5 * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee
8 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
9 * confused with the original package (see point 3 below). Thanks, Henry!
12 /* Additional note: this code is very heavily munged from Henry's version
13 * in places. In some spots I've traded clarity for efficiency, so don't
14 * blame Henry for some of the lack of readability.
17 /* The names of the functions have been changed from regcomp and
18 * regexec to pregcomp and pregexec in order to avoid conflicts
19 * with the POSIX routines of the same names.
22 #ifdef PERL_EXT_RE_BUILD
23 /* need to replace pregcomp et al, so enable that */
24 # ifndef PERL_IN_XSUB_RE
25 # define PERL_IN_XSUB_RE
27 /* need access to debugger hooks */
28 # if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
33 #ifdef PERL_IN_XSUB_RE
34 /* We *really* need to overwrite these symbols: */
35 # define Perl_pregcomp my_regcomp
36 # define Perl_regdump my_regdump
37 # define Perl_regprop my_regprop
38 # define Perl_pregfree my_regfree
39 # define Perl_re_intuit_string my_re_intuit_string
40 /* *These* symbols are masked to allow static link. */
41 # define Perl_regnext my_regnext
42 # define Perl_save_re_context my_save_re_context
43 # define Perl_reginitcolors my_reginitcolors
45 # define PERL_NO_GET_CONTEXT
50 * pregcomp and pregexec -- regsub and regerror are not used in perl
52 * Copyright (c) 1986 by University of Toronto.
53 * Written by Henry Spencer. Not derived from licensed software.
55 * Permission is granted to anyone to use this software for any
56 * purpose on any computer system, and to redistribute it freely,
57 * subject to the following restrictions:
59 * 1. The author is not responsible for the consequences of use of
60 * this software, no matter how awful, even if they arise
63 * 2. The origin of this software must not be misrepresented, either
64 * by explicit claim or by omission.
66 * 3. Altered versions must be plainly marked as such, and must not
67 * be misrepresented as being the original software.
70 **** Alterations to Henry's code are...
72 **** Copyright (c) 1991-2001, Larry Wall
74 **** You may distribute under the terms of either the GNU General Public
75 **** License or the Artistic License, as specified in the README file.
78 * Beware that some of this code is subtly aware of the way operator
79 * precedence is structured in regular expressions. Serious changes in
80 * regular-expression syntax might require a total rethink.
83 #define PERL_IN_REGCOMP_C
86 #ifdef PERL_IN_XSUB_RE
87 # if defined(PERL_CAPI) || defined(PERL_OBJECT)
102 # if defined(BUGGY_MSC6)
103 /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
104 # pragma optimize("a",off)
105 /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
106 # pragma optimize("w",on )
107 # endif /* BUGGY_MSC6 */
111 #define STATIC static
114 typedef struct RExC_state_t {
115 U16 flags16; /* are we folding, multilining? */
116 char *precomp; /* uncompiled string. */
118 char *start; /* Start of input for compile */
119 char *end; /* End of input for compile */
120 char *parse; /* Input-scan pointer. */
121 I32 whilem_seen; /* number of WHILEM in this expr */
122 regnode *emit_start; /* Start of emitted-code area */
123 regnode *emit; /* Code-emit pointer; ®dummy = don't = compiling */
124 I32 naughty; /* How bad is this pattern? */
125 I32 sawback; /* Did we see \1, ...? */
127 I32 size; /* Code size. */
128 I32 npar; /* () count. */
134 char *starttry; /* -Dr: where regtry was called. */
135 #define RExC_starttry (pRExC_state->starttry)
139 #define RExC_flags16 (pRExC_state->flags16)
140 #define RExC_precomp (pRExC_state->precomp)
141 #define RExC_rx (pRExC_state->rx)
142 #define RExC_start (pRExC_state->start)
143 #define RExC_end (pRExC_state->end)
144 #define RExC_parse (pRExC_state->parse)
145 #define RExC_whilem_seen (pRExC_state->whilem_seen)
146 #define RExC_offsets (pRExC_state->rx->offsets) /* I am not like the others */
147 #define RExC_emit (pRExC_state->emit)
148 #define RExC_emit_start (pRExC_state->emit_start)
149 #define RExC_naughty (pRExC_state->naughty)
150 #define RExC_sawback (pRExC_state->sawback)
151 #define RExC_seen (pRExC_state->seen)
152 #define RExC_size (pRExC_state->size)
153 #define RExC_npar (pRExC_state->npar)
154 #define RExC_extralen (pRExC_state->extralen)
155 #define RExC_seen_zerolen (pRExC_state->seen_zerolen)
156 #define RExC_seen_evals (pRExC_state->seen_evals)
157 #define RExC_utf8 (pRExC_state->utf8)
159 #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
160 #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
161 ((*s) == '{' && regcurly(s)))
164 #undef SPSTART /* dratted cpp namespace... */
167 * Flags to be passed up and down.
169 #define WORST 0 /* Worst case. */
170 #define HASWIDTH 0x1 /* Known to match non-null strings. */
171 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
172 #define SPSTART 0x4 /* Starts with * or +. */
173 #define TRYAGAIN 0x8 /* Weeded out a declaration. */
175 /* Length of a variant. */
177 typedef struct scan_data_t {
183 I32 last_end; /* min value, <0 unless valid. */
186 SV **longest; /* Either &l_fixed, or &l_float. */
190 I32 offset_float_min;
191 I32 offset_float_max;
195 struct regnode_charclass_class *start_class;
199 * Forward declarations for pregcomp()'s friends.
202 static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
205 #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
206 #define SF_BEFORE_SEOL 0x1
207 #define SF_BEFORE_MEOL 0x2
208 #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
209 #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
212 # define SF_FIX_SHIFT_EOL (0+2)
213 # define SF_FL_SHIFT_EOL (0+4)
215 # define SF_FIX_SHIFT_EOL (+2)
216 # define SF_FL_SHIFT_EOL (+4)
219 #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
220 #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
222 #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
223 #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
224 #define SF_IS_INF 0x40
225 #define SF_HAS_PAR 0x80
226 #define SF_IN_PAR 0x100
227 #define SF_HAS_EVAL 0x200
228 #define SCF_DO_SUBSTR 0x400
229 #define SCF_DO_STCLASS_AND 0x0800
230 #define SCF_DO_STCLASS_OR 0x1000
231 #define SCF_DO_STCLASS (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
232 #define SCF_WHILEM_VISITED_POS 0x2000
234 #define UTF RExC_utf8
235 #define LOC (RExC_flags16 & PMf_LOCALE)
236 #define FOLD (RExC_flags16 & PMf_FOLD)
238 #define OOB_UNICODE 12345678
239 #define OOB_NAMEDCLASS -1
241 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
242 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
245 /* length of regex to show in messages that don't mark a position within */
246 #define RegexLengthToShowInErrorMessages 127
249 * If MARKER[12] are adjusted, be sure to adjust the constants at the top
250 * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
251 * op/pragma/warn/regcomp.
253 #define MARKER1 "HERE" /* marker as it appears in the description */
254 #define MARKER2 " << HERE " /* marker as it appears within the regex */
256 #define REPORT_LOCATION " before " MARKER1 " mark in regex m/%.*s" MARKER2 "%s/"
259 * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
260 * arg. Show regex, up to a maximum length. If it's too long, chop and add
265 char *ellipses = ""; \
266 unsigned len = strlen(RExC_precomp); \
269 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
271 if (len > RegexLengthToShowInErrorMessages) { \
272 /* chop 10 shorter than the max, to ensure meaning of "..." */ \
273 len = RegexLengthToShowInErrorMessages - 10; \
276 Perl_croak(aTHX_ "%s in regex m/%.*s%s/", \
277 msg, (int)len, RExC_precomp, ellipses); \
281 * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
282 * args. Show regex, up to a maximum length. If it's too long, chop and add
285 #define FAIL2(pat,msg) \
287 char *ellipses = ""; \
288 unsigned len = strlen(RExC_precomp); \
291 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
293 if (len > RegexLengthToShowInErrorMessages) { \
294 /* chop 10 shorter than the max, to ensure meaning of "..." */ \
295 len = RegexLengthToShowInErrorMessages - 10; \
298 S_re_croak2(aTHX_ pat, " in regex m/%.*s%s/", \
299 msg, (int)len, RExC_precomp, ellipses); \
304 * Simple_vFAIL -- like FAIL, but marks the current location in the scan
306 #define Simple_vFAIL(m) \
308 unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
310 Perl_croak(aTHX_ "%s" REPORT_LOCATION, \
311 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
315 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
320 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
325 * Like Simple_vFAIL(), but accepts two arguments.
327 #define Simple_vFAIL2(m,a1) \
329 unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
331 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, \
332 (int)offset, RExC_precomp, RExC_precomp + offset); \
336 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
338 #define vFAIL2(m,a1) \
341 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
342 Simple_vFAIL2(m, a1); \
347 * Like Simple_vFAIL(), but accepts three arguments.
349 #define Simple_vFAIL3(m, a1, a2) \
351 unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
353 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, \
354 (int)offset, RExC_precomp, RExC_precomp + offset); \
358 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
360 #define vFAIL3(m,a1,a2) \
363 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
364 Simple_vFAIL3(m, a1, a2); \
368 * Like Simple_vFAIL(), but accepts four arguments.
370 #define Simple_vFAIL4(m, a1, a2, a3) \
372 unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
374 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3,\
375 (int)offset, RExC_precomp, RExC_precomp + offset); \
379 * Like Simple_vFAIL(), but accepts five arguments.
381 #define Simple_vFAIL5(m, a1, a2, a3, a4) \
383 unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
384 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, a4,\
385 (int)offset, RExC_precomp, RExC_precomp + offset); \
389 #define vWARN(loc,m) \
391 unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc)); \
392 Perl_warner(aTHX_ WARN_REGEXP, "%s" REPORT_LOCATION,\
393 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
397 #define vWARN2(loc, m, a1) \
399 unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc)); \
400 Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
402 (int)offset, RExC_precomp, RExC_precomp + offset); \
405 #define vWARN3(loc, m, a1, a2) \
407 unsigned offset = strlen(RExC_precomp) - (RExC_end - (loc)); \
408 Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION, \
410 (int)offset, RExC_precomp, RExC_precomp + offset); \
413 #define vWARN4(loc, m, a1, a2, a3) \
415 unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc)); \
416 Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
418 (int)offset, RExC_precomp, RExC_precomp + offset); \
422 /* Allow for side effects in s */
423 #define REGC(c,s) STMT_START { if (!SIZE_ONLY) *(s) = (c); else (s);} STMT_END
425 /* Macros for recording node offsets. 20001227 mjd@plover.com
426 * Nodes are numbered 1, 2, 3, 4. Node #n's position is recorded in
427 * element 2*n-1 of the array. Element #2n holds the byte length node #n.
428 * Element 0 holds the number n.
431 #define MJD_OFFSET_DEBUG(x)
432 /* #define MJD_OFFSET_DEBUG(x) fprintf x */
435 # define Set_Node_Offset_To_R(node,byte) \
439 Perl_croak(aTHX_ "value of node is %d in Offset macro", node); \
441 RExC_offsets[2*(node)-1] = (byte); \
446 # define Set_Node_Offset(node,byte) Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
447 # define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
449 # define Set_Node_Length_To_R(node,len) \
452 MJD_OFFSET_DEBUG((stderr, "** (%d) size of node %d is %d.\n", __LINE__, (node), (len))); \
454 Perl_croak(aTHX_ "value of node is %d in Length macro", node); \
456 RExC_offsets[2*(node)] = (len); \
461 # define Set_Node_Length(node,len) Set_Node_Length_To_R((node)-RExC_emit_start, len)
462 # define Set_Cur_Node_Length(len) Set_Node_Length(RExC_emit, len)
463 # define Set_Node_Cur_Length(node) Set_Node_Length(node, RExC_parse - parse_start)
465 /* Get offsets and lengths */
466 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
467 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
469 static void clear_re(pTHXo_ void *r);
471 /* Mark that we cannot extend a found fixed substring at this point.
472 Updata the longest found anchored substring and the longest found
473 floating substrings if needed. */
476 S_scan_commit(pTHX_ RExC_state_t *pRExC_state, scan_data_t *data)
478 STRLEN l = CHR_SVLEN(data->last_found);
479 STRLEN old_l = CHR_SVLEN(*data->longest);
481 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
482 sv_setsv(*data->longest, data->last_found);
483 if (*data->longest == data->longest_fixed) {
484 data->offset_fixed = l ? data->last_start_min : data->pos_min;
485 if (data->flags & SF_BEFORE_EOL)
487 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
489 data->flags &= ~SF_FIX_BEFORE_EOL;
492 data->offset_float_min = l ? data->last_start_min : data->pos_min;
493 data->offset_float_max = (l
494 ? data->last_start_max
495 : data->pos_min + data->pos_delta);
496 if (data->flags & SF_BEFORE_EOL)
498 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
500 data->flags &= ~SF_FL_BEFORE_EOL;
503 SvCUR_set(data->last_found, 0);
505 data->flags &= ~SF_BEFORE_EOL;
508 /* Can match anything (initialization) */
510 S_cl_anything(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
514 ANYOF_CLASS_ZERO(cl);
515 for (value = 0; value < 256; ++value)
516 ANYOF_BITMAP_SET(cl, value);
517 cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
519 cl->flags |= ANYOF_LOCALE;
522 /* Can match anything (initialization) */
524 S_cl_is_anything(pTHX_ struct regnode_charclass_class *cl)
528 for (value = 0; value <= ANYOF_MAX; value += 2)
529 if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
531 if (!(cl->flags & ANYOF_UNICODE_ALL))
533 for (value = 0; value < 256; ++value)
534 if (!ANYOF_BITMAP_TEST(cl, value))
539 /* Can match anything (initialization) */
541 S_cl_init(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
543 Zero(cl, 1, struct regnode_charclass_class);
545 cl_anything(pRExC_state, cl);
549 S_cl_init_zero(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
551 Zero(cl, 1, struct regnode_charclass_class);
553 cl_anything(pRExC_state, cl);
555 cl->flags |= ANYOF_LOCALE;
558 /* 'And' a given class with another one. Can create false positives */
559 /* We assume that cl is not inverted */
561 S_cl_and(pTHX_ struct regnode_charclass_class *cl,
562 struct regnode_charclass_class *and_with)
564 if (!(and_with->flags & ANYOF_CLASS)
565 && !(cl->flags & ANYOF_CLASS)
566 && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
567 && !(and_with->flags & ANYOF_FOLD)
568 && !(cl->flags & ANYOF_FOLD)) {
571 if (and_with->flags & ANYOF_INVERT)
572 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
573 cl->bitmap[i] &= ~and_with->bitmap[i];
575 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
576 cl->bitmap[i] &= and_with->bitmap[i];
577 } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
578 if (!(and_with->flags & ANYOF_EOS))
579 cl->flags &= ~ANYOF_EOS;
581 if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_UNICODE) {
582 cl->flags &= ~ANYOF_UNICODE_ALL;
583 cl->flags |= ANYOF_UNICODE;
584 ARG_SET(cl, ARG(and_with));
586 if (!(and_with->flags & ANYOF_UNICODE_ALL))
587 cl->flags &= ~ANYOF_UNICODE_ALL;
588 if (!(and_with->flags & (ANYOF_UNICODE|ANYOF_UNICODE_ALL)))
589 cl->flags &= ~ANYOF_UNICODE;
592 /* 'OR' a given class with another one. Can create false positives */
593 /* We assume that cl is not inverted */
595 S_cl_or(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with)
597 if (or_with->flags & ANYOF_INVERT) {
599 * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
600 * <= (B1 | !B2) | (CL1 | !CL2)
601 * which is wasteful if CL2 is small, but we ignore CL2:
602 * (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
603 * XXXX Can we handle case-fold? Unclear:
604 * (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
605 * (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
607 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
608 && !(or_with->flags & ANYOF_FOLD)
609 && !(cl->flags & ANYOF_FOLD) ) {
612 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
613 cl->bitmap[i] |= ~or_with->bitmap[i];
614 } /* XXXX: logic is complicated otherwise */
616 cl_anything(pRExC_state, cl);
619 /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
620 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
621 && (!(or_with->flags & ANYOF_FOLD)
622 || (cl->flags & ANYOF_FOLD)) ) {
625 /* OR char bitmap and class bitmap separately */
626 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
627 cl->bitmap[i] |= or_with->bitmap[i];
628 if (or_with->flags & ANYOF_CLASS) {
629 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
630 cl->classflags[i] |= or_with->classflags[i];
631 cl->flags |= ANYOF_CLASS;
634 else { /* XXXX: logic is complicated, leave it along for a moment. */
635 cl_anything(pRExC_state, cl);
638 if (or_with->flags & ANYOF_EOS)
639 cl->flags |= ANYOF_EOS;
641 if (cl->flags & ANYOF_UNICODE && or_with->flags & ANYOF_UNICODE &&
642 ARG(cl) != ARG(or_with)) {
643 cl->flags |= ANYOF_UNICODE_ALL;
644 cl->flags &= ~ANYOF_UNICODE;
646 if (or_with->flags & ANYOF_UNICODE_ALL) {
647 cl->flags |= ANYOF_UNICODE_ALL;
648 cl->flags &= ~ANYOF_UNICODE;
652 /* REx optimizer. Converts nodes into quickier variants "in place".
653 Finds fixed substrings. */
655 /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
656 to the position after last scanned or to NULL. */
659 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
660 /* scanp: Start here (read-write). */
661 /* deltap: Write maxlen-minlen here. */
662 /* last: Stop before this one. */
664 I32 min = 0, pars = 0, code;
665 regnode *scan = *scanp, *next;
667 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
668 int is_inf_internal = 0; /* The studied chunk is infinite */
669 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
670 scan_data_t data_fake;
671 struct regnode_charclass_class and_with; /* Valid if flags & SCF_DO_STCLASS_OR */
673 while (scan && OP(scan) != END && scan < last) {
674 /* Peephole optimizer: */
676 if (PL_regkind[(U8)OP(scan)] == EXACT) {
677 /* Merge several consecutive EXACTish nodes into one. */
678 regnode *n = regnext(scan);
681 regnode *stop = scan;
684 next = scan + NODE_SZ_STR(scan);
685 /* Skip NOTHING, merge EXACT*. */
687 ( PL_regkind[(U8)OP(n)] == NOTHING ||
688 (stringok && (OP(n) == OP(scan))))
690 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
691 if (OP(n) == TAIL || n > next)
693 if (PL_regkind[(U8)OP(n)] == NOTHING) {
694 NEXT_OFF(scan) += NEXT_OFF(n);
695 next = n + NODE_STEP_REGNODE;
703 int oldl = STR_LEN(scan);
704 regnode *nnext = regnext(n);
706 if (oldl + STR_LEN(n) > U8_MAX)
708 NEXT_OFF(scan) += NEXT_OFF(n);
709 STR_LEN(scan) += STR_LEN(n);
710 next = n + NODE_SZ_STR(n);
711 /* Now we can overwrite *n : */
712 Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
721 n = scan + NODE_SZ_STR(scan);
723 if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
731 /* Follow the next-chain of the current node and optimize
732 away all the NOTHINGs from it. */
733 if (OP(scan) != CURLYX) {
734 int max = (reg_off_by_arg[OP(scan)]
736 /* I32 may be smaller than U16 on CRAYs! */
737 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
738 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
742 /* Skip NOTHING and LONGJMP. */
743 while ((n = regnext(n))
744 && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
745 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
748 if (reg_off_by_arg[OP(scan)])
751 NEXT_OFF(scan) = off;
753 /* The principal pseudo-switch. Cannot be a switch, since we
754 look into several different things. */
755 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
756 || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
757 next = regnext(scan);
760 if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
761 I32 max1 = 0, min1 = I32_MAX, num = 0;
762 struct regnode_charclass_class accum;
764 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
765 scan_commit(pRExC_state, data); /* Cannot merge strings after this. */
766 if (flags & SCF_DO_STCLASS)
767 cl_init_zero(pRExC_state, &accum);
768 while (OP(scan) == code) {
769 I32 deltanext, minnext, f = 0, fake;
770 struct regnode_charclass_class this_class;
775 data_fake.whilem_c = data->whilem_c;
776 data_fake.last_closep = data->last_closep;
779 data_fake.last_closep = &fake;
780 next = regnext(scan);
781 scan = NEXTOPER(scan);
783 scan = NEXTOPER(scan);
784 if (flags & SCF_DO_STCLASS) {
785 cl_init(pRExC_state, &this_class);
786 data_fake.start_class = &this_class;
787 f = SCF_DO_STCLASS_AND;
789 if (flags & SCF_WHILEM_VISITED_POS)
790 f |= SCF_WHILEM_VISITED_POS;
791 /* we suppose the run is continuous, last=next...*/
792 minnext = study_chunk(pRExC_state, &scan, &deltanext,
793 next, &data_fake, f);
796 if (max1 < minnext + deltanext)
797 max1 = minnext + deltanext;
798 if (deltanext == I32_MAX)
799 is_inf = is_inf_internal = 1;
801 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
803 if (data && (data_fake.flags & SF_HAS_EVAL))
804 data->flags |= SF_HAS_EVAL;
806 data->whilem_c = data_fake.whilem_c;
807 if (flags & SCF_DO_STCLASS)
808 cl_or(pRExC_state, &accum, &this_class);
812 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
814 if (flags & SCF_DO_SUBSTR) {
815 data->pos_min += min1;
816 data->pos_delta += max1 - min1;
817 if (max1 != min1 || is_inf)
818 data->longest = &(data->longest_float);
821 delta += max1 - min1;
822 if (flags & SCF_DO_STCLASS_OR) {
823 cl_or(pRExC_state, data->start_class, &accum);
825 cl_and(data->start_class, &and_with);
826 flags &= ~SCF_DO_STCLASS;
829 else if (flags & SCF_DO_STCLASS_AND) {
831 cl_and(data->start_class, &accum);
832 flags &= ~SCF_DO_STCLASS;
835 /* Switch to OR mode: cache the old value of
836 * data->start_class */
837 StructCopy(data->start_class, &and_with,
838 struct regnode_charclass_class);
839 flags &= ~SCF_DO_STCLASS_AND;
840 StructCopy(&accum, data->start_class,
841 struct regnode_charclass_class);
842 flags |= SCF_DO_STCLASS_OR;
843 data->start_class->flags |= ANYOF_EOS;
847 else if (code == BRANCHJ) /* single branch is optimized. */
848 scan = NEXTOPER(NEXTOPER(scan));
849 else /* single branch is optimized. */
850 scan = NEXTOPER(scan);
853 else if (OP(scan) == EXACT) {
854 I32 l = STR_LEN(scan);
855 UV uc = *((U8*)STRING(scan));
857 U8 *s = (U8*)STRING(scan);
858 l = utf8_length(s, s + l);
859 uc = utf8_to_uvchr(s, NULL);
862 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
863 /* The code below prefers earlier match for fixed
864 offset, later match for variable offset. */
865 if (data->last_end == -1) { /* Update the start info. */
866 data->last_start_min = data->pos_min;
867 data->last_start_max = is_inf
868 ? I32_MAX : data->pos_min + data->pos_delta;
870 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
871 data->last_end = data->pos_min + l;
872 data->pos_min += l; /* As in the first entry. */
873 data->flags &= ~SF_BEFORE_EOL;
875 if (flags & SCF_DO_STCLASS_AND) {
876 /* Check whether it is compatible with what we know already! */
880 !(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
881 && !ANYOF_BITMAP_TEST(data->start_class, uc)
882 && (!(data->start_class->flags & ANYOF_FOLD)
883 || !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
885 ANYOF_CLASS_ZERO(data->start_class);
886 ANYOF_BITMAP_ZERO(data->start_class);
888 ANYOF_BITMAP_SET(data->start_class, uc);
889 data->start_class->flags &= ~ANYOF_EOS;
891 data->start_class->flags &= ~ANYOF_UNICODE_ALL;
893 else if (flags & SCF_DO_STCLASS_OR) {
894 /* false positive possible if the class is case-folded */
896 ANYOF_BITMAP_SET(data->start_class, uc);
898 data->start_class->flags |= ANYOF_UNICODE_ALL;
899 data->start_class->flags &= ~ANYOF_EOS;
900 cl_and(data->start_class, &and_with);
902 flags &= ~SCF_DO_STCLASS;
904 else if (PL_regkind[(U8)OP(scan)] == EXACT) { /* But OP != EXACT! */
905 I32 l = STR_LEN(scan);
906 UV uc = *((U8*)STRING(scan));
908 /* Search for fixed substrings supports EXACT only. */
909 if (flags & SCF_DO_SUBSTR)
910 scan_commit(pRExC_state, data);
912 U8 *s = (U8 *)STRING(scan);
913 l = utf8_length(s, s + l);
914 uc = utf8_to_uvchr(s, NULL);
917 if (data && (flags & SCF_DO_SUBSTR))
919 if (flags & SCF_DO_STCLASS_AND) {
920 /* Check whether it is compatible with what we know already! */
924 !(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
925 && !ANYOF_BITMAP_TEST(data->start_class, uc)
926 && !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc]))
928 ANYOF_CLASS_ZERO(data->start_class);
929 ANYOF_BITMAP_ZERO(data->start_class);
931 ANYOF_BITMAP_SET(data->start_class, uc);
932 data->start_class->flags &= ~ANYOF_EOS;
933 data->start_class->flags |= ANYOF_FOLD;
934 if (OP(scan) == EXACTFL)
935 data->start_class->flags |= ANYOF_LOCALE;
938 else if (flags & SCF_DO_STCLASS_OR) {
939 if (data->start_class->flags & ANYOF_FOLD) {
940 /* false positive possible if the class is case-folded.
941 Assume that the locale settings are the same... */
943 ANYOF_BITMAP_SET(data->start_class, uc);
944 data->start_class->flags &= ~ANYOF_EOS;
946 cl_and(data->start_class, &and_with);
948 flags &= ~SCF_DO_STCLASS;
950 else if (strchr((char*)PL_varies,OP(scan))) {
951 I32 mincount, maxcount, minnext, deltanext, fl;
952 I32 f = flags, pos_before = 0;
953 regnode *oscan = scan;
954 struct regnode_charclass_class this_class;
955 struct regnode_charclass_class *oclass = NULL;
957 switch (PL_regkind[(U8)OP(scan)]) {
958 case WHILEM: /* End of (?:...)* . */
959 scan = NEXTOPER(scan);
962 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
963 next = NEXTOPER(scan);
964 if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
966 maxcount = REG_INFTY;
967 next = regnext(scan);
968 scan = NEXTOPER(scan);
972 if (flags & SCF_DO_SUBSTR)
977 if (flags & SCF_DO_STCLASS) {
979 maxcount = REG_INFTY;
980 next = regnext(scan);
981 scan = NEXTOPER(scan);
984 is_inf = is_inf_internal = 1;
985 scan = regnext(scan);
986 if (flags & SCF_DO_SUBSTR) {
987 scan_commit(pRExC_state, data); /* Cannot extend fixed substrings */
988 data->longest = &(data->longest_float);
990 goto optimize_curly_tail;
992 mincount = ARG1(scan);
993 maxcount = ARG2(scan);
994 next = regnext(scan);
995 if (OP(scan) == CURLYX) {
996 I32 lp = (data ? *(data->last_closep) : 0);
998 scan->flags = ((lp <= U8_MAX) ? lp : U8_MAX);
1000 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
1002 if (flags & SCF_DO_SUBSTR) {
1003 if (mincount == 0) scan_commit(pRExC_state,data); /* Cannot extend fixed substrings */
1004 pos_before = data->pos_min;
1008 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
1010 data->flags |= SF_IS_INF;
1012 if (flags & SCF_DO_STCLASS) {
1013 cl_init(pRExC_state, &this_class);
1014 oclass = data->start_class;
1015 data->start_class = &this_class;
1016 f |= SCF_DO_STCLASS_AND;
1017 f &= ~SCF_DO_STCLASS_OR;
1019 /* These are the cases when once a subexpression
1020 fails at a particular position, it cannot succeed
1021 even after backtracking at the enclosing scope.
1023 XXXX what if minimal match and we are at the
1024 initial run of {n,m}? */
1025 if ((mincount != maxcount - 1) && (maxcount != REG_INFTY))
1026 f &= ~SCF_WHILEM_VISITED_POS;
1028 /* This will finish on WHILEM, setting scan, or on NULL: */
1029 minnext = study_chunk(pRExC_state, &scan, &deltanext, last, data,
1031 ? (f & ~SCF_DO_SUBSTR) : f);
1033 if (flags & SCF_DO_STCLASS)
1034 data->start_class = oclass;
1035 if (mincount == 0 || minnext == 0) {
1036 if (flags & SCF_DO_STCLASS_OR) {
1037 cl_or(pRExC_state, data->start_class, &this_class);
1039 else if (flags & SCF_DO_STCLASS_AND) {
1040 /* Switch to OR mode: cache the old value of
1041 * data->start_class */
1042 StructCopy(data->start_class, &and_with,
1043 struct regnode_charclass_class);
1044 flags &= ~SCF_DO_STCLASS_AND;
1045 StructCopy(&this_class, data->start_class,
1046 struct regnode_charclass_class);
1047 flags |= SCF_DO_STCLASS_OR;
1048 data->start_class->flags |= ANYOF_EOS;
1050 } else { /* Non-zero len */
1051 if (flags & SCF_DO_STCLASS_OR) {
1052 cl_or(pRExC_state, data->start_class, &this_class);
1053 cl_and(data->start_class, &and_with);
1055 else if (flags & SCF_DO_STCLASS_AND)
1056 cl_and(data->start_class, &this_class);
1057 flags &= ~SCF_DO_STCLASS;
1059 if (!scan) /* It was not CURLYX, but CURLY. */
1061 if (ckWARN(WARN_REGEXP) && (minnext + deltanext == 0)
1062 && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
1063 && maxcount <= REG_INFTY/3) /* Complement check for big count */
1066 "Quantifier unexpected on zero-length expression");
1069 min += minnext * mincount;
1070 is_inf_internal |= ((maxcount == REG_INFTY
1071 && (minnext + deltanext) > 0)
1072 || deltanext == I32_MAX);
1073 is_inf |= is_inf_internal;
1074 delta += (minnext + deltanext) * maxcount - minnext * mincount;
1076 /* Try powerful optimization CURLYX => CURLYN. */
1077 if ( OP(oscan) == CURLYX && data
1078 && data->flags & SF_IN_PAR
1079 && !(data->flags & SF_HAS_EVAL)
1080 && !deltanext && minnext == 1 ) {
1081 /* Try to optimize to CURLYN. */
1082 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
1083 regnode *nxt1 = nxt, *nxt2;
1087 if (!strchr((char*)PL_simple,OP(nxt))
1088 && !(PL_regkind[(U8)OP(nxt)] == EXACT
1089 && STR_LEN(nxt) == 1))
1093 if (OP(nxt) != CLOSE)
1095 /* Now we know that nxt2 is the only contents: */
1096 oscan->flags = ARG(nxt);
1098 OP(nxt1) = NOTHING; /* was OPEN. */
1100 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1101 NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
1102 NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
1103 OP(nxt) = OPTIMIZED; /* was CLOSE. */
1104 OP(nxt + 1) = OPTIMIZED; /* was count. */
1105 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
1110 /* Try optimization CURLYX => CURLYM. */
1111 if ( OP(oscan) == CURLYX && data
1112 && !(data->flags & SF_HAS_PAR)
1113 && !(data->flags & SF_HAS_EVAL)
1115 /* XXXX How to optimize if data == 0? */
1116 /* Optimize to a simpler form. */
1117 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
1121 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
1122 && (OP(nxt2) != WHILEM))
1124 OP(nxt2) = SUCCEED; /* Whas WHILEM */
1125 /* Need to optimize away parenths. */
1126 if (data->flags & SF_IN_PAR) {
1127 /* Set the parenth number. */
1128 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
1130 if (OP(nxt) != CLOSE)
1131 FAIL("Panic opt close");
1132 oscan->flags = ARG(nxt);
1133 OP(nxt1) = OPTIMIZED; /* was OPEN. */
1134 OP(nxt) = OPTIMIZED; /* was CLOSE. */
1136 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1137 OP(nxt + 1) = OPTIMIZED; /* was count. */
1138 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
1139 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
1142 while ( nxt1 && (OP(nxt1) != WHILEM)) {
1143 regnode *nnxt = regnext(nxt1);
1146 if (reg_off_by_arg[OP(nxt1)])
1147 ARG_SET(nxt1, nxt2 - nxt1);
1148 else if (nxt2 - nxt1 < U16_MAX)
1149 NEXT_OFF(nxt1) = nxt2 - nxt1;
1151 OP(nxt) = NOTHING; /* Cannot beautify */
1156 /* Optimize again: */
1157 study_chunk(pRExC_state, &nxt1, &deltanext, nxt,
1163 else if ((OP(oscan) == CURLYX)
1164 && (flags & SCF_WHILEM_VISITED_POS)
1165 /* See the comment on a similar expression above.
1166 However, this time it not a subexpression
1167 we care about, but the expression itself. */
1168 && (maxcount == REG_INFTY)
1169 && data && ++data->whilem_c < 16) {
1170 /* This stays as CURLYX, we can put the count/of pair. */
1171 /* Find WHILEM (as in regexec.c) */
1172 regnode *nxt = oscan + NEXT_OFF(oscan);
1174 if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
1176 PREVOPER(nxt)->flags = data->whilem_c
1177 | (RExC_whilem_seen << 4); /* On WHILEM */
1179 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
1181 if (flags & SCF_DO_SUBSTR) {
1182 SV *last_str = Nullsv;
1183 int counted = mincount != 0;
1185 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
1186 I32 b = pos_before >= data->last_start_min
1187 ? pos_before : data->last_start_min;
1189 char *s = SvPV(data->last_found, l);
1190 I32 old = b - data->last_start_min;
1193 old = utf8_hop((U8*)s, old) - (U8*)s;
1196 /* Get the added string: */
1197 last_str = newSVpvn(s + old, l);
1198 if (deltanext == 0 && pos_before == b) {
1199 /* What was added is a constant string */
1201 SvGROW(last_str, (mincount * l) + 1);
1202 repeatcpy(SvPVX(last_str) + l,
1203 SvPVX(last_str), l, mincount - 1);
1204 SvCUR(last_str) *= mincount;
1205 /* Add additional parts. */
1206 SvCUR_set(data->last_found,
1207 SvCUR(data->last_found) - l);
1208 sv_catsv(data->last_found, last_str);
1209 data->last_end += l * (mincount - 1);
1212 /* start offset must point into the last copy */
1213 data->last_start_min += minnext * (mincount - 1);
1214 data->last_start_max += is_inf ? 0 : (maxcount - 1)
1215 * (minnext + data->pos_delta);
1218 /* It is counted once already... */
1219 data->pos_min += minnext * (mincount - counted);
1220 data->pos_delta += - counted * deltanext +
1221 (minnext + deltanext) * maxcount - minnext * mincount;
1222 if (mincount != maxcount) {
1223 /* Cannot extend fixed substrings found inside
1225 scan_commit(pRExC_state,data);
1226 if (mincount && last_str) {
1227 sv_setsv(data->last_found, last_str);
1228 data->last_end = data->pos_min;
1229 data->last_start_min =
1230 data->pos_min - CHR_SVLEN(last_str);
1231 data->last_start_max = is_inf
1233 : data->pos_min + data->pos_delta
1234 - CHR_SVLEN(last_str);
1236 data->longest = &(data->longest_float);
1238 SvREFCNT_dec(last_str);
1240 if (data && (fl & SF_HAS_EVAL))
1241 data->flags |= SF_HAS_EVAL;
1242 optimize_curly_tail:
1243 if (OP(oscan) != CURLYX) {
1244 while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
1246 NEXT_OFF(oscan) += NEXT_OFF(next);
1249 default: /* REF and CLUMP only? */
1250 if (flags & SCF_DO_SUBSTR) {
1251 scan_commit(pRExC_state,data); /* Cannot expect anything... */
1252 data->longest = &(data->longest_float);
1254 is_inf = is_inf_internal = 1;
1255 if (flags & SCF_DO_STCLASS_OR)
1256 cl_anything(pRExC_state, data->start_class);
1257 flags &= ~SCF_DO_STCLASS;
1261 else if (strchr((char*)PL_simple,OP(scan))) {
1264 if (flags & SCF_DO_SUBSTR) {
1265 scan_commit(pRExC_state,data);
1269 if (flags & SCF_DO_STCLASS) {
1270 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
1272 /* Some of the logic below assumes that switching
1273 locale on will only add false positives. */
1274 switch (PL_regkind[(U8)OP(scan)]) {
1278 /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
1279 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1280 cl_anything(pRExC_state, data->start_class);
1283 if (OP(scan) == SANY)
1285 if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
1286 value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
1287 || (data->start_class->flags & ANYOF_CLASS));
1288 cl_anything(pRExC_state, data->start_class);
1290 if (flags & SCF_DO_STCLASS_AND || !value)
1291 ANYOF_BITMAP_CLEAR(data->start_class,'\n');
1294 if (flags & SCF_DO_STCLASS_AND)
1295 cl_and(data->start_class,
1296 (struct regnode_charclass_class*)scan);
1298 cl_or(pRExC_state, data->start_class,
1299 (struct regnode_charclass_class*)scan);
1302 if (flags & SCF_DO_STCLASS_AND) {
1303 if (!(data->start_class->flags & ANYOF_LOCALE)) {
1304 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1305 for (value = 0; value < 256; value++)
1306 if (!isALNUM(value))
1307 ANYOF_BITMAP_CLEAR(data->start_class, value);
1311 if (data->start_class->flags & ANYOF_LOCALE)
1312 ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1314 for (value = 0; value < 256; value++)
1316 ANYOF_BITMAP_SET(data->start_class, value);
1321 if (flags & SCF_DO_STCLASS_AND) {
1322 if (data->start_class->flags & ANYOF_LOCALE)
1323 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1326 ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1327 data->start_class->flags |= ANYOF_LOCALE;
1331 if (flags & SCF_DO_STCLASS_AND) {
1332 if (!(data->start_class->flags & ANYOF_LOCALE)) {
1333 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1334 for (value = 0; value < 256; value++)
1336 ANYOF_BITMAP_CLEAR(data->start_class, value);
1340 if (data->start_class->flags & ANYOF_LOCALE)
1341 ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1343 for (value = 0; value < 256; value++)
1344 if (!isALNUM(value))
1345 ANYOF_BITMAP_SET(data->start_class, value);
1350 if (flags & SCF_DO_STCLASS_AND) {
1351 if (data->start_class->flags & ANYOF_LOCALE)
1352 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1355 data->start_class->flags |= ANYOF_LOCALE;
1356 ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1360 if (flags & SCF_DO_STCLASS_AND) {
1361 if (!(data->start_class->flags & ANYOF_LOCALE)) {
1362 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1363 for (value = 0; value < 256; value++)
1364 if (!isSPACE(value))
1365 ANYOF_BITMAP_CLEAR(data->start_class, value);
1369 if (data->start_class->flags & ANYOF_LOCALE)
1370 ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1372 for (value = 0; value < 256; value++)
1374 ANYOF_BITMAP_SET(data->start_class, value);
1379 if (flags & SCF_DO_STCLASS_AND) {
1380 if (data->start_class->flags & ANYOF_LOCALE)
1381 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1384 data->start_class->flags |= ANYOF_LOCALE;
1385 ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1389 if (flags & SCF_DO_STCLASS_AND) {
1390 if (!(data->start_class->flags & ANYOF_LOCALE)) {
1391 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1392 for (value = 0; value < 256; value++)
1394 ANYOF_BITMAP_CLEAR(data->start_class, value);
1398 if (data->start_class->flags & ANYOF_LOCALE)
1399 ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1401 for (value = 0; value < 256; value++)
1402 if (!isSPACE(value))
1403 ANYOF_BITMAP_SET(data->start_class, value);
1408 if (flags & SCF_DO_STCLASS_AND) {
1409 if (data->start_class->flags & ANYOF_LOCALE) {
1410 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1411 for (value = 0; value < 256; value++)
1412 if (!isSPACE(value))
1413 ANYOF_BITMAP_CLEAR(data->start_class, value);
1417 data->start_class->flags |= ANYOF_LOCALE;
1418 ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1422 if (flags & SCF_DO_STCLASS_AND) {
1423 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
1424 for (value = 0; value < 256; value++)
1425 if (!isDIGIT(value))
1426 ANYOF_BITMAP_CLEAR(data->start_class, value);
1429 if (data->start_class->flags & ANYOF_LOCALE)
1430 ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
1432 for (value = 0; value < 256; value++)
1434 ANYOF_BITMAP_SET(data->start_class, value);
1439 if (flags & SCF_DO_STCLASS_AND) {
1440 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT);
1441 for (value = 0; value < 256; value++)
1443 ANYOF_BITMAP_CLEAR(data->start_class, value);
1446 if (data->start_class->flags & ANYOF_LOCALE)
1447 ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT);
1449 for (value = 0; value < 256; value++)
1450 if (!isDIGIT(value))
1451 ANYOF_BITMAP_SET(data->start_class, value);
1456 if (flags & SCF_DO_STCLASS_OR)
1457 cl_and(data->start_class, &and_with);
1458 flags &= ~SCF_DO_STCLASS;
1461 else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
1462 data->flags |= (OP(scan) == MEOL
1466 else if ( PL_regkind[(U8)OP(scan)] == BRANCHJ
1467 /* Lookbehind, or need to calculate parens/evals/stclass: */
1468 && (scan->flags || data || (flags & SCF_DO_STCLASS))
1469 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
1470 /* Lookahead/lookbehind */
1471 I32 deltanext, minnext, fake = 0;
1473 struct regnode_charclass_class intrnl;
1476 data_fake.flags = 0;
1478 data_fake.whilem_c = data->whilem_c;
1479 data_fake.last_closep = data->last_closep;
1482 data_fake.last_closep = &fake;
1483 if ( flags & SCF_DO_STCLASS && !scan->flags
1484 && OP(scan) == IFMATCH ) { /* Lookahead */
1485 cl_init(pRExC_state, &intrnl);
1486 data_fake.start_class = &intrnl;
1487 f |= SCF_DO_STCLASS_AND;
1489 if (flags & SCF_WHILEM_VISITED_POS)
1490 f |= SCF_WHILEM_VISITED_POS;
1491 next = regnext(scan);
1492 nscan = NEXTOPER(NEXTOPER(scan));
1493 minnext = study_chunk(pRExC_state, &nscan, &deltanext, last, &data_fake, f);
1496 vFAIL("Variable length lookbehind not implemented");
1498 else if (minnext > U8_MAX) {
1499 vFAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
1501 scan->flags = minnext;
1503 if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
1505 if (data && (data_fake.flags & SF_HAS_EVAL))
1506 data->flags |= SF_HAS_EVAL;
1508 data->whilem_c = data_fake.whilem_c;
1509 if (f & SCF_DO_STCLASS_AND) {
1510 int was = (data->start_class->flags & ANYOF_EOS);
1512 cl_and(data->start_class, &intrnl);
1514 data->start_class->flags |= ANYOF_EOS;
1517 else if (OP(scan) == OPEN) {
1520 else if (OP(scan) == CLOSE) {
1521 if (ARG(scan) == is_par) {
1522 next = regnext(scan);
1524 if ( next && (OP(next) != WHILEM) && next < last)
1525 is_par = 0; /* Disable optimization */
1528 *(data->last_closep) = ARG(scan);
1530 else if (OP(scan) == EVAL) {
1532 data->flags |= SF_HAS_EVAL;
1534 else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded follows */
1535 if (flags & SCF_DO_SUBSTR) {
1536 scan_commit(pRExC_state,data);
1537 data->longest = &(data->longest_float);
1539 is_inf = is_inf_internal = 1;
1540 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1541 cl_anything(pRExC_state, data->start_class);
1542 flags &= ~SCF_DO_STCLASS;
1544 /* Else: zero-length, ignore. */
1545 scan = regnext(scan);
1550 *deltap = is_inf_internal ? I32_MAX : delta;
1551 if (flags & SCF_DO_SUBSTR && is_inf)
1552 data->pos_delta = I32_MAX - data->pos_min;
1553 if (is_par > U8_MAX)
1555 if (is_par && pars==1 && data) {
1556 data->flags |= SF_IN_PAR;
1557 data->flags &= ~SF_HAS_PAR;
1559 else if (pars && data) {
1560 data->flags |= SF_HAS_PAR;
1561 data->flags &= ~SF_IN_PAR;
1563 if (flags & SCF_DO_STCLASS_OR)
1564 cl_and(data->start_class, &and_with);
1569 S_add_data(pTHX_ RExC_state_t *pRExC_state, I32 n, char *s)
1571 if (RExC_rx->data) {
1572 Renewc(RExC_rx->data,
1573 sizeof(*RExC_rx->data) + sizeof(void*) * (RExC_rx->data->count + n - 1),
1574 char, struct reg_data);
1575 Renew(RExC_rx->data->what, RExC_rx->data->count + n, U8);
1576 RExC_rx->data->count += n;
1579 Newc(1207, RExC_rx->data, sizeof(*RExC_rx->data) + sizeof(void*) * (n - 1),
1580 char, struct reg_data);
1581 New(1208, RExC_rx->data->what, n, U8);
1582 RExC_rx->data->count = n;
1584 Copy(s, RExC_rx->data->what + RExC_rx->data->count - n, n, U8);
1585 return RExC_rx->data->count - n;
1589 Perl_reginitcolors(pTHX)
1592 char *s = PerlEnv_getenv("PERL_RE_COLORS");
1595 PL_colors[0] = s = savepv(s);
1597 s = strchr(s, '\t');
1603 PL_colors[i] = s = "";
1607 PL_colors[i++] = "";
1614 - pregcomp - compile a regular expression into internal code
1616 * We can't allocate space until we know how big the compiled form will be,
1617 * but we can't compile it (and thus know how big it is) until we've got a
1618 * place to put the code. So we cheat: we compile it twice, once with code
1619 * generation turned off and size counting turned on, and once "for real".
1620 * This also means that we don't allocate space until we are sure that the
1621 * thing really will compile successfully, and we never have to move the
1622 * code and thus invalidate pointers into it. (Note that it has to be in
1623 * one piece because free() must be able to free it all.) [NB: not true in perl]
1625 * Beware that the optimization-preparation code in here knows about some
1626 * of the structure of the compiled regexp. [I'll say.]
1629 Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
1639 RExC_state_t RExC_state;
1640 RExC_state_t *pRExC_state = &RExC_state;
1643 FAIL("NULL regexp argument");
1645 /* XXXX This looks very suspicious... */
1646 if (pm->op_pmdynflags & PMdf_CMP_UTF8)
1652 DEBUG_r(if (!PL_colorset) reginitcolors());
1653 DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n",
1654 PL_colors[4],PL_colors[5],PL_colors[0],
1655 (int)(xend - exp), RExC_precomp, PL_colors[1]));
1656 RExC_flags16 = pm->op_pmflags;
1660 RExC_seen_zerolen = *exp == '^' ? -1 : 0;
1661 RExC_seen_evals = 0;
1664 /* First pass: determine size, legality. */
1671 RExC_emit = &PL_regdummy;
1672 RExC_whilem_seen = 0;
1673 #if 0 /* REGC() is (currently) a NOP at the first pass.
1674 * Clever compilers notice this and complain. --jhi */
1675 REGC((U8)REG_MAGIC, (char*)RExC_emit);
1677 if (reg(pRExC_state, 0, &flags) == NULL) {
1678 RExC_precomp = Nullch;
1681 DEBUG_r(PerlIO_printf(Perl_debug_log, "size %"IVdf" ", (IV)RExC_size));
1683 /* Small enough for pointer-storage convention?
1684 If extralen==0, this means that we will not need long jumps. */
1685 if (RExC_size >= 0x10000L && RExC_extralen)
1686 RExC_size += RExC_extralen;
1689 if (RExC_whilem_seen > 15)
1690 RExC_whilem_seen = 15;
1692 /* Allocate space and initialize. */
1693 Newc(1001, r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode),
1696 FAIL("Regexp out of space");
1699 /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
1700 Zero(r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode), char);
1703 r->prelen = xend - exp;
1704 r->precomp = savepvn(RExC_precomp, r->prelen);
1706 r->reganch = pm->op_pmflags & PMf_COMPILETIME;
1707 r->nparens = RExC_npar - 1; /* set early to validate backrefs */
1709 r->substrs = 0; /* Useful during FAIL. */
1710 r->startp = 0; /* Useful during FAIL. */
1711 r->endp = 0; /* Useful during FAIL. */
1713 Newz(1304, r->offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
1715 r->offsets[0] = RExC_size;
1717 DEBUG_r(PerlIO_printf(Perl_debug_log,
1718 "%s %u bytes for offset annotations.\n",
1719 r->offsets ? "Got" : "Couldn't get",
1720 (2*RExC_size+1) * sizeof(U32)));
1724 /* Second pass: emit code. */
1729 RExC_emit_start = r->program;
1730 RExC_emit = r->program;
1731 /* Store the count of eval-groups for security checks: */
1732 RExC_emit->next_off = ((RExC_seen_evals > U16_MAX) ? U16_MAX : RExC_seen_evals);
1733 REGC((U8)REG_MAGIC, (char*) RExC_emit++);
1735 if (reg(pRExC_state, 0, &flags) == NULL)
1738 /* Dig out information for optimizations. */
1739 r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
1740 pm->op_pmflags = RExC_flags16;
1742 r->reganch |= ROPT_UTF8;
1743 r->regstclass = NULL;
1744 if (RExC_naughty >= 10) /* Probably an expensive pattern. */
1745 r->reganch |= ROPT_NAUGHTY;
1746 scan = r->program + 1; /* First BRANCH. */
1748 /* XXXX To minimize changes to RE engine we always allocate
1749 3-units-long substrs field. */
1750 Newz(1004, r->substrs, 1, struct reg_substr_data);
1752 StructCopy(&zero_scan_data, &data, scan_data_t);
1753 /* XXXX Should not we check for something else? Usually it is OPEN1... */
1754 if (OP(scan) != BRANCH) { /* Only one top-level choice. */
1756 STRLEN longest_float_length, longest_fixed_length;
1757 struct regnode_charclass_class ch_class;
1762 /* Skip introductions and multiplicators >= 1. */
1763 while ((OP(first) == OPEN && (sawopen = 1)) ||
1764 /* An OR of *one* alternative - should not happen now. */
1765 (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
1766 (OP(first) == PLUS) ||
1767 (OP(first) == MINMOD) ||
1768 /* An {n,m} with n>0 */
1769 (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
1770 if (OP(first) == PLUS)
1773 first += regarglen[(U8)OP(first)];
1774 first = NEXTOPER(first);
1777 /* Starting-point info. */
1779 if (PL_regkind[(U8)OP(first)] == EXACT) {
1780 if (OP(first) == EXACT)
1781 ; /* Empty, get anchored substr later. */
1782 else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
1783 r->regstclass = first;
1785 else if (strchr((char*)PL_simple,OP(first)))
1786 r->regstclass = first;
1787 else if (PL_regkind[(U8)OP(first)] == BOUND ||
1788 PL_regkind[(U8)OP(first)] == NBOUND)
1789 r->regstclass = first;
1790 else if (PL_regkind[(U8)OP(first)] == BOL) {
1791 r->reganch |= (OP(first) == MBOL
1793 : (OP(first) == SBOL
1796 first = NEXTOPER(first);
1799 else if (OP(first) == GPOS) {
1800 r->reganch |= ROPT_ANCH_GPOS;
1801 first = NEXTOPER(first);
1804 else if ((OP(first) == STAR &&
1805 PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
1806 !(r->reganch & ROPT_ANCH) )
1808 /* turn .* into ^.* with an implied $*=1 */
1809 int type = OP(NEXTOPER(first));
1811 if (type == REG_ANY)
1812 type = ROPT_ANCH_MBOL;
1814 type = ROPT_ANCH_SBOL;
1816 r->reganch |= type | ROPT_IMPLICIT;
1817 first = NEXTOPER(first);
1820 if (sawplus && (!sawopen || !RExC_sawback)
1821 && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
1822 /* x+ must match at the 1st pos of run of x's */
1823 r->reganch |= ROPT_SKIP;
1825 /* Scan is after the zeroth branch, first is atomic matcher. */
1826 DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
1827 (IV)(first - scan + 1)));
1829 * If there's something expensive in the r.e., find the
1830 * longest literal string that must appear and make it the
1831 * regmust. Resolve ties in favor of later strings, since
1832 * the regstart check works with the beginning of the r.e.
1833 * and avoiding duplication strengthens checking. Not a
1834 * strong reason, but sufficient in the absence of others.
1835 * [Now we resolve ties in favor of the earlier string if
1836 * it happens that c_offset_min has been invalidated, since the
1837 * earlier string may buy us something the later one won't.]
1841 data.longest_fixed = newSVpvn("",0);
1842 data.longest_float = newSVpvn("",0);
1843 data.last_found = newSVpvn("",0);
1844 data.longest = &(data.longest_fixed);
1846 if (!r->regstclass) {
1847 cl_init(pRExC_state, &ch_class);
1848 data.start_class = &ch_class;
1849 stclass_flag = SCF_DO_STCLASS_AND;
1850 } else /* XXXX Check for BOUND? */
1852 data.last_closep = &last_close;
1854 minlen = study_chunk(pRExC_state, &first, &fake, scan + RExC_size, /* Up to end */
1855 &data, SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag);
1856 if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
1857 && data.last_start_min == 0 && data.last_end > 0
1858 && !RExC_seen_zerolen
1859 && (!(RExC_seen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
1860 r->reganch |= ROPT_CHECK_ALL;
1861 scan_commit(pRExC_state, &data);
1862 SvREFCNT_dec(data.last_found);
1864 longest_float_length = CHR_SVLEN(data.longest_float);
1865 if (longest_float_length
1866 || (data.flags & SF_FL_BEFORE_EOL
1867 && (!(data.flags & SF_FL_BEFORE_MEOL)
1868 || (RExC_flags16 & PMf_MULTILINE)))) {
1871 if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */
1872 && data.offset_fixed == data.offset_float_min
1873 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
1874 goto remove_float; /* As in (a)+. */
1876 r->float_substr = data.longest_float;
1877 r->float_min_offset = data.offset_float_min;
1878 r->float_max_offset = data.offset_float_max;
1879 t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
1880 && (!(data.flags & SF_FL_BEFORE_MEOL)
1881 || (RExC_flags16 & PMf_MULTILINE)));
1882 fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0);
1886 r->float_substr = Nullsv;
1887 SvREFCNT_dec(data.longest_float);
1888 longest_float_length = 0;
1891 longest_fixed_length = CHR_SVLEN(data.longest_fixed);
1892 if (longest_fixed_length
1893 || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1894 && (!(data.flags & SF_FIX_BEFORE_MEOL)
1895 || (RExC_flags16 & PMf_MULTILINE)))) {
1898 r->anchored_substr = data.longest_fixed;
1899 r->anchored_offset = data.offset_fixed;
1900 t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
1901 && (!(data.flags & SF_FIX_BEFORE_MEOL)
1902 || (RExC_flags16 & PMf_MULTILINE)));
1903 fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0);
1906 r->anchored_substr = Nullsv;
1907 SvREFCNT_dec(data.longest_fixed);
1908 longest_fixed_length = 0;
1911 && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == SANY))
1912 r->regstclass = NULL;
1913 if ((!r->anchored_substr || r->anchored_offset) && stclass_flag
1914 && !(data.start_class->flags & ANYOF_EOS)
1915 && !cl_is_anything(data.start_class)) {
1917 I32 n = add_data(pRExC_state, 1, "f");
1919 New(1006, RExC_rx->data->data[n], 1,
1920 struct regnode_charclass_class);
1921 StructCopy(data.start_class,
1922 (struct regnode_charclass_class*)RExC_rx->data->data[n],
1923 struct regnode_charclass_class);
1924 r->regstclass = (regnode*)RExC_rx->data->data[n];
1925 r->reganch &= ~ROPT_SKIP; /* Used in find_byclass(). */
1926 PL_regdata = r->data; /* for regprop() */
1927 DEBUG_r((sv = sv_newmortal(),
1928 regprop(sv, (regnode*)data.start_class),
1929 PerlIO_printf(Perl_debug_log, "synthetic stclass `%s'.\n",
1933 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
1934 if (longest_fixed_length > longest_float_length) {
1935 r->check_substr = r->anchored_substr;
1936 r->check_offset_min = r->check_offset_max = r->anchored_offset;
1937 if (r->reganch & ROPT_ANCH_SINGLE)
1938 r->reganch |= ROPT_NOSCAN;
1941 r->check_substr = r->float_substr;
1942 r->check_offset_min = data.offset_float_min;
1943 r->check_offset_max = data.offset_float_max;
1945 /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
1946 This should be changed ASAP! */
1947 if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) {
1948 r->reganch |= RE_USE_INTUIT;
1949 if (SvTAIL(r->check_substr))
1950 r->reganch |= RE_INTUIT_TAIL;
1954 /* Several toplevels. Best we can is to set minlen. */
1956 struct regnode_charclass_class ch_class;
1959 DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
1960 scan = r->program + 1;
1961 cl_init(pRExC_state, &ch_class);
1962 data.start_class = &ch_class;
1963 data.last_closep = &last_close;
1964 minlen = study_chunk(pRExC_state, &scan, &fake, scan + RExC_size, &data, SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS);
1965 r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
1966 if (!(data.start_class->flags & ANYOF_EOS)
1967 && !cl_is_anything(data.start_class)) {
1969 I32 n = add_data(pRExC_state, 1, "f");
1971 New(1006, RExC_rx->data->data[n], 1,
1972 struct regnode_charclass_class);
1973 StructCopy(data.start_class,
1974 (struct regnode_charclass_class*)RExC_rx->data->data[n],
1975 struct regnode_charclass_class);
1976 r->regstclass = (regnode*)RExC_rx->data->data[n];
1977 r->reganch &= ~ROPT_SKIP; /* Used in find_byclass(). */
1978 DEBUG_r((sv = sv_newmortal(),
1979 regprop(sv, (regnode*)data.start_class),
1980 PerlIO_printf(Perl_debug_log, "synthetic stclass `%s'.\n",
1986 if (RExC_seen & REG_SEEN_GPOS)
1987 r->reganch |= ROPT_GPOS_SEEN;
1988 if (RExC_seen & REG_SEEN_LOOKBEHIND)
1989 r->reganch |= ROPT_LOOKBEHIND_SEEN;
1990 if (RExC_seen & REG_SEEN_EVAL)
1991 r->reganch |= ROPT_EVAL_SEEN;
1992 if (RExC_seen & REG_SEEN_SANY)
1993 r->reganch |= ROPT_SANY_SEEN;
1994 Newz(1002, r->startp, RExC_npar, I32);
1995 Newz(1002, r->endp, RExC_npar, I32);
1996 PL_regdata = r->data; /* for regprop() */
1997 DEBUG_r(regdump(r));
2002 - reg - regular expression, i.e. main body or parenthesized thing
2004 * Caller must absorb opening parenthesis.
2006 * Combining parenthesis handling with the base level of regular expression
2007 * is a trifle forced, but the need to tie the tails of the branches to what
2008 * follows makes it hard to avoid.
2011 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp)
2012 /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
2014 register regnode *ret; /* Will be the head of the group. */
2015 register regnode *br;
2016 register regnode *lastbr;
2017 register regnode *ender = 0;
2018 register I32 parno = 0;
2019 I32 flags, oregflags = RExC_flags16, have_branch = 0, open = 0;
2020 char * parse_start = RExC_parse; /* MJD */
2021 char *oregcomp_parse = RExC_parse;
2024 *flagp = 0; /* Tentatively. */
2026 /* Make an OPEN node, if parenthesized. */
2028 if (*RExC_parse == '?') { /* (?...) */
2029 U16 posflags = 0, negflags = 0;
2030 U16 *flagsp = &posflags;
2032 char *seqstart = RExC_parse;
2035 paren = *RExC_parse++;
2036 ret = NULL; /* For look-ahead/behind. */
2038 case '<': /* (?<...) */
2039 RExC_seen |= REG_SEEN_LOOKBEHIND;
2040 if (*RExC_parse == '!')
2042 if (*RExC_parse != '=' && *RExC_parse != '!')
2045 case '=': /* (?=...) */
2046 case '!': /* (?!...) */
2047 RExC_seen_zerolen++;
2048 case ':': /* (?:...) */
2049 case '>': /* (?>...) */
2051 case '$': /* (?$...) */
2052 case '@': /* (?@...) */
2053 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
2055 case '#': /* (?#...) */
2056 while (*RExC_parse && *RExC_parse != ')')
2058 if (*RExC_parse != ')')
2059 FAIL("Sequence (?#... not terminated");
2060 nextchar(pRExC_state);
2063 case 'p': /* (?p...) */
2065 vWARN(RExC_parse, "(?p{}) is deprecated - use (??{})");
2067 case '?': /* (??...) */
2069 paren = *RExC_parse++;
2071 case '{': /* (?{...}) */
2073 I32 count = 1, n = 0;
2075 char *s = RExC_parse;
2077 OP_4tree *sop, *rop;
2079 RExC_seen_zerolen++;
2080 RExC_seen |= REG_SEEN_EVAL;
2081 while (count && (c = *RExC_parse)) {
2082 if (c == '\\' && RExC_parse[1])
2090 if (*RExC_parse != ')')
2093 vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
2098 if (RExC_parse - 1 - s)
2099 sv = newSVpvn(s, RExC_parse - 1 - s);
2101 sv = newSVpvn("", 0);
2104 Perl_save_re_context(aTHX);
2105 rop = sv_compile_2op(sv, &sop, "re", &av);
2108 n = add_data(pRExC_state, 3, "nop");
2109 RExC_rx->data->data[n] = (void*)rop;
2110 RExC_rx->data->data[n+1] = (void*)sop;
2111 RExC_rx->data->data[n+2] = (void*)av;
2114 else { /* First pass */
2115 if (PL_reginterp_cnt < ++RExC_seen_evals
2116 && PL_curcop != &PL_compiling)
2117 /* No compiled RE interpolated, has runtime
2118 components ===> unsafe. */
2119 FAIL("Eval-group not allowed at runtime, use re 'eval'");
2121 FAIL("Eval-group in insecure regular expression");
2124 nextchar(pRExC_state);
2126 ret = reg_node(pRExC_state, LOGICAL);
2129 regtail(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
2130 /* deal with the length of this later - MJD */
2133 return reganode(pRExC_state, EVAL, n);
2135 case '(': /* (?(?{...})...) and (?(?=...)...) */
2137 if (RExC_parse[0] == '?') { /* (?(?...)) */
2138 if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
2139 || RExC_parse[1] == '<'
2140 || RExC_parse[1] == '{') { /* Lookahead or eval. */
2143 ret = reg_node(pRExC_state, LOGICAL);
2146 regtail(pRExC_state, ret, reg(pRExC_state, 1, &flag));
2150 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
2152 parno = atoi(RExC_parse++);
2154 while (isDIGIT(*RExC_parse))
2156 ret = reganode(pRExC_state, GROUPP, parno);
2158 if ((c = *nextchar(pRExC_state)) != ')')
2159 vFAIL("Switch condition not recognized");
2161 regtail(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
2162 br = regbranch(pRExC_state, &flags, 1);
2164 br = reganode(pRExC_state, LONGJMP, 0);
2166 regtail(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
2167 c = *nextchar(pRExC_state);
2171 lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
2172 regbranch(pRExC_state, &flags, 1);
2173 regtail(pRExC_state, ret, lastbr);
2176 c = *nextchar(pRExC_state);
2181 vFAIL("Switch (?(condition)... contains too many branches");
2182 ender = reg_node(pRExC_state, TAIL);
2183 regtail(pRExC_state, br, ender);
2185 regtail(pRExC_state, lastbr, ender);
2186 regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
2189 regtail(pRExC_state, ret, ender);
2193 vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
2197 RExC_parse--; /* for vFAIL to print correctly */
2198 vFAIL("Sequence (? incomplete");
2202 parse_flags: /* (?i) */
2203 while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
2204 if (*RExC_parse != 'o')
2205 pmflag(flagsp, *RExC_parse);
2208 if (*RExC_parse == '-') {
2213 RExC_flags16 |= posflags;
2214 RExC_flags16 &= ~negflags;
2215 if (*RExC_parse == ':') {
2221 if (*RExC_parse != ')') {
2223 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
2225 nextchar(pRExC_state);
2233 ret = reganode(pRExC_state, OPEN, parno);
2234 Set_Node_Length(ret, 1); /* MJD */
2235 Set_Node_Offset(ret, RExC_parse); /* MJD */
2242 /* Pick up the branches, linking them together. */
2243 parse_start = RExC_parse; /* MJD */
2244 br = regbranch(pRExC_state, &flags, 1);
2245 /* branch_len = (paren != 0); */
2249 if (*RExC_parse == '|') {
2250 if (!SIZE_ONLY && RExC_extralen) {
2251 reginsert(pRExC_state, BRANCHJ, br);
2254 reginsert(pRExC_state, BRANCH, br);
2255 Set_Node_Length(br, paren != 0);
2256 Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
2260 RExC_extralen += 1; /* For BRANCHJ-BRANCH. */
2262 else if (paren == ':') {
2263 *flagp |= flags&SIMPLE;
2265 if (open) { /* Starts with OPEN. */
2266 regtail(pRExC_state, ret, br); /* OPEN -> first. */
2268 else if (paren != '?') /* Not Conditional */
2272 *flagp |= flags&SPSTART;
2274 while (*RExC_parse == '|') {
2275 if (!SIZE_ONLY && RExC_extralen) {
2276 ender = reganode(pRExC_state, LONGJMP,0);
2277 regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
2280 RExC_extralen += 2; /* Account for LONGJMP. */
2281 nextchar(pRExC_state);
2282 br = regbranch(pRExC_state, &flags, 0);
2286 regtail(pRExC_state, lastbr, br); /* BRANCH -> BRANCH. */
2290 *flagp |= flags&SPSTART;
2293 if (have_branch || paren != ':') {
2294 /* Make a closing node, and hook it on the end. */
2297 ender = reg_node(pRExC_state, TAIL);
2300 ender = reganode(pRExC_state, CLOSE, parno);
2301 Set_Node_Offset(ender,RExC_parse+1); /* MJD */
2302 Set_Node_Length(ender,1); /* MJD */
2308 *flagp &= ~HASWIDTH;
2311 ender = reg_node(pRExC_state, SUCCEED);
2314 ender = reg_node(pRExC_state, END);
2317 regtail(pRExC_state, lastbr, ender);
2320 /* Hook the tails of the branches to the closing node. */
2321 for (br = ret; br != NULL; br = regnext(br)) {
2322 regoptail(pRExC_state, br, ender);
2329 static char parens[] = "=!<,>";
2331 if (paren && (p = strchr(parens, paren))) {
2332 int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
2333 int flag = (p - parens) > 1;
2336 node = SUSPEND, flag = 0;
2337 reginsert(pRExC_state, node,ret);
2339 regtail(pRExC_state, ret, reg_node(pRExC_state, TAIL));
2343 /* Check for proper termination. */
2345 RExC_flags16 = oregflags;
2346 if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
2347 RExC_parse = oregcomp_parse;
2348 vFAIL("Unmatched (");
2351 else if (!paren && RExC_parse < RExC_end) {
2352 if (*RExC_parse == ')') {
2354 vFAIL("Unmatched )");
2357 FAIL("Junk on end of regexp"); /* "Can't happen". */
2365 - regbranch - one alternative of an | operator
2367 * Implements the concatenation operator.
2370 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first)
2372 register regnode *ret;
2373 register regnode *chain = NULL;
2374 register regnode *latest;
2375 I32 flags = 0, c = 0;
2380 if (!SIZE_ONLY && RExC_extralen)
2381 ret = reganode(pRExC_state, BRANCHJ,0);
2383 ret = reg_node(pRExC_state, BRANCH);
2384 Set_Node_Length(ret, 1);
2388 if (!first && SIZE_ONLY)
2389 RExC_extralen += 1; /* BRANCHJ */
2391 *flagp = WORST; /* Tentatively. */
2394 nextchar(pRExC_state);
2395 while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
2397 latest = regpiece(pRExC_state, &flags);
2398 if (latest == NULL) {
2399 if (flags & TRYAGAIN)
2403 else if (ret == NULL)
2405 *flagp |= flags&HASWIDTH;
2406 if (chain == NULL) /* First piece. */
2407 *flagp |= flags&SPSTART;
2410 regtail(pRExC_state, chain, latest);
2415 if (chain == NULL) { /* Loop ran zero times. */
2416 chain = reg_node(pRExC_state, NOTHING);
2421 *flagp |= flags&SIMPLE;
2428 - regpiece - something followed by possible [*+?]
2430 * Note that the branching code sequences used for ? and the general cases
2431 * of * and + are somewhat optimized: they use the same NOTHING node as
2432 * both the endmarker for their branch list and the body of the last branch.
2433 * It might seem that this node could be dispensed with entirely, but the
2434 * endmarker role is not redundant.
2437 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2439 register regnode *ret;
2441 register char *next;
2443 char *origparse = RExC_parse;
2446 I32 max = REG_INFTY;
2449 ret = regatom(pRExC_state, &flags);
2451 if (flags & TRYAGAIN)
2458 if (op == '{' && regcurly(RExC_parse)) {
2459 parse_start = RExC_parse; /* MJD */
2460 next = RExC_parse + 1;
2462 while (isDIGIT(*next) || *next == ',') {
2471 if (*next == '}') { /* got one */
2475 min = atoi(RExC_parse);
2479 maxpos = RExC_parse;
2481 if (!max && *maxpos != '0')
2482 max = REG_INFTY; /* meaning "infinity" */
2483 else if (max >= REG_INFTY)
2484 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
2486 nextchar(pRExC_state);
2489 if ((flags&SIMPLE)) {
2490 RExC_naughty += 2 + RExC_naughty / 2;
2491 reginsert(pRExC_state, CURLY, ret);
2492 Set_Node_Offset(ret, parse_start+1); /* MJD */
2493 Set_Node_Cur_Length(ret);
2496 regnode *w = reg_node(pRExC_state, WHILEM);
2499 regtail(pRExC_state, ret, w);
2500 if (!SIZE_ONLY && RExC_extralen) {
2501 reginsert(pRExC_state, LONGJMP,ret);
2502 reginsert(pRExC_state, NOTHING,ret);
2503 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
2505 reginsert(pRExC_state, CURLYX,ret);
2507 Set_Node_Offset(ret, parse_start+1);
2508 Set_Node_Length(ret,
2509 op == '{' ? (RExC_parse - parse_start) : 1);
2511 if (!SIZE_ONLY && RExC_extralen)
2512 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
2513 regtail(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
2515 RExC_whilem_seen++, RExC_extralen += 3;
2516 RExC_naughty += 4 + RExC_naughty; /* compound interest */
2524 if (max && max < min)
2525 vFAIL("Can't do {n,m} with n > m");
2540 #if 0 /* Now runtime fix should be reliable. */
2542 /* if this is reinstated, don't forget to put this back into perldiag:
2544 =item Regexp *+ operand could be empty at {#} in regex m/%s/
2546 (F) The part of the regexp subject to either the * or + quantifier
2547 could match an empty string. The {#} shows in the regular
2548 expression about where the problem was discovered.
2552 if (!(flags&HASWIDTH) && op != '?')
2553 vFAIL("Regexp *+ operand could be empty");
2556 parse_start = RExC_parse;
2557 nextchar(pRExC_state);
2559 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
2561 if (op == '*' && (flags&SIMPLE)) {
2562 reginsert(pRExC_state, STAR, ret);
2566 else if (op == '*') {
2570 else if (op == '+' && (flags&SIMPLE)) {
2571 reginsert(pRExC_state, PLUS, ret);
2575 else if (op == '+') {
2579 else if (op == '?') {
2584 if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
2586 "%.*s matches null string many times",
2587 RExC_parse - origparse,
2591 if (*RExC_parse == '?') {
2592 nextchar(pRExC_state);
2593 reginsert(pRExC_state, MINMOD, ret);
2594 regtail(pRExC_state, ret, ret + NODE_STEP_REGNODE);
2596 if (ISMULT2(RExC_parse)) {
2598 vFAIL("Nested quantifiers");
2605 - regatom - the lowest level
2607 * Optimization: gobbles an entire sequence of ordinary characters so that
2608 * it can turn them into a single node, which is smaller to store and
2609 * faster to run. Backslashed characters are exceptions, each becoming a
2610 * separate node; the code is simpler that way and it's not worth fixing.
2612 * [Yes, it is worth fixing, some scripts can run twice the speed.] */
2614 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2616 register regnode *ret = 0;
2618 char *parse_start = RExC_parse;
2620 *flagp = WORST; /* Tentatively. */
2623 switch (*RExC_parse) {
2625 RExC_seen_zerolen++;
2626 nextchar(pRExC_state);
2627 if (RExC_flags16 & PMf_MULTILINE)
2628 ret = reg_node(pRExC_state, MBOL);
2629 else if (RExC_flags16 & PMf_SINGLELINE)
2630 ret = reg_node(pRExC_state, SBOL);
2632 ret = reg_node(pRExC_state, BOL);
2633 Set_Node_Length(ret, 1); /* MJD */
2636 nextchar(pRExC_state);
2638 RExC_seen_zerolen++;
2639 if (RExC_flags16 & PMf_MULTILINE)
2640 ret = reg_node(pRExC_state, MEOL);
2641 else if (RExC_flags16 & PMf_SINGLELINE)
2642 ret = reg_node(pRExC_state, SEOL);
2644 ret = reg_node(pRExC_state, EOL);
2645 Set_Node_Length(ret, 1); /* MJD */
2648 nextchar(pRExC_state);
2649 if (RExC_flags16 & PMf_SINGLELINE)
2650 ret = reg_node(pRExC_state, SANY);
2652 ret = reg_node(pRExC_state, REG_ANY);
2653 *flagp |= HASWIDTH|SIMPLE;
2655 Set_Node_Length(ret, 1); /* MJD */
2659 char *oregcomp_parse = ++RExC_parse;
2660 ret = regclass(pRExC_state);
2661 if (*RExC_parse != ']') {
2662 RExC_parse = oregcomp_parse;
2663 vFAIL("Unmatched [");
2665 nextchar(pRExC_state);
2666 *flagp |= HASWIDTH|SIMPLE;
2667 Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
2671 nextchar(pRExC_state);
2672 ret = reg(pRExC_state, 1, &flags);
2674 if (flags & TRYAGAIN) {
2675 if (RExC_parse == RExC_end) {
2676 /* Make parent create an empty node if needed. */
2684 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
2688 if (flags & TRYAGAIN) {
2692 vFAIL("Internal urp");
2693 /* Supposed to be caught earlier. */
2696 if (!regcurly(RExC_parse)) {
2705 vFAIL("Quantifier follows nothing");
2708 switch (*++RExC_parse) {
2710 RExC_seen_zerolen++;
2711 ret = reg_node(pRExC_state, SBOL);
2713 nextchar(pRExC_state);
2714 Set_Node_Length(ret, 2); /* MJD */
2717 ret = reg_node(pRExC_state, GPOS);
2718 RExC_seen |= REG_SEEN_GPOS;
2720 nextchar(pRExC_state);
2721 Set_Node_Length(ret, 2); /* MJD */
2724 ret = reg_node(pRExC_state, SEOL);
2726 nextchar(pRExC_state);
2729 ret = reg_node(pRExC_state, EOS);
2731 RExC_seen_zerolen++; /* Do not optimize RE away */
2732 nextchar(pRExC_state);
2733 Set_Node_Length(ret, 2); /* MJD */
2736 ret = reg_node(pRExC_state, SANY);
2737 RExC_seen |= REG_SEEN_SANY;
2738 *flagp |= HASWIDTH|SIMPLE;
2739 nextchar(pRExC_state);
2740 Set_Node_Length(ret, 2); /* MJD */
2743 ret = reg_node(pRExC_state, CLUMP);
2745 nextchar(pRExC_state);
2746 Set_Node_Length(ret, 2); /* MJD */
2749 ret = reg_node(pRExC_state, LOC ? ALNUML : ALNUM);
2750 *flagp |= HASWIDTH|SIMPLE;
2751 nextchar(pRExC_state);
2752 Set_Node_Length(ret, 2); /* MJD */
2755 ret = reg_node(pRExC_state, LOC ? NALNUML : NALNUM);
2756 *flagp |= HASWIDTH|SIMPLE;
2757 nextchar(pRExC_state);
2758 Set_Node_Length(ret, 2); /* MJD */
2761 RExC_seen_zerolen++;
2762 RExC_seen |= REG_SEEN_LOOKBEHIND;
2763 ret = reg_node(pRExC_state, LOC ? BOUNDL : BOUND);
2765 nextchar(pRExC_state);
2766 Set_Node_Length(ret, 2); /* MJD */
2769 RExC_seen_zerolen++;
2770 RExC_seen |= REG_SEEN_LOOKBEHIND;
2771 ret = reg_node(pRExC_state, LOC ? NBOUNDL : NBOUND);
2773 nextchar(pRExC_state);
2774 Set_Node_Length(ret, 2); /* MJD */
2777 ret = reg_node(pRExC_state, LOC ? SPACEL : SPACE);
2778 *flagp |= HASWIDTH|SIMPLE;
2779 nextchar(pRExC_state);
2780 Set_Node_Length(ret, 2); /* MJD */
2783 ret = reg_node(pRExC_state, LOC ? NSPACEL : NSPACE);
2784 *flagp |= HASWIDTH|SIMPLE;
2785 nextchar(pRExC_state);
2786 Set_Node_Length(ret, 2); /* MJD */
2789 ret = reg_node(pRExC_state, DIGIT);
2790 *flagp |= HASWIDTH|SIMPLE;
2791 nextchar(pRExC_state);
2792 Set_Node_Length(ret, 2); /* MJD */
2795 ret = reg_node(pRExC_state, NDIGIT);
2796 *flagp |= HASWIDTH|SIMPLE;
2797 nextchar(pRExC_state);
2798 Set_Node_Length(ret, 2); /* MJD */
2803 char* oldregxend = RExC_end;
2804 char* parse_start = RExC_parse;
2806 if (RExC_parse[1] == '{') {
2807 /* a lovely hack--pretend we saw [\pX] instead */
2808 RExC_end = strchr(RExC_parse, '}');
2811 RExC_end = oldregxend;
2812 vFAIL("Missing right brace on \\p{}");
2817 RExC_end = RExC_parse + 2;
2820 ret = regclass(pRExC_state);
2822 RExC_end = oldregxend;
2824 Set_Node_Cur_Length(ret); /* MJD */
2825 nextchar(pRExC_state);
2826 *flagp |= HASWIDTH|SIMPLE;
2839 case '1': case '2': case '3': case '4':
2840 case '5': case '6': case '7': case '8': case '9':
2842 I32 num = atoi(RExC_parse);
2844 if (num > 9 && num >= RExC_npar)
2847 char * parse_start = RExC_parse - 1; /* MJD */
2848 while (isDIGIT(*RExC_parse))
2851 if (!SIZE_ONLY && num > RExC_rx->nparens)
2852 vFAIL("Reference to nonexistent group");
2854 ret = reganode(pRExC_state, FOLD
2855 ? (LOC ? REFFL : REFF)
2859 /* override incorrect value set in reganode MJD */
2860 Set_Node_Offset(ret, parse_start+1);
2861 Set_Node_Cur_Length(ret); /* MJD */
2863 nextchar(pRExC_state);
2868 if (RExC_parse >= RExC_end)
2869 FAIL("Trailing \\");
2872 /* Do not generate `unrecognized' warnings here, we fall
2873 back into the quick-grab loop below */
2879 if (RExC_flags16 & PMf_EXTENDED) {
2880 while (RExC_parse < RExC_end && *RExC_parse != '\n') RExC_parse++;
2881 if (RExC_parse < RExC_end)
2887 register STRLEN len;
2892 char *parse_start = RExC_parse - 1;
2897 ret = reg_node(pRExC_state, FOLD
2898 ? (LOC ? EXACTFL : EXACTF)
2901 for (len = 0, p = RExC_parse - 1;
2902 len < 127 && p < RExC_end;
2907 if (RExC_flags16 & PMf_EXTENDED)
2908 p = regwhite(p, RExC_end);
2953 ender = ASCII_TO_NATIVE('\033');
2957 ender = ASCII_TO_NATIVE('\007');
2962 char* e = strchr(p, '}');
2966 vFAIL("Missing right brace on \\x{}");
2969 numlen = 1; /* allow underscores */
2970 ender = (UV)scan_hex(p + 1, e - p - 1, &numlen);
2973 /* numlen is generous */
2974 if (numlen + len >= 127) {
2982 numlen = 0; /* disallow underscores */
2983 ender = (UV)scan_hex(p, 2, &numlen);
2989 ender = UCHARAT(p++);
2990 ender = toCTRL(ender);
2992 case '0': case '1': case '2': case '3':case '4':
2993 case '5': case '6': case '7': case '8':case '9':
2995 (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
2996 numlen = 0; /* disallow underscores */
2997 ender = (UV)scan_oct(p, 3, &numlen);
3007 FAIL("Trailing \\");
3010 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p))
3011 vWARN2(p +1, "Unrecognized escape \\%c passed through", *p);
3012 goto normal_default;
3017 if (UTF8_IS_START(*p) && UTF) {
3018 ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
3026 if (RExC_flags16 & PMf_EXTENDED)
3027 p = regwhite(p, RExC_end);
3030 ender = toLOWER_LC_uvchr(ender);
3032 ender = toLOWER_uni(ender);
3034 if (ISMULT2(p)) { /* Back off on ?+*. */
3037 else if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3038 reguni(pRExC_state, ender, s, &numlen);
3048 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3049 reguni(pRExC_state, ender, s, &numlen);
3058 Set_Node_Cur_Length(ret); /* MJD */
3059 nextchar(pRExC_state);
3061 /* len is STRLEN which is unsigned, need to copy to signed */
3064 vFAIL("Internal disaster");
3073 RExC_size += STR_SZ(len);
3075 RExC_emit += STR_SZ(len);
3084 S_regwhite(pTHX_ char *p, char *e)
3089 else if (*p == '#') {
3092 } while (p < e && *p != '\n');
3100 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
3101 Character classes ([:foo:]) can also be negated ([:^foo:]).
3102 Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
3103 Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
3104 but trigger warnings because they are currently unimplemented. */
3106 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
3109 I32 namedclass = OOB_NAMEDCLASS;
3111 if (value == '[' && RExC_parse + 1 < RExC_end &&
3112 /* I smell either [: or [= or [. -- POSIX has been here, right? */
3113 (*RExC_parse == ':' ||
3114 *RExC_parse == '=' ||
3115 *RExC_parse == '.')) {
3116 char c = *RExC_parse;
3117 char* s = RExC_parse++;
3119 while (RExC_parse < RExC_end && *RExC_parse != c)
3121 if (RExC_parse == RExC_end)
3122 /* Grandfather lone [:, [=, [. */
3125 char* t = RExC_parse++; /* skip over the c */
3127 if (*RExC_parse == ']') {
3128 RExC_parse++; /* skip over the ending ] */
3131 I32 complement = *posixcc == '^' ? *posixcc++ : 0;
3132 I32 skip = 5; /* the most common skip */
3136 if (strnEQ(posixcc, "alnum", 5))
3138 complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
3139 else if (strnEQ(posixcc, "alpha", 5))
3141 complement ? ANYOF_NALPHA : ANYOF_ALPHA;
3142 else if (strnEQ(posixcc, "ascii", 5))
3144 complement ? ANYOF_NASCII : ANYOF_ASCII;
3147 if (strnEQ(posixcc, "blank", 5))
3149 complement ? ANYOF_NBLANK : ANYOF_BLANK;
3152 if (strnEQ(posixcc, "cntrl", 5))
3154 complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
3157 if (strnEQ(posixcc, "digit", 5))
3159 complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
3162 if (strnEQ(posixcc, "graph", 5))
3164 complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
3167 if (strnEQ(posixcc, "lower", 5))
3169 complement ? ANYOF_NLOWER : ANYOF_LOWER;
3172 if (strnEQ(posixcc, "print", 5))
3174 complement ? ANYOF_NPRINT : ANYOF_PRINT;
3175 else if (strnEQ(posixcc, "punct", 5))
3177 complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
3180 if (strnEQ(posixcc, "space", 5))
3182 complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
3185 if (strnEQ(posixcc, "upper", 5))
3187 complement ? ANYOF_NUPPER : ANYOF_UPPER;
3189 case 'w': /* this is not POSIX, this is the Perl \w */
3190 if (strnEQ(posixcc, "word", 4)) {
3192 complement ? ANYOF_NALNUM : ANYOF_ALNUM;
3197 if (strnEQ(posixcc, "xdigit", 6)) {
3199 complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
3204 if (namedclass == OOB_NAMEDCLASS ||
3205 posixcc[skip] != ':' ||
3206 posixcc[skip+1] != ']')
3208 Simple_vFAIL3("POSIX class [:%.*s:] unknown",
3211 } else if (!SIZE_ONLY) {
3212 /* [[=foo=]] and [[.foo.]] are still future. */
3214 /* adjust RExC_parse so the warning shows after
3216 while (*RExC_parse && *RExC_parse != ']')
3218 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3221 /* Maternal grandfather:
3222 * "[:" ending in ":" but not in ":]" */
3232 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
3234 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) &&
3235 (*RExC_parse == ':' ||
3236 *RExC_parse == '=' ||
3237 *RExC_parse == '.')) {
3238 char *s = RExC_parse;
3241 while(*s && isALNUM(*s))
3243 if (*s && c == *s && s[1] == ']') {
3244 vWARN3(s+2, "POSIX syntax [%c %c] belongs inside character classes", c, c);
3246 /* [[=foo=]] and [[.foo.]] are still future. */
3247 if (c == '=' || c == '.')
3249 /* adjust RExC_parse so the error shows after
3251 while (*RExC_parse && *RExC_parse++ != ']')
3253 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3260 S_regclass(pTHX_ RExC_state_t *pRExC_state)
3263 register IV prevvalue = OOB_UNICODE;
3264 register IV range = 0;
3265 register regnode *ret;
3269 bool need_class = 0;
3272 char *parse_start = RExC_parse; /* MJD */
3274 bool optimize_invert = TRUE;
3276 ret = reganode(pRExC_state, ANYOF, 0);
3279 ANYOF_FLAGS(ret) = 0;
3281 if (*RExC_parse == '^') { /* Complement of range. */
3285 ANYOF_FLAGS(ret) |= ANYOF_INVERT;
3289 RExC_size += ANYOF_SKIP;
3291 RExC_emit += ANYOF_SKIP;
3293 ANYOF_FLAGS(ret) |= ANYOF_FOLD;
3295 ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
3296 ANYOF_BITMAP_ZERO(ret);
3297 listsv = newSVpvn("# comment\n", 10);
3300 if (!SIZE_ONLY && ckWARN(WARN_REGEXP))
3301 checkposixcc(pRExC_state);
3303 if (*RExC_parse == ']' || *RExC_parse == '-')
3304 goto charclassloop; /* allow 1st char to be ] or - */
3306 while (RExC_parse < RExC_end && *RExC_parse != ']') {
3310 namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
3313 rangebegin = RExC_parse;
3315 value = utf8n_to_uvchr((U8*)RExC_parse,
3316 RExC_end - RExC_parse,
3318 RExC_parse += numlen;
3321 value = UCHARAT(RExC_parse++);
3323 namedclass = regpposixcc(pRExC_state, value);
3324 else if (value == '\\') {
3326 value = utf8n_to_uvchr((U8*)RExC_parse,
3327 RExC_end - RExC_parse,
3329 RExC_parse += numlen;
3332 value = UCHARAT(RExC_parse++);
3333 /* Some compilers cannot handle switching on 64-bit integer
3334 * values, therefore value cannot be an UV. Yes, this will
3335 * be a problem later if we want switch on Unicode.
3336 * A similar issue a little bit later when switching on
3337 * namedclass. --jhi */
3338 switch ((I32)value) {
3339 case 'w': namedclass = ANYOF_ALNUM; break;
3340 case 'W': namedclass = ANYOF_NALNUM; break;
3341 case 's': namedclass = ANYOF_SPACE; break;
3342 case 'S': namedclass = ANYOF_NSPACE; break;
3343 case 'd': namedclass = ANYOF_DIGIT; break;
3344 case 'D': namedclass = ANYOF_NDIGIT; break;
3347 if (*RExC_parse == '{') {
3348 e = strchr(RExC_parse++, '}');
3350 vFAIL("Missing right brace on \\p{}");
3359 Perl_sv_catpvf(aTHX_ listsv,
3360 "+utf8::%.*s\n", (int)n, RExC_parse);
3362 Perl_sv_catpvf(aTHX_ listsv,
3363 "!utf8::%.*s\n", (int)n, RExC_parse);
3366 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3368 case 'n': value = '\n'; break;
3369 case 'r': value = '\r'; break;
3370 case 't': value = '\t'; break;
3371 case 'f': value = '\f'; break;
3372 case 'b': value = '\b'; break;
3373 case 'e': value = ASCII_TO_NATIVE('\033');break;
3374 case 'a': value = ASCII_TO_NATIVE('\007');break;
3376 if (*RExC_parse == '{') {
3377 e = strchr(RExC_parse++, '}');
3379 vFAIL("Missing right brace on \\x{}");
3380 numlen = 1; /* allow underscores */
3381 value = (UV)scan_hex(RExC_parse,
3387 numlen = 0; /* disallow underscores */
3388 value = (UV)scan_hex(RExC_parse, 2, &numlen);
3389 RExC_parse += numlen;
3393 value = UCHARAT(RExC_parse++);
3394 value = toCTRL(value);
3396 case '0': case '1': case '2': case '3': case '4':
3397 case '5': case '6': case '7': case '8': case '9':
3398 numlen = 0; /* disallow underscores */
3399 value = (UV)scan_oct(--RExC_parse, 3, &numlen);
3400 RExC_parse += numlen;
3403 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
3405 "Unrecognized escape \\%c in character class passed through",
3409 } /* end of \blah */
3411 if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
3413 if (!SIZE_ONLY && !need_class)
3414 ANYOF_CLASS_ZERO(ret);
3418 /* a bad range like a-\d, a-[:digit:] ? */
3421 if (ckWARN(WARN_REGEXP))
3423 "False [] range \"%*.*s\"",
3424 RExC_parse - rangebegin,
3425 RExC_parse - rangebegin,
3427 if (prevvalue < 256) {
3428 ANYOF_BITMAP_SET(ret, prevvalue);
3429 ANYOF_BITMAP_SET(ret, '-');
3432 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3433 Perl_sv_catpvf(aTHX_ listsv,
3434 "%04"UVxf"\n%04"UVxf"\n", (UV)prevvalue, (UV) '-');
3438 range = 0; /* this was not a true range */
3442 if (namedclass > OOB_NAMEDCLASS)
3443 optimize_invert = FALSE;
3444 /* Possible truncation here but in some 64-bit environments
3445 * the compiler gets heartburn about switch on 64-bit values.
3446 * A similar issue a little earlier when switching on value.
3448 switch ((I32)namedclass) {
3451 ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
3453 for (value = 0; value < 256; value++)
3455 ANYOF_BITMAP_SET(ret, value);
3457 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n");
3461 ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
3463 for (value = 0; value < 256; value++)
3464 if (!isALNUM(value))
3465 ANYOF_BITMAP_SET(ret, value);
3467 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n");
3471 ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
3473 for (value = 0; value < 256; value++)
3474 if (isALNUMC(value))
3475 ANYOF_BITMAP_SET(ret, value);
3477 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n");
3481 ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
3483 for (value = 0; value < 256; value++)
3484 if (!isALNUMC(value))
3485 ANYOF_BITMAP_SET(ret, value);
3487 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n");
3491 ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
3493 for (value = 0; value < 256; value++)
3495 ANYOF_BITMAP_SET(ret, value);
3497 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n");
3501 ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
3503 for (value = 0; value < 256; value++)
3504 if (!isALPHA(value))
3505 ANYOF_BITMAP_SET(ret, value);
3507 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n");
3511 ANYOF_CLASS_SET(ret, ANYOF_ASCII);
3514 for (value = 0; value < 128; value++)
3515 ANYOF_BITMAP_SET(ret, value);
3517 for (value = 0; value < 256; value++) {
3518 if (PL_hints & HINT_RE_ASCIIR) {
3519 if (NATIVE_TO_ASCII(value) < 128)
3520 ANYOF_BITMAP_SET(ret, value);
3524 ANYOF_BITMAP_SET(ret, value);
3529 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n");
3533 ANYOF_CLASS_SET(ret, ANYOF_NASCII);
3536 for (value = 128; value < 256; value++)
3537 ANYOF_BITMAP_SET(ret, value);
3539 for (value = 0; value < 256; value++) {
3540 if (PL_hints & HINT_RE_ASCIIR) {
3541 if (NATIVE_TO_ASCII(value) >= 128)
3542 ANYOF_BITMAP_SET(ret, value);
3545 if (!isASCII(value))
3546 ANYOF_BITMAP_SET(ret, value);
3551 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n");
3555 ANYOF_CLASS_SET(ret, ANYOF_BLANK);
3557 for (value = 0; value < 256; value++)
3559 ANYOF_BITMAP_SET(ret, value);
3561 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsBlank\n");
3565 ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
3567 for (value = 0; value < 256; value++)
3568 if (!isBLANK(value))
3569 ANYOF_BITMAP_SET(ret, value);
3571 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsBlank\n");
3575 ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
3577 for (value = 0; value < 256; value++)
3579 ANYOF_BITMAP_SET(ret, value);
3581 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n");
3585 ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
3587 for (value = 0; value < 256; value++)
3588 if (!isCNTRL(value))
3589 ANYOF_BITMAP_SET(ret, value);
3591 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n");
3595 ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
3597 /* consecutive digits assumed */
3598 for (value = '0'; value <= '9'; value++)
3599 ANYOF_BITMAP_SET(ret, value);
3601 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n");
3605 ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
3607 /* consecutive digits assumed */
3608 for (value = 0; value < '0'; value++)
3609 ANYOF_BITMAP_SET(ret, value);
3610 for (value = '9' + 1; value < 256; value++)
3611 ANYOF_BITMAP_SET(ret, value);
3613 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n");
3617 ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
3619 for (value = 0; value < 256; value++)
3621 ANYOF_BITMAP_SET(ret, value);
3623 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n");
3627 ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
3629 for (value = 0; value < 256; value++)
3630 if (!isGRAPH(value))
3631 ANYOF_BITMAP_SET(ret, value);
3633 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n");
3637 ANYOF_CLASS_SET(ret, ANYOF_LOWER);
3639 for (value = 0; value < 256; value++)
3641 ANYOF_BITMAP_SET(ret, value);
3643 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n");
3647 ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
3649 for (value = 0; value < 256; value++)
3650 if (!isLOWER(value))
3651 ANYOF_BITMAP_SET(ret, value);
3653 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n");
3657 ANYOF_CLASS_SET(ret, ANYOF_PRINT);
3659 for (value = 0; value < 256; value++)
3661 ANYOF_BITMAP_SET(ret, value);
3663 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n");
3667 ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
3669 for (value = 0; value < 256; value++)
3670 if (!isPRINT(value))
3671 ANYOF_BITMAP_SET(ret, value);
3673 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n");
3677 ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
3679 for (value = 0; value < 256; value++)
3680 if (isPSXSPC(value))
3681 ANYOF_BITMAP_SET(ret, value);
3683 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n");
3687 ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
3689 for (value = 0; value < 256; value++)
3690 if (!isPSXSPC(value))
3691 ANYOF_BITMAP_SET(ret, value);
3693 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n");
3697 ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
3699 for (value = 0; value < 256; value++)
3701 ANYOF_BITMAP_SET(ret, value);
3703 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n");
3707 ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
3709 for (value = 0; value < 256; value++)
3710 if (!isPUNCT(value))
3711 ANYOF_BITMAP_SET(ret, value);
3713 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n");
3717 ANYOF_CLASS_SET(ret, ANYOF_SPACE);
3719 for (value = 0; value < 256; value++)
3721 ANYOF_BITMAP_SET(ret, value);
3723 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpacePerl\n");
3727 ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
3729 for (value = 0; value < 256; value++)
3730 if (!isSPACE(value))
3731 ANYOF_BITMAP_SET(ret, value);
3733 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpacePerl\n");
3737 ANYOF_CLASS_SET(ret, ANYOF_UPPER);
3739 for (value = 0; value < 256; value++)
3741 ANYOF_BITMAP_SET(ret, value);
3743 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n");
3747 ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
3749 for (value = 0; value < 256; value++)
3750 if (!isUPPER(value))
3751 ANYOF_BITMAP_SET(ret, value);
3753 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n");
3757 ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
3759 for (value = 0; value < 256; value++)
3760 if (isXDIGIT(value))
3761 ANYOF_BITMAP_SET(ret, value);
3763 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n");
3767 ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
3769 for (value = 0; value < 256; value++)
3770 if (!isXDIGIT(value))
3771 ANYOF_BITMAP_SET(ret, value);
3773 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n");
3776 vFAIL("Invalid [::] class");
3780 ANYOF_FLAGS(ret) |= ANYOF_CLASS;
3783 } /* end of namedclass \blah */
3786 if (((prevvalue > value) && !(PL_hints & HINT_RE_ASCIIR)) ||
3787 ((NATIVE_TO_UNI(prevvalue) > NATIVE_TO_UNI(value)) &&
3788 (PL_hints & HINT_RE_ASCIIR))) /* b-a */ {
3789 Simple_vFAIL4("Invalid [] range \"%*.*s\"",
3790 RExC_parse - rangebegin,
3791 RExC_parse - rangebegin,
3793 range = 0; /* not a valid range */
3797 prevvalue = value; /* save the beginning of the range */
3798 if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
3799 RExC_parse[1] != ']') {
3802 /* a bad range like \w-, [:word:]- ? */
3803 if (namedclass > OOB_NAMEDCLASS) {
3804 if (ckWARN(WARN_REGEXP))
3806 "False [] range \"%*.*s\"",
3807 RExC_parse - rangebegin,
3808 RExC_parse - rangebegin,
3811 ANYOF_BITMAP_SET(ret, '-');
3813 range = 1; /* yeah, it's a range! */
3814 continue; /* but do it the next time */
3818 /* now is the next time */
3822 if (prevvalue < 256) {
3823 IV ceilvalue = value < 256 ? value : 255;
3826 /* New style scheme for ranges:
3828 * do ranges in ASCII/Unicode space
3830 for (i = NATIVE_TO_ASCII(prevvalue);
3831 i <= NATIVE_TO_ASCII(ceilvalue);
3833 ANYOF_BITMAP_SET(ret, ASCII_TO_NATIVE(i));
3835 else if ((isLOWER(prevvalue) && isLOWER(ceilvalue)) ||
3836 (isUPPER(prevvalue) && isUPPER(ceilvalue)))
3838 if (isLOWER(prevvalue)) {
3839 for (i = prevvalue; i <= ceilvalue; i++)
3841 ANYOF_BITMAP_SET(ret, i);
3843 for (i = prevvalue; i <= ceilvalue; i++)
3845 ANYOF_BITMAP_SET(ret, i);
3850 for (i = prevvalue; i <= ceilvalue; i++)
3851 ANYOF_BITMAP_SET(ret, i);
3854 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3855 if (prevvalue < value)
3856 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
3857 (UV)prevvalue, (UV)value);
3858 else if (prevvalue == value)
3859 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
3864 range = 0; /* this range (if it was one) is done now */
3868 ANYOF_FLAGS(ret) |= ANYOF_LARGE;
3870 RExC_size += ANYOF_CLASS_ADD_SKIP;
3872 RExC_emit += ANYOF_CLASS_ADD_SKIP;
3875 /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
3878 /* If the only flag is folding (plus possibly inversion). */
3879 (ANYOF_FLAGS_ALL ^ ANYOF_INVERT) == ANYOF_FOLD)) {
3880 for (value = 0; value < 256; ++value) {
3881 if (ANYOF_BITMAP_TEST(ret, value)) {
3882 IV fold = PL_fold[value];
3885 ANYOF_BITMAP_SET(ret, fold);
3888 ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
3891 /* optimize inverted simple patterns (e.g. [^a-z]) */
3892 if (!SIZE_ONLY && optimize_invert &&
3893 /* If the only flag is inversion. */
3894 (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
3895 for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
3896 ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
3897 ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
3904 av_store(av, 0, listsv);
3905 av_store(av, 1, NULL);
3906 rv = newRV_noinc((SV*)av);
3907 n = add_data(pRExC_state, 1, "s");
3908 RExC_rx->data->data[n] = (void*)rv;
3916 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
3918 char* retval = RExC_parse++;
3921 if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
3922 RExC_parse[2] == '#') {
3923 while (*RExC_parse && *RExC_parse != ')')
3928 if (RExC_flags16 & PMf_EXTENDED) {
3929 if (isSPACE(*RExC_parse)) {
3933 else if (*RExC_parse == '#') {
3934 while (*RExC_parse && *RExC_parse != '\n')
3945 - reg_node - emit a node
3947 STATIC regnode * /* Location. */
3948 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
3950 register regnode *ret;
3951 register regnode *ptr;
3955 SIZE_ALIGN(RExC_size);
3960 NODE_ALIGN_FILL(ret);
3962 FILL_ADVANCE_NODE(ptr, op);
3963 if (RExC_offsets) { /* MJD */
3964 MJD_OFFSET_DEBUG((stderr, "%s:%u: (op %s) %s %u <- %u (len %u) (max %u).\n",
3965 "reg_node", __LINE__,
3967 RExC_emit - RExC_emit_start > RExC_offsets[0]
3968 ? "Overwriting end of array!\n" : "OK",
3969 RExC_emit - RExC_emit_start,
3970 RExC_parse - RExC_start,
3972 Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
3981 - reganode - emit a node with an argument
3983 STATIC regnode * /* Location. */
3984 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
3986 register regnode *ret;
3987 register regnode *ptr;
3991 SIZE_ALIGN(RExC_size);
3996 NODE_ALIGN_FILL(ret);
3998 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
3999 if (RExC_offsets) { /* MJD */
4000 MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n",
4002 RExC_emit - RExC_emit_start > RExC_offsets[0] ?
4003 "Overwriting end of array!\n" : "OK",
4004 RExC_emit - RExC_emit_start,
4005 RExC_parse - RExC_start,
4007 Set_Cur_Node_Offset;
4016 - reguni - emit (if appropriate) a Unicode character
4019 S_reguni(pTHX_ RExC_state_t *pRExC_state, UV uv, char* s, STRLEN* lenp)
4021 *lenp = SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
4025 - reginsert - insert an operator in front of already-emitted operand
4027 * Means relocating the operand.
4030 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd)
4032 register regnode *src;
4033 register regnode *dst;
4034 register regnode *place;
4035 register int offset = regarglen[(U8)op];
4037 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
4040 RExC_size += NODE_STEP_REGNODE + offset;
4045 RExC_emit += NODE_STEP_REGNODE + offset;
4047 while (src > opnd) {
4048 StructCopy(--src, --dst, regnode);
4049 if (RExC_offsets) { /* MJD 20010112 */
4050 MJD_OFFSET_DEBUG((stderr, "%s: %s copy %u -> %u (max %u).\n",
4052 dst - RExC_emit_start > RExC_offsets[0]
4053 ? "Overwriting end of array!\n" : "OK",
4054 src - RExC_emit_start,
4055 dst - RExC_emit_start,
4057 Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
4058 Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
4063 place = opnd; /* Op node, where operand used to be. */
4064 if (RExC_offsets) { /* MJD */
4065 MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n",
4067 place - RExC_emit_start > RExC_offsets[0]
4068 ? "Overwriting end of array!\n" : "OK",
4069 place - RExC_emit_start,
4070 RExC_parse - RExC_start,
4072 Set_Node_Offset(place, RExC_parse);
4074 src = NEXTOPER(place);
4075 FILL_ADVANCE_NODE(place, op);
4076 Zero(src, offset, regnode);
4080 - regtail - set the next-pointer at the end of a node chain of p to val.
4083 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4085 register regnode *scan;
4086 register regnode *temp;
4091 /* Find last node. */
4094 temp = regnext(scan);
4100 if (reg_off_by_arg[OP(scan)]) {
4101 ARG_SET(scan, val - scan);
4104 NEXT_OFF(scan) = val - scan;
4109 - regoptail - regtail on operand of first argument; nop if operandless
4112 S_regoptail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4114 /* "Operandless" and "op != BRANCH" are synonymous in practice. */
4115 if (p == NULL || SIZE_ONLY)
4117 if (PL_regkind[(U8)OP(p)] == BRANCH) {
4118 regtail(pRExC_state, NEXTOPER(p), val);
4120 else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
4121 regtail(pRExC_state, NEXTOPER(NEXTOPER(p)), val);
4128 - regcurly - a little FSA that accepts {\d+,?\d*}
4131 S_regcurly(pTHX_ register char *s)
4150 S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
4153 register U8 op = EXACT; /* Arbitrary non-END op. */
4154 register regnode *next;
4156 while (op != END && (!last || node < last)) {
4157 /* While that wasn't END last time... */
4163 next = regnext(node);
4165 if (OP(node) == OPTIMIZED)
4168 PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
4169 (int)(2*l + 1), "", SvPVX(sv));
4170 if (next == NULL) /* Next ptr. */
4171 PerlIO_printf(Perl_debug_log, "(0)");
4173 PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
4174 (void)PerlIO_putc(Perl_debug_log, '\n');
4176 if (PL_regkind[(U8)op] == BRANCHJ) {
4177 register regnode *nnode = (OP(next) == LONGJMP
4180 if (last && nnode > last)
4182 node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
4184 else if (PL_regkind[(U8)op] == BRANCH) {
4185 node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
4187 else if ( op == CURLY) { /* `next' might be very big: optimizer */
4188 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4189 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
4191 else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
4192 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4195 else if ( op == PLUS || op == STAR) {
4196 node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
4198 else if (op == ANYOF) {
4199 /* arglen 1 + class block */
4200 node += 1 + ((ANYOF_FLAGS(node) & ANYOF_LARGE)
4201 ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
4202 node = NEXTOPER(node);
4204 else if (PL_regkind[(U8)op] == EXACT) {
4205 /* Literal string, where present. */
4206 node += NODE_SZ_STR(node) - 1;
4207 node = NEXTOPER(node);
4210 node = NEXTOPER(node);
4211 node += regarglen[(U8)op];
4213 if (op == CURLYX || op == OPEN)
4215 else if (op == WHILEM)
4218 #endif /* DEBUGGING */
4223 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
4226 Perl_regdump(pTHX_ regexp *r)
4229 SV *sv = sv_newmortal();
4231 (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
4233 /* Header fields of interest. */
4234 if (r->anchored_substr)
4235 PerlIO_printf(Perl_debug_log,
4236 "anchored `%s%.*s%s'%s at %"IVdf" ",
4238 (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)),
4239 SvPVX(r->anchored_substr),
4241 SvTAIL(r->anchored_substr) ? "$" : "",
4242 (IV)r->anchored_offset);
4243 if (r->float_substr)
4244 PerlIO_printf(Perl_debug_log,
4245 "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ",
4247 (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)),
4248 SvPVX(r->float_substr),
4250 SvTAIL(r->float_substr) ? "$" : "",
4251 (IV)r->float_min_offset, (UV)r->float_max_offset);
4252 if (r->check_substr)
4253 PerlIO_printf(Perl_debug_log,
4254 r->check_substr == r->float_substr
4255 ? "(checking floating" : "(checking anchored");
4256 if (r->reganch & ROPT_NOSCAN)
4257 PerlIO_printf(Perl_debug_log, " noscan");
4258 if (r->reganch & ROPT_CHECK_ALL)
4259 PerlIO_printf(Perl_debug_log, " isall");
4260 if (r->check_substr)
4261 PerlIO_printf(Perl_debug_log, ") ");
4263 if (r->regstclass) {
4264 regprop(sv, r->regstclass);
4265 PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
4267 if (r->reganch & ROPT_ANCH) {
4268 PerlIO_printf(Perl_debug_log, "anchored");
4269 if (r->reganch & ROPT_ANCH_BOL)
4270 PerlIO_printf(Perl_debug_log, "(BOL)");
4271 if (r->reganch & ROPT_ANCH_MBOL)
4272 PerlIO_printf(Perl_debug_log, "(MBOL)");
4273 if (r->reganch & ROPT_ANCH_SBOL)
4274 PerlIO_printf(Perl_debug_log, "(SBOL)");
4275 if (r->reganch & ROPT_ANCH_GPOS)
4276 PerlIO_printf(Perl_debug_log, "(GPOS)");
4277 PerlIO_putc(Perl_debug_log, ' ');
4279 if (r->reganch & ROPT_GPOS_SEEN)
4280 PerlIO_printf(Perl_debug_log, "GPOS ");
4281 if (r->reganch & ROPT_SKIP)
4282 PerlIO_printf(Perl_debug_log, "plus ");
4283 if (r->reganch & ROPT_IMPLICIT)
4284 PerlIO_printf(Perl_debug_log, "implicit ");
4285 PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
4286 if (r->reganch & ROPT_EVAL_SEEN)
4287 PerlIO_printf(Perl_debug_log, "with eval ");
4288 PerlIO_printf(Perl_debug_log, "\n");
4291 U32 len = r->offsets[0];
4292 PerlIO_printf(Perl_debug_log, "Offsets: [%u]\n\t", r->offsets[0]);
4293 for (i = 1; i <= len; i++)
4294 PerlIO_printf(Perl_debug_log, "%u[%u] ",
4297 PerlIO_printf(Perl_debug_log, "\n");
4299 #endif /* DEBUGGING */
4303 S_put_byte(pTHX_ SV *sv, int c)
4305 if (isCNTRL(c) || c == 127 || c == 255 || !isPRINT(c))
4306 Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
4307 else if (c == '-' || c == ']' || c == '\\' || c == '^')
4308 Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
4310 Perl_sv_catpvf(aTHX_ sv, "%c", c);
4314 - regprop - printable representation of opcode
4317 Perl_regprop(pTHX_ SV *sv, regnode *o)
4322 sv_setpvn(sv, "", 0);
4323 if (OP(o) >= reg_num) /* regnode.type is unsigned */
4324 /* It would be nice to FAIL() here, but this may be called from
4325 regexec.c, and it would be hard to supply pRExC_state. */
4326 Perl_croak(aTHX_ "Corrupted regexp opcode");
4327 sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
4329 k = PL_regkind[(U8)OP(o)];
4332 Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0],
4333 STR_LEN(o), STRING(o), PL_colors[1]);
4334 else if (k == CURLY) {
4335 if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
4336 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
4337 Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
4339 else if (k == WHILEM && o->flags) /* Ordinal/of */
4340 Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
4341 else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
4342 Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o)); /* Parenth number */
4343 else if (k == LOGICAL)
4344 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* 2: embedded, otherwise 1 */
4345 else if (k == ANYOF) {
4346 int i, rangestart = -1;
4347 U8 flags = ANYOF_FLAGS(o);
4348 const char * const anyofs[] = { /* Should be syncronized with
4349 * ANYOF_ #xdefines in regcomp.h */
4382 if (flags & ANYOF_LOCALE)
4383 sv_catpv(sv, "{loc}");
4384 if (flags & ANYOF_FOLD)
4385 sv_catpv(sv, "{i}");
4386 Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
4387 if (flags & ANYOF_INVERT)
4389 for (i = 0; i <= 256; i++) {
4390 if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
4391 if (rangestart == -1)
4393 } else if (rangestart != -1) {
4394 if (i <= rangestart + 3)
4395 for (; rangestart < i; rangestart++)
4396 put_byte(sv, rangestart);
4398 put_byte(sv, rangestart);
4400 put_byte(sv, i - 1);
4406 if (o->flags & ANYOF_CLASS)
4407 for (i = 0; i < sizeof(anyofs)/sizeof(char*); i++)
4408 if (ANYOF_CLASS_TEST(o,i))
4409 sv_catpv(sv, anyofs[i]);
4411 if (flags & ANYOF_UNICODE)
4412 sv_catpv(sv, "{unicode}");
4413 else if (flags & ANYOF_UNICODE_ALL)
4414 sv_catpv(sv, "{all-unicode}");
4418 SV *sw = regclass_swash(o, FALSE, &lv);
4423 U8 s[UTF8_MAXLEN+1];
4425 for (i = 0; i <= 256; i++) { /* just the first 256 */
4426 U8 *e = uvchr_to_utf8(s, i);
4428 if (i < 256 && swash_fetch(sw, s, TRUE)) {
4429 if (rangestart == -1)
4431 } else if (rangestart != -1) {
4434 if (i <= rangestart + 3)
4435 for (; rangestart < i; rangestart++) {
4436 for(e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4440 for (e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4443 for (e = uvchr_to_utf8(s, i - 1), p = s; p < e; p++)
4450 sv_catpv(sv, "..."); /* et cetera */
4454 char *s = savepv(SvPVX(lv));
4457 while(*s && *s != '\n') s++;
4478 Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
4480 else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
4481 Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
4482 #endif /* DEBUGGING */
4486 Perl_re_intuit_string(pTHX_ regexp *prog)
4487 { /* Assume that RE_INTUIT is set */
4490 char *s = SvPV(prog->check_substr,n_a);
4492 if (!PL_colorset) reginitcolors();
4493 PerlIO_printf(Perl_debug_log,
4494 "%sUsing REx substr:%s `%s%.60s%s%s'\n",
4495 PL_colors[4],PL_colors[5],PL_colors[0],
4498 (strlen(s) > 60 ? "..." : ""));
4501 return prog->check_substr;
4505 Perl_pregfree(pTHX_ struct regexp *r)
4507 DEBUG_r(if (!PL_colorset) reginitcolors());
4509 if (!r || (--r->refcnt > 0))
4511 DEBUG_r(PerlIO_printf(Perl_debug_log,
4512 "%sFreeing REx:%s `%s%.60s%s%s'\n",
4513 PL_colors[4],PL_colors[5],PL_colors[0],
4516 (strlen(r->precomp) > 60 ? "..." : "")));
4519 Safefree(r->precomp);
4520 if (r->offsets) /* 20010421 MJD */
4521 Safefree(r->offsets);
4522 if (RX_MATCH_COPIED(r))
4523 Safefree(r->subbeg);
4525 if (r->anchored_substr)
4526 SvREFCNT_dec(r->anchored_substr);
4527 if (r->float_substr)
4528 SvREFCNT_dec(r->float_substr);
4529 Safefree(r->substrs);
4532 int n = r->data->count;
4533 AV* new_comppad = NULL;
4538 switch (r->data->what[n]) {
4540 SvREFCNT_dec((SV*)r->data->data[n]);
4543 Safefree(r->data->data[n]);
4546 new_comppad = (AV*)r->data->data[n];
4549 if (new_comppad == NULL)
4550 Perl_croak(aTHX_ "panic: pregfree comppad");
4551 old_comppad = PL_comppad;
4552 old_curpad = PL_curpad;
4553 /* Watch out for global destruction's random ordering. */
4554 if (SvTYPE(new_comppad) == SVt_PVAV) {
4555 PL_comppad = new_comppad;
4556 PL_curpad = AvARRAY(new_comppad);
4560 op_free((OP_4tree*)r->data->data[n]);
4561 PL_comppad = old_comppad;
4562 PL_curpad = old_curpad;
4563 SvREFCNT_dec((SV*)new_comppad);
4569 Perl_croak(aTHX_ "panic: regfree data code '%c'", r->data->what[n]);
4572 Safefree(r->data->what);
4575 Safefree(r->startp);
4581 - regnext - dig the "next" pointer out of a node
4583 * [Note, when REGALIGN is defined there are two places in regmatch()
4584 * that bypass this code for speed.]
4587 Perl_regnext(pTHX_ register regnode *p)
4589 register I32 offset;
4591 if (p == &PL_regdummy)
4594 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
4602 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
4605 STRLEN l1 = strlen(pat1);
4606 STRLEN l2 = strlen(pat2);
4615 Copy(pat1, buf, l1 , char);
4616 Copy(pat2, buf + l1, l2 , char);
4617 buf[l1 + l2] = '\n';
4618 buf[l1 + l2 + 1] = '\0';
4620 /* ANSI variant takes additional second argument */
4621 va_start(args, pat2);
4625 msv = vmess(buf, &args);
4627 message = SvPV(msv,l1);
4630 Copy(message, buf, l1 , char);
4631 buf[l1] = '\0'; /* Overwrite \n */
4632 Perl_croak(aTHX_ "%s", buf);
4635 /* XXX Here's a total kludge. But we need to re-enter for swash routines. */
4638 Perl_save_re_context(pTHX)
4641 SAVEPPTR(RExC_precomp); /* uncompiled string. */
4642 SAVEI32(RExC_npar); /* () count. */
4643 SAVEI32(RExC_size); /* Code size. */
4644 SAVEI16(RExC_flags16); /* are we folding, multilining? */
4645 SAVEVPTR(RExC_rx); /* from regcomp.c */
4646 SAVEI32(RExC_seen); /* from regcomp.c */
4647 SAVEI32(RExC_sawback); /* Did we see \1, ...? */
4648 SAVEI32(RExC_naughty); /* How bad is this pattern? */
4649 SAVEVPTR(RExC_emit); /* Code-emit pointer; ®dummy = don't */
4650 SAVEPPTR(RExC_end); /* End of input for compile */
4651 SAVEPPTR(RExC_parse); /* Input-scan pointer. */
4654 SAVEI32(PL_reg_flags); /* from regexec.c */
4656 SAVEPPTR(PL_reginput); /* String-input pointer. */
4657 SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */
4658 SAVEPPTR(PL_regeol); /* End of input, for $ check. */
4659 SAVEVPTR(PL_regstartp); /* Pointer to startp array. */
4660 SAVEVPTR(PL_regendp); /* Ditto for endp. */
4661 SAVEVPTR(PL_reglastparen); /* Similarly for lastparen. */
4662 SAVEPPTR(PL_regtill); /* How far we are required to go. */
4663 SAVEGENERICPV(PL_reg_start_tmp); /* from regexec.c */
4664 PL_reg_start_tmp = 0;
4665 SAVEI32(PL_reg_start_tmpl); /* from regexec.c */
4666 PL_reg_start_tmpl = 0;
4667 SAVEVPTR(PL_regdata);
4668 SAVEI32(PL_reg_eval_set); /* from regexec.c */
4669 SAVEI32(PL_regnarrate); /* from regexec.c */
4670 SAVEVPTR(PL_regprogram); /* from regexec.c */
4671 SAVEINT(PL_regindent); /* from regexec.c */
4672 SAVEVPTR(PL_regcc); /* from regexec.c */
4673 SAVEVPTR(PL_curcop);
4674 SAVEVPTR(PL_reg_call_cc); /* from regexec.c */
4675 SAVEVPTR(PL_reg_re); /* from regexec.c */
4676 SAVEPPTR(PL_reg_ganch); /* from regexec.c */
4677 SAVESPTR(PL_reg_sv); /* from regexec.c */
4678 SAVEVPTR(PL_reg_magic); /* from regexec.c */
4679 SAVEI32(PL_reg_oldpos); /* from regexec.c */
4680 SAVEVPTR(PL_reg_oldcurpm); /* from regexec.c */
4681 SAVEVPTR(PL_reg_curpm); /* from regexec.c */
4682 SAVEI32(PL_regnpar); /* () count. */
4684 SAVEPPTR(PL_reg_starttry); /* from regexec.c */
4695 clear_re(pTHXo_ void *r)
4697 ReREFCNT_dec((regexp *)r);
projects / p5sagit/p5-mst-13.2.git / blob