This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
document several optree construction functions
[perl5.git] / regexec.c
CommitLineData
a0d0e21e
LW
1/* regexec.c
2 */
3
4/*
4ac71550
TC
5 * One Ring to rule them all, One Ring to find them
6 &
7 * [p.v of _The Lord of the Rings_, opening poem]
8 * [p.50 of _The Lord of the Rings_, I/iii: "The Shadow of the Past"]
9 * [p.254 of _The Lord of the Rings_, II/ii: "The Council of Elrond"]
a0d0e21e
LW
10 */
11
61296642
DM
12/* This file contains functions for executing a regular expression. See
13 * also regcomp.c which funnily enough, contains functions for compiling
166f8a29 14 * a regular expression.
e4a054ea
DM
15 *
16 * This file is also copied at build time to ext/re/re_exec.c, where
17 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
18 * This causes the main functions to be compiled under new names and with
19 * debugging support added, which makes "use re 'debug'" work.
166f8a29
DM
20 */
21
a687059c
LW
22/* NOTE: this is derived from Henry Spencer's regexp code, and should not
23 * confused with the original package (see point 3 below). Thanks, Henry!
24 */
25
26/* Additional note: this code is very heavily munged from Henry's version
27 * in places. In some spots I've traded clarity for efficiency, so don't
28 * blame Henry for some of the lack of readability.
29 */
30
e50aee73
AD
31/* The names of the functions have been changed from regcomp and
32 * regexec to pregcomp and pregexec in order to avoid conflicts
33 * with the POSIX routines of the same names.
34*/
35
b9d5759e 36#ifdef PERL_EXT_RE_BUILD
54df2634 37#include "re_top.h"
9041c2e3 38#endif
56953603 39
a687059c 40/*
e50aee73 41 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
42 *
43 * Copyright (c) 1986 by University of Toronto.
44 * Written by Henry Spencer. Not derived from licensed software.
45 *
46 * Permission is granted to anyone to use this software for any
47 * purpose on any computer system, and to redistribute it freely,
48 * subject to the following restrictions:
49 *
50 * 1. The author is not responsible for the consequences of use of
51 * this software, no matter how awful, even if they arise
52 * from defects in it.
53 *
54 * 2. The origin of this software must not be misrepresented, either
55 * by explicit claim or by omission.
56 *
57 * 3. Altered versions must be plainly marked as such, and must not
58 * be misrepresented as being the original software.
59 *
60 **** Alterations to Henry's code are...
61 ****
4bb101f2 62 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
1129b882
NC
63 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
64 **** by Larry Wall and others
a687059c 65 ****
9ef589d8
LW
66 **** You may distribute under the terms of either the GNU General Public
67 **** License or the Artistic License, as specified in the README file.
a687059c
LW
68 *
69 * Beware that some of this code is subtly aware of the way operator
70 * precedence is structured in regular expressions. Serious changes in
71 * regular-expression syntax might require a total rethink.
72 */
73#include "EXTERN.h"
864dbfa3 74#define PERL_IN_REGEXEC_C
a687059c 75#include "perl.h"
0f5d15d6 76
54df2634
NC
77#ifdef PERL_IN_XSUB_RE
78# include "re_comp.h"
79#else
80# include "regcomp.h"
81#endif
a687059c 82
c277df42
IZ
83#define RF_tainted 1 /* tainted information used? */
84#define RF_warned 2 /* warned about big count? */
faec1544 85
ab3bbdeb 86#define RF_utf8 8 /* Pattern contains multibyte chars? */
a0ed51b3 87
f2ed9b32 88#define UTF_PATTERN ((PL_reg_flags & RF_utf8) != 0)
ce862d02
IZ
89
90#define RS_init 1 /* eval environment created */
91#define RS_set 2 /* replsv value is set */
c277df42 92
a687059c
LW
93#ifndef STATIC
94#define STATIC static
95#endif
96
32fc9b6a 97#define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
7d3e948e 98
c277df42
IZ
99/*
100 * Forwards.
101 */
102
f2ed9b32 103#define CHR_SVLEN(sv) (utf8_target ? sv_len_utf8(sv) : SvCUR(sv))
53c4c00c 104#define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)
a0ed51b3 105
3dab1dad
YO
106#define HOPc(pos,off) \
107 (char *)(PL_reg_match_utf8 \
52657f30 108 ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \
3dab1dad
YO
109 : (U8*)(pos + off))
110#define HOPBACKc(pos, off) \
07be1b83
YO
111 (char*)(PL_reg_match_utf8\
112 ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \
113 : (pos - off >= PL_bostr) \
8e11feef 114 ? (U8*)pos - off \
3dab1dad 115 : NULL)
efb30f32 116
e7409c1b 117#define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
1aa99e6b 118#define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
1aa99e6b 119
20d0b1e9 120/* these are unrolled below in the CCC_TRY_XXX defined */
1a4fad37
AL
121#define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \
122 if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END
37e2e78e
KW
123
124/* Doesn't do an assert to verify that is correct */
125#define LOAD_UTF8_CHARCLASS_NO_CHECK(class) STMT_START { \
126 if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)" "); LEAVE; } } STMT_END
127
1a4fad37
AL
128#define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
129#define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
130#define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ")
51371543 131
37e2e78e 132#define LOAD_UTF8_CHARCLASS_GCB() /* Grapheme cluster boundaries */ \
d275fa5e
CB
133 LOAD_UTF8_CHARCLASS(X_begin, " "); \
134 LOAD_UTF8_CHARCLASS(X_non_hangul, "A"); \
37e2e78e
KW
135 /* These are utf8 constants, and not utf-ebcdic constants, so the \
136 * assert should likely and hopefully fail on an EBCDIC machine */ \
d275fa5e 137 LOAD_UTF8_CHARCLASS(X_extend, "\xcc\x80"); /* U+0300 */ \
37e2e78e
KW
138 \
139 /* No asserts are done for these, in case called on an early \
140 * Unicode version in which they map to nothing */ \
d275fa5e
CB
141 LOAD_UTF8_CHARCLASS_NO_CHECK(X_prepend);/* U+0E40 "\xe0\xb9\x80" */ \
142 LOAD_UTF8_CHARCLASS_NO_CHECK(X_L); /* U+1100 "\xe1\x84\x80" */ \
143 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LV); /* U+AC00 "\xea\xb0\x80" */ \
144 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LVT); /* U+AC01 "\xea\xb0\x81" */ \
145 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LV_LVT_V);/* U+AC01 "\xea\xb0\x81" */\
146 LOAD_UTF8_CHARCLASS_NO_CHECK(X_T); /* U+11A8 "\xe1\x86\xa8" */ \
37e2e78e 147 LOAD_UTF8_CHARCLASS_NO_CHECK(X_V) /* U+1160 "\xe1\x85\xa0" */
20d0b1e9 148
d1eb3177
YO
149/*
150 We dont use PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS as the direct test
151 so that it is possible to override the option here without having to
152 rebuild the entire core. as we are required to do if we change regcomp.h
153 which is where PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS is defined.
154*/
155#if PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS
156#define BROKEN_UNICODE_CHARCLASS_MAPPINGS
157#endif
158
159#ifdef BROKEN_UNICODE_CHARCLASS_MAPPINGS
160#define LOAD_UTF8_CHARCLASS_PERL_WORD() LOAD_UTF8_CHARCLASS_ALNUM()
161#define LOAD_UTF8_CHARCLASS_PERL_SPACE() LOAD_UTF8_CHARCLASS_SPACE()
162#define LOAD_UTF8_CHARCLASS_POSIX_DIGIT() LOAD_UTF8_CHARCLASS_DIGIT()
163#define RE_utf8_perl_word PL_utf8_alnum
164#define RE_utf8_perl_space PL_utf8_space
165#define RE_utf8_posix_digit PL_utf8_digit
166#define perl_word alnum
167#define perl_space space
168#define posix_digit digit
169#else
170#define LOAD_UTF8_CHARCLASS_PERL_WORD() LOAD_UTF8_CHARCLASS(perl_word,"a")
171#define LOAD_UTF8_CHARCLASS_PERL_SPACE() LOAD_UTF8_CHARCLASS(perl_space," ")
172#define LOAD_UTF8_CHARCLASS_POSIX_DIGIT() LOAD_UTF8_CHARCLASS(posix_digit,"0")
173#define RE_utf8_perl_word PL_utf8_perl_word
174#define RE_utf8_perl_space PL_utf8_perl_space
175#define RE_utf8_posix_digit PL_utf8_posix_digit
176#endif
177
178
20d0b1e9
YO
179#define CCC_TRY_AFF(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC,LCFUNC) \
180 case NAMEL: \
181 PL_reg_flags |= RF_tainted; \
182 /* FALL THROUGH */ \
183 case NAME: \
184 if (!nextchr) \
185 sayNO; \
f2ed9b32 186 if (utf8_target && UTF8_IS_CONTINUED(nextchr)) { \
20d0b1e9
YO
187 if (!CAT2(PL_utf8_,CLASS)) { \
188 bool ok; \
189 ENTER; \
190 save_re_context(); \
191 ok=CAT2(is_utf8_,CLASS)((const U8*)STR); \
192 assert(ok); \
193 LEAVE; \
194 } \
195 if (!(OP(scan) == NAME \
f2ed9b32 196 ? cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), (U8*)locinput, utf8_target)) \
20d0b1e9
YO
197 : LCFUNC_utf8((U8*)locinput))) \
198 { \
199 sayNO; \
200 } \
201 locinput += PL_utf8skip[nextchr]; \
202 nextchr = UCHARAT(locinput); \
203 break; \
204 } \
205 if (!(OP(scan) == NAME ? FUNC(nextchr) : LCFUNC(nextchr))) \
206 sayNO; \
207 nextchr = UCHARAT(++locinput); \
208 break
209
210#define CCC_TRY_NEG(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC,LCFUNC) \
211 case NAMEL: \
212 PL_reg_flags |= RF_tainted; \
213 /* FALL THROUGH */ \
214 case NAME : \
215 if (!nextchr && locinput >= PL_regeol) \
216 sayNO; \
f2ed9b32 217 if (utf8_target && UTF8_IS_CONTINUED(nextchr)) { \
20d0b1e9
YO
218 if (!CAT2(PL_utf8_,CLASS)) { \
219 bool ok; \
220 ENTER; \
221 save_re_context(); \
222 ok=CAT2(is_utf8_,CLASS)((const U8*)STR); \
223 assert(ok); \
224 LEAVE; \
225 } \
226 if ((OP(scan) == NAME \
f2ed9b32 227 ? cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), (U8*)locinput, utf8_target)) \
20d0b1e9
YO
228 : LCFUNC_utf8((U8*)locinput))) \
229 { \
230 sayNO; \
231 } \
232 locinput += PL_utf8skip[nextchr]; \
233 nextchr = UCHARAT(locinput); \
234 break; \
235 } \
236 if ((OP(scan) == NAME ? FUNC(nextchr) : LCFUNC(nextchr))) \
237 sayNO; \
238 nextchr = UCHARAT(++locinput); \
239 break
240
241
d1eb3177
YO
242
243
244
3dab1dad
YO
245/* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
246
5f80c4cf 247/* for use after a quantifier and before an EXACT-like node -- japhy */
3e901dc0
YO
248/* it would be nice to rework regcomp.sym to generate this stuff. sigh */
249#define JUMPABLE(rn) ( \
250 OP(rn) == OPEN || \
251 (OP(rn) == CLOSE && (!cur_eval || cur_eval->u.eval.close_paren != ARG(rn))) || \
252 OP(rn) == EVAL || \
cca55fe3
JP
253 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
254 OP(rn) == PLUS || OP(rn) == MINMOD || \
ee9b8eae 255 OP(rn) == KEEPS || (PL_regkind[OP(rn)] == VERB) || \
3dab1dad 256 (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
e2d8ce26 257)
ee9b8eae 258#define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT)
e2d8ce26 259
ee9b8eae
YO
260#define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF )
261
262#if 0
263/* Currently these are only used when PL_regkind[OP(rn)] == EXACT so
264 we don't need this definition. */
265#define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF )
266#define IS_TEXTF(rn) ( OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF )
267#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL )
268
269#else
270/* ... so we use this as its faster. */
271#define IS_TEXT(rn) ( OP(rn)==EXACT )
272#define IS_TEXTF(rn) ( OP(rn)==EXACTF )
273#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL )
274
275#endif
e2d8ce26 276
a84d97b6
HS
277/*
278 Search for mandatory following text node; for lookahead, the text must
279 follow but for lookbehind (rn->flags != 0) we skip to the next step.
280*/
cca55fe3 281#define FIND_NEXT_IMPT(rn) STMT_START { \
3dab1dad
YO
282 while (JUMPABLE(rn)) { \
283 const OPCODE type = OP(rn); \
284 if (type == SUSPEND || PL_regkind[type] == CURLY) \
e2d8ce26 285 rn = NEXTOPER(NEXTOPER(rn)); \
3dab1dad 286 else if (type == PLUS) \
cca55fe3 287 rn = NEXTOPER(rn); \
3dab1dad 288 else if (type == IFMATCH) \
a84d97b6 289 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
e2d8ce26 290 else rn += NEXT_OFF(rn); \
3dab1dad 291 } \
5f80c4cf 292} STMT_END
74750237 293
c476f425 294
acfe0abc 295static void restore_pos(pTHX_ void *arg);
51371543 296
620d5b66
NC
297#define REGCP_PAREN_ELEMS 4
298#define REGCP_OTHER_ELEMS 5
e0fa7e2b 299#define REGCP_FRAME_ELEMS 1
620d5b66
NC
300/* REGCP_FRAME_ELEMS are not part of the REGCP_OTHER_ELEMS and
301 * are needed for the regexp context stack bookkeeping. */
302
76e3520e 303STATIC CHECKPOINT
cea2e8a9 304S_regcppush(pTHX_ I32 parenfloor)
a0d0e21e 305{
97aff369 306 dVAR;
a3b680e6 307 const int retval = PL_savestack_ix;
a3b680e6 308 const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
e0fa7e2b
NC
309 const UV total_elems = paren_elems_to_push + REGCP_OTHER_ELEMS;
310 const UV elems_shifted = total_elems << SAVE_TIGHT_SHIFT;
a0d0e21e 311 int p;
40a82448 312 GET_RE_DEBUG_FLAGS_DECL;
a0d0e21e 313
e49a9654
IH
314 if (paren_elems_to_push < 0)
315 Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
316
e0fa7e2b
NC
317 if ((elems_shifted >> SAVE_TIGHT_SHIFT) != total_elems)
318 Perl_croak(aTHX_ "panic: paren_elems_to_push offset %"UVuf
5df417d0
JH
319 " out of range (%lu-%ld)",
320 total_elems, (unsigned long)PL_regsize, (long)parenfloor);
e0fa7e2b 321
620d5b66 322 SSGROW(total_elems + REGCP_FRAME_ELEMS);
7f69552c 323
3280af22 324 for (p = PL_regsize; p > parenfloor; p--) {
b1ce53c5 325/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
f0ab9afb
NC
326 SSPUSHINT(PL_regoffs[p].end);
327 SSPUSHINT(PL_regoffs[p].start);
3280af22 328 SSPUSHPTR(PL_reg_start_tmp[p]);
a0d0e21e 329 SSPUSHINT(p);
e7707071 330 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
40a82448 331 " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n",
f0ab9afb 332 (UV)p, (IV)PL_regoffs[p].start,
40a82448 333 (IV)(PL_reg_start_tmp[p] - PL_bostr),
f0ab9afb 334 (IV)PL_regoffs[p].end
40a82448 335 ));
a0d0e21e 336 }
b1ce53c5 337/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
f0ab9afb 338 SSPUSHPTR(PL_regoffs);
3280af22
NIS
339 SSPUSHINT(PL_regsize);
340 SSPUSHINT(*PL_reglastparen);
a01268b5 341 SSPUSHINT(*PL_reglastcloseparen);
3280af22 342 SSPUSHPTR(PL_reginput);
e0fa7e2b 343 SSPUSHUV(SAVEt_REGCONTEXT | elems_shifted); /* Magic cookie. */
41123dfd 344
a0d0e21e
LW
345 return retval;
346}
347
c277df42 348/* These are needed since we do not localize EVAL nodes: */
ab3bbdeb
YO
349#define REGCP_SET(cp) \
350 DEBUG_STATE_r( \
ab3bbdeb 351 PerlIO_printf(Perl_debug_log, \
e4f74956 352 " Setting an EVAL scope, savestack=%"IVdf"\n", \
ab3bbdeb
YO
353 (IV)PL_savestack_ix)); \
354 cp = PL_savestack_ix
c3464db5 355
ab3bbdeb 356#define REGCP_UNWIND(cp) \
e4f74956 357 DEBUG_STATE_r( \
ab3bbdeb 358 if (cp != PL_savestack_ix) \
e4f74956
YO
359 PerlIO_printf(Perl_debug_log, \
360 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
ab3bbdeb
YO
361 (IV)(cp), (IV)PL_savestack_ix)); \
362 regcpblow(cp)
c277df42 363
76e3520e 364STATIC char *
097eb12c 365S_regcppop(pTHX_ const regexp *rex)
a0d0e21e 366{
97aff369 367 dVAR;
e0fa7e2b 368 UV i;
a0d0e21e 369 char *input;
a3621e74
YO
370 GET_RE_DEBUG_FLAGS_DECL;
371
7918f24d
NC
372 PERL_ARGS_ASSERT_REGCPPOP;
373
b1ce53c5 374 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
c6bf6a65 375 i = SSPOPUV;
e0fa7e2b
NC
376 assert((i & SAVE_MASK) == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
377 i >>= SAVE_TIGHT_SHIFT; /* Parentheses elements to pop. */
a0d0e21e 378 input = (char *) SSPOPPTR;
a01268b5 379 *PL_reglastcloseparen = SSPOPINT;
3280af22
NIS
380 *PL_reglastparen = SSPOPINT;
381 PL_regsize = SSPOPINT;
f0ab9afb 382 PL_regoffs=(regexp_paren_pair *) SSPOPPTR;
b1ce53c5 383
620d5b66 384 i -= REGCP_OTHER_ELEMS;
b1ce53c5 385 /* Now restore the parentheses context. */
620d5b66 386 for ( ; i > 0; i -= REGCP_PAREN_ELEMS) {
1df70142 387 I32 tmps;
097eb12c 388 U32 paren = (U32)SSPOPINT;
3280af22 389 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
f0ab9afb 390 PL_regoffs[paren].start = SSPOPINT;
cf93c79d 391 tmps = SSPOPINT;
3280af22 392 if (paren <= *PL_reglastparen)
f0ab9afb 393 PL_regoffs[paren].end = tmps;
e7707071 394 DEBUG_BUFFERS_r(
c3464db5 395 PerlIO_printf(Perl_debug_log,
b900a521 396 " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
f0ab9afb 397 (UV)paren, (IV)PL_regoffs[paren].start,
b900a521 398 (IV)(PL_reg_start_tmp[paren] - PL_bostr),
f0ab9afb 399 (IV)PL_regoffs[paren].end,
3280af22 400 (paren > *PL_reglastparen ? "(no)" : ""));
c277df42 401 );
a0d0e21e 402 }
e7707071 403 DEBUG_BUFFERS_r(
bb7a0f54 404 if (*PL_reglastparen + 1 <= rex->nparens) {
c3464db5 405 PerlIO_printf(Perl_debug_log,
faccc32b 406 " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
4f639d21 407 (IV)(*PL_reglastparen + 1), (IV)rex->nparens);
c277df42
IZ
408 }
409 );
daf18116 410#if 1
dafc8851
JH
411 /* It would seem that the similar code in regtry()
412 * already takes care of this, and in fact it is in
413 * a better location to since this code can #if 0-ed out
414 * but the code in regtry() is needed or otherwise tests
415 * requiring null fields (pat.t#187 and split.t#{13,14}
daf18116
JH
416 * (as of patchlevel 7877) will fail. Then again,
417 * this code seems to be necessary or otherwise
225593e1
DM
418 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
419 * --jhi updated by dapm */
3b6647e0 420 for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) {
097eb12c 421 if (i > PL_regsize)
f0ab9afb
NC
422 PL_regoffs[i].start = -1;
423 PL_regoffs[i].end = -1;
a0d0e21e 424 }
dafc8851 425#endif
a0d0e21e
LW
426 return input;
427}
428
02db2b7b 429#define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
a0d0e21e 430
a687059c 431/*
e50aee73 432 * pregexec and friends
a687059c
LW
433 */
434
76234dfb 435#ifndef PERL_IN_XSUB_RE
a687059c 436/*
c277df42 437 - pregexec - match a regexp against a string
a687059c 438 */
c277df42 439I32
49d7dfbc 440Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, register char *strend,
c3464db5 441 char *strbeg, I32 minend, SV *screamer, U32 nosave)
c277df42
IZ
442/* strend: pointer to null at end of string */
443/* strbeg: real beginning of string */
444/* minend: end of match must be >=minend after stringarg. */
445/* nosave: For optimizations. */
446{
7918f24d
NC
447 PERL_ARGS_ASSERT_PREGEXEC;
448
c277df42 449 return
9041c2e3 450 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
c277df42
IZ
451 nosave ? 0 : REXEC_COPY_STR);
452}
76234dfb 453#endif
22e551b9 454
9041c2e3 455/*
cad2e5aa
JH
456 * Need to implement the following flags for reg_anch:
457 *
458 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
459 * USE_INTUIT_ML
460 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
461 * INTUIT_AUTORITATIVE_ML
462 * INTUIT_ONCE_NOML - Intuit can match in one location only.
463 * INTUIT_ONCE_ML
464 *
465 * Another flag for this function: SECOND_TIME (so that float substrs
466 * with giant delta may be not rechecked).
467 */
468
469/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
470
3f7c398e 471/* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend.
cad2e5aa
JH
472 Otherwise, only SvCUR(sv) is used to get strbeg. */
473
474/* XXXX We assume that strpos is strbeg unless sv. */
475
6eb5f6b9
JH
476/* XXXX Some places assume that there is a fixed substring.
477 An update may be needed if optimizer marks as "INTUITable"
478 RExen without fixed substrings. Similarly, it is assumed that
479 lengths of all the strings are no more than minlen, thus they
480 cannot come from lookahead.
40d049e4
YO
481 (Or minlen should take into account lookahead.)
482 NOTE: Some of this comment is not correct. minlen does now take account
483 of lookahead/behind. Further research is required. -- demerphq
484
485*/
6eb5f6b9 486
2c2d71f5
JH
487/* A failure to find a constant substring means that there is no need to make
488 an expensive call to REx engine, thus we celebrate a failure. Similarly,
489 finding a substring too deep into the string means that less calls to
30944b6d
IZ
490 regtry() should be needed.
491
492 REx compiler's optimizer found 4 possible hints:
493 a) Anchored substring;
494 b) Fixed substring;
495 c) Whether we are anchored (beginning-of-line or \G);
496 d) First node (of those at offset 0) which may distingush positions;
6eb5f6b9 497 We use a)b)d) and multiline-part of c), and try to find a position in the
30944b6d
IZ
498 string which does not contradict any of them.
499 */
2c2d71f5 500
6eb5f6b9
JH
501/* Most of decisions we do here should have been done at compile time.
502 The nodes of the REx which we used for the search should have been
503 deleted from the finite automaton. */
504
cad2e5aa 505char *
288b8c02 506Perl_re_intuit_start(pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a 507 char *strend, const U32 flags, re_scream_pos_data *data)
cad2e5aa 508{
97aff369 509 dVAR;
288b8c02 510 struct regexp *const prog = (struct regexp *)SvANY(rx);
b7953727 511 register I32 start_shift = 0;
cad2e5aa 512 /* Should be nonnegative! */
b7953727 513 register I32 end_shift = 0;
2c2d71f5
JH
514 register char *s;
515 register SV *check;
a1933d95 516 char *strbeg;
cad2e5aa 517 char *t;
f2ed9b32 518 const bool utf8_target = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
cad2e5aa 519 I32 ml_anch;
bd61b366
SS
520 register char *other_last = NULL; /* other substr checked before this */
521 char *check_at = NULL; /* check substr found at this pos */
bbe252da 522 const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
f8fc2ecf 523 RXi_GET_DECL(prog,progi);
30944b6d 524#ifdef DEBUGGING
890ce7af 525 const char * const i_strpos = strpos;
30944b6d 526#endif
a3621e74
YO
527 GET_RE_DEBUG_FLAGS_DECL;
528
7918f24d
NC
529 PERL_ARGS_ASSERT_RE_INTUIT_START;
530
f2ed9b32 531 RX_MATCH_UTF8_set(rx,utf8_target);
cad2e5aa 532
3c8556c3 533 if (RX_UTF8(rx)) {
b8d68ded
JH
534 PL_reg_flags |= RF_utf8;
535 }
ab3bbdeb 536 DEBUG_EXECUTE_r(
f2ed9b32 537 debug_start_match(rx, utf8_target, strpos, strend,
1de06328
YO
538 sv ? "Guessing start of match in sv for"
539 : "Guessing start of match in string for");
2a782b5b 540 );
cad2e5aa 541
c344f387
JH
542 /* CHR_DIST() would be more correct here but it makes things slow. */
543 if (prog->minlen > strend - strpos) {
a3621e74 544 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
a72c7584 545 "String too short... [re_intuit_start]\n"));
cad2e5aa 546 goto fail;
2c2d71f5 547 }
1de06328 548
a1933d95 549 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
1aa99e6b 550 PL_regeol = strend;
f2ed9b32 551 if (utf8_target) {
33b8afdf
JH
552 if (!prog->check_utf8 && prog->check_substr)
553 to_utf8_substr(prog);
554 check = prog->check_utf8;
555 } else {
556 if (!prog->check_substr && prog->check_utf8)
557 to_byte_substr(prog);
558 check = prog->check_substr;
559 }
1de06328 560 if (check == &PL_sv_undef) {
a3621e74 561 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1de06328 562 "Non-utf8 string cannot match utf8 check string\n"));
33b8afdf
JH
563 goto fail;
564 }
bbe252da
YO
565 if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */
566 ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE)
567 || ( (prog->extflags & RXf_ANCH_BOL)
7fba1cd6 568 && !multiline ) ); /* Check after \n? */
cad2e5aa 569
7e25d62c 570 if (!ml_anch) {
bbe252da
YO
571 if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */
572 && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */
3f7c398e 573 /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
7e25d62c
JH
574 && sv && !SvROK(sv)
575 && (strpos != strbeg)) {
a3621e74 576 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
7e25d62c
JH
577 goto fail;
578 }
579 if (prog->check_offset_min == prog->check_offset_max &&
bbe252da 580 !(prog->extflags & RXf_CANY_SEEN)) {
2c2d71f5 581 /* Substring at constant offset from beg-of-str... */
cad2e5aa
JH
582 I32 slen;
583
1aa99e6b 584 s = HOP3c(strpos, prog->check_offset_min, strend);
1de06328 585
653099ff
GS
586 if (SvTAIL(check)) {
587 slen = SvCUR(check); /* >= 1 */
cad2e5aa 588
9041c2e3 589 if ( strend - s > slen || strend - s < slen - 1
2c2d71f5 590 || (strend - s == slen && strend[-1] != '\n')) {
a3621e74 591 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
2c2d71f5 592 goto fail_finish;
cad2e5aa
JH
593 }
594 /* Now should match s[0..slen-2] */
595 slen--;
3f7c398e 596 if (slen && (*SvPVX_const(check) != *s
cad2e5aa 597 || (slen > 1
3f7c398e 598 && memNE(SvPVX_const(check), s, slen)))) {
2c2d71f5 599 report_neq:
a3621e74 600 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
2c2d71f5
JH
601 goto fail_finish;
602 }
cad2e5aa 603 }
3f7c398e 604 else if (*SvPVX_const(check) != *s
653099ff 605 || ((slen = SvCUR(check)) > 1
3f7c398e 606 && memNE(SvPVX_const(check), s, slen)))
2c2d71f5 607 goto report_neq;
c315bfe8 608 check_at = s;
2c2d71f5 609 goto success_at_start;
7e25d62c 610 }
cad2e5aa 611 }
2c2d71f5 612 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
cad2e5aa 613 s = strpos;
2c2d71f5 614 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
1de06328
YO
615 end_shift = prog->check_end_shift;
616
2c2d71f5 617 if (!ml_anch) {
a3b680e6 618 const I32 end = prog->check_offset_max + CHR_SVLEN(check)
653099ff 619 - (SvTAIL(check) != 0);
a3b680e6 620 const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
2c2d71f5
JH
621
622 if (end_shift < eshift)
623 end_shift = eshift;
624 }
cad2e5aa 625 }
2c2d71f5 626 else { /* Can match at random position */
cad2e5aa
JH
627 ml_anch = 0;
628 s = strpos;
1de06328
YO
629 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
630 end_shift = prog->check_end_shift;
631
632 /* end shift should be non negative here */
cad2e5aa
JH
633 }
634
bcdf7404 635#ifdef QDEBUGGING /* 7/99: reports of failure (with the older version) */
0033605d 636 if (end_shift < 0)
1de06328 637 Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
220fc49f 638 (IV)end_shift, RX_PRECOMP(prog));
2c2d71f5
JH
639#endif
640
2c2d71f5
JH
641 restart:
642 /* Find a possible match in the region s..strend by looking for
643 the "check" substring in the region corrected by start/end_shift. */
1de06328
YO
644
645 {
646 I32 srch_start_shift = start_shift;
647 I32 srch_end_shift = end_shift;
648 if (srch_start_shift < 0 && strbeg - s > srch_start_shift) {
649 srch_end_shift -= ((strbeg - s) - srch_start_shift);
650 srch_start_shift = strbeg - s;
651 }
6bda09f9 652 DEBUG_OPTIMISE_MORE_r({
1de06328
YO
653 PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n",
654 (IV)prog->check_offset_min,
655 (IV)srch_start_shift,
656 (IV)srch_end_shift,
657 (IV)prog->check_end_shift);
658 });
659
cad2e5aa 660 if (flags & REXEC_SCREAM) {
cad2e5aa 661 I32 p = -1; /* Internal iterator of scream. */
a3b680e6 662 I32 * const pp = data ? data->scream_pos : &p;
cad2e5aa 663
2c2d71f5
JH
664 if (PL_screamfirst[BmRARE(check)] >= 0
665 || ( BmRARE(check) == '\n'
85c508c3 666 && (BmPREVIOUS(check) == SvCUR(check) - 1)
2c2d71f5 667 && SvTAIL(check) ))
9041c2e3 668 s = screaminstr(sv, check,
1de06328 669 srch_start_shift + (s - strbeg), srch_end_shift, pp, 0);
cad2e5aa 670 else
2c2d71f5 671 goto fail_finish;
4addbd3b 672 /* we may be pointing at the wrong string */
07bc277f 673 if (s && RXp_MATCH_COPIED(prog))
3f7c398e 674 s = strbeg + (s - SvPVX_const(sv));
cad2e5aa
JH
675 if (data)
676 *data->scream_olds = s;
677 }
1de06328
YO
678 else {
679 U8* start_point;
680 U8* end_point;
bbe252da 681 if (prog->extflags & RXf_CANY_SEEN) {
1de06328
YO
682 start_point= (U8*)(s + srch_start_shift);
683 end_point= (U8*)(strend - srch_end_shift);
684 } else {
685 start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend);
686 end_point= HOP3(strend, -srch_end_shift, strbeg);
687 }
6bda09f9 688 DEBUG_OPTIMISE_MORE_r({
56570a2c 689 PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n",
1de06328 690 (int)(end_point - start_point),
fc8cd66c 691 (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point),
1de06328
YO
692 start_point);
693 });
694
695 s = fbm_instr( start_point, end_point,
7fba1cd6 696 check, multiline ? FBMrf_MULTILINE : 0);
1de06328
YO
697 }
698 }
cad2e5aa
JH
699 /* Update the count-of-usability, remove useless subpatterns,
700 unshift s. */
2c2d71f5 701
ab3bbdeb 702 DEBUG_EXECUTE_r({
f2ed9b32 703 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
ab3bbdeb
YO
704 SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
705 PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s",
2c2d71f5 706 (s ? "Found" : "Did not find"),
f2ed9b32 707 (check == (utf8_target ? prog->anchored_utf8 : prog->anchored_substr)
ab3bbdeb
YO
708 ? "anchored" : "floating"),
709 quoted,
710 RE_SV_TAIL(check),
711 (s ? " at offset " : "...\n") );
712 });
2c2d71f5
JH
713
714 if (!s)
715 goto fail_finish;
2c2d71f5 716 /* Finish the diagnostic message */
a3621e74 717 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
2c2d71f5 718
1de06328
YO
719 /* XXX dmq: first branch is for positive lookbehind...
720 Our check string is offset from the beginning of the pattern.
721 So we need to do any stclass tests offset forward from that
722 point. I think. :-(
723 */
724
725
726
727 check_at=s;
728
729
2c2d71f5
JH
730 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
731 Start with the other substr.
732 XXXX no SCREAM optimization yet - and a very coarse implementation
a0288114 733 XXXX /ttx+/ results in anchored="ttx", floating="x". floating will
2c2d71f5
JH
734 *always* match. Probably should be marked during compile...
735 Probably it is right to do no SCREAM here...
736 */
737
f2ed9b32 738 if (utf8_target ? (prog->float_utf8 && prog->anchored_utf8)
1de06328
YO
739 : (prog->float_substr && prog->anchored_substr))
740 {
30944b6d 741 /* Take into account the "other" substring. */
2c2d71f5
JH
742 /* XXXX May be hopelessly wrong for UTF... */
743 if (!other_last)
6eb5f6b9 744 other_last = strpos;
f2ed9b32 745 if (check == (utf8_target ? prog->float_utf8 : prog->float_substr)) {
30944b6d
IZ
746 do_other_anchored:
747 {
890ce7af
AL
748 char * const last = HOP3c(s, -start_shift, strbeg);
749 char *last1, *last2;
be8e71aa 750 char * const saved_s = s;
33b8afdf 751 SV* must;
2c2d71f5 752
2c2d71f5
JH
753 t = s - prog->check_offset_max;
754 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
f2ed9b32 755 && (!utf8_target
0ce71af7 756 || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos))
2c2d71f5 757 && t > strpos)))
6f207bd3 758 NOOP;
2c2d71f5
JH
759 else
760 t = strpos;
1aa99e6b 761 t = HOP3c(t, prog->anchored_offset, strend);
6eb5f6b9
JH
762 if (t < other_last) /* These positions already checked */
763 t = other_last;
1aa99e6b 764 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
2c2d71f5
JH
765 if (last < last1)
766 last1 = last;
1de06328
YO
767 /* XXXX It is not documented what units *_offsets are in.
768 We assume bytes, but this is clearly wrong.
769 Meaning this code needs to be carefully reviewed for errors.
770 dmq.
771 */
772
2c2d71f5 773 /* On end-of-str: see comment below. */
f2ed9b32 774 must = utf8_target ? prog->anchored_utf8 : prog->anchored_substr;
33b8afdf
JH
775 if (must == &PL_sv_undef) {
776 s = (char*)NULL;
1de06328 777 DEBUG_r(must = prog->anchored_utf8); /* for debug */
33b8afdf
JH
778 }
779 else
780 s = fbm_instr(
781 (unsigned char*)t,
782 HOP3(HOP3(last1, prog->anchored_offset, strend)
783 + SvCUR(must), -(SvTAIL(must)!=0), strbeg),
784 must,
7fba1cd6 785 multiline ? FBMrf_MULTILINE : 0
33b8afdf 786 );
ab3bbdeb 787 DEBUG_EXECUTE_r({
f2ed9b32 788 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
ab3bbdeb
YO
789 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
790 PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s",
2c2d71f5 791 (s ? "Found" : "Contradicts"),
ab3bbdeb
YO
792 quoted, RE_SV_TAIL(must));
793 });
794
795
2c2d71f5
JH
796 if (!s) {
797 if (last1 >= last2) {
a3621e74 798 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2c2d71f5
JH
799 ", giving up...\n"));
800 goto fail_finish;
801 }
a3621e74 802 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2c2d71f5 803 ", trying floating at offset %ld...\n",
be8e71aa 804 (long)(HOP3c(saved_s, 1, strend) - i_strpos)));
1aa99e6b
IH
805 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
806 s = HOP3c(last, 1, strend);
2c2d71f5
JH
807 goto restart;
808 }
809 else {
a3621e74 810 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
30944b6d 811 (long)(s - i_strpos)));
1aa99e6b
IH
812 t = HOP3c(s, -prog->anchored_offset, strbeg);
813 other_last = HOP3c(s, 1, strend);
be8e71aa 814 s = saved_s;
2c2d71f5
JH
815 if (t == strpos)
816 goto try_at_start;
2c2d71f5
JH
817 goto try_at_offset;
818 }
30944b6d 819 }
2c2d71f5
JH
820 }
821 else { /* Take into account the floating substring. */
33b8afdf 822 char *last, *last1;
be8e71aa 823 char * const saved_s = s;
33b8afdf
JH
824 SV* must;
825
826 t = HOP3c(s, -start_shift, strbeg);
827 last1 = last =
828 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
829 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
830 last = HOP3c(t, prog->float_max_offset, strend);
831 s = HOP3c(t, prog->float_min_offset, strend);
832 if (s < other_last)
833 s = other_last;
2c2d71f5 834 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
f2ed9b32 835 must = utf8_target ? prog->float_utf8 : prog->float_substr;
33b8afdf
JH
836 /* fbm_instr() takes into account exact value of end-of-str
837 if the check is SvTAIL(ed). Since false positives are OK,
838 and end-of-str is not later than strend we are OK. */
839 if (must == &PL_sv_undef) {
840 s = (char*)NULL;
1de06328 841 DEBUG_r(must = prog->float_utf8); /* for debug message */
33b8afdf
JH
842 }
843 else
2c2d71f5 844 s = fbm_instr((unsigned char*)s,
33b8afdf
JH
845 (unsigned char*)last + SvCUR(must)
846 - (SvTAIL(must)!=0),
7fba1cd6 847 must, multiline ? FBMrf_MULTILINE : 0);
ab3bbdeb 848 DEBUG_EXECUTE_r({
f2ed9b32 849 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
ab3bbdeb
YO
850 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
851 PerlIO_printf(Perl_debug_log, "%s floating substr %s%s",
33b8afdf 852 (s ? "Found" : "Contradicts"),
ab3bbdeb
YO
853 quoted, RE_SV_TAIL(must));
854 });
33b8afdf
JH
855 if (!s) {
856 if (last1 == last) {
a3621e74 857 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf
JH
858 ", giving up...\n"));
859 goto fail_finish;
2c2d71f5 860 }
a3621e74 861 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf 862 ", trying anchored starting at offset %ld...\n",
be8e71aa 863 (long)(saved_s + 1 - i_strpos)));
33b8afdf
JH
864 other_last = last;
865 s = HOP3c(t, 1, strend);
866 goto restart;
867 }
868 else {
a3621e74 869 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
33b8afdf
JH
870 (long)(s - i_strpos)));
871 other_last = s; /* Fix this later. --Hugo */
be8e71aa 872 s = saved_s;
33b8afdf
JH
873 if (t == strpos)
874 goto try_at_start;
875 goto try_at_offset;
876 }
2c2d71f5 877 }
cad2e5aa 878 }
2c2d71f5 879
1de06328 880
9ef43ace 881 t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos);
1de06328 882
6bda09f9 883 DEBUG_OPTIMISE_MORE_r(
1de06328
YO
884 PerlIO_printf(Perl_debug_log,
885 "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n",
886 (IV)prog->check_offset_min,
887 (IV)prog->check_offset_max,
888 (IV)(s-strpos),
889 (IV)(t-strpos),
890 (IV)(t-s),
891 (IV)(strend-strpos)
892 )
893 );
894
2c2d71f5 895 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
f2ed9b32 896 && (!utf8_target
9ef43ace 897 || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos)))
1de06328
YO
898 && t > strpos)))
899 {
2c2d71f5
JH
900 /* Fixed substring is found far enough so that the match
901 cannot start at strpos. */
902 try_at_offset:
cad2e5aa 903 if (ml_anch && t[-1] != '\n') {
30944b6d
IZ
904 /* Eventually fbm_*() should handle this, but often
905 anchored_offset is not 0, so this check will not be wasted. */
906 /* XXXX In the code below we prefer to look for "^" even in
907 presence of anchored substrings. And we search even
908 beyond the found float position. These pessimizations
909 are historical artefacts only. */
910 find_anchor:
2c2d71f5 911 while (t < strend - prog->minlen) {
cad2e5aa 912 if (*t == '\n') {
4ee3650e 913 if (t < check_at - prog->check_offset_min) {
f2ed9b32 914 if (utf8_target ? prog->anchored_utf8 : prog->anchored_substr) {
4ee3650e
GS
915 /* Since we moved from the found position,
916 we definitely contradict the found anchored
30944b6d
IZ
917 substr. Due to the above check we do not
918 contradict "check" substr.
919 Thus we can arrive here only if check substr
920 is float. Redo checking for "other"=="fixed".
921 */
9041c2e3 922 strpos = t + 1;
a3621e74 923 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
e4584336 924 PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
30944b6d
IZ
925 goto do_other_anchored;
926 }
4ee3650e
GS
927 /* We don't contradict the found floating substring. */
928 /* XXXX Why not check for STCLASS? */
cad2e5aa 929 s = t + 1;
a3621e74 930 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
e4584336 931 PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
cad2e5aa
JH
932 goto set_useful;
933 }
4ee3650e
GS
934 /* Position contradicts check-string */
935 /* XXXX probably better to look for check-string
936 than for "\n", so one should lower the limit for t? */
a3621e74 937 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
e4584336 938 PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
0e41cd87 939 other_last = strpos = s = t + 1;
cad2e5aa
JH
940 goto restart;
941 }
942 t++;
943 }
a3621e74 944 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
e4584336 945 PL_colors[0], PL_colors[1]));
2c2d71f5 946 goto fail_finish;
cad2e5aa 947 }
f5952150 948 else {
a3621e74 949 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
e4584336 950 PL_colors[0], PL_colors[1]));
f5952150 951 }
cad2e5aa
JH
952 s = t;
953 set_useful:
f2ed9b32 954 ++BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
cad2e5aa
JH
955 }
956 else {
f5952150 957 /* The found string does not prohibit matching at strpos,
2c2d71f5 958 - no optimization of calling REx engine can be performed,
f5952150
GS
959 unless it was an MBOL and we are not after MBOL,
960 or a future STCLASS check will fail this. */
2c2d71f5
JH
961 try_at_start:
962 /* Even in this situation we may use MBOL flag if strpos is offset
963 wrt the start of the string. */
05b4157f 964 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
a1933d95 965 && (strpos != strbeg) && strpos[-1] != '\n'
d506a20d 966 /* May be due to an implicit anchor of m{.*foo} */
bbe252da 967 && !(prog->intflags & PREGf_IMPLICIT))
d506a20d 968 {
cad2e5aa
JH
969 t = strpos;
970 goto find_anchor;
971 }
a3621e74 972 DEBUG_EXECUTE_r( if (ml_anch)
f5952150 973 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
70685ca0 974 (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
30944b6d 975 );
2c2d71f5 976 success_at_start:
bbe252da 977 if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */
f2ed9b32 978 && (utf8_target ? (
33b8afdf
JH
979 prog->check_utf8 /* Could be deleted already */
980 && --BmUSEFUL(prog->check_utf8) < 0
981 && (prog->check_utf8 == prog->float_utf8)
982 ) : (
983 prog->check_substr /* Could be deleted already */
984 && --BmUSEFUL(prog->check_substr) < 0
985 && (prog->check_substr == prog->float_substr)
986 )))
66e933ab 987 {
cad2e5aa 988 /* If flags & SOMETHING - do not do it many times on the same match */
a3621e74 989 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
f2ed9b32
KW
990 /* XXX Does the destruction order has to change with utf8_target? */
991 SvREFCNT_dec(utf8_target ? prog->check_utf8 : prog->check_substr);
992 SvREFCNT_dec(utf8_target ? prog->check_substr : prog->check_utf8);
a0714e2c
SS
993 prog->check_substr = prog->check_utf8 = NULL; /* disable */
994 prog->float_substr = prog->float_utf8 = NULL; /* clear */
995 check = NULL; /* abort */
cad2e5aa 996 s = strpos;
c9415951
YO
997 /* XXXX If the check string was an implicit check MBOL, then we need to unset the relevent flag
998 see http://bugs.activestate.com/show_bug.cgi?id=87173 */
999 if (prog->intflags & PREGf_IMPLICIT)
1000 prog->extflags &= ~RXf_ANCH_MBOL;
3cf5c195
IZ
1001 /* XXXX This is a remnant of the old implementation. It
1002 looks wasteful, since now INTUIT can use many
6eb5f6b9 1003 other heuristics. */
bbe252da 1004 prog->extflags &= ~RXf_USE_INTUIT;
c9415951 1005 /* XXXX What other flags might need to be cleared in this branch? */
cad2e5aa
JH
1006 }
1007 else
1008 s = strpos;
1009 }
1010
6eb5f6b9
JH
1011 /* Last resort... */
1012 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
1de06328
YO
1013 /* trie stclasses are too expensive to use here, we are better off to
1014 leave it to regmatch itself */
f8fc2ecf 1015 if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
6eb5f6b9
JH
1016 /* minlen == 0 is possible if regstclass is \b or \B,
1017 and the fixed substr is ''$.
1018 Since minlen is already taken into account, s+1 is before strend;
1019 accidentally, minlen >= 1 guaranties no false positives at s + 1
1020 even for \b or \B. But (minlen? 1 : 0) below assumes that
1021 regstclass does not come from lookahead... */
1022 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
1023 This leaves EXACTF only, which is dealt with in find_byclass(). */
f8fc2ecf
YO
1024 const U8* const str = (U8*)STRING(progi->regstclass);
1025 const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
1026 ? CHR_DIST(str+STR_LEN(progi->regstclass), str)
66e933ab 1027 : 1);
1de06328
YO
1028 char * endpos;
1029 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
1030 endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend);
1031 else if (prog->float_substr || prog->float_utf8)
1032 endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend);
1033 else
1034 endpos= strend;
1035
70685ca0
JH
1036 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n",
1037 (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg)));
1de06328 1038
6eb5f6b9 1039 t = s;
f8fc2ecf 1040 s = find_byclass(prog, progi->regstclass, s, endpos, NULL);
6eb5f6b9
JH
1041 if (!s) {
1042#ifdef DEBUGGING
cbbf8932 1043 const char *what = NULL;
6eb5f6b9
JH
1044#endif
1045 if (endpos == strend) {
a3621e74 1046 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6eb5f6b9
JH
1047 "Could not match STCLASS...\n") );
1048 goto fail;
1049 }
a3621e74 1050 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab 1051 "This position contradicts STCLASS...\n") );
bbe252da 1052 if ((prog->extflags & RXf_ANCH) && !ml_anch)
653099ff 1053 goto fail;
6eb5f6b9 1054 /* Contradict one of substrings */
33b8afdf 1055 if (prog->anchored_substr || prog->anchored_utf8) {
f2ed9b32 1056 if ((utf8_target ? prog->anchored_utf8 : prog->anchored_substr) == check) {
a3621e74 1057 DEBUG_EXECUTE_r( what = "anchored" );
6eb5f6b9 1058 hop_and_restart:
1aa99e6b 1059 s = HOP3c(t, 1, strend);
66e933ab
GS
1060 if (s + start_shift + end_shift > strend) {
1061 /* XXXX Should be taken into account earlier? */
a3621e74 1062 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab
GS
1063 "Could not match STCLASS...\n") );
1064 goto fail;
1065 }
5e39e1e5
HS
1066 if (!check)
1067 goto giveup;
a3621e74 1068 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 1069 "Looking for %s substr starting at offset %ld...\n",
6eb5f6b9
JH
1070 what, (long)(s + start_shift - i_strpos)) );
1071 goto restart;
1072 }
66e933ab 1073 /* Have both, check_string is floating */
6eb5f6b9
JH
1074 if (t + start_shift >= check_at) /* Contradicts floating=check */
1075 goto retry_floating_check;
1076 /* Recheck anchored substring, but not floating... */
9041c2e3 1077 s = check_at;
5e39e1e5
HS
1078 if (!check)
1079 goto giveup;
a3621e74 1080 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 1081 "Looking for anchored substr starting at offset %ld...\n",
6eb5f6b9
JH
1082 (long)(other_last - i_strpos)) );
1083 goto do_other_anchored;
1084 }
60e71179
GS
1085 /* Another way we could have checked stclass at the
1086 current position only: */
1087 if (ml_anch) {
1088 s = t = t + 1;
5e39e1e5
HS
1089 if (!check)
1090 goto giveup;
a3621e74 1091 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 1092 "Looking for /%s^%s/m starting at offset %ld...\n",
e4584336 1093 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
60e71179 1094 goto try_at_offset;
66e933ab 1095 }
f2ed9b32 1096 if (!(utf8_target ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
60e71179 1097 goto fail;
6eb5f6b9
JH
1098 /* Check is floating subtring. */
1099 retry_floating_check:
1100 t = check_at - start_shift;
a3621e74 1101 DEBUG_EXECUTE_r( what = "floating" );
6eb5f6b9
JH
1102 goto hop_and_restart;
1103 }
b7953727 1104 if (t != s) {
a3621e74 1105 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
6eb5f6b9 1106 "By STCLASS: moving %ld --> %ld\n",
b7953727
JH
1107 (long)(t - i_strpos), (long)(s - i_strpos))
1108 );
1109 }
1110 else {
a3621e74 1111 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
b7953727
JH
1112 "Does not contradict STCLASS...\n");
1113 );
1114 }
6eb5f6b9 1115 }
5e39e1e5 1116 giveup:
a3621e74 1117 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
5e39e1e5
HS
1118 PL_colors[4], (check ? "Guessed" : "Giving up"),
1119 PL_colors[5], (long)(s - i_strpos)) );
cad2e5aa 1120 return s;
2c2d71f5
JH
1121
1122 fail_finish: /* Substring not found */
33b8afdf 1123 if (prog->check_substr || prog->check_utf8) /* could be removed already */
f2ed9b32 1124 BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
cad2e5aa 1125 fail:
a3621e74 1126 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
e4584336 1127 PL_colors[4], PL_colors[5]));
bd61b366 1128 return NULL;
cad2e5aa 1129}
9661b544 1130
a0a388a1
YO
1131#define DECL_TRIE_TYPE(scan) \
1132 const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold } \
1133 trie_type = (scan->flags != EXACT) \
f2ed9b32
KW
1134 ? (utf8_target ? trie_utf8_fold : (UTF_PATTERN ? trie_latin_utf8_fold : trie_plain)) \
1135 : (utf8_target ? trie_utf8 : trie_plain)
3b0527fe 1136
55eed653
NC
1137#define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \
1138uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \
4cadc6a9
YO
1139 switch (trie_type) { \
1140 case trie_utf8_fold: \
1141 if ( foldlen>0 ) { \
0abd0d78 1142 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
4cadc6a9
YO
1143 foldlen -= len; \
1144 uscan += len; \
1145 len=0; \
1146 } else { \
0abd0d78 1147 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
4cadc6a9
YO
1148 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
1149 foldlen -= UNISKIP( uvc ); \
1150 uscan = foldbuf + UNISKIP( uvc ); \
1151 } \
1152 break; \
a0a388a1
YO
1153 case trie_latin_utf8_fold: \
1154 if ( foldlen>0 ) { \
1155 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
1156 foldlen -= len; \
1157 uscan += len; \
1158 len=0; \
1159 } else { \
1160 len = 1; \
1161 uvc = to_uni_fold( *(U8*)uc, foldbuf, &foldlen ); \
1162 foldlen -= UNISKIP( uvc ); \
1163 uscan = foldbuf + UNISKIP( uvc ); \
1164 } \
1165 break; \
4cadc6a9
YO
1166 case trie_utf8: \
1167 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
1168 break; \
1169 case trie_plain: \
1170 uvc = (UV)*uc; \
1171 len = 1; \
1172 } \
4cadc6a9
YO
1173 if (uvc < 256) { \
1174 charid = trie->charmap[ uvc ]; \
1175 } \
1176 else { \
1177 charid = 0; \
55eed653
NC
1178 if (widecharmap) { \
1179 SV** const svpp = hv_fetch(widecharmap, \
4cadc6a9
YO
1180 (char*)&uvc, sizeof(UV), 0); \
1181 if (svpp) \
1182 charid = (U16)SvIV(*svpp); \
1183 } \
1184 } \
1185} STMT_END
1186
a0a388a1
YO
1187#define REXEC_FBC_EXACTISH_CHECK(CoNd) \
1188{ \
1189 char *my_strend= (char *)strend; \
4cadc6a9
YO
1190 if ( (CoNd) \
1191 && (ln == len || \
f2ed9b32
KW
1192 foldEQ_utf8(s, &my_strend, 0, utf8_target, \
1193 m, NULL, ln, cBOOL(UTF_PATTERN))) \
a0a388a1 1194 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1195 goto got_it; \
1196 else { \
1197 U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \
1198 uvchr_to_utf8(tmpbuf, c); \
1199 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \
1200 if ( f != c \
1201 && (f == c1 || f == c2) \
a0a388a1 1202 && (ln == len || \
f2ed9b32
KW
1203 foldEQ_utf8(s, &my_strend, 0, utf8_target,\
1204 m, NULL, ln, cBOOL(UTF_PATTERN)))\
a0a388a1 1205 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1206 goto got_it; \
1207 } \
a0a388a1
YO
1208} \
1209s += len
4cadc6a9
YO
1210
1211#define REXEC_FBC_EXACTISH_SCAN(CoNd) \
1212STMT_START { \
1213 while (s <= e) { \
1214 if ( (CoNd) \
4c1b470c
KW
1215 && (ln == 1 || (OP(c) == EXACTF \
1216 ? foldEQ(s, m, ln) \
1217 : foldEQ_locale(s, m, ln))) \
24b23f37 1218 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1219 goto got_it; \
1220 s++; \
1221 } \
1222} STMT_END
1223
1224#define REXEC_FBC_UTF8_SCAN(CoDe) \
1225STMT_START { \
1226 while (s + (uskip = UTF8SKIP(s)) <= strend) { \
1227 CoDe \
1228 s += uskip; \
1229 } \
1230} STMT_END
1231
1232#define REXEC_FBC_SCAN(CoDe) \
1233STMT_START { \
1234 while (s < strend) { \
1235 CoDe \
1236 s++; \
1237 } \
1238} STMT_END
1239
1240#define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \
1241REXEC_FBC_UTF8_SCAN( \
1242 if (CoNd) { \
24b23f37 1243 if (tmp && (!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1244 goto got_it; \
1245 else \
1246 tmp = doevery; \
1247 } \
1248 else \
1249 tmp = 1; \
1250)
1251
1252#define REXEC_FBC_CLASS_SCAN(CoNd) \
1253REXEC_FBC_SCAN( \
1254 if (CoNd) { \
24b23f37 1255 if (tmp && (!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1256 goto got_it; \
1257 else \
1258 tmp = doevery; \
1259 } \
1260 else \
1261 tmp = 1; \
1262)
1263
1264#define REXEC_FBC_TRYIT \
24b23f37 1265if ((!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1266 goto got_it
1267
e1d1eefb 1268#define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \
f2ed9b32 1269 if (utf8_target) { \
e1d1eefb
YO
1270 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1271 } \
1272 else { \
1273 REXEC_FBC_CLASS_SCAN(CoNd); \
1274 } \
1275 break
1276
4cadc6a9 1277#define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
f2ed9b32 1278 if (utf8_target) { \
4cadc6a9
YO
1279 UtFpReLoAd; \
1280 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1281 } \
1282 else { \
1283 REXEC_FBC_CLASS_SCAN(CoNd); \
1284 } \
1285 break
1286
1287#define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \
1288 PL_reg_flags |= RF_tainted; \
f2ed9b32 1289 if (utf8_target) { \
4cadc6a9
YO
1290 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1291 } \
1292 else { \
1293 REXEC_FBC_CLASS_SCAN(CoNd); \
1294 } \
1295 break
1296
786e8c11
YO
1297#define DUMP_EXEC_POS(li,s,doutf8) \
1298 dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
1299
1300/* We know what class REx starts with. Try to find this position... */
1301/* if reginfo is NULL, its a dryrun */
1302/* annoyingly all the vars in this routine have different names from their counterparts
1303 in regmatch. /grrr */
1304
3c3eec57 1305STATIC char *
07be1b83 1306S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
24b23f37 1307 const char *strend, regmatch_info *reginfo)
a687059c 1308{
27da23d5 1309 dVAR;
bbe252da 1310 const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
6eb5f6b9 1311 char *m;
d8093b23 1312 STRLEN ln;
5dab1207 1313 STRLEN lnc;
078c425b 1314 register STRLEN uskip;
d8093b23 1315 unsigned int c1;
1316 unsigned int c2;
6eb5f6b9
JH
1317 char *e;
1318 register I32 tmp = 1; /* Scratch variable? */
f2ed9b32 1319 register const bool utf8_target = PL_reg_match_utf8;
f8fc2ecf 1320 RXi_GET_DECL(prog,progi);
7918f24d
NC
1321
1322 PERL_ARGS_ASSERT_FIND_BYCLASS;
f8fc2ecf 1323
6eb5f6b9
JH
1324 /* We know what class it must start with. */
1325 switch (OP(c)) {
6eb5f6b9 1326 case ANYOF:
f2ed9b32 1327 if (utf8_target) {
4cadc6a9 1328 REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
388cc4de 1329 !UTF8_IS_INVARIANT((U8)s[0]) ?
f2ed9b32 1330 reginclass(prog, c, (U8*)s, 0, utf8_target) :
4cadc6a9 1331 REGINCLASS(prog, c, (U8*)s));
388cc4de
HS
1332 }
1333 else {
1334 while (s < strend) {
1335 STRLEN skip = 1;
1336
32fc9b6a 1337 if (REGINCLASS(prog, c, (U8*)s) ||
388cc4de
HS
1338 (ANYOF_FOLD_SHARP_S(c, s, strend) &&
1339 /* The assignment of 2 is intentional:
1340 * for the folded sharp s, the skip is 2. */
1341 (skip = SHARP_S_SKIP))) {
24b23f37 1342 if (tmp && (!reginfo || regtry(reginfo, &s)))
388cc4de
HS
1343 goto got_it;
1344 else
1345 tmp = doevery;
1346 }
1347 else
1348 tmp = 1;
1349 s += skip;
1350 }
a0d0e21e 1351 }
6eb5f6b9 1352 break;
f33976b4 1353 case CANY:
4cadc6a9 1354 REXEC_FBC_SCAN(
24b23f37 1355 if (tmp && (!reginfo || regtry(reginfo, &s)))
f33976b4
DB
1356 goto got_it;
1357 else
1358 tmp = doevery;
4cadc6a9 1359 );
f33976b4 1360 break;
6eb5f6b9 1361 case EXACTF:
5dab1207
NIS
1362 m = STRING(c);
1363 ln = STR_LEN(c); /* length to match in octets/bytes */
1364 lnc = (I32) ln; /* length to match in characters */
f2ed9b32 1365 if (UTF_PATTERN) {
a2a2844f 1366 STRLEN ulen1, ulen2;
5dab1207 1367 U8 *sm = (U8 *) m;
89ebb4a3
JH
1368 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
1369 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
97dc7d3e
RGS
1370 /* used by commented-out code below */
1371 /*const U32 uniflags = UTF8_ALLOW_DEFAULT;*/
a0a388a1
YO
1372
1373 /* XXX: Since the node will be case folded at compile
1374 time this logic is a little odd, although im not
1375 sure that its actually wrong. --dmq */
1376
1377 c1 = to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1378 c2 = to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1379
1380 /* XXX: This is kinda strange. to_utf8_XYZ returns the
1381 codepoint of the first character in the converted
1382 form, yet originally we did the extra step.
1383 No tests fail by commenting this code out however
1384 so Ive left it out. -- dmq.
1385
89ebb4a3 1386 c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
041457d9 1387 0, uniflags);
89ebb4a3 1388 c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
041457d9 1389 0, uniflags);
a0a388a1
YO
1390 */
1391
5dab1207
NIS
1392 lnc = 0;
1393 while (sm < ((U8 *) m + ln)) {
1394 lnc++;
1395 sm += UTF8SKIP(sm);
1396 }
1aa99e6b
IH
1397 }
1398 else {
1399 c1 = *(U8*)m;
1400 c2 = PL_fold[c1];
1401 }
6eb5f6b9
JH
1402 goto do_exactf;
1403 case EXACTFL:
5dab1207
NIS
1404 m = STRING(c);
1405 ln = STR_LEN(c);
1406 lnc = (I32) ln;
d8093b23 1407 c1 = *(U8*)m;
6eb5f6b9
JH
1408 c2 = PL_fold_locale[c1];
1409 do_exactf:
db12adc6 1410 e = HOP3c(strend, -((I32)lnc), s);
b3c9acc1 1411
3b0527fe 1412 if (!reginfo && e < s)
6eb5f6b9 1413 e = s; /* Due to minlen logic of intuit() */
1aa99e6b 1414
60a8b682
JH
1415 /* The idea in the EXACTF* cases is to first find the
1416 * first character of the EXACTF* node and then, if
1417 * necessary, case-insensitively compare the full
1418 * text of the node. The c1 and c2 are the first
1419 * characters (though in Unicode it gets a bit
1420 * more complicated because there are more cases
7f16dd3d
JH
1421 * than just upper and lower: one needs to use
1422 * the so-called folding case for case-insensitive
1423 * matching (called "loose matching" in Unicode).
4c1b470c 1424 * foldEQ_utf8() will do just that. */
60a8b682 1425
f2ed9b32 1426 if (utf8_target || UTF_PATTERN) {
575cac57 1427 UV c, f;
89ebb4a3 1428 U8 tmpbuf [UTF8_MAXBYTES+1];
a0a388a1
YO
1429 STRLEN len = 1;
1430 STRLEN foldlen;
4ad0818d 1431 const U32 uniflags = UTF8_ALLOW_DEFAULT;
09091399 1432 if (c1 == c2) {
5dab1207
NIS
1433 /* Upper and lower of 1st char are equal -
1434 * probably not a "letter". */
1aa99e6b 1435 while (s <= e) {
f2ed9b32 1436 if (utf8_target) {
a0a388a1 1437 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
041457d9 1438 uniflags);
a0a388a1
YO
1439 } else {
1440 c = *((U8*)s);
1441 }
4cadc6a9 1442 REXEC_FBC_EXACTISH_CHECK(c == c1);
1aa99e6b 1443 }
09091399
JH
1444 }
1445 else {
1aa99e6b 1446 while (s <= e) {
f2ed9b32 1447 if (utf8_target) {
a0a388a1 1448 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
041457d9 1449 uniflags);
a0a388a1
YO
1450 } else {
1451 c = *((U8*)s);
1452 }
80aecb99 1453
60a8b682 1454 /* Handle some of the three Greek sigmas cases.
8c01da3c
JH
1455 * Note that not all the possible combinations
1456 * are handled here: some of them are handled
1457 * by the standard folding rules, and some of
1458 * them (the character class or ANYOF cases)
1459 * are handled during compiletime in
1460 * regexec.c:S_regclass(). */
880bd946
JH
1461 if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
1462 c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
1463 c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
80aecb99 1464
4cadc6a9 1465 REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2);
1aa99e6b 1466 }
09091399 1467 }
1aa99e6b
IH
1468 }
1469 else {
a0a388a1 1470 /* Neither pattern nor string are UTF8 */
1aa99e6b 1471 if (c1 == c2)
4cadc6a9 1472 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
1aa99e6b 1473 else
4cadc6a9 1474 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
b3c9acc1
IZ
1475 }
1476 break;
bbce6d69 1477 case BOUNDL:
3280af22 1478 PL_reg_flags |= RF_tainted;
bbce6d69 1479 /* FALL THROUGH */
a0d0e21e 1480 case BOUND:
f2ed9b32 1481 if (utf8_target) {
12d33761 1482 if (s == PL_bostr)
ffc61ed2
JH
1483 tmp = '\n';
1484 else {
6136c704 1485 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
4ad0818d 1486 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
ffc61ed2
JH
1487 }
1488 tmp = ((OP(c) == BOUND ?
9041c2e3 1489 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1a4fad37 1490 LOAD_UTF8_CHARCLASS_ALNUM();
4cadc6a9 1491 REXEC_FBC_UTF8_SCAN(
ffc61ed2 1492 if (tmp == !(OP(c) == BOUND ?
f2ed9b32 1493 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)) :
ffc61ed2
JH
1494 isALNUM_LC_utf8((U8*)s)))
1495 {
1496 tmp = !tmp;
4cadc6a9 1497 REXEC_FBC_TRYIT;
a687059c 1498 }
4cadc6a9 1499 );
a0d0e21e 1500 }
667bb95a 1501 else {
12d33761 1502 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2 1503 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
4cadc6a9 1504 REXEC_FBC_SCAN(
ffc61ed2
JH
1505 if (tmp ==
1506 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
1507 tmp = !tmp;
4cadc6a9 1508 REXEC_FBC_TRYIT;
a0ed51b3 1509 }
4cadc6a9 1510 );
a0ed51b3 1511 }
24b23f37 1512 if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))
a0ed51b3
LW
1513 goto got_it;
1514 break;
bbce6d69 1515 case NBOUNDL:
3280af22 1516 PL_reg_flags |= RF_tainted;
bbce6d69 1517 /* FALL THROUGH */
a0d0e21e 1518 case NBOUND:
f2ed9b32 1519 if (utf8_target) {
12d33761 1520 if (s == PL_bostr)
ffc61ed2
JH
1521 tmp = '\n';
1522 else {
6136c704 1523 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
4ad0818d 1524 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
ffc61ed2
JH
1525 }
1526 tmp = ((OP(c) == NBOUND ?
9041c2e3 1527 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1a4fad37 1528 LOAD_UTF8_CHARCLASS_ALNUM();
4cadc6a9 1529 REXEC_FBC_UTF8_SCAN(
ffc61ed2 1530 if (tmp == !(OP(c) == NBOUND ?
f2ed9b32 1531 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)) :
ffc61ed2
JH
1532 isALNUM_LC_utf8((U8*)s)))
1533 tmp = !tmp;
4cadc6a9
YO
1534 else REXEC_FBC_TRYIT;
1535 );
a0d0e21e 1536 }
667bb95a 1537 else {
12d33761 1538 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2
JH
1539 tmp = ((OP(c) == NBOUND ?
1540 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
4cadc6a9 1541 REXEC_FBC_SCAN(
ffc61ed2
JH
1542 if (tmp ==
1543 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1544 tmp = !tmp;
4cadc6a9
YO
1545 else REXEC_FBC_TRYIT;
1546 );
a0ed51b3 1547 }
24b23f37 1548 if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s)))
a0ed51b3
LW
1549 goto got_it;
1550 break;
a0d0e21e 1551 case ALNUM:
4cadc6a9 1552 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1553 LOAD_UTF8_CHARCLASS_PERL_WORD(),
f2ed9b32 1554 swash_fetch(RE_utf8_perl_word, (U8*)s, utf8_target),
4cadc6a9
YO
1555 isALNUM(*s)
1556 );
bbce6d69 1557 case ALNUML:
4cadc6a9
YO
1558 REXEC_FBC_CSCAN_TAINT(
1559 isALNUM_LC_utf8((U8*)s),
1560 isALNUM_LC(*s)
1561 );
a0d0e21e 1562 case NALNUM:
4cadc6a9 1563 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1564 LOAD_UTF8_CHARCLASS_PERL_WORD(),
f2ed9b32 1565 !swash_fetch(RE_utf8_perl_word, (U8*)s, utf8_target),
4cadc6a9
YO
1566 !isALNUM(*s)
1567 );
bbce6d69 1568 case NALNUML:
4cadc6a9
YO
1569 REXEC_FBC_CSCAN_TAINT(
1570 !isALNUM_LC_utf8((U8*)s),
1571 !isALNUM_LC(*s)
1572 );
a0d0e21e 1573 case SPACE:
4cadc6a9 1574 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1575 LOAD_UTF8_CHARCLASS_PERL_SPACE(),
f2ed9b32 1576 *s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, utf8_target),
4cadc6a9
YO
1577 isSPACE(*s)
1578 );
bbce6d69 1579 case SPACEL:
4cadc6a9
YO
1580 REXEC_FBC_CSCAN_TAINT(
1581 *s == ' ' || isSPACE_LC_utf8((U8*)s),
1582 isSPACE_LC(*s)
1583 );
a0d0e21e 1584 case NSPACE:
4cadc6a9 1585 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1586 LOAD_UTF8_CHARCLASS_PERL_SPACE(),
f2ed9b32 1587 !(*s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, utf8_target)),
4cadc6a9
YO
1588 !isSPACE(*s)
1589 );
bbce6d69 1590 case NSPACEL:
4cadc6a9
YO
1591 REXEC_FBC_CSCAN_TAINT(
1592 !(*s == ' ' || isSPACE_LC_utf8((U8*)s)),
1593 !isSPACE_LC(*s)
1594 );
a0d0e21e 1595 case DIGIT:
4cadc6a9 1596 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1597 LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
f2ed9b32 1598 swash_fetch(RE_utf8_posix_digit,(U8*)s, utf8_target),
4cadc6a9
YO
1599 isDIGIT(*s)
1600 );
b8c5462f 1601 case DIGITL:
4cadc6a9
YO
1602 REXEC_FBC_CSCAN_TAINT(
1603 isDIGIT_LC_utf8((U8*)s),
1604 isDIGIT_LC(*s)
1605 );
a0d0e21e 1606 case NDIGIT:
4cadc6a9 1607 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1608 LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
f2ed9b32 1609 !swash_fetch(RE_utf8_posix_digit,(U8*)s, utf8_target),
4cadc6a9
YO
1610 !isDIGIT(*s)
1611 );
b8c5462f 1612 case NDIGITL:
4cadc6a9
YO
1613 REXEC_FBC_CSCAN_TAINT(
1614 !isDIGIT_LC_utf8((U8*)s),
1615 !isDIGIT_LC(*s)
1616 );
e1d1eefb
YO
1617 case LNBREAK:
1618 REXEC_FBC_CSCAN(
1619 is_LNBREAK_utf8(s),
1620 is_LNBREAK_latin1(s)
1621 );
1622 case VERTWS:
1623 REXEC_FBC_CSCAN(
1624 is_VERTWS_utf8(s),
1625 is_VERTWS_latin1(s)
1626 );
1627 case NVERTWS:
1628 REXEC_FBC_CSCAN(
1629 !is_VERTWS_utf8(s),
1630 !is_VERTWS_latin1(s)
1631 );
1632 case HORIZWS:
1633 REXEC_FBC_CSCAN(
1634 is_HORIZWS_utf8(s),
1635 is_HORIZWS_latin1(s)
1636 );
1637 case NHORIZWS:
1638 REXEC_FBC_CSCAN(
1639 !is_HORIZWS_utf8(s),
1640 !is_HORIZWS_latin1(s)
1641 );
1de06328
YO
1642 case AHOCORASICKC:
1643 case AHOCORASICK:
07be1b83 1644 {
a0a388a1 1645 DECL_TRIE_TYPE(c);
07be1b83
YO
1646 /* what trie are we using right now */
1647 reg_ac_data *aho
f8fc2ecf 1648 = (reg_ac_data*)progi->data->data[ ARG( c ) ];
3251b653
NC
1649 reg_trie_data *trie
1650 = (reg_trie_data*)progi->data->data[ aho->trie ];
85fbaab2 1651 HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]);
07be1b83
YO
1652
1653 const char *last_start = strend - trie->minlen;
6148ee25 1654#ifdef DEBUGGING
07be1b83 1655 const char *real_start = s;
6148ee25 1656#endif
07be1b83 1657 STRLEN maxlen = trie->maxlen;
be8e71aa
YO
1658 SV *sv_points;
1659 U8 **points; /* map of where we were in the input string
786e8c11 1660 when reading a given char. For ASCII this
be8e71aa 1661 is unnecessary overhead as the relationship
38a44b82
NC
1662 is always 1:1, but for Unicode, especially
1663 case folded Unicode this is not true. */
f9e705e8 1664 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
786e8c11
YO
1665 U8 *bitmap=NULL;
1666
07be1b83
YO
1667
1668 GET_RE_DEBUG_FLAGS_DECL;
1669
be8e71aa
YO
1670 /* We can't just allocate points here. We need to wrap it in
1671 * an SV so it gets freed properly if there is a croak while
1672 * running the match */
1673 ENTER;
1674 SAVETMPS;
1675 sv_points=newSV(maxlen * sizeof(U8 *));
1676 SvCUR_set(sv_points,
1677 maxlen * sizeof(U8 *));
1678 SvPOK_on(sv_points);
1679 sv_2mortal(sv_points);
1680 points=(U8**)SvPV_nolen(sv_points );
1de06328
YO
1681 if ( trie_type != trie_utf8_fold
1682 && (trie->bitmap || OP(c)==AHOCORASICKC) )
1683 {
786e8c11
YO
1684 if (trie->bitmap)
1685 bitmap=(U8*)trie->bitmap;
1686 else
1687 bitmap=(U8*)ANYOF_BITMAP(c);
07be1b83 1688 }
786e8c11
YO
1689 /* this is the Aho-Corasick algorithm modified a touch
1690 to include special handling for long "unknown char"
1691 sequences. The basic idea being that we use AC as long
1692 as we are dealing with a possible matching char, when
1693 we encounter an unknown char (and we have not encountered
1694 an accepting state) we scan forward until we find a legal
1695 starting char.
1696 AC matching is basically that of trie matching, except
1697 that when we encounter a failing transition, we fall back
1698 to the current states "fail state", and try the current char
1699 again, a process we repeat until we reach the root state,
1700 state 1, or a legal transition. If we fail on the root state
1701 then we can either terminate if we have reached an accepting
1702 state previously, or restart the entire process from the beginning
1703 if we have not.
1704
1705 */
07be1b83
YO
1706 while (s <= last_start) {
1707 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1708 U8 *uc = (U8*)s;
1709 U16 charid = 0;
1710 U32 base = 1;
1711 U32 state = 1;
1712 UV uvc = 0;
1713 STRLEN len = 0;
1714 STRLEN foldlen = 0;
1715 U8 *uscan = (U8*)NULL;
1716 U8 *leftmost = NULL;
786e8c11
YO
1717#ifdef DEBUGGING
1718 U32 accepted_word= 0;
1719#endif
07be1b83
YO
1720 U32 pointpos = 0;
1721
1722 while ( state && uc <= (U8*)strend ) {
1723 int failed=0;
786e8c11
YO
1724 U32 word = aho->states[ state ].wordnum;
1725
1de06328
YO
1726 if( state==1 ) {
1727 if ( bitmap ) {
1728 DEBUG_TRIE_EXECUTE_r(
1729 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1730 dump_exec_pos( (char *)uc, c, strend, real_start,
f2ed9b32 1731 (char *)uc, utf8_target );
1de06328
YO
1732 PerlIO_printf( Perl_debug_log,
1733 " Scanning for legal start char...\n");
1734 }
1735 );
1736 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1737 uc++;
786e8c11 1738 }
1de06328 1739 s= (char *)uc;
786e8c11 1740 }
786e8c11
YO
1741 if (uc >(U8*)last_start) break;
1742 }
1743
1744 if ( word ) {
2e64971a 1745 U8 *lpos= points[ (pointpos - trie->wordinfo[word].len) % maxlen ];
786e8c11
YO
1746 if (!leftmost || lpos < leftmost) {
1747 DEBUG_r(accepted_word=word);
07be1b83 1748 leftmost= lpos;
786e8c11 1749 }
07be1b83 1750 if (base==0) break;
786e8c11 1751
07be1b83
YO
1752 }
1753 points[pointpos++ % maxlen]= uc;
55eed653
NC
1754 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
1755 uscan, len, uvc, charid, foldlen,
1756 foldbuf, uniflags);
786e8c11
YO
1757 DEBUG_TRIE_EXECUTE_r({
1758 dump_exec_pos( (char *)uc, c, strend, real_start,
f2ed9b32 1759 s, utf8_target );
07be1b83 1760 PerlIO_printf(Perl_debug_log,
786e8c11
YO
1761 " Charid:%3u CP:%4"UVxf" ",
1762 charid, uvc);
1763 });
07be1b83
YO
1764
1765 do {
6148ee25 1766#ifdef DEBUGGING
786e8c11 1767 word = aho->states[ state ].wordnum;
6148ee25 1768#endif
07be1b83
YO
1769 base = aho->states[ state ].trans.base;
1770
786e8c11
YO
1771 DEBUG_TRIE_EXECUTE_r({
1772 if (failed)
1773 dump_exec_pos( (char *)uc, c, strend, real_start,
f2ed9b32 1774 s, utf8_target );
07be1b83 1775 PerlIO_printf( Perl_debug_log,
786e8c11
YO
1776 "%sState: %4"UVxf", word=%"UVxf,
1777 failed ? " Fail transition to " : "",
1778 (UV)state, (UV)word);
1779 });
07be1b83
YO
1780 if ( base ) {
1781 U32 tmp;
6dd2be57 1782 I32 offset;
07be1b83 1783 if (charid &&
6dd2be57
DM
1784 ( ((offset = base + charid
1785 - 1 - trie->uniquecharcount)) >= 0)
1786 && ((U32)offset < trie->lasttrans)
1787 && trie->trans[offset].check == state
1788 && (tmp=trie->trans[offset].next))
07be1b83 1789 {
786e8c11
YO
1790 DEBUG_TRIE_EXECUTE_r(
1791 PerlIO_printf( Perl_debug_log," - legal\n"));
07be1b83
YO
1792 state = tmp;
1793 break;
1794 }
1795 else {
786e8c11
YO
1796 DEBUG_TRIE_EXECUTE_r(
1797 PerlIO_printf( Perl_debug_log," - fail\n"));
1798 failed = 1;
1799 state = aho->fail[state];
07be1b83
YO
1800 }
1801 }
1802 else {
1803 /* we must be accepting here */
786e8c11
YO
1804 DEBUG_TRIE_EXECUTE_r(
1805 PerlIO_printf( Perl_debug_log," - accepting\n"));
1806 failed = 1;
07be1b83
YO
1807 break;
1808 }
1809 } while(state);
786e8c11 1810 uc += len;
07be1b83
YO
1811 if (failed) {
1812 if (leftmost)
1813 break;
786e8c11 1814 if (!state) state = 1;
07be1b83
YO
1815 }
1816 }
1817 if ( aho->states[ state ].wordnum ) {
2e64971a 1818 U8 *lpos = points[ (pointpos - trie->wordinfo[aho->states[ state ].wordnum].len) % maxlen ];
786e8c11
YO
1819 if (!leftmost || lpos < leftmost) {
1820 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
07be1b83 1821 leftmost = lpos;
786e8c11 1822 }
07be1b83 1823 }
07be1b83
YO
1824 if (leftmost) {
1825 s = (char*)leftmost;
786e8c11
YO
1826 DEBUG_TRIE_EXECUTE_r({
1827 PerlIO_printf(
70685ca0
JH
1828 Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
1829 (UV)accepted_word, (IV)(s - real_start)
786e8c11
YO
1830 );
1831 });
24b23f37 1832 if (!reginfo || regtry(reginfo, &s)) {
be8e71aa
YO
1833 FREETMPS;
1834 LEAVE;
07be1b83 1835 goto got_it;
be8e71aa 1836 }
07be1b83 1837 s = HOPc(s,1);
786e8c11
YO
1838 DEBUG_TRIE_EXECUTE_r({
1839 PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
1840 });
07be1b83 1841 } else {
786e8c11
YO
1842 DEBUG_TRIE_EXECUTE_r(
1843 PerlIO_printf( Perl_debug_log,"No match.\n"));
07be1b83
YO
1844 break;
1845 }
1846 }
be8e71aa
YO
1847 FREETMPS;
1848 LEAVE;
07be1b83
YO
1849 }
1850 break;
b3c9acc1 1851 default:
3c3eec57
GS
1852 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
1853 break;
d6a28714 1854 }
6eb5f6b9
JH
1855 return 0;
1856 got_it:
1857 return s;
1858}
1859
fae667d5 1860
6eb5f6b9
JH
1861/*
1862 - regexec_flags - match a regexp against a string
1863 */
1864I32
288b8c02 1865Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, register char *strend,
6eb5f6b9
JH
1866 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
1867/* strend: pointer to null at end of string */
1868/* strbeg: real beginning of string */
1869/* minend: end of match must be >=minend after stringarg. */
58e23c8d
YO
1870/* data: May be used for some additional optimizations.
1871 Currently its only used, with a U32 cast, for transmitting
1872 the ganch offset when doing a /g match. This will change */
6eb5f6b9
JH
1873/* nosave: For optimizations. */
1874{
97aff369 1875 dVAR;
288b8c02 1876 struct regexp *const prog = (struct regexp *)SvANY(rx);
24b23f37 1877 /*register*/ char *s;
6eb5f6b9 1878 register regnode *c;
24b23f37 1879 /*register*/ char *startpos = stringarg;
6eb5f6b9
JH
1880 I32 minlen; /* must match at least this many chars */
1881 I32 dontbother = 0; /* how many characters not to try at end */
6eb5f6b9
JH
1882 I32 end_shift = 0; /* Same for the end. */ /* CC */
1883 I32 scream_pos = -1; /* Internal iterator of scream. */
ccac19ea 1884 char *scream_olds = NULL;
f2ed9b32 1885 const bool utf8_target = cBOOL(DO_UTF8(sv));
2757e526 1886 I32 multiline;
f8fc2ecf 1887 RXi_GET_DECL(prog,progi);
3b0527fe 1888 regmatch_info reginfo; /* create some info to pass to regtry etc */
e9105d30 1889 regexp_paren_pair *swap = NULL;
a3621e74
YO
1890 GET_RE_DEBUG_FLAGS_DECL;
1891
7918f24d 1892 PERL_ARGS_ASSERT_REGEXEC_FLAGS;
9d4ba2ae 1893 PERL_UNUSED_ARG(data);
6eb5f6b9
JH
1894
1895 /* Be paranoid... */
1896 if (prog == NULL || startpos == NULL) {
1897 Perl_croak(aTHX_ "NULL regexp parameter");
1898 return 0;
1899 }
1900
bbe252da 1901 multiline = prog->extflags & RXf_PMf_MULTILINE;
288b8c02 1902 reginfo.prog = rx; /* Yes, sorry that this is confusing. */
2757e526 1903
f2ed9b32 1904 RX_MATCH_UTF8_set(rx, utf8_target);
1de06328 1905 DEBUG_EXECUTE_r(
f2ed9b32 1906 debug_start_match(rx, utf8_target, startpos, strend,
1de06328
YO
1907 "Matching");
1908 );
bac06658 1909
6eb5f6b9 1910 minlen = prog->minlen;
1de06328
YO
1911
1912 if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
a3621e74 1913 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
a72c7584
JH
1914 "String too short [regexec_flags]...\n"));
1915 goto phooey;
1aa99e6b 1916 }
6eb5f6b9 1917
1de06328 1918
6eb5f6b9 1919 /* Check validity of program. */
f8fc2ecf 1920 if (UCHARAT(progi->program) != REG_MAGIC) {
6eb5f6b9
JH
1921 Perl_croak(aTHX_ "corrupted regexp program");
1922 }
1923
1924 PL_reg_flags = 0;
1925 PL_reg_eval_set = 0;
1926 PL_reg_maxiter = 0;
1927
3c8556c3 1928 if (RX_UTF8(rx))
6eb5f6b9
JH
1929 PL_reg_flags |= RF_utf8;
1930
1931 /* Mark beginning of line for ^ and lookbehind. */
3b0527fe 1932 reginfo.bol = startpos; /* XXX not used ??? */
6eb5f6b9 1933 PL_bostr = strbeg;
3b0527fe 1934 reginfo.sv = sv;
6eb5f6b9
JH
1935
1936 /* Mark end of line for $ (and such) */
1937 PL_regeol = strend;
1938
1939 /* see how far we have to get to not match where we matched before */
3b0527fe 1940 reginfo.till = startpos+minend;
6eb5f6b9 1941
6eb5f6b9
JH
1942 /* If there is a "must appear" string, look for it. */
1943 s = startpos;
1944
bbe252da 1945 if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */
6eb5f6b9 1946 MAGIC *mg;
2c296965 1947 if (flags & REXEC_IGNOREPOS){ /* Means: check only at start */
58e23c8d 1948 reginfo.ganch = startpos + prog->gofs;
2c296965 1949 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
ed549f2e 1950 "GPOS IGNOREPOS: reginfo.ganch = startpos + %"UVxf"\n",(UV)prog->gofs));
2c296965 1951 } else if (sv && SvTYPE(sv) >= SVt_PVMG
6eb5f6b9 1952 && SvMAGIC(sv)
14befaf4
DM
1953 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
1954 && mg->mg_len >= 0) {
3b0527fe 1955 reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */
2c296965 1956 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
ed549f2e 1957 "GPOS MAGIC: reginfo.ganch = strbeg + %"IVdf"\n",(IV)mg->mg_len));
2c296965 1958
bbe252da 1959 if (prog->extflags & RXf_ANCH_GPOS) {
3b0527fe 1960 if (s > reginfo.ganch)
6eb5f6b9 1961 goto phooey;
58e23c8d 1962 s = reginfo.ganch - prog->gofs;
2c296965 1963 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
ed549f2e 1964 "GPOS ANCH_GPOS: s = ganch - %"UVxf"\n",(UV)prog->gofs));
c584a96e
YO
1965 if (s < strbeg)
1966 goto phooey;
6eb5f6b9
JH
1967 }
1968 }
58e23c8d 1969 else if (data) {
70685ca0 1970 reginfo.ganch = strbeg + PTR2UV(data);
2c296965
YO
1971 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1972 "GPOS DATA: reginfo.ganch= strbeg + %"UVxf"\n",PTR2UV(data)));
1973
1974 } else { /* pos() not defined */
3b0527fe 1975 reginfo.ganch = strbeg;
2c296965
YO
1976 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1977 "GPOS: reginfo.ganch = strbeg\n"));
1978 }
6eb5f6b9 1979 }
288b8c02 1980 if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
e9105d30
GG
1981 /* We have to be careful. If the previous successful match
1982 was from this regex we don't want a subsequent partially
1983 successful match to clobber the old results.
1984 So when we detect this possibility we add a swap buffer
1985 to the re, and switch the buffer each match. If we fail
1986 we switch it back, otherwise we leave it swapped.
1987 */
1988 swap = prog->offs;
1989 /* do we need a save destructor here for eval dies? */
1990 Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair);
c74340f9 1991 }
a0714e2c 1992 if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
6eb5f6b9
JH
1993 re_scream_pos_data d;
1994
1995 d.scream_olds = &scream_olds;
1996 d.scream_pos = &scream_pos;
288b8c02 1997 s = re_intuit_start(rx, sv, s, strend, flags, &d);
3fa9c3d7 1998 if (!s) {
a3621e74 1999 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
6eb5f6b9 2000 goto phooey; /* not present */
3fa9c3d7 2001 }
6eb5f6b9
JH
2002 }
2003
1de06328 2004
6eb5f6b9
JH
2005
2006 /* Simplest case: anchored match need be tried only once. */
2007 /* [unless only anchor is BOL and multiline is set] */
bbe252da 2008 if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) {
24b23f37 2009 if (s == startpos && regtry(&reginfo, &startpos))
6eb5f6b9 2010 goto got_it;
bbe252da
YO
2011 else if (multiline || (prog->intflags & PREGf_IMPLICIT)
2012 || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */
6eb5f6b9
JH
2013 {
2014 char *end;
2015
2016 if (minlen)
2017 dontbother = minlen - 1;
1aa99e6b 2018 end = HOP3c(strend, -dontbother, strbeg) - 1;
6eb5f6b9 2019 /* for multiline we only have to try after newlines */
33b8afdf 2020 if (prog->check_substr || prog->check_utf8) {
6eb5f6b9
JH
2021 if (s == startpos)
2022 goto after_try;
2023 while (1) {
24b23f37 2024 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2025 goto got_it;
2026 after_try:
5339e136 2027 if (s > end)
6eb5f6b9 2028 goto phooey;
bbe252da 2029 if (prog->extflags & RXf_USE_INTUIT) {
288b8c02 2030 s = re_intuit_start(rx, sv, s + 1, strend, flags, NULL);
6eb5f6b9
JH
2031 if (!s)
2032 goto phooey;
2033 }
2034 else
2035 s++;
2036 }
2037 } else {
2038 if (s > startpos)
2039 s--;
2040 while (s < end) {
2041 if (*s++ == '\n') { /* don't need PL_utf8skip here */
24b23f37 2042 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2043 goto got_it;
2044 }
2045 }
2046 }
2047 }
2048 goto phooey;
bbe252da 2049 } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK))
f9f4320a
YO
2050 {
2051 /* the warning about reginfo.ganch being used without intialization
bbe252da 2052 is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN
f9f4320a 2053 and we only enter this block when the same bit is set. */
58e23c8d 2054 char *tmp_s = reginfo.ganch - prog->gofs;
c584a96e
YO
2055
2056 if (tmp_s >= strbeg && regtry(&reginfo, &tmp_s))
6eb5f6b9
JH
2057 goto got_it;
2058 goto phooey;
2059 }
2060
2061 /* Messy cases: unanchored match. */
bbe252da 2062 if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
6eb5f6b9 2063 /* we have /x+whatever/ */
f2ed9b32 2064 /* it must be a one character string (XXXX Except UTF_PATTERN?) */
33b8afdf 2065 char ch;
bf93d4cc
GS
2066#ifdef DEBUGGING
2067 int did_match = 0;
2068#endif
f2ed9b32
KW
2069 if (!(utf8_target ? prog->anchored_utf8 : prog->anchored_substr))
2070 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2071 ch = SvPVX_const(utf8_target ? prog->anchored_utf8 : prog->anchored_substr)[0];
bf93d4cc 2072
f2ed9b32 2073 if (utf8_target) {
4cadc6a9 2074 REXEC_FBC_SCAN(
6eb5f6b9 2075 if (*s == ch) {
a3621e74 2076 DEBUG_EXECUTE_r( did_match = 1 );
24b23f37 2077 if (regtry(&reginfo, &s)) goto got_it;
6eb5f6b9
JH
2078 s += UTF8SKIP(s);
2079 while (s < strend && *s == ch)
2080 s += UTF8SKIP(s);
2081 }
4cadc6a9 2082 );
6eb5f6b9
JH
2083 }
2084 else {
4cadc6a9 2085 REXEC_FBC_SCAN(
6eb5f6b9 2086 if (*s == ch) {
a3621e74 2087 DEBUG_EXECUTE_r( did_match = 1 );
24b23f37 2088 if (regtry(&reginfo, &s)) goto got_it;
6eb5f6b9
JH
2089 s++;
2090 while (s < strend && *s == ch)
2091 s++;
2092 }
4cadc6a9 2093 );
6eb5f6b9 2094 }
a3621e74 2095 DEBUG_EXECUTE_r(if (!did_match)
bf93d4cc 2096 PerlIO_printf(Perl_debug_log,
b7953727
JH
2097 "Did not find anchored character...\n")
2098 );
6eb5f6b9 2099 }
a0714e2c
SS
2100 else if (prog->anchored_substr != NULL
2101 || prog->anchored_utf8 != NULL
2102 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
33b8afdf
JH
2103 && prog->float_max_offset < strend - s)) {
2104 SV *must;
2105 I32 back_max;
2106 I32 back_min;
2107 char *last;
6eb5f6b9 2108 char *last1; /* Last position checked before */
bf93d4cc
GS
2109#ifdef DEBUGGING
2110 int did_match = 0;
2111#endif
33b8afdf 2112 if (prog->anchored_substr || prog->anchored_utf8) {
f2ed9b32
KW
2113 if (!(utf8_target ? prog->anchored_utf8 : prog->anchored_substr))
2114 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2115 must = utf8_target ? prog->anchored_utf8 : prog->anchored_substr;
33b8afdf
JH
2116 back_max = back_min = prog->anchored_offset;
2117 } else {
f2ed9b32
KW
2118 if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
2119 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2120 must = utf8_target ? prog->float_utf8 : prog->float_substr;
33b8afdf
JH
2121 back_max = prog->float_max_offset;
2122 back_min = prog->float_min_offset;
2123 }
1de06328
YO
2124
2125
33b8afdf
JH
2126 if (must == &PL_sv_undef)
2127 /* could not downgrade utf8 check substring, so must fail */
2128 goto phooey;
2129
1de06328
YO
2130 if (back_min<0) {
2131 last = strend;
2132 } else {
2133 last = HOP3c(strend, /* Cannot start after this */
2134 -(I32)(CHR_SVLEN(must)
2135 - (SvTAIL(must) != 0) + back_min), strbeg);
2136 }
6eb5f6b9
JH
2137 if (s > PL_bostr)
2138 last1 = HOPc(s, -1);
2139 else
2140 last1 = s - 1; /* bogus */
2141
a0288114 2142 /* XXXX check_substr already used to find "s", can optimize if
6eb5f6b9
JH
2143 check_substr==must. */
2144 scream_pos = -1;
2145 dontbother = end_shift;
2146 strend = HOPc(strend, -dontbother);
2147 while ( (s <= last) &&
9041c2e3 2148 ((flags & REXEC_SCREAM)
1de06328 2149 ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg,
6eb5f6b9 2150 end_shift, &scream_pos, 0))
1de06328 2151 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)),
9041c2e3 2152 (unsigned char*)strend, must,
7fba1cd6 2153 multiline ? FBMrf_MULTILINE : 0))) ) {
4addbd3b 2154 /* we may be pointing at the wrong string */
07bc277f 2155 if ((flags & REXEC_SCREAM) && RXp_MATCH_COPIED(prog))
3f7c398e 2156 s = strbeg + (s - SvPVX_const(sv));
a3621e74 2157 DEBUG_EXECUTE_r( did_match = 1 );
6eb5f6b9
JH
2158 if (HOPc(s, -back_max) > last1) {
2159 last1 = HOPc(s, -back_min);
2160 s = HOPc(s, -back_max);
2161 }
2162 else {
52657f30 2163 char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
6eb5f6b9
JH
2164
2165 last1 = HOPc(s, -back_min);
52657f30 2166 s = t;
6eb5f6b9 2167 }
f2ed9b32 2168 if (utf8_target) {
6eb5f6b9 2169 while (s <= last1) {
24b23f37 2170 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2171 goto got_it;
2172 s += UTF8SKIP(s);
2173 }
2174 }
2175 else {
2176 while (s <= last1) {
24b23f37 2177 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2178 goto got_it;
2179 s++;
2180 }
2181 }
2182 }
ab3bbdeb 2183 DEBUG_EXECUTE_r(if (!did_match) {
f2ed9b32 2184 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
ab3bbdeb
YO
2185 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
2186 PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
33b8afdf 2187 ((must == prog->anchored_substr || must == prog->anchored_utf8)
bf93d4cc 2188 ? "anchored" : "floating"),
ab3bbdeb
YO
2189 quoted, RE_SV_TAIL(must));
2190 });
6eb5f6b9
JH
2191 goto phooey;
2192 }
f8fc2ecf 2193 else if ( (c = progi->regstclass) ) {
f14c76ed 2194 if (minlen) {
f8fc2ecf 2195 const OPCODE op = OP(progi->regstclass);
66e933ab 2196 /* don't bother with what can't match */
786e8c11 2197 if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
f14c76ed
RGS
2198 strend = HOPc(strend, -(minlen - 1));
2199 }
a3621e74 2200 DEBUG_EXECUTE_r({
be8e71aa 2201 SV * const prop = sv_newmortal();
32fc9b6a 2202 regprop(prog, prop, c);
0df25f3d 2203 {
f2ed9b32 2204 RE_PV_QUOTED_DECL(quoted,utf8_target,PERL_DEBUG_PAD_ZERO(1),
ab3bbdeb 2205 s,strend-s,60);
0df25f3d 2206 PerlIO_printf(Perl_debug_log,
1c8f8eb1 2207 "Matching stclass %.*s against %s (%d bytes)\n",
e4f74956 2208 (int)SvCUR(prop), SvPVX_const(prop),
ab3bbdeb 2209 quoted, (int)(strend - s));
0df25f3d 2210 }
ffc61ed2 2211 });
3b0527fe 2212 if (find_byclass(prog, c, s, strend, &reginfo))
6eb5f6b9 2213 goto got_it;
07be1b83 2214 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
d6a28714
JH
2215 }
2216 else {
2217 dontbother = 0;
a0714e2c 2218 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
33b8afdf 2219 /* Trim the end. */
d6a28714 2220 char *last;
33b8afdf
JH
2221 SV* float_real;
2222
f2ed9b32
KW
2223 if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
2224 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2225 float_real = utf8_target ? prog->float_utf8 : prog->float_substr;
d6a28714
JH
2226
2227 if (flags & REXEC_SCREAM) {
33b8afdf 2228 last = screaminstr(sv, float_real, s - strbeg,
d6a28714
JH
2229 end_shift, &scream_pos, 1); /* last one */
2230 if (!last)
ffc61ed2 2231 last = scream_olds; /* Only one occurrence. */
4addbd3b 2232 /* we may be pointing at the wrong string */
07bc277f 2233 else if (RXp_MATCH_COPIED(prog))
3f7c398e 2234 s = strbeg + (s - SvPVX_const(sv));
b8c5462f 2235 }
d6a28714
JH
2236 else {
2237 STRLEN len;
cfd0369c 2238 const char * const little = SvPV_const(float_real, len);
d6a28714 2239
33b8afdf 2240 if (SvTAIL(float_real)) {
d6a28714
JH
2241 if (memEQ(strend - len + 1, little, len - 1))
2242 last = strend - len + 1;
7fba1cd6 2243 else if (!multiline)
9041c2e3 2244 last = memEQ(strend - len, little, len)
bd61b366 2245 ? strend - len : NULL;
b8c5462f 2246 else
d6a28714
JH
2247 goto find_last;
2248 } else {
2249 find_last:
9041c2e3 2250 if (len)
d6a28714 2251 last = rninstr(s, strend, little, little + len);
b8c5462f 2252 else
a0288114 2253 last = strend; /* matching "$" */
b8c5462f 2254 }
b8c5462f 2255 }
bf93d4cc 2256 if (last == NULL) {
6bda09f9
YO
2257 DEBUG_EXECUTE_r(
2258 PerlIO_printf(Perl_debug_log,
2259 "%sCan't trim the tail, match fails (should not happen)%s\n",
2260 PL_colors[4], PL_colors[5]));
bf93d4cc
GS
2261 goto phooey; /* Should not happen! */
2262 }
d6a28714
JH
2263 dontbother = strend - last + prog->float_min_offset;
2264 }
2265 if (minlen && (dontbother < minlen))
2266 dontbother = minlen - 1;
2267 strend -= dontbother; /* this one's always in bytes! */
2268 /* We don't know much -- general case. */
f2ed9b32 2269 if (utf8_target) {
d6a28714 2270 for (;;) {
24b23f37 2271 if (regtry(&reginfo, &s))
d6a28714
JH
2272 goto got_it;
2273 if (s >= strend)
2274 break;
b8c5462f 2275 s += UTF8SKIP(s);
d6a28714
JH
2276 };
2277 }
2278 else {
2279 do {
24b23f37 2280 if (regtry(&reginfo, &s))
d6a28714
JH
2281 goto got_it;
2282 } while (s++ < strend);
2283 }
2284 }
2285
2286 /* Failure. */
2287 goto phooey;
2288
2289got_it:
e9105d30 2290 Safefree(swap);
288b8c02 2291 RX_MATCH_TAINTED_set(rx, PL_reg_flags & RF_tainted);
d6a28714 2292
19b95bf0 2293 if (PL_reg_eval_set)
4f639d21 2294 restore_pos(aTHX_ prog);
5daac39c
NC
2295 if (RXp_PAREN_NAMES(prog))
2296 (void)hv_iterinit(RXp_PAREN_NAMES(prog));
d6a28714
JH
2297
2298 /* make sure $`, $&, $', and $digit will work later */
2299 if ( !(flags & REXEC_NOT_FIRST) ) {
288b8c02 2300 RX_MATCH_COPY_FREE(rx);
d6a28714 2301 if (flags & REXEC_COPY_STR) {
be8e71aa 2302 const I32 i = PL_regeol - startpos + (stringarg - strbeg);
f8c7b90f 2303#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
2304 if ((SvIsCOW(sv)
2305 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2306 if (DEBUG_C_TEST) {
2307 PerlIO_printf(Perl_debug_log,
2308 "Copy on write: regexp capture, type %d\n",
2309 (int) SvTYPE(sv));
2310 }
2311 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
d5263905 2312 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
ed252734
NC
2313 assert (SvPOKp(prog->saved_copy));
2314 } else
2315#endif
2316 {
288b8c02 2317 RX_MATCH_COPIED_on(rx);
ed252734
NC
2318 s = savepvn(strbeg, i);
2319 prog->subbeg = s;
2320 }
d6a28714 2321 prog->sublen = i;
d6a28714
JH
2322 }
2323 else {
2324 prog->subbeg = strbeg;
2325 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2326 }
2327 }
9041c2e3 2328
d6a28714
JH
2329 return 1;
2330
2331phooey:
a3621e74 2332 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
e4584336 2333 PL_colors[4], PL_colors[5]));
d6a28714 2334 if (PL_reg_eval_set)
4f639d21 2335 restore_pos(aTHX_ prog);
e9105d30 2336 if (swap) {
c74340f9 2337 /* we failed :-( roll it back */
e9105d30
GG
2338 Safefree(prog->offs);
2339 prog->offs = swap;
2340 }
2341
d6a28714
JH
2342 return 0;
2343}
2344
6bda09f9 2345
d6a28714
JH
2346/*
2347 - regtry - try match at specific point
2348 */
2349STATIC I32 /* 0 failure, 1 success */
24b23f37 2350S_regtry(pTHX_ regmatch_info *reginfo, char **startpos)
d6a28714 2351{
97aff369 2352 dVAR;
d6a28714 2353 CHECKPOINT lastcp;
288b8c02
NC
2354 REGEXP *const rx = reginfo->prog;
2355 regexp *const prog = (struct regexp *)SvANY(rx);
f8fc2ecf 2356 RXi_GET_DECL(prog,progi);
a3621e74 2357 GET_RE_DEBUG_FLAGS_DECL;
7918f24d
NC
2358
2359 PERL_ARGS_ASSERT_REGTRY;
2360
24b23f37 2361 reginfo->cutpoint=NULL;
d6a28714 2362
bbe252da 2363 if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) {
d6a28714
JH
2364 MAGIC *mg;
2365
2366 PL_reg_eval_set = RS_init;
a3621e74 2367 DEBUG_EXECUTE_r(DEBUG_s(
b900a521
JH
2368 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
2369 (IV)(PL_stack_sp - PL_stack_base));
d6a28714 2370 ));
ea8d6ae1 2371 SAVESTACK_CXPOS();
d6a28714
JH
2372 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
2373 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
2374 SAVETMPS;
2375 /* Apparently this is not needed, judging by wantarray. */
e8347627 2376 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
d6a28714
JH
2377 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
2378
3b0527fe 2379 if (reginfo->sv) {
d6a28714 2380 /* Make $_ available to executed code. */
3b0527fe 2381 if (reginfo->sv != DEFSV) {
59f00321 2382 SAVE_DEFSV;
414bf5ae 2383 DEFSV_set(reginfo->sv);
b8c5462f 2384 }
d6a28714 2385
3b0527fe
DM
2386 if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
2387 && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
d6a28714 2388 /* prepare for quick setting of pos */
d300d9fa 2389#ifdef PERL_OLD_COPY_ON_WRITE
51a9ea20
NC
2390 if (SvIsCOW(reginfo->sv))
2391 sv_force_normal_flags(reginfo->sv, 0);
d300d9fa 2392#endif
3dab1dad 2393 mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global,
d300d9fa 2394 &PL_vtbl_mglob, NULL, 0);
d6a28714 2395 mg->mg_len = -1;
b8c5462f 2396 }
d6a28714
JH
2397 PL_reg_magic = mg;
2398 PL_reg_oldpos = mg->mg_len;
4f639d21 2399 SAVEDESTRUCTOR_X(restore_pos, prog);
d6a28714 2400 }
09687e5a 2401 if (!PL_reg_curpm) {
a02a5408 2402 Newxz(PL_reg_curpm, 1, PMOP);
09687e5a
AB
2403#ifdef USE_ITHREADS
2404 {
14a49a24 2405 SV* const repointer = &PL_sv_undef;
92313705
NC
2406 /* this regexp is also owned by the new PL_reg_curpm, which
2407 will try to free it. */
d2ece331 2408 av_push(PL_regex_padav, repointer);
09687e5a
AB
2409 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2410 PL_regex_pad = AvARRAY(PL_regex_padav);
2411 }
2412#endif
2413 }
86c29d75
NC
2414#ifdef USE_ITHREADS
2415 /* It seems that non-ithreads works both with and without this code.
2416 So for efficiency reasons it seems best not to have the code
2417 compiled when it is not needed. */
92313705
NC
2418 /* This is safe against NULLs: */
2419 ReREFCNT_dec(PM_GETRE(PL_reg_curpm));
2420 /* PM_reg_curpm owns a reference to this regexp. */
2421 ReREFCNT_inc(rx);
86c29d75 2422#endif
288b8c02 2423 PM_SETRE(PL_reg_curpm, rx);
d6a28714
JH
2424 PL_reg_oldcurpm = PL_curpm;
2425 PL_curpm = PL_reg_curpm;
07bc277f 2426 if (RXp_MATCH_COPIED(prog)) {
d6a28714
JH
2427 /* Here is a serious problem: we cannot rewrite subbeg,
2428 since it may be needed if this match fails. Thus
2429 $` inside (?{}) could fail... */
2430 PL_reg_oldsaved = prog->subbeg;
2431 PL_reg_oldsavedlen = prog->sublen;
f8c7b90f 2432#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
2433 PL_nrs = prog->saved_copy;
2434#endif
07bc277f 2435 RXp_MATCH_COPIED_off(prog);
d6a28714
JH
2436 }
2437 else
bd61b366 2438 PL_reg_oldsaved = NULL;
d6a28714
JH
2439 prog->subbeg = PL_bostr;
2440 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2441 }
24b23f37 2442 DEBUG_EXECUTE_r(PL_reg_starttry = *startpos);
f0ab9afb 2443 prog->offs[0].start = *startpos - PL_bostr;
24b23f37 2444 PL_reginput = *startpos;
d6a28714 2445 PL_reglastparen = &prog->lastparen;
a01268b5 2446 PL_reglastcloseparen = &prog->lastcloseparen;
d6a28714 2447 prog->lastparen = 0;
03994de8 2448 prog->lastcloseparen = 0;
d6a28714 2449 PL_regsize = 0;
f0ab9afb 2450 PL_regoffs = prog->offs;
d6a28714
JH
2451 if (PL_reg_start_tmpl <= prog->nparens) {
2452 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
2453 if(PL_reg_start_tmp)
2454 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2455 else
a02a5408 2456 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
d6a28714
JH
2457 }
2458
2459 /* XXXX What this code is doing here?!!! There should be no need
2460 to do this again and again, PL_reglastparen should take care of
3dd2943c 2461 this! --ilya*/
dafc8851
JH
2462
2463 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2464 * Actually, the code in regcppop() (which Ilya may be meaning by
daf18116 2465 * PL_reglastparen), is not needed at all by the test suite
225593e1
DM
2466 * (op/regexp, op/pat, op/split), but that code is needed otherwise
2467 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
2468 * Meanwhile, this code *is* needed for the
daf18116
JH
2469 * above-mentioned test suite tests to succeed. The common theme
2470 * on those tests seems to be returning null fields from matches.
225593e1 2471 * --jhi updated by dapm */
dafc8851 2472#if 1
d6a28714 2473 if (prog->nparens) {
f0ab9afb 2474 regexp_paren_pair *pp = PL_regoffs;
097eb12c 2475 register I32 i;
eb160463 2476 for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
f0ab9afb
NC
2477 ++pp;
2478 pp->start = -1;
2479 pp->end = -1;
d6a28714
JH
2480 }
2481 }
dafc8851 2482#endif
02db2b7b 2483 REGCP_SET(lastcp);
f8fc2ecf 2484 if (regmatch(reginfo, progi->program + 1)) {
f0ab9afb 2485 PL_regoffs[0].end = PL_reginput - PL_bostr;
d6a28714
JH
2486 return 1;
2487 }
24b23f37
YO
2488 if (reginfo->cutpoint)
2489 *startpos= reginfo->cutpoint;
02db2b7b 2490 REGCP_UNWIND(lastcp);
d6a28714
JH
2491 return 0;
2492}
2493
02db2b7b 2494
8ba1375e
MJD
2495#define sayYES goto yes
2496#define sayNO goto no
262b90c4 2497#define sayNO_SILENT goto no_silent
8ba1375e 2498
f9f4320a
YO
2499/* we dont use STMT_START/END here because it leads to
2500 "unreachable code" warnings, which are bogus, but distracting. */
2501#define CACHEsayNO \
c476f425
DM
2502 if (ST.cache_mask) \
2503 PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \
f9f4320a 2504 sayNO
3298f257 2505
a3621e74 2506/* this is used to determine how far from the left messages like
265c4333
YO
2507 'failed...' are printed. It should be set such that messages
2508 are inline with the regop output that created them.
a3621e74 2509*/
265c4333 2510#define REPORT_CODE_OFF 32
a3621e74
YO
2511
2512
40a82448
DM
2513#define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
2514#define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
9e137952 2515
86545054
DM
2516#define SLAB_FIRST(s) (&(s)->states[0])
2517#define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
2518
5d9a96ca
DM
2519/* grab a new slab and return the first slot in it */
2520
2521STATIC regmatch_state *
2522S_push_slab(pTHX)
2523{
a35a87e7 2524#if PERL_VERSION < 9 && !defined(PERL_CORE)
54df2634
NC
2525 dMY_CXT;
2526#endif
5d9a96ca
DM
2527 regmatch_slab *s = PL_regmatch_slab->next;
2528 if (!s) {
2529 Newx(s, 1, regmatch_slab);
2530 s->prev = PL_regmatch_slab;
2531 s->next = NULL;
2532 PL_regmatch_slab->next = s;
2533 }
2534 PL_regmatch_slab = s;
86545054 2535 return SLAB_FIRST(s);
5d9a96ca 2536}
5b47454d 2537
95b24440 2538
40a82448
DM
2539/* push a new state then goto it */
2540
2541#define PUSH_STATE_GOTO(state, node) \
2542 scan = node; \
2543 st->resume_state = state; \
2544 goto push_state;
2545
2546/* push a new state with success backtracking, then goto it */
2547
2548#define PUSH_YES_STATE_GOTO(state, node) \
2549 scan = node; \
2550 st->resume_state = state; \
2551 goto push_yes_state;
2552
aa283a38 2553
aa283a38 2554
d6a28714 2555/*
95b24440 2556
bf1f174e
DM
2557regmatch() - main matching routine
2558
2559This is basically one big switch statement in a loop. We execute an op,
2560set 'next' to point the next op, and continue. If we come to a point which
2561we may need to backtrack to on failure such as (A|B|C), we push a
2562backtrack state onto the backtrack stack. On failure, we pop the top
2563state, and re-enter the loop at the state indicated. If there are no more
2564states to pop, we return failure.
2565
2566Sometimes we also need to backtrack on success; for example /A+/, where
2567after successfully matching one A, we need to go back and try to
2568match another one; similarly for lookahead assertions: if the assertion
2569completes successfully, we backtrack to the state just before the assertion
2570and then carry on. In these cases, the pushed state is marked as
2571'backtrack on success too'. This marking is in fact done by a chain of
2572pointers, each pointing to the previous 'yes' state. On success, we pop to
2573the nearest yes state, discarding any intermediate failure-only states.
2574Sometimes a yes state is pushed just to force some cleanup code to be
2575called at the end of a successful match or submatch; e.g. (??{$re}) uses
2576it to free the inner regex.
2577
2578Note that failure backtracking rewinds the cursor position, while
2579success backtracking leaves it alone.
2580
2581A pattern is complete when the END op is executed, while a subpattern
2582such as (?=foo) is complete when the SUCCESS op is executed. Both of these
2583ops trigger the "pop to last yes state if any, otherwise return true"
2584behaviour.
2585
2586A common convention in this function is to use A and B to refer to the two
2587subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
2588the subpattern to be matched possibly multiple times, while B is the entire
2589rest of the pattern. Variable and state names reflect this convention.
2590
2591The states in the main switch are the union of ops and failure/success of
2592substates associated with with that op. For example, IFMATCH is the op
2593that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
2594'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
2595successfully matched A and IFMATCH_A_fail is a state saying that we have
2596just failed to match A. Resume states always come in pairs. The backtrack
2597state we push is marked as 'IFMATCH_A', but when that is popped, we resume
2598at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
2599on success or failure.
2600
2601The struct that holds a backtracking state is actually a big union, with
2602one variant for each major type of op. The variable st points to the
2603top-most backtrack struct. To make the code clearer, within each
2604block of code we #define ST to alias the relevant union.
2605
2606Here's a concrete example of a (vastly oversimplified) IFMATCH
2607implementation:
2608
2609 switch (state) {
2610 ....
2611
2612#define ST st->u.ifmatch
2613
2614 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2615 ST.foo = ...; // some state we wish to save
95b24440 2616 ...
bf1f174e
DM
2617 // push a yes backtrack state with a resume value of
2618 // IFMATCH_A/IFMATCH_A_fail, then continue execution at the
2619 // first node of A:
2620 PUSH_YES_STATE_GOTO(IFMATCH_A, A);
2621 // NOTREACHED
2622
2623 case IFMATCH_A: // we have successfully executed A; now continue with B
2624 next = B;
2625 bar = ST.foo; // do something with the preserved value
2626 break;
2627
2628 case IFMATCH_A_fail: // A failed, so the assertion failed
2629 ...; // do some housekeeping, then ...
2630 sayNO; // propagate the failure
2631
2632#undef ST
95b24440 2633
bf1f174e
DM
2634 ...
2635 }
95b24440 2636
bf1f174e
DM
2637For any old-timers reading this who are familiar with the old recursive
2638approach, the code above is equivalent to:
95b24440 2639
bf1f174e
DM
2640 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2641 {
2642 int foo = ...
95b24440 2643 ...
bf1f174e
DM
2644 if (regmatch(A)) {
2645 next = B;
2646 bar = foo;
2647 break;
95b24440 2648 }
bf1f174e
DM
2649 ...; // do some housekeeping, then ...
2650 sayNO; // propagate the failure
95b24440 2651 }
bf1f174e
DM
2652
2653The topmost backtrack state, pointed to by st, is usually free. If you
2654want to claim it, populate any ST.foo fields in it with values you wish to
2655save, then do one of
2656
2657 PUSH_STATE_GOTO(resume_state, node);
2658 PUSH_YES_STATE_GOTO(resume_state, node);
2659
2660which sets that backtrack state's resume value to 'resume_state', pushes a
2661new free entry to the top of the backtrack stack, then goes to 'node'.
2662On backtracking, the free slot is popped, and the saved state becomes the
2663new free state. An ST.foo field in this new top state can be temporarily
2664accessed to retrieve values, but once the main loop is re-entered, it
2665becomes available for reuse.
2666
2667Note that the depth of the backtrack stack constantly increases during the
2668left-to-right execution of the pattern, rather than going up and down with
2669the pattern nesting. For example the stack is at its maximum at Z at the
2670end of the pattern, rather than at X in the following:
2671
2672 /(((X)+)+)+....(Y)+....Z/
2673
2674The only exceptions to this are lookahead/behind assertions and the cut,
2675(?>A), which pop all the backtrack states associated with A before
2676continuing.
2677
2678Bascktrack state structs are allocated in slabs of about 4K in size.
2679PL_regmatch_state and st always point to the currently active state,
2680and PL_regmatch_slab points to the slab currently containing
2681PL_regmatch_state. The first time regmatch() is called, the first slab is
2682allocated, and is never freed until interpreter destruction. When the slab
2683is full, a new one is allocated and chained to the end. At exit from
2684regmatch(), slabs allocated since entry are freed.
2685
2686*/
95b24440 2687
40a82448 2688
5bc10b2c 2689#define DEBUG_STATE_pp(pp) \
265c4333 2690 DEBUG_STATE_r({ \
f2ed9b32 2691 DUMP_EXEC_POS(locinput, scan, utf8_target); \
5bc10b2c 2692 PerlIO_printf(Perl_debug_log, \
5d458dd8 2693 " %*s"pp" %s%s%s%s%s\n", \
5bc10b2c 2694 depth*2, "", \
13d6edb4 2695 PL_reg_name[st->resume_state], \
5d458dd8
YO
2696 ((st==yes_state||st==mark_state) ? "[" : ""), \
2697 ((st==yes_state) ? "Y" : ""), \
2698 ((st==mark_state) ? "M" : ""), \
2699 ((st==yes_state||st==mark_state) ? "]" : "") \
2700 ); \
265c4333 2701 });
5bc10b2c 2702
40a82448 2703
3dab1dad 2704#define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)
95b24440 2705
3df15adc 2706#ifdef DEBUGGING
5bc10b2c 2707
ab3bbdeb 2708STATIC void
f2ed9b32 2709S_debug_start_match(pTHX_ const REGEXP *prog, const bool utf8_target,
ab3bbdeb
YO
2710 const char *start, const char *end, const char *blurb)
2711{
efd26800 2712 const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;
7918f24d
NC
2713
2714 PERL_ARGS_ASSERT_DEBUG_START_MATCH;
2715
ab3bbdeb
YO
2716 if (!PL_colorset)
2717 reginitcolors();
2718 {
2719 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
d2c6dc5e 2720 RX_PRECOMP_const(prog), RX_PRELEN(prog), 60);
ab3bbdeb 2721
f2ed9b32 2722 RE_PV_QUOTED_DECL(s1, utf8_target, PERL_DEBUG_PAD_ZERO(1),
ab3bbdeb
YO
2723 start, end - start, 60);
2724
2725 PerlIO_printf(Perl_debug_log,
2726 "%s%s REx%s %s against %s\n",
2727 PL_colors[4], blurb, PL_colors[5], s0, s1);
2728
f2ed9b32 2729 if (utf8_target||utf8_pat)
1de06328
YO
2730 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
2731 utf8_pat ? "pattern" : "",
f2ed9b32
KW
2732 utf8_pat && utf8_target ? " and " : "",
2733 utf8_target ? "string" : ""
ab3bbdeb
YO
2734 );
2735 }
2736}
3df15adc
YO
2737
2738STATIC void
786e8c11
YO
2739S_dump_exec_pos(pTHX_ const char *locinput,
2740 const regnode *scan,
2741 const char *loc_regeol,
2742 const char *loc_bostr,
2743 const char *loc_reg_starttry,
f2ed9b32 2744 const bool utf8_target)
07be1b83 2745{
786e8c11 2746 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
07be1b83 2747 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
786e8c11 2748 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
07be1b83
YO
2749 /* The part of the string before starttry has one color
2750 (pref0_len chars), between starttry and current
2751 position another one (pref_len - pref0_len chars),
2752 after the current position the third one.
2753 We assume that pref0_len <= pref_len, otherwise we
2754 decrease pref0_len. */
786e8c11
YO
2755 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
2756 ? (5 + taill) - l : locinput - loc_bostr;
07be1b83
YO
2757 int pref0_len;
2758
7918f24d
NC
2759 PERL_ARGS_ASSERT_DUMP_EXEC_POS;
2760
f2ed9b32 2761 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
07be1b83 2762 pref_len++;
786e8c11
YO
2763 pref0_len = pref_len - (locinput - loc_reg_starttry);
2764 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
2765 l = ( loc_regeol - locinput > (5 + taill) - pref_len
2766 ? (5 + taill) - pref_len : loc_regeol - locinput);
f2ed9b32 2767 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
07be1b83
YO
2768 l--;
2769 if (pref0_len < 0)
2770 pref0_len = 0;
2771 if (pref0_len > pref_len)
2772 pref0_len = pref_len;
2773 {
f2ed9b32 2774 const int is_uni = (utf8_target && OP(scan) != CANY) ? 1 : 0;
0df25f3d 2775
ab3bbdeb 2776 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
1de06328 2777 (locinput - pref_len),pref0_len, 60, 4, 5);
0df25f3d 2778
ab3bbdeb 2779 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
3df15adc 2780 (locinput - pref_len + pref0_len),
1de06328 2781 pref_len - pref0_len, 60, 2, 3);
0df25f3d 2782
ab3bbdeb 2783 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
1de06328 2784 locinput, loc_regeol - locinput, 10, 0, 1);
0df25f3d 2785
1de06328 2786 const STRLEN tlen=len0+len1+len2;
3df15adc 2787 PerlIO_printf(Perl_debug_log,
ab3bbdeb 2788 "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
786e8c11 2789 (IV)(locinput - loc_bostr),
07be1b83 2790 len0, s0,
07be1b83 2791 len1, s1,
07be1b83 2792 (docolor ? "" : "> <"),
07be1b83 2793 len2, s2,
f9f4320a 2794 (int)(tlen > 19 ? 0 : 19 - tlen),
07be1b83
YO
2795 "");
2796 }
2797}
3df15adc 2798
07be1b83
YO
2799#endif
2800
0a4db386
YO
2801/* reg_check_named_buff_matched()
2802 * Checks to see if a named buffer has matched. The data array of
2803 * buffer numbers corresponding to the buffer is expected to reside
2804 * in the regexp->data->data array in the slot stored in the ARG() of
2805 * node involved. Note that this routine doesn't actually care about the
2806 * name, that information is not preserved from compilation to execution.
2807 * Returns the index of the leftmost defined buffer with the given name
2808 * or 0 if non of the buffers matched.
2809 */
2810STATIC I32
7918f24d
NC
2811S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan)
2812{
0a4db386 2813 I32 n;
f8fc2ecf 2814 RXi_GET_DECL(rex,rexi);
ad64d0ec 2815 SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
0a4db386 2816 I32 *nums=(I32*)SvPVX(sv_dat);
7918f24d
NC
2817
2818 PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED;
2819
0a4db386
YO
2820 for ( n=0; n<SvIVX(sv_dat); n++ ) {
2821 if ((I32)*PL_reglastparen >= nums[n] &&
f0ab9afb 2822 PL_regoffs[nums[n]].end != -1)
0a4db386
YO
2823 {
2824 return nums[n];
2825 }
2826 }
2827 return 0;
2828}
2829
2f554ef7
DM
2830
2831/* free all slabs above current one - called during LEAVE_SCOPE */
2832
2833STATIC void
2834S_clear_backtrack_stack(pTHX_ void *p)
2835{
2836 regmatch_slab *s = PL_regmatch_slab->next;
2837 PERL_UNUSED_ARG(p);
2838
2839 if (!s)
2840 return;
2841 PL_regmatch_slab->next = NULL;
2842 while (s) {
2843 regmatch_slab * const osl = s;
2844 s = s->next;
2845 Safefree(osl);
2846 }
2847}
2848
2849
28d8d7f4
YO
2850#define SETREX(Re1,Re2) \
2851 if (PL_reg_eval_set) PM_SETRE((PL_reg_curpm), (Re2)); \
2852 Re1 = (Re2)
2853
d6a28714 2854STATIC I32 /* 0 failure, 1 success */
24b23f37 2855S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
d6a28714 2856{
a35a87e7 2857#if PERL_VERSION < 9 && !defined(PERL_CORE)
54df2634
NC
2858 dMY_CXT;
2859#endif
27da23d5 2860 dVAR;
f2ed9b32 2861 register const bool utf8_target = PL_reg_match_utf8;
4ad0818d 2862 const U32 uniflags = UTF8_ALLOW_DEFAULT;
288b8c02
NC
2863 REGEXP *rex_sv = reginfo->prog;
2864 regexp *rex = (struct regexp *)SvANY(rex_sv);
f8fc2ecf 2865 RXi_GET_DECL(rex,rexi);
2f554ef7 2866 I32 oldsave;
5d9a96ca
DM
2867 /* the current state. This is a cached copy of PL_regmatch_state */
2868 register regmatch_state *st;
5d9a96ca
DM
2869 /* cache heavy used fields of st in registers */
2870 register regnode *scan;
2871 register regnode *next;
438e9bae 2872 register U32 n = 0; /* general value; init to avoid compiler warning */
24d3c4a9 2873 register I32 ln = 0; /* len or last; init to avoid compiler warning */
5d9a96ca 2874 register char *locinput = PL_reginput;
5d9a96ca 2875 register I32 nextchr; /* is always set to UCHARAT(locinput) */
24d3c4a9 2876
b69b0499 2877 bool result = 0; /* return value of S_regmatch */
24d3c4a9 2878 int depth = 0; /* depth of backtrack stack */
4b196cd4
YO
2879 U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
2880 const U32 max_nochange_depth =
2881 (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
2882 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
77cb431f
DM
2883 regmatch_state *yes_state = NULL; /* state to pop to on success of
2884 subpattern */
e2e6a0f1
YO
2885 /* mark_state piggy backs on the yes_state logic so that when we unwind
2886 the stack on success we can update the mark_state as we go */
2887 regmatch_state *mark_state = NULL; /* last mark state we have seen */
faec1544 2888 regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
b8591aee 2889 struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */
40a82448 2890 U32 state_num;
5d458dd8
YO
2891 bool no_final = 0; /* prevent failure from backtracking? */
2892 bool do_cutgroup = 0; /* no_final only until next branch/trie entry */
e2e6a0f1 2893 char *startpoint = PL_reginput;
5d458dd8
YO
2894 SV *popmark = NULL; /* are we looking for a mark? */
2895 SV *sv_commit = NULL; /* last mark name seen in failure */
2896 SV *sv_yes_mark = NULL; /* last mark name we have seen
2897 during a successfull match */
2898 U32 lastopen = 0; /* last open we saw */
2899 bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;
19b95bf0 2900 SV* const oreplsv = GvSV(PL_replgv);
24d3c4a9
DM
2901 /* these three flags are set by various ops to signal information to
2902 * the very next op. They have a useful lifetime of exactly one loop
2903 * iteration, and are not preserved or restored by state pushes/pops
2904 */
2905 bool sw = 0; /* the condition value in (?(cond)a|b) */
2906 bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */
2907 int logical = 0; /* the following EVAL is:
2908 0: (?{...})
2909 1: (?(?{...})X|Y)
2910 2: (??{...})
2911 or the following IFMATCH/UNLESSM is:
2912 false: plain (?=foo)
2913 true: used as a condition: (?(?=foo))
2914 */
95b24440 2915#ifdef DEBUGGING
e68ec53f 2916 GET_RE_DEBUG_FLAGS_DECL;
d6a28714
JH
2917#endif
2918
7918f24d
NC
2919 PERL_ARGS_ASSERT_REGMATCH;
2920
3b57cd43 2921 DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
24b23f37 2922 PerlIO_printf(Perl_debug_log,"regmatch start\n");
3b57cd43 2923 }));
5d9a96ca
DM
2924 /* on first ever call to regmatch, allocate first slab */
2925 if (!PL_regmatch_slab) {
2926 Newx(PL_regmatch_slab, 1, regmatch_slab);
2927 PL_regmatch_slab->prev = NULL;
2928 PL_regmatch_slab->next = NULL;
86545054 2929 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
5d9a96ca
DM
2930 }
2931
2f554ef7
DM
2932 oldsave = PL_savestack_ix;
2933 SAVEDESTRUCTOR_X(S_clear_backtrack_stack, NULL);
2934 SAVEVPTR(PL_regmatch_slab);
2935 SAVEVPTR(PL_regmatch_state);
5d9a96ca
DM
2936
2937 /* grab next free state slot */
2938 st = ++PL_regmatch_state;
86545054 2939 if (st > SLAB_LAST(PL_regmatch_slab))
5d9a96ca
DM
2940 st = PL_regmatch_state = S_push_slab(aTHX);
2941
d6a28714
JH
2942 /* Note that nextchr is a byte even in UTF */
2943 nextchr = UCHARAT(locinput);
2944 scan = prog;
2945 while (scan != NULL) {
8ba1375e 2946
a3621e74 2947 DEBUG_EXECUTE_r( {
6136c704 2948 SV * const prop = sv_newmortal();
1de06328 2949 regnode *rnext=regnext(scan);
f2ed9b32 2950 DUMP_EXEC_POS( locinput, scan, utf8_target );
32fc9b6a 2951 regprop(rex, prop, scan);
07be1b83
YO
2952
2953 PerlIO_printf(Perl_debug_log,
2954 "%3"IVdf":%*s%s(%"IVdf")\n",
f8fc2ecf 2955 (IV)(scan - rexi->program), depth*2, "",
07be1b83 2956 SvPVX_const(prop),
1de06328 2957 (PL_regkind[OP(scan)] == END || !rnext) ?
f8fc2ecf 2958 0 : (IV)(rnext - rexi->program));
2a782b5b 2959 });
d6a28714
JH
2960
2961 next = scan + NEXT_OFF(scan);
2962 if (next == scan)
2963 next = NULL;
40a82448 2964 state_num = OP(scan);
d6a28714 2965
40a82448 2966 reenter_switch:
34a81e2b
B
2967
2968 assert(PL_reglastparen == &rex->lastparen);
2969 assert(PL_reglastcloseparen == &rex->lastcloseparen);
2970 assert(PL_regoffs == rex->offs);
2971
40a82448 2972 switch (state_num) {
d6a28714 2973 case BOL:
7fba1cd6 2974 if (locinput == PL_bostr)
d6a28714 2975 {
3b0527fe 2976 /* reginfo->till = reginfo->bol; */
b8c5462f
JH
2977 break;
2978 }
d6a28714
JH
2979 sayNO;
2980 case MBOL:
12d33761
HS
2981 if (locinput == PL_bostr ||
2982 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
d6a28714 2983 {
b8c5462f
JH
2984 break;
2985 }
d6a28714
JH
2986 sayNO;
2987 case SBOL:
c2a73568 2988 if (locinput == PL_bostr)
b8c5462f 2989 break;
d6a28714
JH
2990 sayNO;
2991 case GPOS:
3b0527fe 2992 if (locinput == reginfo->ganch)
d6a28714
JH
2993 break;
2994 sayNO;
ee9b8eae
YO
2995
2996 case KEEPS:
2997 /* update the startpoint */
f0ab9afb 2998 st->u.keeper.val = PL_regoffs[0].start;
ee9b8eae 2999 PL_reginput = locinput;
f0ab9afb 3000 PL_regoffs[0].start = locinput - PL_bostr;
ee9b8eae
YO
3001 PUSH_STATE_GOTO(KEEPS_next, next);
3002 /*NOT-REACHED*/
3003 case KEEPS_next_fail:
3004 /* rollback the start point change */
f0ab9afb 3005 PL_regoffs[0].start = st->u.keeper.val;
ee9b8eae
YO
3006 sayNO_SILENT;
3007 /*NOT-REACHED*/
d6a28714 3008 case EOL:
d6a28714
JH
3009 goto seol;
3010 case MEOL:
d6a28714 3011 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 3012 sayNO;
b8c5462f 3013 break;
d6a28714
JH
3014 case SEOL:
3015 seol:
3016 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 3017 sayNO;
d6a28714 3018 if (PL_regeol - locinput > 1)
b8c5462f 3019 sayNO;
b8c5462f 3020 break;
d6a28714
JH
3021 case EOS:
3022 if (PL_regeol != locinput)
b8c5462f 3023 sayNO;
d6a28714 3024 break;
ffc61ed2 3025 case SANY:
d6a28714 3026 if (!nextchr && locinput >= PL_regeol)
4633a7c4 3027 sayNO;
f2ed9b32 3028 if (utf8_target) {
f33976b4
DB
3029 locinput += PL_utf8skip[nextchr];
3030 if (locinput > PL_regeol)
3031 sayNO;
3032 nextchr = UCHARAT(locinput);
3033 }
3034 else
3035 nextchr = UCHARAT(++locinput);
3036 break;
3037 case CANY:
3038 if (!nextchr && locinput >= PL_regeol)
3039 sayNO;
b8c5462f 3040 nextchr = UCHARAT(++locinput);
a0d0e21e 3041 break;
ffc61ed2 3042 case REG_ANY:
1aa99e6b
IH
3043 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
3044 sayNO;
f2ed9b32 3045 if (utf8_target) {
b8c5462f 3046 locinput += PL_utf8skip[nextchr];
d6a28714
JH
3047 if (locinput > PL_regeol)
3048 sayNO;
a0ed51b3 3049 nextchr = UCHARAT(locinput);
a0ed51b3 3050 }
1aa99e6b
IH
3051 else
3052 nextchr = UCHARAT(++locinput);
a0ed51b3 3053 break;
166ba7cd
DM
3054
3055#undef ST
3056#define ST st->u.trie
786e8c11
YO
3057 case TRIEC:
3058 /* In this case the charclass data is available inline so
3059 we can fail fast without a lot of extra overhead.
3060 */
f2ed9b32 3061 if (scan->flags == EXACT || !utf8_target) {
786e8c11
YO
3062 if(!ANYOF_BITMAP_TEST(scan, *locinput)) {
3063 DEBUG_EXECUTE_r(
3064 PerlIO_printf(Perl_debug_log,
3065 "%*s %sfailed to match trie start class...%s\n",
5bc10b2c 3066 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
786e8c11
YO
3067 );
3068 sayNO_SILENT;
3069 /* NOTREACHED */
3070 }
3071 }
3072 /* FALL THROUGH */
5b47454d 3073 case TRIE:
2e64971a
DM
3074 /* the basic plan of execution of the trie is:
3075 * At the beginning, run though all the states, and
3076 * find the longest-matching word. Also remember the position
3077 * of the shortest matching word. For example, this pattern:
3078 * 1 2 3 4 5
3079 * ab|a|x|abcd|abc
3080 * when matched against the string "abcde", will generate
3081 * accept states for all words except 3, with the longest
3082 * matching word being 4, and the shortest being 1 (with
3083 * the position being after char 1 of the string).
3084 *
3085 * Then for each matching word, in word order (i.e. 1,2,4,5),
3086 * we run the remainder of the pattern; on each try setting
3087 * the current position to the character following the word,
3088 * returning to try the next word on failure.
3089 *
3090 * We avoid having to build a list of words at runtime by
3091 * using a compile-time structure, wordinfo[].prev, which
3092 * gives, for each word, the previous accepting word (if any).
3093 * In the case above it would contain the mappings 1->2, 2->0,
3094 * 3->0, 4->5, 5->1. We can use this table to generate, from
3095 * the longest word (4 above), a list of all words, by
3096 * following the list of prev pointers; this gives us the
3097 * unordered list 4,5,1,2. Then given the current word we have
3098 * just tried, we can go through the list and find the
3099 * next-biggest word to try (so if we just failed on word 2,
3100 * the next in the list is 4).
3101 *
3102 * Since at runtime we don't record the matching position in
3103 * the string for each word, we have to work that out for
3104 * each word we're about to process. The wordinfo table holds
3105 * the character length of each word; given that we recorded
3106 * at the start: the position of the shortest word and its
3107 * length in chars, we just need to move the pointer the
3108 * difference between the two char lengths. Depending on
3109 * Unicode status and folding, that's cheap or expensive.
3110 *
3111 * This algorithm is optimised for the case where are only a
3112 * small number of accept states, i.e. 0,1, or maybe 2.
3113 * With lots of accepts states, and having to try all of them,
3114 * it becomes quadratic on number of accept states to find all
3115 * the next words.
3116 */
3117
3dab1dad 3118 {
07be1b83 3119 /* what type of TRIE am I? (utf8 makes this contextual) */
a0a388a1 3120 DECL_TRIE_TYPE(scan);
3dab1dad
YO
3121
3122 /* what trie are we using right now */
be8e71aa 3123 reg_trie_data * const trie
f8fc2ecf 3124 = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
85fbaab2 3125 HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]);
3dab1dad 3126 U32 state = trie->startstate;
166ba7cd 3127
3dab1dad
YO
3128 if (trie->bitmap && trie_type != trie_utf8_fold &&
3129 !TRIE_BITMAP_TEST(trie,*locinput)
3130 ) {
3131 if (trie->states[ state ].wordnum) {
3132 DEBUG_EXECUTE_r(
3133 PerlIO_printf(Perl_debug_log,
3134 "%*s %smatched empty string...%s\n",
5bc10b2c 3135 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3dab1dad
YO
3136 );
3137 break;
3138 } else {
3139 DEBUG_EXECUTE_r(
3140 PerlIO_printf(Perl_debug_log,
786e8c11 3141 "%*s %sfailed to match trie start class...%s\n",
5bc10b2c 3142 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3dab1dad
YO
3143 );
3144 sayNO_SILENT;
3145 }
3146 }
166ba7cd 3147
786e8c11
YO
3148 {
3149 U8 *uc = ( U8* )locinput;
3150
3151 STRLEN len = 0;
3152 STRLEN foldlen = 0;
3153 U8 *uscan = (U8*)NULL;
786e8c11 3154 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2e64971a
DM
3155 U32 charcount = 0; /* how many input chars we have matched */
3156 U32 accepted = 0; /* have we seen any accepting states? */
786e8c11 3157
786e8c11
YO
3158 ST.B = next;
3159 ST.jump = trie->jump;
786e8c11 3160 ST.me = scan;
2e64971a
DM
3161 ST.firstpos = NULL;
3162 ST.longfold = FALSE; /* char longer if folded => it's harder */
3163 ST.nextword = 0;
3164
3165 /* fully traverse the TRIE; note the position of the
3166 shortest accept state and the wordnum of the longest
3167 accept state */
07be1b83 3168
a3621e74 3169 while ( state && uc <= (U8*)PL_regeol ) {
786e8c11 3170 U32 base = trie->states[ state ].trans.base;
f9f4320a 3171 UV uvc = 0;
acb909b4 3172 U16 charid = 0;
2e64971a
DM
3173 U16 wordnum;
3174 wordnum = trie->states[ state ].wordnum;
3175
3176 if (wordnum) { /* it's an accept state */
3177 if (!accepted) {
3178 accepted = 1;
3179 /* record first match position */
3180 if (ST.longfold) {
3181 ST.firstpos = (U8*)locinput;
3182 ST.firstchars = 0;
5b47454d 3183 }
2e64971a
DM
3184 else {
3185 ST.firstpos = uc;
3186 ST.firstchars = charcount;
3187 }
3188 }
3189 if (!ST.nextword || wordnum < ST.nextword)
3190 ST.nextword = wordnum;
3191 ST.topword = wordnum;
786e8c11 3192 }
a3621e74 3193
07be1b83 3194 DEBUG_TRIE_EXECUTE_r({
f2ed9b32 3195 DUMP_EXEC_POS( (char *)uc, scan, utf8_target );
a3621e74 3196 PerlIO_printf( Perl_debug_log,
2e64971a 3197 "%*s %sState: %4"UVxf" Accepted: %c ",
5bc10b2c 3198 2+depth * 2, "", PL_colors[4],
2e64971a 3199 (UV)state, (accepted ? 'Y' : 'N'));
07be1b83 3200 });
a3621e74 3201
2e64971a 3202 /* read a char and goto next state */
a3621e74 3203 if ( base ) {
6dd2be57 3204 I32 offset;
55eed653
NC
3205 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
3206 uscan, len, uvc, charid, foldlen,
3207 foldbuf, uniflags);
2e64971a
DM
3208 charcount++;
3209 if (foldlen>0)
3210 ST.longfold = TRUE;
5b47454d 3211 if (charid &&
6dd2be57
DM
3212 ( ((offset =
3213 base + charid - 1 - trie->uniquecharcount)) >= 0)
3214
3215 && ((U32)offset < trie->lasttrans)
3216 && trie->trans[offset].check == state)
5b47454d 3217 {
6dd2be57 3218 state = trie->trans[offset].next;
5b47454d
DM
3219 }
3220 else {
3221 state = 0;
3222 }
3223 uc += len;
3224
3225 }
3226 else {
a3621e74
YO
3227 state = 0;
3228 }
3229 DEBUG_TRIE_EXECUTE_r(
e4584336 3230 PerlIO_printf( Perl_debug_log,
786e8c11 3231 "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
e4584336 3232 charid, uvc, (UV)state, PL_colors[5] );
a3621e74
YO
3233 );
3234 }
2e64971a 3235 if (!accepted)
a3621e74 3236 sayNO;
a3621e74 3237
2e64971a
DM
3238 /* calculate total number of accept states */
3239 {
3240 U16 w = ST.topword;
3241 accepted = 0;
3242 while (w) {
3243 w = trie->wordinfo[w].prev;
3244 accepted++;
3245 }
3246 ST.accepted = accepted;
3247 }
3248
166ba7cd
DM
3249 DEBUG_EXECUTE_r(
3250 PerlIO_printf( Perl_debug_log,
3251 "%*s %sgot %"IVdf" possible matches%s\n",
5bc10b2c 3252 REPORT_CODE_OFF + depth * 2, "",
166ba7cd
DM
3253 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
3254 );
2e64971a 3255 goto trie_first_try; /* jump into the fail handler */
786e8c11 3256 }}
fae667d5 3257 /* NOTREACHED */
2e64971a
DM
3258
3259 case TRIE_next_fail: /* we failed - try next alternative */
fae667d5
YO
3260 if ( ST.jump) {
3261 REGCP_UNWIND(ST.cp);
3262 for (n = *PL_reglastparen; n > ST.lastparen; n--)
f0ab9afb 3263 PL_regoffs[n].end = -1;
fae667d5
YO
3264 *PL_reglastparen = n;
3265 }
2e64971a
DM
3266 if (!--ST.accepted) {
3267 DEBUG_EXECUTE_r({
3268 PerlIO_printf( Perl_debug_log,
3269 "%*s %sTRIE failed...%s\n",
3270 REPORT_CODE_OFF+depth*2, "",
3271 PL_colors[4],
3272 PL_colors[5] );
3273 });
3274 sayNO_SILENT;
3275 }
3276 {
3277 /* Find next-highest word to process. Note that this code
3278 * is O(N^2) per trie run (O(N) per branch), so keep tight */
d9a396a3
DM
3279 register U16 min = 0;
3280 register U16 word;
2e64971a
DM
3281 register U16 const nextword = ST.nextword;
3282 register reg_trie_wordinfo * const wordinfo
3283 = ((reg_trie_data*)rexi->data->data[ARG(ST.me)])->wordinfo;
3284 for (word=ST.topword; word; word=wordinfo[word].prev) {
3285 if (word > nextword && (!min || word < min))
3286 min = word;
3287 }
3288 ST.nextword = min;
3289 }
3290
fae667d5 3291 trie_first_try:
5d458dd8
YO
3292 if (do_cutgroup) {
3293 do_cutgroup = 0;
3294 no_final = 0;
3295 }
fae667d5
YO
3296
3297 if ( ST.jump) {
3298 ST.lastparen = *PL_reglastparen;
3299 REGCP_SET(ST.cp);
2e64971a 3300 }
a3621e74 3301
2e64971a 3302 /* find start char of end of current word */
166ba7cd 3303 {
2e64971a
DM
3304 U32 chars; /* how many chars to skip */
3305 U8 *uc = ST.firstpos;
3306 reg_trie_data * const trie
3307 = (reg_trie_data*)rexi->data->data[ARG(ST.me)];
3308
3309 assert((trie->wordinfo[ST.nextword].len - trie->prefixlen)
3310 >= ST.firstchars);
3311 chars = (trie->wordinfo[ST.nextword].len - trie->prefixlen)
3312 - ST.firstchars;
3313
3314 if (ST.longfold) {
3315 /* the hard option - fold each char in turn and find
3316 * its folded length (which may be different */
3317 U8 foldbuf[UTF8_MAXBYTES_CASE + 1];
3318 STRLEN foldlen;
3319 STRLEN len;
d9a396a3 3320 UV uvc;
2e64971a
DM
3321 U8 *uscan;
3322
3323 while (chars) {
f2ed9b32 3324 if (utf8_target) {
2e64971a
DM
3325 uvc = utf8n_to_uvuni((U8*)uc, UTF8_MAXLEN, &len,
3326 uniflags);
3327 uc += len;
3328 }
3329 else {
3330 uvc = *uc;
3331 uc++;
3332 }
3333 uvc = to_uni_fold(uvc, foldbuf, &foldlen);
3334 uscan = foldbuf;
3335 while (foldlen) {
3336 if (!--chars)
3337 break;
3338 uvc = utf8n_to_uvuni(uscan, UTF8_MAXLEN, &len,
3339 uniflags);
3340 uscan += len;
3341 foldlen -= len;
3342 }
3343 }
a3621e74 3344 }
2e64971a 3345 else {
f2ed9b32 3346 if (utf8_target)
2e64971a
DM
3347 while (chars--)
3348 uc += UTF8SKIP(uc);
3349 else
3350 uc += chars;
3351 }
3352 PL_reginput = (char *)uc;
3353 }
166ba7cd 3354
2e64971a
DM
3355 scan = (ST.jump && ST.jump[ST.nextword])
3356 ? ST.me + ST.jump[ST.nextword]
3357 : ST.B;
166ba7cd 3358
2e64971a
DM
3359 DEBUG_EXECUTE_r({
3360 PerlIO_printf( Perl_debug_log,
3361 "%*s %sTRIE matched word #%d, continuing%s\n",
3362 REPORT_CODE_OFF+depth*2, "",
3363 PL_colors[4],
3364 ST.nextword,
3365 PL_colors[5]
3366 );
3367 });
3368
3369 if (ST.accepted > 1 || has_cutgroup) {
3370 PUSH_STATE_GOTO(TRIE_next, scan);
3371 /* NOTREACHED */
166ba7cd 3372 }
2e64971a
DM
3373 /* only one choice left - just continue */
3374 DEBUG_EXECUTE_r({
3375 AV *const trie_words
3376 = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
3377 SV ** const tmp = av_fetch( trie_words,
3378 ST.nextword-1, 0 );
3379 SV *sv= tmp ? sv_newmortal() : NULL;
3380