5 * "One Ring to rule them all, One Ring to find them..."
8 /* This file contains functions for executing a regular expression. See
9 * also regcomp.c which funnily enough, contains functions for compiling
10 * a regular expression.
12 * This file is also copied at build time to ext/re/re_exec.c, where
13 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
14 * This causes the main functions to be compiled under new names and with
15 * debugging support added, which makes "use re 'debug'" work.
19 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
20 * confused with the original package (see point 3 below). Thanks, Henry!
23 /* Additional note: this code is very heavily munged from Henry's version
24 * in places. In some spots I've traded clarity for efficiency, so don't
25 * blame Henry for some of the lack of readability.
28 /* The names of the functions have been changed from regcomp and
29 * regexec to pregcomp and pregexec in order to avoid conflicts
30 * with the POSIX routines of the same names.
33 #ifdef PERL_EXT_RE_BUILD
34 /* need to replace pregcomp et al, so enable that */
35 # ifndef PERL_IN_XSUB_RE
36 # define PERL_IN_XSUB_RE
38 /* need access to debugger hooks */
39 # if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
44 #ifdef PERL_IN_XSUB_RE
45 /* We *really* need to overwrite these symbols: */
46 # define Perl_regexec_flags my_regexec
47 # define Perl_regdump my_regdump
48 # define Perl_regprop my_regprop
49 # define Perl_re_intuit_start my_re_intuit_start
50 /* *These* symbols are masked to allow static link. */
51 # define Perl_pregexec my_pregexec
52 # define Perl_reginitcolors my_reginitcolors
53 # define Perl_regclass_swash my_regclass_swash
55 # define PERL_NO_GET_CONTEXT
60 * pregcomp and pregexec -- regsub and regerror are not used in perl
62 * Copyright (c) 1986 by University of Toronto.
63 * Written by Henry Spencer. Not derived from licensed software.
65 * Permission is granted to anyone to use this software for any
66 * purpose on any computer system, and to redistribute it freely,
67 * subject to the following restrictions:
69 * 1. The author is not responsible for the consequences of use of
70 * this software, no matter how awful, even if they arise
73 * 2. The origin of this software must not be misrepresented, either
74 * by explicit claim or by omission.
76 * 3. Altered versions must be plainly marked as such, and must not
77 * be misrepresented as being the original software.
79 **** Alterations to Henry's code are...
81 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
82 **** 2000, 2001, 2002, 2003, 2004, 2005, by Larry Wall and others
84 **** You may distribute under the terms of either the GNU General Public
85 **** License or the Artistic License, as specified in the README file.
87 * Beware that some of this code is subtly aware of the way operator
88 * precedence is structured in regular expressions. Serious changes in
89 * regular-expression syntax might require a total rethink.
92 #define PERL_IN_REGEXEC_C
97 #define RF_tainted 1 /* tainted information used? */
98 #define RF_warned 2 /* warned about big count? */
99 #define RF_evaled 4 /* Did an EVAL with setting? */
100 #define RF_utf8 8 /* String contains multibyte chars? */
102 #define UTF ((PL_reg_flags & RF_utf8) != 0)
104 #define RS_init 1 /* eval environment created */
105 #define RS_set 2 /* replsv value is set */
108 #define STATIC static
111 #define REGINCLASS(p,c) (ANYOF_FLAGS(p) ? reginclass(p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
117 #define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv))
118 #define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)
120 #define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
121 #define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
122 #define HOP(pos,off) (PL_reg_match_utf8 ? reghop((U8*)pos, off) : (U8*)(pos + off))
123 #define HOPMAYBE(pos,off) (PL_reg_match_utf8 ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
124 #define HOPc(pos,off) ((char*)HOP(pos,off))
125 #define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
127 #define HOPBACK(pos, off) ( \
128 (PL_reg_match_utf8) \
129 ? reghopmaybe((U8*)pos, -off) \
130 : (pos - off >= PL_bostr) \
134 #define HOPBACKc(pos, off) (char*)HOPBACK(pos, off)
136 #define reghop3_c(pos,off,lim) ((char*)reghop3((U8*)pos, off, (U8*)lim))
137 #define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim))
138 #define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
139 #define HOPMAYBE3(pos,off,lim) (PL_reg_match_utf8 ? reghopmaybe3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
140 #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
141 #define HOPMAYBE3c(pos,off,lim) ((char*)HOPMAYBE3(pos,off,lim))
143 #define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) { ENTER; save_re_context(); (void)CAT2(is_utf8_, a)((const U8*)b); LEAVE; } } STMT_END
145 /* for use after a quantifier and before an EXACT-like node -- japhy */
146 #define JUMPABLE(rn) ( \
147 OP(rn) == OPEN || OP(rn) == CLOSE || OP(rn) == EVAL || \
148 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
149 OP(rn) == PLUS || OP(rn) == MINMOD || \
150 (PL_regkind[(U8)OP(rn)] == CURLY && ARG1(rn) > 0) \
153 #define HAS_TEXT(rn) ( \
154 PL_regkind[(U8)OP(rn)] == EXACT || PL_regkind[(U8)OP(rn)] == REF \
158 Search for mandatory following text node; for lookahead, the text must
159 follow but for lookbehind (rn->flags != 0) we skip to the next step.
161 #define FIND_NEXT_IMPT(rn) STMT_START { \
162 while (JUMPABLE(rn)) \
163 if (OP(rn) == SUSPEND || PL_regkind[(U8)OP(rn)] == CURLY) \
164 rn = NEXTOPER(NEXTOPER(rn)); \
165 else if (OP(rn) == PLUS) \
167 else if (OP(rn) == IFMATCH) \
168 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
169 else rn += NEXT_OFF(rn); \
172 static void restore_pos(pTHX_ void *arg);
175 S_regcppush(pTHX_ I32 parenfloor)
177 const int retval = PL_savestack_ix;
178 #define REGCP_PAREN_ELEMS 4
179 const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
182 if (paren_elems_to_push < 0)
183 Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
185 #define REGCP_OTHER_ELEMS 6
186 SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
187 for (p = PL_regsize; p > parenfloor; p--) {
188 /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
189 SSPUSHINT(PL_regendp[p]);
190 SSPUSHINT(PL_regstartp[p]);
191 SSPUSHPTR(PL_reg_start_tmp[p]);
194 /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
195 SSPUSHINT(PL_regsize);
196 SSPUSHINT(*PL_reglastparen);
197 SSPUSHINT(*PL_reglastcloseparen);
198 SSPUSHPTR(PL_reginput);
199 #define REGCP_FRAME_ELEMS 2
200 /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and
201 * are needed for the regexp context stack bookkeeping. */
202 SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
203 SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */
208 /* These are needed since we do not localize EVAL nodes: */
209 # define REGCP_SET(cp) DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, \
210 " Setting an EVAL scope, savestack=%"IVdf"\n", \
211 (IV)PL_savestack_ix)); cp = PL_savestack_ix
213 # define REGCP_UNWIND(cp) DEBUG_EXECUTE_r(cp != PL_savestack_ix ? \
214 PerlIO_printf(Perl_debug_log, \
215 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
216 (IV)(cp), (IV)PL_savestack_ix) : 0); regcpblow(cp)
225 GET_RE_DEBUG_FLAGS_DECL;
227 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
229 assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
230 i = SSPOPINT; /* Parentheses elements to pop. */
231 input = (char *) SSPOPPTR;
232 *PL_reglastcloseparen = SSPOPINT;
233 *PL_reglastparen = SSPOPINT;
234 PL_regsize = SSPOPINT;
236 /* Now restore the parentheses context. */
237 for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
238 i > 0; i -= REGCP_PAREN_ELEMS) {
240 paren = (U32)SSPOPINT;
241 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
242 PL_regstartp[paren] = SSPOPINT;
244 if (paren <= *PL_reglastparen)
245 PL_regendp[paren] = tmps;
247 PerlIO_printf(Perl_debug_log,
248 " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
249 (UV)paren, (IV)PL_regstartp[paren],
250 (IV)(PL_reg_start_tmp[paren] - PL_bostr),
251 (IV)PL_regendp[paren],
252 (paren > *PL_reglastparen ? "(no)" : ""));
256 if ((I32)(*PL_reglastparen + 1) <= PL_regnpar) {
257 PerlIO_printf(Perl_debug_log,
258 " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
259 (IV)(*PL_reglastparen + 1), (IV)PL_regnpar);
263 /* It would seem that the similar code in regtry()
264 * already takes care of this, and in fact it is in
265 * a better location to since this code can #if 0-ed out
266 * but the code in regtry() is needed or otherwise tests
267 * requiring null fields (pat.t#187 and split.t#{13,14}
268 * (as of patchlevel 7877) will fail. Then again,
269 * this code seems to be necessary or otherwise
270 * building DynaLoader will fail:
271 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
273 for (paren = *PL_reglastparen + 1; (I32)paren <= PL_regnpar; paren++) {
274 if ((I32)paren > PL_regsize)
275 PL_regstartp[paren] = -1;
276 PL_regendp[paren] = -1;
283 S_regcp_set_to(pTHX_ I32 ss)
285 const I32 tmp = PL_savestack_ix;
287 PL_savestack_ix = ss;
289 PL_savestack_ix = tmp;
293 typedef struct re_cc_state
297 struct re_cc_state *prev;
302 #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
304 #define TRYPAREN(paren, n, input) { \
307 PL_regstartp[paren] = HOPc(input, -1) - PL_bostr; \
308 PL_regendp[paren] = input - PL_bostr; \
311 PL_regendp[paren] = -1; \
313 if (regmatch(next)) \
316 PL_regendp[paren] = -1; \
321 * pregexec and friends
325 - pregexec - match a regexp against a string
328 Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend,
329 char *strbeg, I32 minend, SV *screamer, U32 nosave)
330 /* strend: pointer to null at end of string */
331 /* strbeg: real beginning of string */
332 /* minend: end of match must be >=minend after stringarg. */
333 /* nosave: For optimizations. */
336 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
337 nosave ? 0 : REXEC_COPY_STR);
341 S_cache_re(pTHX_ regexp *prog)
343 PL_regprecomp = prog->precomp; /* Needed for FAIL. */
345 PL_regprogram = prog->program;
347 PL_regnpar = prog->nparens;
348 PL_regdata = prog->data;
353 * Need to implement the following flags for reg_anch:
355 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
357 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
358 * INTUIT_AUTORITATIVE_ML
359 * INTUIT_ONCE_NOML - Intuit can match in one location only.
362 * Another flag for this function: SECOND_TIME (so that float substrs
363 * with giant delta may be not rechecked).
366 /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
368 /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend.
369 Otherwise, only SvCUR(sv) is used to get strbeg. */
371 /* XXXX We assume that strpos is strbeg unless sv. */
373 /* XXXX Some places assume that there is a fixed substring.
374 An update may be needed if optimizer marks as "INTUITable"
375 RExen without fixed substrings. Similarly, it is assumed that
376 lengths of all the strings are no more than minlen, thus they
377 cannot come from lookahead.
378 (Or minlen should take into account lookahead.) */
380 /* A failure to find a constant substring means that there is no need to make
381 an expensive call to REx engine, thus we celebrate a failure. Similarly,
382 finding a substring too deep into the string means that less calls to
383 regtry() should be needed.
385 REx compiler's optimizer found 4 possible hints:
386 a) Anchored substring;
388 c) Whether we are anchored (beginning-of-line or \G);
389 d) First node (of those at offset 0) which may distingush positions;
390 We use a)b)d) and multiline-part of c), and try to find a position in the
391 string which does not contradict any of them.
394 /* Most of decisions we do here should have been done at compile time.
395 The nodes of the REx which we used for the search should have been
396 deleted from the finite automaton. */
399 Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
400 char *strend, U32 flags, re_scream_pos_data *data)
402 register I32 start_shift = 0;
403 /* Should be nonnegative! */
404 register I32 end_shift = 0;
409 const int do_utf8 = sv ? SvUTF8(sv) : 0; /* if no sv we have to assume bytes */
411 register char *other_last = Nullch; /* other substr checked before this */
412 char *check_at = Nullch; /* check substr found at this pos */
413 const I32 multiline = prog->reganch & PMf_MULTILINE;
415 char *i_strpos = strpos;
416 SV *dsv = PERL_DEBUG_PAD_ZERO(0);
419 GET_RE_DEBUG_FLAGS_DECL;
421 RX_MATCH_UTF8_set(prog,do_utf8);
423 if (prog->reganch & ROPT_UTF8) {
424 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
425 "UTF-8 regex...\n"));
426 PL_reg_flags |= RF_utf8;
430 const char *s = PL_reg_match_utf8 ?
431 sv_uni_display(dsv, sv, 60, UNI_DISPLAY_REGEX) :
433 const int len = PL_reg_match_utf8 ?
434 strlen(s) : strend - strpos;
437 if (PL_reg_match_utf8)
438 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
439 "UTF-8 target...\n"));
440 PerlIO_printf(Perl_debug_log,
441 "%sGuessing start of match, REx%s \"%s%.60s%s%s\" against \"%s%.*s%s%s\"...\n",
442 PL_colors[4], PL_colors[5], PL_colors[0],
445 (strlen(prog->precomp) > 60 ? "..." : ""),
447 (int)(len > 60 ? 60 : len),
449 (len > 60 ? "..." : "")
453 /* CHR_DIST() would be more correct here but it makes things slow. */
454 if (prog->minlen > strend - strpos) {
455 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
456 "String too short... [re_intuit_start]\n"));
459 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
462 if (!prog->check_utf8 && prog->check_substr)
463 to_utf8_substr(prog);
464 check = prog->check_utf8;
466 if (!prog->check_substr && prog->check_utf8)
467 to_byte_substr(prog);
468 check = prog->check_substr;
470 if (check == &PL_sv_undef) {
471 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
472 "Non-utf string cannot match utf check string\n"));
475 if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */
476 ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE)
477 || ( (prog->reganch & ROPT_ANCH_BOL)
478 && !multiline ) ); /* Check after \n? */
481 if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */
482 | ROPT_IMPLICIT)) /* not a real BOL */
483 /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
485 && (strpos != strbeg)) {
486 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
489 if (prog->check_offset_min == prog->check_offset_max &&
490 !(prog->reganch & ROPT_CANY_SEEN)) {
491 /* Substring at constant offset from beg-of-str... */
494 s = HOP3c(strpos, prog->check_offset_min, strend);
496 slen = SvCUR(check); /* >= 1 */
498 if ( strend - s > slen || strend - s < slen - 1
499 || (strend - s == slen && strend[-1] != '\n')) {
500 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
503 /* Now should match s[0..slen-2] */
505 if (slen && (*SvPVX_const(check) != *s
507 && memNE(SvPVX_const(check), s, slen)))) {
509 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
513 else if (*SvPVX_const(check) != *s
514 || ((slen = SvCUR(check)) > 1
515 && memNE(SvPVX_const(check), s, slen)))
517 goto success_at_start;
520 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
522 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
523 end_shift = prog->minlen - start_shift -
524 CHR_SVLEN(check) + (SvTAIL(check) != 0);
526 const I32 end = prog->check_offset_max + CHR_SVLEN(check)
527 - (SvTAIL(check) != 0);
528 const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
530 if (end_shift < eshift)
534 else { /* Can match at random position */
537 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
538 /* Should be nonnegative! */
539 end_shift = prog->minlen - start_shift -
540 CHR_SVLEN(check) + (SvTAIL(check) != 0);
543 #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */
545 Perl_croak(aTHX_ "panic: end_shift");
549 /* Find a possible match in the region s..strend by looking for
550 the "check" substring in the region corrected by start/end_shift. */
551 if (flags & REXEC_SCREAM) {
552 I32 p = -1; /* Internal iterator of scream. */
553 I32 * const pp = data ? data->scream_pos : &p;
555 if (PL_screamfirst[BmRARE(check)] >= 0
556 || ( BmRARE(check) == '\n'
557 && (BmPREVIOUS(check) == SvCUR(check) - 1)
559 s = screaminstr(sv, check,
560 start_shift + (s - strbeg), end_shift, pp, 0);
563 /* we may be pointing at the wrong string */
564 if (s && RX_MATCH_COPIED(prog))
565 s = strbeg + (s - SvPVX_const(sv));
567 *data->scream_olds = s;
569 else if (prog->reganch & ROPT_CANY_SEEN)
570 s = fbm_instr((U8*)(s + start_shift),
571 (U8*)(strend - end_shift),
572 check, multiline ? FBMrf_MULTILINE : 0);
574 s = fbm_instr(HOP3(s, start_shift, strend),
575 HOP3(strend, -end_shift, strbeg),
576 check, multiline ? FBMrf_MULTILINE : 0);
578 /* Update the count-of-usability, remove useless subpatterns,
581 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s %s substr \"%s%.*s%s\"%s%s",
582 (s ? "Found" : "Did not find"),
583 (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) ? "anchored" : "floating"),
585 (int)(SvCUR(check) - (SvTAIL(check)!=0)),
587 PL_colors[1], (SvTAIL(check) ? "$" : ""),
588 (s ? " at offset " : "...\n") ) );
595 /* Finish the diagnostic message */
596 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
598 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
599 Start with the other substr.
600 XXXX no SCREAM optimization yet - and a very coarse implementation
601 XXXX /ttx+/ results in anchored="ttx", floating="x". floating will
602 *always* match. Probably should be marked during compile...
603 Probably it is right to do no SCREAM here...
606 if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) : (prog->float_substr && prog->anchored_substr)) {
607 /* Take into account the "other" substring. */
608 /* XXXX May be hopelessly wrong for UTF... */
611 if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) {
614 char *last = HOP3c(s, -start_shift, strbeg), *last1, *last2;
618 t = s - prog->check_offset_max;
619 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
621 || ((t = reghopmaybe3_c(s, -(prog->check_offset_max), strpos))
626 t = HOP3c(t, prog->anchored_offset, strend);
627 if (t < other_last) /* These positions already checked */
629 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
632 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
633 /* On end-of-str: see comment below. */
634 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
635 if (must == &PL_sv_undef) {
637 DEBUG_EXECUTE_r(must = prog->anchored_utf8); /* for debug */
642 HOP3(HOP3(last1, prog->anchored_offset, strend)
643 + SvCUR(must), -(SvTAIL(must)!=0), strbeg),
645 multiline ? FBMrf_MULTILINE : 0
647 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
648 "%s anchored substr \"%s%.*s%s\"%s",
649 (s ? "Found" : "Contradicts"),
652 - (SvTAIL(must)!=0)),
654 PL_colors[1], (SvTAIL(must) ? "$" : "")));
656 if (last1 >= last2) {
657 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
658 ", giving up...\n"));
661 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
662 ", trying floating at offset %ld...\n",
663 (long)(HOP3c(s1, 1, strend) - i_strpos)));
664 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
665 s = HOP3c(last, 1, strend);
669 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
670 (long)(s - i_strpos)));
671 t = HOP3c(s, -prog->anchored_offset, strbeg);
672 other_last = HOP3c(s, 1, strend);
680 else { /* Take into account the floating substring. */
685 t = HOP3c(s, -start_shift, strbeg);
687 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
688 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
689 last = HOP3c(t, prog->float_max_offset, strend);
690 s = HOP3c(t, prog->float_min_offset, strend);
693 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
694 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
695 /* fbm_instr() takes into account exact value of end-of-str
696 if the check is SvTAIL(ed). Since false positives are OK,
697 and end-of-str is not later than strend we are OK. */
698 if (must == &PL_sv_undef) {
700 DEBUG_EXECUTE_r(must = prog->float_utf8); /* for debug message */
703 s = fbm_instr((unsigned char*)s,
704 (unsigned char*)last + SvCUR(must)
706 must, multiline ? FBMrf_MULTILINE : 0);
707 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s floating substr \"%s%.*s%s\"%s",
708 (s ? "Found" : "Contradicts"),
710 (int)(SvCUR(must) - (SvTAIL(must)!=0)),
712 PL_colors[1], (SvTAIL(must) ? "$" : "")));
715 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
716 ", giving up...\n"));
719 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
720 ", trying anchored starting at offset %ld...\n",
721 (long)(s1 + 1 - i_strpos)));
723 s = HOP3c(t, 1, strend);
727 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
728 (long)(s - i_strpos)));
729 other_last = s; /* Fix this later. --Hugo */
738 t = s - prog->check_offset_max;
739 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
741 || ((t = reghopmaybe3_c(s, -prog->check_offset_max, strpos))
743 /* Fixed substring is found far enough so that the match
744 cannot start at strpos. */
746 if (ml_anch && t[-1] != '\n') {
747 /* Eventually fbm_*() should handle this, but often
748 anchored_offset is not 0, so this check will not be wasted. */
749 /* XXXX In the code below we prefer to look for "^" even in
750 presence of anchored substrings. And we search even
751 beyond the found float position. These pessimizations
752 are historical artefacts only. */
754 while (t < strend - prog->minlen) {
756 if (t < check_at - prog->check_offset_min) {
757 if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) {
758 /* Since we moved from the found position,
759 we definitely contradict the found anchored
760 substr. Due to the above check we do not
761 contradict "check" substr.
762 Thus we can arrive here only if check substr
763 is float. Redo checking for "other"=="fixed".
766 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
767 PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
768 goto do_other_anchored;
770 /* We don't contradict the found floating substring. */
771 /* XXXX Why not check for STCLASS? */
773 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
774 PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
777 /* Position contradicts check-string */
778 /* XXXX probably better to look for check-string
779 than for "\n", so one should lower the limit for t? */
780 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
781 PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
782 other_last = strpos = s = t + 1;
787 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
788 PL_colors[0], PL_colors[1]));
792 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
793 PL_colors[0], PL_colors[1]));
797 ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
800 /* The found string does not prohibit matching at strpos,
801 - no optimization of calling REx engine can be performed,
802 unless it was an MBOL and we are not after MBOL,
803 or a future STCLASS check will fail this. */
805 /* Even in this situation we may use MBOL flag if strpos is offset
806 wrt the start of the string. */
807 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
808 && (strpos != strbeg) && strpos[-1] != '\n'
809 /* May be due to an implicit anchor of m{.*foo} */
810 && !(prog->reganch & ROPT_IMPLICIT))
815 DEBUG_EXECUTE_r( if (ml_anch)
816 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
817 (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
820 if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */
822 prog->check_utf8 /* Could be deleted already */
823 && --BmUSEFUL(prog->check_utf8) < 0
824 && (prog->check_utf8 == prog->float_utf8)
826 prog->check_substr /* Could be deleted already */
827 && --BmUSEFUL(prog->check_substr) < 0
828 && (prog->check_substr == prog->float_substr)
831 /* If flags & SOMETHING - do not do it many times on the same match */
832 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
833 SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr);
834 if (do_utf8 ? prog->check_substr : prog->check_utf8)
835 SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8);
836 prog->check_substr = prog->check_utf8 = Nullsv; /* disable */
837 prog->float_substr = prog->float_utf8 = Nullsv; /* clear */
838 check = Nullsv; /* abort */
840 /* XXXX This is a remnant of the old implementation. It
841 looks wasteful, since now INTUIT can use many
843 prog->reganch &= ~RE_USE_INTUIT;
850 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
851 if (prog->regstclass) {
852 /* minlen == 0 is possible if regstclass is \b or \B,
853 and the fixed substr is ''$.
854 Since minlen is already taken into account, s+1 is before strend;
855 accidentally, minlen >= 1 guaranties no false positives at s + 1
856 even for \b or \B. But (minlen? 1 : 0) below assumes that
857 regstclass does not come from lookahead... */
858 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
859 This leaves EXACTF only, which is dealt with in find_byclass(). */
860 const U8* str = (U8*)STRING(prog->regstclass);
861 const int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT
862 ? CHR_DIST(str+STR_LEN(prog->regstclass), str)
864 const char * const endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
865 ? HOP3c(s, (prog->minlen ? cl_l : 0), strend)
866 : (prog->float_substr || prog->float_utf8
867 ? HOP3c(HOP3c(check_at, -start_shift, strbeg),
873 s = find_byclass(prog, prog->regstclass, s, endpos, 1);
876 const char *what = 0;
878 if (endpos == strend) {
879 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
880 "Could not match STCLASS...\n") );
883 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
884 "This position contradicts STCLASS...\n") );
885 if ((prog->reganch & ROPT_ANCH) && !ml_anch)
887 /* Contradict one of substrings */
888 if (prog->anchored_substr || prog->anchored_utf8) {
889 if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) {
890 DEBUG_EXECUTE_r( what = "anchored" );
892 s = HOP3c(t, 1, strend);
893 if (s + start_shift + end_shift > strend) {
894 /* XXXX Should be taken into account earlier? */
895 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
896 "Could not match STCLASS...\n") );
901 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
902 "Looking for %s substr starting at offset %ld...\n",
903 what, (long)(s + start_shift - i_strpos)) );
906 /* Have both, check_string is floating */
907 if (t + start_shift >= check_at) /* Contradicts floating=check */
908 goto retry_floating_check;
909 /* Recheck anchored substring, but not floating... */
913 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
914 "Looking for anchored substr starting at offset %ld...\n",
915 (long)(other_last - i_strpos)) );
916 goto do_other_anchored;
918 /* Another way we could have checked stclass at the
919 current position only: */
924 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
925 "Looking for /%s^%s/m starting at offset %ld...\n",
926 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
929 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
931 /* Check is floating subtring. */
932 retry_floating_check:
933 t = check_at - start_shift;
934 DEBUG_EXECUTE_r( what = "floating" );
935 goto hop_and_restart;
938 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
939 "By STCLASS: moving %ld --> %ld\n",
940 (long)(t - i_strpos), (long)(s - i_strpos))
944 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
945 "Does not contradict STCLASS...\n");
950 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
951 PL_colors[4], (check ? "Guessed" : "Giving up"),
952 PL_colors[5], (long)(s - i_strpos)) );
955 fail_finish: /* Substring not found */
956 if (prog->check_substr || prog->check_utf8) /* could be removed already */
957 BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
959 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
960 PL_colors[4], PL_colors[5]));
964 /* We know what class REx starts with. Try to find this position... */
966 S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, const char *strend, I32 norun)
969 const I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
973 register STRLEN uskip;
977 register I32 tmp = 1; /* Scratch variable? */
978 register const bool do_utf8 = PL_reg_match_utf8;
980 /* We know what class it must start with. */
984 while (s + (uskip = UTF8SKIP(s)) <= strend) {
985 if ((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
986 !UTF8_IS_INVARIANT((U8)s[0]) ?
987 reginclass(c, (U8*)s, 0, do_utf8) :
988 REGINCLASS(c, (U8*)s)) {
989 if (tmp && (norun || regtry(prog, s)))
1000 while (s < strend) {
1003 if (REGINCLASS(c, (U8*)s) ||
1004 (ANYOF_FOLD_SHARP_S(c, s, strend) &&
1005 /* The assignment of 2 is intentional:
1006 * for the folded sharp s, the skip is 2. */
1007 (skip = SHARP_S_SKIP))) {
1008 if (tmp && (norun || regtry(prog, s)))
1020 while (s < strend) {
1021 if (tmp && (norun || regtry(prog, s)))
1030 ln = STR_LEN(c); /* length to match in octets/bytes */
1031 lnc = (I32) ln; /* length to match in characters */
1033 STRLEN ulen1, ulen2;
1035 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
1036 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
1038 to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1039 to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1041 c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
1042 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
1043 c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
1044 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
1046 while (sm < ((U8 *) m + ln)) {
1061 c2 = PL_fold_locale[c1];
1063 e = HOP3c(strend, -((I32)lnc), s);
1066 e = s; /* Due to minlen logic of intuit() */
1068 /* The idea in the EXACTF* cases is to first find the
1069 * first character of the EXACTF* node and then, if
1070 * necessary, case-insensitively compare the full
1071 * text of the node. The c1 and c2 are the first
1072 * characters (though in Unicode it gets a bit
1073 * more complicated because there are more cases
1074 * than just upper and lower: one needs to use
1075 * the so-called folding case for case-insensitive
1076 * matching (called "loose matching" in Unicode).
1077 * ibcmp_utf8() will do just that. */
1081 U8 tmpbuf [UTF8_MAXBYTES+1];
1082 STRLEN len, foldlen;
1085 /* Upper and lower of 1st char are equal -
1086 * probably not a "letter". */
1088 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
1090 0 : UTF8_ALLOW_ANY);
1093 ibcmp_utf8(s, (char **)0, 0, do_utf8,
1094 m, (char **)0, ln, (bool)UTF))
1095 && (norun || regtry(prog, s)) )
1098 U8 foldbuf[UTF8_MAXBYTES_CASE+1];
1099 uvchr_to_utf8(tmpbuf, c);
1100 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen);
1102 && (f == c1 || f == c2)
1103 && (ln == foldlen ||
1104 !ibcmp_utf8((char *) foldbuf,
1105 (char **)0, foldlen, do_utf8,
1107 (char **)0, ln, (bool)UTF))
1108 && (norun || regtry(prog, s)) )
1116 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
1118 0 : UTF8_ALLOW_ANY);
1120 /* Handle some of the three Greek sigmas cases.
1121 * Note that not all the possible combinations
1122 * are handled here: some of them are handled
1123 * by the standard folding rules, and some of
1124 * them (the character class or ANYOF cases)
1125 * are handled during compiletime in
1126 * regexec.c:S_regclass(). */
1127 if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
1128 c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
1129 c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
1131 if ( (c == c1 || c == c2)
1133 ibcmp_utf8(s, (char **)0, 0, do_utf8,
1134 m, (char **)0, ln, (bool)UTF))
1135 && (norun || regtry(prog, s)) )
1138 U8 foldbuf[UTF8_MAXBYTES_CASE+1];
1139 uvchr_to_utf8(tmpbuf, c);
1140 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen);
1142 && (f == c1 || f == c2)
1143 && (ln == foldlen ||
1144 !ibcmp_utf8((char *) foldbuf,
1145 (char **)0, foldlen, do_utf8,
1147 (char **)0, ln, (bool)UTF))
1148 && (norun || regtry(prog, s)) )
1159 && (ln == 1 || !(OP(c) == EXACTF
1161 : ibcmp_locale(s, m, ln)))
1162 && (norun || regtry(prog, s)) )
1168 if ( (*(U8*)s == c1 || *(U8*)s == c2)
1169 && (ln == 1 || !(OP(c) == EXACTF
1171 : ibcmp_locale(s, m, ln)))
1172 && (norun || regtry(prog, s)) )
1179 PL_reg_flags |= RF_tainted;
1186 U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr);
1188 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
1190 tmp = ((OP(c) == BOUND ?
1191 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1192 LOAD_UTF8_CHARCLASS(alnum,"a");
1193 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1194 if (tmp == !(OP(c) == BOUND ?
1195 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
1196 isALNUM_LC_utf8((U8*)s)))
1199 if ((norun || regtry(prog, s)))
1206 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
1207 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1208 while (s < strend) {
1210 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
1212 if ((norun || regtry(prog, s)))
1218 if ((!prog->minlen && tmp) && (norun || regtry(prog, s)))
1222 PL_reg_flags |= RF_tainted;
1229 U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr);
1231 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
1233 tmp = ((OP(c) == NBOUND ?
1234 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1235 LOAD_UTF8_CHARCLASS(alnum,"a");
1236 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1237 if (tmp == !(OP(c) == NBOUND ?
1238 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
1239 isALNUM_LC_utf8((U8*)s)))
1241 else if ((norun || regtry(prog, s)))
1247 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
1248 tmp = ((OP(c) == NBOUND ?
1249 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1250 while (s < strend) {
1252 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1254 else if ((norun || regtry(prog, s)))
1259 if ((!prog->minlen && !tmp) && (norun || regtry(prog, s)))
1264 LOAD_UTF8_CHARCLASS(alnum,"a");
1265 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1266 if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
1267 if (tmp && (norun || regtry(prog, s)))
1278 while (s < strend) {
1280 if (tmp && (norun || regtry(prog, s)))
1292 PL_reg_flags |= RF_tainted;
1294 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1295 if (isALNUM_LC_utf8((U8*)s)) {
1296 if (tmp && (norun || regtry(prog, s)))
1307 while (s < strend) {
1308 if (isALNUM_LC(*s)) {
1309 if (tmp && (norun || regtry(prog, s)))
1322 LOAD_UTF8_CHARCLASS(alnum,"a");
1323 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1324 if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
1325 if (tmp && (norun || regtry(prog, s)))
1336 while (s < strend) {
1338 if (tmp && (norun || regtry(prog, s)))
1350 PL_reg_flags |= RF_tainted;
1352 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1353 if (!isALNUM_LC_utf8((U8*)s)) {
1354 if (tmp && (norun || regtry(prog, s)))
1365 while (s < strend) {
1366 if (!isALNUM_LC(*s)) {
1367 if (tmp && (norun || regtry(prog, s)))
1380 LOAD_UTF8_CHARCLASS(space," ");
1381 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1382 if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) {
1383 if (tmp && (norun || regtry(prog, s)))
1394 while (s < strend) {
1396 if (tmp && (norun || regtry(prog, s)))
1408 PL_reg_flags |= RF_tainted;
1410 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1411 if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
1412 if (tmp && (norun || regtry(prog, s)))
1423 while (s < strend) {
1424 if (isSPACE_LC(*s)) {
1425 if (tmp && (norun || regtry(prog, s)))
1438 LOAD_UTF8_CHARCLASS(space," ");
1439 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1440 if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) {
1441 if (tmp && (norun || regtry(prog, s)))
1452 while (s < strend) {
1454 if (tmp && (norun || regtry(prog, s)))
1466 PL_reg_flags |= RF_tainted;
1468 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1469 if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
1470 if (tmp && (norun || regtry(prog, s)))
1481 while (s < strend) {
1482 if (!isSPACE_LC(*s)) {
1483 if (tmp && (norun || regtry(prog, s)))
1496 LOAD_UTF8_CHARCLASS(digit,"0");
1497 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1498 if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
1499 if (tmp && (norun || regtry(prog, s)))
1510 while (s < strend) {
1512 if (tmp && (norun || regtry(prog, s)))
1524 PL_reg_flags |= RF_tainted;
1526 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1527 if (isDIGIT_LC_utf8((U8*)s)) {
1528 if (tmp && (norun || regtry(prog, s)))
1539 while (s < strend) {
1540 if (isDIGIT_LC(*s)) {
1541 if (tmp && (norun || regtry(prog, s)))
1554 LOAD_UTF8_CHARCLASS(digit,"0");
1555 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1556 if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
1557 if (tmp && (norun || regtry(prog, s)))
1568 while (s < strend) {
1570 if (tmp && (norun || regtry(prog, s)))
1582 PL_reg_flags |= RF_tainted;
1584 while (s + (uskip = UTF8SKIP(s)) <= strend) {
1585 if (!isDIGIT_LC_utf8((U8*)s)) {
1586 if (tmp && (norun || regtry(prog, s)))
1597 while (s < strend) {
1598 if (!isDIGIT_LC(*s)) {
1599 if (tmp && (norun || regtry(prog, s)))
1611 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
1620 - regexec_flags - match a regexp against a string
1623 Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend,
1624 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
1625 /* strend: pointer to null at end of string */
1626 /* strbeg: real beginning of string */
1627 /* minend: end of match must be >=minend after stringarg. */
1628 /* data: May be used for some additional optimizations. */
1629 /* nosave: For optimizations. */
1632 register regnode *c;
1633 register char *startpos = stringarg;
1634 I32 minlen; /* must match at least this many chars */
1635 I32 dontbother = 0; /* how many characters not to try at end */
1636 I32 end_shift = 0; /* Same for the end. */ /* CC */
1637 I32 scream_pos = -1; /* Internal iterator of scream. */
1639 SV* oreplsv = GvSV(PL_replgv);
1640 const bool do_utf8 = DO_UTF8(sv);
1641 const I32 multiline = prog->reganch & PMf_MULTILINE;
1643 SV *dsv0 = PERL_DEBUG_PAD_ZERO(0);
1644 SV *dsv1 = PERL_DEBUG_PAD_ZERO(1);
1647 GET_RE_DEBUG_FLAGS_DECL;
1649 (void)data; /* Currently unused */
1650 RX_MATCH_UTF8_set(prog,do_utf8);
1656 PL_regnarrate = DEBUG_r_TEST;
1659 /* Be paranoid... */
1660 if (prog == NULL || startpos == NULL) {
1661 Perl_croak(aTHX_ "NULL regexp parameter");
1665 minlen = prog->minlen;
1666 if (strend - startpos < minlen) {
1667 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1668 "String too short [regexec_flags]...\n"));
1672 /* Check validity of program. */
1673 if (UCHARAT(prog->program) != REG_MAGIC) {
1674 Perl_croak(aTHX_ "corrupted regexp program");
1678 PL_reg_eval_set = 0;
1681 if (prog->reganch & ROPT_UTF8)
1682 PL_reg_flags |= RF_utf8;
1684 /* Mark beginning of line for ^ and lookbehind. */
1685 PL_regbol = startpos;
1689 /* Mark end of line for $ (and such) */
1692 /* see how far we have to get to not match where we matched before */
1693 PL_regtill = startpos+minend;
1695 /* We start without call_cc context. */
1698 /* If there is a "must appear" string, look for it. */
1701 if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to have PL_reg_ganch */
1704 if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
1705 PL_reg_ganch = startpos;
1706 else if (sv && SvTYPE(sv) >= SVt_PVMG
1708 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
1709 && mg->mg_len >= 0) {
1710 PL_reg_ganch = strbeg + mg->mg_len; /* Defined pos() */
1711 if (prog->reganch & ROPT_ANCH_GPOS) {
1712 if (s > PL_reg_ganch)
1717 else /* pos() not defined */
1718 PL_reg_ganch = strbeg;
1721 if (!(flags & REXEC_CHECKED) && (prog->check_substr != Nullsv || prog->check_utf8 != Nullsv)) {
1722 re_scream_pos_data d;
1724 d.scream_olds = &scream_olds;
1725 d.scream_pos = &scream_pos;
1726 s = re_intuit_start(prog, sv, s, strend, flags, &d);
1728 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
1729 goto phooey; /* not present */
1734 const char * const s0 = UTF
1735 ? pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60,
1738 const int len0 = UTF ? SvCUR(dsv0) : prog->prelen;
1739 const char * const s1 = do_utf8 ? sv_uni_display(dsv1, sv, 60,
1740 UNI_DISPLAY_REGEX) : startpos;
1741 const int len1 = do_utf8 ? SvCUR(dsv1) : strend - startpos;
1744 PerlIO_printf(Perl_debug_log,
1745 "%sMatching REx%s \"%s%*.*s%s%s\" against \"%s%.*s%s%s\"\n",
1746 PL_colors[4], PL_colors[5], PL_colors[0],
1749 len0 > 60 ? "..." : "",
1751 (int)(len1 > 60 ? 60 : len1),
1753 (len1 > 60 ? "..." : "")
1757 /* Simplest case: anchored match need be tried only once. */
1758 /* [unless only anchor is BOL and multiline is set] */
1759 if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
1760 if (s == startpos && regtry(prog, startpos))
1762 else if (multiline || (prog->reganch & ROPT_IMPLICIT)
1763 || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
1768 dontbother = minlen - 1;
1769 end = HOP3c(strend, -dontbother, strbeg) - 1;
1770 /* for multiline we only have to try after newlines */
1771 if (prog->check_substr || prog->check_utf8) {
1775 if (regtry(prog, s))
1780 if (prog->reganch & RE_USE_INTUIT) {
1781 s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
1792 if (*s++ == '\n') { /* don't need PL_utf8skip here */
1793 if (regtry(prog, s))
1800 } else if (prog->reganch & ROPT_ANCH_GPOS) {
1801 if (regtry(prog, PL_reg_ganch))
1806 /* Messy cases: unanchored match. */
1807 if ((prog->anchored_substr || prog->anchored_utf8) && prog->reganch & ROPT_SKIP) {
1808 /* we have /x+whatever/ */
1809 /* it must be a one character string (XXXX Except UTF?) */
1814 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1815 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1816 ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
1819 while (s < strend) {
1821 DEBUG_EXECUTE_r( did_match = 1 );
1822 if (regtry(prog, s)) goto got_it;
1824 while (s < strend && *s == ch)
1831 while (s < strend) {
1833 DEBUG_EXECUTE_r( did_match = 1 );
1834 if (regtry(prog, s)) goto got_it;
1836 while (s < strend && *s == ch)
1842 DEBUG_EXECUTE_r(if (!did_match)
1843 PerlIO_printf(Perl_debug_log,
1844 "Did not find anchored character...\n")
1848 else if (prog->anchored_substr != Nullsv
1849 || prog->anchored_utf8 != Nullsv
1850 || ((prog->float_substr != Nullsv || prog->float_utf8 != Nullsv)
1851 && prog->float_max_offset < strend - s)) {
1856 char *last1; /* Last position checked before */
1860 if (prog->anchored_substr || prog->anchored_utf8) {
1861 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1862 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1863 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
1864 back_max = back_min = prog->anchored_offset;
1866 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1867 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1868 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
1869 back_max = prog->float_max_offset;
1870 back_min = prog->float_min_offset;
1872 if (must == &PL_sv_undef)
1873 /* could not downgrade utf8 check substring, so must fail */
1876 last = HOP3c(strend, /* Cannot start after this */
1877 -(I32)(CHR_SVLEN(must)
1878 - (SvTAIL(must) != 0) + back_min), strbeg);
1881 last1 = HOPc(s, -1);
1883 last1 = s - 1; /* bogus */
1885 /* XXXX check_substr already used to find "s", can optimize if
1886 check_substr==must. */
1888 dontbother = end_shift;
1889 strend = HOPc(strend, -dontbother);
1890 while ( (s <= last) &&
1891 ((flags & REXEC_SCREAM)
1892 ? (s = screaminstr(sv, must, HOP3c(s, back_min, strend) - strbeg,
1893 end_shift, &scream_pos, 0))
1894 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend),
1895 (unsigned char*)strend, must,
1896 multiline ? FBMrf_MULTILINE : 0))) ) {
1897 /* we may be pointing at the wrong string */
1898 if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog))
1899 s = strbeg + (s - SvPVX_const(sv));
1900 DEBUG_EXECUTE_r( did_match = 1 );
1901 if (HOPc(s, -back_max) > last1) {
1902 last1 = HOPc(s, -back_min);
1903 s = HOPc(s, -back_max);
1906 char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
1908 last1 = HOPc(s, -back_min);
1912 while (s <= last1) {
1913 if (regtry(prog, s))
1919 while (s <= last1) {
1920 if (regtry(prog, s))
1926 DEBUG_EXECUTE_r(if (!did_match)
1927 PerlIO_printf(Perl_debug_log,
1928 "Did not find %s substr \"%s%.*s%s\"%s...\n",
1929 ((must == prog->anchored_substr || must == prog->anchored_utf8)
1930 ? "anchored" : "floating"),
1932 (int)(SvCUR(must) - (SvTAIL(must)!=0)),
1934 PL_colors[1], (SvTAIL(must) ? "$" : ""))
1938 else if ((c = prog->regstclass)) {
1940 I32 op = (U8)OP(prog->regstclass);
1941 /* don't bother with what can't match */
1942 if (PL_regkind[op] != EXACT && op != CANY)
1943 strend = HOPc(strend, -(minlen - 1));
1946 SV *prop = sv_newmortal();
1954 pv_uni_display(dsv0, (U8*)SvPVX_const(prop), SvCUR(prop), 60,
1955 UNI_DISPLAY_REGEX) :
1957 len0 = UTF ? SvCUR(dsv0) : SvCUR(prop);
1959 sv_uni_display(dsv1, sv, 60, UNI_DISPLAY_REGEX) : s;
1960 len1 = UTF ? SvCUR(dsv1) : strend - s;
1961 PerlIO_printf(Perl_debug_log,
1962 "Matching stclass \"%*.*s\" against \"%*.*s\"\n",
1966 if (find_byclass(prog, c, s, strend, 0))
1968 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n"));
1972 if (prog->float_substr != Nullsv || prog->float_utf8 != Nullsv) {
1977 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1978 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1979 float_real = do_utf8 ? prog->float_utf8 : prog->float_substr;
1981 if (flags & REXEC_SCREAM) {
1982 last = screaminstr(sv, float_real, s - strbeg,
1983 end_shift, &scream_pos, 1); /* last one */
1985 last = scream_olds; /* Only one occurrence. */
1986 /* we may be pointing at the wrong string */
1987 else if (RX_MATCH_COPIED(prog))
1988 s = strbeg + (s - SvPVX_const(sv));
1992 const char * const little = SvPV(float_real, len);
1994 if (SvTAIL(float_real)) {
1995 if (memEQ(strend - len + 1, little, len - 1))
1996 last = strend - len + 1;
1997 else if (!multiline)
1998 last = memEQ(strend - len, little, len)
1999 ? strend - len : Nullch;
2005 last = rninstr(s, strend, little, little + len);
2007 last = strend; /* matching "$" */
2011 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2012 "%sCan't trim the tail, match fails (should not happen)%s\n",
2013 PL_colors[4], PL_colors[5]));
2014 goto phooey; /* Should not happen! */
2016 dontbother = strend - last + prog->float_min_offset;
2018 if (minlen && (dontbother < minlen))
2019 dontbother = minlen - 1;
2020 strend -= dontbother; /* this one's always in bytes! */
2021 /* We don't know much -- general case. */
2024 if (regtry(prog, s))
2033 if (regtry(prog, s))
2035 } while (s++ < strend);
2043 RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
2045 if (PL_reg_eval_set) {
2046 /* Preserve the current value of $^R */
2047 if (oreplsv != GvSV(PL_replgv))
2048 sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
2049 restored, the value remains
2051 restore_pos(aTHX_ 0);
2054 /* make sure $`, $&, $', and $digit will work later */
2055 if ( !(flags & REXEC_NOT_FIRST) ) {
2056 RX_MATCH_COPY_FREE(prog);
2057 if (flags & REXEC_COPY_STR) {
2058 I32 i = PL_regeol - startpos + (stringarg - strbeg);
2059 #ifdef PERL_COPY_ON_WRITE
2061 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2063 PerlIO_printf(Perl_debug_log,
2064 "Copy on write: regexp capture, type %d\n",
2067 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
2068 prog->subbeg = SvPVX(prog->saved_copy);
2069 assert (SvPOKp(prog->saved_copy));
2073 RX_MATCH_COPIED_on(prog);
2074 s = savepvn(strbeg, i);
2080 prog->subbeg = strbeg;
2081 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2088 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
2089 PL_colors[4], PL_colors[5]));
2090 if (PL_reg_eval_set)
2091 restore_pos(aTHX_ 0);
2096 - regtry - try match at specific point
2098 STATIC I32 /* 0 failure, 1 success */
2099 S_regtry(pTHX_ regexp *prog, char *startpos)
2105 GET_RE_DEBUG_FLAGS_DECL;
2108 PL_regindent = 0; /* XXXX Not good when matches are reenterable... */
2110 if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
2113 PL_reg_eval_set = RS_init;
2114 DEBUG_EXECUTE_r(DEBUG_s(
2115 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
2116 (IV)(PL_stack_sp - PL_stack_base));
2118 SAVEI32(cxstack[cxstack_ix].blk_oldsp);
2119 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
2120 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
2122 /* Apparently this is not needed, judging by wantarray. */
2123 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
2124 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
2127 /* Make $_ available to executed code. */
2128 if (PL_reg_sv != DEFSV) {
2133 if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
2134 && (mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global)))) {
2135 /* prepare for quick setting of pos */
2136 sv_magic(PL_reg_sv, (SV*)0,
2137 PERL_MAGIC_regex_global, Nullch, 0);
2138 mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global);
2142 PL_reg_oldpos = mg->mg_len;
2143 SAVEDESTRUCTOR_X(restore_pos, 0);
2145 if (!PL_reg_curpm) {
2146 Newz(22, PL_reg_curpm, 1, PMOP);
2149 SV* repointer = newSViv(0);
2150 /* so we know which PL_regex_padav element is PL_reg_curpm */
2151 SvFLAGS(repointer) |= SVf_BREAK;
2152 av_push(PL_regex_padav,repointer);
2153 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2154 PL_regex_pad = AvARRAY(PL_regex_padav);
2158 PM_SETRE(PL_reg_curpm, prog);
2159 PL_reg_oldcurpm = PL_curpm;
2160 PL_curpm = PL_reg_curpm;
2161 if (RX_MATCH_COPIED(prog)) {
2162 /* Here is a serious problem: we cannot rewrite subbeg,
2163 since it may be needed if this match fails. Thus
2164 $` inside (?{}) could fail... */
2165 PL_reg_oldsaved = prog->subbeg;
2166 PL_reg_oldsavedlen = prog->sublen;
2167 #ifdef PERL_COPY_ON_WRITE
2168 PL_nrs = prog->saved_copy;
2170 RX_MATCH_COPIED_off(prog);
2173 PL_reg_oldsaved = Nullch;
2174 prog->subbeg = PL_bostr;
2175 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2177 prog->startp[0] = startpos - PL_bostr;
2178 PL_reginput = startpos;
2179 PL_regstartp = prog->startp;
2180 PL_regendp = prog->endp;
2181 PL_reglastparen = &prog->lastparen;
2182 PL_reglastcloseparen = &prog->lastcloseparen;
2183 prog->lastparen = 0;
2184 prog->lastcloseparen = 0;
2186 DEBUG_EXECUTE_r(PL_reg_starttry = startpos);
2187 if (PL_reg_start_tmpl <= prog->nparens) {
2188 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
2189 if(PL_reg_start_tmp)
2190 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2192 New(22, PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2195 /* XXXX What this code is doing here?!!! There should be no need
2196 to do this again and again, PL_reglastparen should take care of
2199 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2200 * Actually, the code in regcppop() (which Ilya may be meaning by
2201 * PL_reglastparen), is not needed at all by the test suite
2202 * (op/regexp, op/pat, op/split), but that code is needed, oddly
2203 * enough, for building DynaLoader, or otherwise this
2204 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
2205 * will happen. Meanwhile, this code *is* needed for the
2206 * above-mentioned test suite tests to succeed. The common theme
2207 * on those tests seems to be returning null fields from matches.
2212 if (prog->nparens) {
2213 for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
2220 if (regmatch(prog->program + 1)) {
2221 prog->endp[0] = PL_reginput - PL_bostr;
2224 REGCP_UNWIND(lastcp);
2228 #define RE_UNWIND_BRANCH 1
2229 #define RE_UNWIND_BRANCHJ 2
2233 typedef struct { /* XX: makes sense to enlarge it... */
2237 } re_unwind_generic_t;
2250 } re_unwind_branch_t;
2252 typedef union re_unwind_t {
2254 re_unwind_generic_t generic;
2255 re_unwind_branch_t branch;
2258 #define sayYES goto yes
2259 #define sayNO goto no
2260 #define sayNO_ANYOF goto no_anyof
2261 #define sayYES_FINAL goto yes_final
2262 #define sayYES_LOUD goto yes_loud
2263 #define sayNO_FINAL goto no_final
2264 #define sayNO_SILENT goto do_no
2265 #define saySAME(x) if (x) goto yes; else goto no
2267 #define POSCACHE_SUCCESS 0 /* caching success rather than failure */
2268 #define POSCACHE_SEEN 1 /* we know what we're caching */
2269 #define POSCACHE_START 2 /* the real cache: this bit maps to pos 0 */
2270 #define CACHEsayYES STMT_START { \
2271 if (cache_offset | cache_bit) { \
2272 if (!(PL_reg_poscache[0] & (1<<POSCACHE_SEEN))) \
2273 PL_reg_poscache[0] |= (1<<POSCACHE_SUCCESS) || (1<<POSCACHE_SEEN); \
2274 else if (!(PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))) { \
2275 /* cache records failure, but this is success */ \
2277 PerlIO_printf(Perl_debug_log, \
2278 "%*s (remove success from failure cache)\n", \
2279 REPORT_CODE_OFF+PL_regindent*2, "") \
2281 PL_reg_poscache[cache_offset] &= ~(1<<cache_bit); \
2286 #define CACHEsayNO STMT_START { \
2287 if (cache_offset | cache_bit) { \
2288 if (!(PL_reg_poscache[0] & (1<<POSCACHE_SEEN))) \
2289 PL_reg_poscache[0] |= (1<<POSCACHE_SEEN); \
2290 else if ((PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))) { \
2291 /* cache records success, but this is failure */ \
2293 PerlIO_printf(Perl_debug_log, \
2294 "%*s (remove failure from success cache)\n", \
2295 REPORT_CODE_OFF+PL_regindent*2, "") \
2297 PL_reg_poscache[cache_offset] &= ~(1<<cache_bit); \
2303 /* this is used to determine how far from the left messages like
2304 'failed...' are printed. Currently 29 makes these messages line
2305 up with the opcode they refer to. Earlier perls used 25 which
2306 left these messages outdented making reviewing a debug output
2309 #define REPORT_CODE_OFF 29
2312 /* Make sure there is a test for this +1 options in re_tests */
2313 #define TRIE_INITAL_ACCEPT_BUFFLEN 4;
2315 #define TRIE_CHECK_STATE_IS_ACCEPTING STMT_START { \
2316 if ( trie->states[ state ].wordnum ) { \
2317 if ( !accepted ) { \
2320 bufflen = TRIE_INITAL_ACCEPT_BUFFLEN ; \
2321 sv_accept_buff=NEWSV( 1234, \
2322 bufflen * sizeof(reg_trie_accepted) - 1 ); \
2323 SvCUR_set( sv_accept_buff, sizeof(reg_trie_accepted) ); \
2324 SvPOK_on( sv_accept_buff ); \
2325 sv_2mortal( sv_accept_buff ); \
2326 accept_buff = (reg_trie_accepted*)SvPV_nolen( sv_accept_buff );\
2328 if ( accepted >= bufflen ) { \
2330 accept_buff =(reg_trie_accepted*)SvGROW( sv_accept_buff, \
2331 bufflen * sizeof(reg_trie_accepted) ); \
2333 SvCUR_set( sv_accept_buff,SvCUR( sv_accept_buff ) \
2334 + sizeof( reg_trie_accepted ) ); \
2336 accept_buff[ accepted ].wordnum = trie->states[ state ].wordnum; \
2337 accept_buff[ accepted ].endpos = uc; \
2341 #define TRIE_HANDLE_CHAR STMT_START { \
2342 if ( uvc < 256 ) { \
2343 charid = trie->charmap[ uvc ]; \
2346 if( trie->widecharmap ) { \
2347 SV** svpp = (SV**)NULL; \
2348 svpp = hv_fetch( trie->widecharmap, (char*)&uvc, \
2349 sizeof( UV ), 0 ); \
2351 charid = (U16)SvIV( *svpp ); \
2356 ( base + charid > trie->uniquecharcount ) && \
2357 ( base + charid - 1 - trie->uniquecharcount < trie->lasttrans) && \
2358 trie->trans[ base + charid - 1 - trie->uniquecharcount ].check == state ) \
2360 state = trie->trans[ base + charid - 1 - trie->uniquecharcount ].next; \
2368 - regmatch - main matching routine
2370 * Conceptually the strategy is simple: check to see whether the current
2371 * node matches, call self recursively to see whether the rest matches,
2372 * and then act accordingly. In practice we make some effort to avoid
2373 * recursion, in particular by going through "ordinary" nodes (that don't
2374 * need to know whether the rest of the match failed) by a loop instead of
2377 /* [lwall] I've hoisted the register declarations to the outer block in order to
2378 * maybe save a little bit of pushing and popping on the stack. It also takes
2379 * advantage of machines that use a register save mask on subroutine entry.
2381 STATIC I32 /* 0 failure, 1 success */
2382 S_regmatch(pTHX_ regnode *prog)
2385 register regnode *scan; /* Current node. */
2386 regnode *next; /* Next node. */
2387 regnode *inner; /* Next node in internal branch. */
2388 register I32 nextchr; /* renamed nextchr - nextchar colides with
2389 function of same name */
2390 register I32 n; /* no or next */
2391 register I32 ln = 0; /* len or last */
2392 register char *s = Nullch; /* operand or save */
2393 register char *locinput = PL_reginput;
2394 register I32 c1 = 0, c2 = 0, paren; /* case fold search, parenth */
2395 int minmod = 0, sw = 0, logical = 0;
2398 /* used by the trie code */
2399 SV *sv_accept_buff = 0; /* accepting states we have traversed */
2400 reg_trie_accepted *accept_buff = 0; /* "" */
2401 reg_trie_data *trie; /* what trie are we using right now */
2402 U32 accepted = 0; /* how many accepting states we have seen*/
2405 I32 firstcp = PL_savestack_ix;
2407 const register bool do_utf8 = PL_reg_match_utf8;
2409 SV *dsv0 = PERL_DEBUG_PAD_ZERO(0);
2410 SV *dsv1 = PERL_DEBUG_PAD_ZERO(1);
2411 SV *dsv2 = PERL_DEBUG_PAD_ZERO(2);
2413 SV *re_debug_flags = NULL;
2423 /* Note that nextchr is a byte even in UTF */
2424 nextchr = UCHARAT(locinput);
2426 while (scan != NULL) {
2429 SV *prop = sv_newmortal();
2430 const int docolor = *PL_colors[0];
2431 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
2432 int l = (PL_regeol - locinput) > taill ? taill : (PL_regeol - locinput);
2433 /* The part of the string before starttry has one color
2434 (pref0_len chars), between starttry and current
2435 position another one (pref_len - pref0_len chars),
2436 after the current position the third one.
2437 We assume that pref0_len <= pref_len, otherwise we
2438 decrease pref0_len. */
2439 int pref_len = (locinput - PL_bostr) > (5 + taill) - l
2440 ? (5 + taill) - l : locinput - PL_bostr;
2443 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
2445 pref0_len = pref_len - (locinput - PL_reg_starttry);
2446 if (l + pref_len < (5 + taill) && l < PL_regeol - locinput)
2447 l = ( PL_regeol - locinput > (5 + taill) - pref_len
2448 ? (5 + taill) - pref_len : PL_regeol - locinput);
2449 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
2453 if (pref0_len > pref_len)
2454 pref0_len = pref_len;
2455 regprop(prop, scan);
2457 const char * const s0 =
2458 do_utf8 && OP(scan) != CANY ?
2459 pv_uni_display(dsv0, (U8*)(locinput - pref_len),
2460 pref0_len, 60, UNI_DISPLAY_REGEX) :
2461 locinput - pref_len;
2462 const int len0 = do_utf8 ? strlen(s0) : pref0_len;
2463 const char * const s1 = do_utf8 && OP(scan) != CANY ?
2464 pv_uni_display(dsv1, (U8*)(locinput - pref_len + pref0_len),
2465 pref_len - pref0_len, 60, UNI_DISPLAY_REGEX) :
2466 locinput - pref_len + pref0_len;
2467 const int len1 = do_utf8 ? strlen(s1) : pref_len - pref0_len;
2468 const char * const s2 = do_utf8 && OP(scan) != CANY ?
2469 pv_uni_display(dsv2, (U8*)locinput,
2470 PL_regeol - locinput, 60, UNI_DISPLAY_REGEX) :
2472 const int len2 = do_utf8 ? strlen(s2) : l;
2473 PerlIO_printf(Perl_debug_log,
2474 "%4"IVdf" <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3"IVdf":%*s%s\n",
2475 (IV)(locinput - PL_bostr),
2482 (docolor ? "" : "> <"),
2486 15 - l - pref_len + 1,
2488 (IV)(scan - PL_regprogram), PL_regindent*2, "",
2493 next = scan + NEXT_OFF(scan);
2499 if (locinput == PL_bostr)
2501 /* regtill = regbol; */
2506 if (locinput == PL_bostr ||
2507 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
2513 if (locinput == PL_bostr)
2517 if (locinput == PL_reg_ganch)
2523 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
2528 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
2530 if (PL_regeol - locinput > 1)
2534 if (PL_regeol != locinput)
2538 if (!nextchr && locinput >= PL_regeol)
2541 locinput += PL_utf8skip[nextchr];
2542 if (locinput > PL_regeol)
2544 nextchr = UCHARAT(locinput);
2547 nextchr = UCHARAT(++locinput);
2550 if (!nextchr && locinput >= PL_regeol)
2552 nextchr = UCHARAT(++locinput);
2555 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
2558 locinput += PL_utf8skip[nextchr];
2559 if (locinput > PL_regeol)
2561 nextchr = UCHARAT(locinput);
2564 nextchr = UCHARAT(++locinput);
2570 traverse the TRIE keeping track of all accepting states
2571 we transition through until we get to a failing node.
2573 we use two slightly different pieces of code to handle
2574 the traversal depending on whether its case sensitive or
2575 not. we reuse the accept code however. (this should probably
2576 be turned into a macro.)
2583 const U32 uniflags = ckWARN( WARN_UTF8 ) ? 0 : UTF8_ALLOW_ANY;
2584 U8 *uc = ( U8* )locinput;
2591 U8 *uscan = (U8*)NULL;
2595 trie = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ];
2597 while ( state && uc <= (U8*)PL_regeol ) {
2599 TRIE_CHECK_STATE_IS_ACCEPTING;
2601 base = trie->states[ state ].trans.base;
2603 DEBUG_TRIE_EXECUTE_r(
2604 PerlIO_printf( Perl_debug_log,
2605 "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ",
2606 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
2607 (UV)state, (UV)base, (UV)accepted );
2612 if ( do_utf8 || UTF ) {
2614 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags );
2619 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2620 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags );
2621 uvc = to_uni_fold( uvc, foldbuf, &foldlen );
2622 foldlen -= UNISKIP( uvc );
2623 uscan = foldbuf + UNISKIP( uvc );
2635 DEBUG_TRIE_EXECUTE_r(
2636 PerlIO_printf( Perl_debug_log,
2637 "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n",
2638 charid, uvc, (UV)state, PL_colors[5] );
2647 /* unreached codepoint: we jump into the middle of the next case
2648 from previous if blocks */
2651 const U32 uniflags = ckWARN( WARN_UTF8 ) ? 0 : UTF8_ALLOW_ANY;
2652 U8 *uc = (U8*)locinput;
2661 trie = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ];
2663 while ( state && uc <= (U8*)PL_regeol ) {
2665 TRIE_CHECK_STATE_IS_ACCEPTING;
2667 base = trie->states[ state ].trans.base;
2669 DEBUG_TRIE_EXECUTE_r(
2670 PerlIO_printf( Perl_debug_log,
2671 "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ",
2672 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
2673 (UV)state, (UV)base, (UV)accepted );
2678 if ( do_utf8 || UTF ) {
2679 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags );
2690 DEBUG_TRIE_EXECUTE_r(
2691 PerlIO_printf( Perl_debug_log,
2692 "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n",
2693 charid, uvc, (UV)state, PL_colors[5] );
2703 There was at least one accepting state that we
2704 transitioned through. Presumably the number of accepting
2705 states is going to be low, typically one or two. So we
2706 simply scan through to find the one with lowest wordnum.
2707 Once we find it, we swap the last state into its place
2708 and decrement the size. We then try to match the rest of
2709 the pattern at the point where the word ends, if we
2710 succeed then we end the loop, otherwise the loop
2711 eventually terminates once all of the accepting states
2718 if ( accepted == 1 ) {
2720 SV **tmp = av_fetch( trie->words, accept_buff[ 0 ].wordnum-1, 0 );
2721 PerlIO_printf( Perl_debug_log,
2722 "%*s %sonly one match : #%d <%s>%s\n",
2723 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],
2724 accept_buff[ 0 ].wordnum,
2725 tmp ? SvPV_nolen( *tmp ) : "not compiled under -Dr",
2728 PL_reginput = (char *)accept_buff[ 0 ].endpos;
2729 /* in this case we free tmps/leave before we call regmatch
2730 as we wont be using accept_buff again. */
2733 gotit = regmatch( scan + NEXT_OFF( scan ) );
2736 PerlIO_printf( Perl_debug_log,"%*s %sgot %"IVdf" possible matches%s\n",
2737 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], (IV)accepted,
2740 while ( !gotit && accepted-- ) {
2743 for( cur = 1 ; cur <= accepted ; cur++ ) {
2744 DEBUG_TRIE_EXECUTE_r(
2745 PerlIO_printf( Perl_debug_log,
2746 "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n",
2747 REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
2748 (IV)best, accept_buff[ best ].wordnum, (IV)cur,
2749 accept_buff[ cur ].wordnum, PL_colors[5] );
2752 if ( accept_buff[ cur ].wordnum < accept_buff[ best ].wordnum )
2756 SV **tmp = av_fetch( trie->words, accept_buff[ best ].wordnum - 1, 0 );
2757 PerlIO_printf( Perl_debug_log, "%*s %strying alternation #%d <%s> at 0x%p%s\n",
2758 REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],
2759 accept_buff[best].wordnum,
2760 tmp ? SvPV_nolen( *tmp ) : "not compiled under -Dr",scan,
2763 if ( best<accepted ) {
2764 reg_trie_accepted tmp = accept_buff[ best ];
2765 accept_buff[ best ] = accept_buff[ accepted ];
2766 accept_buff[ accepted ] = tmp;
2769 PL_reginput = (char *)accept_buff[ best ].endpos;
2772 as far as I can tell we only need the SAVETMPS/FREETMPS
2773 for re's with EVAL in them but I'm leaving them in for
2774 all until I can be sure.
2777 gotit = regmatch( scan + NEXT_OFF( scan ) ) ;
2790 /* unreached codepoint */
2794 if (do_utf8 != UTF) {
2795 /* The target and the pattern have differing utf8ness. */
2797 const char *e = s + ln;
2800 /* The target is utf8, the pattern is not utf8. */
2805 if (NATIVE_TO_UNI(*(U8*)s) !=
2806 utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
2808 0 : UTF8_ALLOW_ANY))
2815 /* The target is not utf8, the pattern is utf8. */
2820 if (NATIVE_TO_UNI(*((U8*)l)) !=
2821 utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
2823 0 : UTF8_ALLOW_ANY))
2830 nextchr = UCHARAT(locinput);
2833 /* The target and the pattern have the same utf8ness. */
2834 /* Inline the first character, for speed. */
2835 if (UCHARAT(s) != nextchr)
2837 if (PL_regeol - locinput < ln)
2839 if (ln > 1 && memNE(s, locinput, ln))
2842 nextchr = UCHARAT(locinput);
2845 PL_reg_flags |= RF_tainted;
2851 if (do_utf8 || UTF) {
2852 /* Either target or the pattern are utf8. */
2854 char *e = PL_regeol;
2856 if (ibcmp_utf8(s, 0, ln, (bool)UTF,
2857 l, &e, 0, do_utf8)) {
2858 /* One more case for the sharp s:
2859 * pack("U0U*", 0xDF) =~ /ss/i,
2860 * the 0xC3 0x9F are the UTF-8
2861 * byte sequence for the U+00DF. */
2863 toLOWER(s[0]) == 's' &&
2865 toLOWER(s[1]) == 's' &&
2872 nextchr = UCHARAT(locinput);
2876 /* Neither the target and the pattern are utf8. */
2878 /* Inline the first character, for speed. */
2879 if (UCHARAT(s) != nextchr &&
2880 UCHARAT(s) != ((OP(scan) == EXACTF)
2881 ? PL_fold : PL_fold_locale)[nextchr])
2883 if (PL_regeol - locinput < ln)
2885 if (ln > 1 && (OP(scan) == EXACTF
2886 ? ibcmp(s, locinput, ln)
2887 : ibcmp_locale(s, locinput, ln)))
2890 nextchr = UCHARAT(locinput);
2894 STRLEN inclasslen = PL_regeol - locinput;
2896 if (!reginclass(scan, (U8*)locinput, &inclasslen, do_utf8))
2898 if (locinput >= PL_regeol)
2900 locinput += inclasslen ? inclasslen : UTF8SKIP(locinput);
2901 nextchr = UCHARAT(locinput);
2906 nextchr = UCHARAT(locinput);
2907 if (!REGINCLASS(scan, (U8*)locinput))
2909 if (!nextchr && locinput >= PL_regeol)
2911 nextchr = UCHARAT(++locinput);
2915 /* If we might have the case of the German sharp s
2916 * in a casefolding Unicode character class. */
2918 if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
2919 locinput += SHARP_S_SKIP;
2920 nextchr = UCHARAT(locinput);
2926 PL_reg_flags |= RF_tainted;
2932 LOAD_UTF8_CHARCLASS(alnum,"a");
2933 if (!(OP(scan) == ALNUM
2934 ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
2935 : isALNUM_LC_utf8((U8*)locinput)))
2939 locinput += PL_utf8skip[nextchr];
2940 nextchr = UCHARAT(locinput);
2943 if (!(OP(scan) == ALNUM
2944 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
2946 nextchr = UCHARAT(++locinput);
2949 PL_reg_flags |= RF_tainted;
2952 if (!nextchr && locinput >= PL_regeol)
2955 LOAD_UTF8_CHARCLASS(alnum,"a");
2956 if (OP(scan) == NALNUM
2957 ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
2958 : isALNUM_LC_utf8((U8*)locinput))
2962 locinput += PL_utf8skip[nextchr];
2963 nextchr = UCHARAT(locinput);
2966 if (OP(scan) == NALNUM
2967 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
2969 nextchr = UCHARAT(++locinput);
2973 PL_reg_flags |= RF_tainted;
2977 /* was last char in word? */
2979 if (locinput == PL_bostr)
2982 const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
2984 ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
2986 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
2987 ln = isALNUM_uni(ln);
2988 LOAD_UTF8_CHARCLASS(alnum,"a");
2989 n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8);
2992 ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln));
2993 n = isALNUM_LC_utf8((U8*)locinput);
2997 ln = (locinput != PL_bostr) ?
2998 UCHARAT(locinput - 1) : '\n';
2999 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
3001 n = isALNUM(nextchr);
3004 ln = isALNUM_LC(ln);
3005 n = isALNUM_LC(nextchr);
3008 if (((!ln) == (!n)) == (OP(scan) == BOUND ||
3009 OP(scan) == BOUNDL))
3013 PL_reg_flags |= RF_tainted;
3019 if (UTF8_IS_CONTINUED(nextchr)) {
3020 LOAD_UTF8_CHARCLASS(space," ");
3021 if (!(OP(scan) == SPACE
3022 ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
3023 : isSPACE_LC_utf8((U8*)locinput)))
3027 locinput += PL_utf8skip[nextchr];
3028 nextchr = UCHARAT(locinput);
3031 if (!(OP(scan) == SPACE
3032 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
3034 nextchr = UCHARAT(++locinput);
3037 if (!(OP(scan) == SPACE
3038 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
3040 nextchr = UCHARAT(++locinput);
3044 PL_reg_flags |= RF_tainted;
3047 if (!nextchr && locinput >= PL_regeol)
3050 LOAD_UTF8_CHARCLASS(space," ");
3051 if (OP(scan) == NSPACE
3052 ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
3053 : isSPACE_LC_utf8((U8*)locinput))
3057 locinput += PL_utf8skip[nextchr];
3058 nextchr = UCHARAT(locinput);
3061 if (OP(scan) == NSPACE
3062 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
3064 nextchr = UCHARAT(++locinput);
3067 PL_reg_flags |= RF_tainted;
3073 LOAD_UTF8_CHARCLASS(digit,"0");
3074 if (!(OP(scan) == DIGIT
3075 ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
3076 : isDIGIT_LC_utf8((U8*)locinput)))
3080 locinput += PL_utf8skip[nextchr];
3081 nextchr = UCHARAT(locinput);
3084 if (!(OP(scan) == DIGIT
3085 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
3087 nextchr = UCHARAT(++locinput);
3090 PL_reg_flags |= RF_tainted;
3093 if (!nextchr && locinput >= PL_regeol)
3096 LOAD_UTF8_CHARCLASS(digit,"0");
3097 if (OP(scan) == NDIGIT
3098 ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
3099 : isDIGIT_LC_utf8((U8*)locinput))
3103 locinput += PL_utf8skip[nextchr];
3104 nextchr = UCHARAT(locinput);
3107 if (OP(scan) == NDIGIT
3108 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
3110 nextchr = UCHARAT(++locinput);
3113 if (locinput >= PL_regeol)
3116 LOAD_UTF8_CHARCLASS(mark,"~");
3117 if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
3119 locinput += PL_utf8skip[nextchr];
3120 while (locinput < PL_regeol &&
3121 swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
3122 locinput += UTF8SKIP(locinput);
3123 if (locinput > PL_regeol)
3128 nextchr = UCHARAT(locinput);
3131 PL_reg_flags |= RF_tainted;
3135 n = ARG(scan); /* which paren pair */
3136 ln = PL_regstartp[n];
3137 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
3138 if ((I32)*PL_reglastparen < n || ln == -1)
3139 sayNO; /* Do not match unless seen CLOSEn. */
3140 if (ln == PL_regendp[n])
3144 if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */
3146 const char *e = PL_bostr + PL_regendp[n];
3148 * Note that we can't do the "other character" lookup trick as
3149 * in the 8-bit case (no pun intended) because in Unicode we
3150 * have to map both upper and title case to lower case.
3152 if (OP(scan) == REFF) {
3154 STRLEN ulen1, ulen2;
3155 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
3156 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
3160 toLOWER_utf8((U8*)s, tmpbuf1, &ulen1);
3161 toLOWER_utf8((U8*)l, tmpbuf2, &ulen2);
3162 if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
3169 nextchr = UCHARAT(locinput);
3173 /* Inline the first character, for speed. */
3174 if (UCHARAT(s) != nextchr &&
3176 (UCHARAT(s) != ((OP(scan) == REFF
3177 ? PL_fold : PL_fold_locale)[nextchr]))))
3179 ln = PL_regendp[n] - ln;
3180 if (locinput + ln > PL_regeol)
3182 if (ln > 1 && (OP(scan) == REF
3183 ? memNE(s, locinput, ln)
3185 ? ibcmp(s, locinput, ln)
3186 : ibcmp_locale(s, locinput, ln))))
3189 nextchr = UCHARAT(locinput);
3200 OP_4tree *oop = PL_op;
3201 COP *ocurcop = PL_curcop;
3204 struct regexp *oreg = PL_reg_re;
3207 PL_op = (OP_4tree*)PL_regdata->data[n];
3208 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
3209 PAD_SAVE_LOCAL(old_comppad, (PAD*)PL_regdata->data[n + 2]);
3210 PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
3214 CALLRUNOPS(aTHX); /* Scalar context. */
3217 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
3225 PAD_RESTORE_LOCAL(old_comppad);
3226 PL_curcop = ocurcop;
3228 if (logical == 2) { /* Postponed subexpression. */
3230 MAGIC *mg = Null(MAGIC*);
3232 CHECKPOINT cp, lastcp;
3236 if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret)))
3237 mg = mg_find(sv, PERL_MAGIC_qr);
3238 else if (SvSMAGICAL(ret)) {
3239 if (SvGMAGICAL(ret))
3240 sv_unmagic(ret, PERL_MAGIC_qr);
3242 mg = mg_find(ret, PERL_MAGIC_qr);
3246 re = (regexp *)mg->mg_obj;
3247 (void)ReREFCNT_inc(re);
3251 char *t = SvPV(ret, len);
3253 char * const oprecomp = PL_regprecomp;
3254 const I32 osize = PL_regsize;
3255 const I32 onpar = PL_regnpar;
3258 if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8;
3259 re = CALLREGCOMP(aTHX_ t, t + len, &pm);
3261 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
3263 sv_magic(ret,(SV*)ReREFCNT_inc(re),
3265 PL_regprecomp = oprecomp;
3270 PerlIO_printf(Perl_debug_log,
3271 "Entering embedded \"%s%.60s%s%s\"\n",
3275 (strlen(re->precomp) > 60 ? "..." : ""))
3278 state.prev = PL_reg_call_cc;
3279 state.cc = PL_regcc;
3280 state.re = PL_reg_re;
3284 cp = regcppush(0); /* Save *all* the positions. */
3287 state.ss = PL_savestack_ix;
3288 *PL_reglastparen = 0;
3289 *PL_reglastcloseparen = 0;
3290 PL_reg_call_cc = &state;
3291 PL_reginput = locinput;
3292 toggleutf = ((PL_reg_flags & RF_utf8) != 0) ^
3293 ((re->reganch & ROPT_UTF8) != 0);
3294 if (toggleutf) PL_reg_flags ^= RF_utf8;
3296 /* XXXX This is too dramatic a measure... */
3299 if (regmatch(re->program + 1)) {
3300 /* Even though we succeeded, we need to restore
3301 global variables, since we may be wrapped inside
3302 SUSPEND, thus the match may be not finished yet. */
3304 /* XXXX Do this only if SUSPENDed? */
3305 PL_reg_call_cc = state.prev;
3306 PL_regcc = state.cc;
3307 PL_reg_re = state.re;
3308 cache_re(PL_reg_re);
3309 if (toggleutf) PL_reg_flags ^= RF_utf8;
3311 /* XXXX This is too dramatic a measure... */
3314 /* These are needed even if not SUSPEND. */
3320 REGCP_UNWIND(lastcp);
3322 PL_reg_call_cc = state.prev;
3323 PL_regcc = state.cc;
3324 PL_reg_re = state.re;
3325 cache_re(PL_reg_re);
3326 if (toggleutf) PL_reg_flags ^= RF_utf8;
3328 /* XXXX This is too dramatic a measure... */
3338 sv_setsv(save_scalar(PL_replgv), ret);
3344 n = ARG(scan); /* which paren pair */
3345 PL_reg_start_tmp[n] = locinput;
3350 n = ARG(scan); /* which paren pair */
3351 PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr;
3352 PL_regendp[n] = locinput - PL_bostr;
3353 if (n > (I32)*PL_reglastparen)
3354 *PL_reglastparen = n;
3355 *PL_reglastcloseparen = n;
3358 n = ARG(scan); /* which paren pair */
3359 sw = ((I32)*PL_reglastparen >= n && PL_regendp[n] != -1);
3362 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
3364 next = NEXTOPER(NEXTOPER(scan));
3366 next = scan + ARG(scan);
3367 if (OP(next) == IFTHEN) /* Fake one. */
3368 next = NEXTOPER(NEXTOPER(next));
3372 logical = scan->flags;
3374 /*******************************************************************
3375 PL_regcc contains infoblock about the innermost (...)* loop, and
3376 a pointer to the next outer infoblock.
3378 Here is how Y(A)*Z is processed (if it is compiled into CURLYX/WHILEM):
3380 1) After matching X, regnode for CURLYX is processed;
3382 2) This regnode creates infoblock on the stack, and calls
3383 regmatch() recursively with the starting point at WHILEM node;
3385 3) Each hit of WHILEM node tries to match A and Z (in the order
3386 depending on the current iteration, min/max of {min,max} and
3387 greediness). The information about where are nodes for "A"
3388 and "Z" is read from the infoblock, as is info on how many times "A"
3389 was already matched, and greediness.
3391 4) After A matches, the same WHILEM node is hit again.
3393 5) Each time WHILEM is hit, PL_regcc is the infoblock created by CURLYX
3394 of the same pair. Thus when WHILEM tries to match Z, it temporarily
3395 resets PL_regcc, since this Y(A)*Z can be a part of some other loop:
3396 as in (Y(A)*Z)*. If Z matches, the automaton will hit the WHILEM node
3397 of the external loop.
3399 Currently present infoblocks form a tree with a stem formed by PL_curcc
3400 and whatever it mentions via ->next, and additional attached trees
3401 corresponding to temporarily unset infoblocks as in "5" above.
3403 In the following picture infoblocks for outer loop of
3404 (Y(A)*?Z)*?T are denoted O, for inner I. NULL starting block
3405 is denoted by x. The matched string is YAAZYAZT. Temporarily postponed
3406 infoblocks are drawn below the "reset" infoblock.
3408 In fact in the picture below we do not show failed matches for Z and T
3409 by WHILEM blocks. [We illustrate minimal matches, since for them it is
3410 more obvious *why* one needs to *temporary* unset infoblocks.]
3412 Matched REx position InfoBlocks Comment
3416 Y A)*?Z)*?T x <- O <- I
3417 YA )*?Z)*?T x <- O <- I
3418 YA A)*?Z)*?T x <- O <- I
3419 YAA )*?Z)*?T x <- O <- I
3420 YAA Z)*?T x <- O # Temporary unset I
3423 YAAZ Y(A)*?Z)*?T x <- O
3426 YAAZY (A)*?Z)*?T x <- O
3429 YAAZY A)*?Z)*?T x <- O <- I
3432 YAAZYA )*?Z)*?T x <- O <- I
3435 YAAZYA Z)*?T x <- O # Temporary unset I
3441 YAAZYAZ T x # Temporary unset O
3448 *******************************************************************/
3451 CHECKPOINT cp = PL_savestack_ix;
3452 /* No need to save/restore up to this paren */
3453 I32 parenfloor = scan->flags;
3455 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
3457 cc.oldcc = PL_regcc;
3459 /* XXXX Probably it is better to teach regpush to support
3460 parenfloor > PL_regsize... */
3461 if (parenfloor > (I32)*PL_reglastparen)
3462 parenfloor = *PL_reglastparen; /* Pessimization... */
3463 cc.parenfloor = parenfloor;
3465 cc.min = ARG1(scan);
3466 cc.max = ARG2(scan);
3467 cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
3471 PL_reginput = locinput;
3472 n = regmatch(PREVOPER(next)); /* start on the WHILEM */
3474 PL_regcc = cc.oldcc;
3480 * This is really hard to understand, because after we match
3481 * what we're trying to match, we must make sure the rest of
3482 * the REx is going to match for sure, and to do that we have
3483 * to go back UP the parse tree by recursing ever deeper. And
3484 * if it fails, we have to reset our parent's current state
3485 * that we can try again after backing off.
3488 CHECKPOINT cp, lastcp;
3489 CURCUR* cc = PL_regcc;
3490 char *lastloc = cc->lastloc; /* Detection of 0-len. */
3491 I32 cache_offset = 0, cache_bit = 0;
3493 n = cc->cur + 1; /* how many we know we matched */
3494 PL_reginput = locinput;
3497 PerlIO_printf(Perl_debug_log,
3498 "%*s %ld out of %ld..%ld cc=%"UVxf"\n",
3499 REPORT_CODE_OFF+PL_regindent*2, "",
3500 (long)n, (long)cc->min,
3501 (long)cc->max, PTR2UV(cc))
3504 /* If degenerate scan matches "", assume scan done. */
3506 if (locinput == cc->lastloc && n >= cc->min) {
3507 PL_regcc = cc->oldcc;
3511 PerlIO_printf(Perl_debug_log,
3512 "%*s empty match detected, try continuation...\n",
3513 REPORT_CODE_OFF+PL_regindent*2, "")
3515 if (regmatch(cc->next))
3523 /* First just match a string of min scans. */
3527 cc->lastloc = locinput;
3528 if (regmatch(cc->scan))
3531 cc->lastloc = lastloc;
3536 /* Check whether we already were at this position.
3537 Postpone detection until we know the match is not
3538 *that* much linear. */
3539 if (!PL_reg_maxiter) {
3540 PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4);
3541 PL_reg_leftiter = PL_reg_maxiter;
3543 if (PL_reg_leftiter-- == 0) {
3544 const I32 size = (PL_reg_maxiter + 7 + POSCACHE_START)/8;
3545 if (PL_reg_poscache) {
3546 if ((I32)PL_reg_poscache_size < size) {
3547 Renew(PL_reg_poscache, size, char);
3548 PL_reg_poscache_size = size;
3550 Zero(PL_reg_poscache, size, char);
3553 PL_reg_poscache_size = size;
3554 Newz(29, PL_reg_poscache, size, char);
3557 PerlIO_printf(Perl_debug_log,
3558 "%sDetected a super-linear match, switching on caching%s...\n",
3559 PL_colors[4], PL_colors[5])
3562 if (PL_reg_leftiter < 0) {
3563 cache_offset = locinput - PL_bostr;
3565 cache_offset = (scan->flags & 0xf) - 1 + POSCACHE_START
3566 + cache_offset * (scan->flags>>4);
3567 cache_bit = cache_offset % 8;
3569 if (PL_reg_poscache[cache_offset] & (1<<cache_bit)) {
3571 PerlIO_printf(Perl_debug_log,
3572 "%*s already tried at this position...\n",
3573 REPORT_CODE_OFF+PL_regindent*2, "")
3575 if (PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))
3576 /* cache records success */
3579 /* cache records failure */
3582 PL_reg_poscache[cache_offset] |= (1<<cache_bit);
3586 /* Prefer next over scan for minimal matching. */
3589 PL_regcc = cc->oldcc;
3592 cp = regcppush(cc->parenfloor);
3594 if (regmatch(cc->next)) {
3596 CACHEsayYES; /* All done. */
3598 REGCP_UNWIND(lastcp);
3604 if (n >= cc->max) { /* Maximum greed exceeded? */
3605 if (ckWARN(WARN_REGEXP) && n >= REG_INFTY
3606 && !(PL_reg_flags & RF_warned)) {
3607 PL_reg_flags |= RF_warned;
3608 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded",
3609 "Complex regular subexpression recursion",
3616 PerlIO_printf(Perl_debug_log,
3617 "%*s trying longer...\n",
3618 REPORT_CODE_OFF+PL_regindent*2, "")
3620 /* Try scanning more and see if it helps. */
3621 PL_reginput = locinput;
3623 cc->lastloc = locinput;
3624 cp = regcppush(cc->parenfloor);
3626 if (regmatch(cc->scan)) {
3630 REGCP_UNWIND(lastcp);
3633 cc->lastloc = lastloc;
3637 /* Prefer scan over next for maximal matching. */
3639 if (n < cc->max) { /* More greed allowed? */
3640 cp = regcppush(cc->parenfloor);
3642 cc->lastloc = locinput;
3644 if (regmatch(cc->scan)) {
3648 REGCP_UNWIND(lastcp);
3649 regcppop(); /* Restore some previous $<digit>s? */
3650 PL_reginput = locinput;
3652 PerlIO_printf(Perl_debug_log,
3653 "%*s failed, try continuation...\n",
3654 REPORT_CODE_OFF+PL_regindent*2, "")
3657 if (ckWARN(WARN_REGEXP) && n >= REG_INFTY
3658 && !(PL_reg_flags & RF_warned)) {
3659 PL_reg_flags |= RF_warned;
3660 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded",
3661 "Complex regular subexpression recursion",
3665 /* Failed deeper matches of scan, so see if this one works. */
3666 PL_regcc = cc->oldcc;
3669 if (regmatch(cc->next))
3675 cc->lastloc = lastloc;
3680 next = scan + ARG(scan);
3683 inner = NEXTOPER(NEXTOPER(scan));
3686 inner = NEXTOPER(scan);
3690 if (OP(next) != c1) /* No choice. */
3691 next = inner; /* Avoid recursion. */
3693 const I32 lastparen = *PL_reglastparen;
3695 re_unwind_branch_t *uw;
3697 /* Put unwinding data on stack */
3698 unwind1 = SSNEWt(1,re_unwind_branch_t);
3699 uw = SSPTRt(unwind1,re_unwind_branch_t);
3702 uw->type = ((c1 == BRANCH)
3704 : RE_UNWIND_BRANCHJ);
3705 uw->lastparen = lastparen;
3707 uw->locinput = locinput;
3708 uw->nextchr = nextchr;
3710 uw->regindent = ++PL_regindent;
3713 REGCP_SET(uw->lastcp);
3715 /* Now go into the first branch */
3728 /* We suppose that the next guy does not need
3729 backtracking: in particular, it is of constant non-zero length,
3730 and has no parenths to influence future backrefs. */
3731 ln = ARG1(scan); /* min to match */
3732 n = ARG2(scan); /* max to match */
3733 paren = scan->flags;
3735 if (paren > PL_regsize)
3737 if (paren > (I32)*PL_reglastparen)
3738 *PL_reglastparen = paren;
3740 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
3742 scan += NEXT_OFF(scan); /* Skip former OPEN. */
3743 PL_reginput = locinput;
3746 if (ln && regrepeat_hard(scan, ln, &l) < ln)
3748 locinput = PL_reginput;
3749 if (HAS_TEXT(next) || JUMPABLE(next)) {
3750 regnode *text_node = next;
3752 if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
3754 if (! HAS_TEXT(text_node)) c1 = c2 = -1000;
3756 if (PL_regkind[(U8)OP(text_node)] == REF) {
3760 else { c1 = (U8)*STRING(text_node); }
3761 if (OP(text_node) == EXACTF || OP(text_node) == REFF)
3763 else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
3764 c2 = PL_fold_locale[c1];
3773 while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
3774 /* If it could work, try it. */
3776 UCHARAT(PL_reginput) == c1 ||
3777 UCHARAT(PL_reginput) == c2)
3781 PL_regstartp[paren] =
3782 HOPc(PL_reginput, -l) - PL_bostr;
3783 PL_regendp[paren] = PL_reginput - PL_bostr;
3786 PL_regendp[paren] = -1;
3790 REGCP_UNWIND(lastcp);
3792 /* Couldn't or didn't -- move forward. */
3793 PL_reginput = locinput;
3794 if (regrepeat_hard(scan, 1, &l)) {
3796 locinput = PL_reginput;
3803 n = regrepeat_hard(scan, n, &l);
3804 locinput = PL_reginput;
3806 PerlIO_printf(Perl_debug_log,
3807 "%*s matched %"IVdf" times, len=%"IVdf"...\n",
3808 (int)(REPORT_CODE_OFF+PL_regindent*2), "",
3812 if (HAS_TEXT(next) || JUMPABLE(next)) {
3813 regnode *text_node = next;
3815 if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
3817 if (! HAS_TEXT(text_node)) c1 = c2 = -1000;
3819 if (PL_regkind[(U8)OP(text_node)] == REF) {
3823 else { c1 = (U8)*STRING(text_node); }
3825 if (OP(text_node) == EXACTF || OP(text_node) == REFF)
3827 else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
3828 c2 = PL_fold_locale[c1];
3839 /* If it could work, try it. */
3841 UCHARAT(PL_reginput) == c1 ||
3842 UCHARAT(PL_reginput) == c2)
3845 PerlIO_printf(Perl_debug_log,
3846 "%*s trying tail with n=%"IVdf"...\n",
3847 (int)(REPORT_CODE_OFF+PL_regindent*2), "", (IV)n)
3851 PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr;
3852 PL_regendp[paren] = PL_reginput - PL_bostr;
3855 PL_regendp[paren] = -1;