This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Update Term-ANSIColor to CPAN version 3.00
[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;
3cf5c195
IZ
997 /* XXXX This is a remnant of the old implementation. It
998 looks wasteful, since now INTUIT can use many
6eb5f6b9 999 other heuristics. */
bbe252da 1000 prog->extflags &= ~RXf_USE_INTUIT;
cad2e5aa
JH
1001 }
1002 else
1003 s = strpos;
1004 }
1005
6eb5f6b9
JH
1006 /* Last resort... */
1007 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
1de06328
YO
1008 /* trie stclasses are too expensive to use here, we are better off to
1009 leave it to regmatch itself */
f8fc2ecf 1010 if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
6eb5f6b9
JH
1011 /* minlen == 0 is possible if regstclass is \b or \B,
1012 and the fixed substr is ''$.
1013 Since minlen is already taken into account, s+1 is before strend;
1014 accidentally, minlen >= 1 guaranties no false positives at s + 1
1015 even for \b or \B. But (minlen? 1 : 0) below assumes that
1016 regstclass does not come from lookahead... */
1017 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
1018 This leaves EXACTF only, which is dealt with in find_byclass(). */
f8fc2ecf
YO
1019 const U8* const str = (U8*)STRING(progi->regstclass);
1020 const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
1021 ? CHR_DIST(str+STR_LEN(progi->regstclass), str)
66e933ab 1022 : 1);
1de06328
YO
1023 char * endpos;
1024 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
1025 endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend);
1026 else if (prog->float_substr || prog->float_utf8)
1027 endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend);
1028 else
1029 endpos= strend;
1030
70685ca0
JH
1031 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n",
1032 (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg)));
1de06328 1033
6eb5f6b9 1034 t = s;
f8fc2ecf 1035 s = find_byclass(prog, progi->regstclass, s, endpos, NULL);
6eb5f6b9
JH
1036 if (!s) {
1037#ifdef DEBUGGING
cbbf8932 1038 const char *what = NULL;
6eb5f6b9
JH
1039#endif
1040 if (endpos == strend) {
a3621e74 1041 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6eb5f6b9
JH
1042 "Could not match STCLASS...\n") );
1043 goto fail;
1044 }
a3621e74 1045 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab 1046 "This position contradicts STCLASS...\n") );
bbe252da 1047 if ((prog->extflags & RXf_ANCH) && !ml_anch)
653099ff 1048 goto fail;
6eb5f6b9 1049 /* Contradict one of substrings */
33b8afdf 1050 if (prog->anchored_substr || prog->anchored_utf8) {
f2ed9b32 1051 if ((utf8_target ? prog->anchored_utf8 : prog->anchored_substr) == check) {
a3621e74 1052 DEBUG_EXECUTE_r( what = "anchored" );
6eb5f6b9 1053 hop_and_restart:
1aa99e6b 1054 s = HOP3c(t, 1, strend);
66e933ab
GS
1055 if (s + start_shift + end_shift > strend) {
1056 /* XXXX Should be taken into account earlier? */
a3621e74 1057 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab
GS
1058 "Could not match STCLASS...\n") );
1059 goto fail;
1060 }
5e39e1e5
HS
1061 if (!check)
1062 goto giveup;
a3621e74 1063 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 1064 "Looking for %s substr starting at offset %ld...\n",
6eb5f6b9
JH
1065 what, (long)(s + start_shift - i_strpos)) );
1066 goto restart;
1067 }
66e933ab 1068 /* Have both, check_string is floating */
6eb5f6b9
JH
1069 if (t + start_shift >= check_at) /* Contradicts floating=check */
1070 goto retry_floating_check;
1071 /* Recheck anchored substring, but not floating... */
9041c2e3 1072 s = check_at;
5e39e1e5
HS
1073 if (!check)
1074 goto giveup;
a3621e74 1075 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 1076 "Looking for anchored substr starting at offset %ld...\n",
6eb5f6b9
JH
1077 (long)(other_last - i_strpos)) );
1078 goto do_other_anchored;
1079 }
60e71179
GS
1080 /* Another way we could have checked stclass at the
1081 current position only: */
1082 if (ml_anch) {
1083 s = t = t + 1;
5e39e1e5
HS
1084 if (!check)
1085 goto giveup;
a3621e74 1086 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 1087 "Looking for /%s^%s/m starting at offset %ld...\n",
e4584336 1088 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
60e71179 1089 goto try_at_offset;
66e933ab 1090 }
f2ed9b32 1091 if (!(utf8_target ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
60e71179 1092 goto fail;
6eb5f6b9
JH
1093 /* Check is floating subtring. */
1094 retry_floating_check:
1095 t = check_at - start_shift;
a3621e74 1096 DEBUG_EXECUTE_r( what = "floating" );
6eb5f6b9
JH
1097 goto hop_and_restart;
1098 }
b7953727 1099 if (t != s) {
a3621e74 1100 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
6eb5f6b9 1101 "By STCLASS: moving %ld --> %ld\n",
b7953727
JH
1102 (long)(t - i_strpos), (long)(s - i_strpos))
1103 );
1104 }
1105 else {
a3621e74 1106 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
b7953727
JH
1107 "Does not contradict STCLASS...\n");
1108 );
1109 }
6eb5f6b9 1110 }
5e39e1e5 1111 giveup:
a3621e74 1112 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
5e39e1e5
HS
1113 PL_colors[4], (check ? "Guessed" : "Giving up"),
1114 PL_colors[5], (long)(s - i_strpos)) );
cad2e5aa 1115 return s;
2c2d71f5
JH
1116
1117 fail_finish: /* Substring not found */
33b8afdf 1118 if (prog->check_substr || prog->check_utf8) /* could be removed already */
f2ed9b32 1119 BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
cad2e5aa 1120 fail:
a3621e74 1121 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
e4584336 1122 PL_colors[4], PL_colors[5]));
bd61b366 1123 return NULL;
cad2e5aa 1124}
9661b544 1125
a0a388a1
YO
1126#define DECL_TRIE_TYPE(scan) \
1127 const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold } \
1128 trie_type = (scan->flags != EXACT) \
f2ed9b32
KW
1129 ? (utf8_target ? trie_utf8_fold : (UTF_PATTERN ? trie_latin_utf8_fold : trie_plain)) \
1130 : (utf8_target ? trie_utf8 : trie_plain)
3b0527fe 1131
55eed653
NC
1132#define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \
1133uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \
4cadc6a9
YO
1134 switch (trie_type) { \
1135 case trie_utf8_fold: \
1136 if ( foldlen>0 ) { \
0abd0d78 1137 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
4cadc6a9
YO
1138 foldlen -= len; \
1139 uscan += len; \
1140 len=0; \
1141 } else { \
0abd0d78 1142 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
4cadc6a9
YO
1143 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
1144 foldlen -= UNISKIP( uvc ); \
1145 uscan = foldbuf + UNISKIP( uvc ); \
1146 } \
1147 break; \
a0a388a1
YO
1148 case trie_latin_utf8_fold: \
1149 if ( foldlen>0 ) { \
1150 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
1151 foldlen -= len; \
1152 uscan += len; \
1153 len=0; \
1154 } else { \
1155 len = 1; \
1156 uvc = to_uni_fold( *(U8*)uc, foldbuf, &foldlen ); \
1157 foldlen -= UNISKIP( uvc ); \
1158 uscan = foldbuf + UNISKIP( uvc ); \
1159 } \
1160 break; \
4cadc6a9
YO
1161 case trie_utf8: \
1162 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
1163 break; \
1164 case trie_plain: \
1165 uvc = (UV)*uc; \
1166 len = 1; \
1167 } \
4cadc6a9
YO
1168 if (uvc < 256) { \
1169 charid = trie->charmap[ uvc ]; \
1170 } \
1171 else { \
1172 charid = 0; \
55eed653
NC
1173 if (widecharmap) { \
1174 SV** const svpp = hv_fetch(widecharmap, \
4cadc6a9
YO
1175 (char*)&uvc, sizeof(UV), 0); \
1176 if (svpp) \
1177 charid = (U16)SvIV(*svpp); \
1178 } \
1179 } \
1180} STMT_END
1181
a0a388a1
YO
1182#define REXEC_FBC_EXACTISH_CHECK(CoNd) \
1183{ \
1184 char *my_strend= (char *)strend; \
4cadc6a9
YO
1185 if ( (CoNd) \
1186 && (ln == len || \
f2ed9b32
KW
1187 foldEQ_utf8(s, &my_strend, 0, utf8_target, \
1188 m, NULL, ln, cBOOL(UTF_PATTERN))) \
a0a388a1 1189 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1190 goto got_it; \
1191 else { \
1192 U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \
1193 uvchr_to_utf8(tmpbuf, c); \
1194 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \
1195 if ( f != c \
1196 && (f == c1 || f == c2) \
a0a388a1 1197 && (ln == len || \
f2ed9b32
KW
1198 foldEQ_utf8(s, &my_strend, 0, utf8_target,\
1199 m, NULL, ln, cBOOL(UTF_PATTERN)))\
a0a388a1 1200 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1201 goto got_it; \
1202 } \
a0a388a1
YO
1203} \
1204s += len
4cadc6a9
YO
1205
1206#define REXEC_FBC_EXACTISH_SCAN(CoNd) \
1207STMT_START { \
1208 while (s <= e) { \
1209 if ( (CoNd) \
4c1b470c
KW
1210 && (ln == 1 || (OP(c) == EXACTF \
1211 ? foldEQ(s, m, ln) \
1212 : foldEQ_locale(s, m, ln))) \
24b23f37 1213 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1214 goto got_it; \
1215 s++; \
1216 } \
1217} STMT_END
1218
1219#define REXEC_FBC_UTF8_SCAN(CoDe) \
1220STMT_START { \
1221 while (s + (uskip = UTF8SKIP(s)) <= strend) { \
1222 CoDe \
1223 s += uskip; \
1224 } \
1225} STMT_END
1226
1227#define REXEC_FBC_SCAN(CoDe) \
1228STMT_START { \
1229 while (s < strend) { \
1230 CoDe \
1231 s++; \
1232 } \
1233} STMT_END
1234
1235#define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \
1236REXEC_FBC_UTF8_SCAN( \
1237 if (CoNd) { \
24b23f37 1238 if (tmp && (!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1239 goto got_it; \
1240 else \
1241 tmp = doevery; \
1242 } \
1243 else \
1244 tmp = 1; \
1245)
1246
1247#define REXEC_FBC_CLASS_SCAN(CoNd) \
1248REXEC_FBC_SCAN( \
1249 if (CoNd) { \
24b23f37 1250 if (tmp && (!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1251 goto got_it; \
1252 else \
1253 tmp = doevery; \
1254 } \
1255 else \
1256 tmp = 1; \
1257)
1258
1259#define REXEC_FBC_TRYIT \
24b23f37 1260if ((!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1261 goto got_it
1262
e1d1eefb 1263#define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \
f2ed9b32 1264 if (utf8_target) { \
e1d1eefb
YO
1265 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1266 } \
1267 else { \
1268 REXEC_FBC_CLASS_SCAN(CoNd); \
1269 } \
1270 break
1271
4cadc6a9 1272#define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
f2ed9b32 1273 if (utf8_target) { \
4cadc6a9
YO
1274 UtFpReLoAd; \
1275 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1276 } \
1277 else { \
1278 REXEC_FBC_CLASS_SCAN(CoNd); \
1279 } \
1280 break
1281
1282#define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \
1283 PL_reg_flags |= RF_tainted; \
f2ed9b32 1284 if (utf8_target) { \
4cadc6a9
YO
1285 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1286 } \
1287 else { \
1288 REXEC_FBC_CLASS_SCAN(CoNd); \
1289 } \
1290 break
1291
786e8c11
YO
1292#define DUMP_EXEC_POS(li,s,doutf8) \
1293 dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
1294
1295/* We know what class REx starts with. Try to find this position... */
1296/* if reginfo is NULL, its a dryrun */
1297/* annoyingly all the vars in this routine have different names from their counterparts
1298 in regmatch. /grrr */
1299
3c3eec57 1300STATIC char *
07be1b83 1301S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
24b23f37 1302 const char *strend, regmatch_info *reginfo)
a687059c 1303{
27da23d5 1304 dVAR;
bbe252da 1305 const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
6eb5f6b9 1306 char *m;
d8093b23 1307 STRLEN ln;
5dab1207 1308 STRLEN lnc;
078c425b 1309 register STRLEN uskip;
d8093b23
G
1310 unsigned int c1;
1311 unsigned int c2;
6eb5f6b9
JH
1312 char *e;
1313 register I32 tmp = 1; /* Scratch variable? */
f2ed9b32 1314 register const bool utf8_target = PL_reg_match_utf8;
f8fc2ecf 1315 RXi_GET_DECL(prog,progi);
7918f24d
NC
1316
1317 PERL_ARGS_ASSERT_FIND_BYCLASS;
f8fc2ecf 1318
6eb5f6b9
JH
1319 /* We know what class it must start with. */
1320 switch (OP(c)) {
6eb5f6b9 1321 case ANYOF:
f2ed9b32 1322 if (utf8_target) {
4cadc6a9 1323 REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
388cc4de 1324 !UTF8_IS_INVARIANT((U8)s[0]) ?
f2ed9b32 1325 reginclass(prog, c, (U8*)s, 0, utf8_target) :
4cadc6a9 1326 REGINCLASS(prog, c, (U8*)s));
388cc4de
HS
1327 }
1328 else {
1329 while (s < strend) {
1330 STRLEN skip = 1;
1331
32fc9b6a 1332 if (REGINCLASS(prog, c, (U8*)s) ||
388cc4de
HS
1333 (ANYOF_FOLD_SHARP_S(c, s, strend) &&
1334 /* The assignment of 2 is intentional:
1335 * for the folded sharp s, the skip is 2. */
1336 (skip = SHARP_S_SKIP))) {
24b23f37 1337 if (tmp && (!reginfo || regtry(reginfo, &s)))
388cc4de
HS
1338 goto got_it;
1339 else
1340 tmp = doevery;
1341 }
1342 else
1343 tmp = 1;
1344 s += skip;
1345 }
a0d0e21e 1346 }
6eb5f6b9 1347 break;
f33976b4 1348 case CANY:
4cadc6a9 1349 REXEC_FBC_SCAN(
24b23f37 1350 if (tmp && (!reginfo || regtry(reginfo, &s)))
f33976b4
DB
1351 goto got_it;
1352 else
1353 tmp = doevery;
4cadc6a9 1354 );
f33976b4 1355 break;
6eb5f6b9 1356 case EXACTF:
5dab1207
NIS
1357 m = STRING(c);
1358 ln = STR_LEN(c); /* length to match in octets/bytes */
1359 lnc = (I32) ln; /* length to match in characters */
f2ed9b32 1360 if (UTF_PATTERN) {
a2a2844f 1361 STRLEN ulen1, ulen2;
5dab1207 1362 U8 *sm = (U8 *) m;
89ebb4a3
JH
1363 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
1364 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
97dc7d3e
RGS
1365 /* used by commented-out code below */
1366 /*const U32 uniflags = UTF8_ALLOW_DEFAULT;*/
a0a388a1
YO
1367
1368 /* XXX: Since the node will be case folded at compile
1369 time this logic is a little odd, although im not
1370 sure that its actually wrong. --dmq */
1371
1372 c1 = to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1373 c2 = to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1374
1375 /* XXX: This is kinda strange. to_utf8_XYZ returns the
1376 codepoint of the first character in the converted
1377 form, yet originally we did the extra step.
1378 No tests fail by commenting this code out however
1379 so Ive left it out. -- dmq.
1380
89ebb4a3 1381 c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
041457d9 1382 0, uniflags);
89ebb4a3 1383 c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
041457d9 1384 0, uniflags);
a0a388a1
YO
1385 */
1386
5dab1207
NIS
1387 lnc = 0;
1388 while (sm < ((U8 *) m + ln)) {
1389 lnc++;
1390 sm += UTF8SKIP(sm);
1391 }
1aa99e6b
IH
1392 }
1393 else {
1394 c1 = *(U8*)m;
1395 c2 = PL_fold[c1];
1396 }
6eb5f6b9
JH
1397 goto do_exactf;
1398 case EXACTFL:
5dab1207
NIS
1399 m = STRING(c);
1400 ln = STR_LEN(c);
1401 lnc = (I32) ln;
d8093b23 1402 c1 = *(U8*)m;
6eb5f6b9
JH
1403 c2 = PL_fold_locale[c1];
1404 do_exactf:
db12adc6 1405 e = HOP3c(strend, -((I32)lnc), s);
b3c9acc1 1406
3b0527fe 1407 if (!reginfo && e < s)
6eb5f6b9 1408 e = s; /* Due to minlen logic of intuit() */
1aa99e6b 1409
60a8b682
JH
1410 /* The idea in the EXACTF* cases is to first find the
1411 * first character of the EXACTF* node and then, if
1412 * necessary, case-insensitively compare the full
1413 * text of the node. The c1 and c2 are the first
1414 * characters (though in Unicode it gets a bit
1415 * more complicated because there are more cases
7f16dd3d
JH
1416 * than just upper and lower: one needs to use
1417 * the so-called folding case for case-insensitive
1418 * matching (called "loose matching" in Unicode).
4c1b470c 1419 * foldEQ_utf8() will do just that. */
60a8b682 1420
f2ed9b32 1421 if (utf8_target || UTF_PATTERN) {
575cac57 1422 UV c, f;
89ebb4a3 1423 U8 tmpbuf [UTF8_MAXBYTES+1];
a0a388a1
YO
1424 STRLEN len = 1;
1425 STRLEN foldlen;
4ad0818d 1426 const U32 uniflags = UTF8_ALLOW_DEFAULT;
09091399 1427 if (c1 == c2) {
5dab1207
NIS
1428 /* Upper and lower of 1st char are equal -
1429 * probably not a "letter". */
1aa99e6b 1430 while (s <= e) {
f2ed9b32 1431 if (utf8_target) {
a0a388a1 1432 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
041457d9 1433 uniflags);
a0a388a1
YO
1434 } else {
1435 c = *((U8*)s);
1436 }
4cadc6a9 1437 REXEC_FBC_EXACTISH_CHECK(c == c1);
1aa99e6b 1438 }
09091399
JH
1439 }
1440 else {
1aa99e6b 1441 while (s <= e) {
f2ed9b32 1442 if (utf8_target) {
a0a388a1 1443 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
041457d9 1444 uniflags);
a0a388a1
YO
1445 } else {
1446 c = *((U8*)s);
1447 }
80aecb99 1448
60a8b682 1449 /* Handle some of the three Greek sigmas cases.
8c01da3c
JH
1450 * Note that not all the possible combinations
1451 * are handled here: some of them are handled
1452 * by the standard folding rules, and some of
1453 * them (the character class or ANYOF cases)
1454 * are handled during compiletime in
1455 * regexec.c:S_regclass(). */
880bd946
JH
1456 if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
1457 c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
1458 c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
80aecb99 1459
4cadc6a9 1460 REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2);
1aa99e6b 1461 }
09091399 1462 }
1aa99e6b
IH
1463 }
1464 else {
a0a388a1 1465 /* Neither pattern nor string are UTF8 */
1aa99e6b 1466 if (c1 == c2)
4cadc6a9 1467 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
1aa99e6b 1468 else
4cadc6a9 1469 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
b3c9acc1
IZ
1470 }
1471 break;
bbce6d69 1472 case BOUNDL:
3280af22 1473 PL_reg_flags |= RF_tainted;
bbce6d69 1474 /* FALL THROUGH */
a0d0e21e 1475 case BOUND:
f2ed9b32 1476 if (utf8_target) {
12d33761 1477 if (s == PL_bostr)
ffc61ed2
JH
1478 tmp = '\n';
1479 else {
6136c704 1480 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
4ad0818d 1481 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
ffc61ed2
JH
1482 }
1483 tmp = ((OP(c) == BOUND ?
9041c2e3 1484 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1a4fad37 1485 LOAD_UTF8_CHARCLASS_ALNUM();
4cadc6a9 1486 REXEC_FBC_UTF8_SCAN(
ffc61ed2 1487 if (tmp == !(OP(c) == BOUND ?
f2ed9b32 1488 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)) :
ffc61ed2
JH
1489 isALNUM_LC_utf8((U8*)s)))
1490 {
1491 tmp = !tmp;
4cadc6a9 1492 REXEC_FBC_TRYIT;
a687059c 1493 }
4cadc6a9 1494 );
a0d0e21e 1495 }
667bb95a 1496 else {
12d33761 1497 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2 1498 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
4cadc6a9 1499 REXEC_FBC_SCAN(
ffc61ed2
JH
1500 if (tmp ==
1501 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
1502 tmp = !tmp;
4cadc6a9 1503 REXEC_FBC_TRYIT;
a0ed51b3 1504 }
4cadc6a9 1505 );
a0ed51b3 1506 }
24b23f37 1507 if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))
a0ed51b3
LW
1508 goto got_it;
1509 break;
bbce6d69 1510 case NBOUNDL:
3280af22 1511 PL_reg_flags |= RF_tainted;
bbce6d69 1512 /* FALL THROUGH */
a0d0e21e 1513 case NBOUND:
f2ed9b32 1514 if (utf8_target) {
12d33761 1515 if (s == PL_bostr)
ffc61ed2
JH
1516 tmp = '\n';
1517 else {
6136c704 1518 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
4ad0818d 1519 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
ffc61ed2
JH
1520 }
1521 tmp = ((OP(c) == NBOUND ?
9041c2e3 1522 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1a4fad37 1523 LOAD_UTF8_CHARCLASS_ALNUM();
4cadc6a9 1524 REXEC_FBC_UTF8_SCAN(
ffc61ed2 1525 if (tmp == !(OP(c) == NBOUND ?
f2ed9b32 1526 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)) :
ffc61ed2
JH
1527 isALNUM_LC_utf8((U8*)s)))
1528 tmp = !tmp;
4cadc6a9
YO
1529 else REXEC_FBC_TRYIT;
1530 );
a0d0e21e 1531 }
667bb95a 1532 else {
12d33761 1533 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2
JH
1534 tmp = ((OP(c) == NBOUND ?
1535 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
4cadc6a9 1536 REXEC_FBC_SCAN(
ffc61ed2
JH
1537 if (tmp ==
1538 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1539 tmp = !tmp;
4cadc6a9
YO
1540 else REXEC_FBC_TRYIT;
1541 );
a0ed51b3 1542 }
24b23f37 1543 if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s)))
a0ed51b3
LW
1544 goto got_it;
1545 break;
a0d0e21e 1546 case ALNUM:
4cadc6a9 1547 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1548 LOAD_UTF8_CHARCLASS_PERL_WORD(),
f2ed9b32 1549 swash_fetch(RE_utf8_perl_word, (U8*)s, utf8_target),
4cadc6a9
YO
1550 isALNUM(*s)
1551 );
bbce6d69 1552 case ALNUML:
4cadc6a9
YO
1553 REXEC_FBC_CSCAN_TAINT(
1554 isALNUM_LC_utf8((U8*)s),
1555 isALNUM_LC(*s)
1556 );
a0d0e21e 1557 case NALNUM:
4cadc6a9 1558 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1559 LOAD_UTF8_CHARCLASS_PERL_WORD(),
f2ed9b32 1560 !swash_fetch(RE_utf8_perl_word, (U8*)s, utf8_target),
4cadc6a9
YO
1561 !isALNUM(*s)
1562 );
bbce6d69 1563 case NALNUML:
4cadc6a9
YO
1564 REXEC_FBC_CSCAN_TAINT(
1565 !isALNUM_LC_utf8((U8*)s),
1566 !isALNUM_LC(*s)
1567 );
a0d0e21e 1568 case SPACE:
4cadc6a9 1569 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1570 LOAD_UTF8_CHARCLASS_PERL_SPACE(),
f2ed9b32 1571 *s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, utf8_target),
4cadc6a9
YO
1572 isSPACE(*s)
1573 );
bbce6d69 1574 case SPACEL:
4cadc6a9
YO
1575 REXEC_FBC_CSCAN_TAINT(
1576 *s == ' ' || isSPACE_LC_utf8((U8*)s),
1577 isSPACE_LC(*s)
1578 );
a0d0e21e 1579 case NSPACE:
4cadc6a9 1580 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1581 LOAD_UTF8_CHARCLASS_PERL_SPACE(),
f2ed9b32 1582 !(*s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, utf8_target)),
4cadc6a9
YO
1583 !isSPACE(*s)
1584 );
bbce6d69 1585 case NSPACEL:
4cadc6a9
YO
1586 REXEC_FBC_CSCAN_TAINT(
1587 !(*s == ' ' || isSPACE_LC_utf8((U8*)s)),
1588 !isSPACE_LC(*s)
1589 );
a0d0e21e 1590 case DIGIT:
4cadc6a9 1591 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1592 LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
f2ed9b32 1593 swash_fetch(RE_utf8_posix_digit,(U8*)s, utf8_target),
4cadc6a9
YO
1594 isDIGIT(*s)
1595 );
b8c5462f 1596 case DIGITL:
4cadc6a9
YO
1597 REXEC_FBC_CSCAN_TAINT(
1598 isDIGIT_LC_utf8((U8*)s),
1599 isDIGIT_LC(*s)
1600 );
a0d0e21e 1601 case NDIGIT:
4cadc6a9 1602 REXEC_FBC_CSCAN_PRELOAD(
d1eb3177 1603 LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
f2ed9b32 1604 !swash_fetch(RE_utf8_posix_digit,(U8*)s, utf8_target),
4cadc6a9
YO
1605 !isDIGIT(*s)
1606 );
b8c5462f 1607 case NDIGITL:
4cadc6a9
YO
1608 REXEC_FBC_CSCAN_TAINT(
1609 !isDIGIT_LC_utf8((U8*)s),
1610 !isDIGIT_LC(*s)
1611 );
e1d1eefb
YO
1612 case LNBREAK:
1613 REXEC_FBC_CSCAN(
1614 is_LNBREAK_utf8(s),
1615 is_LNBREAK_latin1(s)
1616 );
1617 case VERTWS:
1618 REXEC_FBC_CSCAN(
1619 is_VERTWS_utf8(s),
1620 is_VERTWS_latin1(s)
1621 );
1622 case NVERTWS:
1623 REXEC_FBC_CSCAN(
1624 !is_VERTWS_utf8(s),
1625 !is_VERTWS_latin1(s)
1626 );
1627 case HORIZWS:
1628 REXEC_FBC_CSCAN(
1629 is_HORIZWS_utf8(s),
1630 is_HORIZWS_latin1(s)
1631 );
1632 case NHORIZWS:
1633 REXEC_FBC_CSCAN(
1634 !is_HORIZWS_utf8(s),
1635 !is_HORIZWS_latin1(s)
1636 );
1de06328
YO
1637 case AHOCORASICKC:
1638 case AHOCORASICK:
07be1b83 1639 {
a0a388a1 1640 DECL_TRIE_TYPE(c);
07be1b83
YO
1641 /* what trie are we using right now */
1642 reg_ac_data *aho
f8fc2ecf 1643 = (reg_ac_data*)progi->data->data[ ARG( c ) ];
3251b653
NC
1644 reg_trie_data *trie
1645 = (reg_trie_data*)progi->data->data[ aho->trie ];
85fbaab2 1646 HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]);
07be1b83
YO
1647
1648 const char *last_start = strend - trie->minlen;
6148ee25 1649#ifdef DEBUGGING
07be1b83 1650 const char *real_start = s;
6148ee25 1651#endif
07be1b83 1652 STRLEN maxlen = trie->maxlen;
be8e71aa
YO
1653 SV *sv_points;
1654 U8 **points; /* map of where we were in the input string
786e8c11 1655 when reading a given char. For ASCII this
be8e71aa 1656 is unnecessary overhead as the relationship
38a44b82
NC
1657 is always 1:1, but for Unicode, especially
1658 case folded Unicode this is not true. */
f9e705e8 1659 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
786e8c11
YO
1660 U8 *bitmap=NULL;
1661
07be1b83
YO
1662
1663 GET_RE_DEBUG_FLAGS_DECL;
1664
be8e71aa
YO
1665 /* We can't just allocate points here. We need to wrap it in
1666 * an SV so it gets freed properly if there is a croak while
1667 * running the match */
1668 ENTER;
1669 SAVETMPS;
1670 sv_points=newSV(maxlen * sizeof(U8 *));
1671 SvCUR_set(sv_points,
1672 maxlen * sizeof(U8 *));
1673 SvPOK_on(sv_points);
1674 sv_2mortal(sv_points);
1675 points=(U8**)SvPV_nolen(sv_points );
1de06328
YO
1676 if ( trie_type != trie_utf8_fold
1677 && (trie->bitmap || OP(c)==AHOCORASICKC) )
1678 {
786e8c11
YO
1679 if (trie->bitmap)
1680 bitmap=(U8*)trie->bitmap;
1681 else
1682 bitmap=(U8*)ANYOF_BITMAP(c);
07be1b83 1683 }
786e8c11
YO
1684 /* this is the Aho-Corasick algorithm modified a touch
1685 to include special handling for long "unknown char"
1686 sequences. The basic idea being that we use AC as long
1687 as we are dealing with a possible matching char, when
1688 we encounter an unknown char (and we have not encountered
1689 an accepting state) we scan forward until we find a legal
1690 starting char.
1691 AC matching is basically that of trie matching, except
1692 that when we encounter a failing transition, we fall back
1693 to the current states "fail state", and try the current char
1694 again, a process we repeat until we reach the root state,
1695 state 1, or a legal transition. If we fail on the root state
1696 then we can either terminate if we have reached an accepting
1697 state previously, or restart the entire process from the beginning
1698 if we have not.
1699
1700 */
07be1b83
YO
1701 while (s <= last_start) {
1702 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1703 U8 *uc = (U8*)s;
1704 U16 charid = 0;
1705 U32 base = 1;
1706 U32 state = 1;
1707 UV uvc = 0;
1708 STRLEN len = 0;
1709 STRLEN foldlen = 0;
1710 U8 *uscan = (U8*)NULL;
1711 U8 *leftmost = NULL;
786e8c11
YO
1712#ifdef DEBUGGING
1713 U32 accepted_word= 0;
1714#endif
07be1b83
YO
1715 U32 pointpos = 0;
1716
1717 while ( state && uc <= (U8*)strend ) {
1718 int failed=0;
786e8c11
YO
1719 U32 word = aho->states[ state ].wordnum;
1720
1de06328
YO
1721 if( state==1 ) {
1722 if ( bitmap ) {
1723 DEBUG_TRIE_EXECUTE_r(
1724 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1725 dump_exec_pos( (char *)uc, c, strend, real_start,
f2ed9b32 1726 (char *)uc, utf8_target );
1de06328
YO
1727 PerlIO_printf( Perl_debug_log,
1728 " Scanning for legal start char...\n");
1729 }
1730 );
1731 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1732 uc++;
786e8c11 1733 }
1de06328 1734 s= (char *)uc;
786e8c11 1735 }
786e8c11
YO
1736 if (uc >(U8*)last_start) break;
1737 }
1738
1739 if ( word ) {
2e64971a 1740 U8 *lpos= points[ (pointpos - trie->wordinfo[word].len) % maxlen ];
786e8c11
YO
1741 if (!leftmost || lpos < leftmost) {
1742 DEBUG_r(accepted_word=word);
07be1b83 1743 leftmost= lpos;
786e8c11 1744 }
07be1b83 1745 if (base==0) break;
786e8c11 1746
07be1b83
YO
1747 }
1748 points[pointpos++ % maxlen]= uc;
55eed653
NC
1749 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
1750 uscan, len, uvc, charid, foldlen,
1751 foldbuf, uniflags);
786e8c11
YO
1752 DEBUG_TRIE_EXECUTE_r({
1753 dump_exec_pos( (char *)uc, c, strend, real_start,
f2ed9b32 1754 s, utf8_target );
07be1b83 1755 PerlIO_printf(Perl_debug_log,
786e8c11
YO
1756 " Charid:%3u CP:%4"UVxf" ",
1757 charid, uvc);
1758 });
07be1b83
YO
1759
1760 do {
6148ee25 1761#ifdef DEBUGGING
786e8c11 1762 word = aho->states[ state ].wordnum;
6148ee25 1763#endif
07be1b83
YO
1764 base = aho->states[ state ].trans.base;
1765
786e8c11
YO
1766 DEBUG_TRIE_EXECUTE_r({
1767 if (failed)
1768 dump_exec_pos( (char *)uc, c, strend, real_start,
f2ed9b32 1769 s, utf8_target );
07be1b83 1770 PerlIO_printf( Perl_debug_log,
786e8c11
YO
1771 "%sState: %4"UVxf", word=%"UVxf,
1772 failed ? " Fail transition to " : "",
1773 (UV)state, (UV)word);
1774 });
07be1b83
YO
1775 if ( base ) {
1776 U32 tmp;
6dd2be57 1777 I32 offset;
07be1b83 1778 if (charid &&
6dd2be57
DM
1779 ( ((offset = base + charid
1780 - 1 - trie->uniquecharcount)) >= 0)
1781 && ((U32)offset < trie->lasttrans)
1782 && trie->trans[offset].check == state
1783 && (tmp=trie->trans[offset].next))
07be1b83 1784 {
786e8c11
YO
1785 DEBUG_TRIE_EXECUTE_r(
1786 PerlIO_printf( Perl_debug_log," - legal\n"));
07be1b83
YO
1787 state = tmp;
1788 break;
1789 }
1790 else {
786e8c11
YO
1791 DEBUG_TRIE_EXECUTE_r(
1792 PerlIO_printf( Perl_debug_log," - fail\n"));
1793 failed = 1;
1794 state = aho->fail[state];
07be1b83
YO
1795 }
1796 }
1797 else {
1798 /* we must be accepting here */
786e8c11
YO
1799 DEBUG_TRIE_EXECUTE_r(
1800 PerlIO_printf( Perl_debug_log," - accepting\n"));
1801 failed = 1;
07be1b83
YO
1802 break;
1803 }
1804 } while(state);
786e8c11 1805 uc += len;
07be1b83
YO
1806 if (failed) {
1807 if (leftmost)
1808 break;
786e8c11 1809 if (!state) state = 1;
07be1b83
YO
1810 }
1811 }
1812 if ( aho->states[ state ].wordnum ) {
2e64971a 1813 U8 *lpos = points[ (pointpos - trie->wordinfo[aho->states[ state ].wordnum].len) % maxlen ];
786e8c11
YO
1814 if (!leftmost || lpos < leftmost) {
1815 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
07be1b83 1816 leftmost = lpos;
786e8c11 1817 }
07be1b83 1818 }
07be1b83
YO
1819 if (leftmost) {
1820 s = (char*)leftmost;
786e8c11
YO
1821 DEBUG_TRIE_EXECUTE_r({
1822 PerlIO_printf(
70685ca0
JH
1823 Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
1824 (UV)accepted_word, (IV)(s - real_start)
786e8c11
YO
1825 );
1826 });
24b23f37 1827 if (!reginfo || regtry(reginfo, &s)) {
be8e71aa
YO
1828 FREETMPS;
1829 LEAVE;
07be1b83 1830 goto got_it;
be8e71aa 1831 }
07be1b83 1832 s = HOPc(s,1);
786e8c11
YO
1833 DEBUG_TRIE_EXECUTE_r({
1834 PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
1835 });
07be1b83 1836 } else {
786e8c11
YO
1837 DEBUG_TRIE_EXECUTE_r(
1838 PerlIO_printf( Perl_debug_log,"No match.\n"));
07be1b83
YO
1839 break;
1840 }
1841 }
be8e71aa
YO
1842 FREETMPS;
1843 LEAVE;
07be1b83
YO
1844 }
1845 break;
b3c9acc1 1846 default:
3c3eec57
GS
1847 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
1848 break;
d6a28714 1849 }
6eb5f6b9
JH
1850 return 0;
1851 got_it:
1852 return s;
1853}
1854
fae667d5 1855
6eb5f6b9
JH
1856/*
1857 - regexec_flags - match a regexp against a string
1858 */
1859I32
288b8c02 1860Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, register char *strend,
6eb5f6b9
JH
1861 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
1862/* strend: pointer to null at end of string */
1863/* strbeg: real beginning of string */
1864/* minend: end of match must be >=minend after stringarg. */
58e23c8d
YO
1865/* data: May be used for some additional optimizations.
1866 Currently its only used, with a U32 cast, for transmitting
1867 the ganch offset when doing a /g match. This will change */
6eb5f6b9
JH
1868/* nosave: For optimizations. */
1869{
97aff369 1870 dVAR;
288b8c02 1871 struct regexp *const prog = (struct regexp *)SvANY(rx);
24b23f37 1872 /*register*/ char *s;
6eb5f6b9 1873 register regnode *c;
24b23f37 1874 /*register*/ char *startpos = stringarg;
6eb5f6b9
JH
1875 I32 minlen; /* must match at least this many chars */
1876 I32 dontbother = 0; /* how many characters not to try at end */
6eb5f6b9
JH
1877 I32 end_shift = 0; /* Same for the end. */ /* CC */
1878 I32 scream_pos = -1; /* Internal iterator of scream. */
ccac19ea 1879 char *scream_olds = NULL;
f2ed9b32 1880 const bool utf8_target = cBOOL(DO_UTF8(sv));
2757e526 1881 I32 multiline;
f8fc2ecf 1882 RXi_GET_DECL(prog,progi);
3b0527fe 1883 regmatch_info reginfo; /* create some info to pass to regtry etc */
e9105d30 1884 regexp_paren_pair *swap = NULL;
a3621e74
YO
1885 GET_RE_DEBUG_FLAGS_DECL;
1886
7918f24d 1887 PERL_ARGS_ASSERT_REGEXEC_FLAGS;
9d4ba2ae 1888 PERL_UNUSED_ARG(data);
6eb5f6b9
JH
1889
1890 /* Be paranoid... */
1891 if (prog == NULL || startpos == NULL) {
1892 Perl_croak(aTHX_ "NULL regexp parameter");
1893 return 0;
1894 }
1895
bbe252da 1896 multiline = prog->extflags & RXf_PMf_MULTILINE;
288b8c02 1897 reginfo.prog = rx; /* Yes, sorry that this is confusing. */
2757e526 1898
f2ed9b32 1899 RX_MATCH_UTF8_set(rx, utf8_target);
1de06328 1900 DEBUG_EXECUTE_r(
f2ed9b32 1901 debug_start_match(rx, utf8_target, startpos, strend,
1de06328
YO
1902 "Matching");
1903 );
bac06658 1904
6eb5f6b9 1905 minlen = prog->minlen;
1de06328
YO
1906
1907 if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
a3621e74 1908 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
a72c7584
JH
1909 "String too short [regexec_flags]...\n"));
1910 goto phooey;
1aa99e6b 1911 }
6eb5f6b9 1912
1de06328 1913
6eb5f6b9 1914 /* Check validity of program. */
f8fc2ecf 1915 if (UCHARAT(progi->program) != REG_MAGIC) {
6eb5f6b9
JH
1916 Perl_croak(aTHX_ "corrupted regexp program");
1917 }
1918
1919 PL_reg_flags = 0;
1920 PL_reg_eval_set = 0;
1921 PL_reg_maxiter = 0;
1922
3c8556c3 1923 if (RX_UTF8(rx))
6eb5f6b9
JH
1924 PL_reg_flags |= RF_utf8;
1925
1926 /* Mark beginning of line for ^ and lookbehind. */
3b0527fe 1927 reginfo.bol = startpos; /* XXX not used ??? */
6eb5f6b9 1928 PL_bostr = strbeg;
3b0527fe 1929 reginfo.sv = sv;
6eb5f6b9
JH
1930
1931 /* Mark end of line for $ (and such) */
1932 PL_regeol = strend;
1933
1934 /* see how far we have to get to not match where we matched before */
3b0527fe 1935 reginfo.till = startpos+minend;
6eb5f6b9 1936
6eb5f6b9
JH
1937 /* If there is a "must appear" string, look for it. */
1938 s = startpos;
1939
bbe252da 1940 if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */
6eb5f6b9 1941 MAGIC *mg;
2c296965 1942 if (flags & REXEC_IGNOREPOS){ /* Means: check only at start */
58e23c8d 1943 reginfo.ganch = startpos + prog->gofs;
2c296965 1944 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
ed549f2e 1945 "GPOS IGNOREPOS: reginfo.ganch = startpos + %"UVxf"\n",(UV)prog->gofs));
2c296965 1946 } else if (sv && SvTYPE(sv) >= SVt_PVMG
6eb5f6b9 1947 && SvMAGIC(sv)
14befaf4
DM
1948 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
1949 && mg->mg_len >= 0) {
3b0527fe 1950 reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */
2c296965 1951 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
ed549f2e 1952 "GPOS MAGIC: reginfo.ganch = strbeg + %"IVdf"\n",(IV)mg->mg_len));
2c296965 1953
bbe252da 1954 if (prog->extflags & RXf_ANCH_GPOS) {
3b0527fe 1955 if (s > reginfo.ganch)
6eb5f6b9 1956 goto phooey;
58e23c8d 1957 s = reginfo.ganch - prog->gofs;
2c296965 1958 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
ed549f2e 1959 "GPOS ANCH_GPOS: s = ganch - %"UVxf"\n",(UV)prog->gofs));
c584a96e
YO
1960 if (s < strbeg)
1961 goto phooey;
6eb5f6b9
JH
1962 }
1963 }
58e23c8d 1964 else if (data) {
70685ca0 1965 reginfo.ganch = strbeg + PTR2UV(data);
2c296965
YO
1966 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1967 "GPOS DATA: reginfo.ganch= strbeg + %"UVxf"\n",PTR2UV(data)));
1968
1969 } else { /* pos() not defined */
3b0527fe 1970 reginfo.ganch = strbeg;
2c296965
YO
1971 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1972 "GPOS: reginfo.ganch = strbeg\n"));
1973 }
6eb5f6b9 1974 }
288b8c02 1975 if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
e9105d30
GG
1976 /* We have to be careful. If the previous successful match
1977 was from this regex we don't want a subsequent partially
1978 successful match to clobber the old results.
1979 So when we detect this possibility we add a swap buffer
1980 to the re, and switch the buffer each match. If we fail
1981 we switch it back, otherwise we leave it swapped.
1982 */
1983 swap = prog->offs;
1984 /* do we need a save destructor here for eval dies? */
1985 Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair);
c74340f9 1986 }
a0714e2c 1987 if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
6eb5f6b9
JH
1988 re_scream_pos_data d;
1989
1990 d.scream_olds = &scream_olds;
1991 d.scream_pos = &scream_pos;
288b8c02 1992 s = re_intuit_start(rx, sv, s, strend, flags, &d);
3fa9c3d7 1993 if (!s) {
a3621e74 1994 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
6eb5f6b9 1995 goto phooey; /* not present */
3fa9c3d7 1996 }
6eb5f6b9
JH
1997 }
1998
1de06328 1999
6eb5f6b9
JH
2000
2001 /* Simplest case: anchored match need be tried only once. */
2002 /* [unless only anchor is BOL and multiline is set] */
bbe252da 2003 if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) {
24b23f37 2004 if (s == startpos && regtry(&reginfo, &startpos))
6eb5f6b9 2005 goto got_it;
bbe252da
YO
2006 else if (multiline || (prog->intflags & PREGf_IMPLICIT)
2007 || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */
6eb5f6b9
JH
2008 {
2009 char *end;
2010
2011 if (minlen)
2012 dontbother = minlen - 1;
1aa99e6b 2013 end = HOP3c(strend, -dontbother, strbeg) - 1;
6eb5f6b9 2014 /* for multiline we only have to try after newlines */
33b8afdf 2015 if (prog->check_substr || prog->check_utf8) {
6eb5f6b9
JH
2016 if (s == startpos)
2017 goto after_try;
2018 while (1) {
24b23f37 2019 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2020 goto got_it;
2021 after_try:
5339e136 2022 if (s > end)
6eb5f6b9 2023 goto phooey;
bbe252da 2024 if (prog->extflags & RXf_USE_INTUIT) {
288b8c02 2025 s = re_intuit_start(rx, sv, s + 1, strend, flags, NULL);
6eb5f6b9
JH
2026 if (!s)
2027 goto phooey;
2028 }
2029 else
2030 s++;
2031 }
2032 } else {
2033 if (s > startpos)
2034 s--;
2035 while (s < end) {
2036 if (*s++ == '\n') { /* don't need PL_utf8skip here */
24b23f37 2037 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2038 goto got_it;
2039 }
2040 }
2041 }
2042 }
2043 goto phooey;
bbe252da 2044 } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK))
f9f4320a
YO
2045 {
2046 /* the warning about reginfo.ganch being used without intialization
bbe252da 2047 is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN
f9f4320a 2048 and we only enter this block when the same bit is set. */
58e23c8d 2049 char *tmp_s = reginfo.ganch - prog->gofs;
c584a96e
YO
2050
2051 if (tmp_s >= strbeg && regtry(&reginfo, &tmp_s))
6eb5f6b9
JH
2052 goto got_it;
2053 goto phooey;
2054 }
2055
2056 /* Messy cases: unanchored match. */
bbe252da 2057 if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
6eb5f6b9 2058 /* we have /x+whatever/ */
f2ed9b32 2059 /* it must be a one character string (XXXX Except UTF_PATTERN?) */
33b8afdf 2060 char ch;
bf93d4cc
GS
2061#ifdef DEBUGGING
2062 int did_match = 0;
2063#endif
f2ed9b32
KW
2064 if (!(utf8_target ? prog->anchored_utf8 : prog->anchored_substr))
2065 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2066 ch = SvPVX_const(utf8_target ? prog->anchored_utf8 : prog->anchored_substr)[0];
bf93d4cc 2067
f2ed9b32 2068 if (utf8_target) {
4cadc6a9 2069 REXEC_FBC_SCAN(
6eb5f6b9 2070 if (*s == ch) {
a3621e74 2071 DEBUG_EXECUTE_r( did_match = 1 );
24b23f37 2072 if (regtry(&reginfo, &s)) goto got_it;
6eb5f6b9
JH
2073 s += UTF8SKIP(s);
2074 while (s < strend && *s == ch)
2075 s += UTF8SKIP(s);
2076 }
4cadc6a9 2077 );
6eb5f6b9
JH
2078 }
2079 else {
4cadc6a9 2080 REXEC_FBC_SCAN(
6eb5f6b9 2081 if (*s == ch) {
a3621e74 2082 DEBUG_EXECUTE_r( did_match = 1 );
24b23f37 2083 if (regtry(&reginfo, &s)) goto got_it;
6eb5f6b9
JH
2084 s++;
2085 while (s < strend && *s == ch)
2086 s++;
2087 }
4cadc6a9 2088 );
6eb5f6b9 2089 }
a3621e74 2090 DEBUG_EXECUTE_r(if (!did_match)
bf93d4cc 2091 PerlIO_printf(Perl_debug_log,
b7953727
JH
2092 "Did not find anchored character...\n")
2093 );
6eb5f6b9 2094 }
a0714e2c
SS
2095 else if (prog->anchored_substr != NULL
2096 || prog->anchored_utf8 != NULL
2097 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
33b8afdf
JH
2098 && prog->float_max_offset < strend - s)) {
2099 SV *must;
2100 I32 back_max;
2101 I32 back_min;
2102 char *last;
6eb5f6b9 2103 char *last1; /* Last position checked before */
bf93d4cc
GS
2104#ifdef DEBUGGING
2105 int did_match = 0;
2106#endif
33b8afdf 2107 if (prog->anchored_substr || prog->anchored_utf8) {
f2ed9b32
KW
2108 if (!(utf8_target ? prog->anchored_utf8 : prog->anchored_substr))
2109 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2110 must = utf8_target ? prog->anchored_utf8 : prog->anchored_substr;
33b8afdf
JH
2111 back_max = back_min = prog->anchored_offset;
2112 } else {
f2ed9b32
KW
2113 if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
2114 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2115 must = utf8_target ? prog->float_utf8 : prog->float_substr;
33b8afdf
JH
2116 back_max = prog->float_max_offset;
2117 back_min = prog->float_min_offset;
2118 }
1de06328
YO
2119
2120
33b8afdf
JH
2121 if (must == &PL_sv_undef)
2122 /* could not downgrade utf8 check substring, so must fail */
2123 goto phooey;
2124
1de06328
YO
2125 if (back_min<0) {
2126 last = strend;
2127 } else {
2128 last = HOP3c(strend, /* Cannot start after this */
2129 -(I32)(CHR_SVLEN(must)
2130 - (SvTAIL(must) != 0) + back_min), strbeg);
2131 }
6eb5f6b9
JH
2132 if (s > PL_bostr)
2133 last1 = HOPc(s, -1);
2134 else
2135 last1 = s - 1; /* bogus */
2136
a0288114 2137 /* XXXX check_substr already used to find "s", can optimize if
6eb5f6b9
JH
2138 check_substr==must. */
2139 scream_pos = -1;
2140 dontbother = end_shift;
2141 strend = HOPc(strend, -dontbother);
2142 while ( (s <= last) &&
9041c2e3 2143 ((flags & REXEC_SCREAM)
1de06328 2144 ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg,
6eb5f6b9 2145 end_shift, &scream_pos, 0))
1de06328 2146 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)),
9041c2e3 2147 (unsigned char*)strend, must,
7fba1cd6 2148 multiline ? FBMrf_MULTILINE : 0))) ) {
4addbd3b 2149 /* we may be pointing at the wrong string */
07bc277f 2150 if ((flags & REXEC_SCREAM) && RXp_MATCH_COPIED(prog))
3f7c398e 2151 s = strbeg + (s - SvPVX_const(sv));
a3621e74 2152 DEBUG_EXECUTE_r( did_match = 1 );
6eb5f6b9
JH
2153 if (HOPc(s, -back_max) > last1) {
2154 last1 = HOPc(s, -back_min);
2155 s = HOPc(s, -back_max);
2156 }
2157 else {
52657f30 2158 char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
6eb5f6b9
JH
2159
2160 last1 = HOPc(s, -back_min);
52657f30 2161 s = t;
6eb5f6b9 2162 }
f2ed9b32 2163 if (utf8_target) {
6eb5f6b9 2164 while (s <= last1) {
24b23f37 2165 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2166 goto got_it;
2167 s += UTF8SKIP(s);
2168 }
2169 }
2170 else {
2171 while (s <= last1) {
24b23f37 2172 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2173 goto got_it;
2174 s++;
2175 }
2176 }
2177 }
ab3bbdeb 2178 DEBUG_EXECUTE_r(if (!did_match) {
f2ed9b32 2179 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
ab3bbdeb
YO
2180 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
2181 PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
33b8afdf 2182 ((must == prog->anchored_substr || must == prog->anchored_utf8)
bf93d4cc 2183 ? "anchored" : "floating"),
ab3bbdeb
YO
2184 quoted, RE_SV_TAIL(must));
2185 });
6eb5f6b9
JH
2186 goto phooey;
2187 }
f8fc2ecf 2188 else if ( (c = progi->regstclass) ) {
f14c76ed 2189 if (minlen) {
f8fc2ecf 2190 const OPCODE op = OP(progi->regstclass);
66e933ab 2191 /* don't bother with what can't match */
786e8c11 2192 if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
f14c76ed
RGS
2193 strend = HOPc(strend, -(minlen - 1));
2194 }
a3621e74 2195 DEBUG_EXECUTE_r({
be8e71aa 2196 SV * const prop = sv_newmortal();
32fc9b6a 2197 regprop(prog, prop, c);
0df25f3d 2198 {
f2ed9b32 2199 RE_PV_QUOTED_DECL(quoted,utf8_target,PERL_DEBUG_PAD_ZERO(1),
ab3bbdeb 2200 s,strend-s,60);
0df25f3d 2201 PerlIO_printf(Perl_debug_log,
1c8f8eb1 2202 "Matching stclass %.*s against %s (%d bytes)\n",
e4f74956 2203 (int)SvCUR(prop), SvPVX_const(prop),
ab3bbdeb 2204 quoted, (int)(strend - s));
0df25f3d 2205 }
ffc61ed2 2206 });
3b0527fe 2207 if (find_byclass(prog, c, s, strend, &reginfo))
6eb5f6b9 2208 goto got_it;
07be1b83 2209 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
d6a28714
JH
2210 }
2211 else {
2212 dontbother = 0;
a0714e2c 2213 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
33b8afdf 2214 /* Trim the end. */
d6a28714 2215 char *last;
33b8afdf
JH
2216 SV* float_real;
2217
f2ed9b32
KW
2218 if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
2219 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2220 float_real = utf8_target ? prog->float_utf8 : prog->float_substr;
d6a28714
JH
2221
2222 if (flags & REXEC_SCREAM) {
33b8afdf 2223 last = screaminstr(sv, float_real, s - strbeg,
d6a28714
JH
2224 end_shift, &scream_pos, 1); /* last one */
2225 if (!last)
ffc61ed2 2226 last = scream_olds; /* Only one occurrence. */
4addbd3b 2227 /* we may be pointing at the wrong string */
07bc277f 2228 else if (RXp_MATCH_COPIED(prog))
3f7c398e 2229 s = strbeg + (s - SvPVX_const(sv));
b8c5462f 2230 }
d6a28714
JH
2231 else {
2232 STRLEN len;
cfd0369c 2233 const char * const little = SvPV_const(float_real, len);
d6a28714 2234
33b8afdf 2235 if (SvTAIL(float_real)) {
d6a28714
JH
2236 if (memEQ(strend - len + 1, little, len - 1))
2237 last = strend - len + 1;
7fba1cd6 2238 else if (!multiline)
9041c2e3 2239 last = memEQ(strend - len, little, len)
bd61b366 2240 ? strend - len : NULL;
b8c5462f 2241 else
d6a28714
JH
2242 goto find_last;
2243 } else {
2244 find_last:
9041c2e3 2245 if (len)
d6a28714 2246 last = rninstr(s, strend, little, little + len);
b8c5462f 2247 else
a0288114 2248 last = strend; /* matching "$" */
b8c5462f 2249 }
b8c5462f 2250 }
bf93d4cc 2251 if (last == NULL) {
6bda09f9
YO
2252 DEBUG_EXECUTE_r(
2253 PerlIO_printf(Perl_debug_log,
2254 "%sCan't trim the tail, match fails (should not happen)%s\n",
2255 PL_colors[4], PL_colors[5]));
bf93d4cc
GS
2256 goto phooey; /* Should not happen! */
2257 }
d6a28714
JH
2258 dontbother = strend - last + prog->float_min_offset;
2259 }
2260 if (minlen && (dontbother < minlen))
2261 dontbother = minlen - 1;
2262 strend -= dontbother; /* this one's always in bytes! */
2263 /* We don't know much -- general case. */
f2ed9b32 2264 if (utf8_target) {
d6a28714 2265 for (;;) {
24b23f37 2266 if (regtry(&reginfo, &s))
d6a28714
JH
2267 goto got_it;
2268 if (s >= strend)
2269 break;
b8c5462f 2270 s += UTF8SKIP(s);
d6a28714
JH
2271 };
2272 }
2273 else {
2274 do {
24b23f37 2275 if (regtry(&reginfo, &s))
d6a28714
JH
2276 goto got_it;
2277 } while (s++ < strend);
2278 }
2279 }
2280
2281 /* Failure. */
2282 goto phooey;
2283
2284got_it:
e9105d30 2285 Safefree(swap);
288b8c02 2286 RX_MATCH_TAINTED_set(rx, PL_reg_flags & RF_tainted);
d6a28714 2287
19b95bf0 2288 if (PL_reg_eval_set)
4f639d21 2289 restore_pos(aTHX_ prog);
5daac39c
NC
2290 if (RXp_PAREN_NAMES(prog))
2291 (void)hv_iterinit(RXp_PAREN_NAMES(prog));
d6a28714
JH
2292
2293 /* make sure $`, $&, $', and $digit will work later */
2294 if ( !(flags & REXEC_NOT_FIRST) ) {
288b8c02 2295 RX_MATCH_COPY_FREE(rx);
d6a28714 2296 if (flags & REXEC_COPY_STR) {
be8e71aa 2297 const I32 i = PL_regeol - startpos + (stringarg - strbeg);
f8c7b90f 2298#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
2299 if ((SvIsCOW(sv)
2300 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2301 if (DEBUG_C_TEST) {
2302 PerlIO_printf(Perl_debug_log,
2303 "Copy on write: regexp capture, type %d\n",
2304 (int) SvTYPE(sv));
2305 }
2306 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
d5263905 2307 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
ed252734
NC
2308 assert (SvPOKp(prog->saved_copy));
2309 } else
2310#endif
2311 {
288b8c02 2312 RX_MATCH_COPIED_on(rx);
ed252734
NC
2313 s = savepvn(strbeg, i);
2314 prog->subbeg = s;
2315 }
d6a28714 2316 prog->sublen = i;
d6a28714
JH
2317 }
2318 else {
2319 prog->subbeg = strbeg;
2320 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2321 }
2322 }
9041c2e3 2323
d6a28714
JH
2324 return 1;
2325
2326phooey:
a3621e74 2327 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
e4584336 2328 PL_colors[4], PL_colors[5]));
d6a28714 2329 if (PL_reg_eval_set)
4f639d21 2330 restore_pos(aTHX_ prog);
e9105d30 2331 if (swap) {
c74340f9 2332 /* we failed :-( roll it back */
e9105d30
GG
2333 Safefree(prog->offs);
2334 prog->offs = swap;
2335 }
2336
d6a28714
JH
2337 return 0;
2338}
2339
6bda09f9 2340
d6a28714
JH
2341/*
2342 - regtry - try match at specific point
2343 */
2344STATIC I32 /* 0 failure, 1 success */
24b23f37 2345S_regtry(pTHX_ regmatch_info *reginfo, char **startpos)
d6a28714 2346{
97aff369 2347 dVAR;
d6a28714 2348 CHECKPOINT lastcp;
288b8c02
NC
2349 REGEXP *const rx = reginfo->prog;
2350 regexp *const prog = (struct regexp *)SvANY(rx);
f8fc2ecf 2351 RXi_GET_DECL(prog,progi);
a3621e74 2352 GET_RE_DEBUG_FLAGS_DECL;
7918f24d
NC
2353
2354 PERL_ARGS_ASSERT_REGTRY;
2355
24b23f37 2356 reginfo->cutpoint=NULL;
d6a28714 2357
bbe252da 2358 if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) {
d6a28714
JH
2359 MAGIC *mg;
2360
2361 PL_reg_eval_set = RS_init;
a3621e74 2362 DEBUG_EXECUTE_r(DEBUG_s(
b900a521
JH
2363 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
2364 (IV)(PL_stack_sp - PL_stack_base));
d6a28714 2365 ));
ea8d6ae1 2366 SAVESTACK_CXPOS();
d6a28714
JH
2367 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
2368 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
2369 SAVETMPS;
2370 /* Apparently this is not needed, judging by wantarray. */
e8347627 2371 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
d6a28714
JH
2372 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
2373
3b0527fe 2374 if (reginfo->sv) {
d6a28714 2375 /* Make $_ available to executed code. */
3b0527fe 2376 if (reginfo->sv != DEFSV) {
59f00321 2377 SAVE_DEFSV;
414bf5ae 2378 DEFSV_set(reginfo->sv);
b8c5462f 2379 }
d6a28714 2380
3b0527fe
DM
2381 if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
2382 && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
d6a28714 2383 /* prepare for quick setting of pos */
d300d9fa 2384#ifdef PERL_OLD_COPY_ON_WRITE
51a9ea20
NC
2385 if (SvIsCOW(reginfo->sv))
2386 sv_force_normal_flags(reginfo->sv, 0);
d300d9fa 2387#endif
3dab1dad 2388 mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global,
d300d9fa 2389 &PL_vtbl_mglob, NULL, 0);
d6a28714 2390 mg->mg_len = -1;
b8c5462f 2391 }
d6a28714
JH
2392 PL_reg_magic = mg;
2393 PL_reg_oldpos = mg->mg_len;
4f639d21 2394 SAVEDESTRUCTOR_X(restore_pos, prog);
d6a28714 2395 }
09687e5a 2396 if (!PL_reg_curpm) {
a02a5408 2397 Newxz(PL_reg_curpm, 1, PMOP);
09687e5a
AB
2398#ifdef USE_ITHREADS
2399 {
14a49a24 2400 SV* const repointer = &PL_sv_undef;
92313705
NC
2401 /* this regexp is also owned by the new PL_reg_curpm, which
2402 will try to free it. */
d2ece331 2403 av_push(PL_regex_padav, repointer);
09687e5a
AB
2404 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2405 PL_regex_pad = AvARRAY(PL_regex_padav);
2406 }
2407#endif
2408 }
86c29d75
NC
2409#ifdef USE_ITHREADS
2410 /* It seems that non-ithreads works both with and without this code.
2411 So for efficiency reasons it seems best not to have the code
2412 compiled when it is not needed. */
92313705
NC
2413 /* This is safe against NULLs: */
2414 ReREFCNT_dec(PM_GETRE(PL_reg_curpm));
2415 /* PM_reg_curpm owns a reference to this regexp. */
2416 ReREFCNT_inc(rx);
86c29d75 2417#endif
288b8c02 2418 PM_SETRE(PL_reg_curpm, rx);
d6a28714
JH
2419 PL_reg_oldcurpm = PL_curpm;
2420 PL_curpm = PL_reg_curpm;
07bc277f 2421 if (RXp_MATCH_COPIED(prog)) {
d6a28714
JH
2422 /* Here is a serious problem: we cannot rewrite subbeg,
2423 since it may be needed if this match fails. Thus
2424 $` inside (?{}) could fail... */
2425 PL_reg_oldsaved = prog->subbeg;
2426 PL_reg_oldsavedlen = prog->sublen;
f8c7b90f 2427#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
2428 PL_nrs = prog->saved_copy;
2429#endif
07bc277f 2430 RXp_MATCH_COPIED_off(prog);
d6a28714
JH
2431 }
2432 else
bd61b366 2433 PL_reg_oldsaved = NULL;
d6a28714
JH
2434 prog->subbeg = PL_bostr;
2435 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2436 }
24b23f37 2437 DEBUG_EXECUTE_r(PL_reg_starttry = *startpos);
f0ab9afb 2438 prog->offs[0].start = *startpos - PL_bostr;
24b23f37 2439 PL_reginput = *startpos;
d6a28714 2440 PL_reglastparen = &prog->lastparen;
a01268b5 2441 PL_reglastcloseparen = &prog->lastcloseparen;
d6a28714 2442 prog->lastparen = 0;
03994de8 2443 prog->lastcloseparen = 0;
d6a28714 2444 PL_regsize = 0;
f0ab9afb 2445 PL_regoffs = prog->offs;
d6a28714
JH
2446 if (PL_reg_start_tmpl <= prog->nparens) {
2447 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
2448 if(PL_reg_start_tmp)
2449 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2450 else
a02a5408 2451 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
d6a28714
JH
2452 }
2453
2454 /* XXXX What this code is doing here?!!! There should be no need
2455 to do this again and again, PL_reglastparen should take care of
3dd2943c 2456 this! --ilya*/
dafc8851
JH
2457
2458 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2459 * Actually, the code in regcppop() (which Ilya may be meaning by
daf18116 2460 * PL_reglastparen), is not needed at all by the test suite
225593e1
DM
2461 * (op/regexp, op/pat, op/split), but that code is needed otherwise
2462 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
2463 * Meanwhile, this code *is* needed for the
daf18116
JH
2464 * above-mentioned test suite tests to succeed. The common theme
2465 * on those tests seems to be returning null fields from matches.
225593e1 2466 * --jhi updated by dapm */
dafc8851 2467#if 1
d6a28714 2468 if (prog->nparens) {
f0ab9afb 2469 regexp_paren_pair *pp = PL_regoffs;
097eb12c 2470 register I32 i;
eb160463 2471 for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
f0ab9afb
NC
2472 ++pp;
2473 pp->start = -1;
2474 pp->end = -1;
d6a28714
JH
2475 }
2476 }
dafc8851 2477#endif
02db2b7b 2478 REGCP_SET(lastcp);
f8fc2ecf 2479 if (regmatch(reginfo, progi->program + 1)) {
f0ab9afb 2480 PL_regoffs[0].end = PL_reginput - PL_bostr;
d6a28714
JH
2481 return 1;
2482 }
24b23f37
YO
2483 if (reginfo->cutpoint)
2484 *startpos= reginfo->cutpoint;
02db2b7b 2485 REGCP_UNWIND(lastcp);
d6a28714
JH
2486 return 0;
2487}
2488
02db2b7b 2489
8ba1375e
MJD
2490#define sayYES goto yes
2491#define sayNO goto no
262b90c4 2492#define sayNO_SILENT goto no_silent
8ba1375e 2493
f9f4320a
YO
2494/* we dont use STMT_START/END here because it leads to
2495 "unreachable code" warnings, which are bogus, but distracting. */
2496#define CACHEsayNO \
c476f425
DM
2497 if (ST.cache_mask) \
2498 PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \
f9f4320a 2499 sayNO
3298f257 2500
a3621e74 2501/* this is used to determine how far from the left messages like
265c4333
YO
2502 'failed...' are printed. It should be set such that messages
2503 are inline with the regop output that created them.
a3621e74 2504*/
265c4333 2505#define REPORT_CODE_OFF 32
a3621e74
YO
2506
2507
40a82448
DM
2508#define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
2509#define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
9e137952 2510
86545054
DM
2511#define SLAB_FIRST(s) (&(s)->states[0])
2512#define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
2513
5d9a96ca
DM
2514/* grab a new slab and return the first slot in it */
2515
2516STATIC regmatch_state *
2517S_push_slab(pTHX)
2518{
a35a87e7 2519#if PERL_VERSION < 9 && !defined(PERL_CORE)
54df2634
NC
2520 dMY_CXT;
2521#endif
5d9a96ca
DM
2522 regmatch_slab *s = PL_regmatch_slab->next;
2523 if (!s) {
2524 Newx(s, 1, regmatch_slab);
2525 s->prev = PL_regmatch_slab;
2526 s->next = NULL;
2527 PL_regmatch_slab->next = s;
2528 }
2529 PL_regmatch_slab = s;
86545054 2530 return SLAB_FIRST(s);
5d9a96ca 2531}
5b47454d 2532
95b24440 2533
40a82448
DM
2534/* push a new state then goto it */
2535
2536#define PUSH_STATE_GOTO(state, node) \
2537 scan = node; \
2538 st->resume_state = state; \
2539 goto push_state;
2540
2541/* push a new state with success backtracking, then goto it */
2542
2543#define PUSH_YES_STATE_GOTO(state, node) \
2544 scan = node; \
2545 st->resume_state = state; \
2546 goto push_yes_state;
2547
aa283a38 2548
aa283a38 2549
d6a28714 2550/*
95b24440 2551
bf1f174e
DM
2552regmatch() - main matching routine
2553
2554This is basically one big switch statement in a loop. We execute an op,
2555set 'next' to point the next op, and continue. If we come to a point which
2556we may need to backtrack to on failure such as (A|B|C), we push a
2557backtrack state onto the backtrack stack. On failure, we pop the top
2558state, and re-enter the loop at the state indicated. If there are no more
2559states to pop, we return failure.
2560
2561Sometimes we also need to backtrack on success; for example /A+/, where
2562after successfully matching one A, we need to go back and try to
2563match another one; similarly for lookahead assertions: if the assertion
2564completes successfully, we backtrack to the state just before the assertion
2565and then carry on. In these cases, the pushed state is marked as
2566'backtrack on success too'. This marking is in fact done by a chain of
2567pointers, each pointing to the previous 'yes' state. On success, we pop to
2568the nearest yes state, discarding any intermediate failure-only states.
2569Sometimes a yes state is pushed just to force some cleanup code to be
2570called at the end of a successful match or submatch; e.g. (??{$re}) uses
2571it to free the inner regex.
2572
2573Note that failure backtracking rewinds the cursor position, while
2574success backtracking leaves it alone.
2575
2576A pattern is complete when the END op is executed, while a subpattern
2577such as (?=foo) is complete when the SUCCESS op is executed. Both of these
2578ops trigger the "pop to last yes state if any, otherwise return true"
2579behaviour.
2580
2581A common convention in this function is to use A and B to refer to the two
2582subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
2583the subpattern to be matched possibly multiple times, while B is the entire
2584rest of the pattern. Variable and state names reflect this convention.
2585
2586The states in the main switch are the union of ops and failure/success of
2587substates associated with with that op. For example, IFMATCH is the op
2588that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
2589'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
2590successfully matched A and IFMATCH_A_fail is a state saying that we have
2591just failed to match A. Resume states always come in pairs. The backtrack
2592state we push is marked as 'IFMATCH_A', but when that is popped, we resume
2593at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
2594on success or failure.
2595
2596The struct that holds a backtracking state is actually a big union, with
2597one variant for each major type of op. The variable st points to the
2598top-most backtrack struct. To make the code clearer, within each
2599block of code we #define ST to alias the relevant union.
2600
2601Here's a concrete example of a (vastly oversimplified) IFMATCH
2602implementation:
2603
2604 switch (state) {
2605 ....
2606
2607#define ST st->u.ifmatch
2608
2609 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2610 ST.foo = ...; // some state we wish to save
95b24440 2611 ...
bf1f174e
DM
2612 // push a yes backtrack state with a resume value of
2613 // IFMATCH_A/IFMATCH_A_fail, then continue execution at the
2614 // first node of A:
2615 PUSH_YES_STATE_GOTO(IFMATCH_A, A);
2616 // NOTREACHED
2617
2618 case IFMATCH_A: // we have successfully executed A; now continue with B
2619 next = B;
2620 bar = ST.foo; // do something with the preserved value
2621 break;
2622
2623 case IFMATCH_A_fail: // A failed, so the assertion failed
2624 ...; // do some housekeeping, then ...
2625 sayNO; // propagate the failure
2626
2627#undef ST
95b24440 2628
bf1f174e
DM
2629 ...
2630 }
95b24440 2631
bf1f174e
DM
2632For any old-timers reading this who are familiar with the old recursive
2633approach, the code above is equivalent to:
95b24440 2634
bf1f174e
DM
2635 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2636 {
2637 int foo = ...
95b24440 2638 ...
bf1f174e
DM
2639 if (regmatch(A)) {
2640 next = B;
2641 bar = foo;
2642 break;
95b24440 2643 }
bf1f174e
DM
2644 ...; // do some housekeeping, then ...
2645 sayNO; // propagate the failure
95b24440 2646 }
bf1f174e
DM
2647
2648The topmost backtrack state, pointed to by st, is usually free. If you
2649want to claim it, populate any ST.foo fields in it with values you wish to
2650save, then do one of
2651
2652 PUSH_STATE_GOTO(resume_state, node);
2653 PUSH_YES_STATE_GOTO(resume_state, node);
2654
2655which sets that backtrack state's resume value to 'resume_state', pushes a
2656new free entry to the top of the backtrack stack, then goes to 'node'.
2657On backtracking, the free slot is popped, and the saved state becomes the
2658new free state. An ST.foo field in this new top state can be temporarily
2659accessed to retrieve values, but once the main loop is re-entered, it
2660becomes available for reuse.
2661
2662Note that the depth of the backtrack stack constantly increases during the
2663left-to-right execution of the pattern, rather than going up and down with
2664the pattern nesting. For example the stack is at its maximum at Z at the
2665end of the pattern, rather than at X in the following:
2666
2667 /(((X)+)+)+....(Y)+....Z/
2668
2669The only exceptions to this are lookahead/behind assertions and the cut,
2670(?>A), which pop all the backtrack states associated with A before
2671continuing.
2672
2673Bascktrack state structs are allocated in slabs of about 4K in size.
2674PL_regmatch_state and st always point to the currently active state,
2675and PL_regmatch_slab points to the slab currently containing
2676PL_regmatch_state. The first time regmatch() is called, the first slab is
2677allocated, and is never freed until interpreter destruction. When the slab
2678is full, a new one is allocated and chained to the end. At exit from
2679regmatch(), slabs allocated since entry are freed.
2680
2681*/
95b24440 2682
40a82448 2683
5bc10b2c 2684#define DEBUG_STATE_pp(pp) \
265c4333 2685 DEBUG_STATE_r({ \
f2ed9b32 2686 DUMP_EXEC_POS(locinput, scan, utf8_target); \
5bc10b2c 2687 PerlIO_printf(Perl_debug_log, \
5d458dd8 2688 " %*s"pp" %s%s%s%s%s\n", \
5bc10b2c 2689 depth*2, "", \
13d6edb4 2690 PL_reg_name[st->resume_state], \
5d458dd8
YO
2691 ((st==yes_state||st==mark_state) ? "[" : ""), \
2692 ((st==yes_state) ? "Y" : ""), \
2693 ((st==mark_state) ? "M" : ""), \
2694 ((st==yes_state||st==mark_state) ? "]" : "") \
2695 ); \
265c4333 2696 });
5bc10b2c 2697
40a82448 2698
3dab1dad 2699#define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)
95b24440 2700
3df15adc 2701#ifdef DEBUGGING
5bc10b2c 2702
ab3bbdeb 2703STATIC void
f2ed9b32 2704S_debug_start_match(pTHX_ const REGEXP *prog, const bool utf8_target,
ab3bbdeb
YO
2705 const char *start, const char *end, const char *blurb)
2706{
efd26800 2707 const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;
7918f24d
NC
2708
2709 PERL_ARGS_ASSERT_DEBUG_START_MATCH;
2710
ab3bbdeb
YO
2711 if (!PL_colorset)
2712 reginitcolors();
2713 {
2714 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
d2c6dc5e 2715 RX_PRECOMP_const(prog), RX_PRELEN(prog), 60);
ab3bbdeb 2716
f2ed9b32 2717 RE_PV_QUOTED_DECL(s1, utf8_target, PERL_DEBUG_PAD_ZERO(1),
ab3bbdeb
YO
2718 start, end - start, 60);
2719
2720 PerlIO_printf(Perl_debug_log,
2721 "%s%s REx%s %s against %s\n",
2722 PL_colors[4], blurb, PL_colors[5], s0, s1);
2723
f2ed9b32 2724 if (utf8_target||utf8_pat)
1de06328
YO
2725 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
2726 utf8_pat ? "pattern" : "",
f2ed9b32
KW
2727 utf8_pat && utf8_target ? " and " : "",
2728 utf8_target ? "string" : ""
ab3bbdeb
YO
2729 );
2730 }
2731}
3df15adc
YO
2732
2733STATIC void
786e8c11
YO
2734S_dump_exec_pos(pTHX_ const char *locinput,
2735 const regnode *scan,
2736 const char *loc_regeol,
2737 const char *loc_bostr,
2738 const char *loc_reg_starttry,
f2ed9b32 2739 const bool utf8_target)
07be1b83 2740{
786e8c11 2741 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
07be1b83 2742 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
786e8c11 2743 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
07be1b83
YO
2744 /* The part of the string before starttry has one color
2745 (pref0_len chars), between starttry and current
2746 position another one (pref_len - pref0_len chars),
2747 after the current position the third one.
2748 We assume that pref0_len <= pref_len, otherwise we
2749 decrease pref0_len. */
786e8c11
YO
2750 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
2751 ? (5 + taill) - l : locinput - loc_bostr;
07be1b83
YO
2752 int pref0_len;
2753
7918f24d
NC
2754 PERL_ARGS_ASSERT_DUMP_EXEC_POS;
2755
f2ed9b32 2756 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
07be1b83 2757 pref_len++;
786e8c11
YO
2758 pref0_len = pref_len - (locinput - loc_reg_starttry);
2759 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
2760 l = ( loc_regeol - locinput > (5 + taill) - pref_len
2761 ? (5 + taill) - pref_len : loc_regeol - locinput);
f2ed9b32 2762 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
07be1b83
YO
2763 l--;
2764 if (pref0_len < 0)
2765 pref0_len = 0;
2766 if (pref0_len > pref_len)
2767 pref0_len = pref_len;
2768 {
f2ed9b32 2769 const int is_uni = (utf8_target && OP(scan) != CANY) ? 1 : 0;
0df25f3d 2770
ab3bbdeb 2771 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
1de06328 2772 (locinput - pref_len),pref0_len, 60, 4, 5);
0df25f3d 2773
ab3bbdeb 2774 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
3df15adc 2775 (locinput - pref_len + pref0_len),
1de06328 2776 pref_len - pref0_len, 60, 2, 3);
0df25f3d 2777
ab3bbdeb 2778 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
1de06328 2779 locinput, loc_regeol - locinput, 10, 0, 1);
0df25f3d 2780
1de06328 2781 const STRLEN tlen=len0+len1+len2;
3df15adc 2782 PerlIO_printf(Perl_debug_log,
ab3bbdeb 2783 "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
786e8c11 2784 (IV)(locinput - loc_bostr),
07be1b83 2785 len0, s0,
07be1b83 2786 len1, s1,
07be1b83 2787 (docolor ? "" : "> <"),
07be1b83 2788 len2, s2,
f9f4320a 2789 (int)(tlen > 19 ? 0 : 19 - tlen),
07be1b83
YO
2790 "");
2791 }
2792}
3df15adc 2793
07be1b83
YO
2794#endif
2795
0a4db386
YO
2796/* reg_check_named_buff_matched()
2797 * Checks to see if a named buffer has matched. The data array of
2798 * buffer numbers corresponding to the buffer is expected to reside
2799 * in the regexp->data->data array in the slot stored in the ARG() of
2800 * node involved. Note that this routine doesn't actually care about the
2801 * name, that information is not preserved from compilation to execution.
2802 * Returns the index of the leftmost defined buffer with the given name
2803 * or 0 if non of the buffers matched.
2804 */
2805STATIC I32
7918f24d
NC
2806S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan)
2807{
0a4db386 2808 I32 n;
f8fc2ecf 2809 RXi_GET_DECL(rex,rexi);
ad64d0ec 2810 SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
0a4db386 2811 I32 *nums=(I32*)SvPVX(sv_dat);
7918f24d
NC
2812
2813 PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED;
2814
0a4db386
YO
2815 for ( n=0; n<SvIVX(sv_dat); n++ ) {
2816 if ((I32)*PL_reglastparen >= nums[n] &&
f0ab9afb 2817 PL_regoffs[nums[n]].end != -1)
0a4db386
YO
2818 {
2819 return nums[n];
2820 }
2821 }
2822 return 0;
2823}
2824
2f554ef7
DM
2825
2826/* free all slabs above current one - called during LEAVE_SCOPE */
2827
2828STATIC void
2829S_clear_backtrack_stack(pTHX_ void *p)
2830{
2831 regmatch_slab *s = PL_regmatch_slab->next;
2832 PERL_UNUSED_ARG(p);
2833
2834 if (!s)
2835 return;
2836 PL_regmatch_slab->next = NULL;
2837 while (s) {
2838 regmatch_slab * const osl = s;
2839 s = s->next;
2840 Safefree(osl);
2841 }
2842}
2843
2844
28d8d7f4
YO
2845#define SETREX(Re1,Re2) \
2846 if (PL_reg_eval_set) PM_SETRE((PL_reg_curpm), (Re2)); \
2847 Re1 = (Re2)
2848
d6a28714 2849STATIC I32 /* 0 failure, 1 success */
24b23f37 2850S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
d6a28714 2851{
a35a87e7 2852#if PERL_VERSION < 9 && !defined(PERL_CORE)
54df2634
NC
2853 dMY_CXT;
2854#endif
27da23d5 2855 dVAR;
f2ed9b32 2856 register const bool utf8_target = PL_reg_match_utf8;
4ad0818d 2857 const U32 uniflags = UTF8_ALLOW_DEFAULT;
288b8c02
NC
2858 REGEXP *rex_sv = reginfo->prog;
2859 regexp *rex = (struct regexp *)SvANY(rex_sv);
f8fc2ecf 2860 RXi_GET_DECL(rex,rexi);
2f554ef7 2861 I32 oldsave;
5d9a96ca
DM
2862 /* the current state. This is a cached copy of PL_regmatch_state */
2863 register regmatch_state *st;
5d9a96ca
DM
2864 /* cache heavy used fields of st in registers */
2865 register regnode *scan;
2866 register regnode *next;
438e9bae 2867 register U32 n = 0; /* general value; init to avoid compiler warning */
24d3c4a9 2868 register I32 ln = 0; /* len or last; init to avoid compiler warning */
5d9a96ca 2869 register char *locinput = PL_reginput;
5d9a96ca 2870 register I32 nextchr; /* is always set to UCHARAT(locinput) */
24d3c4a9 2871
b69b0499 2872 bool result = 0; /* return value of S_regmatch */
24d3c4a9 2873 int depth = 0; /* depth of backtrack stack */
4b196cd4
YO
2874 U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
2875 const U32 max_nochange_depth =
2876 (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
2877 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
77cb431f
DM
2878 regmatch_state *yes_state = NULL; /* state to pop to on success of
2879 subpattern */
e2e6a0f1
YO
2880 /* mark_state piggy backs on the yes_state logic so that when we unwind
2881 the stack on success we can update the mark_state as we go */
2882 regmatch_state *mark_state = NULL; /* last mark state we have seen */
faec1544 2883 regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
b8591aee 2884 struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */
40a82448 2885 U32 state_num;
5d458dd8
YO
2886 bool no_final = 0; /* prevent failure from backtracking? */
2887 bool do_cutgroup = 0; /* no_final only until next branch/trie entry */
e2e6a0f1 2888 char *startpoint = PL_reginput;
5d458dd8
YO
2889 SV *popmark = NULL; /* are we looking for a mark? */
2890 SV *sv_commit = NULL; /* last mark name seen in failure */
2891 SV *sv_yes_mark = NULL; /* last mark name we have seen
2892 during a successfull match */
2893 U32 lastopen = 0; /* last open we saw */
2894 bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;
19b95bf0 2895 SV* const oreplsv = GvSV(PL_replgv);
24d3c4a9
DM
2896 /* these three flags are set by various ops to signal information to
2897 * the very next op. They have a useful lifetime of exactly one loop
2898 * iteration, and are not preserved or restored by state pushes/pops
2899 */
2900 bool sw = 0; /* the condition value in (?(cond)a|b) */
2901 bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */
2902 int logical = 0; /* the following EVAL is:
2903 0: (?{...})
2904 1: (?(?{...})X|Y)
2905 2: (??{...})
2906 or the following IFMATCH/UNLESSM is:
2907 false: plain (?=foo)
2908 true: used as a condition: (?(?=foo))
2909 */
95b24440 2910#ifdef DEBUGGING
e68ec53f 2911 GET_RE_DEBUG_FLAGS_DECL;
d6a28714
JH
2912#endif
2913
7918f24d
NC
2914 PERL_ARGS_ASSERT_REGMATCH;
2915
3b57cd43 2916 DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
24b23f37 2917 PerlIO_printf(Perl_debug_log,"regmatch start\n");
3b57cd43 2918 }));
5d9a96ca
DM
2919 /* on first ever call to regmatch, allocate first slab */
2920 if (!PL_regmatch_slab) {
2921 Newx(PL_regmatch_slab, 1, regmatch_slab);
2922 PL_regmatch_slab->prev = NULL;
2923 PL_regmatch_slab->next = NULL;
86545054 2924 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
5d9a96ca
DM
2925 }
2926
2f554ef7
DM
2927 oldsave = PL_savestack_ix;
2928 SAVEDESTRUCTOR_X(S_clear_backtrack_stack, NULL);
2929 SAVEVPTR(PL_regmatch_slab);
2930 SAVEVPTR(PL_regmatch_state);
5d9a96ca
DM
2931
2932 /* grab next free state slot */
2933 st = ++PL_regmatch_state;
86545054 2934 if (st > SLAB_LAST(PL_regmatch_slab))
5d9a96ca
DM
2935 st = PL_regmatch_state = S_push_slab(aTHX);
2936
d6a28714
JH
2937 /* Note that nextchr is a byte even in UTF */
2938 nextchr = UCHARAT(locinput);
2939 scan = prog;
2940 while (scan != NULL) {
8ba1375e 2941
a3621e74 2942 DEBUG_EXECUTE_r( {
6136c704 2943 SV * const prop = sv_newmortal();
1de06328 2944 regnode *rnext=regnext(scan);
f2ed9b32 2945 DUMP_EXEC_POS( locinput, scan, utf8_target );
32fc9b6a 2946 regprop(rex, prop, scan);
07be1b83
YO
2947
2948 PerlIO_printf(Perl_debug_log,
2949 "%3"IVdf":%*s%s(%"IVdf")\n",
f8fc2ecf 2950 (IV)(scan - rexi->program), depth*2, "",
07be1b83 2951 SvPVX_const(prop),
1de06328 2952 (PL_regkind[OP(scan)] == END || !rnext) ?
f8fc2ecf 2953 0 : (IV)(rnext - rexi->program));
2a782b5b 2954 });
d6a28714
JH
2955
2956 next = scan + NEXT_OFF(scan);
2957 if (next == scan)
2958 next = NULL;
40a82448 2959 state_num = OP(scan);
d6a28714 2960
40a82448 2961 reenter_switch:
34a81e2b
B
2962
2963 assert(PL_reglastparen == &rex->lastparen);
2964 assert(PL_reglastcloseparen == &rex->lastcloseparen);
2965 assert(PL_regoffs == rex->offs);
2966
40a82448 2967 switch (state_num) {
d6a28714 2968 case BOL:
7fba1cd6 2969 if (locinput == PL_bostr)
d6a28714 2970 {
3b0527fe 2971 /* reginfo->till = reginfo->bol; */
b8c5462f
JH
2972 break;
2973 }
d6a28714
JH
2974 sayNO;
2975 case MBOL:
12d33761
HS
2976 if (locinput == PL_bostr ||
2977 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
d6a28714 2978 {
b8c5462f
JH
2979 break;
2980 }
d6a28714
JH
2981 sayNO;
2982 case SBOL:
c2a73568 2983 if (locinput == PL_bostr)
b8c5462f 2984 break;
d6a28714
JH
2985 sayNO;
2986 case GPOS:
3b0527fe 2987 if (locinput == reginfo->ganch)
d6a28714
JH
2988 break;
2989 sayNO;
ee9b8eae
YO
2990
2991 case KEEPS:
2992 /* update the startpoint */
f0ab9afb 2993 st->u.keeper.val = PL_regoffs[0].start;
ee9b8eae 2994 PL_reginput = locinput;
f0ab9afb 2995 PL_regoffs[0].start = locinput - PL_bostr;
ee9b8eae
YO
2996 PUSH_STATE_GOTO(KEEPS_next, next);
2997 /*NOT-REACHED*/
2998 case KEEPS_next_fail:
2999 /* rollback the start point change */
f0ab9afb 3000 PL_regoffs[0].start = st->u.keeper.val;
ee9b8eae
YO
3001 sayNO_SILENT;
3002 /*NOT-REACHED*/
d6a28714 3003 case EOL:
d6a28714
JH
3004 goto seol;
3005 case MEOL:
d6a28714 3006 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 3007 sayNO;
b8c5462f 3008 break;
d6a28714
JH
3009 case SEOL:
3010 seol:
3011 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 3012 sayNO;
d6a28714 3013 if (PL_regeol - locinput > 1)
b8c5462f 3014 sayNO;
b8c5462f 3015 break;
d6a28714
JH
3016 case EOS:
3017 if (PL_regeol != locinput)
b8c5462f 3018 sayNO;
d6a28714 3019 break;
ffc61ed2 3020 case SANY:
d6a28714 3021 if (!nextchr && locinput >= PL_regeol)
4633a7c4 3022 sayNO;
f2ed9b32 3023 if (utf8_target) {
f33976b4
DB
3024 locinput += PL_utf8skip[nextchr];
3025 if (locinput > PL_regeol)
3026 sayNO;
3027 nextchr = UCHARAT(locinput);
3028 }
3029 else
3030 nextchr = UCHARAT(++locinput);
3031 break;
3032 case CANY:
3033 if (!nextchr && locinput >= PL_regeol)
3034 sayNO;
b8c5462f 3035 nextchr = UCHARAT(++locinput);
a0d0e21e 3036 break;
ffc61ed2 3037 case REG_ANY:
1aa99e6b
IH
3038 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
3039 sayNO;
f2ed9b32 3040 if (utf8_target) {
b8c5462f 3041 locinput += PL_utf8skip[nextchr];
d6a28714
JH
3042 if (locinput > PL_regeol)
3043 sayNO;
a0ed51b3 3044 nextchr = UCHARAT(locinput);
a0ed51b3 3045 }
1aa99e6b
IH
3046 else
3047 nextchr = UCHARAT(++locinput);
a0ed51b3 3048 break;
166ba7cd
DM
3049
3050#undef ST
3051#define ST st->u.trie
786e8c11
YO
3052 case TRIEC:
3053 /* In this case the charclass data is available inline so
3054 we can fail fast without a lot of extra overhead.
3055 */
f2ed9b32 3056 if (scan->flags == EXACT || !utf8_target) {
786e8c11
YO
3057 if(!ANYOF_BITMAP_TEST(scan, *locinput)) {
3058 DEBUG_EXECUTE_r(
3059 PerlIO_printf(Perl_debug_log,
3060 "%*s %sfailed to match trie start class...%s\n",
5bc10b2c 3061 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
786e8c11
YO
3062 );
3063 sayNO_SILENT;
3064 /* NOTREACHED */
3065 }
3066 }
3067 /* FALL THROUGH */
5b47454d 3068 case TRIE:
2e64971a
DM
3069 /* the basic plan of execution of the trie is:
3070 * At the beginning, run though all the states, and
3071 * find the longest-matching word. Also remember the position
3072 * of the shortest matching word. For example, this pattern:
3073 * 1 2 3 4 5
3074 * ab|a|x|abcd|abc
3075 * when matched against the string "abcde", will generate
3076 * accept states for all words except 3, with the longest
3077 * matching word being 4, and the shortest being 1 (with
3078 * the position being after char 1 of the string).
3079 *
3080 * Then for each matching word, in word order (i.e. 1,2,4,5),
3081 * we run the remainder of the pattern; on each try setting
3082 * the current position to the character following the word,
3083 * returning to try the next word on failure.
3084 *
3085 * We avoid having to build a list of words at runtime by
3086 * using a compile-time structure, wordinfo[].prev, which
3087 * gives, for each word, the previous accepting word (if any).
3088 * In the case above it would contain the mappings 1->2, 2->0,
3089 * 3->0, 4->5, 5->1. We can use this table to generate, from
3090 * the longest word (4 above), a list of all words, by
3091 * following the list of prev pointers; this gives us the
3092 * unordered list 4,5,1,2. Then given the current word we have
3093 * just tried, we can go through the list and find the
3094 * next-biggest word to try (so if we just failed on word 2,
3095 * the next in the list is 4).
3096 *
3097 * Since at runtime we don't record the matching position in
3098 * the string for each word, we have to work that out for
3099 * each word we're about to process. The wordinfo table holds
3100 * the character length of each word; given that we recorded
3101 * at the start: the position of the shortest word and its
3102 * length in chars, we just need to move the pointer the
3103 * difference between the two char lengths. Depending on
3104 * Unicode status and folding, that's cheap or expensive.
3105 *
3106 * This algorithm is optimised for the case where are only a
3107 * small number of accept states, i.e. 0,1, or maybe 2.
3108 * With lots of accepts states, and having to try all of them,
3109 * it becomes quadratic on number of accept states to find all
3110 * the next words.
3111 */
3112
3dab1dad 3113 {
07be1b83 3114 /* what type of TRIE am I? (utf8 makes this contextual) */
a0a388a1 3115 DECL_TRIE_TYPE(scan);
3dab1dad
YO
3116
3117 /* what trie are we using right now */
be8e71aa 3118 reg_trie_data * const trie
f8fc2ecf 3119 = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
85fbaab2 3120 HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]);
3dab1dad 3121 U32 state = trie->startstate;
166ba7cd 3122
3dab1dad
YO
3123 if (trie->bitmap && trie_type != trie_utf8_fold &&
3124 !TRIE_BITMAP_TEST(trie,*locinput)
3125 ) {
3126 if (trie->states[ state ].wordnum) {
3127 DEBUG_EXECUTE_r(
3128 PerlIO_printf(Perl_debug_log,
3129 "%*s %smatched empty string...%s\n",
5bc10b2c 3130 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3dab1dad
YO
3131 );
3132 break;
3133 } else {
3134 DEBUG_EXECUTE_r(
3135 PerlIO_printf(Perl_debug_log,
786e8c11 3136 "%*s %sfailed to match trie start class...%s\n",
5bc10b2c 3137 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3dab1dad
YO
3138 );
3139 sayNO_SILENT;
3140 }
3141 }
166ba7cd 3142
786e8c11
YO
3143 {
3144 U8 *uc = ( U8* )locinput;
3145
3146 STRLEN len = 0;
3147 STRLEN foldlen = 0;
3148 U8 *uscan = (U8*)NULL;
786e8c11 3149 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2e64971a
DM
3150 U32 charcount = 0; /* how many input chars we have matched */
3151 U32 accepted = 0; /* have we seen any accepting states? */
786e8c11 3152
786e8c11
YO
3153 ST.B = next;
3154 ST.jump = trie->jump;
786e8c11 3155 ST.me = scan;
2e64971a
DM
3156 ST.firstpos = NULL;
3157 ST.longfold = FALSE; /* char longer if folded => it's harder */
3158 ST.nextword = 0;
3159
3160 /* fully traverse the TRIE; note the position of the
3161 shortest accept state and the wordnum of the longest
3162 accept state */
07be1b83 3163
a3621e74 3164 while ( state && uc <= (U8*)PL_regeol ) {
786e8c11 3165 U32 base = trie->states[ state ].trans.base;
f9f4320a 3166 UV uvc = 0;
786e8c11 3167 U16 charid;
2e64971a
DM
3168 U16 wordnum;
3169 wordnum = trie->states[ state ].wordnum;
3170
3171 if (wordnum) { /* it's an accept state */
3172 if (!accepted) {
3173 accepted = 1;
3174 /* record first match position */
3175 if (ST.longfold) {
3176 ST.firstpos = (U8*)locinput;
3177 ST.firstchars = 0;
5b47454d 3178 }
2e64971a
DM
3179 else {
3180 ST.firstpos = uc;
3181 ST.firstchars = charcount;
3182 }
3183 }
3184 if (!ST.nextword || wordnum < ST.nextword)
3185 ST.nextword = wordnum;
3186 ST.topword = wordnum;
786e8c11 3187 }
a3621e74 3188
07be1b83 3189 DEBUG_TRIE_EXECUTE_r({
f2ed9b32 3190 DUMP_EXEC_POS( (char *)uc, scan, utf8_target );
a3621e74 3191 PerlIO_printf( Perl_debug_log,
2e64971a 3192 "%*s %sState: %4"UVxf" Accepted: %c ",
5bc10b2c 3193 2+depth * 2, "", PL_colors[4],
2e64971a 3194 (UV)state, (accepted ? 'Y' : 'N'));
07be1b83 3195 });
a3621e74 3196
2e64971a 3197 /* read a char and goto next state */
a3621e74 3198 if ( base ) {
6dd2be57 3199 I32 offset;
55eed653
NC
3200 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
3201 uscan, len, uvc, charid, foldlen,
3202 foldbuf, uniflags);
2e64971a
DM
3203 charcount++;
3204 if (foldlen>0)
3205 ST.longfold = TRUE;
5b47454d 3206 if (charid &&
6dd2be57
DM
3207 ( ((offset =
3208 base + charid - 1 - trie->uniquecharcount)) >= 0)
3209
3210 && ((U32)offset < trie->lasttrans)
3211 && trie->trans[offset].check == state)
5b47454d 3212 {
6dd2be57 3213 state = trie->trans[offset].next;
5b47454d
DM
3214 }
3215 else {
3216 state = 0;
3217 }
3218 uc += len;
3219
3220 }
3221 else {
a3621e74
YO
3222 state = 0;
3223 }
3224 DEBUG_TRIE_EXECUTE_r(
e4584336 3225 PerlIO_printf( Perl_debug_log,
786e8c11 3226 "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
e4584336 3227 charid, uvc, (UV)state, PL_colors[5] );
a3621e74
YO
3228 );
3229 }
2e64971a 3230 if (!accepted)
a3621e74 3231 sayNO;
a3621e74 3232
2e64971a
DM
3233 /* calculate total number of accept states */
3234 {
3235 U16 w = ST.topword;
3236 accepted = 0;
3237 while (w) {
3238 w = trie->wordinfo[w].prev;
3239 accepted++;
3240 }
3241 ST.accepted = accepted;
3242 }
3243
166ba7cd
DM
3244 DEBUG_EXECUTE_r(
3245 PerlIO_printf( Perl_debug_log,
3246 "%*s %sgot %"IVdf" possible matches%s\n",
5bc10b2c 3247 REPORT_CODE_OFF + depth * 2, "",
166ba7cd
DM
3248 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
3249 );
2e64971a 3250 goto trie_first_try; /* jump into the fail handler */
786e8c11 3251 }}
fae667d5 3252 /* NOTREACHED */
2e64971a
DM
3253
3254 case TRIE_next_fail: /* we failed - try next alternative */
fae667d5
YO
3255 if ( ST.jump) {
3256 REGCP_UNWIND(ST.cp);
3257 for (n = *PL_reglastparen; n > ST.lastparen; n--)
f0ab9afb 3258 PL_regoffs[n].end = -1;
fae667d5
YO
3259 *PL_reglastparen = n;
3260 }
2e64971a
DM
3261 if (!--ST.accepted) {
3262 DEBUG_EXECUTE_r({
3263 PerlIO_printf( Perl_debug_log,
3264 "%*s %sTRIE failed...%s\n",
3265 REPORT_CODE_OFF+depth*2, "",
3266 PL_colors[4],
3267 PL_colors[5] );
3268 });
3269 sayNO_SILENT;
3270 }
3271 {
3272 /* Find next-highest word to process. Note that this code
3273 * is O(N^2) per trie run (O(N) per branch), so keep tight */
d9a396a3
DM
3274 register U16 min = 0;
3275 register U16 word;
2e64971a
DM
3276 register U16 const nextword = ST.nextword;
3277 register reg_trie_wordinfo * const wordinfo
3278 = ((reg_trie_data*)rexi->data->data[ARG(ST.me)])->wordinfo;
3279 for (word=ST.topword; word; word=wordinfo[word].prev) {
3280 if (word > nextword && (!min || word < min))
3281 min = word;
3282 }
3283 ST.nextword = min;
3284 }
3285
fae667d5 3286 trie_first_try:
5d458dd8
YO
3287 if (do_cutgroup) {
3288 do_cutgroup = 0;
3289 no_final = 0;
3290 }
fae667d5
YO
3291
3292 if ( ST.jump) {
3293 ST.lastparen = *PL_reglastparen;
3294 REGCP_SET(ST.cp);
2e64971a 3295 }
a3621e74 3296
2e64971a 3297 /* find start char of end of current word */
166ba7cd 3298 {
2e64971a
DM
3299 U32 chars; /* how many chars to skip */
3300 U8 *uc = ST.firstpos;
3301 reg_trie_data * const trie
3302 = (reg_trie_data*)rexi->data->data[ARG(ST.me)];
3303
3304 assert((trie->wordinfo[ST.nextword].len - trie->prefixlen)
3305 >= ST.firstchars);
3306 chars = (trie->wordinfo[ST.nextword].len - trie->prefixlen)
3307 - ST.firstchars;
3308
3309 if (ST.longfold) {
3310 /* the hard option - fold each char in turn and find
3311 * its folded length (which may be different */
3312 U8 foldbuf[UTF8_MAXBYTES_CASE + 1];
3313 STRLEN foldlen;
3314 STRLEN len;
d9a396a3 3315 UV uvc;
2e64971a
DM
3316 U8 *uscan;
3317
3318 while (chars) {
f2ed9b32 3319 if (utf8_target) {
2e64971a
DM
3320 uvc = utf8n_to_uvuni((U8*)uc, UTF8_MAXLEN, &len,
3321 uniflags);
3322 uc += len;
3323 }
3324 else {
3325 uvc = *uc;
3326 uc++;
3327 }
3328 uvc = to_uni_fold(uvc, foldbuf, &foldlen);
3329 uscan = foldbuf;
3330 while (foldlen) {
3331 if (!--chars)
3332 break;
3333 uvc = utf8n_to_uvuni(uscan, UTF8_MAXLEN, &len,
3334 uniflags);
3335 uscan += len;
3336 foldlen -= len;
3337 }
3338 }
a3621e74 3339 }
2e64971a 3340 else {
f2ed9b32 3341 if (utf8_target)
2e64971a
DM
3342 while (chars--)
3343 uc += UTF8SKIP(uc);
3344 else
3345 uc += chars;
3346 }
3347 PL_reginput = (char *)uc;
3348 }
166ba7cd 3349
2e64971a
DM
3350 scan = (ST.jump && ST.jump[ST.nextword])
3351 ? ST.me + ST.jump[ST.nextword]
3352 : ST.B;
166ba7cd 3353
2e64971a
DM
3354 DEBUG_EXECUTE_r({
3355 PerlIO_printf( Perl_debug_log,
3356 "%*s %sTRIE matched word #%d, continuing%s\n",
3357 REPORT_CODE_OFF+depth*2, "",
3358 PL_colors[4],
3359 ST.nextword,
3360 PL_colors[5]
3361 );
3362 });
3363
3364 if (ST.accepted > 1 || has_cutgroup) {
3365 PUSH_STATE_GOTO(TRIE_next, scan);
3366 /* NOTREACHED */
166ba7cd 3367 }
2e64971a
DM
3368 /* only one choice left - just continue */
3369 DEBUG_EXECUTE_r({
3370 AV *const trie_words
3371 = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
3372 SV ** const tmp = av_fetch( trie_words,
3373 ST.nextword-1, 0 );
3374 SV *sv= tmp ? sv_newmortal() : NULL;
3375
3376 PerlIO_printf( Perl_debug_log,
3377 "%*s %sonly one match left, short-circuiting: #%d <%s>%s\n",
3378 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
3379 ST.nextword,
3380 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
3381 PL_colors[0], PL_colors[1],
3382 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
3383 )
3384 : "not compiled under -Dr",
3385 PL_colors[5] );
3386 });
3387
3388 locinput = PL_reginput;
3389 nextchr = UCHARAT(locinput);
3390 continue; /* execute rest of RE */
166ba7cd 3391 /* NOTREACHED */
166ba7cd
DM
3392#undef ST
3393
95b24440
DM
3394 case EXACT: {
3395 char *s = STRING(scan);
24d3c4a9 3396 ln = STR_LEN(scan);
f2ed9b32 3397 if (utf8_target != UTF_PATTERN) {
bc517b45 3398 /* The target and the pattern have differing utf8ness. */
1aa99e6b 3399 char *l = locinput;
24d3c4a9 3400 const char * const e = s + ln;
a72c7584 3401
f2ed9b32 3402 if (utf8_target) {
5ff6fc6d 3403 /* The target is utf8, the pattern is not utf8. */
1aa99e6b 3404 while (s < e) {
a3b680e6 3405 STRLEN ulen;
1aa99e6b 3406 if (l >= PL_regeol)
5ff6fc6d
JH
3407 sayNO;
3408 if (NATIVE_TO_UNI(*(U8*)s) !=
89ebb4a3 3409 utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
041457d9 3410 uniflags))
5ff6fc6d 3411 sayNO;
bc517b45 3412 l += ulen;
5ff6fc6d 3413 s ++;
1aa99e6b 3414 }
5ff6fc6d
JH
3415 }
3416 else {
3417 /* The target is not utf8, the pattern is utf8. */
1aa99e6b 3418 while (s < e) {
a3b680e6 3419 STRLEN ulen;
1aa99e6b
IH
3420 if (l >= PL_regeol)
3421 sayNO;
5ff6fc6d 3422 if (NATIVE_TO_UNI(*((U8*)l)) !=
89ebb4a3 3423 utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
041457d9 3424 uniflags))
1aa99e6b 3425 sayNO;
bc517b45 3426 s += ulen;
a72c7584 3427 l ++;
1aa99e6b 3428 }
5ff6fc6d 3429 }
1aa99e6b
IH
3430 locinput = l;
3431 nextchr = UCHARAT(locinput);
3432 break;
3433 }
bc517b45 3434 /* The target and the pattern have the same utf8ness. */
d6a28714
JH
3435 /* Inline the first character, for speed. */
3436 if (UCHARAT(s) != nextchr)
3437 sayNO;
24d3c4a9 3438 if (PL_regeol - locinput < ln)
d6a28714 3439 sayNO;
24d3c4a9 3440 if (ln > 1 && memNE(s, locinput, ln))
d6a28714 3441 sayNO;
24d3c4a9 3442 locinput += ln;
d6a28714
JH
3443 nextchr = UCHARAT(locinput);
3444 break;
95b24440 3445 }
d6a28714 3446 case EXACTFL:
b8c5462f
JH
3447 PL_reg_flags |= RF_tainted;
3448 /* FALL THROUGH */
95b24440 3449 case EXACTF: {
be8e71aa 3450 char * const s = STRING(scan);
24d3c4a9 3451 ln = STR_LEN(scan);
d6a28714 3452
f2ed9b32 3453 if (utf8_target || UTF_PATTERN) {
d07ddd77 3454 /* Either target or the pattern are utf8. */
be8e71aa 3455 const char * const l = locinput;
d07ddd77 3456 char *e = PL_regeol;
bc517b45 3457
f2ed9b32
KW
3458 if (! foldEQ_utf8(s, 0, ln, cBOOL(UTF_PATTERN),
3459 l, &e, 0, utf8_target)) {
5486206c
JH
3460 /* One more case for the sharp s:
3461 * pack("U0U*", 0xDF) =~ /ss/i,
3462 * the 0xC3 0x9F are the UTF-8
3463 * byte sequence for the U+00DF. */
e1d1eefb 3464
f2ed9b32 3465 if (!(utf8_target &&
e1d1eefb 3466 toLOWER(s[0]) == 's' &&
24d3c4a9 3467 ln >= 2 &&
5486206c
JH
3468 toLOWER(s[1]) == 's' &&
3469 (U8)l[0] == 0xC3 &&
3470 e - l >= 2 &&
3471 (U8)l[1] == 0x9F))
3472 sayNO;
3473 }
d07ddd77
JH
3474 locinput = e;
3475 nextchr = UCHARAT(locinput);
3476 break;
a0ed51b3 3477 }
d6a28714 3478
bc517b45
JH
3479 /* Neither the target and the pattern are utf8. */
3480
d6a28714
JH
3481 /* Inline the first character, for speed. */
3482 if (UCHARAT(s) != nextchr &&
3483 UCHARAT(s) != ((OP(scan) == EXACTF)
3484 ? PL_fold : PL_fold_locale)[nextchr])
a0ed51b3 3485 sayNO;
24d3c4a9 3486 if (PL_regeol - locinput < ln)
b8c5462f 3487 sayNO;
24d3c4a9 3488 if (ln > 1 && (OP(scan) == EXACTF
4c1b470c
KW
3489 ? ! foldEQ(s, locinput, ln)
3490 : ! foldEQ_locale(s, locinput, ln)))
4633a7c4 3491 sayNO;
24d3c4a9 3492 locinput += ln;
d6a28714 3493 nextchr = UCHARAT(locinput);
a0d0e21e 3494 break;
95b24440 3495 }
b2680017
YO
3496 case BOUNDL:
3497 case NBOUNDL:
3498 PL_reg_flags |= RF_tainted;
3499 /* FALL THROUGH */
3500 case BOUND:
3501 case NBOUND:
3502 /* was last char in word? */
f2ed9b32 3503 if (utf8_target) {
b2680017
YO
3504 if (locinput == PL_bostr)
3505 ln = '\n';
3506 else {
3507 const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
3508
3509 ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags);
3510 }
3511 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
3512 ln = isALNUM_uni(ln);
3513 LOAD_UTF8_CHARCLASS_ALNUM();
f2ed9b32 3514 n = swash_fetch(PL_utf8_alnum, (U8*)locinput, utf8_target);
b2680017
YO
3515 }
3516 else {
3517 ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln));
3518 n = isALNUM_LC_utf8((U8*)locinput);
3519 }
3520 }
3521 else {
3522 ln = (locinput != PL_bostr) ?
3523 UCHARAT(locinput - 1) : '\n';
3524 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
3525 ln = isALNUM(ln);
3526 n = isALNUM(nextchr);
3527 }
3528 else {
3529 ln = isALNUM_LC(ln);
3530 n = isALNUM_LC(nextchr);
3531 }
3532 }
3533 if (((!ln) == (!n)) == (OP(scan) == BOUND ||
3534 OP(scan) == BOUNDL))
3535 sayNO;
3536 break;
d6a28714 3537 case ANYOF:
f2ed9b32 3538 if (utf8_target) {
9e55ce06
JH
3539 STRLEN inclasslen = PL_regeol - locinput;
3540
f2ed9b32 3541 if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, utf8_target))
262b90c4 3542 goto anyof_fail;
ffc61ed2
JH
3543 if (locinput >= PL_regeol)
3544 sayNO;
0f0076b4 3545 locinput += inclasslen ? inclasslen : UTF8SKIP(locinput);
b8c5462f 3546 nextchr = UCHARAT(locinput);
e0f9d4a8 3547 break;
ffc61ed2
JH
3548 }
3549 else {
3550 if (nextchr < 0)
3551 nextchr = UCHARAT(locinput);
32fc9b6a 3552 if (!REGINCLASS(rex, scan, (U8*)locinput))
262b90c4 3553 goto anyof_fail;
ffc61ed2
JH
3554 if (!nextchr && locinput >= PL_regeol)
3555 sayNO;
3556 nextchr = UCHARAT(++locinput);
e0f9d4a8
JH
3557 break;
3558 }
262b90c4 3559 anyof_fail:
e0f9d4a8
JH
3560 /* If we might have the case of the German sharp s
3561 * in a casefolding Unicode character class. */
3562
ebc501f0
JH
3563 if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
3564 locinput += SHARP_S_SKIP;
e0f9d4a8 3565 nextchr = UCHARAT(locinput);
ffc61ed2 3566 }
e0f9d4a8
JH
3567 else
3568 sayNO;
b8c5462f 3569 break;
20d0b1e9 3570 /* Special char classes - The defines start on line 129 or so */
d1eb3177
YO
3571 CCC_TRY_AFF( ALNUM, ALNUML, perl_word, "a", isALNUM_LC_utf8, isALNUM, isALNUM_LC);
3572 CCC_TRY_NEG(NALNUM, NALNUML, perl_word, "a", isALNUM_LC_utf8, isALNUM, isALNUM_LC);
20d0b1e9 3573
d1eb3177
YO
3574 CCC_TRY_AFF( SPACE, SPACEL, perl_space, " ", isSPACE_LC_utf8, isSPACE, isSPACE_LC);
3575 CCC_TRY_NEG(NSPACE, NSPACEL, perl_space, " ", isSPACE_LC_utf8, isSPACE, isSPACE_LC);
20d0b1e9 3576
d1eb3177
YO
3577 CCC_TRY_AFF( DIGIT, DIGITL, posix_digit, "0", isDIGIT_LC_utf8, isDIGIT, isDIGIT_LC);
3578 CCC_TRY_NEG(NDIGIT, NDIGITL, posix_digit, "0", isDIGIT_LC_utf8, isDIGIT, isDIGIT_LC);
20d0b1e9 3579
37e2e78e
KW
3580 case CLUMP: /* Match \X: logical Unicode character. This is defined as
3581 a Unicode extended Grapheme Cluster */
3582 /* From http://www.unicode.org/reports/tr29 (5.2 version). An
3583 extended Grapheme Cluster is:
3584
3585 CR LF
3586 | Prepend* Begin Extend*
3587 | .
3588
3589 Begin is (Hangul-syllable | ! Control)
3590 Extend is (Grapheme_Extend | Spacing_Mark)
3591 Control is [ GCB_Control CR LF ]
3592
3593 The discussion below shows how the code for CLUMP is derived
3594 from this regex. Note that most of these concepts are from
3595 property values of the Grapheme Cluster Boundary (GCB) property.
3596 No code point can have multiple property values for a given
3597 property. Thus a code point in Prepend can't be in Control, but
3598 it must be in !Control. This is why Control above includes
3599 GCB_Control plus CR plus LF. The latter two are used in the GCB
3600 property separately, and so can't be in GCB_Control, even though
3601 they logically are controls. Control is not the same as gc=cc,
3602 but includes format and other characters as well.
3603
3604 The Unicode definition of Hangul-syllable is:
3605 L+
3606 | (L* ( ( V | LV ) V* | LVT ) T*)
3607 | T+
3608 )
3609 Each of these is a value for the GCB property, and hence must be
3610 disjoint, so the order they are tested is immaterial, so the
3611 above can safely be changed to
3612 T+
3613 | L+
3614 | (L* ( LVT | ( V | LV ) V*) T*)
3615
3616 The last two terms can be combined like this:
3617 L* ( L
3618 | (( LVT | ( V | LV ) V*) T*))
3619
3620 And refactored into this:
3621 L* (L | LVT T* | V V* T* | LV V* T*)
3622
3623 That means that if we have seen any L's at all we can quit
3624 there, but if the next character is a LVT, a V or and LV we
3625 should keep going.
3626
3627 There is a subtlety with Prepend* which showed up in testing.
3628 Note that the Begin, and only the Begin is required in:
3629 | Prepend* Begin Extend*
3630 Also, Begin contains '! Control'. A Prepend must be a '!
3631 Control', which means it must be a Begin. What it comes down to
3632 is that if we match Prepend* and then find no suitable Begin
3633 afterwards, that if we backtrack the last Prepend, that one will
3634 be a suitable Begin.
3635 */
3636
b7c83a7e 3637 if (locinput >= PL_regeol)
a0ed51b3 3638 sayNO;
f2ed9b32 3639 if (! utf8_target) {
37e2e78e
KW
3640
3641 /* Match either CR LF or '.', as all the other possibilities
3642 * require utf8 */
3643 locinput++; /* Match the . or CR */
3644 if (nextchr == '\r'
3645 && locinput < PL_regeol
3646 && UCHARAT(locinput) == '\n') locinput++;
3647 }
3648 else {
3649
3650 /* Utf8: See if is ( CR LF ); already know that locinput <
3651 * PL_regeol, so locinput+1 is in bounds */
3652 if (nextchr == '\r' && UCHARAT(locinput + 1) == '\n') {
3653 locinput += 2;
3654 }
3655 else {
3656 /* In case have to backtrack to beginning, then match '.' */
3657 char *starting = locinput;
3658
3659 /* In case have to backtrack the last prepend */
3660 char *previous_prepend = 0;
3661
3662 LOAD_UTF8_CHARCLASS_GCB();
3663
3664 /* Match (prepend)* */
3665 while (locinput < PL_regeol
3666 && swash_fetch(PL_utf8_X_prepend,
f2ed9b32 3667 (U8*)locinput, utf8_target))
37e2e78e
KW
3668 {
3669 previous_prepend = locinput;
3670 locinput += UTF8SKIP(locinput);
3671 }
3672
3673 /* As noted above, if we matched a prepend character, but
3674 * the next thing won't match, back off the last prepend we
3675 * matched, as it is guaranteed to match the begin */
3676 if (previous_prepend
3677 && (locinput >= PL_regeol
3678 || ! swash_fetch(PL_utf8_X_begin,
f2ed9b32 3679 (U8*)locinput, utf8_target)))
37e2e78e
KW
3680 {
3681 locinput = previous_prepend;
3682 }
3683
3684 /* Note that here we know PL_regeol > locinput, as we
3685 * tested that upon input to this switch case, and if we
3686 * moved locinput forward, we tested the result just above
3687 * and it either passed, or we backed off so that it will
3688 * now pass */
f2ed9b32 3689 if (! swash_fetch(PL_utf8_X_begin, (U8*)locinput, utf8_target)) {
37e2e78e
KW
3690
3691 /* Here did not match the required 'Begin' in the
3692 * second term. So just match the very first
3693 * character, the '.' of the final term of the regex */
3694 locinput = starting + UTF8SKIP(starting);
3695 } else {
3696
3697 /* Here is the beginning of a character that can have
3698 * an extender. It is either a hangul syllable, or a
3699 * non-control */
3700 if (swash_fetch(PL_utf8_X_non_hangul,
f2ed9b32 3701 (U8*)locinput, utf8_target))
37e2e78e
KW
3702 {
3703
3704 /* Here not a Hangul syllable, must be a
3705 * ('! * Control') */
3706 locinput += UTF8SKIP(locinput);
3707 } else {
3708
3709 /* Here is a Hangul syllable. It can be composed
3710 * of several individual characters. One
3711 * possibility is T+ */
3712 if (swash_fetch(PL_utf8_X_T,
f2ed9b32 3713 (U8*)locinput, utf8_target))
37e2e78e
KW
3714 {
3715 while (locinput < PL_regeol
3716 && swash_fetch(PL_utf8_X_T,
f2ed9b32 3717 (U8*)locinput, utf8_target))
37e2e78e
KW
3718 {
3719 locinput += UTF8SKIP(locinput);
3720 }
3721 } else {
3722
3723 /* Here, not T+, but is a Hangul. That means
3724 * it is one of the others: L, LV, LVT or V,
3725 * and matches:
3726 * L* (L | LVT T* | V V* T* | LV V* T*) */
3727
3728 /* Match L* */
3729 while (locinput < PL_regeol
3730 && swash_fetch(PL_utf8_X_L,
f2ed9b32 3731 (U8*)locinput, utf8_target))
37e2e78e
KW
3732 {
3733 locinput += UTF8SKIP(locinput);
3734 }
3735
3736 /* Here, have exhausted L*. If the next
3737 * character is not an LV, LVT nor V, it means
3738 * we had to have at least one L, so matches L+
3739 * in the original equation, we have a complete
3740 * hangul syllable. Are done. */
3741
3742 if (locinput < PL_regeol
3743 && swash_fetch(PL_utf8_X_LV_LVT_V,
f2ed9b32 3744 (U8*)locinput, utf8_target))
37e2e78e
KW
3745 {
3746
3747 /* Otherwise keep going. Must be LV, LVT
3748 * or V. See if LVT */
3749 if (swash_fetch(PL_utf8_X_LVT,
f2ed9b32 3750 (U8*)locinput, utf8_target))
37e2e78e
KW
3751 {
3752 locinput += UTF8SKIP(locinput);
3753 } else {
3754
3755 /* Must be V or LV. Take it, then
3756 * match V* */
3757 locinput += UTF8SKIP(locinput);
3758 while (locinput < PL_regeol
3759 && swash_fetch(PL_utf8_X_V,
f2ed9b32 3760 (U8*)locinput, utf8_target))
37e2e78e
KW
3761 {
3762 locinput += UTF8SKIP(locinput);
3763 }
3764 }
3765
3766 /* And any of LV, LVT, or V can be followed
3767 * by T* */
3768 while (locinput < PL_regeol
3769 && swash_fetch(PL_utf8_X_T,
3770 (U8*)locinput,
f2ed9b32 3771 utf8_target))
37e2e78e
KW
3772 {
3773 locinput += UTF8SKIP(locinput);
3774 }
3775 }
3776 }
3777 }
3778
3779 /* Match any extender */
3780 while (locinput < PL_regeol
3781 && swash_fetch(PL_utf8_X_extend,
f2ed9b32 3782 (U8*)locinput, utf8_target))
37e2e78e
KW
3783 {
3784 locinput += UTF8SKIP(locinput);
3785 }
3786 }
3787 }
3788 if (locinput > PL_regeol) sayNO;
3789 }
a0ed51b3
LW
3790 nextchr = UCHARAT(locinput);
3791 break;
81714fb9
YO
3792
3793 case NREFFL:
3794 {
3795 char *s;
ff1157ca 3796 char type;
81714fb9
YO
3797 PL_reg_flags |= RF_tainted;
3798 /* FALL THROUGH */
3799 case NREF:
3800 case NREFF:
ff1157ca 3801 type = OP(scan);
0a4db386
YO
3802 n = reg_check_named_buff_matched(rex,scan);
3803
3804 if ( n ) {
3805 type = REF + ( type - NREF );
3806 goto do_ref;
3807 } else {
81714fb9 3808 sayNO;
0a4db386
YO
3809 }
3810 /* unreached */
c8756f30 3811 case REFFL:
3280af22 3812 PL_reg_flags |= RF_tainted;
c8756f30 3813 /* FALL THROUGH */
c277df42 3814 case REF:
81714fb9 3815 case REFF:
c277df42 3816 n = ARG(scan); /* which paren pair */
81714fb9
YO
3817 type = OP(scan);
3818 do_ref:
f0ab9afb 3819 ln = PL_regoffs[n].start;
2c2d71f5 3820 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
3b6647e0 3821 if (*PL_reglastparen < n || ln == -1)
af3f8c16 3822 sayNO; /* Do not match unless seen CLOSEn. */
f0ab9afb 3823 if (ln == PL_regoffs[n].end)
a0d0e21e 3824 break;
a0ed51b3 3825
24d3c4a9 3826 s = PL_bostr + ln;
f2ed9b32 3827 if (utf8_target && type != REF) { /* REF can do byte comparison */
a0ed51b3 3828 char *l = locinput;
f0ab9afb 3829 const char *e = PL_bostr + PL_regoffs[n].end;
a0ed51b3
LW
3830 /*
3831 * Note that we can't do the "other character" lookup trick as
3832 * in the 8-bit case (no pun intended) because in Unicode we
3833 * have to map both upper and title case to lower case.
3834 */
81714fb9 3835 if (type == REFF) {
a0ed51b3 3836 while (s < e) {
a3b680e6
AL
3837 STRLEN ulen1, ulen2;
3838 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
3839 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
3840
a0ed51b3
LW
3841 if (l >= PL_regeol)
3842 sayNO;
a2a2844f
JH
3843 toLOWER_utf8((U8*)s, tmpbuf1, &ulen1);
3844 toLOWER_utf8((U8*)l, tmpbuf2, &ulen2);
7114a2d2 3845 if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
a0ed51b3 3846 sayNO;
a2a2844f
JH
3847 s += ulen1;
3848 l += ulen2;
a0ed51b3
LW
3849 }
3850 }
3851 locinput = l;
3852 nextchr = UCHARAT(locinput);
3853 break;
3854 }
3855
a0d0e21e 3856 /* Inline the first character, for speed. */
76e3520e 3857 if (UCHARAT(s) != nextchr &&
81714fb9
YO
3858 (type == REF ||
3859 (UCHARAT(s) != (type == REFF
3860 ? PL_fold : PL_fold_locale)[nextchr])))
4633a7c4 3861 sayNO;
f0ab9afb 3862 ln = PL_regoffs[n].end - ln;
24d3c4a9 3863 if (locinput + ln > PL_regeol)
4633a7c4 3864 sayNO;
81714fb9 3865 if (ln > 1 && (type == REF
24d3c4a9 3866 ? memNE(s, locinput, ln)
81714fb9 3867 : (type == REFF
4c1b470c
KW
3868 ? ! foldEQ(s, locinput, ln)
3869 : ! foldEQ_locale(s, locinput, ln))))
4633a7c4 3870 sayNO;
24d3c4a9 3871 locinput += ln;
76e3520e 3872 nextchr = UCHARAT(locinput);
a0d0e21e 3873 break;
81714fb9 3874 }
a0d0e21e 3875 case NOTHING:
c277df42 3876 case TAIL:
a0d0e21e
LW
3877 break;
3878 case BACK:
3879 break;
40a82448
DM
3880
3881#undef ST
3882#define ST st->u.eval
c277df42 3883 {
c277df42 3884 SV *ret;
d2f13c59 3885 REGEXP *re_sv;
6bda09f9 3886 regexp *re;
f8fc2ecf 3887 regexp_internal *rei;
1a147d38
YO
3888 regnode *startpoint;
3889
3890 case GOSTART:
e7707071
YO
3891 case GOSUB: /* /(...(?1))/ /(...(?&foo))/ */
3892 if (cur_eval && cur_eval->locinput==locinput) {
24b23f37 3893 if (cur_eval->u.eval.close_paren == (U32)ARG(scan))
1a147d38 3894 Perl_croak(aTHX_ "Infinite recursion in regex");
4b196cd4 3895 if ( ++nochange_depth > max_nochange_depth )
1a147d38
YO
3896 Perl_croak(aTHX_
3897 "Pattern subroutine nesting without pos change"
3898 " exceeded limit in regex");
6bda09f9
YO
3899 } else {
3900 nochange_depth = 0;
1a147d38 3901 }
288b8c02 3902 re_sv = rex_sv;
6bda09f9 3903 re = rex;
f8fc2ecf 3904 rei = rexi;
288b8c02 3905 (void)ReREFCNT_inc(rex_sv);
1a147d38 3906 if (OP(scan)==GOSUB) {
6bda09f9
YO
3907 startpoint = scan + ARG2L(scan);
3908 ST.close_paren = ARG(scan);
3909 } else {
f8fc2ecf 3910 startpoint = rei->program+1;
6bda09f9
YO
3911 ST.close_paren = 0;
3912 }
3913 goto eval_recurse_doit;
3914 /* NOTREACHED */
3915 case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */
3916 if (cur_eval && cur_eval->locinput==locinput) {
4b196cd4 3917 if ( ++nochange_depth > max_nochange_depth )
1a147d38 3918 Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex");
6bda09f9
YO
3919 } else {
3920 nochange_depth = 0;
3921 }
8e5e9ebe 3922 {
4aabdb9b
DM
3923 /* execute the code in the {...} */
3924 dSP;
6136c704 3925 SV ** const before = SP;
4aabdb9b
DM
3926 OP_4tree * const oop = PL_op;
3927 COP * const ocurcop = PL_curcop;
3928 PAD *old_comppad;
d80618d2 3929 char *saved_regeol = PL_regeol;
4aabdb9b
DM
3930
3931 n = ARG(scan);
f8fc2ecf 3932 PL_op = (OP_4tree*)rexi->data->data[n];
24b23f37
YO
3933 DEBUG_STATE_r( PerlIO_printf(Perl_debug_log,
3934 " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
f8fc2ecf 3935 PAD_SAVE_LOCAL(old_comppad, (PAD*)rexi->data->data[n + 2]);
f0ab9afb 3936 PL_regoffs[0].end = PL_reg_magic->mg_len = locinput - PL_bostr;
4aabdb9b 3937
2bf803e2
YO
3938 if (sv_yes_mark) {
3939 SV *sv_mrk = get_sv("REGMARK", 1);
3940 sv_setsv(sv_mrk, sv_yes_mark);
3941 }
3942
8e5e9ebe
RGS
3943 CALLRUNOPS(aTHX); /* Scalar context. */
3944 SPAGAIN;
3945 if (SP == before)
075aa684 3946 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
8e5e9ebe
RGS
3947 else {
3948 ret = POPs;
3949 PUTBACK;
3950 }
4aabdb9b
DM
3951
3952 PL_op = oop;
3953 PAD_RESTORE_LOCAL(old_comppad);
3954 PL_curcop = ocurcop;
d80618d2 3955 PL_regeol = saved_regeol;
24d3c4a9 3956 if (!logical) {
4aabdb9b
DM
3957 /* /(?{...})/ */
3958 sv_setsv(save_scalar(PL_replgv), ret);
4aabdb9b
DM
3959 break;
3960 }
8e5e9ebe 3961 }
24d3c4a9
DM
3962 if (logical == 2) { /* Postponed subexpression: /(??{...})/ */
3963 logical = 0;
4aabdb9b 3964 {
4f639d21
DM
3965 /* extract RE object from returned value; compiling if
3966 * necessary */
6136c704 3967 MAGIC *mg = NULL;
288b8c02 3968 REGEXP *rx = NULL;
5c35adbb
NC
3969
3970 if (SvROK(ret)) {
288b8c02 3971 SV *const sv = SvRV(ret);
5c35adbb
NC
3972
3973 if (SvTYPE(sv) == SVt_REGEXP) {
d2f13c59 3974 rx = (REGEXP*) sv;
5c35adbb
NC
3975 } else if (SvSMAGICAL(sv)) {
3976 mg = mg_find(sv, PERL_MAGIC_qr);
3977 assert(mg);
3978 }
3979 } else if (SvTYPE(ret) == SVt_REGEXP) {
d2f13c59 3980 rx = (REGEXP*) ret;
5c35adbb 3981 } else if (SvSMAGICAL(ret)) {
124ee91a
NC
3982 if (SvGMAGICAL(ret)) {
3983 /* I don't believe that there is ever qr magic
3984 here. */
3985 assert(!mg_find(ret, PERL_MAGIC_qr));
faf82a0b 3986 sv_unmagic(ret, PERL_MAGIC_qr);
124ee91a
NC
3987 }
3988 else {
faf82a0b 3989 mg = mg_find(ret, PERL_MAGIC_qr);
124ee91a
NC
3990 /* testing suggests mg only ends up non-NULL for
3991 scalars who were upgraded and compiled in the
3992 else block below. In turn, this is only
3993 triggered in the "postponed utf8 string" tests
3994 in t/op/pat.t */
3995 }
0f5d15d6 3996 }
faf82a0b 3997
0f5d15d6 3998 if (mg) {
d2f13c59 3999 rx = (REGEXP *) mg->mg_obj; /*XXX:dmq*/
e2560c33 4000 assert(rx);
0f5d15d6 4001 }
288b8c02 4002 if (rx) {
f0826785 4003 rx = reg_temp_copy(NULL, rx);
288b8c02 4004 }
0f5d15d6 4005 else {
c737faaf 4006 U32 pm_flags = 0;
a3b680e6 4007 const I32 osize = PL_regsize;
0f5d15d6 4008
b9ad30b4
NC
4009 if (DO_UTF8(ret)) {
4010 assert (SvUTF8(ret));
4011 } else if (SvUTF8(ret)) {
4012 /* Not doing UTF-8, despite what the SV says. Is
4013 this only if we're trapped in use 'bytes'? */
4014 /* Make a copy of the octet sequence, but without
4015 the flag on, as the compiler now honours the
4016 SvUTF8 flag on ret. */
4017 STRLEN len;
4018 const char *const p = SvPV(ret, len);
4019 ret = newSVpvn_flags(p, len, SVs_TEMP);
4020 }
288b8c02 4021 rx = CALLREGCOMP(ret, pm_flags);
9041c2e3 4022 if (!(SvFLAGS(ret)
faf82a0b 4023 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
3ce3ed55 4024 | SVs_GMG))) {
a2794585
NC
4025 /* This isn't a first class regexp. Instead, it's
4026 caching a regexp onto an existing, Perl visible
4027 scalar. */
ad64d0ec 4028 sv_magic(ret, MUTABLE_SV(rx), PERL_MAGIC_qr, 0, 0);
3ce3ed55 4029 }
0f5d15d6 4030 PL_regsize = osize;
0f5d15d6 4031 }
288b8c02
NC
4032 re_sv = rx;
4033 re = (struct regexp *)SvANY(rx);
4aabdb9b 4034 }
07bc277f 4035 RXp_MATCH_COPIED_off(re);
28d8d7f4
YO
4036 re->subbeg = rex->subbeg;
4037 re->sublen = rex->sublen;
f8fc2ecf 4038 rei = RXi_GET(re);
6bda09f9 4039 DEBUG_EXECUTE_r(
f2ed9b32 4040 debug_start_match(re_sv, utf8_target, locinput, PL_regeol,
6bda09f9
YO
4041 "Matching embedded");
4042 );
f8fc2ecf 4043 startpoint = rei->program + 1;
1a147d38 4044 ST.close_paren = 0; /* only used for GOSUB */
6bda09f9
YO
4045 /* borrowed from regtry */
4046 if (PL_reg_start_tmpl <= re->nparens) {
4047 PL_reg_start_tmpl = re->nparens*3/2 + 3;
4048 if(PL_reg_start_tmp)
4049 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
4050 else
4051 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
dd5def09 4052 }
aa283a38 4053
1a147d38 4054 eval_recurse_doit: /* Share code with GOSUB below this line */
aa283a38 4055 /* run the pattern returned from (??{...}) */
40a82448
DM
4056 ST.cp = regcppush(0); /* Save *all* the positions. */
4057 REGCP_SET(ST.lastcp);
6bda09f9 4058
f0ab9afb 4059 PL_regoffs = re->offs; /* essentially NOOP on GOSUB */
6bda09f9 4060
0357f1fd
ML
4061 /* see regtry, specifically PL_reglast(?:close)?paren is a pointer! (i dont know why) :dmq */
4062 PL_reglastparen = &re->lastparen;
4063 PL_reglastcloseparen = &re->lastcloseparen;
4064 re->lastparen = 0;
4065 re->lastcloseparen = 0;
4066
4aabdb9b 4067 PL_reginput = locinput;
ae0beba1 4068 PL_regsize = 0;
4aabdb9b
DM
4069
4070 /* XXXX This is too dramatic a measure... */
4071 PL_reg_maxiter = 0;
4072
faec1544 4073 ST.toggle_reg_flags = PL_reg_flags;
3c8556c3 4074 if (RX_UTF8(re_sv))
faec1544
DM
4075 PL_reg_flags |= RF_utf8;
4076 else
4077 PL_reg_flags &= ~RF_utf8;
4078 ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */
4079
288b8c02 4080 ST.prev_rex = rex_sv;
faec1544 4081 ST.prev_curlyx = cur_curlyx;
288b8c02
NC
4082 SETREX(rex_sv,re_sv);
4083 rex = re;
f8fc2ecf 4084 rexi = rei;
faec1544 4085 cur_curlyx = NULL;
40a82448 4086 ST.B = next;
faec1544
DM
4087 ST.prev_eval = cur_eval;
4088 cur_eval = st;
faec1544 4089 /* now continue from first node in postoned RE */
6bda09f9 4090 PUSH_YES_STATE_GOTO(EVAL_AB, startpoint);
4aabdb9b 4091 /* NOTREACHED */
a0ed51b3 4092 }
24d3c4a9 4093 /* logical is 1, /(?(?{...})X|Y)/ */
f2338a2e 4094 sw = cBOOL(SvTRUE(ret));
24d3c4a9 4095 logical = 0;
c277df42
IZ
4096 break;
4097 }
40a82448 4098
faec1544
DM
4099 case EVAL_AB: /* cleanup after a successful (??{A})B */
4100 /* note: this is called twice; first after popping B, then A */
4101 PL_reg_flags ^= ST.toggle_reg_flags;
288b8c02
NC
4102 ReREFCNT_dec(rex_sv);
4103 SETREX(rex_sv,ST.prev_rex);
4104 rex = (struct regexp *)SvANY(rex_sv);
f8fc2ecf 4105 rexi = RXi_GET(rex);
faec1544
DM
4106 regcpblow(ST.cp);
4107 cur_eval = ST.prev_eval;
4108 cur_curlyx = ST.prev_curlyx;
34a81e2b
B
4109
4110 /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
0357f1fd
ML
4111 PL_reglastparen = &rex->lastparen;
4112 PL_reglastcloseparen = &rex->lastcloseparen;
34a81e2b
B
4113 /* also update PL_regoffs */
4114 PL_regoffs = rex->offs;
0357f1fd 4115
40a82448
DM
4116 /* XXXX This is too dramatic a measure... */
4117 PL_reg_maxiter = 0;
e7707071 4118 if ( nochange_depth )
4b196cd4 4119 nochange_depth--;
262b90c4 4120 sayYES;
40a82448 4121
40a82448 4122
faec1544
DM
4123 case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
4124 /* note: this is called twice; first after popping B, then A */
4125 PL_reg_flags ^= ST.toggle_reg_flags;
288b8c02
NC
4126 ReREFCNT_dec(rex_sv);
4127 SETREX(rex_sv,ST.prev_rex);
4128 rex = (struct regexp *)SvANY(rex_sv);
f8fc2ecf 4129 rexi = RXi_GET(rex);
34a81e2b 4130 /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
0357f1fd
ML
4131 PL_reglastparen = &rex->lastparen;
4132 PL_reglastcloseparen = &rex->lastcloseparen;
4133
40a82448
DM
4134 PL_reginput = locinput;
4135 REGCP_UNWIND(ST.lastcp);
4136 regcppop(rex);
faec1544
DM
4137 cur_eval = ST.prev_eval;
4138 cur_curlyx = ST.prev_curlyx;
4139 /* XXXX This is too dramatic a measure... */
4140 PL_reg_maxiter = 0;
e7707071 4141 if ( nochange_depth )
4b196cd4 4142 nochange_depth--;
40a82448 4143 sayNO_SILENT;
40a82448
DM
4144#undef ST
4145
a0d0e21e 4146 case OPEN:
c277df42 4147 n = ARG(scan); /* which paren pair */
3280af22
NIS
4148 PL_reg_start_tmp[n] = locinput;
4149 if (n > PL_regsize)
4150 PL_regsize = n;
e2e6a0f1 4151 lastopen = n;
a0d0e21e
LW
4152 break;
4153 case CLOSE:
c277df42 4154 n = ARG(scan); /* which paren pair */
f0ab9afb
NC
4155 PL_regoffs[n].start = PL_reg_start_tmp[n] - PL_bostr;
4156 PL_regoffs[n].end = locinput - PL_bostr;
7f69552c
YO
4157 /*if (n > PL_regsize)
4158 PL_regsize = n;*/
3b6647e0 4159 if (n > *PL_reglastparen)
3280af22 4160 *PL_reglastparen = n;
a01268b5 4161 *PL_reglastcloseparen = n;
3b6647e0 4162 if (cur_eval && cur_eval->u.eval.close_paren == n) {
6bda09f9
YO
4163 goto fake_end;
4164 }
a0d0e21e 4165 break;
e2e6a0f1
YO
4166 case ACCEPT:
4167 if (ARG(scan)){
4168 regnode *cursor;
4169 for (cursor=scan;
4170 cursor && OP(cursor)!=END;
4171 cursor=regnext(cursor))
4172 {
4173 if ( OP(cursor)==CLOSE ){
4174 n = ARG(cursor);
4175 if ( n <= lastopen ) {
f0ab9afb
NC
4176 PL_regoffs[n].start
4177 = PL_reg_start_tmp[n] - PL_bostr;
4178 PL_regoffs[n].end = locinput - PL_bostr;
e2e6a0f1
YO
4179 /*if (n > PL_regsize)
4180 PL_regsize = n;*/
3b6647e0 4181 if (n > *PL_reglastparen)
e2e6a0f1
YO
4182 *PL_reglastparen = n;
4183 *PL_reglastcloseparen = n;
3b6647e0
RB
4184 if ( n == ARG(scan) || (cur_eval &&
4185 cur_eval->u.eval.close_paren == n))
e2e6a0f1
YO
4186 break;
4187 }
4188 }
4189 }
4190 }
4191 goto fake_end;
4192 /*NOTREACHED*/
c277df42
IZ
4193 case GROUPP:
4194 n = ARG(scan); /* which paren pair */
f2338a2e 4195 sw = cBOOL(*PL_reglastparen >= n && PL_regoffs[n].end != -1);
c277df42 4196 break;
0a4db386
YO
4197 case NGROUPP:
4198 /* reg_check_named_buff_matched returns 0 for no match */
f2338a2e 4199 sw = cBOOL(0 < reg_check_named_buff_matched(rex,scan));
0a4db386 4200 break;
1a147d38 4201 case INSUBP:
0a4db386 4202 n = ARG(scan);
3b6647e0 4203 sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n));
0a4db386
YO
4204 break;
4205 case DEFINEP:
4206 sw = 0;
4207 break;
c277df42 4208 case IFTHEN:
2c2d71f5 4209 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
24d3c4a9 4210 if (sw)
c277df42
IZ
4211 next = NEXTOPER(NEXTOPER(scan));
4212 else {
4213 next = scan + ARG(scan);
4214 if (OP(next) == IFTHEN) /* Fake one. */
4215 next = NEXTOPER(NEXTOPER(next));
4216 }
4217 break;
4218 case LOGICAL:
24d3c4a9 4219 logical = scan->flags;
c277df42 4220 break;
c476f425 4221
2ab05381 4222/*******************************************************************
2ab05381 4223
c476f425
DM
4224The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/
4225pattern, where A and B are subpatterns. (For simple A, CURLYM or
4226STAR/PLUS/CURLY/CURLYN are used instead.)
2ab05381 4227
c476f425 4228A*B is compiled as <CURLYX><A><WHILEM><B>
2ab05381 4229
c476f425
DM
4230On entry to the subpattern, CURLYX is called. This pushes a CURLYX
4231state, which contains the current count, initialised to -1. It also sets
4232cur_curlyx to point to this state, with any previous value saved in the
4233state block.
2ab05381 4234
c476f425
DM
4235CURLYX then jumps straight to the WHILEM op, rather than executing A,
4236since the pattern may possibly match zero times (i.e. it's a while {} loop
4237rather than a do {} while loop).
2ab05381 4238
c476f425
DM
4239Each entry to WHILEM represents a successful match of A. The count in the
4240CURLYX block is incremented, another WHILEM state is pushed, and execution
4241passes to A or B depending on greediness and the current count.
2ab05381 4242
c476f425
DM
4243For example, if matching against the string a1a2a3b (where the aN are
4244substrings that match /A/), then the match progresses as follows: (the
4245pushed states are interspersed with the bits of strings matched so far):
2ab05381 4246
c476f425
DM
4247 <CURLYX cnt=-1>
4248 <CURLYX cnt=0><WHILEM>
4249 <CURLYX cnt=1><WHILEM> a1 <WHILEM>
4250 <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM>
4251 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM>
4252 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b
2ab05381 4253
c476f425
DM
4254(Contrast this with something like CURLYM, which maintains only a single
4255backtrack state:
2ab05381 4256
c476f425
DM
4257 <CURLYM cnt=0> a1
4258 a1 <CURLYM cnt=1> a2
4259 a1 a2 <CURLYM cnt=2> a3
4260 a1 a2 a3 <CURLYM cnt=3> b
4261)
2ab05381 4262
c476f425
DM
4263Each WHILEM state block marks a point to backtrack to upon partial failure
4264of A or B, and also contains some minor state data related to that
4265iteration. The CURLYX block, pointed to by cur_curlyx, contains the
4266overall state, such as the count, and pointers to the A and B ops.
2ab05381 4267
c476f425
DM
4268This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx
4269must always point to the *current* CURLYX block, the rules are:
2ab05381 4270
c476f425
DM
4271When executing CURLYX, save the old cur_curlyx in the CURLYX state block,
4272and set cur_curlyx to point the new block.
2ab05381 4273
c476f425
DM
4274When popping the CURLYX block after a successful or unsuccessful match,
4275restore the previous cur_curlyx.
2ab05381 4276
c476f425
DM
4277When WHILEM is about to execute B, save the current cur_curlyx, and set it
4278to the outer one saved in the CURLYX block.
2ab05381 4279
c476f425
DM
4280When popping the WHILEM block after a successful or unsuccessful B match,
4281restore the previous cur_curlyx.
2ab05381 4282
c476f425
DM
4283Here's an example for the pattern (AI* BI)*BO
4284I and O refer to inner and outer, C and W refer to CURLYX and WHILEM:
2ab05381 4285
c476f425
DM
4286cur_
4287curlyx backtrack stack
4288------ ---------------
4289NULL
4290CO <CO prev=NULL> <WO>
4291CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
4292CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
4293NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo
2ab05381 4294
c476f425
DM
4295At this point the pattern succeeds, and we work back down the stack to
4296clean up, restoring as we go:
95b24440 4297
c476f425
DM
4298CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
4299CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
4300CO <CO prev=NULL> <WO>
4301NULL
a0374537 4302
c476f425
DM
4303*******************************************************************/
4304
4305#define ST st->u.curlyx
4306
4307 case CURLYX: /* start of /A*B/ (for complex A) */
4308 {
4309 /* No need to save/restore up to this paren */
4310 I32 parenfloor = scan->flags;
4311
4312 assert(next); /* keep Coverity happy */
4313 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
4314 next += ARG(next);
4315
4316 /* XXXX Probably it is better to teach regpush to support
4317 parenfloor > PL_regsize... */
4318 if (parenfloor > (I32)*PL_reglastparen)
4319 parenfloor = *PL_reglastparen; /* Pessimization... */
4320
4321 ST.prev_curlyx= cur_curlyx;
4322 cur_curlyx = st;
4323 ST.cp = PL_savestack_ix;
4324
4325 /* these fields contain the state of the current curly.
4326 * they are accessed by subsequent WHILEMs */
4327 ST.parenfloor = parenfloor;
d02d6d97 4328 ST.me = scan;
c476f425 4329 ST.B = next;
24d3c4a9
DM
4330 ST.minmod = minmod;
4331 minmod = 0;
c476f425
DM
4332 ST.count = -1; /* this will be updated by WHILEM */
4333 ST.lastloc = NULL; /* this will be updated by WHILEM */
4334
4335 PL_reginput = locinput;
4336 PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next));
5f66b61c 4337 /* NOTREACHED */
c476f425 4338 }
a0d0e21e 4339
c476f425 4340 case CURLYX_end: /* just finished matching all of A*B */
c476f425
DM
4341 cur_curlyx = ST.prev_curlyx;
4342 sayYES;
4343 /* NOTREACHED */
a0d0e21e 4344
c476f425
DM
4345 case CURLYX_end_fail: /* just failed to match all of A*B */
4346 regcpblow(ST.cp);
4347 cur_curlyx = ST.prev_curlyx;
4348 sayNO;
4349 /* NOTREACHED */
4633a7c4 4350
a0d0e21e 4351
c476f425
DM
4352#undef ST
4353#define ST st->u.whilem
4354
4355 case WHILEM: /* just matched an A in /A*B/ (for complex A) */
4356 {
4357 /* see the discussion above about CURLYX/WHILEM */
c476f425 4358 I32 n;
d02d6d97
DM
4359 int min = ARG1(cur_curlyx->u.curlyx.me);
4360 int max = ARG2(cur_curlyx->u.curlyx.me);
4361 regnode *A = NEXTOPER(cur_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS;
4362
c476f425
DM
4363 assert(cur_curlyx); /* keep Coverity happy */
4364 n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
4365 ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
4366 ST.cache_offset = 0;
4367 ST.cache_mask = 0;
4368
4369 PL_reginput = locinput;
4370
4371 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
d02d6d97
DM
4372 "%*s whilem: matched %ld out of %d..%d\n",
4373 REPORT_CODE_OFF+depth*2, "", (long)n, min, max)
c476f425 4374 );
a0d0e21e 4375
c476f425 4376 /* First just match a string of min A's. */
a0d0e21e 4377
d02d6d97 4378 if (n < min) {
c476f425 4379 cur_curlyx->u.curlyx.lastloc = locinput;
d02d6d97 4380 PUSH_STATE_GOTO(WHILEM_A_pre, A);
c476f425
DM
4381 /* NOTREACHED */
4382 }
4383
4384 /* If degenerate A matches "", assume A done. */
4385
4386 if (locinput == cur_curlyx->u.curlyx.lastloc) {
4387 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4388 "%*s whilem: empty match detected, trying continuation...\n",
4389 REPORT_CODE_OFF+depth*2, "")
4390 );
4391 goto do_whilem_B_max;
4392 }
4393
4394 /* super-linear cache processing */
4395
4396 if (scan->flags) {
a0d0e21e 4397
2c2d71f5 4398 if (!PL_reg_maxiter) {
c476f425
DM
4399 /* start the countdown: Postpone detection until we
4400 * know the match is not *that* much linear. */
2c2d71f5 4401 PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4);
66bf836d
DM
4402 /* possible overflow for long strings and many CURLYX's */
4403 if (PL_reg_maxiter < 0)
4404 PL_reg_maxiter = I32_MAX;
2c2d71f5
JH
4405 PL_reg_leftiter = PL_reg_maxiter;
4406 }
c476f425 4407
2c2d71f5 4408 if (PL_reg_leftiter-- == 0) {
c476f425 4409 /* initialise cache */
3298f257 4410 const I32 size = (PL_reg_maxiter + 7)/8;
2c2d71f5 4411 if (PL_reg_poscache) {
eb160463 4412 if ((I32)PL_reg_poscache_size < size) {
2c2d71f5
JH
4413 Renew(PL_reg_poscache, size, char);
4414 PL_reg_poscache_size = size;
4415 }
4416 Zero(PL_reg_poscache, size, char);
4417 }
4418 else {
4419 PL_reg_poscache_size = size;
a02a5408 4420 Newxz(PL_reg_poscache, size, char);
2c2d71f5 4421 }
c476f425
DM
4422 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4423 "%swhilem: Detected a super-linear match, switching on caching%s...\n",
4424 PL_colors[4], PL_colors[5])
4425 );
2c2d71f5 4426 }
c476f425 4427
2c2d71f5 4428 if (PL_reg_leftiter < 0) {
c476f425
DM
4429 /* have we already failed at this position? */
4430 I32 offset, mask;
4431 offset = (scan->flags & 0xf) - 1
4432 + (locinput - PL_bostr) * (scan->flags>>4);
4433 mask = 1 << (offset % 8);
4434 offset /= 8;
4435 if (PL_reg_poscache[offset] & mask) {
4436 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4437 "%*s whilem: (cache) already tried at this position...\n",
4438 REPORT_CODE_OFF+depth*2, "")
2c2d71f5 4439 );
3298f257 4440 sayNO; /* cache records failure */
2c2d71f5 4441 }
c476f425
DM
4442 ST.cache_offset = offset;
4443 ST.cache_mask = mask;
2c2d71f5 4444 }
c476f425 4445 }
2c2d71f5 4446
c476f425 4447 /* Prefer B over A for minimal matching. */
a687059c 4448
c476f425
DM
4449 if (cur_curlyx->u.curlyx.minmod) {
4450 ST.save_curlyx = cur_curlyx;
4451 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
4452 ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor);
4453 REGCP_SET(ST.lastcp);
4454 PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B);
4455 /* NOTREACHED */
4456 }
a0d0e21e 4457
c476f425
DM
4458 /* Prefer A over B for maximal matching. */
4459
d02d6d97 4460 if (n < max) { /* More greed allowed? */
c476f425
DM
4461 ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
4462 cur_curlyx->u.curlyx.lastloc = locinput;
4463 REGCP_SET(ST.lastcp);
d02d6d97 4464 PUSH_STATE_GOTO(WHILEM_A_max, A);
c476f425
DM
4465 /* NOTREACHED */
4466 }
4467 goto do_whilem_B_max;
4468 }
4469 /* NOTREACHED */
4470
4471 case WHILEM_B_min: /* just matched B in a minimal match */
4472 case WHILEM_B_max: /* just matched B in a maximal match */
4473 cur_curlyx = ST.save_curlyx;
4474 sayYES;
4475 /* NOTREACHED */
4476
4477 case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
4478 cur_curlyx = ST.save_curlyx;
4479 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
4480 cur_curlyx->u.curlyx.count--;
4481 CACHEsayNO;
4482 /* NOTREACHED */
4483
4484 case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
4485 REGCP_UNWIND(ST.lastcp);
4486 regcppop(rex);
4487 /* FALL THROUGH */
4488 case WHILEM_A_pre_fail: /* just failed to match even minimal A */
4489 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
4490 cur_curlyx->u.curlyx.count--;
4491 CACHEsayNO;
4492 /* NOTREACHED */
4493
4494 case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
4495 REGCP_UNWIND(ST.lastcp);
4496 regcppop(rex); /* Restore some previous $<digit>s? */
4497 PL_reginput = locinput;
4498 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
4499 "%*s whilem: failed, trying continuation...\n",
4500 REPORT_CODE_OFF+depth*2, "")
4501 );
4502 do_whilem_B_max:
4503 if (cur_curlyx->u.curlyx.count >= REG_INFTY
4504 && ckWARN(WARN_REGEXP)
4505 && !(PL_reg_flags & RF_warned))
4506 {
4507 PL_reg_flags |= RF_warned;
4508 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded",
4509 "Complex regular subexpression recursion",
4510 REG_INFTY - 1);
4511 }
4512
4513 /* now try B */
4514 ST.save_curlyx = cur_curlyx;
4515 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
4516 PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B);
4517 /* NOTREACHED */
4518
4519 case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
4520 cur_curlyx = ST.save_curlyx;
4521 REGCP_UNWIND(ST.lastcp);
4522 regcppop(rex);
4523
d02d6d97 4524 if (cur_curlyx->u.curlyx.count >= /*max*/ARG2(cur_curlyx->u.curlyx.me)) {
c476f425
DM
4525 /* Maximum greed exceeded */
4526 if (cur_curlyx->u.curlyx.count >= REG_INFTY
4527 && ckWARN(WARN_REGEXP)
4528 && !(PL_reg_flags & RF_warned))
4529 {
3280af22 4530 PL_reg_flags |= RF_warned;
c476f425
DM
4531 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
4532 "%s limit (%d) exceeded",
4533 "Complex regular subexpression recursion",
4534 REG_INFTY - 1);
a0d0e21e 4535 }
c476f425 4536 cur_curlyx->u.curlyx.count--;
3ab3c9b4 4537 CACHEsayNO;
a0d0e21e 4538 }
c476f425
DM
4539
4540 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
4541 "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "")
4542 );
4543 /* Try grabbing another A and see if it helps. */
4544 PL_reginput = locinput;
4545 cur_curlyx->u.curlyx.lastloc = locinput;
4546 ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
4547 REGCP_SET(ST.lastcp);
d02d6d97
DM
4548 PUSH_STATE_GOTO(WHILEM_A_min,
4549 /*A*/ NEXTOPER(ST.save_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS);
5f66b61c 4550 /* NOTREACHED */
40a82448
DM
4551
4552#undef ST
4553#define ST st->u.branch
4554
4555 case BRANCHJ: /* /(...|A|...)/ with long next pointer */
c277df42
IZ
4556 next = scan + ARG(scan);
4557 if (next == scan)
4558 next = NULL;
40a82448
DM
4559 scan = NEXTOPER(scan);
4560 /* FALL THROUGH */
c277df42 4561
40a82448
DM
4562 case BRANCH: /* /(...|A|...)/ */
4563 scan = NEXTOPER(scan); /* scan now points to inner node */
40a82448
DM
4564 ST.lastparen = *PL_reglastparen;
4565 ST.next_branch = next;
4566 REGCP_SET(ST.cp);
4567 PL_reginput = locinput;
02db2b7b 4568
40a82448 4569 /* Now go into the branch */
5d458dd8
YO
4570 if (has_cutgroup) {
4571 PUSH_YES_STATE_GOTO(BRANCH_next, scan);
4572 } else {
4573 PUSH_STATE_GOTO(BRANCH_next, scan);
4574 }
40a82448 4575 /* NOTREACHED */
5d458dd8
YO
4576 case CUTGROUP:
4577 PL_reginput = locinput;
4578 sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL :
ad64d0ec 4579 MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5d458dd8
YO
4580 PUSH_STATE_GOTO(CUTGROUP_next,next);
4581 /* NOTREACHED */
4582 case CUTGROUP_next_fail:
4583 do_cutgroup = 1;
4584 no_final = 1;
4585 if (st->u.mark.mark_name)
4586 sv_commit = st->u.mark.mark_name;
4587 sayNO;
4588 /* NOTREACHED */
4589 case BRANCH_next:
4590 sayYES;
4591 /* NOTREACHED */
40a82448 4592 case BRANCH_next_fail: /* that branch failed; try the next, if any */
5d458dd8
YO
4593 if (do_cutgroup) {
4594 do_cutgroup = 0;
4595 no_final = 0;
4596 }
40a82448
DM
4597 REGCP_UNWIND(ST.cp);
4598 for (n = *PL_reglastparen; n > ST.lastparen; n--)
f0ab9afb 4599 PL_regoffs[n].end = -1;
40a82448 4600 *PL_reglastparen = n;
0a4db386 4601 /*dmq: *PL_reglastcloseparen = n; */
40a82448
DM
4602 scan = ST.next_branch;
4603 /* no more branches? */
5d458dd8
YO
4604 if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) {
4605 DEBUG_EXECUTE_r({
4606 PerlIO_printf( Perl_debug_log,
4607 "%*s %sBRANCH failed...%s\n",
4608 REPORT_CODE_OFF+depth*2, "",
4609 PL_colors[4],
4610 PL_colors[5] );
4611 });
4612 sayNO_SILENT;
4613 }
40a82448
DM
4614 continue; /* execute next BRANCH[J] op */
4615 /* NOTREACHED */
4616
a0d0e21e 4617 case MINMOD:
24d3c4a9 4618 minmod = 1;
a0d0e21e 4619 break;
40a82448
DM
4620
4621#undef ST
4622#define ST st->u.curlym
4623
4624 case CURLYM: /* /A{m,n}B/ where A is fixed-length */
4625
4626 /* This is an optimisation of CURLYX that enables us to push
84d2fa14 4627 * only a single backtracking state, no matter how many matches
40a82448
DM
4628 * there are in {m,n}. It relies on the pattern being constant
4629 * length, with no parens to influence future backrefs
4630 */
4631
4632 ST.me = scan;
dc45a647 4633 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
40a82448
DM
4634
4635 /* if paren positive, emulate an OPEN/CLOSE around A */
4636 if (ST.me->flags) {
3b6647e0 4637 U32 paren = ST.me->flags;
40a82448
DM
4638 if (paren > PL_regsize)
4639 PL_regsize = paren;
3b6647e0 4640 if (paren > *PL_reglastparen)
40a82448 4641 *PL_reglastparen = paren;
c277df42 4642 scan += NEXT_OFF(scan); /* Skip former OPEN. */
6407bf3b 4643 }
40a82448
DM
4644 ST.A = scan;
4645 ST.B = next;
4646 ST.alen = 0;
4647 ST.count = 0;
24d3c4a9
DM
4648 ST.minmod = minmod;
4649 minmod = 0;
40a82448
DM
4650 ST.c1 = CHRTEST_UNINIT;
4651 REGCP_SET(ST.cp);
6407bf3b 4652
40a82448
DM
4653 if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */
4654 goto curlym_do_B;
4655
4656 curlym_do_A: /* execute the A in /A{m,n}B/ */
6407bf3b 4657 PL_reginput = locinput;
40a82448
DM
4658 PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */
4659 /* NOTREACHED */
5f80c4cf 4660
40a82448
DM
4661 case CURLYM_A: /* we've just matched an A */
4662 locinput = st->locinput;
4663 nextchr = UCHARAT(locinput);
4664
4665 ST.count++;
4666 /* after first match, determine A's length: u.curlym.alen */
4667 if (ST.count == 1) {
4668 if (PL_reg_match_utf8) {
4669 char *s = locinput;
4670 while (s < PL_reginput) {
4671 ST.alen++;
4672 s += UTF8SKIP(s);
4673 }
4674 }
4675 else {
4676 ST.alen = PL_reginput - locinput;
4677 }
4678 if (ST.alen == 0)
4679 ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me);
4680 }
0cadcf80
DM
4681 DEBUG_EXECUTE_r(
4682 PerlIO_printf(Perl_debug_log,
40a82448 4683 "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n",
5bc10b2c 4684 (int)(REPORT_CODE_OFF+(depth*2)), "",
40a82448 4685 (IV) ST.count, (IV)ST.alen)
0cadcf80
DM
4686 );
4687
40a82448 4688 locinput = PL_reginput;
0a4db386
YO
4689
4690 if (cur_eval && cur_eval->u.eval.close_paren &&
24b23f37 4691 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
0a4db386
YO
4692 goto fake_end;
4693
c966426a
DM
4694 {
4695 I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me));
4696 if ( max == REG_INFTY || ST.count < max )
4697 goto curlym_do_A; /* try to match another A */
4698 }
40a82448 4699 goto curlym_do_B; /* try to match B */
5f80c4cf 4700
40a82448
DM
4701 case CURLYM_A_fail: /* just failed to match an A */
4702 REGCP_UNWIND(ST.cp);
0a4db386
YO
4703
4704 if (ST.minmod || ST.count < ARG1(ST.me) /* min*/
4705 || (cur_eval && cur_eval->u.eval.close_paren &&
24b23f37 4706 cur_eval->u.eval.close_paren == (U32)ST.me->flags))
40a82448 4707 sayNO;
0cadcf80 4708
40a82448
DM
4709 curlym_do_B: /* execute the B in /A{m,n}B/ */
4710 PL_reginput = locinput;
4711 if (ST.c1 == CHRTEST_UNINIT) {
4712 /* calculate c1 and c2 for possible match of 1st char
4713 * following curly */
4714 ST.c1 = ST.c2 = CHRTEST_VOID;
4715 if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
4716 regnode *text_node = ST.B;
4717 if (! HAS_TEXT(text_node))
4718 FIND_NEXT_IMPT(text_node);
ee9b8eae
YO
4719 /* this used to be
4720
4721 (HAS_TEXT(text_node) && PL_regkind[OP(text_node)] == EXACT)
4722
4723 But the former is redundant in light of the latter.
4724
4725 if this changes back then the macro for
4726 IS_TEXT and friends need to change.
4727 */
4728 if (PL_regkind[OP(text_node)] == EXACT)
40a82448 4729 {
ee9b8eae 4730
40a82448
DM
4731 ST.c1 = (U8)*STRING(text_node);
4732 ST.c2 =
ee9b8eae 4733 (IS_TEXTF(text_node))
40a82448 4734 ? PL_fold[ST.c1]
ee9b8eae 4735 : (IS_TEXTFL(text_node))
40a82448
DM
4736 ? PL_fold_locale[ST.c1]
4737 : ST.c1;
c277df42 4738 }
c277df42 4739 }
40a82448
DM
4740 }
4741
4742 DEBUG_EXECUTE_r(
4743 PerlIO_printf(Perl_debug_log,
4744 "%*s CURLYM trying tail with matches=%"IVdf"...\n",
5bc10b2c 4745 (int)(REPORT_CODE_OFF+(depth*2)),
40a82448
DM
4746 "", (IV)ST.count)
4747 );
4748 if (ST.c1 != CHRTEST_VOID
4749 && UCHARAT(PL_reginput) != ST.c1
4750 && UCHARAT(PL_reginput) != ST.c2)
4751 {
4752 /* simulate B failing */
3e901dc0
YO
4753 DEBUG_OPTIMISE_r(
4754 PerlIO_printf(Perl_debug_log,
4755 "%*s CURLYM Fast bail c1=%"IVdf" c2=%"IVdf"\n",
4756 (int)(REPORT_CODE_OFF+(depth*2)),"",
4757 (IV)ST.c1,(IV)ST.c2
4758 ));
40a82448
DM
4759 state_num = CURLYM_B_fail;
4760 goto reenter_switch;
4761 }
4762
4763 if (ST.me->flags) {
4764 /* mark current A as captured */
4765 I32 paren = ST.me->flags;
4766 if (ST.count) {
f0ab9afb 4767 PL_regoffs[paren].start
40a82448 4768 = HOPc(PL_reginput, -ST.alen) - PL_bostr;
f0ab9afb 4769 PL_regoffs[paren].end = PL_reginput - PL_bostr;
0a4db386 4770 /*dmq: *PL_reglastcloseparen = paren; */
c277df42 4771 }
40a82448 4772 else
f0ab9afb 4773 PL_regoffs[paren].end = -1;
0a4db386 4774 if (cur_eval && cur_eval->u.eval.close_paren &&
24b23f37 4775 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
0a4db386
YO
4776 {
4777 if (ST.count)
4778 goto fake_end;
4779 else
4780 sayNO;
4781 }
c277df42 4782 }
0a4db386 4783
40a82448 4784 PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */
5f66b61c 4785 /* NOTREACHED */
40a82448
DM
4786
4787 case CURLYM_B_fail: /* just failed to match a B */
4788 REGCP_UNWIND(ST.cp);
4789 if (ST.minmod) {
84d2fa14
HS
4790 I32 max = ARG2(ST.me);
4791 if (max != REG_INFTY && ST.count == max)
40a82448
DM
4792 sayNO;
4793 goto curlym_do_A; /* try to match a further A */
4794 }
4795 /* backtrack one A */
4796 if (ST.count == ARG1(ST.me) /* min */)
4797 sayNO;
4798 ST.count--;
4799 locinput = HOPc(locinput, -ST.alen);
4800 goto curlym_do_B; /* try to match B */
4801
c255a977
DM
4802#undef ST
4803#define ST st->u.curly
40a82448 4804
c255a977
DM
4805#define CURLY_SETPAREN(paren, success) \
4806 if (paren) { \
4807 if (success) { \
f0ab9afb
NC
4808 PL_regoffs[paren].start = HOPc(locinput, -1) - PL_bostr; \
4809 PL_regoffs[paren].end = locinput - PL_bostr; \
0a4db386 4810 *PL_reglastcloseparen = paren; \
c255a977
DM
4811 } \
4812 else \
f0ab9afb 4813 PL_regoffs[paren].end = -1; \
c255a977
DM
4814 }
4815
4816 case STAR: /* /A*B/ where A is width 1 */
4817 ST.paren = 0;
4818 ST.min = 0;
4819 ST.max = REG_INFTY;
a0d0e21e
LW
4820 scan = NEXTOPER(scan);
4821 goto repeat;
c255a977
DM
4822 case PLUS: /* /A+B/ where A is width 1 */
4823 ST.paren = 0;
4824 ST.min = 1;
4825 ST.max = REG_INFTY;
c277df42 4826 scan = NEXTOPER(scan);
c255a977
DM
4827 goto repeat;
4828 case CURLYN: /* /(A){m,n}B/ where A is width 1 */
4829 ST.paren = scan->flags; /* Which paren to set */
4830 if (ST.paren > PL_regsize)
4831 PL_regsize = ST.paren;
3b6647e0 4832 if (ST.paren > *PL_reglastparen)
c255a977
DM
4833 *PL_reglastparen = ST.paren;
4834 ST.min = ARG1(scan); /* min to match */
4835 ST.max = ARG2(scan); /* max to match */
0a4db386 4836 if (cur_eval && cur_eval->u.eval.close_paren &&
86413ec0 4837 cur_eval->u.eval.close_paren == (U32)ST.paren) {
0a4db386
YO
4838 ST.min=1;
4839 ST.max=1;
4840 }
c255a977
DM
4841 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
4842 goto repeat;
4843 case CURLY: /* /A{m,n}B/ where A is width 1 */
4844 ST.paren = 0;
4845 ST.min = ARG1(scan); /* min to match */
4846 ST.max = ARG2(scan); /* max to match */
4847 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
c277df42 4848 repeat:
a0d0e21e
LW
4849 /*
4850 * Lookahead to avoid useless match attempts
4851 * when we know what character comes next.
c255a977 4852 *
5f80c4cf
JP
4853 * Used to only do .*x and .*?x, but now it allows
4854 * for )'s, ('s and (?{ ... })'s to be in the way
4855 * of the quantifier and the EXACT-like node. -- japhy
4856 */
4857
c255a977
DM
4858 if (ST.min > ST.max) /* XXX make this a compile-time check? */
4859 sayNO;
cca55fe3 4860 if (HAS_TEXT(next) || JUMPABLE(next)) {
5f80c4cf
JP
4861 U8 *s;
4862 regnode *text_node = next;
4863
3dab1dad
YO
4864 if (! HAS_TEXT(text_node))
4865 FIND_NEXT_IMPT(text_node);
5f80c4cf 4866
9e137952 4867 if (! HAS_TEXT(text_node))
c255a977 4868 ST.c1 = ST.c2 = CHRTEST_VOID;
5f80c4cf 4869 else {
ee9b8eae 4870 if ( PL_regkind[OP(text_node)] != EXACT ) {
c255a977 4871 ST.c1 = ST.c2 = CHRTEST_VOID;
44a68960 4872 goto assume_ok_easy;
cca55fe3 4873 }
be8e71aa
YO
4874 else
4875 s = (U8*)STRING(text_node);
ee9b8eae
YO
4876
4877 /* Currently we only get here when
4878
4879 PL_rekind[OP(text_node)] == EXACT
4880
4881 if this changes back then the macro for IS_TEXT and
4882 friends need to change. */
f2ed9b32 4883 if (!UTF_PATTERN) {
c255a977 4884 ST.c2 = ST.c1 = *s;
ee9b8eae 4885 if (IS_TEXTF(text_node))
c255a977 4886 ST.c2 = PL_fold[ST.c1];
ee9b8eae 4887 else if (IS_TEXTFL(text_node))
c255a977 4888 ST.c2 = PL_fold_locale[ST.c1];
1aa99e6b 4889 }
f2ed9b32 4890 else { /* UTF_PATTERN */
ee9b8eae 4891 if (IS_TEXTF(text_node)) {
a2a2844f 4892 STRLEN ulen1, ulen2;
89ebb4a3
JH
4893 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
4894 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
a2a2844f
JH
4895
4896 to_utf8_lower((U8*)s, tmpbuf1, &ulen1);
4897 to_utf8_upper((U8*)s, tmpbuf2, &ulen2);
e294cc5d
JH
4898#ifdef EBCDIC
4899 ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0,
4900 ckWARN(WARN_UTF8) ?
4901 0 : UTF8_ALLOW_ANY);
4902 ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0,
4903 ckWARN(WARN_UTF8) ?
4904 0 : UTF8_ALLOW_ANY);
4905#else
c255a977 4906 ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0,
e294cc5d 4907 uniflags);
c255a977 4908 ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0,
e294cc5d
JH
4909 uniflags);
4910#endif
5f80c4cf
JP
4911 }
4912 else {
c255a977 4913 ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0,
041457d9 4914 uniflags);
5f80c4cf 4915 }
1aa99e6b
IH
4916 }
4917 }
bbce6d69 4918 }
a0d0e21e 4919 else
c255a977 4920 ST.c1 = ST.c2 = CHRTEST_VOID;
cca55fe3 4921 assume_ok_easy:
c255a977
DM
4922
4923 ST.A = scan;
4924 ST.B = next;
3280af22 4925 PL_reginput = locinput;
24d3c4a9
DM
4926 if (minmod) {
4927 minmod = 0;
e2e6a0f1 4928 if (ST.min && regrepeat(rex, ST.A, ST.min, depth) < ST.min)
4633a7c4 4929 sayNO;
c255a977 4930 ST.count = ST.min;
a0ed51b3 4931 locinput = PL_reginput;
c255a977
DM
4932 REGCP_SET(ST.cp);
4933 if (ST.c1 == CHRTEST_VOID)
4934 goto curly_try_B_min;
4935
4936 ST.oldloc = locinput;
4937
4938 /* set ST.maxpos to the furthest point along the
4939 * string that could possibly match */
4940 if (ST.max == REG_INFTY) {
4941 ST.maxpos = PL_regeol - 1;
f2ed9b32 4942 if (utf8_target)
c255a977
DM
4943 while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos))
4944 ST.maxpos--;
4945 }
f2ed9b32 4946 else if (utf8_target) {
c255a977
DM
4947 int m = ST.max - ST.min;
4948 for (ST.maxpos = locinput;
4949 m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--)
4950 ST.maxpos += UTF8SKIP(ST.maxpos);
4951 }
4952 else {
4953 ST.maxpos = locinput + ST.max - ST.min;
4954 if (ST.maxpos >= PL_regeol)
4955 ST.maxpos = PL_regeol - 1;
4956 }
4957 goto curly_try_B_min_known;
4958
4959 }
4960 else {
e2e6a0f1 4961 ST.count = regrepeat(rex, ST.A, ST.max, depth);
c255a977
DM
4962 locinput = PL_reginput;
4963 if (ST.count < ST.min)
4964 sayNO;
4965 if ((ST.count > ST.min)
4966 && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL))
4967 {
4968 /* A{m,n} must come at the end of the string, there's
4969 * no point in backing off ... */
4970 ST.min = ST.count;
4971 /* ...except that $ and \Z can match before *and* after
4972 newline at the end. Consider "\n\n" =~ /\n+\Z\n/.
4973 We may back off by one in this case. */
4974 if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS)
4975 ST.min--;
4976 }
4977 REGCP_SET(ST.cp);
4978 goto curly_try_B_max;
4979 }
4980 /* NOTREACHED */
4981
4982
4983 case CURLY_B_min_known_fail:
4984 /* failed to find B in a non-greedy match where c1,c2 valid */
4985 if (ST.paren && ST.count)
f0ab9afb 4986 PL_regoffs[ST.paren].end = -1;
c255a977
DM
4987
4988 PL_reginput = locinput; /* Could be reset... */
4989 REGCP_UNWIND(ST.cp);
4990 /* Couldn't or didn't -- move forward. */
4991 ST.oldloc = locinput;
f2ed9b32 4992 if (utf8_target)
c255a977
DM
4993 locinput += UTF8SKIP(locinput);
4994 else
4995 locinput++;
4996 ST.count++;
4997 curly_try_B_min_known:
4998 /* find the next place where 'B' could work, then call B */
4999 {
5000 int n;
f2ed9b32 5001 if (utf8_target) {
c255a977
DM
5002 n = (ST.oldloc == locinput) ? 0 : 1;
5003 if (ST.c1 == ST.c2) {
5004 STRLEN len;
5005 /* set n to utf8_distance(oldloc, locinput) */
5006 while (locinput <= ST.maxpos &&
5007 utf8n_to_uvchr((U8*)locinput,
5008 UTF8_MAXBYTES, &len,
5009 uniflags) != (UV)ST.c1) {
5010 locinput += len;
5011 n++;
5012 }
1aa99e6b
IH
5013 }
5014 else {
c255a977
DM
5015 /* set n to utf8_distance(oldloc, locinput) */
5016 while (locinput <= ST.maxpos) {
5017 STRLEN len;
5018 const UV c = utf8n_to_uvchr((U8*)locinput,
5019 UTF8_MAXBYTES, &len,
5020 uniflags);
5021 if (c == (UV)ST.c1 || c == (UV)ST.c2)
5022 break;
5023 locinput += len;
5024 n++;
1aa99e6b 5025 }
0fe9bf95
IZ
5026 }
5027 }
c255a977
DM
5028 else {
5029 if (ST.c1 == ST.c2) {
5030 while (locinput <= ST.maxpos &&
5031 UCHARAT(locinput) != ST.c1)
5032 locinput++;
bbce6d69 5033 }
c255a977
DM
5034 else {
5035 while (locinput <= ST.maxpos
5036 && UCHARAT(locinput) != ST.c1
5037 && UCHARAT(locinput) != ST.c2)
5038 locinput++;
a0ed51b3 5039 }
c255a977
DM
5040 n = locinput - ST.oldloc;
5041 }
5042 if (locinput > ST.maxpos)
5043 sayNO;
5044 /* PL_reginput == oldloc now */
5045 if (n) {
5046 ST.count += n;
e2e6a0f1 5047 if (regrepeat(rex, ST.A, n, depth) < n)
4633a7c4 5048 sayNO;
a0d0e21e 5049 }
c255a977
DM
5050 PL_reginput = locinput;
5051 CURLY_SETPAREN(ST.paren, ST.count);
0a4db386 5052 if (cur_eval && cur_eval->u.eval.close_paren &&
86413ec0 5053 cur_eval->u.eval.close_paren == (U32)ST.paren) {
0a4db386
YO
5054 goto fake_end;
5055 }
c255a977 5056 PUSH_STATE_GOTO(CURLY_B_min_known, ST.B);
a0d0e21e 5057 }
c255a977
DM
5058 /* NOTREACHED */
5059
5060
5061 case CURLY_B_min_fail:
5062 /* failed to find B in a non-greedy match where c1,c2 invalid */
5063 if (ST.paren && ST.count)
f0ab9afb 5064 PL_regoffs[ST.paren].end = -1;
c255a977
DM
5065
5066 REGCP_UNWIND(ST.cp);
5067 /* failed -- move forward one */
5068 PL_reginput = locinput;
e2e6a0f1 5069 if (regrepeat(rex, ST.A, 1, depth)) {
c255a977 5070 ST.count++;
a0ed51b3 5071 locinput = PL_reginput;
c255a977
DM
5072 if (ST.count <= ST.max || (ST.max == REG_INFTY &&
5073 ST.count > 0)) /* count overflow ? */
15272685 5074 {
c255a977
DM
5075 curly_try_B_min:
5076 CURLY_SETPAREN(ST.paren, ST.count);
0a4db386 5077 if (cur_eval && cur_eval->u.eval.close_paren &&
86413ec0 5078 cur_eval->u.eval.close_paren == (U32)ST.paren) {
0a4db386
YO
5079 goto fake_end;
5080 }
c255a977 5081 PUSH_STATE_GOTO(CURLY_B_min, ST.B);
a0d0e21e
LW
5082 }
5083 }
4633a7c4 5084 sayNO;
c255a977
DM
5085 /* NOTREACHED */
5086
5087
5088 curly_try_B_max:
5089 /* a successful greedy match: now try to match B */
40d049e4 5090 if (cur_eval && cur_eval->u.eval.close_paren &&
86413ec0 5091 cur_eval->u.eval.close_paren == (U32)ST.paren) {
40d049e4
YO
5092 goto fake_end;
5093 }
c255a977
DM
5094 {
5095 UV c = 0;
5096 if (ST.c1 != CHRTEST_VOID)
f2ed9b32 5097 c = utf8_target ? utf8n_to_uvchr((U8*)PL_reginput,
c255a977 5098 UTF8_MAXBYTES, 0, uniflags)
466787eb 5099 : (UV) UCHARAT(PL_reginput);
c255a977
DM
5100 /* If it could work, try it. */
5101 if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) {
5102 CURLY_SETPAREN(ST.paren, ST.count);
5103 PUSH_STATE_GOTO(CURLY_B_max, ST.B);
5104 /* NOTREACHED */
5105 }
5106 }
5107 /* FALL THROUGH */
5108 case CURLY_B_max_fail:
5109 /* failed to find B in a greedy match */
5110 if (ST.paren && ST.count)
f0ab9afb 5111 PL_regoffs[ST.paren].end = -1;
c255a977
DM
5112
5113 REGCP_UNWIND(ST.cp);
5114 /* back up. */
5115 if (--ST.count < ST.min)
5116 sayNO;
5117 PL_reginput = locinput = HOPc(locinput, -1);
5118 goto curly_try_B_max;
5119
5120#undef ST
5121
a0d0e21e 5122 case END:
6bda09f9 5123 fake_end:
faec1544
DM
5124 if (cur_eval) {
5125 /* we've just finished A in /(??{A})B/; now continue with B */
5126 I32 tmpix;
faec1544
DM
5127 st->u.eval.toggle_reg_flags
5128 = cur_eval->u.eval.toggle_reg_flags;
5129 PL_reg_flags ^= st->u.eval.toggle_reg_flags;
5130
288b8c02
NC
5131 st->u.eval.prev_rex = rex_sv; /* inner */
5132 SETREX(rex_sv,cur_eval->u.eval.prev_rex);
5133 rex = (struct regexp *)SvANY(rex_sv);
f8fc2ecf 5134 rexi = RXi_GET(rex);
faec1544 5135 cur_curlyx = cur_eval->u.eval.prev_curlyx;
288b8c02 5136 ReREFCNT_inc(rex_sv);
faec1544 5137 st->u.eval.cp = regcppush(0); /* Save *all* the positions. */
34a81e2b
B
5138
5139 /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
5140 PL_reglastparen = &rex->lastparen;
5141 PL_reglastcloseparen = &rex->lastcloseparen;
5142
faec1544
DM
5143 REGCP_SET(st->u.eval.lastcp);
5144 PL_reginput = locinput;
5145
5146 /* Restore parens of the outer rex without popping the
5147 * savestack */
5148 tmpix = PL_savestack_ix;
5149 PL_savestack_ix = cur_eval->u.eval.lastcp;
5150 regcppop(rex);
5151 PL_savestack_ix = tmpix;
5152
5153 st->u.eval.prev_eval = cur_eval;
5154 cur_eval = cur_eval->u.eval.prev_eval;
5155 DEBUG_EXECUTE_r(
2a49f0f5
JH
5156 PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %"UVxf"\n",
5157 REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval)););
e7707071
YO
5158 if ( nochange_depth )
5159 nochange_depth--;
5160
5161 PUSH_YES_STATE_GOTO(EVAL_AB,
faec1544
DM
5162 st->u.eval.prev_eval->u.eval.B); /* match B */
5163 }
5164
3b0527fe 5165 if (locinput < reginfo->till) {
a3621e74 5166 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
7821416a
IZ
5167 "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
5168 PL_colors[4],
5169 (long)(locinput - PL_reg_starttry),
3b0527fe 5170 (long)(reginfo->till - PL_reg_starttry),
7821416a 5171 PL_colors[5]));
58e23c8d 5172
262b90c4 5173 sayNO_SILENT; /* Cannot match: too short. */
7821416a
IZ
5174 }
5175 PL_reginput = locinput; /* put where regtry can find it */
262b90c4 5176 sayYES; /* Success! */
dad79028
DM
5177
5178 case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */
5179 DEBUG_EXECUTE_r(
5180 PerlIO_printf(Perl_debug_log,
5181 "%*s %ssubpattern success...%s\n",
5bc10b2c 5182 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]));
3280af22 5183 PL_reginput = locinput; /* put where regtry can find it */
262b90c4 5184 sayYES; /* Success! */
dad79028 5185
40a82448
DM
5186#undef ST
5187#define ST st->u.ifmatch
5188
5189 case SUSPEND: /* (?>A) */
5190 ST.wanted = 1;
9fe1d20c 5191 PL_reginput = locinput;
9041c2e3 5192 goto do_ifmatch;
dad79028 5193
40a82448
DM
5194 case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */
5195 ST.wanted = 0;
dad79028
DM
5196 goto ifmatch_trivial_fail_test;
5197
40a82448
DM
5198 case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */
5199 ST.wanted = 1;
dad79028 5200 ifmatch_trivial_fail_test:
a0ed51b3 5201 if (scan->flags) {
52657f30 5202 char * const s = HOPBACKc(locinput, scan->flags);
dad79028
DM
5203 if (!s) {
5204 /* trivial fail */
24d3c4a9
DM
5205 if (logical) {
5206 logical = 0;
f2338a2e 5207 sw = 1 - cBOOL(ST.wanted);
dad79028 5208 }
40a82448 5209 else if (ST.wanted)
dad79028
DM
5210 sayNO;
5211 next = scan + ARG(scan);
5212 if (next == scan)
5213 next = NULL;
5214 break;
5215 }
efb30f32 5216 PL_reginput = s;
a0ed51b3
LW
5217 }
5218 else
5219 PL_reginput = locinput;
5220
c277df42 5221 do_ifmatch:
40a82448 5222 ST.me = scan;
24d3c4a9 5223 ST.logical = logical;
24d786f4
YO
5224 logical = 0; /* XXX: reset state of logical once it has been saved into ST */
5225
40a82448
DM
5226 /* execute body of (?...A) */
5227 PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan)));
5228 /* NOTREACHED */
5229
5230 case IFMATCH_A_fail: /* body of (?...A) failed */
5231 ST.wanted = !ST.wanted;
5232 /* FALL THROUGH */
5233
5234 case IFMATCH_A: /* body of (?...A) succeeded */
24d3c4a9 5235 if (ST.logical) {
f2338a2e 5236 sw = cBOOL(ST.wanted);
40a82448
DM
5237 }
5238 else if (!ST.wanted)
5239 sayNO;
5240
5241 if (OP(ST.me) == SUSPEND)
5242 locinput = PL_reginput;
5243 else {
5244 locinput = PL_reginput = st->locinput;
5245 nextchr = UCHARAT(locinput);
5246 }
5247 scan = ST.me + ARG(ST.me);
5248 if (scan == ST.me)
5249 scan = NULL;
5250 continue; /* execute B */
5251
5252#undef ST
dad79028 5253
c277df42 5254 case LONGJMP:
c277df42
IZ
5255 next = scan + ARG(scan);
5256 if (next == scan)
5257 next = NULL;
a0d0e21e 5258 break;
54612592 5259 case COMMIT:
e2e6a0f1
YO
5260 reginfo->cutpoint = PL_regeol;
5261 /* FALLTHROUGH */
5d458dd8 5262 case PRUNE:
24b23f37 5263 PL_reginput = locinput;
e2e6a0f1 5264 if (!scan->flags)
ad64d0ec 5265 sv_yes_mark = sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
54612592
YO
5266 PUSH_STATE_GOTO(COMMIT_next,next);
5267 /* NOTREACHED */
5268 case COMMIT_next_fail:
5269 no_final = 1;
5270 /* FALLTHROUGH */
7f69552c
YO
5271 case OPFAIL:
5272 sayNO;
e2e6a0f1
YO
5273 /* NOTREACHED */
5274
5275#define ST st->u.mark
5276 case MARKPOINT:
5277 ST.prev_mark = mark_state;
5d458dd8 5278 ST.mark_name = sv_commit = sv_yes_mark
ad64d0ec 5279 = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
e2e6a0f1
YO
5280 mark_state = st;
5281 ST.mark_loc = PL_reginput = locinput;
5282 PUSH_YES_STATE_GOTO(MARKPOINT_next,next);
5283 /* NOTREACHED */
5284 case MARKPOINT_next:
5285 mark_state = ST.prev_mark;
5286 sayYES;
5287 /* NOTREACHED */
5288 case MARKPOINT_next_fail:
5d458dd8 5289 if (popmark && sv_eq(ST.mark_name,popmark))
e2e6a0f1
YO
5290 {
5291 if (ST.mark_loc > startpoint)
5292 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
5293 popmark = NULL; /* we found our mark */
5294 sv_commit = ST.mark_name;
5295
5296 DEBUG_EXECUTE_r({
5d458dd8 5297 PerlIO_printf(Perl_debug_log,
e2e6a0f1
YO
5298 "%*s %ssetting cutpoint to mark:%"SVf"...%s\n",
5299 REPORT_CODE_OFF+depth*2, "",
be2597df 5300 PL_colors[4], SVfARG(sv_commit), PL_colors[5]);
e2e6a0f1
YO
5301 });
5302 }
5303 mark_state = ST.prev_mark;
5d458dd8
YO
5304 sv_yes_mark = mark_state ?
5305 mark_state->u.mark.mark_name : NULL;
e2e6a0f1
YO
5306 sayNO;
5307 /* NOTREACHED */
5d458dd8
YO
5308 case SKIP:
5309 PL_reginput = locinput;
5310 if (scan->flags) {
2bf803e2 5311 /* (*SKIP) : if we fail we cut here*/
5d458dd8 5312 ST.mark_name = NULL;
e2e6a0f1 5313 ST.mark_loc = locinput;
5d458dd8
YO
5314 PUSH_STATE_GOTO(SKIP_next,next);
5315 } else {
2bf803e2 5316 /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was,
5d458dd8
YO
5317 otherwise do nothing. Meaning we need to scan
5318 */
5319 regmatch_state *cur = mark_state;
ad64d0ec 5320 SV *find = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5d458dd8
YO
5321
5322 while (cur) {
5323 if ( sv_eq( cur->u.mark.mark_name,
5324 find ) )
5325 {
5326 ST.mark_name = find;
5327 PUSH_STATE_GOTO( SKIP_next, next );
5328 }
5329 cur = cur->u.mark.prev_mark;
5330 }
e2e6a0f1 5331 }
2bf803e2 5332 /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */
5d458dd8
YO
5333 break;
5334 case SKIP_next_fail:
5335 if (ST.mark_name) {
5336 /* (*CUT:NAME) - Set up to search for the name as we
5337 collapse the stack*/
5338 popmark = ST.mark_name;
5339 } else {
5340 /* (*CUT) - No name, we cut here.*/
e2e6a0f1
YO
5341 if (ST.mark_loc > startpoint)
5342 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
5d458dd8
YO
5343 /* but we set sv_commit to latest mark_name if there
5344 is one so they can test to see how things lead to this
5345 cut */
5346 if (mark_state)
5347 sv_commit=mark_state->u.mark.mark_name;
5348 }
e2e6a0f1
YO
5349 no_final = 1;
5350 sayNO;
5351 /* NOTREACHED */
5352#undef ST
32e6a07c
YO
5353 case FOLDCHAR:
5354 n = ARG(scan);
f2ed9b32 5355 if ( n == (U32)what_len_TRICKYFOLD(locinput,utf8_target,ln) ) {
e64b1bd1 5356 locinput += ln;
f2ed9b32 5357 } else if ( 0xDF == n && !utf8_target && !UTF_PATTERN ) {
e64b1bd1
YO
5358 sayNO;
5359 } else {
5360 U8 folded[UTF8_MAXBYTES_CASE+1];
5361 STRLEN foldlen;
5362 const char * const l = locinput;
5363 char *e = PL_regeol;
5364 to_uni_fold(n, folded, &foldlen);
5365
4c1b470c 5366 if (! foldEQ_utf8((const char*) folded, 0, foldlen, 1,
f2ed9b32 5367 l, &e, 0, utf8_target)) {
32e6a07c 5368 sayNO;
e64b1bd1
YO
5369 }
5370 locinput = e;
32e6a07c
YO
5371 }
5372 nextchr = UCHARAT(locinput);
5373 break;
e1d1eefb 5374 case LNBREAK:
f2ed9b32 5375 if ((n=is_LNBREAK(locinput,utf8_target))) {
e1d1eefb
YO
5376 locinput += n;
5377 nextchr = UCHARAT(locinput);
5378 } else
5379 sayNO;
5380 break;
5381
5382#define CASE_CLASS(nAmE) \
5383 case nAmE: \
f2ed9b32 5384 if ((n=is_##nAmE(locinput,utf8_target))) { \
e1d1eefb
YO
5385 locinput += n; \
5386 nextchr = UCHARAT(locinput); \
5387 } else \
5388 sayNO; \
5389 break; \
5390 case N##nAmE: \
f2ed9b32 5391 if ((n=is_##nAmE(locinput,utf8_target))) { \
e1d1eefb
YO
5392 sayNO; \
5393 } else { \
5394 locinput += UTF8SKIP(locinput); \
5395 nextchr = UCHARAT(locinput); \
5396 } \
5397 break
5398
5399 CASE_CLASS(VERTWS);
5400 CASE_CLASS(HORIZWS);
5401#undef CASE_CLASS
5402
a0d0e21e 5403 default:
b900a521 5404 PerlIO_printf(Perl_error_log, "%"UVxf" %d\n",
d7d93a81 5405 PTR2UV(scan), OP(scan));
cea2e8a9 5406 Perl_croak(aTHX_ "regexp memory corruption");
5d458dd8
YO
5407
5408 } /* end switch */
95b24440 5409
5d458dd8
YO
5410 /* switch break jumps here */
5411 scan = next; /* prepare to execute the next op and ... */
5412 continue; /* ... jump back to the top, reusing st */
95b24440
DM
5413 /* NOTREACHED */
5414
40a82448
DM
5415 push_yes_state:
5416 /* push a state that backtracks on success */
5417 st->u.yes.prev_yes_state = yes_state;
5418 yes_state = st;
5419 /* FALL THROUGH */
5420 push_state:
5421 /* push a new regex state, then continue at scan */
5422 {
5423 regmatch_state *newst;
5424
24b23f37
YO
5425 DEBUG_STACK_r({
5426 regmatch_state *cur = st;
5427 regmatch_state *curyes = yes_state;
5428 int curd = depth;
5429 regmatch_slab *slab = PL_regmatch_slab;
5430 for (;curd > -1;cur--,curd--) {
5431 if (cur < SLAB_FIRST(slab)) {
5432 slab = slab->prev;
5433 cur = SLAB_LAST(slab);
5434 }
5435 PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n",
5436 REPORT_CODE_OFF + 2 + depth * 2,"",
13d6edb4 5437 curd, PL_reg_name[cur->resume_state],
24b23f37
YO
5438 (curyes == cur) ? "yes" : ""
5439 );
5440 if (curyes == cur)
5441 curyes = cur->u.yes.prev_yes_state;
5442 }
5443 } else
5444 DEBUG_STATE_pp("push")
5445 );
40a82448 5446 depth++;
40a82448
DM
5447 st->locinput = locinput;
5448 newst = st+1;
5449 if (newst > SLAB_LAST(PL_regmatch_slab))
5450 newst = S_push_slab(aTHX);
5451 PL_regmatch_state = newst;
786e8c11 5452
40a82448
DM
5453 locinput = PL_reginput;
5454 nextchr = UCHARAT(locinput);
5455 st = newst;
5456 continue;
5457 /* NOTREACHED */
5458 }
a0d0e21e 5459 }
a687059c 5460
a0d0e21e
LW
5461 /*
5462 * We get here only if there's trouble -- normally "case END" is
5463 * the terminating point.
5464 */
cea2e8a9 5465 Perl_croak(aTHX_ "corrupted regexp pointers");
a0d0e21e 5466 /*NOTREACHED*/
4633a7c4
LW
5467 sayNO;
5468
262b90c4 5469yes:
77cb431f
DM
5470 if (yes_state) {
5471 /* we have successfully completed a subexpression, but we must now
5472 * pop to the state marked by yes_state and continue from there */
77cb431f 5473 assert(st != yes_state);
5bc10b2c
DM
5474#ifdef DEBUGGING
5475 while (st != yes_state) {
5476 st--;
5477 if (st < SLAB_FIRST(PL_regmatch_slab)) {
5478 PL_regmatch_slab = PL_regmatch_slab->prev;
5479 st = SLAB_LAST(PL_regmatch_slab);
5480 }
e2e6a0f1 5481 DEBUG_STATE_r({
54612592
YO
5482 if (no_final) {
5483 DEBUG_STATE_pp("pop (no final)");
5484 } else {
5485 DEBUG_STATE_pp("pop (yes)");
5486 }
e2e6a0f1 5487 });
5bc10b2c
DM
5488 depth--;
5489 }
5490#else
77cb431f
DM
5491 while (yes_state < SLAB_FIRST(PL_regmatch_slab)
5492 || yes_state > SLAB_LAST(PL_regmatch_slab))
5493 {
5494 /* not in this slab, pop slab */
5495 depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1);
5496 PL_regmatch_slab = PL_regmatch_slab->prev;
5497 st = SLAB_LAST(PL_regmatch_slab);
5498 }
5499 depth -= (st - yes_state);
5bc10b2c 5500#endif
77cb431f
DM
5501 st = yes_state;
5502 yes_state = st->u.yes.prev_yes_state;
5503 PL_regmatch_state = st;
24b23f37 5504
5d458dd8
YO
5505 if (no_final) {
5506 locinput= st->locinput;
5507 nextchr = UCHARAT(locinput);
5508 }
54612592 5509 state_num = st->resume_state + no_final;
24d3c4a9 5510 goto reenter_switch;
77cb431f
DM
5511 }
5512
a3621e74 5513 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
e4584336 5514 PL_colors[4], PL_colors[5]));
02db2b7b 5515
19b95bf0
DM
5516 if (PL_reg_eval_set) {
5517 /* each successfully executed (?{...}) block does the equivalent of
5518 * local $^R = do {...}
5519 * When popping the save stack, all these locals would be undone;
5520 * bypass this by setting the outermost saved $^R to the latest
5521 * value */
5522 if (oreplsv != GvSV(PL_replgv))
5523 sv_setsv(oreplsv, GvSV(PL_replgv));
5524 }
95b24440 5525 result = 1;
aa283a38 5526 goto final_exit;
4633a7c4
LW
5527
5528no:
a3621e74 5529 DEBUG_EXECUTE_r(
7821416a 5530 PerlIO_printf(Perl_debug_log,
786e8c11 5531 "%*s %sfailed...%s\n",
5bc10b2c 5532 REPORT_CODE_OFF+depth*2, "",
786e8c11 5533 PL_colors[4], PL_colors[5])
7821416a 5534 );
aa283a38 5535
262b90c4 5536no_silent:
54612592
YO
5537 if (no_final) {
5538 if (yes_state) {
5539 goto yes;
5540 } else {
5541 goto final_exit;
5542 }
5543 }
aa283a38
DM
5544 if (depth) {
5545 /* there's a previous state to backtrack to */
40a82448
DM
5546 st--;
5547 if (st < SLAB_FIRST(PL_regmatch_slab)) {
5548 PL_regmatch_slab = PL_regmatch_slab->prev;
5549 st = SLAB_LAST(PL_regmatch_slab);
5550 }
5551 PL_regmatch_state = st;
40a82448
DM
5552 locinput= st->locinput;
5553 nextchr = UCHARAT(locinput);
5554
5bc10b2c
DM
5555 DEBUG_STATE_pp("pop");
5556 depth--;
262b90c4
DM
5557 if (yes_state == st)
5558 yes_state = st->u.yes.prev_yes_state;
5bc10b2c 5559
24d3c4a9
DM
5560 state_num = st->resume_state + 1; /* failure = success + 1 */
5561 goto reenter_switch;
95b24440 5562 }
24d3c4a9 5563 result = 0;
aa283a38 5564
262b90c4 5565 final_exit:
bbe252da 5566 if (rex->intflags & PREGf_VERBARG_SEEN) {
5d458dd8
YO
5567 SV *sv_err = get_sv("REGERROR", 1);
5568 SV *sv_mrk = get_sv("REGMARK", 1);
5569 if (result) {
e2e6a0f1 5570 sv_commit = &PL_sv_no;
5d458dd8
YO
5571 if (!sv_yes_mark)
5572 sv_yes_mark = &PL_sv_yes;
5573 } else {
5574 if (!sv_commit)
5575 sv_commit = &PL_sv_yes;
5576 sv_yes_mark = &PL_sv_no;
5577 }
5578 sv_setsv(sv_err, sv_commit);
5579 sv_setsv(sv_mrk, sv_yes_mark);
e2e6a0f1 5580 }
19b95bf0 5581
2f554ef7
DM
5582 /* clean up; in particular, free all slabs above current one */
5583 LEAVE_SCOPE(oldsave);
5d9a96ca 5584
95b24440 5585 return result;
a687059c
LW
5586}
5587
5588/*
5589 - regrepeat - repeatedly match something simple, report how many
5590 */
5591/*
5592 * [This routine now assumes that it will only match on things of length 1.
5593 * That was true before, but now we assume scan - reginput is the count,
a0ed51b3 5594 * rather than incrementing count on every character. [Er, except utf8.]]
a687059c 5595 */
76e3520e 5596STATIC I32
e2e6a0f1 5597S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth)
a687059c 5598{
27da23d5 5599 dVAR;
a0d0e21e 5600 register char *scan;
a0d0e21e 5601 register I32 c;
3280af22 5602 register char *loceol = PL_regeol;
a0ed51b3 5603 register I32 hardcount = 0;
f2ed9b32 5604 register bool utf8_target = PL_reg_match_utf8;
4f55667c
SP
5605#ifndef DEBUGGING
5606 PERL_UNUSED_ARG(depth);
5607#endif
a0d0e21e 5608
7918f24d
NC
5609 PERL_ARGS_ASSERT_REGREPEAT;
5610
3280af22 5611 scan = PL_reginput;
faf11cac
HS
5612 if (max == REG_INFTY)
5613 max = I32_MAX;
5614 else if (max < loceol - scan)
7f596f4c 5615 loceol = scan + max;
a0d0e21e 5616 switch (OP(p)) {
22c35a8c 5617 case REG_ANY:
f2ed9b32 5618 if (utf8_target) {
ffc61ed2 5619 loceol = PL_regeol;
1aa99e6b 5620 while (scan < loceol && hardcount < max && *scan != '\n') {
ffc61ed2
JH
5621 scan += UTF8SKIP(scan);
5622 hardcount++;
5623 }
5624 } else {
5625 while (scan < loceol && *scan != '\n')
5626 scan++;
a0ed51b3
LW
5627 }
5628 break;
ffc61ed2 5629 case SANY:
f2ed9b32 5630 if (utf8_target) {
def8e4ea 5631 loceol = PL_regeol;
a0804c9e 5632 while (scan < loceol && hardcount < max) {
def8e4ea
JH
5633 scan += UTF8SKIP(scan);
5634 hardcount++;
5635 }
5636 }
5637 else
5638 scan = loceol;
a0ed51b3 5639 break;
f33976b4
DB
5640 case CANY:
5641 scan = loceol;
5642 break;
090f7165
JH
5643 case EXACT: /* length of string is 1 */
5644 c = (U8)*STRING(p);
5645 while (scan < loceol && UCHARAT(scan) == c)
5646 scan++;
bbce6d69 5647 break;
5648 case EXACTF: /* length of string is 1 */
cd439c50 5649 c = (U8)*STRING(p);
bbce6d69 5650 while (scan < loceol &&
22c35a8c 5651 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c]))
bbce6d69 5652 scan++;
5653 break;
5654 case EXACTFL: /* length of string is 1 */
3280af22 5655 PL_reg_flags |= RF_tainted;
cd439c50 5656 c = (U8)*STRING(p);
bbce6d69 5657 while (scan < loceol &&
22c35a8c 5658 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c]))
a0d0e21e
LW
5659 scan++;
5660 break;
5661 case ANYOF:
f2ed9b32 5662 if (utf8_target) {
ffc61ed2 5663 loceol = PL_regeol;
cfc92286 5664 while (hardcount < max && scan < loceol &&
f2ed9b32 5665 reginclass(prog, p, (U8*)scan, 0, utf8_target)) {
ffc61ed2
JH
5666 scan += UTF8SKIP(scan);
5667 hardcount++;
5668 }
5669 } else {
32fc9b6a 5670 while (scan < loceol && REGINCLASS(prog, p, (U8*)scan))
ffc61ed2
JH
5671 scan++;
5672 }
a0d0e21e
LW
5673 break;
5674 case ALNUM:
f2ed9b32 5675 if (utf8_target) {
ffc61ed2 5676 loceol = PL_regeol;
1a4fad37 5677 LOAD_UTF8_CHARCLASS_ALNUM();
1aa99e6b 5678 while (hardcount < max && scan < loceol &&
f2ed9b32 5679 swash_fetch(PL_utf8_alnum, (U8*)scan, utf8_target)) {
ffc61ed2
JH
5680 scan += UTF8SKIP(scan);
5681 hardcount++;
5682 }
5683 } else {
5684 while (scan < loceol && isALNUM(*scan))
5685 scan++;
a0ed51b3
LW
5686 }
5687 break;
bbce6d69 5688 case ALNUML:
3280af22 5689 PL_reg_flags |= RF_tainted;
f2ed9b32 5690 if (utf8_target) {
ffc61ed2 5691 loceol = PL_regeol;
1aa99e6b
IH
5692 while (hardcount < max && scan < loceol &&
5693 isALNUM_LC_utf8((U8*)scan)) {
ffc61ed2
JH
5694 scan += UTF8SKIP(scan);
5695 hardcount++;
5696 }
5697 } else {
5698 while (scan < loceol && isALNUM_LC(*scan))
5699 scan++;
a0ed51b3
LW
5700 }
5701 break;
a0d0e21e 5702 case NALNUM:
f2ed9b32 5703 if (utf8_target) {
ffc61ed2 5704 loceol = PL_regeol;
1a4fad37 5705 LOAD_UTF8_CHARCLASS_ALNUM();
1aa99e6b 5706 while (hardcount < max && scan < loceol &&
f2ed9b32 5707 !swash_fetch(PL_utf8_alnum, (U8*)scan, utf8_target)) {
ffc61ed2
JH
5708 scan += UTF8SKIP(scan);
5709 hardcount++;
5710 }
5711 } else {
5712 while (scan < loceol && !isALNUM(*scan))
5713 scan++;
a0ed51b3
LW
5714 }
5715 break;
bbce6d69 5716 case NALNUML:
3280af22 5717 PL_reg_flags |= RF_tainted;
f2ed9b32 5718 if (utf8_target) {
ffc61ed2 5719 loceol = PL_regeol;
1aa99e6b
IH
5720 while (hardcount < max && scan < loceol &&
5721 !isALNUM_LC_utf8((U8*)scan)) {
ffc61ed2
JH
5722 scan += UTF8SKIP(scan);
5723 hardcount++;
5724 }
5725 } else {
5726 while (scan < loceol && !isALNUM_LC(*scan))
5727 scan++;
a0ed51b3
LW
5728 }
5729 break;
a0d0e21e 5730 case SPACE:
f2ed9b32 5731 if (utf8_target) {
ffc61ed2 5732 loceol = PL_regeol;
1a4fad37 5733 LOAD_UTF8_CHARCLASS_SPACE();
1aa99e6b 5734 while (hardcount < max && scan < loceol &&
3568d838 5735 (*scan == ' ' ||
f2ed9b32 5736 swash_fetch(PL_utf8_space,(U8*)scan, utf8_target))) {
ffc61ed2
JH
5737 scan += UTF8SKIP(scan);
5738 hardcount++;
5739 }
5740 } else {
5741 while (scan < loceol && isSPACE(*scan))
5742 scan++;
a0ed51b3
LW
5743 }
5744 break;
bbce6d69 5745 case SPACEL:
3280af22 5746 PL_reg_flags |= RF_tainted;
f2ed9b32 5747 if (utf8_target) {
ffc61ed2 5748 loceol = PL_regeol;
1aa99e6b 5749 while (hardcount < max && scan < loceol &&
ffc61ed2
JH
5750 (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
5751 scan += UTF8SKIP(scan);
5752 hardcount++;
5753 }
5754 } else {
5755 while (scan < loceol && isSPACE_LC(*scan))
5756 scan++;
a0ed51b3
LW
5757 }
5758 break;
a0d0e21e 5759 case NSPACE:
f2ed9b32 5760 if (utf8_target) {
ffc61ed2 5761 loceol = PL_regeol;
1a4fad37 5762 LOAD_UTF8_CHARCLASS_SPACE();
1aa99e6b 5763 while (hardcount < max && scan < loceol &&
3568d838 5764 !(*scan == ' ' ||
f2ed9b32 5765 swash_fetch(PL_utf8_space,(U8*)scan, utf8_target))) {
ffc61ed2
JH
5766 scan += UTF8SKIP(scan);
5767 hardcount++;
5768 }
5769 } else {
5770 while (scan < loceol && !isSPACE(*scan))
5771 scan++;
a0ed51b3 5772 }
0008a298 5773 break;
bbce6d69 5774 case NSPACEL:
3280af22 5775 PL_reg_flags |= RF_tainted;
f2ed9b32 5776 if (utf8_target) {
ffc61ed2 5777 loceol = PL_regeol;
1aa99e6b 5778 while (hardcount < max && scan < loceol &&
ffc61ed2
JH
5779 !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
5780 scan += UTF8SKIP(scan);
5781 hardcount++;
5782 }
5783 } else {
5784 while (scan < loceol && !isSPACE_LC(*scan))
5785 scan++;
a0ed51b3
LW
5786 }
5787 break;
a0d0e21e 5788 case DIGIT:
f2ed9b32 5789 if (utf8_target) {
ffc61ed2 5790 loceol = PL_regeol;
1a4fad37 5791 LOAD_UTF8_CHARCLASS_DIGIT();
1aa99e6b 5792 while (hardcount < max && scan < loceol &&
f2ed9b32 5793 swash_fetch(PL_utf8_digit, (U8*)scan, utf8_target)) {
ffc61ed2
JH
5794 scan += UTF8SKIP(scan);
5795 hardcount++;
5796 }
5797 } else {
5798 while (scan < loceol && isDIGIT(*scan))
5799 scan++;
a0ed51b3
LW
5800 }
5801 break;
a0d0e21e 5802 case NDIGIT:
f2ed9b32 5803 if (utf8_target) {
ffc61ed2 5804 loceol = PL_regeol;
1a4fad37 5805 LOAD_UTF8_CHARCLASS_DIGIT();
1aa99e6b 5806 while (hardcount < max && scan < loceol &&
f2ed9b32 5807 !swash_fetch(PL_utf8_digit, (U8*)scan, utf8_target)) {
ffc61ed2
JH
5808 scan += UTF8SKIP(scan);
5809 hardcount++;
5810 }
5811 } else {
5812 while (scan < loceol && !isDIGIT(*scan))
5813 scan++;
a0ed51b3 5814 }
e1d1eefb 5815 case LNBREAK:
f2ed9b32 5816 if (utf8_target) {
e1d1eefb
YO
5817 loceol = PL_regeol;
5818 while (hardcount < max && scan < loceol && (c=is_LNBREAK_utf8(scan))) {
5819 scan += c;
5820 hardcount++;
5821 }
5822 } else {
5823 /*
5824 LNBREAK can match two latin chars, which is ok,
5825 because we have a null terminated string, but we
5826 have to use hardcount in this situation
5827 */
5828 while (scan < loceol && (c=is_LNBREAK_latin1(scan))) {
5829 scan+=c;
5830 hardcount++;
5831 }
5832 }
5833 break;
5834 case HORIZWS:
f2ed9b32 5835 if (utf8_target) {
e1d1eefb
YO
5836 loceol = PL_regeol;
5837 while (hardcount < max && scan < loceol && (c=is_HORIZWS_utf8(scan))) {
5838 scan += c;
5839 hardcount++;
5840 }
5841 } else {
5842 while (scan < loceol && is_HORIZWS_latin1(scan))
5843 scan++;
5844 }
a0ed51b3 5845 break;
e1d1eefb 5846 case NHORIZWS:
f2ed9b32 5847 if (utf8_target) {
e1d1eefb
YO
5848 loceol = PL_regeol;
5849 while (hardcount < max && scan < loceol && !is_HORIZWS_utf8(scan)) {
5850 scan += UTF8SKIP(scan);
5851 hardcount++;
5852 }
5853 } else {
5854 while (scan < loceol && !is_HORIZWS_latin1(scan))
5855 scan++;
5856
5857 }
5858 break;
5859 case VERTWS:
f2ed9b32 5860 if (utf8_target) {
e1d1eefb
YO
5861 loceol = PL_regeol;
5862 while (hardcount < max && scan < loceol && (c=is_VERTWS_utf8(scan))) {
5863 scan += c;
5864 hardcount++;
5865 }
5866 } else {
5867 while (scan < loceol && is_VERTWS_latin1(scan))
5868 scan++;
5869
5870 }
5871 break;
5872 case NVERTWS:
f2ed9b32 5873 if (utf8_target) {
e1d1eefb
YO
5874 loceol = PL_regeol;
5875 while (hardcount < max && scan < loceol && !is_VERTWS_utf8(scan)) {
5876 scan += UTF8SKIP(scan);
5877 hardcount++;
5878 }
5879 } else {
5880 while (scan < loceol && !is_VERTWS_latin1(scan))
5881 scan++;
5882
5883 }
5884 break;
5885
a0d0e21e
LW
5886 default: /* Called on something of 0 width. */
5887 break; /* So match right here or not at all. */
5888 }
a687059c 5889
a0ed51b3
LW
5890 if (hardcount)
5891 c = hardcount;
5892 else
5893 c = scan - PL_reginput;
3280af22 5894 PL_reginput = scan;
a687059c 5895
a3621e74 5896 DEBUG_r({
e68ec53f 5897 GET_RE_DEBUG_FLAGS_DECL;
be8e71aa 5898 DEBUG_EXECUTE_r({
e68ec53f
YO
5899 SV * const prop = sv_newmortal();
5900 regprop(prog, prop, p);
5901 PerlIO_printf(Perl_debug_log,
be8e71aa 5902 "%*s %s can match %"IVdf" times out of %"IVdf"...\n",
e2e6a0f1 5903 REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max);
a3621e74 5904 });
be8e71aa 5905 });
9041c2e3 5906
a0d0e21e 5907 return(c);
a687059c
LW
5908}
5909
c277df42 5910
be8e71aa 5911#if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION)
c277df42 5912/*
ffc61ed2
JH
5913- regclass_swash - prepare the utf8 swash
5914*/
5915
5916SV *
32fc9b6a 5917Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
ffc61ed2 5918{
97aff369 5919 dVAR;
9e55ce06
JH
5920 SV *sw = NULL;
5921 SV *si = NULL;
5922 SV *alt = NULL;
f8fc2ecf
YO
5923 RXi_GET_DECL(prog,progi);
5924 const struct reg_data * const data = prog ? progi->data : NULL;
ffc61ed2 5925
7918f24d
NC
5926 PERL_ARGS_ASSERT_REGCLASS_SWASH;
5927
4f639d21 5928 if (data && data->count) {
a3b680e6 5929 const U32 n = ARG(node);
ffc61ed2 5930
4f639d21 5931 if (data->what[n] == 's') {
ad64d0ec
NC
5932 SV * const rv = MUTABLE_SV(data->data[n]);
5933 AV * const av = MUTABLE_AV(SvRV(rv));
2d03de9c 5934 SV **const ary = AvARRAY(av);
9e55ce06 5935 SV **a, **b;
9041c2e3 5936
711a919c 5937 /* See the end of regcomp.c:S_regclass() for
9e55ce06
JH
5938 * documentation of these array elements. */
5939
b11f357e 5940 si = *ary;
fe5bfecd
JH
5941 a = SvROK(ary[1]) ? &ary[1] : NULL;
5942 b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : NULL;
b11f357e 5943
ffc61ed2
JH
5944 if (a)
5945 sw = *a;
5946 else if (si && doinit) {
5947 sw = swash_init("utf8", "", si, 1, 0);
5948 (void)av_store(av, 1, sw);
5949 }
9e55ce06
JH
5950 if (b)
5951 alt = *b;
ffc61ed2
JH
5952 }
5953 }
5954
9e55ce06
JH
5955 if (listsvp)
5956 *listsvp = si;
5957 if (altsvp)
5958 *altsvp = alt;
ffc61ed2
JH
5959
5960 return sw;
5961}
76234dfb 5962#endif
ffc61ed2
JH
5963
5964/*
ba7b4546 5965 - reginclass - determine if a character falls into a character class
832705d4
JH
5966
5967 The n is the ANYOF regnode, the p is the target string, lenp
5968 is pointer to the maximum length of how far to go in the p
5969 (if the lenp is zero, UTF8SKIP(p) is used),
f2ed9b32 5970 utf8_target tells whether the target string is in UTF-8.
832705d4 5971
bbce6d69 5972 */
5973
76e3520e 5974STATIC bool
f2ed9b32 5975S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool utf8_target)
bbce6d69 5976{
27da23d5 5977 dVAR;
a3b680e6 5978 const char flags = ANYOF_FLAGS(n);
bbce6d69 5979 bool match = FALSE;
cc07378b 5980 UV c = *p;
ae9ddab8 5981 STRLEN len = 0;
9e55ce06 5982 STRLEN plen;
1aa99e6b 5983
7918f24d
NC
5984 PERL_ARGS_ASSERT_REGINCLASS;
5985
f2ed9b32 5986 if (utf8_target && !UTF8_IS_INVARIANT(c)) {
19f67299 5987 c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len,
6182169b
KW
5988 (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV)
5989 | UTF8_ALLOW_FFFF | UTF8_CHECK_ONLY);
5990 /* see [perl #37836] for UTF8_ALLOW_ANYUV; [perl #38293] for
5991 * UTF8_ALLOW_FFFF */
e8a70c6f
SP
5992 if (len == (STRLEN)-1)
5993 Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)");
19f67299 5994 }
bbce6d69 5995
0f0076b4 5996 plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c));
f2ed9b32 5997 if (utf8_target || (flags & ANYOF_UNICODE)) {
9e55ce06
JH
5998 if (lenp)
5999 *lenp = 0;
f2ed9b32 6000 if (utf8_target && !ANYOF_RUNTIME(n)) {
ffc61ed2
JH
6001 if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c))
6002 match = TRUE;
bbce6d69 6003 }
f2ed9b32 6004 if (!match && utf8_target && (flags & ANYOF_UNICODE_ALL) && c >= 256)
1aa99e6b 6005 match = TRUE;
ffc61ed2 6006 if (!match) {
9e55ce06 6007 AV *av;
32fc9b6a 6008 SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av);
ffc61ed2
JH
6009
6010 if (sw) {
3f0c5693 6011 U8 * utf8_p;
f2ed9b32 6012 if (utf8_target) {
3f0c5693
KW
6013 utf8_p = (U8 *) p;
6014 } else {
6015 STRLEN len = 1;
6016 utf8_p = bytes_to_utf8(p, &len);
6017 }
6018 if (swash_fetch(sw, utf8_p, 1))
ffc61ed2
JH
6019 match = TRUE;
6020 else if (flags & ANYOF_FOLD) {
9e55ce06
JH
6021 if (!match && lenp && av) {
6022 I32 i;
9e55ce06 6023 for (i = 0; i <= av_len(av); i++) {
890ce7af 6024 SV* const sv = *av_fetch(av, i, FALSE);
9e55ce06 6025 STRLEN len;
890ce7af 6026 const char * const s = SvPV_const(sv, len);
3f0c5693 6027 if (len <= plen && memEQ(s, (char*)utf8_p, len)) {
9e55ce06
JH
6028 *lenp = len;
6029 match = TRUE;
6030 break;
6031 }
6032 }
6033 }
6034 if (!match) {
89ebb4a3 6035 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
4a623e43 6036
3f0c5693
KW
6037 STRLEN tmplen;
6038 to_utf8_fold(utf8_p, tmpbuf, &tmplen);
6039 if (swash_fetch(sw, tmpbuf, 1))
9e55ce06
JH
6040 match = TRUE;
6041 }
ffc61ed2 6042 }
b3a04dd3
KW
6043
6044 /* If we allocated a string above, free it */
f2ed9b32 6045 if (! utf8_target) Safefree(utf8_p);
ffc61ed2 6046 }
bbce6d69 6047 }
9e55ce06 6048 if (match && lenp && *lenp == 0)
0f0076b4 6049 *lenp = UNISKIP(NATIVE_TO_UNI(c));
bbce6d69 6050 }
1aa99e6b 6051 if (!match && c < 256) {
ffc61ed2
JH
6052 if (ANYOF_BITMAP_TEST(n, c))
6053 match = TRUE;
6054 else if (flags & ANYOF_FOLD) {
eb160463 6055 U8 f;
a0ed51b3 6056
ffc61ed2
JH
6057 if (flags & ANYOF_LOCALE) {
6058 PL_reg_flags |= RF_tainted;
6059 f = PL_fold_locale[c];
6060 }
6061 else
6062 f = PL_fold[c];
6063 if (f != c && ANYOF_BITMAP_TEST(n, f))
6064 match = TRUE;
6065 }
6066
6067 if (!match && (flags & ANYOF_CLASS)) {
a0ed51b3 6068 PL_reg_flags |= RF_tainted;
ffc61ed2
JH
6069 if (
6070 (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) ||
6071 (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) ||
6072 (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) ||
6073 (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) ||
6074 (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) ||
6075 (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) ||
6076 (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) ||
6077 (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
6078 (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) ||
6079 (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) ||
6080 (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) ||
6081 (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) ||
6082 (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) ||
6083 (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) ||
6084 (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) ||
6085 (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) ||
6086 (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) ||
6087 (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) ||
6088 (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) ||
6089 (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) ||
6090 (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) ||
6091 (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) ||
6092 (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) ||
6093 (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
6094 (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) ||
6095 (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) ||
6096 (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) ||
6097 (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) ||
6098 (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) ||
6099 (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c))
6100 ) /* How's that for a conditional? */
6101 {
6102 match = TRUE;
6103 }
a0ed51b3 6104 }
a0ed51b3
LW
6105 }
6106
a0ed51b3
LW
6107 return (flags & ANYOF_INVERT) ? !match : match;
6108}
161b471a 6109
dfe13c55 6110STATIC U8 *
0ce71af7 6111S_reghop3(U8 *s, I32 off, const U8* lim)
9041c2e3 6112{
97aff369 6113 dVAR;
7918f24d
NC
6114
6115 PERL_ARGS_ASSERT_REGHOP3;
6116
a0ed51b3 6117 if (off >= 0) {
1aa99e6b 6118 while (off-- && s < lim) {
ffc61ed2 6119 /* XXX could check well-formedness here */
a0ed51b3 6120 s += UTF8SKIP(s);
ffc61ed2 6121 }
a0ed51b3
LW
6122 }
6123 else {
1de06328
YO
6124 while (off++ && s > lim) {
6125 s--;
6126 if (UTF8_IS_CONTINUED(*s)) {
6127 while (s > lim && UTF8_IS_CONTINUATION(*s))
6128 s--;
a0ed51b3 6129 }
1de06328 6130 /* XXX could check well-formedness here */
a0ed51b3
LW
6131 }
6132 }
6133 return s;
6134}
161b471a 6135
f9f4320a
YO
6136#ifdef XXX_dmq
6137/* there are a bunch of places where we use two reghop3's that should
6138 be replaced with this routine. but since thats not done yet
6139 we ifdef it out - dmq
6140*/
dfe13c55 6141STATIC U8 *
1de06328
YO
6142S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim)
6143{
6144 dVAR;
7918f24d
NC
6145
6146 PERL_ARGS_ASSERT_REGHOP4;
6147
1de06328
YO
6148 if (off >= 0) {
6149 while (off-- && s < rlim) {
6150 /* XXX could check well-formedness here */
6151 s += UTF8SKIP(s);
6152 }
6153 }
6154 else {
6155 while (off++ && s > llim) {
6156 s--;
6157 if (UTF8_IS_CONTINUED(*s)) {
6158 while (s > llim && UTF8_IS_CONTINUATION(*s))
6159 s--;
6160 }
6161 /* XXX could check well-formedness here */
6162 }
6163 }
6164 return s;
6165}
f9f4320a 6166#endif
1de06328
YO
6167
6168STATIC U8 *
0ce71af7 6169S_reghopmaybe3(U8* s, I32 off, const U8* lim)
a0ed51b3 6170{
97aff369 6171 dVAR;
7918f24d
NC
6172
6173 PERL_ARGS_ASSERT_REGHOPMAYBE3;
6174
a0ed51b3 6175 if (off >= 0) {
1aa99e6b 6176 while (off-- && s < lim) {
ffc61ed2 6177 /* XXX could check well-formedness here */
a0ed51b3 6178 s += UTF8SKIP(s);
ffc61ed2 6179 }
a0ed51b3 6180 if (off >= 0)
3dab1dad 6181 return NULL;
a0ed51b3
LW
6182 }
6183 else {
1de06328
YO
6184 while (off++ && s > lim) {
6185 s--;
6186 if (UTF8_IS_CONTINUED(*s)) {
6187 while (s > lim && UTF8_IS_CONTINUATION(*s))
6188 s--;
a0ed51b3 6189 }
1de06328 6190 /* XXX could check well-formedness here */
a0ed51b3
LW
6191 }
6192 if (off <= 0)
3dab1dad 6193 return NULL;
a0ed51b3
LW
6194 }
6195 return s;
6196}
51371543 6197
51371543 6198static void
acfe0abc 6199restore_pos(pTHX_ void *arg)
51371543 6200{
97aff369 6201 dVAR;
097eb12c 6202 regexp * const rex = (regexp *)arg;
51371543
GS
6203 if (PL_reg_eval_set) {
6204 if (PL_reg_oldsaved) {
4f639d21
DM
6205 rex->subbeg = PL_reg_oldsaved;
6206 rex->sublen = PL_reg_oldsavedlen;
f8c7b90f 6207#ifdef PERL_OLD_COPY_ON_WRITE
4f639d21 6208 rex->saved_copy = PL_nrs;
ed252734 6209#endif
07bc277f 6210 RXp_MATCH_COPIED_on(rex);
51371543
GS
6211 }
6212 PL_reg_magic->mg_len = PL_reg_oldpos;
6213 PL_reg_eval_set = 0;
6214 PL_curpm = PL_reg_oldcurpm;
6215 }
6216}
33b8afdf
JH
6217
6218STATIC void
6219S_to_utf8_substr(pTHX_ register regexp *prog)
6220{
a1cac82e 6221 int i = 1;
7918f24d
NC
6222
6223 PERL_ARGS_ASSERT_TO_UTF8_SUBSTR;
6224
a1cac82e
NC
6225 do {
6226 if (prog->substrs->data[i].substr
6227 && !prog->substrs->data[i].utf8_substr) {
6228 SV* const sv = newSVsv(prog->substrs->data[i].substr);
6229 prog->substrs->data[i].utf8_substr = sv;
6230 sv_utf8_upgrade(sv);
610460f9
NC
6231 if (SvVALID(prog->substrs->data[i].substr)) {
6232 const U8 flags = BmFLAGS(prog->substrs->data[i].substr);
6233 if (flags & FBMcf_TAIL) {
6234 /* Trim the trailing \n that fbm_compile added last
6235 time. */
6236 SvCUR_set(sv, SvCUR(sv) - 1);
6237 /* Whilst this makes the SV technically "invalid" (as its
6238 buffer is no longer followed by "\0") when fbm_compile()
6239 adds the "\n" back, a "\0" is restored. */
6240 }
6241 fbm_compile(sv, flags);
6242 }
a1cac82e
NC
6243 if (prog->substrs->data[i].substr == prog->check_substr)
6244 prog->check_utf8 = sv;
6245 }
6246 } while (i--);
33b8afdf
JH
6247}
6248
6249STATIC void
6250S_to_byte_substr(pTHX_ register regexp *prog)
6251{
97aff369 6252 dVAR;
a1cac82e 6253 int i = 1;
7918f24d
NC
6254
6255 PERL_ARGS_ASSERT_TO_BYTE_SUBSTR;
6256
a1cac82e
NC
6257 do {
6258 if (prog->substrs->data[i].utf8_substr
6259 && !prog->substrs->data[i].substr) {
6260 SV* sv = newSVsv(prog->substrs->data[i].utf8_substr);
6261 if (sv_utf8_downgrade(sv, TRUE)) {
610460f9
NC
6262 if (SvVALID(prog->substrs->data[i].utf8_substr)) {
6263 const U8 flags
6264 = BmFLAGS(prog->substrs->data[i].utf8_substr);
6265 if (flags & FBMcf_TAIL) {
6266 /* Trim the trailing \n that fbm_compile added last
6267 time. */
6268 SvCUR_set(sv, SvCUR(sv) - 1);
6269 }
6270 fbm_compile(sv, flags);
6271 }
a1cac82e
NC
6272 } else {
6273 SvREFCNT_dec(sv);
6274 sv = &PL_sv_undef;
6275 }
6276 prog->substrs->data[i].substr = sv;
6277 if (prog->substrs->data[i].utf8_substr == prog->check_utf8)
6278 prog->check_substr = sv;
33b8afdf 6279 }
a1cac82e 6280 } while (i--);
33b8afdf 6281}
66610fdd
RGS
6282
6283/*
6284 * Local variables:
6285 * c-indentation-style: bsd
6286 * c-basic-offset: 4
6287 * indent-tabs-mode: t
6288 * End:
6289 *
37442d52
RGS
6290 * ex: set ts=8 sts=4 sw=4 noet:
6291 */