1 /* regexec.c
2 */
3
4 /*
5 * One Ring to rule them all, One Ring to find them
6 *
7 * [p.v of _The Lord of the Rings_, opening poem]
8 * [p.50 of _The Lord of the Rings_, I/iii: "The Shadow of the Past"]
9 * [p.254 of _The Lord of the Rings_, II/ii: "The Council of Elrond"]
10 */
11
12 /* This file contains functions for executing a regular expression. See
13 * also regcomp.c which funnily enough, contains functions for compiling
14 * a regular expression.
15 *
16 * This file is also copied at build time to ext/re/re_exec.c, where
17 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
18 * This causes the main functions to be compiled under new names and with
19 * debugging support added, which makes "use re 'debug'" work.
20 */
21
22 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
23 * confused with the original package (see point 3 below). Thanks, Henry!
24 */
25
26 /* Additional note: this code is very heavily munged from Henry's version
27 * in places. In some spots I've traded clarity for efficiency, so don't
28 * blame Henry for some of the lack of readability.
29 */
30
31 /* The names of the functions have been changed from regcomp and
32 * regexec to pregcomp and pregexec in order to avoid conflicts
33 * with the POSIX routines of the same names.
34 */
35
36 #ifdef PERL_EXT_RE_BUILD
37 #include "re_top.h"
38 #endif
39
40 /*
41 * pregcomp and pregexec -- regsub and regerror are not used in perl
42 *
43 * Copyright (c) 1986 by University of Toronto.
44 * Written by Henry Spencer. Not derived from licensed software.
45 *
46 * Permission is granted to anyone to use this software for any
47 * purpose on any computer system, and to redistribute it freely,
48 * subject to the following restrictions:
49 *
50 * 1. The author is not responsible for the consequences of use of
51 * this software, no matter how awful, even if they arise
52 * from defects in it.
53 *
54 * 2. The origin of this software must not be misrepresented, either
55 * by explicit claim or by omission.
56 *
57 * 3. Altered versions must be plainly marked as such, and must not
58 * be misrepresented as being the original software.
59 *
60 **** Alterations to Henry's code are...
61 ****
62 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
63 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
64 **** by Larry Wall and others
65 ****
66 **** You may distribute under the terms of either the GNU General Public
67 **** License or the Artistic License, as specified in the README file.
68 *
69 * Beware that some of this code is subtly aware of the way operator
70 * precedence is structured in regular expressions. Serious changes in
71 * regular-expression syntax might require a total rethink.
72 */
73 #include "EXTERN.h"
74 #define PERL_IN_REGEXEC_C
75 #include "perl.h"
76
77 #ifdef PERL_IN_XSUB_RE
78 # include "re_comp.h"
79 #else
80 # include "regcomp.h"
81 #endif
82
83 #include "invlist_inline.h"
84 #include "unicode_constants.h"
85
86 #define B_ON_NON_UTF8_LOCALE_IS_WRONG \
87 "Use of \\b{} or \\B{} for non-UTF-8 locale is wrong. Assuming a UTF-8 locale"
88
89 static const char utf8_locale_required[] =
90 "Use of (?[ ]) for non-UTF-8 locale is wrong. Assuming a UTF-8 locale";
91
92 #ifdef DEBUGGING
93 /* At least one required character in the target string is expressible only in
94 * UTF-8. */
95 static const char non_utf8_target_but_utf8_required[]
96 = "Can't match, because target string needs to be in UTF-8\n";
97 #endif
98
99 #define NON_UTF8_TARGET_BUT_UTF8_REQUIRED(target) STMT_START { \
100 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_ "%s", non_utf8_target_but_utf8_required));\
101 goto target; \
102 } STMT_END
103
104 #define HAS_NONLATIN1_FOLD_CLOSURE(i) _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)
105
106 #ifndef STATIC
107 #define STATIC static
108 #endif
109
110 /*
111 * Forwards.
112 */
113
114 #define CHR_SVLEN(sv) (utf8_target ? sv_len_utf8(sv) : SvCUR(sv))
115
116 #define HOPc(pos,off) \
117 (char *)(reginfo->is_utf8_target \
118 ? reghop3((U8*)pos, off, \
119 (U8*)(off >= 0 ? reginfo->strend : reginfo->strbeg)) \
120 : (U8*)(pos + off))
121
122 /* like HOPMAYBE3 but backwards. lim must be +ve. Returns NULL on overshoot */
123 #define HOPBACK3(pos, off, lim) \
124 (reginfo->is_utf8_target \
125 ? reghopmaybe3((U8*)pos, (SSize_t)0-off, (U8*)(lim)) \
126 : (pos - off >= lim) \
127 ? (U8*)pos - off \
128 : NULL)
129
130 #define HOPBACKc(pos, off) ((char*)HOPBACK3(pos, off, reginfo->strbeg))
131
132 #define HOP3(pos,off,lim) (reginfo->is_utf8_target ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
133 #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
134
135 /* lim must be +ve. Returns NULL on overshoot */
136 #define HOPMAYBE3(pos,off,lim) \
137 (reginfo->is_utf8_target \
138 ? reghopmaybe3((U8*)pos, off, (U8*)(lim)) \
139 : ((U8*)pos + off <= lim) \
140 ? (U8*)pos + off \
141 : NULL)
142
143 /* like HOP3, but limits the result to <= lim even for the non-utf8 case.
144 * off must be >=0; args should be vars rather than expressions */
145 #define HOP3lim(pos,off,lim) (reginfo->is_utf8_target \
146 ? reghop3((U8*)(pos), off, (U8*)(lim)) \
147 : (U8*)((pos + off) > lim ? lim : (pos + off)))
148 #define HOP3clim(pos,off,lim) ((char*)HOP3lim(pos,off,lim))
149
150 #define HOP4(pos,off,llim, rlim) (reginfo->is_utf8_target \
151 ? reghop4((U8*)(pos), off, (U8*)(llim), (U8*)(rlim)) \
152 : (U8*)(pos + off))
153 #define HOP4c(pos,off,llim, rlim) ((char*)HOP4(pos,off,llim, rlim))
154
155 #define PLACEHOLDER /* Something for the preprocessor to grab onto */
156 /* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
157
158 /* for use after a quantifier and before an EXACT-like node -- japhy */
159 /* it would be nice to rework regcomp.sym to generate this stuff. sigh
160 *
161 * NOTE that *nothing* that affects backtracking should be in here, specifically
162 * VERBS must NOT be included. JUMPABLE is used to determine if we can ignore a
163 * node that is in between two EXACT like nodes when ascertaining what the required
164 * "follow" character is. This should probably be moved to regex compile time
165 * although it may be done at run time beause of the REF possibility - more
166 * investigation required. -- demerphq
167 */
168 #define JUMPABLE(rn) ( \
169 OP(rn) == OPEN || \
170 (OP(rn) == CLOSE && \
171 !EVAL_CLOSE_PAREN_IS(cur_eval,ARG(rn)) ) || \
172 OP(rn) == EVAL || \
173 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
174 OP(rn) == PLUS || OP(rn) == MINMOD || \
175 OP(rn) == KEEPS || \
176 (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
177 )
178 #define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT)
179
180 #define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF )
181
182 /*
183 Search for mandatory following text node; for lookahead, the text must
184 follow but for lookbehind (rn->flags != 0) we skip to the next step.
185 */
186 #define FIND_NEXT_IMPT(rn) STMT_START { \
187 while (JUMPABLE(rn)) { \
188 const OPCODE type = OP(rn); \
189 if (type == SUSPEND || PL_regkind[type] == CURLY) \
190 rn = NEXTOPER(NEXTOPER(rn)); \
191 else if (type == PLUS) \
192 rn = NEXTOPER(rn); \
193 else if (type == IFMATCH) \
194 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
195 else rn += NEXT_OFF(rn); \
196 } \
197 } STMT_END
198
199 #define SLAB_FIRST(s) (&(s)->states[0])
200 #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
201
202 static void S_setup_eval_state(pTHX_ regmatch_info *const reginfo);
203 static void S_cleanup_regmatch_info_aux(pTHX_ void *arg);
204 static regmatch_state * S_push_slab(pTHX);
205
206 #define REGCP_PAREN_ELEMS 3
207 #define REGCP_OTHER_ELEMS 3
208 #define REGCP_FRAME_ELEMS 1
209 /* REGCP_FRAME_ELEMS are not part of the REGCP_OTHER_ELEMS and
210 * are needed for the regexp context stack bookkeeping. */
211
212 STATIC CHECKPOINT
S_regcppush(pTHX_ const regexp * rex,I32 parenfloor,U32 maxopenparen _pDEPTH)213 S_regcppush(pTHX_ const regexp *rex, I32 parenfloor, U32 maxopenparen _pDEPTH)
214 {
215 const int retval = PL_savestack_ix;
216 const int paren_elems_to_push =
217 (maxopenparen - parenfloor) * REGCP_PAREN_ELEMS;
218 const UV total_elems = paren_elems_to_push + REGCP_OTHER_ELEMS;
219 const UV elems_shifted = total_elems << SAVE_TIGHT_SHIFT;
220 I32 p;
221 GET_RE_DEBUG_FLAGS_DECL;
222
223 PERL_ARGS_ASSERT_REGCPPUSH;
224
225 if (paren_elems_to_push < 0)
226 Perl_croak(aTHX_ "panic: paren_elems_to_push, %i < 0, maxopenparen: %i parenfloor: %i REGCP_PAREN_ELEMS: %u",
227 (int)paren_elems_to_push, (int)maxopenparen,
228 (int)parenfloor, (unsigned)REGCP_PAREN_ELEMS);
229
230 if ((elems_shifted >> SAVE_TIGHT_SHIFT) != total_elems)
231 Perl_croak(aTHX_ "panic: paren_elems_to_push offset %" UVuf
232 " out of range (%lu-%ld)",
233 total_elems,
234 (unsigned long)maxopenparen,
235 (long)parenfloor);
236
237 SSGROW(total_elems + REGCP_FRAME_ELEMS);
238
239 DEBUG_BUFFERS_r(
240 if ((int)maxopenparen > (int)parenfloor)
241 Perl_re_exec_indentf( aTHX_
242 "rex=0x%" UVxf " offs=0x%" UVxf ": saving capture indices:\n",
243 depth,
244 PTR2UV(rex),
245 PTR2UV(rex->offs)
246 );
247 );
248 for (p = parenfloor+1; p <= (I32)maxopenparen; p++) {
249 /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
250 SSPUSHIV(rex->offs[p].end);
251 SSPUSHIV(rex->offs[p].start);
252 SSPUSHINT(rex->offs[p].start_tmp);
253 DEBUG_BUFFERS_r(Perl_re_exec_indentf( aTHX_
254 " \\%" UVuf ": %" IVdf "(%" IVdf ")..%" IVdf "\n",
255 depth,
256 (UV)p,
257 (IV)rex->offs[p].start,
258 (IV)rex->offs[p].start_tmp,
259 (IV)rex->offs[p].end
260 ));
261 }
262 /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
263 SSPUSHINT(maxopenparen);
264 SSPUSHINT(rex->lastparen);
265 SSPUSHINT(rex->lastcloseparen);
266 SSPUSHUV(SAVEt_REGCONTEXT | elems_shifted); /* Magic cookie. */
267
268 return retval;
269 }
270
271 /* These are needed since we do not localize EVAL nodes: */
272 #define REGCP_SET(cp) \
273 DEBUG_STATE_r( \
274 Perl_re_exec_indentf( aTHX_ \
275 "Setting an EVAL scope, savestack=%" IVdf ",\n", \
276 depth, (IV)PL_savestack_ix \
277 ) \
278 ); \
279 cp = PL_savestack_ix
280
281 #define REGCP_UNWIND(cp) \
282 DEBUG_STATE_r( \
283 if (cp != PL_savestack_ix) \
284 Perl_re_exec_indentf( aTHX_ \
285 "Clearing an EVAL scope, savestack=%" \
286 IVdf "..%" IVdf "\n", \
287 depth, (IV)(cp), (IV)PL_savestack_ix \
288 ) \
289 ); \
290 regcpblow(cp)
291
292 /* set the start and end positions of capture ix */
293 #define CLOSE_CAPTURE(ix, s, e) \
294 rex->offs[ix].start = s; \
295 rex->offs[ix].end = e; \
296 if (ix > rex->lastparen) \
297 rex->lastparen = ix; \
298 rex->lastcloseparen = ix; \
299 DEBUG_BUFFERS_r(Perl_re_exec_indentf( aTHX_ \
300 "CLOSE: rex=0x%" UVxf " offs=0x%" UVxf ": \\%" UVuf ": set %" IVdf "..%" IVdf " max: %" UVuf "\n", \
301 depth, \
302 PTR2UV(rex), \
303 PTR2UV(rex->offs), \
304 (UV)ix, \
305 (IV)rex->offs[ix].start, \
306 (IV)rex->offs[ix].end, \
307 (UV)rex->lastparen \
308 ))
309
310 #define UNWIND_PAREN(lp, lcp) \
311 DEBUG_BUFFERS_r(Perl_re_exec_indentf( aTHX_ \
312 "UNWIND_PAREN: rex=0x%" UVxf " offs=0x%" UVxf ": invalidate (%" UVuf "..%" UVuf "] set lcp: %" UVuf "\n", \
313 depth, \
314 PTR2UV(rex), \
315 PTR2UV(rex->offs), \
316 (UV)(lp), \
317 (UV)(rex->lastparen), \
318 (UV)(lcp) \
319 )); \
320 for (n = rex->lastparen; n > lp; n--) \
321 rex->offs[n].end = -1; \
322 rex->lastparen = n; \
323 rex->lastcloseparen = lcp;
324
325
326 STATIC void
S_regcppop(pTHX_ regexp * rex,U32 * maxopenparen_p _pDEPTH)327 S_regcppop(pTHX_ regexp *rex, U32 *maxopenparen_p _pDEPTH)
328 {
329 UV i;
330 U32 paren;
331 GET_RE_DEBUG_FLAGS_DECL;
332
333 PERL_ARGS_ASSERT_REGCPPOP;
334
335 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
336 i = SSPOPUV;
337 assert((i & SAVE_MASK) == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
338 i >>= SAVE_TIGHT_SHIFT; /* Parentheses elements to pop. */
339 rex->lastcloseparen = SSPOPINT;
340 rex->lastparen = SSPOPINT;
341 *maxopenparen_p = SSPOPINT;
342
343 i -= REGCP_OTHER_ELEMS;
344 /* Now restore the parentheses context. */
345 DEBUG_BUFFERS_r(
346 if (i || rex->lastparen + 1 <= rex->nparens)
347 Perl_re_exec_indentf( aTHX_
348 "rex=0x%" UVxf " offs=0x%" UVxf ": restoring capture indices to:\n",
349 depth,
350 PTR2UV(rex),
351 PTR2UV(rex->offs)
352 );
353 );
354 paren = *maxopenparen_p;
355 for ( ; i > 0; i -= REGCP_PAREN_ELEMS) {
356 SSize_t tmps;
357 rex->offs[paren].start_tmp = SSPOPINT;
358 rex->offs[paren].start = SSPOPIV;
359 tmps = SSPOPIV;
360 if (paren <= rex->lastparen)
361 rex->offs[paren].end = tmps;
362 DEBUG_BUFFERS_r( Perl_re_exec_indentf( aTHX_
363 " \\%" UVuf ": %" IVdf "(%" IVdf ")..%" IVdf "%s\n",
364 depth,
365 (UV)paren,
366 (IV)rex->offs[paren].start,
367 (IV)rex->offs[paren].start_tmp,
368 (IV)rex->offs[paren].end,
369 (paren > rex->lastparen ? "(skipped)" : ""));
370 );
371 paren--;
372 }
373 #if 1
374 /* It would seem that the similar code in regtry()
375 * already takes care of this, and in fact it is in
376 * a better location to since this code can #if 0-ed out
377 * but the code in regtry() is needed or otherwise tests
378 * requiring null fields (pat.t#187 and split.t#{13,14}
379 * (as of patchlevel 7877) will fail. Then again,
380 * this code seems to be necessary or otherwise
381 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
382 * --jhi updated by dapm */
383 for (i = rex->lastparen + 1; i <= rex->nparens; i++) {
384 if (i > *maxopenparen_p)
385 rex->offs[i].start = -1;
386 rex->offs[i].end = -1;
387 DEBUG_BUFFERS_r( Perl_re_exec_indentf( aTHX_
388 " \\%" UVuf ": %s ..-1 undeffing\n",
389 depth,
390 (UV)i,
391 (i > *maxopenparen_p) ? "-1" : " "
392 ));
393 }
394 #endif
395 }
396
397 /* restore the parens and associated vars at savestack position ix,
398 * but without popping the stack */
399
400 STATIC void
S_regcp_restore(pTHX_ regexp * rex,I32 ix,U32 * maxopenparen_p _pDEPTH)401 S_regcp_restore(pTHX_ regexp *rex, I32 ix, U32 *maxopenparen_p _pDEPTH)
402 {
403 I32 tmpix = PL_savestack_ix;
404 PERL_ARGS_ASSERT_REGCP_RESTORE;
405
406 PL_savestack_ix = ix;
407 regcppop(rex, maxopenparen_p);
408 PL_savestack_ix = tmpix;
409 }
410
411 #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
412
413 #ifndef PERL_IN_XSUB_RE
414
415 bool
Perl_isFOO_lc(pTHX_ const U8 classnum,const U8 character)416 Perl_isFOO_lc(pTHX_ const U8 classnum, const U8 character)
417 {
418 /* Returns a boolean as to whether or not 'character' is a member of the
419 * Posix character class given by 'classnum' that should be equivalent to a
420 * value in the typedef '_char_class_number'.
421 *
422 * Ideally this could be replaced by a just an array of function pointers
423 * to the C library functions that implement the macros this calls.
424 * However, to compile, the precise function signatures are required, and
425 * these may vary from platform to to platform. To avoid having to figure
426 * out what those all are on each platform, I (khw) am using this method,
427 * which adds an extra layer of function call overhead (unless the C
428 * optimizer strips it away). But we don't particularly care about
429 * performance with locales anyway. */
430
431 switch ((_char_class_number) classnum) {
432 case _CC_ENUM_ALPHANUMERIC: return isALPHANUMERIC_LC(character);
433 case _CC_ENUM_ALPHA: return isALPHA_LC(character);
434 case _CC_ENUM_ASCII: return isASCII_LC(character);
435 case _CC_ENUM_BLANK: return isBLANK_LC(character);
436 case _CC_ENUM_CASED: return isLOWER_LC(character)
437 || isUPPER_LC(character);
438 case _CC_ENUM_CNTRL: return isCNTRL_LC(character);
439 case _CC_ENUM_DIGIT: return isDIGIT_LC(character);
440 case _CC_ENUM_GRAPH: return isGRAPH_LC(character);
441 case _CC_ENUM_LOWER: return isLOWER_LC(character);
442 case _CC_ENUM_PRINT: return isPRINT_LC(character);
443 case _CC_ENUM_PUNCT: return isPUNCT_LC(character);
444 case _CC_ENUM_SPACE: return isSPACE_LC(character);
445 case _CC_ENUM_UPPER: return isUPPER_LC(character);
446 case _CC_ENUM_WORDCHAR: return isWORDCHAR_LC(character);
447 case _CC_ENUM_XDIGIT: return isXDIGIT_LC(character);
448 default: /* VERTSPACE should never occur in locales */
449 Perl_croak(aTHX_ "panic: isFOO_lc() has an unexpected character class '%d'", classnum);
450 }
451
452 NOT_REACHED; /* NOTREACHED */
453 return FALSE;
454 }
455
456 #endif
457
458 PERL_STATIC_INLINE I32
S_foldEQ_latin1_s2_folded(const char * s1,const char * s2,I32 len)459 S_foldEQ_latin1_s2_folded(const char *s1, const char *s2, I32 len)
460 {
461 /* Compare non-UTF-8 using Unicode (Latin1) semantics. s2 must already be
462 * folded. Works on all folds representable without UTF-8, except for
463 * LATIN_SMALL_LETTER_SHARP_S, and does not check for this. Nor does it
464 * check that the strings each have at least 'len' characters.
465 *
466 * There is almost an identical API function where s2 need not be folded:
467 * Perl_foldEQ_latin1() */
468
469 const U8 *a = (const U8 *)s1;
470 const U8 *b = (const U8 *)s2;
471
472 PERL_ARGS_ASSERT_FOLDEQ_LATIN1_S2_FOLDED;
473
474 assert(len >= 0);
475
476 while (len--) {
477 assert(! isUPPER_L1(*b));
478 if (toLOWER_L1(*a) != *b) {
479 return 0;
480 }
481 a++, b++;
482 }
483 return 1;
484 }
485
486 STATIC bool
S_isFOO_utf8_lc(pTHX_ const U8 classnum,const U8 * character,const U8 * e)487 S_isFOO_utf8_lc(pTHX_ const U8 classnum, const U8* character, const U8* e)
488 {
489 /* Returns a boolean as to whether or not the (well-formed) UTF-8-encoded
490 * 'character' is a member of the Posix character class given by 'classnum'
491 * that should be equivalent to a value in the typedef
492 * '_char_class_number'.
493 *
494 * This just calls isFOO_lc on the code point for the character if it is in
495 * the range 0-255. Outside that range, all characters use Unicode
496 * rules, ignoring any locale. So use the Unicode function if this class
497 * requires an inversion list, and use the Unicode macro otherwise. */
498
499 dVAR;
500
501 PERL_ARGS_ASSERT_ISFOO_UTF8_LC;
502
503 if (UTF8_IS_INVARIANT(*character)) {
504 return isFOO_lc(classnum, *character);
505 }
506 else if (UTF8_IS_DOWNGRADEABLE_START(*character)) {
507 return isFOO_lc(classnum,
508 EIGHT_BIT_UTF8_TO_NATIVE(*character, *(character + 1)));
509 }
510
511 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(character, e);
512
513 switch ((_char_class_number) classnum) {
514 case _CC_ENUM_SPACE: return is_XPERLSPACE_high(character);
515 case _CC_ENUM_BLANK: return is_HORIZWS_high(character);
516 case _CC_ENUM_XDIGIT: return is_XDIGIT_high(character);
517 case _CC_ENUM_VERTSPACE: return is_VERTWS_high(character);
518 default:
519 return _invlist_contains_cp(PL_XPosix_ptrs[classnum],
520 utf8_to_uvchr_buf(character, e, NULL));
521 }
522
523 return FALSE; /* Things like CNTRL are always below 256 */
524 }
525
526 STATIC U8 *
S_find_span_end(U8 * s,const U8 * send,const U8 span_byte)527 S_find_span_end(U8 * s, const U8 * send, const U8 span_byte)
528 {
529 /* Returns the position of the first byte in the sequence between 's' and
530 * 'send-1' inclusive that isn't 'span_byte'; returns 'send' if none found.
531 * */
532
533 PERL_ARGS_ASSERT_FIND_SPAN_END;
534
535 assert(send >= s);
536
537 if ((STRLEN) (send - s) >= PERL_WORDSIZE
538 + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(s)
539 - (PTR2nat(s) & PERL_WORD_BOUNDARY_MASK))
540 {
541 PERL_UINTMAX_T span_word;
542
543 /* Process per-byte until reach word boundary. XXX This loop could be
544 * eliminated if we knew that this platform had fast unaligned reads */
545 while (PTR2nat(s) & PERL_WORD_BOUNDARY_MASK) {
546 if (*s != span_byte) {
547 return s;
548 }
549 s++;
550 }
551
552 /* Create a word filled with the bytes we are spanning */
553 span_word = PERL_COUNT_MULTIPLIER * span_byte;
554
555 /* Process per-word as long as we have at least a full word left */
556 do {
557
558 /* Keep going if the whole word is composed of 'span_byte's */
559 if ((* (PERL_UINTMAX_T *) s) == span_word) {
560 s += PERL_WORDSIZE;
561 continue;
562 }
563
564 /* Here, at least one byte in the word isn't 'span_byte'. */
565
566 #ifdef EBCDIC
567
568 break;
569
570 #else
571
572 /* This xor leaves 1 bits only in those non-matching bytes */
573 span_word ^= * (PERL_UINTMAX_T *) s;
574
575 /* Make sure the upper bit of each non-matching byte is set. This
576 * makes each such byte look like an ASCII platform variant byte */
577 span_word |= span_word << 1;
578 span_word |= span_word << 2;
579 span_word |= span_word << 4;
580
581 /* That reduces the problem to what this function solves */
582 return s + _variant_byte_number(span_word);
583
584 #endif
585
586 } while (s + PERL_WORDSIZE <= send);
587 }
588
589 /* Process the straggler bytes beyond the final word boundary */
590 while (s < send) {
591 if (*s != span_byte) {
592 return s;
593 }
594 s++;
595 }
596
597 return s;
598 }
599
600 STATIC U8 *
S_find_next_masked(U8 * s,const U8 * send,const U8 byte,const U8 mask)601 S_find_next_masked(U8 * s, const U8 * send, const U8 byte, const U8 mask)
602 {
603 /* Returns the position of the first byte in the sequence between 's'
604 * and 'send-1' inclusive that when ANDed with 'mask' yields 'byte';
605 * returns 'send' if none found. It uses word-level operations instead of
606 * byte to speed up the process */
607
608 PERL_ARGS_ASSERT_FIND_NEXT_MASKED;
609
610 assert(send >= s);
611 assert((byte & mask) == byte);
612
613 #ifndef EBCDIC
614
615 if ((STRLEN) (send - s) >= PERL_WORDSIZE
616 + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(s)
617 - (PTR2nat(s) & PERL_WORD_BOUNDARY_MASK))
618 {
619 PERL_UINTMAX_T word, mask_word;
620
621 while (PTR2nat(s) & PERL_WORD_BOUNDARY_MASK) {
622 if (((*s) & mask) == byte) {
623 return s;
624 }
625 s++;
626 }
627
628 word = PERL_COUNT_MULTIPLIER * byte;
629 mask_word = PERL_COUNT_MULTIPLIER * mask;
630
631 do {
632 PERL_UINTMAX_T masked = (* (PERL_UINTMAX_T *) s) & mask_word;
633
634 /* If 'masked' contains bytes with the bit pattern of 'byte' within
635 * it, xoring with 'word' will leave each of the 8 bits in such
636 * bytes be 0, and no byte containing any other bit pattern will be
637 * 0. */
638 masked ^= word;
639
640 /* This causes the most significant bit to be set to 1 for any
641 * bytes in the word that aren't completely 0 */
642 masked |= masked << 1;
643 masked |= masked << 2;
644 masked |= masked << 4;
645
646 /* The msbits are the same as what marks a byte as variant, so we
647 * can use this mask. If all msbits are 1, the word doesn't
648 * contain 'byte' */
649 if ((masked & PERL_VARIANTS_WORD_MASK) == PERL_VARIANTS_WORD_MASK) {
650 s += PERL_WORDSIZE;
651 continue;
652 }
653
654 /* Here, the msbit of bytes in the word that aren't 'byte' are 1,
655 * and any that are, are 0. Complement and re-AND to swap that */
656 masked = ~ masked;
657 masked &= PERL_VARIANTS_WORD_MASK;
658
659 /* This reduces the problem to that solved by this function */
660 s += _variant_byte_number(masked);
661 return s;
662
663 } while (s + PERL_WORDSIZE <= send);
664 }
665
666 #endif
667
668 while (s < send) {
669 if (((*s) & mask) == byte) {
670 return s;
671 }
672 s++;
673 }
674
675 return s;
676 }
677
678 STATIC U8 *
S_find_span_end_mask(U8 * s,const U8 * send,const U8 span_byte,const U8 mask)679 S_find_span_end_mask(U8 * s, const U8 * send, const U8 span_byte, const U8 mask)
680 {
681 /* Returns the position of the first byte in the sequence between 's' and
682 * 'send-1' inclusive that when ANDed with 'mask' isn't 'span_byte'.
683 * 'span_byte' should have been ANDed with 'mask' in the call of this
684 * function. Returns 'send' if none found. Works like find_span_end(),
685 * except for the AND */
686
687 PERL_ARGS_ASSERT_FIND_SPAN_END_MASK;
688
689 assert(send >= s);
690 assert((span_byte & mask) == span_byte);
691
692 if ((STRLEN) (send - s) >= PERL_WORDSIZE
693 + PERL_WORDSIZE * PERL_IS_SUBWORD_ADDR(s)
694 - (PTR2nat(s) & PERL_WORD_BOUNDARY_MASK))
695 {
696 PERL_UINTMAX_T span_word, mask_word;
697
698 while (PTR2nat(s) & PERL_WORD_BOUNDARY_MASK) {
699 if (((*s) & mask) != span_byte) {
700 return s;
701 }
702 s++;
703 }
704
705 span_word = PERL_COUNT_MULTIPLIER * span_byte;
706 mask_word = PERL_COUNT_MULTIPLIER * mask;
707
708 do {
709 PERL_UINTMAX_T masked = (* (PERL_UINTMAX_T *) s) & mask_word;
710
711 if (masked == span_word) {
712 s += PERL_WORDSIZE;
713 continue;
714 }
715
716 #ifdef EBCDIC
717
718 break;
719
720 #else
721
722 masked ^= span_word;
723 masked |= masked << 1;
724 masked |= masked << 2;
725 masked |= masked << 4;
726 return s + _variant_byte_number(masked);
727
728 #endif
729
730 } while (s + PERL_WORDSIZE <= send);
731 }
732
733 while (s < send) {
734 if (((*s) & mask) != span_byte) {
735 return s;
736 }
737 s++;
738 }
739
740 return s;
741 }
742
743 /*
744 * pregexec and friends
745 */
746
747 #ifndef PERL_IN_XSUB_RE
748 /*
749 - pregexec - match a regexp against a string
750 */
751 I32
Perl_pregexec(pTHX_ REGEXP * const prog,char * stringarg,char * strend,char * strbeg,SSize_t minend,SV * screamer,U32 nosave)752 Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, char *strend,
753 char *strbeg, SSize_t minend, SV *screamer, U32 nosave)
754 /* stringarg: the point in the string at which to begin matching */
755 /* strend: pointer to null at end of string */
756 /* strbeg: real beginning of string */
757 /* minend: end of match must be >= minend bytes after stringarg. */
758 /* screamer: SV being matched: only used for utf8 flag, pos() etc; string
759 * itself is accessed via the pointers above */
760 /* nosave: For optimizations. */
761 {
762 PERL_ARGS_ASSERT_PREGEXEC;
763
764 return
765 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
766 nosave ? 0 : REXEC_COPY_STR);
767 }
768 #endif
769
770
771
772 /* re_intuit_start():
773 *
774 * Based on some optimiser hints, try to find the earliest position in the
775 * string where the regex could match.
776 *
777 * rx: the regex to match against
778 * sv: the SV being matched: only used for utf8 flag; the string
779 * itself is accessed via the pointers below. Note that on
780 * something like an overloaded SV, SvPOK(sv) may be false
781 * and the string pointers may point to something unrelated to
782 * the SV itself.
783 * strbeg: real beginning of string
784 * strpos: the point in the string at which to begin matching
785 * strend: pointer to the byte following the last char of the string
786 * flags currently unused; set to 0
787 * data: currently unused; set to NULL
788 *
789 * The basic idea of re_intuit_start() is to use some known information
790 * about the pattern, namely:
791 *
792 * a) the longest known anchored substring (i.e. one that's at a
793 * constant offset from the beginning of the pattern; but not
794 * necessarily at a fixed offset from the beginning of the
795 * string);
796 * b) the longest floating substring (i.e. one that's not at a constant
797 * offset from the beginning of the pattern);
798 * c) Whether the pattern is anchored to the string; either
799 * an absolute anchor: /^../, or anchored to \n: /^.../m,
800 * or anchored to pos(): /\G/;
801 * d) A start class: a real or synthetic character class which
802 * represents which characters are legal at the start of the pattern;
803 *
804 * to either quickly reject the match, or to find the earliest position
805 * within the string at which the pattern might match, thus avoiding
806 * running the full NFA engine at those earlier locations, only to
807 * eventually fail and retry further along.
808 *
809 * Returns NULL if the pattern can't match, or returns the address within
810 * the string which is the earliest place the match could occur.
811 *
812 * The longest of the anchored and floating substrings is called 'check'
813 * and is checked first. The other is called 'other' and is checked
814 * second. The 'other' substring may not be present. For example,
815 *
816 * /(abc|xyz)ABC\d{0,3}DEFG/
817 *
818 * will have
819 *
820 * check substr (float) = "DEFG", offset 6..9 chars
821 * other substr (anchored) = "ABC", offset 3..3 chars
822 * stclass = [ax]
823 *
824 * Be aware that during the course of this function, sometimes 'anchored'
825 * refers to a substring being anchored relative to the start of the
826 * pattern, and sometimes to the pattern itself being anchored relative to
827 * the string. For example:
828 *
829 * /\dabc/: "abc" is anchored to the pattern;
830 * /^\dabc/: "abc" is anchored to the pattern and the string;
831 * /\d+abc/: "abc" is anchored to neither the pattern nor the string;
832 * /^\d+abc/: "abc" is anchored to neither the pattern nor the string,
833 * but the pattern is anchored to the string.
834 */
835
836 char *
Perl_re_intuit_start(pTHX_ REGEXP * const rx,SV * sv,const char * const strbeg,char * strpos,char * strend,const U32 flags,re_scream_pos_data * data)837 Perl_re_intuit_start(pTHX_
838 REGEXP * const rx,
839 SV *sv,
840 const char * const strbeg,
841 char *strpos,
842 char *strend,
843 const U32 flags,
844 re_scream_pos_data *data)
845 {
846 struct regexp *const prog = ReANY(rx);
847 SSize_t start_shift = prog->check_offset_min;
848 /* Should be nonnegative! */
849 SSize_t end_shift = 0;
850 /* current lowest pos in string where the regex can start matching */
851 char *rx_origin = strpos;
852 SV *check;
853 const bool utf8_target = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
854 U8 other_ix = 1 - prog->substrs->check_ix;
855 bool ml_anch = 0;
856 char *other_last = strpos;/* latest pos 'other' substr already checked to */
857 char *check_at = NULL; /* check substr found at this pos */
858 const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
859 RXi_GET_DECL(prog,progi);
860 regmatch_info reginfo_buf; /* create some info to pass to find_byclass */
861 regmatch_info *const reginfo = ®info_buf;
862 GET_RE_DEBUG_FLAGS_DECL;
863
864 PERL_ARGS_ASSERT_RE_INTUIT_START;
865 PERL_UNUSED_ARG(flags);
866 PERL_UNUSED_ARG(data);
867
868 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
869 "Intuit: trying to determine minimum start position...\n"));
870
871 /* for now, assume that all substr offsets are positive. If at some point
872 * in the future someone wants to do clever things with lookbehind and
873 * -ve offsets, they'll need to fix up any code in this function
874 * which uses these offsets. See the thread beginning
875 * <20140113145929.GF27210@iabyn.com>
876 */
877 assert(prog->substrs->data[0].min_offset >= 0);
878 assert(prog->substrs->data[0].max_offset >= 0);
879 assert(prog->substrs->data[1].min_offset >= 0);
880 assert(prog->substrs->data[1].max_offset >= 0);
881 assert(prog->substrs->data[2].min_offset >= 0);
882 assert(prog->substrs->data[2].max_offset >= 0);
883
884 /* for now, assume that if both present, that the floating substring
885 * doesn't start before the anchored substring.
886 * If you break this assumption (e.g. doing better optimisations
887 * with lookahead/behind), then you'll need to audit the code in this
888 * function carefully first
889 */
890 assert(
891 ! ( (prog->anchored_utf8 || prog->anchored_substr)
892 && (prog->float_utf8 || prog->float_substr))
893 || (prog->float_min_offset >= prog->anchored_offset));
894
895 /* byte rather than char calculation for efficiency. It fails
896 * to quickly reject some cases that can't match, but will reject
897 * them later after doing full char arithmetic */
898 if (prog->minlen > strend - strpos) {
899 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
900 " String too short...\n"));
901 goto fail;
902 }
903
904 RXp_MATCH_UTF8_set(prog, utf8_target);
905 reginfo->is_utf8_target = cBOOL(utf8_target);
906 reginfo->info_aux = NULL;
907 reginfo->strbeg = strbeg;
908 reginfo->strend = strend;
909 reginfo->is_utf8_pat = cBOOL(RX_UTF8(rx));
910 reginfo->intuit = 1;
911 /* not actually used within intuit, but zero for safety anyway */
912 reginfo->poscache_maxiter = 0;
913
914 if (utf8_target) {
915 if ((!prog->anchored_utf8 && prog->anchored_substr)
916 || (!prog->float_utf8 && prog->float_substr))
917 to_utf8_substr(prog);
918 check = prog->check_utf8;
919 } else {
920 if (!prog->check_substr && prog->check_utf8) {
921 if (! to_byte_substr(prog)) {
922 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(fail);
923 }
924 }
925 check = prog->check_substr;
926 }
927
928 /* dump the various substring data */
929 DEBUG_OPTIMISE_MORE_r({
930 int i;
931 for (i=0; i<=2; i++) {
932 SV *sv = (utf8_target ? prog->substrs->data[i].utf8_substr
933 : prog->substrs->data[i].substr);
934 if (!sv)
935 continue;
936
937 Perl_re_printf( aTHX_
938 " substrs[%d]: min=%" IVdf " max=%" IVdf " end shift=%" IVdf
939 " useful=%" IVdf " utf8=%d [%s]\n",
940 i,
941 (IV)prog->substrs->data[i].min_offset,
942 (IV)prog->substrs->data[i].max_offset,
943 (IV)prog->substrs->data[i].end_shift,
944 BmUSEFUL(sv),
945 utf8_target ? 1 : 0,
946 SvPEEK(sv));
947 }
948 });
949
950 if (prog->intflags & PREGf_ANCH) { /* Match at \G, beg-of-str or after \n */
951
952 /* ml_anch: check after \n?
953 *
954 * A note about PREGf_IMPLICIT: on an un-anchored pattern beginning
955 * with /.*.../, these flags will have been added by the
956 * compiler:
957 * /.*abc/, /.*abc/m: PREGf_IMPLICIT | PREGf_ANCH_MBOL
958 * /.*abc/s: PREGf_IMPLICIT | PREGf_ANCH_SBOL
959 */
960 ml_anch = (prog->intflags & PREGf_ANCH_MBOL)
961 && !(prog->intflags & PREGf_IMPLICIT);
962
963 if (!ml_anch && !(prog->intflags & PREGf_IMPLICIT)) {
964 /* we are only allowed to match at BOS or \G */
965
966 /* trivially reject if there's a BOS anchor and we're not at BOS.
967 *
968 * Note that we don't try to do a similar quick reject for
969 * \G, since generally the caller will have calculated strpos
970 * based on pos() and gofs, so the string is already correctly
971 * anchored by definition; and handling the exceptions would
972 * be too fiddly (e.g. REXEC_IGNOREPOS).
973 */
974 if ( strpos != strbeg
975 && (prog->intflags & PREGf_ANCH_SBOL))
976 {
977 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
978 " Not at start...\n"));
979 goto fail;
980 }
981
982 /* in the presence of an anchor, the anchored (relative to the
983 * start of the regex) substr must also be anchored relative
984 * to strpos. So quickly reject if substr isn't found there.
985 * This works for \G too, because the caller will already have
986 * subtracted gofs from pos, and gofs is the offset from the
987 * \G to the start of the regex. For example, in /.abc\Gdef/,
988 * where substr="abcdef", pos()=3, gofs=4, offset_min=1:
989 * caller will have set strpos=pos()-4; we look for the substr
990 * at position pos()-4+1, which lines up with the "a" */
991
992 if (prog->check_offset_min == prog->check_offset_max) {
993 /* Substring at constant offset from beg-of-str... */
994 SSize_t slen = SvCUR(check);
995 char *s = HOP3c(strpos, prog->check_offset_min, strend);
996
997 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
998 " Looking for check substr at fixed offset %" IVdf "...\n",
999 (IV)prog->check_offset_min));
1000
1001 if (SvTAIL(check)) {
1002 /* In this case, the regex is anchored at the end too.
1003 * Unless it's a multiline match, the lengths must match
1004 * exactly, give or take a \n. NB: slen >= 1 since
1005 * the last char of check is \n */
1006 if (!multiline
1007 && ( strend - s > slen
1008 || strend - s < slen - 1
1009 || (strend - s == slen && strend[-1] != '\n')))
1010 {
1011 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1012 " String too long...\n"));
1013 goto fail_finish;
1014 }
1015 /* Now should match s[0..slen-2] */
1016 slen--;
1017 }
1018 if (slen && (strend - s < slen
1019 || *SvPVX_const(check) != *s
1020 || (slen > 1 && (memNE(SvPVX_const(check), s, slen)))))
1021 {
1022 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1023 " String not equal...\n"));
1024 goto fail_finish;
1025 }
1026
1027 check_at = s;
1028 goto success_at_start;
1029 }
1030 }
1031 }
1032
1033 end_shift = prog->check_end_shift;
1034
1035 #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */
1036 if (end_shift < 0)
1037 Perl_croak(aTHX_ "panic: end_shift: %" IVdf " pattern:\n%s\n ",
1038 (IV)end_shift, RX_PRECOMP(rx));
1039 #endif
1040
1041 restart:
1042
1043 /* This is the (re)entry point of the main loop in this function.
1044 * The goal of this loop is to:
1045 * 1) find the "check" substring in the region rx_origin..strend
1046 * (adjusted by start_shift / end_shift). If not found, reject
1047 * immediately.
1048 * 2) If it exists, look for the "other" substr too if defined; for
1049 * example, if the check substr maps to the anchored substr, then
1050 * check the floating substr, and vice-versa. If not found, go
1051 * back to (1) with rx_origin suitably incremented.
1052 * 3) If we find an rx_origin position that doesn't contradict
1053 * either of the substrings, then check the possible additional
1054 * constraints on rx_origin of /^.../m or a known start class.
1055 * If these fail, then depending on which constraints fail, jump
1056 * back to here, or to various other re-entry points further along
1057 * that skip some of the first steps.
1058 * 4) If we pass all those tests, update the BmUSEFUL() count on the
1059 * substring. If the start position was determined to be at the
1060 * beginning of the string - so, not rejected, but not optimised,
1061 * since we have to run regmatch from position 0 - decrement the
1062 * BmUSEFUL() count. Otherwise increment it.
1063 */
1064
1065
1066 /* first, look for the 'check' substring */
1067
1068 {
1069 U8* start_point;
1070 U8* end_point;
1071
1072 DEBUG_OPTIMISE_MORE_r({
1073 Perl_re_printf( aTHX_
1074 " At restart: rx_origin=%" IVdf " Check offset min: %" IVdf
1075 " Start shift: %" IVdf " End shift %" IVdf
1076 " Real end Shift: %" IVdf "\n",
1077 (IV)(rx_origin - strbeg),
1078 (IV)prog->check_offset_min,
1079 (IV)start_shift,
1080 (IV)end_shift,
1081 (IV)prog->check_end_shift);
1082 });
1083
1084 end_point = HOPBACK3(strend, end_shift, rx_origin);
1085 if (!end_point)
1086 goto fail_finish;
1087 start_point = HOPMAYBE3(rx_origin, start_shift, end_point);
1088 if (!start_point)
1089 goto fail_finish;
1090
1091
1092 /* If the regex is absolutely anchored to either the start of the
1093 * string (SBOL) or to pos() (ANCH_GPOS), then
1094 * check_offset_max represents an upper bound on the string where
1095 * the substr could start. For the ANCH_GPOS case, we assume that
1096 * the caller of intuit will have already set strpos to
1097 * pos()-gofs, so in this case strpos + offset_max will still be
1098 * an upper bound on the substr.
1099 */
1100 if (!ml_anch
1101 && prog->intflags & PREGf_ANCH
1102 && prog->check_offset_max != SSize_t_MAX)
1103 {
1104 SSize_t check_len = SvCUR(check) - !!SvTAIL(check);
1105 const char * const anchor =
1106 (prog->intflags & PREGf_ANCH_GPOS ? strpos : strbeg);
1107 SSize_t targ_len = (char*)end_point - anchor;
1108
1109 if (check_len > targ_len) {
1110 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1111 "Target string too short to match required substring...\n"));
1112 goto fail_finish;
1113 }
1114
1115 /* do a bytes rather than chars comparison. It's conservative;
1116 * so it skips doing the HOP if the result can't possibly end
1117 * up earlier than the old value of end_point.
1118 */
1119 assert(anchor + check_len <= (char *)end_point);
1120 if (prog->check_offset_max + check_len < targ_len) {
1121 end_point = HOP3lim((U8*)anchor,
1122 prog->check_offset_max,
1123 end_point - check_len
1124 )
1125 + check_len;
1126 if (end_point < start_point)
1127 goto fail_finish;
1128 }
1129 }
1130
1131 check_at = fbm_instr( start_point, end_point,
1132 check, multiline ? FBMrf_MULTILINE : 0);
1133
1134 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1135 " doing 'check' fbm scan, [%" IVdf "..%" IVdf "] gave %" IVdf "\n",
1136 (IV)((char*)start_point - strbeg),
1137 (IV)((char*)end_point - strbeg),
1138 (IV)(check_at ? check_at - strbeg : -1)
1139 ));
1140
1141 /* Update the count-of-usability, remove useless subpatterns,
1142 unshift s. */
1143
1144 DEBUG_EXECUTE_r({
1145 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
1146 SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
1147 Perl_re_printf( aTHX_ " %s %s substr %s%s%s",
1148 (check_at ? "Found" : "Did not find"),
1149 (check == (utf8_target ? prog->anchored_utf8 : prog->anchored_substr)
1150 ? "anchored" : "floating"),
1151 quoted,
1152 RE_SV_TAIL(check),
1153 (check_at ? " at offset " : "...\n") );
1154 });
1155
1156 if (!check_at)
1157 goto fail_finish;
1158 /* set rx_origin to the minimum position where the regex could start
1159 * matching, given the constraint of the just-matched check substring.
1160 * But don't set it lower than previously.
1161 */
1162
1163 if (check_at - rx_origin > prog->check_offset_max)
1164 rx_origin = HOP3c(check_at, -prog->check_offset_max, rx_origin);
1165 /* Finish the diagnostic message */
1166 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1167 "%ld (rx_origin now %" IVdf ")...\n",
1168 (long)(check_at - strbeg),
1169 (IV)(rx_origin - strbeg)
1170 ));
1171 }
1172
1173
1174 /* now look for the 'other' substring if defined */
1175
1176 if (prog->substrs->data[other_ix].utf8_substr
1177 || prog->substrs->data[other_ix].substr)
1178 {
1179 /* Take into account the "other" substring. */
1180 char *last, *last1;
1181 char *s;
1182 SV* must;
1183 struct reg_substr_datum *other;
1184
1185 do_other_substr:
1186 other = &prog->substrs->data[other_ix];
1187 if (!utf8_target && !other->substr) {
1188 if (!to_byte_substr(prog)) {
1189 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(fail);
1190 }
1191 }
1192
1193 /* if "other" is anchored:
1194 * we've previously found a floating substr starting at check_at.
1195 * This means that the regex origin must lie somewhere
1196 * between min (rx_origin): HOP3(check_at, -check_offset_max)
1197 * and max: HOP3(check_at, -check_offset_min)
1198 * (except that min will be >= strpos)
1199 * So the fixed substr must lie somewhere between
1200 * HOP3(min, anchored_offset)
1201 * HOP3(max, anchored_offset) + SvCUR(substr)
1202 */
1203
1204 /* if "other" is floating
1205 * Calculate last1, the absolute latest point where the
1206 * floating substr could start in the string, ignoring any
1207 * constraints from the earlier fixed match. It is calculated
1208 * as follows:
1209 *
1210 * strend - prog->minlen (in chars) is the absolute latest
1211 * position within the string where the origin of the regex
1212 * could appear. The latest start point for the floating
1213 * substr is float_min_offset(*) on from the start of the
1214 * regex. last1 simply combines thee two offsets.
1215 *
1216 * (*) You might think the latest start point should be
1217 * float_max_offset from the regex origin, and technically
1218 * you'd be correct. However, consider
1219 * /a\d{2,4}bcd\w/
1220 * Here, float min, max are 3,5 and minlen is 7.
1221 * This can match either
1222 * /a\d\dbcd\w/
1223 * /a\d\d\dbcd\w/
1224 * /a\d\d\d\dbcd\w/
1225 * In the first case, the regex matches minlen chars; in the
1226 * second, minlen+1, in the third, minlen+2.
1227 * In the first case, the floating offset is 3 (which equals
1228 * float_min), in the second, 4, and in the third, 5 (which
1229 * equals float_max). In all cases, the floating string bcd
1230 * can never start more than 4 chars from the end of the
1231 * string, which equals minlen - float_min. As the substring
1232 * starts to match more than float_min from the start of the
1233 * regex, it makes the regex match more than minlen chars,
1234 * and the two cancel each other out. So we can always use
1235 * float_min - minlen, rather than float_max - minlen for the
1236 * latest position in the string.
1237 *
1238 * Note that -minlen + float_min_offset is equivalent (AFAIKT)
1239 * to CHR_SVLEN(must) - !!SvTAIL(must) + prog->float_end_shift
1240 */
1241
1242 assert(prog->minlen >= other->min_offset);
1243 last1 = HOP3c(strend,
1244 other->min_offset - prog->minlen, strbeg);
1245
1246 if (other_ix) {/* i.e. if (other-is-float) */
1247 /* last is the latest point where the floating substr could
1248 * start, *given* any constraints from the earlier fixed
1249 * match. This constraint is that the floating string starts
1250 * <= float_max_offset chars from the regex origin (rx_origin).
1251 * If this value is less than last1, use it instead.
1252 */
1253 assert(rx_origin <= last1);
1254 last =
1255 /* this condition handles the offset==infinity case, and
1256 * is a short-cut otherwise. Although it's comparing a
1257 * byte offset to a char length, it does so in a safe way,
1258 * since 1 char always occupies 1 or more bytes,
1259 * so if a string range is (last1 - rx_origin) bytes,
1260 * it will be less than or equal to (last1 - rx_origin)
1261 * chars; meaning it errs towards doing the accurate HOP3
1262 * rather than just using last1 as a short-cut */
1263 (last1 - rx_origin) < other->max_offset
1264 ? last1
1265 : (char*)HOP3lim(rx_origin, other->max_offset, last1);
1266 }
1267 else {
1268 assert(strpos + start_shift <= check_at);
1269 last = HOP4c(check_at, other->min_offset - start_shift,
1270 strbeg, strend);
1271 }
1272
1273 s = HOP3c(rx_origin, other->min_offset, strend);
1274 if (s < other_last) /* These positions already checked */
1275 s = other_last;
1276
1277 must = utf8_target ? other->utf8_substr : other->substr;
1278 assert(SvPOK(must));
1279 {
1280 char *from = s;
1281 char *to = last + SvCUR(must) - (SvTAIL(must)!=0);
1282
1283 if (to > strend)
1284 to = strend;
1285 if (from > to) {
1286 s = NULL;
1287 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1288 " skipping 'other' fbm scan: %" IVdf " > %" IVdf "\n",
1289 (IV)(from - strbeg),
1290 (IV)(to - strbeg)
1291 ));
1292 }
1293 else {
1294 s = fbm_instr(
1295 (unsigned char*)from,
1296 (unsigned char*)to,
1297 must,
1298 multiline ? FBMrf_MULTILINE : 0
1299 );
1300 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1301 " doing 'other' fbm scan, [%" IVdf "..%" IVdf "] gave %" IVdf "\n",
1302 (IV)(from - strbeg),
1303 (IV)(to - strbeg),
1304 (IV)(s ? s - strbeg : -1)
1305 ));
1306 }
1307 }
1308
1309 DEBUG_EXECUTE_r({
1310 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
1311 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
1312 Perl_re_printf( aTHX_ " %s %s substr %s%s",
1313 s ? "Found" : "Contradicts",
1314 other_ix ? "floating" : "anchored",
1315 quoted, RE_SV_TAIL(must));
1316 });
1317
1318
1319 if (!s) {
1320 /* last1 is latest possible substr location. If we didn't
1321 * find it before there, we never will */
1322 if (last >= last1) {
1323 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1324 "; giving up...\n"));
1325 goto fail_finish;
1326 }
1327
1328 /* try to find the check substr again at a later
1329 * position. Maybe next time we'll find the "other" substr
1330 * in range too */
1331 other_last = HOP3c(last, 1, strend) /* highest failure */;
1332 rx_origin =
1333 other_ix /* i.e. if other-is-float */
1334 ? HOP3c(rx_origin, 1, strend)
1335 : HOP4c(last, 1 - other->min_offset, strbeg, strend);
1336 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1337 "; about to retry %s at offset %ld (rx_origin now %" IVdf ")...\n",
1338 (other_ix ? "floating" : "anchored"),
1339 (long)(HOP3c(check_at, 1, strend) - strbeg),
1340 (IV)(rx_origin - strbeg)
1341 ));
1342 goto restart;
1343 }
1344 else {
1345 if (other_ix) { /* if (other-is-float) */
1346 /* other_last is set to s, not s+1, since its possible for
1347 * a floating substr to fail first time, then succeed
1348 * second time at the same floating position; e.g.:
1349 * "-AB--AABZ" =~ /\wAB\d*Z/
1350 * The first time round, anchored and float match at
1351 * "-(AB)--AAB(Z)" then fail on the initial \w character
1352 * class. Second time round, they match at "-AB--A(AB)(Z)".
1353 */
1354 other_last = s;
1355 }
1356 else {
1357 rx_origin = HOP3c(s, -other->min_offset, strbeg);
1358 other_last = HOP3c(s, 1, strend);
1359 }
1360 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1361 " at offset %ld (rx_origin now %" IVdf ")...\n",
1362 (long)(s - strbeg),
1363 (IV)(rx_origin - strbeg)
1364 ));
1365
1366 }
1367 }
1368 else {
1369 DEBUG_OPTIMISE_MORE_r(
1370 Perl_re_printf( aTHX_
1371 " Check-only match: offset min:%" IVdf " max:%" IVdf
1372 " check_at:%" IVdf " rx_origin:%" IVdf " rx_origin-check_at:%" IVdf
1373 " strend:%" IVdf "\n",
1374 (IV)prog->check_offset_min,
1375 (IV)prog->check_offset_max,
1376 (IV)(check_at-strbeg),
1377 (IV)(rx_origin-strbeg),
1378 (IV)(rx_origin-check_at),
1379 (IV)(strend-strbeg)
1380 )
1381 );
1382 }
1383
1384 postprocess_substr_matches:
1385
1386 /* handle the extra constraint of /^.../m if present */
1387
1388 if (ml_anch && rx_origin != strbeg && rx_origin[-1] != '\n') {
1389 char *s;
1390
1391 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1392 " looking for /^/m anchor"));
1393
1394 /* we have failed the constraint of a \n before rx_origin.
1395 * Find the next \n, if any, even if it's beyond the current
1396 * anchored and/or floating substrings. Whether we should be
1397 * scanning ahead for the next \n or the next substr is debatable.
1398 * On the one hand you'd expect rare substrings to appear less
1399 * often than \n's. On the other hand, searching for \n means
1400 * we're effectively flipping between check_substr and "\n" on each
1401 * iteration as the current "rarest" string candidate, which
1402 * means for example that we'll quickly reject the whole string if
1403 * hasn't got a \n, rather than trying every substr position
1404 * first
1405 */
1406
1407 s = HOP3c(strend, - prog->minlen, strpos);
1408 if (s <= rx_origin ||
1409 ! ( rx_origin = (char *)memchr(rx_origin, '\n', s - rx_origin)))
1410 {
1411 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1412 " Did not find /%s^%s/m...\n",
1413 PL_colors[0], PL_colors[1]));
1414 goto fail_finish;
1415 }
1416
1417 /* earliest possible origin is 1 char after the \n.
1418 * (since *rx_origin == '\n', it's safe to ++ here rather than
1419 * HOP(rx_origin, 1)) */
1420 rx_origin++;
1421
1422 if (prog->substrs->check_ix == 0 /* check is anchored */
1423 || rx_origin >= HOP3c(check_at, - prog->check_offset_min, strpos))
1424 {
1425 /* Position contradicts check-string; either because
1426 * check was anchored (and thus has no wiggle room),
1427 * or check was float and rx_origin is above the float range */
1428 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1429 " Found /%s^%s/m, about to restart lookup for check-string with rx_origin %ld...\n",
1430 PL_colors[0], PL_colors[1], (long)(rx_origin - strbeg)));
1431 goto restart;
1432 }
1433
1434 /* if we get here, the check substr must have been float,
1435 * is in range, and we may or may not have had an anchored
1436 * "other" substr which still contradicts */
1437 assert(prog->substrs->check_ix); /* check is float */
1438
1439 if (utf8_target ? prog->anchored_utf8 : prog->anchored_substr) {
1440 /* whoops, the anchored "other" substr exists, so we still
1441 * contradict. On the other hand, the float "check" substr
1442 * didn't contradict, so just retry the anchored "other"
1443 * substr */
1444 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1445 " Found /%s^%s/m, rescanning for anchored from offset %" IVdf " (rx_origin now %" IVdf ")...\n",
1446 PL_colors[0], PL_colors[1],
1447 (IV)(rx_origin - strbeg + prog->anchored_offset),
1448 (IV)(rx_origin - strbeg)
1449 ));
1450 goto do_other_substr;
1451 }
1452
1453 /* success: we don't contradict the found floating substring
1454 * (and there's no anchored substr). */
1455 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1456 " Found /%s^%s/m with rx_origin %ld...\n",
1457 PL_colors[0], PL_colors[1], (long)(rx_origin - strbeg)));
1458 }
1459 else {
1460 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1461 " (multiline anchor test skipped)\n"));
1462 }
1463
1464 success_at_start:
1465
1466
1467 /* if we have a starting character class, then test that extra constraint.
1468 * (trie stclasses are too expensive to use here, we are better off to
1469 * leave it to regmatch itself) */
1470
1471 if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
1472 const U8* const str = (U8*)STRING(progi->regstclass);
1473
1474 /* XXX this value could be pre-computed */
1475 const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
1476 ? (reginfo->is_utf8_pat
1477 ? utf8_distance(str + STR_LEN(progi->regstclass), str)
1478 : STR_LEN(progi->regstclass))
1479 : 1);
1480 char * endpos;
1481 char *s;
1482 /* latest pos that a matching float substr constrains rx start to */
1483 char *rx_max_float = NULL;
1484
1485 /* if the current rx_origin is anchored, either by satisfying an
1486 * anchored substring constraint, or a /^.../m constraint, then we
1487 * can reject the current origin if the start class isn't found
1488 * at the current position. If we have a float-only match, then
1489 * rx_origin is constrained to a range; so look for the start class
1490 * in that range. if neither, then look for the start class in the
1491 * whole rest of the string */
1492
1493 /* XXX DAPM it's not clear what the minlen test is for, and why
1494 * it's not used in the floating case. Nothing in the test suite
1495 * causes minlen == 0 here. See <20140313134639.GS12844@iabyn.com>.
1496 * Here are some old comments, which may or may not be correct:
1497 *
1498 * minlen == 0 is possible if regstclass is \b or \B,
1499 * and the fixed substr is ''$.
1500 * Since minlen is already taken into account, rx_origin+1 is
1501 * before strend; accidentally, minlen >= 1 guaranties no false
1502 * positives at rx_origin + 1 even for \b or \B. But (minlen? 1 :
1503 * 0) below assumes that regstclass does not come from lookahead...
1504 * If regstclass takes bytelength more than 1: If charlength==1, OK.
1505 * This leaves EXACTF-ish only, which are dealt with in
1506 * find_byclass().
1507 */
1508
1509 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
1510 endpos = HOP3clim(rx_origin, (prog->minlen ? cl_l : 0), strend);
1511 else if (prog->float_substr || prog->float_utf8) {
1512 rx_max_float = HOP3c(check_at, -start_shift, strbeg);
1513 endpos = HOP3clim(rx_max_float, cl_l, strend);
1514 }
1515 else
1516 endpos= strend;
1517
1518 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1519 " looking for class: start_shift: %" IVdf " check_at: %" IVdf
1520 " rx_origin: %" IVdf " endpos: %" IVdf "\n",
1521 (IV)start_shift, (IV)(check_at - strbeg),
1522 (IV)(rx_origin - strbeg), (IV)(endpos - strbeg)));
1523
1524 s = find_byclass(prog, progi->regstclass, rx_origin, endpos,
1525 reginfo);
1526 if (!s) {
1527 if (endpos == strend) {
1528 DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
1529 " Could not match STCLASS...\n") );
1530 goto fail;
1531 }
1532 DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
1533 " This position contradicts STCLASS...\n") );
1534 if ((prog->intflags & PREGf_ANCH) && !ml_anch
1535 && !(prog->intflags & PREGf_IMPLICIT))
1536 goto fail;
1537
1538 /* Contradict one of substrings */
1539 if (prog->anchored_substr || prog->anchored_utf8) {
1540 if (prog->substrs->check_ix == 1) { /* check is float */
1541 /* Have both, check_string is floating */
1542 assert(rx_origin + start_shift <= check_at);
1543 if (rx_origin + start_shift != check_at) {
1544 /* not at latest position float substr could match:
1545 * Recheck anchored substring, but not floating.
1546 * The condition above is in bytes rather than
1547 * chars for efficiency. It's conservative, in
1548 * that it errs on the side of doing 'goto
1549 * do_other_substr'. In this case, at worst,
1550 * an extra anchored search may get done, but in
1551 * practice the extra fbm_instr() is likely to
1552 * get skipped anyway. */
1553 DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
1554 " about to retry anchored at offset %ld (rx_origin now %" IVdf ")...\n",
1555 (long)(other_last - strbeg),
1556 (IV)(rx_origin - strbeg)
1557 ));
1558 goto do_other_substr;
1559 }
1560 }
1561 }
1562 else {
1563 /* float-only */
1564
1565 if (ml_anch) {
1566 /* In the presence of ml_anch, we might be able to
1567 * find another \n without breaking the current float
1568 * constraint. */
1569
1570 /* strictly speaking this should be HOP3c(..., 1, ...),
1571 * but since we goto a block of code that's going to
1572 * search for the next \n if any, its safe here */
1573 rx_origin++;
1574 DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
1575 " about to look for /%s^%s/m starting at rx_origin %ld...\n",
1576 PL_colors[0], PL_colors[1],
1577 (long)(rx_origin - strbeg)) );
1578 goto postprocess_substr_matches;
1579 }
1580
1581 /* strictly speaking this can never be true; but might
1582 * be if we ever allow intuit without substrings */
1583 if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
1584 goto fail;
1585
1586 rx_origin = rx_max_float;
1587 }
1588
1589 /* at this point, any matching substrings have been
1590 * contradicted. Start again... */
1591
1592 rx_origin = HOP3c(rx_origin, 1, strend);
1593
1594 /* uses bytes rather than char calculations for efficiency.
1595 * It's conservative: it errs on the side of doing 'goto restart',
1596 * where there is code that does a proper char-based test */
1597 if (rx_origin + start_shift + end_shift > strend) {
1598 DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
1599 " Could not match STCLASS...\n") );
1600 goto fail;
1601 }
1602 DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
1603 " about to look for %s substr starting at offset %ld (rx_origin now %" IVdf ")...\n",
1604 (prog->substrs->check_ix ? "floating" : "anchored"),
1605 (long)(rx_origin + start_shift - strbeg),
1606 (IV)(rx_origin - strbeg)
1607 ));
1608 goto restart;
1609 }
1610
1611 /* Success !!! */
1612
1613 if (rx_origin != s) {
1614 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1615 " By STCLASS: moving %ld --> %ld\n",
1616 (long)(rx_origin - strbeg), (long)(s - strbeg))
1617 );
1618 }
1619 else {
1620 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1621 " Does not contradict STCLASS...\n");
1622 );
1623 }
1624 }
1625
1626 /* Decide whether using the substrings helped */
1627
1628 if (rx_origin != strpos) {
1629 /* Fixed substring is found far enough so that the match
1630 cannot start at strpos. */
1631
1632 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_ " try at offset...\n"));
1633 ++BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
1634 }
1635 else {
1636 /* The found rx_origin position does not prohibit matching at
1637 * strpos, so calling intuit didn't gain us anything. Decrement
1638 * the BmUSEFUL() count on the check substring, and if we reach
1639 * zero, free it. */
1640 if (!(prog->intflags & PREGf_NAUGHTY)
1641 && (utf8_target ? (
1642 prog->check_utf8 /* Could be deleted already */
1643 && --BmUSEFUL(prog->check_utf8) < 0
1644 && (prog->check_utf8 == prog->float_utf8)
1645 ) : (
1646 prog->check_substr /* Could be deleted already */
1647 && --BmUSEFUL(prog->check_substr) < 0
1648 && (prog->check_substr == prog->float_substr)
1649 )))
1650 {
1651 /* If flags & SOMETHING - do not do it many times on the same match */
1652 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_ " ... Disabling check substring...\n"));
1653 /* XXX Does the destruction order has to change with utf8_target? */
1654 SvREFCNT_dec(utf8_target ? prog->check_utf8 : prog->check_substr);
1655 SvREFCNT_dec(utf8_target ? prog->check_substr : prog->check_utf8);
1656 prog->check_substr = prog->check_utf8 = NULL; /* disable */
1657 prog->float_substr = prog->float_utf8 = NULL; /* clear */
1658 check = NULL; /* abort */
1659 /* XXXX This is a remnant of the old implementation. It
1660 looks wasteful, since now INTUIT can use many
1661 other heuristics. */
1662 prog->extflags &= ~RXf_USE_INTUIT;
1663 }
1664 }
1665
1666 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
1667 "Intuit: %sSuccessfully guessed:%s match at offset %ld\n",
1668 PL_colors[4], PL_colors[5], (long)(rx_origin - strbeg)) );
1669
1670 return rx_origin;
1671
1672 fail_finish: /* Substring not found */
1673 if (prog->check_substr || prog->check_utf8) /* could be removed already */
1674 BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
1675 fail:
1676 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_ "%sMatch rejected by optimizer%s\n",
1677 PL_colors[4], PL_colors[5]));
1678 return NULL;
1679 }
1680
1681
1682 #define DECL_TRIE_TYPE(scan) \
1683 const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold, \
1684 trie_utf8_exactfa_fold, trie_latin_utf8_exactfa_fold, \
1685 trie_utf8l, trie_flu8, trie_flu8_latin } \
1686 trie_type = ((scan->flags == EXACT) \
1687 ? (utf8_target ? trie_utf8 : trie_plain) \
1688 : (scan->flags == EXACTL) \
1689 ? (utf8_target ? trie_utf8l : trie_plain) \
1690 : (scan->flags == EXACTFAA) \
1691 ? (utf8_target \
1692 ? trie_utf8_exactfa_fold \
1693 : trie_latin_utf8_exactfa_fold) \
1694 : (scan->flags == EXACTFLU8 \
1695 ? (utf8_target \
1696 ? trie_flu8 \
1697 : trie_flu8_latin) \
1698 : (utf8_target \
1699 ? trie_utf8_fold \
1700 : trie_latin_utf8_fold)))
1701
1702 /* 'uscan' is set to foldbuf, and incremented, so below the end of uscan is
1703 * 'foldbuf+sizeof(foldbuf)' */
1704 #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uc_end, uscan, len, uvc, charid, foldlen, foldbuf, uniflags) \
1705 STMT_START { \
1706 STRLEN skiplen; \
1707 U8 flags = FOLD_FLAGS_FULL; \
1708 switch (trie_type) { \
1709 case trie_flu8: \
1710 _CHECK_AND_WARN_PROBLEMATIC_LOCALE; \
1711 if (UTF8_IS_ABOVE_LATIN1(*uc)) { \
1712 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(uc, uc_end); \
1713 } \
1714 goto do_trie_utf8_fold; \
1715 case trie_utf8_exactfa_fold: \
1716 flags |= FOLD_FLAGS_NOMIX_ASCII; \
1717 /* FALLTHROUGH */ \
1718 case trie_utf8_fold: \
1719 do_trie_utf8_fold: \
1720 if ( foldlen>0 ) { \
1721 uvc = utf8n_to_uvchr( (const U8*) uscan, foldlen, &len, uniflags ); \
1722 foldlen -= len; \
1723 uscan += len; \
1724 len=0; \
1725 } else { \
1726 uvc = _toFOLD_utf8_flags( (const U8*) uc, uc_end, foldbuf, &foldlen, \
1727 flags); \
1728 len = UTF8_SAFE_SKIP(uc, uc_end); \
1729 skiplen = UVCHR_SKIP( uvc ); \
1730 foldlen -= skiplen; \
1731 uscan = foldbuf + skiplen; \
1732 } \
1733 break; \
1734 case trie_flu8_latin: \
1735 _CHECK_AND_WARN_PROBLEMATIC_LOCALE; \
1736 goto do_trie_latin_utf8_fold; \
1737 case trie_latin_utf8_exactfa_fold: \
1738 flags |= FOLD_FLAGS_NOMIX_ASCII; \
1739 /* FALLTHROUGH */ \
1740 case trie_latin_utf8_fold: \
1741 do_trie_latin_utf8_fold: \
1742 if ( foldlen>0 ) { \
1743 uvc = utf8n_to_uvchr( (const U8*) uscan, foldlen, &len, uniflags ); \
1744 foldlen -= len; \
1745 uscan += len; \
1746 len=0; \
1747 } else { \
1748 len = 1; \
1749 uvc = _to_fold_latin1( (U8) *uc, foldbuf, &foldlen, flags); \
1750 skiplen = UVCHR_SKIP( uvc ); \
1751 foldlen -= skiplen; \
1752 uscan = foldbuf + skiplen; \
1753 } \
1754 break; \
1755 case trie_utf8l: \
1756 _CHECK_AND_WARN_PROBLEMATIC_LOCALE; \
1757 if (utf8_target && UTF8_IS_ABOVE_LATIN1(*uc)) { \
1758 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(uc, uc_end); \
1759 } \
1760 /* FALLTHROUGH */ \
1761 case trie_utf8: \
1762 uvc = utf8n_to_uvchr( (const U8*) uc, uc_end - uc, &len, uniflags ); \
1763 break; \
1764 case trie_plain: \
1765 uvc = (UV)*uc; \
1766 len = 1; \
1767 } \
1768 if (uvc < 256) { \
1769 charid = trie->charmap[ uvc ]; \
1770 } \
1771 else { \
1772 charid = 0; \
1773 if (widecharmap) { \
1774 SV** const svpp = hv_fetch(widecharmap, \
1775 (char*)&uvc, sizeof(UV), 0); \
1776 if (svpp) \
1777 charid = (U16)SvIV(*svpp); \
1778 } \
1779 } \
1780 } STMT_END
1781
1782 #define DUMP_EXEC_POS(li,s,doutf8,depth) \
1783 dump_exec_pos(li,s,(reginfo->strend),(reginfo->strbeg), \
1784 startpos, doutf8, depth)
1785
1786 #define REXEC_FBC_SCAN(UTF8, CODE) \
1787 STMT_START { \
1788 while (s < strend) { \
1789 CODE \
1790 s += ((UTF8) \
1791 ? UTF8_SAFE_SKIP(s, reginfo->strend) \
1792 : 1); \
1793 } \
1794 } STMT_END
1795
1796 #define REXEC_FBC_CLASS_SCAN(UTF8, COND) \
1797 STMT_START { \
1798 while (s < strend) { \
1799 REXEC_FBC_CLASS_SCAN_GUTS(UTF8, COND) \
1800 } \
1801 } STMT_END
1802
1803 #define REXEC_FBC_CLASS_SCAN_GUTS(UTF8, COND) \
1804 if (COND) { \
1805 FBC_CHECK_AND_TRY \
1806 s += ((UTF8) ? UTF8_SAFE_SKIP(s, reginfo->strend) : 1);\
1807 previous_occurrence_end = s; \
1808 } \
1809 else { \
1810 s += ((UTF8) ? UTF8SKIP(s) : 1); \
1811 }
1812
1813 #define REXEC_FBC_CSCAN(CONDUTF8,COND) \
1814 if (utf8_target) { \
1815 REXEC_FBC_CLASS_SCAN(1, CONDUTF8); \
1816 } \
1817 else { \
1818 REXEC_FBC_CLASS_SCAN(0, COND); \
1819 }
1820
1821 /* We keep track of where the next character should start after an occurrence
1822 * of the one we're looking for. Knowing that, we can see right away if the
1823 * next occurrence is adjacent to the previous. When 'doevery' is FALSE, we
1824 * don't accept the 2nd and succeeding adjacent occurrences */
1825 #define FBC_CHECK_AND_TRY \
1826 if ( ( doevery \
1827 || s != previous_occurrence_end) \
1828 && ( reginfo->intuit \
1829 || (s <= reginfo->strend && regtry(reginfo, &s)))) \
1830 { \
1831 goto got_it; \
1832 }
1833
1834
1835 /* This differs from the above macros in that it calls a function which returns
1836 * the next occurrence of the thing being looked for in 's'; and 'strend' if
1837 * there is no such occurrence. */
1838 #define REXEC_FBC_FIND_NEXT_SCAN(UTF8, f) \
1839 while (s < strend) { \
1840 s = (f); \
1841 if (s >= strend) { \
1842 break; \
1843 } \
1844 \
1845 FBC_CHECK_AND_TRY \
1846 s += (UTF8) ? UTF8SKIP(s) : 1; \
1847 previous_occurrence_end = s; \
1848 }
1849
1850 /* This differs from the above macros in that it is passed a single byte that
1851 * is known to begin the next occurrence of the thing being looked for in 's'.
1852 * It does a memchr to find the next occurrence of 'byte', before trying 'COND'
1853 * at that position. */
1854 #define REXEC_FBC_FIND_NEXT_UTF8_BYTE_SCAN(byte, COND) \
1855 while (s < strend) { \
1856 s = (char *) memchr(s, byte, strend -s); \
1857 if (s == NULL) { \
1858 s = (char *) strend; \
1859 break; \
1860 } \
1861 \
1862 if (COND) { \
1863 FBC_CHECK_AND_TRY \
1864 s += UTF8_SAFE_SKIP(s, reginfo->strend); \
1865 previous_occurrence_end = s; \
1866 } \
1867 else { \
1868 s += UTF8SKIP(s); \
1869 } \
1870 }
1871
1872 /* The three macros below are slightly different versions of the same logic.
1873 *
1874 * The first is for /a and /aa when the target string is UTF-8. This can only
1875 * match ascii, but it must advance based on UTF-8. The other two handle the
1876 * non-UTF-8 and the more generic UTF-8 cases. In all three, we are looking
1877 * for the boundary (or non-boundary) between a word and non-word character.
1878 * The utf8 and non-utf8 cases have the same logic, but the details must be
1879 * different. Find the "wordness" of the character just prior to this one, and
1880 * compare it with the wordness of this one. If they differ, we have a
1881 * boundary. At the beginning of the string, pretend that the previous
1882 * character was a new-line.
1883 *
1884 * All these macros uncleanly have side-effects with each other and outside
1885 * variables. So far it's been too much trouble to clean-up
1886 *
1887 * TEST_NON_UTF8 is the macro or function to call to test if its byte input is
1888 * a word character or not.
1889 * IF_SUCCESS is code to do if it finds that we are at a boundary between
1890 * word/non-word
1891 * IF_FAIL is code to do if we aren't at a boundary between word/non-word
1892 *
1893 * Exactly one of the two IF_FOO parameters is a no-op, depending on whether we
1894 * are looking for a boundary or for a non-boundary. If we are looking for a
1895 * boundary, we want IF_FAIL to be the no-op, and for IF_SUCCESS to go out and
1896 * see if this tentative match actually works, and if so, to quit the loop
1897 * here. And vice-versa if we are looking for a non-boundary.
1898 *
1899 * 'tmp' below in the next three macros in the REXEC_FBC_SCAN and
1900 * REXEC_FBC_SCAN loops is a loop invariant, a bool giving the return of
1901 * TEST_NON_UTF8(s-1). To see this, note that that's what it is defined to be
1902 * at entry to the loop, and to get to the IF_FAIL branch, tmp must equal
1903 * TEST_NON_UTF8(s), and in the opposite branch, IF_SUCCESS, tmp is that
1904 * complement. But in that branch we complement tmp, meaning that at the
1905 * bottom of the loop tmp is always going to be equal to TEST_NON_UTF8(s),
1906 * which means at the top of the loop in the next iteration, it is
1907 * TEST_NON_UTF8(s-1) */
1908 #define FBC_UTF8_A(TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
1909 tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n'; \
1910 tmp = TEST_NON_UTF8(tmp); \
1911 REXEC_FBC_SCAN(1, /* 1=>is-utf8; advances s while s < strend */ \
1912 if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
1913 tmp = !tmp; \
1914 IF_SUCCESS; /* Is a boundary if values for s-1 and s differ */ \
1915 } \
1916 else { \
1917 IF_FAIL; \
1918 } \
1919 ); \
1920
1921 /* Like FBC_UTF8_A, but TEST_UV is a macro which takes a UV as its input, and
1922 * TEST_UTF8 is a macro that for the same input code points returns identically
1923 * to TEST_UV, but takes a pointer to a UTF-8 encoded string instead */
1924 #define FBC_UTF8(TEST_UV, TEST_UTF8, IF_SUCCESS, IF_FAIL) \
1925 if (s == reginfo->strbeg) { \
1926 tmp = '\n'; \
1927 } \
1928 else { /* Back-up to the start of the previous character */ \
1929 U8 * const r = reghop3((U8*)s, -1, (U8*)reginfo->strbeg); \
1930 tmp = utf8n_to_uvchr(r, (U8*) reginfo->strend - r, \
1931 0, UTF8_ALLOW_DEFAULT); \
1932 } \
1933 tmp = TEST_UV(tmp); \
1934 REXEC_FBC_SCAN(1, /* 1=>is-utf8; advances s while s < strend */ \
1935 if (tmp == ! (TEST_UTF8((U8 *) s, (U8 *) reginfo->strend))) { \
1936 tmp = !tmp; \
1937 IF_SUCCESS; \
1938 } \
1939 else { \
1940 IF_FAIL; \
1941 } \
1942 );
1943
1944 /* Like the above two macros. UTF8_CODE is the complete code for handling
1945 * UTF-8. Common to the BOUND and NBOUND cases, set-up by the FBC_BOUND, etc
1946 * macros below */
1947 #define FBC_BOUND_COMMON(UTF8_CODE, TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
1948 if (utf8_target) { \
1949 UTF8_CODE \
1950 } \
1951 else { /* Not utf8 */ \
1952 tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n'; \
1953 tmp = TEST_NON_UTF8(tmp); \
1954 REXEC_FBC_SCAN(0, /* 0=>not-utf8; advances s while s < strend */ \
1955 if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
1956 IF_SUCCESS; \
1957 tmp = !tmp; \
1958 } \
1959 else { \
1960 IF_FAIL; \
1961 } \
1962 ); \
1963 } \
1964 /* Here, things have been set up by the previous code so that tmp is the \
1965 * return of TEST_NON_UTF(s-1) or TEST_UTF8(s-1) (depending on the \
1966 * utf8ness of the target). We also have to check if this matches against \
1967 * the EOS, which we treat as a \n (which is the same value in both UTF-8 \
1968 * or non-UTF8, so can use the non-utf8 test condition even for a UTF-8 \
1969 * string */ \
1970 if (tmp == ! TEST_NON_UTF8('\n')) { \
1971 IF_SUCCESS; \
1972 } \
1973 else { \
1974 IF_FAIL; \
1975 }
1976
1977 /* This is the macro to use when we want to see if something that looks like it
1978 * could match, actually does, and if so exits the loop. It needs to be used
1979 * only for bounds checking macros, as it allows for matching beyond the end of
1980 * string (which should be zero length without having to look at the string
1981 * contents) */
1982 #define REXEC_FBC_TRYIT \
1983 if (reginfo->intuit || (s <= reginfo->strend && regtry(reginfo, &s))) \
1984 goto got_it
1985
1986 /* The only difference between the BOUND and NBOUND cases is that
1987 * REXEC_FBC_TRYIT is called when matched in BOUND, and when non-matched in
1988 * NBOUND. This is accomplished by passing it as either the if or else clause,
1989 * with the other one being empty (PLACEHOLDER is defined as empty).
1990 *
1991 * The TEST_FOO parameters are for operating on different forms of input, but
1992 * all should be ones that return identically for the same underlying code
1993 * points */
1994 #define FBC_BOUND(TEST_NON_UTF8, TEST_UV, TEST_UTF8) \
1995 FBC_BOUND_COMMON( \
1996 FBC_UTF8(TEST_UV, TEST_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), \
1997 TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
1998
1999 #define FBC_BOUND_A(TEST_NON_UTF8) \
2000 FBC_BOUND_COMMON( \
2001 FBC_UTF8_A(TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), \
2002 TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
2003
2004 #define FBC_NBOUND(TEST_NON_UTF8, TEST_UV, TEST_UTF8) \
2005 FBC_BOUND_COMMON( \
2006 FBC_UTF8(TEST_UV, TEST_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), \
2007 TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
2008
2009 #define FBC_NBOUND_A(TEST_NON_UTF8) \
2010 FBC_BOUND_COMMON( \
2011 FBC_UTF8_A(TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), \
2012 TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
2013
2014 #ifdef DEBUGGING
2015 static IV
S_get_break_val_cp_checked(SV * const invlist,const UV cp_in)2016 S_get_break_val_cp_checked(SV* const invlist, const UV cp_in) {
2017 IV cp_out = _invlist_search(invlist, cp_in);
2018 assert(cp_out >= 0);
2019 return cp_out;
2020 }
2021 # define _generic_GET_BREAK_VAL_CP_CHECKED(invlist, invmap, cp) \
2022 invmap[S_get_break_val_cp_checked(invlist, cp)]
2023 #else
2024 # define _generic_GET_BREAK_VAL_CP_CHECKED(invlist, invmap, cp) \
2025 invmap[_invlist_search(invlist, cp)]
2026 #endif
2027
2028 /* Takes a pointer to an inversion list, a pointer to its corresponding
2029 * inversion map, and a code point, and returns the code point's value
2030 * according to the two arrays. It assumes that all code points have a value.
2031 * This is used as the base macro for macros for particular properties */
2032 #define _generic_GET_BREAK_VAL_CP(invlist, invmap, cp) \
2033 _generic_GET_BREAK_VAL_CP_CHECKED(invlist, invmap, cp)
2034
2035 /* Same as above, but takes begin, end ptrs to a UTF-8 encoded string instead
2036 * of a code point, returning the value for the first code point in the string.
2037 * And it takes the particular macro name that finds the desired value given a
2038 * code point. Merely convert the UTF-8 to code point and call the cp macro */
2039 #define _generic_GET_BREAK_VAL_UTF8(cp_macro, pos, strend) \
2040 (__ASSERT_(pos < strend) \
2041 /* Note assumes is valid UTF-8 */ \
2042 (cp_macro(utf8_to_uvchr_buf((pos), (strend), NULL))))
2043
2044 /* Returns the GCB value for the input code point */
2045 #define getGCB_VAL_CP(cp) \
2046 _generic_GET_BREAK_VAL_CP( \
2047 PL_GCB_invlist, \
2048 _Perl_GCB_invmap, \
2049 (cp))
2050
2051 /* Returns the GCB value for the first code point in the UTF-8 encoded string
2052 * bounded by pos and strend */
2053 #define getGCB_VAL_UTF8(pos, strend) \
2054 _generic_GET_BREAK_VAL_UTF8(getGCB_VAL_CP, pos, strend)
2055
2056 /* Returns the LB value for the input code point */
2057 #define getLB_VAL_CP(cp) \
2058 _generic_GET_BREAK_VAL_CP( \
2059 PL_LB_invlist, \
2060 _Perl_LB_invmap, \
2061 (cp))
2062
2063 /* Returns the LB value for the first code point in the UTF-8 encoded string
2064 * bounded by pos and strend */
2065 #define getLB_VAL_UTF8(pos, strend) \
2066 _generic_GET_BREAK_VAL_UTF8(getLB_VAL_CP, pos, strend)
2067
2068
2069 /* Returns the SB value for the input code point */
2070 #define getSB_VAL_CP(cp) \
2071 _generic_GET_BREAK_VAL_CP( \
2072 PL_SB_invlist, \
2073 _Perl_SB_invmap, \
2074 (cp))
2075
2076 /* Returns the SB value for the first code point in the UTF-8 encoded string
2077 * bounded by pos and strend */
2078 #define getSB_VAL_UTF8(pos, strend) \
2079 _generic_GET_BREAK_VAL_UTF8(getSB_VAL_CP, pos, strend)
2080
2081 /* Returns the WB value for the input code point */
2082 #define getWB_VAL_CP(cp) \
2083 _generic_GET_BREAK_VAL_CP( \
2084 PL_WB_invlist, \
2085 _Perl_WB_invmap, \
2086 (cp))
2087
2088 /* Returns the WB value for the first code point in the UTF-8 encoded string
2089 * bounded by pos and strend */
2090 #define getWB_VAL_UTF8(pos, strend) \
2091 _generic_GET_BREAK_VAL_UTF8(getWB_VAL_CP, pos, strend)
2092
2093 /* We know what class REx starts with. Try to find this position... */
2094 /* if reginfo->intuit, its a dryrun */
2095 /* annoyingly all the vars in this routine have different names from their counterparts
2096 in regmatch. /grrr */
2097 STATIC char *
S_find_byclass(pTHX_ regexp * prog,const regnode * c,char * s,const char * strend,regmatch_info * reginfo)2098 S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
2099 const char *strend, regmatch_info *reginfo)
2100 {
2101 dVAR;
2102
2103 /* TRUE if x+ need not match at just the 1st pos of run of x's */
2104 const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
2105
2106 char *pat_string; /* The pattern's exactish string */
2107 char *pat_end; /* ptr to end char of pat_string */
2108 re_fold_t folder; /* Function for computing non-utf8 folds */
2109 const U8 *fold_array; /* array for folding ords < 256 */
2110 STRLEN ln;
2111 STRLEN lnc;
2112 U8 c1;
2113 U8 c2;
2114 char *e = NULL;
2115
2116 /* In some cases we accept only the first occurence of 'x' in a sequence of
2117 * them. This variable points to just beyond the end of the previous
2118 * occurrence of 'x', hence we can tell if we are in a sequence. (Having
2119 * it point to beyond the 'x' allows us to work for UTF-8 without having to
2120 * hop back.) */
2121 char * previous_occurrence_end = 0;
2122
2123 I32 tmp; /* Scratch variable */
2124 const bool utf8_target = reginfo->is_utf8_target;
2125 UV utf8_fold_flags = 0;
2126 const bool is_utf8_pat = reginfo->is_utf8_pat;
2127 bool to_complement = FALSE; /* Invert the result? Taking the xor of this
2128 with a result inverts that result, as 0^1 =
2129 1 and 1^1 = 0 */
2130 _char_class_number classnum;
2131
2132 RXi_GET_DECL(prog,progi);
2133
2134 PERL_ARGS_ASSERT_FIND_BYCLASS;
2135
2136 /* We know what class it must start with. */
2137 switch (OP(c)) {
2138 case ANYOFPOSIXL:
2139 case ANYOFL:
2140 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
2141
2142 if (ANYOFL_UTF8_LOCALE_REQD(FLAGS(c)) && ! IN_UTF8_CTYPE_LOCALE) {
2143 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE), utf8_locale_required);
2144 }
2145
2146 /* FALLTHROUGH */
2147 case ANYOFD:
2148 case ANYOF:
2149 if (utf8_target) {
2150 REXEC_FBC_CLASS_SCAN(1, /* 1=>is-utf8 */
2151 reginclass(prog, c, (U8*)s, (U8*) strend, utf8_target));
2152 }
2153 else if (ANYOF_FLAGS(c) & ~ ANYOF_MATCHES_ALL_ABOVE_BITMAP) {
2154 /* We know that s is in the bitmap range since the target isn't
2155 * UTF-8, so what happens for out-of-range values is not relevant,
2156 * so exclude that from the flags */
2157 REXEC_FBC_CLASS_SCAN(0, reginclass(prog,c, (U8*)s, (U8*)s+1, 0));
2158 }
2159 else {
2160 REXEC_FBC_CLASS_SCAN(0, ANYOF_BITMAP_TEST(c, *((U8*)s)));
2161 }
2162 break;
2163
2164 case ANYOFM: /* ARG() is the base byte; FLAGS() the mask byte */
2165 /* UTF-8ness doesn't matter because only matches UTF-8 invariants, so
2166 * use 0 */
2167 REXEC_FBC_FIND_NEXT_SCAN(0,
2168 (char *) find_next_masked((U8 *) s, (U8 *) strend,
2169 (U8) ARG(c), FLAGS(c)));
2170 break;
2171
2172 case NANYOFM: /* UTF-8ness does matter because can match UTF-8 variants.
2173 */
2174 REXEC_FBC_FIND_NEXT_SCAN(utf8_target,
2175 (char *) find_span_end_mask((U8 *) s, (U8 *) strend,
2176 (U8) ARG(c), FLAGS(c)));
2177 break;
2178
2179 case ANYOFH:
2180 if (utf8_target) { /* Can't possibly match a non-UTF-8 target */
2181 U8 first_byte = FLAGS(c);
2182
2183 if (first_byte) { /* We know what the first byte of any matched
2184 string should be */
2185 REXEC_FBC_FIND_NEXT_UTF8_BYTE_SCAN(first_byte,
2186 reginclass(prog, c, (U8*)s, (U8*) strend, utf8_target));
2187 }
2188 else {
2189 REXEC_FBC_CLASS_SCAN(TRUE,
2190 reginclass(prog, c, (U8*)s, (U8*) strend, utf8_target));
2191 }
2192 }
2193 break;
2194
2195 case EXACTFAA_NO_TRIE: /* This node only generated for non-utf8 patterns */
2196 assert(! is_utf8_pat);
2197 /* FALLTHROUGH */
2198 case EXACTFAA:
2199 if (is_utf8_pat) {
2200 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII
2201 |FOLDEQ_S2_ALREADY_FOLDED|FOLDEQ_S2_FOLDS_SANE;
2202 goto do_exactf_utf8;
2203 }
2204 else if (utf8_target) {
2205
2206 /* Here, and elsewhere in this file, the reason we can't consider a
2207 * non-UTF-8 pattern already folded in the presence of a UTF-8
2208 * target is because any MICRO SIGN in the pattern won't be folded.
2209 * Since the fold of the MICRO SIGN requires UTF-8 to represent, we
2210 * can consider a non-UTF-8 pattern folded when matching a
2211 * non-UTF-8 target */
2212 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
2213 goto do_exactf_utf8;
2214 }
2215
2216 /* Latin1 folds are not affected by /a, except it excludes the sharp s,
2217 * which these functions don't handle anyway */
2218 fold_array = PL_fold_latin1;
2219 folder = foldEQ_latin1_s2_folded;
2220 goto do_exactf_non_utf8;
2221
2222 case EXACTF: /* This node only generated for non-utf8 patterns */
2223 assert(! is_utf8_pat);
2224 if (utf8_target) {
2225 goto do_exactf_utf8;
2226 }
2227 fold_array = PL_fold;
2228 folder = foldEQ;
2229 goto do_exactf_non_utf8;
2230
2231 case EXACTFL:
2232 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
2233 if (is_utf8_pat || utf8_target || IN_UTF8_CTYPE_LOCALE) {
2234 utf8_fold_flags = FOLDEQ_LOCALE;
2235 goto do_exactf_utf8;
2236 }
2237 fold_array = PL_fold_locale;
2238 folder = foldEQ_locale;
2239 goto do_exactf_non_utf8;
2240
2241 case EXACTFUP: /* Problematic even though pattern isn't UTF-8. Use
2242 full functionality normally not done except for
2243 UTF-8 */
2244 assert(! is_utf8_pat);
2245 goto do_exactf_utf8;
2246
2247 case EXACTFLU8:
2248 if (! utf8_target) { /* All code points in this node require
2249 UTF-8 to express. */
2250 break;
2251 }
2252 utf8_fold_flags = FOLDEQ_LOCALE | FOLDEQ_S2_ALREADY_FOLDED
2253 | FOLDEQ_S2_FOLDS_SANE;
2254 goto do_exactf_utf8;
2255
2256 case EXACTFU_ONLY8:
2257 if (! utf8_target) {
2258 break;
2259 }
2260 assert(is_utf8_pat);
2261 utf8_fold_flags = FOLDEQ_S2_ALREADY_FOLDED;
2262 goto do_exactf_utf8;
2263
2264 case EXACTFU:
2265 if (is_utf8_pat || utf8_target) {
2266 utf8_fold_flags = FOLDEQ_S2_ALREADY_FOLDED;
2267 goto do_exactf_utf8;
2268 }
2269
2270 /* Any 'ss' in the pattern should have been replaced by regcomp,
2271 * so we don't have to worry here about this single special case
2272 * in the Latin1 range */
2273 fold_array = PL_fold_latin1;
2274 folder = foldEQ_latin1_s2_folded;
2275
2276 /* FALLTHROUGH */
2277
2278 do_exactf_non_utf8: /* Neither pattern nor string are UTF8, and there
2279 are no glitches with fold-length differences
2280 between the target string and pattern */
2281
2282 /* The idea in the non-utf8 EXACTF* cases is to first find the
2283 * first character of the EXACTF* node and then, if necessary,
2284 * case-insensitively compare the full text of the node. c1 is the
2285 * first character. c2 is its fold. This logic will not work for
2286 * Unicode semantics and the german sharp ss, which hence should
2287 * not be compiled into a node that gets here. */
2288 pat_string = STRING(c);
2289 ln = STR_LEN(c); /* length to match in octets/bytes */
2290
2291 /* We know that we have to match at least 'ln' bytes (which is the
2292 * same as characters, since not utf8). If we have to match 3
2293 * characters, and there are only 2 availabe, we know without
2294 * trying that it will fail; so don't start a match past the
2295 * required minimum number from the far end */
2296 e = HOP3c(strend, -((SSize_t)ln), s);
2297 if (e < s)
2298 break;
2299
2300 c1 = *pat_string;
2301 c2 = fold_array[c1];
2302 if (c1 == c2) { /* If char and fold are the same */
2303 while (s <= e) {
2304 s = (char *) memchr(s, c1, e + 1 - s);
2305 if (s == NULL) {
2306 break;
2307 }
2308
2309 /* Check that the rest of the node matches */
2310 if ( (ln == 1 || folder(s + 1, pat_string + 1, ln - 1))
2311 && (reginfo->intuit || regtry(reginfo, &s)) )
2312 {
2313 goto got_it;
2314 }
2315 s++;
2316 }
2317 }
2318 else {
2319 U8 bits_differing = c1 ^ c2;
2320
2321 /* If the folds differ in one bit position only, we can mask to
2322 * match either of them, and can use this faster find method. Both
2323 * ASCII and EBCDIC tend to have their case folds differ in only
2324 * one position, so this is very likely */
2325 if (LIKELY(PL_bitcount[bits_differing] == 1)) {
2326 bits_differing = ~ bits_differing;
2327 while (s <= e) {
2328 s = (char *) find_next_masked((U8 *) s, (U8 *) e + 1,
2329 (c1 & bits_differing), bits_differing);
2330 if (s > e) {
2331 break;
2332 }
2333
2334 if ( (ln == 1 || folder(s + 1, pat_string + 1, ln - 1))
2335 && (reginfo->intuit || regtry(reginfo, &s)) )
2336 {
2337 goto got_it;
2338 }
2339 s++;
2340 }
2341 }
2342 else { /* Otherwise, stuck with looking byte-at-a-time. This
2343 should actually happen only in EXACTFL nodes */
2344 while (s <= e) {
2345 if ( (*(U8*)s == c1 || *(U8*)s == c2)
2346 && (ln == 1 || folder(s + 1, pat_string + 1, ln - 1))
2347 && (reginfo->intuit || regtry(reginfo, &s)) )
2348 {
2349 goto got_it;
2350 }
2351 s++;
2352 }
2353 }
2354 }
2355 break;
2356
2357 do_exactf_utf8:
2358 {
2359 unsigned expansion;
2360
2361 /* If one of the operands is in utf8, we can't use the simpler folding
2362 * above, due to the fact that many different characters can have the
2363 * same fold, or portion of a fold, or different- length fold */
2364 pat_string = STRING(c);
2365 ln = STR_LEN(c); /* length to match in octets/bytes */
2366 pat_end = pat_string + ln;
2367 lnc = is_utf8_pat /* length to match in characters */
2368 ? utf8_length((U8 *) pat_string, (U8 *) pat_end)
2369 : ln;
2370
2371 /* We have 'lnc' characters to match in the pattern, but because of
2372 * multi-character folding, each character in the target can match
2373 * up to 3 characters (Unicode guarantees it will never exceed
2374 * this) if it is utf8-encoded; and up to 2 if not (based on the
2375 * fact that the Latin 1 folds are already determined, and the
2376 * only multi-char fold in that range is the sharp-s folding to
2377 * 'ss'. Thus, a pattern character can match as little as 1/3 of a
2378 * string character. Adjust lnc accordingly, rounding up, so that
2379 * if we need to match at least 4+1/3 chars, that really is 5. */
2380 expansion = (utf8_target) ? UTF8_MAX_FOLD_CHAR_EXPAND : 2;
2381 lnc = (lnc + expansion - 1) / expansion;
2382
2383 /* As in the non-UTF8 case, if we have to match 3 characters, and
2384 * only 2 are left, it's guaranteed to fail, so don't start a
2385 * match that would require us to go beyond the end of the string
2386 */
2387 e = HOP3c(strend, -((SSize_t)lnc), s);
2388
2389 /* XXX Note that we could recalculate e to stop the loop earlier,
2390 * as the worst case expansion above will rarely be met, and as we
2391 * go along we would usually find that e moves further to the left.
2392 * This would happen only after we reached the point in the loop
2393 * where if there were no expansion we should fail. Unclear if
2394 * worth the expense */
2395
2396 while (s <= e) {
2397 char *my_strend= (char *)strend;
2398 if (foldEQ_utf8_flags(s, &my_strend, 0, utf8_target,
2399 pat_string, NULL, ln, is_utf8_pat, utf8_fold_flags)
2400 && (reginfo->intuit || regtry(reginfo, &s)) )
2401 {
2402 goto got_it;
2403 }
2404 s += (utf8_target) ? UTF8_SAFE_SKIP(s, reginfo->strend) : 1;
2405 }
2406 break;
2407 }
2408
2409 case BOUNDL:
2410 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
2411 if (FLAGS(c) != TRADITIONAL_BOUND) {
2412 if (! IN_UTF8_CTYPE_LOCALE) {
2413 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
2414 B_ON_NON_UTF8_LOCALE_IS_WRONG);
2415 }
2416 goto do_boundu;
2417 }
2418
2419 FBC_BOUND(isWORDCHAR_LC, isWORDCHAR_LC_uvchr, isWORDCHAR_LC_utf8_safe);
2420 break;
2421
2422 case NBOUNDL:
2423 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
2424 if (FLAGS(c) != TRADITIONAL_BOUND) {
2425 if (! IN_UTF8_CTYPE_LOCALE) {
2426 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
2427 B_ON_NON_UTF8_LOCALE_IS_WRONG);
2428 }
2429 goto do_nboundu;
2430 }
2431
2432 FBC_NBOUND(isWORDCHAR_LC, isWORDCHAR_LC_uvchr, isWORDCHAR_LC_utf8_safe);
2433 break;
2434
2435 case BOUND: /* regcomp.c makes sure that this only has the traditional \b
2436 meaning */
2437 assert(FLAGS(c) == TRADITIONAL_BOUND);
2438
2439 FBC_BOUND(isWORDCHAR, isWORDCHAR_uni, isWORDCHAR_utf8_safe);
2440 break;
2441
2442 case BOUNDA: /* regcomp.c makes sure that this only has the traditional \b
2443 meaning */
2444 assert(FLAGS(c) == TRADITIONAL_BOUND);
2445
2446 FBC_BOUND_A(isWORDCHAR_A);
2447 break;
2448
2449 case NBOUND: /* regcomp.c makes sure that this only has the traditional \b
2450 meaning */
2451 assert(FLAGS(c) == TRADITIONAL_BOUND);
2452
2453 FBC_NBOUND(isWORDCHAR, isWORDCHAR_uni, isWORDCHAR_utf8_safe);
2454 break;
2455
2456 case NBOUNDA: /* regcomp.c makes sure that this only has the traditional \b
2457 meaning */
2458 assert(FLAGS(c) == TRADITIONAL_BOUND);
2459
2460 FBC_NBOUND_A(isWORDCHAR_A);
2461 break;
2462
2463 case NBOUNDU:
2464 if ((bound_type) FLAGS(c) == TRADITIONAL_BOUND) {
2465 FBC_NBOUND(isWORDCHAR_L1, isWORDCHAR_uni, isWORDCHAR_utf8_safe);
2466 break;
2467 }
2468
2469 do_nboundu:
2470
2471 to_complement = 1;
2472 /* FALLTHROUGH */
2473
2474 case BOUNDU:
2475 do_boundu:
2476 switch((bound_type) FLAGS(c)) {
2477 case TRADITIONAL_BOUND:
2478 FBC_BOUND(isWORDCHAR_L1, isWORDCHAR_uni, isWORDCHAR_utf8_safe);
2479 break;
2480 case GCB_BOUND:
2481 if (s == reginfo->strbeg) {
2482 if (reginfo->intuit || regtry(reginfo, &s))
2483 {
2484 goto got_it;
2485 }
2486
2487 /* Didn't match. Try at the next position (if there is one) */
2488 s += (utf8_target) ? UTF8_SAFE_SKIP(s, reginfo->strend) : 1;
2489 if (UNLIKELY(s >= reginfo->strend)) {
2490 break;
2491 }
2492 }
2493
2494 if (utf8_target) {
2495 GCB_enum before = getGCB_VAL_UTF8(
2496 reghop3((U8*)s, -1,
2497 (U8*)(reginfo->strbeg)),
2498 (U8*) reginfo->strend);
2499 while (s < strend) {
2500 GCB_enum after = getGCB_VAL_UTF8((U8*) s,
2501 (U8*) reginfo->strend);
2502 if ( (to_complement ^ isGCB(before,
2503 after,
2504 (U8*) reginfo->strbeg,
2505 (U8*) s,
2506 utf8_target))
2507 && (reginfo->intuit || regtry(reginfo, &s)))
2508 {
2509 goto got_it;
2510 }
2511 before = after;
2512 s += UTF8_SAFE_SKIP(s, reginfo->strend);
2513 }
2514 }
2515 else { /* Not utf8. Everything is a GCB except between CR and
2516 LF */
2517 while (s < strend) {
2518 if ((to_complement ^ ( UCHARAT(s - 1) != '\r'
2519 || UCHARAT(s) != '\n'))
2520 && (reginfo->intuit || regtry(reginfo, &s)))
2521 {
2522 goto got_it;
2523 }
2524 s++;
2525 }
2526 }
2527
2528 /* And, since this is a bound, it can match after the final
2529 * character in the string */
2530 if ( reginfo->intuit
2531 || (s <= reginfo->strend && regtry(reginfo, &s)))
2532 {
2533 goto got_it;
2534 }
2535 break;
2536
2537 case LB_BOUND:
2538 if (s == reginfo->strbeg) {
2539 if (reginfo->intuit || regtry(reginfo, &s)) {
2540 goto got_it;
2541 }
2542 s += (utf8_target) ? UTF8_SAFE_SKIP(s, reginfo->strend) : 1;
2543 if (UNLIKELY(s >= reginfo->strend)) {
2544 break;
2545 }
2546 }
2547
2548 if (utf8_target) {
2549 LB_enum before = getLB_VAL_UTF8(reghop3((U8*)s,
2550 -1,
2551 (U8*)(reginfo->strbeg)),
2552 (U8*) reginfo->strend);
2553 while (s < strend) {
2554 LB_enum after = getLB_VAL_UTF8((U8*) s, (U8*) reginfo->strend);
2555 if (to_complement ^ isLB(before,
2556 after,
2557 (U8*) reginfo->strbeg,
2558 (U8*) s,
2559 (U8*) reginfo->strend,
2560 utf8_target)
2561 && (reginfo->intuit || regtry(reginfo, &s)))
2562 {
2563 goto got_it;
2564 }
2565 before = after;
2566 s += UTF8_SAFE_SKIP(s, reginfo->strend);
2567 }
2568 }
2569 else { /* Not utf8. */
2570 LB_enum before = getLB_VAL_CP((U8) *(s -1));
2571 while (s < strend) {
2572 LB_enum after = getLB_VAL_CP((U8) *s);
2573 if (to_complement ^ isLB(before,
2574 after,
2575 (U8*) reginfo->strbeg,
2576 (U8*) s,
2577 (U8*) reginfo->strend,
2578 utf8_target)
2579 && (reginfo->intuit || regtry(reginfo, &s)))
2580 {
2581 goto got_it;
2582 }
2583 before = after;
2584 s++;
2585 }
2586 }
2587
2588 if ( reginfo->intuit
2589 || (s <= reginfo->strend && regtry(reginfo, &s)))
2590 {
2591 goto got_it;
2592 }
2593
2594 break;
2595
2596 case SB_BOUND:
2597 if (s == reginfo->strbeg) {
2598 if (reginfo->intuit || regtry(reginfo, &s)) {
2599 goto got_it;
2600 }
2601 s += (utf8_target) ? UTF8_SAFE_SKIP(s, reginfo->strend) : 1;
2602 if (UNLIKELY(s >= reginfo->strend)) {
2603 break;
2604 }
2605 }
2606
2607 if (utf8_target) {
2608 SB_enum before = getSB_VAL_UTF8(reghop3((U8*)s,
2609 -1,
2610 (U8*)(reginfo->strbeg)),
2611 (U8*) reginfo->strend);
2612 while (s < strend) {
2613 SB_enum after = getSB_VAL_UTF8((U8*) s,
2614 (U8*) reginfo->strend);
2615 if ((to_complement ^ isSB(before,
2616 after,
2617 (U8*) reginfo->strbeg,
2618 (U8*) s,
2619 (U8*) reginfo->strend,
2620 utf8_target))
2621 && (reginfo->intuit || regtry(reginfo, &s)))
2622 {
2623 goto got_it;
2624 }
2625 before = after;
2626 s += UTF8_SAFE_SKIP(s, reginfo->strend);
2627 }
2628 }
2629 else { /* Not utf8. */
2630 SB_enum before = getSB_VAL_CP((U8) *(s -1));
2631 while (s < strend) {
2632 SB_enum after = getSB_VAL_CP((U8) *s);
2633 if ((to_complement ^ isSB(before,
2634 after,
2635 (U8*) reginfo->strbeg,
2636 (U8*) s,
2637 (U8*) reginfo->strend,
2638 utf8_target))
2639 && (reginfo->intuit || regtry(reginfo, &s)))
2640 {
2641 goto got_it;
2642 }
2643 before = after;
2644 s++;
2645 }
2646 }
2647
2648 /* Here are at the final position in the target string. The SB
2649 * value is always true here, so matches, depending on other
2650 * constraints */
2651 if ( reginfo->intuit
2652 || (s <= reginfo->strend && regtry(reginfo, &s)))
2653 {
2654 goto got_it;
2655 }
2656
2657 break;
2658
2659 case WB_BOUND:
2660 if (s == reginfo->strbeg) {
2661 if (reginfo->intuit || regtry(reginfo, &s)) {
2662 goto got_it;
2663 }
2664 s += (utf8_target) ? UTF8_SAFE_SKIP(s, reginfo->strend) : 1;
2665 if (UNLIKELY(s >= reginfo->strend)) {
2666 break;
2667 }
2668 }
2669
2670 if (utf8_target) {
2671 /* We are at a boundary between char_sub_0 and char_sub_1.
2672 * We also keep track of the value for char_sub_-1 as we
2673 * loop through the line. Context may be needed to make a
2674 * determination, and if so, this can save having to
2675 * recalculate it */
2676 WB_enum previous = WB_UNKNOWN;
2677 WB_enum before = getWB_VAL_UTF8(
2678 reghop3((U8*)s,
2679 -1,
2680 (U8*)(reginfo->strbeg)),
2681 (U8*) reginfo->strend);
2682 while (s < strend) {
2683 WB_enum after = getWB_VAL_UTF8((U8*) s,
2684 (U8*) reginfo->strend);
2685 if ((to_complement ^ isWB(previous,
2686 before,
2687 after,
2688 (U8*) reginfo->strbeg,
2689 (U8*) s,
2690 (U8*) reginfo->strend,
2691 utf8_target))
2692 && (reginfo->intuit || regtry(reginfo, &s)))
2693 {
2694 goto got_it;
2695 }
2696 previous = before;
2697 before = after;
2698 s += UTF8_SAFE_SKIP(s, reginfo->strend);
2699 }
2700 }
2701 else { /* Not utf8. */
2702 WB_enum previous = WB_UNKNOWN;
2703 WB_enum before = getWB_VAL_CP((U8) *(s -1));
2704 while (s < strend) {
2705 WB_enum after = getWB_VAL_CP((U8) *s);
2706 if ((to_complement ^ isWB(previous,
2707 before,
2708 after,
2709 (U8*) reginfo->strbeg,
2710 (U8*) s,
2711 (U8*) reginfo->strend,
2712 utf8_target))
2713 && (reginfo->intuit || regtry(reginfo, &s)))
2714 {
2715 goto got_it;
2716 }
2717 previous = before;
2718 before = after;
2719 s++;
2720 }
2721 }
2722
2723 if ( reginfo->intuit
2724 || (s <= reginfo->strend && regtry(reginfo, &s)))
2725 {
2726 goto got_it;
2727 }
2728 }
2729 break;
2730
2731 case LNBREAK:
2732 REXEC_FBC_CSCAN(is_LNBREAK_utf8_safe(s, strend),
2733 is_LNBREAK_latin1_safe(s, strend)
2734 );
2735 break;
2736
2737 /* The argument to all the POSIX node types is the class number to pass to
2738 * _generic_isCC() to build a mask for searching in PL_charclass[] */
2739
2740 case NPOSIXL:
2741 to_complement = 1;
2742 /* FALLTHROUGH */
2743
2744 case POSIXL:
2745 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
2746 REXEC_FBC_CSCAN(to_complement ^ cBOOL(isFOO_utf8_lc(FLAGS(c), (U8 *) s, (U8 *) strend)),
2747 to_complement ^ cBOOL(isFOO_lc(FLAGS(c), *s)));
2748 break;
2749
2750 case NPOSIXD:
2751 to_complement = 1;
2752 /* FALLTHROUGH */
2753
2754 case POSIXD:
2755 if (utf8_target) {
2756 goto posix_utf8;
2757 }
2758 goto posixa;
2759
2760 case NPOSIXA:
2761 if (utf8_target) {
2762 /* The complement of something that matches only ASCII matches all
2763 * non-ASCII, plus everything in ASCII that isn't in the class. */
2764 REXEC_FBC_CLASS_SCAN(1, ! isASCII_utf8_safe(s, strend)
2765 || ! _generic_isCC_A(*s, FLAGS(c)));
2766 break;
2767 }
2768
2769 to_complement = 1;
2770 goto posixa;
2771
2772 case POSIXA:
2773 /* Don't need to worry about utf8, as it can match only a single
2774 * byte invariant character. But we do anyway for performance reasons,
2775 * as otherwise we would have to examine all the continuation
2776 * characters */
2777 if (utf8_target) {
2778 REXEC_FBC_CLASS_SCAN(1, _generic_isCC_A(*s, FLAGS(c)));
2779 break;
2780 }
2781
2782 posixa:
2783 REXEC_FBC_CLASS_SCAN(0, /* 0=>not-utf8 */
2784 to_complement ^ cBOOL(_generic_isCC_A(*s, FLAGS(c))));
2785 break;
2786
2787 case NPOSIXU:
2788 to_complement = 1;
2789 /* FALLTHROUGH */
2790
2791 case POSIXU:
2792 if (! utf8_target) {
2793 REXEC_FBC_CLASS_SCAN(0, /* 0=>not-utf8 */
2794 to_complement ^ cBOOL(_generic_isCC(*s,
2795 FLAGS(c))));
2796 }
2797 else {
2798
2799 posix_utf8:
2800 classnum = (_char_class_number) FLAGS(c);
2801 switch (classnum) {
2802 default:
2803 REXEC_FBC_CLASS_SCAN(1, /* 1=>is-utf8 */
2804 to_complement ^ cBOOL(_invlist_contains_cp(
2805 PL_XPosix_ptrs[classnum],
2806 utf8_to_uvchr_buf((U8 *) s,
2807 (U8 *) strend,
2808 NULL))));
2809 break;
2810 case _CC_ENUM_SPACE:
2811 REXEC_FBC_CLASS_SCAN(1, /* 1=>is-utf8 */
2812 to_complement ^ cBOOL(isSPACE_utf8_safe(s, strend)));
2813 break;
2814
2815 case _CC_ENUM_BLANK:
2816 REXEC_FBC_CLASS_SCAN(1,
2817 to_complement ^ cBOOL(isBLANK_utf8_safe(s, strend)));
2818 break;
2819
2820 case _CC_ENUM_XDIGIT:
2821 REXEC_FBC_CLASS_SCAN(1,
2822 to_complement ^ cBOOL(isXDIGIT_utf8_safe(s, strend)));
2823 break;
2824
2825 case _CC_ENUM_VERTSPACE:
2826 REXEC_FBC_CLASS_SCAN(1,
2827 to_complement ^ cBOOL(isVERTWS_utf8_safe(s, strend)));
2828 break;
2829
2830 case _CC_ENUM_CNTRL:
2831 REXEC_FBC_CLASS_SCAN(1,
2832 to_complement ^ cBOOL(isCNTRL_utf8_safe(s, strend)));
2833 break;
2834 }
2835 }
2836 break;
2837
2838 case AHOCORASICKC:
2839 case AHOCORASICK:
2840 {
2841 DECL_TRIE_TYPE(c);
2842 /* what trie are we using right now */
2843 reg_ac_data *aho = (reg_ac_data*)progi->data->data[ ARG( c ) ];
2844 reg_trie_data *trie = (reg_trie_data*)progi->data->data[ aho->trie ];
2845 HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]);
2846
2847 const char *last_start = strend - trie->minlen;
2848 #ifdef DEBUGGING
2849 const char *real_start = s;
2850 #endif
2851 STRLEN maxlen = trie->maxlen;
2852 SV *sv_points;
2853 U8 **points; /* map of where we were in the input string
2854 when reading a given char. For ASCII this
2855 is unnecessary overhead as the relationship
2856 is always 1:1, but for Unicode, especially
2857 case folded Unicode this is not true. */
2858 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2859 U8 *bitmap=NULL;
2860
2861
2862 GET_RE_DEBUG_FLAGS_DECL;
2863
2864 /* We can't just allocate points here. We need to wrap it in
2865 * an SV so it gets freed properly if there is a croak while
2866 * running the match */
2867 ENTER;
2868 SAVETMPS;
2869 sv_points=newSV(maxlen * sizeof(U8 *));
2870 SvCUR_set(sv_points,
2871 maxlen * sizeof(U8 *));
2872 SvPOK_on(sv_points);
2873 sv_2mortal(sv_points);
2874 points=(U8**)SvPV_nolen(sv_points );
2875 if ( trie_type != trie_utf8_fold
2876 && (trie->bitmap || OP(c)==AHOCORASICKC) )
2877 {
2878 if (trie->bitmap)
2879 bitmap=(U8*)trie->bitmap;
2880 else
2881 bitmap=(U8*)ANYOF_BITMAP(c);
2882 }
2883 /* this is the Aho-Corasick algorithm modified a touch
2884 to include special handling for long "unknown char" sequences.
2885 The basic idea being that we use AC as long as we are dealing
2886 with a possible matching char, when we encounter an unknown char
2887 (and we have not encountered an accepting state) we scan forward
2888 until we find a legal starting char.
2889 AC matching is basically that of trie matching, except that when
2890 we encounter a failing transition, we fall back to the current
2891 states "fail state", and try the current char again, a process
2892 we repeat until we reach the root state, state 1, or a legal
2893 transition. If we fail on the root state then we can either
2894 terminate if we have reached an accepting state previously, or
2895 restart the entire process from the beginning if we have not.
2896
2897 */
2898 while (s <= last_start) {
2899 const U32 uniflags = UTF8_ALLOW_DEFAULT;
2900 U8 *uc = (U8*)s;
2901 U16 charid = 0;
2902 U32 base = 1;
2903 U32 state = 1;
2904 UV uvc = 0;
2905 STRLEN len = 0;
2906 STRLEN foldlen = 0;
2907 U8 *uscan = (U8*)NULL;
2908 U8 *leftmost = NULL;
2909 #ifdef DEBUGGING
2910 U32 accepted_word= 0;
2911 #endif
2912 U32 pointpos = 0;
2913
2914 while ( state && uc <= (U8*)strend ) {
2915 int failed=0;
2916 U32 word = aho->states[ state ].wordnum;
2917
2918 if( state==1 ) {
2919 if ( bitmap ) {
2920 DEBUG_TRIE_EXECUTE_r(
2921 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
2922 dump_exec_pos( (char *)uc, c, strend, real_start,
2923 (char *)uc, utf8_target, 0 );
2924 Perl_re_printf( aTHX_
2925 " Scanning for legal start char...\n");
2926 }
2927 );
2928 if (utf8_target) {
2929 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
2930 uc += UTF8SKIP(uc);
2931 }
2932 } else {
2933 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
2934 uc++;
2935 }
2936 }
2937 s= (char *)uc;
2938 }
2939 if (uc >(U8*)last_start) break;
2940 }
2941
2942 if ( word ) {
2943 U8 *lpos= points[ (pointpos - trie->wordinfo[word].len) % maxlen ];
2944 if (!leftmost || lpos < leftmost) {
2945 DEBUG_r(accepted_word=word);
2946 leftmost= lpos;
2947 }
2948 if (base==0) break;
2949
2950 }
2951 points[pointpos++ % maxlen]= uc;
2952 if (foldlen || uc < (U8*)strend) {
2953 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
2954 (U8 *) strend, uscan, len, uvc,
2955 charid, foldlen, foldbuf,
2956 uniflags);
2957 DEBUG_TRIE_EXECUTE_r({
2958 dump_exec_pos( (char *)uc, c, strend,
2959 real_start, s, utf8_target, 0);
2960 Perl_re_printf( aTHX_
2961 " Charid:%3u CP:%4" UVxf " ",
2962 charid, uvc);
2963 });
2964 }
2965 else {
2966 len = 0;
2967 charid = 0;
2968 }
2969
2970
2971 do {
2972 #ifdef DEBUGGING
2973 word = aho->states[ state ].wordnum;
2974 #endif
2975 base = aho->states[ state ].trans.base;
2976
2977 DEBUG_TRIE_EXECUTE_r({
2978 if (failed)
2979 dump_exec_pos( (char *)uc, c, strend, real_start,
2980 s, utf8_target, 0 );
2981 Perl_re_printf( aTHX_
2982 "%sState: %4" UVxf ", word=%" UVxf,
2983 failed ? " Fail transition to " : "",
2984 (UV)state, (UV)word);
2985 });
2986 if ( base ) {
2987 U32 tmp;
2988 I32 offset;
2989 if (charid &&
2990 ( ((offset = base + charid
2991 - 1 - trie->uniquecharcount)) >= 0)
2992 && ((U32)offset < trie->lasttrans)
2993 && trie->trans[offset].check == state
2994 && (tmp=trie->trans[offset].next))
2995 {
2996 DEBUG_TRIE_EXECUTE_r(
2997 Perl_re_printf( aTHX_ " - legal\n"));
2998 state = tmp;
2999 break;
3000 }
3001 else {
3002 DEBUG_TRIE_EXECUTE_r(
3003 Perl_re_printf( aTHX_ " - fail\n"));
3004 failed = 1;
3005 state = aho->fail[state];
3006 }
3007 }
3008 else {
3009 /* we must be accepting here */
3010 DEBUG_TRIE_EXECUTE_r(
3011 Perl_re_printf( aTHX_ " - accepting\n"));
3012 failed = 1;
3013 break;
3014 }
3015 } while(state);
3016 uc += len;
3017 if (failed) {
3018 if (leftmost)
3019 break;
3020 if (!state) state = 1;
3021 }
3022 }
3023 if ( aho->states[ state ].wordnum ) {
3024 U8 *lpos = points[ (pointpos - trie->wordinfo[aho->states[ state ].wordnum].len) % maxlen ];
3025 if (!leftmost || lpos < leftmost) {
3026 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
3027 leftmost = lpos;
3028 }
3029 }
3030 if (leftmost) {
3031 s = (char*)leftmost;
3032 DEBUG_TRIE_EXECUTE_r({
3033 Perl_re_printf( aTHX_ "Matches word #%" UVxf " at position %" IVdf ". Trying full pattern...\n",
3034 (UV)accepted_word, (IV)(s - real_start)
3035 );
3036 });
3037 if (reginfo->intuit || regtry(reginfo, &s)) {
3038 FREETMPS;
3039 LEAVE;
3040 goto got_it;
3041 }
3042 if (s < reginfo->strend) {
3043 s = HOPc(s,1);
3044 }
3045 DEBUG_TRIE_EXECUTE_r({
3046 Perl_re_printf( aTHX_ "Pattern failed. Looking for new start point...\n");
3047 });
3048 } else {
3049 DEBUG_TRIE_EXECUTE_r(
3050 Perl_re_printf( aTHX_ "No match.\n"));
3051 break;
3052 }
3053 }
3054 FREETMPS;
3055 LEAVE;
3056 }
3057 break;
3058 default:
3059 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
3060 }
3061 return 0;
3062 got_it:
3063 return s;
3064 }
3065
3066 /* set RX_SAVED_COPY, RX_SUBBEG etc.
3067 * flags have same meanings as with regexec_flags() */
3068
3069 static void
S_reg_set_capture_string(pTHX_ REGEXP * const rx,char * strbeg,char * strend,SV * sv,U32 flags,bool utf8_target)3070 S_reg_set_capture_string(pTHX_ REGEXP * const rx,
3071 char *strbeg,
3072 char *strend,
3073 SV *sv,
3074 U32 flags,
3075 bool utf8_target)
3076 {
3077 struct regexp *const prog = ReANY(rx);
3078
3079 if (flags & REXEC_COPY_STR) {
3080 #ifdef PERL_ANY_COW
3081 if (SvCANCOW(sv)) {
3082 DEBUG_C(Perl_re_printf( aTHX_
3083 "Copy on write: regexp capture, type %d\n",
3084 (int) SvTYPE(sv)));
3085 /* Create a new COW SV to share the match string and store
3086 * in saved_copy, unless the current COW SV in saved_copy
3087 * is valid and suitable for our purpose */
3088 if (( prog->saved_copy
3089 && SvIsCOW(prog->saved_copy)
3090 && SvPOKp(prog->saved_copy)
3091 && SvIsCOW(sv)
3092 && SvPOKp(sv)
3093 && SvPVX(sv) == SvPVX(prog->saved_copy)))
3094 {
3095 /* just reuse saved_copy SV */
3096 if (RXp_MATCH_COPIED(prog)) {
3097 Safefree(prog->subbeg);
3098 RXp_MATCH_COPIED_off(prog);
3099 }
3100 }
3101 else {
3102 /* create new COW SV to share string */
3103 RXp_MATCH_COPY_FREE(prog);
3104 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
3105 }
3106 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
3107 assert (SvPOKp(prog->saved_copy));
3108 prog->sublen = strend - strbeg;
3109 prog->suboffset = 0;
3110 prog->subcoffset = 0;
3111 } else
3112 #endif
3113 {
3114 SSize_t min = 0;
3115 SSize_t max = strend - strbeg;
3116 SSize_t sublen;
3117
3118 if ( (flags & REXEC_COPY_SKIP_POST)
3119 && !(prog->extflags & RXf_PMf_KEEPCOPY) /* //p */
3120 && !(PL_sawampersand & SAWAMPERSAND_RIGHT)
3121 ) { /* don't copy $' part of string */
3122 U32 n = 0;
3123 max = -1;
3124 /* calculate the right-most part of the string covered
3125 * by a capture. Due to lookahead, this may be to
3126 * the right of $&, so we have to scan all captures */
3127 while (n <= prog->lastparen) {
3128 if (prog->offs[n].end > max)
3129 max = prog->offs[n].end;
3130 n++;
3131 }
3132 if (max == -1)
3133 max = (PL_sawampersand & SAWAMPERSAND_LEFT)
3134 ? prog->offs[0].start
3135 : 0;
3136 assert(max >= 0 && max <= strend - strbeg);
3137 }
3138
3139 if ( (flags & REXEC_COPY_SKIP_PRE)
3140 && !(prog->extflags & RXf_PMf_KEEPCOPY) /* //p */
3141 && !(PL_sawampersand & SAWAMPERSAND_LEFT)
3142 ) { /* don't copy $` part of string */
3143 U32 n = 0;
3144 min = max;
3145 /* calculate the left-most part of the string covered
3146 * by a capture. Due to lookbehind, this may be to
3147 * the left of $&, so we have to scan all captures */
3148 while (min && n <= prog->lastparen) {
3149 if ( prog->offs[n].start != -1
3150 && prog->offs[n].start < min)
3151 {
3152 min = prog->offs[n].start;
3153 }
3154 n++;
3155 }
3156 if ((PL_sawampersand & SAWAMPERSAND_RIGHT)
3157 && min > prog->offs[0].end
3158 )
3159 min = prog->offs[0].end;
3160
3161 }
3162
3163 assert(min >= 0 && min <= max && min <= strend - strbeg);
3164 sublen = max - min;
3165
3166 if (RXp_MATCH_COPIED(prog)) {
3167 if (sublen > prog->sublen)
3168 prog->subbeg =
3169 (char*)saferealloc(prog->subbeg, sublen+1);
3170 }
3171 else
3172 prog->subbeg = (char*)safemalloc(sublen+1);
3173 Copy(strbeg + min, prog->subbeg, sublen, char);
3174 prog->subbeg[sublen] = '\0';
3175 prog->suboffset = min;
3176 prog->sublen = sublen;
3177 RXp_MATCH_COPIED_on(prog);
3178 }
3179 prog->subcoffset = prog->suboffset;
3180 if (prog->suboffset && utf8_target) {
3181 /* Convert byte offset to chars.
3182 * XXX ideally should only compute this if @-/@+
3183 * has been seen, a la PL_sawampersand ??? */
3184
3185 /* If there's a direct correspondence between the
3186 * string which we're matching and the original SV,
3187 * then we can use the utf8 len cache associated with
3188 * the SV. In particular, it means that under //g,
3189 * sv_pos_b2u() will use the previously cached
3190 * position to speed up working out the new length of
3191 * subcoffset, rather than counting from the start of
3192 * the string each time. This stops
3193 * $x = "\x{100}" x 1E6; 1 while $x =~ /(.)/g;
3194 * from going quadratic */
3195 if (SvPOKp(sv) && SvPVX(sv) == strbeg)
3196 prog->subcoffset = sv_pos_b2u_flags(sv, prog->subcoffset,
3197 SV_GMAGIC|SV_CONST_RETURN);
3198 else
3199 prog->subcoffset = utf8_length((U8*)strbeg,
3200 (U8*)(strbeg+prog->suboffset));
3201 }
3202 }
3203 else {
3204 RXp_MATCH_COPY_FREE(prog);
3205 prog->subbeg = strbeg;
3206 prog->suboffset = 0;
3207 prog->subcoffset = 0;
3208 prog->sublen = strend - strbeg;
3209 }
3210 }
3211
3212
3213
3214
3215 /*
3216 - regexec_flags - match a regexp against a string
3217 */
3218 I32
Perl_regexec_flags(pTHX_ REGEXP * const rx,char * stringarg,char * strend,char * strbeg,SSize_t minend,SV * sv,void * data,U32 flags)3219 Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, char *strend,
3220 char *strbeg, SSize_t minend, SV *sv, void *data, U32 flags)
3221 /* stringarg: the point in the string at which to begin matching */
3222 /* strend: pointer to null at end of string */
3223 /* strbeg: real beginning of string */
3224 /* minend: end of match must be >= minend bytes after stringarg. */
3225 /* sv: SV being matched: only used for utf8 flag, pos() etc; string
3226 * itself is accessed via the pointers above */
3227 /* data: May be used for some additional optimizations.
3228 Currently unused. */
3229 /* flags: For optimizations. See REXEC_* in regexp.h */
3230
3231 {
3232 struct regexp *const prog = ReANY(rx);
3233 char *s;
3234 regnode *c;
3235 char *startpos;
3236 SSize_t minlen; /* must match at least this many chars */
3237 SSize_t dontbother = 0; /* how many characters not to try at end */
3238 const bool utf8_target = cBOOL(DO_UTF8(sv));
3239 I32 multiline;
3240 RXi_GET_DECL(prog,progi);
3241 regmatch_info reginfo_buf; /* create some info to pass to regtry etc */
3242 regmatch_info *const reginfo = ®info_buf;
3243 regexp_paren_pair *swap = NULL;
3244 I32 oldsave;
3245 GET_RE_DEBUG_FLAGS_DECL;
3246
3247 PERL_ARGS_ASSERT_REGEXEC_FLAGS;
3248 PERL_UNUSED_ARG(data);
3249
3250 /* Be paranoid... */
3251 if (prog == NULL) {
3252 Perl_croak(aTHX_ "NULL regexp parameter");
3253 }
3254
3255 DEBUG_EXECUTE_r(
3256 debug_start_match(rx, utf8_target, stringarg, strend,
3257 "Matching");
3258 );
3259
3260 startpos = stringarg;
3261
3262 /* set these early as they may be used by the HOP macros below */
3263 reginfo->strbeg = strbeg;
3264 reginfo->strend = strend;
3265 reginfo->is_utf8_target = cBOOL(utf8_target);
3266
3267 if (prog->intflags & PREGf_GPOS_SEEN) {
3268 MAGIC *mg;
3269
3270 /* set reginfo->ganch, the position where \G can match */
3271
3272 reginfo->ganch =
3273 (flags & REXEC_IGNOREPOS)
3274 ? stringarg /* use start pos rather than pos() */
3275 : ((mg = mg_find_mglob(sv)) && mg->mg_len >= 0)
3276 /* Defined pos(): */
3277 ? strbeg + MgBYTEPOS(mg, sv, strbeg, strend-strbeg)
3278 : strbeg; /* pos() not defined; use start of string */
3279
3280 DEBUG_GPOS_r(Perl_re_printf( aTHX_
3281 "GPOS ganch set to strbeg[%" IVdf "]\n", (IV)(reginfo->ganch - strbeg)));
3282
3283 /* in the presence of \G, we may need to start looking earlier in
3284 * the string than the suggested start point of stringarg:
3285 * if prog->gofs is set, then that's a known, fixed minimum
3286 * offset, such as
3287 * /..\G/: gofs = 2
3288 * /ab|c\G/: gofs = 1
3289 * or if the minimum offset isn't known, then we have to go back
3290 * to the start of the string, e.g. /w+\G/
3291 */
3292
3293 if (prog->intflags & PREGf_ANCH_GPOS) {
3294 if (prog->gofs) {
3295 startpos = HOPBACKc(reginfo->ganch, prog->gofs);
3296 if (!startpos ||
3297 ((flags & REXEC_FAIL_ON_UNDERFLOW) && startpos < stringarg))
3298 {
3299 DEBUG_r(Perl_re_printf( aTHX_
3300 "fail: ganch-gofs before earliest possible start\n"));
3301 return 0;
3302 }
3303 }
3304 else
3305 startpos = reginfo->ganch;
3306 }
3307 else if (prog->gofs) {
3308 startpos = HOPBACKc(startpos, prog->gofs);
3309 if (!startpos)
3310 startpos = strbeg;
3311 }
3312 else if (prog->intflags & PREGf_GPOS_FLOAT)
3313 startpos = strbeg;
3314 }
3315
3316 minlen = prog->minlen;
3317 if ((startpos + minlen) > strend || startpos < strbeg) {
3318 DEBUG_r(Perl_re_printf( aTHX_
3319 "Regex match can't succeed, so not even tried\n"));
3320 return 0;
3321 }
3322
3323 /* at the end of this function, we'll do a LEAVE_SCOPE(oldsave),
3324 * which will call destuctors to reset PL_regmatch_state, free higher
3325 * PL_regmatch_slabs, and clean up regmatch_info_aux and
3326 * regmatch_info_aux_eval */
3327
3328 oldsave = PL_savestack_ix;
3329
3330 s = startpos;
3331
3332 if ((prog->extflags & RXf_USE_INTUIT)
3333 && !(flags & REXEC_CHECKED))
3334 {
3335 s = re_intuit_start(rx, sv, strbeg, startpos, strend,
3336 flags, NULL);
3337 if (!s)
3338 return 0;
3339
3340 if (prog->extflags & RXf_CHECK_ALL) {
3341 /* we can match based purely on the result of INTUIT.
3342 * Set up captures etc just for $& and $-[0]
3343 * (an intuit-only match wont have $1,$2,..) */
3344 assert(!prog->nparens);
3345
3346 /* s/// doesn't like it if $& is earlier than where we asked it to
3347 * start searching (which can happen on something like /.\G/) */
3348 if ( (flags & REXEC_FAIL_ON_UNDERFLOW)
3349 && (s < stringarg))
3350 {
3351 /* this should only be possible under \G */
3352 assert(prog->intflags & PREGf_GPOS_SEEN);
3353 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
3354 "matched, but failing for REXEC_FAIL_ON_UNDERFLOW\n"));
3355 goto phooey;
3356 }
3357
3358 /* match via INTUIT shouldn't have any captures.
3359 * Let @-, @+, $^N know */
3360 prog->lastparen = prog->lastcloseparen = 0;
3361 RXp_MATCH_UTF8_set(prog, utf8_target);
3362 prog->offs[0].start = s - strbeg;
3363 prog->offs[0].end = utf8_target
3364 ? (char*)utf8_hop_forward((U8*)s, prog->minlenret, (U8 *) strend) - strbeg
3365 : s - strbeg + prog->minlenret;
3366 if ( !(flags & REXEC_NOT_FIRST) )
3367 S_reg_set_capture_string(aTHX_ rx,
3368 strbeg, strend,
3369 sv, flags, utf8_target);
3370
3371 return 1;
3372 }
3373 }
3374
3375 multiline = prog->extflags & RXf_PMf_MULTILINE;
3376
3377 if (strend - s < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
3378 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
3379 "String too short [regexec_flags]...\n"));
3380 goto phooey;
3381 }
3382
3383 /* Check validity of program. */
3384 if (UCHARAT(progi->program) != REG_MAGIC) {
3385 Perl_croak(aTHX_ "corrupted regexp program");
3386 }
3387
3388 RXp_MATCH_TAINTED_off(prog);
3389 RXp_MATCH_UTF8_set(prog, utf8_target);
3390
3391 reginfo->prog = rx; /* Yes, sorry that this is confusing. */
3392 reginfo->intuit = 0;
3393 reginfo->is_utf8_pat = cBOOL(RX_UTF8(rx));
3394 reginfo->warned = FALSE;
3395 reginfo->sv = sv;
3396 reginfo->poscache_maxiter = 0; /* not yet started a countdown */
3397 /* see how far we have to get to not match where we matched before */
3398 reginfo->till = stringarg + minend;
3399
3400 if (prog->extflags & RXf_EVAL_SEEN && SvPADTMP(sv)) {
3401 /* SAVEFREESV, not sv_mortalcopy, as this SV must last until after
3402 S_cleanup_regmatch_info_aux has executed (registered by
3403 SAVEDESTRUCTOR_X below). S_cleanup_regmatch_info_aux modifies
3404 magic belonging to this SV.
3405 Not newSVsv, either, as it does not COW.
3406 */
3407 reginfo->sv = newSV(0);
3408 SvSetSV_nosteal(reginfo->sv, sv);
3409 SAVEFREESV(reginfo->sv);
3410 }
3411
3412 /* reserve next 2 or 3 slots in PL_regmatch_state:
3413 * slot N+0: may currently be in use: skip it
3414 * slot N+1: use for regmatch_info_aux struct
3415 * slot N+2: use for regmatch_info_aux_eval struct if we have (?{})'s
3416 * slot N+3: ready for use by regmatch()
3417 */
3418
3419 {
3420 regmatch_state *old_regmatch_state;
3421 regmatch_slab *old_regmatch_slab;
3422 int i, max = (prog->extflags & RXf_EVAL_SEEN) ? 2 : 1;
3423
3424 /* on first ever match, allocate first slab */
3425 if (!PL_regmatch_slab) {
3426 Newx(PL_regmatch_slab, 1, regmatch_slab);
3427 PL_regmatch_slab->prev = NULL;
3428 PL_regmatch_slab->next = NULL;
3429 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
3430 }
3431
3432 old_regmatch_state = PL_regmatch_state;
3433 old_regmatch_slab = PL_regmatch_slab;
3434
3435 for (i=0; i <= max; i++) {
3436 if (i == 1)
3437 reginfo->info_aux = &(PL_regmatch_state->u.info_aux);
3438 else if (i ==2)
3439 reginfo->info_aux_eval =
3440 reginfo->info_aux->info_aux_eval =
3441 &(PL_regmatch_state->u.info_aux_eval);
3442
3443 if (++PL_regmatch_state > SLAB_LAST(PL_regmatch_slab))
3444 PL_regmatch_state = S_push_slab(aTHX);
3445 }
3446
3447 /* note initial PL_regmatch_state position; at end of match we'll
3448 * pop back to there and free any higher slabs */
3449
3450 reginfo->info_aux->old_regmatch_state = old_regmatch_state;
3451 reginfo->info_aux->old_regmatch_slab = old_regmatch_slab;
3452 reginfo->info_aux->poscache = NULL;
3453
3454 SAVEDESTRUCTOR_X(S_cleanup_regmatch_info_aux, reginfo->info_aux);
3455
3456 if ((prog->extflags & RXf_EVAL_SEEN))
3457 S_setup_eval_state(aTHX_ reginfo);
3458 else
3459 reginfo->info_aux_eval = reginfo->info_aux->info_aux_eval = NULL;
3460 }
3461
3462 /* If there is a "must appear" string, look for it. */
3463
3464 if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
3465 /* We have to be careful. If the previous successful match
3466 was from this regex we don't want a subsequent partially
3467 successful match to clobber the old results.
3468 So when we detect this possibility we add a swap buffer
3469 to the re, and switch the buffer each match. If we fail,
3470 we switch it back; otherwise we leave it swapped.
3471 */
3472 swap = prog->offs;
3473 /* avoid leak if we die, or clean up anyway if match completes */
3474 SAVEFREEPV(swap);
3475 Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair);
3476 DEBUG_BUFFERS_r(Perl_re_exec_indentf( aTHX_
3477 "rex=0x%" UVxf " saving offs: orig=0x%" UVxf " new=0x%" UVxf "\n",
3478 0,
3479 PTR2UV(prog),
3480 PTR2UV(swap),
3481 PTR2UV(prog->offs)
3482 ));
3483 }
3484
3485 if (prog->recurse_locinput)
3486 Zero(prog->recurse_locinput,prog->nparens + 1, char *);
3487
3488 /* Simplest case: anchored match need be tried only once, or with
3489 * MBOL, only at the beginning of each line.
3490 *
3491 * Note that /.*.../ sets PREGf_IMPLICIT|MBOL, while /.*.../s sets
3492 * PREGf_IMPLICIT|SBOL. The idea is that with /.*.../s, if it doesn't
3493 * match at the start of the string then it won't match anywhere else
3494 * either; while with /.*.../, if it doesn't match at the beginning,
3495 * the earliest it could match is at the start of the next line */
3496
3497 if (prog->intflags & (PREGf_ANCH & ~PREGf_ANCH_GPOS)) {
3498 char *end;
3499
3500 if (regtry(reginfo, &s))
3501 goto got_it;
3502
3503 if (!(prog->intflags & PREGf_ANCH_MBOL))
3504 goto phooey;
3505
3506 /* didn't match at start, try at other newline positions */
3507
3508 if (minlen)
3509 dontbother = minlen - 1;
3510 end = HOP3c(strend, -dontbother, strbeg) - 1;
3511
3512 /* skip to next newline */
3513
3514 while (s <= end) { /* note it could be possible to match at the end of the string */
3515 /* NB: newlines are the same in unicode as they are in latin */
3516 if (*s++ != '\n')
3517 continue;
3518 if (prog->check_substr || prog->check_utf8) {
3519 /* note that with PREGf_IMPLICIT, intuit can only fail
3520 * or return the start position, so it's of limited utility.
3521 * Nevertheless, I made the decision that the potential for
3522 * quick fail was still worth it - DAPM */
3523 s = re_intuit_start(rx, sv, strbeg, s, strend, flags, NULL);
3524 if (!s)
3525 goto phooey;
3526 }
3527 if (regtry(reginfo, &s))
3528 goto got_it;
3529 }
3530 goto phooey;
3531 } /* end anchored search */
3532
3533 if (prog->intflags & PREGf_ANCH_GPOS)
3534 {
3535 /* PREGf_ANCH_GPOS should never be true if PREGf_GPOS_SEEN is not true */
3536 assert(prog->intflags & PREGf_GPOS_SEEN);
3537 /* For anchored \G, the only position it can match from is
3538 * (ganch-gofs); we already set startpos to this above; if intuit
3539 * moved us on from there, we can't possibly succeed */
3540 assert(startpos == HOPBACKc(reginfo->ganch, prog->gofs));
3541 if (s == startpos && regtry(reginfo, &s))
3542 goto got_it;
3543 goto phooey;
3544 }
3545
3546 /* Messy cases: unanchored match. */
3547 if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
3548 /* we have /x+whatever/ */
3549 /* it must be a one character string (XXXX Except is_utf8_pat?) */
3550 char ch;
3551 #ifdef DEBUGGING
3552 int did_match = 0;
3553 #endif
3554 if (utf8_target) {
3555 if (! prog->anchored_utf8) {
3556 to_utf8_substr(prog);
3557 }
3558 ch = SvPVX_const(prog->anchored_utf8)[0];
3559 REXEC_FBC_SCAN(0, /* 0=>not-utf8 */
3560 if (*s == ch) {
3561 DEBUG_EXECUTE_r( did_match = 1 );
3562 if (regtry(reginfo, &s)) goto got_it;
3563 s += UTF8_SAFE_SKIP(s, strend);
3564 while (s < strend && *s == ch)
3565 s += UTF8SKIP(s);
3566 }
3567 );
3568
3569 }
3570 else {
3571 if (! prog->anchored_substr) {
3572 if (! to_byte_substr(prog)) {
3573 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
3574 }
3575 }
3576 ch = SvPVX_const(prog->anchored_substr)[0];
3577 REXEC_FBC_SCAN(0, /* 0=>not-utf8 */
3578 if (*s == ch) {
3579 DEBUG_EXECUTE_r( did_match = 1 );
3580 if (regtry(reginfo, &s)) goto got_it;
3581 s++;
3582 while (s < strend && *s == ch)
3583 s++;
3584 }
3585 );
3586 }
3587 DEBUG_EXECUTE_r(if (!did_match)
3588 Perl_re_printf( aTHX_
3589 "Did not find anchored character...\n")
3590 );
3591 }
3592 else if (prog->anchored_substr != NULL
3593 || prog->anchored_utf8 != NULL
3594 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
3595 && prog->float_max_offset < strend - s)) {
3596 SV *must;
3597 SSize_t back_max;
3598 SSize_t back_min;
3599 char *last;
3600 char *last1; /* Last position checked before */
3601 #ifdef DEBUGGING
3602 int did_match = 0;
3603 #endif
3604 if (prog->anchored_substr || prog->anchored_utf8) {
3605 if (utf8_target) {
3606 if (! prog->anchored_utf8) {
3607 to_utf8_substr(prog);
3608 }
3609 must = prog->anchored_utf8;
3610 }
3611 else {
3612 if (! prog->anchored_substr) {
3613 if (! to_byte_substr(prog)) {
3614 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
3615 }
3616 }
3617 must = prog->anchored_substr;
3618 }
3619 back_max = back_min = prog->anchored_offset;
3620 } else {
3621 if (utf8_target) {
3622 if (! prog->float_utf8) {
3623 to_utf8_substr(prog);
3624 }
3625 must = prog->float_utf8;
3626 }
3627 else {
3628 if (! prog->float_substr) {
3629 if (! to_byte_substr(prog)) {
3630 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
3631 }
3632 }
3633 must = prog->float_substr;
3634 }
3635 back_max = prog->float_max_offset;
3636 back_min = prog->float_min_offset;
3637 }
3638
3639 if (back_min<0) {
3640 last = strend;
3641 } else {
3642 last = HOP3c(strend, /* Cannot start after this */
3643 -(SSize_t)(CHR_SVLEN(must)
3644 - (SvTAIL(must) != 0) + back_min), strbeg);
3645 }
3646 if (s > reginfo->strbeg)
3647 last1 = HOPc(s, -1);
3648 else
3649 last1 = s - 1; /* bogus */
3650
3651 /* XXXX check_substr already used to find "s", can optimize if
3652 check_substr==must. */
3653 dontbother = 0;
3654 strend = HOPc(strend, -dontbother);
3655 while ( (s <= last) &&
3656 (s = fbm_instr((unsigned char*)HOP4c(s, back_min, strbeg, strend),
3657 (unsigned char*)strend, must,
3658 multiline ? FBMrf_MULTILINE : 0)) ) {
3659 DEBUG_EXECUTE_r( did_match = 1 );
3660 if (HOPc(s, -back_max) > last1) {
3661 last1 = HOPc(s, -back_min);
3662 s = HOPc(s, -back_max);
3663 }
3664 else {
3665 char * const t = (last1 >= reginfo->strbeg)
3666 ? HOPc(last1, 1) : last1 + 1;
3667
3668 last1 = HOPc(s, -back_min);
3669 s = t;
3670 }
3671 if (utf8_target) {
3672 while (s <= last1) {
3673 if (regtry(reginfo, &s))
3674 goto got_it;
3675 if (s >= last1) {
3676 s++; /* to break out of outer loop */
3677 break;
3678 }
3679 s += UTF8SKIP(s);
3680 }
3681 }
3682 else {
3683 while (s <= last1) {
3684 if (regtry(reginfo, &s))
3685 goto got_it;
3686 s++;
3687 }
3688 }
3689 }
3690 DEBUG_EXECUTE_r(if (!did_match) {
3691 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
3692 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
3693 Perl_re_printf( aTHX_ "Did not find %s substr %s%s...\n",
3694 ((must == prog->anchored_substr || must == prog->anchored_utf8)
3695 ? "anchored" : "floating"),
3696 quoted, RE_SV_TAIL(must));
3697 });
3698 goto phooey;
3699 }
3700 else if ( (c = progi->regstclass) ) {
3701 if (minlen) {
3702 const OPCODE op = OP(progi->regstclass);
3703 /* don't bother with what can't match */
3704 if (PL_regkind[op] != EXACT && PL_regkind[op] != TRIE)
3705 strend = HOPc(strend, -(minlen - 1));
3706 }
3707 DEBUG_EXECUTE_r({
3708 SV * const prop = sv_newmortal();
3709 regprop(prog, prop, c, reginfo, NULL);
3710 {
3711 RE_PV_QUOTED_DECL(quoted,utf8_target,PERL_DEBUG_PAD_ZERO(1),
3712 s,strend-s,PL_dump_re_max_len);
3713 Perl_re_printf( aTHX_
3714 "Matching stclass %.*s against %s (%d bytes)\n",
3715 (int)SvCUR(prop), SvPVX_const(prop),
3716 quoted, (int)(strend - s));
3717 }
3718 });
3719 if (find_byclass(prog, c, s, strend, reginfo))
3720 goto got_it;
3721 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_ "Contradicts stclass... [regexec_flags]\n"));
3722 }
3723 else {
3724 dontbother = 0;
3725 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
3726 /* Trim the end. */
3727 char *last= NULL;
3728 SV* float_real;
3729 STRLEN len;
3730 const char *little;
3731
3732 if (utf8_target) {
3733 if (! prog->float_utf8) {
3734 to_utf8_substr(prog);
3735 }
3736 float_real = prog->float_utf8;
3737 }
3738 else {
3739 if (! prog->float_substr) {
3740 if (! to_byte_substr(prog)) {
3741 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
3742 }
3743 }
3744 float_real = prog->float_substr;
3745 }
3746
3747 little = SvPV_const(float_real, len);
3748 if (SvTAIL(float_real)) {
3749 /* This means that float_real contains an artificial \n on
3750 * the end due to the presence of something like this:
3751 * /foo$/ where we can match both "foo" and "foo\n" at the
3752 * end of the string. So we have to compare the end of the
3753 * string first against the float_real without the \n and
3754 * then against the full float_real with the string. We
3755 * have to watch out for cases where the string might be
3756 * smaller than the float_real or the float_real without
3757 * the \n. */
3758 char *checkpos= strend - len;
3759 DEBUG_OPTIMISE_r(
3760 Perl_re_printf( aTHX_
3761 "%sChecking for float_real.%s\n",
3762 PL_colors[4], PL_colors[5]));
3763 if (checkpos + 1 < strbeg) {
3764 /* can't match, even if we remove the trailing \n
3765 * string is too short to match */
3766 DEBUG_EXECUTE_r(
3767 Perl_re_printf( aTHX_
3768 "%sString shorter than required trailing substring, cannot match.%s\n",
3769 PL_colors[4], PL_colors[5]));
3770 goto phooey;
3771 } else if (memEQ(checkpos + 1, little, len - 1)) {
3772 /* can match, the end of the string matches without the
3773 * "\n" */
3774 last = checkpos + 1;
3775 } else if (checkpos < strbeg) {
3776 /* cant match, string is too short when the "\n" is
3777 * included */
3778 DEBUG_EXECUTE_r(
3779 Perl_re_printf( aTHX_
3780 "%sString does not contain required trailing substring, cannot match.%s\n",
3781 PL_colors[4], PL_colors[5]));
3782 goto phooey;
3783 } else if (!multiline) {
3784 /* non multiline match, so compare with the "\n" at the
3785 * end of the string */
3786 if (memEQ(checkpos, little, len)) {
3787 last= checkpos;
3788 } else {
3789 DEBUG_EXECUTE_r(
3790 Perl_re_printf( aTHX_
3791 "%sString does not contain required trailing substring, cannot match.%s\n",
3792 PL_colors[4], PL_colors[5]));
3793 goto phooey;
3794 }
3795 } else {
3796 /* multiline match, so we have to search for a place
3797 * where the full string is located */
3798 goto find_last;
3799 }
3800 } else {
3801 find_last:
3802 if (len)
3803 last = rninstr(s, strend, little, little + len);
3804 else
3805 last = strend; /* matching "$" */
3806 }
3807 if (!last) {
3808 /* at one point this block contained a comment which was
3809 * probably incorrect, which said that this was a "should not
3810 * happen" case. Even if it was true when it was written I am
3811 * pretty sure it is not anymore, so I have removed the comment
3812 * and replaced it with this one. Yves */
3813 DEBUG_EXECUTE_r(
3814 Perl_re_printf( aTHX_
3815 "%sString does not contain required substring, cannot match.%s\n",
3816 PL_colors[4], PL_colors[5]
3817 ));
3818 goto phooey;
3819 }
3820 dontbother = strend - last + prog->float_min_offset;
3821 }
3822 if (minlen && (dontbother < minlen))
3823 dontbother = minlen - 1;
3824 strend -= dontbother; /* this one's always in bytes! */
3825 /* We don't know much -- general case. */
3826 if (utf8_target) {
3827 for (;;) {
3828 if (regtry(reginfo, &s))
3829 goto got_it;
3830 if (s >= strend)
3831 break;
3832 s += UTF8SKIP(s);
3833 };
3834 }
3835 else {
3836 do {
3837 if (regtry(reginfo, &s))
3838 goto got_it;
3839 } while (s++ < strend);
3840 }
3841 }
3842
3843 /* Failure. */
3844 goto phooey;
3845
3846 got_it:
3847 /* s/// doesn't like it if $& is earlier than where we asked it to
3848 * start searching (which can happen on something like /.\G/) */
3849 if ( (flags & REXEC_FAIL_ON_UNDERFLOW)
3850 && (prog->offs[0].start < stringarg - strbeg))
3851 {
3852 /* this should only be possible under \G */
3853 assert(prog->intflags & PREGf_GPOS_SEEN);
3854 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
3855 "matched, but failing for REXEC_FAIL_ON_UNDERFLOW\n"));
3856 goto phooey;
3857 }
3858
3859 /* clean up; this will trigger destructors that will free all slabs
3860 * above the current one, and cleanup the regmatch_info_aux
3861 * and regmatch_info_aux_eval sructs */
3862
3863 LEAVE_SCOPE(oldsave);
3864
3865 if (RXp_PAREN_NAMES(prog))
3866 (void)hv_iterinit(RXp_PAREN_NAMES(prog));
3867
3868 /* make sure $`, $&, $', and $digit will work later */
3869 if ( !(flags & REXEC_NOT_FIRST) )
3870 S_reg_set_capture_string(aTHX_ rx,
3871 strbeg, reginfo->strend,
3872 sv, flags, utf8_target);
3873
3874 return 1;
3875
3876 phooey:
3877 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_ "%sMatch failed%s\n",
3878 PL_colors[4], PL_colors[5]));
3879
3880 if (swap) {
3881 /* we failed :-( roll it back.
3882 * Since the swap buffer will be freed on scope exit which follows
3883 * shortly, restore the old captures by copying 'swap's original
3884 * data to the new offs buffer
3885 */
3886 DEBUG_BUFFERS_r(Perl_re_exec_indentf( aTHX_
3887 "rex=0x%" UVxf " rolling back offs: 0x%" UVxf " will be freed; restoring data to =0x%" UVxf "\n",
3888 0,
3889 PTR2UV(prog),
3890 PTR2UV(prog->offs),
3891 PTR2UV(swap)
3892 ));
3893
3894 Copy(swap, prog->offs, prog->nparens + 1, regexp_paren_pair);
3895 }
3896
3897 /* clean up; this will trigger destructors that will free all slabs
3898 * above the current one, and cleanup the regmatch_info_aux
3899 * and regmatch_info_aux_eval sructs */
3900
3901 LEAVE_SCOPE(oldsave);
3902
3903 return 0;
3904 }
3905
3906
3907 /* Set which rex is pointed to by PL_reg_curpm, handling ref counting.
3908 * Do inc before dec, in case old and new rex are the same */
3909 #define SET_reg_curpm(Re2) \
3910 if (reginfo->info_aux_eval) { \
3911 (void)ReREFCNT_inc(Re2); \
3912 ReREFCNT_dec(PM_GETRE(PL_reg_curpm)); \
3913 PM_SETRE((PL_reg_curpm), (Re2)); \
3914 }
3915
3916
3917 /*
3918 - regtry - try match at specific point
3919 */
3920 STATIC bool /* 0 failure, 1 success */
S_regtry(pTHX_ regmatch_info * reginfo,char ** startposp)3921 S_regtry(pTHX_ regmatch_info *reginfo, char **startposp)
3922 {
3923 CHECKPOINT lastcp;
3924 REGEXP *const rx = reginfo->prog;
3925 regexp *const prog = ReANY(rx);
3926 SSize_t result;
3927 #ifdef DEBUGGING
3928 U32 depth = 0; /* used by REGCP_SET */
3929 #endif
3930 RXi_GET_DECL(prog,progi);
3931 GET_RE_DEBUG_FLAGS_DECL;
3932
3933 PERL_ARGS_ASSERT_REGTRY;
3934
3935 reginfo->cutpoint=NULL;
3936
3937 prog->offs[0].start = *startposp - reginfo->strbeg;
3938 prog->lastparen = 0;
3939 prog->lastcloseparen = 0;
3940
3941 /* XXXX What this code is doing here?!!! There should be no need
3942 to do this again and again, prog->lastparen should take care of
3943 this! --ilya*/
3944
3945 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
3946 * Actually, the code in regcppop() (which Ilya may be meaning by
3947 * prog->lastparen), is not needed at all by the test suite
3948 * (op/regexp, op/pat, op/split), but that code is needed otherwise
3949 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
3950 * Meanwhile, this code *is* needed for the
3951 * above-mentioned test suite tests to succeed. The common theme
3952 * on those tests seems to be returning null fields from matches.
3953 * --jhi updated by dapm */
3954
3955 /* After encountering a variant of the issue mentioned above I think
3956 * the point Ilya was making is that if we properly unwind whenever
3957 * we set lastparen to a smaller value then we should not need to do
3958 * this every time, only when needed. So if we have tests that fail if
3959 * we remove this, then it suggests somewhere else we are improperly
3960 * unwinding the lastparen/paren buffers. See UNWIND_PARENS() and
3961 * places it is called, and related regcp() routines. - Yves */
3962 #if 1
3963 if (prog->nparens) {
3964 regexp_paren_pair *pp = prog->offs;
3965 I32 i;
3966 for (i = prog->nparens; i > (I32)prog->lastparen; i--) {
3967 ++pp;
3968 pp->start = -1;
3969 pp->end = -1;
3970 }
3971 }
3972 #endif
3973 REGCP_SET(lastcp);
3974 result = regmatch(reginfo, *startposp, progi->program + 1);
3975 if (result != -1) {
3976 prog->offs[0].end = result;
3977 return 1;
3978 }
3979 if (reginfo->cutpoint)
3980 *startposp= reginfo->cutpoint;
3981 REGCP_UNWIND(lastcp);
3982 return 0;
3983 }
3984
3985 /* this is used to determine how far from the left messages like
3986 'failed...' are printed in regexec.c. It should be set such that
3987 messages are inline with the regop output that created them.
3988 */
3989 #define REPORT_CODE_OFF 29
3990 #define INDENT_CHARS(depth) ((int)(depth) % 20)
3991 #ifdef DEBUGGING
3992 int
Perl_re_exec_indentf(pTHX_ const char * fmt,U32 depth,...)3993 Perl_re_exec_indentf(pTHX_ const char *fmt, U32 depth, ...)
3994 {
3995 va_list ap;
3996 int result;
3997 PerlIO *f= Perl_debug_log;
3998 PERL_ARGS_ASSERT_RE_EXEC_INDENTF;
3999 va_start(ap, depth);
4000 PerlIO_printf(f, "%*s|%4" UVuf "| %*s", REPORT_CODE_OFF, "", (UV)depth, INDENT_CHARS(depth), "" );
4001 result = PerlIO_vprintf(f, fmt, ap);
4002 va_end(ap);
4003 return result;
4004 }
4005 #endif /* DEBUGGING */
4006
4007 /* grab a new slab and return the first slot in it */
4008
4009 STATIC regmatch_state *
S_push_slab(pTHX)4010 S_push_slab(pTHX)
4011 {
4012 regmatch_slab *s = PL_regmatch_slab->next;
4013 if (!s) {
4014 Newx(s, 1, regmatch_slab);
4015 s->prev = PL_regmatch_slab;
4016 s->next = NULL;
4017 PL_regmatch_slab->next = s;
4018 }
4019 PL_regmatch_slab = s;
4020 return SLAB_FIRST(s);
4021 }
4022
4023 #ifdef DEBUGGING
4024
4025 STATIC void
S_debug_start_match(pTHX_ const REGEXP * prog,const bool utf8_target,const char * start,const char * end,const char * blurb)4026 S_debug_start_match(pTHX_ const REGEXP *prog, const bool utf8_target,
4027 const char *start, const char *end, const char *blurb)
4028 {
4029 const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;
4030
4031 PERL_ARGS_ASSERT_DEBUG_START_MATCH;
4032
4033 if (!PL_colorset)
4034 reginitcolors();
4035 {
4036 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
4037 RX_PRECOMP_const(prog), RX_PRELEN(prog), PL_dump_re_max_len);
4038
4039 RE_PV_QUOTED_DECL(s1, utf8_target, PERL_DEBUG_PAD_ZERO(1),
4040 start, end - start, PL_dump_re_max_len);
4041
4042 Perl_re_printf( aTHX_
4043 "%s%s REx%s %s against %s\n",
4044 PL_colors[4], blurb, PL_colors[5], s0, s1);
4045
4046 if (utf8_target||utf8_pat)
4047 Perl_re_printf( aTHX_ "UTF-8 %s%s%s...\n",
4048 utf8_pat ? "pattern" : "",
4049 utf8_pat && utf8_target ? " and " : "",
4050 utf8_target ? "string" : ""
4051 );
4052 }
4053 }
4054
4055 STATIC void
S_dump_exec_pos(pTHX_ const char * locinput,const regnode * scan,const char * loc_regeol,const char * loc_bostr,const char * loc_reg_starttry,const bool utf8_target,const U32 depth)4056 S_dump_exec_pos(pTHX_ const char *locinput,
4057 const regnode *scan,
4058 const char *loc_regeol,
4059 const char *loc_bostr,
4060 const char *loc_reg_starttry,
4061 const bool utf8_target,
4062 const U32 depth
4063 )
4064 {
4065 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
4066 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
4067 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
4068 /* The part of the string before starttry has one color
4069 (pref0_len chars), between starttry and current
4070 position another one (pref_len - pref0_len chars),
4071 after the current position the third one.
4072 We assume that pref0_len <= pref_len, otherwise we
4073 decrease pref0_len. */
4074 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
4075 ? (5 + taill) - l : locinput - loc_bostr;
4076 int pref0_len;
4077
4078 PERL_ARGS_ASSERT_DUMP_EXEC_POS;
4079
4080 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
4081 pref_len++;
4082 pref0_len = pref_len - (locinput - loc_reg_starttry);
4083 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
4084 l = ( loc_regeol - locinput > (5 + taill) - pref_len
4085 ? (5 + taill) - pref_len : loc_regeol - locinput);
4086 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
4087 l--;
4088 if (pref0_len < 0)
4089 pref0_len = 0;
4090 if (pref0_len > pref_len)
4091 pref0_len = pref_len;
4092 {
4093 const int is_uni = utf8_target ? 1 : 0;
4094
4095 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
4096 (locinput - pref_len),pref0_len, PL_dump_re_max_len, 4, 5);
4097
4098 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
4099 (locinput - pref_len + pref0_len),
4100 pref_len - pref0_len, PL_dump_re_max_len, 2, 3);
4101
4102 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
4103 locinput, loc_regeol - locinput, 10, 0, 1);
4104
4105 const STRLEN tlen=len0+len1+len2;
4106 Perl_re_printf( aTHX_
4107 "%4" IVdf " <%.*s%.*s%s%.*s>%*s|%4u| ",
4108 (IV)(locinput - loc_bostr),
4109 len0, s0,
4110 len1, s1,
4111 (docolor ? "" : "> <"),
4112 len2, s2,
4113 (int)(tlen > 19 ? 0 : 19 - tlen),
4114 "",
4115 depth);
4116 }
4117 }
4118
4119 #endif
4120
4121 /* reg_check_named_buff_matched()
4122 * Checks to see if a named buffer has matched. The data array of
4123 * buffer numbers corresponding to the buffer is expected to reside
4124 * in the regexp->data->data array in the slot stored in the ARG() of
4125 * node involved. Note that this routine doesn't actually care about the
4126 * name, that information is not preserved from compilation to execution.
4127 * Returns the index of the leftmost defined buffer with the given name
4128 * or 0 if non of the buffers matched.
4129 */
4130 STATIC I32
S_reg_check_named_buff_matched(const regexp * rex,const regnode * scan)4131 S_reg_check_named_buff_matched(const regexp *rex, const regnode *scan)
4132 {
4133 I32 n;
4134 RXi_GET_DECL(rex,rexi);
4135 SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
4136 I32 *nums=(I32*)SvPVX(sv_dat);
4137
4138 PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED;
4139
4140 for ( n=0; n<SvIVX(sv_dat); n++ ) {
4141 if ((I32)rex->lastparen >= nums[n] &&
4142 rex->offs[nums[n]].end != -1)
4143 {
4144 return nums[n];
4145 }
4146 }
4147 return 0;
4148 }
4149
4150 #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
4151 #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
4152 #define CHRTEST_NOT_A_CP_1 -999
4153 #define CHRTEST_NOT_A_CP_2 -998
4154
4155 static bool
S_setup_EXACTISH_ST_c1_c2(pTHX_ const regnode * const text_node,int * c1p,U8 * c1_utf8,int * c2p,U8 * c2_utf8,regmatch_info * reginfo)4156 S_setup_EXACTISH_ST_c1_c2(pTHX_ const regnode * const text_node, int *c1p,
4157 U8* c1_utf8, int *c2p, U8* c2_utf8, regmatch_info *reginfo)
4158 {
4159 /* This function determines if there are zero, one, two, or more characters
4160 * that match the first character of the passed-in EXACTish node
4161 * <text_node>, and if there are one or two, it returns them in the
4162 * passed-in pointers.
4163 *
4164 * If it determines that no possible character in the target string can
4165 * match, it returns FALSE; otherwise TRUE. (The FALSE situation occurs if
4166 * the first character in <text_node> requires UTF-8 to represent, and the
4167 * target string isn't in UTF-8.)
4168 *
4169 * If there are more than two characters that could match the beginning of
4170 * <text_node>, or if more context is required to determine a match or not,
4171 * it sets both *<c1p> and *<c2p> to CHRTEST_VOID.
4172 *
4173 * The motiviation behind this function is to allow the caller to set up
4174 * tight loops for matching. If <text_node> is of type EXACT, there is
4175 * only one possible character that can match its first character, and so
4176 * the situation is quite simple. But things get much more complicated if
4177 * folding is involved. It may be that the first character of an EXACTFish
4178 * node doesn't participate in any possible fold, e.g., punctuation, so it
4179 * can be matched only by itself. The vast majority of characters that are
4180 * in folds match just two things, their lower and upper-case equivalents.
4181 * But not all are like that; some have multiple possible matches, or match
4182 * sequences of more than one character. This function sorts all that out.
4183 *
4184 * Consider the patterns A*B or A*?B where A and B are arbitrary. In a
4185 * loop of trying to match A*, we know we can't exit where the thing
4186 * following it isn't a B. And something can't be a B unless it is the
4187 * beginning of B. By putting a quick test for that beginning in a tight
4188 * loop, we can rule out things that can't possibly be B without having to
4189 * break out of the loop, thus avoiding work. Similarly, if A is a single
4190 * character, we can make a tight loop matching A*, using the outputs of
4191 * this function.
4192 *
4193 * If the target string to match isn't in UTF-8, and there aren't
4194 * complications which require CHRTEST_VOID, *<c1p> and *<c2p> are set to
4195 * the one or two possible octets (which are characters in this situation)
4196 * that can match. In all cases, if there is only one character that can
4197 * match, *<c1p> and *<c2p> will be identical.
4198 *
4199 * If the target string is in UTF-8, the buffers pointed to by <c1_utf8>
4200 * and <c2_utf8> will contain the one or two UTF-8 sequences of bytes that
4201 * can match the beginning of <text_node>. They should be declared with at
4202 * least length UTF8_MAXBYTES+1. (If the target string isn't in UTF-8, it is
4203 * undefined what these contain.) If one or both of the buffers are
4204 * invariant under UTF-8, *<c1p>, and *<c2p> will also be set to the
4205 * corresponding invariant. If variant, the corresponding *<c1p> and/or
4206 * *<c2p> will be set to a negative number(s) that shouldn't match any code
4207 * point (unless inappropriately coerced to unsigned). *<c1p> will equal
4208 * *<c2p> if and only if <c1_utf8> and <c2_utf8> are the same. */
4209
4210 const bool utf8_target = reginfo->is_utf8_target;
4211
4212 UV c1 = (UV)CHRTEST_NOT_A_CP_1;
4213 UV c2 = (UV)CHRTEST_NOT_A_CP_2;
4214 bool use_chrtest_void = FALSE;
4215 const bool is_utf8_pat = reginfo->is_utf8_pat;
4216
4217 /* Used when we have both utf8 input and utf8 output, to avoid converting
4218 * to/from code points */
4219 bool utf8_has_been_setup = FALSE;
4220
4221 dVAR;
4222
4223 U8 *pat = (U8*)STRING(text_node);
4224 U8 folded[UTF8_MAX_FOLD_CHAR_EXPAND * UTF8_MAXBYTES_CASE + 1] = { '\0' };
4225
4226 if ( OP(text_node) == EXACT
4227 || OP(text_node) == EXACT_ONLY8
4228 || OP(text_node) == EXACTL)
4229 {
4230
4231 /* In an exact node, only one thing can be matched, that first
4232 * character. If both the pat and the target are UTF-8, we can just
4233 * copy the input to the output, avoiding finding the code point of
4234 * that character */
4235 if (!is_utf8_pat) {
4236 assert(OP(text_node) != EXACT_ONLY8);
4237 c2 = c1 = *pat;
4238 }
4239 else if (utf8_target) {
4240 Copy(pat, c1_utf8, UTF8SKIP(pat), U8);
4241 Copy(pat, c2_utf8, UTF8SKIP(pat), U8);
4242 utf8_has_been_setup = TRUE;
4243 }
4244 else if (OP(text_node) == EXACT_ONLY8) {
4245 return FALSE; /* Can only match UTF-8 target */
4246 }
4247 else {
4248 c2 = c1 = valid_utf8_to_uvchr(pat, NULL);
4249 }
4250 }
4251 else { /* an EXACTFish node */
4252 U8 *pat_end = pat + STR_LEN(text_node);
4253
4254 /* An EXACTFL node has at least some characters unfolded, because what
4255 * they match is not known until now. So, now is the time to fold
4256 * the first few of them, as many as are needed to determine 'c1' and
4257 * 'c2' later in the routine. If the pattern isn't UTF-8, we only need
4258 * to fold if in a UTF-8 locale, and then only the Sharp S; everything
4259 * else is 1-1 and isn't assumed to be folded. In a UTF-8 pattern, we
4260 * need to fold as many characters as a single character can fold to,
4261 * so that later we can check if the first ones are such a multi-char
4262 * fold. But, in such a pattern only locale-problematic characters
4263 * aren't folded, so we can skip this completely if the first character
4264 * in the node isn't one of the tricky ones */
4265 if (OP(text_node) == EXACTFL) {
4266
4267 if (! is_utf8_pat) {
4268 if (IN_UTF8_CTYPE_LOCALE && *pat == LATIN_SMALL_LETTER_SHARP_S)
4269 {
4270 folded[0] = folded[1] = 's';
4271 pat = folded;
4272 pat_end = folded + 2;
4273 }
4274 }
4275 else if (is_PROBLEMATIC_LOCALE_FOLDEDS_START_utf8(pat)) {
4276 U8 *s = pat;
4277 U8 *d = folded;
4278 int i;
4279
4280 for (i = 0; i < UTF8_MAX_FOLD_CHAR_EXPAND && s < pat_end; i++) {
4281 if (isASCII(*s) && LIKELY(! PL_in_utf8_turkic_locale)) {
4282 *(d++) = (U8) toFOLD_LC(*s);
4283 s++;
4284 }
4285 else {
4286 STRLEN len;
4287 _toFOLD_utf8_flags(s,
4288 pat_end,
4289 d,
4290 &len,
4291 FOLD_FLAGS_FULL | FOLD_FLAGS_LOCALE);
4292 d += len;
4293 s += UTF8SKIP(s);
4294 }
4295 }
4296
4297 pat = folded;
4298 pat_end = d;
4299 }
4300 }
4301
4302 if ( ( is_utf8_pat && is_MULTI_CHAR_FOLD_utf8_safe(pat, pat_end))
4303 || (!is_utf8_pat && is_MULTI_CHAR_FOLD_latin1_safe(pat, pat_end)))
4304 {
4305 /* Multi-character folds require more context to sort out. Also
4306 * PL_utf8_foldclosures used below doesn't handle them, so have to
4307 * be handled outside this routine */
4308 use_chrtest_void = TRUE;
4309 }
4310 else { /* an EXACTFish node which doesn't begin with a multi-char fold */
4311 c1 = is_utf8_pat ? valid_utf8_to_uvchr(pat, NULL) : *pat;
4312
4313 if ( UNLIKELY(PL_in_utf8_turkic_locale)
4314 && OP(text_node) == EXACTFL
4315 && UNLIKELY( c1 == 'i' || c1 == 'I'
4316 || c1 == LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE
4317 || c1 == LATIN_SMALL_LETTER_DOTLESS_I))
4318 { /* Hard-coded Turkish locale rules for these 4 characters
4319 override normal rules */
4320 if (c1 == 'i') {
4321 c2 = LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE;
4322 }
4323 else if (c1 == 'I') {
4324 c2 = LATIN_SMALL_LETTER_DOTLESS_I;
4325 }
4326 else if (c1 == LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE) {
4327 c2 = 'i';
4328 }
4329 else if (c1 == LATIN_SMALL_LETTER_DOTLESS_I) {
4330 c2 = 'I';
4331 }
4332 }
4333 else if (c1 > 255) {
4334 const unsigned int * remaining_folds;
4335 unsigned int first_fold;
4336
4337 /* Look up what code points (besides c1) fold to c1; e.g.,
4338 * [ 'K', KELVIN_SIGN ] both fold to 'k'. */
4339 Size_t folds_count = _inverse_folds(c1, &first_fold,
4340 &remaining_folds);
4341 if (folds_count == 0) {
4342 c2 = c1; /* there is only a single character that could
4343 match */
4344 }
4345 else if (folds_count != 1) {
4346 /* If there aren't exactly two folds to this (itself and
4347 * another), it is outside the scope of this function */
4348 use_chrtest_void = TRUE;
4349 }
4350 else { /* There are two. We already have one, get the other */
4351 c2 = first_fold;
4352
4353 /* Folds that cross the 255/256 boundary are forbidden if
4354 * EXACTFL (and isnt a UTF8 locale), or EXACTFAA and one is
4355 * ASCIII. The only other match to c1 is c2, and since c1
4356 * is above 255, c2 better be as well under these
4357 * circumstances. If it isn't, it means the only legal
4358 * match of c1 is itself. */
4359 if ( c2 < 256
4360 && ( ( OP(text_node) == EXACTFL
4361 && ! IN_UTF8_CTYPE_LOCALE)
4362 || (( OP(text_node) == EXACTFAA
4363 || OP(text_node) == EXACTFAA_NO_TRIE)
4364 && (isASCII(c1) || isASCII(c2)))))
4365 {
4366 c2 = c1;
4367 }
4368 }
4369 }
4370 else /* Here, c1 is <= 255 */
4371 if ( utf8_target
4372 && HAS_NONLATIN1_FOLD_CLOSURE(c1)
4373 && ( ! (OP(text_node) == EXACTFL && ! IN_UTF8_CTYPE_LOCALE))
4374 && ( ( OP(text_node) != EXACTFAA
4375 && OP(text_node) != EXACTFAA_NO_TRIE)
4376 || ! isASCII(c1)))
4377 {
4378 /* Here, there could be something above Latin1 in the target
4379 * which folds to this character in the pattern. All such
4380 * cases except LATIN SMALL LETTER Y WITH DIAERESIS have more
4381 * than two characters involved in their folds, so are outside
4382 * the scope of this function */
4383 if (UNLIKELY(c1 == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS)) {
4384 c2 = LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS;
4385 }
4386 else {
4387 use_chrtest_void = TRUE;
4388 }
4389 }
4390 else { /* Here nothing above Latin1 can fold to the pattern
4391 character */
4392 switch (OP(text_node)) {
4393
4394 case EXACTFL: /* /l rules */
4395 c2 = PL_fold_locale[c1];
4396 break;
4397
4398 case EXACTF: /* This node only generated for non-utf8
4399 patterns */
4400 assert(! is_utf8_pat);
4401 if (! utf8_target) { /* /d rules */
4402 c2 = PL_fold[c1];
4403 break;
4404 }
4405 /* FALLTHROUGH */
4406 /* /u rules for all these. This happens to work for
4407 * EXACTFAA as nothing in Latin1 folds to ASCII */
4408 case EXACTFAA_NO_TRIE: /* This node only generated for
4409 non-utf8 patterns */
4410 assert(! is_utf8_pat);
4411 /* FALLTHROUGH */
4412 case EXACTFAA:
4413 case EXACTFUP:
4414 case EXACTFU:
4415 c2 = PL_fold_latin1[c1];
4416 break;
4417 case EXACTFU_ONLY8:
4418 return FALSE;
4419 NOT_REACHED; /* NOTREACHED */
4420
4421 default:
4422 Perl_croak(aTHX_ "panic: Unexpected op %u", OP(text_node));
4423 NOT_REACHED; /* NOTREACHED */
4424 }
4425 }
4426 }
4427 }
4428
4429 /* Here have figured things out. Set up the returns */
4430 if (use_chrtest_void) {
4431 *c2p = *c1p = CHRTEST_VOID;
4432 }
4433 else if (utf8_target) {
4434 if (! utf8_has_been_setup) { /* Don't have the utf8; must get it */
4435 uvchr_to_utf8(c1_utf8, c1);
4436 uvchr_to_utf8(c2_utf8, c2);
4437 }
4438
4439 /* Invariants are stored in both the utf8 and byte outputs; Use
4440 * negative numbers otherwise for the byte ones. Make sure that the
4441 * byte ones are the same iff the utf8 ones are the same */
4442 *c1p = (UTF8_IS_INVARIANT(*c1_utf8)) ? *c1_utf8 : CHRTEST_NOT_A_CP_1;
4443 *c2p = (UTF8_IS_INVARIANT(*c2_utf8))
4444 ? *c2_utf8
4445 : (c1 == c2)
4446 ? CHRTEST_NOT_A_CP_1
4447 : CHRTEST_NOT_A_CP_2;
4448 }
4449 else if (c1 > 255) {
4450 if (c2 > 255) { /* both possibilities are above what a non-utf8 string
4451 can represent */
4452 return FALSE;
4453 }
4454
4455 *c1p = *c2p = c2; /* c2 is the only representable value */
4456 }
4457 else { /* c1 is representable; see about c2 */
4458 *c1p = c1;
4459 *c2p = (c2 < 256) ? c2 : c1;
4460 }
4461
4462 return TRUE;
4463 }
4464
4465 STATIC bool
S_isGCB(pTHX_ const GCB_enum before,const GCB_enum after,const U8 * const strbeg,const U8 * const curpos,const bool utf8_target)4466 S_isGCB(pTHX_ const GCB_enum before, const GCB_enum after, const U8 * const strbeg, const U8 * const curpos, const bool utf8_target)
4467 {
4468 /* returns a boolean indicating if there is a Grapheme Cluster Boundary
4469 * between the inputs. See http://www.unicode.org/reports/tr29/. */
4470
4471 PERL_ARGS_ASSERT_ISGCB;
4472
4473 switch (GCB_table[before][after]) {
4474 case GCB_BREAKABLE:
4475 return TRUE;
4476
4477 case GCB_NOBREAK:
4478 return FALSE;
4479
4480 case GCB_RI_then_RI:
4481 {
4482 int RI_count = 1;
4483 U8 * temp_pos = (U8 *) curpos;
4484
4485 /* Do not break within emoji flag sequences. That is, do not
4486 * break between regional indicator (RI) symbols if there is an
4487 * odd number of RI characters before the break point.
4488 * GB12 sot (RI RI)* RI × RI
4489 * GB13 [^RI] (RI RI)* RI × RI */
4490
4491 while (backup_one_GCB(strbeg,
4492 &temp_pos,
4493 utf8_target) == GCB_Regional_Indicator)
4494 {
4495 RI_count++;
4496 }
4497
4498 return RI_count % 2 != 1;
4499 }
4500
4501 case GCB_EX_then_EM:
4502
4503 /* GB10 ( E_Base | E_Base_GAZ ) Extend* × E_Modifier */
4504 {
4505 U8 * temp_pos = (U8 *) curpos;
4506 GCB_enum prev;
4507
4508 do {
4509 prev = backup_one_GCB(strbeg, &temp_pos, utf8_target);
4510 }
4511 while (prev == GCB_Extend);
4512
4513 return prev != GCB_E_Base && prev != GCB_E_Base_GAZ;
4514 }
4515
4516 case GCB_Maybe_Emoji_NonBreak:
4517
4518 {
4519
4520 /* Do not break within emoji modifier sequences or emoji zwj sequences.
4521 GB11 \p{Extended_Pictographic} Extend* ZWJ × \p{Extended_Pictographic}
4522 */
4523 U8 * temp_pos = (U8 *) curpos;
4524 GCB_enum prev;
4525
4526 do {
4527 prev = backup_one_GCB(strbeg, &temp_pos, utf8_target);
4528 }
4529 while (prev == GCB_Extend);
4530
4531 return prev != GCB_XPG_XX;
4532 }
4533
4534 default:
4535 break;
4536 }
4537
4538 #ifdef DEBUGGING
4539 Perl_re_printf( aTHX_ "Unhandled GCB pair: GCB_table[%d, %d] = %d\n",
4540 before, after, GCB_table[before][after]);
4541 assert(0);
4542 #endif
4543 return TRUE;
4544 }
4545
4546 STATIC GCB_enum
S_backup_one_GCB(pTHX_ const U8 * const strbeg,U8 ** curpos,const bool utf8_target)4547 S_backup_one_GCB(pTHX_ const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
4548 {
4549 dVAR;
4550 GCB_enum gcb;
4551
4552 PERL_ARGS_ASSERT_BACKUP_ONE_GCB;
4553
4554 if (*curpos < strbeg) {
4555 return GCB_EDGE;
4556 }
4557
4558 if (utf8_target) {
4559 U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
4560 U8 * prev_prev_char_pos;
4561
4562 if (! prev_char_pos) {
4563 return GCB_EDGE;
4564 }
4565
4566 if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos, -1, strbeg))) {
4567 gcb = getGCB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
4568 *curpos = prev_char_pos;
4569 prev_char_pos = prev_prev_char_pos;
4570 }
4571 else {
4572 *curpos = (U8 *) strbeg;
4573 return GCB_EDGE;
4574 }
4575 }
4576 else {
4577 if (*curpos - 2 < strbeg) {
4578 *curpos = (U8 *) strbeg;
4579 return GCB_EDGE;
4580 }
4581 (*curpos)--;
4582 gcb = getGCB_VAL_CP(*(*curpos - 1));
4583 }
4584
4585 return gcb;
4586 }
4587
4588 /* Combining marks attach to most classes that precede them, but this defines
4589 * the exceptions (from TR14) */
4590 #define LB_CM_ATTACHES_TO(prev) ( ! ( prev == LB_EDGE \
4591 || prev == LB_Mandatory_Break \
4592 || prev == LB_Carriage_Return \
4593 || prev == LB_Line_Feed \
4594 || prev == LB_Next_Line \
4595 || prev == LB_Space \
4596 || prev == LB_ZWSpace))
4597
4598 STATIC bool
S_isLB(pTHX_ LB_enum before,LB_enum after,const U8 * const strbeg,const U8 * const curpos,const U8 * const strend,const bool utf8_target)4599 S_isLB(pTHX_ LB_enum before,
4600 LB_enum after,
4601 const U8 * const strbeg,
4602 const U8 * const curpos,
4603 const U8 * const strend,
4604 const bool utf8_target)
4605 {
4606 U8 * temp_pos = (U8 *) curpos;
4607 LB_enum prev = before;
4608
4609 /* Is the boundary between 'before' and 'after' line-breakable?
4610 * Most of this is just a table lookup of a generated table from Unicode
4611 * rules. But some rules require context to decide, and so have to be
4612 * implemented in code */
4613
4614 PERL_ARGS_ASSERT_ISLB;
4615
4616 /* Rule numbers in the comments below are as of Unicode 9.0 */
4617
4618 redo:
4619 before = prev;
4620 switch (LB_table[before][after]) {
4621 case LB_BREAKABLE:
4622 return TRUE;
4623
4624 case LB_NOBREAK:
4625 case LB_NOBREAK_EVEN_WITH_SP_BETWEEN:
4626 return FALSE;
4627
4628 case LB_SP_foo + LB_BREAKABLE:
4629 case LB_SP_foo + LB_NOBREAK:
4630 case LB_SP_foo + LB_NOBREAK_EVEN_WITH_SP_BETWEEN:
4631
4632 /* When we have something following a SP, we have to look at the
4633 * context in order to know what to do.
4634 *
4635 * SP SP should not reach here because LB7: Do not break before
4636 * spaces. (For two spaces in a row there is nothing that
4637 * overrides that) */
4638 assert(after != LB_Space);
4639
4640 /* Here we have a space followed by a non-space. Mostly this is a
4641 * case of LB18: "Break after spaces". But there are complications
4642 * as the handling of spaces is somewhat tricky. They are in a
4643 * number of rules, which have to be applied in priority order, but
4644 * something earlier in the string can cause a rule to be skipped
4645 * and a lower priority rule invoked. A prime example is LB7 which
4646 * says don't break before a space. But rule LB8 (lower priority)
4647 * says that the first break opportunity after a ZW is after any
4648 * span of spaces immediately after it. If a ZW comes before a SP
4649 * in the input, rule LB8 applies, and not LB7. Other such rules
4650 * involve combining marks which are rules 9 and 10, but they may
4651 * override higher priority rules if they come earlier in the
4652 * string. Since we're doing random access into the middle of the
4653 * string, we have to look for rules that should get applied based
4654 * on both string position and priority. Combining marks do not
4655 * attach to either ZW nor SP, so we don't have to consider them
4656 * until later.
4657 *
4658 * To check for LB8, we have to find the first non-space character
4659 * before this span of spaces */
4660 do {
4661 prev = backup_one_LB(strbeg, &temp_pos, utf8_target);
4662 }
4663 while (prev == LB_Space);
4664
4665 /* LB8 Break before any character following a zero-width space,
4666 * even if one or more spaces intervene.
4667 * ZW SP* ÷
4668 * So if we have a ZW just before this span, and to get here this
4669 * is the final space in the span. */
4670 if (prev == LB_ZWSpace) {
4671 return TRUE;
4672 }
4673
4674 /* Here, not ZW SP+. There are several rules that have higher
4675 * priority than LB18 and can be resolved now, as they don't depend
4676 * on anything earlier in the string (except ZW, which we have
4677 * already handled). One of these rules is LB11 Do not break
4678 * before Word joiner, but we have specially encoded that in the
4679 * lookup table so it is caught by the single test below which
4680 * catches the other ones. */
4681 if (LB_table[LB_Space][after] - LB_SP_foo
4682 == LB_NOBREAK_EVEN_WITH_SP_BETWEEN)
4683 {
4684 return FALSE;
4685 }
4686
4687 /* If we get here, we have to XXX consider combining marks. */
4688 if (prev == LB_Combining_Mark) {
4689
4690 /* What happens with these depends on the character they
4691 * follow. */
4692 do {
4693 prev = backup_one_LB(strbeg, &temp_pos, utf8_target);
4694 }
4695 while (prev == LB_Combining_Mark);
4696
4697 /* Most times these attach to and inherit the characteristics
4698 * of that character, but not always, and when not, they are to
4699 * be treated as AL by rule LB10. */
4700 if (! LB_CM_ATTACHES_TO(prev)) {
4701 prev = LB_Alphabetic;
4702 }
4703 }
4704
4705 /* Here, we have the character preceding the span of spaces all set
4706 * up. We follow LB18: "Break after spaces" unless the table shows
4707 * that is overriden */
4708 return LB_table[prev][after] != LB_NOBREAK_EVEN_WITH_SP_BETWEEN;
4709
4710 case LB_CM_ZWJ_foo:
4711
4712 /* We don't know how to treat the CM except by looking at the first
4713 * non-CM character preceding it. ZWJ is treated as CM */
4714 do {
4715 prev = backup_one_LB(strbeg, &temp_pos, utf8_target);
4716 }
4717 while (prev == LB_Combining_Mark || prev == LB_ZWJ);
4718
4719 /* Here, 'prev' is that first earlier non-CM character. If the CM
4720 * attatches to it, then it inherits the behavior of 'prev'. If it
4721 * doesn't attach, it is to be treated as an AL */
4722 if (! LB_CM_ATTACHES_TO(prev)) {
4723 prev = LB_Alphabetic;
4724 }
4725
4726 goto redo;
4727
4728 case LB_HY_or_BA_then_foo + LB_BREAKABLE:
4729 case LB_HY_or_BA_then_foo + LB_NOBREAK:
4730
4731 /* LB21a Don't break after Hebrew + Hyphen.
4732 * HL (HY | BA) × */
4733
4734 if (backup_one_LB(strbeg, &temp_pos, utf8_target)
4735 == LB_Hebrew_Letter)
4736 {
4737 return FALSE;
4738 }
4739
4740 return LB_table[prev][after] - LB_HY_or_BA_then_foo == LB_BREAKABLE;
4741
4742 case LB_PR_or_PO_then_OP_or_HY + LB_BREAKABLE:
4743 case LB_PR_or_PO_then_OP_or_HY + LB_NOBREAK:
4744
4745 /* LB25a (PR | PO) × ( OP | HY )? NU */
4746 if (advance_one_LB(&temp_pos, strend, utf8_target) == LB_Numeric) {
4747 return FALSE;
4748 }
4749
4750 return LB_table[prev][after] - LB_PR_or_PO_then_OP_or_HY
4751 == LB_BREAKABLE;
4752
4753 case LB_SY_or_IS_then_various + LB_BREAKABLE:
4754 case LB_SY_or_IS_then_various + LB_NOBREAK:
4755 {
4756 /* LB25d NU (SY | IS)* × (NU | SY | IS | CL | CP ) */
4757
4758 LB_enum temp = prev;
4759 do {
4760 temp = backup_one_LB(strbeg, &temp_pos, utf8_target);
4761 }
4762 while (temp == LB_Break_Symbols || temp == LB_Infix_Numeric);
4763 if (temp == LB_Numeric) {
4764 return FALSE;
4765 }
4766
4767 return LB_table[prev][after] - LB_SY_or_IS_then_various
4768 == LB_BREAKABLE;
4769 }
4770
4771 case LB_various_then_PO_or_PR + LB_BREAKABLE:
4772 case LB_various_then_PO_or_PR + LB_NOBREAK:
4773 {
4774 /* LB25e NU (SY | IS)* (CL | CP)? × (PO | PR) */
4775
4776 LB_enum temp = prev;
4777 if (temp == LB_Close_Punctuation || temp == LB_Close_Parenthesis)
4778 {
4779 temp = backup_one_LB(strbeg, &temp_pos, utf8_target);
4780 }
4781 while (temp == LB_Break_Symbols || temp == LB_Infix_Numeric) {
4782 temp = backup_one_LB(strbeg, &temp_pos, utf8_target);
4783 }
4784 if (temp == LB_Numeric) {
4785 return FALSE;
4786 }
4787 return LB_various_then_PO_or_PR;
4788 }
4789
4790 case LB_RI_then_RI + LB_NOBREAK:
4791 case LB_RI_then_RI + LB_BREAKABLE:
4792 {
4793 int RI_count = 1;
4794
4795 /* LB30a Break between two regional indicator symbols if and
4796 * only if there are an even number of regional indicators
4797 * preceding the position of the break.
4798 *
4799 * sot (RI RI)* RI × RI
4800 * [^RI] (RI RI)* RI × RI */
4801
4802 while (backup_one_LB(strbeg,
4803 &temp_pos,
4804 utf8_target) == LB_Regional_Indicator)
4805 {
4806 RI_count++;
4807 }
4808
4809 return RI_count % 2 == 0;
4810 }
4811
4812 default:
4813 break;
4814 }
4815
4816 #ifdef DEBUGGING
4817 Perl_re_printf( aTHX_ "Unhandled LB pair: LB_table[%d, %d] = %d\n",
4818 before, after, LB_table[before][after]);
4819 assert(0);
4820 #endif
4821 return TRUE;
4822 }
4823
4824 STATIC LB_enum
S_advance_one_LB(pTHX_ U8 ** curpos,const U8 * const strend,const bool utf8_target)4825 S_advance_one_LB(pTHX_ U8 ** curpos, const U8 * const strend, const bool utf8_target)
4826 {
4827 dVAR;
4828
4829 LB_enum lb;
4830
4831 PERL_ARGS_ASSERT_ADVANCE_ONE_LB;
4832
4833 if (*curpos >= strend) {
4834 return LB_EDGE;
4835 }
4836
4837 if (utf8_target) {
4838 *curpos += UTF8SKIP(*curpos);
4839 if (*curpos >= strend) {
4840 return LB_EDGE;
4841 }
4842 lb = getLB_VAL_UTF8(*curpos, strend);
4843 }
4844 else {
4845 (*curpos)++;
4846 if (*curpos >= strend) {
4847 return LB_EDGE;
4848 }
4849 lb = getLB_VAL_CP(**curpos);
4850 }
4851
4852 return lb;
4853 }
4854
4855 STATIC LB_enum
S_backup_one_LB(pTHX_ const U8 * const strbeg,U8 ** curpos,const bool utf8_target)4856 S_backup_one_LB(pTHX_ const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
4857 {
4858 dVAR;
4859 LB_enum lb;
4860
4861 PERL_ARGS_ASSERT_BACKUP_ONE_LB;
4862
4863 if (*curpos < strbeg) {
4864 return LB_EDGE;
4865 }
4866
4867 if (utf8_target) {
4868 U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
4869 U8 * prev_prev_char_pos;
4870
4871 if (! prev_char_pos) {
4872 return LB_EDGE;
4873 }
4874
4875 if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos, -1, strbeg))) {
4876 lb = getLB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
4877 *curpos = prev_char_pos;
4878 prev_char_pos = prev_prev_char_pos;
4879 }
4880 else {
4881 *curpos = (U8 *) strbeg;
4882 return LB_EDGE;
4883 }
4884 }
4885 else {
4886 if (*curpos - 2 < strbeg) {
4887 *curpos = (U8 *) strbeg;
4888 return LB_EDGE;
4889 }
4890 (*curpos)--;
4891 lb = getLB_VAL_CP(*(*curpos - 1));
4892 }
4893
4894 return lb;
4895 }
4896
4897 STATIC bool
S_isSB(pTHX_ SB_enum before,SB_enum after,const U8 * const strbeg,const U8 * const curpos,const U8 * const strend,const bool utf8_target)4898 S_isSB(pTHX_ SB_enum before,
4899 SB_enum after,
4900 const U8 * const strbeg,
4901 const U8 * const curpos,
4902 const U8 * const strend,
4903 const bool utf8_target)
4904 {
4905 /* returns a boolean indicating if there is a Sentence Boundary Break
4906 * between the inputs. See http://www.unicode.org/reports/tr29/ */
4907
4908 U8 * lpos = (U8 *) curpos;
4909 bool has_para_sep = FALSE;
4910 bool has_sp = FALSE;
4911
4912 PERL_ARGS_ASSERT_ISSB;
4913
4914 /* Break at the start and end of text.
4915 SB1. sot ÷
4916 SB2. ÷ eot
4917 But unstated in Unicode is don't break if the text is empty */
4918 if (before == SB_EDGE || after == SB_EDGE) {
4919 return before != after;
4920 }
4921
4922 /* SB 3: Do not break within CRLF. */
4923 if (before == SB_CR && after == SB_LF) {
4924 return FALSE;
4925 }
4926
4927 /* Break after paragraph separators. CR and LF are considered
4928 * so because Unicode views text as like word processing text where there
4929 * are no newlines except between paragraphs, and the word processor takes
4930 * care of wrapping without there being hard line-breaks in the text *./
4931 SB4. Sep | CR | LF ÷ */
4932 if (before == SB_Sep || before == SB_CR || before == SB_LF) {
4933 return TRUE;
4934 }
4935
4936 /* Ignore Format and Extend characters, except after sot, Sep, CR, or LF.
4937 * (See Section 6.2, Replacing Ignore Rules.)
4938 SB5. X (Extend | Format)* → X */
4939 if (after == SB_Extend || after == SB_Format) {
4940
4941 /* Implied is that the these characters attach to everything
4942 * immediately prior to them except for those separator-type
4943 * characters. And the rules earlier have already handled the case
4944 * when one of those immediately precedes the extend char */
4945 return FALSE;
4946 }
4947
4948 if (before == SB_Extend || before == SB_Format) {
4949 U8 * temp_pos = lpos;
4950 const SB_enum backup = backup_one_SB(strbeg, &temp_pos, utf8_target);
4951 if ( backup != SB_EDGE
4952 && backup != SB_Sep
4953 && backup != SB_CR
4954 && backup != SB_LF)
4955 {
4956 before = backup;
4957 lpos = temp_pos;
4958 }
4959
4960 /* Here, both 'before' and 'backup' are these types; implied is that we
4961 * don't break between them */
4962 if (backup == SB_Extend || backup == SB_Format) {
4963 return FALSE;
4964 }
4965 }
4966
4967 /* Do not break after ambiguous terminators like period, if they are
4968 * immediately followed by a number or lowercase letter, if they are
4969 * between uppercase letters, if the first following letter (optionally
4970 * after certain punctuation) is lowercase, or if they are followed by
4971 * "continuation" punctuation such as comma, colon, or semicolon. For
4972 * example, a period may be an abbreviation or numeric period, and thus may
4973 * not mark the end of a sentence.
4974
4975 * SB6. ATerm × Numeric */
4976 if (before == SB_ATerm && after == SB_Numeric) {
4977 return FALSE;
4978 }
4979
4980 /* SB7. (Upper | Lower) ATerm × Upper */
4981 if (before == SB_ATerm && after == SB_Upper) {
4982 U8 * temp_pos = lpos;
4983 SB_enum backup = backup_one_SB(strbeg, &temp_pos, utf8_target);
4984 if (backup == SB_Upper || backup == SB_Lower) {
4985 return FALSE;
4986 }
4987 }
4988
4989 /* The remaining rules that aren't the final one, all require an STerm or
4990 * an ATerm after having backed up over some Close* Sp*, and in one case an
4991 * optional Paragraph separator, although one rule doesn't have any Sp's in it.
4992 * So do that backup now, setting flags if either Sp or a paragraph
4993 * separator are found */
4994
4995 if (before == SB_Sep || before == SB_CR || before == SB_LF) {
4996 has_para_sep = TRUE;
4997 before = backup_one_SB(strbeg, &lpos, utf8_target);
4998 }
4999
5000 if (before == SB_Sp) {
5001 has_sp = TRUE;
5002 do {
5003 before = backup_one_SB(strbeg, &lpos, utf8_target);
5004 }
5005 while (before == SB_Sp);
5006 }
5007
5008 while (before == SB_Close) {
5009 before = backup_one_SB(strbeg, &lpos, utf8_target);
5010 }
5011
5012 /* The next few rules apply only when the backed-up-to is an ATerm, and in
5013 * most cases an STerm */
5014 if (before == SB_STerm || before == SB_ATerm) {
5015
5016 /* So, here the lhs matches
5017 * (STerm | ATerm) Close* Sp* (Sep | CR | LF)?
5018 * and we have set flags if we found an Sp, or the optional Sep,CR,LF.
5019 * The rules that apply here are:
5020 *
5021 * SB8 ATerm Close* Sp* × ( ¬(OLetter | Upper | Lower | Sep | CR
5022 | LF | STerm | ATerm) )* Lower
5023 SB8a (STerm | ATerm) Close* Sp* × (SContinue | STerm | ATerm)
5024 SB9 (STerm | ATerm) Close* × (Close | Sp | Sep | CR | LF)
5025 SB10 (STerm | ATerm) Close* Sp* × (Sp | Sep | CR | LF)
5026 SB11 (STerm | ATerm) Close* Sp* (Sep | CR | LF)? ÷
5027 */
5028
5029 /* And all but SB11 forbid having seen a paragraph separator */
5030 if (! has_para_sep) {
5031 if (before == SB_ATerm) { /* SB8 */
5032 U8 * rpos = (U8 *) curpos;
5033 SB_enum later = after;
5034
5035 while ( later != SB_OLetter
5036 && later != SB_Upper
5037 && later != SB_Lower
5038 && later != SB_Sep
5039 && later != SB_CR
5040 && later != SB_LF
5041 && later != SB_STerm
5042 && later != SB_ATerm
5043 && later != SB_EDGE)
5044 {
5045 later = advance_one_SB(&rpos, strend, utf8_target);
5046 }
5047 if (later == SB_Lower) {
5048 return FALSE;
5049 }
5050 }
5051
5052 if ( after == SB_SContinue /* SB8a */
5053 || after == SB_STerm
5054 || after == SB_ATerm)
5055 {
5056 return FALSE;
5057 }
5058
5059 if (! has_sp) { /* SB9 applies only if there was no Sp* */
5060 if ( after == SB_Close
5061 || after == SB_Sp
5062 || after == SB_Sep
5063 || after == SB_CR
5064 || after == SB_LF)
5065 {
5066 return FALSE;
5067 }
5068 }
5069
5070 /* SB10. This and SB9 could probably be combined some way, but khw
5071 * has decided to follow the Unicode rule book precisely for
5072 * simplified maintenance */
5073 if ( after == SB_Sp
5074 || after == SB_Sep
5075 || after == SB_CR
5076 || after == SB_LF)
5077 {
5078 return FALSE;
5079 }
5080 }
5081
5082 /* SB11. */
5083 return TRUE;
5084 }
5085
5086 /* Otherwise, do not break.
5087 SB12. Any × Any */
5088
5089 return FALSE;
5090 }
5091
5092 STATIC SB_enum
S_advance_one_SB(pTHX_ U8 ** curpos,const U8 * const strend,const bool utf8_target)5093 S_advance_one_SB(pTHX_ U8 ** curpos, const U8 * const strend, const bool utf8_target)
5094 {
5095 dVAR;
5096 SB_enum sb;
5097
5098 PERL_ARGS_ASSERT_ADVANCE_ONE_SB;
5099
5100 if (*curpos >= strend) {
5101 return SB_EDGE;
5102 }
5103
5104 if (utf8_target) {
5105 do {
5106 *curpos += UTF8SKIP(*curpos);
5107 if (*curpos >= strend) {
5108 return SB_EDGE;
5109 }
5110 sb = getSB_VAL_UTF8(*curpos, strend);
5111 } while (sb == SB_Extend || sb == SB_Format);
5112 }
5113 else {
5114 do {
5115 (*curpos)++;
5116 if (*curpos >= strend) {
5117 return SB_EDGE;
5118 }
5119 sb = getSB_VAL_CP(**curpos);
5120 } while (sb == SB_Extend || sb == SB_Format);
5121 }
5122
5123 return sb;
5124 }
5125
5126 STATIC SB_enum
S_backup_one_SB(pTHX_ const U8 * const strbeg,U8 ** curpos,const bool utf8_target)5127 S_backup_one_SB(pTHX_ const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
5128 {
5129 dVAR;
5130 SB_enum sb;
5131
5132 PERL_ARGS_ASSERT_BACKUP_ONE_SB;
5133
5134 if (*curpos < strbeg) {
5135 return SB_EDGE;
5136 }
5137
5138 if (utf8_target) {
5139 U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
5140 if (! prev_char_pos) {
5141 return SB_EDGE;
5142 }
5143
5144 /* Back up over Extend and Format. curpos is always just to the right
5145 * of the characater whose value we are getting */
5146 do {
5147 U8 * prev_prev_char_pos;
5148 if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos, -1,
5149 strbeg)))
5150 {
5151 sb = getSB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
5152 *curpos = prev_char_pos;
5153 prev_char_pos = prev_prev_char_pos;
5154 }
5155 else {
5156 *curpos = (U8 *) strbeg;
5157 return SB_EDGE;
5158 }
5159 } while (sb == SB_Extend || sb == SB_Format);
5160 }
5161 else {
5162 do {
5163 if (*curpos - 2 < strbeg) {
5164 *curpos = (U8 *) strbeg;
5165 return SB_EDGE;
5166 }
5167 (*curpos)--;
5168 sb = getSB_VAL_CP(*(*curpos - 1));
5169 } while (sb == SB_Extend || sb == SB_Format);
5170 }
5171
5172 return sb;
5173 }
5174
5175 STATIC bool
S_isWB(pTHX_ WB_enum previous,WB_enum before,WB_enum after,const U8 * const strbeg,const U8 * const curpos,const U8 * const strend,const bool utf8_target)5176 S_isWB(pTHX_ WB_enum previous,
5177 WB_enum before,
5178 WB_enum after,
5179 const U8 * const strbeg,
5180 const U8 * const curpos,
5181 const U8 * const strend,
5182 const bool utf8_target)
5183 {
5184 /* Return a boolean as to if the boundary between 'before' and 'after' is
5185 * a Unicode word break, using their published algorithm, but tailored for
5186 * Perl by treating spans of white space as one unit. Context may be
5187 * needed to make this determination. If the value for the character
5188 * before 'before' is known, it is passed as 'previous'; otherwise that
5189 * should be set to WB_UNKNOWN. The other input parameters give the
5190 * boundaries and current position in the matching of the string. That
5191 * is, 'curpos' marks the position where the character whose wb value is
5192 * 'after' begins. See http://www.unicode.org/reports/tr29/ */
5193
5194 U8 * before_pos = (U8 *) curpos;
5195 U8 * after_pos = (U8 *) curpos;
5196 WB_enum prev = before;
5197 WB_enum next;
5198
5199 PERL_ARGS_ASSERT_ISWB;
5200
5201 /* Rule numbers in the comments below are as of Unicode 9.0 */
5202
5203 redo:
5204 before = prev;
5205 switch (WB_table[before][after]) {
5206 case WB_BREAKABLE:
5207 return TRUE;
5208
5209 case WB_NOBREAK:
5210 return FALSE;
5211
5212 case WB_hs_then_hs: /* 2 horizontal spaces in a row */
5213 next = advance_one_WB(&after_pos, strend, utf8_target,
5214 FALSE /* Don't skip Extend nor Format */ );
5215 /* A space immediately preceeding an Extend or Format is attached
5216 * to by them, and hence gets separated from previous spaces.
5217 * Otherwise don't break between horizontal white space */
5218 return next == WB_Extend || next == WB_Format;
5219
5220 /* WB4 Ignore Format and Extend characters, except when they appear at
5221 * the beginning of a region of text. This code currently isn't
5222 * general purpose, but it works as the rules are currently and likely
5223 * to be laid out. The reason it works is that when 'they appear at
5224 * the beginning of a region of text', the rule is to break before
5225 * them, just like any other character. Therefore, the default rule
5226 * applies and we don't have to look in more depth. Should this ever
5227 * change, we would have to have 2 'case' statements, like in the rules
5228 * below, and backup a single character (not spacing over the extend
5229 * ones) and then see if that is one of the region-end characters and
5230 * go from there */
5231 case WB_Ex_or_FO_or_ZWJ_then_foo:
5232 prev = backup_one_WB(&previous, strbeg, &before_pos, utf8_target);
5233 goto redo;
5234
5235 case WB_DQ_then_HL + WB_BREAKABLE:
5236 case WB_DQ_then_HL + WB_NOBREAK:
5237
5238 /* WB7c Hebrew_Letter Double_Quote × Hebrew_Letter */
5239
5240 if (backup_one_WB(&previous, strbeg, &before_pos, utf8_target)
5241 == WB_Hebrew_Letter)
5242 {
5243 return FALSE;
5244 }
5245
5246 return WB_table[before][after] - WB_DQ_then_HL == WB_BREAKABLE;
5247
5248 case WB_HL_then_DQ + WB_BREAKABLE:
5249 case WB_HL_then_DQ + WB_NOBREAK:
5250
5251 /* WB7b Hebrew_Letter × Double_Quote Hebrew_Letter */
5252
5253 if (advance_one_WB(&after_pos, strend, utf8_target,
5254 TRUE /* Do skip Extend and Format */ )
5255 == WB_Hebrew_Letter)
5256 {
5257 return FALSE;
5258 }
5259
5260 return WB_table[before][after] - WB_HL_then_DQ == WB_BREAKABLE;
5261
5262 case WB_LE_or_HL_then_MB_or_ML_or_SQ + WB_NOBREAK:
5263 case WB_LE_or_HL_then_MB_or_ML_or_SQ + WB_BREAKABLE:
5264
5265 /* WB6 (ALetter | Hebrew_Letter) × (MidLetter | MidNumLet
5266 * | Single_Quote) (ALetter | Hebrew_Letter) */
5267
5268 next = advance_one_WB(&after_pos, strend, utf8_target,
5269 TRUE /* Do skip Extend and Format */ );
5270
5271 if (next == WB_ALetter || next == WB_Hebrew_Letter)
5272 {
5273 return FALSE;
5274 }
5275
5276 return WB_table[before][after]
5277 - WB_LE_or_HL_then_MB_or_ML_or_SQ == WB_BREAKABLE;
5278
5279 case WB_MB_or_ML_or_SQ_then_LE_or_HL + WB_NOBREAK:
5280 case WB_MB_or_ML_or_SQ_then_LE_or_HL + WB_BREAKABLE:
5281
5282 /* WB7 (ALetter | Hebrew_Letter) (MidLetter | MidNumLet
5283 * | Single_Quote) × (ALetter | Hebrew_Letter) */
5284
5285 prev = backup_one_WB(&previous, strbeg, &before_pos, utf8_target);
5286 if (prev == WB_ALetter || prev == WB_Hebrew_Letter)
5287 {
5288 return FALSE;
5289 }
5290
5291 return WB_table[before][after]
5292 - WB_MB_or_ML_or_SQ_then_LE_or_HL == WB_BREAKABLE;
5293
5294 case WB_MB_or_MN_or_SQ_then_NU + WB_NOBREAK:
5295 case WB_MB_or_MN_or_SQ_then_NU + WB_BREAKABLE:
5296
5297 /* WB11 Numeric (MidNum | (MidNumLet | Single_Quote)) × Numeric
5298 * */
5299
5300 if (backup_one_WB(&previous, strbeg, &before_pos, utf8_target)
5301 == WB_Numeric)
5302 {
5303 return FALSE;
5304 }
5305
5306 return WB_table[before][after]
5307 - WB_MB_or_MN_or_SQ_then_NU == WB_BREAKABLE;
5308
5309 case WB_NU_then_MB_or_MN_or_SQ + WB_NOBREAK:
5310 case WB_NU_then_MB_or_MN_or_SQ + WB_BREAKABLE:
5311
5312 /* WB12 Numeric × (MidNum | MidNumLet | Single_Quote) Numeric */
5313
5314 if (advance_one_WB(&after_pos, strend, utf8_target,
5315 TRUE /* Do skip Extend and Format */ )
5316 == WB_Numeric)
5317 {
5318 return FALSE;
5319 }
5320
5321 return WB_table[before][after]
5322 - WB_NU_then_MB_or_MN_or_SQ == WB_BREAKABLE;
5323
5324 case WB_RI_then_RI + WB_NOBREAK:
5325 case WB_RI_then_RI + WB_BREAKABLE:
5326 {
5327 int RI_count = 1;
5328
5329 /* Do not break within emoji flag sequences. That is, do not
5330 * break between regional indicator (RI) symbols if there is an
5331 * odd number of RI characters before the potential break
5332 * point.
5333 *
5334 * WB15 sot (RI RI)* RI × RI
5335 * WB16 [^RI] (RI RI)* RI × RI */
5336
5337 while (backup_one_WB(&previous,
5338 strbeg,
5339 &before_pos,
5340 utf8_target) == WB_Regional_Indicator)
5341 {
5342 RI_count++;
5343 }
5344
5345 return RI_count % 2 != 1;
5346 }
5347
5348 default:
5349 break;
5350 }
5351
5352 #ifdef DEBUGGING
5353 Perl_re_printf( aTHX_ "Unhandled WB pair: WB_table[%d, %d] = %d\n",
5354 before, after, WB_table[before][after]);
5355 assert(0);
5356 #endif
5357 return TRUE;
5358 }
5359
5360 STATIC WB_enum
S_advance_one_WB(pTHX_ U8 ** curpos,const U8 * const strend,const bool utf8_target,const bool skip_Extend_Format)5361 S_advance_one_WB(pTHX_ U8 ** curpos,
5362 const U8 * const strend,
5363 const bool utf8_target,
5364 const bool skip_Extend_Format)
5365 {
5366 dVAR;
5367 WB_enum wb;
5368
5369 PERL_ARGS_ASSERT_ADVANCE_ONE_WB;
5370
5371 if (*curpos >= strend) {
5372 return WB_EDGE;
5373 }
5374
5375 if (utf8_target) {
5376
5377 /* Advance over Extend and Format */
5378 do {
5379 *curpos += UTF8SKIP(*curpos);
5380 if (*curpos >= strend) {
5381 return WB_EDGE;
5382 }
5383 wb = getWB_VAL_UTF8(*curpos, strend);
5384 } while ( skip_Extend_Format
5385 && (wb == WB_Extend || wb == WB_Format));
5386 }
5387 else {
5388 do {
5389 (*curpos)++;
5390 if (*curpos >= strend) {
5391 return WB_EDGE;
5392 }
5393 wb = getWB_VAL_CP(**curpos);
5394 } while ( skip_Extend_Format
5395 && (wb == WB_Extend || wb == WB_Format));
5396 }
5397
5398 return wb;
5399 }
5400
5401 STATIC WB_enum
S_backup_one_WB(pTHX_ WB_enum * previous,const U8 * const strbeg,U8 ** curpos,const bool utf8_target)5402 S_backup_one_WB(pTHX_ WB_enum * previous, const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
5403 {
5404 dVAR;
5405 WB_enum wb;
5406
5407 PERL_ARGS_ASSERT_BACKUP_ONE_WB;
5408
5409 /* If we know what the previous character's break value is, don't have
5410 * to look it up */
5411 if (*previous != WB_UNKNOWN) {
5412 wb = *previous;
5413
5414 /* But we need to move backwards by one */
5415 if (utf8_target) {
5416 *curpos = reghopmaybe3(*curpos, -1, strbeg);
5417 if (! *curpos) {
5418 *previous = WB_EDGE;
5419 *curpos = (U8 *) strbeg;
5420 }
5421 else {
5422 *previous = WB_UNKNOWN;
5423 }
5424 }
5425 else {
5426 (*curpos)--;
5427 *previous = (*curpos <= strbeg) ? WB_EDGE : WB_UNKNOWN;
5428 }
5429
5430 /* And we always back up over these three types */
5431 if (wb != WB_Extend && wb != WB_Format && wb != WB_ZWJ) {
5432 return wb;
5433 }
5434 }
5435
5436 if (*curpos < strbeg) {
5437 return WB_EDGE;
5438 }
5439
5440 if (utf8_target) {
5441 U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
5442 if (! prev_char_pos) {
5443 return WB_EDGE;
5444 }
5445
5446 /* Back up over Extend and Format. curpos is always just to the right
5447 * of the characater whose value we are getting */
5448 do {
5449 U8 * prev_prev_char_pos;
5450 if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos,
5451 -1,
5452 strbeg)))
5453 {
5454 wb = getWB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
5455 *curpos = prev_char_pos;
5456 prev_char_pos = prev_prev_char_pos;
5457 }
5458 else {
5459 *curpos = (U8 *) strbeg;
5460 return WB_EDGE;
5461 }
5462 } while (wb == WB_Extend || wb == WB_Format || wb == WB_ZWJ);
5463 }
5464 else {
5465 do {
5466 if (*curpos - 2 < strbeg) {
5467 *curpos = (U8 *) strbeg;
5468 return WB_EDGE;
5469 }
5470 (*curpos)--;
5471 wb = getWB_VAL_CP(*(*curpos - 1));
5472 } while (wb == WB_Extend || wb == WB_Format);
5473 }
5474
5475 return wb;
5476 }
5477
5478 /* Macros for regmatch(), using its internal variables */
5479 #define NEXTCHR_EOS -10 /* nextchr has fallen off the end */
5480 #define NEXTCHR_IS_EOS (nextchr < 0)
5481
5482 #define SET_nextchr \
5483 nextchr = ((locinput < reginfo->strend) ? UCHARAT(locinput) : NEXTCHR_EOS)
5484
5485 #define SET_locinput(p) \
5486 locinput = (p); \
5487 SET_nextchr
5488
5489 #define sayYES goto yes
5490 #define sayNO goto no
5491 #define sayNO_SILENT goto no_silent
5492
5493 /* we dont use STMT_START/END here because it leads to
5494 "unreachable code" warnings, which are bogus, but distracting. */
5495 #define CACHEsayNO \
5496 if (ST.cache_mask) \
5497 reginfo->info_aux->poscache[ST.cache_offset] |= ST.cache_mask; \
5498 sayNO
5499
5500 #define EVAL_CLOSE_PAREN_IS(st,expr) \
5501 ( \
5502 ( ( st ) ) && \
5503 ( ( st )->u.eval.close_paren ) && \
5504 ( ( ( st )->u.eval.close_paren ) == ( (expr) + 1 ) ) \
5505 )
5506
5507 #define EVAL_CLOSE_PAREN_IS_TRUE(st,expr) \
5508 ( \
5509 ( ( st ) ) && \
5510 ( ( st )->u.eval.close_paren ) && \
5511 ( ( expr ) ) && \
5512 ( ( ( st )->u.eval.close_paren ) == ( (expr) + 1 ) ) \
5513 )
5514
5515
5516 #define EVAL_CLOSE_PAREN_SET(st,expr) \
5517 (st)->u.eval.close_paren = ( (expr) + 1 )
5518
5519 #define EVAL_CLOSE_PAREN_CLEAR(st) \
5520 (st)->u.eval.close_paren = 0
5521
5522 /* push a new state then goto it */
5523
5524 #define PUSH_STATE_GOTO(state, node, input, eol, sr0) \
5525 pushinput = input; \
5526 pusheol = eol; \
5527 pushsr0 = sr0; \
5528 scan = node; \
5529 st->resume_state = state; \
5530 goto push_state;
5531
5532 /* push a new state with success backtracking, then goto it */
5533
5534 #define PUSH_YES_STATE_GOTO(state, node, input, eol, sr0) \
5535 pushinput = input; \
5536 pusheol = eol; \
5537 pushsr0 = sr0; \
5538 scan = node; \
5539 st->resume_state = state; \
5540 goto push_yes_state;
5541
5542 #define DEBUG_STATE_pp(pp) \
5543 DEBUG_STATE_r({ \
5544 DUMP_EXEC_POS(locinput, scan, utf8_target,depth); \
5545 Perl_re_printf( aTHX_ \
5546 "%*s" pp " %s%s%s%s%s\n", \
5547 INDENT_CHARS(depth), "", \
5548 PL_reg_name[st->resume_state], \
5549 ((st==yes_state||st==mark_state) ? "[" : ""), \
5550 ((st==yes_state) ? "Y" : ""), \
5551 ((st==mark_state) ? "M" : ""), \
5552 ((st==yes_state||st==mark_state) ? "]" : "") \
5553 ); \
5554 });
5555
5556 /*
5557
5558 regmatch() - main matching routine
5559
5560 This is basically one big switch statement in a loop. We execute an op,
5561 set 'next' to point the next op, and continue. If we come to a point which
5562 we may need to backtrack to on failure such as (A|B|C), we push a
5563 backtrack state onto the backtrack stack. On failure, we pop the top
5564 state, and re-enter the loop at the state indicated. If there are no more
5565 states to pop, we return failure.
5566
5567 Sometimes we also need to backtrack on success; for example /A+/, where
5568 after successfully matching one A, we need to go back and try to
5569 match another one; similarly for lookahead assertions: if the assertion
5570 completes successfully, we backtrack to the state just before the assertion
5571 and then carry on. In these cases, the pushed state is marked as
5572 'backtrack on success too'. This marking is in fact done by a chain of
5573 pointers, each pointing to the previous 'yes' state. On success, we pop to
5574 the nearest yes state, discarding any intermediate failure-only states.
5575 Sometimes a yes state is pushed just to force some cleanup code to be
5576 called at the end of a successful match or submatch; e.g. (??{$re}) uses
5577 it to free the inner regex.
5578
5579 Note that failure backtracking rewinds the cursor position, while
5580 success backtracking leaves it alone.
5581
5582 A pattern is complete when the END op is executed, while a subpattern
5583 such as (?=foo) is complete when the SUCCESS op is executed. Both of these
5584 ops trigger the "pop to last yes state if any, otherwise return true"
5585 behaviour.
5586
5587 A common convention in this function is to use A and B to refer to the two
5588 subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
5589 the subpattern to be matched possibly multiple times, while B is the entire
5590 rest of the pattern. Variable and state names reflect this convention.
5591
5592 The states in the main switch are the union of ops and failure/success of
5593 substates associated with with that op. For example, IFMATCH is the op
5594 that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
5595 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
5596 successfully matched A and IFMATCH_A_fail is a state saying that we have
5597 just failed to match A. Resume states always come in pairs. The backtrack
5598 state we push is marked as 'IFMATCH_A', but when that is popped, we resume
5599 at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
5600 on success or failure.
5601
5602 The struct that holds a backtracking state is actually a big union, with
5603 one variant for each major type of op. The variable st points to the
5604 top-most backtrack struct. To make the code clearer, within each
5605 block of code we #define ST to alias the relevant union.
5606
5607 Here's a concrete example of a (vastly oversimplified) IFMATCH
5608 implementation:
5609
5610 switch (state) {
5611 ....
5612
5613 #define ST st->u.ifmatch
5614
5615 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
5616 ST.foo = ...; // some state we wish to save
5617 ...
5618 // push a yes backtrack state with a resume value of
5619 // IFMATCH_A/IFMATCH_A_fail, then continue execution at the
5620 // first node of A:
5621 PUSH_YES_STATE_GOTO(IFMATCH_A, A, newinput);
5622 // NOTREACHED
5623
5624 case IFMATCH_A: // we have successfully executed A; now continue with B
5625 next = B;
5626 bar = ST.foo; // do something with the preserved value
5627 break;
5628
5629 case IFMATCH_A_fail: // A failed, so the assertion failed
5630 ...; // do some housekeeping, then ...
5631 sayNO; // propagate the failure
5632
5633 #undef ST
5634
5635 ...
5636 }
5637
5638 For any old-timers reading this who are familiar with the old recursive
5639 approach, the code above is equivalent to:
5640
5641 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
5642 {
5643 int foo = ...
5644 ...
5645 if (regmatch(A)) {
5646 next = B;
5647 bar = foo;
5648 break;
5649 }
5650 ...; // do some housekeeping, then ...
5651 sayNO; // propagate the failure
5652 }
5653
5654 The topmost backtrack state, pointed to by st, is usually free. If you
5655 want to claim it, populate any ST.foo fields in it with values you wish to
5656 save, then do one of
5657
5658 PUSH_STATE_GOTO(resume_state, node, newinput, new_eol);
5659 PUSH_YES_STATE_GOTO(resume_state, node, newinput, new_eol);
5660
5661 which sets that backtrack state's resume value to 'resume_state', pushes a
5662 new free entry to the top of the backtrack stack, then goes to 'node'.
5663 On backtracking, the free slot is popped, and the saved state becomes the
5664 new free state. An ST.foo field in this new top state can be temporarily
5665 accessed to retrieve values, but once the main loop is re-entered, it
5666 becomes available for reuse.
5667
5668 Note that the depth of the backtrack stack constantly increases during the
5669 left-to-right execution of the pattern, rather than going up and down with
5670 the pattern nesting. For example the stack is at its maximum at Z at the
5671 end of the pattern, rather than at X in the following:
5672
5673 /(((X)+)+)+....(Y)+....Z/
5674
5675 The only exceptions to this are lookahead/behind assertions and the cut,
5676 (?>A), which pop all the backtrack states associated with A before
5677 continuing.
5678
5679 Backtrack state structs are allocated in slabs of about 4K in size.
5680 PL_regmatch_state and st always point to the currently active state,
5681 and PL_regmatch_slab points to the slab currently containing
5682 PL_regmatch_state. The first time regmatch() is called, the first slab is
5683 allocated, and is never freed until interpreter destruction. When the slab
5684 is full, a new one is allocated and chained to the end. At exit from
5685 regmatch(), slabs allocated since entry are freed.
5686
5687 In order to work with variable length lookbehinds, an upper limit is placed on
5688 lookbehinds which is set to where the match position is at the end of where the
5689 lookbehind would get to. Nothing in the lookbehind should match above that,
5690 except we should be able to look beyond if for things like \b, which need the
5691 next character in the string to be able to determine if this is a boundary or
5692 not. We also can't match the end of string/line unless we are also at the end
5693 of the entire string, so NEXTCHR_IS_EOS remains the same, and for those OPs
5694 that match a width, we have to add a condition that they are within the legal
5695 bounds of our window into the string.
5696
5697 */
5698
5699 /* returns -1 on failure, $+[0] on success */
5700 STATIC SSize_t
S_regmatch(pTHX_ regmatch_info * reginfo,char * startpos,regnode * prog)5701 S_regmatch(pTHX_ regmatch_info *reginfo, char *startpos, regnode *prog)
5702 {
5703 dVAR;
5704 const bool utf8_target = reginfo->is_utf8_target;
5705 const U32 uniflags = UTF8_ALLOW_DEFAULT;
5706 REGEXP *rex_sv = reginfo->prog;
5707 regexp *rex = ReANY(rex_sv);
5708 RXi_GET_DECL(rex,rexi);
5709 /* the current state. This is a cached copy of PL_regmatch_state */
5710 regmatch_state *st;
5711 /* cache heavy used fields of st in registers */
5712 regnode *scan;
5713 regnode *next;
5714 U32 n = 0; /* general value; init to avoid compiler warning */
5715 SSize_t ln = 0; /* len or last; init to avoid compiler warning */
5716 SSize_t endref = 0; /* offset of end of backref when ln is start */
5717 char *locinput = startpos;
5718 char *loceol = reginfo->strend;
5719 char *pushinput; /* where to continue after a PUSH */
5720 char *pusheol; /* where to stop matching (loceol) after a PUSH */
5721 U8 *pushsr0; /* save starting pos of script run */
5722 I32 nextchr; /* is always set to UCHARAT(locinput), or -1 at EOS */
5723
5724 bool result = 0; /* return value of S_regmatch */
5725 U32 depth = 0; /* depth of backtrack stack */
5726 U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
5727 const U32 max_nochange_depth =
5728 (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
5729 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
5730 regmatch_state *yes_state = NULL; /* state to pop to on success of
5731 subpattern */
5732 /* mark_state piggy backs on the yes_state logic so that when we unwind
5733 the stack on success we can update the mark_state as we go */
5734 regmatch_state *mark_state = NULL; /* last mark state we have seen */
5735 regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
5736 struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */
5737 U32 state_num;
5738 bool no_final = 0; /* prevent failure from backtracking? */
5739 bool do_cutgroup = 0; /* no_final only until next branch/trie entry */
5740 char *startpoint = locinput;
5741 SV *popmark = NULL; /* are we looking for a mark? */
5742 SV *sv_commit = NULL; /* last mark name seen in failure */
5743 SV *sv_yes_mark = NULL; /* last mark name we have seen
5744 during a successful match */
5745 U32 lastopen = 0; /* last open we saw */
5746 bool has_cutgroup = RXp_HAS_CUTGROUP(rex) ? 1 : 0;
5747 SV* const oreplsv = GvSVn(PL_replgv);
5748 /* these three flags are set by various ops to signal information to
5749 * the very next op. They have a useful lifetime of exactly one loop
5750 * iteration, and are not preserved or restored by state pushes/pops
5751 */
5752 bool sw = 0; /* the condition value in (?(cond)a|b) */
5753 bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */
5754 int logical = 0; /* the following EVAL is:
5755 0: (?{...})
5756 1: (?(?{...})X|Y)
5757 2: (??{...})
5758 or the following IFMATCH/UNLESSM is:
5759 false: plain (?=foo)
5760 true: used as a condition: (?(?=foo))
5761 */
5762 PAD* last_pad = NULL;
5763 dMULTICALL;
5764 U8 gimme = G_SCALAR;
5765 CV *caller_cv = NULL; /* who called us */
5766 CV *last_pushed_cv = NULL; /* most recently called (?{}) CV */
5767 U32 maxopenparen = 0; /* max '(' index seen so far */
5768 int to_complement; /* Invert the result? */
5769 _char_class_number classnum;
5770 bool is_utf8_pat = reginfo->is_utf8_pat;
5771 bool match = FALSE;
5772 I32 orig_savestack_ix = PL_savestack_ix;
5773 U8 * script_run_begin = NULL;
5774
5775 /* Solaris Studio 12.3 messes up fetching PL_charclass['\n'] */
5776 #if (defined(__SUNPRO_C) && (__SUNPRO_C == 0x5120) && defined(__x86_64) && defined(USE_64_BIT_ALL))
5777 # define SOLARIS_BAD_OPTIMIZER
5778 const U32 *pl_charclass_dup = PL_charclass;
5779 # define PL_charclass pl_charclass_dup
5780 #endif
5781
5782 #ifdef DEBUGGING
5783 GET_RE_DEBUG_FLAGS_DECL;
5784 #endif
5785
5786 /* protect against undef(*^R) */
5787 SAVEFREESV(SvREFCNT_inc_simple_NN(oreplsv));
5788
5789 /* shut up 'may be used uninitialized' compiler warnings for dMULTICALL */
5790 multicall_oldcatch = 0;
5791 PERL_UNUSED_VAR(multicall_cop);
5792
5793 PERL_ARGS_ASSERT_REGMATCH;
5794
5795 st = PL_regmatch_state;
5796
5797 /* Note that nextchr is a byte even in UTF */
5798 SET_nextchr;
5799 scan = prog;
5800
5801 DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
5802 DUMP_EXEC_POS( locinput, scan, utf8_target, depth );
5803 Perl_re_printf( aTHX_ "regmatch start\n" );
5804 }));
5805
5806 while (scan != NULL) {
5807 next = scan + NEXT_OFF(scan);
5808 if (next == scan)
5809 next = NULL;
5810 state_num = OP(scan);
5811
5812 reenter_switch:
5813 DEBUG_EXECUTE_r(
5814 if (state_num <= REGNODE_MAX) {
5815 SV * const prop = sv_newmortal();
5816 regnode *rnext = regnext(scan);
5817
5818 DUMP_EXEC_POS( locinput, scan, utf8_target, depth );
5819 regprop(rex, prop, scan, reginfo, NULL);
5820 Perl_re_printf( aTHX_
5821 "%*s%" IVdf ":%s(%" IVdf ")\n",
5822 INDENT_CHARS(depth), "",
5823 (IV)(scan - rexi->program),
5824 SvPVX_const(prop),
5825 (PL_regkind[OP(scan)] == END || !rnext) ?
5826 0 : (IV)(rnext - rexi->program));
5827 }
5828 );
5829
5830 to_complement = 0;
5831
5832 SET_nextchr;
5833 assert(nextchr < 256 && (nextchr >= 0 || nextchr == NEXTCHR_EOS));
5834
5835 switch (state_num) {
5836 case SBOL: /* /^../ and /\A../ */
5837 if (locinput == reginfo->strbeg)
5838 break;
5839 sayNO;
5840
5841 case MBOL: /* /^../m */
5842 if (locinput == reginfo->strbeg ||
5843 (!NEXTCHR_IS_EOS && locinput[-1] == '\n'))
5844 {
5845 break;
5846 }
5847 sayNO;
5848
5849 case GPOS: /* \G */
5850 if (locinput == reginfo->ganch)
5851 break;
5852 sayNO;
5853
5854 case KEEPS: /* \K */
5855 /* update the startpoint */
5856 st->u.keeper.val = rex->offs[0].start;
5857 rex->offs[0].start = locinput - reginfo->strbeg;
5858 PUSH_STATE_GOTO(KEEPS_next, next, locinput, loceol,
5859 script_run_begin);
5860 NOT_REACHED; /* NOTREACHED */
5861
5862 case KEEPS_next_fail:
5863 /* rollback the start point change */
5864 rex->offs[0].start = st->u.keeper.val;
5865 sayNO_SILENT;
5866 NOT_REACHED; /* NOTREACHED */
5867
5868 case MEOL: /* /..$/m */
5869 if (!NEXTCHR_IS_EOS && nextchr != '\n')
5870 sayNO;
5871 break;
5872
5873 case SEOL: /* /..$/ */
5874 if (!NEXTCHR_IS_EOS && nextchr != '\n')
5875 sayNO;
5876 if (reginfo->strend - locinput > 1)
5877 sayNO;
5878 break;
5879
5880 case EOS: /* \z */
5881 if (!NEXTCHR_IS_EOS)
5882 sayNO;
5883 break;
5884
5885 case SANY: /* /./s */
5886 if (NEXTCHR_IS_EOS || locinput >= loceol)
5887 sayNO;
5888 goto increment_locinput;
5889
5890 case REG_ANY: /* /./ */
5891 if ( NEXTCHR_IS_EOS
5892 || locinput >= loceol
5893 || nextchr == '\n')
5894 {
5895 sayNO;
5896 }
5897 goto increment_locinput;
5898
5899
5900 #undef ST
5901 #define ST st->u.trie
5902 case TRIEC: /* (ab|cd) with known charclass */
5903 /* In this case the charclass data is available inline so
5904 we can fail fast without a lot of extra overhead.
5905 */
5906 if ( ! NEXTCHR_IS_EOS
5907 && locinput < loceol
5908 && ! ANYOF_BITMAP_TEST(scan, nextchr))
5909 {
5910 DEBUG_EXECUTE_r(
5911 Perl_re_exec_indentf( aTHX_ "%sTRIE: failed to match trie start class...%s\n",
5912 depth, PL_colors[4], PL_colors[5])
5913 );
5914 sayNO_SILENT;
5915 NOT_REACHED; /* NOTREACHED */
5916 }
5917 /* FALLTHROUGH */
5918 case TRIE: /* (ab|cd) */
5919 /* the basic plan of execution of the trie is:
5920 * At the beginning, run though all the states, and
5921 * find the longest-matching word. Also remember the position
5922 * of the shortest matching word. For example, this pattern:
5923 * 1 2 3 4 5
5924 * ab|a|x|abcd|abc
5925 * when matched against the string "abcde", will generate
5926 * accept states for all words except 3, with the longest
5927 * matching word being 4, and the shortest being 2 (with
5928 * the position being after char 1 of the string).
5929 *
5930 * Then for each matching word, in word order (i.e. 1,2,4,5),
5931 * we run the remainder of the pattern; on each try setting
5932 * the current position to the character following the word,
5933 * returning to try the next word on failure.
5934 *
5935 * We avoid having to build a list of words at runtime by
5936 * using a compile-time structure, wordinfo[].prev, which
5937 * gives, for each word, the previous accepting word (if any).
5938 * In the case above it would contain the mappings 1->2, 2->0,
5939 * 3->0, 4->5, 5->1. We can use this table to generate, from
5940 * the longest word (4 above), a list of all words, by
5941 * following the list of prev pointers; this gives us the
5942 * unordered list 4,5,1,2. Then given the current word we have
5943 * just tried, we can go through the list and find the
5944 * next-biggest word to try (so if we just failed on word 2,
5945 * the next in the list is 4).
5946 *
5947 * Since at runtime we don't record the matching position in
5948 * the string for each word, we have to work that out for
5949 * each word we're about to process. The wordinfo table holds
5950 * the character length of each word; given that we recorded
5951 * at the start: the position of the shortest word and its
5952 * length in chars, we just need to move the pointer the
5953 * difference between the two char lengths. Depending on
5954 * Unicode status and folding, that's cheap or expensive.
5955 *
5956 * This algorithm is optimised for the case where are only a
5957 * small number of accept states, i.e. 0,1, or maybe 2.
5958 * With lots of accepts states, and having to try all of them,
5959 * it becomes quadratic on number of accept states to find all
5960 * the next words.
5961 */
5962
5963 {
5964 /* what type of TRIE am I? (utf8 makes this contextual) */
5965 DECL_TRIE_TYPE(scan);
5966
5967 /* what trie are we using right now */
5968 reg_trie_data * const trie
5969 = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
5970 HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]);
5971 U32 state = trie->startstate;
5972
5973 if (scan->flags == EXACTL || scan->flags == EXACTFLU8) {
5974 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
5975 if (utf8_target
5976 && ! NEXTCHR_IS_EOS
5977 && UTF8_IS_ABOVE_LATIN1(nextchr)
5978 && scan->flags == EXACTL)
5979 {
5980 /* We only output for EXACTL, as we let the folder
5981 * output this message for EXACTFLU8 to avoid
5982 * duplication */
5983 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput,
5984 reginfo->strend);
5985 }
5986 }
5987 if ( trie->bitmap
5988 && ( NEXTCHR_IS_EOS
5989 || locinput >= loceol
5990 || ! TRIE_BITMAP_TEST(trie, nextchr)))
5991 {
5992 if (trie->states[ state ].wordnum) {
5993 DEBUG_EXECUTE_r(
5994 Perl_re_exec_indentf( aTHX_ "%sTRIE: matched empty string...%s\n",
5995 depth, PL_colors[4], PL_colors[5])
5996 );
5997 if (!trie->jump)
5998 break;
5999 } else {
6000 DEBUG_EXECUTE_r(
6001 Perl_re_exec_indentf( aTHX_ "%sTRIE: failed to match trie start class...%s\n",
6002 depth, PL_colors[4], PL_colors[5])
6003 );
6004 sayNO_SILENT;
6005 }
6006 }
6007
6008 {
6009 U8 *uc = ( U8* )locinput;
6010
6011 STRLEN len = 0;
6012 STRLEN foldlen = 0;
6013 U8 *uscan = (U8*)NULL;
6014 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
6015 U32 charcount = 0; /* how many input chars we have matched */
6016 U32 accepted = 0; /* have we seen any accepting states? */
6017
6018 ST.jump = trie->jump;
6019 ST.me = scan;
6020 ST.firstpos = NULL;
6021 ST.longfold = FALSE; /* char longer if folded => it's harder */
6022 ST.nextword = 0;
6023
6024 /* fully traverse the TRIE; note the position of the
6025 shortest accept state and the wordnum of the longest
6026 accept state */
6027
6028 while ( state && uc <= (U8*)(loceol) ) {
6029 U32 base = trie->states[ state ].trans.base;
6030 UV uvc = 0;
6031 U16 charid = 0;
6032 U16 wordnum;
6033 wordnum = trie->states[ state ].wordnum;
6034
6035 if (wordnum) { /* it's an accept state */
6036 if (!accepted) {
6037 accepted = 1;
6038 /* record first match position */
6039 if (ST.longfold) {
6040 ST.firstpos = (U8*)locinput;
6041 ST.firstchars = 0;
6042 }
6043 else {
6044 ST.firstpos = uc;
6045 ST.firstchars = charcount;
6046 }
6047 }
6048 if (!ST.nextword || wordnum < ST.nextword)
6049 ST.nextword = wordnum;
6050 ST.topword = wordnum;
6051 }
6052
6053 DEBUG_TRIE_EXECUTE_r({
6054 DUMP_EXEC_POS( (char *)uc, scan, utf8_target, depth );
6055 /* HERE */
6056 PerlIO_printf( Perl_debug_log,
6057 "%*s%sTRIE: State: %4" UVxf " Accepted: %c ",
6058 INDENT_CHARS(depth), "", PL_colors[4],
6059 (UV)state, (accepted ? 'Y' : 'N'));
6060 });
6061
6062 /* read a char and goto next state */
6063 if ( base && (foldlen || uc < (U8*)(loceol))) {
6064 I32 offset;
6065 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
6066 (U8 *) loceol, uscan,
6067 len, uvc, charid, foldlen,
6068 foldbuf, uniflags);
6069 charcount++;
6070 if (foldlen>0)
6071 ST.longfold = TRUE;
6072 if (charid &&
6073 ( ((offset =
6074 base + charid - 1 - trie->uniquecharcount)) >= 0)
6075
6076 && ((U32)offset < trie->lasttrans)
6077 && trie->trans[offset].check == state)
6078 {
6079 state = trie->trans[offset].next;
6080 }
6081 else {
6082 state = 0;
6083 }
6084 uc += len;
6085
6086 }
6087 else {
6088 state = 0;
6089 }
6090 DEBUG_TRIE_EXECUTE_r(
6091 Perl_re_printf( aTHX_
6092 "TRIE: Charid:%3x CP:%4" UVxf " After State: %4" UVxf "%s\n",
6093 charid, uvc, (UV)state, PL_colors[5] );
6094 );
6095 }
6096 if (!accepted)
6097 sayNO;
6098
6099 /* calculate total number of accept states */
6100 {
6101 U16 w = ST.topword;
6102 accepted = 0;
6103 while (w) {
6104 w = trie->wordinfo[w].prev;
6105 accepted++;
6106 }
6107 ST.accepted = accepted;
6108 }
6109
6110 DEBUG_EXECUTE_r(
6111 Perl_re_exec_indentf( aTHX_ "%sTRIE: got %" IVdf " possible matches%s\n",
6112 depth,
6113 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
6114 );
6115 goto trie_first_try; /* jump into the fail handler */
6116 }}
6117 NOT_REACHED; /* NOTREACHED */
6118
6119 case TRIE_next_fail: /* we failed - try next alternative */
6120 {
6121 U8 *uc;
6122 if ( ST.jump ) {
6123 /* undo any captures done in the tail part of a branch,
6124 * e.g.
6125 * /(?:X(.)(.)|Y(.)).../
6126 * where the trie just matches X then calls out to do the
6127 * rest of the branch */
6128 REGCP_UNWIND(ST.cp);
6129 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
6130 }
6131 if (!--ST.accepted) {
6132 DEBUG_EXECUTE_r({
6133 Perl_re_exec_indentf( aTHX_ "%sTRIE failed...%s\n",
6134 depth,
6135 PL_colors[4],
6136 PL_colors[5] );
6137 });
6138 sayNO_SILENT;
6139 }
6140 {
6141 /* Find next-highest word to process. Note that this code
6142 * is O(N^2) per trie run (O(N) per branch), so keep tight */
6143 U16 min = 0;
6144 U16 word;
6145 U16 const nextword = ST.nextword;
6146 reg_trie_wordinfo * const wordinfo
6147 = ((reg_trie_data*)rexi->data->data[ARG(ST.me)])->wordinfo;
6148 for (word=ST.topword; word; word=wordinfo[word].prev) {
6149 if (word > nextword && (!min || word < min))
6150 min = word;
6151 }
6152 ST.nextword = min;
6153 }
6154
6155 trie_first_try:
6156 if (do_cutgroup) {
6157 do_cutgroup = 0;
6158 no_final = 0;
6159 }
6160
6161 if ( ST.jump ) {
6162 ST.lastparen = rex->lastparen;
6163 ST.lastcloseparen = rex->lastcloseparen;
6164 REGCP_SET(ST.cp);
6165 }
6166
6167 /* find start char of end of current word */
6168 {
6169 U32 chars; /* how many chars to skip */
6170 reg_trie_data * const trie
6171 = (reg_trie_data*)rexi->data->data[ARG(ST.me)];
6172
6173 assert((trie->wordinfo[ST.nextword].len - trie->prefixlen)
6174 >= ST.firstchars);
6175 chars = (trie->wordinfo[ST.nextword].len - trie->prefixlen)
6176 - ST.firstchars;
6177 uc = ST.firstpos;
6178
6179 if (ST.longfold) {
6180 /* the hard option - fold each char in turn and find
6181 * its folded length (which may be different */
6182 U8 foldbuf[UTF8_MAXBYTES_CASE + 1];
6183 STRLEN foldlen;
6184 STRLEN len;
6185 UV uvc;
6186 U8 *uscan;
6187
6188 while (chars) {
6189 if (utf8_target) {
6190 /* XXX This assumes the length is well-formed, as
6191 * does the UTF8SKIP below */
6192 uvc = utf8n_to_uvchr((U8*)uc, UTF8_MAXLEN, &len,
6193 uniflags);
6194 uc += len;
6195 }
6196 else {
6197 uvc = *uc;
6198 uc++;
6199 }
6200 uvc = to_uni_fold(uvc, foldbuf, &foldlen);
6201 uscan = foldbuf;
6202 while (foldlen) {
6203 if (!--chars)
6204 break;
6205 uvc = utf8n_to_uvchr(uscan, foldlen, &len,
6206 uniflags);
6207 uscan += len;
6208 foldlen -= len;
6209 }
6210 }
6211 }
6212 else {
6213 if (utf8_target)
6214 while (chars--)
6215 uc += UTF8SKIP(uc);
6216 else
6217 uc += chars;
6218 }
6219 }
6220
6221 scan = ST.me + ((ST.jump && ST.jump[ST.nextword])
6222 ? ST.jump[ST.nextword]
6223 : NEXT_OFF(ST.me));
6224
6225 DEBUG_EXECUTE_r({
6226 Perl_re_exec_indentf( aTHX_ "%sTRIE matched word #%d, continuing%s\n",
6227 depth,
6228 PL_colors[4],
6229 ST.nextword,
6230 PL_colors[5]
6231 );
6232 });
6233
6234 if ( ST.accepted > 1 || has_cutgroup || ST.jump ) {
6235 PUSH_STATE_GOTO(TRIE_next, scan, (char*)uc, loceol,
6236 script_run_begin);
6237 NOT_REACHED; /* NOTREACHED */
6238 }
6239 /* only one choice left - just continue */
6240 DEBUG_EXECUTE_r({
6241 AV *const trie_words
6242 = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
6243 SV ** const tmp = trie_words
6244 ? av_fetch(trie_words, ST.nextword - 1, 0) : NULL;
6245 SV *sv= tmp ? sv_newmortal() : NULL;
6246
6247 Perl_re_exec_indentf( aTHX_ "%sTRIE: only one match left, short-circuiting: #%d <%s>%s\n",
6248 depth, PL_colors[4],
6249 ST.nextword,
6250 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
6251 PL_colors[0], PL_colors[1],
6252 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)|PERL_PV_ESCAPE_NONASCII
6253 )
6254 : "not compiled under -Dr",
6255 PL_colors[5] );
6256 });
6257
6258 locinput = (char*)uc;
6259 continue; /* execute rest of RE */
6260 /* NOTREACHED */
6261 }
6262 #undef ST
6263
6264 case EXACTL: /* /abc/l */
6265 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
6266
6267 /* Complete checking would involve going through every character
6268 * matched by the string to see if any is above latin1. But the
6269 * comparision otherwise might very well be a fast assembly
6270 * language routine, and I (khw) don't think slowing things down
6271 * just to check for this warning is worth it. So this just checks
6272 * the first character */
6273 if (utf8_target && UTF8_IS_ABOVE_LATIN1(*locinput)) {
6274 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput, reginfo->strend);
6275 }
6276 goto do_exact;
6277 case EXACT_ONLY8:
6278 if (! utf8_target) {
6279 sayNO;
6280 }
6281 /* FALLTHROUGH */
6282 case EXACT: { /* /abc/ */
6283 char *s;
6284 do_exact:
6285 s = STRING(scan);
6286 ln = STR_LEN(scan);
6287 if (utf8_target != is_utf8_pat) {
6288 /* The target and the pattern have differing utf8ness. */
6289 char *l = locinput;
6290 const char * const e = s + ln;
6291
6292 if (utf8_target) {
6293 /* The target is utf8, the pattern is not utf8.
6294 * Above-Latin1 code points can't match the pattern;
6295 * invariants match exactly, and the other Latin1 ones need
6296 * to be downgraded to a single byte in order to do the
6297 * comparison. (If we could be confident that the target
6298 * is not malformed, this could be refactored to have fewer
6299 * tests by just assuming that if the first bytes match, it
6300 * is an invariant, but there are tests in the test suite
6301 * dealing with (??{...}) which violate this) */
6302 while (s < e) {
6303 if ( l >= loceol
6304 || UTF8_IS_ABOVE_LATIN1(* (U8*) l))
6305 {
6306 sayNO;
6307 }
6308 if (UTF8_IS_INVARIANT(*(U8*)l)) {
6309 if (*l != *s) {
6310 sayNO;
6311 }
6312 l++;
6313 }
6314 else {
6315 if (EIGHT_BIT_UTF8_TO_NATIVE(*l, *(l+1)) != * (U8*) s)
6316 {
6317 sayNO;
6318 }
6319 l += 2;
6320 }
6321 s++;
6322 }
6323 }
6324 else {
6325 /* The target is not utf8, the pattern is utf8. */
6326 while (s < e) {
6327 if ( l >= loceol
6328 || UTF8_IS_ABOVE_LATIN1(* (U8*) s))
6329 {
6330 sayNO;
6331 }
6332 if (UTF8_IS_INVARIANT(*(U8*)s)) {
6333 if (*s != *l) {
6334 sayNO;
6335 }
6336 s++;
6337 }
6338 else {
6339 if (EIGHT_BIT_UTF8_TO_NATIVE(*s, *(s+1)) != * (U8*) l)
6340 {
6341 sayNO;
6342 }
6343 s += 2;
6344 }
6345 l++;
6346 }
6347 }
6348 locinput = l;
6349 }
6350 else {
6351 /* The target and the pattern have the same utf8ness. */
6352 /* Inline the first character, for speed. */
6353 if ( loceol - locinput < ln
6354 || UCHARAT(s) != nextchr
6355 || (ln > 1 && memNE(s, locinput, ln)))
6356 {
6357 sayNO;
6358 }
6359 locinput += ln;
6360 }
6361 break;
6362 }
6363
6364 case EXACTFL: /* /abc/il */
6365 {
6366 re_fold_t folder;
6367 const U8 * fold_array;
6368 const char * s;
6369 U32 fold_utf8_flags;
6370
6371 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
6372 folder = foldEQ_locale;
6373 fold_array = PL_fold_locale;
6374 fold_utf8_flags = FOLDEQ_LOCALE;
6375 goto do_exactf;
6376
6377 case EXACTFLU8: /* /abc/il; but all 'abc' are above 255, so
6378 is effectively /u; hence to match, target
6379 must be UTF-8. */
6380 if (! utf8_target) {
6381 sayNO;
6382 }
6383 fold_utf8_flags = FOLDEQ_LOCALE | FOLDEQ_S2_ALREADY_FOLDED
6384 | FOLDEQ_S2_FOLDS_SANE;
6385 folder = foldEQ_latin1_s2_folded;
6386 fold_array = PL_fold_latin1;
6387 goto do_exactf;
6388
6389 case EXACTFU_ONLY8: /* /abc/iu with something in /abc/ > 255 */
6390 if (! utf8_target) {
6391 sayNO;
6392 }
6393 assert(is_utf8_pat);
6394 fold_utf8_flags = FOLDEQ_S2_ALREADY_FOLDED;
6395 goto do_exactf;
6396
6397 case EXACTFUP: /* /foo/iu, and something is problematic in
6398 'foo' so can't take shortcuts. */
6399 assert(! is_utf8_pat);
6400 folder = foldEQ_latin1;
6401 fold_array = PL_fold_latin1;
6402 fold_utf8_flags = 0;
6403 goto do_exactf;
6404
6405 case EXACTFU: /* /abc/iu */
6406 folder = foldEQ_latin1_s2_folded;
6407 fold_array = PL_fold_latin1;
6408 fold_utf8_flags = FOLDEQ_S2_ALREADY_FOLDED;
6409 goto do_exactf;
6410
6411 case EXACTFAA_NO_TRIE: /* This node only generated for non-utf8
6412 patterns */
6413 assert(! is_utf8_pat);
6414 /* FALLTHROUGH */
6415 case EXACTFAA: /* /abc/iaa */
6416 folder = foldEQ_latin1_s2_folded;
6417 fold_array = PL_fold_latin1;
6418 fold_utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
6419 if (is_utf8_pat || ! utf8_target) {
6420
6421 /* The possible presence of a MICRO SIGN in the pattern forbids
6422 * us to view a non-UTF-8 pattern as folded when there is a
6423 * UTF-8 target */
6424 fold_utf8_flags |= FOLDEQ_S2_ALREADY_FOLDED
6425 |FOLDEQ_S2_FOLDS_SANE;
6426 }
6427 goto do_exactf;
6428
6429
6430 case EXACTF: /* /abc/i This node only generated for
6431 non-utf8 patterns */
6432 assert(! is_utf8_pat);
6433 folder = foldEQ;
6434 fold_array = PL_fold;
6435 fold_utf8_flags = 0;
6436
6437 do_exactf:
6438 s = STRING(scan);
6439 ln = STR_LEN(scan);
6440
6441 if ( utf8_target
6442 || is_utf8_pat
6443 || state_num == EXACTFUP
6444 || (state_num == EXACTFL && IN_UTF8_CTYPE_LOCALE))
6445 {
6446 /* Either target or the pattern are utf8, or has the issue where
6447 * the fold lengths may differ. */
6448 const char * const l = locinput;
6449 char *e = loceol;
6450
6451 if (! foldEQ_utf8_flags(l, &e, 0, utf8_target,
6452 s, 0, ln, is_utf8_pat,fold_utf8_flags))
6453 {
6454 sayNO;
6455 }
6456 locinput = e;
6457 break;
6458 }
6459
6460 /* Neither the target nor the pattern are utf8 */
6461 if (UCHARAT(s) != nextchr
6462 && !NEXTCHR_IS_EOS
6463 && UCHARAT(s) != fold_array[nextchr])
6464 {
6465 sayNO;
6466 }
6467 if (loceol - locinput < ln)
6468 sayNO;
6469 if (ln > 1 && ! folder(locinput, s, ln))
6470 sayNO;
6471 locinput += ln;
6472 break;
6473 }
6474
6475 case NBOUNDL: /* /\B/l */
6476 to_complement = 1;
6477 /* FALLTHROUGH */
6478
6479 case BOUNDL: /* /\b/l */
6480 {
6481 bool b1, b2;
6482 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
6483
6484 if (FLAGS(scan) != TRADITIONAL_BOUND) {
6485 if (! IN_UTF8_CTYPE_LOCALE) {
6486 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
6487 B_ON_NON_UTF8_LOCALE_IS_WRONG);
6488 }
6489 goto boundu;
6490 }
6491
6492 if (utf8_target) {
6493 if (locinput == reginfo->strbeg)
6494 b1 = isWORDCHAR_LC('\n');
6495 else {
6496 b1 = isWORDCHAR_LC_utf8_safe(reghop3((U8*)locinput, -1,
6497 (U8*)(reginfo->strbeg)),
6498 (U8*)(reginfo->strend));
6499 }
6500 b2 = (NEXTCHR_IS_EOS)
6501 ? isWORDCHAR_LC('\n')
6502 : isWORDCHAR_LC_utf8_safe((U8*) locinput,
6503 (U8*) reginfo->strend);
6504 }
6505 else { /* Here the string isn't utf8 */
6506 b1 = (locinput == reginfo->strbeg)
6507 ? isWORDCHAR_LC('\n')
6508 : isWORDCHAR_LC(UCHARAT(locinput - 1));
6509 b2 = (NEXTCHR_IS_EOS)
6510 ? isWORDCHAR_LC('\n')
6511 : isWORDCHAR_LC(nextchr);
6512 }
6513 if (to_complement ^ (b1 == b2)) {
6514 sayNO;
6515 }
6516 break;
6517 }
6518
6519 case NBOUND: /* /\B/ */
6520 to_complement = 1;
6521 /* FALLTHROUGH */
6522
6523 case BOUND: /* /\b/ */
6524 if (utf8_target) {
6525 goto bound_utf8;
6526 }
6527 goto bound_ascii_match_only;
6528
6529 case NBOUNDA: /* /\B/a */
6530 to_complement = 1;
6531 /* FALLTHROUGH */
6532
6533 case BOUNDA: /* /\b/a */
6534 {
6535 bool b1, b2;
6536
6537 bound_ascii_match_only:
6538 /* Here the string isn't utf8, or is utf8 and only ascii characters
6539 * are to match \w. In the latter case looking at the byte just
6540 * prior to the current one may be just the final byte of a
6541 * multi-byte character. This is ok. There are two cases:
6542 * 1) it is a single byte character, and then the test is doing
6543 * just what it's supposed to.
6544 * 2) it is a multi-byte character, in which case the final byte is
6545 * never mistakable for ASCII, and so the test will say it is
6546 * not a word character, which is the correct answer. */
6547 b1 = (locinput == reginfo->strbeg)
6548 ? isWORDCHAR_A('\n')
6549 : isWORDCHAR_A(UCHARAT(locinput - 1));
6550 b2 = (NEXTCHR_IS_EOS)
6551 ? isWORDCHAR_A('\n')
6552 : isWORDCHAR_A(nextchr);
6553 if (to_complement ^ (b1 == b2)) {
6554 sayNO;
6555 }
6556 break;
6557 }
6558
6559 case NBOUNDU: /* /\B/u */
6560 to_complement = 1;
6561 /* FALLTHROUGH */
6562
6563 case BOUNDU: /* /\b/u */
6564
6565 boundu:
6566 if (UNLIKELY(reginfo->strbeg >= reginfo->strend)) {
6567 match = FALSE;
6568 }
6569 else if (utf8_target) {
6570 bound_utf8:
6571 switch((bound_type) FLAGS(scan)) {
6572 case TRADITIONAL_BOUND:
6573 {
6574 bool b1, b2;
6575 b1 = (locinput == reginfo->strbeg)
6576 ? 0 /* isWORDCHAR_L1('\n') */
6577 : isWORDCHAR_utf8_safe(
6578 reghop3((U8*)locinput,
6579 -1,
6580 (U8*)(reginfo->strbeg)),
6581 (U8*) reginfo->strend);
6582 b2 = (NEXTCHR_IS_EOS)
6583 ? 0 /* isWORDCHAR_L1('\n') */
6584 : isWORDCHAR_utf8_safe((U8*)locinput,
6585 (U8*) reginfo->strend);
6586 match = cBOOL(b1 != b2);
6587 break;
6588 }
6589 case GCB_BOUND:
6590 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
6591 match = TRUE; /* GCB always matches at begin and
6592 end */
6593 }
6594 else {
6595 /* Find the gcb values of previous and current
6596 * chars, then see if is a break point */
6597 match = isGCB(getGCB_VAL_UTF8(
6598 reghop3((U8*)locinput,
6599 -1,
6600 (U8*)(reginfo->strbeg)),
6601 (U8*) reginfo->strend),
6602 getGCB_VAL_UTF8((U8*) locinput,
6603 (U8*) reginfo->strend),
6604 (U8*) reginfo->strbeg,
6605 (U8*) locinput,
6606 utf8_target);
6607 }
6608 break;
6609
6610 case LB_BOUND:
6611 if (locinput == reginfo->strbeg) {
6612 match = FALSE;
6613 }
6614 else if (NEXTCHR_IS_EOS) {
6615 match = TRUE;
6616 }
6617 else {
6618 match = isLB(getLB_VAL_UTF8(
6619 reghop3((U8*)locinput,
6620 -1,
6621 (U8*)(reginfo->strbeg)),
6622 (U8*) reginfo->strend),
6623 getLB_VAL_UTF8((U8*) locinput,
6624 (U8*) reginfo->strend),
6625 (U8*) reginfo->strbeg,
6626 (U8*) locinput,
6627 (U8*) reginfo->strend,
6628 utf8_target);
6629 }
6630 break;
6631
6632 case SB_BOUND: /* Always matches at begin and end */
6633 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
6634 match = TRUE;
6635 }
6636 else {
6637 match = isSB(getSB_VAL_UTF8(
6638 reghop3((U8*)locinput,
6639 -1,
6640 (U8*)(reginfo->strbeg)),
6641 (U8*) reginfo->strend),
6642 getSB_VAL_UTF8((U8*) locinput,
6643 (U8*) reginfo->strend),
6644 (U8*) reginfo->strbeg,
6645 (U8*) locinput,
6646 (U8*) reginfo->strend,
6647 utf8_target);
6648 }
6649 break;
6650
6651 case WB_BOUND:
6652 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
6653 match = TRUE;
6654 }
6655 else {
6656 match = isWB(WB_UNKNOWN,
6657 getWB_VAL_UTF8(
6658 reghop3((U8*)locinput,
6659 -1,
6660 (U8*)(reginfo->strbeg)),
6661 (U8*) reginfo->strend),
6662 getWB_VAL_UTF8((U8*) locinput,
6663 (U8*) reginfo->strend),
6664 (U8*) reginfo->strbeg,
6665 (U8*) locinput,
6666 (U8*) reginfo->strend,
6667 utf8_target);
6668 }
6669 break;
6670 }
6671 }
6672 else { /* Not utf8 target */
6673 switch((bound_type) FLAGS(scan)) {
6674 case TRADITIONAL_BOUND:
6675 {
6676 bool b1, b2;
6677 b1 = (locinput == reginfo->strbeg)
6678 ? 0 /* isWORDCHAR_L1('\n') */
6679 : isWORDCHAR_L1(UCHARAT(locinput - 1));
6680 b2 = (NEXTCHR_IS_EOS)
6681 ? 0 /* isWORDCHAR_L1('\n') */
6682 : isWORDCHAR_L1(nextchr);
6683 match = cBOOL(b1 != b2);
6684 break;
6685 }
6686
6687 case GCB_BOUND:
6688 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
6689 match = TRUE; /* GCB always matches at begin and
6690 end */
6691 }
6692 else { /* Only CR-LF combo isn't a GCB in 0-255
6693 range */
6694 match = UCHARAT(locinput - 1) != '\r'
6695 || UCHARAT(locinput) != '\n';
6696 }
6697 break;
6698
6699 case LB_BOUND:
6700 if (locinput == reginfo->strbeg) {
6701 match = FALSE;
6702 }
6703 else if (NEXTCHR_IS_EOS) {
6704 match = TRUE;
6705 }
6706 else {
6707 match = isLB(getLB_VAL_CP(UCHARAT(locinput -1)),
6708 getLB_VAL_CP(UCHARAT(locinput)),
6709 (U8*) reginfo->strbeg,
6710 (U8*) locinput,
6711 (U8*) reginfo->strend,
6712 utf8_target);
6713 }
6714 break;
6715
6716 case SB_BOUND: /* Always matches at begin and end */
6717 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
6718 match = TRUE;
6719 }
6720 else {
6721 match = isSB(getSB_VAL_CP(UCHARAT(locinput -1)),
6722 getSB_VAL_CP(UCHARAT(locinput)),
6723 (U8*) reginfo->strbeg,
6724 (U8*) locinput,
6725 (U8*) reginfo->strend,
6726 utf8_target);
6727 }
6728 break;
6729
6730 case WB_BOUND:
6731 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
6732 match = TRUE;
6733 }
6734 else {
6735 match = isWB(WB_UNKNOWN,
6736 getWB_VAL_CP(UCHARAT(locinput -1)),
6737 getWB_VAL_CP(UCHARAT(locinput)),
6738 (U8*) reginfo->strbeg,
6739 (U8*) locinput,
6740 (U8*) reginfo->strend,
6741 utf8_target);
6742 }
6743 break;
6744 }
6745 }
6746
6747 if (to_complement ^ ! match) {
6748 sayNO;
6749 }
6750 break;
6751
6752 case ANYOFPOSIXL:
6753 case ANYOFL: /* /[abc]/l */
6754 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
6755
6756 if (ANYOFL_UTF8_LOCALE_REQD(FLAGS(scan)) && ! IN_UTF8_CTYPE_LOCALE)
6757 {
6758 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE), utf8_locale_required);
6759 }
6760 /* FALLTHROUGH */
6761 case ANYOFD: /* /[abc]/d */
6762 case ANYOF: /* /[abc]/ */
6763 if (NEXTCHR_IS_EOS || locinput >= loceol)
6764 sayNO;
6765 if ( (! utf8_target || UTF8_IS_INVARIANT(*locinput))
6766 && ! (ANYOF_FLAGS(scan) & ~ ANYOF_MATCHES_ALL_ABOVE_BITMAP))
6767 {
6768 if (! ANYOF_BITMAP_TEST(scan, * (U8 *) (locinput))) {
6769 sayNO;
6770 }
6771 locinput++;
6772 }
6773 else {
6774 if (!reginclass(rex, scan, (U8*)locinput, (U8*) loceol,
6775 utf8_target))
6776 {
6777 sayNO;
6778 }
6779 goto increment_locinput;
6780 }
6781 break;
6782
6783 case ANYOFM:
6784 if ( NEXTCHR_IS_EOS
6785 || (UCHARAT(locinput) & FLAGS(scan)) != ARG(scan)
6786 || locinput >= loceol)
6787 {
6788 sayNO;
6789 }
6790 locinput++; /* ANYOFM is always single byte */
6791 break;
6792
6793 case NANYOFM:
6794 if ( NEXTCHR_IS_EOS
6795 || (UCHARAT(locinput) & FLAGS(scan)) == ARG(scan)
6796 || locinput >= loceol)
6797 {
6798 sayNO;
6799 }
6800 goto increment_locinput;
6801 break;
6802
6803 case ANYOFH:
6804 if ( ! utf8_target
6805 || NEXTCHR_IS_EOS
6806 || ( ANYOF_FLAGS(scan) != 0
6807 && ANYOF_FLAGS(scan) != (U8) *locinput)
6808 || ! reginclass(rex, scan, (U8*)locinput, (U8*) loceol,
6809 utf8_target))
6810 {
6811 sayNO;
6812 }
6813 goto increment_locinput;
6814 break;
6815
6816 /* The argument (FLAGS) to all the POSIX node types is the class number
6817 * */
6818
6819 case NPOSIXL: /* \W or [:^punct:] etc. under /l */
6820 to_complement = 1;
6821 /* FALLTHROUGH */
6822
6823 case POSIXL: /* \w or [:punct:] etc. under /l */
6824 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
6825 if (NEXTCHR_IS_EOS || locinput >= loceol)
6826 sayNO;
6827
6828 /* Use isFOO_lc() for characters within Latin1. (Note that
6829 * UTF8_IS_INVARIANT works even on non-UTF-8 strings, or else
6830 * wouldn't be invariant) */
6831 if (UTF8_IS_INVARIANT(nextchr) || ! utf8_target) {
6832 if (! (to_complement ^ cBOOL(isFOO_lc(FLAGS(scan), (U8) nextchr)))) {
6833 sayNO;
6834 }
6835
6836 locinput++;
6837 break;
6838 }
6839
6840 if (! UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(locinput, reginfo->strend)) {
6841 /* An above Latin-1 code point, or malformed */
6842 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput,
6843 reginfo->strend);
6844 goto utf8_posix_above_latin1;
6845 }
6846
6847 /* Here is a UTF-8 variant code point below 256 and the target is
6848 * UTF-8 */
6849 if (! (to_complement ^ cBOOL(isFOO_lc(FLAGS(scan),
6850 EIGHT_BIT_UTF8_TO_NATIVE(nextchr,
6851 *(locinput + 1))))))
6852 {
6853 sayNO;
6854 }
6855
6856 goto increment_locinput;
6857
6858 case NPOSIXD: /* \W or [:^punct:] etc. under /d */
6859 to_complement = 1;
6860 /* FALLTHROUGH */
6861
6862 case POSIXD: /* \w or [:punct:] etc. under /d */
6863 if (utf8_target) {
6864 goto utf8_posix;
6865 }
6866 goto posixa;
6867
6868 case NPOSIXA: /* \W or [:^punct:] etc. under /a */
6869
6870 if (NEXTCHR_IS_EOS || locinput >= loceol) {
6871 sayNO;
6872 }
6873
6874 /* All UTF-8 variants match */
6875 if (! UTF8_IS_INVARIANT(nextchr)) {
6876 goto increment_locinput;
6877 }
6878
6879 to_complement = 1;
6880 goto join_nposixa;
6881
6882 case POSIXA: /* \w or [:punct:] etc. under /a */
6883
6884 posixa:
6885 /* We get here through POSIXD, NPOSIXD, and NPOSIXA when not in
6886 * UTF-8, and also from NPOSIXA even in UTF-8 when the current
6887 * character is a single byte */
6888
6889 if (NEXTCHR_IS_EOS || locinput >= loceol) {
6890 sayNO;
6891 }
6892
6893 join_nposixa:
6894
6895 if (! (to_complement ^ cBOOL(_generic_isCC_A(nextchr,
6896 FLAGS(scan)))))
6897 {
6898 sayNO;
6899 }
6900
6901 /* Here we are either not in utf8, or we matched a utf8-invariant,
6902 * so the next char is the next byte */
6903 locinput++;
6904 break;
6905
6906 case NPOSIXU: /* \W or [:^punct:] etc. under /u */
6907 to_complement = 1;
6908 /* FALLTHROUGH */
6909
6910 case POSIXU: /* \w or [:punct:] etc. under /u */
6911 utf8_posix:
6912 if (NEXTCHR_IS_EOS || locinput >= loceol) {
6913 sayNO;
6914 }
6915
6916 /* Use _generic_isCC() for characters within Latin1. (Note that
6917 * UTF8_IS_INVARIANT works even on non-UTF-8 strings, or else
6918 * wouldn't be invariant) */
6919 if (UTF8_IS_INVARIANT(nextchr) || ! utf8_target) {
6920 if (! (to_complement ^ cBOOL(_generic_isCC(nextchr,
6921 FLAGS(scan)))))
6922 {
6923 sayNO;
6924 }
6925 locinput++;
6926 }
6927 else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(locinput, reginfo->strend)) {
6928 if (! (to_complement
6929 ^ cBOOL(_generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(nextchr,
6930 *(locinput + 1)),
6931 FLAGS(scan)))))
6932 {
6933 sayNO;
6934 }
6935 locinput += 2;
6936 }
6937 else { /* Handle above Latin-1 code points */
6938 utf8_posix_above_latin1:
6939 classnum = (_char_class_number) FLAGS(scan);
6940 switch (classnum) {
6941 default:
6942 if (! (to_complement
6943 ^ cBOOL(_invlist_contains_cp(
6944 PL_XPosix_ptrs[classnum],
6945 utf8_to_uvchr_buf((U8 *) locinput,
6946 (U8 *) reginfo->strend,
6947 NULL)))))
6948 {
6949 sayNO;
6950 }
6951 break;
6952 case _CC_ENUM_SPACE:
6953 if (! (to_complement
6954 ^ cBOOL(is_XPERLSPACE_high(locinput))))
6955 {
6956 sayNO;
6957 }
6958 break;
6959 case _CC_ENUM_BLANK:
6960 if (! (to_complement
6961 ^ cBOOL(is_HORIZWS_high(locinput))))
6962 {
6963 sayNO;
6964 }
6965 break;
6966 case _CC_ENUM_XDIGIT:
6967 if (! (to_complement
6968 ^ cBOOL(is_XDIGIT_high(locinput))))
6969 {
6970 sayNO;
6971 }
6972 break;
6973 case _CC_ENUM_VERTSPACE:
6974 if (! (to_complement
6975 ^ cBOOL(is_VERTWS_high(locinput))))
6976 {
6977 sayNO;
6978 }
6979 break;
6980 case _CC_ENUM_CNTRL: /* These can't match above Latin1 */
6981 case _CC_ENUM_ASCII:
6982 if (! to_complement) {
6983 sayNO;
6984 }
6985 break;
6986 }
6987 locinput += UTF8_SAFE_SKIP(locinput, reginfo->strend);
6988 }
6989 break;
6990
6991 case CLUMP: /* Match \X: logical Unicode character. This is defined as
6992 a Unicode extended Grapheme Cluster */
6993 if (NEXTCHR_IS_EOS || locinput >= loceol)
6994 sayNO;
6995 if (! utf8_target) {
6996
6997 /* Match either CR LF or '.', as all the other possibilities
6998 * require utf8 */
6999 locinput++; /* Match the . or CR */
7000 if (nextchr == '\r' /* And if it was CR, and the next is LF,
7001 match the LF */
7002 && locinput < loceol
7003 && UCHARAT(locinput) == '\n')
7004 {
7005 locinput++;
7006 }
7007 }
7008 else {
7009
7010 /* Get the gcb type for the current character */
7011 GCB_enum prev_gcb = getGCB_VAL_UTF8((U8*) locinput,
7012 (U8*) reginfo->strend);
7013
7014 /* Then scan through the input until we get to the first
7015 * character whose type is supposed to be a gcb with the
7016 * current character. (There is always a break at the
7017 * end-of-input) */
7018 locinput += UTF8SKIP(locinput);
7019 while (locinput < loceol) {
7020 GCB_enum cur_gcb = getGCB_VAL_UTF8((U8*) locinput,
7021 (U8*) reginfo->strend);
7022 if (isGCB(prev_gcb, cur_gcb,
7023 (U8*) reginfo->strbeg, (U8*) locinput,
7024 utf8_target))
7025 {
7026 break;
7027 }
7028
7029 prev_gcb = cur_gcb;
7030 locinput += UTF8SKIP(locinput);
7031 }
7032
7033
7034 }
7035 break;
7036
7037 case NREFFL: /* /\g{name}/il */
7038 { /* The capture buffer cases. The ones beginning with N for the
7039 named buffers just convert to the equivalent numbered and
7040 pretend they were called as the corresponding numbered buffer
7041 op. */
7042 /* don't initialize these in the declaration, it makes C++
7043 unhappy */
7044 const char *s;
7045 char type;
7046 re_fold_t folder;
7047 const U8 *fold_array;
7048 UV utf8_fold_flags;
7049
7050 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
7051 folder = foldEQ_locale;
7052 fold_array = PL_fold_locale;
7053 type = REFFL;
7054 utf8_fold_flags = FOLDEQ_LOCALE;
7055 goto do_nref;
7056
7057 case NREFFA: /* /\g{name}/iaa */
7058 folder = foldEQ_latin1;
7059 fold_array = PL_fold_latin1;
7060 type = REFFA;
7061 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
7062 goto do_nref;
7063
7064 case NREFFU: /* /\g{name}/iu */
7065 folder = foldEQ_latin1;
7066 fold_array = PL_fold_latin1;
7067 type = REFFU;
7068 utf8_fold_flags = 0;
7069 goto do_nref;
7070
7071 case NREFF: /* /\g{name}/i */
7072 folder = foldEQ;
7073 fold_array = PL_fold;
7074 type = REFF;
7075 utf8_fold_flags = 0;
7076 goto do_nref;
7077
7078 case NREF: /* /\g{name}/ */
7079 type = REF;
7080 folder = NULL;
7081 fold_array = NULL;
7082 utf8_fold_flags = 0;
7083 do_nref:
7084
7085 /* For the named back references, find the corresponding buffer
7086 * number */
7087 n = reg_check_named_buff_matched(rex,scan);
7088
7089 if ( ! n ) {
7090 sayNO;
7091 }
7092 goto do_nref_ref_common;
7093
7094 case REFFL: /* /\1/il */
7095 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
7096 folder = foldEQ_locale;
7097 fold_array = PL_fold_locale;
7098 utf8_fold_flags = FOLDEQ_LOCALE;
7099 goto do_ref;
7100
7101 case REFFA: /* /\1/iaa */
7102 folder = foldEQ_latin1;
7103 fold_array = PL_fold_latin1;
7104 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
7105 goto do_ref;
7106
7107 case REFFU: /* /\1/iu */
7108 folder = foldEQ_latin1;
7109 fold_array = PL_fold_latin1;
7110 utf8_fold_flags = 0;
7111 goto do_ref;
7112
7113 case REFF: /* /\1/i */
7114 folder = foldEQ;
7115 fold_array = PL_fold;
7116 utf8_fold_flags = 0;
7117 goto do_ref;
7118
7119 case REF: /* /\1/ */
7120 folder = NULL;
7121 fold_array = NULL;
7122 utf8_fold_flags = 0;
7123
7124 do_ref:
7125 type = OP(scan);
7126 n = ARG(scan); /* which paren pair */
7127
7128 do_nref_ref_common:
7129 ln = rex->offs[n].start;
7130 endref = rex->offs[n].end;
7131 reginfo->poscache_iter = reginfo->poscache_maxiter; /* Void cache */
7132 if (rex->lastparen < n || ln == -1 || endref == -1)
7133 sayNO; /* Do not match unless seen CLOSEn. */
7134 if (ln == endref)
7135 break;
7136
7137 s = reginfo->strbeg + ln;
7138 if (type != REF /* REF can do byte comparison */
7139 && (utf8_target || type == REFFU || type == REFFL))
7140 {
7141 char * limit = loceol;
7142
7143 /* This call case insensitively compares the entire buffer
7144 * at s, with the current input starting at locinput, but
7145 * not going off the end given by loceol, and
7146 * returns in <limit> upon success, how much of the
7147 * current input was matched */
7148 if (! foldEQ_utf8_flags(s, NULL, endref - ln, utf8_target,
7149 locinput, &limit, 0, utf8_target, utf8_fold_flags))
7150 {
7151 sayNO;
7152 }
7153 locinput = limit;
7154 break;
7155 }
7156
7157 /* Not utf8: Inline the first character, for speed. */
7158 if ( ! NEXTCHR_IS_EOS
7159 && locinput < loceol
7160 && UCHARAT(s) != nextchr
7161 && ( type == REF
7162 || UCHARAT(s) != fold_array[nextchr]))
7163 {
7164 sayNO;
7165 }
7166 ln = endref - ln;
7167 if (locinput + ln > loceol)
7168 sayNO;
7169 if (ln > 1 && (type == REF
7170 ? memNE(s, locinput, ln)
7171 : ! folder(locinput, s, ln)))
7172 sayNO;
7173 locinput += ln;
7174 break;
7175 }
7176
7177 case NOTHING: /* null op; e.g. the 'nothing' following
7178 * the '*' in m{(a+|b)*}' */
7179 break;
7180 case TAIL: /* placeholder while compiling (A|B|C) */
7181 break;
7182
7183 #undef ST
7184 #define ST st->u.eval
7185 #define CUR_EVAL cur_eval->u.eval
7186
7187 {
7188 SV *ret;
7189 REGEXP *re_sv;
7190 regexp *re;
7191 regexp_internal *rei;
7192 regnode *startpoint;
7193 U32 arg;
7194
7195 case GOSUB: /* /(...(?1))/ /(...(?&foo))/ */
7196 arg= (U32)ARG(scan);
7197 if (cur_eval && cur_eval->locinput == locinput) {
7198 if ( ++nochange_depth > max_nochange_depth )
7199 Perl_croak(aTHX_
7200 "Pattern subroutine nesting without pos change"
7201 " exceeded limit in regex");
7202 } else {
7203 nochange_depth = 0;
7204 }
7205 re_sv = rex_sv;
7206 re = rex;
7207 rei = rexi;
7208 startpoint = scan + ARG2L(scan);
7209 EVAL_CLOSE_PAREN_SET( st, arg );
7210 /* Detect infinite recursion
7211 *
7212 * A pattern like /(?R)foo/ or /(?<x>(?&y)foo)(?<y>(?&x)bar)/
7213 * or "a"=~/(.(?2))((?<=(?=(?1)).))/ could recurse forever.
7214 * So we track the position in the string we are at each time
7215 * we recurse and if we try to enter the same routine twice from
7216 * the same position we throw an error.
7217 */
7218 if ( rex->recurse_locinput[arg] == locinput ) {
7219 /* FIXME: we should show the regop that is failing as part
7220 * of the error message. */
7221 Perl_croak(aTHX_ "Infinite recursion in regex");
7222 } else {
7223 ST.prev_recurse_locinput= rex->recurse_locinput[arg];
7224 rex->recurse_locinput[arg]= locinput;
7225
7226 DEBUG_r({
7227 GET_RE_DEBUG_FLAGS_DECL;
7228 DEBUG_STACK_r({
7229 Perl_re_exec_indentf( aTHX_
7230 "entering GOSUB, prev_recurse_locinput=%p recurse_locinput[%d]=%p\n",
7231 depth, ST.prev_recurse_locinput, arg, rex->recurse_locinput[arg]
7232 );
7233 });
7234 });
7235 }
7236
7237 /* Save all the positions seen so far. */
7238 ST.cp = regcppush(rex, 0, maxopenparen);
7239 REGCP_SET(ST.lastcp);
7240
7241 /* and then jump to the code we share with EVAL */
7242 goto eval_recurse_doit;
7243 /* NOTREACHED */
7244
7245 case EVAL: /* /(?{...})B/ /(??{A})B/ and /(?(?{...})X|Y)B/ */
7246 if (logical == 2 && cur_eval && cur_eval->locinput==locinput) {
7247 if ( ++nochange_depth > max_nochange_depth )
7248 Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex");
7249 } else {
7250 nochange_depth = 0;
7251 }
7252 {
7253 /* execute the code in the {...} */
7254
7255 dSP;
7256 IV before;
7257 OP * const oop = PL_op;
7258 COP * const ocurcop = PL_curcop;
7259 OP *nop;
7260 CV *newcv;
7261
7262 /* save *all* paren positions */
7263 regcppush(rex, 0, maxopenparen);
7264 REGCP_SET(ST.lastcp);
7265
7266 if (!caller_cv)
7267 caller_cv = find_runcv(NULL);
7268
7269 n = ARG(scan);
7270
7271 if (rexi->data->what[n] == 'r') { /* code from an external qr */
7272 newcv = (ReANY(
7273 (REGEXP*)(rexi->data->data[n])
7274 ))->qr_anoncv;
7275 nop = (OP*)rexi->data->data[n+1];
7276 }
7277 else if (rexi->data->what[n] == 'l') { /* literal code */
7278 newcv = caller_cv;
7279 nop = (OP*)rexi->data->data[n];
7280 assert(CvDEPTH(newcv));
7281 }
7282 else {
7283 /* literal with own CV */
7284 assert(rexi->data->what[n] == 'L');
7285 newcv = rex->qr_anoncv;
7286 nop = (OP*)rexi->data->data[n];
7287 }
7288
7289 /* Some notes about MULTICALL and the context and save stacks.
7290 *
7291 * In something like
7292 * /...(?{ my $x)}...(?{ my $y)}...(?{ my $z)}.../
7293 * since codeblocks don't introduce a new scope (so that
7294 * local() etc accumulate), at the end of a successful
7295 * match there will be a SAVEt_CLEARSV on the savestack
7296 * for each of $x, $y, $z. If the three code blocks above
7297 * happen to have come from different CVs (e.g. via
7298 * embedded qr//s), then we must ensure that during any
7299 * savestack unwinding, PL_comppad always points to the
7300 * right pad at each moment. We achieve this by
7301 * interleaving SAVEt_COMPPAD's on the savestack whenever
7302 * there is a change of pad.
7303 * In theory whenever we call a code block, we should
7304 * push a CXt_SUB context, then pop it on return from
7305 * that code block. This causes a bit of an issue in that
7306 * normally popping a context also clears the savestack
7307 * back to cx->blk_oldsaveix, but here we specifically
7308 * don't want to clear the save stack on exit from the
7309 * code block.
7310 * Also for efficiency we don't want to keep pushing and
7311 * popping the single SUB context as we backtrack etc.
7312 * So instead, we push a single context the first time
7313 * we need, it, then hang onto it until the end of this
7314 * function. Whenever we encounter a new code block, we
7315 * update the CV etc if that's changed. During the times
7316 * in this function where we're not executing a code
7317 * block, having the SUB context still there is a bit
7318 * naughty - but we hope that no-one notices.
7319 * When the SUB context is initially pushed, we fake up
7320 * cx->blk_oldsaveix to be as if we'd pushed this context
7321 * on first entry to S_regmatch rather than at some random
7322 * point during the regexe execution. That way if we
7323 * croak, popping the context stack will ensure that
7324 * *everything* SAVEd by this function is undone and then
7325 * the context popped, rather than e.g., popping the
7326 * context (and restoring the original PL_comppad) then
7327 * popping more of the savestack and restoring a bad
7328 * PL_comppad.
7329 */
7330
7331 /* If this is the first EVAL, push a MULTICALL. On
7332 * subsequent calls, if we're executing a different CV, or
7333 * if PL_comppad has got messed up from backtracking
7334 * through SAVECOMPPADs, then refresh the context.
7335 */
7336 if (newcv != last_pushed_cv || PL_comppad != last_pad)
7337 {
7338 U8 flags = (CXp_SUB_RE |
7339 ((newcv == caller_cv) ? CXp_SUB_RE_FAKE : 0));
7340 SAVECOMPPAD();
7341 if (last_pushed_cv) {
7342 CHANGE_MULTICALL_FLAGS(newcv, flags);
7343 }
7344 else {
7345 PUSH_MULTICALL_FLAGS(newcv, flags);
7346 }
7347 /* see notes above */
7348 CX_CUR()->blk_oldsaveix = orig_savestack_ix;
7349
7350 last_pushed_cv = newcv;
7351 }
7352 else {
7353 /* these assignments are just to silence compiler
7354 * warnings */
7355 multicall_cop = NULL;
7356 }
7357 last_pad = PL_comppad;
7358
7359 /* the initial nextstate you would normally execute
7360 * at the start of an eval (which would cause error
7361 * messages to come from the eval), may be optimised
7362 * away from the execution path in the regex code blocks;
7363 * so manually set PL_curcop to it initially */
7364 {
7365 OP *o = cUNOPx(nop)->op_first;
7366 assert(o->op_type == OP_NULL);
7367 if (o->op_targ == OP_SCOPE) {
7368 o = cUNOPo->op_first;
7369 }
7370 else {
7371 assert(o->op_targ == OP_LEAVE);
7372 o = cUNOPo->op_first;
7373 assert(o->op_type == OP_ENTER);
7374 o = OpSIBLING(o);
7375 }
7376
7377 if (o->op_type != OP_STUB) {
7378 assert( o->op_type == OP_NEXTSTATE
7379 || o->op_type == OP_DBSTATE
7380 || (o->op_type == OP_NULL
7381 && ( o->op_targ == OP_NEXTSTATE
7382 || o->op_targ == OP_DBSTATE
7383 )
7384 )
7385 );
7386 PL_curcop = (COP*)o;
7387 }
7388 }
7389 nop = nop->op_next;
7390
7391 DEBUG_STATE_r( Perl_re_printf( aTHX_
7392 " re EVAL PL_op=0x%" UVxf "\n", PTR2UV(nop)) );
7393
7394 rex->offs[0].end = locinput - reginfo->strbeg;
7395 if (reginfo->info_aux_eval->pos_magic)
7396 MgBYTEPOS_set(reginfo->info_aux_eval->pos_magic,
7397 reginfo->sv, reginfo->strbeg,
7398 locinput - reginfo->strbeg);
7399
7400 if (sv_yes_mark) {
7401 SV *sv_mrk = get_sv("REGMARK", 1);
7402 sv_setsv(sv_mrk, sv_yes_mark);
7403 }
7404
7405 /* we don't use MULTICALL here as we want to call the
7406 * first op of the block of interest, rather than the
7407 * first op of the sub. Also, we don't want to free
7408 * the savestack frame */
7409 before = (IV)(SP-PL_stack_base);
7410 PL_op = nop;
7411 CALLRUNOPS(aTHX); /* Scalar context. */
7412 SPAGAIN;
7413 if ((IV)(SP-PL_stack_base) == before)
7414 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
7415 else {
7416 ret = POPs;
7417 PUTBACK;
7418 }
7419
7420 /* before restoring everything, evaluate the returned
7421 * value, so that 'uninit' warnings don't use the wrong
7422 * PL_op or pad. Also need to process any magic vars
7423 * (e.g. $1) *before* parentheses are restored */
7424
7425 PL_op = NULL;
7426
7427 re_sv = NULL;
7428 if (logical == 0) { /* (?{})/ */
7429 SV *replsv = save_scalar(PL_replgv);
7430 sv_setsv(replsv, ret); /* $^R */
7431 SvSETMAGIC(replsv);
7432 }
7433 else if (logical == 1) { /* /(?(?{...})X|Y)/ */
7434 sw = cBOOL(SvTRUE_NN(ret));
7435 logical = 0;
7436 }
7437 else { /* /(??{}) */
7438 /* if its overloaded, let the regex compiler handle
7439 * it; otherwise extract regex, or stringify */
7440 if (SvGMAGICAL(ret))
7441 ret = sv_mortalcopy(ret);
7442 if (!SvAMAGIC(ret)) {
7443 SV *sv = ret;
7444 if (SvROK(sv))
7445 sv = SvRV(sv);
7446 if (SvTYPE(sv) == SVt_REGEXP)
7447 re_sv = (REGEXP*) sv;
7448 else if (SvSMAGICAL(ret)) {
7449 MAGIC *mg = mg_find(ret, PERL_MAGIC_qr);
7450 if (mg)
7451 re_sv = (REGEXP *) mg->mg_obj;
7452 }
7453
7454 /* force any undef warnings here */
7455 if (!re_sv && !SvPOK(ret) && !SvNIOK(ret)) {
7456 ret = sv_mortalcopy(ret);
7457 (void) SvPV_force_nolen(ret);
7458 }
7459 }
7460
7461 }
7462
7463 /* *** Note that at this point we don't restore
7464 * PL_comppad, (or pop the CxSUB) on the assumption it may
7465 * be used again soon. This is safe as long as nothing
7466 * in the regexp code uses the pad ! */
7467 PL_op = oop;
7468 PL_curcop = ocurcop;
7469 regcp_restore(rex, ST.lastcp, &maxopenparen);
7470 PL_curpm_under = PL_curpm;
7471 PL_curpm = PL_reg_curpm;
7472
7473 if (logical != 2) {
7474 PUSH_STATE_GOTO(EVAL_B, next, locinput, loceol,
7475 script_run_begin);
7476 /* NOTREACHED */
7477 }
7478 }
7479
7480 /* only /(??{})/ from now on */
7481 logical = 0;
7482 {
7483 /* extract RE object from returned value; compiling if
7484 * necessary */
7485
7486 if (re_sv) {
7487 re_sv = reg_temp_copy(NULL, re_sv);
7488 }
7489 else {
7490 U32 pm_flags = 0;
7491
7492 if (SvUTF8(ret) && IN_BYTES) {
7493 /* In use 'bytes': make a copy of the octet
7494 * sequence, but without the flag on */
7495 STRLEN len;
7496 const char *const p = SvPV(ret, len);
7497 ret = newSVpvn_flags(p, len, SVs_TEMP);
7498 }
7499 if (rex->intflags & PREGf_USE_RE_EVAL)
7500 pm_flags |= PMf_USE_RE_EVAL;
7501
7502 /* if we got here, it should be an engine which
7503 * supports compiling code blocks and stuff */
7504 assert(rex->engine && rex->engine->op_comp);
7505 assert(!(scan->flags & ~RXf_PMf_COMPILETIME));
7506 re_sv = rex->engine->op_comp(aTHX_ &ret, 1, NULL,
7507 rex->engine, NULL, NULL,
7508 /* copy /msixn etc to inner pattern */
7509 ARG2L(scan),
7510 pm_flags);
7511
7512 if (!(SvFLAGS(ret)
7513 & (SVs_TEMP | SVs_GMG | SVf_ROK))
7514 && (!SvPADTMP(ret) || SvREADONLY(ret))) {
7515 /* This isn't a first class regexp. Instead, it's
7516 caching a regexp onto an existing, Perl visible
7517 scalar. */
7518 sv_magic(ret, MUTABLE_SV(re_sv), PERL_MAGIC_qr, 0, 0);
7519 }
7520 }
7521 SAVEFREESV(re_sv);
7522 re = ReANY(re_sv);
7523 }
7524 RXp_MATCH_COPIED_off(re);
7525 re->subbeg = rex->subbeg;
7526 re->sublen = rex->sublen;
7527 re->suboffset = rex->suboffset;
7528 re->subcoffset = rex->subcoffset;
7529 re->lastparen = 0;
7530 re->lastcloseparen = 0;
7531 rei = RXi_GET(re);
7532 DEBUG_EXECUTE_r(
7533 debug_start_match(re_sv, utf8_target, locinput,
7534 reginfo->strend, "EVAL/GOSUB: Matching embedded");
7535 );
7536 startpoint = rei->program + 1;
7537 EVAL_CLOSE_PAREN_CLEAR(st); /* ST.close_paren = 0;
7538 * close_paren only for GOSUB */
7539 ST.prev_recurse_locinput= NULL; /* only used for GOSUB */
7540 /* Save all the seen positions so far. */
7541 ST.cp = regcppush(rex, 0, maxopenparen);
7542 REGCP_SET(ST.lastcp);
7543 /* and set maxopenparen to 0, since we are starting a "fresh" match */
7544 maxopenparen = 0;
7545 /* run the pattern returned from (??{...}) */
7546
7547 eval_recurse_doit: /* Share code with GOSUB below this line
7548 * At this point we expect the stack context to be
7549 * set up correctly */
7550
7551 /* invalidate the S-L poscache. We're now executing a
7552 * different set of WHILEM ops (and their associated
7553 * indexes) against the same string, so the bits in the
7554 * cache are meaningless. Setting maxiter to zero forces
7555 * the cache to be invalidated and zeroed before reuse.
7556 * XXX This is too dramatic a measure. Ideally we should
7557 * save the old cache and restore when running the outer
7558 * pattern again */
7559 reginfo->poscache_maxiter = 0;
7560
7561 /* the new regexp might have a different is_utf8_pat than we do */
7562 is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(re_sv));
7563
7564 ST.prev_rex = rex_sv;
7565 ST.prev_curlyx = cur_curlyx;
7566 rex_sv = re_sv;
7567 SET_reg_curpm(rex_sv);
7568 rex = re;
7569 rexi = rei;
7570 cur_curlyx = NULL;
7571 ST.B = next;
7572 ST.prev_eval = cur_eval;
7573 cur_eval = st;
7574 /* now continue from first node in postoned RE */
7575 PUSH_YES_STATE_GOTO(EVAL_postponed_AB, startpoint, locinput,
7576 loceol, script_run_begin);
7577 NOT_REACHED; /* NOTREACHED */
7578 }
7579
7580 case EVAL_postponed_AB: /* cleanup after a successful (??{A})B */
7581 /* note: this is called twice; first after popping B, then A */
7582 DEBUG_STACK_r({
7583 Perl_re_exec_indentf( aTHX_ "EVAL_AB cur_eval=%p prev_eval=%p\n",
7584 depth, cur_eval, ST.prev_eval);
7585 });
7586
7587 #define SET_RECURSE_LOCINPUT(STR,VAL)\
7588 if ( cur_eval && CUR_EVAL.close_paren ) {\
7589 DEBUG_STACK_r({ \
7590 Perl_re_exec_indentf( aTHX_ STR " GOSUB%d ce=%p recurse_locinput=%p\n",\
7591 depth, \
7592 CUR_EVAL.close_paren - 1,\
7593 cur_eval, \
7594 VAL); \
7595 }); \
7596 rex->recurse_locinput[CUR_EVAL.close_paren - 1] = VAL;\
7597 }
7598
7599 SET_RECURSE_LOCINPUT("EVAL_AB[before]", CUR_EVAL.prev_recurse_locinput);
7600
7601 rex_sv = ST.prev_rex;
7602 is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(rex_sv));
7603 SET_reg_curpm(rex_sv);
7604 rex = ReANY(rex_sv);
7605 rexi = RXi_GET(rex);
7606 {
7607 /* preserve $^R across LEAVE's. See Bug 121070. */
7608 SV *save_sv= GvSV(PL_replgv);
7609 SV *replsv;
7610 SvREFCNT_inc(save_sv);
7611 regcpblow(ST.cp); /* LEAVE in disguise */
7612 /* don't move this initialization up */
7613 replsv = GvSV(PL_replgv);
7614 sv_setsv(replsv, save_sv);
7615 SvSETMAGIC(replsv);
7616 SvREFCNT_dec(save_sv);
7617 }
7618 cur_eval = ST.prev_eval;
7619 cur_curlyx = ST.prev_curlyx;
7620
7621 /* Invalidate cache. See "invalidate" comment above. */
7622 reginfo->poscache_maxiter = 0;
7623 if ( nochange_depth )
7624 nochange_depth--;
7625
7626 SET_RECURSE_LOCINPUT("EVAL_AB[after]", cur_eval->locinput);
7627 sayYES;
7628
7629
7630 case EVAL_B_fail: /* unsuccessful B in (?{...})B */
7631 REGCP_UNWIND(ST.lastcp);
7632 sayNO;
7633
7634 case EVAL_postponed_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
7635 /* note: this is called twice; first after popping B, then A */
7636 DEBUG_STACK_r({
7637 Perl_re_exec_indentf( aTHX_ "EVAL_AB_fail cur_eval=%p prev_eval=%p\n",
7638 depth, cur_eval, ST.prev_eval);
7639 });
7640
7641 SET_RECURSE_LOCINPUT("EVAL_AB_fail[before]", CUR_EVAL.prev_recurse_locinput);
7642
7643 rex_sv = ST.prev_rex;
7644 is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(rex_sv));
7645 SET_reg_curpm(rex_sv);
7646 rex = ReANY(rex_sv);
7647 rexi = RXi_GET(rex);
7648
7649 REGCP_UNWIND(ST.lastcp);
7650 regcppop(rex, &maxopenparen);
7651 cur_eval = ST.prev_eval;
7652 cur_curlyx = ST.prev_curlyx;
7653
7654 /* Invalidate cache. See "invalidate" comment above. */
7655 reginfo->poscache_maxiter = 0;
7656 if ( nochange_depth )
7657 nochange_depth--;
7658
7659 SET_RECURSE_LOCINPUT("EVAL_AB_fail[after]", cur_eval->locinput);
7660 sayNO_SILENT;
7661 #undef ST
7662
7663 case OPEN: /* ( */
7664 n = ARG(scan); /* which paren pair */
7665 rex->offs[n].start_tmp = locinput - reginfo->strbeg;
7666 if (n > maxopenparen)
7667 maxopenparen = n;
7668 DEBUG_BUFFERS_r(Perl_re_exec_indentf( aTHX_
7669 "OPEN: rex=0x%" UVxf " offs=0x%" UVxf ": \\%" UVuf ": set %" IVdf " tmp; maxopenparen=%" UVuf "\n",
7670 depth,
7671 PTR2UV(rex),
7672 PTR2UV(rex->offs),
7673 (UV)n,
7674 (IV)rex->offs[n].start_tmp,
7675 (UV)maxopenparen
7676 ));
7677 lastopen = n;
7678 break;
7679
7680 case SROPEN: /* (*SCRIPT_RUN: */
7681 script_run_begin = (U8 *) locinput;
7682 break;
7683
7684
7685 case CLOSE: /* ) */
7686 n = ARG(scan); /* which paren pair */
7687 CLOSE_CAPTURE(n, rex->offs[n].start_tmp,
7688 locinput - reginfo->strbeg);
7689 if ( EVAL_CLOSE_PAREN_IS( cur_eval, n ) )
7690 goto fake_end;
7691
7692 break;
7693
7694 case SRCLOSE: /* (*SCRIPT_RUN: ... ) */
7695
7696 if (! isSCRIPT_RUN(script_run_begin, (U8 *) locinput, utf8_target))
7697 {
7698 sayNO;
7699 }
7700
7701 break;
7702
7703
7704 case ACCEPT: /* (*ACCEPT) */
7705 if (scan->flags)
7706 sv_yes_mark = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
7707 if (ARG2L(scan)){
7708 regnode *cursor;
7709 for (cursor=scan;
7710 cursor && OP(cursor)!=END;
7711 cursor=regnext(cursor))
7712 {
7713 if ( OP(cursor)==CLOSE ){
7714 n = ARG(cursor);
7715 if ( n <= lastopen ) {
7716 CLOSE_CAPTURE(n, rex->offs[n].start_tmp,
7717 locinput - reginfo->strbeg);
7718 if ( n == ARG(scan) || EVAL_CLOSE_PAREN_IS(cur_eval, n) )
7719 break;
7720 }
7721 }
7722 }
7723 }
7724 goto fake_end;
7725 /* NOTREACHED */
7726
7727 case GROUPP: /* (?(1)) */
7728 n = ARG(scan); /* which paren pair */
7729 sw = cBOOL(rex->lastparen >= n && rex->offs[n].end != -1);
7730 break;
7731
7732 case NGROUPP: /* (?(<name>)) */
7733 /* reg_check_named_buff_matched returns 0 for no match */
7734 sw = cBOOL(0 < reg_check_named_buff_matched(rex,scan));
7735 break;
7736
7737 case INSUBP: /* (?(R)) */
7738 n = ARG(scan);
7739 /* this does not need to use EVAL_CLOSE_PAREN macros, as the arg
7740 * of SCAN is already set up as matches a eval.close_paren */
7741 sw = cur_eval && (n == 0 || CUR_EVAL.close_paren == n);
7742 break;
7743
7744 case DEFINEP: /* (?(DEFINE)) */
7745 sw = 0;
7746 break;
7747
7748 case IFTHEN: /* (?(cond)A|B) */
7749 reginfo->poscache_iter = reginfo->poscache_maxiter; /* Void cache */
7750 if (sw)
7751 next = NEXTOPER(NEXTOPER(scan));
7752 else {
7753 next = scan + ARG(scan);
7754 if (OP(next) == IFTHEN) /* Fake one. */
7755 next = NEXTOPER(NEXTOPER(next));
7756 }
7757 break;
7758
7759 case LOGICAL: /* modifier for EVAL and IFMATCH */
7760 logical = scan->flags;
7761 break;
7762
7763 /*******************************************************************
7764
7765 The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/
7766 pattern, where A and B are subpatterns. (For simple A, CURLYM or
7767 STAR/PLUS/CURLY/CURLYN are used instead.)
7768
7769 A*B is compiled as <CURLYX><A><WHILEM><B>
7770
7771 On entry to the subpattern, CURLYX is called. This pushes a CURLYX
7772 state, which contains the current count, initialised to -1. It also sets
7773 cur_curlyx to point to this state, with any previous value saved in the
7774 state block.
7775
7776 CURLYX then jumps straight to the WHILEM op, rather than executing A,
7777 since the pattern may possibly match zero times (i.e. it's a while {} loop
7778 rather than a do {} while loop).
7779
7780 Each entry to WHILEM represents a successful match of A. The count in the
7781 CURLYX block is incremented, another WHILEM state is pushed, and execution
7782 passes to A or B depending on greediness and the current count.
7783
7784 For example, if matching against the string a1a2a3b (where the aN are
7785 substrings that match /A/), then the match progresses as follows: (the
7786 pushed states are interspersed with the bits of strings matched so far):
7787
7788 <CURLYX cnt=-1>
7789 <CURLYX cnt=0><WHILEM>
7790 <CURLYX cnt=1><WHILEM> a1 <WHILEM>
7791 <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM>
7792 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM>
7793 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b
7794
7795 (Contrast this with something like CURLYM, which maintains only a single
7796 backtrack state:
7797
7798 <CURLYM cnt=0> a1
7799 a1 <CURLYM cnt=1> a2
7800 a1 a2 <CURLYM cnt=2> a3
7801 a1 a2 a3 <CURLYM cnt=3> b
7802 )
7803
7804 Each WHILEM state block marks a point to backtrack to upon partial failure
7805 of A or B, and also contains some minor state data related to that
7806 iteration. The CURLYX block, pointed to by cur_curlyx, contains the
7807 overall state, such as the count, and pointers to the A and B ops.
7808
7809 This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx
7810 must always point to the *current* CURLYX block, the rules are:
7811
7812 When executing CURLYX, save the old cur_curlyx in the CURLYX state block,
7813 and set cur_curlyx to point the new block.
7814
7815 When popping the CURLYX block after a successful or unsuccessful match,
7816 restore the previous cur_curlyx.
7817
7818 When WHILEM is about to execute B, save the current cur_curlyx, and set it
7819 to the outer one saved in the CURLYX block.
7820
7821 When popping the WHILEM block after a successful or unsuccessful B match,
7822 restore the previous cur_curlyx.
7823
7824 Here's an example for the pattern (AI* BI)*BO
7825 I and O refer to inner and outer, C and W refer to CURLYX and WHILEM:
7826
7827 cur_
7828 curlyx backtrack stack
7829 ------ ---------------
7830 NULL
7831 CO <CO prev=NULL> <WO>
7832 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
7833 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
7834 NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo
7835
7836 At this point the pattern succeeds, and we work back down the stack to
7837 clean up, restoring as we go:
7838
7839 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
7840 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
7841 CO <CO prev=NULL> <WO>
7842 NULL
7843
7844 *******************************************************************/
7845
7846 #define ST st->u.curlyx
7847
7848 case CURLYX: /* start of /A*B/ (for complex A) */
7849 {
7850 /* No need to save/restore up to this paren */
7851 I32 parenfloor = scan->flags;
7852
7853 assert(next); /* keep Coverity happy */
7854 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
7855 next += ARG(next);
7856
7857 /* XXXX Probably it is better to teach regpush to support
7858 parenfloor > maxopenparen ... */
7859 if (parenfloor > (I32)rex->lastparen)
7860 parenfloor = rex->lastparen; /* Pessimization... */
7861
7862 ST.prev_curlyx= cur_curlyx;
7863 cur_curlyx = st;
7864 ST.cp = PL_savestack_ix;
7865
7866 /* these fields contain the state of the current curly.
7867 * they are accessed by subsequent WHILEMs */
7868 ST.parenfloor = parenfloor;
7869 ST.me = scan;
7870 ST.B = next;
7871 ST.minmod = minmod;
7872 minmod = 0;
7873 ST.count = -1; /* this will be updated by WHILEM */
7874 ST.lastloc = NULL; /* this will be updated by WHILEM */
7875
7876 PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next), locinput, loceol,
7877 script_run_begin);
7878 NOT_REACHED; /* NOTREACHED */
7879 }
7880
7881 case CURLYX_end: /* just finished matching all of A*B */
7882 cur_curlyx = ST.prev_curlyx;
7883 sayYES;
7884 NOT_REACHED; /* NOTREACHED */
7885
7886 case CURLYX_end_fail: /* just failed to match all of A*B */
7887 regcpblow(ST.cp);
7888 cur_curlyx = ST.prev_curlyx;
7889 sayNO;
7890 NOT_REACHED; /* NOTREACHED */
7891
7892
7893 #undef ST
7894 #define ST st->u.whilem
7895
7896 case WHILEM: /* just matched an A in /A*B/ (for complex A) */
7897 {
7898 /* see the discussion above about CURLYX/WHILEM */
7899 I32 n;
7900 int min, max;
7901 regnode *A;
7902
7903 assert(cur_curlyx); /* keep Coverity happy */
7904
7905 min = ARG1(cur_curlyx->u.curlyx.me);
7906 max = ARG2(cur_curlyx->u.curlyx.me);
7907 A = NEXTOPER(cur_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS;
7908 n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
7909 ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
7910 ST.cache_offset = 0;
7911 ST.cache_mask = 0;
7912
7913
7914 DEBUG_EXECUTE_r( Perl_re_exec_indentf( aTHX_ "WHILEM: matched %ld out of %d..%d\n",
7915 depth, (long)n, min, max)
7916 );
7917
7918 /* First just match a string of min A's. */
7919
7920 if (n < min) {
7921 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor, maxopenparen);
7922 cur_curlyx->u.curlyx.lastloc = locinput;
7923 REGCP_SET(ST.lastcp);
7924
7925 PUSH_STATE_GOTO(WHILEM_A_pre, A, locinput, loceol,
7926 script_run_begin);
7927 NOT_REACHED; /* NOTREACHED */
7928 }
7929
7930 /* If degenerate A matches "", assume A done. */
7931
7932 if (locinput == cur_curlyx->u.curlyx.lastloc) {
7933 DEBUG_EXECUTE_r( Perl_re_exec_indentf( aTHX_ "WHILEM: empty match detected, trying continuation...\n",
7934 depth)
7935 );
7936 goto do_whilem_B_max;
7937 }
7938
7939 /* super-linear cache processing.
7940 *
7941 * The idea here is that for certain types of CURLYX/WHILEM -
7942 * principally those whose upper bound is infinity (and
7943 * excluding regexes that have things like \1 and other very
7944 * non-regular expresssiony things), then if a pattern like
7945 * /....A*.../ fails and we backtrack to the WHILEM, then we
7946 * make a note that this particular WHILEM op was at string
7947 * position 47 (say) when the rest of pattern failed. Then, if
7948 * we ever find ourselves back at that WHILEM, and at string
7949 * position 47 again, we can just fail immediately rather than
7950 * running the rest of the pattern again.
7951 *
7952 * This is very handy when patterns start to go
7953 * 'super-linear', like in (a+)*(a+)*(a+)*, where you end up
7954 * with a combinatorial explosion of backtracking.
7955 *
7956 * The cache is implemented as a bit array, with one bit per
7957 * string byte position per WHILEM op (up to 16) - so its
7958 * between 0.25 and 2x the string size.
7959 *
7960 * To avoid allocating a poscache buffer every time, we do an
7961 * initially countdown; only after we have executed a WHILEM
7962 * op (string-length x #WHILEMs) times do we allocate the
7963 * cache.
7964 *
7965 * The top 4 bits of scan->flags byte say how many different
7966 * relevant CURLLYX/WHILEM op pairs there are, while the
7967 * bottom 4-bits is the identifying index number of this
7968 * WHILEM.
7969 */
7970
7971 if (scan->flags) {
7972
7973 if (!reginfo->poscache_maxiter) {
7974 /* start the countdown: Postpone detection until we
7975 * know the match is not *that* much linear. */
7976 reginfo->poscache_maxiter
7977 = (reginfo->strend - reginfo->strbeg + 1)
7978 * (scan->flags>>4);
7979 /* possible overflow for long strings and many CURLYX's */
7980 if (reginfo->poscache_maxiter < 0)
7981 reginfo->poscache_maxiter = I32_MAX;
7982 reginfo->poscache_iter = reginfo->poscache_maxiter;
7983 }
7984
7985 if (reginfo->poscache_iter-- == 0) {
7986 /* initialise cache */
7987 const SSize_t size = (reginfo->poscache_maxiter + 7)/8;
7988 regmatch_info_aux *const aux = reginfo->info_aux;
7989 if (aux->poscache) {
7990 if ((SSize_t)reginfo->poscache_size < size) {
7991 Renew(aux->poscache, size, char);
7992 reginfo->poscache_size = size;
7993 }
7994 Zero(aux->poscache, size, char);
7995 }
7996 else {
7997 reginfo->poscache_size = size;
7998 Newxz(aux->poscache, size, char);
7999 }
8000 DEBUG_EXECUTE_r( Perl_re_printf( aTHX_
8001 "%sWHILEM: Detected a super-linear match, switching on caching%s...\n",
8002 PL_colors[4], PL_colors[5])
8003 );
8004 }
8005
8006 if (reginfo->poscache_iter < 0) {
8007 /* have we already failed at this position? */
8008 SSize_t offset, mask;
8009
8010 reginfo->poscache_iter = -1; /* stop eventual underflow */
8011 offset = (scan->flags & 0xf) - 1
8012 + (locinput - reginfo->strbeg)
8013 * (scan->flags>>4);
8014 mask = 1 << (offset % 8);
8015 offset /= 8;
8016 if (reginfo->info_aux->poscache[offset] & mask) {
8017 DEBUG_EXECUTE_r( Perl_re_exec_indentf( aTHX_ "WHILEM: (cache) already tried at this position...\n",
8018 depth)
8019 );
8020 cur_curlyx->u.curlyx.count--;
8021 sayNO; /* cache records failure */
8022 }
8023 ST.cache_offset = offset;
8024 ST.cache_mask = mask;
8025 }
8026 }
8027
8028 /* Prefer B over A for minimal matching. */
8029
8030 if (cur_curlyx->u.curlyx.minmod) {
8031 ST.save_curlyx = cur_curlyx;
8032 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
8033 PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B,
8034 locinput, loceol, script_run_begin);
8035 NOT_REACHED; /* NOTREACHED */
8036 }
8037
8038 /* Prefer A over B for maximal matching. */
8039
8040 if (n < max) { /* More greed allowed? */
8041 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor,
8042 maxopenparen);
8043 cur_curlyx->u.curlyx.lastloc = locinput;
8044 REGCP_SET(ST.lastcp);
8045 PUSH_STATE_GOTO(WHILEM_A_max, A, locinput, loceol,
8046 script_run_begin);
8047 NOT_REACHED; /* NOTREACHED */
8048 }
8049 goto do_whilem_B_max;
8050 }
8051 NOT_REACHED; /* NOTREACHED */
8052
8053 case WHILEM_B_min: /* just matched B in a minimal match */
8054 case WHILEM_B_max: /* just matched B in a maximal match */
8055 cur_curlyx = ST.save_curlyx;
8056 sayYES;
8057 NOT_REACHED; /* NOTREACHED */
8058
8059 case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
8060 cur_curlyx = ST.save_curlyx;
8061 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
8062 cur_curlyx->u.curlyx.count--;
8063 CACHEsayNO;
8064 NOT_REACHED; /* NOTREACHED */
8065
8066 case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
8067 /* FALLTHROUGH */
8068 case WHILEM_A_pre_fail: /* just failed to match even minimal A */
8069 REGCP_UNWIND(ST.lastcp);
8070 regcppop(rex, &maxopenparen);
8071 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
8072 cur_curlyx->u.curlyx.count--;
8073 CACHEsayNO;
8074 NOT_REACHED; /* NOTREACHED */
8075
8076 case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
8077 REGCP_UNWIND(ST.lastcp);
8078 regcppop(rex, &maxopenparen); /* Restore some previous $<digit>s? */
8079 DEBUG_EXECUTE_r(Perl_re_exec_indentf( aTHX_ "WHILEM: failed, trying continuation...\n",
8080 depth)
8081 );
8082 do_whilem_B_max:
8083 if (cur_curlyx->u.curlyx.count >= REG_INFTY
8084 && ckWARN(WARN_REGEXP)
8085 && !reginfo->warned)
8086 {
8087 reginfo->warned = TRUE;
8088 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
8089 "Complex regular subexpression recursion limit (%d) "
8090 "exceeded",
8091 REG_INFTY - 1);
8092 }
8093
8094 /* now try B */
8095 ST.save_curlyx = cur_curlyx;
8096 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
8097 PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B,
8098 locinput, loceol, script_run_begin);
8099 NOT_REACHED; /* NOTREACHED */
8100
8101 case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
8102 cur_curlyx = ST.save_curlyx;
8103
8104 if (cur_curlyx->u.curlyx.count >= /*max*/ARG2(cur_curlyx->u.curlyx.me)) {
8105 /* Maximum greed exceeded */
8106 if (cur_curlyx->u.curlyx.count >= REG_INFTY
8107 && ckWARN(WARN_REGEXP)
8108 && !reginfo->warned)
8109 {
8110 reginfo->warned = TRUE;
8111 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
8112 "Complex regular subexpression recursion "
8113 "limit (%d) exceeded",
8114 REG_INFTY - 1);
8115 }
8116 cur_curlyx->u.curlyx.count--;
8117 CACHEsayNO;
8118 }
8119
8120 DEBUG_EXECUTE_r(Perl_re_exec_indentf( aTHX_ "WHILEM: B min fail: trying longer...\n", depth)
8121 );
8122 /* Try grabbing another A and see if it helps. */
8123 cur_curlyx->u.curlyx.lastloc = locinput;
8124 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor,
8125 maxopenparen);
8126 REGCP_SET(ST.lastcp);
8127 PUSH_STATE_GOTO(WHILEM_A_min,
8128 /*A*/ NEXTOPER(ST.save_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS,
8129 locinput, loceol, script_run_begin);
8130 NOT_REACHED; /* NOTREACHED */
8131
8132 #undef ST
8133 #define ST st->u.branch
8134
8135 case BRANCHJ: /* /(...|A|...)/ with long next pointer */
8136 next = scan + ARG(scan);
8137 if (next == scan)
8138 next = NULL;
8139 scan = NEXTOPER(scan);
8140 /* FALLTHROUGH */
8141
8142 case BRANCH: /* /(...|A|...)/ */
8143 scan = NEXTOPER(scan); /* scan now points to inner node */
8144 ST.lastparen = rex->lastparen;
8145 ST.lastcloseparen = rex->lastcloseparen;
8146 ST.next_branch = next;
8147 REGCP_SET(ST.cp);
8148
8149 /* Now go into the branch */
8150 if (has_cutgroup) {
8151 PUSH_YES_STATE_GOTO(BRANCH_next, scan, locinput, loceol,
8152 script_run_begin);
8153 } else {
8154 PUSH_STATE_GOTO(BRANCH_next, scan, locinput, loceol,
8155 script_run_begin);
8156 }
8157 NOT_REACHED; /* NOTREACHED */
8158
8159 case CUTGROUP: /* /(*THEN)/ */
8160 sv_yes_mark = st->u.mark.mark_name = scan->flags
8161 ? MUTABLE_SV(rexi->data->data[ ARG( scan ) ])
8162 : NULL;
8163 PUSH_STATE_GOTO(CUTGROUP_next, next, locinput, loceol,
8164 script_run_begin);
8165 NOT_REACHED; /* NOTREACHED */
8166
8167 case CUTGROUP_next_fail:
8168 do_cutgroup = 1;
8169 no_final = 1;
8170 if (st->u.mark.mark_name)
8171 sv_commit = st->u.mark.mark_name;
8172 sayNO;
8173 NOT_REACHED; /* NOTREACHED */
8174
8175 case BRANCH_next:
8176 sayYES;
8177 NOT_REACHED; /* NOTREACHED */
8178
8179 case BRANCH_next_fail: /* that branch failed; try the next, if any */
8180 if (do_cutgroup) {
8181 do_cutgroup = 0;
8182 no_final = 0;
8183 }
8184 REGCP_UNWIND(ST.cp);
8185 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
8186 scan = ST.next_branch;
8187 /* no more branches? */
8188 if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) {
8189 DEBUG_EXECUTE_r({
8190 Perl_re_exec_indentf( aTHX_ "%sBRANCH failed...%s\n",
8191 depth,
8192 PL_colors[4],
8193 PL_colors[5] );
8194 });
8195 sayNO_SILENT;
8196 }
8197 continue; /* execute next BRANCH[J] op */
8198 /* NOTREACHED */
8199
8200 case MINMOD: /* next op will be non-greedy, e.g. A*? */
8201 minmod = 1;
8202 break;
8203
8204 #undef ST
8205 #define ST st->u.curlym
8206
8207 case CURLYM: /* /A{m,n}B/ where A is fixed-length */
8208
8209 /* This is an optimisation of CURLYX that enables us to push
8210 * only a single backtracking state, no matter how many matches
8211 * there are in {m,n}. It relies on the pattern being constant
8212 * length, with no parens to influence future backrefs
8213 */
8214
8215 ST.me = scan;
8216 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
8217
8218 ST.lastparen = rex->lastparen;
8219 ST.lastcloseparen = rex->lastcloseparen;
8220
8221 /* if paren positive, emulate an OPEN/CLOSE around A */
8222 if (ST.me->flags) {
8223 U32 paren = ST.me->flags;
8224 if (paren > maxopenparen)
8225 maxopenparen = paren;
8226 scan += NEXT_OFF(scan); /* Skip former OPEN. */
8227 }
8228 ST.A = scan;
8229 ST.B = next;
8230 ST.alen = 0;
8231 ST.count = 0;
8232 ST.minmod = minmod;
8233 minmod = 0;
8234 ST.c1 = CHRTEST_UNINIT;
8235 REGCP_SET(ST.cp);
8236
8237 if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */
8238 goto curlym_do_B;
8239
8240 curlym_do_A: /* execute the A in /A{m,n}B/ */
8241 PUSH_YES_STATE_GOTO(CURLYM_A, ST.A, locinput, loceol, /* match A */
8242 script_run_begin);
8243 NOT_REACHED; /* NOTREACHED */
8244
8245 case CURLYM_A: /* we've just matched an A */
8246 ST.count++;
8247 /* after first match, determine A's length: u.curlym.alen */
8248 if (ST.count == 1) {
8249 if (reginfo->is_utf8_target) {
8250 char *s = st->locinput;
8251 while (s < locinput) {
8252 ST.alen++;
8253 s += UTF8SKIP(s);
8254 }
8255 }
8256 else {
8257 ST.alen = locinput - st->locinput;
8258 }
8259 if (ST.alen == 0)
8260 ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me);
8261 }
8262 DEBUG_EXECUTE_r(
8263 Perl_re_exec_indentf( aTHX_ "CURLYM now matched %" IVdf " times, len=%" IVdf "...\n",
8264 depth, (IV) ST.count, (IV)ST.alen)
8265 );
8266
8267 if (EVAL_CLOSE_PAREN_IS_TRUE(cur_eval,(U32)ST.me->flags))
8268 goto fake_end;
8269
8270 {
8271 I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me));
8272 if ( max == REG_INFTY || ST.count < max )
8273 goto curlym_do_A; /* try to match another A */
8274 }
8275 goto curlym_do_B; /* try to match B */
8276
8277 case CURLYM_A_fail: /* just failed to match an A */
8278 REGCP_UNWIND(ST.cp);
8279
8280
8281 if (ST.minmod || ST.count < ARG1(ST.me) /* min*/
8282 || EVAL_CLOSE_PAREN_IS_TRUE(cur_eval,(U32)ST.me->flags))
8283 sayNO;
8284
8285 curlym_do_B: /* execute the B in /A{m,n}B/ */
8286 if (ST.c1 == CHRTEST_UNINIT) {
8287 /* calculate c1 and c2 for possible match of 1st char
8288 * following curly */
8289 ST.c1 = ST.c2 = CHRTEST_VOID;
8290 assert(ST.B);
8291 if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
8292 regnode *text_node = ST.B;
8293 if (! HAS_TEXT(text_node))
8294 FIND_NEXT_IMPT(text_node);
8295 if (PL_regkind[OP(text_node)] == EXACT) {
8296 if (! S_setup_EXACTISH_ST_c1_c2(aTHX_
8297 text_node, &ST.c1, ST.c1_utf8, &ST.c2, ST.c2_utf8,
8298 reginfo))
8299 {
8300 sayNO;
8301 }
8302 }
8303 }
8304 }
8305
8306 DEBUG_EXECUTE_r(
8307 Perl_re_exec_indentf( aTHX_ "CURLYM trying tail with matches=%" IVdf "...\n",
8308 depth, (IV)ST.count)
8309 );
8310 if (! NEXTCHR_IS_EOS && ST.c1 != CHRTEST_VOID) {
8311 if (! UTF8_IS_INVARIANT(nextchr) && utf8_target) {
8312
8313 /* (We can use memEQ and memNE in this file without
8314 * having to worry about one being shorter than the
8315 * other, since the first byte of each gives the
8316 * length of the character) */
8317 if ( memNE(locinput, ST.c1_utf8, UTF8_SAFE_SKIP(locinput,
8318 reginfo->strend))
8319 && memNE(locinput, ST.c2_utf8, UTF8_SAFE_SKIP(locinput,
8320 reginfo->strend)))
8321 {
8322 /* simulate B failing */
8323 DEBUG_OPTIMISE_r(
8324 Perl_re_exec_indentf( aTHX_ "CURLYM Fast bail next target=0x%" UVXf " c1=0x%" UVXf " c2=0x%" UVXf "\n",
8325 depth,
8326 valid_utf8_to_uvchr((U8 *) locinput, NULL),
8327 valid_utf8_to_uvchr(ST.c1_utf8, NULL),
8328 valid_utf8_to_uvchr(ST.c2_utf8, NULL))
8329 );
8330 state_num = CURLYM_B_fail;
8331 goto reenter_switch;
8332 }
8333 }
8334 else if (nextchr != ST.c1 && nextchr != ST.c2) {
8335 /* simulate B failing */
8336 DEBUG_OPTIMISE_r(
8337 Perl_re_exec_indentf( aTHX_ "CURLYM Fast bail next target=0x%X c1=0x%X c2=0x%X\n",
8338 depth,
8339 (int) nextchr, ST.c1, ST.c2)
8340 );
8341 state_num = CURLYM_B_fail;
8342 goto reenter_switch;
8343 }
8344 }
8345
8346 if (ST.me->flags) {
8347 /* emulate CLOSE: mark current A as captured */
8348 U32 paren = (U32)ST.me->flags;
8349 if (ST.count) {
8350 CLOSE_CAPTURE(paren,
8351 HOPc(locinput, -ST.alen) - reginfo->strbeg,
8352 locinput - reginfo->strbeg);
8353 }
8354 else
8355 rex->offs[paren].end = -1;
8356
8357 if (EVAL_CLOSE_PAREN_IS_TRUE(cur_eval,(U32)ST.me->flags))
8358 {
8359 if (ST.count)
8360 goto fake_end;
8361 else
8362 sayNO;
8363 }
8364 }
8365
8366 PUSH_STATE_GOTO(CURLYM_B, ST.B, locinput, loceol, /* match B */
8367 script_run_begin);
8368 NOT_REACHED; /* NOTREACHED */
8369
8370 case CURLYM_B_fail: /* just failed to match a B */
8371 REGCP_UNWIND(ST.cp);
8372 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
8373 if (ST.minmod) {
8374 I32 max = ARG2(ST.me);
8375 if (max != REG_INFTY && ST.count == max)
8376 sayNO;
8377 goto curlym_do_A; /* try to match a further A */
8378 }
8379 /* backtrack one A */
8380 if (ST.count == ARG1(ST.me) /* min */)
8381 sayNO;
8382 ST.count--;
8383 SET_locinput(HOPc(locinput, -ST.alen));
8384 goto curlym_do_B; /* try to match B */
8385
8386 #undef ST
8387 #define ST st->u.curly
8388
8389 #define CURLY_SETPAREN(paren, success) \
8390 if (paren) { \
8391 if (success) { \
8392 CLOSE_CAPTURE(paren, HOPc(locinput, -1) - reginfo->strbeg, \
8393 locinput - reginfo->strbeg); \
8394 } \
8395 else { \
8396 rex->offs[paren].end = -1; \
8397 rex->lastparen = ST.lastparen; \
8398 rex->lastcloseparen = ST.lastcloseparen; \
8399 } \
8400 }
8401
8402 case STAR: /* /A*B/ where A is width 1 char */
8403 ST.paren = 0;
8404 ST.min = 0;
8405 ST.max = REG_INFTY;
8406 scan = NEXTOPER(scan);
8407 goto repeat;
8408
8409 case PLUS: /* /A+B/ where A is width 1 char */
8410 ST.paren = 0;
8411 ST.min = 1;
8412 ST.max = REG_INFTY;
8413 scan = NEXTOPER(scan);
8414 goto repeat;
8415
8416 case CURLYN: /* /(A){m,n}B/ where A is width 1 char */
8417 ST.paren = scan->flags; /* Which paren to set */
8418 ST.lastparen = rex->lastparen;
8419 ST.lastcloseparen = rex->lastcloseparen;
8420 if (ST.paren > maxopenparen)
8421 maxopenparen = ST.paren;
8422 ST.min = ARG1(scan); /* min to match */
8423 ST.max = ARG2(scan); /* max to match */
8424 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
8425
8426 /* handle the single-char capture called as a GOSUB etc */
8427 if (EVAL_CLOSE_PAREN_IS_TRUE(cur_eval,(U32)ST.paren))
8428 {
8429 char *li = locinput;
8430 if (!regrepeat(rex, &li, scan, loceol, reginfo, 1))
8431 sayNO;
8432 SET_locinput(li);
8433 goto fake_end;
8434 }
8435
8436 goto repeat;
8437
8438 case CURLY: /* /A{m,n}B/ where A is width 1 char */
8439 ST.paren = 0;
8440 ST.min = ARG1(scan); /* min to match */
8441 ST.max = ARG2(scan); /* max to match */
8442 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
8443 repeat:
8444 /*
8445 * Lookahead to avoid useless match attempts
8446 * when we know what character comes next.
8447 *
8448 * Used to only do .*x and .*?x, but now it allows
8449 * for )'s, ('s and (?{ ... })'s to be in the way
8450 * of the quantifier and the EXACT-like node. -- japhy
8451 */
8452
8453 assert(ST.min <= ST.max);
8454 if (! HAS_TEXT(next) && ! JUMPABLE(next)) {
8455 ST.c1 = ST.c2 = CHRTEST_VOID;
8456 }
8457 else {
8458 regnode *text_node = next;
8459
8460 if (! HAS_TEXT(text_node))
8461 FIND_NEXT_IMPT(text_node);
8462
8463 if (! HAS_TEXT(text_node))
8464 ST.c1 = ST.c2 = CHRTEST_VOID;
8465 else {
8466 if ( PL_regkind[OP(text_node)] != EXACT ) {
8467 ST.c1 = ST.c2 = CHRTEST_VOID;
8468 }
8469 else {
8470 if (! S_setup_EXACTISH_ST_c1_c2(aTHX_
8471 text_node, &ST.c1, ST.c1_utf8, &ST.c2, ST.c2_utf8,
8472 reginfo))
8473 {
8474 sayNO;
8475 }
8476 }
8477 }
8478 }
8479
8480 ST.A = scan;
8481 ST.B = next;
8482 if (minmod) {
8483 char *li = locinput;
8484 minmod = 0;
8485 if (ST.min &&
8486 regrepeat(rex, &li, ST.A, loceol, reginfo, ST.min)
8487 < ST.min)
8488 sayNO;
8489 SET_locinput(li);
8490 ST.count = ST.min;
8491 REGCP_SET(ST.cp);
8492 if (ST.c1 == CHRTEST_VOID)
8493 goto curly_try_B_min;
8494
8495 ST.oldloc = locinput;
8496
8497 /* set ST.maxpos to the furthest point along the
8498 * string that could possibly match */
8499 if (ST.max == REG_INFTY) {
8500 ST.maxpos = loceol - 1;
8501 if (utf8_target)
8502 while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos))
8503 ST.maxpos--;
8504 }
8505 else if (utf8_target) {
8506 int m = ST.max - ST.min;
8507 for (ST.maxpos = locinput;
8508 m >0 && ST.maxpos < loceol; m--)
8509 ST.maxpos += UTF8SKIP(ST.maxpos);
8510 }
8511 else {
8512 ST.maxpos = locinput + ST.max - ST.min;
8513 if (ST.maxpos >= loceol)
8514 ST.maxpos = loceol - 1;
8515 }
8516 goto curly_try_B_min_known;
8517
8518 }
8519 else {
8520 /* avoid taking address of locinput, so it can remain
8521 * a register var */
8522 char *li = locinput;
8523 ST.count = regrepeat(rex, &li, ST.A, loceol, reginfo, ST.max);
8524 if (ST.count < ST.min)
8525 sayNO;
8526 SET_locinput(li);
8527 if ((ST.count > ST.min)
8528 && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL))
8529 {
8530 /* A{m,n} must come at the end of the string, there's
8531 * no point in backing off ... */
8532 ST.min = ST.count;
8533 /* ...except that $ and \Z can match before *and* after
8534 newline at the end. Consider "\n\n" =~ /\n+\Z\n/.
8535 We may back off by one in this case. */
8536 if (UCHARAT(locinput - 1) == '\n' && OP(ST.B) != EOS)
8537 ST.min--;
8538 }
8539 REGCP_SET(ST.cp);
8540 goto curly_try_B_max;
8541 }
8542 NOT_REACHED; /* NOTREACHED */
8543
8544 case CURLY_B_min_fail:
8545 /* failed to find B in a non-greedy match.
8546 * Handles both cases where c1,c2 valid or not */
8547
8548 REGCP_UNWIND(ST.cp);
8549 if (ST.paren) {
8550 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
8551 }
8552
8553 if (ST.c1 == CHRTEST_VOID) {
8554 /* failed -- move forward one */
8555 char *li = locinput;
8556 if (!regrepeat(rex, &li, ST.A, loceol, reginfo, 1)) {
8557 sayNO;
8558 }
8559 locinput = li;
8560 ST.count++;
8561 if (!( ST.count <= ST.max
8562 /* count overflow ? */
8563 || (ST.max == REG_INFTY && ST.count > 0))
8564 )
8565 sayNO;
8566 }
8567 else {
8568 int n;
8569 /* Couldn't or didn't -- move forward. */
8570 ST.oldloc = locinput;
8571 if (utf8_target)
8572 locinput += UTF8SKIP(locinput);
8573 else
8574 locinput++;
8575 ST.count++;
8576
8577 curly_try_B_min_known:
8578 /* find the next place where 'B' could work, then call B */
8579 if (utf8_target) {
8580 n = (ST.oldloc == locinput) ? 0 : 1;
8581 if (ST.c1 == ST.c2) {
8582 /* set n to utf8_distance(oldloc, locinput) */
8583 while ( locinput <= ST.maxpos
8584 && locinput < loceol
8585 && memNE(locinput, ST.c1_utf8,
8586 UTF8_SAFE_SKIP(locinput, reginfo->strend)))
8587 {
8588 locinput += UTF8_SAFE_SKIP(locinput,
8589 reginfo->strend);
8590 n++;
8591 }
8592 }
8593 else {
8594 /* set n to utf8_distance(oldloc, locinput) */
8595 while ( locinput <= ST.maxpos
8596 && locinput < loceol
8597 && memNE(locinput, ST.c1_utf8,
8598 UTF8_SAFE_SKIP(locinput, reginfo->strend))
8599 && memNE(locinput, ST.c2_utf8,
8600 UTF8_SAFE_SKIP(locinput, reginfo->strend)))
8601 {
8602 locinput += UTF8_SAFE_SKIP(locinput, reginfo->strend);
8603 n++;
8604 }
8605 }
8606 }
8607 else { /* Not utf8_target */
8608 if (ST.c1 == ST.c2) {
8609 locinput = (char *) memchr(locinput,
8610 ST.c1,
8611 ST.maxpos + 1 - locinput);
8612 if (! locinput) {
8613 locinput = ST.maxpos + 1;
8614 }
8615 }
8616 else {
8617 U8 c1_c2_bits_differing = ST.c1 ^ ST.c2;
8618
8619 if (! isPOWER_OF_2(c1_c2_bits_differing)) {
8620 while ( locinput <= ST.maxpos
8621 && UCHARAT(locinput) != ST.c1
8622 && UCHARAT(locinput) != ST.c2)
8623 {
8624 locinput++;
8625 }
8626 }
8627 else {
8628 /* If c1 and c2 only differ by a single bit, we can
8629 * avoid a conditional each time through the loop,
8630 * at the expense of a little preliminary setup and
8631 * an extra mask each iteration. By masking out
8632 * that bit, we match exactly two characters, c1
8633 * and c2, and so we don't have to test for both.
8634 * On both ASCII and EBCDIC platforms, most of the
8635 * ASCII-range and Latin1-range folded equivalents
8636 * differ only in a single bit, so this is actually
8637 * the most common case. (e.g. 'A' 0x41 vs 'a'
8638 * 0x61). */
8639 U8 c1_masked = ST.c1 &~ c1_c2_bits_differing;
8640 U8 c1_c2_mask = ~ c1_c2_bits_differing;
8641 while ( locinput <= ST.maxpos
8642 && (UCHARAT(locinput) & c1_c2_mask)
8643 != c1_masked)
8644 {
8645 locinput++;
8646 }
8647 }
8648 }
8649 n = locinput - ST.oldloc;
8650 }
8651 if (locinput > ST.maxpos)
8652 sayNO;
8653 if (n) {
8654 /* In /a{m,n}b/, ST.oldloc is at "a" x m, locinput is
8655 * at b; check that everything between oldloc and
8656 * locinput matches */
8657 char *li = ST.oldloc;
8658 ST.count += n;
8659 if (regrepeat(rex, &li, ST.A, loceol, reginfo, n) < n)
8660 sayNO;
8661 assert(n == REG_INFTY || locinput == li);
8662 }
8663 }
8664
8665 curly_try_B_min:
8666 CURLY_SETPAREN(ST.paren, ST.count);
8667 PUSH_STATE_GOTO(CURLY_B_min, ST.B, locinput, loceol,
8668 script_run_begin);
8669 NOT_REACHED; /* NOTREACHED */
8670
8671
8672 curly_try_B_max:
8673 /* a successful greedy match: now try to match B */
8674 {
8675 bool could_match = locinput < loceol;
8676
8677 /* If it could work, try it. */
8678 if (ST.c1 != CHRTEST_VOID && could_match) {
8679 if (! UTF8_IS_INVARIANT(UCHARAT(locinput)) && utf8_target)
8680 {
8681 could_match = memEQ(locinput, ST.c1_utf8,
8682 UTF8_SAFE_SKIP(locinput,
8683 reginfo->strend))
8684 || memEQ(locinput, ST.c2_utf8,
8685 UTF8_SAFE_SKIP(locinput,
8686 reginfo->strend));
8687 }
8688 else {
8689 could_match = UCHARAT(locinput) == ST.c1
8690 || UCHARAT(locinput) == ST.c2;
8691 }
8692 }
8693 if (ST.c1 == CHRTEST_VOID || could_match) {
8694 CURLY_SETPAREN(ST.paren, ST.count);
8695 PUSH_STATE_GOTO(CURLY_B_max, ST.B, locinput, loceol,
8696 script_run_begin);
8697 NOT_REACHED; /* NOTREACHED */
8698 }
8699 }
8700 /* FALLTHROUGH */
8701
8702 case CURLY_B_max_fail:
8703 /* failed to find B in a greedy match */
8704
8705 REGCP_UNWIND(ST.cp);
8706 if (ST.paren) {
8707 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
8708 }
8709 /* back up. */
8710 if (--ST.count < ST.min)
8711 sayNO;
8712 locinput = HOPc(locinput, -1);
8713 goto curly_try_B_max;
8714
8715 #undef ST
8716
8717 case END: /* last op of main pattern */
8718 fake_end:
8719 if (cur_eval) {
8720 /* we've just finished A in /(??{A})B/; now continue with B */
8721 SET_RECURSE_LOCINPUT("FAKE-END[before]", CUR_EVAL.prev_recurse_locinput);
8722 st->u.eval.prev_rex = rex_sv; /* inner */
8723
8724 /* Save *all* the positions. */
8725 st->u.eval.cp = regcppush(rex, 0, maxopenparen);
8726 rex_sv = CUR_EVAL.prev_rex;
8727 is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(rex_sv));
8728 SET_reg_curpm(rex_sv);
8729 rex = ReANY(rex_sv);
8730 rexi = RXi_GET(rex);
8731
8732 st->u.eval.prev_curlyx = cur_curlyx;
8733 cur_curlyx = CUR_EVAL.prev_curlyx;
8734
8735 REGCP_SET(st->u.eval.lastcp);
8736
8737 /* Restore parens of the outer rex without popping the
8738 * savestack */
8739 regcp_restore(rex, CUR_EVAL.lastcp, &maxopenparen);
8740
8741 st->u.eval.prev_eval = cur_eval;
8742 cur_eval = CUR_EVAL.prev_eval;
8743 DEBUG_EXECUTE_r(
8744 Perl_re_exec_indentf( aTHX_ "END: EVAL trying tail ... (cur_eval=%p)\n",
8745 depth, cur_eval););
8746 if ( nochange_depth )
8747 nochange_depth--;
8748
8749 SET_RECURSE_LOCINPUT("FAKE-END[after]", cur_eval->locinput);
8750
8751 PUSH_YES_STATE_GOTO(EVAL_postponed_AB, /* match B */
8752 st->u.eval.prev_eval->u.eval.B,
8753 locinput, loceol, script_run_begin);
8754 }
8755
8756 if (locinput < reginfo->till) {
8757 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_
8758 "%sEND: Match possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
8759 PL_colors[4],
8760 (long)(locinput - startpos),
8761 (long)(reginfo->till - startpos),
8762 PL_colors[5]));
8763
8764 sayNO_SILENT; /* Cannot match: too short. */
8765 }
8766 sayYES; /* Success! */
8767
8768 case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */
8769 DEBUG_EXECUTE_r(
8770 Perl_re_exec_indentf( aTHX_ "%sSUCCEED: subpattern success...%s\n",
8771 depth, PL_colors[4], PL_colors[5]));
8772 sayYES; /* Success! */
8773
8774 #undef ST
8775 #define ST st->u.ifmatch
8776
8777 case SUSPEND: /* (?>A) */
8778 ST.wanted = 1;
8779 ST.start = locinput;
8780 ST.end = loceol;
8781 ST.count = 1;
8782 goto do_ifmatch;
8783
8784 case UNLESSM: /* -ve lookaround: (?!A), or with 'flags', (?<!A) */
8785 ST.wanted = 0;
8786 goto ifmatch_trivial_fail_test;
8787
8788 case IFMATCH: /* +ve lookaround: (?=A), or with 'flags', (?<=A) */
8789 ST.wanted = 1;
8790 ifmatch_trivial_fail_test:
8791 ST.count = scan->next_off + 1; /* next_off repurposed to be
8792 lookbehind count, requires
8793 non-zero flags */
8794 if (! scan->flags) { /* 'flags' zero means lookahed */
8795
8796 /* Lookahead starts here and ends at the normal place */
8797 ST.start = locinput;
8798 ST.end = loceol;
8799 }
8800 else {
8801 PERL_UINT_FAST8_T back_count = scan->flags;
8802 char * s;
8803
8804 /* Lookbehind can look beyond the current position */
8805 ST.end = loceol;
8806
8807 /* ... and starts at the first place in the input that is in
8808 * the range of the possible start positions */
8809 for (; ST.count > 0; ST.count--, back_count--) {
8810 s = HOPBACKc(locinput, back_count);
8811 if (s) {
8812 ST.start = s;
8813 goto do_ifmatch;
8814 }
8815 }
8816
8817 /* If the lookbehind doesn't start in the actual string, is a
8818 * trivial match failure */
8819 if (logical) {
8820 logical = 0;
8821 sw = 1 - cBOOL(ST.wanted);
8822 }
8823 else if (ST.wanted)
8824 sayNO;
8825
8826 /* Here, we didn't want it to match, so is actually success */
8827 next = scan + ARG(scan);
8828 if (next == scan)
8829 next = NULL;
8830 break;
8831 }
8832
8833 do_ifmatch:
8834 ST.me = scan;
8835 ST.logical = logical;
8836 logical = 0; /* XXX: reset state of logical once it has been saved into ST */
8837
8838 /* execute body of (?...A) */
8839 PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan)), ST.start,
8840 ST.end, script_run_begin);
8841 NOT_REACHED; /* NOTREACHED */
8842
8843 {
8844 bool matched;
8845
8846 case IFMATCH_A_fail: /* body of (?...A) failed */
8847 if (! ST.logical && ST.count > 1) {
8848
8849 /* It isn't a real failure until we've tried all starting
8850 * positions. Move to the next starting position and retry */
8851 ST.count--;
8852 ST.start = HOPc(ST.start, 1);
8853 scan = ST.me;
8854 logical = ST.logical;
8855 goto do_ifmatch;
8856 }
8857
8858 /* Here, all starting positions have been tried. */
8859 matched = FALSE;
8860 goto ifmatch_done;
8861
8862 case IFMATCH_A: /* body of (?...A) succeeded */
8863 matched = TRUE;
8864 ifmatch_done:
8865 sw = matched == ST.wanted;
8866 if (! ST.logical && !sw) {
8867 sayNO;
8868 }
8869
8870 if (OP(ST.me) != SUSPEND) {
8871 /* restore old position except for (?>...) */
8872 locinput = st->locinput;
8873 loceol = st->loceol;
8874 script_run_begin = st->sr0;
8875 }
8876 scan = ST.me + ARG(ST.me);
8877 if (scan == ST.me)
8878 scan = NULL;
8879 continue; /* execute B */
8880 }
8881
8882 #undef ST
8883
8884 case LONGJMP: /* alternative with many branches compiles to
8885 * (BRANCHJ; EXACT ...; LONGJMP ) x N */
8886 next = scan + ARG(scan);
8887 if (next == scan)
8888 next = NULL;
8889 break;
8890
8891 case COMMIT: /* (*COMMIT) */
8892 reginfo->cutpoint = loceol;
8893 /* FALLTHROUGH */
8894
8895 case PRUNE: /* (*PRUNE) */
8896 if (scan->flags)
8897 sv_yes_mark = sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
8898 PUSH_STATE_GOTO(COMMIT_next, next, locinput, loceol,
8899 script_run_begin);
8900 NOT_REACHED; /* NOTREACHED */
8901
8902 case COMMIT_next_fail:
8903 no_final = 1;
8904 /* FALLTHROUGH */
8905 sayNO;
8906 NOT_REACHED; /* NOTREACHED */
8907
8908 case OPFAIL: /* (*FAIL) */
8909 if (scan->flags)
8910 sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
8911 if (logical) {
8912 /* deal with (?(?!)X|Y) properly,
8913 * make sure we trigger the no branch
8914 * of the trailing IFTHEN structure*/
8915 sw= 0;
8916 break;
8917 } else {
8918 sayNO;
8919 }
8920 NOT_REACHED; /* NOTREACHED */
8921
8922 #define ST st->u.mark
8923 case MARKPOINT: /* (*MARK:foo) */
8924 ST.prev_mark = mark_state;
8925 ST.mark_name = sv_commit = sv_yes_mark
8926 = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
8927 mark_state = st;
8928 ST.mark_loc = locinput;
8929 PUSH_YES_STATE_GOTO(MARKPOINT_next, next, locinput, loceol,
8930 script_run_begin);
8931 NOT_REACHED; /* NOTREACHED */
8932
8933 case MARKPOINT_next:
8934 mark_state = ST.prev_mark;
8935 sayYES;
8936 NOT_REACHED; /* NOTREACHED */
8937
8938 case MARKPOINT_next_fail:
8939 if (popmark && sv_eq(ST.mark_name,popmark))
8940 {
8941 if (ST.mark_loc > startpoint)
8942 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
8943 popmark = NULL; /* we found our mark */
8944 sv_commit = ST.mark_name;
8945
8946 DEBUG_EXECUTE_r({
8947 Perl_re_exec_indentf( aTHX_ "%sMARKPOINT: next fail: setting cutpoint to mark:%" SVf "...%s\n",
8948 depth,
8949 PL_colors[4], SVfARG(sv_commit), PL_colors[5]);
8950 });
8951 }
8952 mark_state = ST.prev_mark;
8953 sv_yes_mark = mark_state ?
8954 mark_state->u.mark.mark_name : NULL;
8955 sayNO;
8956 NOT_REACHED; /* NOTREACHED */
8957
8958 case SKIP: /* (*SKIP) */
8959 if (!scan->flags) {
8960 /* (*SKIP) : if we fail we cut here*/
8961 ST.mark_name = NULL;
8962 ST.mark_loc = locinput;
8963 PUSH_STATE_GOTO(SKIP_next,next, locinput, loceol,
8964 script_run_begin);
8965 } else {
8966 /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was,
8967 otherwise do nothing. Meaning we need to scan
8968 */
8969 regmatch_state *cur = mark_state;
8970 SV *find = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
8971
8972 while (cur) {
8973 if ( sv_eq( cur->u.mark.mark_name,
8974 find ) )
8975 {
8976 ST.mark_name = find;
8977 PUSH_STATE_GOTO( SKIP_next, next, locinput, loceol,
8978 script_run_begin);
8979 }
8980 cur = cur->u.mark.prev_mark;
8981 }
8982 }
8983 /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */
8984 break;
8985
8986 case SKIP_next_fail:
8987 if (ST.mark_name) {
8988 /* (*CUT:NAME) - Set up to search for the name as we
8989 collapse the stack*/
8990 popmark = ST.mark_name;
8991 } else {
8992 /* (*CUT) - No name, we cut here.*/
8993 if (ST.mark_loc > startpoint)
8994 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
8995 /* but we set sv_commit to latest mark_name if there
8996 is one so they can test to see how things lead to this
8997 cut */
8998 if (mark_state)
8999 sv_commit=mark_state->u.mark.mark_name;
9000 }
9001 no_final = 1;
9002 sayNO;
9003 NOT_REACHED; /* NOTREACHED */
9004 #undef ST
9005
9006 case LNBREAK: /* \R */
9007 if ((n=is_LNBREAK_safe(locinput, loceol, utf8_target))) {
9008 locinput += n;
9009 } else
9010 sayNO;
9011 break;
9012
9013 default:
9014 PerlIO_printf(Perl_error_log, "%" UVxf " %d\n",
9015 PTR2UV(scan), OP(scan));
9016 Perl_croak(aTHX_ "regexp memory corruption");
9017
9018 /* this is a point to jump to in order to increment
9019 * locinput by one character */
9020 increment_locinput:
9021 assert(!NEXTCHR_IS_EOS);
9022 if (utf8_target) {
9023 locinput += PL_utf8skip[nextchr];
9024 /* locinput is allowed to go 1 char off the end (signifying
9025 * EOS), but not 2+ */
9026 if (locinput > loceol)
9027 sayNO;
9028 }
9029 else
9030 locinput++;
9031 break;
9032
9033 } /* end switch */
9034
9035 /* switch break jumps here */
9036 scan = next; /* prepare to execute the next op and ... */
9037 continue; /* ... jump back to the top, reusing st */
9038 /* NOTREACHED */
9039
9040 push_yes_state:
9041 /* push a state that backtracks on success */
9042 st->u.yes.prev_yes_state = yes_state;
9043 yes_state = st;
9044 /* FALLTHROUGH */
9045 push_state:
9046 /* push a new regex state, then continue at scan */
9047 {
9048 regmatch_state *newst;
9049
9050 DEBUG_STACK_r({
9051 regmatch_state *cur = st;
9052 regmatch_state *curyes = yes_state;
9053 U32 i;
9054 regmatch_slab *slab = PL_regmatch_slab;
9055 for (i = 0; i < 3 && i <= depth; cur--,i++) {
9056 if (cur < SLAB_FIRST(slab)) {
9057 slab = slab->prev;
9058 cur = SLAB_LAST(slab);
9059 }
9060 Perl_re_exec_indentf( aTHX_ "%4s #%-3d %-10s %s\n",
9061 depth,
9062 i ? " " : "push",
9063 depth - i, PL_reg_name[cur->resume_state],
9064 (curyes == cur) ? "yes" : ""
9065 );
9066 if (curyes == cur)
9067 curyes = cur->u.yes.prev_yes_state;
9068 }
9069 } else
9070 DEBUG_STATE_pp("push")
9071 );
9072 depth++;
9073 st->locinput = locinput;
9074 st->loceol = loceol;
9075 st->sr0 = script_run_begin;
9076 newst = st+1;
9077 if (newst > SLAB_LAST(PL_regmatch_slab))
9078 newst = S_push_slab(aTHX);
9079 PL_regmatch_state = newst;
9080
9081 locinput = pushinput;
9082 loceol = pusheol;
9083 script_run_begin = pushsr0;
9084 st = newst;
9085 continue;
9086 /* NOTREACHED */
9087 }
9088 }
9089 #ifdef SOLARIS_BAD_OPTIMIZER
9090 # undef PL_charclass
9091 #endif
9092
9093 /*
9094 * We get here only if there's trouble -- normally "case END" is
9095 * the terminating point.
9096 */
9097 Perl_croak(aTHX_ "corrupted regexp pointers");
9098 NOT_REACHED; /* NOTREACHED */
9099
9100 yes:
9101 if (yes_state) {
9102 /* we have successfully completed a subexpression, but we must now
9103 * pop to the state marked by yes_state and continue from there */
9104 assert(st != yes_state);
9105 #ifdef DEBUGGING
9106 while (st != yes_state) {
9107 st--;
9108 if (st < SLAB_FIRST(PL_regmatch_slab)) {
9109 PL_regmatch_slab = PL_regmatch_slab->prev;
9110 st = SLAB_LAST(PL_regmatch_slab);
9111 }
9112 DEBUG_STATE_r({
9113 if (no_final) {
9114 DEBUG_STATE_pp("pop (no final)");
9115 } else {
9116 DEBUG_STATE_pp("pop (yes)");
9117 }
9118 });
9119 depth--;
9120 }
9121 #else
9122 while (yes_state < SLAB_FIRST(PL_regmatch_slab)
9123 || yes_state > SLAB_LAST(PL_regmatch_slab))
9124 {
9125 /* not in this slab, pop slab */
9126 depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1);
9127 PL_regmatch_slab = PL_regmatch_slab->prev;
9128 st = SLAB_LAST(PL_regmatch_slab);
9129 }
9130 depth -= (st - yes_state);
9131 #endif
9132 st = yes_state;
9133 yes_state = st->u.yes.prev_yes_state;
9134 PL_regmatch_state = st;
9135
9136 if (no_final) {
9137 locinput= st->locinput;
9138 loceol= st->loceol;
9139 script_run_begin = st->sr0;
9140 }
9141 state_num = st->resume_state + no_final;
9142 goto reenter_switch;
9143 }
9144
9145 DEBUG_EXECUTE_r(Perl_re_printf( aTHX_ "%sMatch successful!%s\n",
9146 PL_colors[4], PL_colors[5]));
9147
9148 if (reginfo->info_aux_eval) {
9149 /* each successfully executed (?{...}) block does the equivalent of
9150 * local $^R = do {...}
9151 * When popping the save stack, all these locals would be undone;
9152 * bypass this by setting the outermost saved $^R to the latest
9153 * value */
9154 /* I dont know if this is needed or works properly now.
9155 * see code related to PL_replgv elsewhere in this file.
9156 * Yves
9157 */
9158 if (oreplsv != GvSV(PL_replgv)) {
9159 sv_setsv(oreplsv, GvSV(PL_replgv));
9160 SvSETMAGIC(oreplsv);
9161 }
9162 }
9163 result = 1;
9164 goto final_exit;
9165
9166 no:
9167 DEBUG_EXECUTE_r(
9168 Perl_re_exec_indentf( aTHX_ "%sfailed...%s\n",
9169 depth,
9170 PL_colors[4], PL_colors[5])
9171 );
9172
9173 no_silent:
9174 if (no_final) {
9175 if (yes_state) {
9176 goto yes;
9177 } else {
9178 goto final_exit;
9179 }
9180 }
9181 if (depth) {
9182 /* there's a previous state to backtrack to */
9183 st--;
9184 if (st < SLAB_FIRST(PL_regmatch_slab)) {
9185 PL_regmatch_slab = PL_regmatch_slab->prev;
9186 st = SLAB_LAST(PL_regmatch_slab);
9187 }
9188 PL_regmatch_state = st;
9189 locinput= st->locinput;
9190 loceol= st->loceol;
9191 script_run_begin = st->sr0;
9192
9193 DEBUG_STATE_pp("pop");
9194 depth--;
9195 if (yes_state == st)
9196 yes_state = st->u.yes.prev_yes_state;
9197
9198 state_num = st->resume_state + 1; /* failure = success + 1 */
9199 PERL_ASYNC_CHECK();
9200 goto reenter_switch;
9201 }
9202 result = 0;
9203
9204 final_exit:
9205 if (rex->intflags & PREGf_VERBARG_SEEN) {
9206 SV *sv_err = get_sv("REGERROR", 1);
9207 SV *sv_mrk = get_sv("REGMARK", 1);
9208 if (result) {
9209 sv_commit = &PL_sv_no;
9210 if (!sv_yes_mark)
9211 sv_yes_mark = &PL_sv_yes;
9212 } else {
9213 if (!sv_commit)
9214 sv_commit = &PL_sv_yes;
9215 sv_yes_mark = &PL_sv_no;
9216 }
9217 assert(sv_err);
9218 assert(sv_mrk);
9219 sv_setsv(sv_err, sv_commit);
9220 sv_setsv(sv_mrk, sv_yes_mark);
9221 }
9222
9223
9224 if (last_pushed_cv) {
9225 dSP;
9226 /* see "Some notes about MULTICALL" above */
9227 POP_MULTICALL;
9228 PERL_UNUSED_VAR(SP);
9229 }
9230 else
9231 LEAVE_SCOPE(orig_savestack_ix);
9232
9233 assert(!result || locinput - reginfo->strbeg >= 0);
9234 return result ? locinput - reginfo->strbeg : -1;
9235 }
9236
9237 /*
9238 - regrepeat - repeatedly match something simple, report how many
9239 *
9240 * What 'simple' means is a node which can be the operand of a quantifier like
9241 * '+', or {1,3}
9242 *
9243 * startposp - pointer to a pointer to the start position. This is updated
9244 * to point to the byte following the highest successful
9245 * match.
9246 * p - the regnode to be repeatedly matched against.
9247 * loceol - pointer to the end position beyond which we aren't supposed to
9248 * look.
9249 * reginfo - struct holding match state, such as utf8_target
9250 * max - maximum number of things to match.
9251 * depth - (for debugging) backtracking depth.
9252 */
9253 STATIC I32
S_regrepeat(pTHX_ regexp * prog,char ** startposp,const regnode * p,char * loceol,regmatch_info * const reginfo,I32 max _pDEPTH)9254 S_regrepeat(pTHX_ regexp *prog, char **startposp, const regnode *p,
9255 char * loceol, regmatch_info *const reginfo, I32 max _pDEPTH)
9256 {
9257 dVAR;
9258 char *scan; /* Pointer to current position in target string */
9259 I32 c;
9260 char *this_eol = loceol; /* potentially adjusted version. */
9261 I32 hardcount = 0; /* How many matches so far */
9262 bool utf8_target = reginfo->is_utf8_target;
9263 unsigned int to_complement = 0; /* Invert the result? */
9264 UV utf8_flags = 0;
9265 _char_class_number classnum;
9266
9267 PERL_ARGS_ASSERT_REGREPEAT;
9268
9269 /* This routine is structured so that we switch on the input OP. Each OP
9270 * case: statement contains a loop to repeatedly apply the OP, advancing
9271 * the input until it fails, or reaches the end of the input, or until it
9272 * reaches the upper limit of matches. */
9273
9274 scan = *startposp;
9275 if (max == REG_INFTY) /* This is a special marker to go to the platform's
9276 max */
9277 max = I32_MAX;
9278 else if (! utf8_target && this_eol - scan > max)
9279 this_eol = scan + max;
9280
9281 /* Here, for the case of a non-UTF-8 target we have adjusted <this_eol> down
9282 * to the maximum of how far we should go in it (leaving it set to the real
9283 * end, if the maximum permissible would take us beyond that). This allows
9284 * us to make the loop exit condition that we haven't gone past <this_eol> to
9285 * also mean that we haven't exceeded the max permissible count, saving a
9286 * test each time through the loops. But it assumes that the OP matches a
9287 * single byte, which is true for most of the OPs below when applied to a
9288 * non-UTF-8 target. Those relatively few OPs that don't have this
9289 * characteristic will have to compensate.
9290 *
9291 * There is no adjustment for UTF-8 targets, as the number of bytes per
9292 * character varies. OPs will have to test both that the count is less
9293 * than the max permissible (using <hardcount> to keep track), and that we
9294 * are still within the bounds of the string (using <this_eol>. A few OPs
9295 * match a single byte no matter what the encoding. They can omit the max
9296 * test if, for the UTF-8 case, they do the adjustment that was skipped
9297 * above.
9298 *
9299 * Thus, the code above sets things up for the common case; and exceptional
9300 * cases need extra work; the common case is to make sure <scan> doesn't
9301 * go past <this_eol>, and for UTF-8 to also use <hardcount> to make sure the
9302 * count doesn't exceed the maximum permissible */
9303
9304 switch (OP(p)) {
9305 case REG_ANY:
9306 if (utf8_target) {
9307 while (scan < this_eol && hardcount < max && *scan != '\n') {
9308 scan += UTF8SKIP(scan);
9309 hardcount++;
9310 }
9311 } else {
9312 scan = (char *) memchr(scan, '\n', this_eol - scan);
9313 if (! scan) {
9314 scan = this_eol;
9315 }
9316 }
9317 break;
9318 case SANY:
9319 if (utf8_target) {
9320 while (scan < this_eol && hardcount < max) {
9321 scan += UTF8SKIP(scan);
9322 hardcount++;
9323 }
9324 }
9325 else
9326 scan = this_eol;
9327 break;
9328 case EXACTL:
9329 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
9330 if (utf8_target && UTF8_IS_ABOVE_LATIN1(*scan)) {
9331 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(scan, loceol);
9332 }
9333 goto do_exact;
9334
9335 case EXACT_ONLY8:
9336 if (! utf8_target) {
9337 break;
9338 }
9339 /* FALLTHROUGH */
9340 case EXACT:
9341 do_exact:
9342 assert(STR_LEN(p) == reginfo->is_utf8_pat ? UTF8SKIP(STRING(p)) : 1);
9343
9344 c = (U8)*STRING(p);
9345
9346 /* Can use a simple find if the pattern char to match on is invariant
9347 * under UTF-8, or both target and pattern aren't UTF-8. Note that we
9348 * can use UTF8_IS_INVARIANT() even if the pattern isn't UTF-8, as it's
9349 * true iff it doesn't matter if the argument is in UTF-8 or not */
9350 if (UTF8_IS_INVARIANT(c) || (! utf8_target && ! reginfo->is_utf8_pat)) {
9351 if (utf8_target && this_eol - scan > max) {
9352 /* We didn't adjust <this_eol> because is UTF-8, but ok to do so,
9353 * since here, to match at all, 1 char == 1 byte */
9354 this_eol = scan + max;
9355 }
9356 scan = (char *) find_span_end((U8 *) scan, (U8 *) this_eol, (U8) c);
9357 }
9358 else if (reginfo->is_utf8_pat) {
9359 if (utf8_target) {
9360 STRLEN scan_char_len;
9361
9362 /* When both target and pattern are UTF-8, we have to do
9363 * string EQ */
9364 while (hardcount < max
9365 && scan < this_eol
9366 && (scan_char_len = UTF8SKIP(scan)) <= STR_LEN(p)
9367 && memEQ(scan, STRING(p), scan_char_len))
9368 {
9369 scan += scan_char_len;
9370 hardcount++;
9371 }
9372 }
9373 else if (! UTF8_IS_ABOVE_LATIN1(c)) {
9374
9375 /* Target isn't utf8; convert the character in the UTF-8
9376 * pattern to non-UTF8, and do a simple find */
9377 c = EIGHT_BIT_UTF8_TO_NATIVE(c, *(STRING(p) + 1));
9378 scan = (char *) find_span_end((U8 *) scan, (U8 *) this_eol, (U8) c);
9379 } /* else pattern char is above Latin1, can't possibly match the
9380 non-UTF-8 target */
9381 }
9382 else {
9383
9384 /* Here, the string must be utf8; pattern isn't, and <c> is
9385 * different in utf8 than not, so can't compare them directly.
9386 * Outside the loop, find the two utf8 bytes that represent c, and
9387 * then look for those in sequence in the utf8 string */
9388 U8 high = UTF8_TWO_BYTE_HI(c);
9389 U8 low = UTF8_TWO_BYTE_LO(c);
9390
9391 while (hardcount < max
9392 && scan + 1 < this_eol
9393 && UCHARAT(scan) == high
9394 && UCHARAT(scan + 1) == low)
9395 {
9396 scan += 2;
9397 hardcount++;
9398 }
9399 }
9400 break;
9401
9402 case EXACTFAA_NO_TRIE: /* This node only generated for non-utf8 patterns */
9403 assert(! reginfo->is_utf8_pat);
9404 /* FALLTHROUGH */
9405 case EXACTFAA:
9406 utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
9407 if (reginfo->is_utf8_pat || ! utf8_target) {
9408
9409 /* The possible presence of a MICRO SIGN in the pattern forbids us
9410 * to view a non-UTF-8 pattern as folded when there is a UTF-8
9411 * target. */
9412 utf8_flags |= FOLDEQ_S2_ALREADY_FOLDED|FOLDEQ_S2_FOLDS_SANE;
9413 }
9414 goto do_exactf;
9415
9416 case EXACTFL:
9417 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
9418 utf8_flags = FOLDEQ_LOCALE;
9419 goto do_exactf;
9420
9421 case EXACTF: /* This node only generated for non-utf8 patterns */
9422 assert(! reginfo->is_utf8_pat);
9423 goto do_exactf;
9424
9425 case EXACTFLU8:
9426 if (! utf8_target) {
9427 break;
9428 }
9429 utf8_flags = FOLDEQ_LOCALE | FOLDEQ_S2_ALREADY_FOLDED
9430 | FOLDEQ_S2_FOLDS_SANE;
9431 goto do_exactf;
9432
9433 case EXACTFU_ONLY8:
9434 if (! utf8_target) {
9435 break;
9436 }
9437 assert(reginfo->is_utf8_pat);
9438 utf8_flags = FOLDEQ_S2_ALREADY_FOLDED;
9439 goto do_exactf;
9440
9441 case EXACTFU:
9442 utf8_flags = FOLDEQ_S2_ALREADY_FOLDED;
9443 /* FALLTHROUGH */
9444
9445 case EXACTFUP:
9446
9447 do_exactf: {
9448 int c1, c2;
9449 U8 c1_utf8[UTF8_MAXBYTES+1], c2_utf8[UTF8_MAXBYTES+1];
9450
9451 assert(STR_LEN(p) == reginfo->is_utf8_pat ? UTF8SKIP(STRING(p)) : 1);
9452
9453 if (S_setup_EXACTISH_ST_c1_c2(aTHX_ p, &c1, c1_utf8, &c2, c2_utf8,
9454 reginfo))
9455 {
9456 if (c1 == CHRTEST_VOID) {
9457 /* Use full Unicode fold matching */
9458 char *tmpeol = loceol;
9459 STRLEN pat_len = reginfo->is_utf8_pat ? UTF8SKIP(STRING(p)) : 1;
9460 while (hardcount < max
9461 && foldEQ_utf8_flags(scan, &tmpeol, 0, utf8_target,
9462 STRING(p), NULL, pat_len,
9463 reginfo->is_utf8_pat, utf8_flags))
9464 {
9465 scan = tmpeol;
9466 tmpeol = loceol;
9467 hardcount++;
9468 }
9469 }
9470 else if (utf8_target) {
9471 if (c1 == c2) {
9472 while (scan < this_eol
9473 && hardcount < max
9474 && memEQ(scan, c1_utf8, UTF8_SAFE_SKIP(scan,
9475 loceol)))
9476 {
9477 scan += UTF8SKIP(c1_utf8);
9478 hardcount++;
9479 }
9480 }
9481 else {
9482 while (scan < this_eol
9483 && hardcount < max
9484 && ( memEQ(scan, c1_utf8, UTF8_SAFE_SKIP(scan,
9485 loceol))
9486 || memEQ(scan, c2_utf8, UTF8_SAFE_SKIP(scan,
9487 loceol))))
9488 {
9489 scan += UTF8_SAFE_SKIP(scan, loceol);
9490 hardcount++;
9491 }
9492 }
9493 }
9494 else if (c1 == c2) {
9495 scan = (char *) find_span_end((U8 *) scan, (U8 *) this_eol, (U8) c1);
9496 }
9497 else {
9498 /* See comments in regmatch() CURLY_B_min_known_fail. We avoid
9499 * a conditional each time through the loop if the characters
9500 * differ only in a single bit, as is the usual situation */
9501 U8 c1_c2_bits_differing = c1 ^ c2;
9502
9503 if (isPOWER_OF_2(c1_c2_bits_differing)) {
9504 U8 c1_c2_mask = ~ c1_c2_bits_differing;
9505
9506 scan = (char *) find_span_end_mask((U8 *) scan,
9507 (U8 *) this_eol,
9508 c1 & c1_c2_mask,
9509 c1_c2_mask);
9510 }
9511 else {
9512 while ( scan < this_eol
9513 && (UCHARAT(scan) == c1 || UCHARAT(scan) == c2))
9514 {
9515 scan++;
9516 }
9517 }
9518 }
9519 }
9520 break;
9521 }
9522 case ANYOFPOSIXL:
9523 case ANYOFL:
9524 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
9525
9526 if (ANYOFL_UTF8_LOCALE_REQD(FLAGS(p)) && ! IN_UTF8_CTYPE_LOCALE) {
9527 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE), utf8_locale_required);
9528 }
9529 /* FALLTHROUGH */
9530 case ANYOFD:
9531 case ANYOF:
9532 if (utf8_target) {
9533 while (hardcount < max
9534 && scan < this_eol
9535 && reginclass(prog, p, (U8*)scan, (U8*) this_eol, utf8_target))
9536 {
9537 scan += UTF8SKIP(scan);
9538 hardcount++;
9539 }
9540 }
9541 else if (ANYOF_FLAGS(p) & ~ ANYOF_MATCHES_ALL_ABOVE_BITMAP) {
9542 while (scan < this_eol
9543 && reginclass(prog, p, (U8*)scan, (U8*)scan+1, 0))
9544 scan++;
9545 }
9546 else {
9547 while (scan < this_eol && ANYOF_BITMAP_TEST(p, *((U8*)scan)))
9548 scan++;
9549 }
9550 break;
9551
9552 case ANYOFM:
9553 if (utf8_target && this_eol - scan > max) {
9554
9555 /* We didn't adjust <this_eol> at the beginning of this routine
9556 * because is UTF-8, but it is actually ok to do so, since here, to
9557 * match, 1 char == 1 byte. */
9558 this_eol = scan + max;
9559 }
9560
9561 scan = (char *) find_span_end_mask((U8 *) scan, (U8 *) this_eol, (U8) ARG(p), FLAGS(p));
9562 break;
9563
9564 case NANYOFM:
9565 if (utf8_target) {
9566 while ( hardcount < max
9567 && scan < this_eol
9568 && (*scan & FLAGS(p)) != ARG(p))
9569 {
9570 scan += UTF8SKIP(scan);
9571 hardcount++;
9572 }
9573 }
9574 else {
9575 scan = (char *) find_next_masked((U8 *) scan, (U8 *) this_eol, (U8) ARG(p), FLAGS(p));
9576 }
9577 break;
9578
9579 case ANYOFH:
9580 if (utf8_target) { /* ANYOFH only can match UTF-8 targets */
9581 if (ANYOF_FLAGS(p)) { /* If we know the first byte of what
9582 matches, we can avoid calling reginclass
9583 */
9584 while ( hardcount < max
9585 && scan < this_eol
9586 && (U8) *scan == ANYOF_FLAGS(p)
9587 && reginclass(prog, p, (U8*)scan, (U8*) this_eol,
9588 TRUE))
9589 {
9590 scan += UTF8SKIP(scan);
9591 hardcount++;
9592 }
9593 }
9594 else while ( hardcount < max
9595 && scan < this_eol
9596 && reginclass(prog, p, (U8*)scan, (U8*) this_eol, TRUE))
9597 {
9598 scan += UTF8SKIP(scan);
9599 hardcount++;
9600 }
9601 }
9602 break;
9603
9604 /* The argument (FLAGS) to all the POSIX node types is the class number */
9605
9606 case NPOSIXL:
9607 to_complement = 1;
9608 /* FALLTHROUGH */
9609
9610 case POSIXL:
9611 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
9612 if (! utf8_target) {
9613 while (scan < this_eol && to_complement ^ cBOOL(isFOO_lc(FLAGS(p),
9614 *scan)))
9615 {
9616 scan++;
9617 }
9618 } else {
9619 while (hardcount < max && scan < this_eol
9620 && to_complement ^ cBOOL(isFOO_utf8_lc(FLAGS(p),
9621 (U8 *) scan,
9622 (U8 *) this_eol)))
9623 {
9624 scan += UTF8SKIP(scan);
9625 hardcount++;
9626 }
9627 }
9628 break;
9629
9630 case POSIXD:
9631 if (utf8_target) {
9632 goto utf8_posix;
9633 }
9634 /* FALLTHROUGH */
9635
9636 case POSIXA:
9637 if (utf8_target && this_eol - scan > max) {
9638
9639 /* We didn't adjust <this_eol> at the beginning of this routine
9640 * because is UTF-8, but it is actually ok to do so, since here, to
9641 * match, 1 char == 1 byte. */
9642 this_eol = scan + max;
9643 }
9644 while (scan < this_eol && _generic_isCC_A((U8) *scan, FLAGS(p))) {
9645 scan++;
9646 }
9647 break;
9648
9649 case NPOSIXD:
9650 if (utf8_target) {
9651 to_complement = 1;
9652 goto utf8_posix;
9653 }
9654 /* FALLTHROUGH */
9655
9656 case NPOSIXA:
9657 if (! utf8_target) {
9658 while (scan < this_eol && ! _generic_isCC_A((U8) *scan, FLAGS(p))) {
9659 scan++;
9660 }
9661 }
9662 else {
9663
9664 /* The complement of something that matches only ASCII matches all
9665 * non-ASCII, plus everything in ASCII that isn't in the class. */
9666 while (hardcount < max && scan < this_eol
9667 && ( ! isASCII_utf8_safe(scan, loceol)
9668 || ! _generic_isCC_A((U8) *scan, FLAGS(p))))
9669 {
9670 scan += UTF8SKIP(scan);
9671 hardcount++;
9672 }
9673 }
9674 break;
9675
9676 case NPOSIXU:
9677 to_complement = 1;
9678 /* FALLTHROUGH */
9679
9680 case POSIXU:
9681 if (! utf8_target) {
9682 while (scan < this_eol && to_complement
9683 ^ cBOOL(_generic_isCC((U8) *scan, FLAGS(p))))
9684 {
9685 scan++;
9686 }
9687 }
9688 else {
9689 utf8_posix:
9690 classnum = (_char_class_number) FLAGS(p);
9691 switch (classnum) {
9692 default:
9693 while ( hardcount < max && scan < this_eol
9694 && to_complement ^ cBOOL(_invlist_contains_cp(
9695 PL_XPosix_ptrs[classnum],
9696 utf8_to_uvchr_buf((U8 *) scan,
9697 (U8 *) this_eol,
9698 NULL))))
9699 {
9700 scan += UTF8SKIP(scan);
9701 hardcount++;
9702 }
9703 break;
9704
9705 /* For the classes below, the knowledge of how to handle
9706 * every code point is compiled in to Perl via a macro.
9707 * This code is written for making the loops as tight as
9708 * possible. It could be refactored to save space instead.
9709 * */
9710
9711 case _CC_ENUM_SPACE:
9712 while (hardcount < max
9713 && scan < this_eol
9714 && (to_complement
9715 ^ cBOOL(isSPACE_utf8_safe(scan, this_eol))))
9716 {
9717 scan += UTF8SKIP(scan);
9718 hardcount++;
9719 }
9720 break;
9721 case _CC_ENUM_BLANK:
9722 while (hardcount < max
9723 && scan < this_eol
9724 && (to_complement
9725 ^ cBOOL(isBLANK_utf8_safe(scan, this_eol))))
9726 {
9727 scan += UTF8SKIP(scan);
9728 hardcount++;
9729 }
9730 break;
9731 case _CC_ENUM_XDIGIT:
9732 while (hardcount < max
9733 && scan < this_eol
9734 && (to_complement
9735 ^ cBOOL(isXDIGIT_utf8_safe(scan, this_eol))))
9736 {
9737 scan += UTF8SKIP(scan);
9738 hardcount++;
9739 }
9740 break;
9741 case _CC_ENUM_VERTSPACE:
9742 while (hardcount < max
9743 && scan < this_eol
9744 && (to_complement
9745 ^ cBOOL(isVERTWS_utf8_safe(scan, this_eol))))
9746 {
9747 scan += UTF8SKIP(scan);
9748 hardcount++;
9749 }
9750 break;
9751 case _CC_ENUM_CNTRL:
9752 while (hardcount < max
9753 && scan < this_eol
9754 && (to_complement
9755 ^ cBOOL(isCNTRL_utf8_safe(scan, this_eol))))
9756 {
9757 scan += UTF8SKIP(scan);
9758 hardcount++;
9759 }
9760 break;
9761 }
9762 }
9763 break;
9764
9765 case LNBREAK:
9766 if (utf8_target) {
9767 while (hardcount < max && scan < this_eol &&
9768 (c=is_LNBREAK_utf8_safe(scan, this_eol))) {
9769 scan += c;
9770 hardcount++;
9771 }
9772 } else {
9773 /* LNBREAK can match one or two latin chars, which is ok, but we
9774 * have to use hardcount in this situation, and throw away the
9775 * adjustment to <this_eol> done before the switch statement */
9776 while (scan < loceol && (c=is_LNBREAK_latin1_safe(scan, loceol))) {
9777 scan+=c;
9778 hardcount++;
9779 }
9780 }
9781 break;
9782
9783 case BOUNDL:
9784 case NBOUNDL:
9785 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
9786 /* FALLTHROUGH */
9787 case BOUND:
9788 case BOUNDA:
9789 case BOUNDU:
9790 case EOS:
9791 case GPOS:
9792 case KEEPS:
9793 case NBOUND:
9794 case NBOUNDA:
9795 case NBOUNDU:
9796 case OPFAIL:
9797 case SBOL:
9798 case SEOL:
9799 /* These are all 0 width, so match right here or not at all. */
9800 break;
9801
9802 default:
9803 Perl_croak(aTHX_ "panic: regrepeat() called with unrecognized node type %d='%s'", OP(p), PL_reg_name[OP(p)]);
9804 NOT_REACHED; /* NOTREACHED */
9805
9806 }
9807
9808 if (hardcount)
9809 c = hardcount;
9810 else
9811 c = scan - *startposp;
9812 *startposp = scan;
9813
9814 DEBUG_r({
9815 GET_RE_DEBUG_FLAGS_DECL;
9816 DEBUG_EXECUTE_r({
9817 SV * const prop = sv_newmortal();
9818 regprop(prog, prop, p, reginfo, NULL);
9819 Perl_re_exec_indentf( aTHX_ "%s can match %" IVdf " times out of %" IVdf "...\n",
9820 depth, SvPVX_const(prop),(IV)c,(IV)max);
9821 });
9822 });
9823
9824 return(c);
9825 }
9826
9827 /*
9828 - reginclass - determine if a character falls into a character class
9829
9830 n is the ANYOF-type regnode
9831 p is the target string
9832 p_end points to one byte beyond the end of the target string
9833 utf8_target tells whether p is in UTF-8.
9834
9835 Returns true if matched; false otherwise.
9836
9837 Note that this can be a synthetic start class, a combination of various
9838 nodes, so things you think might be mutually exclusive, such as locale,
9839 aren't. It can match both locale and non-locale
9840
9841 */
9842
9843 STATIC bool
S_reginclass(pTHX_ regexp * const prog,const regnode * const n,const U8 * const p,const U8 * const p_end,const bool utf8_target)9844 S_reginclass(pTHX_ regexp * const prog, const regnode * const n, const U8* const p, const U8* const p_end, const bool utf8_target)
9845 {
9846 dVAR;
9847 const char flags = (OP(n) == ANYOFH) ? 0 : ANYOF_FLAGS(n);
9848 bool match = FALSE;
9849 UV c = *p;
9850
9851 PERL_ARGS_ASSERT_REGINCLASS;
9852
9853 /* If c is not already the code point, get it. Note that
9854 * UTF8_IS_INVARIANT() works even if not in UTF-8 */
9855 if (! UTF8_IS_INVARIANT(c) && utf8_target) {
9856 STRLEN c_len = 0;
9857 const U32 utf8n_flags = UTF8_ALLOW_DEFAULT;
9858 c = utf8n_to_uvchr(p, p_end - p, &c_len, utf8n_flags | UTF8_CHECK_ONLY);
9859 if (c_len == (STRLEN)-1) {
9860 _force_out_malformed_utf8_message(p, p_end,
9861 utf8n_flags,
9862 1 /* 1 means die */ );
9863 NOT_REACHED; /* NOTREACHED */
9864 }
9865 if ( c > 255
9866 && (OP(n) == ANYOFL || OP(n) == ANYOFPOSIXL)
9867 && ! ANYOFL_UTF8_LOCALE_REQD(flags))
9868 {
9869 _CHECK_AND_OUTPUT_WIDE_LOCALE_CP_MSG(c);
9870 }
9871 }
9872
9873 /* If this character is potentially in the bitmap, check it */
9874 if (c < NUM_ANYOF_CODE_POINTS && OP(n) != ANYOFH) {
9875 if (ANYOF_BITMAP_TEST(n, c))
9876 match = TRUE;
9877 else if ((flags
9878 & ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER)
9879 && OP(n) == ANYOFD
9880 && ! utf8_target
9881 && ! isASCII(c))
9882 {
9883 match = TRUE;
9884 }
9885 else if (flags & ANYOF_LOCALE_FLAGS) {
9886 if ( (flags & ANYOFL_FOLD)
9887 && c < sizeof(PL_fold_locale)
9888 && ANYOF_BITMAP_TEST(n, PL_fold_locale[c]))
9889 {
9890 match = TRUE;
9891 }
9892 else if ( ANYOF_POSIXL_TEST_ANY_SET(n)
9893 && c <= U8_MAX /* param to isFOO_lc() */
9894 ) {
9895
9896 /* The data structure is arranged so bits 0, 2, 4, ... are set
9897 * if the class includes the Posix character class given by
9898 * bit/2; and 1, 3, 5, ... are set if the class includes the
9899 * complemented Posix class given by int(bit/2). So we loop
9900 * through the bits, each time changing whether we complement
9901 * the result or not. Suppose for the sake of illustration
9902 * that bits 0-3 mean respectively, \w, \W, \s, \S. If bit 0
9903 * is set, it means there is a match for this ANYOF node if the
9904 * character is in the class given by the expression (0 / 2 = 0
9905 * = \w). If it is in that class, isFOO_lc() will return 1,
9906 * and since 'to_complement' is 0, the result will stay TRUE,
9907 * and we exit the loop. Suppose instead that bit 0 is 0, but
9908 * bit 1 is 1. That means there is a match if the character
9909 * matches \W. We won't bother to call isFOO_lc() on bit 0,
9910 * but will on bit 1. On the second iteration 'to_complement'
9911 * will be 1, so the exclusive or will reverse things, so we
9912 * are testing for \W. On the third iteration, 'to_complement'
9913 * will be 0, and we would be testing for \s; the fourth
9914 * iteration would test for \S, etc.
9915 *
9916 * Note that this code assumes that all the classes are closed
9917 * under folding. For example, if a character matches \w, then
9918 * its fold does too; and vice versa. This should be true for
9919 * any well-behaved locale for all the currently defined Posix
9920 * classes, except for :lower: and :upper:, which are handled
9921 * by the pseudo-class :cased: which matches if either of the
9922 * other two does. To get rid of this assumption, an outer
9923 * loop could be used below to iterate over both the source
9924 * character, and its fold (if different) */
9925
9926 int count = 0;
9927 int to_complement = 0;
9928
9929 while (count < ANYOF_MAX) {
9930 if (ANYOF_POSIXL_TEST(n, count)
9931 && to_complement ^ cBOOL(isFOO_lc(count/2, (U8) c)))
9932 {
9933 match = TRUE;
9934 break;
9935 }
9936 count++;
9937 to_complement ^= 1;
9938 }
9939 }
9940 }
9941 }
9942
9943
9944 /* If the bitmap didn't (or couldn't) match, and something outside the
9945 * bitmap could match, try that. */
9946 if (!match) {
9947 if (c >= NUM_ANYOF_CODE_POINTS
9948 && (flags & ANYOF_MATCHES_ALL_ABOVE_BITMAP))
9949 {
9950 match = TRUE; /* Everything above the bitmap matches */
9951 }
9952 /* Here doesn't match everything above the bitmap. If there is
9953 * some information available beyond the bitmap, we may find a
9954 * match in it. If so, this is most likely because the code point
9955 * is outside the bitmap range. But rarely, it could be because of
9956 * some other reason. If so, various flags are set to indicate
9957 * this possibility. On ANYOFD nodes, there may be matches that
9958 * happen only when the target string is UTF-8; or for other node
9959 * types, because runtime lookup is needed, regardless of the
9960 * UTF-8ness of the target string. Finally, under /il, there may
9961 * be some matches only possible if the locale is a UTF-8 one. */
9962 else if ( ARG(n) != ANYOF_ONLY_HAS_BITMAP
9963 && ( c >= NUM_ANYOF_CODE_POINTS
9964 || ( (flags & ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP)
9965 && ( UNLIKELY(OP(n) != ANYOFD)
9966 || (utf8_target && ! isASCII_uni(c)
9967 # if NUM_ANYOF_CODE_POINTS > 256
9968 && c < 256
9969 # endif
9970 )))
9971 || ( ANYOFL_SOME_FOLDS_ONLY_IN_UTF8_LOCALE(flags)
9972 && IN_UTF8_CTYPE_LOCALE)))
9973 {
9974 SV* only_utf8_locale = NULL;
9975 SV * const definition = _get_regclass_nonbitmap_data(prog, n, TRUE,
9976 0, &only_utf8_locale, NULL);
9977 if (definition) {
9978 U8 utf8_buffer[2];
9979 U8 * utf8_p;
9980 if (utf8_target) {
9981 utf8_p = (U8 *) p;
9982 } else { /* Convert to utf8 */
9983 utf8_p = utf8_buffer;
9984 append_utf8_from_native_byte(*p, &utf8_p);
9985 utf8_p = utf8_buffer;
9986 }
9987
9988 /* Turkish locales have these hard-coded rules overriding
9989 * normal ones */
9990 if ( UNLIKELY(PL_in_utf8_turkic_locale)
9991 && isALPHA_FOLD_EQ(*p, 'i'))
9992 {
9993 if (*p == 'i') {
9994 if (_invlist_contains_cp(definition,
9995 LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE))
9996 {
9997 match = TRUE;
9998 }
9999 }
10000 else if (*p == 'I') {
10001 if (_invlist_contains_cp(definition,
10002 LATIN_SMALL_LETTER_DOTLESS_I))
10003 {
10004 match = TRUE;
10005 }
10006 }
10007 }
10008 else if (_invlist_contains_cp(definition, c)) {
10009 match = TRUE;
10010 }
10011 }
10012 if (! match && only_utf8_locale && IN_UTF8_CTYPE_LOCALE) {
10013 match = _invlist_contains_cp(only_utf8_locale, c);
10014 }
10015 }
10016
10017 /* In a Turkic locale under folding, hard-code the I i case pair
10018 * matches */
10019 if ( UNLIKELY(PL_in_utf8_turkic_locale)
10020 && ! match
10021 && (flags & ANYOFL_FOLD)
10022 && utf8_target)
10023 {
10024 if (c == LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE) {
10025 if (ANYOF_BITMAP_TEST(n, 'i')) {
10026 match = TRUE;
10027 }
10028 }
10029 else if (c == LATIN_SMALL_LETTER_DOTLESS_I) {
10030 if (ANYOF_BITMAP_TEST(n, 'I')) {
10031 match = TRUE;
10032 }
10033 }
10034 }
10035
10036 if (UNICODE_IS_SUPER(c)
10037 && (flags
10038 & ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER)
10039 && OP(n) != ANYOFD
10040 && ckWARN_d(WARN_NON_UNICODE))
10041 {
10042 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
10043 "Matched non-Unicode code point 0x%04" UVXf " against Unicode property; may not be portable", c);
10044 }
10045 }
10046
10047 #if ANYOF_INVERT != 1
10048 /* Depending on compiler optimization cBOOL takes time, so if don't have to
10049 * use it, don't */
10050 # error ANYOF_INVERT needs to be set to 1, or guarded with cBOOL below,
10051 #endif
10052
10053 /* The xor complements the return if to invert: 1^1 = 0, 1^0 = 1 */
10054 return (flags & ANYOF_INVERT) ^ match;
10055 }
10056
10057 STATIC U8 *
S_reghop3(U8 * s,SSize_t off,const U8 * lim)10058 S_reghop3(U8 *s, SSize_t off, const U8* lim)
10059 {
10060 /* return the position 'off' UTF-8 characters away from 's', forward if
10061 * 'off' >= 0, backwards if negative. But don't go outside of position
10062 * 'lim', which better be < s if off < 0 */
10063
10064 PERL_ARGS_ASSERT_REGHOP3;
10065
10066 if (off >= 0) {
10067 while (off-- && s < lim) {
10068 /* XXX could check well-formedness here */
10069 U8 *new_s = s + UTF8SKIP(s);
10070 if (new_s > lim) /* lim may be in the middle of a long character */
10071 return s;
10072 s = new_s;
10073 }
10074 }
10075 else {
10076 while (off++ && s > lim) {
10077 s--;
10078 if (UTF8_IS_CONTINUED(*s)) {
10079 while (s > lim && UTF8_IS_CONTINUATION(*s))
10080 s--;
10081 if (! UTF8_IS_START(*s)) {
10082 Perl_croak_nocontext("Malformed UTF-8 character (fatal)");
10083 }
10084 }
10085 /* XXX could check well-formedness here */
10086 }
10087 }
10088 return s;
10089 }
10090
10091 STATIC U8 *
S_reghop4(U8 * s,SSize_t off,const U8 * llim,const U8 * rlim)10092 S_reghop4(U8 *s, SSize_t off, const U8* llim, const U8* rlim)
10093 {
10094 PERL_ARGS_ASSERT_REGHOP4;
10095
10096 if (off >= 0) {
10097 while (off-- && s < rlim) {
10098 /* XXX could check well-formedness here */
10099 s += UTF8SKIP(s);
10100 }
10101 }
10102 else {
10103 while (off++ && s > llim) {
10104 s--;
10105 if (UTF8_IS_CONTINUED(*s)) {
10106 while (s > llim && UTF8_IS_CONTINUATION(*s))
10107 s--;
10108 if (! UTF8_IS_START(*s)) {
10109 Perl_croak_nocontext("Malformed UTF-8 character (fatal)");
10110 }
10111 }
10112 /* XXX could check well-formedness here */
10113 }
10114 }
10115 return s;
10116 }
10117
10118 /* like reghop3, but returns NULL on overrun, rather than returning last
10119 * char pos */
10120
10121 STATIC U8 *
S_reghopmaybe3(U8 * s,SSize_t off,const U8 * const lim)10122 S_reghopmaybe3(U8* s, SSize_t off, const U8* const lim)
10123 {
10124 PERL_ARGS_ASSERT_REGHOPMAYBE3;
10125
10126 if (off >= 0) {
10127 while (off-- && s < lim) {
10128 /* XXX could check well-formedness here */
10129 s += UTF8SKIP(s);
10130 }
10131 if (off >= 0)
10132 return NULL;
10133 }
10134 else {
10135 while (off++ && s > lim) {
10136 s--;
10137 if (UTF8_IS_CONTINUED(*s)) {
10138 while (s > lim && UTF8_IS_CONTINUATION(*s))
10139 s--;
10140 if (! UTF8_IS_START(*s)) {
10141 Perl_croak_nocontext("Malformed UTF-8 character (fatal)");
10142 }
10143 }
10144 /* XXX could check well-formedness here */
10145 }
10146 if (off <= 0)
10147 return NULL;
10148 }
10149 return s;
10150 }
10151
10152
10153 /* when executing a regex that may have (?{}), extra stuff needs setting
10154 up that will be visible to the called code, even before the current
10155 match has finished. In particular:
10156
10157 * $_ is localised to the SV currently being matched;
10158 * pos($_) is created if necessary, ready to be updated on each call-out
10159 to code;
10160 * a fake PMOP is created that can be set to PL_curpm (normally PL_curpm
10161 isn't set until the current pattern is successfully finished), so that
10162 $1 etc of the match-so-far can be seen;
10163 * save the old values of subbeg etc of the current regex, and set then
10164 to the current string (again, this is normally only done at the end
10165 of execution)
10166 */
10167
10168 static void
S_setup_eval_state(pTHX_ regmatch_info * const reginfo)10169 S_setup_eval_state(pTHX_ regmatch_info *const reginfo)
10170 {
10171 MAGIC *mg;
10172 regexp *const rex = ReANY(reginfo->prog);
10173 regmatch_info_aux_eval *eval_state = reginfo->info_aux_eval;
10174
10175 eval_state->rex = rex;
10176 eval_state->sv = reginfo->sv;
10177
10178 if (reginfo->sv) {
10179 /* Make $_ available to executed code. */
10180 if (reginfo->sv != DEFSV) {
10181 SAVE_DEFSV;
10182 DEFSV_set(reginfo->sv);
10183 }
10184 /* will be dec'd by S_cleanup_regmatch_info_aux */
10185 SvREFCNT_inc_NN(reginfo->sv);
10186
10187 if (!(mg = mg_find_mglob(reginfo->sv))) {
10188 /* prepare for quick setting of pos */
10189 mg = sv_magicext_mglob(reginfo->sv);
10190 mg->mg_len = -1;
10191 }
10192 eval_state->pos_magic = mg;
10193 eval_state->pos = mg->mg_len;
10194 eval_state->pos_flags = mg->mg_flags;
10195 }
10196 else
10197 eval_state->pos_magic = NULL;
10198
10199 if (!PL_reg_curpm) {
10200 /* PL_reg_curpm is a fake PMOP that we can attach the current
10201 * regex to and point PL_curpm at, so that $1 et al are visible
10202 * within a /(?{})/. It's just allocated once per interpreter the
10203 * first time its needed */
10204 Newxz(PL_reg_curpm, 1, PMOP);
10205 #ifdef USE_ITHREADS
10206 {
10207 SV* const repointer = &PL_sv_undef;
10208 /* this regexp is also owned by the new PL_reg_curpm, which
10209 will try to free it. */
10210 av_push(PL_regex_padav, repointer);
10211 PL_reg_curpm->op_pmoffset = av_tindex(PL_regex_padav);
10212 PL_regex_pad = AvARRAY(PL_regex_padav);
10213 }
10214 #endif
10215 }
10216 SET_reg_curpm(reginfo->prog);
10217 eval_state->curpm = PL_curpm;
10218 PL_curpm_under = PL_curpm;
10219 PL_curpm = PL_reg_curpm;
10220 if (RXp_MATCH_COPIED(rex)) {
10221 /* Here is a serious problem: we cannot rewrite subbeg,
10222 since it may be needed if this match fails. Thus
10223 $` inside (?{}) could fail... */
10224 eval_state->subbeg = rex->subbeg;
10225 eval_state->sublen = rex->sublen;
10226 eval_state->suboffset = rex->suboffset;
10227 eval_state->subcoffset = rex->subcoffset;
10228 #ifdef PERL_ANY_COW
10229 eval_state->saved_copy = rex->saved_copy;
10230 #endif
10231 RXp_MATCH_COPIED_off(rex);
10232 }
10233 else
10234 eval_state->subbeg = NULL;
10235 rex->subbeg = (char *)reginfo->strbeg;
10236 rex->suboffset = 0;
10237 rex->subcoffset = 0;
10238 rex->sublen = reginfo->strend - reginfo->strbeg;
10239 }
10240
10241
10242 /* destructor to clear up regmatch_info_aux and regmatch_info_aux_eval */
10243
10244 static void
S_cleanup_regmatch_info_aux(pTHX_ void * arg)10245 S_cleanup_regmatch_info_aux(pTHX_ void *arg)
10246 {
10247 regmatch_info_aux *aux = (regmatch_info_aux *) arg;
10248 regmatch_info_aux_eval *eval_state = aux->info_aux_eval;
10249 regmatch_slab *s;
10250
10251 Safefree(aux->poscache);
10252
10253 if (eval_state) {
10254
10255 /* undo the effects of S_setup_eval_state() */
10256
10257 if (eval_state->subbeg) {
10258 regexp * const rex = eval_state->rex;
10259 rex->subbeg = eval_state->subbeg;
10260 rex->sublen = eval_state->sublen;
10261 rex->suboffset = eval_state->suboffset;
10262 rex->subcoffset = eval_state->subcoffset;
10263 #ifdef PERL_ANY_COW
10264 rex->saved_copy = eval_state->saved_copy;
10265 #endif
10266 RXp_MATCH_COPIED_on(rex);
10267 }
10268 if (eval_state->pos_magic)
10269 {
10270 eval_state->pos_magic->mg_len = eval_state->pos;
10271 eval_state->pos_magic->mg_flags =
10272 (eval_state->pos_magic->mg_flags & ~MGf_BYTES)
10273 | (eval_state->pos_flags & MGf_BYTES);
10274 }
10275
10276 PL_curpm = eval_state->curpm;
10277 SvREFCNT_dec(eval_state->sv);
10278 }
10279
10280 PL_regmatch_state = aux->old_regmatch_state;
10281 PL_regmatch_slab = aux->old_regmatch_slab;
10282
10283 /* free all slabs above current one - this must be the last action
10284 * of this function, as aux and eval_state are allocated within
10285 * slabs and may be freed here */
10286
10287 s = PL_regmatch_slab->next;
10288 if (s) {
10289 PL_regmatch_slab->next = NULL;
10290 while (s) {
10291 regmatch_slab * const osl = s;
10292 s = s->next;
10293 Safefree(osl);
10294 }
10295 }
10296 }
10297
10298
10299 STATIC void
S_to_utf8_substr(pTHX_ regexp * prog)10300 S_to_utf8_substr(pTHX_ regexp *prog)
10301 {
10302 /* Converts substr fields in prog from bytes to UTF-8, calling fbm_compile
10303 * on the converted value */
10304
10305 int i = 1;
10306
10307 PERL_ARGS_ASSERT_TO_UTF8_SUBSTR;
10308
10309 do {
10310 if (prog->substrs->data[i].substr
10311 && !prog->substrs->data[i].utf8_substr) {
10312 SV* const sv = newSVsv(prog->substrs->data[i].substr);
10313 prog->substrs->data[i].utf8_substr = sv;
10314 sv_utf8_upgrade(sv);
10315 if (SvVALID(prog->substrs->data[i].substr)) {
10316 if (SvTAIL(prog->substrs->data[i].substr)) {
10317 /* Trim the trailing \n that fbm_compile added last
10318 time. */
10319 SvCUR_set(sv, SvCUR(sv) - 1);
10320 /* Whilst this makes the SV technically "invalid" (as its
10321 buffer is no longer followed by "\0") when fbm_compile()
10322 adds the "\n" back, a "\0" is restored. */
10323 fbm_compile(sv, FBMcf_TAIL);
10324 } else
10325 fbm_compile(sv, 0);
10326 }
10327 if (prog->substrs->data[i].substr == prog->check_substr)
10328 prog->check_utf8 = sv;
10329 }
10330 } while (i--);
10331 }
10332
10333 STATIC bool
S_to_byte_substr(pTHX_ regexp * prog)10334 S_to_byte_substr(pTHX_ regexp *prog)
10335 {
10336 /* Converts substr fields in prog from UTF-8 to bytes, calling fbm_compile
10337 * on the converted value; returns FALSE if can't be converted. */
10338
10339 int i = 1;
10340
10341 PERL_ARGS_ASSERT_TO_BYTE_SUBSTR;
10342
10343 do {
10344 if (prog->substrs->data[i].utf8_substr
10345 && !prog->substrs->data[i].substr) {
10346 SV* sv = newSVsv(prog->substrs->data[i].utf8_substr);
10347 if (! sv_utf8_downgrade(sv, TRUE)) {
10348 SvREFCNT_dec_NN(sv);
10349 return FALSE;
10350 }
10351 if (SvVALID(prog->substrs->data[i].utf8_substr)) {
10352 if (SvTAIL(prog->substrs->data[i].utf8_substr)) {
10353 /* Trim the trailing \n that fbm_compile added last
10354 time. */
10355 SvCUR_set(sv, SvCUR(sv) - 1);
10356 fbm_compile(sv, FBMcf_TAIL);
10357 } else
10358 fbm_compile(sv, 0);
10359 }
10360 prog->substrs->data[i].substr = sv;
10361 if (prog->substrs->data[i].utf8_substr == prog->check_utf8)
10362 prog->check_substr = sv;
10363 }
10364 } while (i--);
10365
10366 return TRUE;
10367 }
10368
10369 #ifndef PERL_IN_XSUB_RE
10370
10371 bool
Perl__is_grapheme(pTHX_ const U8 * strbeg,const U8 * s,const U8 * strend,const UV cp)10372 Perl__is_grapheme(pTHX_ const U8 * strbeg, const U8 * s, const U8 * strend, const UV cp)
10373 {
10374 /* Temporary helper function for toke.c. Verify that the code point 'cp'
10375 * is a stand-alone grapheme. The UTF-8 for 'cp' begins at position 's' in
10376 * the larger string bounded by 'strbeg' and 'strend'.
10377 *
10378 * 'cp' needs to be assigned (if not a future version of the Unicode
10379 * Standard could make it something that combines with adjacent characters,
10380 * so code using it would then break), and there has to be a GCB break
10381 * before and after the character. */
10382
10383 dVAR;
10384
10385 GCB_enum cp_gcb_val, prev_cp_gcb_val, next_cp_gcb_val;
10386 const U8 * prev_cp_start;
10387
10388 PERL_ARGS_ASSERT__IS_GRAPHEME;
10389
10390 if ( UNLIKELY(UNICODE_IS_SUPER(cp))
10391 || UNLIKELY(UNICODE_IS_NONCHAR(cp)))
10392 {
10393 /* These are considered graphemes */
10394 return TRUE;
10395 }
10396
10397 /* Otherwise, unassigned code points are forbidden */
10398 if (UNLIKELY(! ELEMENT_RANGE_MATCHES_INVLIST(
10399 _invlist_search(PL_Assigned_invlist, cp))))
10400 {
10401 return FALSE;
10402 }
10403
10404 cp_gcb_val = getGCB_VAL_CP(cp);
10405
10406 /* Find the GCB value of the previous code point in the input */
10407 prev_cp_start = utf8_hop_back(s, -1, strbeg);
10408 if (UNLIKELY(prev_cp_start == s)) {
10409 prev_cp_gcb_val = GCB_EDGE;
10410 }
10411 else {
10412 prev_cp_gcb_val = getGCB_VAL_UTF8(prev_cp_start, strend);
10413 }
10414
10415 /* And check that is a grapheme boundary */
10416 if (! isGCB(prev_cp_gcb_val, cp_gcb_val, strbeg, s,
10417 TRUE /* is UTF-8 encoded */ ))
10418 {
10419 return FALSE;
10420 }
10421
10422 /* Similarly verify there is a break between the current character and the
10423 * following one */
10424 s += UTF8SKIP(s);
10425 if (s >= strend) {
10426 next_cp_gcb_val = GCB_EDGE;
10427 }
10428 else {
10429 next_cp_gcb_val = getGCB_VAL_UTF8(s, strend);
10430 }
10431
10432 return isGCB(cp_gcb_val, next_cp_gcb_val, strbeg, s, TRUE);
10433 }
10434
10435 /*
10436 =head1 Unicode Support
10437
10438 =for apidoc isSCRIPT_RUN
10439
10440 Returns a bool as to whether or not the sequence of bytes from C<s> up to but
10441 not including C<send> form a "script run". C<utf8_target> is TRUE iff the
10442 sequence starting at C<s> is to be treated as UTF-8. To be precise, except for
10443 two degenerate cases given below, this function returns TRUE iff all code
10444 points in it come from any combination of three "scripts" given by the Unicode
10445 "Script Extensions" property: Common, Inherited, and possibly one other.
10446 Additionally all decimal digits must come from the same consecutive sequence of
10447 10.
10448
10449 For example, if all the characters in the sequence are Greek, or Common, or
10450 Inherited, this function will return TRUE, provided any decimal digits in it
10451 are from the same block of digits in Common. (These are the ASCII digits
10452 "0".."9" and additionally a block for full width forms of these, and several
10453 others used in mathematical notation.) For scripts (unlike Greek) that have
10454 their own digits defined this will accept either digits from that set or from
10455 one of the Common digit sets, but not a combination of the two. Some scripts,
10456 such as Arabic, have more than one set of digits. All digits must come from
10457 the same set for this function to return TRUE.
10458
10459 C<*ret_script>, if C<ret_script> is not NULL, will on return of TRUE
10460 contain the script found, using the C<SCX_enum> typedef. Its value will be
10461 C<SCX_INVALID> if the function returns FALSE.
10462
10463 If the sequence is empty, TRUE is returned, but C<*ret_script> (if asked for)
10464 will be C<SCX_INVALID>.
10465
10466 If the sequence contains a single code point which is unassigned to a character
10467 in the version of Unicode being used, the function will return TRUE, and the
10468 script will be C<SCX_Unknown>. Any other combination of unassigned code points
10469 in the input sequence will result in the function treating the input as not
10470 being a script run.
10471
10472 The returned script will be C<SCX_Inherited> iff all the code points in it are
10473 from the Inherited script.
10474
10475 Otherwise, the returned script will be C<SCX_Common> iff all the code points in
10476 it are from the Inherited or Common scripts.
10477
10478 =cut
10479
10480 */
10481
10482 bool
Perl_isSCRIPT_RUN(pTHX_ const U8 * s,const U8 * send,const bool utf8_target)10483 Perl_isSCRIPT_RUN(pTHX_ const U8 * s, const U8 * send, const bool utf8_target)
10484 {
10485 /* Basically, it looks at each character in the sequence to see if the
10486 * above conditions are met; if not it fails. It uses an inversion map to
10487 * find the enum corresponding to the script of each character. But this
10488 * is complicated by the fact that a few code points can be in any of
10489 * several scripts. The data has been constructed so that there are
10490 * additional enum values (all negative) for these situations. The
10491 * absolute value of those is an index into another table which contains
10492 * pointers to auxiliary tables for each such situation. Each aux array
10493 * lists all the scripts for the given situation. There is another,
10494 * parallel, table that gives the number of entries in each aux table.
10495 * These are all defined in charclass_invlists.h */
10496
10497 /* XXX Here are the additional things UTS 39 says could be done:
10498 *
10499 * Forbid sequences of the same nonspacing mark
10500 *
10501 * Check to see that all the characters are in the sets of exemplar
10502 * characters for at least one language in the Unicode Common Locale Data
10503 * Repository [CLDR]. */
10504
10505 dVAR;
10506
10507 /* Things that match /\d/u */
10508 SV * decimals_invlist = PL_XPosix_ptrs[_CC_DIGIT];
10509 UV * decimals_array = invlist_array(decimals_invlist);
10510
10511 /* What code point is the digit '0' of the script run? (0 meaning FALSE if
10512 * not currently known) */
10513 UV zero_of_run = 0;
10514
10515 SCX_enum script_of_run = SCX_INVALID; /* Illegal value */
10516 SCX_enum script_of_char = SCX_INVALID;
10517
10518 /* If the script remains not fully determined from iteration to iteration,
10519 * this is the current intersection of the possiblities. */
10520 SCX_enum * intersection = NULL;
10521 PERL_UINT_FAST8_T intersection_len = 0;
10522
10523 bool retval = TRUE;
10524 SCX_enum * ret_script = NULL;
10525
10526 assert(send >= s);
10527
10528 PERL_ARGS_ASSERT_ISSCRIPT_RUN;
10529
10530 /* All code points in 0..255 are either Common or Latin, so must be a
10531 * script run. We can return immediately unless we need to know which
10532 * script it is. */
10533 if (! utf8_target && LIKELY(send > s)) {
10534 if (ret_script == NULL) {
10535 return TRUE;
10536 }
10537
10538 /* If any character is Latin, the run is Latin */
10539 while (s < send) {
10540 if (isALPHA_L1(*s) && LIKELY(*s != MICRO_SIGN_NATIVE)) {
10541 *ret_script = SCX_Latin;
10542 return TRUE;
10543 }
10544 }
10545
10546 /* Here, all are Common */
10547 *ret_script = SCX_Common;
10548 return TRUE;
10549 }
10550
10551 /* Look at each character in the sequence */
10552 while (s < send) {
10553 /* If the current character being examined is a digit, this is the code
10554 * point of the zero for its sequence of 10 */
10555 UV zero_of_char;
10556
10557 UV cp;
10558
10559 /* The code allows all scripts to use the ASCII digits. This is
10560 * because they are in the Common script. Hence any ASCII ones found
10561 * are ok, unless and until a digit from another set has already been
10562 * encountered. digit ranges in Common are not similarly blessed) */
10563 if (UNLIKELY(isDIGIT(*s))) {
10564 if (UNLIKELY(script_of_run == SCX_Unknown)) {
10565 retval = FALSE;
10566 break;
10567 }
10568 if (zero_of_run) {
10569 if (zero_of_run != '0') {
10570 retval = FALSE;
10571 break;
10572 }
10573 }
10574 else {
10575 zero_of_run = '0';
10576 }
10577 s++;
10578 continue;
10579 }
10580
10581 /* Here, isn't an ASCII digit. Find the code point of the character */
10582 if (! UTF8_IS_INVARIANT(*s)) {
10583 Size_t len;
10584 cp = valid_utf8_to_uvchr((U8 *) s, &len);
10585 s += len;
10586 }
10587 else {
10588 cp = *(s++);
10589 }
10590
10591 /* If is within the range [+0 .. +9] of the script's zero, it also is a
10592 * digit in that script. We can skip the rest of this code for this
10593 * character. */
10594 if (UNLIKELY( zero_of_run
10595 && cp >= zero_of_run
10596 && cp - zero_of_run <= 9))
10597 {
10598 continue;
10599 }
10600
10601 /* Find the character's script. The correct values are hard-coded here
10602 * for small-enough code points. */
10603 if (cp < 0x2B9) { /* From inspection of Unicode db; extremely
10604 unlikely to change */
10605 if ( cp > 255
10606 || ( isALPHA_L1(cp)
10607 && LIKELY(cp != MICRO_SIGN_NATIVE)))
10608 {
10609 script_of_char = SCX_Latin;
10610 }
10611 else {
10612 script_of_char = SCX_Common;
10613 }
10614 }
10615 else {
10616 script_of_char = _Perl_SCX_invmap[
10617 _invlist_search(PL_SCX_invlist, cp)];
10618 }
10619
10620 /* We arbitrarily accept a single unassigned character, but not in
10621 * combination with anything else, and not a run of them. */
10622 if ( UNLIKELY(script_of_run == SCX_Unknown)
10623 || UNLIKELY( script_of_run != SCX_INVALID
10624 && script_of_char == SCX_Unknown))
10625 {
10626 retval = FALSE;
10627 break;
10628 }
10629
10630 /* For the first character, or the run is inherited, the run's script
10631 * is set to the char's */
10632 if ( UNLIKELY(script_of_run == SCX_INVALID)
10633 || UNLIKELY(script_of_run == SCX_Inherited))
10634 {
10635 script_of_run = script_of_char;
10636 }
10637
10638 /* For the character's script to be Unknown, it must be the first
10639 * character in the sequence (for otherwise a test above would have
10640 * prevented us from reaching here), and we have set the run's script
10641 * to it. Nothing further to be done for this character */
10642 if (UNLIKELY(script_of_char == SCX_Unknown)) {
10643 continue;
10644 }
10645
10646 /* We accept 'inherited' script characters currently even at the
10647 * beginning. (We know that no characters in Inherited are digits, or
10648 * we'd have to check for that) */
10649 if (UNLIKELY(script_of_char == SCX_Inherited)) {
10650 continue;
10651 }
10652
10653 /* If the run so far is Common, and the new character isn't, change the
10654 * run's script to that of this character */
10655 if (script_of_run == SCX_Common && script_of_char != SCX_Common) {
10656 script_of_run = script_of_char;
10657 }
10658
10659 /* Now we can see if the script of the new character is the same as
10660 * that of the run */
10661 if (LIKELY(script_of_char == script_of_run)) {
10662 /* By far the most common case */
10663 goto scripts_match;
10664 }
10665
10666 /* Here, the script of the run isn't Common. But characters in Common
10667 * match any script */
10668 if (script_of_char == SCX_Common) {
10669 goto scripts_match;
10670 }
10671
10672 #ifndef HAS_SCX_AUX_TABLES
10673
10674 /* Too early a Unicode version to have a code point belonging to more
10675 * than one script, so, if the scripts don't exactly match, fail */
10676 PERL_UNUSED_VAR(intersection_len);
10677 retval = FALSE;
10678 break;
10679
10680 #else
10681
10682 /* Here there is no exact match between the character's script and the
10683 * run's. And we've handled the special cases of scripts Unknown,
10684 * Inherited, and Common.
10685 *
10686 * Negative script numbers signify that the value may be any of several
10687 * scripts, and we need to look at auxiliary information to make our
10688 * deterimination. But if both are non-negative, we can fail now */
10689 if (LIKELY(script_of_char >= 0)) {
10690 const SCX_enum * search_in;
10691 PERL_UINT_FAST8_T search_in_len;
10692 PERL_UINT_FAST8_T i;
10693
10694 if (LIKELY(script_of_run >= 0)) {
10695 retval = FALSE;
10696 break;
10697 }
10698
10699 /* Use the previously constructed set of possible scripts, if any.
10700 * */
10701 if (intersection) {
10702 search_in = intersection;
10703 search_in_len = intersection_len;
10704 }
10705 else {
10706 search_in = SCX_AUX_TABLE_ptrs[-script_of_run];
10707 search_in_len = SCX_AUX_TABLE_lengths[-script_of_run];
10708 }
10709
10710 for (i = 0; i < search_in_len; i++) {
10711 if (search_in[i] == script_of_char) {
10712 script_of_run = script_of_char;
10713 goto scripts_match;
10714 }
10715 }
10716
10717 retval = FALSE;
10718 break;
10719 }
10720 else if (LIKELY(script_of_run >= 0)) {
10721 /* script of character could be one of several, but run is a single
10722 * script */
10723 const SCX_enum * search_in = SCX_AUX_TABLE_ptrs[-script_of_char];
10724 const PERL_UINT_FAST8_T search_in_len
10725 = SCX_AUX_TABLE_lengths[-script_of_char];
10726 PERL_UINT_FAST8_T i;
10727
10728 for (i = 0; i < search_in_len; i++) {
10729 if (search_in[i] == script_of_run) {
10730 script_of_char = script_of_run;
10731 goto scripts_match;
10732 }
10733 }
10734
10735 retval = FALSE;
10736 break;
10737 }
10738 else {
10739 /* Both run and char could be in one of several scripts. If the
10740 * intersection is empty, then this character isn't in this script
10741 * run. Otherwise, we need to calculate the intersection to use
10742 * for future iterations of the loop, unless we are already at the
10743 * final character */
10744 const SCX_enum * search_char = SCX_AUX_TABLE_ptrs[-script_of_char];
10745 const PERL_UINT_FAST8_T char_len
10746 = SCX_AUX_TABLE_lengths[-script_of_char];
10747 const SCX_enum * search_run;
10748 PERL_UINT_FAST8_T run_len;
10749
10750 SCX_enum * new_overlap = NULL;
10751 PERL_UINT_FAST8_T i, j;
10752
10753 if (intersection) {
10754 search_run = intersection;
10755 run_len = intersection_len;
10756 }
10757 else {
10758 search_run = SCX_AUX_TABLE_ptrs[-script_of_run];
10759 run_len = SCX_AUX_TABLE_lengths[-script_of_run];
10760 }
10761
10762 intersection_len = 0;
10763
10764 for (i = 0; i < run_len; i++) {
10765 for (j = 0; j < char_len; j++) {
10766 if (search_run[i] == search_char[j]) {
10767
10768 /* Here, the script at i,j matches. That means this
10769 * character is in the run. But continue on to find
10770 * the complete intersection, for the next loop
10771 * iteration, and for the digit check after it.
10772 *
10773 * On the first found common script, we malloc space
10774 * for the intersection list for the worst case of the
10775 * intersection, which is the minimum of the number of
10776 * scripts remaining in each set. */
10777 if (intersection_len == 0) {
10778 Newx(new_overlap,
10779 MIN(run_len - i, char_len - j),
10780 SCX_enum);
10781 }
10782 new_overlap[intersection_len++] = search_run[i];
10783 }
10784 }
10785 }
10786
10787 /* Here we've looked through everything. If they have no scripts
10788 * in common, not a run */
10789 if (intersection_len == 0) {
10790 retval = FALSE;
10791 break;
10792 }
10793
10794 /* If there is only a single script in common, set to that.
10795 * Otherwise, use the intersection going forward */
10796 Safefree(intersection);
10797 intersection = NULL;
10798 if (intersection_len == 1) {
10799 script_of_run = script_of_char = new_overlap[0];
10800 Safefree(new_overlap);
10801 new_overlap = NULL;
10802 }
10803 else {
10804 intersection = new_overlap;
10805 }
10806 }
10807
10808 #endif
10809
10810 scripts_match:
10811
10812 /* Here, the script of the character is compatible with that of the
10813 * run. That means that in most cases, it continues the script run.
10814 * Either it and the run match exactly, or one or both can be in any of
10815 * several scripts, and the intersection is not empty. However, if the
10816 * character is a decimal digit, it could still mean failure if it is
10817 * from the wrong sequence of 10. So, we need to look at if it's a
10818 * digit. We've already handled the 10 decimal digits, and the next
10819 * lowest one is this one: */
10820 if (cp < FIRST_NON_ASCII_DECIMAL_DIGIT) {
10821 continue; /* Not a digit; this character is part of the run */
10822 }
10823
10824 /* If we have a definitive '0' for the script of this character, we
10825 * know that for this to be a digit, it must be in the range of +0..+9
10826 * of that zero. */
10827 if ( script_of_char >= 0
10828 && (zero_of_char = script_zeros[script_of_char]))
10829 {
10830 if ( cp < zero_of_char
10831 || cp > zero_of_char + 9)
10832 {
10833 continue; /* Not a digit; this character is part of the run
10834 */
10835 }
10836
10837 }
10838 else { /* Need to look up if this character is a digit or not */
10839 SSize_t index_of_zero_of_char;
10840 index_of_zero_of_char = _invlist_search(decimals_invlist, cp);
10841 if ( UNLIKELY(index_of_zero_of_char < 0)
10842 || ! ELEMENT_RANGE_MATCHES_INVLIST(index_of_zero_of_char))
10843 {
10844 continue; /* Not a digit; this character is part of the run.
10845 */
10846 }
10847
10848 zero_of_char = decimals_array[index_of_zero_of_char];
10849 }
10850
10851 /* Here, the character is a decimal digit, and the zero of its sequence
10852 * of 10 is in 'zero_of_char'. If we already have a zero for this run,
10853 * they better be the same. */
10854 if (zero_of_run) {
10855 if (zero_of_run != zero_of_char) {
10856 retval = FALSE;
10857 break;
10858 }
10859 }
10860 else { /* Otherwise we now have a zero for this run */
10861 zero_of_run = zero_of_char;
10862 }
10863 } /* end of looping through CLOSESR text */
10864
10865 Safefree(intersection);
10866
10867 if (ret_script != NULL) {
10868 if (retval) {
10869 *ret_script = script_of_run;
10870 }
10871 else {
10872 *ret_script = SCX_INVALID;
10873 }
10874 }
10875
10876 return retval;
10877 }
10878
10879 #endif /* ifndef PERL_IN_XSUB_RE */
10880
10881 /*
10882 * ex: set ts=8 sts=4 sw=4 et:
10883 */
10884