAT(s) != nextchr &&
2880 UCHARAT(s) != ((OP(scan) == EXACTF)
2881 ? PL_fold : PL_fold_locale)[nextchr])
a0ed51b3 2882 sayNO;
d6a28714 2883 if (PL_regeol - locinput < ln)
b8c5462f 2884 sayNO;
d6a28714
JH
2885 if (ln > 1 && (OP(scan) == EXACTF
2886 ? ibcmp(s, locinput, ln)
2887 : ibcmp_locale(s, locinput, ln)))
4633a7c4 2888 sayNO;
d6a28714
JH
2889 locinput += ln;
2890 nextchr = UCHARAT(locinput);
a0d0e21e 2891 break;
d6a28714 2892 case ANYOF:
ffc61ed2 2893 if (do_utf8) {
9e55ce06
JH
2894 STRLEN inclasslen = PL_regeol - locinput;
2895
ba7b4546 2896 if (!reginclass(scan, (U8*)locinput, &inclasslen, do_utf8))
e0f9d4a8 2897 sayNO_ANYOF;
ffc61ed2
JH
2898 if (locinput >= PL_regeol)
2899 sayNO;
0f0076b4 2900 locinput += inclasslen ? inclasslen : UTF8SKIP(locinput);
b8c5462f 2901 nextchr = UCHARAT(locinput);
e0f9d4a8 2902 break;
ffc61ed2
JH
2903 }
2904 else {
2905 if (nextchr < 0)
2906 nextchr = UCHARAT(locinput);
7d3e948e 2907 if (!REGINCLASS(scan, (U8*)locinput))
e0f9d4a8 2908 sayNO_ANYOF;
ffc61ed2
JH
2909 if (!nextchr && locinput >= PL_regeol)
2910 sayNO;
2911 nextchr = UCHARAT(++locinput);
e0f9d4a8
JH
2912 break;
2913 }
2914 no_anyof:
2915 /* If we might have the case of the German sharp s
2916 * in a casefolding Unicode character class. */
2917
ebc501f0
JH
2918 if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
2919 locinput += SHARP_S_SKIP;
e0f9d4a8 2920 nextchr = UCHARAT(locinput);
ffc61ed2 2921 }
e0f9d4a8
JH
2922 else
2923 sayNO;
b8c5462f 2924 break;
d6a28714 2925 case ALNUML:
b8c5462f
JH
2926 PL_reg_flags |= RF_tainted;
2927 /* FALL THROUGH */
d6a28714 2928 case ALNUM:
b8c5462f 2929 if (!nextchr)
4633a7c4 2930 sayNO;
ffc61ed2 2931 if (do_utf8) {
ad24be35 2932 LOAD_UTF8_CHARCLASS(alnum,"a");
ffc61ed2 2933 if (!(OP(scan) == ALNUM
3568d838 2934 ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
d6a28714 2935 : isALNUM_LC_utf8((U8*)locinput)))
b8c5462f
JH
2936 {
2937 sayNO;
a0ed51b3 2938 }
b8c5462f 2939 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
2940 nextchr = UCHARAT(locinput);
2941 break;
2942 }
ffc61ed2 2943 if (!(OP(scan) == ALNUM
d6a28714 2944 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
bbce6d69 2945 sayNO;
b8c5462f 2946 nextchr = UCHARAT(++locinput);
a0d0e21e 2947 break;
d6a28714 2948 case NALNUML:
b8c5462f
JH
2949 PL_reg_flags |= RF_tainted;
2950 /* FALL THROUGH */
d6a28714
JH
2951 case NALNUM:
2952 if (!nextchr && locinput >= PL_regeol)
a0ed51b3 2953 sayNO;
ffc61ed2 2954 if (do_utf8) {
8269fa76 2955 LOAD_UTF8_CHARCLASS(alnum,"a");
ffc61ed2 2956 if (OP(scan) == NALNUM
3568d838 2957 ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
d6a28714
JH
2958 : isALNUM_LC_utf8((U8*)locinput))
2959 {
b8c5462f 2960 sayNO;
d6a28714 2961 }
b8c5462f
JH
2962 locinput += PL_utf8skip[nextchr];
2963 nextchr = UCHARAT(locinput);
2964 break;
2965 }
ffc61ed2 2966 if (OP(scan) == NALNUM
d6a28714 2967 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
b8c5462f 2968 sayNO;
76e3520e 2969 nextchr = UCHARAT(++locinput);
a0d0e21e 2970 break;
d6a28714
JH
2971 case BOUNDL:
2972 case NBOUNDL:
3280af22 2973 PL_reg_flags |= RF_tainted;
bbce6d69 2974 /* FALL THROUGH */
d6a28714
JH
2975 case BOUND:
2976 case NBOUND:
2977 /* was last char in word? */
ffc61ed2 2978 if (do_utf8) {
12d33761
HS
2979 if (locinput == PL_bostr)
2980 ln = '\n';
ffc61ed2 2981 else {
a3b680e6 2982 const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
9041c2e3 2983
b4f7163a 2984 ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
ffc61ed2
JH
2985 }
2986 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
2987 ln = isALNUM_uni(ln);
8269fa76 2988 LOAD_UTF8_CHARCLASS(alnum,"a");
3568d838 2989 n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8);
ffc61ed2
JH
2990 }
2991 else {
9041c2e3 2992 ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln));
ffc61ed2
JH
2993 n = isALNUM_LC_utf8((U8*)locinput);
2994 }
a0ed51b3 2995 }
d6a28714 2996 else {
12d33761
HS
2997 ln = (locinput != PL_bostr) ?
2998 UCHARAT(locinput - 1) : '\n';
ffc61ed2
JH
2999 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
3000 ln = isALNUM(ln);
3001 n = isALNUM(nextchr);
3002 }
3003 else {
3004 ln = isALNUM_LC(ln);
3005 n = isALNUM_LC(nextchr);
3006 }
d6a28714 3007 }
ffc61ed2
JH
3008 if (((!ln) == (!n)) == (OP(scan) == BOUND ||
3009 OP(scan) == BOUNDL))
3010 sayNO;
a0ed51b3 3011 break;
d6a28714 3012 case SPACEL:
3280af22 3013 PL_reg_flags |= RF_tainted;
bbce6d69 3014 /* FALL THROUGH */
d6a28714 3015 case SPACE:
9442cb0e 3016 if (!nextchr)
4633a7c4 3017 sayNO;
1aa99e6b 3018 if (do_utf8) {
fd400ab9 3019 if (UTF8_IS_CONTINUED(nextchr)) {
8269fa76 3020 LOAD_UTF8_CHARCLASS(space," ");
ffc61ed2 3021 if (!(OP(scan) == SPACE
3568d838 3022 ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
ffc61ed2
JH
3023 : isSPACE_LC_utf8((U8*)locinput)))
3024 {
3025 sayNO;
3026 }
3027 locinput += PL_utf8skip[nextchr];
3028 nextchr = UCHARAT(locinput);
3029 break;
d6a28714 3030 }
ffc61ed2
JH
3031 if (!(OP(scan) == SPACE
3032 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
3033 sayNO;
3034 nextchr = UCHARAT(++locinput);
3035 }
3036 else {
3037 if (!(OP(scan) == SPACE
3038 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
3039 sayNO;
3040 nextchr = UCHARAT(++locinput);
a0ed51b3 3041 }
a0ed51b3 3042 break;
d6a28714 3043 case NSPACEL:
3280af22 3044 PL_reg_flags |= RF_tainted;
bbce6d69 3045 /* FALL THROUGH */
d6a28714 3046 case NSPACE:
9442cb0e 3047 if (!nextchr && locinput >= PL_regeol)
b8c5462f 3048 sayNO;
1aa99e6b 3049 if (do_utf8) {
8269fa76 3050 LOAD_UTF8_CHARCLASS(space," ");
ffc61ed2 3051 if (OP(scan) == NSPACE
3568d838 3052 ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
d6a28714 3053 : isSPACE_LC_utf8((U8*)locinput))
b8c5462f
JH
3054 {
3055 sayNO;
3056 }
3057 locinput += PL_utf8skip[nextchr];
3058 nextchr = UCHARAT(locinput);
3059 break;
a0ed51b3 3060 }
ffc61ed2 3061 if (OP(scan) == NSPACE
d6a28714 3062 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
4633a7c4 3063 sayNO;
76e3520e 3064 nextchr = UCHARAT(++locinput);
a0d0e21e 3065 break;
d6a28714 3066 case DIGITL:
a0ed51b3
LW
3067 PL_reg_flags |= RF_tainted;
3068 /* FALL THROUGH */
d6a28714 3069 case DIGIT:
9442cb0e 3070 if (!nextchr)
a0ed51b3 3071 sayNO;
1aa99e6b 3072 if (do_utf8) {
8269fa76 3073 LOAD_UTF8_CHARCLASS(digit,"0");
ffc61ed2 3074 if (!(OP(scan) == DIGIT
3568d838 3075 ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
9442cb0e 3076 : isDIGIT_LC_utf8((U8*)locinput)))
dfe13c55 3077 {
a0ed51b3 3078 sayNO;
dfe13c55 3079 }
6f06b55f 3080 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
3081 nextchr = UCHARAT(locinput);
3082 break;
3083 }
ffc61ed2 3084 if (!(OP(scan) == DIGIT
9442cb0e 3085 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
a0ed51b3
LW
3086 sayNO;
3087 nextchr = UCHARAT(++locinput);
3088 break;
d6a28714 3089 case NDIGITL:
b8c5462f
JH
3090 PL_reg_flags |= RF_tainted;
3091 /* FALL THROUGH */
d6a28714 3092 case NDIGIT:
9442cb0e 3093 if (!nextchr && locinput >= PL_regeol)
b8c5462f 3094 sayNO;
1aa99e6b 3095 if (do_utf8) {
8269fa76 3096 LOAD_UTF8_CHARCLASS(digit,"0");
ffc61ed2 3097 if (OP(scan) == NDIGIT
3568d838 3098 ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
9442cb0e
GS
3099 : isDIGIT_LC_utf8((U8*)locinput))
3100 {
a0ed51b3 3101 sayNO;
9442cb0e 3102 }
6f06b55f 3103 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
3104 nextchr = UCHARAT(locinput);
3105 break;
3106 }
ffc61ed2 3107 if (OP(scan) == NDIGIT
9442cb0e 3108 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
a0ed51b3
LW
3109 sayNO;
3110 nextchr = UCHARAT(++locinput);
3111 break;
3112 case CLUMP:
b7c83a7e 3113 if (locinput >= PL_regeol)
a0ed51b3 3114 sayNO;
b7c83a7e
JH
3115 if (do_utf8) {
3116 LOAD_UTF8_CHARCLASS(mark,"~");
3117 if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
3118 sayNO;
3119 locinput += PL_utf8skip[nextchr];
3120 while (locinput < PL_regeol &&
3121 swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
3122 locinput += UTF8SKIP(locinput);
3123 if (locinput > PL_regeol)
3124 sayNO;
eb08e2da
JH
3125 }
3126 else
3127 locinput++;
a0ed51b3
LW
3128 nextchr = UCHARAT(locinput);
3129 break;
c8756f30 3130 case REFFL:
3280af22 3131 PL_reg_flags |= RF_tainted;
c8756f30 3132 /* FALL THROUGH */
c277df42 3133 case REF:
c8756f30 3134 case REFF:
c277df42 3135 n = ARG(scan); /* which paren pair */
cf93c79d 3136 ln = PL_regstartp[n];
2c2d71f5 3137 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
eb160463 3138 if ((I32)*PL_reglastparen < n || ln == -1)
af3f8c16 3139 sayNO; /* Do not match unless seen CLOSEn. */
cf93c79d 3140 if (ln == PL_regendp[n])
a0d0e21e 3141 break;
a0ed51b3 3142
cf93c79d 3143 s = PL_bostr + ln;
1aa99e6b 3144 if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */
a0ed51b3 3145 char *l = locinput;
a3b680e6 3146 const char *e = PL_bostr + PL_regendp[n];
a0ed51b3
LW
3147 /*
3148 * Note that we can't do the "other character" lookup trick as
3149 * in the 8-bit case (no pun intended) because in Unicode we
3150 * have to map both upper and title case to lower case.
3151 */
3152 if (OP(scan) == REFF) {
3153 while (s < e) {
a3b680e6
AL
3154 STRLEN ulen1, ulen2;
3155 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
3156 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
3157
a0ed51b3
LW
3158 if (l >= PL_regeol)
3159 sayNO;
a2a2844f
JH
3160 toLOWER_utf8((U8*)s, tmpbuf1, &ulen1);
3161 toLOWER_utf8((U8*)l, tmpbuf2, &ulen2);
7114a2d2 3162 if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
a0ed51b3 3163 sayNO;
a2a2844f
JH
3164 s += ulen1;
3165 l += ulen2;
a0ed51b3
LW
3166 }
3167 }
3168 locinput = l;
3169 nextchr = UCHARAT(locinput);
3170 break;
3171 }
3172
a0d0e21e 3173 /* Inline the first character, for speed. */
76e3520e 3174 if (UCHARAT(s) != nextchr &&
c8756f30
AK
3175 (OP(scan) == REF ||
3176 (UCHARAT(s) != ((OP(scan) == REFF
22c35a8c 3177 ? PL_fold : PL_fold_locale)[nextchr]))))
4633a7c4 3178 sayNO;
cf93c79d 3179 ln = PL_regendp[n] - ln;
3280af22 3180 if (locinput + ln > PL_regeol)
4633a7c4 3181 sayNO;
c8756f30
AK
3182 if (ln > 1 && (OP(scan) == REF
3183 ? memNE(s, locinput, ln)
3184 : (OP(scan) == REFF
3185 ? ibcmp(s, locinput, ln)
3186 : ibcmp_locale(s, locinput, ln))))
4633a7c4 3187 sayNO;
a0d0e21e 3188 locinput += ln;
76e3520e 3189 nextchr = UCHARAT(locinput);
a0d0e21e
LW
3190 break;
3191
3192 case NOTHING:
c277df42 3193 case TAIL:
a0d0e21e
LW
3194 break;
3195 case BACK:
3196 break;
c277df42
IZ
3197 case EVAL:
3198 {
3199 dSP;
533c011a 3200 OP_4tree *oop = PL_op;
3280af22 3201 COP *ocurcop = PL_curcop;
f3548bdc 3202 PAD *old_comppad;
c277df42 3203 SV *ret;
080c2dec 3204 struct regexp *oreg = PL_reg_re;
9041c2e3 3205
c277df42 3206 n = ARG(scan);
533c011a 3207 PL_op = (OP_4tree*)PL_regdata->data[n];
a3621e74 3208 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
f3548bdc 3209 PAD_SAVE_LOCAL(old_comppad, (PAD*)PL_regdata->data[n + 2]);
cf93c79d 3210 PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
c277df42 3211
8e5e9ebe
RGS
3212 {
3213 SV **before = SP;
3214 CALLRUNOPS(aTHX); /* Scalar context. */
3215 SPAGAIN;
3216 if (SP == before)
075aa684 3217 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
8e5e9ebe
RGS
3218 else {
3219 ret = POPs;
3220 PUTBACK;
3221 }
3222 }
3223
0f5d15d6 3224 PL_op = oop;
f3548bdc 3225 PAD_RESTORE_LOCAL(old_comppad);
0f5d15d6 3226 PL_curcop = ocurcop;
c277df42 3227 if (logical) {
0f5d15d6
IZ
3228 if (logical == 2) { /* Postponed subexpression. */
3229 regexp *re;
22c35a8c 3230 MAGIC *mg = Null(MAGIC*);
0f5d15d6 3231 re_cc_state state;
0f5d15d6 3232 CHECKPOINT cp, lastcp;
cb50f42d 3233 int toggleutf;
faf82a0b 3234 register SV *sv;
0f5d15d6 3235
faf82a0b
AE
3236 if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret)))
3237 mg = mg_find(sv, PERL_MAGIC_qr);
3238 else if (SvSMAGICAL(ret)) {
3239 if (SvGMAGICAL(ret))
3240 sv_unmagic(ret, PERL_MAGIC_qr);
3241 else
3242 mg = mg_find(ret, PERL_MAGIC_qr);
0f5d15d6 3243 }
faf82a0b 3244
0f5d15d6
IZ
3245 if (mg) {
3246 re = (regexp *)mg->mg_obj;
df0003d4 3247 (void)ReREFCNT_inc(re);
0f5d15d6
IZ
3248 }
3249 else {
3250 STRLEN len;
3251 char *t = SvPV(ret, len);
3252 PMOP pm;
a3b680e6
AL
3253 char * const oprecomp = PL_regprecomp;
3254 const I32 osize = PL_regsize;
3255 const I32 onpar = PL_regnpar;
0f5d15d6 3256
5fcd1c1b 3257 Zero(&pm, 1, PMOP);
cb50f42d 3258 if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8;
cea2e8a9 3259 re = CALLREGCOMP(aTHX_ t, t + len, &pm);
9041c2e3 3260 if (!(SvFLAGS(ret)
faf82a0b
AE
3261 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
3262 | SVs_GMG)))
14befaf4
DM
3263 sv_magic(ret,(SV*)ReREFCNT_inc(re),
3264 PERL_MAGIC_qr,0,0);
0f5d15d6
IZ
3265 PL_regprecomp = oprecomp;
3266 PL_regsize = osize;
3267 PL_regnpar = onpar;
3268 }
a3621e74 3269 DEBUG_EXECUTE_r(
9041c2e3 3270 PerlIO_printf(Perl_debug_log,
0f5d15d6
IZ
3271 "Entering embedded `%s%.60s%s%s'\n",
3272 PL_colors[0],
3273 re->precomp,
3274 PL_colors[1],
3275 (strlen(re->precomp) > 60 ? "..." : ""))
3276 );
3277 state.node = next;
3278 state.prev = PL_reg_call_cc;
3279 state.cc = PL_regcc;
3280 state.re = PL_reg_re;
3281
2ab05381 3282 PL_regcc = 0;
9041c2e3 3283
0f5d15d6 3284 cp = regcppush(0); /* Save *all* the positions. */
02db2b7b 3285 REGCP_SET(lastcp);
0f5d15d6
IZ
3286 cache_re(re);
3287 state.ss = PL_savestack_ix;
3288 *PL_reglastparen = 0;
a01268b5 3289 *PL_reglastcloseparen = 0;
0f5d15d6
IZ
3290 PL_reg_call_cc = &state;
3291 PL_reginput = locinput;
cb50f42d
YST
3292 toggleutf = ((PL_reg_flags & RF_utf8) != 0) ^
3293 ((re->reganch & ROPT_UTF8) != 0);
3294 if (toggleutf) PL_reg_flags ^= RF_utf8;
2c2d71f5
JH
3295
3296 /* XXXX This is too dramatic a measure... */
3297 PL_reg_maxiter = 0;
3298
0f5d15d6 3299 if (regmatch(re->program + 1)) {
2c914db6
IZ
3300 /* Even though we succeeded, we need to restore
3301 global variables, since we may be wrapped inside
3302 SUSPEND, thus the match may be not finished yet. */
3303
3304 /* XXXX Do this only if SUSPENDed? */
3305 PL_reg_call_cc = state.prev;
3306 PL_regcc = state.cc;
3307 PL_reg_re = state.re;
3308 cache_re(PL_reg_re);
cb50f42d 3309 if (toggleutf) PL_reg_flags ^= RF_utf8;
2c914db6
IZ
3310
3311 /* XXXX This is too dramatic a measure... */
3312 PL_reg_maxiter = 0;
3313
3314 /* These are needed even if not SUSPEND. */
0f5d15d6
IZ
3315 ReREFCNT_dec(re);
3316 regcpblow(cp);
3317 sayYES;
3318 }
0f5d15d6 3319 ReREFCNT_dec(re);
02db2b7b 3320 REGCP_UNWIND(lastcp);
0f5d15d6
IZ
3321 regcppop();
3322 PL_reg_call_cc = state.prev;
3323 PL_regcc = state.cc;
3324 PL_reg_re = state.re;
d3790889 3325 cache_re(PL_reg_re);
cb50f42d 3326 if (toggleutf) PL_reg_flags ^= RF_utf8;
2c2d71f5
JH
3327
3328 /* XXXX This is too dramatic a measure... */
3329 PL_reg_maxiter = 0;
3330
8e514ae6 3331 logical = 0;
0f5d15d6
IZ
3332 sayNO;
3333 }
c277df42 3334 sw = SvTRUE(ret);
0f5d15d6 3335 logical = 0;
a0ed51b3 3336 }
080c2dec 3337 else {
3280af22 3338 sv_setsv(save_scalar(PL_replgv), ret);
080c2dec
AE
3339 cache_re(oreg);
3340 }
c277df42
IZ
3341 break;
3342 }
a0d0e21e 3343 case OPEN:
c277df42 3344 n = ARG(scan); /* which paren pair */
3280af22
NIS
3345 PL_reg_start_tmp[n] = locinput;
3346 if (n > PL_regsize)
3347 PL_regsize = n;
a0d0e21e
LW
3348 break;
3349 case CLOSE:
c277df42 3350 n = ARG(scan); /* which paren pair */
cf93c79d
IZ
3351 PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr;
3352 PL_regendp[n] = locinput - PL_bostr;
eb160463 3353 if (n > (I32)*PL_reglastparen)
3280af22 3354 *PL_reglastparen = n;
a01268b5 3355 *PL_reglastcloseparen = n;
a0d0e21e 3356 break;
c277df42
IZ
3357 case GROUPP:
3358 n = ARG(scan); /* which paren pair */
eb160463 3359 sw = ((I32)*PL_reglastparen >= n && PL_regendp[n] != -1);
c277df42
IZ
3360 break;
3361 case IFTHEN:
2c2d71f5 3362 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
c277df42
IZ
3363 if (sw)
3364 next = NEXTOPER(NEXTOPER(scan));
3365 else {
3366 next = scan + ARG(scan);
3367 if (OP(next) == IFTHEN) /* Fake one. */
3368 next = NEXTOPER(NEXTOPER(next));
3369 }
3370 break;
3371 case LOGICAL:
0f5d15d6 3372 logical = scan->flags;
c277df42 3373 break;
2ab05381
IZ
3374/*******************************************************************
3375 PL_regcc contains infoblock about the innermost (...)* loop, and
3376 a pointer to the next outer infoblock.
3377
3378 Here is how Y(A)*Z is processed (if it is compiled into CURLYX/WHILEM):
3379
3380 1) After matching X, regnode for CURLYX is processed;
3381
9041c2e3 3382 2) This regnode creates infoblock on the stack, and calls
2ab05381
IZ
3383 regmatch() recursively with the starting point at WHILEM node;
3384
3385 3) Each hit of WHILEM node tries to match A and Z (in the order
3386 depending on the current iteration, min/max of {min,max} and
3387 greediness). The information about where are nodes for "A"
3388 and "Z" is read from the infoblock, as is info on how many times "A"
3389 was already matched, and greediness.
3390
3391 4) After A matches, the same WHILEM node is hit again.
3392
3393 5) Each time WHILEM is hit, PL_regcc is the infoblock created by CURLYX
3394 of the same pair. Thus when WHILEM tries to match Z, it temporarily
3395 resets PL_regcc, since this Y(A)*Z can be a part of some other loop:
3396 as in (Y(A)*Z)*. If Z matches, the automaton will hit the WHILEM node
3397 of the external loop.
3398
3399 Currently present infoblocks form a tree with a stem formed by PL_curcc
3400 and whatever it mentions via ->next, and additional attached trees
3401 corresponding to temporarily unset infoblocks as in "5" above.
3402
9041c2e3 3403 In the following picture infoblocks for outer loop of
2ab05381
IZ
3404 (Y(A)*?Z)*?T are denoted O, for inner I. NULL starting block
3405 is denoted by x. The matched string is YAAZYAZT. Temporarily postponed
3406 infoblocks are drawn below the "reset" infoblock.
3407
3408 In fact in the picture below we do not show failed matches for Z and T
3409 by WHILEM blocks. [We illustrate minimal matches, since for them it is
3410 more obvious *why* one needs to *temporary* unset infoblocks.]
3411
3412 Matched REx position InfoBlocks Comment
3413 (Y(A)*?Z)*?T x
3414 Y(A)*?Z)*?T x <- O
3415 Y (A)*?Z)*?T x <- O
3416 Y A)*?Z)*?T x <- O <- I
3417 YA )*?Z)*?T x <- O <- I
3418 YA A)*?Z)*?T x <- O <- I
3419 YAA )*?Z)*?T x <- O <- I
3420 YAA Z)*?T x <- O # Temporary unset I
3421 I
3422
3423 YAAZ Y(A)*?Z)*?T x <- O
3424 I
3425
3426 YAAZY (A)*?Z)*?T x <- O
3427 I
3428
3429 YAAZY A)*?Z)*?T x <- O <- I
3430 I
3431
3432 YAAZYA )*?Z)*?T x <- O <- I
3433 I
3434
3435 YAAZYA Z)*?T x <- O # Temporary unset I
3436 I,I
3437
3438 YAAZYAZ )*?T x <- O
3439 I,I
3440
3441 YAAZYAZ T x # Temporary unset O
3442 O
3443 I,I
3444
3445 YAAZYAZT x
3446 O
3447 I,I
3448 *******************************************************************/
a0d0e21e
LW
3449 case CURLYX: {
3450 CURCUR cc;
3280af22 3451 CHECKPOINT cp = PL_savestack_ix;
cb434fcc
IZ
3452 /* No need to save/restore up to this paren */
3453 I32 parenfloor = scan->flags;
c277df42
IZ
3454
3455 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
3456 next += ARG(next);
3280af22
NIS
3457 cc.oldcc = PL_regcc;
3458 PL_regcc = &cc;
cb434fcc
IZ
3459 /* XXXX Probably it is better to teach regpush to support
3460 parenfloor > PL_regsize... */
eb160463 3461 if (parenfloor > (I32)*PL_reglastparen)
cb434fcc
IZ
3462 parenfloor = *PL_reglastparen; /* Pessimization... */
3463 cc.parenfloor = parenfloor;
a0d0e21e
LW
3464 cc.cur = -1;
3465 cc.min = ARG1(scan);
3466 cc.max = ARG2(scan);
c277df42 3467 cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
a0d0e21e
LW
3468 cc.next = next;
3469 cc.minmod = minmod;
3470 cc.lastloc = 0;
3280af22 3471 PL_reginput = locinput;
a0d0e21e
LW
3472 n = regmatch(PREVOPER(next)); /* start on the WHILEM */
3473 regcpblow(cp);
3280af22 3474 PL_regcc = cc.oldcc;
4633a7c4 3475 saySAME(n);
a0d0e21e
LW
3476 }
3477 /* NOT REACHED */
3478 case WHILEM: {
3479 /*
3480 * This is really hard to understand, because after we match
3481 * what we're trying to match, we must make sure the rest of
2c2d71f5 3482 * the REx is going to match for sure, and to do that we have
a0d0e21e
LW
3483 * to go back UP the parse tree by recursing ever deeper. And
3484 * if it fails, we have to reset our parent's current state
3485 * that we can try again after backing off.
3486 */
3487
c277df42 3488 CHECKPOINT cp, lastcp;
3280af22 3489 CURCUR* cc = PL_regcc;
c277df42 3490 char *lastloc = cc->lastloc; /* Detection of 0-len. */
3ab3c9b4 3491 I32 cache_offset = 0, cache_bit = 0;
c277df42 3492
4633a7c4 3493 n = cc->cur + 1; /* how many we know we matched */
3280af22 3494 PL_reginput = locinput;
a0d0e21e 3495
a3621e74 3496 DEBUG_EXECUTE_r(
9041c2e3 3497 PerlIO_printf(Perl_debug_log,
91f3b821 3498 "%*s %ld out of %ld..%ld cc=%"UVxf"\n",
3280af22 3499 REPORT_CODE_OFF+PL_regindent*2, "",
9041c2e3 3500 (long)n, (long)cc->min,
2797576d 3501 (long)cc->max, PTR2UV(cc))
c277df42 3502 );
4633a7c4 3503
a0d0e21e
LW
3504 /* If degenerate scan matches "", assume scan done. */
3505
579cf2c3 3506 if (locinput == cc->lastloc && n >= cc->min) {
3280af22 3507 PL_regcc = cc->oldcc;
2ab05381
IZ
3508 if (PL_regcc)
3509 ln = PL_regcc->cur;
a3621e74 3510 DEBUG_EXECUTE_r(
c3464db5
DD
3511 PerlIO_printf(Perl_debug_log,
3512 "%*s empty match detected, try continuation...\n",
3280af22 3513 REPORT_CODE_OFF+PL_regindent*2, "")
c277df42 3514 );
a0d0e21e 3515 if (regmatch(cc->next))
4633a7c4 3516 sayYES;
2ab05381
IZ
3517 if (PL_regcc)
3518 PL_regcc->cur = ln;
3280af22 3519 PL_regcc = cc;
4633a7c4 3520 sayNO;
a0d0e21e
LW
3521 }
3522
3523 /* First just match a string of min scans. */
3524
3525 if (n < cc->min) {