This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Re: \N{...} in regular expression [PATCH]
[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.
166f8a29
DM
16 */
17
a687059c
LW
18/* NOTE: this is derived from Henry Spencer's regexp code, and should not
19 * confused with the original package (see point 3 below). Thanks, Henry!
20 */
21
22/* Additional note: this code is very heavily munged from Henry's version
23 * in places. In some spots I've traded clarity for efficiency, so don't
24 * blame Henry for some of the lack of readability.
25 */
26
e50aee73
AD
27/* The names of the functions have been changed from regcomp and
28 * regexec to pregcomp and pregexec in order to avoid conflicts
29 * with the POSIX routines of the same names.
30*/
31
b9d5759e 32#ifdef PERL_EXT_RE_BUILD
54df2634 33#include "re_top.h"
9041c2e3 34#endif
56953603 35
a687059c 36/*
e50aee73 37 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
38 *
39 * Copyright (c) 1986 by University of Toronto.
40 * Written by Henry Spencer. Not derived from licensed software.
41 *
42 * Permission is granted to anyone to use this software for any
43 * purpose on any computer system, and to redistribute it freely,
44 * subject to the following restrictions:
45 *
46 * 1. The author is not responsible for the consequences of use of
47 * this software, no matter how awful, even if they arise
48 * from defects in it.
49 *
50 * 2. The origin of this software must not be misrepresented, either
51 * by explicit claim or by omission.
52 *
53 * 3. Altered versions must be plainly marked as such, and must not
54 * be misrepresented as being the original software.
55 *
56 **** Alterations to Henry's code are...
57 ****
4bb101f2 58 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
b94e2f88 59 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, by Larry Wall and others
a687059c 60 ****
9ef589d8
LW
61 **** You may distribute under the terms of either the GNU General Public
62 **** License or the Artistic License, as specified in the README file.
a687059c
LW
63 *
64 * Beware that some of this code is subtly aware of the way operator
65 * precedence is structured in regular expressions. Serious changes in
66 * regular-expression syntax might require a total rethink.
67 */
68#include "EXTERN.h"
864dbfa3 69#define PERL_IN_REGEXEC_C
a687059c 70#include "perl.h"
0f5d15d6 71
54df2634
NC
72#ifdef PERL_IN_XSUB_RE
73# include "re_comp.h"
74#else
75# include "regcomp.h"
76#endif
a687059c 77
c277df42
IZ
78#define RF_tainted 1 /* tainted information used? */
79#define RF_warned 2 /* warned about big count? */
ce862d02 80#define RF_evaled 4 /* Did an EVAL with setting? */
ab3bbdeb 81#define RF_utf8 8 /* Pattern contains multibyte chars? */
a0ed51b3 82
eb160463 83#define UTF ((PL_reg_flags & RF_utf8) != 0)
ce862d02
IZ
84
85#define RS_init 1 /* eval environment created */
86#define RS_set 2 /* replsv value is set */
c277df42 87
a687059c
LW
88#ifndef STATIC
89#define STATIC static
90#endif
91
32fc9b6a 92#define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
7d3e948e 93
c277df42
IZ
94/*
95 * Forwards.
96 */
97
33b8afdf 98#define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv))
53c4c00c 99#define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)
a0ed51b3 100
3dab1dad
YO
101#define HOPc(pos,off) \
102 (char *)(PL_reg_match_utf8 \
52657f30 103 ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \
3dab1dad
YO
104 : (U8*)(pos + off))
105#define HOPBACKc(pos, off) \
07be1b83
YO
106 (char*)(PL_reg_match_utf8\
107 ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \
108 : (pos - off >= PL_bostr) \
8e11feef 109 ? (U8*)pos - off \
3dab1dad 110 : NULL)
efb30f32 111
e7409c1b 112#define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
1aa99e6b 113#define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
1aa99e6b 114
1a4fad37
AL
115#define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \
116 if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END
117#define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
118#define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
119#define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ")
120#define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86")
51371543 121
3dab1dad
YO
122/* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
123
5f80c4cf 124/* for use after a quantifier and before an EXACT-like node -- japhy */
e2d8ce26
JP
125#define JUMPABLE(rn) ( \
126 OP(rn) == OPEN || OP(rn) == CLOSE || OP(rn) == EVAL || \
cca55fe3
JP
127 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
128 OP(rn) == PLUS || OP(rn) == MINMOD || \
3dab1dad 129 (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
e2d8ce26
JP
130)
131
cca55fe3 132#define HAS_TEXT(rn) ( \
3dab1dad 133 PL_regkind[OP(rn)] == EXACT || PL_regkind[OP(rn)] == REF \
cca55fe3 134)
e2d8ce26 135
a84d97b6
HS
136/*
137 Search for mandatory following text node; for lookahead, the text must
138 follow but for lookbehind (rn->flags != 0) we skip to the next step.
139*/
cca55fe3 140#define FIND_NEXT_IMPT(rn) STMT_START { \
3dab1dad
YO
141 while (JUMPABLE(rn)) { \
142 const OPCODE type = OP(rn); \
143 if (type == SUSPEND || PL_regkind[type] == CURLY) \
e2d8ce26 144 rn = NEXTOPER(NEXTOPER(rn)); \
3dab1dad 145 else if (type == PLUS) \
cca55fe3 146 rn = NEXTOPER(rn); \
3dab1dad 147 else if (type == IFMATCH) \
a84d97b6 148 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
e2d8ce26 149 else rn += NEXT_OFF(rn); \
3dab1dad 150 } \
5f80c4cf 151} STMT_END
74750237 152
acfe0abc 153static void restore_pos(pTHX_ void *arg);
51371543 154
76e3520e 155STATIC CHECKPOINT
cea2e8a9 156S_regcppush(pTHX_ I32 parenfloor)
a0d0e21e 157{
97aff369 158 dVAR;
a3b680e6 159 const int retval = PL_savestack_ix;
b1ce53c5 160#define REGCP_PAREN_ELEMS 4
a3b680e6 161 const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
a0d0e21e 162 int p;
40a82448 163 GET_RE_DEBUG_FLAGS_DECL;
a0d0e21e 164
e49a9654
IH
165 if (paren_elems_to_push < 0)
166 Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
167
a01268b5 168#define REGCP_OTHER_ELEMS 6
4b3c1a47 169 SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
3280af22 170 for (p = PL_regsize; p > parenfloor; p--) {
b1ce53c5 171/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
cf93c79d
IZ
172 SSPUSHINT(PL_regendp[p]);
173 SSPUSHINT(PL_regstartp[p]);
3280af22 174 SSPUSHPTR(PL_reg_start_tmp[p]);
a0d0e21e 175 SSPUSHINT(p);
40a82448
DM
176 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
177 " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n",
178 (UV)p, (IV)PL_regstartp[p],
179 (IV)(PL_reg_start_tmp[p] - PL_bostr),
180 (IV)PL_regendp[p]
181 ));
a0d0e21e 182 }
b1ce53c5 183/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
3280af22
NIS
184 SSPUSHINT(PL_regsize);
185 SSPUSHINT(*PL_reglastparen);
a01268b5 186 SSPUSHINT(*PL_reglastcloseparen);
3280af22 187 SSPUSHPTR(PL_reginput);
41123dfd
JH
188#define REGCP_FRAME_ELEMS 2
189/* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and
190 * are needed for the regexp context stack bookkeeping. */
191 SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
b1ce53c5 192 SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */
41123dfd 193
a0d0e21e
LW
194 return retval;
195}
196
c277df42 197/* These are needed since we do not localize EVAL nodes: */
ab3bbdeb
YO
198#define REGCP_SET(cp) \
199 DEBUG_STATE_r( \
ab3bbdeb 200 PerlIO_printf(Perl_debug_log, \
e4f74956 201 " Setting an EVAL scope, savestack=%"IVdf"\n", \
ab3bbdeb
YO
202 (IV)PL_savestack_ix)); \
203 cp = PL_savestack_ix
c3464db5 204
ab3bbdeb 205#define REGCP_UNWIND(cp) \
e4f74956 206 DEBUG_STATE_r( \
ab3bbdeb 207 if (cp != PL_savestack_ix) \
e4f74956
YO
208 PerlIO_printf(Perl_debug_log, \
209 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
ab3bbdeb
YO
210 (IV)(cp), (IV)PL_savestack_ix)); \
211 regcpblow(cp)
c277df42 212
76e3520e 213STATIC char *
097eb12c 214S_regcppop(pTHX_ const regexp *rex)
a0d0e21e 215{
97aff369 216 dVAR;
b1ce53c5 217 I32 i;
a0d0e21e 218 char *input;
b1ce53c5 219
a3621e74
YO
220 GET_RE_DEBUG_FLAGS_DECL;
221
b1ce53c5 222 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
a0d0e21e 223 i = SSPOPINT;
b1ce53c5
JH
224 assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
225 i = SSPOPINT; /* Parentheses elements to pop. */
a0d0e21e 226 input = (char *) SSPOPPTR;
a01268b5 227 *PL_reglastcloseparen = SSPOPINT;
3280af22
NIS
228 *PL_reglastparen = SSPOPINT;
229 PL_regsize = SSPOPINT;
b1ce53c5
JH
230
231 /* Now restore the parentheses context. */
41123dfd
JH
232 for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
233 i > 0; i -= REGCP_PAREN_ELEMS) {
1df70142 234 I32 tmps;
097eb12c 235 U32 paren = (U32)SSPOPINT;
3280af22 236 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
cf93c79d
IZ
237 PL_regstartp[paren] = SSPOPINT;
238 tmps = SSPOPINT;
3280af22
NIS
239 if (paren <= *PL_reglastparen)
240 PL_regendp[paren] = tmps;
a3621e74 241 DEBUG_EXECUTE_r(
c3464db5 242 PerlIO_printf(Perl_debug_log,
b900a521 243 " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
9041c2e3 244 (UV)paren, (IV)PL_regstartp[paren],
b900a521 245 (IV)(PL_reg_start_tmp[paren] - PL_bostr),
9041c2e3 246 (IV)PL_regendp[paren],
3280af22 247 (paren > *PL_reglastparen ? "(no)" : ""));
c277df42 248 );
a0d0e21e 249 }
a3621e74 250 DEBUG_EXECUTE_r(
bb7a0f54 251 if (*PL_reglastparen + 1 <= rex->nparens) {
c3464db5 252 PerlIO_printf(Perl_debug_log,
faccc32b 253 " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
4f639d21 254 (IV)(*PL_reglastparen + 1), (IV)rex->nparens);
c277df42
IZ
255 }
256 );
daf18116 257#if 1
dafc8851
JH
258 /* It would seem that the similar code in regtry()
259 * already takes care of this, and in fact it is in
260 * a better location to since this code can #if 0-ed out
261 * but the code in regtry() is needed or otherwise tests
262 * requiring null fields (pat.t#187 and split.t#{13,14}
daf18116
JH
263 * (as of patchlevel 7877) will fail. Then again,
264 * this code seems to be necessary or otherwise
265 * building DynaLoader will fail:
266 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
267 * --jhi */
bb7a0f54 268 for (i = *PL_reglastparen + 1; (U32)i <= rex->nparens; i++) {
097eb12c
AL
269 if (i > PL_regsize)
270 PL_regstartp[i] = -1;
271 PL_regendp[i] = -1;
a0d0e21e 272 }
dafc8851 273#endif
a0d0e21e
LW
274 return input;
275}
276
02db2b7b 277#define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
a0d0e21e 278
a687059c 279/*
e50aee73 280 * pregexec and friends
a687059c
LW
281 */
282
76234dfb 283#ifndef PERL_IN_XSUB_RE
a687059c 284/*
c277df42 285 - pregexec - match a regexp against a string
a687059c 286 */
c277df42 287I32
864dbfa3 288Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend,
c3464db5 289 char *strbeg, I32 minend, SV *screamer, U32 nosave)
c277df42
IZ
290/* strend: pointer to null at end of string */
291/* strbeg: real beginning of string */
292/* minend: end of match must be >=minend after stringarg. */
293/* nosave: For optimizations. */
294{
295 return
9041c2e3 296 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
c277df42
IZ
297 nosave ? 0 : REXEC_COPY_STR);
298}
76234dfb 299#endif
22e551b9 300
9041c2e3 301/*
cad2e5aa
JH
302 * Need to implement the following flags for reg_anch:
303 *
304 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
305 * USE_INTUIT_ML
306 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
307 * INTUIT_AUTORITATIVE_ML
308 * INTUIT_ONCE_NOML - Intuit can match in one location only.
309 * INTUIT_ONCE_ML
310 *
311 * Another flag for this function: SECOND_TIME (so that float substrs
312 * with giant delta may be not rechecked).
313 */
314
315/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
316
3f7c398e 317/* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend.
cad2e5aa
JH
318 Otherwise, only SvCUR(sv) is used to get strbeg. */
319
320/* XXXX We assume that strpos is strbeg unless sv. */
321
6eb5f6b9
JH
322/* XXXX Some places assume that there is a fixed substring.
323 An update may be needed if optimizer marks as "INTUITable"
324 RExen without fixed substrings. Similarly, it is assumed that
325 lengths of all the strings are no more than minlen, thus they
326 cannot come from lookahead.
327 (Or minlen should take into account lookahead.) */
328
2c2d71f5
JH
329/* A failure to find a constant substring means that there is no need to make
330 an expensive call to REx engine, thus we celebrate a failure. Similarly,
331 finding a substring too deep into the string means that less calls to
30944b6d
IZ
332 regtry() should be needed.
333
334 REx compiler's optimizer found 4 possible hints:
335 a) Anchored substring;
336 b) Fixed substring;
337 c) Whether we are anchored (beginning-of-line or \G);
338 d) First node (of those at offset 0) which may distingush positions;
6eb5f6b9 339 We use a)b)d) and multiline-part of c), and try to find a position in the
30944b6d
IZ
340 string which does not contradict any of them.
341 */
2c2d71f5 342
6eb5f6b9
JH
343/* Most of decisions we do here should have been done at compile time.
344 The nodes of the REx which we used for the search should have been
345 deleted from the finite automaton. */
346
cad2e5aa
JH
347char *
348Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
349 char *strend, U32 flags, re_scream_pos_data *data)
350{
97aff369 351 dVAR;
b7953727 352 register I32 start_shift = 0;
cad2e5aa 353 /* Should be nonnegative! */
b7953727 354 register I32 end_shift = 0;
2c2d71f5
JH
355 register char *s;
356 register SV *check;
a1933d95 357 char *strbeg;
cad2e5aa 358 char *t;
1de06328 359 const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
cad2e5aa 360 I32 ml_anch;
bd61b366
SS
361 register char *other_last = NULL; /* other substr checked before this */
362 char *check_at = NULL; /* check substr found at this pos */
1df70142 363 const I32 multiline = prog->reganch & PMf_MULTILINE;
30944b6d 364#ifdef DEBUGGING
890ce7af 365 const char * const i_strpos = strpos;
30944b6d 366#endif
a3621e74
YO
367
368 GET_RE_DEBUG_FLAGS_DECL;
369
a30b2f1f 370 RX_MATCH_UTF8_set(prog,do_utf8);
cad2e5aa 371
b8d68ded 372 if (prog->reganch & ROPT_UTF8) {
b8d68ded
JH
373 PL_reg_flags |= RF_utf8;
374 }
ab3bbdeb
YO
375 DEBUG_EXECUTE_r(
376 debug_start_match(prog, do_utf8, strpos, strend,
1de06328
YO
377 sv ? "Guessing start of match in sv for"
378 : "Guessing start of match in string for");
2a782b5b 379 );
cad2e5aa 380
c344f387
JH
381 /* CHR_DIST() would be more correct here but it makes things slow. */
382 if (prog->minlen > strend - strpos) {
a3621e74 383 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
a72c7584 384 "String too short... [re_intuit_start]\n"));
cad2e5aa 385 goto fail;
2c2d71f5 386 }
1de06328 387
a1933d95 388 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
1aa99e6b 389 PL_regeol = strend;
33b8afdf
JH
390 if (do_utf8) {
391 if (!prog->check_utf8 && prog->check_substr)
392 to_utf8_substr(prog);
393 check = prog->check_utf8;
394 } else {
395 if (!prog->check_substr && prog->check_utf8)
396 to_byte_substr(prog);
397 check = prog->check_substr;
398 }
1de06328 399 if (check == &PL_sv_undef) {
a3621e74 400 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1de06328 401 "Non-utf8 string cannot match utf8 check string\n"));
33b8afdf
JH
402 goto fail;
403 }
2c2d71f5 404 if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */
cad2e5aa
JH
405 ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE)
406 || ( (prog->reganch & ROPT_ANCH_BOL)
7fba1cd6 407 && !multiline ) ); /* Check after \n? */
cad2e5aa 408
7e25d62c
JH
409 if (!ml_anch) {
410 if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */
411 | ROPT_IMPLICIT)) /* not a real BOL */
3f7c398e 412 /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
7e25d62c
JH
413 && sv && !SvROK(sv)
414 && (strpos != strbeg)) {
a3621e74 415 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
7e25d62c
JH
416 goto fail;
417 }
418 if (prog->check_offset_min == prog->check_offset_max &&
cce850e4 419 !(prog->reganch & ROPT_CANY_SEEN)) {
2c2d71f5 420 /* Substring at constant offset from beg-of-str... */
cad2e5aa
JH
421 I32 slen;
422
1aa99e6b 423 s = HOP3c(strpos, prog->check_offset_min, strend);
1de06328 424
653099ff
GS
425 if (SvTAIL(check)) {
426 slen = SvCUR(check); /* >= 1 */
cad2e5aa 427
9041c2e3 428 if ( strend - s > slen || strend - s < slen - 1
2c2d71f5 429 || (strend - s == slen && strend[-1] != '\n')) {
a3621e74 430 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
2c2d71f5 431 goto fail_finish;
cad2e5aa
JH
432 }
433 /* Now should match s[0..slen-2] */
434 slen--;
3f7c398e 435 if (slen && (*SvPVX_const(check) != *s
cad2e5aa 436 || (slen > 1
3f7c398e 437 && memNE(SvPVX_const(check), s, slen)))) {
2c2d71f5 438 report_neq:
a3621e74 439 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
2c2d71f5
JH
440 goto fail_finish;
441 }
cad2e5aa 442 }
3f7c398e 443 else if (*SvPVX_const(check) != *s
653099ff 444 || ((slen = SvCUR(check)) > 1
3f7c398e 445 && memNE(SvPVX_const(check), s, slen)))
2c2d71f5 446 goto report_neq;
c315bfe8 447 check_at = s;
2c2d71f5 448 goto success_at_start;
7e25d62c 449 }
cad2e5aa 450 }
2c2d71f5 451 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
cad2e5aa 452 s = strpos;
2c2d71f5 453 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
1de06328
YO
454 end_shift = prog->check_end_shift;
455
2c2d71f5 456 if (!ml_anch) {
a3b680e6 457 const I32 end = prog->check_offset_max + CHR_SVLEN(check)
653099ff 458 - (SvTAIL(check) != 0);
a3b680e6 459 const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
2c2d71f5
JH
460
461 if (end_shift < eshift)
462 end_shift = eshift;
463 }
cad2e5aa 464 }
2c2d71f5 465 else { /* Can match at random position */
cad2e5aa
JH
466 ml_anch = 0;
467 s = strpos;
1de06328
YO
468 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
469 end_shift = prog->check_end_shift;
470
471 /* end shift should be non negative here */
cad2e5aa
JH
472 }
473
2c2d71f5 474#ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */
0033605d 475 if (end_shift < 0)
1de06328
YO
476 Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
477 end_shift,prog->precomp);
2c2d71f5
JH
478#endif
479
2c2d71f5
JH
480 restart:
481 /* Find a possible match in the region s..strend by looking for
482 the "check" substring in the region corrected by start/end_shift. */
1de06328
YO
483
484 {
485 I32 srch_start_shift = start_shift;
486 I32 srch_end_shift = end_shift;
487 if (srch_start_shift < 0 && strbeg - s > srch_start_shift) {
488 srch_end_shift -= ((strbeg - s) - srch_start_shift);
489 srch_start_shift = strbeg - s;
490 }
491 DEBUG_OPTIMISE_r({
492 PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n",
493 (IV)prog->check_offset_min,
494 (IV)srch_start_shift,
495 (IV)srch_end_shift,
496 (IV)prog->check_end_shift);
497 });
498
cad2e5aa 499 if (flags & REXEC_SCREAM) {
cad2e5aa 500 I32 p = -1; /* Internal iterator of scream. */
a3b680e6 501 I32 * const pp = data ? data->scream_pos : &p;
cad2e5aa 502
2c2d71f5
JH
503 if (PL_screamfirst[BmRARE(check)] >= 0
504 || ( BmRARE(check) == '\n'
505 && (BmPREVIOUS(check) == SvCUR(check) - 1)
506 && SvTAIL(check) ))
9041c2e3 507 s = screaminstr(sv, check,
1de06328 508 srch_start_shift + (s - strbeg), srch_end_shift, pp, 0);
cad2e5aa 509 else
2c2d71f5 510 goto fail_finish;
4addbd3b
HS
511 /* we may be pointing at the wrong string */
512 if (s && RX_MATCH_COPIED(prog))
3f7c398e 513 s = strbeg + (s - SvPVX_const(sv));
cad2e5aa
JH
514 if (data)
515 *data->scream_olds = s;
516 }
1de06328
YO
517 else {
518 U8* start_point;
519 U8* end_point;
520 if (prog->reganch & ROPT_CANY_SEEN) {
521 start_point= (U8*)(s + srch_start_shift);
522 end_point= (U8*)(strend - srch_end_shift);
523 } else {
524 start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend);
525 end_point= HOP3(strend, -srch_end_shift, strbeg);
526 }
527 DEBUG_OPTIMISE_r({
56570a2c 528 PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n",
1de06328 529 (int)(end_point - start_point),
fc8cd66c 530 (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point),
1de06328
YO
531 start_point);
532 });
533
534 s = fbm_instr( start_point, end_point,
7fba1cd6 535 check, multiline ? FBMrf_MULTILINE : 0);
1de06328
YO
536 }
537 }
cad2e5aa
JH
538 /* Update the count-of-usability, remove useless subpatterns,
539 unshift s. */
2c2d71f5 540
ab3bbdeb
YO
541 DEBUG_EXECUTE_r({
542 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
543 SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
544 PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s",
2c2d71f5 545 (s ? "Found" : "Did not find"),
ab3bbdeb
YO
546 (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)
547 ? "anchored" : "floating"),
548 quoted,
549 RE_SV_TAIL(check),
550 (s ? " at offset " : "...\n") );
551 });
2c2d71f5
JH
552
553 if (!s)
554 goto fail_finish;
2c2d71f5 555 /* Finish the diagnostic message */
a3621e74 556 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
2c2d71f5 557
1de06328
YO
558 /* XXX dmq: first branch is for positive lookbehind...
559 Our check string is offset from the beginning of the pattern.
560 So we need to do any stclass tests offset forward from that
561 point. I think. :-(
562 */
563
564
565
566 check_at=s;
567
568
2c2d71f5
JH
569 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
570 Start with the other substr.
571 XXXX no SCREAM optimization yet - and a very coarse implementation
a0288114 572 XXXX /ttx+/ results in anchored="ttx", floating="x". floating will
2c2d71f5
JH
573 *always* match. Probably should be marked during compile...
574 Probably it is right to do no SCREAM here...
575 */
576
1de06328
YO
577 if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8)
578 : (prog->float_substr && prog->anchored_substr))
579 {
30944b6d 580 /* Take into account the "other" substring. */
2c2d71f5
JH
581 /* XXXX May be hopelessly wrong for UTF... */
582 if (!other_last)
6eb5f6b9 583 other_last = strpos;
33b8afdf 584 if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) {
30944b6d
IZ
585 do_other_anchored:
586 {
890ce7af
AL
587 char * const last = HOP3c(s, -start_shift, strbeg);
588 char *last1, *last2;
be8e71aa 589 char * const saved_s = s;
33b8afdf 590 SV* must;
2c2d71f5 591
2c2d71f5
JH
592 t = s - prog->check_offset_max;
593 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
1d86a7f9 594 && (!do_utf8
0ce71af7 595 || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos))
2c2d71f5 596 && t > strpos)))
6f207bd3 597 NOOP;
2c2d71f5
JH
598 else
599 t = strpos;
1aa99e6b 600 t = HOP3c(t, prog->anchored_offset, strend);
6eb5f6b9
JH
601 if (t < other_last) /* These positions already checked */
602 t = other_last;
1aa99e6b 603 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
2c2d71f5
JH
604 if (last < last1)
605 last1 = last;
1de06328
YO
606 /* XXXX It is not documented what units *_offsets are in.
607 We assume bytes, but this is clearly wrong.
608 Meaning this code needs to be carefully reviewed for errors.
609 dmq.
610 */
611
2c2d71f5 612 /* On end-of-str: see comment below. */
33b8afdf
JH
613 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
614 if (must == &PL_sv_undef) {
615 s = (char*)NULL;
1de06328 616 DEBUG_r(must = prog->anchored_utf8); /* for debug */
33b8afdf
JH
617 }
618 else
619 s = fbm_instr(
620 (unsigned char*)t,
621 HOP3(HOP3(last1, prog->anchored_offset, strend)
622 + SvCUR(must), -(SvTAIL(must)!=0), strbeg),
623 must,
7fba1cd6 624 multiline ? FBMrf_MULTILINE : 0
33b8afdf 625 );
ab3bbdeb
YO
626 DEBUG_EXECUTE_r({
627 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
628 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
629 PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s",
2c2d71f5 630 (s ? "Found" : "Contradicts"),
ab3bbdeb
YO
631 quoted, RE_SV_TAIL(must));
632 });
633
634
2c2d71f5
JH
635 if (!s) {
636 if (last1 >= last2) {
a3621e74 637 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2c2d71f5
JH
638 ", giving up...\n"));
639 goto fail_finish;
640 }
a3621e74 641 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2c2d71f5 642 ", trying floating at offset %ld...\n",
be8e71aa 643 (long)(HOP3c(saved_s, 1, strend) - i_strpos)));
1aa99e6b
IH
644 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
645 s = HOP3c(last, 1, strend);
2c2d71f5
JH
646 goto restart;
647 }
648 else {
a3621e74 649 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
30944b6d 650 (long)(s - i_strpos)));
1aa99e6b
IH
651 t = HOP3c(s, -prog->anchored_offset, strbeg);
652 other_last = HOP3c(s, 1, strend);
be8e71aa 653 s = saved_s;
2c2d71f5
JH
654 if (t == strpos)
655 goto try_at_start;
2c2d71f5
JH
656 goto try_at_offset;
657 }
30944b6d 658 }
2c2d71f5
JH
659 }
660 else { /* Take into account the floating substring. */
33b8afdf 661 char *last, *last1;
be8e71aa 662 char * const saved_s = s;
33b8afdf
JH
663 SV* must;
664
665 t = HOP3c(s, -start_shift, strbeg);
666 last1 = last =
667 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
668 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
669 last = HOP3c(t, prog->float_max_offset, strend);
670 s = HOP3c(t, prog->float_min_offset, strend);
671 if (s < other_last)
672 s = other_last;
2c2d71f5 673 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
33b8afdf
JH
674 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
675 /* fbm_instr() takes into account exact value of end-of-str
676 if the check is SvTAIL(ed). Since false positives are OK,
677 and end-of-str is not later than strend we are OK. */
678 if (must == &PL_sv_undef) {
679 s = (char*)NULL;
1de06328 680 DEBUG_r(must = prog->float_utf8); /* for debug message */
33b8afdf
JH
681 }
682 else
2c2d71f5 683 s = fbm_instr((unsigned char*)s,
33b8afdf
JH
684 (unsigned char*)last + SvCUR(must)
685 - (SvTAIL(must)!=0),
7fba1cd6 686 must, multiline ? FBMrf_MULTILINE : 0);
ab3bbdeb
YO
687 DEBUG_EXECUTE_r({
688 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
689 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
690 PerlIO_printf(Perl_debug_log, "%s floating substr %s%s",
33b8afdf 691 (s ? "Found" : "Contradicts"),
ab3bbdeb
YO
692 quoted, RE_SV_TAIL(must));
693 });
33b8afdf
JH
694 if (!s) {
695 if (last1 == last) {
a3621e74 696 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf
JH
697 ", giving up...\n"));
698 goto fail_finish;
2c2d71f5 699 }
a3621e74 700 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf 701 ", trying anchored starting at offset %ld...\n",
be8e71aa 702 (long)(saved_s + 1 - i_strpos)));
33b8afdf
JH
703 other_last = last;
704 s = HOP3c(t, 1, strend);
705 goto restart;
706 }
707 else {
a3621e74 708 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
33b8afdf
JH
709 (long)(s - i_strpos)));
710 other_last = s; /* Fix this later. --Hugo */
be8e71aa 711 s = saved_s;
33b8afdf
JH
712 if (t == strpos)
713 goto try_at_start;
714 goto try_at_offset;
715 }
2c2d71f5 716 }
cad2e5aa 717 }
2c2d71f5 718
1de06328 719
9ef43ace 720 t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos);
1de06328
YO
721
722 DEBUG_OPTIMISE_r(
723 PerlIO_printf(Perl_debug_log,
724 "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n",
725 (IV)prog->check_offset_min,
726 (IV)prog->check_offset_max,
727 (IV)(s-strpos),
728 (IV)(t-strpos),
729 (IV)(t-s),
730 (IV)(strend-strpos)
731 )
732 );
733
2c2d71f5 734 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
1d86a7f9 735 && (!do_utf8
9ef43ace 736 || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos)))
1de06328
YO
737 && t > strpos)))
738 {
2c2d71f5
JH
739 /* Fixed substring is found far enough so that the match
740 cannot start at strpos. */
741 try_at_offset:
cad2e5aa 742 if (ml_anch && t[-1] != '\n') {
30944b6d
IZ
743 /* Eventually fbm_*() should handle this, but often
744 anchored_offset is not 0, so this check will not be wasted. */
745 /* XXXX In the code below we prefer to look for "^" even in
746 presence of anchored substrings. And we search even
747 beyond the found float position. These pessimizations
748 are historical artefacts only. */
749 find_anchor:
2c2d71f5 750 while (t < strend - prog->minlen) {
cad2e5aa 751 if (*t == '\n') {
4ee3650e 752 if (t < check_at - prog->check_offset_min) {
33b8afdf 753 if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) {
4ee3650e
GS
754 /* Since we moved from the found position,
755 we definitely contradict the found anchored
30944b6d
IZ
756 substr. Due to the above check we do not
757 contradict "check" substr.
758 Thus we can arrive here only if check substr
759 is float. Redo checking for "other"=="fixed".
760 */
9041c2e3 761 strpos = t + 1;
a3621e74 762 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
e4584336 763 PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
30944b6d
IZ
764 goto do_other_anchored;
765 }
4ee3650e
GS
766 /* We don't contradict the found floating substring. */
767 /* XXXX Why not check for STCLASS? */
cad2e5aa 768 s = t + 1;
a3621e74 769 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
e4584336 770 PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
cad2e5aa
JH
771 goto set_useful;
772 }
4ee3650e
GS
773 /* Position contradicts check-string */
774 /* XXXX probably better to look for check-string
775 than for "\n", so one should lower the limit for t? */
a3621e74 776 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
e4584336 777 PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
0e41cd87 778 other_last = strpos = s = t + 1;
cad2e5aa
JH
779 goto restart;
780 }
781 t++;
782 }
a3621e74 783 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
e4584336 784 PL_colors[0], PL_colors[1]));
2c2d71f5 785 goto fail_finish;
cad2e5aa 786 }
f5952150 787 else {
a3621e74 788 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
e4584336 789 PL_colors[0], PL_colors[1]));
f5952150 790 }
cad2e5aa
JH
791 s = t;
792 set_useful:
33b8afdf 793 ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
cad2e5aa
JH
794 }
795 else {
f5952150 796 /* The found string does not prohibit matching at strpos,
2c2d71f5 797 - no optimization of calling REx engine can be performed,
f5952150
GS
798 unless it was an MBOL and we are not after MBOL,
799 or a future STCLASS check will fail this. */
2c2d71f5
JH
800 try_at_start:
801 /* Even in this situation we may use MBOL flag if strpos is offset
802 wrt the start of the string. */
05b4157f 803 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
a1933d95 804 && (strpos != strbeg) && strpos[-1] != '\n'
d506a20d
IZ
805 /* May be due to an implicit anchor of m{.*foo} */
806 && !(prog->reganch & ROPT_IMPLICIT))
807 {
cad2e5aa
JH
808 t = strpos;
809 goto find_anchor;
810 }
a3621e74 811 DEBUG_EXECUTE_r( if (ml_anch)
f5952150 812 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
e4584336 813 (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
30944b6d 814 );
2c2d71f5 815 success_at_start:
30944b6d 816 if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */
33b8afdf
JH
817 && (do_utf8 ? (
818 prog->check_utf8 /* Could be deleted already */
819 && --BmUSEFUL(prog->check_utf8) < 0
820 && (prog->check_utf8 == prog->float_utf8)
821 ) : (
822 prog->check_substr /* Could be deleted already */
823 && --BmUSEFUL(prog->check_substr) < 0
824 && (prog->check_substr == prog->float_substr)
825 )))
66e933ab 826 {
cad2e5aa 827 /* If flags & SOMETHING - do not do it many times on the same match */
a3621e74 828 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
33b8afdf
JH
829 SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr);
830 if (do_utf8 ? prog->check_substr : prog->check_utf8)
831 SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8);
a0714e2c
SS
832 prog->check_substr = prog->check_utf8 = NULL; /* disable */
833 prog->float_substr = prog->float_utf8 = NULL; /* clear */
834 check = NULL; /* abort */
cad2e5aa 835 s = strpos;
3cf5c195
IZ
836 /* XXXX This is a remnant of the old implementation. It
837 looks wasteful, since now INTUIT can use many
6eb5f6b9 838 other heuristics. */
cad2e5aa
JH
839 prog->reganch &= ~RE_USE_INTUIT;
840 }
841 else
842 s = strpos;
843 }
844
6eb5f6b9
JH
845 /* Last resort... */
846 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
1de06328
YO
847 /* trie stclasses are too expensive to use here, we are better off to
848 leave it to regmatch itself */
786e8c11 849 if (prog->regstclass && PL_regkind[OP(prog->regstclass)]!=TRIE) {
6eb5f6b9
JH
850 /* minlen == 0 is possible if regstclass is \b or \B,
851 and the fixed substr is ''$.
852 Since minlen is already taken into account, s+1 is before strend;
853 accidentally, minlen >= 1 guaranties no false positives at s + 1
854 even for \b or \B. But (minlen? 1 : 0) below assumes that
855 regstclass does not come from lookahead... */
856 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
857 This leaves EXACTF only, which is dealt with in find_byclass(). */
890ce7af 858 const U8* const str = (U8*)STRING(prog->regstclass);
3dab1dad 859 const int cl_l = (PL_regkind[OP(prog->regstclass)] == EXACT
1aa99e6b 860 ? CHR_DIST(str+STR_LEN(prog->regstclass), str)
66e933ab 861 : 1);
1de06328
YO
862 char * endpos;
863 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
864 endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend);
865 else if (prog->float_substr || prog->float_utf8)
866 endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend);
867 else
868 endpos= strend;
869
56570a2c 870 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %d s: %d endpos: %d\n",
1de06328
YO
871 start_shift,check_at-strbeg,s-strbeg,endpos-strbeg));
872
6eb5f6b9 873 t = s;
3b0527fe 874 s = find_byclass(prog, prog->regstclass, s, endpos, NULL);
6eb5f6b9
JH
875 if (!s) {
876#ifdef DEBUGGING
cbbf8932 877 const char *what = NULL;
6eb5f6b9
JH
878#endif
879 if (endpos == strend) {
a3621e74 880 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6eb5f6b9
JH
881 "Could not match STCLASS...\n") );
882 goto fail;
883 }
a3621e74 884 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab 885 "This position contradicts STCLASS...\n") );
653099ff
GS
886 if ((prog->reganch & ROPT_ANCH) && !ml_anch)
887 goto fail;
6eb5f6b9 888 /* Contradict one of substrings */
33b8afdf
JH
889 if (prog->anchored_substr || prog->anchored_utf8) {
890 if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) {
a3621e74 891 DEBUG_EXECUTE_r( what = "anchored" );
6eb5f6b9 892 hop_and_restart:
1aa99e6b 893 s = HOP3c(t, 1, strend);
66e933ab
GS
894 if (s + start_shift + end_shift > strend) {
895 /* XXXX Should be taken into account earlier? */
a3621e74 896 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab
GS
897 "Could not match STCLASS...\n") );
898 goto fail;
899 }
5e39e1e5
HS
900 if (!check)
901 goto giveup;
a3621e74 902 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 903 "Looking for %s substr starting at offset %ld...\n",
6eb5f6b9
JH
904 what, (long)(s + start_shift - i_strpos)) );
905 goto restart;
906 }
66e933ab 907 /* Have both, check_string is floating */
6eb5f6b9
JH
908 if (t + start_shift >= check_at) /* Contradicts floating=check */
909 goto retry_floating_check;
910 /* Recheck anchored substring, but not floating... */
9041c2e3 911 s = check_at;
5e39e1e5
HS
912 if (!check)
913 goto giveup;
a3621e74 914 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 915 "Looking for anchored substr starting at offset %ld...\n",
6eb5f6b9
JH
916 (long)(other_last - i_strpos)) );
917 goto do_other_anchored;
918 }
60e71179
GS
919 /* Another way we could have checked stclass at the
920 current position only: */
921 if (ml_anch) {
922 s = t = t + 1;
5e39e1e5
HS
923 if (!check)
924 goto giveup;
a3621e74 925 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 926 "Looking for /%s^%s/m starting at offset %ld...\n",
e4584336 927 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
60e71179 928 goto try_at_offset;
66e933ab 929 }
33b8afdf 930 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
60e71179 931 goto fail;
6eb5f6b9
JH
932 /* Check is floating subtring. */
933 retry_floating_check:
934 t = check_at - start_shift;
a3621e74 935 DEBUG_EXECUTE_r( what = "floating" );
6eb5f6b9
JH
936 goto hop_and_restart;
937 }
b7953727 938 if (t != s) {
a3621e74 939 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
6eb5f6b9 940 "By STCLASS: moving %ld --> %ld\n",
b7953727
JH
941 (long)(t - i_strpos), (long)(s - i_strpos))
942 );
943 }
944 else {
a3621e74 945 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
b7953727
JH
946 "Does not contradict STCLASS...\n");
947 );
948 }
6eb5f6b9 949 }
5e39e1e5 950 giveup:
a3621e74 951 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
5e39e1e5
HS
952 PL_colors[4], (check ? "Guessed" : "Giving up"),
953 PL_colors[5], (long)(s - i_strpos)) );
cad2e5aa 954 return s;
2c2d71f5
JH
955
956 fail_finish: /* Substring not found */
33b8afdf
JH
957 if (prog->check_substr || prog->check_utf8) /* could be removed already */
958 BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
cad2e5aa 959 fail:
a3621e74 960 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
e4584336 961 PL_colors[4], PL_colors[5]));
bd61b366 962 return NULL;
cad2e5aa 963}
9661b544 964
786e8c11 965
3b0527fe 966
4cadc6a9
YO
967#define REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, uvc, charid, \
968foldlen, foldbuf, uniflags) STMT_START { \
969 switch (trie_type) { \
970 case trie_utf8_fold: \
971 if ( foldlen>0 ) { \
972 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
973 foldlen -= len; \
974 uscan += len; \
975 len=0; \
976 } else { \
977 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
978 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
979 foldlen -= UNISKIP( uvc ); \
980 uscan = foldbuf + UNISKIP( uvc ); \
981 } \
982 break; \
983 case trie_utf8: \
984 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
985 break; \
986 case trie_plain: \
987 uvc = (UV)*uc; \
988 len = 1; \
989 } \
990 \
991 if (uvc < 256) { \
992 charid = trie->charmap[ uvc ]; \
993 } \
994 else { \
995 charid = 0; \
996 if (trie->widecharmap) { \
997 SV** const svpp = hv_fetch(trie->widecharmap, \
998 (char*)&uvc, sizeof(UV), 0); \
999 if (svpp) \
1000 charid = (U16)SvIV(*svpp); \
1001 } \
1002 } \
1003} STMT_END
1004
1005#define REXEC_FBC_EXACTISH_CHECK(CoNd) \
1006 if ( (CoNd) \
1007 && (ln == len || \
1008 ibcmp_utf8(s, NULL, 0, do_utf8, \
1009 m, NULL, ln, (bool)UTF)) \
1010 && (!reginfo || regtry(reginfo, s)) ) \
1011 goto got_it; \
1012 else { \
1013 U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \
1014 uvchr_to_utf8(tmpbuf, c); \
1015 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \
1016 if ( f != c \
1017 && (f == c1 || f == c2) \
1018 && (ln == foldlen || \
1019 !ibcmp_utf8((char *) foldbuf, \
1020 NULL, foldlen, do_utf8, \
1021 m, \
1022 NULL, ln, (bool)UTF)) \
1023 && (!reginfo || regtry(reginfo, s)) ) \
1024 goto got_it; \
1025 } \
1026 s += len
1027
1028#define REXEC_FBC_EXACTISH_SCAN(CoNd) \
1029STMT_START { \
1030 while (s <= e) { \
1031 if ( (CoNd) \
1032 && (ln == 1 || !(OP(c) == EXACTF \
1033 ? ibcmp(s, m, ln) \
1034 : ibcmp_locale(s, m, ln))) \
1035 && (!reginfo || regtry(reginfo, s)) ) \
1036 goto got_it; \
1037 s++; \
1038 } \
1039} STMT_END
1040
1041#define REXEC_FBC_UTF8_SCAN(CoDe) \
1042STMT_START { \
1043 while (s + (uskip = UTF8SKIP(s)) <= strend) { \
1044 CoDe \
1045 s += uskip; \
1046 } \
1047} STMT_END
1048
1049#define REXEC_FBC_SCAN(CoDe) \
1050STMT_START { \
1051 while (s < strend) { \
1052 CoDe \
1053 s++; \
1054 } \
1055} STMT_END
1056
1057#define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \
1058REXEC_FBC_UTF8_SCAN( \
1059 if (CoNd) { \
1060 if (tmp && (!reginfo || regtry(reginfo, s))) \
1061 goto got_it; \
1062 else \
1063 tmp = doevery; \
1064 } \
1065 else \
1066 tmp = 1; \
1067)
1068
1069#define REXEC_FBC_CLASS_SCAN(CoNd) \
1070REXEC_FBC_SCAN( \
1071 if (CoNd) { \
1072 if (tmp && (!reginfo || regtry(reginfo, s))) \
1073 goto got_it; \
1074 else \
1075 tmp = doevery; \
1076 } \
1077 else \
1078 tmp = 1; \
1079)
1080
1081#define REXEC_FBC_TRYIT \
1082if ((!reginfo || regtry(reginfo, s))) \
1083 goto got_it
1084
1085#define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
1086 if (do_utf8) { \
1087 UtFpReLoAd; \
1088 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1089 } \
1090 else { \
1091 REXEC_FBC_CLASS_SCAN(CoNd); \
1092 } \
1093 break
1094
1095#define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \
1096 PL_reg_flags |= RF_tainted; \
1097 if (do_utf8) { \
1098 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1099 } \
1100 else { \
1101 REXEC_FBC_CLASS_SCAN(CoNd); \
1102 } \
1103 break
1104
786e8c11
YO
1105#define DUMP_EXEC_POS(li,s,doutf8) \
1106 dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
1107
1108/* We know what class REx starts with. Try to find this position... */
1109/* if reginfo is NULL, its a dryrun */
1110/* annoyingly all the vars in this routine have different names from their counterparts
1111 in regmatch. /grrr */
1112
3c3eec57 1113STATIC char *
07be1b83
YO
1114S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
1115 const char *strend, const regmatch_info *reginfo)
a687059c 1116{
27da23d5 1117 dVAR;
1df70142 1118 const I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
6eb5f6b9 1119 char *m;
d8093b23 1120 STRLEN ln;
5dab1207 1121 STRLEN lnc;
078c425b 1122 register STRLEN uskip;
d8093b23 1123 unsigned int c1;
1124 unsigned int c2;
6eb5f6b9
JH
1125 char *e;
1126 register I32 tmp = 1; /* Scratch variable? */
a3b680e6 1127 register const bool do_utf8 = PL_reg_match_utf8;
cad2e5aa 1128
6eb5f6b9
JH
1129 /* We know what class it must start with. */
1130 switch (OP(c)) {
6eb5f6b9 1131 case ANYOF:
388cc4de 1132 if (do_utf8) {
4cadc6a9 1133 REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
388cc4de 1134 !UTF8_IS_INVARIANT((U8)s[0]) ?
32fc9b6a 1135 reginclass(prog, c, (U8*)s, 0, do_utf8) :
4cadc6a9 1136 REGINCLASS(prog, c, (U8*)s));
388cc4de
HS
1137 }
1138 else {
1139 while (s < strend) {
1140 STRLEN skip = 1;
1141
32fc9b6a 1142 if (REGINCLASS(prog, c, (U8*)s) ||
388cc4de
HS
1143 (ANYOF_FOLD_SHARP_S(c, s, strend) &&
1144 /* The assignment of 2 is intentional:
1145 * for the folded sharp s, the skip is 2. */
1146 (skip = SHARP_S_SKIP))) {
3b0527fe 1147 if (tmp && (!reginfo || regtry(reginfo, s)))
388cc4de
HS
1148 goto got_it;
1149 else
1150 tmp = doevery;
1151 }
1152 else
1153 tmp = 1;
1154 s += skip;
1155 }
a0d0e21e 1156 }
6eb5f6b9 1157 break;
f33976b4 1158 case CANY:
4cadc6a9 1159 REXEC_FBC_SCAN(
3b0527fe 1160 if (tmp && (!reginfo || regtry(reginfo, s)))
f33976b4
DB
1161 goto got_it;
1162 else
1163 tmp = doevery;
4cadc6a9 1164 );
f33976b4 1165 break;
6eb5f6b9 1166 case EXACTF:
5dab1207
NIS
1167 m = STRING(c);
1168 ln = STR_LEN(c); /* length to match in octets/bytes */
1169 lnc = (I32) ln; /* length to match in characters */
1aa99e6b 1170 if (UTF) {
a2a2844f 1171 STRLEN ulen1, ulen2;
5dab1207 1172 U8 *sm = (U8 *) m;
89ebb4a3
JH
1173 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
1174 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
4ad0818d 1175 const U32 uniflags = UTF8_ALLOW_DEFAULT;
a2a2844f
JH
1176
1177 to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1178 to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1179
89ebb4a3 1180 c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
041457d9 1181 0, uniflags);
89ebb4a3 1182 c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
041457d9 1183 0, uniflags);
5dab1207
NIS
1184 lnc = 0;
1185 while (sm < ((U8 *) m + ln)) {
1186 lnc++;
1187 sm += UTF8SKIP(sm);
1188 }
1aa99e6b
IH
1189 }
1190 else {
1191 c1 = *(U8*)m;
1192 c2 = PL_fold[c1];
1193 }
6eb5f6b9
JH
1194 goto do_exactf;
1195 case EXACTFL:
5dab1207
NIS
1196 m = STRING(c);
1197 ln = STR_LEN(c);
1198 lnc = (I32) ln;
d8093b23 1199 c1 = *(U8*)m;
6eb5f6b9
JH
1200 c2 = PL_fold_locale[c1];
1201 do_exactf:
db12adc6 1202 e = HOP3c(strend, -((I32)lnc), s);
b3c9acc1 1203
3b0527fe 1204 if (!reginfo && e < s)
6eb5f6b9 1205 e = s; /* Due to minlen logic of intuit() */
1aa99e6b 1206
60a8b682
JH
1207 /* The idea in the EXACTF* cases is to first find the
1208 * first character of the EXACTF* node and then, if
1209 * necessary, case-insensitively compare the full
1210 * text of the node. The c1 and c2 are the first
1211 * characters (though in Unicode it gets a bit
1212 * more complicated because there are more cases
7f16dd3d
JH
1213 * than just upper and lower: one needs to use
1214 * the so-called folding case for case-insensitive
1215 * matching (called "loose matching" in Unicode).
1216 * ibcmp_utf8() will do just that. */
60a8b682 1217
1aa99e6b 1218 if (do_utf8) {
575cac57 1219 UV c, f;
89ebb4a3 1220 U8 tmpbuf [UTF8_MAXBYTES+1];
575cac57 1221 STRLEN len, foldlen;
4ad0818d 1222 const U32 uniflags = UTF8_ALLOW_DEFAULT;
09091399 1223 if (c1 == c2) {
5dab1207
NIS
1224 /* Upper and lower of 1st char are equal -
1225 * probably not a "letter". */
1aa99e6b 1226 while (s <= e) {
89ebb4a3 1227 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
041457d9 1228 uniflags);
4cadc6a9 1229 REXEC_FBC_EXACTISH_CHECK(c == c1);
1aa99e6b 1230 }
09091399
JH
1231 }
1232 else {
1aa99e6b 1233 while (s <= e) {
89ebb4a3 1234 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
041457d9 1235 uniflags);
80aecb99 1236
60a8b682 1237 /* Handle some of the three Greek sigmas cases.
8c01da3c
JH
1238 * Note that not all the possible combinations
1239 * are handled here: some of them are handled
1240 * by the standard folding rules, and some of
1241 * them (the character class or ANYOF cases)
1242 * are handled during compiletime in
1243 * regexec.c:S_regclass(). */
880bd946
JH
1244 if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
1245 c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
1246 c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
80aecb99 1247
4cadc6a9 1248 REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2);
1aa99e6b 1249 }
09091399 1250 }
1aa99e6b
IH
1251 }
1252 else {
1253 if (c1 == c2)
4cadc6a9 1254 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
1aa99e6b 1255 else
4cadc6a9 1256 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
b3c9acc1
IZ
1257 }
1258 break;
bbce6d69 1259 case BOUNDL:
3280af22 1260 PL_reg_flags |= RF_tainted;
bbce6d69 1261 /* FALL THROUGH */
a0d0e21e 1262 case BOUND:
ffc61ed2 1263 if (do_utf8) {
12d33761 1264 if (s == PL_bostr)
ffc61ed2
JH
1265 tmp = '\n';
1266 else {
6136c704 1267 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
4ad0818d 1268 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
ffc61ed2
JH
1269 }
1270 tmp = ((OP(c) == BOUND ?
9041c2e3 1271 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1a4fad37 1272 LOAD_UTF8_CHARCLASS_ALNUM();
4cadc6a9 1273 REXEC_FBC_UTF8_SCAN(
ffc61ed2 1274 if (tmp == !(OP(c) == BOUND ?
bb7a0f54 1275 (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
ffc61ed2
JH
1276 isALNUM_LC_utf8((U8*)s)))
1277 {
1278 tmp = !tmp;
4cadc6a9 1279 REXEC_FBC_TRYIT;
a687059c 1280 }
4cadc6a9 1281 );
a0d0e21e 1282 }
667bb95a 1283 else {
12d33761 1284 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2 1285 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
4cadc6a9 1286 REXEC_FBC_SCAN(
ffc61ed2
JH
1287 if (tmp ==
1288 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
1289 tmp = !tmp;
4cadc6a9 1290 REXEC_FBC_TRYIT;
a0ed51b3 1291 }
4cadc6a9 1292 );
a0ed51b3 1293 }
3b0527fe 1294 if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, s)))
a0ed51b3
LW
1295 goto got_it;
1296 break;
bbce6d69 1297 case NBOUNDL:
3280af22 1298 PL_reg_flags |= RF_tainted;
bbce6d69 1299 /* FALL THROUGH */
a0d0e21e 1300 case NBOUND:
ffc61ed2 1301 if (do_utf8) {
12d33761 1302 if (s == PL_bostr)
ffc61ed2
JH
1303 tmp = '\n';
1304 else {
6136c704 1305 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
4ad0818d 1306 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
ffc61ed2
JH
1307 }
1308 tmp = ((OP(c) == NBOUND ?
9041c2e3 1309 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1a4fad37 1310 LOAD_UTF8_CHARCLASS_ALNUM();
4cadc6a9 1311 REXEC_FBC_UTF8_SCAN(
ffc61ed2 1312 if (tmp == !(OP(c) == NBOUND ?
bb7a0f54 1313 (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
ffc61ed2
JH
1314 isALNUM_LC_utf8((U8*)s)))
1315 tmp = !tmp;
4cadc6a9
YO
1316 else REXEC_FBC_TRYIT;
1317 );
a0d0e21e 1318 }
667bb95a 1319 else {
12d33761 1320 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2
JH
1321 tmp = ((OP(c) == NBOUND ?
1322 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
4cadc6a9 1323 REXEC_FBC_SCAN(
ffc61ed2
JH
1324 if (tmp ==
1325 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1326 tmp = !tmp;
4cadc6a9
YO
1327 else REXEC_FBC_TRYIT;
1328 );
a0ed51b3 1329 }
3b0527fe 1330 if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, s)))
a0ed51b3
LW
1331 goto got_it;
1332 break;
a0d0e21e 1333 case ALNUM:
4cadc6a9
YO
1334 REXEC_FBC_CSCAN_PRELOAD(
1335 LOAD_UTF8_CHARCLASS_ALNUM(),
1336 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8),
1337 isALNUM(*s)
1338 );
bbce6d69 1339 case ALNUML:
4cadc6a9
YO
1340 REXEC_FBC_CSCAN_TAINT(
1341 isALNUM_LC_utf8((U8*)s),
1342 isALNUM_LC(*s)
1343 );
a0d0e21e 1344 case NALNUM:
4cadc6a9
YO
1345 REXEC_FBC_CSCAN_PRELOAD(
1346 LOAD_UTF8_CHARCLASS_ALNUM(),
1347 !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8),
1348 !isALNUM(*s)
1349 );
bbce6d69 1350 case NALNUML:
4cadc6a9
YO
1351 REXEC_FBC_CSCAN_TAINT(
1352 !isALNUM_LC_utf8((U8*)s),
1353 !isALNUM_LC(*s)
1354 );
a0d0e21e 1355 case SPACE:
4cadc6a9
YO
1356 REXEC_FBC_CSCAN_PRELOAD(
1357 LOAD_UTF8_CHARCLASS_SPACE(),
1358 *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8),
1359 isSPACE(*s)
1360 );
bbce6d69 1361 case SPACEL:
4cadc6a9
YO
1362 REXEC_FBC_CSCAN_TAINT(
1363 *s == ' ' || isSPACE_LC_utf8((U8*)s),
1364 isSPACE_LC(*s)
1365 );
a0d0e21e 1366 case NSPACE:
4cadc6a9
YO
1367 REXEC_FBC_CSCAN_PRELOAD(
1368 LOAD_UTF8_CHARCLASS_SPACE(),
1369 !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)),
1370 !isSPACE(*s)
1371 );
bbce6d69 1372 case NSPACEL:
4cadc6a9
YO
1373 REXEC_FBC_CSCAN_TAINT(
1374 !(*s == ' ' || isSPACE_LC_utf8((U8*)s)),
1375 !isSPACE_LC(*s)
1376 );
a0d0e21e 1377 case DIGIT:
4cadc6a9
YO
1378 REXEC_FBC_CSCAN_PRELOAD(
1379 LOAD_UTF8_CHARCLASS_DIGIT(),
1380 swash_fetch(PL_utf8_digit,(U8*)s, do_utf8),
1381 isDIGIT(*s)
1382 );
b8c5462f 1383 case DIGITL:
4cadc6a9
YO
1384 REXEC_FBC_CSCAN_TAINT(
1385 isDIGIT_LC_utf8((U8*)s),
1386 isDIGIT_LC(*s)
1387 );
a0d0e21e 1388 case NDIGIT:
4cadc6a9
YO
1389 REXEC_FBC_CSCAN_PRELOAD(
1390 LOAD_UTF8_CHARCLASS_DIGIT(),
1391 !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8),
1392 !isDIGIT(*s)
1393 );
b8c5462f 1394 case NDIGITL:
4cadc6a9
YO
1395 REXEC_FBC_CSCAN_TAINT(
1396 !isDIGIT_LC_utf8((U8*)s),
1397 !isDIGIT_LC(*s)
1398 );
1de06328
YO
1399 case AHOCORASICKC:
1400 case AHOCORASICK:
07be1b83
YO
1401 {
1402 const enum { trie_plain, trie_utf8, trie_utf8_fold }
1403 trie_type = do_utf8 ?
1404 (c->flags == EXACT ? trie_utf8 : trie_utf8_fold)
1405 : trie_plain;
1406 /* what trie are we using right now */
1407 reg_ac_data *aho
1408 = (reg_ac_data*)prog->data->data[ ARG( c ) ];
1409 reg_trie_data *trie=aho->trie;
1410
1411 const char *last_start = strend - trie->minlen;
6148ee25 1412#ifdef DEBUGGING
07be1b83 1413 const char *real_start = s;
6148ee25 1414#endif
07be1b83 1415 STRLEN maxlen = trie->maxlen;
be8e71aa
YO
1416 SV *sv_points;
1417 U8 **points; /* map of where we were in the input string
786e8c11 1418 when reading a given char. For ASCII this
be8e71aa
YO
1419 is unnecessary overhead as the relationship
1420 is always 1:1, but for unicode, especially
1421 case folded unicode this is not true. */
f9e705e8 1422 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
786e8c11
YO
1423 U8 *bitmap=NULL;
1424
07be1b83
YO
1425
1426 GET_RE_DEBUG_FLAGS_DECL;
1427
be8e71aa
YO
1428 /* We can't just allocate points here. We need to wrap it in
1429 * an SV so it gets freed properly if there is a croak while
1430 * running the match */
1431 ENTER;
1432 SAVETMPS;
1433 sv_points=newSV(maxlen * sizeof(U8 *));
1434 SvCUR_set(sv_points,
1435 maxlen * sizeof(U8 *));
1436 SvPOK_on(sv_points);
1437 sv_2mortal(sv_points);
1438 points=(U8**)SvPV_nolen(sv_points );
1de06328
YO
1439 if ( trie_type != trie_utf8_fold
1440 && (trie->bitmap || OP(c)==AHOCORASICKC) )
1441 {
786e8c11
YO
1442 if (trie->bitmap)
1443 bitmap=(U8*)trie->bitmap;
1444 else
1445 bitmap=(U8*)ANYOF_BITMAP(c);
07be1b83 1446 }
786e8c11
YO
1447 /* this is the Aho-Corasick algorithm modified a touch
1448 to include special handling for long "unknown char"
1449 sequences. The basic idea being that we use AC as long
1450 as we are dealing with a possible matching char, when
1451 we encounter an unknown char (and we have not encountered
1452 an accepting state) we scan forward until we find a legal
1453 starting char.
1454 AC matching is basically that of trie matching, except
1455 that when we encounter a failing transition, we fall back
1456 to the current states "fail state", and try the current char
1457 again, a process we repeat until we reach the root state,
1458 state 1, or a legal transition. If we fail on the root state
1459 then we can either terminate if we have reached an accepting
1460 state previously, or restart the entire process from the beginning
1461 if we have not.
1462
1463 */
07be1b83
YO
1464 while (s <= last_start) {
1465 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1466 U8 *uc = (U8*)s;
1467 U16 charid = 0;
1468 U32 base = 1;
1469 U32 state = 1;
1470 UV uvc = 0;
1471 STRLEN len = 0;
1472 STRLEN foldlen = 0;
1473 U8 *uscan = (U8*)NULL;
1474 U8 *leftmost = NULL;
786e8c11
YO
1475#ifdef DEBUGGING
1476 U32 accepted_word= 0;
1477#endif
07be1b83
YO
1478 U32 pointpos = 0;
1479
1480 while ( state && uc <= (U8*)strend ) {
1481 int failed=0;
786e8c11
YO
1482 U32 word = aho->states[ state ].wordnum;
1483
1de06328
YO
1484 if( state==1 ) {
1485 if ( bitmap ) {
1486 DEBUG_TRIE_EXECUTE_r(
1487 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1488 dump_exec_pos( (char *)uc, c, strend, real_start,
1489 (char *)uc, do_utf8 );
1490 PerlIO_printf( Perl_debug_log,
1491 " Scanning for legal start char...\n");
1492 }
1493 );
1494 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1495 uc++;
786e8c11 1496 }
1de06328 1497 s= (char *)uc;
786e8c11 1498 }
786e8c11
YO
1499 if (uc >(U8*)last_start) break;
1500 }
1501
1502 if ( word ) {
1503 U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ];
1504 if (!leftmost || lpos < leftmost) {
1505 DEBUG_r(accepted_word=word);
07be1b83 1506 leftmost= lpos;
786e8c11 1507 }
07be1b83 1508 if (base==0) break;
786e8c11 1509
07be1b83
YO
1510 }
1511 points[pointpos++ % maxlen]= uc;
4cadc6a9
YO
1512 REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len,
1513 uvc, charid, foldlen, foldbuf, uniflags);
786e8c11
YO
1514 DEBUG_TRIE_EXECUTE_r({
1515 dump_exec_pos( (char *)uc, c, strend, real_start,
1516 s, do_utf8 );
07be1b83 1517 PerlIO_printf(Perl_debug_log,
786e8c11
YO
1518 " Charid:%3u CP:%4"UVxf" ",
1519 charid, uvc);
1520 });
07be1b83
YO
1521
1522 do {
6148ee25 1523#ifdef DEBUGGING
786e8c11 1524 word = aho->states[ state ].wordnum;
6148ee25 1525#endif
07be1b83
YO
1526 base = aho->states[ state ].trans.base;
1527
786e8c11
YO
1528 DEBUG_TRIE_EXECUTE_r({
1529 if (failed)
1530 dump_exec_pos( (char *)uc, c, strend, real_start,
1531 s, do_utf8 );
07be1b83 1532 PerlIO_printf( Perl_debug_log,
786e8c11
YO
1533 "%sState: %4"UVxf", word=%"UVxf,
1534 failed ? " Fail transition to " : "",
1535 (UV)state, (UV)word);
1536 });
07be1b83
YO
1537 if ( base ) {
1538 U32 tmp;
1539 if (charid &&
1540 (base + charid > trie->uniquecharcount )
1541 && (base + charid - 1 - trie->uniquecharcount
1542 < trie->lasttrans)
1543 && trie->trans[base + charid - 1 -
1544 trie->uniquecharcount].check == state
1545 && (tmp=trie->trans[base + charid - 1 -
1546 trie->uniquecharcount ].next))
1547 {
786e8c11
YO
1548 DEBUG_TRIE_EXECUTE_r(
1549 PerlIO_printf( Perl_debug_log," - legal\n"));
07be1b83
YO
1550 state = tmp;
1551 break;
1552 }
1553 else {
786e8c11
YO
1554 DEBUG_TRIE_EXECUTE_r(
1555 PerlIO_printf( Perl_debug_log," - fail\n"));
1556 failed = 1;
1557 state = aho->fail[state];
07be1b83
YO
1558 }
1559 }
1560 else {
1561 /* we must be accepting here */
786e8c11
YO
1562 DEBUG_TRIE_EXECUTE_r(
1563 PerlIO_printf( Perl_debug_log," - accepting\n"));
1564 failed = 1;
07be1b83
YO
1565 break;
1566 }
1567 } while(state);
786e8c11 1568 uc += len;
07be1b83
YO
1569 if (failed) {
1570 if (leftmost)
1571 break;
786e8c11 1572 if (!state) state = 1;
07be1b83
YO
1573 }
1574 }
1575 if ( aho->states[ state ].wordnum ) {
1576 U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ];
786e8c11
YO
1577 if (!leftmost || lpos < leftmost) {
1578 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
07be1b83 1579 leftmost = lpos;
786e8c11 1580 }
07be1b83 1581 }
07be1b83
YO
1582 if (leftmost) {
1583 s = (char*)leftmost;
786e8c11
YO
1584 DEBUG_TRIE_EXECUTE_r({
1585 PerlIO_printf(
1586 Perl_debug_log,"Matches word #%"UVxf" at position %d. Trying full pattern...\n",
1587 (UV)accepted_word, s - real_start
1588 );
1589 });
be8e71aa
YO
1590 if (!reginfo || regtry(reginfo, s)) {
1591 FREETMPS;
1592 LEAVE;
07be1b83 1593 goto got_it;
be8e71aa 1594 }
07be1b83 1595 s = HOPc(s,1);
786e8c11
YO
1596 DEBUG_TRIE_EXECUTE_r({
1597 PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
1598 });
07be1b83 1599 } else {
786e8c11
YO
1600 DEBUG_TRIE_EXECUTE_r(
1601 PerlIO_printf( Perl_debug_log,"No match.\n"));
07be1b83
YO
1602 break;
1603 }
1604 }
be8e71aa
YO
1605 FREETMPS;
1606 LEAVE;
07be1b83
YO
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{
97aff369 1630 dVAR;
6eb5f6b9
JH
1631 register char *s;
1632 register regnode *c;
1633 register char *startpos = stringarg;
6eb5f6b9
JH
1634 I32 minlen; /* must match at least this many chars */
1635 I32 dontbother = 0; /* how many characters not to try at end */
6eb5f6b9
JH
1636 I32 end_shift = 0; /* Same for the end. */ /* CC */
1637 I32 scream_pos = -1; /* Internal iterator of scream. */
ccac19ea 1638 char *scream_olds = NULL;
3dab1dad 1639 SV* const oreplsv = GvSV(PL_replgv);
1df70142 1640 const bool do_utf8 = DO_UTF8(sv);
2757e526 1641 I32 multiline;
0df25f3d 1642
3b0527fe 1643 regmatch_info reginfo; /* create some info to pass to regtry etc */
a3621e74
YO
1644
1645 GET_RE_DEBUG_FLAGS_DECL;
1646
9d4ba2ae 1647 PERL_UNUSED_ARG(data);
6eb5f6b9
JH
1648
1649 /* Be paranoid... */
1650 if (prog == NULL || startpos == NULL) {
1651 Perl_croak(aTHX_ "NULL regexp parameter");
1652 return 0;
1653 }
1654
2757e526 1655 multiline = prog->reganch & PMf_MULTILINE;
3b0527fe 1656 reginfo.prog = prog;
2757e526 1657
bac06658 1658 RX_MATCH_UTF8_set(prog, do_utf8);
1de06328
YO
1659 DEBUG_EXECUTE_r(
1660 debug_start_match(prog, do_utf8, startpos, strend,
1661 "Matching");
1662 );
bac06658 1663
6eb5f6b9 1664 minlen = prog->minlen;
1de06328
YO
1665
1666 if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
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
1de06328 1672
6eb5f6b9
JH
1673 /* Check validity of program. */
1674 if (UCHARAT(prog->program) != REG_MAGIC) {
1675 Perl_croak(aTHX_ "corrupted regexp program");
1676 }
1677
1678 PL_reg_flags = 0;
1679 PL_reg_eval_set = 0;
1680 PL_reg_maxiter = 0;
1681
1682 if (prog->reganch & ROPT_UTF8)
1683 PL_reg_flags |= RF_utf8;
1684
1685 /* Mark beginning of line for ^ and lookbehind. */
3b0527fe 1686 reginfo.bol = startpos; /* XXX not used ??? */
6eb5f6b9 1687 PL_bostr = strbeg;
3b0527fe 1688 reginfo.sv = sv;
6eb5f6b9
JH
1689
1690 /* Mark end of line for $ (and such) */
1691 PL_regeol = strend;
1692
1693 /* see how far we have to get to not match where we matched before */
3b0527fe 1694 reginfo.till = startpos+minend;
6eb5f6b9 1695
6eb5f6b9
JH
1696 /* If there is a "must appear" string, look for it. */
1697 s = startpos;
1698
3b0527fe 1699 if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to set reginfo->ganch */
6eb5f6b9
JH
1700 MAGIC *mg;
1701
1702 if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
3b0527fe 1703 reginfo.ganch = startpos;
6eb5f6b9
JH
1704 else if (sv && SvTYPE(sv) >= SVt_PVMG
1705 && SvMAGIC(sv)
14befaf4
DM
1706 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
1707 && mg->mg_len >= 0) {
3b0527fe 1708 reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */
6eb5f6b9 1709 if (prog->reganch & ROPT_ANCH_GPOS) {
3b0527fe 1710 if (s > reginfo.ganch)
6eb5f6b9 1711 goto phooey;
3b0527fe 1712 s = reginfo.ganch;
6eb5f6b9
JH
1713 }
1714 }
1715 else /* pos() not defined */
3b0527fe 1716 reginfo.ganch = strbeg;
6eb5f6b9
JH
1717 }
1718
a0714e2c 1719 if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
6eb5f6b9
JH
1720 re_scream_pos_data d;
1721
1722 d.scream_olds = &scream_olds;
1723 d.scream_pos = &scream_pos;
1724 s = re_intuit_start(prog, sv, s, strend, flags, &d);
3fa9c3d7 1725 if (!s) {
a3621e74 1726 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
6eb5f6b9 1727 goto phooey; /* not present */
3fa9c3d7 1728 }
6eb5f6b9
JH
1729 }
1730
1de06328 1731
6eb5f6b9
JH
1732
1733 /* Simplest case: anchored match need be tried only once. */
1734 /* [unless only anchor is BOL and multiline is set] */
1735 if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
3b0527fe 1736 if (s == startpos && regtry(&reginfo, startpos))
6eb5f6b9 1737 goto got_it;
7fba1cd6 1738 else if (multiline || (prog->reganch & ROPT_IMPLICIT)
6eb5f6b9
JH
1739 || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
1740 {
1741 char *end;
1742
1743 if (minlen)
1744 dontbother = minlen - 1;
1aa99e6b 1745 end = HOP3c(strend, -dontbother, strbeg) - 1;
6eb5f6b9 1746 /* for multiline we only have to try after newlines */
33b8afdf 1747 if (prog->check_substr || prog->check_utf8) {
6eb5f6b9
JH
1748 if (s == startpos)
1749 goto after_try;
1750 while (1) {
3b0527fe 1751 if (regtry(&reginfo, s))
6eb5f6b9
JH
1752 goto got_it;
1753 after_try:
1754 if (s >= end)
1755 goto phooey;
1756 if (prog->reganch & RE_USE_INTUIT) {
1757 s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
1758 if (!s)
1759 goto phooey;
1760 }
1761 else
1762 s++;
1763 }
1764 } else {
1765 if (s > startpos)
1766 s--;
1767 while (s < end) {
1768 if (*s++ == '\n') { /* don't need PL_utf8skip here */
3b0527fe 1769 if (regtry(&reginfo, s))
6eb5f6b9
JH
1770 goto got_it;
1771 }
1772 }
1773 }
1774 }
1775 goto phooey;
1776 } else if (prog->reganch & ROPT_ANCH_GPOS) {
3b0527fe 1777 if (regtry(&reginfo, reginfo.ganch))
6eb5f6b9
JH
1778 goto got_it;
1779 goto phooey;
1780 }
1781
1782 /* Messy cases: unanchored match. */
33b8afdf 1783 if ((prog->anchored_substr || prog->anchored_utf8) && prog->reganch & ROPT_SKIP) {
6eb5f6b9
JH
1784 /* we have /x+whatever/ */
1785 /* it must be a one character string (XXXX Except UTF?) */
33b8afdf 1786 char ch;
bf93d4cc
GS
1787#ifdef DEBUGGING
1788 int did_match = 0;
1789#endif
33b8afdf
JH
1790 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1791 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
3f7c398e 1792 ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
bf93d4cc 1793
1aa99e6b 1794 if (do_utf8) {
4cadc6a9 1795 REXEC_FBC_SCAN(
6eb5f6b9 1796 if (*s == ch) {
a3621e74 1797 DEBUG_EXECUTE_r( did_match = 1 );
3b0527fe 1798 if (regtry(&reginfo, s)) goto got_it;
6eb5f6b9
JH
1799 s += UTF8SKIP(s);
1800 while (s < strend && *s == ch)
1801 s += UTF8SKIP(s);
1802 }
4cadc6a9 1803 );
6eb5f6b9
JH
1804 }
1805 else {
4cadc6a9 1806 REXEC_FBC_SCAN(
6eb5f6b9 1807 if (*s == ch) {
a3621e74 1808 DEBUG_EXECUTE_r( did_match = 1 );
3b0527fe 1809 if (regtry(&reginfo, s)) goto got_it;
6eb5f6b9
JH
1810 s++;
1811 while (s < strend && *s == ch)
1812 s++;
1813 }
4cadc6a9 1814 );
6eb5f6b9 1815 }
a3621e74 1816 DEBUG_EXECUTE_r(if (!did_match)
bf93d4cc 1817 PerlIO_printf(Perl_debug_log,
b7953727
JH
1818 "Did not find anchored character...\n")
1819 );
6eb5f6b9 1820 }
a0714e2c
SS
1821 else if (prog->anchored_substr != NULL
1822 || prog->anchored_utf8 != NULL
1823 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
33b8afdf
JH
1824 && prog->float_max_offset < strend - s)) {
1825 SV *must;
1826 I32 back_max;
1827 I32 back_min;
1828 char *last;
6eb5f6b9 1829 char *last1; /* Last position checked before */
bf93d4cc
GS
1830#ifdef DEBUGGING
1831 int did_match = 0;
1832#endif
33b8afdf
JH
1833 if (prog->anchored_substr || prog->anchored_utf8) {
1834 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1835 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1836 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
1837 back_max = back_min = prog->anchored_offset;
1838 } else {
1839 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1840 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1841 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
1842 back_max = prog->float_max_offset;
1843 back_min = prog->float_min_offset;
1844 }
1de06328
YO
1845
1846
33b8afdf
JH
1847 if (must == &PL_sv_undef)
1848 /* could not downgrade utf8 check substring, so must fail */
1849 goto phooey;
1850
1de06328
YO
1851 if (back_min<0) {
1852 last = strend;
1853 } else {
1854 last = HOP3c(strend, /* Cannot start after this */
1855 -(I32)(CHR_SVLEN(must)
1856 - (SvTAIL(must) != 0) + back_min), strbeg);
1857 }
6eb5f6b9
JH
1858 if (s > PL_bostr)
1859 last1 = HOPc(s, -1);
1860 else
1861 last1 = s - 1; /* bogus */
1862
a0288114 1863 /* XXXX check_substr already used to find "s", can optimize if
6eb5f6b9
JH
1864 check_substr==must. */
1865 scream_pos = -1;
1866 dontbother = end_shift;
1867 strend = HOPc(strend, -dontbother);
1868 while ( (s <= last) &&
9041c2e3 1869 ((flags & REXEC_SCREAM)
1de06328 1870 ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg,
6eb5f6b9 1871 end_shift, &scream_pos, 0))
1de06328 1872 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)),
9041c2e3 1873 (unsigned char*)strend, must,
7fba1cd6 1874 multiline ? FBMrf_MULTILINE : 0))) ) {
4addbd3b
HS
1875 /* we may be pointing at the wrong string */
1876 if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog))
3f7c398e 1877 s = strbeg + (s - SvPVX_const(sv));
a3621e74 1878 DEBUG_EXECUTE_r( did_match = 1 );
6eb5f6b9
JH
1879 if (HOPc(s, -back_max) > last1) {
1880 last1 = HOPc(s, -back_min);
1881 s = HOPc(s, -back_max);
1882 }
1883 else {
52657f30 1884 char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
6eb5f6b9
JH
1885
1886 last1 = HOPc(s, -back_min);
52657f30 1887 s = t;
6eb5f6b9 1888 }
1aa99e6b 1889 if (do_utf8) {
6eb5f6b9 1890 while (s <= last1) {
3b0527fe 1891 if (regtry(&reginfo, s))
6eb5f6b9
JH
1892 goto got_it;
1893 s += UTF8SKIP(s);
1894 }
1895 }
1896 else {
1897 while (s <= last1) {
3b0527fe 1898 if (regtry(&reginfo, s))
6eb5f6b9
JH
1899 goto got_it;
1900 s++;
1901 }
1902 }
1903 }
ab3bbdeb
YO
1904 DEBUG_EXECUTE_r(if (!did_match) {
1905 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
1906 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
1907 PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
33b8afdf 1908 ((must == prog->anchored_substr || must == prog->anchored_utf8)
bf93d4cc 1909 ? "anchored" : "floating"),
ab3bbdeb
YO
1910 quoted, RE_SV_TAIL(must));
1911 });
6eb5f6b9
JH
1912 goto phooey;
1913 }
786e8c11 1914 else if ( (c = prog->regstclass) ) {
f14c76ed 1915 if (minlen) {
be8e71aa 1916 const OPCODE op = OP(prog->regstclass);
66e933ab 1917 /* don't bother with what can't match */
786e8c11 1918 if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
f14c76ed
RGS
1919 strend = HOPc(strend, -(minlen - 1));
1920 }
a3621e74 1921 DEBUG_EXECUTE_r({
be8e71aa 1922 SV * const prop = sv_newmortal();
32fc9b6a 1923 regprop(prog, prop, c);
0df25f3d 1924 {
ab3bbdeb
YO
1925 RE_PV_QUOTED_DECL(quoted,UTF,PERL_DEBUG_PAD_ZERO(1),
1926 s,strend-s,60);
0df25f3d 1927 PerlIO_printf(Perl_debug_log,
ab3bbdeb 1928 "Matching stclass %.*s against %s (%d chars)\n",
e4f74956 1929 (int)SvCUR(prop), SvPVX_const(prop),
ab3bbdeb 1930 quoted, (int)(strend - s));
0df25f3d 1931 }
ffc61ed2 1932 });
3b0527fe 1933 if (find_byclass(prog, c, s, strend, &reginfo))
6eb5f6b9 1934 goto got_it;
07be1b83 1935 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
d6a28714
JH
1936 }
1937 else {
1938 dontbother = 0;
a0714e2c 1939 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
33b8afdf 1940 /* Trim the end. */
d6a28714 1941 char *last;
33b8afdf
JH
1942 SV* float_real;
1943
1944 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1945 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1946 float_real = do_utf8 ? prog->float_utf8 : prog->float_substr;
d6a28714
JH
1947
1948 if (flags & REXEC_SCREAM) {
33b8afdf 1949 last = screaminstr(sv, float_real, s - strbeg,
d6a28714
JH
1950 end_shift, &scream_pos, 1); /* last one */
1951 if (!last)
ffc61ed2 1952 last = scream_olds; /* Only one occurrence. */
4addbd3b
HS
1953 /* we may be pointing at the wrong string */
1954 else if (RX_MATCH_COPIED(prog))
3f7c398e 1955 s = strbeg + (s - SvPVX_const(sv));
b8c5462f 1956 }
d6a28714
JH
1957 else {
1958 STRLEN len;
cfd0369c 1959 const char * const little = SvPV_const(float_real, len);
d6a28714 1960
33b8afdf 1961 if (SvTAIL(float_real)) {
d6a28714
JH
1962 if (memEQ(strend - len + 1, little, len - 1))
1963 last = strend - len + 1;
7fba1cd6 1964 else if (!multiline)
9041c2e3 1965 last = memEQ(strend - len, little, len)
bd61b366 1966 ? strend - len : NULL;
b8c5462f 1967 else
d6a28714
JH
1968 goto find_last;
1969 } else {
1970 find_last:
9041c2e3 1971 if (len)
d6a28714 1972 last = rninstr(s, strend, little, little + len);
b8c5462f 1973 else
a0288114 1974 last = strend; /* matching "$" */
b8c5462f 1975 }
b8c5462f 1976 }
bf93d4cc 1977 if (last == NULL) {
a3621e74 1978 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
bf93d4cc 1979 "%sCan't trim the tail, match fails (should not happen)%s\n",
e4584336 1980 PL_colors[4], PL_colors[5]));
bf93d4cc
GS
1981 goto phooey; /* Should not happen! */
1982 }
d6a28714
JH
1983 dontbother = strend - last + prog->float_min_offset;
1984 }
1985 if (minlen && (dontbother < minlen))
1986 dontbother = minlen - 1;
1987 strend -= dontbother; /* this one's always in bytes! */
1988 /* We don't know much -- general case. */
1aa99e6b 1989 if (do_utf8) {
d6a28714 1990 for (;;) {
3b0527fe 1991 if (regtry(&reginfo, s))
d6a28714
JH
1992 goto got_it;
1993 if (s >= strend)
1994 break;
b8c5462f 1995 s += UTF8SKIP(s);
d6a28714
JH
1996 };
1997 }
1998 else {
1999 do {
3b0527fe 2000 if (regtry(&reginfo, s))
d6a28714
JH
2001 goto got_it;
2002 } while (s++ < strend);
2003 }
2004 }
2005
2006 /* Failure. */
2007 goto phooey;
2008
2009got_it:
2010 RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
2011
2012 if (PL_reg_eval_set) {
2013 /* Preserve the current value of $^R */
2014 if (oreplsv != GvSV(PL_replgv))
2015 sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
2016 restored, the value remains
2017 the same. */
4f639d21 2018 restore_pos(aTHX_ prog);
d6a28714
JH
2019 }
2020
2021 /* make sure $`, $&, $', and $digit will work later */
2022 if ( !(flags & REXEC_NOT_FIRST) ) {
ed252734 2023 RX_MATCH_COPY_FREE(prog);
d6a28714 2024 if (flags & REXEC_COPY_STR) {
be8e71aa 2025 const I32 i = PL_regeol - startpos + (stringarg - strbeg);
f8c7b90f 2026#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
2027 if ((SvIsCOW(sv)
2028 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2029 if (DEBUG_C_TEST) {
2030 PerlIO_printf(Perl_debug_log,
2031 "Copy on write: regexp capture, type %d\n",
2032 (int) SvTYPE(sv));
2033 }
2034 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
d5263905 2035 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
ed252734
NC
2036 assert (SvPOKp(prog->saved_copy));
2037 } else
2038#endif
2039 {
2040 RX_MATCH_COPIED_on(prog);
2041 s = savepvn(strbeg, i);
2042 prog->subbeg = s;
2043 }
d6a28714 2044 prog->sublen = i;
d6a28714
JH
2045 }
2046 else {
2047 prog->subbeg = strbeg;
2048 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2049 }
2050 }
9041c2e3 2051
d6a28714
JH
2052 return 1;
2053
2054phooey:
a3621e74 2055 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
e4584336 2056 PL_colors[4], PL_colors[5]));
d6a28714 2057 if (PL_reg_eval_set)
4f639d21 2058 restore_pos(aTHX_ prog);
d6a28714
JH
2059 return 0;
2060}
2061
2062/*
2063 - regtry - try match at specific point
2064 */
2065STATIC I32 /* 0 failure, 1 success */
3b0527fe 2066S_regtry(pTHX_ const regmatch_info *reginfo, char *startpos)
d6a28714 2067{
97aff369 2068 dVAR;
d6a28714
JH
2069 register I32 *sp;
2070 register I32 *ep;
2071 CHECKPOINT lastcp;
3b0527fe 2072 regexp *prog = reginfo->prog;
a3621e74 2073 GET_RE_DEBUG_FLAGS_DECL;
d6a28714 2074
02db2b7b
IZ
2075#ifdef DEBUGGING
2076 PL_regindent = 0; /* XXXX Not good when matches are reenterable... */
2077#endif
d6a28714
JH
2078 if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
2079 MAGIC *mg;
2080
2081 PL_reg_eval_set = RS_init;
a3621e74 2082 DEBUG_EXECUTE_r(DEBUG_s(
b900a521
JH
2083 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
2084 (IV)(PL_stack_sp - PL_stack_base));
d6a28714 2085 ));
e8347627 2086 SAVEI32(cxstack[cxstack_ix].blk_oldsp);
d6a28714
JH
2087 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
2088 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
2089 SAVETMPS;
2090 /* Apparently this is not needed, judging by wantarray. */
e8347627 2091 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
d6a28714
JH
2092 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
2093
3b0527fe 2094 if (reginfo->sv) {
d6a28714 2095 /* Make $_ available to executed code. */
3b0527fe 2096 if (reginfo->sv != DEFSV) {
59f00321 2097 SAVE_DEFSV;
3b0527fe 2098 DEFSV = reginfo->sv;
b8c5462f 2099 }
d6a28714 2100
3b0527fe
DM
2101 if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
2102 && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
d6a28714 2103 /* prepare for quick setting of pos */
d300d9fa
NC
2104#ifdef PERL_OLD_COPY_ON_WRITE
2105 if (SvIsCOW(sv))
2106 sv_force_normal_flags(sv, 0);
2107#endif
3dab1dad 2108 mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global,
d300d9fa 2109 &PL_vtbl_mglob, NULL, 0);
d6a28714 2110 mg->mg_len = -1;
b8c5462f 2111 }
d6a28714
JH
2112 PL_reg_magic = mg;
2113 PL_reg_oldpos = mg->mg_len;
4f639d21 2114 SAVEDESTRUCTOR_X(restore_pos, prog);
d6a28714 2115 }
09687e5a 2116 if (!PL_reg_curpm) {
a02a5408 2117 Newxz(PL_reg_curpm, 1, PMOP);
09687e5a
AB
2118#ifdef USE_ITHREADS
2119 {
be8e71aa 2120 SV* const repointer = newSViv(0);
577e12cc 2121 /* so we know which PL_regex_padav element is PL_reg_curpm */
35061a7e 2122 SvFLAGS(repointer) |= SVf_BREAK;
09687e5a
AB
2123 av_push(PL_regex_padav,repointer);
2124 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2125 PL_regex_pad = AvARRAY(PL_regex_padav);
2126 }
2127#endif
2128 }
aaa362c4 2129 PM_SETRE(PL_reg_curpm, prog);
d6a28714
JH
2130 PL_reg_oldcurpm = PL_curpm;
2131 PL_curpm = PL_reg_curpm;
2132 if (RX_MATCH_COPIED(prog)) {
2133 /* Here is a serious problem: we cannot rewrite subbeg,
2134 since it may be needed if this match fails. Thus
2135 $` inside (?{}) could fail... */
2136 PL_reg_oldsaved = prog->subbeg;
2137 PL_reg_oldsavedlen = prog->sublen;
f8c7b90f 2138#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
2139 PL_nrs = prog->saved_copy;
2140#endif
d6a28714
JH
2141 RX_MATCH_COPIED_off(prog);
2142 }
2143 else
bd61b366 2144 PL_reg_oldsaved = NULL;
d6a28714
JH
2145 prog->subbeg = PL_bostr;
2146 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2147 }
973dddac 2148 prog->startp[0] = startpos - PL_bostr;
d6a28714
JH
2149 PL_reginput = startpos;
2150 PL_regstartp = prog->startp;
2151 PL_regendp = prog->endp;
2152 PL_reglastparen = &prog->lastparen;
a01268b5 2153 PL_reglastcloseparen = &prog->lastcloseparen;
d6a28714 2154 prog->lastparen = 0;
03994de8 2155 prog->lastcloseparen = 0;
d6a28714 2156 PL_regsize = 0;
a3621e74 2157 DEBUG_EXECUTE_r(PL_reg_starttry = startpos);
d6a28714
JH
2158 if (PL_reg_start_tmpl <= prog->nparens) {
2159 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
2160 if(PL_reg_start_tmp)
2161 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2162 else
a02a5408 2163 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
d6a28714
JH
2164 }
2165
2166 /* XXXX What this code is doing here?!!! There should be no need
2167 to do this again and again, PL_reglastparen should take care of
3dd2943c 2168 this! --ilya*/
dafc8851
JH
2169
2170 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2171 * Actually, the code in regcppop() (which Ilya may be meaning by
daf18116
JH
2172 * PL_reglastparen), is not needed at all by the test suite
2173 * (op/regexp, op/pat, op/split), but that code is needed, oddly
2174 * enough, for building DynaLoader, or otherwise this
2175 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
2176 * will happen. Meanwhile, this code *is* needed for the
2177 * above-mentioned test suite tests to succeed. The common theme
2178 * on those tests seems to be returning null fields from matches.
2179 * --jhi */
dafc8851 2180#if 1
d6a28714
JH
2181 sp = prog->startp;
2182 ep = prog->endp;
2183 if (prog->nparens) {
097eb12c 2184 register I32 i;
eb160463 2185 for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
d6a28714
JH
2186 *++sp = -1;
2187 *++ep = -1;
2188 }
2189 }
dafc8851 2190#endif
02db2b7b 2191 REGCP_SET(lastcp);
3b0527fe 2192 if (regmatch(reginfo, prog->program + 1)) {
d6a28714
JH
2193 prog->endp[0] = PL_reginput - PL_bostr;
2194 return 1;
2195 }
02db2b7b 2196 REGCP_UNWIND(lastcp);
d6a28714
JH
2197 return 0;
2198}
2199
02db2b7b 2200
8ba1375e
MJD
2201#define sayYES goto yes
2202#define sayNO goto no
e0f9d4a8 2203#define sayNO_ANYOF goto no_anyof
8ba1375e 2204#define sayYES_FINAL goto yes_final
8ba1375e
MJD
2205#define sayNO_FINAL goto no_final
2206#define sayNO_SILENT goto do_no
2207#define saySAME(x) if (x) goto yes; else goto no
2208
3ab3c9b4 2209#define CACHEsayNO STMT_START { \
3298f257
DM
2210 if (st->u.whilem.cache_offset | st->u.whilem.cache_bit) \
2211 PL_reg_poscache[st->u.whilem.cache_offset] |= \
2212 (1<<st->u.whilem.cache_bit); \
3ab3c9b4
HS
2213 sayNO; \
2214} STMT_END
2215
3298f257 2216
a3621e74
YO
2217/* this is used to determine how far from the left messages like
2218 'failed...' are printed. Currently 29 makes these messages line
2219 up with the opcode they refer to. Earlier perls used 25 which
2220 left these messages outdented making reviewing a debug output
2221 quite difficult.
2222*/
2223#define REPORT_CODE_OFF 29
2224
2225
2226/* Make sure there is a test for this +1 options in re_tests */
2227#define TRIE_INITAL_ACCEPT_BUFFLEN 4;
2228
40a82448
DM
2229#define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
2230#define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
9e137952 2231
86545054
DM
2232#define SLAB_FIRST(s) (&(s)->states[0])
2233#define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
2234
5d9a96ca
DM
2235/* grab a new slab and return the first slot in it */
2236
2237STATIC regmatch_state *
2238S_push_slab(pTHX)
2239{
54df2634
NC
2240#if PERL_VERSION < 9
2241 dMY_CXT;
2242#endif
5d9a96ca
DM
2243 regmatch_slab *s = PL_regmatch_slab->next;
2244 if (!s) {
2245 Newx(s, 1, regmatch_slab);
2246 s->prev = PL_regmatch_slab;
2247 s->next = NULL;
2248 PL_regmatch_slab->next = s;
2249 }
2250 PL_regmatch_slab = s;
86545054 2251 return SLAB_FIRST(s);
5d9a96ca 2252}
5b47454d 2253
95b24440
DM
2254/* simulate a recursive call to regmatch */
2255
2256#define REGMATCH(ns, where) \
5d9a96ca
DM
2257 st->scan = scan; \
2258 scan = (ns); \
2259 st->resume_state = resume_##where; \
95b24440
DM
2260 goto start_recurse; \
2261 resume_point_##where:
2262
40a82448
DM
2263/* push a new state then goto it */
2264
2265#define PUSH_STATE_GOTO(state, node) \
2266 scan = node; \
2267 st->resume_state = state; \
2268 goto push_state;
2269
2270/* push a new state with success backtracking, then goto it */
2271
2272#define PUSH_YES_STATE_GOTO(state, node) \
2273 scan = node; \
2274 st->resume_state = state; \
2275 goto push_yes_state;
2276
aa283a38 2277
aa283a38 2278
d6a28714
JH
2279/*
2280 - regmatch - main matching routine
2281 *
2282 * Conceptually the strategy is simple: check to see whether the current
2283 * node matches, call self recursively to see whether the rest matches,
2284 * and then act accordingly. In practice we make some effort to avoid
2285 * recursion, in particular by going through "ordinary" nodes (that don't
2286 * need to know whether the rest of the match failed) by a loop instead of
2287 * by recursion.
2288 */
2289/* [lwall] I've hoisted the register declarations to the outer block in order to
2290 * maybe save a little bit of pushing and popping on the stack. It also takes
2291 * advantage of machines that use a register save mask on subroutine entry.
95b24440
DM
2292 *
2293 * This function used to be heavily recursive, but since this had the
2294 * effect of blowing the CPU stack on complex regexes, it has been
2295 * restructured to be iterative, and to save state onto the heap rather
2296 * than the stack. Essentially whereever regmatch() used to be called, it
2297 * pushes the current state, notes where to return, then jumps back into
2298 * the main loop.
2299 *
2300 * Originally the structure of this function used to look something like
2301
2302 S_regmatch() {
2303 int a = 1, b = 2;
2304 ...
2305 while (scan != NULL) {
5d9a96ca 2306 a++; // do stuff with a and b
95b24440
DM
2307 ...
2308 switch (OP(scan)) {
2309 case FOO: {
2310 int local = 3;
2311 ...
2312 if (regmatch(...)) // recurse
2313 goto yes;
2314 }
2315 ...
2316 }
2317 }
2318 yes:
2319 return 1;
2320 }
2321
2322 * Now it looks something like this:
2323
5d9a96ca 2324 typedef struct {
95b24440
DM
2325 int a, b, local;
2326 int resume_state;
5d9a96ca 2327 } regmatch_state;
95b24440
DM
2328
2329 S_regmatch() {
5d9a96ca
DM
2330 regmatch_state *st = new();
2331 int depth=0;
2332 st->a++; // do stuff with a and b
95b24440
DM
2333 ...
2334 while (scan != NULL) {
2335 ...
2336 switch (OP(scan)) {
2337 case FOO: {
5d9a96ca 2338 st->local = 3;
95b24440 2339 ...
5d9a96ca
DM
2340 st->scan = scan;
2341 scan = ...;
2342 st->resume_state = resume_FOO;
2343 goto start_recurse; // recurse
95b24440 2344
5d9a96ca
DM
2345 resume_point_FOO:
2346 if (result)
95b24440
DM
2347 goto yes;
2348 }
2349 ...
2350 }
5d9a96ca
DM
2351 start_recurse:
2352 st = new(); push a new state
2353 st->a = 1; st->b = 2;
2354 depth++;
95b24440 2355 }
5d9a96ca 2356 yes:
95b24440 2357 result = 1;
5d9a96ca
DM
2358 if (depth--) {
2359 st = pop();
95b24440
DM
2360 switch (resume_state) {
2361 case resume_FOO:
2362 goto resume_point_FOO;
2363 ...
2364 }
2365 }
2366 return result
2367 }
2368
2369 * WARNING: this means that any line in this function that contains a
2370 * REGMATCH() or TRYPAREN() is actually simulating a recursive call to
2371 * regmatch() using gotos instead. Thus the values of any local variables
2372 * not saved in the regmatch_state structure will have been lost when
2373 * execution resumes on the next line .
5d9a96ca
DM
2374 *
2375 * States (ie the st pointer) are allocated in slabs of about 4K in size.
2376 * PL_regmatch_state always points to the currently active state, and
2377 * PL_regmatch_slab points to the slab currently containing PL_regmatch_state.
2378 * The first time regmatch is called, the first slab is allocated, and is
2379 * never freed until interpreter desctruction. When the slab is full,
2380 * a new one is allocated chained to the end. At exit from regmatch, slabs
2381 * allocated since entry are freed.
d6a28714 2382 */
95b24440 2383
40a82448 2384/* *** every FOO_fail should = FOO+1 */
c255a977
DM
2385#define TRIE_next (REGNODE_MAX+1)
2386#define TRIE_next_fail (REGNODE_MAX+2)
2387#define EVAL_A (REGNODE_MAX+3)
2388#define EVAL_A_fail (REGNODE_MAX+4)
2389#define resume_CURLYX (REGNODE_MAX+5)
2390#define resume_WHILEM1 (REGNODE_MAX+6)
2391#define resume_WHILEM2 (REGNODE_MAX+7)
2392#define resume_WHILEM3 (REGNODE_MAX+8)
2393#define resume_WHILEM4 (REGNODE_MAX+9)
2394#define resume_WHILEM5 (REGNODE_MAX+10)
2395#define resume_WHILEM6 (REGNODE_MAX+11)
2396#define BRANCH_next (REGNODE_MAX+12)
2397#define BRANCH_next_fail (REGNODE_MAX+13)
2398#define CURLYM_A (REGNODE_MAX+14)
2399#define CURLYM_A_fail (REGNODE_MAX+15)
2400#define CURLYM_B (REGNODE_MAX+16)
2401#define CURLYM_B_fail (REGNODE_MAX+17)
2402#define IFMATCH_A (REGNODE_MAX+18)
2403#define IFMATCH_A_fail (REGNODE_MAX+19)
2404#define CURLY_B_min_known (REGNODE_MAX+20)
2405#define CURLY_B_min_known_fail (REGNODE_MAX+21)
2406#define CURLY_B_min (REGNODE_MAX+22)
2407#define CURLY_B_min_fail (REGNODE_MAX+23)
2408#define CURLY_B_max (REGNODE_MAX+24)
2409#define CURLY_B_max_fail (REGNODE_MAX+25)
40a82448
DM
2410
2411
3dab1dad 2412#define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)
95b24440 2413
3df15adc 2414#ifdef DEBUGGING
ab3bbdeb
YO
2415STATIC void
2416S_debug_start_match(pTHX_ const regexp *prog, const bool do_utf8,
2417 const char *start, const char *end, const char *blurb)
2418{
2419 const bool utf8_pat= prog->reganch & ROPT_UTF8 ? 1 : 0;
2420 if (!PL_colorset)
2421 reginitcolors();
2422 {
2423 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
2424 prog->precomp, prog->prelen, 60);
2425
2426 RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1),
2427 start, end - start, 60);
2428
2429 PerlIO_printf(Perl_debug_log,
2430 "%s%s REx%s %s against %s\n",
2431 PL_colors[4], blurb, PL_colors[5], s0, s1);
2432
2433 if (do_utf8||utf8_pat)
1de06328
YO
2434 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
2435 utf8_pat ? "pattern" : "",
2436 utf8_pat && do_utf8 ? " and " : "",
2437 do_utf8 ? "string" : ""
ab3bbdeb
YO
2438 );
2439 }
2440}
3df15adc
YO
2441
2442STATIC void
786e8c11
YO
2443S_dump_exec_pos(pTHX_ const char *locinput,
2444 const regnode *scan,
2445 const char *loc_regeol,
2446 const char *loc_bostr,
2447 const char *loc_reg_starttry,
2448 const bool do_utf8)
07be1b83 2449{
786e8c11 2450 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
07be1b83 2451 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
786e8c11 2452 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
07be1b83
YO
2453 /* The part of the string before starttry has one color
2454 (pref0_len chars), between starttry and current
2455 position another one (pref_len - pref0_len chars),
2456 after the current position the third one.
2457 We assume that pref0_len <= pref_len, otherwise we
2458 decrease pref0_len. */
786e8c11
YO
2459 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
2460 ? (5 + taill) - l : locinput - loc_bostr;
07be1b83
YO
2461 int pref0_len;
2462
2463 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
2464 pref_len++;
786e8c11
YO
2465 pref0_len = pref_len - (locinput - loc_reg_starttry);
2466 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
2467 l = ( loc_regeol - locinput > (5 + taill) - pref_len
2468 ? (5 + taill) - pref_len : loc_regeol - locinput);
07be1b83
YO
2469 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
2470 l--;
2471 if (pref0_len < 0)
2472 pref0_len = 0;
2473 if (pref0_len > pref_len)
2474 pref0_len = pref_len;
2475 {
3df15adc 2476 const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0;
0df25f3d 2477
ab3bbdeb 2478 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
1de06328 2479 (locinput - pref_len),pref0_len, 60, 4, 5);
0df25f3d 2480
ab3bbdeb 2481 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
3df15adc 2482 (locinput - pref_len + pref0_len),
1de06328 2483 pref_len - pref0_len, 60, 2, 3);
0df25f3d 2484
ab3bbdeb 2485 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
1de06328 2486 locinput, loc_regeol - locinput, 10, 0, 1);
0df25f3d 2487
1de06328 2488 const STRLEN tlen=len0+len1+len2;
3df15adc 2489 PerlIO_printf(Perl_debug_log,
ab3bbdeb 2490 "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
786e8c11 2491 (IV)(locinput - loc_bostr),
07be1b83 2492 len0, s0,
07be1b83 2493 len1, s1,
07be1b83 2494 (docolor ? "" : "> <"),
07be1b83 2495 len2, s2,
1de06328 2496 tlen > 19 ? 0 : 19 - tlen,
07be1b83
YO
2497 "");
2498 }
2499}
3df15adc 2500
07be1b83
YO
2501#endif
2502
d6a28714 2503STATIC I32 /* 0 failure, 1 success */
3b0527fe 2504S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
d6a28714 2505{
54df2634
NC
2506#if PERL_VERSION < 9
2507 dMY_CXT;
2508#endif
27da23d5 2509 dVAR;
95b24440 2510 register const bool do_utf8 = PL_reg_match_utf8;
4ad0818d 2511 const U32 uniflags = UTF8_ALLOW_DEFAULT;
95b24440 2512
3b0527fe
DM
2513 regexp *rex = reginfo->prog;
2514
5d9a96ca
DM
2515 regmatch_slab *orig_slab;
2516 regmatch_state *orig_state;
a3621e74 2517
5d9a96ca
DM
2518 /* the current state. This is a cached copy of PL_regmatch_state */
2519 register regmatch_state *st;
95b24440 2520
5d9a96ca
DM
2521 /* cache heavy used fields of st in registers */
2522 register regnode *scan;
2523 register regnode *next;
2524 register I32 n = 0; /* initialize to shut up compiler warning */
2525 register char *locinput = PL_reginput;
95b24440 2526
5d9a96ca
DM
2527 /* these variables are NOT saved during a recusive RFEGMATCH: */
2528 register I32 nextchr; /* is always set to UCHARAT(locinput) */
b69b0499 2529 bool result = 0; /* return value of S_regmatch */
5d9a96ca 2530 int depth = 0; /* depth of recursion */
77cb431f
DM
2531 regmatch_state *yes_state = NULL; /* state to pop to on success of
2532 subpattern */
40a82448 2533 U32 state_num;
95b24440 2534
10edeb5d
JH
2535 I32 parenfloor = 0;
2536
95b24440 2537#ifdef DEBUGGING
e68ec53f 2538 GET_RE_DEBUG_FLAGS_DECL;
d6a28714
JH
2539 PL_regindent++;
2540#endif
2541
5d9a96ca
DM
2542 /* on first ever call to regmatch, allocate first slab */
2543 if (!PL_regmatch_slab) {
2544 Newx(PL_regmatch_slab, 1, regmatch_slab);
2545 PL_regmatch_slab->prev = NULL;
2546 PL_regmatch_slab->next = NULL;
86545054 2547 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
5d9a96ca
DM
2548 }
2549
2550 /* remember current high-water mark for exit */
2551 /* XXX this should be done with SAVE* instead */
2552 orig_slab = PL_regmatch_slab;
2553 orig_state = PL_regmatch_state;
2554
2555 /* grab next free state slot */
2556 st = ++PL_regmatch_state;
86545054 2557 if (st > SLAB_LAST(PL_regmatch_slab))
5d9a96ca
DM
2558 st = PL_regmatch_state = S_push_slab(aTHX);
2559
2560 st->minmod = 0;
2561 st->sw = 0;
2562 st->logical = 0;
5d9a96ca 2563 st->cc = NULL;
786e8c11 2564
d6a28714
JH
2565 /* Note that nextchr is a byte even in UTF */
2566 nextchr = UCHARAT(locinput);
2567 scan = prog;
2568 while (scan != NULL) {
8ba1375e 2569
a3621e74 2570 DEBUG_EXECUTE_r( {
6136c704 2571 SV * const prop = sv_newmortal();
1de06328 2572 regnode *rnext=regnext(scan);
786e8c11 2573 DUMP_EXEC_POS( locinput, scan, do_utf8 );
32fc9b6a 2574 regprop(rex, prop, scan);
07be1b83
YO
2575
2576 PerlIO_printf(Perl_debug_log,
2577 "%3"IVdf":%*s%s(%"IVdf")\n",
2578 (IV)(scan - rex->program), PL_regindent*2, "",
2579 SvPVX_const(prop),
1de06328
YO
2580 (PL_regkind[OP(scan)] == END || !rnext) ?
2581 0 : (IV)(rnext - rex->program));
2a782b5b 2582 });
d6a28714
JH
2583
2584 next = scan + NEXT_OFF(scan);
2585 if (next == scan)
2586 next = NULL;
40a82448 2587 state_num = OP(scan);
d6a28714 2588
40a82448
DM
2589 reenter_switch:
2590 switch (state_num) {
d6a28714 2591 case BOL:
7fba1cd6 2592 if (locinput == PL_bostr)
d6a28714 2593 {
3b0527fe 2594 /* reginfo->till = reginfo->bol; */
b8c5462f
JH
2595 break;
2596 }
d6a28714
JH
2597 sayNO;
2598 case MBOL:
12d33761
HS
2599 if (locinput == PL_bostr ||
2600 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
d6a28714 2601 {
b8c5462f
JH
2602 break;
2603 }
d6a28714
JH
2604 sayNO;
2605 case SBOL:
c2a73568 2606 if (locinput == PL_bostr)
b8c5462f 2607 break;
d6a28714
JH
2608 sayNO;
2609 case GPOS:
3b0527fe 2610 if (locinput == reginfo->ganch)
d6a28714
JH
2611 break;
2612 sayNO;
2613 case EOL:
d6a28714
JH
2614 goto seol;
2615 case MEOL:
d6a28714 2616 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 2617 sayNO;
b8c5462f 2618 break;
d6a28714
JH
2619 case SEOL:
2620 seol:
2621 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 2622 sayNO;
d6a28714 2623 if (PL_regeol - locinput > 1)
b8c5462f 2624 sayNO;
b8c5462f 2625 break;
d6a28714
JH
2626 case EOS:
2627 if (PL_regeol != locinput)
b8c5462f 2628 sayNO;
d6a28714 2629 break;
ffc61ed2 2630 case SANY:
d6a28714 2631 if (!nextchr && locinput >= PL_regeol)
4633a7c4 2632 sayNO;
f33976b4
DB
2633 if (do_utf8) {
2634 locinput += PL_utf8skip[nextchr];
2635 if (locinput > PL_regeol)
2636 sayNO;
2637 nextchr = UCHARAT(locinput);
2638 }
2639 else
2640 nextchr = UCHARAT(++locinput);
2641 break;
2642 case CANY:
2643 if (!nextchr && locinput >= PL_regeol)
2644 sayNO;
b8c5462f 2645 nextchr = UCHARAT(++locinput);
a0d0e21e 2646 break;
ffc61ed2 2647 case REG_ANY:
1aa99e6b
IH
2648 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
2649 sayNO;
2650 if (do_utf8) {
b8c5462f 2651 locinput += PL_utf8skip[nextchr];
d6a28714
JH
2652 if (locinput > PL_regeol)
2653 sayNO;
a0ed51b3 2654 nextchr = UCHARAT(locinput);
a0ed51b3 2655 }
1aa99e6b
IH
2656 else
2657 nextchr = UCHARAT(++locinput);
a0ed51b3 2658 break;
166ba7cd
DM
2659
2660#undef ST
2661#define ST st->u.trie
786e8c11
YO
2662 case TRIEC:
2663 /* In this case the charclass data is available inline so
2664 we can fail fast without a lot of extra overhead.
2665 */
2666 if (scan->flags == EXACT || !do_utf8) {
2667 if(!ANYOF_BITMAP_TEST(scan, *locinput)) {
2668 DEBUG_EXECUTE_r(
2669 PerlIO_printf(Perl_debug_log,
2670 "%*s %sfailed to match trie start class...%s\n",
2671 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5])
2672 );
2673 sayNO_SILENT;
2674 /* NOTREACHED */
2675 }
2676 }
2677 /* FALL THROUGH */
5b47454d 2678 case TRIE:
3dab1dad 2679 {
07be1b83 2680 /* what type of TRIE am I? (utf8 makes this contextual) */
3dab1dad
YO
2681 const enum { trie_plain, trie_utf8, trie_utf8_fold }
2682 trie_type = do_utf8 ?
2683 (scan->flags == EXACT ? trie_utf8 : trie_utf8_fold)
2684 : trie_plain;
2685
2686 /* what trie are we using right now */
be8e71aa 2687 reg_trie_data * const trie
3dab1dad
YO
2688 = (reg_trie_data*)rex->data->data[ ARG( scan ) ];
2689 U32 state = trie->startstate;
166ba7cd 2690
3dab1dad
YO
2691 if (trie->bitmap && trie_type != trie_utf8_fold &&
2692 !TRIE_BITMAP_TEST(trie,*locinput)
2693 ) {
2694 if (trie->states[ state ].wordnum) {
2695 DEBUG_EXECUTE_r(
2696 PerlIO_printf(Perl_debug_log,
2697 "%*s %smatched empty string...%s\n",
2698 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5])
2699 );
2700 break;
2701 } else {
2702 DEBUG_EXECUTE_r(
2703 PerlIO_printf(Perl_debug_log,
786e8c11 2704 "%*s %sfailed to match trie start class...%s\n",
3dab1dad
YO
2705 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5])
2706 );
2707 sayNO_SILENT;
2708 }
2709 }
166ba7cd 2710
786e8c11
YO
2711 {
2712 U8 *uc = ( U8* )locinput;
2713
2714 STRLEN len = 0;
2715 STRLEN foldlen = 0;
2716 U8 *uscan = (U8*)NULL;
2717 STRLEN bufflen=0;
2718 SV *sv_accept_buff = NULL;
2719 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2720
2721 ST.accepted = 0; /* how many accepting states we have seen */
2722 ST.B = next;
2723 ST.jump = trie->jump;
2724
2725#ifdef DEBUGGING
2726 ST.me = scan;
2727#endif
2728
2729
2730
07be1b83
YO
2731 /*
2732 traverse the TRIE keeping track of all accepting states
2733 we transition through until we get to a failing node.
2734 */
2735
a3621e74 2736 while ( state && uc <= (U8*)PL_regeol ) {
786e8c11
YO
2737 U32 base = trie->states[ state ].trans.base;
2738 UV uvc;
2739 U16 charid;
2740 /* We use charid to hold the wordnum as we don't use it
2741 for charid until after we have done the wordnum logic.
2742 We define an alias just so that the wordnum logic reads
2743 more naturally. */
2744
2745#define got_wordnum charid
2746 got_wordnum = trie->states[ state ].wordnum;
2747
2748 if ( got_wordnum ) {
2749 if ( ! ST.accepted ) {
5b47454d
DM
2750 ENTER;
2751 SAVETMPS;
2752 bufflen = TRIE_INITAL_ACCEPT_BUFFLEN;
2753 sv_accept_buff=newSV(bufflen *
2754 sizeof(reg_trie_accepted) - 1);
786e8c11 2755 SvCUR_set(sv_accept_buff, 0);
5b47454d
DM
2756 SvPOK_on(sv_accept_buff);
2757 sv_2mortal(sv_accept_buff);
166ba7cd
DM
2758 SAVETMPS;
2759 ST.accept_buff =
5b47454d
DM
2760 (reg_trie_accepted*)SvPV_nolen(sv_accept_buff );
2761 }
786e8c11 2762 do {
166ba7cd 2763 if (ST.accepted >= bufflen) {
5b47454d 2764 bufflen *= 2;
166ba7cd 2765 ST.accept_buff =(reg_trie_accepted*)
5b47454d
DM
2766 SvGROW(sv_accept_buff,
2767 bufflen * sizeof(reg_trie_accepted));
2768 }
2769 SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff)
2770 + sizeof(reg_trie_accepted));
a3621e74 2771
786e8c11
YO
2772
2773 ST.accept_buff[ST.accepted].wordnum = got_wordnum;
2774 ST.accept_buff[ST.accepted].endpos = uc;
2775 ++ST.accepted;
2776 } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum]));
2777 }
2778#undef got_wordnum
a3621e74 2779
07be1b83 2780 DEBUG_TRIE_EXECUTE_r({
786e8c11 2781 DUMP_EXEC_POS( (char *)uc, scan, do_utf8 );
a3621e74 2782 PerlIO_printf( Perl_debug_log,
786e8c11 2783 "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ",
07be1b83 2784 2+PL_regindent * 2, "", PL_colors[4],
786e8c11 2785 (UV)state, (UV)ST.accepted );
07be1b83 2786 });
a3621e74
YO
2787
2788 if ( base ) {
4cadc6a9
YO
2789 REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len,
2790 uvc, charid, foldlen, foldbuf, uniflags);
a3621e74 2791
5b47454d
DM
2792 if (charid &&
2793 (base + charid > trie->uniquecharcount )
2794 && (base + charid - 1 - trie->uniquecharcount
2795 < trie->lasttrans)
2796 && trie->trans[base + charid - 1 -
2797 trie->uniquecharcount].check == state)
2798 {
2799 state = trie->trans[base + charid - 1 -
2800 trie->uniquecharcount ].next;
2801 }
2802 else {
2803 state = 0;
2804 }
2805 uc += len;
2806
2807 }
2808 else {
a3621e74
YO
2809 state = 0;
2810 }
2811 DEBUG_TRIE_EXECUTE_r(
e4584336 2812 PerlIO_printf( Perl_debug_log,
786e8c11 2813 "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
e4584336 2814 charid, uvc, (UV)state, PL_colors[5] );
a3621e74
YO
2815 );
2816 }
166ba7cd 2817 if (!ST.accepted )
a3621e74 2818 sayNO;
a3621e74 2819
166ba7cd
DM
2820 DEBUG_EXECUTE_r(
2821 PerlIO_printf( Perl_debug_log,
2822 "%*s %sgot %"IVdf" possible matches%s\n",
2823 REPORT_CODE_OFF + PL_regindent * 2, "",
2824 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
2825 );
786e8c11 2826 }}
166ba7cd
DM
2827
2828 /* FALL THROUGH */
2829
2830 case TRIE_next_fail: /* we failed - try next alterative */
2831
2832 if ( ST.accepted == 1 ) {
2833 /* only one choice left - just continue */
2834 DEBUG_EXECUTE_r({
2835 reg_trie_data * const trie
2836 = (reg_trie_data*)rex->data->data[ ARG(ST.me) ];
2837 SV ** const tmp = RX_DEBUG(reginfo->prog)
2838 ? av_fetch( trie->words, ST.accept_buff[ 0 ].wordnum-1, 0 )
2839 : NULL;
2840 PerlIO_printf( Perl_debug_log,
2841 "%*s %sonly one match left: #%d <%s>%s\n",
2842 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],
2843 ST.accept_buff[ 0 ].wordnum,
2844 tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr",
2845 PL_colors[5] );
2846 });
2847 PL_reginput = (char *)ST.accept_buff[ 0 ].endpos;
2848 /* in this case we free tmps/leave before we call regmatch
2849 as we wont be using accept_buff again. */
2850 FREETMPS;
2851 LEAVE;
2852 locinput = PL_reginput;
2853 nextchr = UCHARAT(locinput);
786e8c11
YO
2854
2855 if ( !ST.jump )
2856 scan = ST.B;
2857 else
2858 scan = ST.B - ST.jump[ST.accept_buff[0].wordnum];
2859
166ba7cd
DM
2860 continue; /* execute rest of RE */
2861 }
2862
2863 if (!ST.accepted-- ) {
2864 FREETMPS;
2865 LEAVE;
2866 sayNO;
2867 }
2868
a3621e74 2869 /*
166ba7cd
DM
2870 There are at least two accepting states left. Presumably
2871 the number of accepting states is going to be low,
2872 typically two. So we simply scan through to find the one
2873 with lowest wordnum. Once we find it, we swap the last
2874 state into its place and decrement the size. We then try to
2875 match the rest of the pattern at the point where the word
2876 ends. If we succeed, control just continues along the
2877 regex; if we fail we return here to try the next accepting
2878 state
2879 */
a3621e74 2880
166ba7cd
DM
2881 {
2882 U32 best = 0;
2883 U32 cur;
2884 for( cur = 1 ; cur <= ST.accepted ; cur++ ) {
2885 DEBUG_TRIE_EXECUTE_r(
f2278c82 2886 PerlIO_printf( Perl_debug_log,
166ba7cd
DM
2887 "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n",
2888 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
2889 (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur,
2890 ST.accept_buff[ cur ].wordnum, PL_colors[5] );
2891 );
2892
2893 if (ST.accept_buff[cur].wordnum <
2894 ST.accept_buff[best].wordnum)
2895 best = cur;
a3621e74 2896 }
166ba7cd
DM
2897
2898 DEBUG_EXECUTE_r({
2899 reg_trie_data * const trie
2900 = (reg_trie_data*)rex->data->data[ ARG(ST.me) ];
2901 SV ** const tmp = RX_DEBUG(reginfo->prog)
2902 ? av_fetch( trie->words, ST.accept_buff[ best ].wordnum - 1, 0 )
2903 : NULL;
2904 PerlIO_printf( Perl_debug_log, "%*s %strying alternation #%d <%s> at node #%d %s\n",
2905 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],
2906 ST.accept_buff[best].wordnum,
2907 tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", REG_NODE_NUM(scan),
2908 PL_colors[5] );
2909 });
2910
2911 if ( best<ST.accepted ) {
2912 reg_trie_accepted tmp = ST.accept_buff[ best ];
2913 ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ];
2914 ST.accept_buff[ ST.accepted ] = tmp;
2915 best = ST.accepted;
a3621e74 2916 }
166ba7cd 2917 PL_reginput = (char *)ST.accept_buff[ best ].endpos;
786e8c11
YO
2918 if ( !ST.jump ) {
2919 PUSH_STATE_GOTO(TRIE_next, ST.B);
2920 /* NOTREACHED */
2921 } else {
2922 PUSH_STATE_GOTO(TRIE_next, ST.B - ST.jump[ST.accept_buff[best].wordnum]);
2923 /* NOTREACHED */
2924 }
2925 /* NOTREACHED */
166ba7cd 2926 }
166ba7cd
DM
2927 /* NOTREACHED */
2928
2929#undef ST
2930
95b24440
DM
2931 case EXACT: {
2932 char *s = STRING(scan);
5d9a96ca 2933 st->ln = STR_LEN(scan);
eb160463 2934 if (do_utf8 != UTF) {
bc517b45 2935 /* The target and the pattern have differing utf8ness. */
1aa99e6b 2936 char *l = locinput;
be8e71aa 2937 const char * const e = s + st->ln;
a72c7584 2938
5ff6fc6d
JH
2939 if (do_utf8) {
2940 /* The target is utf8, the pattern is not utf8. */
1aa99e6b 2941 while (s < e) {
a3b680e6 2942 STRLEN ulen;
1aa99e6b 2943 if (l >= PL_regeol)
5ff6fc6d
JH
2944 sayNO;
2945 if (NATIVE_TO_UNI(*(U8*)s) !=
89ebb4a3 2946 utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
041457d9 2947 uniflags))
5ff6fc6d 2948 sayNO;
bc517b45 2949 l += ulen;
5ff6fc6d 2950 s ++;
1aa99e6b 2951 }
5ff6fc6d
JH
2952 }
2953 else {
2954 /* The target is not utf8, the pattern is utf8. */
1aa99e6b 2955 while (s < e) {
a3b680e6 2956 STRLEN ulen;
1aa99e6b
IH
2957 if (l >= PL_regeol)
2958 sayNO;
5ff6fc6d 2959 if (NATIVE_TO_UNI(*((U8*)l)) !=
89ebb4a3 2960 utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
041457d9 2961 uniflags))
1aa99e6b 2962 sayNO;
bc517b45 2963 s += ulen;
a72c7584 2964 l ++;
1aa99e6b 2965 }
5ff6fc6d 2966 }
1aa99e6b
IH
2967 locinput = l;
2968 nextchr = UCHARAT(locinput);
2969 break;
2970 }
bc517b45 2971 /* The target and the pattern have the same utf8ness. */
d6a28714
JH
2972 /* Inline the first character, for speed. */
2973 if (UCHARAT(s) != nextchr)
2974 sayNO;
5d9a96ca 2975 if (PL_regeol - locinput < st->ln)
d6a28714 2976 sayNO;
5d9a96ca 2977 if (st->ln > 1 && memNE(s, locinput, st->ln))
d6a28714 2978 sayNO;
5d9a96ca 2979 locinput += st->ln;
d6a28714
JH
2980 nextchr = UCHARAT(locinput);
2981 break;
95b24440 2982 }
d6a28714 2983 case EXACTFL:
b8c5462f
JH
2984 PL_reg_flags |= RF_tainted;
2985 /* FALL THROUGH */
95b24440 2986 case EXACTF: {
be8e71aa 2987 char * const s = STRING(scan);
5d9a96ca 2988 st->ln = STR_LEN(scan);
d6a28714 2989
d07ddd77
JH
2990 if (do_utf8 || UTF) {
2991 /* Either target or the pattern are utf8. */
be8e71aa 2992 const char * const l = locinput;
d07ddd77 2993 char *e = PL_regeol;
bc517b45 2994
5d9a96ca 2995 if (ibcmp_utf8(s, 0, st->ln, (bool)UTF,
1feea2c7 2996 l, &e, 0, do_utf8)) {
5486206c
JH
2997 /* One more case for the sharp s:
2998 * pack("U0U*", 0xDF) =~ /ss/i,
2999 * the 0xC3 0x9F are the UTF-8
3000 * byte sequence for the U+00DF. */
3001 if (!(do_utf8 &&
3002 toLOWER(s[0]) == 's' &&
5d9a96ca 3003 st->ln >= 2 &&
5486206c
JH
3004 toLOWER(s[1]) == 's' &&
3005 (U8)l[0] == 0xC3 &&
3006 e - l >= 2 &&
3007 (U8)l[1] == 0x9F))
3008 sayNO;
3009 }
d07ddd77
JH
3010 locinput = e;
3011 nextchr = UCHARAT(locinput);
3012 break;
a0ed51b3 3013 }
d6a28714 3014
bc517b45
JH
3015 /* Neither the target and the pattern are utf8. */
3016
d6a28714
JH
3017 /* Inline the first character, for speed. */
3018 if (UCHARAT(s) != nextchr &&
3019 UCHARAT(s) != ((OP(scan) == EXACTF)
3020 ? PL_fold : PL_fold_locale)[nextchr])
a0ed51b3 3021 sayNO;
5d9a96ca 3022 if (PL_regeol - locinput < st->ln)
b8c5462f 3023 sayNO;
5d9a96ca
DM
3024 if (st->ln > 1 && (OP(scan) == EXACTF
3025 ? ibcmp(s, locinput, st->ln)
3026 : ibcmp_locale(s, locinput, st->ln)))
4633a7c4 3027 sayNO;
5d9a96ca 3028 locinput += st->ln;
d6a28714 3029 nextchr = UCHARAT(locinput);
a0d0e21e 3030 break;
95b24440 3031 }
d6a28714 3032 case ANYOF:
ffc61ed2 3033 if (do_utf8) {
9e55ce06
JH
3034 STRLEN inclasslen = PL_regeol - locinput;
3035
32fc9b6a 3036 if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8))
e0f9d4a8 3037 sayNO_ANYOF;
ffc61ed2
JH
3038 if (locinput >= PL_regeol)
3039 sayNO;
0f0076b4 3040 locinput += inclasslen ? inclasslen : UTF8SKIP(locinput);
b8c5462f 3041 nextchr = UCHARAT(locinput);
e0f9d4a8 3042 break;
ffc61ed2
JH
3043 }
3044 else {
3045 if (nextchr < 0)
3046 nextchr = UCHARAT(locinput);
32fc9b6a 3047 if (!REGINCLASS(rex, scan, (U8*)locinput))
e0f9d4a8 3048 sayNO_ANYOF;
ffc61ed2
JH
3049 if (!nextchr && locinput >= PL_regeol)
3050 sayNO;
3051 nextchr = UCHARAT(++locinput);
e0f9d4a8
JH
3052 break;
3053 }
3054 no_anyof:
3055 /* If we might have the case of the German sharp s
3056 * in a casefolding Unicode character class. */
3057
ebc501f0
JH
3058 if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
3059 locinput += SHARP_S_SKIP;
e0f9d4a8 3060 nextchr = UCHARAT(locinput);
ffc61ed2 3061 }
e0f9d4a8
JH
3062 else
3063 sayNO;
b8c5462f 3064 break;
d6a28714 3065 case ALNUML:
b8c5462f
JH
3066 PL_reg_flags |= RF_tainted;
3067 /* FALL THROUGH */
d6a28714 3068 case ALNUM:
b8c5462f 3069 if (!nextchr)
4633a7c4 3070 sayNO;
ffc61ed2 3071 if (do_utf8) {
1a4fad37 3072 LOAD_UTF8_CHARCLASS_ALNUM();
ffc61ed2 3073 if (!(OP(scan) == ALNUM
bb7a0f54 3074 ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
d6a28714 3075 : isALNUM_LC_utf8((U8*)locinput)))
b8c5462f
JH
3076 {
3077 sayNO;
a0ed51b3 3078 }
b8c5462f 3079 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
3080 nextchr = UCHARAT(locinput);
3081 break;
3082 }
ffc61ed2 3083 if (!(OP(scan) == ALNUM
d6a28714 3084 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
bbce6d69 3085 sayNO;
b8c5462f 3086 nextchr = UCHARAT(++locinput);
a0d0e21e 3087 break;
d6a28714 3088 case NALNUML:
b8c5462f
JH
3089 PL_reg_flags |= RF_tainted;
3090 /* FALL THROUGH */
d6a28714
JH
3091 case NALNUM:
3092 if (!nextchr && locinput >= PL_regeol)
a0ed51b3 3093 sayNO;
ffc61ed2 3094 if (do_utf8) {
1a4fad37 3095 LOAD_UTF8_CHARCLASS_ALNUM();
ffc61ed2 3096 if (OP(scan) == NALNUM
bb7a0f54 3097 ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
d6a28714
JH
3098 : isALNUM_LC_utf8((U8*)locinput))
3099 {
b8c5462f 3100 sayNO;
d6a28714 3101 }
b8c5462f
JH
3102 locinput += PL_utf8skip[nextchr];
3103 nextchr = UCHARAT(locinput);
3104 break;
3105 }
ffc61ed2 3106 if (OP(scan) == NALNUM
d6a28714 3107 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
b8c5462f 3108 sayNO;
76e3520e 3109 nextchr = UCHARAT(++locinput);
a0d0e21e 3110 break;
d6a28714
JH
3111 case BOUNDL:
3112 case NBOUNDL:
3280af22 3113 PL_reg_flags |= RF_tainted;
bbce6d69 3114 /* FALL THROUGH */
d6a28714
JH
3115 case BOUND:
3116 case NBOUND:
3117 /* was last char in word? */
ffc61ed2 3118 if (do_utf8) {
12d33761 3119 if (locinput == PL_bostr)
5d9a96ca 3120 st->ln = '\n';
ffc61ed2 3121 else {
a3b680e6 3122 const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
9041c2e3 3123
4ad0818d 3124 st->ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags);
ffc61ed2
JH
3125 }
3126 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
5d9a96ca 3127 st->ln = isALNUM_uni(st->ln);
1a4fad37 3128 LOAD_UTF8_CHARCLASS_ALNUM();
3568d838 3129 n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8);
ffc61ed2
JH
3130 }
3131 else {
5d9a96ca 3132 st->ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(st->ln));
ffc61ed2
JH
3133 n = isALNUM_LC_utf8((U8*)locinput);
3134 }
a0ed51b3 3135 }
d6a28714 3136 else {
5d9a96ca 3137 st->ln = (locinput != PL_bostr) ?
12d33761 3138 UCHARAT(locinput - 1) : '\n';
ffc61ed2 3139 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
5d9a96ca 3140 st->ln = isALNUM(st->ln);
ffc61ed2
JH
3141 n = isALNUM(nextchr);
3142 }
3143 else {
5d9a96ca 3144 st->ln = isALNUM_LC(st->ln);
ffc61ed2
JH
3145 n = isALNUM_LC(nextchr);
3146 }
d6a28714 3147 }
5d9a96ca 3148 if (((!st->ln) == (!n)) == (OP(scan) == BOUND ||
ffc61ed2
JH
3149 OP(scan) == BOUNDL))
3150 sayNO;
a0ed51b3 3151 break;
d6a28714 3152 case SPACEL:
3280af22 3153 PL_reg_flags |= RF_tainted;
bbce6d69 3154 /* FALL THROUGH */
d6a28714 3155 case SPACE:
9442cb0e 3156 if (!nextchr)
4633a7c4 3157 sayNO;
1aa99e6b 3158 if (do_utf8) {
fd400ab9 3159 if (UTF8_IS_CONTINUED(nextchr)) {
1a4fad37 3160 LOAD_UTF8_CHARCLASS_SPACE();
ffc61ed2 3161 if (!(OP(scan) == SPACE
bb7a0f54 3162 ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
ffc61ed2
JH
3163 : isSPACE_LC_utf8((U8*)locinput)))
3164 {
3165 sayNO;
3166 }
3167 locinput += PL_utf8skip[nextchr];
3168 nextchr = UCHARAT(locinput);
3169 break;
d6a28714 3170 }
ffc61ed2
JH
3171 if (!(OP(scan) == SPACE
3172 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
3173 sayNO;
3174 nextchr = UCHARAT(++locinput);
3175 }
3176 else {
3177 if (!(OP(scan) == SPACE
3178 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
3179 sayNO;
3180 nextchr = UCHARAT(++locinput);
a0ed51b3 3181 }
a0ed51b3 3182 break;
d6a28714 3183 case NSPACEL:
3280af22 3184 PL_reg_flags |= RF_tainted;
bbce6d69 3185 /* FALL THROUGH */
d6a28714 3186 case NSPACE:
9442cb0e 3187 if (!nextchr && locinput >= PL_regeol)
b8c5462f 3188 sayNO;
1aa99e6b 3189 if (do_utf8) {
1a4fad37 3190 LOAD_UTF8_CHARCLASS_SPACE();
ffc61ed2 3191 if (OP(scan) == NSPACE
bb7a0f54 3192 ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
d6a28714 3193 : isSPACE_LC_utf8((U8*)locinput))
b8c5462f
JH
3194 {
3195 sayNO;
3196 }
3197 locinput += PL_utf8skip[nextchr];
3198 nextchr = UCHARAT(locinput);
3199 break;
a0ed51b3 3200 }
ffc61ed2 3201 if (OP(scan) == NSPACE
d6a28714 3202 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
4633a7c4 3203 sayNO;
76e3520e 3204 nextchr = UCHARAT(++locinput);
a0d0e21e 3205 break;
d6a28714 3206 case DIGITL:
a0ed51b3
LW
3207 PL_reg_flags |= RF_tainted;
3208 /* FALL THROUGH */
d6a28714 3209 case DIGIT:
9442cb0e 3210 if (!nextchr)
a0ed51b3 3211 sayNO;
1aa99e6b 3212 if (do_utf8) {
1a4fad37 3213 LOAD_UTF8_CHARCLASS_DIGIT();
ffc61ed2 3214 if (!(OP(scan) == DIGIT
bb7a0f54 3215 ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
9442cb0e 3216 : isDIGIT_LC_utf8((U8*)locinput)))
dfe13c55 3217 {
a0ed51b3 3218 sayNO;
dfe13c55 3219 }
6f06b55f 3220 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
3221 nextchr = UCHARAT(locinput);
3222 break;
3223 }
ffc61ed2 3224 if (!(OP(scan) == DIGIT
9442cb0e 3225 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
a0ed51b3
LW
3226 sayNO;
3227 nextchr = UCHARAT(++locinput);
3228 break;
d6a28714 3229 case NDIGITL:
b8c5462f
JH
3230 PL_reg_flags |= RF_tainted;
3231 /* FALL THROUGH */
d6a28714 3232 case NDIGIT:
9442cb0e 3233 if (!nextchr && locinput >= PL_regeol)
b8c5462f 3234 sayNO;
1aa99e6b 3235 if (do_utf8) {
1a4fad37 3236 LOAD_UTF8_CHARCLASS_DIGIT();
ffc61ed2 3237 if (OP(scan) == NDIGIT
bb7a0f54 3238 ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
9442cb0e
GS
3239 : isDIGIT_LC_utf8((U8*)locinput))
3240 {
a0ed51b3 3241 sayNO;
9442cb0e 3242 }
6f06b55f 3243 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
3244 nextchr = UCHARAT(locinput);
3245 break;
3246 }
ffc61ed2 3247 if (OP(scan) == NDIGIT
9442cb0e 3248 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
a0ed51b3
LW
3249 sayNO;
3250 nextchr = UCHARAT(++locinput);
3251 break;
3252 case CLUMP:
b7c83a7e 3253 if (locinput >= PL_regeol)
a0ed51b3 3254 sayNO;
b7c83a7e 3255 if (do_utf8) {
1a4fad37 3256 LOAD_UTF8_CHARCLASS_MARK();
b7c83a7e
JH
3257 if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
3258 sayNO;
3259 locinput += PL_utf8skip[nextchr];
3260 while (locinput < PL_regeol &&
3261 swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
3262 locinput += UTF8SKIP(locinput);
3263 if (locinput > PL_regeol)
3264 sayNO;
eb08e2da
JH
3265 }
3266 else
3267 locinput++;
a0ed51b3
LW
3268 nextchr = UCHARAT(locinput);
3269 break;
c8756f30 3270 case REFFL:
3280af22 3271 PL_reg_flags |= RF_tainted;
c8756f30 3272 /* FALL THROUGH */
c277df42 3273 case REF:
95b24440
DM
3274 case REFF: {
3275 char *s;
c277df42 3276 n = ARG(scan); /* which paren pair */
5d9a96ca 3277 st->ln = PL_regstartp[n];
2c2d71f5 3278 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
5d9a96ca 3279 if ((I32)*PL_reglastparen < n || st->ln == -1)
af3f8c16 3280 sayNO; /* Do not match unless seen CLOSEn. */
5d9a96ca 3281 if (st->ln == PL_regendp[n])
a0d0e21e 3282 break;
a0ed51b3 3283
5d9a96ca 3284 s = PL_bostr + st->ln;
1aa99e6b 3285 if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */
a0ed51b3 3286 char *l = locinput;
a3b680e6 3287 const char *e = PL_bostr + PL_regendp[n];
a0ed51b3
LW
3288 /*
3289 * Note that we can't do the "other character" lookup trick as
3290 * in the 8-bit case (no pun intended) because in Unicode we
3291 * have to map both upper and title case to lower case.
3292 */
3293 if (OP(scan) == REFF) {
3294 while (s < e) {
a3b680e6
AL
3295 STRLEN ulen1, ulen2;
3296 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
3297 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
3298
a0ed51b3
LW
3299 if (l >= PL_regeol)
3300 sayNO;
a2a2844f
JH
3301 toLOWER_utf8((U8*)s, tmpbuf1, &ulen1);
3302 toLOWER_utf8((U8*)l, tmpbuf2, &ulen2);
7114a2d2 3303 if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
a0ed51b3 3304 sayNO;
a2a2844f
JH
3305 s += ulen1;
3306 l += ulen2;
a0ed51b3
LW
3307 }
3308 }
3309 locinput = l;
3310 nextchr = UCHARAT(locinput);
3311 break;
3312 }
3313
a0d0e21e 3314 /* Inline the first character, for speed. */
76e3520e 3315 if (UCHARAT(s) != nextchr &&
c8756f30
AK
3316 (OP(scan) == REF ||
3317 (UCHARAT(s) != ((OP(scan) == REFF
22c35a8c 3318 ? PL_fold : PL_fold_locale)[nextchr]))))
4633a7c4 3319 sayNO;
5d9a96ca
DM
3320 st->ln = PL_regendp[n] - st->ln;
3321 if (locinput + st->ln > PL_regeol)
4633a7c4 3322 sayNO;
5d9a96ca
DM
3323 if (st->ln > 1 && (OP(scan) == REF
3324 ? memNE(s, locinput, st->ln)
c8756f30 3325 : (OP(scan) == REFF
5d9a96ca
DM
3326 ? ibcmp(s, locinput, st->ln)
3327 : ibcmp_locale(s, locinput, st->ln))))
4633a7c4 3328 sayNO;
5d9a96ca 3329 locinput += st->ln;
76e3520e 3330 nextchr = UCHARAT(locinput);
a0d0e21e 3331 break;
95b24440 3332 }
a0d0e21e
LW
3333
3334 case NOTHING:
c277df42 3335 case TAIL:
a0d0e21e
LW
3336 break;
3337 case BACK:
3338 break;
40a82448
DM
3339
3340#undef ST
3341#define ST st->u.eval
3342
3343 case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */
c277df42 3344 {
c277df42 3345 SV *ret;
8e5e9ebe 3346 {
4aabdb9b
DM
3347 /* execute the code in the {...} */
3348 dSP;
6136c704 3349 SV ** const before = SP;
4aabdb9b
DM
3350 OP_4tree * const oop = PL_op;
3351 COP * const ocurcop = PL_curcop;
3352 PAD *old_comppad;
4aabdb9b
DM
3353
3354 n = ARG(scan);
32fc9b6a 3355 PL_op = (OP_4tree*)rex->data->data[n];
4aabdb9b 3356 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
32fc9b6a 3357 PAD_SAVE_LOCAL(old_comppad, (PAD*)rex->data->data[n + 2]);
4aabdb9b
DM
3358 PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
3359
8e5e9ebe
RGS
3360 CALLRUNOPS(aTHX); /* Scalar context. */
3361 SPAGAIN;
3362 if (SP == before)
075aa684 3363 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
8e5e9ebe
RGS
3364 else {
3365 ret = POPs;
3366 PUTBACK;
3367 }
4aabdb9b
DM
3368
3369 PL_op = oop;
3370 PAD_RESTORE_LOCAL(old_comppad);
3371 PL_curcop = ocurcop;
3372 if (!st->logical) {
3373 /* /(?{...})/ */
3374 sv_setsv(save_scalar(PL_replgv), ret);
4aabdb9b
DM
3375 break;
3376 }
8e5e9ebe 3377 }
4aabdb9b
DM
3378 if (st->logical == 2) { /* Postponed subexpression: /(??{...})/ */
3379 regexp *re;
4aabdb9b 3380 {
4f639d21
DM
3381 /* extract RE object from returned value; compiling if
3382 * necessary */
3383
6136c704 3384 MAGIC *mg = NULL;
be8e71aa 3385 const SV *sv;
faf82a0b
AE
3386 if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret)))
3387 mg = mg_find(sv, PERL_MAGIC_qr);
3388 else if (SvSMAGICAL(ret)) {
3389 if (SvGMAGICAL(ret))
3390 sv_unmagic(ret, PERL_MAGIC_qr);
3391 else
3392 mg = mg_find(ret, PERL_MAGIC_qr);
0f5d15d6 3393 }
faf82a0b 3394
0f5d15d6
IZ
3395 if (mg) {
3396 re = (regexp *)mg->mg_obj;
df0003d4 3397 (void)ReREFCNT_inc(re);
0f5d15d6
IZ
3398 }
3399 else {
3400 STRLEN len;
6136c704 3401 const char * const t = SvPV_const(ret, len);
0f5d15d6 3402 PMOP pm;
a3b680e6 3403 const I32 osize = PL_regsize;
0f5d15d6 3404
5fcd1c1b 3405 Zero(&pm, 1, PMOP);
4aabdb9b 3406 if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8;
83003860 3407 re = CALLREGCOMP(aTHX_ (char*)t, (char*)t + len, &pm);
9041c2e3 3408 if (!(SvFLAGS(ret)
faf82a0b
AE
3409 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
3410 | SVs_GMG)))
14befaf4
DM
3411 sv_magic(ret,(SV*)ReREFCNT_inc(re),
3412 PERL_MAGIC_qr,0,0);
0f5d15d6 3413 PL_regsize = osize;
0f5d15d6 3414 }
4aabdb9b 3415 }
aa283a38
DM
3416
3417 /* run the pattern returned from (??{...}) */
40a82448
DM
3418 ST.cp = regcppush(0); /* Save *all* the positions. */
3419 REGCP_SET(ST.lastcp);
4aabdb9b
DM
3420 *PL_reglastparen = 0;
3421 *PL_reglastcloseparen = 0;
4aabdb9b 3422 PL_reginput = locinput;
4aabdb9b
DM
3423
3424 /* XXXX This is too dramatic a measure... */
3425 PL_reg_maxiter = 0;
3426
5d9a96ca 3427 st->logical = 0;
40a82448 3428 ST.toggleutf = ((PL_reg_flags & RF_utf8) != 0) ^
aa283a38 3429 ((re->reganch & ROPT_UTF8) != 0);
40a82448
DM
3430 if (ST.toggleutf) PL_reg_flags ^= RF_utf8;
3431 ST.prev_rex = rex;
aa283a38 3432 rex = re;
aa283a38 3433
40a82448 3434 ST.B = next;
1de06328
YO
3435 DEBUG_EXECUTE_r(
3436 debug_start_match(re, do_utf8, locinput, PL_regeol,
3437 "Matching embedded");
3438 );
aa283a38 3439 /* now continue from first node in postoned RE */
40a82448 3440 PUSH_YES_STATE_GOTO(EVAL_A, re->program + 1);
4aabdb9b 3441 /* NOTREACHED */
a0ed51b3 3442 }
4aabdb9b
DM
3443 /* /(?(?{...})X|Y)/ */
3444 st->sw = SvTRUE(ret);
3445 st->logical = 0;