This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
reduce size of regmatch_state.u.curlyx by 2 words
[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
eb160463 88#define UTF ((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
33b8afdf 103#define CHR_SVLEN(sv) (do_utf8 ? 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; \
186 if (do_utf8 && UTF8_IS_CONTINUED(nextchr)) { \
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 \
f2338a2e 196 ? cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), (U8*)locinput, do_utf8)) \
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; \
217 if (do_utf8 && UTF8_IS_CONTINUED(nextchr)) { \
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 \
f2338a2e 227 ? cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), (U8*)locinput, do_utf8)) \
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;
1de06328 518 const bool do_utf8 = (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
288b8c02 531 RX_MATCH_UTF8_set(rx,do_utf8);
cad2e5aa 532
3c8556c3 533 if (RX_UTF8(rx)) {
b8d68ded
JH
534 PL_reg_flags |= RF_utf8;
535 }
ab3bbdeb 536 DEBUG_EXECUTE_r(
efd26800 537 debug_start_match(rx, do_utf8, 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;
33b8afdf
JH
551 if (do_utf8) {
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
YO
702 DEBUG_EXECUTE_r({
703 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
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"),
ab3bbdeb
YO
707 (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)
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
1de06328
YO
738 if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8)
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;
33b8afdf 745 if (check == (do_utf8 ? 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 */
1d86a7f9 755 && (!do_utf8
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. */
33b8afdf
JH
774 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
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
YO
787 DEBUG_EXECUTE_r({
788 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
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. */
33b8afdf
JH
835 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
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
YO
848 DEBUG_EXECUTE_r({
849 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
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 */
1d86a7f9 896 && (!do_utf8
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) {
33b8afdf 914 if (do_utf8 ? 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:
33b8afdf 954 ++BmUSEFUL(do_utf8 ? 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? */
33b8afdf
JH
978 && (do_utf8 ? (
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"));
ef8d46e8 990 /* XXX Does the destruction order has to change with do_utf8? */
33b8afdf 991 SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr);
ef8d46e8 992 SvREFCNT_dec(do_utf8 ? 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
JH
1050 if (prog->anchored_substr || prog->anchored_utf8) {
1051 if ((do_utf8 ? 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 }
33b8afdf 1091 if (!(do_utf8 ? 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
JH
1118 if (prog->check_substr || prog->check_utf8) /* could be removed already */
1119 BmUSEFUL(do_utf8 ? 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) \
1129 ? (do_utf8 ? trie_utf8_fold : (UTF ? trie_latin_utf8_fold : trie_plain)) \
1130 : (do_utf8 ? 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 || \
4c1b470c 1187 foldEQ_utf8(s, &my_strend, 0, do_utf8, \
f2338a2e 1188 m, NULL, ln, cBOOL(UTF))) \
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 || \
4c1b470c 1198 foldEQ_utf8(s, &my_strend, 0, do_utf8,\
f2338a2e 1199 m, NULL, ln, cBOOL(UTF)))\
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
YO
1263#define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \
1264 if (do_utf8) { \
1265 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1266 } \
1267 else { \
1268 REXEC_FBC_CLASS_SCAN(CoNd); \
1269 } \
1270 break
1271
4cadc6a9
YO
1272#define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
1273 if (do_utf8) { \
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; \
1284 if (do_utf8) { \
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 1310 unsigned int c1;
1311 unsigned int c2;
6eb5f6b9
JH
1312 char *e;
1313 register I32 tmp = 1; /* Scratch variable? */
a3b680e6 1314 register const bool do_utf8 = 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:
388cc4de 1322 if (do_utf8) {
4cadc6a9 1323 REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
388cc4de 1324 !UTF8_IS_INVARIANT((U8)s[0]) ?
32fc9b6a 1325 reginclass(prog, c, (U8*)s, 0, do_utf8) :
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 */
1aa99e6b 1360 if (UTF) {
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
a0a388a1 1421 if (do_utf8 || UTF) {
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) {
a0a388a1
YO
1431 if (do_utf8) {
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) {
a0a388a1
YO
1442 if (do_utf8) {
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:
ffc61ed2 1476 if (do_utf8) {
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 ?
f2338a2e 1488 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) :
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:
ffc61ed2 1514 if (do_utf8) {
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 ?
f2338a2e 1526 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) :
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
YO
1548 LOAD_UTF8_CHARCLASS_PERL_WORD(),
1549 swash_fetch(RE_utf8_perl_word, (U8*)s, do_utf8),
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
YO
1559 LOAD_UTF8_CHARCLASS_PERL_WORD(),
1560 !swash_fetch(RE_utf8_perl_word, (U8*)s, do_utf8),
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
YO
1570 LOAD_UTF8_CHARCLASS_PERL_SPACE(),
1571 *s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, do_utf8),
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
YO
1581 LOAD_UTF8_CHARCLASS_PERL_SPACE(),
1582 !(*s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, do_utf8)),
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
YO
1592 LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
1593 swash_fetch(RE_utf8_posix_digit,(U8*)s, do_utf8),
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
YO
1603 LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
1604 !swash_fetch(RE_utf8_posix_digit,(U8*)s, do_utf8),
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,
1726 (char *)uc, do_utf8 );
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,
1754 s, do_utf8 );
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,
1769 s, do_utf8 );
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;
f2338a2e 1880 const bool do_utf8 = 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
288b8c02 1899 RX_MATCH_UTF8_set(rx, do_utf8);
1de06328 1900 DEBUG_EXECUTE_r(
efd26800 1901 debug_start_match(rx, do_utf8, 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
JH
2058 /* we have /x+whatever/ */
2059 /* it must be a one character string (XXXX Except UTF?) */
33b8afdf 2060 char ch;
bf93d4cc
GS
2061#ifdef DEBUGGING
2062 int did_match = 0;
2063#endif
33b8afdf
JH
2064 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
2065 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
3f7c398e 2066 ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
bf93d4cc 2067
1aa99e6b 2068 if (do_utf8) {
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
JH
2107 if (prog->anchored_substr || prog->anchored_utf8) {
2108 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
2109 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
2110 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
2111 back_max = back_min = prog->anchored_offset;
2112 } else {
2113 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
2114 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
2115 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
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 }
1aa99e6b 2163 if (do_utf8) {
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
YO
2178 DEBUG_EXECUTE_r(if (!did_match) {
2179 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
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 {
02daf0ab 2199 RE_PV_QUOTED_DECL(quoted,do_utf8,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
2218 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
2219 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
2220 float_real = do_utf8 ? 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. */
1aa99e6b 2264 if (do_utf8) {
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({ \
5bc10b2c
DM
2686 DUMP_EXEC_POS(locinput, scan, do_utf8); \
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
efd26800 2704S_debug_start_match(pTHX_ const REGEXP *prog, const bool do_utf8,
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
YO
2716
2717 RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1),
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
2724 if (do_utf8||utf8_pat)
1de06328
YO
2725 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
2726 utf8_pat ? "pattern" : "",
2727 utf8_pat && do_utf8 ? " and " : "",
2728 do_utf8 ? "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,
2739 const bool do_utf8)
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
07be1b83
YO
2756 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
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);
07be1b83
YO
2762 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
2763 l--;
2764 if (pref0_len < 0)
2765 pref0_len = 0;
2766 if (pref0_len > pref_len)
2767 pref0_len = pref_len;
2768 {
3df15adc 2769 const int is_uni = (do_utf8 && 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;
95b24440 2856 register const bool do_utf8 = 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);
786e8c11 2945 DUMP_EXEC_POS( locinput, scan, do_utf8 );
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;
f33976b4
DB
3023 if (do_utf8) {
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;
3040 if (do_utf8) {
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 */
3056 if (scan->flags == EXACT || !do_utf8) {
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({
786e8c11 3190 DUMP_EXEC_POS( (char *)uc, scan, do_utf8 );
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) {
3319 if (do_utf8) {
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
DM
3340 else {
3341 if (do_utf8)
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);
eb160463 3397 if (do_utf8 != UTF) {
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
5ff6fc6d
JH
3402 if (do_utf8) {
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