This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Convert all accesses of the member paren_names of struct regexp to
[perl5.git] / regexec.c
CommitLineData
a0d0e21e
LW
1/* regexec.c
2 */
3
4/*
5 * "One Ring to rule them all, One Ring to find them..."
6 */
7
61296642
DM
8/* This file contains functions for executing a regular expression. See
9 * also regcomp.c which funnily enough, contains functions for compiling
166f8a29 10 * a regular expression.
e4a054ea
DM
11 *
12 * This file is also copied at build time to ext/re/re_exec.c, where
13 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
14 * This causes the main functions to be compiled under new names and with
15 * debugging support added, which makes "use re 'debug'" work.
166f8a29
DM
16 */
17
a687059c
LW
18/* NOTE: this is derived from Henry Spencer's regexp code, and should not
19 * confused with the original package (see point 3 below). Thanks, Henry!
20 */
21
22/* Additional note: this code is very heavily munged from Henry's version
23 * in places. In some spots I've traded clarity for efficiency, so don't
24 * blame Henry for some of the lack of readability.
25 */
26
e50aee73
AD
27/* The names of the functions have been changed from regcomp and
28 * regexec to pregcomp and pregexec in order to avoid conflicts
29 * with the POSIX routines of the same names.
30*/
31
b9d5759e 32#ifdef PERL_EXT_RE_BUILD
54df2634 33#include "re_top.h"
9041c2e3 34#endif
56953603 35
a687059c 36/*
e50aee73 37 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
38 *
39 * Copyright (c) 1986 by University of Toronto.
40 * Written by Henry Spencer. Not derived from licensed software.
41 *
42 * Permission is granted to anyone to use this software for any
43 * purpose on any computer system, and to redistribute it freely,
44 * subject to the following restrictions:
45 *
46 * 1. The author is not responsible for the consequences of use of
47 * this software, no matter how awful, even if they arise
48 * from defects in it.
49 *
50 * 2. The origin of this software must not be misrepresented, either
51 * by explicit claim or by omission.
52 *
53 * 3. Altered versions must be plainly marked as such, and must not
54 * be misrepresented as being the original software.
55 *
56 **** Alterations to Henry's code are...
57 ****
4bb101f2 58 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
fdf8c088 59 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007 by Larry Wall and others
a687059c 60 ****
9ef589d8
LW
61 **** You may distribute under the terms of either the GNU General Public
62 **** License or the Artistic License, as specified in the README file.
a687059c
LW
63 *
64 * Beware that some of this code is subtly aware of the way operator
65 * precedence is structured in regular expressions. Serious changes in
66 * regular-expression syntax might require a total rethink.
67 */
68#include "EXTERN.h"
864dbfa3 69#define PERL_IN_REGEXEC_C
a687059c 70#include "perl.h"
0f5d15d6 71
54df2634
NC
72#ifdef PERL_IN_XSUB_RE
73# include "re_comp.h"
74#else
75# include "regcomp.h"
76#endif
a687059c 77
c277df42
IZ
78#define RF_tainted 1 /* tainted information used? */
79#define RF_warned 2 /* warned about big count? */
faec1544 80
ab3bbdeb 81#define RF_utf8 8 /* Pattern contains multibyte chars? */
a0ed51b3 82
eb160463 83#define UTF ((PL_reg_flags & RF_utf8) != 0)
ce862d02
IZ
84
85#define RS_init 1 /* eval environment created */
86#define RS_set 2 /* replsv value is set */
c277df42 87
a687059c
LW
88#ifndef STATIC
89#define STATIC static
90#endif
91
32fc9b6a 92#define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
7d3e948e 93
c277df42
IZ
94/*
95 * Forwards.
96 */
97
33b8afdf 98#define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv))
53c4c00c 99#define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)
a0ed51b3 100
3dab1dad
YO
101#define HOPc(pos,off) \
102 (char *)(PL_reg_match_utf8 \
52657f30 103 ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \
3dab1dad
YO
104 : (U8*)(pos + off))
105#define HOPBACKc(pos, off) \
07be1b83
YO
106 (char*)(PL_reg_match_utf8\
107 ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \
108 : (pos - off >= PL_bostr) \
8e11feef 109 ? (U8*)pos - off \
3dab1dad 110 : NULL)
efb30f32 111
e7409c1b 112#define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
1aa99e6b 113#define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
1aa99e6b 114
1a4fad37
AL
115#define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \
116 if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END
117#define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
118#define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
119#define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ")
120#define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86")
51371543 121
3dab1dad
YO
122/* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
123
5f80c4cf 124/* for use after a quantifier and before an EXACT-like node -- japhy */
3e901dc0
YO
125/* it would be nice to rework regcomp.sym to generate this stuff. sigh */
126#define JUMPABLE(rn) ( \
127 OP(rn) == OPEN || \
128 (OP(rn) == CLOSE && (!cur_eval || cur_eval->u.eval.close_paren != ARG(rn))) || \
129 OP(rn) == EVAL || \
cca55fe3
JP
130 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
131 OP(rn) == PLUS || OP(rn) == MINMOD || \
ee9b8eae 132 OP(rn) == KEEPS || (PL_regkind[OP(rn)] == VERB) || \
3dab1dad 133 (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
e2d8ce26 134)
ee9b8eae 135#define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT)
e2d8ce26 136
ee9b8eae
YO
137#define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF )
138
139#if 0
140/* Currently these are only used when PL_regkind[OP(rn)] == EXACT so
141 we don't need this definition. */
142#define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF )
143#define IS_TEXTF(rn) ( OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF )
144#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL )
145
146#else
147/* ... so we use this as its faster. */
148#define IS_TEXT(rn) ( OP(rn)==EXACT )
149#define IS_TEXTF(rn) ( OP(rn)==EXACTF )
150#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL )
151
152#endif
e2d8ce26 153
a84d97b6
HS
154/*
155 Search for mandatory following text node; for lookahead, the text must
156 follow but for lookbehind (rn->flags != 0) we skip to the next step.
157*/
cca55fe3 158#define FIND_NEXT_IMPT(rn) STMT_START { \
3dab1dad
YO
159 while (JUMPABLE(rn)) { \
160 const OPCODE type = OP(rn); \
161 if (type == SUSPEND || PL_regkind[type] == CURLY) \
e2d8ce26 162 rn = NEXTOPER(NEXTOPER(rn)); \
3dab1dad 163 else if (type == PLUS) \
cca55fe3 164 rn = NEXTOPER(rn); \
3dab1dad 165 else if (type == IFMATCH) \
a84d97b6 166 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
e2d8ce26 167 else rn += NEXT_OFF(rn); \
3dab1dad 168 } \
5f80c4cf 169} STMT_END
74750237 170
c476f425 171
acfe0abc 172static void restore_pos(pTHX_ void *arg);
51371543 173
76e3520e 174STATIC CHECKPOINT
cea2e8a9 175S_regcppush(pTHX_ I32 parenfloor)
a0d0e21e 176{
97aff369 177 dVAR;
a3b680e6 178 const int retval = PL_savestack_ix;
b1ce53c5 179#define REGCP_PAREN_ELEMS 4
a3b680e6 180 const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
a0d0e21e 181 int p;
40a82448 182 GET_RE_DEBUG_FLAGS_DECL;
a0d0e21e 183
e49a9654
IH
184 if (paren_elems_to_push < 0)
185 Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
186
f0ab9afb 187#define REGCP_OTHER_ELEMS 7
4b3c1a47 188 SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
7f69552c 189
3280af22 190 for (p = PL_regsize; p > parenfloor; p--) {
b1ce53c5 191/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
f0ab9afb
NC
192 SSPUSHINT(PL_regoffs[p].end);
193 SSPUSHINT(PL_regoffs[p].start);
3280af22 194 SSPUSHPTR(PL_reg_start_tmp[p]);
a0d0e21e 195 SSPUSHINT(p);
e7707071 196 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
40a82448 197 " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n",
f0ab9afb 198 (UV)p, (IV)PL_regoffs[p].start,
40a82448 199 (IV)(PL_reg_start_tmp[p] - PL_bostr),
f0ab9afb 200 (IV)PL_regoffs[p].end
40a82448 201 ));
a0d0e21e 202 }
b1ce53c5 203/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
f0ab9afb 204 SSPUSHPTR(PL_regoffs);
3280af22
NIS
205 SSPUSHINT(PL_regsize);
206 SSPUSHINT(*PL_reglastparen);
a01268b5 207 SSPUSHINT(*PL_reglastcloseparen);
3280af22 208 SSPUSHPTR(PL_reginput);
41123dfd
JH
209#define REGCP_FRAME_ELEMS 2
210/* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and
211 * are needed for the regexp context stack bookkeeping. */
212 SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
b1ce53c5 213 SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */
41123dfd 214
a0d0e21e
LW
215 return retval;
216}
217
c277df42 218/* These are needed since we do not localize EVAL nodes: */
ab3bbdeb
YO
219#define REGCP_SET(cp) \
220 DEBUG_STATE_r( \
ab3bbdeb 221 PerlIO_printf(Perl_debug_log, \
e4f74956 222 " Setting an EVAL scope, savestack=%"IVdf"\n", \
ab3bbdeb
YO
223 (IV)PL_savestack_ix)); \
224 cp = PL_savestack_ix
c3464db5 225
ab3bbdeb 226#define REGCP_UNWIND(cp) \
e4f74956 227 DEBUG_STATE_r( \
ab3bbdeb 228 if (cp != PL_savestack_ix) \
e4f74956
YO
229 PerlIO_printf(Perl_debug_log, \
230 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
ab3bbdeb
YO
231 (IV)(cp), (IV)PL_savestack_ix)); \
232 regcpblow(cp)
c277df42 233
76e3520e 234STATIC char *
097eb12c 235S_regcppop(pTHX_ const regexp *rex)
a0d0e21e 236{
97aff369 237 dVAR;
3b6647e0 238 U32 i;
a0d0e21e 239 char *input;
b1ce53c5 240
a3621e74
YO
241 GET_RE_DEBUG_FLAGS_DECL;
242
b1ce53c5 243 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
a0d0e21e 244 i = SSPOPINT;
b1ce53c5
JH
245 assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
246 i = SSPOPINT; /* Parentheses elements to pop. */
a0d0e21e 247 input = (char *) SSPOPPTR;
a01268b5 248 *PL_reglastcloseparen = SSPOPINT;
3280af22
NIS
249 *PL_reglastparen = SSPOPINT;
250 PL_regsize = SSPOPINT;
f0ab9afb 251 PL_regoffs=(regexp_paren_pair *) SSPOPPTR;
b1ce53c5 252
6bda09f9 253
b1ce53c5 254 /* Now restore the parentheses context. */
41123dfd
JH
255 for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
256 i > 0; i -= REGCP_PAREN_ELEMS) {
1df70142 257 I32 tmps;
097eb12c 258 U32 paren = (U32)SSPOPINT;
3280af22 259 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
f0ab9afb 260 PL_regoffs[paren].start = SSPOPINT;
cf93c79d 261 tmps = SSPOPINT;
3280af22 262 if (paren <= *PL_reglastparen)
f0ab9afb 263 PL_regoffs[paren].end = tmps;
e7707071 264 DEBUG_BUFFERS_r(
c3464db5 265 PerlIO_printf(Perl_debug_log,
b900a521 266 " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
f0ab9afb 267 (UV)paren, (IV)PL_regoffs[paren].start,
b900a521 268 (IV)(PL_reg_start_tmp[paren] - PL_bostr),
f0ab9afb 269 (IV)PL_regoffs[paren].end,
3280af22 270 (paren > *PL_reglastparen ? "(no)" : ""));
c277df42 271 );
a0d0e21e 272 }
e7707071 273 DEBUG_BUFFERS_r(
bb7a0f54 274 if (*PL_reglastparen + 1 <= rex->nparens) {
c3464db5 275 PerlIO_printf(Perl_debug_log,
faccc32b 276 " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
4f639d21 277 (IV)(*PL_reglastparen + 1), (IV)rex->nparens);
c277df42
IZ
278 }
279 );
daf18116 280#if 1
dafc8851
JH
281 /* It would seem that the similar code in regtry()
282 * already takes care of this, and in fact it is in
283 * a better location to since this code can #if 0-ed out
284 * but the code in regtry() is needed or otherwise tests
285 * requiring null fields (pat.t#187 and split.t#{13,14}
daf18116
JH
286 * (as of patchlevel 7877) will fail. Then again,
287 * this code seems to be necessary or otherwise
225593e1
DM
288 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
289 * --jhi updated by dapm */
3b6647e0 290 for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) {
097eb12c 291 if (i > PL_regsize)
f0ab9afb
NC
292 PL_regoffs[i].start = -1;
293 PL_regoffs[i].end = -1;
a0d0e21e 294 }
dafc8851 295#endif
a0d0e21e
LW
296 return input;
297}
298
02db2b7b 299#define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
a0d0e21e 300
a687059c 301/*
e50aee73 302 * pregexec and friends
a687059c
LW
303 */
304
76234dfb 305#ifndef PERL_IN_XSUB_RE
a687059c 306/*
c277df42 307 - pregexec - match a regexp against a string
a687059c 308 */
c277df42 309I32
49d7dfbc 310Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, register char *strend,
c3464db5 311 char *strbeg, I32 minend, SV *screamer, U32 nosave)
c277df42
IZ
312/* strend: pointer to null at end of string */
313/* strbeg: real beginning of string */
314/* minend: end of match must be >=minend after stringarg. */
315/* nosave: For optimizations. */
316{
317 return
9041c2e3 318 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
c277df42
IZ
319 nosave ? 0 : REXEC_COPY_STR);
320}
76234dfb 321#endif
22e551b9 322
9041c2e3 323/*
cad2e5aa
JH
324 * Need to implement the following flags for reg_anch:
325 *
326 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
327 * USE_INTUIT_ML
328 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
329 * INTUIT_AUTORITATIVE_ML
330 * INTUIT_ONCE_NOML - Intuit can match in one location only.
331 * INTUIT_ONCE_ML
332 *
333 * Another flag for this function: SECOND_TIME (so that float substrs
334 * with giant delta may be not rechecked).
335 */
336
337/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
338
3f7c398e 339/* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend.
cad2e5aa
JH
340 Otherwise, only SvCUR(sv) is used to get strbeg. */
341
342/* XXXX We assume that strpos is strbeg unless sv. */
343
6eb5f6b9
JH
344/* XXXX Some places assume that there is a fixed substring.
345 An update may be needed if optimizer marks as "INTUITable"
346 RExen without fixed substrings. Similarly, it is assumed that
347 lengths of all the strings are no more than minlen, thus they
348 cannot come from lookahead.
40d049e4
YO
349 (Or minlen should take into account lookahead.)
350 NOTE: Some of this comment is not correct. minlen does now take account
351 of lookahead/behind. Further research is required. -- demerphq
352
353*/
6eb5f6b9 354
2c2d71f5
JH
355/* A failure to find a constant substring means that there is no need to make
356 an expensive call to REx engine, thus we celebrate a failure. Similarly,
357 finding a substring too deep into the string means that less calls to
30944b6d
IZ
358 regtry() should be needed.
359
360 REx compiler's optimizer found 4 possible hints:
361 a) Anchored substring;
362 b) Fixed substring;
363 c) Whether we are anchored (beginning-of-line or \G);
364 d) First node (of those at offset 0) which may distingush positions;
6eb5f6b9 365 We use a)b)d) and multiline-part of c), and try to find a position in the
30944b6d
IZ
366 string which does not contradict any of them.
367 */
2c2d71f5 368
6eb5f6b9
JH
369/* Most of decisions we do here should have been done at compile time.
370 The nodes of the REx which we used for the search should have been
371 deleted from the finite automaton. */
372
cad2e5aa 373char *
288b8c02 374Perl_re_intuit_start(pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a 375 char *strend, const U32 flags, re_scream_pos_data *data)
cad2e5aa 376{
97aff369 377 dVAR;
288b8c02 378 struct regexp *const prog = (struct regexp *)SvANY(rx);
b7953727 379 register I32 start_shift = 0;
cad2e5aa 380 /* Should be nonnegative! */
b7953727 381 register I32 end_shift = 0;
2c2d71f5
JH
382 register char *s;
383 register SV *check;
a1933d95 384 char *strbeg;
cad2e5aa 385 char *t;
1de06328 386 const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
cad2e5aa 387 I32 ml_anch;
bd61b366
SS
388 register char *other_last = NULL; /* other substr checked before this */
389 char *check_at = NULL; /* check substr found at this pos */
bbe252da 390 const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
f8fc2ecf 391 RXi_GET_DECL(prog,progi);
30944b6d 392#ifdef DEBUGGING
890ce7af 393 const char * const i_strpos = strpos;
30944b6d 394#endif
a3621e74
YO
395
396 GET_RE_DEBUG_FLAGS_DECL;
397
288b8c02 398 RX_MATCH_UTF8_set(rx,do_utf8);
cad2e5aa 399
3c8556c3 400 if (RX_UTF8(rx)) {
b8d68ded
JH
401 PL_reg_flags |= RF_utf8;
402 }
ab3bbdeb 403 DEBUG_EXECUTE_r(
efd26800 404 debug_start_match(rx, do_utf8, strpos, strend,
1de06328
YO
405 sv ? "Guessing start of match in sv for"
406 : "Guessing start of match in string for");
2a782b5b 407 );
cad2e5aa 408
c344f387
JH
409 /* CHR_DIST() would be more correct here but it makes things slow. */
410 if (prog->minlen > strend - strpos) {
a3621e74 411 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
a72c7584 412 "String too short... [re_intuit_start]\n"));
cad2e5aa 413 goto fail;
2c2d71f5 414 }
1de06328 415
a1933d95 416 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
1aa99e6b 417 PL_regeol = strend;
33b8afdf
JH
418 if (do_utf8) {
419 if (!prog->check_utf8 && prog->check_substr)
420 to_utf8_substr(prog);
421 check = prog->check_utf8;
422 } else {
423 if (!prog->check_substr && prog->check_utf8)
424 to_byte_substr(prog);
425 check = prog->check_substr;
426 }
1de06328 427 if (check == &PL_sv_undef) {
a3621e74 428 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1de06328 429 "Non-utf8 string cannot match utf8 check string\n"));
33b8afdf
JH
430 goto fail;
431 }
bbe252da
YO
432 if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */
433 ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE)
434 || ( (prog->extflags & RXf_ANCH_BOL)
7fba1cd6 435 && !multiline ) ); /* Check after \n? */
cad2e5aa 436
7e25d62c 437 if (!ml_anch) {
bbe252da
YO
438 if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */
439 && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */
3f7c398e 440 /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
7e25d62c
JH
441 && sv && !SvROK(sv)
442 && (strpos != strbeg)) {
a3621e74 443 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
7e25d62c
JH
444 goto fail;
445 }
446 if (prog->check_offset_min == prog->check_offset_max &&
bbe252da 447 !(prog->extflags & RXf_CANY_SEEN)) {
2c2d71f5 448 /* Substring at constant offset from beg-of-str... */
cad2e5aa
JH
449 I32 slen;
450
1aa99e6b 451 s = HOP3c(strpos, prog->check_offset_min, strend);
1de06328 452
653099ff
GS
453 if (SvTAIL(check)) {
454 slen = SvCUR(check); /* >= 1 */
cad2e5aa 455
9041c2e3 456 if ( strend - s > slen || strend - s < slen - 1
2c2d71f5 457 || (strend - s == slen && strend[-1] != '\n')) {
a3621e74 458 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
2c2d71f5 459 goto fail_finish;
cad2e5aa
JH
460 }
461 /* Now should match s[0..slen-2] */
462 slen--;
3f7c398e 463 if (slen && (*SvPVX_const(check) != *s
cad2e5aa 464 || (slen > 1
3f7c398e 465 && memNE(SvPVX_const(check), s, slen)))) {
2c2d71f5 466 report_neq:
a3621e74 467 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
2c2d71f5
JH
468 goto fail_finish;
469 }
cad2e5aa 470 }
3f7c398e 471 else if (*SvPVX_const(check) != *s
653099ff 472 || ((slen = SvCUR(check)) > 1
3f7c398e 473 && memNE(SvPVX_const(check), s, slen)))
2c2d71f5 474 goto report_neq;
c315bfe8 475 check_at = s;
2c2d71f5 476 goto success_at_start;
7e25d62c 477 }
cad2e5aa 478 }
2c2d71f5 479 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
cad2e5aa 480 s = strpos;
2c2d71f5 481 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
1de06328
YO
482 end_shift = prog->check_end_shift;
483
2c2d71f5 484 if (!ml_anch) {
a3b680e6 485 const I32 end = prog->check_offset_max + CHR_SVLEN(check)
653099ff 486 - (SvTAIL(check) != 0);
a3b680e6 487 const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
2c2d71f5
JH
488
489 if (end_shift < eshift)
490 end_shift = eshift;
491 }
cad2e5aa 492 }
2c2d71f5 493 else { /* Can match at random position */
cad2e5aa
JH
494 ml_anch = 0;
495 s = strpos;
1de06328
YO
496 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
497 end_shift = prog->check_end_shift;
498
499 /* end shift should be non negative here */
cad2e5aa
JH
500 }
501
bcdf7404 502#ifdef QDEBUGGING /* 7/99: reports of failure (with the older version) */
0033605d 503 if (end_shift < 0)
1de06328 504 Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
220fc49f 505 (IV)end_shift, RX_PRECOMP(prog));
2c2d71f5
JH
506#endif
507
2c2d71f5
JH
508 restart:
509 /* Find a possible match in the region s..strend by looking for
510 the "check" substring in the region corrected by start/end_shift. */
1de06328
YO
511
512 {
513 I32 srch_start_shift = start_shift;
514 I32 srch_end_shift = end_shift;
515 if (srch_start_shift < 0 && strbeg - s > srch_start_shift) {
516 srch_end_shift -= ((strbeg - s) - srch_start_shift);
517 srch_start_shift = strbeg - s;
518 }
6bda09f9 519 DEBUG_OPTIMISE_MORE_r({
1de06328
YO
520 PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n",
521 (IV)prog->check_offset_min,
522 (IV)srch_start_shift,
523 (IV)srch_end_shift,
524 (IV)prog->check_end_shift);
525 });
526
cad2e5aa 527 if (flags & REXEC_SCREAM) {
cad2e5aa 528 I32 p = -1; /* Internal iterator of scream. */
a3b680e6 529 I32 * const pp = data ? data->scream_pos : &p;
cad2e5aa 530
2c2d71f5
JH
531 if (PL_screamfirst[BmRARE(check)] >= 0
532 || ( BmRARE(check) == '\n'
85c508c3 533 && (BmPREVIOUS(check) == SvCUR(check) - 1)
2c2d71f5 534 && SvTAIL(check) ))
9041c2e3 535 s = screaminstr(sv, check,
1de06328 536 srch_start_shift + (s - strbeg), srch_end_shift, pp, 0);
cad2e5aa 537 else
2c2d71f5 538 goto fail_finish;
4addbd3b 539 /* we may be pointing at the wrong string */
07bc277f 540 if (s && RXp_MATCH_COPIED(prog))
3f7c398e 541 s = strbeg + (s - SvPVX_const(sv));
cad2e5aa
JH
542 if (data)
543 *data->scream_olds = s;
544 }
1de06328
YO
545 else {
546 U8* start_point;
547 U8* end_point;
bbe252da 548 if (prog->extflags & RXf_CANY_SEEN) {
1de06328
YO
549 start_point= (U8*)(s + srch_start_shift);
550 end_point= (U8*)(strend - srch_end_shift);
551 } else {
552 start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend);
553 end_point= HOP3(strend, -srch_end_shift, strbeg);
554 }
6bda09f9 555 DEBUG_OPTIMISE_MORE_r({
56570a2c 556 PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n",
1de06328 557 (int)(end_point - start_point),
fc8cd66c 558 (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point),
1de06328
YO
559 start_point);
560 });
561
562 s = fbm_instr( start_point, end_point,
7fba1cd6 563 check, multiline ? FBMrf_MULTILINE : 0);
1de06328
YO
564 }
565 }
cad2e5aa
JH
566 /* Update the count-of-usability, remove useless subpatterns,
567 unshift s. */
2c2d71f5 568
ab3bbdeb
YO
569 DEBUG_EXECUTE_r({
570 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
571 SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
572 PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s",
2c2d71f5 573 (s ? "Found" : "Did not find"),
ab3bbdeb
YO
574 (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)
575 ? "anchored" : "floating"),
576 quoted,
577 RE_SV_TAIL(check),
578 (s ? " at offset " : "...\n") );
579 });
2c2d71f5
JH
580
581 if (!s)
582 goto fail_finish;
2c2d71f5 583 /* Finish the diagnostic message */
a3621e74 584 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
2c2d71f5 585
1de06328
YO
586 /* XXX dmq: first branch is for positive lookbehind...
587 Our check string is offset from the beginning of the pattern.
588 So we need to do any stclass tests offset forward from that
589 point. I think. :-(
590 */
591
592
593
594 check_at=s;
595
596
2c2d71f5
JH
597 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
598 Start with the other substr.
599 XXXX no SCREAM optimization yet - and a very coarse implementation
a0288114 600 XXXX /ttx+/ results in anchored="ttx", floating="x". floating will
2c2d71f5
JH
601 *always* match. Probably should be marked during compile...
602 Probably it is right to do no SCREAM here...
603 */
604
1de06328
YO
605 if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8)
606 : (prog->float_substr && prog->anchored_substr))
607 {
30944b6d 608 /* Take into account the "other" substring. */
2c2d71f5
JH
609 /* XXXX May be hopelessly wrong for UTF... */
610 if (!other_last)
6eb5f6b9 611 other_last = strpos;
33b8afdf 612 if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) {
30944b6d
IZ
613 do_other_anchored:
614 {
890ce7af
AL
615 char * const last = HOP3c(s, -start_shift, strbeg);
616 char *last1, *last2;
be8e71aa 617 char * const saved_s = s;
33b8afdf 618 SV* must;
2c2d71f5 619
2c2d71f5
JH
620 t = s - prog->check_offset_max;
621 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
1d86a7f9 622 && (!do_utf8
0ce71af7 623 || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos))
2c2d71f5 624 && t > strpos)))
6f207bd3 625 NOOP;
2c2d71f5
JH
626 else
627 t = strpos;
1aa99e6b 628 t = HOP3c(t, prog->anchored_offset, strend);
6eb5f6b9
JH
629 if (t < other_last) /* These positions already checked */
630 t = other_last;
1aa99e6b 631 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
2c2d71f5
JH
632 if (last < last1)
633 last1 = last;
1de06328
YO
634 /* XXXX It is not documented what units *_offsets are in.
635 We assume bytes, but this is clearly wrong.
636 Meaning this code needs to be carefully reviewed for errors.
637 dmq.
638 */
639
2c2d71f5 640 /* On end-of-str: see comment below. */
33b8afdf
JH
641 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
642 if (must == &PL_sv_undef) {
643 s = (char*)NULL;
1de06328 644 DEBUG_r(must = prog->anchored_utf8); /* for debug */
33b8afdf
JH
645 }
646 else
647 s = fbm_instr(
648 (unsigned char*)t,
649 HOP3(HOP3(last1, prog->anchored_offset, strend)
650 + SvCUR(must), -(SvTAIL(must)!=0), strbeg),
651 must,
7fba1cd6 652 multiline ? FBMrf_MULTILINE : 0
33b8afdf 653 );
ab3bbdeb
YO
654 DEBUG_EXECUTE_r({
655 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
656 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
657 PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s",
2c2d71f5 658 (s ? "Found" : "Contradicts"),
ab3bbdeb
YO
659 quoted, RE_SV_TAIL(must));
660 });
661
662
2c2d71f5
JH
663 if (!s) {
664 if (last1 >= last2) {
a3621e74 665 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2c2d71f5
JH
666 ", giving up...\n"));
667 goto fail_finish;
668 }
a3621e74 669 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2c2d71f5 670 ", trying floating at offset %ld...\n",
be8e71aa 671 (long)(HOP3c(saved_s, 1, strend) - i_strpos)));
1aa99e6b
IH
672 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
673 s = HOP3c(last, 1, strend);
2c2d71f5
JH
674 goto restart;
675 }
676 else {
a3621e74 677 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
30944b6d 678 (long)(s - i_strpos)));
1aa99e6b
IH
679 t = HOP3c(s, -prog->anchored_offset, strbeg);
680 other_last = HOP3c(s, 1, strend);
be8e71aa 681 s = saved_s;
2c2d71f5
JH
682 if (t == strpos)
683 goto try_at_start;
2c2d71f5
JH
684 goto try_at_offset;
685 }
30944b6d 686 }
2c2d71f5
JH
687 }
688 else { /* Take into account the floating substring. */
33b8afdf 689 char *last, *last1;
be8e71aa 690 char * const saved_s = s;
33b8afdf
JH
691 SV* must;
692
693 t = HOP3c(s, -start_shift, strbeg);
694 last1 = last =
695 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
696 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
697 last = HOP3c(t, prog->float_max_offset, strend);
698 s = HOP3c(t, prog->float_min_offset, strend);
699 if (s < other_last)
700 s = other_last;
2c2d71f5 701 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
33b8afdf
JH
702 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
703 /* fbm_instr() takes into account exact value of end-of-str
704 if the check is SvTAIL(ed). Since false positives are OK,
705 and end-of-str is not later than strend we are OK. */
706 if (must == &PL_sv_undef) {
707 s = (char*)NULL;
1de06328 708 DEBUG_r(must = prog->float_utf8); /* for debug message */
33b8afdf
JH
709 }
710 else
2c2d71f5 711 s = fbm_instr((unsigned char*)s,
33b8afdf
JH
712 (unsigned char*)last + SvCUR(must)
713 - (SvTAIL(must)!=0),
7fba1cd6 714 must, multiline ? FBMrf_MULTILINE : 0);
ab3bbdeb
YO
715 DEBUG_EXECUTE_r({
716 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
717 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
718 PerlIO_printf(Perl_debug_log, "%s floating substr %s%s",
33b8afdf 719 (s ? "Found" : "Contradicts"),
ab3bbdeb
YO
720 quoted, RE_SV_TAIL(must));
721 });
33b8afdf
JH
722 if (!s) {
723 if (last1 == last) {
a3621e74 724 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf
JH
725 ", giving up...\n"));
726 goto fail_finish;
2c2d71f5 727 }
a3621e74 728 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
33b8afdf 729 ", trying anchored starting at offset %ld...\n",
be8e71aa 730 (long)(saved_s + 1 - i_strpos)));
33b8afdf
JH
731 other_last = last;
732 s = HOP3c(t, 1, strend);
733 goto restart;
734 }
735 else {
a3621e74 736 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
33b8afdf
JH
737 (long)(s - i_strpos)));
738 other_last = s; /* Fix this later. --Hugo */
be8e71aa 739 s = saved_s;
33b8afdf
JH
740 if (t == strpos)
741 goto try_at_start;
742 goto try_at_offset;
743 }
2c2d71f5 744 }
cad2e5aa 745 }
2c2d71f5 746
1de06328 747
9ef43ace 748 t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos);
1de06328 749
6bda09f9 750 DEBUG_OPTIMISE_MORE_r(
1de06328
YO
751 PerlIO_printf(Perl_debug_log,
752 "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n",
753 (IV)prog->check_offset_min,
754 (IV)prog->check_offset_max,
755 (IV)(s-strpos),
756 (IV)(t-strpos),
757 (IV)(t-s),
758 (IV)(strend-strpos)
759 )
760 );
761
2c2d71f5 762 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
1d86a7f9 763 && (!do_utf8
9ef43ace 764 || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos)))
1de06328
YO
765 && t > strpos)))
766 {
2c2d71f5
JH
767 /* Fixed substring is found far enough so that the match
768 cannot start at strpos. */
769 try_at_offset:
cad2e5aa 770 if (ml_anch && t[-1] != '\n') {
30944b6d
IZ
771 /* Eventually fbm_*() should handle this, but often
772 anchored_offset is not 0, so this check will not be wasted. */
773 /* XXXX In the code below we prefer to look for "^" even in
774 presence of anchored substrings. And we search even
775 beyond the found float position. These pessimizations
776 are historical artefacts only. */
777 find_anchor:
2c2d71f5 778 while (t < strend - prog->minlen) {
cad2e5aa 779 if (*t == '\n') {
4ee3650e 780 if (t < check_at - prog->check_offset_min) {
33b8afdf 781 if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) {
4ee3650e
GS
782 /* Since we moved from the found position,
783 we definitely contradict the found anchored
30944b6d
IZ
784 substr. Due to the above check we do not
785 contradict "check" substr.
786 Thus we can arrive here only if check substr
787 is float. Redo checking for "other"=="fixed".
788 */
9041c2e3 789 strpos = t + 1;
a3621e74 790 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
e4584336 791 PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
30944b6d
IZ
792 goto do_other_anchored;
793 }
4ee3650e
GS
794 /* We don't contradict the found floating substring. */
795 /* XXXX Why not check for STCLASS? */
cad2e5aa 796 s = t + 1;
a3621e74 797 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
e4584336 798 PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
cad2e5aa
JH
799 goto set_useful;
800 }
4ee3650e
GS
801 /* Position contradicts check-string */
802 /* XXXX probably better to look for check-string
803 than for "\n", so one should lower the limit for t? */
a3621e74 804 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
e4584336 805 PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
0e41cd87 806 other_last = strpos = s = t + 1;
cad2e5aa
JH
807 goto restart;
808 }
809 t++;
810 }
a3621e74 811 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
e4584336 812 PL_colors[0], PL_colors[1]));
2c2d71f5 813 goto fail_finish;
cad2e5aa 814 }
f5952150 815 else {
a3621e74 816 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
e4584336 817 PL_colors[0], PL_colors[1]));
f5952150 818 }
cad2e5aa
JH
819 s = t;
820 set_useful:
33b8afdf 821 ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
cad2e5aa
JH
822 }
823 else {
f5952150 824 /* The found string does not prohibit matching at strpos,
2c2d71f5 825 - no optimization of calling REx engine can be performed,
f5952150
GS
826 unless it was an MBOL and we are not after MBOL,
827 or a future STCLASS check will fail this. */
2c2d71f5
JH
828 try_at_start:
829 /* Even in this situation we may use MBOL flag if strpos is offset
830 wrt the start of the string. */
05b4157f 831 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
a1933d95 832 && (strpos != strbeg) && strpos[-1] != '\n'
d506a20d 833 /* May be due to an implicit anchor of m{.*foo} */
bbe252da 834 && !(prog->intflags & PREGf_IMPLICIT))
d506a20d 835 {
cad2e5aa
JH
836 t = strpos;
837 goto find_anchor;
838 }
a3621e74 839 DEBUG_EXECUTE_r( if (ml_anch)
f5952150 840 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
70685ca0 841 (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
30944b6d 842 );
2c2d71f5 843 success_at_start:
bbe252da 844 if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */
33b8afdf
JH
845 && (do_utf8 ? (
846 prog->check_utf8 /* Could be deleted already */
847 && --BmUSEFUL(prog->check_utf8) < 0
848 && (prog->check_utf8 == prog->float_utf8)
849 ) : (
850 prog->check_substr /* Could be deleted already */
851 && --BmUSEFUL(prog->check_substr) < 0
852 && (prog->check_substr == prog->float_substr)
853 )))
66e933ab 854 {
cad2e5aa 855 /* If flags & SOMETHING - do not do it many times on the same match */
a3621e74 856 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
33b8afdf
JH
857 SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr);
858 if (do_utf8 ? prog->check_substr : prog->check_utf8)
859 SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8);
a0714e2c
SS
860 prog->check_substr = prog->check_utf8 = NULL; /* disable */
861 prog->float_substr = prog->float_utf8 = NULL; /* clear */
862 check = NULL; /* abort */
cad2e5aa 863 s = strpos;
3cf5c195
IZ
864 /* XXXX This is a remnant of the old implementation. It
865 looks wasteful, since now INTUIT can use many
6eb5f6b9 866 other heuristics. */
bbe252da 867 prog->extflags &= ~RXf_USE_INTUIT;
cad2e5aa
JH
868 }
869 else
870 s = strpos;
871 }
872
6eb5f6b9
JH
873 /* Last resort... */
874 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
1de06328
YO
875 /* trie stclasses are too expensive to use here, we are better off to
876 leave it to regmatch itself */
f8fc2ecf 877 if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
6eb5f6b9
JH
878 /* minlen == 0 is possible if regstclass is \b or \B,
879 and the fixed substr is ''$.
880 Since minlen is already taken into account, s+1 is before strend;
881 accidentally, minlen >= 1 guaranties no false positives at s + 1
882 even for \b or \B. But (minlen? 1 : 0) below assumes that
883 regstclass does not come from lookahead... */
884 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
885 This leaves EXACTF only, which is dealt with in find_byclass(). */
f8fc2ecf
YO
886 const U8* const str = (U8*)STRING(progi->regstclass);
887 const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
888 ? CHR_DIST(str+STR_LEN(progi->regstclass), str)
66e933ab 889 : 1);
1de06328
YO
890 char * endpos;
891 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
892 endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend);
893 else if (prog->float_substr || prog->float_utf8)
894 endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend);
895 else
896 endpos= strend;
897
70685ca0
JH
898 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n",
899 (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg)));
1de06328 900
6eb5f6b9 901 t = s;
f8fc2ecf 902 s = find_byclass(prog, progi->regstclass, s, endpos, NULL);
6eb5f6b9
JH
903 if (!s) {
904#ifdef DEBUGGING
cbbf8932 905 const char *what = NULL;
6eb5f6b9
JH
906#endif
907 if (endpos == strend) {
a3621e74 908 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6eb5f6b9
JH
909 "Could not match STCLASS...\n") );
910 goto fail;
911 }
a3621e74 912 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab 913 "This position contradicts STCLASS...\n") );
bbe252da 914 if ((prog->extflags & RXf_ANCH) && !ml_anch)
653099ff 915 goto fail;
6eb5f6b9 916 /* Contradict one of substrings */
33b8afdf
JH
917 if (prog->anchored_substr || prog->anchored_utf8) {
918 if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) {
a3621e74 919 DEBUG_EXECUTE_r( what = "anchored" );
6eb5f6b9 920 hop_and_restart:
1aa99e6b 921 s = HOP3c(t, 1, strend);
66e933ab
GS
922 if (s + start_shift + end_shift > strend) {
923 /* XXXX Should be taken into account earlier? */
a3621e74 924 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
66e933ab
GS
925 "Could not match STCLASS...\n") );
926 goto fail;
927 }
5e39e1e5
HS
928 if (!check)
929 goto giveup;
a3621e74 930 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 931 "Looking for %s substr starting at offset %ld...\n",
6eb5f6b9
JH
932 what, (long)(s + start_shift - i_strpos)) );
933 goto restart;
934 }
66e933ab 935 /* Have both, check_string is floating */
6eb5f6b9
JH
936 if (t + start_shift >= check_at) /* Contradicts floating=check */
937 goto retry_floating_check;
938 /* Recheck anchored substring, but not floating... */
9041c2e3 939 s = check_at;
5e39e1e5
HS
940 if (!check)
941 goto giveup;
a3621e74 942 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 943 "Looking for anchored substr starting at offset %ld...\n",
6eb5f6b9
JH
944 (long)(other_last - i_strpos)) );
945 goto do_other_anchored;
946 }
60e71179
GS
947 /* Another way we could have checked stclass at the
948 current position only: */
949 if (ml_anch) {
950 s = t = t + 1;
5e39e1e5
HS
951 if (!check)
952 goto giveup;
a3621e74 953 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
f5952150 954 "Looking for /%s^%s/m starting at offset %ld...\n",
e4584336 955 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
60e71179 956 goto try_at_offset;
66e933ab 957 }
33b8afdf 958 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
60e71179 959 goto fail;
6eb5f6b9
JH
960 /* Check is floating subtring. */
961 retry_floating_check:
962 t = check_at - start_shift;
a3621e74 963 DEBUG_EXECUTE_r( what = "floating" );
6eb5f6b9
JH
964 goto hop_and_restart;
965 }
b7953727 966 if (t != s) {
a3621e74 967 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
6eb5f6b9 968 "By STCLASS: moving %ld --> %ld\n",
b7953727
JH
969 (long)(t - i_strpos), (long)(s - i_strpos))
970 );
971 }
972 else {
a3621e74 973 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
b7953727
JH
974 "Does not contradict STCLASS...\n");
975 );
976 }
6eb5f6b9 977 }
5e39e1e5 978 giveup:
a3621e74 979 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
5e39e1e5
HS
980 PL_colors[4], (check ? "Guessed" : "Giving up"),
981 PL_colors[5], (long)(s - i_strpos)) );
cad2e5aa 982 return s;
2c2d71f5
JH
983
984 fail_finish: /* Substring not found */
33b8afdf
JH
985 if (prog->check_substr || prog->check_utf8) /* could be removed already */
986 BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
cad2e5aa 987 fail:
a3621e74 988 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
e4584336 989 PL_colors[4], PL_colors[5]));
bd61b366 990 return NULL;
cad2e5aa 991}
9661b544 992
a0a388a1
YO
993#define DECL_TRIE_TYPE(scan) \
994 const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold } \
995 trie_type = (scan->flags != EXACT) \
996 ? (do_utf8 ? trie_utf8_fold : (UTF ? trie_latin_utf8_fold : trie_plain)) \
997 : (do_utf8 ? trie_utf8 : trie_plain)
3b0527fe 998
55eed653
NC
999#define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \
1000uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \
4cadc6a9
YO
1001 switch (trie_type) { \
1002 case trie_utf8_fold: \
1003 if ( foldlen>0 ) { \
1004 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
1005 foldlen -= len; \
1006 uscan += len; \
1007 len=0; \
1008 } else { \
1009 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
1010 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
1011 foldlen -= UNISKIP( uvc ); \
1012 uscan = foldbuf + UNISKIP( uvc ); \
1013 } \
1014 break; \
a0a388a1
YO
1015 case trie_latin_utf8_fold: \
1016 if ( foldlen>0 ) { \
1017 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
1018 foldlen -= len; \
1019 uscan += len; \
1020 len=0; \
1021 } else { \
1022 len = 1; \
1023 uvc = to_uni_fold( *(U8*)uc, foldbuf, &foldlen ); \
1024 foldlen -= UNISKIP( uvc ); \
1025 uscan = foldbuf + UNISKIP( uvc ); \
1026 } \
1027 break; \
4cadc6a9
YO
1028 case trie_utf8: \
1029 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
1030 break; \
1031 case trie_plain: \
1032 uvc = (UV)*uc; \
1033 len = 1; \
1034 } \
1035 \
1036 if (uvc < 256) { \
1037 charid = trie->charmap[ uvc ]; \
1038 } \
1039 else { \
1040 charid = 0; \
55eed653
NC
1041 if (widecharmap) { \
1042 SV** const svpp = hv_fetch(widecharmap, \
4cadc6a9
YO
1043 (char*)&uvc, sizeof(UV), 0); \
1044 if (svpp) \
1045 charid = (U16)SvIV(*svpp); \
1046 } \
1047 } \
1048} STMT_END
1049
a0a388a1
YO
1050#define REXEC_FBC_EXACTISH_CHECK(CoNd) \
1051{ \
1052 char *my_strend= (char *)strend; \
4cadc6a9
YO
1053 if ( (CoNd) \
1054 && (ln == len || \
a0a388a1 1055 !ibcmp_utf8(s, &my_strend, 0, do_utf8, \
4cadc6a9 1056 m, NULL, ln, (bool)UTF)) \
a0a388a1 1057 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1058 goto got_it; \
1059 else { \
1060 U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \
1061 uvchr_to_utf8(tmpbuf, c); \
1062 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \
1063 if ( f != c \
1064 && (f == c1 || f == c2) \
a0a388a1
YO
1065 && (ln == len || \
1066 !ibcmp_utf8(s, &my_strend, 0, do_utf8,\
1067 m, NULL, ln, (bool)UTF)) \
1068 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1069 goto got_it; \
1070 } \
a0a388a1
YO
1071} \
1072s += len
4cadc6a9
YO
1073
1074#define REXEC_FBC_EXACTISH_SCAN(CoNd) \
1075STMT_START { \
1076 while (s <= e) { \
1077 if ( (CoNd) \
1078 && (ln == 1 || !(OP(c) == EXACTF \
1079 ? ibcmp(s, m, ln) \
1080 : ibcmp_locale(s, m, ln))) \
24b23f37 1081 && (!reginfo || regtry(reginfo, &s)) ) \
4cadc6a9
YO
1082 goto got_it; \
1083 s++; \
1084 } \
1085} STMT_END
1086
1087#define REXEC_FBC_UTF8_SCAN(CoDe) \
1088STMT_START { \
1089 while (s + (uskip = UTF8SKIP(s)) <= strend) { \
1090 CoDe \
1091 s += uskip; \
1092 } \
1093} STMT_END
1094
1095#define REXEC_FBC_SCAN(CoDe) \
1096STMT_START { \
1097 while (s < strend) { \
1098 CoDe \
1099 s++; \
1100 } \
1101} STMT_END
1102
1103#define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \
1104REXEC_FBC_UTF8_SCAN( \
1105 if (CoNd) { \
24b23f37 1106 if (tmp && (!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1107 goto got_it; \
1108 else \
1109 tmp = doevery; \
1110 } \
1111 else \
1112 tmp = 1; \
1113)
1114
1115#define REXEC_FBC_CLASS_SCAN(CoNd) \
1116REXEC_FBC_SCAN( \
1117 if (CoNd) { \
24b23f37 1118 if (tmp && (!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1119 goto got_it; \
1120 else \
1121 tmp = doevery; \
1122 } \
1123 else \
1124 tmp = 1; \
1125)
1126
1127#define REXEC_FBC_TRYIT \
24b23f37 1128if ((!reginfo || regtry(reginfo, &s))) \
4cadc6a9
YO
1129 goto got_it
1130
e1d1eefb
YO
1131#define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \
1132 if (do_utf8) { \
1133 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1134 } \
1135 else { \
1136 REXEC_FBC_CLASS_SCAN(CoNd); \
1137 } \
1138 break
1139
4cadc6a9
YO
1140#define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
1141 if (do_utf8) { \
1142 UtFpReLoAd; \
1143 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1144 } \
1145 else { \
1146 REXEC_FBC_CLASS_SCAN(CoNd); \
1147 } \
1148 break
1149
1150#define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \
1151 PL_reg_flags |= RF_tainted; \
1152 if (do_utf8) { \
1153 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1154 } \
1155 else { \
1156 REXEC_FBC_CLASS_SCAN(CoNd); \
1157 } \
1158 break
1159
786e8c11
YO
1160#define DUMP_EXEC_POS(li,s,doutf8) \
1161 dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
1162
1163/* We know what class REx starts with. Try to find this position... */
1164/* if reginfo is NULL, its a dryrun */
1165/* annoyingly all the vars in this routine have different names from their counterparts
1166 in regmatch. /grrr */
1167
3c3eec57 1168STATIC char *
07be1b83 1169S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
24b23f37 1170 const char *strend, regmatch_info *reginfo)
a687059c 1171{
27da23d5 1172 dVAR;
bbe252da 1173 const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
6eb5f6b9 1174 char *m;
d8093b23 1175 STRLEN ln;
5dab1207 1176 STRLEN lnc;
078c425b 1177 register STRLEN uskip;
d8093b23 1178 unsigned int c1;
1179 unsigned int c2;
6eb5f6b9
JH
1180 char *e;
1181 register I32 tmp = 1; /* Scratch variable? */
a3b680e6 1182 register const bool do_utf8 = PL_reg_match_utf8;
f8fc2ecf
YO
1183 RXi_GET_DECL(prog,progi);
1184
6eb5f6b9
JH
1185 /* We know what class it must start with. */
1186 switch (OP(c)) {
6eb5f6b9 1187 case ANYOF:
388cc4de 1188 if (do_utf8) {
4cadc6a9 1189 REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
388cc4de 1190 !UTF8_IS_INVARIANT((U8)s[0]) ?
32fc9b6a 1191 reginclass(prog, c, (U8*)s, 0, do_utf8) :
4cadc6a9 1192 REGINCLASS(prog, c, (U8*)s));
388cc4de
HS
1193 }
1194 else {
1195 while (s < strend) {
1196 STRLEN skip = 1;
1197
32fc9b6a 1198 if (REGINCLASS(prog, c, (U8*)s) ||
388cc4de
HS
1199 (ANYOF_FOLD_SHARP_S(c, s, strend) &&
1200 /* The assignment of 2 is intentional:
1201 * for the folded sharp s, the skip is 2. */
1202 (skip = SHARP_S_SKIP))) {
24b23f37 1203 if (tmp && (!reginfo || regtry(reginfo, &s)))
388cc4de
HS
1204 goto got_it;
1205 else
1206 tmp = doevery;
1207 }
1208 else
1209 tmp = 1;
1210 s += skip;
1211 }
a0d0e21e 1212 }
6eb5f6b9 1213 break;
f33976b4 1214 case CANY:
4cadc6a9 1215 REXEC_FBC_SCAN(
24b23f37 1216 if (tmp && (!reginfo || regtry(reginfo, &s)))
f33976b4
DB
1217 goto got_it;
1218 else
1219 tmp = doevery;
4cadc6a9 1220 );
f33976b4 1221 break;
6eb5f6b9 1222 case EXACTF:
5dab1207
NIS
1223 m = STRING(c);
1224 ln = STR_LEN(c); /* length to match in octets/bytes */
1225 lnc = (I32) ln; /* length to match in characters */
1aa99e6b 1226 if (UTF) {
a2a2844f 1227 STRLEN ulen1, ulen2;
5dab1207 1228 U8 *sm = (U8 *) m;
89ebb4a3
JH
1229 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
1230 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
97dc7d3e
RGS
1231 /* used by commented-out code below */
1232 /*const U32 uniflags = UTF8_ALLOW_DEFAULT;*/
a0a388a1
YO
1233
1234 /* XXX: Since the node will be case folded at compile
1235 time this logic is a little odd, although im not
1236 sure that its actually wrong. --dmq */
1237
1238 c1 = to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1239 c2 = to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1240
1241 /* XXX: This is kinda strange. to_utf8_XYZ returns the
1242 codepoint of the first character in the converted
1243 form, yet originally we did the extra step.
1244 No tests fail by commenting this code out however
1245 so Ive left it out. -- dmq.
1246
89ebb4a3 1247 c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
041457d9 1248 0, uniflags);
89ebb4a3 1249 c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
041457d9 1250 0, uniflags);
a0a388a1
YO
1251 */
1252
5dab1207
NIS
1253 lnc = 0;
1254 while (sm < ((U8 *) m + ln)) {
1255 lnc++;
1256 sm += UTF8SKIP(sm);
1257 }
1aa99e6b
IH
1258 }
1259 else {
1260 c1 = *(U8*)m;
1261 c2 = PL_fold[c1];
1262 }
6eb5f6b9
JH
1263 goto do_exactf;
1264 case EXACTFL:
5dab1207
NIS
1265 m = STRING(c);
1266 ln = STR_LEN(c);
1267 lnc = (I32) ln;
d8093b23 1268 c1 = *(U8*)m;
6eb5f6b9
JH
1269 c2 = PL_fold_locale[c1];
1270 do_exactf:
db12adc6 1271 e = HOP3c(strend, -((I32)lnc), s);
b3c9acc1 1272
3b0527fe 1273 if (!reginfo && e < s)
6eb5f6b9 1274 e = s; /* Due to minlen logic of intuit() */
1aa99e6b 1275
60a8b682
JH
1276 /* The idea in the EXACTF* cases is to first find the
1277 * first character of the EXACTF* node and then, if
1278 * necessary, case-insensitively compare the full
1279 * text of the node. The c1 and c2 are the first
1280 * characters (though in Unicode it gets a bit
1281 * more complicated because there are more cases
7f16dd3d
JH
1282 * than just upper and lower: one needs to use
1283 * the so-called folding case for case-insensitive
1284 * matching (called "loose matching" in Unicode).
1285 * ibcmp_utf8() will do just that. */
60a8b682 1286
a0a388a1 1287 if (do_utf8 || UTF) {
575cac57 1288 UV c, f;
89ebb4a3 1289 U8 tmpbuf [UTF8_MAXBYTES+1];
a0a388a1
YO
1290 STRLEN len = 1;
1291 STRLEN foldlen;
4ad0818d 1292 const U32 uniflags = UTF8_ALLOW_DEFAULT;
09091399 1293 if (c1 == c2) {
5dab1207
NIS
1294 /* Upper and lower of 1st char are equal -
1295 * probably not a "letter". */
1aa99e6b 1296 while (s <= e) {
a0a388a1
YO
1297 if (do_utf8) {
1298 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
041457d9 1299 uniflags);
a0a388a1
YO
1300 } else {
1301 c = *((U8*)s);
1302 }
4cadc6a9 1303 REXEC_FBC_EXACTISH_CHECK(c == c1);
1aa99e6b 1304 }
09091399
JH
1305 }
1306 else {
1aa99e6b 1307 while (s <= e) {
a0a388a1
YO
1308 if (do_utf8) {
1309 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
041457d9 1310 uniflags);
a0a388a1
YO
1311 } else {
1312 c = *((U8*)s);
1313 }
80aecb99 1314
60a8b682 1315 /* Handle some of the three Greek sigmas cases.
8c01da3c
JH
1316 * Note that not all the possible combinations
1317 * are handled here: some of them are handled
1318 * by the standard folding rules, and some of
1319 * them (the character class or ANYOF cases)
1320 * are handled during compiletime in
1321 * regexec.c:S_regclass(). */
880bd946
JH
1322 if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
1323 c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
1324 c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
80aecb99 1325
4cadc6a9 1326 REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2);
1aa99e6b 1327 }
09091399 1328 }
1aa99e6b
IH
1329 }
1330 else {
a0a388a1 1331 /* Neither pattern nor string are UTF8 */
1aa99e6b 1332 if (c1 == c2)
4cadc6a9 1333 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
1aa99e6b 1334 else
4cadc6a9 1335 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
b3c9acc1
IZ
1336 }
1337 break;
bbce6d69 1338 case BOUNDL:
3280af22 1339 PL_reg_flags |= RF_tainted;
bbce6d69 1340 /* FALL THROUGH */
a0d0e21e 1341 case BOUND:
ffc61ed2 1342 if (do_utf8) {
12d33761 1343 if (s == PL_bostr)
ffc61ed2
JH
1344 tmp = '\n';
1345 else {
6136c704 1346 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
4ad0818d 1347 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
ffc61ed2
JH
1348 }
1349 tmp = ((OP(c) == BOUND ?
9041c2e3 1350 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1a4fad37 1351 LOAD_UTF8_CHARCLASS_ALNUM();
4cadc6a9 1352 REXEC_FBC_UTF8_SCAN(
ffc61ed2 1353 if (tmp == !(OP(c) == BOUND ?
bb7a0f54 1354 (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
ffc61ed2
JH
1355 isALNUM_LC_utf8((U8*)s)))
1356 {
1357 tmp = !tmp;
4cadc6a9 1358 REXEC_FBC_TRYIT;
a687059c 1359 }
4cadc6a9 1360 );
a0d0e21e 1361 }
667bb95a 1362 else {
12d33761 1363 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2 1364 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
4cadc6a9 1365 REXEC_FBC_SCAN(
ffc61ed2
JH
1366 if (tmp ==
1367 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
1368 tmp = !tmp;
4cadc6a9 1369 REXEC_FBC_TRYIT;
a0ed51b3 1370 }
4cadc6a9 1371 );
a0ed51b3 1372 }
24b23f37 1373 if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))
a0ed51b3
LW
1374 goto got_it;
1375 break;
bbce6d69 1376 case NBOUNDL:
3280af22 1377 PL_reg_flags |= RF_tainted;
bbce6d69 1378 /* FALL THROUGH */
a0d0e21e 1379 case NBOUND:
ffc61ed2 1380 if (do_utf8) {
12d33761 1381 if (s == PL_bostr)
ffc61ed2
JH
1382 tmp = '\n';
1383 else {
6136c704 1384 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
4ad0818d 1385 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
ffc61ed2
JH
1386 }
1387 tmp = ((OP(c) == NBOUND ?
9041c2e3 1388 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1a4fad37 1389 LOAD_UTF8_CHARCLASS_ALNUM();
4cadc6a9 1390 REXEC_FBC_UTF8_SCAN(
ffc61ed2 1391 if (tmp == !(OP(c) == NBOUND ?
bb7a0f54 1392 (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
ffc61ed2
JH
1393 isALNUM_LC_utf8((U8*)s)))
1394 tmp = !tmp;
4cadc6a9
YO
1395 else REXEC_FBC_TRYIT;
1396 );
a0d0e21e 1397 }
667bb95a 1398 else {
12d33761 1399 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
ffc61ed2
JH
1400 tmp = ((OP(c) == NBOUND ?
1401 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
4cadc6a9 1402 REXEC_FBC_SCAN(
ffc61ed2
JH
1403 if (tmp ==
1404 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1405 tmp = !tmp;
4cadc6a9
YO
1406 else REXEC_FBC_TRYIT;
1407 );
a0ed51b3 1408 }
24b23f37 1409 if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s)))
a0ed51b3
LW
1410 goto got_it;
1411 break;
a0d0e21e 1412 case ALNUM:
4cadc6a9
YO
1413 REXEC_FBC_CSCAN_PRELOAD(
1414 LOAD_UTF8_CHARCLASS_ALNUM(),
1415 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8),
1416 isALNUM(*s)
1417 );
bbce6d69 1418 case ALNUML:
4cadc6a9
YO
1419 REXEC_FBC_CSCAN_TAINT(
1420 isALNUM_LC_utf8((U8*)s),
1421 isALNUM_LC(*s)
1422 );
a0d0e21e 1423 case NALNUM:
4cadc6a9
YO
1424 REXEC_FBC_CSCAN_PRELOAD(
1425 LOAD_UTF8_CHARCLASS_ALNUM(),
1426 !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8),
1427 !isALNUM(*s)
1428 );
bbce6d69 1429 case NALNUML:
4cadc6a9
YO
1430 REXEC_FBC_CSCAN_TAINT(
1431 !isALNUM_LC_utf8((U8*)s),
1432 !isALNUM_LC(*s)
1433 );
a0d0e21e 1434 case SPACE:
4cadc6a9
YO
1435 REXEC_FBC_CSCAN_PRELOAD(
1436 LOAD_UTF8_CHARCLASS_SPACE(),
1437 *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8),
1438 isSPACE(*s)
1439 );
bbce6d69 1440 case SPACEL:
4cadc6a9
YO
1441 REXEC_FBC_CSCAN_TAINT(
1442 *s == ' ' || isSPACE_LC_utf8((U8*)s),
1443 isSPACE_LC(*s)
1444 );
a0d0e21e 1445 case NSPACE:
4cadc6a9
YO
1446 REXEC_FBC_CSCAN_PRELOAD(
1447 LOAD_UTF8_CHARCLASS_SPACE(),
1448 !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)),
1449 !isSPACE(*s)
1450 );
bbce6d69 1451 case NSPACEL:
4cadc6a9
YO
1452 REXEC_FBC_CSCAN_TAINT(
1453 !(*s == ' ' || isSPACE_LC_utf8((U8*)s)),
1454 !isSPACE_LC(*s)
1455 );
a0d0e21e 1456 case DIGIT:
4cadc6a9
YO
1457 REXEC_FBC_CSCAN_PRELOAD(
1458 LOAD_UTF8_CHARCLASS_DIGIT(),
1459 swash_fetch(PL_utf8_digit,(U8*)s, do_utf8),
1460 isDIGIT(*s)
1461 );
b8c5462f 1462 case DIGITL:
4cadc6a9
YO
1463 REXEC_FBC_CSCAN_TAINT(
1464 isDIGIT_LC_utf8((U8*)s),
1465 isDIGIT_LC(*s)
1466 );
a0d0e21e 1467 case NDIGIT:
4cadc6a9
YO
1468 REXEC_FBC_CSCAN_PRELOAD(
1469 LOAD_UTF8_CHARCLASS_DIGIT(),
1470 !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8),
1471 !isDIGIT(*s)
1472 );
b8c5462f 1473 case NDIGITL:
4cadc6a9
YO
1474 REXEC_FBC_CSCAN_TAINT(
1475 !isDIGIT_LC_utf8((U8*)s),
1476 !isDIGIT_LC(*s)
1477 );
e1d1eefb
YO
1478 case LNBREAK:
1479 REXEC_FBC_CSCAN(
1480 is_LNBREAK_utf8(s),
1481 is_LNBREAK_latin1(s)
1482 );
1483 case VERTWS:
1484 REXEC_FBC_CSCAN(
1485 is_VERTWS_utf8(s),
1486 is_VERTWS_latin1(s)
1487 );
1488 case NVERTWS:
1489 REXEC_FBC_CSCAN(
1490 !is_VERTWS_utf8(s),
1491 !is_VERTWS_latin1(s)
1492 );
1493 case HORIZWS:
1494 REXEC_FBC_CSCAN(
1495 is_HORIZWS_utf8(s),
1496 is_HORIZWS_latin1(s)
1497 );
1498 case NHORIZWS:
1499 REXEC_FBC_CSCAN(
1500 !is_HORIZWS_utf8(s),
1501 !is_HORIZWS_latin1(s)
1502 );
1de06328
YO
1503 case AHOCORASICKC:
1504 case AHOCORASICK:
07be1b83 1505 {
a0a388a1 1506 DECL_TRIE_TYPE(c);
07be1b83
YO
1507 /* what trie are we using right now */
1508 reg_ac_data *aho
f8fc2ecf 1509 = (reg_ac_data*)progi->data->data[ ARG( c ) ];
3251b653
NC
1510 reg_trie_data *trie
1511 = (reg_trie_data*)progi->data->data[ aho->trie ];
55eed653 1512 HV *widecharmap = (HV*) progi->data->data[ aho->trie + 1 ];
07be1b83
YO
1513
1514 const char *last_start = strend - trie->minlen;
6148ee25 1515#ifdef DEBUGGING
07be1b83 1516 const char *real_start = s;
6148ee25 1517#endif
07be1b83 1518 STRLEN maxlen = trie->maxlen;
be8e71aa
YO
1519 SV *sv_points;
1520 U8 **points; /* map of where we were in the input string
786e8c11 1521 when reading a given char. For ASCII this
be8e71aa 1522 is unnecessary overhead as the relationship
38a44b82
NC
1523 is always 1:1, but for Unicode, especially
1524 case folded Unicode this is not true. */
f9e705e8 1525 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
786e8c11
YO
1526 U8 *bitmap=NULL;
1527
07be1b83
YO
1528
1529 GET_RE_DEBUG_FLAGS_DECL;
1530
be8e71aa
YO
1531 /* We can't just allocate points here. We need to wrap it in
1532 * an SV so it gets freed properly if there is a croak while
1533 * running the match */
1534 ENTER;
1535 SAVETMPS;
1536 sv_points=newSV(maxlen * sizeof(U8 *));
1537 SvCUR_set(sv_points,
1538 maxlen * sizeof(U8 *));
1539 SvPOK_on(sv_points);
1540 sv_2mortal(sv_points);
1541 points=(U8**)SvPV_nolen(sv_points );
1de06328
YO
1542 if ( trie_type != trie_utf8_fold
1543 && (trie->bitmap || OP(c)==AHOCORASICKC) )
1544 {
786e8c11
YO
1545 if (trie->bitmap)
1546 bitmap=(U8*)trie->bitmap;
1547 else
1548 bitmap=(U8*)ANYOF_BITMAP(c);
07be1b83 1549 }
786e8c11
YO
1550 /* this is the Aho-Corasick algorithm modified a touch
1551 to include special handling for long "unknown char"
1552 sequences. The basic idea being that we use AC as long
1553 as we are dealing with a possible matching char, when
1554 we encounter an unknown char (and we have not encountered
1555 an accepting state) we scan forward until we find a legal
1556 starting char.
1557 AC matching is basically that of trie matching, except
1558 that when we encounter a failing transition, we fall back
1559 to the current states "fail state", and try the current char
1560 again, a process we repeat until we reach the root state,
1561 state 1, or a legal transition. If we fail on the root state
1562 then we can either terminate if we have reached an accepting
1563 state previously, or restart the entire process from the beginning
1564 if we have not.
1565
1566 */
07be1b83
YO
1567 while (s <= last_start) {
1568 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1569 U8 *uc = (U8*)s;
1570 U16 charid = 0;
1571 U32 base = 1;
1572 U32 state = 1;
1573 UV uvc = 0;
1574 STRLEN len = 0;
1575 STRLEN foldlen = 0;
1576 U8 *uscan = (U8*)NULL;
1577 U8 *leftmost = NULL;
786e8c11
YO
1578#ifdef DEBUGGING
1579 U32 accepted_word= 0;
1580#endif
07be1b83
YO
1581 U32 pointpos = 0;
1582
1583 while ( state && uc <= (U8*)strend ) {
1584 int failed=0;
786e8c11
YO
1585 U32 word = aho->states[ state ].wordnum;
1586
1de06328
YO
1587 if( state==1 ) {
1588 if ( bitmap ) {
1589 DEBUG_TRIE_EXECUTE_r(
1590 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1591 dump_exec_pos( (char *)uc, c, strend, real_start,
1592 (char *)uc, do_utf8 );
1593 PerlIO_printf( Perl_debug_log,
1594 " Scanning for legal start char...\n");
1595 }
1596 );
1597 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1598 uc++;
786e8c11 1599 }
1de06328 1600 s= (char *)uc;
786e8c11 1601 }
786e8c11
YO
1602 if (uc >(U8*)last_start) break;
1603 }
1604
1605 if ( word ) {
1606 U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ];
1607 if (!leftmost || lpos < leftmost) {
1608 DEBUG_r(accepted_word=word);
07be1b83 1609 leftmost= lpos;
786e8c11 1610 }
07be1b83 1611 if (base==0) break;
786e8c11 1612
07be1b83
YO
1613 }
1614 points[pointpos++ % maxlen]= uc;
55eed653
NC
1615 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
1616 uscan, len, uvc, charid, foldlen,
1617 foldbuf, uniflags);
786e8c11
YO
1618 DEBUG_TRIE_EXECUTE_r({
1619 dump_exec_pos( (char *)uc, c, strend, real_start,
1620 s, do_utf8 );
07be1b83 1621 PerlIO_printf(Perl_debug_log,
786e8c11
YO
1622 " Charid:%3u CP:%4"UVxf" ",
1623 charid, uvc);
1624 });
07be1b83
YO
1625
1626 do {
6148ee25 1627#ifdef DEBUGGING
786e8c11 1628 word = aho->states[ state ].wordnum;
6148ee25 1629#endif
07be1b83
YO
1630 base = aho->states[ state ].trans.base;
1631
786e8c11
YO
1632 DEBUG_TRIE_EXECUTE_r({
1633 if (failed)
1634 dump_exec_pos( (char *)uc, c, strend, real_start,
1635 s, do_utf8 );
07be1b83 1636 PerlIO_printf( Perl_debug_log,
786e8c11
YO
1637 "%sState: %4"UVxf", word=%"UVxf,
1638 failed ? " Fail transition to " : "",
1639 (UV)state, (UV)word);
1640 });
07be1b83
YO
1641 if ( base ) {
1642 U32 tmp;
1643 if (charid &&
1644 (base + charid > trie->uniquecharcount )
1645 && (base + charid - 1 - trie->uniquecharcount
1646 < trie->lasttrans)
1647 && trie->trans[base + charid - 1 -
1648 trie->uniquecharcount].check == state
1649 && (tmp=trie->trans[base + charid - 1 -
1650 trie->uniquecharcount ].next))
1651 {
786e8c11
YO
1652 DEBUG_TRIE_EXECUTE_r(
1653 PerlIO_printf( Perl_debug_log," - legal\n"));
07be1b83
YO
1654 state = tmp;
1655 break;
1656 }
1657 else {
786e8c11
YO
1658 DEBUG_TRIE_EXECUTE_r(
1659 PerlIO_printf( Perl_debug_log," - fail\n"));
1660 failed = 1;
1661 state = aho->fail[state];
07be1b83
YO
1662 }
1663 }
1664 else {
1665 /* we must be accepting here */
786e8c11
YO
1666 DEBUG_TRIE_EXECUTE_r(
1667 PerlIO_printf( Perl_debug_log," - accepting\n"));
1668 failed = 1;
07be1b83
YO
1669 break;
1670 }
1671 } while(state);
786e8c11 1672 uc += len;
07be1b83
YO
1673 if (failed) {
1674 if (leftmost)
1675 break;
786e8c11 1676 if (!state) state = 1;
07be1b83
YO
1677 }
1678 }
1679 if ( aho->states[ state ].wordnum ) {
1680 U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ];
786e8c11
YO
1681 if (!leftmost || lpos < leftmost) {
1682 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
07be1b83 1683 leftmost = lpos;
786e8c11 1684 }
07be1b83 1685 }
07be1b83
YO
1686 if (leftmost) {
1687 s = (char*)leftmost;
786e8c11
YO
1688 DEBUG_TRIE_EXECUTE_r({
1689 PerlIO_printf(
70685ca0
JH
1690 Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
1691 (UV)accepted_word, (IV)(s - real_start)
786e8c11
YO
1692 );
1693 });
24b23f37 1694 if (!reginfo || regtry(reginfo, &s)) {
be8e71aa
YO
1695 FREETMPS;
1696 LEAVE;
07be1b83 1697 goto got_it;
be8e71aa 1698 }
07be1b83 1699 s = HOPc(s,1);
786e8c11
YO
1700 DEBUG_TRIE_EXECUTE_r({
1701 PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
1702 });
07be1b83 1703 } else {
786e8c11
YO
1704 DEBUG_TRIE_EXECUTE_r(
1705 PerlIO_printf( Perl_debug_log,"No match.\n"));
07be1b83
YO
1706 break;
1707 }
1708 }
be8e71aa
YO
1709 FREETMPS;
1710 LEAVE;
07be1b83
YO
1711 }
1712 break;
b3c9acc1 1713 default:
3c3eec57
GS
1714 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
1715 break;
d6a28714 1716 }
6eb5f6b9
JH
1717 return 0;
1718 got_it:
1719 return s;
1720}
1721
e6aad7ab 1722static void
fae667d5 1723S_swap_match_buff (pTHX_ regexp *prog) {
f0ab9afb 1724 regexp_paren_pair *t;
fae667d5 1725
28d8d7f4 1726 if (!prog->swap) {
fae667d5
YO
1727 /* We have to be careful. If the previous successful match
1728 was from this regex we don't want a subsequent paritally
1729 successful match to clobber the old results.
1730 So when we detect this possibility we add a swap buffer
1731 to the re, and switch the buffer each match. If we fail
1732 we switch it back, otherwise we leave it swapped.
1733 */
f0ab9afb 1734 Newxz(prog->swap, (prog->nparens + 1), regexp_paren_pair);
fae667d5 1735 }
f0ab9afb
NC
1736 t = prog->swap;
1737 prog->swap = prog->offs;
1738 prog->offs = t;
fae667d5
YO
1739}
1740
1741
6eb5f6b9
JH
1742/*
1743 - regexec_flags - match a regexp against a string
1744 */
1745I32
288b8c02 1746Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, register char *strend,
6eb5f6b9
JH
1747 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
1748/* strend: pointer to null at end of string */
1749/* strbeg: real beginning of string */
1750/* minend: end of match must be >=minend after stringarg. */
58e23c8d
YO
1751/* data: May be used for some additional optimizations.
1752 Currently its only used, with a U32 cast, for transmitting
1753 the ganch offset when doing a /g match. This will change */
6eb5f6b9
JH
1754/* nosave: For optimizations. */
1755{
97aff369 1756 dVAR;
288b8c02 1757 struct regexp *const prog = (struct regexp *)SvANY(rx);
24b23f37 1758 /*register*/ char *s;
6eb5f6b9 1759 register regnode *c;
24b23f37 1760 /*register*/ char *startpos = stringarg;
6eb5f6b9
JH
1761 I32 minlen; /* must match at least this many chars */
1762 I32 dontbother = 0; /* how many characters not to try at end */
6eb5f6b9
JH
1763 I32 end_shift = 0; /* Same for the end. */ /* CC */
1764 I32 scream_pos = -1; /* Internal iterator of scream. */
ccac19ea 1765 char *scream_olds = NULL;
f9f4320a 1766 const bool do_utf8 = (bool)DO_UTF8(sv);
2757e526 1767 I32 multiline;
f8fc2ecf 1768 RXi_GET_DECL(prog,progi);
3b0527fe 1769 regmatch_info reginfo; /* create some info to pass to regtry etc */
fae667d5 1770 bool swap_on_fail = 0;
a3621e74
YO
1771
1772 GET_RE_DEBUG_FLAGS_DECL;
1773
9d4ba2ae 1774 PERL_UNUSED_ARG(data);
6eb5f6b9
JH
1775
1776 /* Be paranoid... */
1777 if (prog == NULL || startpos == NULL) {
1778 Perl_croak(aTHX_ "NULL regexp parameter");
1779 return 0;
1780 }
1781
bbe252da 1782 multiline = prog->extflags & RXf_PMf_MULTILINE;
288b8c02 1783 reginfo.prog = rx; /* Yes, sorry that this is confusing. */
2757e526 1784
288b8c02 1785 RX_MATCH_UTF8_set(rx, do_utf8);
1de06328 1786 DEBUG_EXECUTE_r(
efd26800 1787 debug_start_match(rx, do_utf8, startpos, strend,
1de06328
YO
1788 "Matching");
1789 );
bac06658 1790
6eb5f6b9 1791 minlen = prog->minlen;
1de06328
YO
1792
1793 if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
a3621e74 1794 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
a72c7584
JH
1795 "String too short [regexec_flags]...\n"));
1796 goto phooey;
1aa99e6b 1797 }
6eb5f6b9 1798
1de06328 1799
6eb5f6b9 1800 /* Check validity of program. */
f8fc2ecf 1801 if (UCHARAT(progi->program) != REG_MAGIC) {
6eb5f6b9
JH
1802 Perl_croak(aTHX_ "corrupted regexp program");
1803 }
1804
1805 PL_reg_flags = 0;
1806 PL_reg_eval_set = 0;
1807 PL_reg_maxiter = 0;
1808
3c8556c3 1809 if (RX_UTF8(rx))
6eb5f6b9
JH
1810 PL_reg_flags |= RF_utf8;
1811
1812 /* Mark beginning of line for ^ and lookbehind. */
3b0527fe 1813 reginfo.bol = startpos; /* XXX not used ??? */
6eb5f6b9 1814 PL_bostr = strbeg;
3b0527fe 1815 reginfo.sv = sv;
6eb5f6b9
JH
1816
1817 /* Mark end of line for $ (and such) */
1818 PL_regeol = strend;
1819
1820 /* see how far we have to get to not match where we matched before */
3b0527fe 1821 reginfo.till = startpos+minend;
6eb5f6b9 1822
6eb5f6b9
JH
1823 /* If there is a "must appear" string, look for it. */
1824 s = startpos;
1825
bbe252da 1826 if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */
6eb5f6b9
JH
1827 MAGIC *mg;
1828
1829 if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
58e23c8d 1830 reginfo.ganch = startpos + prog->gofs;
6eb5f6b9
JH
1831 else if (sv && SvTYPE(sv) >= SVt_PVMG
1832 && SvMAGIC(sv)
14befaf4
DM
1833 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
1834 && mg->mg_len >= 0) {
3b0527fe 1835 reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */
bbe252da 1836 if (prog->extflags & RXf_ANCH_GPOS) {
3b0527fe 1837 if (s > reginfo.ganch)
6eb5f6b9 1838 goto phooey;
58e23c8d 1839 s = reginfo.ganch - prog->gofs;
6eb5f6b9
JH
1840 }
1841 }
58e23c8d 1842 else if (data) {
70685ca0 1843 reginfo.ganch = strbeg + PTR2UV(data);
58e23c8d 1844 } else /* pos() not defined */
3b0527fe 1845 reginfo.ganch = strbeg;
6eb5f6b9 1846 }
288b8c02 1847 if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
fae667d5
YO
1848 swap_on_fail = 1;
1849 swap_match_buff(prog); /* do we need a save destructor here for
1850 eval dies? */
c74340f9 1851 }
a0714e2c 1852 if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
6eb5f6b9
JH
1853 re_scream_pos_data d;
1854
1855 d.scream_olds = &scream_olds;
1856 d.scream_pos = &scream_pos;
288b8c02 1857 s = re_intuit_start(rx, sv, s, strend, flags, &d);
3fa9c3d7 1858 if (!s) {
a3621e74 1859 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
6eb5f6b9 1860 goto phooey; /* not present */
3fa9c3d7 1861 }
6eb5f6b9
JH
1862 }
1863
1de06328 1864
6eb5f6b9
JH
1865
1866 /* Simplest case: anchored match need be tried only once. */
1867 /* [unless only anchor is BOL and multiline is set] */
bbe252da 1868 if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) {
24b23f37 1869 if (s == startpos && regtry(&reginfo, &startpos))
6eb5f6b9 1870 goto got_it;
bbe252da
YO
1871 else if (multiline || (prog->intflags & PREGf_IMPLICIT)
1872 || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */
6eb5f6b9
JH
1873 {
1874 char *end;
1875
1876 if (minlen)
1877 dontbother = minlen - 1;
1aa99e6b 1878 end = HOP3c(strend, -dontbother, strbeg) - 1;
6eb5f6b9 1879 /* for multiline we only have to try after newlines */
33b8afdf 1880 if (prog->check_substr || prog->check_utf8) {
6eb5f6b9
JH
1881 if (s == startpos)
1882 goto after_try;
1883 while (1) {
24b23f37 1884 if (regtry(&reginfo, &s))
6eb5f6b9
JH
1885 goto got_it;
1886 after_try:
5339e136 1887 if (s > end)
6eb5f6b9 1888 goto phooey;
bbe252da 1889 if (prog->extflags & RXf_USE_INTUIT) {
288b8c02 1890 s = re_intuit_start(rx, sv, s + 1, strend, flags, NULL);
6eb5f6b9
JH
1891 if (!s)
1892 goto phooey;
1893 }
1894 else
1895 s++;
1896 }
1897 } else {
1898 if (s > startpos)
1899 s--;
1900 while (s < end) {
1901 if (*s++ == '\n') { /* don't need PL_utf8skip here */
24b23f37 1902 if (regtry(&reginfo, &s))
6eb5f6b9
JH
1903 goto got_it;
1904 }
1905 }
1906 }
1907 }
1908 goto phooey;
bbe252da 1909 } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK))
f9f4320a
YO
1910 {
1911 /* the warning about reginfo.ganch being used without intialization
bbe252da 1912 is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN
f9f4320a 1913 and we only enter this block when the same bit is set. */
58e23c8d
YO
1914 char *tmp_s = reginfo.ganch - prog->gofs;
1915 if (regtry(&reginfo, &tmp_s))
6eb5f6b9
JH
1916 goto got_it;
1917 goto phooey;
1918 }
1919
1920 /* Messy cases: unanchored match. */
bbe252da 1921 if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
6eb5f6b9
JH
1922 /* we have /x+whatever/ */
1923 /* it must be a one character string (XXXX Except UTF?) */
33b8afdf 1924 char ch;
bf93d4cc
GS
1925#ifdef DEBUGGING
1926 int did_match = 0;
1927#endif
33b8afdf
JH
1928 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1929 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
3f7c398e 1930 ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
bf93d4cc 1931
1aa99e6b 1932 if (do_utf8) {
4cadc6a9 1933 REXEC_FBC_SCAN(
6eb5f6b9 1934 if (*s == ch) {
a3621e74 1935 DEBUG_EXECUTE_r( did_match = 1 );
24b23f37 1936 if (regtry(&reginfo, &s)) goto got_it;
6eb5f6b9
JH
1937 s += UTF8SKIP(s);
1938 while (s < strend && *s == ch)
1939 s += UTF8SKIP(s);
1940 }
4cadc6a9 1941 );
6eb5f6b9
JH
1942 }
1943 else {
4cadc6a9 1944 REXEC_FBC_SCAN(
6eb5f6b9 1945 if (*s == ch) {
a3621e74 1946 DEBUG_EXECUTE_r( did_match = 1 );
24b23f37 1947 if (regtry(&reginfo, &s)) goto got_it;
6eb5f6b9
JH
1948 s++;
1949 while (s < strend && *s == ch)
1950 s++;
1951 }
4cadc6a9 1952 );
6eb5f6b9 1953 }
a3621e74 1954 DEBUG_EXECUTE_r(if (!did_match)
bf93d4cc 1955 PerlIO_printf(Perl_debug_log,
b7953727
JH
1956 "Did not find anchored character...\n")
1957 );
6eb5f6b9 1958 }
a0714e2c
SS
1959 else if (prog->anchored_substr != NULL
1960 || prog->anchored_utf8 != NULL
1961 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
33b8afdf
JH
1962 && prog->float_max_offset < strend - s)) {
1963 SV *must;
1964 I32 back_max;
1965 I32 back_min;
1966 char *last;
6eb5f6b9 1967 char *last1; /* Last position checked before */
bf93d4cc
GS
1968#ifdef DEBUGGING
1969 int did_match = 0;
1970#endif
33b8afdf
JH
1971 if (prog->anchored_substr || prog->anchored_utf8) {
1972 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1973 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1974 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
1975 back_max = back_min = prog->anchored_offset;
1976 } else {
1977 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1978 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1979 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
1980 back_max = prog->float_max_offset;
1981 back_min = prog->float_min_offset;
1982 }
1de06328
YO
1983
1984
33b8afdf
JH
1985 if (must == &PL_sv_undef)
1986 /* could not downgrade utf8 check substring, so must fail */
1987 goto phooey;
1988
1de06328
YO
1989 if (back_min<0) {
1990 last = strend;
1991 } else {
1992 last = HOP3c(strend, /* Cannot start after this */
1993 -(I32)(CHR_SVLEN(must)
1994 - (SvTAIL(must) != 0) + back_min), strbeg);
1995 }
6eb5f6b9
JH
1996 if (s > PL_bostr)
1997 last1 = HOPc(s, -1);
1998 else
1999 last1 = s - 1; /* bogus */
2000
a0288114 2001 /* XXXX check_substr already used to find "s", can optimize if
6eb5f6b9
JH
2002 check_substr==must. */
2003 scream_pos = -1;
2004 dontbother = end_shift;
2005 strend = HOPc(strend, -dontbother);
2006 while ( (s <= last) &&
9041c2e3 2007 ((flags & REXEC_SCREAM)
1de06328 2008 ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg,
6eb5f6b9 2009 end_shift, &scream_pos, 0))
1de06328 2010 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)),
9041c2e3 2011 (unsigned char*)strend, must,
7fba1cd6 2012 multiline ? FBMrf_MULTILINE : 0))) ) {
4addbd3b 2013 /* we may be pointing at the wrong string */
07bc277f 2014 if ((flags & REXEC_SCREAM) && RXp_MATCH_COPIED(prog))
3f7c398e 2015 s = strbeg + (s - SvPVX_const(sv));
a3621e74 2016 DEBUG_EXECUTE_r( did_match = 1 );
6eb5f6b9
JH
2017 if (HOPc(s, -back_max) > last1) {
2018 last1 = HOPc(s, -back_min);
2019 s = HOPc(s, -back_max);
2020 }
2021 else {
52657f30 2022 char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
6eb5f6b9
JH
2023
2024 last1 = HOPc(s, -back_min);
52657f30 2025 s = t;
6eb5f6b9 2026 }
1aa99e6b 2027 if (do_utf8) {
6eb5f6b9 2028 while (s <= last1) {
24b23f37 2029 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2030 goto got_it;
2031 s += UTF8SKIP(s);
2032 }
2033 }
2034 else {
2035 while (s <= last1) {
24b23f37 2036 if (regtry(&reginfo, &s))
6eb5f6b9
JH
2037 goto got_it;
2038 s++;
2039 }
2040 }
2041 }
ab3bbdeb
YO
2042 DEBUG_EXECUTE_r(if (!did_match) {
2043 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
2044 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
2045 PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
33b8afdf 2046 ((must == prog->anchored_substr || must == prog->anchored_utf8)
bf93d4cc 2047 ? "anchored" : "floating"),
ab3bbdeb
YO
2048 quoted, RE_SV_TAIL(must));
2049 });
6eb5f6b9
JH
2050 goto phooey;
2051 }
f8fc2ecf 2052 else if ( (c = progi->regstclass) ) {
f14c76ed 2053 if (minlen) {
f8fc2ecf 2054 const OPCODE op = OP(progi->regstclass);
66e933ab 2055 /* don't bother with what can't match */
786e8c11 2056 if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
f14c76ed
RGS
2057 strend = HOPc(strend, -(minlen - 1));
2058 }
a3621e74 2059 DEBUG_EXECUTE_r({
be8e71aa 2060 SV * const prop = sv_newmortal();
32fc9b6a 2061 regprop(prog, prop, c);
0df25f3d 2062 {
02daf0ab 2063 RE_PV_QUOTED_DECL(quoted,do_utf8,PERL_DEBUG_PAD_ZERO(1),
ab3bbdeb 2064 s,strend-s,60);
0df25f3d 2065 PerlIO_printf(Perl_debug_log,
ab3bbdeb 2066 "Matching stclass %.*s against %s (%d chars)\n",
e4f74956 2067 (int)SvCUR(prop), SvPVX_const(prop),
ab3bbdeb 2068 quoted, (int)(strend - s));
0df25f3d 2069 }
ffc61ed2 2070 });
3b0527fe 2071 if (find_byclass(prog, c, s, strend, &reginfo))
6eb5f6b9 2072 goto got_it;
07be1b83 2073 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
d6a28714
JH
2074 }
2075 else {
2076 dontbother = 0;
a0714e2c 2077 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
33b8afdf 2078 /* Trim the end. */
d6a28714 2079 char *last;
33b8afdf
JH
2080 SV* float_real;
2081
2082 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
2083 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
2084 float_real = do_utf8 ? prog->float_utf8 : prog->float_substr;
d6a28714
JH
2085
2086 if (flags & REXEC_SCREAM) {
33b8afdf 2087 last = screaminstr(sv, float_real, s - strbeg,
d6a28714
JH
2088 end_shift, &scream_pos, 1); /* last one */
2089 if (!last)
ffc61ed2 2090 last = scream_olds; /* Only one occurrence. */
4addbd3b 2091 /* we may be pointing at the wrong string */
07bc277f 2092 else if (RXp_MATCH_COPIED(prog))
3f7c398e 2093 s = strbeg + (s - SvPVX_const(sv));
b8c5462f 2094 }
d6a28714
JH
2095 else {
2096 STRLEN len;
cfd0369c 2097 const char * const little = SvPV_const(float_real, len);
d6a28714 2098
33b8afdf 2099 if (SvTAIL(float_real)) {
d6a28714
JH
2100 if (memEQ(strend - len + 1, little, len - 1))
2101 last = strend - len + 1;
7fba1cd6 2102 else if (!multiline)
9041c2e3 2103 last = memEQ(strend - len, little, len)
bd61b366 2104 ? strend - len : NULL;
b8c5462f 2105 else
d6a28714
JH
2106 goto find_last;
2107 } else {
2108 find_last:
9041c2e3 2109 if (len)
d6a28714 2110 last = rninstr(s, strend, little, little + len);
b8c5462f 2111 else
a0288114 2112 last = strend; /* matching "$" */
b8c5462f 2113 }
b8c5462f 2114 }
bf93d4cc 2115 if (last == NULL) {
6bda09f9
YO
2116 DEBUG_EXECUTE_r(
2117 PerlIO_printf(Perl_debug_log,
2118 "%sCan't trim the tail, match fails (should not happen)%s\n",
2119 PL_colors[4], PL_colors[5]));
bf93d4cc
GS
2120 goto phooey; /* Should not happen! */
2121 }
d6a28714
JH
2122 dontbother = strend - last + prog->float_min_offset;
2123 }
2124 if (minlen && (dontbother < minlen))
2125 dontbother = minlen - 1;
2126 strend -= dontbother; /* this one's always in bytes! */
2127 /* We don't know much -- general case. */
1aa99e6b 2128 if (do_utf8) {
d6a28714 2129 for (;;) {
24b23f37 2130 if (regtry(&reginfo, &s))
d6a28714
JH
2131 goto got_it;
2132 if (s >= strend)
2133 break;
b8c5462f 2134 s += UTF8SKIP(s);
d6a28714
JH
2135 };
2136 }
2137 else {
2138 do {
24b23f37 2139 if (regtry(&reginfo, &s))
d6a28714
JH
2140 goto got_it;
2141 } while (s++ < strend);
2142 }
2143 }
2144
2145 /* Failure. */
2146 goto phooey;
2147
2148got_it:
288b8c02 2149 RX_MATCH_TAINTED_set(rx, PL_reg_flags & RF_tainted);
d6a28714 2150
19b95bf0 2151 if (PL_reg_eval_set)
4f639d21 2152 restore_pos(aTHX_ prog);
5daac39c
NC
2153 if (RXp_PAREN_NAMES(prog))
2154 (void)hv_iterinit(RXp_PAREN_NAMES(prog));
d6a28714
JH
2155
2156 /* make sure $`, $&, $', and $digit will work later */
2157 if ( !(flags & REXEC_NOT_FIRST) ) {
288b8c02 2158 RX_MATCH_COPY_FREE(rx);
d6a28714 2159 if (flags & REXEC_COPY_STR) {
be8e71aa 2160 const I32 i = PL_regeol - startpos + (stringarg - strbeg);
f8c7b90f 2161#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
2162 if ((SvIsCOW(sv)
2163 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2164 if (DEBUG_C_TEST) {
2165 PerlIO_printf(Perl_debug_log,
2166 "Copy on write: regexp capture, type %d\n",
2167 (int) SvTYPE(sv));
2168 }
2169 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
d5263905 2170 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
ed252734
NC
2171 assert (SvPOKp(prog->saved_copy));
2172 } else
2173#endif
2174 {
288b8c02 2175 RX_MATCH_COPIED_on(rx);
ed252734
NC
2176 s = savepvn(strbeg, i);
2177 prog->subbeg = s;
2178 }
d6a28714 2179 prog->sublen = i;
d6a28714
JH
2180 }
2181 else {
2182 prog->subbeg = strbeg;
2183 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2184 }
2185 }
9041c2e3 2186
d6a28714
JH
2187 return 1;
2188
2189phooey:
a3621e74 2190 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
e4584336 2191 PL_colors[4], PL_colors[5]));
d6a28714 2192 if (PL_reg_eval_set)
4f639d21 2193 restore_pos(aTHX_ prog);
fae667d5 2194 if (swap_on_fail)
c74340f9 2195 /* we failed :-( roll it back */
fae667d5
YO
2196 swap_match_buff(prog);
2197
d6a28714
JH
2198 return 0;
2199}
2200
6bda09f9 2201
d6a28714
JH
2202/*
2203 - regtry - try match at specific point
2204 */
2205STATIC I32 /* 0 failure, 1 success */
24b23f37 2206S_regtry(pTHX_ regmatch_info *reginfo, char **startpos)
d6a28714 2207{
97aff369 2208 dVAR;
d6a28714 2209 CHECKPOINT lastcp;
288b8c02
NC
2210 REGEXP *const rx = reginfo->prog;
2211 regexp *const prog = (struct regexp *)SvANY(rx);
f8fc2ecf 2212 RXi_GET_DECL(prog,progi);
a3621e74 2213 GET_RE_DEBUG_FLAGS_DECL;
24b23f37 2214 reginfo->cutpoint=NULL;
d6a28714 2215
bbe252da 2216 if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) {
d6a28714
JH
2217 MAGIC *mg;
2218
2219 PL_reg_eval_set = RS_init;
a3621e74 2220 DEBUG_EXECUTE_r(DEBUG_s(
b900a521
JH
2221 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
2222 (IV)(PL_stack_sp - PL_stack_base));
d6a28714 2223 ));
ea8d6ae1 2224 SAVESTACK_CXPOS();
d6a28714
JH
2225 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
2226 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
2227 SAVETMPS;
2228 /* Apparently this is not needed, judging by wantarray. */
e8347627 2229 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
d6a28714
JH
2230 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
2231
3b0527fe 2232 if (reginfo->sv) {
d6a28714 2233 /* Make $_ available to executed code. */
3b0527fe 2234 if (reginfo->sv != DEFSV) {
59f00321 2235 SAVE_DEFSV;
3b0527fe 2236 DEFSV = reginfo->sv;
b8c5462f 2237 }
d6a28714 2238
3b0527fe
DM
2239 if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
2240 && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
d6a28714 2241 /* prepare for quick setting of pos */
d300d9fa 2242#ifdef PERL_OLD_COPY_ON_WRITE
51a9ea20
NC
2243 if (SvIsCOW(reginfo->sv))
2244 sv_force_normal_flags(reginfo->sv, 0);
d300d9fa 2245#endif
3dab1dad 2246 mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global,
d300d9fa 2247 &PL_vtbl_mglob, NULL, 0);
d6a28714 2248 mg->mg_len = -1;
b8c5462f 2249 }
d6a28714
JH
2250 PL_reg_magic = mg;
2251 PL_reg_oldpos = mg->mg_len;
4f639d21 2252 SAVEDESTRUCTOR_X(restore_pos, prog);
d6a28714 2253 }
09687e5a 2254 if (!PL_reg_curpm) {
a02a5408 2255 Newxz(PL_reg_curpm, 1, PMOP);
09687e5a
AB
2256#ifdef USE_ITHREADS
2257 {
be8e71aa 2258 SV* const repointer = newSViv(0);
577e12cc 2259 /* so we know which PL_regex_padav element is PL_reg_curpm */
35061a7e 2260 SvFLAGS(repointer) |= SVf_BREAK;
09687e5a
AB
2261 av_push(PL_regex_padav,repointer);
2262 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2263 PL_regex_pad = AvARRAY(PL_regex_padav);
2264 }
2265#endif
2266 }
288b8c02 2267 PM_SETRE(PL_reg_curpm, rx);
d6a28714
JH
2268 PL_reg_oldcurpm = PL_curpm;
2269 PL_curpm = PL_reg_curpm;
07bc277f 2270 if (RXp_MATCH_COPIED(prog)) {
d6a28714
JH
2271 /* Here is a serious problem: we cannot rewrite subbeg,
2272 since it may be needed if this match fails. Thus
2273 $` inside (?{}) could fail... */
2274 PL_reg_oldsaved = prog->subbeg;
2275 PL_reg_oldsavedlen = prog->sublen;
f8c7b90f 2276#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
2277 PL_nrs = prog->saved_copy;
2278#endif
07bc277f 2279 RXp_MATCH_COPIED_off(prog);
d6a28714
JH
2280 }
2281 else
bd61b366 2282 PL_reg_oldsaved = NULL;
d6a28714
JH
2283 prog->subbeg = PL_bostr;
2284 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2285 }
24b23f37 2286 DEBUG_EXECUTE_r(PL_reg_starttry = *startpos);
f0ab9afb 2287 prog->offs[0].start = *startpos - PL_bostr;
24b23f37 2288 PL_reginput = *startpos;
d6a28714 2289 PL_reglastparen = &prog->lastparen;
a01268b5 2290 PL_reglastcloseparen = &prog->lastcloseparen;
d6a28714 2291 prog->lastparen = 0;
03994de8 2292 prog->lastcloseparen = 0;
d6a28714 2293 PL_regsize = 0;
f0ab9afb 2294 PL_regoffs = prog->offs;
d6a28714
JH
2295 if (PL_reg_start_tmpl <= prog->nparens) {
2296 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
2297 if(PL_reg_start_tmp)
2298 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2299 else
a02a5408 2300 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
d6a28714
JH
2301 }
2302
2303 /* XXXX What this code is doing here?!!! There should be no need
2304 to do this again and again, PL_reglastparen should take care of
3dd2943c 2305 this! --ilya*/
dafc8851
JH
2306
2307 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2308 * Actually, the code in regcppop() (which Ilya may be meaning by
daf18116 2309 * PL_reglastparen), is not needed at all by the test suite
225593e1
DM
2310 * (op/regexp, op/pat, op/split), but that code is needed otherwise
2311 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
2312 * Meanwhile, this code *is* needed for the
daf18116
JH
2313 * above-mentioned test suite tests to succeed. The common theme
2314 * on those tests seems to be returning null fields from matches.
225593e1 2315 * --jhi updated by dapm */
dafc8851 2316#if 1
d6a28714 2317 if (prog->nparens) {
f0ab9afb 2318 regexp_paren_pair *pp = PL_regoffs;
097eb12c 2319 register I32 i;
eb160463 2320 for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
f0ab9afb
NC
2321 ++pp;
2322 pp->start = -1;
2323 pp->end = -1;
d6a28714
JH
2324 }
2325 }
dafc8851 2326#endif
02db2b7b 2327 REGCP_SET(lastcp);
f8fc2ecf 2328 if (regmatch(reginfo, progi->program + 1)) {
f0ab9afb 2329 PL_regoffs[0].end = PL_reginput - PL_bostr;
d6a28714
JH
2330 return 1;
2331 }
24b23f37
YO
2332 if (reginfo->cutpoint)
2333 *startpos= reginfo->cutpoint;
02db2b7b 2334 REGCP_UNWIND(lastcp);
d6a28714
JH
2335 return 0;
2336}
2337
02db2b7b 2338
8ba1375e
MJD
2339#define sayYES goto yes
2340#define sayNO goto no
262b90c4 2341#define sayNO_SILENT goto no_silent
8ba1375e 2342
f9f4320a
YO
2343/* we dont use STMT_START/END here because it leads to
2344 "unreachable code" warnings, which are bogus, but distracting. */
2345#define CACHEsayNO \
c476f425
DM
2346 if (ST.cache_mask) \
2347 PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \
f9f4320a 2348 sayNO
3298f257 2349
a3621e74 2350/* this is used to determine how far from the left messages like
265c4333
YO
2351 'failed...' are printed. It should be set such that messages
2352 are inline with the regop output that created them.
a3621e74 2353*/
265c4333 2354#define REPORT_CODE_OFF 32
a3621e74
YO
2355
2356
2357/* Make sure there is a test for this +1 options in re_tests */
2358#define TRIE_INITAL_ACCEPT_BUFFLEN 4;
2359
40a82448
DM
2360#define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
2361#define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
9e137952 2362
86545054
DM
2363#define SLAB_FIRST(s) (&(s)->states[0])
2364#define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
2365
5d9a96ca
DM
2366/* grab a new slab and return the first slot in it */
2367
2368STATIC regmatch_state *
2369S_push_slab(pTHX)
2370{
a35a87e7 2371#if PERL_VERSION < 9 && !defined(PERL_CORE)
54df2634
NC
2372 dMY_CXT;
2373#endif
5d9a96ca
DM
2374 regmatch_slab *s = PL_regmatch_slab->next;
2375 if (!s) {
2376 Newx(s, 1, regmatch_slab);
2377 s->prev = PL_regmatch_slab;
2378 s->next = NULL;
2379 PL_regmatch_slab->next = s;
2380 }
2381 PL_regmatch_slab = s;
86545054 2382 return SLAB_FIRST(s);
5d9a96ca 2383}
5b47454d 2384
95b24440 2385
40a82448
DM
2386/* push a new state then goto it */
2387
2388#define PUSH_STATE_GOTO(state, node) \
2389 scan = node; \
2390 st->resume_state = state; \
2391 goto push_state;
2392
2393/* push a new state with success backtracking, then goto it */
2394
2395#define PUSH_YES_STATE_GOTO(state, node) \
2396 scan = node; \
2397 st->resume_state = state; \
2398 goto push_yes_state;
2399
aa283a38 2400
aa283a38 2401
d6a28714 2402/*
95b24440 2403
bf1f174e
DM
2404regmatch() - main matching routine
2405
2406This is basically one big switch statement in a loop. We execute an op,
2407set 'next' to point the next op, and continue. If we come to a point which
2408we may need to backtrack to on failure such as (A|B|C), we push a
2409backtrack state onto the backtrack stack. On failure, we pop the top
2410state, and re-enter the loop at the state indicated. If there are no more
2411states to pop, we return failure.
2412
2413Sometimes we also need to backtrack on success; for example /A+/, where
2414after successfully matching one A, we need to go back and try to
2415match another one; similarly for lookahead assertions: if the assertion
2416completes successfully, we backtrack to the state just before the assertion
2417and then carry on. In these cases, the pushed state is marked as
2418'backtrack on success too'. This marking is in fact done by a chain of
2419pointers, each pointing to the previous 'yes' state. On success, we pop to
2420the nearest yes state, discarding any intermediate failure-only states.
2421Sometimes a yes state is pushed just to force some cleanup code to be
2422called at the end of a successful match or submatch; e.g. (??{$re}) uses
2423it to free the inner regex.
2424
2425Note that failure backtracking rewinds the cursor position, while
2426success backtracking leaves it alone.
2427
2428A pattern is complete when the END op is executed, while a subpattern
2429such as (?=foo) is complete when the SUCCESS op is executed. Both of these
2430ops trigger the "pop to last yes state if any, otherwise return true"
2431behaviour.
2432
2433A common convention in this function is to use A and B to refer to the two
2434subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
2435the subpattern to be matched possibly multiple times, while B is the entire
2436rest of the pattern. Variable and state names reflect this convention.
2437
2438The states in the main switch are the union of ops and failure/success of
2439substates associated with with that op. For example, IFMATCH is the op
2440that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
2441'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
2442successfully matched A and IFMATCH_A_fail is a state saying that we have
2443just failed to match A. Resume states always come in pairs. The backtrack
2444state we push is marked as 'IFMATCH_A', but when that is popped, we resume
2445at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
2446on success or failure.
2447
2448The struct that holds a backtracking state is actually a big union, with
2449one variant for each major type of op. The variable st points to the
2450top-most backtrack struct. To make the code clearer, within each
2451block of code we #define ST to alias the relevant union.
2452
2453Here's a concrete example of a (vastly oversimplified) IFMATCH
2454implementation:
2455
2456 switch (state) {
2457 ....
2458
2459#define ST st->u.ifmatch
2460
2461 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2462 ST.foo = ...; // some state we wish to save
95b24440 2463 ...
bf1f174e
DM
2464 // push a yes backtrack state with a resume value of
2465 // IFMATCH_A/IFMATCH_A_fail, then continue execution at the
2466 // first node of A:
2467 PUSH_YES_STATE_GOTO(IFMATCH_A, A);
2468 // NOTREACHED
2469
2470 case IFMATCH_A: // we have successfully executed A; now continue with B
2471 next = B;
2472 bar = ST.foo; // do something with the preserved value
2473 break;
2474
2475 case IFMATCH_A_fail: // A failed, so the assertion failed
2476 ...; // do some housekeeping, then ...
2477 sayNO; // propagate the failure
2478
2479#undef ST
95b24440 2480
bf1f174e
DM
2481 ...
2482 }
95b24440 2483
bf1f174e
DM
2484For any old-timers reading this who are familiar with the old recursive
2485approach, the code above is equivalent to:
95b24440 2486
bf1f174e
DM
2487 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2488 {
2489 int foo = ...
95b24440 2490 ...
bf1f174e
DM
2491 if (regmatch(A)) {
2492 next = B;
2493 bar = foo;
2494 break;
95b24440 2495 }
bf1f174e
DM
2496 ...; // do some housekeeping, then ...
2497 sayNO; // propagate the failure
95b24440 2498 }
bf1f174e
DM
2499
2500The topmost backtrack state, pointed to by st, is usually free. If you
2501want to claim it, populate any ST.foo fields in it with values you wish to
2502save, then do one of
2503
2504 PUSH_STATE_GOTO(resume_state, node);
2505 PUSH_YES_STATE_GOTO(resume_state, node);
2506
2507which sets that backtrack state's resume value to 'resume_state', pushes a
2508new free entry to the top of the backtrack stack, then goes to 'node'.
2509On backtracking, the free slot is popped, and the saved state becomes the
2510new free state. An ST.foo field in this new top state can be temporarily
2511accessed to retrieve values, but once the main loop is re-entered, it
2512becomes available for reuse.
2513
2514Note that the depth of the backtrack stack constantly increases during the
2515left-to-right execution of the pattern, rather than going up and down with
2516the pattern nesting. For example the stack is at its maximum at Z at the
2517end of the pattern, rather than at X in the following:
2518
2519 /(((X)+)+)+....(Y)+....Z/
2520
2521The only exceptions to this are lookahead/behind assertions and the cut,
2522(?>A), which pop all the backtrack states associated with A before
2523continuing.
2524
2525Bascktrack state structs are allocated in slabs of about 4K in size.
2526PL_regmatch_state and st always point to the currently active state,
2527and PL_regmatch_slab points to the slab currently containing
2528PL_regmatch_state. The first time regmatch() is called, the first slab is
2529allocated, and is never freed until interpreter destruction. When the slab
2530is full, a new one is allocated and chained to the end. At exit from
2531regmatch(), slabs allocated since entry are freed.
2532
2533*/
95b24440 2534
40a82448 2535
5bc10b2c 2536#define DEBUG_STATE_pp(pp) \
265c4333 2537 DEBUG_STATE_r({ \
5bc10b2c
DM
2538 DUMP_EXEC_POS(locinput, scan, do_utf8); \
2539 PerlIO_printf(Perl_debug_log, \
5d458dd8 2540 " %*s"pp" %s%s%s%s%s\n", \
5bc10b2c 2541 depth*2, "", \
13d6edb4 2542 PL_reg_name[st->resume_state], \
5d458dd8
YO
2543 ((st==yes_state||st==mark_state) ? "[" : ""), \
2544 ((st==yes_state) ? "Y" : ""), \
2545 ((st==mark_state) ? "M" : ""), \
2546 ((st==yes_state||st==mark_state) ? "]" : "") \
2547 ); \
265c4333 2548 });
5bc10b2c 2549
40a82448 2550
3dab1dad 2551#define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)
95b24440 2552
3df15adc 2553#ifdef DEBUGGING
5bc10b2c 2554
ab3bbdeb 2555STATIC void
efd26800 2556S_debug_start_match(pTHX_ const REGEXP *prog, const bool do_utf8,
ab3bbdeb
YO
2557 const char *start, const char *end, const char *blurb)
2558{
efd26800 2559 const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;
ab3bbdeb
YO
2560 if (!PL_colorset)
2561 reginitcolors();
2562 {
2563 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
efd26800 2564 RX_PRECOMP(prog), RX_PRELEN(prog), 60);
ab3bbdeb
YO
2565
2566 RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1),
2567 start, end - start, 60);
2568
2569 PerlIO_printf(Perl_debug_log,
2570 "%s%s REx%s %s against %s\n",
2571 PL_colors[4], blurb, PL_colors[5], s0, s1);
2572
2573 if (do_utf8||utf8_pat)
1de06328
YO
2574 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
2575 utf8_pat ? "pattern" : "",
2576 utf8_pat && do_utf8 ? " and " : "",
2577 do_utf8 ? "string" : ""
ab3bbdeb
YO
2578 );
2579 }
2580}
3df15adc
YO
2581
2582STATIC void
786e8c11
YO
2583S_dump_exec_pos(pTHX_ const char *locinput,
2584 const regnode *scan,
2585 const char *loc_regeol,
2586 const char *loc_bostr,
2587 const char *loc_reg_starttry,
2588 const bool do_utf8)
07be1b83 2589{
786e8c11 2590 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
07be1b83 2591 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
786e8c11 2592 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
07be1b83
YO
2593 /* The part of the string before starttry has one color
2594 (pref0_len chars), between starttry and current
2595 position another one (pref_len - pref0_len chars),
2596 after the current position the third one.
2597 We assume that pref0_len <= pref_len, otherwise we
2598 decrease pref0_len. */
786e8c11
YO
2599 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
2600 ? (5 + taill) - l : locinput - loc_bostr;
07be1b83
YO
2601 int pref0_len;
2602
2603 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
2604 pref_len++;
786e8c11
YO
2605 pref0_len = pref_len - (locinput - loc_reg_starttry);
2606 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
2607 l = ( loc_regeol - locinput > (5 + taill) - pref_len
2608 ? (5 + taill) - pref_len : loc_regeol - locinput);
07be1b83
YO
2609 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
2610 l--;
2611 if (pref0_len < 0)
2612 pref0_len = 0;
2613 if (pref0_len > pref_len)
2614 pref0_len = pref_len;
2615 {
3df15adc 2616 const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0;
0df25f3d 2617
ab3bbdeb 2618 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
1de06328 2619 (locinput - pref_len),pref0_len, 60, 4, 5);
0df25f3d 2620
ab3bbdeb 2621 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
3df15adc 2622 (locinput - pref_len + pref0_len),
1de06328 2623 pref_len - pref0_len, 60, 2, 3);
0df25f3d 2624
ab3bbdeb 2625 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
1de06328 2626 locinput, loc_regeol - locinput, 10, 0, 1);
0df25f3d 2627
1de06328 2628 const STRLEN tlen=len0+len1+len2;
3df15adc 2629 PerlIO_printf(Perl_debug_log,
ab3bbdeb 2630 "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
786e8c11 2631 (IV)(locinput - loc_bostr),
07be1b83 2632 len0, s0,
07be1b83 2633 len1, s1,
07be1b83 2634 (docolor ? "" : "> <"),
07be1b83 2635 len2, s2,
f9f4320a 2636 (int)(tlen > 19 ? 0 : 19 - tlen),
07be1b83
YO
2637 "");
2638 }
2639}
3df15adc 2640
07be1b83
YO
2641#endif
2642
0a4db386
YO
2643/* reg_check_named_buff_matched()
2644 * Checks to see if a named buffer has matched. The data array of
2645 * buffer numbers corresponding to the buffer is expected to reside
2646 * in the regexp->data->data array in the slot stored in the ARG() of
2647 * node involved. Note that this routine doesn't actually care about the
2648 * name, that information is not preserved from compilation to execution.
2649 * Returns the index of the leftmost defined buffer with the given name
2650 * or 0 if non of the buffers matched.
2651 */
2652STATIC I32
2653S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan) {
2654 I32 n;
f8fc2ecf
YO
2655 RXi_GET_DECL(rex,rexi);
2656 SV *sv_dat=(SV*)rexi->data->data[ ARG( scan ) ];
0a4db386
YO
2657 I32 *nums=(I32*)SvPVX(sv_dat);
2658 for ( n=0; n<SvIVX(sv_dat); n++ ) {
2659 if ((I32)*PL_reglastparen >= nums[n] &&
f0ab9afb 2660 PL_regoffs[nums[n]].end != -1)
0a4db386
YO
2661 {
2662 return nums[n];
2663 }
2664 }
2665 return 0;
2666}
2667
2f554ef7
DM
2668
2669/* free all slabs above current one - called during LEAVE_SCOPE */
2670
2671STATIC void
2672S_clear_backtrack_stack(pTHX_ void *p)
2673{
2674 regmatch_slab *s = PL_regmatch_slab->next;
2675 PERL_UNUSED_ARG(p);
2676
2677 if (!s)
2678 return;
2679 PL_regmatch_slab->next = NULL;
2680 while (s) {
2681 regmatch_slab * const osl = s;
2682 s = s->next;
2683 Safefree(osl);
2684 }
2685}
2686
2687
28d8d7f4
YO
2688#define SETREX(Re1,Re2) \
2689 if (PL_reg_eval_set) PM_SETRE((PL_reg_curpm), (Re2)); \
2690 Re1 = (Re2)
2691
d6a28714 2692STATIC I32 /* 0 failure, 1 success */
24b23f37 2693S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
d6a28714 2694{
a35a87e7 2695#if PERL_VERSION < 9 && !defined(PERL_CORE)
54df2634
NC
2696 dMY_CXT;
2697#endif
27da23d5 2698 dVAR;
95b24440 2699 register const bool do_utf8 = PL_reg_match_utf8;
4ad0818d 2700 const U32 uniflags = UTF8_ALLOW_DEFAULT;
95b24440 2701
288b8c02
NC
2702 REGEXP *rex_sv = reginfo->prog;
2703 regexp *rex = (struct regexp *)SvANY(rex_sv);
f8fc2ecf
YO
2704 RXi_GET_DECL(rex,rexi);
2705
2f554ef7 2706 I32 oldsave;
a3621e74 2707
5d9a96ca
DM
2708 /* the current state. This is a cached copy of PL_regmatch_state */
2709 register regmatch_state *st;
95b24440 2710
5d9a96ca
DM
2711 /* cache heavy used fields of st in registers */
2712 register regnode *scan;
2713 register regnode *next;
438e9bae 2714 register U32 n = 0; /* general value; init to avoid compiler warning */
24d3c4a9 2715 register I32 ln = 0; /* len or last; init to avoid compiler warning */
5d9a96ca 2716 register char *locinput = PL_reginput;
5d9a96ca 2717 register I32 nextchr; /* is always set to UCHARAT(locinput) */
24d3c4a9 2718
b69b0499 2719 bool result = 0; /* return value of S_regmatch */
24d3c4a9 2720 int depth = 0; /* depth of backtrack stack */
4b196cd4
YO
2721 U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
2722 const U32 max_nochange_depth =
2723 (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
2724 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
2725
77cb431f
DM
2726 regmatch_state *yes_state = NULL; /* state to pop to on success of
2727 subpattern */
e2e6a0f1
YO
2728 /* mark_state piggy backs on the yes_state logic so that when we unwind
2729 the stack on success we can update the mark_state as we go */
2730 regmatch_state *mark_state = NULL; /* last mark state we have seen */
5d458dd8 2731
faec1544 2732 regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
b8591aee 2733 struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */
40a82448 2734 U32 state_num;
5d458dd8
YO
2735 bool no_final = 0; /* prevent failure from backtracking? */
2736 bool do_cutgroup = 0; /* no_final only until next branch/trie entry */
e2e6a0f1 2737 char *startpoint = PL_reginput;
5d458dd8
YO
2738 SV *popmark = NULL; /* are we looking for a mark? */
2739 SV *sv_commit = NULL; /* last mark name seen in failure */
2740 SV *sv_yes_mark = NULL; /* last mark name we have seen
2741 during a successfull match */
2742 U32 lastopen = 0; /* last open we saw */
2743 bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;
19b95bf0
DM
2744
2745 SV* const oreplsv = GvSV(PL_replgv);
58e23c8d 2746
5d458dd8 2747
24d3c4a9
DM
2748 /* these three flags are set by various ops to signal information to
2749 * the very next op. They have a useful lifetime of exactly one loop
2750 * iteration, and are not preserved or restored by state pushes/pops
2751 */
2752 bool sw = 0; /* the condition value in (?(cond)a|b) */
2753 bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */
2754 int logical = 0; /* the following EVAL is:
2755 0: (?{...})
2756 1: (?(?{...})X|Y)
2757 2: (??{...})
2758 or the following IFMATCH/UNLESSM is:
2759 false: plain (?=foo)
2760 true: used as a condition: (?(?=foo))
2761 */
2762
95b24440 2763#ifdef DEBUGGING
e68ec53f 2764 GET_RE_DEBUG_FLAGS_DECL;
d6a28714
JH
2765#endif
2766
3b57cd43 2767 DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
24b23f37 2768 PerlIO_printf(Perl_debug_log,"regmatch start\n");
3b57cd43 2769 }));
5d9a96ca
DM
2770 /* on first ever call to regmatch, allocate first slab */
2771 if (!PL_regmatch_slab) {
2772 Newx(PL_regmatch_slab, 1, regmatch_slab);
2773 PL_regmatch_slab->prev = NULL;
2774 PL_regmatch_slab->next = NULL;
86545054 2775 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
5d9a96ca
DM
2776 }
2777
2f554ef7
DM
2778 oldsave = PL_savestack_ix;
2779 SAVEDESTRUCTOR_X(S_clear_backtrack_stack, NULL);
2780 SAVEVPTR(PL_regmatch_slab);
2781 SAVEVPTR(PL_regmatch_state);
5d9a96ca
DM
2782
2783 /* grab next free state slot */
2784 st = ++PL_regmatch_state;
86545054 2785 if (st > SLAB_LAST(PL_regmatch_slab))
5d9a96ca
DM
2786 st = PL_regmatch_state = S_push_slab(aTHX);
2787
d6a28714
JH
2788 /* Note that nextchr is a byte even in UTF */
2789 nextchr = UCHARAT(locinput);
2790 scan = prog;
2791 while (scan != NULL) {
8ba1375e 2792
a3621e74 2793 DEBUG_EXECUTE_r( {
6136c704 2794 SV * const prop = sv_newmortal();
1de06328 2795 regnode *rnext=regnext(scan);
786e8c11 2796 DUMP_EXEC_POS( locinput, scan, do_utf8 );
32fc9b6a 2797 regprop(rex, prop, scan);
07be1b83
YO
2798
2799 PerlIO_printf(Perl_debug_log,
2800 "%3"IVdf":%*s%s(%"IVdf")\n",
f8fc2ecf 2801 (IV)(scan - rexi->program), depth*2, "",
07be1b83 2802 SvPVX_const(prop),
1de06328 2803 (PL_regkind[OP(scan)] == END || !rnext) ?
f8fc2ecf 2804 0 : (IV)(rnext - rexi->program));
2a782b5b 2805 });
d6a28714
JH
2806
2807 next = scan + NEXT_OFF(scan);
2808 if (next == scan)
2809 next = NULL;
40a82448 2810 state_num = OP(scan);
d6a28714 2811
40a82448
DM
2812 reenter_switch:
2813 switch (state_num) {
d6a28714 2814 case BOL:
7fba1cd6 2815 if (locinput == PL_bostr)
d6a28714 2816 {
3b0527fe 2817 /* reginfo->till = reginfo->bol; */
b8c5462f
JH
2818 break;
2819 }
d6a28714
JH
2820 sayNO;
2821 case MBOL:
12d33761
HS
2822 if (locinput == PL_bostr ||
2823 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
d6a28714 2824 {
b8c5462f
JH
2825 break;
2826 }
d6a28714
JH
2827 sayNO;
2828 case SBOL:
c2a73568 2829 if (locinput == PL_bostr)
b8c5462f 2830 break;
d6a28714
JH
2831 sayNO;
2832 case GPOS:
3b0527fe 2833 if (locinput == reginfo->ganch)
d6a28714
JH
2834 break;
2835 sayNO;
ee9b8eae
YO
2836
2837 case KEEPS:
2838 /* update the startpoint */
f0ab9afb 2839 st->u.keeper.val = PL_regoffs[0].start;
ee9b8eae 2840 PL_reginput = locinput;
f0ab9afb 2841 PL_regoffs[0].start = locinput - PL_bostr;
ee9b8eae
YO
2842 PUSH_STATE_GOTO(KEEPS_next, next);
2843 /*NOT-REACHED*/
2844 case KEEPS_next_fail:
2845 /* rollback the start point change */
f0ab9afb 2846 PL_regoffs[0].start = st->u.keeper.val;
ee9b8eae
YO
2847 sayNO_SILENT;
2848 /*NOT-REACHED*/
d6a28714 2849 case EOL:
d6a28714
JH
2850 goto seol;
2851 case MEOL:
d6a28714 2852 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 2853 sayNO;
b8c5462f 2854 break;
d6a28714
JH
2855 case SEOL:
2856 seol:
2857 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
b8c5462f 2858 sayNO;
d6a28714 2859 if (PL_regeol - locinput > 1)
b8c5462f 2860 sayNO;
b8c5462f 2861 break;
d6a28714
JH
2862 case EOS:
2863 if (PL_regeol != locinput)
b8c5462f 2864 sayNO;
d6a28714 2865 break;
ffc61ed2 2866 case SANY:
d6a28714 2867 if (!nextchr && locinput >= PL_regeol)
4633a7c4 2868 sayNO;
f33976b4
DB
2869 if (do_utf8) {
2870 locinput += PL_utf8skip[nextchr];
2871 if (locinput > PL_regeol)
2872 sayNO;
2873 nextchr = UCHARAT(locinput);
2874 }
2875 else
2876 nextchr = UCHARAT(++locinput);
2877 break;
2878 case CANY:
2879 if (!nextchr && locinput >= PL_regeol)
2880 sayNO;
b8c5462f 2881 nextchr = UCHARAT(++locinput);
a0d0e21e 2882 break;
ffc61ed2 2883 case REG_ANY:
1aa99e6b
IH
2884 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
2885 sayNO;
2886 if (do_utf8) {
b8c5462f 2887 locinput += PL_utf8skip[nextchr];
d6a28714
JH
2888 if (locinput > PL_regeol)
2889 sayNO;
a0ed51b3 2890 nextchr = UCHARAT(locinput);
a0ed51b3 2891 }
1aa99e6b
IH
2892 else
2893 nextchr = UCHARAT(++locinput);
a0ed51b3 2894 break;
166ba7cd
DM
2895
2896#undef ST
2897#define ST st->u.trie
786e8c11
YO
2898 case TRIEC:
2899 /* In this case the charclass data is available inline so
2900 we can fail fast without a lot of extra overhead.
2901 */
2902 if (scan->flags == EXACT || !do_utf8) {
2903 if(!ANYOF_BITMAP_TEST(scan, *locinput)) {
2904 DEBUG_EXECUTE_r(
2905 PerlIO_printf(Perl_debug_log,
2906 "%*s %sfailed to match trie start class...%s\n",
5bc10b2c 2907 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
786e8c11
YO
2908 );
2909 sayNO_SILENT;
2910 /* NOTREACHED */
2911 }
2912 }
2913 /* FALL THROUGH */
5b47454d 2914 case TRIE:
3dab1dad 2915 {
07be1b83 2916 /* what type of TRIE am I? (utf8 makes this contextual) */
a0a388a1 2917 DECL_TRIE_TYPE(scan);
3dab1dad
YO
2918
2919 /* what trie are we using right now */
be8e71aa 2920 reg_trie_data * const trie
f8fc2ecf 2921 = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
55eed653 2922 HV * widecharmap = (HV *)rexi->data->data[ ARG( scan ) + 1 ];
3dab1dad 2923 U32 state = trie->startstate;
166ba7cd 2924
3dab1dad
YO
2925 if (trie->bitmap && trie_type != trie_utf8_fold &&
2926 !TRIE_BITMAP_TEST(trie,*locinput)
2927 ) {
2928 if (trie->states[ state ].wordnum) {
2929 DEBUG_EXECUTE_r(
2930 PerlIO_printf(Perl_debug_log,
2931 "%*s %smatched empty string...%s\n",
5bc10b2c 2932 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3dab1dad
YO
2933 );
2934 break;
2935 } else {
2936 DEBUG_EXECUTE_r(
2937 PerlIO_printf(Perl_debug_log,
786e8c11 2938 "%*s %sfailed to match trie start class...%s\n",
5bc10b2c 2939 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3dab1dad
YO
2940 );
2941 sayNO_SILENT;
2942 }
2943 }
166ba7cd 2944
786e8c11
YO
2945 {
2946 U8 *uc = ( U8* )locinput;
2947
2948 STRLEN len = 0;
2949 STRLEN foldlen = 0;
2950 U8 *uscan = (U8*)NULL;
2951 STRLEN bufflen=0;
2952 SV *sv_accept_buff = NULL;
2953 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2954
2955 ST.accepted = 0; /* how many accepting states we have seen */
2956 ST.B = next;
2957 ST.jump = trie->jump;
786e8c11 2958 ST.me = scan;
07be1b83
YO
2959 /*
2960 traverse the TRIE keeping track of all accepting states
2961 we transition through until we get to a failing node.
2962 */
2963
a3621e74 2964 while ( state && uc <= (U8*)PL_regeol ) {
786e8c11 2965 U32 base = trie->states[ state ].trans.base;
f9f4320a 2966 UV uvc = 0;
786e8c11
YO
2967 U16 charid;
2968 /* We use charid to hold the wordnum as we don't use it
2969 for charid until after we have done the wordnum logic.
2970 We define an alias just so that the wordnum logic reads
2971 more naturally. */
2972
2973#define got_wordnum charid
2974 got_wordnum = trie->states[ state ].wordnum;
2975
2976 if ( got_wordnum ) {
2977 if ( ! ST.accepted ) {
5b47454d 2978 ENTER;
6b173516 2979 /* SAVETMPS; */ /* XXX is this necessary? dmq */
5b47454d
DM
2980 bufflen = TRIE_INITAL_ACCEPT_BUFFLEN;
2981 sv_accept_buff=newSV(bufflen *
2982 sizeof(reg_trie_accepted) - 1);
786e8c11 2983 SvCUR_set(sv_accept_buff, 0);
5b47454d
DM
2984 SvPOK_on(sv_accept_buff);
2985 sv_2mortal(sv_accept_buff);
166ba7cd
DM
2986 SAVETMPS;
2987 ST.accept_buff =
5b47454d
DM
2988 (reg_trie_accepted*)SvPV_nolen(sv_accept_buff );
2989 }
786e8c11 2990 do {
166ba7cd 2991 if (ST.accepted >= bufflen) {
5b47454d 2992 bufflen *= 2;
166ba7cd 2993 ST.accept_buff =(reg_trie_accepted*)
5b47454d
DM
2994 SvGROW(sv_accept_buff,
2995 bufflen * sizeof(reg_trie_accepted));
2996 }
2997 SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff)
2998 + sizeof(reg_trie_accepted));
a3621e74 2999
786e8c11
YO
3000
3001 ST.accept_buff[ST.accepted].wordnum = got_wordnum;
3002 ST.accept_buff[ST.accepted].endpos = uc;
3003 ++ST.accepted;
3004 } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum]));
3005 }
3006#undef got_wordnum
a3621e74 3007
07be1b83 3008 DEBUG_TRIE_EXECUTE_r({
786e8c11 3009 DUMP_EXEC_POS( (char *)uc, scan, do_utf8 );
a3621e74 3010 PerlIO_printf( Perl_debug_log,
786e8c11 3011 "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ",
5bc10b2c 3012 2+depth * 2, "", PL_colors[4],
786e8c11 3013 (UV)state, (UV)ST.accepted );
07be1b83 3014 });
a3621e74
YO
3015
3016 if ( base ) {
55eed653
NC
3017 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
3018 uscan, len, uvc, charid, foldlen,
3019 foldbuf, uniflags);
a3621e74 3020
5b47454d
DM
3021 if (charid &&
3022 (base + charid > trie->uniquecharcount )
3023 && (base + charid - 1 - trie->uniquecharcount
3024 < trie->lasttrans)
3025 && trie->trans[base + charid - 1 -
3026 trie->uniquecharcount].check == state)
3027 {
3028 state = trie->trans[base + charid - 1 -
3029 trie->uniquecharcount ].next;
3030 }
3031 else {
3032 state = 0;
3033 }
3034 uc += len;
3035
3036 }
3037 else {
a3621e74
YO
3038 state = 0;
3039 }
3040 DEBUG_TRIE_EXECUTE_r(
e4584336 3041 PerlIO_printf( Perl_debug_log,
786e8c11 3042 "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
e4584336 3043 charid, uvc, (UV)state, PL_colors[5] );
a3621e74
YO
3044 );
3045 }
166ba7cd 3046 if (!ST.accepted )
a3621e74 3047 sayNO;
a3621e74 3048
166ba7cd
DM
3049 DEBUG_EXECUTE_r(
3050 PerlIO_printf( Perl_debug_log,
3051 "%*s %sgot %"IVdf" possible matches%s\n",
5bc10b2c 3052 REPORT_CODE_OFF + depth * 2, "",
166ba7cd
DM
3053 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
3054 );
786e8c11 3055 }}
fae667d5
YO
3056 goto trie_first_try; /* jump into the fail handler */
3057 /* NOTREACHED */
166ba7cd 3058 case TRIE_next_fail: /* we failed - try next alterative */
fae667d5
YO
3059 if ( ST.jump) {
3060 REGCP_UNWIND(ST.cp);
3061 for (n = *PL_reglastparen; n > ST.lastparen; n--)
f0ab9afb 3062 PL_regoffs[n].end = -1;
fae667d5
YO
3063 *PL_reglastparen = n;
3064 }
3065 trie_first_try:
5d458dd8
YO
3066 if (do_cutgroup) {
3067 do_cutgroup = 0;
3068 no_final = 0;
3069 }
fae667d5
YO
3070
3071 if ( ST.jump) {
3072 ST.lastparen = *PL_reglastparen;
3073 REGCP_SET(ST.cp);
3074 }
166ba7cd
DM
3075 if ( ST.accepted == 1 ) {
3076 /* only one choice left - just continue */
3077 DEBUG_EXECUTE_r({
2b8b4781
NC
3078 AV *const trie_words
3079 = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET];
3080 SV ** const tmp = av_fetch( trie_words,
5c9f2f80 3081 ST.accept_buff[ 0 ].wordnum-1, 0 );
de734bd5
A
3082 SV *sv= tmp ? sv_newmortal() : NULL;
3083
166ba7cd
DM
3084 PerlIO_printf( Perl_debug_log,
3085 "%*s %sonly one match left: #%d <%s>%s\n",
5bc10b2c 3086 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
166ba7cd 3087 ST.accept_buff[ 0 ].wordnum,
de734bd5
A
3088 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
3089 PL_colors[0], PL_colors[1],
3090 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
3091 )
3092 : "not compiled under -Dr",
166ba7cd
DM
3093 PL_colors[5] );
3094 });
3095 PL_reginput = (char *)ST.accept_buff[ 0 ].endpos;
3096 /* in this case we free tmps/leave before we call regmatch
3097 as we wont be using accept_buff again. */
5d458dd8 3098
166ba7cd
DM
3099 locinput = PL_reginput;
3100 nextchr = UCHARAT(locinput);
5d458dd8
YO
3101 if ( !ST.jump || !ST.jump[ST.accept_buff[0].wordnum])
3102 scan = ST.B;
3103 else
3104 scan = ST.me + ST.jump[ST.accept_buff[0].wordnum];
3105 if (!has_cutgroup) {
3106 FREETMPS;
3107 LEAVE;
3108 } else {
3109 ST.accepted--;
3110 PUSH_YES_STATE_GOTO(TRIE_next, scan);
3111 }
786e8c11 3112
166ba7cd
DM
3113 continue; /* execute rest of RE */
3114 }
fae667d5
YO
3115
3116 if ( !ST.accepted-- ) {
5d458dd8
YO
3117 DEBUG_EXECUTE_r({
3118 PerlIO_printf( Perl_debug_log,
3119 "%*s %sTRIE failed...%s\n",
3120 REPORT_CODE_OFF+depth*2, "",
3121 PL_colors[4],
3122 PL_colors[5] );
3123 });
166ba7cd
DM
3124 FREETMPS;
3125 LEAVE;
5d458dd8 3126 sayNO_SILENT;
fae667d5
YO
3127 /*NOTREACHED*/
3128 }
166ba7cd 3129
a3621e74 3130 /*
166ba7cd
DM
3131 There are at least two accepting states left. Presumably
3132 the number of accepting states is going to be low,
3133 typically two. So we simply scan through to find the one
3134 with lowest wordnum. Once we find it, we swap the last
3135 state into its place and decrement the size. We then try to
3136 match the rest of the pattern at the point where the word
3137 ends. If we succeed, control just continues along the
3138 regex; if we fail we return here to try the next accepting
3139 state
3140 */
a3621e74 3141
166ba7cd
DM
3142 {
3143 U32 best = 0;
3144 U32 cur;
3145 for( cur = 1 ; cur <= ST.accepted ; cur++ ) {
3146 DEBUG_TRIE_EXECUTE_r(
f2278c82 3147 PerlIO_printf( Perl_debug_log,
166ba7cd 3148 "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n",
5bc10b2c 3149 REPORT_CODE_OFF + depth * 2, "", PL_colors[4],
166ba7cd
DM
3150 (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur,
3151 ST.accept_buff[ cur ].wordnum, PL_colors[5] );
3152 );
3153
3154 if (ST.accept_buff[cur].wordnum <
3155 ST.accept_buff[best].wordnum)
3156 best = cur;
a3621e74 3157 }
166ba7cd
DM
3158
3159 DEBUG_EXECUTE_r({
2b8b4781
NC
3160 AV *const trie_words
3161 = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET];
3162 SV ** const tmp = av_fetch( trie_words,
5c9f2f80 3163 ST.accept_buff[ best ].wordnum - 1, 0 );
7f69552c 3164 regnode *nextop=(!ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) ?
265c4333 3165 ST.B :
7f69552c 3166 ST.me + ST.jump[ST.accept_buff[best].wordnum];
de734bd5
A
3167 SV *sv= tmp ? sv_newmortal() : NULL;
3168
265c4333
YO
3169 PerlIO_printf( Perl_debug_log,
3170 "%*s %strying alternation #%d <%s> at node #%d %s\n",
5bc10b2c 3171 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
166ba7cd 3172 ST.accept_buff[best].wordnum,
de734bd5
A
3173 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
3174 PL_colors[0], PL_colors[1],
3175 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
3176 ) : "not compiled under -Dr",
265c4333 3177 REG_NODE_NUM(nextop),
166ba7cd
DM
3178 PL_colors[5] );
3179 });
3180
3181 if ( best<ST.accepted ) {
3182 reg_trie_accepted tmp = ST.accept_buff[ best ];
3183 ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ];
3184 ST.accept_buff[ ST.accepted ] = tmp;
3185 best = ST.accepted;
a3621e74 3186 }
166ba7cd 3187 PL_reginput = (char *)ST.accept_buff[ best ].endpos;
7f69552c 3188 if ( !ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) {
5d458dd8 3189 scan = ST.B;
786e8c11 3190 } else {
5d458dd8 3191 scan = ST.me + ST.jump[ST.accept_buff[best].wordnum];
5d458dd8 3192 }
6b173516 3193 PUSH_YES_STATE_GOTO(TRIE_next, scan);
786e8c11 3194 /* NOTREACHED */
166ba7cd 3195 }
166ba7cd 3196 /* NOTREACHED */
5d458dd8
YO
3197 case TRIE_next:
3198 FREETMPS;
3199 LEAVE;
3200 sayYES;
166ba7cd
DM
3201#undef ST
3202
95b24440
DM
3203 case EXACT: {
3204 char *s = STRING(scan);
24d3c4a9 3205 ln = STR_LEN(scan);
eb160463 3206 if (do_utf8 != UTF) {
bc517b45 3207 /* The target and the pattern have differing utf8ness. */
1aa99e6b 3208 char *l = locinput;
24d3c4a9 3209 const char * const e = s + ln;
a72c7584 3210
5ff6fc6d
JH
3211 if (do_utf8) {
3212 /* The target is utf8, the pattern is not utf8. */
1aa99e6b 3213 while (s < e) {
a3b680e6 3214 STRLEN ulen;
1aa99e6b 3215 if (l >= PL_regeol)
5ff6fc6d
JH
3216 sayNO;
3217 if (NATIVE_TO_UNI(*(U8*)s) !=
89ebb4a3 3218 utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
041457d9 3219 uniflags))
5ff6fc6d 3220 sayNO;
bc517b45 3221 l += ulen;
5ff6fc6d 3222 s ++;
1aa99e6b 3223 }
5ff6fc6d
JH
3224 }
3225 else {
3226 /* The target is not utf8, the pattern is utf8. */
1aa99e6b 3227 while (s < e) {
a3b680e6 3228 STRLEN ulen;
1aa99e6b
IH
3229 if (l >= PL_regeol)
3230 sayNO;
5ff6fc6d 3231 if (NATIVE_TO_UNI(*((U8*)l)) !=
89ebb4a3 3232 utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
041457d9 3233 uniflags))
1aa99e6b 3234 sayNO;
bc517b45 3235 s += ulen;
a72c7584 3236 l ++;
1aa99e6b 3237 }
5ff6fc6d 3238 }
1aa99e6b
IH
3239 locinput = l;
3240 nextchr = UCHARAT(locinput);
3241 break;
3242 }
bc517b45 3243 /* The target and the pattern have the same utf8ness. */
d6a28714
JH
3244 /* Inline the first character, for speed. */
3245 if (UCHARAT(s) != nextchr)
3246 sayNO;
24d3c4a9 3247 if (PL_regeol - locinput < ln)
d6a28714 3248 sayNO;
24d3c4a9 3249 if (ln > 1 && memNE(s, locinput, ln))
d6a28714 3250 sayNO;
24d3c4a9 3251 locinput += ln;
d6a28714
JH
3252 nextchr = UCHARAT(locinput);
3253 break;
95b24440 3254 }
d6a28714 3255 case EXACTFL:
b8c5462f
JH
3256 PL_reg_flags |= RF_tainted;
3257 /* FALL THROUGH */
95b24440 3258 case EXACTF: {
be8e71aa 3259 char * const s = STRING(scan);
24d3c4a9 3260 ln = STR_LEN(scan);
d6a28714 3261
d07ddd77
JH
3262 if (do_utf8 || UTF) {
3263 /* Either target or the pattern are utf8. */
be8e71aa 3264 const char * const l = locinput;
d07ddd77 3265 char *e = PL_regeol;
bc517b45 3266
24d3c4a9 3267 if (ibcmp_utf8(s, 0, ln, (bool)UTF,
1feea2c7 3268 l, &e, 0, do_utf8)) {
5486206c
JH
3269 /* One more case for the sharp s:
3270 * pack("U0U*", 0xDF) =~ /ss/i,
3271 * the 0xC3 0x9F are the UTF-8
3272 * byte sequence for the U+00DF. */
e1d1eefb 3273
5486206c 3274 if (!(do_utf8 &&
e1d1eefb 3275 toLOWER(s[0]) == 's' &&
24d3c4a9 3276 ln >= 2 &&
5486206c
JH
3277 toLOWER(s[1]) == 's' &&
3278 (U8)l[0] == 0xC3 &&
3279 e - l >= 2 &&
3280 (U8)l[1] == 0x9F))
3281 sayNO;
3282 }
d07ddd77
JH
3283 locinput = e;
3284 nextchr = UCHARAT(locinput);
3285 break;
a0ed51b3 3286 }
d6a28714 3287
bc517b45
JH
3288 /* Neither the target and the pattern are utf8. */
3289
d6a28714
JH
3290 /* Inline the first character, for speed. */
3291 if (UCHARAT(s) != nextchr &&
3292 UCHARAT(s) != ((OP(scan) == EXACTF)
3293 ? PL_fold : PL_fold_locale)[nextchr])
a0ed51b3 3294 sayNO;
24d3c4a9 3295 if (PL_regeol - locinput < ln)
b8c5462f 3296 sayNO;
24d3c4a9
DM
3297 if (ln > 1 && (OP(scan) == EXACTF
3298 ? ibcmp(s, locinput, ln)
3299 : ibcmp_locale(s, locinput, ln)))
4633a7c4 3300 sayNO;
24d3c4a9 3301 locinput += ln;
d6a28714 3302 nextchr = UCHARAT(locinput);
a0d0e21e 3303 break;
95b24440 3304 }
d6a28714 3305 case ANYOF:
ffc61ed2 3306 if (do_utf8) {
9e55ce06
JH
3307 STRLEN inclasslen = PL_regeol - locinput;
3308
32fc9b6a 3309 if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8))
262b90c4 3310 goto anyof_fail;
ffc61ed2
JH
3311 if (locinput >= PL_regeol)
3312 sayNO;
0f0076b4 3313 locinput += inclasslen ? inclasslen : UTF8SKIP(locinput);
b8c5462f 3314 nextchr = UCHARAT(locinput);
e0f9d4a8 3315 break;
ffc61ed2
JH
3316 }
3317 else {
3318 if (nextchr < 0)
3319 nextchr = UCHARAT(locinput);
32fc9b6a 3320 if (!REGINCLASS(rex, scan, (U8*)locinput))
262b90c4 3321 goto anyof_fail;
ffc61ed2
JH
3322 if (!nextchr && locinput >= PL_regeol)
3323 sayNO;
3324 nextchr = UCHARAT(++locinput);
e0f9d4a8
JH
3325 break;
3326 }
262b90c4 3327 anyof_fail:
e0f9d4a8
JH
3328 /* If we might have the case of the German sharp s
3329 * in a casefolding Unicode character class. */
3330
ebc501f0
JH
3331 if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
3332 locinput += SHARP_S_SKIP;
e0f9d4a8 3333 nextchr = UCHARAT(locinput);
ffc61ed2 3334 }
e0f9d4a8
JH
3335 else
3336 sayNO;
b8c5462f 3337 break;
d6a28714 3338 case ALNUML:
b8c5462f
JH
3339 PL_reg_flags |= RF_tainted;
3340 /* FALL THROUGH */
d6a28714 3341 case ALNUM:
b8c5462f 3342 if (!nextchr)
4633a7c4 3343 sayNO;
ffc61ed2 3344 if (do_utf8) {
1a4fad37 3345 LOAD_UTF8_CHARCLASS_ALNUM();
ffc61ed2 3346 if (!(OP(scan) == ALNUM
bb7a0f54 3347 ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
d6a28714 3348 : isALNUM_LC_utf8((U8*)locinput)))
b8c5462f
JH
3349 {
3350 sayNO;
a0ed51b3 3351 }
b8c5462f 3352 locinput += PL_utf8skip[nextchr];
a0ed51b3
LW
3353 nextchr = UCHARAT(locinput);
3354 break;
3355 }
ffc61ed2 3356 if (!(OP(scan) == ALNUM
d6a28714 3357 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
bbce6d69 3358 sayNO;
b8c5462f 3359 nextchr = UCHARAT(++locinput);
a0d0e21e 3360 break;
d6a28714 3361 case NALNUML:
b8c5462f
JH
3362 PL_reg_flags |= RF_tainted;
3363 /* FALL THROUGH */
d6a28714
JH
3364 case NALNUM:
3365 if (!nextchr && locinput >= PL_regeol)
a0ed51b3 3366 sayNO;
ffc61ed2 3367 if (do_utf8) {
1a4fad37 3368 LOAD_UTF8_CHARCLASS_ALNUM();
ffc61ed2 3369 if (OP(scan) == NALNUM
bb7a0f54 3370 ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
d6a28714
JH
3371 : isALNUM_LC_utf8((U8*)locinput))
3372 {
b8c5462f 3373 sayNO;
d6a28714 3374 }
b8c5462f
JH
3375 locinput += PL_utf8skip[nextchr];
3376 nextchr = UCHARAT(locinput);
3377 break;
3378 }
ffc61ed2 3379 if (OP(scan) == NALNUM
d6a28714 3380 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
b8c5462f 3381 sayNO;
76e3520e 3382 nextchr = UCHARAT(++locinput);
a0d0e21e 3383 break;
d6a28714
JH