projects / p5sagit/p5-mst-13.2.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | history | HEAD
Allow option to treat barewords at start of Unix paths as simple words rather than...
[p5sagit/p5-mst-13.2.git] / regcomp.c
CommitLineData
a0d0e21e 1/* regcomp.c
2 */
3
4/*
5 * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee
6 */
7
a687059c 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!
10 */
11
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.
15 */
16
e50aee73 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.
20*/
21
b9d5759e 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
26# endif
27/* need access to debugger hooks */
cad2e5aa 28# if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
b9d5759e 29# define DEBUGGING
30# endif
31#endif
32
33#ifdef PERL_IN_XSUB_RE
d06ea78c 34/* We *really* need to overwrite these symbols: */
56953603 35# define Perl_pregcomp my_regcomp
36# define Perl_regdump my_regdump
37# define Perl_regprop my_regprop
d06ea78c 38# define Perl_pregfree my_regfree
cad2e5aa 39# define Perl_re_intuit_string my_re_intuit_string
40/* *These* symbols are masked to allow static link. */
d06ea78c 41# define Perl_regnext my_regnext
f0b8d043 42# define Perl_save_re_context my_save_re_context
d88dccdf 43# define Perl_reginitcolors my_reginitcolors
c5be433b 44
45# define PERL_NO_GET_CONTEXT
56953603 46#endif
47
f0fcb552 48/*SUPPRESS 112*/
a687059c 49/*
e50aee73 50 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c 51 *
52 * Copyright (c) 1986 by University of Toronto.
53 * Written by Henry Spencer. Not derived from licensed software.
54 *
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:
58 *
59 * 1. The author is not responsible for the consequences of use of
60 * this software, no matter how awful, even if they arise
61 * from defects in it.
62 *
63 * 2. The origin of this software must not be misrepresented, either
64 * by explicit claim or by omission.
65 *
66 * 3. Altered versions must be plainly marked as such, and must not
67 * be misrepresented as being the original software.
68 *
69 *
70 **** Alterations to Henry's code are...
71 ****
4eb8286e 72 **** Copyright (c) 1991-1999, Larry Wall
a687059c 73 ****
9ef589d8 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.
76
a687059c 77 *
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.
81 */
82#include "EXTERN.h"
864dbfa3 83#define PERL_IN_REGCOMP_C
a687059c 84#include "perl.h"
d06ea78c 85
c5be433b 86#ifdef PERL_IN_XSUB_RE
87# if defined(PERL_CAPI) || defined(PERL_OBJECT)
88# include "XSUB.h"
89# endif
90#else
d06ea78c 91# include "INTERN.h"
92#endif
c277df42 93
94#define REG_COMP_C
a687059c 95#include "regcomp.h"
96
d4cce5f1 97#ifdef op
11343788 98#undef op
d4cce5f1 99#endif /* op */
11343788 100
fe14fcc3 101#ifdef MSDOS
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 */
108#endif /* MSDOS */
109
a687059c 110#ifndef STATIC
111#define STATIC static
112#endif
113
114#define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
115#define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
116 ((*s) == '{' && regcurly(s)))
2b69d0c2 117#ifdef atarist
118#define PERL_META "^$.[()|?+*\\"
119#else
a687059c 120#define META "^$.[()|?+*\\"
2b69d0c2 121#endif
a687059c 122
35c8bce7 123#ifdef SPSTART
124#undef SPSTART /* dratted cpp namespace... */
125#endif
a687059c 126/*
127 * Flags to be passed up and down.
128 */
a687059c 129#define WORST 0 /* Worst case. */
821b33a5 130#define HASWIDTH 0x1 /* Known to match non-null strings. */
a0d0e21e 131#define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
132#define SPSTART 0x4 /* Starts with * or +. */
133#define TRYAGAIN 0x8 /* Weeded out a declaration. */
a687059c 134
82ba1be6 135/* Length of a variant. */
136
137typedef struct scan_data_t {
138 I32 len_min;
139 I32 len_delta;
140 I32 pos_min;
141 I32 pos_delta;
142 SV *last_found;
143 I32 last_end; /* min value, <0 unless valid. */
144 I32 last_start_min;
145 I32 last_start_max;
146 SV **longest; /* Either &l_fixed, or &l_float. */
147 SV *longest_fixed;
148 I32 offset_fixed;
149 SV *longest_float;
150 I32 offset_float_min;
151 I32 offset_float_max;
152 I32 flags;
153 I32 whilem_c;
154} scan_data_t;
155
a687059c 156/*
e50aee73 157 * Forward declarations for pregcomp()'s friends.
a687059c 158 */
a0d0e21e 159
cd488c12 160static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
82ba1be6 161 0, 0, 0, 0 };
c277df42 162
163#define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
164#define SF_BEFORE_SEOL 0x1
165#define SF_BEFORE_MEOL 0x2
166#define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
167#define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
168
09b7f37c 169#ifdef NO_UNARY_PLUS
170# define SF_FIX_SHIFT_EOL (0+2)
171# define SF_FL_SHIFT_EOL (0+4)
172#else
173# define SF_FIX_SHIFT_EOL (+2)
174# define SF_FL_SHIFT_EOL (+4)
175#endif
c277df42 176
177#define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
178#define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
179
180#define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
181#define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
182#define SF_IS_INF 0x40
183#define SF_HAS_PAR 0x80
184#define SF_IN_PAR 0x100
185#define SF_HAS_EVAL 0x200
4bfe0158 186#define SCF_DO_SUBSTR 0x400
c277df42 187
a0ed51b3 188#define RF_utf8 8
189#define UTF (PL_reg_flags & RF_utf8)
190#define LOC (PL_regflags & PMf_LOCALE)
191#define FOLD (PL_regflags & PMf_FOLD)
192
b8c5462f 193#define OOB_CHAR8 1234
194#define OOB_UTF8 123456
195
a0ed51b3 196#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
197#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
198
51371543 199static void clear_re(pTHXo_ void *r);
4327152a 200
201STATIC void
cea2e8a9 202S_scan_commit(pTHX_ scan_data_t *data)
c277df42 203{
c485e607 204 dTHR;
a0ed51b3 205 STRLEN l = CHR_SVLEN(data->last_found);
206 STRLEN old_l = CHR_SVLEN(*data->longest);
c277df42 207
208 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
209 sv_setsv(*data->longest, data->last_found);
210 if (*data->longest == data->longest_fixed) {
211 data->offset_fixed = l ? data->last_start_min : data->pos_min;
212 if (data->flags & SF_BEFORE_EOL)
213 data->flags
214 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
215 else
216 data->flags &= ~SF_FIX_BEFORE_EOL;
a0ed51b3 217 }
218 else {
c277df42 219 data->offset_float_min = l ? data->last_start_min : data->pos_min;
220 data->offset_float_max = (l
221 ? data->last_start_max
222 : data->pos_min + data->pos_delta);
223 if (data->flags & SF_BEFORE_EOL)
224 data->flags
225 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
226 else
227 data->flags &= ~SF_FL_BEFORE_EOL;
228 }
229 }
230 SvCUR_set(data->last_found, 0);
231 data->last_end = -1;
232 data->flags &= ~SF_BEFORE_EOL;
233}
234
c277df42 235/* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
236 to the position after last scanned or to NULL. */
237
76e3520e 238STATIC I32
cea2e8a9 239S_study_chunk(pTHX_ regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
c277df42 240 /* scanp: Start here (read-write). */
241 /* deltap: Write maxlen-minlen here. */
242 /* last: Stop before this one. */
243{
5c0ca799 244 dTHR;
c277df42 245 I32 min = 0, pars = 0, code;
246 regnode *scan = *scanp, *next;
247 I32 delta = 0;
248 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
aca2d497 249 int is_inf_internal = 0; /* The studied chunk is infinite */
c277df42 250 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
251 scan_data_t data_fake;
252
253 while (scan && OP(scan) != END && scan < last) {
254 /* Peephole optimizer: */
255
22c35a8c 256 if (PL_regkind[(U8)OP(scan)] == EXACT) {
c277df42 257 regnode *n = regnext(scan);
258 U32 stringok = 1;
259#ifdef DEBUGGING
260 regnode *stop = scan;
261#endif
262
263 next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
264 /* Skip NOTHING, merge EXACT*. */
265 while (n &&
22c35a8c 266 ( PL_regkind[(U8)OP(n)] == NOTHING ||
c277df42 267 (stringok && (OP(n) == OP(scan))))
268 && NEXT_OFF(n)
269 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
270 if (OP(n) == TAIL || n > next)
271 stringok = 0;
22c35a8c 272 if (PL_regkind[(U8)OP(n)] == NOTHING) {
c277df42 273 NEXT_OFF(scan) += NEXT_OFF(n);
274 next = n + NODE_STEP_REGNODE;
275#ifdef DEBUGGING
276 if (stringok)
277 stop = n;
278#endif
279 n = regnext(n);
a0ed51b3 280 }
281 else {
c277df42 282 int oldl = *OPERAND(scan);
283 regnode *nnext = regnext(n);
284
285 if (oldl + *OPERAND(n) > U8_MAX)
286 break;
287 NEXT_OFF(scan) += NEXT_OFF(n);
288 *OPERAND(scan) += *OPERAND(n);
289 next = n + (*OPERAND(n) + 2 - 1)/sizeof(regnode) + 2;
290 /* Now we can overwrite *n : */
291 Move(OPERAND(n) + 1, OPERAND(scan) + oldl + 1,
292 *OPERAND(n) + 1, char);
293#ifdef DEBUGGING
294 if (stringok)
295 stop = next - 1;
296#endif
297 n = nnext;
298 }
299 }
300#ifdef DEBUGGING
301 /* Allow dumping */
302 n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
303 while (n <= stop) {
ca04da08 304 /* Purify reports a benign UMR here sometimes, because we
305 * don't initialize the OP() slot of a node when that node
306 * is occupied by just the trailing null of the string in
307 * an EXACT node */
22c35a8c 308 if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
c277df42 309 OP(n) = OPTIMIZED;
310 NEXT_OFF(n) = 0;
311 }
312 n++;
313 }
314#endif
315
316 }
317 if (OP(scan) != CURLYX) {
048cfca1 318 int max = (reg_off_by_arg[OP(scan)]
319 ? I32_MAX
320 /* I32 may be smaller than U16 on CRAYs! */
321 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
c277df42 322 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
323 int noff;
324 regnode *n = scan;
325
326 /* Skip NOTHING and LONGJMP. */
327 while ((n = regnext(n))
22c35a8c 328 && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
c277df42 329 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
330 && off + noff < max)
331 off += noff;
332 if (reg_off_by_arg[OP(scan)])
333 ARG(scan) = off;
334 else
335 NEXT_OFF(scan) = off;
336 }
337 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
338 || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
339 next = regnext(scan);
340 code = OP(scan);
341
342 if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
343 I32 max1 = 0, min1 = I32_MAX, num = 0;
344
345 if (flags & SCF_DO_SUBSTR)
346 scan_commit(data);
347 while (OP(scan) == code) {
348 I32 deltanext, minnext;
349
350 num++;
351 data_fake.flags = 0;
82ba1be6 352 if (data)
353 data_fake.whilem_c = data->whilem_c;
c277df42 354 next = regnext(scan);
355 scan = NEXTOPER(scan);
356 if (code != BRANCH)
357 scan = NEXTOPER(scan);
358 /* We suppose the run is continuous, last=next...*/
359 minnext = study_chunk(&scan, &deltanext, next,
360 &data_fake, 0);
361 if (min1 > minnext)
362 min1 = minnext;
363 if (max1 < minnext + deltanext)
364 max1 = minnext + deltanext;
365 if (deltanext == I32_MAX)
aca2d497 366 is_inf = is_inf_internal = 1;
c277df42 367 scan = next;
368 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
369 pars++;
405ff068 370 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42 371 data->flags |= SF_HAS_EVAL;
82ba1be6 372 if (data)
373 data->whilem_c = data_fake.whilem_c;
c277df42 374 if (code == SUSPEND)
375 break;
376 }
377 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
378 min1 = 0;
379 if (flags & SCF_DO_SUBSTR) {
380 data->pos_min += min1;
381 data->pos_delta += max1 - min1;
382 if (max1 != min1 || is_inf)
383 data->longest = &(data->longest_float);
384 }
385 min += min1;
386 delta += max1 - min1;
a0ed51b3 387 }
388 else if (code == BRANCHJ) /* single branch is optimized. */
c277df42 389 scan = NEXTOPER(NEXTOPER(scan));
390 else /* single branch is optimized. */
391 scan = NEXTOPER(scan);
392 continue;
a0ed51b3 393 }
394 else if (OP(scan) == EXACT) {
395 I32 l = *OPERAND(scan);
396 if (UTF) {
397 unsigned char *s = (unsigned char *)(OPERAND(scan)+1);
398 unsigned char *e = s + l;
399 I32 newl = 0;
400 while (s < e) {
401 newl++;
402 s += UTF8SKIP(s);
403 }
404 l = newl;
405 }
406 min += l;
c277df42 407 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
c277df42 408 /* The code below prefers earlier match for fixed
409 offset, later match for variable offset. */
410 if (data->last_end == -1) { /* Update the start info. */
411 data->last_start_min = data->pos_min;
412 data->last_start_max = is_inf
413 ? I32_MAX : data->pos_min + data->pos_delta;
414 }
a0ed51b3 415 sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), *OPERAND(scan));
c277df42 416 data->last_end = data->pos_min + l;
417 data->pos_min += l; /* As in the first entry. */
418 data->flags &= ~SF_BEFORE_EOL;
419 }
a0ed51b3 420 }
22c35a8c 421 else if (PL_regkind[(U8)OP(scan)] == EXACT) {
a0ed51b3 422 I32 l = *OPERAND(scan);
c277df42 423 if (flags & SCF_DO_SUBSTR)
424 scan_commit(data);
a0ed51b3 425 if (UTF) {
426 unsigned char *s = (unsigned char *)(OPERAND(scan)+1);
427 unsigned char *e = s + l;
428 I32 newl = 0;
429 while (s < e) {
430 newl++;
431 s += UTF8SKIP(s);
432 }
433 l = newl;
434 }
435 min += l;
c277df42 436 if (data && (flags & SCF_DO_SUBSTR))
a0ed51b3 437 data->pos_min += l;
438 }
4d61ec05 439 else if (strchr((char*)PL_varies,OP(scan))) {
c277df42 440 I32 mincount, maxcount, minnext, deltanext, pos_before, fl;
441 regnode *oscan = scan;
442
22c35a8c 443 switch (PL_regkind[(U8)OP(scan)]) {
c277df42 444 case WHILEM:
445 scan = NEXTOPER(scan);
446 goto finish;
447 case PLUS:
448 if (flags & SCF_DO_SUBSTR) {
449 next = NEXTOPER(scan);
450 if (OP(next) == EXACT) {
451 mincount = 1;
452 maxcount = REG_INFTY;
453 next = regnext(scan);
454 scan = NEXTOPER(scan);
455 goto do_curly;
456 }
457 }
458 if (flags & SCF_DO_SUBSTR)
459 data->pos_min++;
460 min++;
461 /* Fall through. */
462 case STAR:
aca2d497 463 is_inf = is_inf_internal = 1;
c277df42 464 scan = regnext(scan);
465 if (flags & SCF_DO_SUBSTR) {
466 scan_commit(data);
467 data->longest = &(data->longest_float);
468 }
469 goto optimize_curly_tail;
470 case CURLY:
471 mincount = ARG1(scan);
472 maxcount = ARG2(scan);
473 next = regnext(scan);
474 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
475 do_curly:
476 if (flags & SCF_DO_SUBSTR) {
477 if (mincount == 0) scan_commit(data);
478 pos_before = data->pos_min;
479 }
480 if (data) {
481 fl = data->flags;
482 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
483 if (is_inf)
484 data->flags |= SF_IS_INF;
485 }
486 /* This will finish on WHILEM, setting scan, or on NULL: */
487 minnext = study_chunk(&scan, &deltanext, last, data,
488 mincount == 0
489 ? (flags & ~SCF_DO_SUBSTR) : flags);
490 if (!scan) /* It was not CURLYX, but CURLY. */
491 scan = next;
599cee73 492 if (ckWARN(WARN_UNSAFE) && (minnext + deltanext == 0)
821b33a5 493 && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
17feb5d5 494 && maxcount <= REG_INFTY/3) /* Complement check for big count */
c5be433b 495 Perl_warner(aTHX_ WARN_UNSAFE,
496 "Strange *+?{} on zero-length expression");
c277df42 497 min += minnext * mincount;
aca2d497 498 is_inf_internal |= (maxcount == REG_INFTY
499 && (minnext + deltanext) > 0
500 || deltanext == I32_MAX);
501 is_inf |= is_inf_internal;
c277df42 502 delta += (minnext + deltanext) * maxcount - minnext * mincount;
503
504 /* Try powerful optimization CURLYX => CURLYN. */
c277df42 505 if ( OP(oscan) == CURLYX && data
506 && data->flags & SF_IN_PAR
507 && !(data->flags & SF_HAS_EVAL)
508 && !deltanext && minnext == 1 ) {
509 /* Try to optimize to CURLYN. */
510 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
511 regnode *nxt1 = nxt, *nxt2;
512
513 /* Skip open. */
514 nxt = regnext(nxt);
4d61ec05 515 if (!strchr((char*)PL_simple,OP(nxt))
22c35a8c 516 && !(PL_regkind[(U8)OP(nxt)] == EXACT
c277df42 517 && *OPERAND(nxt) == 1))
518 goto nogo;
519 nxt2 = nxt;
520 nxt = regnext(nxt);
521 if (OP(nxt) != CLOSE)
522 goto nogo;
523 /* Now we know that nxt2 is the only contents: */
524 oscan->flags = ARG(nxt);
525 OP(oscan) = CURLYN;
526 OP(nxt1) = NOTHING; /* was OPEN. */
527#ifdef DEBUGGING
528 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
529 NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
530 NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
531 OP(nxt) = OPTIMIZED; /* was CLOSE. */
532 OP(nxt + 1) = OPTIMIZED; /* was count. */
533 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
534#endif
535 }
c277df42 536 nogo:
537
538 /* Try optimization CURLYX => CURLYM. */
539 if ( OP(oscan) == CURLYX && data
c277df42 540 && !(data->flags & SF_HAS_PAR)
c277df42 541 && !(data->flags & SF_HAS_EVAL)
542 && !deltanext ) {
543 /* XXXX How to optimize if data == 0? */
544 /* Optimize to a simpler form. */
545 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
546 regnode *nxt2;
547
548 OP(oscan) = CURLYM;
549 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
550 && (OP(nxt2) != WHILEM))
551 nxt = nxt2;
552 OP(nxt2) = SUCCEED; /* Whas WHILEM */
c277df42 553 /* Need to optimize away parenths. */
554 if (data->flags & SF_IN_PAR) {
555 /* Set the parenth number. */
556 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
557
558 if (OP(nxt) != CLOSE)
559 FAIL("panic opt close");
560 oscan->flags = ARG(nxt);
561 OP(nxt1) = OPTIMIZED; /* was OPEN. */
562 OP(nxt) = OPTIMIZED; /* was CLOSE. */
563#ifdef DEBUGGING
564 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
565 OP(nxt + 1) = OPTIMIZED; /* was count. */
566 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
567 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
568#endif
569#if 0
570 while ( nxt1 && (OP(nxt1) != WHILEM)) {
571 regnode *nnxt = regnext(nxt1);
572
573 if (nnxt == nxt) {
574 if (reg_off_by_arg[OP(nxt1)])
575 ARG_SET(nxt1, nxt2 - nxt1);
576 else if (nxt2 - nxt1 < U16_MAX)
577 NEXT_OFF(nxt1) = nxt2 - nxt1;
578 else
579 OP(nxt) = NOTHING; /* Cannot beautify */
580 }
581 nxt1 = nnxt;
582 }
583#endif
584 /* Optimize again: */
585 study_chunk(&nxt1, &deltanext, nxt, NULL, 0);
a0ed51b3 586 }
587 else
c277df42 588 oscan->flags = 0;
c277df42 589 }
82ba1be6 590 else if (OP(oscan) == CURLYX && data && ++data->whilem_c < 16) {
591 /* This stays as CURLYX, and can put the count/of pair. */
592 /* Find WHILEM (as in regexec.c) */
593 regnode *nxt = oscan + NEXT_OFF(oscan);
594
595 if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
596 nxt += ARG(nxt);
597 PREVOPER(nxt)->flags = data->whilem_c
598 | (PL_reg_whilem_seen << 4); /* On WHILEM */
599 }
c277df42 600 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
601 pars++;
602 if (flags & SCF_DO_SUBSTR) {
603 SV *last_str = Nullsv;
604 int counted = mincount != 0;
605
606 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
607 I32 b = pos_before >= data->last_start_min
608 ? pos_before : data->last_start_min;
609 STRLEN l;
610 char *s = SvPV(data->last_found, l);
a0ed51b3 611 I32 old = b - data->last_start_min;
612
613 if (UTF)
614 old = utf8_hop((U8*)s, old) - (U8*)s;
c277df42 615
a0ed51b3 616 l -= old;
c277df42 617 /* Get the added string: */
79cb57f6 618 last_str = newSVpvn(s + old, l);
c277df42 619 if (deltanext == 0 && pos_before == b) {
620 /* What was added is a constant string */
621 if (mincount > 1) {
622 SvGROW(last_str, (mincount * l) + 1);
623 repeatcpy(SvPVX(last_str) + l,
624 SvPVX(last_str), l, mincount - 1);
625 SvCUR(last_str) *= mincount;
626 /* Add additional parts. */
627 SvCUR_set(data->last_found,
628 SvCUR(data->last_found) - l);
629 sv_catsv(data->last_found, last_str);
630 data->last_end += l * (mincount - 1);
631 }
632 }
633 }
634 /* It is counted once already... */
635 data->pos_min += minnext * (mincount - counted);
636 data->pos_delta += - counted * deltanext +
637 (minnext + deltanext) * maxcount - minnext * mincount;
638 if (mincount != maxcount) {
639 scan_commit(data);
640 if (mincount && last_str) {
641 sv_setsv(data->last_found, last_str);
642 data->last_end = data->pos_min;
643 data->last_start_min =
a0ed51b3 644 data->pos_min - CHR_SVLEN(last_str);
c277df42 645 data->last_start_max = is_inf
646 ? I32_MAX
647 : data->pos_min + data->pos_delta
a0ed51b3 648 - CHR_SVLEN(last_str);
c277df42 649 }
650 data->longest = &(data->longest_float);
651 }
aca2d497 652 SvREFCNT_dec(last_str);
c277df42 653 }
405ff068 654 if (data && (fl & SF_HAS_EVAL))
c277df42 655 data->flags |= SF_HAS_EVAL;
656 optimize_curly_tail:
c277df42 657 if (OP(oscan) != CURLYX) {
22c35a8c 658 while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
c277df42 659 && NEXT_OFF(next))
660 NEXT_OFF(oscan) += NEXT_OFF(next);
661 }
c277df42 662 continue;
663 default: /* REF only? */
664 if (flags & SCF_DO_SUBSTR) {
665 scan_commit(data);
666 data->longest = &(data->longest_float);
667 }
aca2d497 668 is_inf = is_inf_internal = 1;
c277df42 669 break;
670 }
a0ed51b3 671 }
4d61ec05 672 else if (strchr((char*)PL_simple,OP(scan)) || PL_regkind[(U8)OP(scan)] == ANYUTF8) {
c277df42 673 if (flags & SCF_DO_SUBSTR) {
674 scan_commit(data);
675 data->pos_min++;
676 }
677 min++;
a0ed51b3 678 }
22c35a8c 679 else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
c277df42 680 data->flags |= (OP(scan) == MEOL
681 ? SF_BEFORE_MEOL
682 : SF_BEFORE_SEOL);
a0ed51b3 683 }
22c35a8c 684 else if (PL_regkind[(U8)OP(scan)] == BRANCHJ
c277df42 685 && (scan->flags || data)
686 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
687 I32 deltanext, minnext;
688 regnode *nscan;
689
690 data_fake.flags = 0;
82ba1be6 691 if (data)
692 data_fake.whilem_c = data->whilem_c;
c277df42 693 next = regnext(scan);
694 nscan = NEXTOPER(NEXTOPER(scan));
695 minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0);
696 if (scan->flags) {
697 if (deltanext) {
698 FAIL("variable length lookbehind not implemented");
a0ed51b3 699 }
700 else if (minnext > U8_MAX) {
c277df42 701 FAIL2("lookbehind longer than %d not implemented", U8_MAX);
702 }
703 scan->flags = minnext;
704 }
705 if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
706 pars++;
405ff068 707 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42 708 data->flags |= SF_HAS_EVAL;
82ba1be6 709 if (data)
710 data->whilem_c = data_fake.whilem_c;
a0ed51b3 711 }
712 else if (OP(scan) == OPEN) {
c277df42 713 pars++;
a0ed51b3 714 }
715 else if (OP(scan) == CLOSE && ARG(scan) == is_par) {
c277df42 716 next = regnext(scan);
717
718 if ( next && (OP(next) != WHILEM) && next < last)
c277df42 719 is_par = 0; /* Disable optimization */
a0ed51b3 720 }
721 else if (OP(scan) == EVAL) {
c277df42 722 if (data)
723 data->flags |= SF_HAS_EVAL;
724 }
0f5d15d6 725 else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */
726 if (flags & SCF_DO_SUBSTR) {
727 scan_commit(data);
728 data->longest = &(data->longest_float);
729 }
730 is_inf = is_inf_internal = 1;
731 }
c277df42 732 /* Else: zero-length, ignore. */
733 scan = regnext(scan);
734 }
735
736 finish:
737 *scanp = scan;
aca2d497 738 *deltap = is_inf_internal ? I32_MAX : delta;
c277df42 739 if (flags & SCF_DO_SUBSTR && is_inf)
740 data->pos_delta = I32_MAX - data->pos_min;
741 if (is_par > U8_MAX)
742 is_par = 0;
743 if (is_par && pars==1 && data) {
744 data->flags |= SF_IN_PAR;
745 data->flags &= ~SF_HAS_PAR;
a0ed51b3 746 }
747 else if (pars && data) {
c277df42 748 data->flags |= SF_HAS_PAR;
749 data->flags &= ~SF_IN_PAR;
750 }
751 return min;
752}
753
76e3520e 754STATIC I32
cea2e8a9 755S_add_data(pTHX_ I32 n, char *s)
c277df42 756{
5c0ca799 757 dTHR;
3280af22 758 if (PL_regcomp_rx->data) {
759 Renewc(PL_regcomp_rx->data,
760 sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1),
c277df42 761 char, struct reg_data);
3280af22 762 Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8);
763 PL_regcomp_rx->data->count += n;
a0ed51b3 764 }
765 else {
3280af22 766 Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1),
c277df42 767 char, struct reg_data);
3280af22 768 New(1208, PL_regcomp_rx->data->what, n, U8);
769 PL_regcomp_rx->data->count = n;
c277df42 770 }
3280af22 771 Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8);
772 return PL_regcomp_rx->data->count - n;
c277df42 773}
774
d88dccdf 775void
864dbfa3 776Perl_reginitcolors(pTHX)
d88dccdf 777{
778 dTHR;
779 int i = 0;
780 char *s = PerlEnv_getenv("PERL_RE_COLORS");
781
782 if (s) {
783 PL_colors[0] = s = savepv(s);
784 while (++i < 6) {
785 s = strchr(s, '\t');
786 if (s) {
787 *s = '\0';
788 PL_colors[i] = ++s;
789 }
790 else
c712d376 791 PL_colors[i] = s = "";
d88dccdf 792 }
793 } else {
794 while (i < 6)
795 PL_colors[i++] = "";
796 }
797 PL_colorset = 1;
798}
799
a687059c 800/*
e50aee73 801 - pregcomp - compile a regular expression into internal code
a687059c 802 *
803 * We can't allocate space until we know how big the compiled form will be,
804 * but we can't compile it (and thus know how big it is) until we've got a
805 * place to put the code. So we cheat: we compile it twice, once with code
806 * generation turned off and size counting turned on, and once "for real".
807 * This also means that we don't allocate space until we are sure that the
808 * thing really will compile successfully, and we never have to move the
809 * code and thus invalidate pointers into it. (Note that it has to be in
810 * one piece because free() must be able to free it all.) [NB: not true in perl]
811 *
812 * Beware that the optimization-preparation code in here knows about some
813 * of the structure of the compiled regexp. [I'll say.]
814 */
815regexp *
864dbfa3 816Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
a687059c 817{
5c0ca799 818 dTHR;
a0d0e21e 819 register regexp *r;
c277df42 820 regnode *scan;
821 SV **longest;
822 SV *longest_fixed;
823 SV *longest_float;
824 regnode *first;
a0d0e21e 825 I32 flags;
a0d0e21e 826 I32 minlen = 0;
827 I32 sawplus = 0;
828 I32 sawopen = 0;
82ba1be6 829 scan_data_t data;
a0d0e21e 830
831 if (exp == NULL)
c277df42 832 FAIL("NULL regexp argument");
a0d0e21e 833
e24b16f9 834 if (PL_curcop == &PL_compiling ? (PL_hints & HINT_UTF8) : IN_UTF8)
a0ed51b3 835 PL_reg_flags |= RF_utf8;
836 else
837 PL_reg_flags = 0;
838
3280af22 839 PL_regprecomp = savepvn(exp, xend - exp);
35ef4773 840 DEBUG_r(if (!PL_colorset) reginitcolors());
82ba1be6 841 DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n",
d88dccdf 842 PL_colors[4],PL_colors[5],PL_colors[0],
843 xend - exp, PL_regprecomp, PL_colors[1]));
3280af22 844 PL_regflags = pm->op_pmflags;
845 PL_regsawback = 0;
bbce6d69 846
3280af22 847 PL_regseen = 0;
848 PL_seen_zerolen = *exp == '^' ? -1 : 0;
849 PL_seen_evals = 0;
850 PL_extralen = 0;
c277df42 851
bbce6d69 852 /* First pass: determine size, legality. */
3280af22 853 PL_regcomp_parse = exp;
854 PL_regxend = xend;
855 PL_regnaughty = 0;
856 PL_regnpar = 1;
857 PL_regsize = 0L;
858 PL_regcode = &PL_regdummy;
82ba1be6 859 PL_reg_whilem_seen = 0;
22c35a8c 860 regc((U8)REG_MAGIC, (char*)PL_regcode);
a0d0e21e 861 if (reg(0, &flags) == NULL) {
3280af22 862 Safefree(PL_regprecomp);
863 PL_regprecomp = Nullch;
a0d0e21e 864 return(NULL);
865 }
3280af22 866 DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", PL_regsize));
c277df42 867
c277df42 868 /* Small enough for pointer-storage convention?
869 If extralen==0, this means that we will not need long jumps. */
3280af22 870 if (PL_regsize >= 0x10000L && PL_extralen)
871 PL_regsize += PL_extralen;
c277df42 872 else
3280af22 873 PL_extralen = 0;
82ba1be6 874 if (PL_reg_whilem_seen > 15)
875 PL_reg_whilem_seen = 15;
a0d0e21e 876
bbce6d69 877 /* Allocate space and initialize. */
3280af22 878 Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode),
c277df42 879 char, regexp);
a0d0e21e 880 if (r == NULL)
881 FAIL("regexp out of space");
c277df42 882 r->refcnt = 1;
bbce6d69 883 r->prelen = xend - exp;
3280af22 884 r->precomp = PL_regprecomp;
cf93c79d 885 r->subbeg = NULL;
886 r->reganch = pm->op_pmflags & PMf_COMPILETIME;
4327152a 887 r->nparens = PL_regnpar - 1; /* set early to validate backrefs */
888
889 r->substrs = 0; /* Useful during FAIL. */
890 r->startp = 0; /* Useful during FAIL. */
891 r->endp = 0; /* Useful during FAIL. */
892
3280af22 893 PL_regcomp_rx = r;
bbce6d69 894
895 /* Second pass: emit code. */
3280af22 896 PL_regcomp_parse = exp;
897 PL_regxend = xend;
898 PL_regnaughty = 0;
899 PL_regnpar = 1;
900 PL_regcode = r->program;
2cd61cdb 901 /* Store the count of eval-groups for security checks: */
3280af22 902 PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals);
22c35a8c 903 regc((U8)REG_MAGIC, (char*) PL_regcode++);
c277df42 904 r->data = 0;
a0d0e21e 905 if (reg(0, &flags) == NULL)
906 return(NULL);
907
908 /* Dig out information for optimizations. */
cf93c79d 909 r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
3280af22 910 pm->op_pmflags = PL_regflags;
a0ed51b3 911 if (UTF)
912 r->reganch |= ROPT_UTF8;
c277df42 913 r->regstclass = NULL;
a0ed51b3 914 if (PL_regnaughty >= 10) /* Probably an expensive pattern. */
915 r->reganch |= ROPT_NAUGHTY;
c277df42 916 scan = r->program + 1; /* First BRANCH. */
2779dcf1 917
918 /* XXXX To minimize changes to RE engine we always allocate
919 3-units-long substrs field. */
920 Newz(1004, r->substrs, 1, struct reg_substr_data);
921
82ba1be6 922 StructCopy(&zero_scan_data, &data, scan_data_t);
c277df42 923 if (OP(scan) != BRANCH) { /* Only one top-level choice. */
c277df42 924 I32 fake;
c5254dd6 925 STRLEN longest_float_length, longest_fixed_length;
a0d0e21e 926
927 first = scan;
c277df42 928 /* Skip introductions and multiplicators >= 1. */
a0d0e21e 929 while ((OP(first) == OPEN && (sawopen = 1)) ||
930 (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
931 (OP(first) == PLUS) ||
932 (OP(first) == MINMOD) ||
22c35a8c 933 (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
a0d0e21e 934 if (OP(first) == PLUS)
935 sawplus = 1;
936 else
937 first += regarglen[(U8)OP(first)];
938 first = NEXTOPER(first);
a687059c 939 }
940
a0d0e21e 941 /* Starting-point info. */
942 again:
c277df42 943 if (OP(first) == EXACT); /* Empty, get anchored substr later. */
4d61ec05 944 else if (strchr((char*)PL_simple+4,OP(first)))
a0d0e21e 945 r->regstclass = first;
22c35a8c 946 else if (PL_regkind[(U8)OP(first)] == BOUND ||
947 PL_regkind[(U8)OP(first)] == NBOUND)
a0d0e21e 948 r->regstclass = first;
22c35a8c 949 else if (PL_regkind[(U8)OP(first)] == BOL) {
cad2e5aa 950 r->reganch |= (OP(first) == MBOL
951 ? ROPT_ANCH_MBOL
952 : (OP(first) == SBOL
953 ? ROPT_ANCH_SBOL
954 : ROPT_ANCH_BOL));
a0d0e21e 955 first = NEXTOPER(first);
774d564b 956 goto again;
957 }
958 else if (OP(first) == GPOS) {
959 r->reganch |= ROPT_ANCH_GPOS;
960 first = NEXTOPER(first);
961 goto again;
a0d0e21e 962 }
963 else if ((OP(first) == STAR &&
22c35a8c 964 PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
a0d0e21e 965 !(r->reganch & ROPT_ANCH) )
966 {
967 /* turn .* into ^.* with an implied $*=1 */
cad2e5aa 968 int type = OP(NEXTOPER(first));
969
970 if (type == REG_ANY || type == ANYUTF8)
971 type = ROPT_ANCH_MBOL;
972 else
973 type = ROPT_ANCH_SBOL;
974
975 r->reganch |= type | ROPT_IMPLICIT;
a0d0e21e 976 first = NEXTOPER(first);
774d564b 977 goto again;
a0d0e21e 978 }
cad2e5aa 979 if (sawplus && (!sawopen || !PL_regsawback)
980 && !(PL_regseen & REG_SEEN_EVAL)) /* May examine pos and $& */
981 /* x+ must match at the 1st pos of run of x's */
982 r->reganch |= ROPT_SKIP;
a0d0e21e 983
c277df42 984 /* Scan is after the zeroth branch, first is atomic matcher. */
985 DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n",
986 first - scan + 1));
a0d0e21e 987 /*
988 * If there's something expensive in the r.e., find the
989 * longest literal string that must appear and make it the
990 * regmust. Resolve ties in favor of later strings, since
991 * the regstart check works with the beginning of the r.e.
992 * and avoiding duplication strengthens checking. Not a
993 * strong reason, but sufficient in the absence of others.
994 * [Now we resolve ties in favor of the earlier string if
c277df42 995 * it happens that c_offset_min has been invalidated, since the
a0d0e21e 996 * earlier string may buy us something the later one won't.]
997 */
a0d0e21e 998 minlen = 0;
a687059c 999
79cb57f6 1000 data.longest_fixed = newSVpvn("",0);
1001 data.longest_float = newSVpvn("",0);
1002 data.last_found = newSVpvn("",0);
c277df42 1003 data.longest = &(data.longest_fixed);
1004 first = scan;
1005
3280af22 1006 minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */
c277df42 1007 &data, SCF_DO_SUBSTR);
3280af22 1008 if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed)
c277df42 1009 && data.last_start_min == 0 && data.last_end > 0
3280af22 1010 && !PL_seen_zerolen
1011 && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
c277df42 1012 r->reganch |= ROPT_CHECK_ALL;
1013 scan_commit(&data);
1014 SvREFCNT_dec(data.last_found);
1015
a0ed51b3 1016 longest_float_length = CHR_SVLEN(data.longest_float);
c5254dd6 1017 if (longest_float_length
c277df42 1018 || (data.flags & SF_FL_BEFORE_EOL
1019 && (!(data.flags & SF_FL_BEFORE_MEOL)
3280af22 1020 || (PL_regflags & PMf_MULTILINE)))) {
cf93c79d 1021 int t;
1022
a0ed51b3 1023 if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */
aca2d497 1024 && data.offset_fixed == data.offset_float_min
1025 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
1026 goto remove_float; /* As in (a)+. */
1027
c277df42 1028 r->float_substr = data.longest_float;
1029 r->float_min_offset = data.offset_float_min;
1030 r->float_max_offset = data.offset_float_max;
cf93c79d 1031 t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
1032 && (!(data.flags & SF_FL_BEFORE_MEOL)
1033 || (PL_regflags & PMf_MULTILINE)));
1034 fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0);
a0ed51b3 1035 }
1036 else {
aca2d497 1037 remove_float:
c277df42 1038 r->float_substr = Nullsv;
1039 SvREFCNT_dec(data.longest_float);
c5254dd6 1040 longest_float_length = 0;
a0d0e21e 1041 }
c277df42 1042
a0ed51b3 1043 longest_fixed_length = CHR_SVLEN(data.longest_fixed);
c5254dd6 1044 if (longest_fixed_length
c277df42 1045 || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1046 && (!(data.flags & SF_FIX_BEFORE_MEOL)
3280af22 1047 || (PL_regflags & PMf_MULTILINE)))) {
cf93c79d 1048 int t;
1049
c277df42 1050 r->anchored_substr = data.longest_fixed;
1051 r->anchored_offset = data.offset_fixed;
cf93c79d 1052 t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
1053 && (!(data.flags & SF_FIX_BEFORE_MEOL)
1054 || (PL_regflags & PMf_MULTILINE)));
1055 fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0);
a0ed51b3 1056 }
1057 else {
c277df42 1058 r->anchored_substr = Nullsv;
1059 SvREFCNT_dec(data.longest_fixed);
c5254dd6 1060 longest_fixed_length = 0;
a0d0e21e 1061 }
c277df42 1062
1063 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
c5254dd6 1064 if (longest_fixed_length > longest_float_length) {
c277df42 1065 r->check_substr = r->anchored_substr;
1066 r->check_offset_min = r->check_offset_max = r->anchored_offset;
1067 if (r->reganch & ROPT_ANCH_SINGLE)
1068 r->reganch |= ROPT_NOSCAN;
a0ed51b3 1069 }
1070 else {
c277df42 1071 r->check_substr = r->float_substr;
1072 r->check_offset_min = data.offset_float_min;
1073 r->check_offset_max = data.offset_float_max;
a0d0e21e 1074 }
cad2e5aa 1075 if (r->check_substr) {
1076 r->reganch |= RE_USE_INTUIT;
1077 if (SvTAIL(r->check_substr))
1078 r->reganch |= RE_INTUIT_TAIL;
1079 }
a0ed51b3 1080 }
1081 else {
c277df42 1082 /* Several toplevels. Best we can is to set minlen. */
1083 I32 fake;
1084
1085 DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
1086 scan = r->program + 1;
82ba1be6 1087 minlen = study_chunk(&scan, &fake, scan + PL_regsize, &data, 0);
c277df42 1088 r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
a0d0e21e 1089 }
1090
a0d0e21e 1091 r->minlen = minlen;
3280af22 1092 if (PL_regseen & REG_SEEN_GPOS)
c277df42 1093 r->reganch |= ROPT_GPOS_SEEN;
3280af22 1094 if (PL_regseen & REG_SEEN_LOOKBEHIND)
c277df42 1095 r->reganch |= ROPT_LOOKBEHIND_SEEN;
3280af22 1096 if (PL_regseen & REG_SEEN_EVAL)
ce862d02 1097 r->reganch |= ROPT_EVAL_SEEN;
cf93c79d 1098 Newz(1002, r->startp, PL_regnpar, I32);
1099 Newz(1002, r->endp, PL_regnpar, I32);
a0d0e21e 1100 DEBUG_r(regdump(r));
1101 return(r);
a687059c 1102}
1103
1104/*
1105 - reg - regular expression, i.e. main body or parenthesized thing
1106 *
1107 * Caller must absorb opening parenthesis.
1108 *
1109 * Combining parenthesis handling with the base level of regular expression
1110 * is a trifle forced, but the need to tie the tails of the branches to what
1111 * follows makes it hard to avoid.
1112 */
76e3520e 1113STATIC regnode *
cea2e8a9 1114S_reg(pTHX_ I32 paren, I32 *flagp)
c277df42 1115 /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
a687059c 1116{
5c0ca799 1117 dTHR;
c277df42 1118 register regnode *ret; /* Will be the head of the group. */
1119 register regnode *br;
1120 register regnode *lastbr;
1121 register regnode *ender = 0;
a0d0e21e 1122 register I32 parno = 0;
3280af22 1123 I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0;
c277df42 1124 char c;
a0d0e21e 1125
821b33a5 1126 *flagp = 0; /* Tentatively. */
a0d0e21e 1127
1128 /* Make an OPEN node, if parenthesized. */
1129 if (paren) {
3280af22 1130 if (*PL_regcomp_parse == '?') {
ca9dfc88 1131 U16 posflags = 0, negflags = 0;
1132 U16 *flagsp = &posflags;
0f5d15d6 1133 int logical = 0;
ca9dfc88 1134
3280af22 1135 PL_regcomp_parse++;
1136 paren = *PL_regcomp_parse++;
c277df42 1137 ret = NULL; /* For look-ahead/behind. */
a0d0e21e 1138 switch (paren) {
c277df42 1139 case '<':
3280af22 1140 PL_regseen |= REG_SEEN_LOOKBEHIND;
1141 if (*PL_regcomp_parse == '!')
c277df42 1142 paren = ',';
3280af22 1143 if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!')
c277df42 1144 goto unknown;
3280af22 1145 PL_regcomp_parse++;
a0d0e21e 1146 case '=':
1147 case '!':
3280af22 1148 PL_seen_zerolen++;
c277df42 1149 case ':':
1150 case '>':
a0d0e21e 1151 break;
1152 case '$':
1153 case '@':
c277df42 1154 FAIL2("Sequence (?%c...) not implemented", (int)paren);
a0d0e21e 1155 break;
1156 case '#':
3280af22 1157 while (*PL_regcomp_parse && *PL_regcomp_parse != ')')
1158 PL_regcomp_parse++;
1159 if (*PL_regcomp_parse != ')')
c277df42 1160 FAIL("Sequence (?#... not terminated");
a0d0e21e 1161 nextchar();
1162 *flagp = TRYAGAIN;
1163 return NULL;
0f5d15d6 1164 case 'p':
1165 logical = 1;
1166 paren = *PL_regcomp_parse++;
1167 /* FALL THROUGH */
c277df42 1168 case '{':
1169 {
1170 dTHR;
1171 I32 count = 1, n = 0;
1172 char c;
3280af22 1173 char *s = PL_regcomp_parse;
c277df42 1174 SV *sv;
1175 OP_4tree *sop, *rop;
1176
3280af22 1177 PL_seen_zerolen++;
1178 PL_regseen |= REG_SEEN_EVAL;
1179 while (count && (c = *PL_regcomp_parse)) {
1180 if (c == '\\' && PL_regcomp_parse[1])
1181 PL_regcomp_parse++;
c277df42 1182 else if (c == '{')
1183 count++;
1184 else if (c == '}')
1185 count--;
3280af22 1186 PL_regcomp_parse++;
c277df42 1187 }
3280af22 1188 if (*PL_regcomp_parse != ')')
c277df42 1189 FAIL("Sequence (?{...}) not terminated or not {}-balanced");
1190 if (!SIZE_ONLY) {
1191 AV *av;
1192
3280af22 1193 if (PL_regcomp_parse - 1 - s)
79cb57f6 1194 sv = newSVpvn(s, PL_regcomp_parse - 1 - s);
c277df42 1195 else
79cb57f6 1196 sv = newSVpvn("", 0);
c277df42 1197
1198 rop = sv_compile_2op(sv, &sop, "re", &av);
1199
dfad63ad 1200 n = add_data(3, "nop");
3280af22 1201 PL_regcomp_rx->data->data[n] = (void*)rop;
dfad63ad 1202 PL_regcomp_rx->data->data[n+1] = (void*)sop;
1203 PL_regcomp_rx->data->data[n+2] = (void*)av;
c277df42 1204 SvREFCNT_dec(sv);
a0ed51b3 1205 }
e24b16f9 1206 else { /* First pass */
1207 if (PL_reginterp_cnt < ++PL_seen_evals
1208 && PL_curcop != &PL_compiling)
2cd61cdb 1209 /* No compiled RE interpolated, has runtime
1210 components ===> unsafe. */
1211 FAIL("Eval-group not allowed at runtime, use re 'eval'");
3280af22 1212 if (PL_tainted)
cc6b7395 1213 FAIL("Eval-group in insecure regular expression");
c277df42 1214 }
1215
1216 nextchar();
0f5d15d6 1217 if (logical) {
1218 ret = reg_node(LOGICAL);
1219 if (!SIZE_ONLY)
1220 ret->flags = 2;
1221 regtail(ret, reganode(EVAL, n));
1222 return ret;
1223 }
c277df42 1224 return reganode(EVAL, n);
1225 }
1226 case '(':
1227 {
3280af22 1228 if (PL_regcomp_parse[0] == '?') {
1229 if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!'
1230 || PL_regcomp_parse[1] == '<'
1231 || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */
c277df42 1232 I32 flag;
1233
1234 ret = reg_node(LOGICAL);
0f5d15d6 1235 if (!SIZE_ONLY)
1236 ret->flags = 1;
c277df42 1237 regtail(ret, reg(1, &flag));
1238 goto insert_if;
1239 }
a0ed51b3 1240 }
1241 else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) {
3280af22 1242 parno = atoi(PL_regcomp_parse++);
c277df42 1243
3280af22 1244 while (isDIGIT(*PL_regcomp_parse))
1245 PL_regcomp_parse++;
c277df42 1246 ret = reganode(GROUPP, parno);
1247 if ((c = *nextchar()) != ')')
1248 FAIL2("Switch (?(number%c not recognized", c);
1249 insert_if:
1250 regtail(ret, reganode(IFTHEN, 0));
1251 br = regbranch(&flags, 1);
1252 if (br == NULL)
1253 br = reganode(LONGJMP, 0);
1254 else
1255 regtail(br, reganode(LONGJMP, 0));
1256 c = *nextchar();
d1b80229 1257 if (flags&HASWIDTH)
1258 *flagp |= HASWIDTH;
c277df42 1259 if (c == '|') {
1260 lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */
1261 regbranch(&flags, 1);
1262 regtail(ret, lastbr);
d1b80229 1263 if (flags&HASWIDTH)
1264 *flagp |= HASWIDTH;
c277df42 1265 c = *nextchar();
a0ed51b3 1266 }
1267 else
c277df42 1268 lastbr = NULL;
1269 if (c != ')')
1270 FAIL("Switch (?(condition)... contains too many branches");
1271 ender = reg_node(TAIL);
1272 regtail(br, ender);
1273 if (lastbr) {
1274 regtail(lastbr, ender);
1275 regtail(NEXTOPER(NEXTOPER(lastbr)), ender);
a0ed51b3 1276 }
1277 else
c277df42 1278 regtail(ret, ender);
1279 return ret;
a0ed51b3 1280 }
1281 else {
3280af22 1282 FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse);
c277df42 1283 }
1284 }
1b1626e4 1285 case 0:
c277df42 1286 FAIL("Sequence (? incomplete");
1b1626e4 1287 break;
a0d0e21e 1288 default:
3280af22 1289 --PL_regcomp_parse;
ca9dfc88 1290 parse_flags:
3280af22 1291 while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) {
1292 if (*PL_regcomp_parse != 'o')
1293 pmflag(flagsp, *PL_regcomp_parse);
1294 ++PL_regcomp_parse;
ca9dfc88 1295 }
3280af22 1296 if (*PL_regcomp_parse == '-') {
ca9dfc88 1297 flagsp = &negflags;
3280af22 1298 ++PL_regcomp_parse;
ca9dfc88 1299 goto parse_flags;
48c036b1 1300 }
3280af22 1301 PL_regflags |= posflags;
1302 PL_regflags &= ~negflags;
1303 if (*PL_regcomp_parse == ':') {
1304 PL_regcomp_parse++;
ca9dfc88 1305 paren = ':';
1306 break;
1307 }
c277df42 1308 unknown:
3280af22 1309 if (*PL_regcomp_parse != ')')
1310 FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse);
a0d0e21e 1311 nextchar();
1312 *flagp = TRYAGAIN;
1313 return NULL;
1314 }
1315 }
1316 else {
3280af22 1317 parno = PL_regnpar;
1318 PL_regnpar++;
a0d0e21e 1319 ret = reganode(OPEN, parno);
c277df42 1320 open = 1;
a0d0e21e 1321 }
a0ed51b3 1322 }
1323 else
a0d0e21e 1324 ret = NULL;
1325
1326 /* Pick up the branches, linking them together. */
c277df42 1327 br = regbranch(&flags, 1);
a0d0e21e 1328 if (br == NULL)
1329 return(NULL);
3280af22 1330 if (*PL_regcomp_parse == '|') {
1331 if (!SIZE_ONLY && PL_extralen) {
c277df42 1332 reginsert(BRANCHJ, br);
a0ed51b3 1333 }
1334 else
c277df42 1335 reginsert(BRANCH, br);
1336 have_branch = 1;
1337 if (SIZE_ONLY)
3280af22 1338 PL_extralen += 1; /* For BRANCHJ-BRANCH. */
a0ed51b3 1339 }
1340 else if (paren == ':') {
c277df42 1341 *flagp |= flags&SIMPLE;
1342 }
1343 if (open) { /* Starts with OPEN. */
1344 regtail(ret, br); /* OPEN -> first. */
a0ed51b3 1345 }
1346 else if (paren != '?') /* Not Conditional */
a0d0e21e 1347 ret = br;
821b33a5 1348 if (flags&HASWIDTH)
1349 *flagp |= HASWIDTH;
a0d0e21e 1350 *flagp |= flags&SPSTART;
c277df42 1351 lastbr = br;
3280af22 1352 while (*PL_regcomp_parse == '|') {
1353 if (!SIZE_ONLY && PL_extralen) {
c277df42 1354 ender = reganode(LONGJMP,0);
1355 regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
1356 }
1357 if (SIZE_ONLY)
3280af22 1358 PL_extralen += 2; /* Account for LONGJMP. */
a0d0e21e 1359 nextchar();
c277df42 1360 br = regbranch(&flags, 0);
a687059c 1361 if (br == NULL)
a0d0e21e 1362 return(NULL);
c277df42 1363 regtail(lastbr, br); /* BRANCH -> BRANCH. */
1364 lastbr = br;
821b33a5 1365 if (flags&HASWIDTH)
1366 *flagp |= HASWIDTH;
a687059c 1367 *flagp |= flags&SPSTART;
a0d0e21e 1368 }
1369
c277df42 1370 if (have_branch || paren != ':') {
1371 /* Make a closing node, and hook it on the end. */
1372 switch (paren) {
1373 case ':':
1374 ender = reg_node(TAIL);
1375 break;
1376 case 1:
1377 ender = reganode(CLOSE, parno);
1378 break;
1379 case '<':
c277df42 1380 case ',':
1381 case '=':
1382 case '!':
c277df42 1383 *flagp &= ~HASWIDTH;
821b33a5 1384 /* FALL THROUGH */
1385 case '>':
1386 ender = reg_node(SUCCEED);
c277df42 1387 break;
1388 case 0:
1389 ender = reg_node(END);
1390 break;
1391 }
1392 regtail(lastbr, ender);
a0d0e21e 1393
c277df42 1394 if (have_branch) {
1395 /* Hook the tails of the branches to the closing node. */
1396 for (br = ret; br != NULL; br = regnext(br)) {
1397 regoptail(br, ender);
1398 }
1399 }
a0d0e21e 1400 }
c277df42 1401
1402 {
1403 char *p;
1404 static char parens[] = "=!<,>";
1405
1406 if (paren && (p = strchr(parens, paren))) {
1407 int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
1408 int flag = (p - parens) > 1;
1409
1410 if (paren == '>')
1411 node = SUSPEND, flag = 0;
1412 reginsert(node,ret);
c277df42 1413 ret->flags = flag;
c277df42 1414 regtail(ret, reg_node(TAIL));
1415 }
a0d0e21e 1416 }
1417
1418 /* Check for proper termination. */
ce3e6498 1419 if (paren) {
1420 PL_regflags = oregflags;
1421 if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') {
1422 FAIL("unmatched () in regexp");
1423 }
a0ed51b3 1424 }
1425 else if (!paren && PL_regcomp_parse < PL_regxend) {
3280af22 1426 if (*PL_regcomp_parse == ')') {
a0d0e21e 1427 FAIL("unmatched () in regexp");
a0ed51b3 1428 }
1429 else
a0d0e21e 1430 FAIL("junk on end of regexp"); /* "Can't happen". */
1431 /* NOTREACHED */
1432 }
a687059c 1433
a0d0e21e 1434 return(ret);
a687059c 1435}
1436
1437/*
1438 - regbranch - one alternative of an | operator
1439 *
1440 * Implements the concatenation operator.
1441 */
76e3520e 1442STATIC regnode *
cea2e8a9 1443S_regbranch(pTHX_ I32 *flagp, I32 first)
a687059c 1444{
5c0ca799 1445 dTHR;
c277df42 1446 register regnode *ret;
1447 register regnode *chain = NULL;
1448 register regnode *latest;
1449 I32 flags = 0, c = 0;
a0d0e21e 1450
c277df42 1451 if (first)
1452 ret = NULL;
1453 else {
3280af22 1454 if (!SIZE_ONLY && PL_extralen)
c277df42 1455 ret = reganode(BRANCHJ,0);
1456 else
1457 ret = reg_node(BRANCH);
1458 }
1459
1460 if (!first && SIZE_ONLY)
3280af22 1461 PL_extralen += 1; /* BRANCHJ */
c277df42 1462
1463 *flagp = WORST; /* Tentatively. */
a0d0e21e 1464
3280af22 1465 PL_regcomp_parse--;
a0d0e21e 1466 nextchar();
3280af22 1467 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') {
a0d0e21e 1468 flags &= ~TRYAGAIN;
1469 latest = regpiece(&flags);
1470 if (latest == NULL) {
1471 if (flags & TRYAGAIN)
1472 continue;
1473 return(NULL);
a0ed51b3 1474 }
1475 else if (ret == NULL)
c277df42 1476 ret = latest;
a0d0e21e 1477 *flagp |= flags&HASWIDTH;
c277df42 1478 if (chain == NULL) /* First piece. */
a0d0e21e 1479 *flagp |= flags&SPSTART;
1480 else {
3280af22 1481 PL_regnaughty++;
a0d0e21e 1482 regtail(chain, latest);
a687059c 1483 }
a0d0e21e 1484 chain = latest;
c277df42 1485 c++;
1486 }
1487 if (chain == NULL) { /* Loop ran zero times. */
1488 chain = reg_node(NOTHING);
1489 if (ret == NULL)
1490 ret = chain;
1491 }
1492 if (c == 1) {
1493 *flagp |= flags&SIMPLE;
a0d0e21e 1494 }
a687059c 1495
a0d0e21e 1496 return(ret);
a687059c 1497}
1498
1499/*
1500 - regpiece - something followed by possible [*+?]
1501 *
1502 * Note that the branching code sequences used for ? and the general cases
1503 * of * and + are somewhat optimized: they use the same NOTHING node as
1504 * both the endmarker for their branch list and the body of the last branch.
1505 * It might seem that this node could be dispensed with entirely, but the
1506 * endmarker role is not redundant.
1507 */
76e3520e 1508STATIC regnode *
cea2e8a9 1509S_regpiece(pTHX_ I32 *flagp)
a687059c 1510{
5c0ca799 1511 dTHR;
c277df42 1512 register regnode *ret;
a0d0e21e 1513 register char op;
1514 register char *next;
1515 I32 flags;
3280af22 1516 char *origparse = PL_regcomp_parse;
a0d0e21e 1517 char *maxpos;
1518 I32 min;
c277df42 1519 I32 max = REG_INFTY;
a0d0e21e 1520
1521 ret = regatom(&flags);
1522 if (ret == NULL) {
1523 if (flags & TRYAGAIN)
1524 *flagp |= TRYAGAIN;
1525 return(NULL);
1526 }
1527
3280af22 1528 op = *PL_regcomp_parse;
a0d0e21e 1529
3280af22 1530 if (op == '{' && regcurly(PL_regcomp_parse)) {
1531 next = PL_regcomp_parse + 1;
a0d0e21e 1532 maxpos = Nullch;
1533 while (isDIGIT(*next) || *next == ',') {
1534 if (*next == ',') {
1535 if (maxpos)
1536 break;
1537 else
1538 maxpos = next;
a687059c 1539 }
a0d0e21e 1540 next++;
1541 }
1542 if (*next == '}') { /* got one */
1543 if (!maxpos)
1544 maxpos = next;
3280af22 1545 PL_regcomp_parse++;
1546 min = atoi(PL_regcomp_parse);
a0d0e21e 1547 if (*maxpos == ',')
1548 maxpos++;
1549 else
3280af22 1550 maxpos = PL_regcomp_parse;
a0d0e21e 1551 max = atoi(maxpos);
1552 if (!max && *maxpos != '0')
c277df42 1553 max = REG_INFTY; /* meaning "infinity" */
1554 else if (max >= REG_INFTY)
1555 FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
3280af22 1556 PL_regcomp_parse = next;
a0d0e21e 1557 nextchar();
1558
1559 do_curly:
1560 if ((flags&SIMPLE)) {
3280af22 1561 PL_regnaughty += 2 + PL_regnaughty / 2;
a0d0e21e 1562 reginsert(CURLY, ret);
1563 }
1564 else {
82ba1be6 1565 regnode *w = reg_node(WHILEM);
1566
1567 w->flags = 0;
1568 regtail(ret, w);
3280af22 1569 if (!SIZE_ONLY && PL_extralen) {
c277df42 1570 reginsert(LONGJMP,ret);
1571 reginsert(NOTHING,ret);
1572 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
1573 }
a0d0e21e 1574 reginsert(CURLYX,ret);
3280af22 1575 if (!SIZE_ONLY && PL_extralen)
c277df42 1576 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
1577 regtail(ret, reg_node(NOTHING));
1578 if (SIZE_ONLY)
82ba1be6 1579 PL_reg_whilem_seen++, PL_extralen += 3;
1580 PL_regnaughty += 4 + PL_regnaughty; /* compound interest */
a0d0e21e 1581 }
c277df42 1582 ret->flags = 0;
a0d0e21e 1583
1584 if (min > 0)
821b33a5 1585 *flagp = WORST;
1586 if (max > 0)
1587 *flagp |= HASWIDTH;
a0d0e21e 1588 if (max && max < min)
c277df42 1589 FAIL("Can't do {n,m} with n > m");
1590 if (!SIZE_ONLY) {
1591 ARG1_SET(ret, min);
1592 ARG2_SET(ret, max);
a687059c 1593 }
a687059c 1594
a0d0e21e 1595 goto nest_check;
a687059c 1596 }
a0d0e21e 1597 }
a687059c 1598
a0d0e21e 1599 if (!ISMULT1(op)) {
1600 *flagp = flags;
a687059c 1601 return(ret);
a0d0e21e 1602 }
bb20fd44 1603
c277df42 1604#if 0 /* Now runtime fix should be reliable. */
bb20fd44 1605 if (!(flags&HASWIDTH) && op != '?')
c277df42 1606 FAIL("regexp *+ operand could be empty");
1607#endif
bb20fd44 1608
a0d0e21e 1609 nextchar();
1610
821b33a5 1611 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
a0d0e21e 1612
1613 if (op == '*' && (flags&SIMPLE)) {
1614 reginsert(STAR, ret);
c277df42 1615 ret->flags = 0;
3280af22 1616 PL_regnaughty += 4;
a0d0e21e 1617 }
1618 else if (op == '*') {
1619 min = 0;
1620 goto do_curly;
a0ed51b3 1621 }
1622 else if (op == '+' && (flags&SIMPLE)) {
a0d0e21e 1623 reginsert(PLUS, ret);
c277df42 1624 ret->flags = 0;
3280af22 1625 PL_regnaughty += 3;
a0d0e21e 1626 }
1627 else if (op == '+') {
1628 min = 1;
1629 goto do_curly;
a0ed51b3 1630 }
1631 else if (op == '?') {
a0d0e21e 1632 min = 0; max = 1;
1633 goto do_curly;
1634 }
1635 nest_check:
17feb5d5 1636 if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
cea2e8a9 1637 Perl_warner(aTHX_ WARN_UNSAFE, "%.*s matches null string many times",
3280af22 1638 PL_regcomp_parse - origparse, origparse);
a0d0e21e 1639 }
1640
3280af22 1641 if (*PL_regcomp_parse == '?') {
a0d0e21e 1642 nextchar();
1643 reginsert(MINMOD, ret);
c277df42 1644 regtail(ret, ret + NODE_STEP_REGNODE);
a0d0e21e 1645 }
3280af22 1646 if (ISMULT2(PL_regcomp_parse))
a0d0e21e 1647 FAIL("nested *?+ in regexp");
1648
1649 return(ret);
a687059c 1650}
1651
1652/*
1653 - regatom - the lowest level
1654 *
1655 * Optimization: gobbles an entire sequence of ordinary characters so that
1656 * it can turn them into a single node, which is smaller to store and
1657 * faster to run. Backslashed characters are exceptions, each becoming a
1658 * separate node; the code is simpler that way and it's not worth fixing.
1659 *
1660 * [Yes, it is worth fixing, some scripts can run twice the speed.]
1661 */
76e3520e 1662STATIC regnode *
cea2e8a9 1663S_regatom(pTHX_ I32 *flagp)
a687059c 1664{
5c0ca799 1665 dTHR;
c277df42 1666 register regnode *ret = 0;
a0d0e21e 1667 I32 flags;
1668
1669 *flagp = WORST; /* Tentatively. */
1670
1671tryagain:
3280af22 1672 switch (*PL_regcomp_parse) {
a0d0e21e 1673 case '^':
3280af22 1674 PL_seen_zerolen++;
a0d0e21e 1675 nextchar();
3280af22 1676 if (PL_regflags & PMf_MULTILINE)
c277df42 1677 ret = reg_node(MBOL);
3280af22 1678 else if (PL_regflags & PMf_SINGLELINE)
c277df42 1679 ret = reg_node(SBOL);
a0d0e21e 1680 else
c277df42 1681 ret = reg_node(BOL);
a0d0e21e 1682 break;
1683 case '$':
3280af22 1684 if (PL_regcomp_parse[1])
1685 PL_seen_zerolen++;
a0d0e21e 1686 nextchar();
3280af22 1687 if (PL_regflags & PMf_MULTILINE)
c277df42 1688 ret = reg_node(MEOL);
3280af22 1689 else if (PL_regflags & PMf_SINGLELINE)
c277df42 1690 ret = reg_node(SEOL);
a0d0e21e 1691 else
c277df42 1692 ret = reg_node(EOL);
a0d0e21e 1693 break;
1694 case '.':
1695 nextchar();
a0ed51b3 1696 if (UTF) {
1697 if (PL_regflags & PMf_SINGLELINE)
1698 ret = reg_node(SANYUTF8);
1699 else
1700 ret = reg_node(ANYUTF8);
1701 *flagp |= HASWIDTH;
1702 }
1703 else {
1704 if (PL_regflags & PMf_SINGLELINE)
1705 ret = reg_node(SANY);
1706 else
22c35a8c 1707 ret = reg_node(REG_ANY);
a0ed51b3 1708 *flagp |= HASWIDTH|SIMPLE;
1709 }
3280af22 1710 PL_regnaughty++;
a0d0e21e 1711 break;
1712 case '[':
3280af22 1713 PL_regcomp_parse++;
a0ed51b3 1714 ret = (UTF ? regclassutf8() : regclass());
a14b48bc 1715 if (*PL_regcomp_parse != ']')
1716 FAIL("unmatched [] in regexp");
1717 nextchar();
a0d0e21e 1718 *flagp |= HASWIDTH|SIMPLE;
1719 break;
1720 case '(':
1721 nextchar();
1722 ret = reg(1, &flags);
1723 if (ret == NULL) {
1724 if (flags & TRYAGAIN)
1725 goto tryagain;
1726 return(NULL);
1727 }
c277df42 1728 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
a0d0e21e 1729 break;
1730 case '|':
1731 case ')':
1732 if (flags & TRYAGAIN) {
1733 *flagp |= TRYAGAIN;
1734 return NULL;
1735 }
3280af22 1736 FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse);
a0d0e21e 1737 /* Supposed to be caught earlier. */
1738 break;
85afd4ae 1739 case '{':
3280af22 1740 if (!regcurly(PL_regcomp_parse)) {
1741 PL_regcomp_parse++;
85afd4ae 1742 goto defchar;
1743 }
1744 /* FALL THROUGH */
a0d0e21e 1745 case '?':
1746 case '+':
1747 case '*':
3115e423 1748 FAIL("?+*{} follows nothing in regexp");
a0d0e21e 1749 break;
1750 case '\\':
3280af22 1751 switch (*++PL_regcomp_parse) {
a0d0e21e 1752 case 'A':
3280af22 1753 PL_seen_zerolen++;
c277df42 1754 ret = reg_node(SBOL);
a0d0e21e 1755 *flagp |= SIMPLE;
1756 nextchar();
1757 break;
1758 case 'G':
c277df42 1759 ret = reg_node(GPOS);
3280af22 1760 PL_regseen |= REG_SEEN_GPOS;
a0d0e21e 1761 *flagp |= SIMPLE;
1762 nextchar();
1763 break;
1764 case 'Z':
c277df42 1765 ret = reg_node(SEOL);
a0d0e21e 1766 *flagp |= SIMPLE;
1767 nextchar();
1768 break;
b85d18e9 1769 case 'z':
1770 ret = reg_node(EOS);
1771 *flagp |= SIMPLE;
3280af22 1772 PL_seen_zerolen++; /* Do not optimize RE away */
b85d18e9 1773 nextchar();
1774 break;
a0ed51b3 1775 case 'C':
1776 ret = reg_node(SANY);
1777 *flagp |= HASWIDTH|SIMPLE;
1778 nextchar();
1779 break;
1780 case 'X':
1781 ret = reg_node(CLUMP);
1782 *flagp |= HASWIDTH;
1783 nextchar();
1784 if (UTF && !PL_utf8_mark)
dfe13c55 1785 is_utf8_mark((U8*)"~"); /* preload table */
a0ed51b3 1786 break;
a0d0e21e 1787 case 'w':
a0ed51b3 1788 ret = reg_node(
1789 UTF
1790 ? (LOC ? ALNUMLUTF8 : ALNUMUTF8)
1791 : (LOC ? ALNUML : ALNUM));
a0d0e21e 1792 *flagp |= HASWIDTH|SIMPLE;
1793 nextchar();
a0ed51b3 1794 if (UTF && !PL_utf8_alnum)
dfe13c55 1795 is_utf8_alnum((U8*)"a"); /* preload table */
a0d0e21e 1796 break;
1797 case 'W':
a0ed51b3 1798 ret = reg_node(
1799 UTF
1800 ? (LOC ? NALNUMLUTF8 : NALNUMUTF8)
1801 : (LOC ? NALNUML : NALNUM));
a0d0e21e 1802 *flagp |= HASWIDTH|SIMPLE;
1803 nextchar();
a0ed51b3 1804 if (UTF && !PL_utf8_alnum)
dfe13c55 1805 is_utf8_alnum((U8*)"a"); /* preload table */
a0d0e21e 1806 break;
1807 case 'b':
3280af22 1808 PL_seen_zerolen++;
f5c9036e 1809 PL_regseen |= REG_SEEN_LOOKBEHIND;
a0ed51b3 1810 ret = reg_node(
1811 UTF
1812 ? (LOC ? BOUNDLUTF8 : BOUNDUTF8)
1813 : (LOC ? BOUNDL : BOUND));
a0d0e21e 1814 *flagp |= SIMPLE;
1815 nextchar();
a0ed51b3 1816 if (UTF && !PL_utf8_alnum)
dfe13c55 1817 is_utf8_alnum((U8*)"a"); /* preload table */
a0d0e21e 1818 break;
1819 case 'B':
3280af22 1820 PL_seen_zerolen++;
f5c9036e 1821 PL_regseen |= REG_SEEN_LOOKBEHIND;
a0ed51b3 1822 ret = reg_node(
1823 UTF
1824 ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8)
1825 : (LOC ? NBOUNDL : NBOUND));
a0d0e21e 1826 *flagp |= SIMPLE;
1827 nextchar();
a0ed51b3 1828 if (UTF && !PL_utf8_alnum)
dfe13c55 1829 is_utf8_alnum((U8*)"a"); /* preload table */
a0d0e21e 1830 break;
1831 case 's':
a0ed51b3 1832 ret = reg_node(
1833 UTF
1834 ? (LOC ? SPACELUTF8 : SPACEUTF8)
1835 : (LOC ? SPACEL : SPACE));
a0d0e21e 1836 *flagp |= HASWIDTH|SIMPLE;
1837 nextchar();
a0ed51b3 1838 if (UTF && !PL_utf8_space)
dfe13c55 1839 is_utf8_space((U8*)" "); /* preload table */
a0d0e21e 1840 break;
1841 case 'S':
a0ed51b3 1842 ret = reg_node(
1843 UTF
1844 ? (LOC ? NSPACELUTF8 : NSPACEUTF8)
1845 : (LOC ? NSPACEL : NSPACE));
a0d0e21e 1846 *flagp |= HASWIDTH|SIMPLE;
1847 nextchar();
a0ed51b3 1848 if (UTF && !PL_utf8_space)
dfe13c55 1849 is_utf8_space((U8*)" "); /* preload table */
a0d0e21e 1850 break;
1851 case 'd':
a0ed51b3 1852 ret = reg_node(UTF ? DIGITUTF8 : DIGIT);
a0d0e21e 1853 *flagp |= HASWIDTH|SIMPLE;
1854 nextchar();
a0ed51b3 1855 if (UTF && !PL_utf8_digit)
dfe13c55 1856 is_utf8_digit((U8*)"1"); /* preload table */
a0d0e21e 1857 break;
1858 case 'D':
a0ed51b3 1859 ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT);
a0d0e21e 1860 *flagp |= HASWIDTH|SIMPLE;
1861 nextchar();
a0ed51b3 1862 if (UTF && !PL_utf8_digit)
dfe13c55 1863 is_utf8_digit((U8*)"1"); /* preload table */
a0d0e21e 1864 break;
a14b48bc 1865 case 'p':
1866 case 'P':
1867 { /* a lovely hack--pretend we saw [\pX] instead */
1868 char* oldregxend = PL_regxend;
1869
1870 if (PL_regcomp_parse[1] == '{') {
1871 PL_regxend = strchr(PL_regcomp_parse, '}');
1872 if (!PL_regxend)
1873 FAIL("Missing right brace on \\p{}");
1874 PL_regxend++;
1875 }
1876 else
1877 PL_regxend = PL_regcomp_parse + 2;
1878 PL_regcomp_parse--;
1879
1880 ret = regclassutf8();
1881
1882 PL_regxend = oldregxend;
1883 PL_regcomp_parse--;
1884 nextchar();
1885 *flagp |= HASWIDTH|SIMPLE;
1886 }
1887 break;
a0d0e21e 1888 case 'n':
1889 case 'r':
1890 case 't':
1891 case 'f':
1892 case 'e':
1893 case 'a':
1894 case 'x':
1895 case 'c':
1896 case '0':
1897 goto defchar;
1898 case '1': case '2': case '3': case '4':
1899 case '5': case '6': case '7': case '8': case '9':
1900 {
3280af22 1901 I32 num = atoi(PL_regcomp_parse);
a0d0e21e 1902
3280af22 1903 if (num > 9 && num >= PL_regnpar)
a0d0e21e 1904 goto defchar;
1905 else {
3280af22 1906 if (!SIZE_ONLY && num > PL_regcomp_rx->nparens)
ef64f398 1907 FAIL("reference to nonexistent group");
3280af22 1908 PL_regsawback = 1;
a0ed51b3 1909 ret = reganode(FOLD
1910 ? (LOC ? REFFL : REFF)
c8756f30 1911 : REF, num);
a0d0e21e 1912 *flagp |= HASWIDTH;
3280af22 1913 while (isDIGIT(*PL_regcomp_parse))
1914 PL_regcomp_parse++;
1915 PL_regcomp_parse--;
a0d0e21e 1916 nextchar();
1917 }
1918 }
1919 break;
1920 case '\0':
3280af22 1921 if (PL_regcomp_parse >= PL_regxend)
a0d0e21e 1922 FAIL("trailing \\ in regexp");
1923 /* FALL THROUGH */
1924 default:
c9f97d15 1925 /* Do not generate `unrecognized' warnings here, we fall
1926 back into the quick-grab loop below */
a0d0e21e 1927 goto defchar;
1928 }
1929 break;
4633a7c4 1930
1931 case '#':
3280af22 1932 if (PL_regflags & PMf_EXTENDED) {
1933 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++;
1934 if (PL_regcomp_parse < PL_regxend)
4633a7c4 1935 goto tryagain;
1936 }
1937 /* FALL THROUGH */
1938
a0d0e21e 1939 default: {
1940 register I32 len;
a0ed51b3 1941 register UV ender;
a0d0e21e 1942 register char *p;
c277df42 1943 char *oldp, *s;
a0d0e21e 1944 I32 numlen;
1945
3280af22 1946 PL_regcomp_parse++;
a0d0e21e 1947
1948 defchar:
a0ed51b3 1949 ret = reg_node(FOLD
1950 ? (LOC ? EXACTFL : EXACTF)
bbce6d69 1951 : EXACT);
161b471a 1952 s = (char *) OPERAND(ret);
c277df42 1953 regc(0, s++); /* save spot for len */
3280af22 1954 for (len = 0, p = PL_regcomp_parse - 1;
1955 len < 127 && p < PL_regxend;
a0d0e21e 1956 len++)
1957 {
1958 oldp = p;
5b5a24f7 1959
3280af22 1960 if (PL_regflags & PMf_EXTENDED)
1961 p = regwhite(p, PL_regxend);
a0d0e21e 1962 switch (*p) {
1963 case '^':
1964 case '$':
1965 case '.':
1966 case '[':
1967 case '(':
1968 case ')':
1969 case '|':
1970 goto loopdone;
1971 case '\\':
1972 switch (*++p) {
1973 case 'A':
1974 case 'G':
1975 case 'Z':
b85d18e9 1976 case 'z':
a0d0e21e 1977 case 'w':
1978 case 'W':
1979 case 'b':
1980 case 'B':
1981 case 's':
1982 case 'S':
1983 case 'd':
1984 case 'D':
a14b48bc 1985 case 'p':
1986 case 'P':
a0d0e21e 1987 --p;
1988 goto loopdone;
1989 case 'n':
1990 ender = '\n';
1991 p++;
a687059c 1992 break;
a0d0e21e 1993 case 'r':
1994 ender = '\r';
1995 p++;
a687059c 1996 break;
a0d0e21e 1997 case 't':
1998 ender = '\t';
1999 p++;
a687059c 2000 break;
a0d0e21e 2001 case 'f':
2002 ender = '\f';
2003 p++;
a687059c 2004 break;
a0d0e21e 2005 case 'e':
2006 ender = '\033';
2007 p++;
a687059c 2008 break;
a0d0e21e 2009 case 'a':
2010 ender = '\007';
2011 p++;
a687059c 2012 break;
a0d0e21e 2013 case 'x':
a0ed51b3 2014 if (*++p == '{') {
2015 char* e = strchr(p, '}');
2016
2017 if (!e)
2018 FAIL("Missing right brace on \\x{}");
2019 else if (UTF) {
2020 ender = scan_hex(p + 1, e - p, &numlen);
2021 if (numlen + len >= 127) { /* numlen is generous */
2022 p--;
2023 goto loopdone;
2024 }
2025 p = e + 1;
2026 }
2027 else
2028 FAIL("Can't use \\x{} without 'use utf8' declaration");
2029 }
2030 else {
2031 ender = scan_hex(p, 2, &numlen);
2032 p += numlen;
2033 }
a687059c 2034 break;
a0d0e21e 2035 case 'c':
2036 p++;
bbce6d69 2037 ender = UCHARAT(p++);
2038 ender = toCTRL(ender);
a687059c 2039 break;
a0d0e21e 2040 case '0': case '1': case '2': case '3':case '4':
2041 case '5': case '6': case '7': case '8':case '9':
2042 if (*p == '0' ||
3280af22 2043 (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) {
a0d0e21e 2044 ender = scan_oct(p, 3, &numlen);
2045 p += numlen;
2046 }
2047 else {
2048 --p;
2049 goto loopdone;
a687059c 2050 }
2051 break;
a0d0e21e 2052 case '\0':
3280af22 2053 if (p >= PL_regxend)
a687059c 2054 FAIL("trailing \\ in regexp");
2055 /* FALL THROUGH */
a0d0e21e 2056 default:
c9f97d15 2057 if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p))
cea2e8a9 2058 Perl_warner(aTHX_ WARN_UNSAFE,
c9f97d15 2059 "/%.127s/: Unrecognized escape \\%c passed through",
2060 PL_regprecomp,
2061 *p);
a0ed51b3 2062 goto normal_default;
a0d0e21e 2063 }
2064 break;
a687059c 2065 default:
a0ed51b3 2066 normal_default:
2067 if ((*p & 0xc0) == 0xc0 && UTF) {
dfe13c55 2068 ender = utf8_to_uv((U8*)p, &numlen);
a0ed51b3 2069 p += numlen;
2070 }
2071 else
2072 ender = *p++;
a0d0e21e 2073 break;
a687059c 2074 }
3280af22 2075 if (PL_regflags & PMf_EXTENDED)
2076 p = regwhite(p, PL_regxend);
a0ed51b3 2077 if (UTF && FOLD) {
2078 if (LOC)
2079 ender = toLOWER_LC_uni(ender);
2080 else
2081 ender = toLOWER_uni(ender);
2082 }
a0d0e21e 2083 if (ISMULT2(p)) { /* Back off on ?+*. */
2084 if (len)
2085 p = oldp;
a0ed51b3 2086 else if (ender >= 0x80 && UTF) {
2087 reguni(ender, s, &numlen);
2088 s += numlen;
2089 len += numlen;
2090 }
a0d0e21e 2091 else {
2092 len++;
c277df42 2093 regc(ender, s++);
a0d0e21e 2094 }
2095 break;
a687059c 2096 }
a0ed51b3 2097 if (ender >= 0x80 && UTF) {
2098 reguni(ender, s, &numlen);
2099 s += numlen;
2100 len += numlen - 1;
2101 }
2102 else
2103 regc(ender, s++);
a0d0e21e 2104 }
2105 loopdone:
3280af22 2106 PL_regcomp_parse = p - 1;
a0d0e21e 2107 nextchar();
2108 if (len < 0)
2109 FAIL("internal disaster in regexp");
2110 if (len > 0)
2111 *flagp |= HASWIDTH;
2112 if (len == 1)
2113 *flagp |= SIMPLE;
c277df42 2114 if (!SIZE_ONLY)
a0d0e21e 2115 *OPERAND(ret) = len;
c277df42 2116 regc('\0', s++);
2117 if (SIZE_ONLY) {
3280af22 2118 PL_regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
a0ed51b3 2119 }
2120 else {
3280af22 2121 PL_regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
c277df42 2122 }
a687059c 2123 }
a0d0e21e 2124 break;
2125 }
a687059c 2126
a0d0e21e 2127 return(ret);
a687059c 2128}
2129
873ef191 2130STATIC char *
cea2e8a9 2131S_regwhite(pTHX_ char *p, char *e)
5b5a24f7 2132{
2133 while (p < e) {
2134 if (isSPACE(*p))
2135 ++p;
2136 else if (*p == '#') {
2137 do {
2138 p++;
2139 } while (p < e && *p != '\n');
2140 }
2141 else
2142 break;
2143 }
2144 return p;
2145}
2146
b8c5462f 2147/* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
2148 Character classes ([:foo:]) can also be negated ([:^foo:]).
2149 Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
2150 Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
2151 but trigger warnings because they are currently unimplemented. */
2152STATIC I32
cea2e8a9 2153S_regpposixcc(pTHX_ I32 value)
620e46c5 2154{
11b8faa4 2155 dTHR;
620e46c5 2156 char *posixcc = 0;
b8c5462f 2157 I32 namedclass = -1;
620e46c5 2158
2159 if (value == '[' && PL_regcomp_parse + 1 < PL_regxend &&
2160 /* I smell either [: or [= or [. -- POSIX has been here, right? */
2161 (*PL_regcomp_parse == ':' ||
2162 *PL_regcomp_parse == '=' ||
2163 *PL_regcomp_parse == '.')) {
2164 char c = *PL_regcomp_parse;
2165 char* s = PL_regcomp_parse++;
2166
2167 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c)
2168 PL_regcomp_parse++;
2169 if (PL_regcomp_parse == PL_regxend)
2170 /* Grandfather lone [:, [=, [. */
2171 PL_regcomp_parse = s;
2172 else {
b8c5462f 2173 char* t = PL_regcomp_parse++; /* skip over the c */
2174
2175 if (*PL_regcomp_parse == ']') {
2176 PL_regcomp_parse++; /* skip over the ending ] */
2177 posixcc = s + 1;
2178 if (*s == ':') {
2179 I32 complement = *posixcc == '^' ? *posixcc++ : 0;
2180 I32 skip = 5; /* the most common skip */
2181
2182 switch (*posixcc) {
2183 case 'a':
2184 if (strnEQ(posixcc, "alnum", 5))
2185 namedclass =
2186 complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
2187 else if (strnEQ(posixcc, "alpha", 5))
2188 namedclass =
2189 complement ? ANYOF_NALPHA : ANYOF_ALPHA;
2190 else if (strnEQ(posixcc, "ascii", 5))
2191 namedclass =
2192 complement ? ANYOF_NASCII : ANYOF_ASCII;
2193 break;
2194 case 'c':
2195 if (strnEQ(posixcc, "cntrl", 5))
2196 namedclass =
2197 complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
2198 break;
2199 case 'd':
2200 if (strnEQ(posixcc, "digit", 5))
2201 namedclass =
2202 complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
2203 break;
2204 case 'g':
2205 if (strnEQ(posixcc, "graph", 5))
2206 namedclass =
2207 complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
2208 break;
2209 case 'l':
2210 if (strnEQ(posixcc, "lower", 5))
2211 namedclass =
2212 complement ? ANYOF_NLOWER : ANYOF_LOWER;
2213 break;
2214 case 'p':
2215 if (strnEQ(posixcc, "print", 5))
2216 namedclass =
2217 complement ? ANYOF_NPRINT : ANYOF_PRINT;
2218 else if (strnEQ(posixcc, "punct", 5))
2219 namedclass =
2220 complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
2221 break;
2222 case 's':
2223 if (strnEQ(posixcc, "space", 5))
2224 namedclass =
2225 complement ? ANYOF_NSPACE : ANYOF_SPACE;
2226 case 'u':
2227 if (strnEQ(posixcc, "upper", 5))
2228 namedclass =
2229 complement ? ANYOF_NUPPER : ANYOF_UPPER;
2230 break;
2231 case 'w': /* this is not POSIX, this is the Perl \w */
2232 if (strnEQ(posixcc, "word", 4)) {
2233 namedclass =
2234 complement ? ANYOF_NALNUM : ANYOF_ALNUM;
2235 skip = 4;
2236 }
2237 break;
2238 case 'x':
2239 if (strnEQ(posixcc, "xdigit", 6)) {
2240 namedclass =
2241 complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
2242 skip = 6;
2243 }
2244 break;
2245 }
2246 if ((namedclass == -1 ||
2247 !(posixcc + skip + 2 < PL_regxend &&
2248 (posixcc[skip] == ':' &&
2249 posixcc[skip + 1] == ']'))))
2250 Perl_croak(aTHX_ "Character class [:%.*s:] unknown",
2251 t - s - 1, s + 1);
2252 } else if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
2253 /* [[=foo=]] and [[.foo.]] are still future. */
cea2e8a9 2254 Perl_warner(aTHX_ WARN_UNSAFE,
b8c5462f 2255 "Character class syntax [%c %c] is reserved for future extensions", c, c);
2256 } else {
2257 /* Maternal grandfather:
2258 * "[:" ending in ":" but not in ":]" */
767d463e 2259 PL_regcomp_parse = s;
2260 }
620e46c5 2261 }
2262 }
2263
b8c5462f 2264 return namedclass;
2265}
2266
2267STATIC void
2268S_checkposixcc(pTHX)
2269{
2270 if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY &&
2271 (*PL_regcomp_parse == ':' ||
2272 *PL_regcomp_parse == '=' ||
2273 *PL_regcomp_parse == '.')) {
2274 char *s = PL_regcomp_parse;
2275 char c = *s++;
2276
2277 while(*s && isALNUM(*s))
2278 s++;
2279 if (*s && c == *s && s[1] == ']') {
2280 Perl_warner(aTHX_ WARN_UNSAFE,
2281 "Character class syntax [%c %c] belongs inside character classes", c, c);
2282 if (c == '=' || c == '.')
2283 Perl_warner(aTHX_ WARN_UNSAFE,
2284 "Character class syntax [%c %c] is reserved for future extensions", c, c);
2285 }
2286 }
620e46c5 2287}
2288
76e3520e 2289STATIC regnode *
cea2e8a9 2290S_regclass(pTHX)
a687059c 2291{
5c0ca799 2292 dTHR;
c277df42 2293 register char *opnd, *s;
a0ed51b3 2294 register I32 value;
b8c5462f 2295 register I32 lastvalue = OOB_CHAR8;
a0d0e21e 2296 register I32 range = 0;
c277df42 2297 register regnode *ret;
a0d0e21e 2298 register I32 def;
2299 I32 numlen;
b8c5462f 2300 I32 namedclass;
a0d0e21e 2301
3280af22 2302 s = opnd = (char *) OPERAND(PL_regcode);
c277df42 2303 ret = reg_node(ANYOF);
b8c5462f 2304 for (value = 0; value < ANYOF_SIZE; value++)
c277df42 2305 regc(0, s++);
3280af22 2306 if (*PL_regcomp_parse == '^') { /* Complement of range. */
2307 PL_regnaughty++;
2308 PL_regcomp_parse++;
c277df42 2309 if (!SIZE_ONLY)
b8c5462f 2310 ANYOF_FLAGS(opnd) |= ANYOF_INVERT;
bbce6d69 2311 }
c277df42 2312 if (!SIZE_ONLY) {
3280af22 2313 PL_regcode += ANY_SKIP;
a0ed51b3 2314 if (FOLD)
b8c5462f 2315 ANYOF_FLAGS(opnd) |= ANYOF_FOLD;
a0ed51b3 2316 if (LOC)
b8c5462f 2317 ANYOF_FLAGS(opnd) |= ANYOF_LOCALE;
a0ed51b3 2318 }
2319 else {
3280af22 2320 PL_regsize += ANY_SKIP;
a0d0e21e 2321 }
b8c5462f 2322
2323 checkposixcc();
2324
3280af22 2325 if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-')
a0d0e21e 2326 goto skipcond; /* allow 1st char to be ] or - */
3280af22 2327 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') {
a0d0e21e 2328 skipcond:
b8c5462f 2329 namedclass = -1;
a0ed51b3 2330 value = UCHARAT(PL_regcomp_parse++);
620e46c5 2331 if (value == '[')
b8c5462f 2332 namedclass = regpposixcc(value);
620e46c5 2333 else if (value == '\\') {
a0ed51b3 2334 value = UCHARAT(PL_regcomp_parse++);
2335 switch (value) {
b8c5462f 2336 case 'w': namedclass = ANYOF_ALNUM; break;
2337 case 'W': namedclass = ANYOF_NALNUM; break;
2338 case 's': namedclass = ANYOF_SPACE; break;
2339 case 'S': namedclass = ANYOF_NSPACE; break;
2340 case 'd': namedclass = ANYOF_DIGIT; break;
2341 case 'D': namedclass = ANYOF_NDIGIT; break;
2342 case 'n': value = '\n'; break;
2343 case 'r': value = '\r'; break;
2344 case 't': value = '\t'; break;
2345 case 'f': value = '\f'; break;
2346 case 'b': value = '\b'; break;
2347 case 'e': value = '\033'; break;
2348 case 'a': value = '\007'; break;
2349 case 'x':
2350 value = scan_hex(PL_regcomp_parse, 2, &numlen);
2351 PL_regcomp_parse += numlen;
2352 break;
2353 case 'c':
2354 value = UCHARAT(PL_regcomp_parse++);
2355 value = toCTRL(value);
2356 break;
2357 case '0': case '1': case '2': case '3': case '4':
2358 case '5': case '6': case '7': case '8': case '9':
2359 value = scan_oct(--PL_regcomp_parse, 3, &numlen);
2360 PL_regcomp_parse += numlen;
2361 break;
2362 }
2363 }
2364 if (!SIZE_ONLY && namedclass > -1) {
2365 switch (namedclass) {
2366 case ANYOF_ALNUM:
2367 if (LOC)
2368 ANYOF_CLASS_SET(opnd, ANYOF_ALNUM);
2369 else {
2370 for (value = 0; value < 256; value++)
2371 if (isALNUM(value))
2372 ANYOF_BITMAP_SET(opnd, value);
bbce6d69 2373 }
b8c5462f 2374 break;
2375 case ANYOF_NALNUM:
2376 if (LOC)
2377 ANYOF_CLASS_SET(opnd, ANYOF_NALNUM);
2378 else {
2379 for (value = 0; value < 256; value++)
2380 if (!isALNUM(value))
2381 ANYOF_BITMAP_SET(opnd, value);
bbce6d69 2382 }
b8c5462f 2383 break;
2384 case ANYOF_SPACE:
2385 if (LOC)
2386 ANYOF_CLASS_SET(opnd, ANYOF_SPACE);
2387 else {
2388 for (value = 0; value < 256; value++)
2389 if (isSPACE(value))
2390 ANYOF_BITMAP_SET(opnd, value);
bbce6d69 2391 }
b8c5462f 2392 break;
2393 case ANYOF_NSPACE:
2394 if (LOC)
2395 ANYOF_CLASS_SET(opnd, ANYOF_NSPACE);
2396 else {
2397 for (value = 0; value < 256; value++)
2398 if (!isSPACE(value))
2399 ANYOF_BITMAP_SET(opnd, value);
bbce6d69 2400 }
b8c5462f 2401 break;
2402 case ANYOF_DIGIT:
2403 if (LOC)
2404 ANYOF_CLASS_SET(opnd, ANYOF_DIGIT);
2405 else {
a0ed51b3 2406 for (value = '0'; value <= '9'; value++)
b8c5462f 2407 ANYOF_BITMAP_SET(opnd, value);
ae5c130c 2408 }
b8c5462f 2409 break;
2410 case ANYOF_NDIGIT:
2411 if (LOC)
2412 ANYOF_CLASS_SET(opnd, ANYOF_NDIGIT);
2413 else {
a0ed51b3 2414 for (value = 0; value < '0'; value++)
b8c5462f 2415 ANYOF_BITMAP_SET(opnd, value);
a0ed51b3 2416 for (value = '9' + 1; value < 256; value++)
b8c5462f 2417 ANYOF_BITMAP_SET(opnd, value);
2418 }
2419 break;
2420 case ANYOF_NALNUMC:
2421 if (LOC)
2422 ANYOF_CLASS_SET(opnd, ANYOF_NALNUMC);
2423 else {
2424 for (value = 0; value < 256; value++)
2425 if (!isALNUMC(value))
2426 ANYOF_BITMAP_SET(opnd, value);
2427 }
2428 break;
2429 case ANYOF_ALNUMC:
2430 if (LOC)
2431 ANYOF_CLASS_SET(opnd, ANYOF_ALNUMC);
2432 else {
2433 for (value = 0; value < 256; value++)
2434 if (isALNUMC(value))
2435 ANYOF_BITMAP_SET(opnd, value);
2436 }
2437 break;
2438 case ANYOF_ALPHA:
2439 if (LOC)
2440 ANYOF_CLASS_SET(opnd, ANYOF_ALPHA);
2441 else {
2442 for (value = 0; value < 256; value++)
2443 if (isALPHA(value))
2444 ANYOF_BITMAP_SET(opnd, value);
ae5c130c 2445 }
a0d0e21e 2446 break;
b8c5462f 2447 case ANYOF_NALPHA:
2448 if (LOC)
2449 ANYOF_CLASS_SET(opnd, ANYOF_NALPHA);
2450 else {
2451 for (value = 0; value < 256; value++)
2452 if (!isALPHA(value))
2453 ANYOF_BITMAP_SET(opnd, value);
2454 }
a0d0e21e 2455 break;
b8c5462f 2456 case ANYOF_ASCII:
2457 if (LOC)
2458 ANYOF_CLASS_SET(opnd, ANYOF_ASCII);
2459 else {
2460 for (value = 0; value < 128; value++)
2461 ANYOF_BITMAP_SET(opnd, value);
2462 }
a0d0e21e 2463 break;
b8c5462f 2464 case ANYOF_NASCII:
2465 if (LOC)
2466 ANYOF_CLASS_SET(opnd, ANYOF_NASCII);
2467 else {
2468 for (value = 128; value < 256; value++)
2469 ANYOF_BITMAP_SET(opnd, value);
2470 }
a0d0e21e 2471 break;
b8c5462f 2472 case ANYOF_CNTRL:
2473 if (LOC)
2474 ANYOF_CLASS_SET(opnd, ANYOF_CNTRL);
2475 else {
2476 for (value = 0; value < 256; value++)
2477 if (isCNTRL(value))
2478 ANYOF_BITMAP_SET(opnd, value);
2479 }
2480 lastvalue = OOB_CHAR8;
a0d0e21e 2481 break;
b8c5462f 2482 case ANYOF_NCNTRL:
2483 if (LOC)
2484 ANYOF_CLASS_SET(opnd, ANYOF_NCNTRL);
2485 else {
2486 for (value = 0; value < 256; value++)
2487 if (!isCNTRL(value))
2488 ANYOF_BITMAP_SET(opnd, value);
2489 }
a0d0e21e 2490 break;
b8c5462f 2491 case ANYOF_GRAPH:
2492 if (LOC)
2493 ANYOF_CLASS_SET(opnd, ANYOF_GRAPH);
2494 else {
2495 for (value = 0; value < 256; value++)
2496 if (isGRAPH(value))
2497 ANYOF_BITMAP_SET(opnd, value);
2498 }
a0d0e21e 2499 break;
b8c5462f 2500 case ANYOF_NGRAPH:
2501 if (LOC)
2502 ANYOF_CLASS_SET(opnd, ANYOF_NGRAPH);
2503 else {
2504 for (value = 0; value < 256; value++)
2505 if (!isGRAPH(value))
2506 ANYOF_BITMAP_SET(opnd, value);
2507 }
a0d0e21e 2508 break;
b8c5462f 2509 case ANYOF_LOWER:
2510 if (LOC)
2511 ANYOF_CLASS_SET(opnd, ANYOF_LOWER);
2512 else {
2513 for (value = 0; value < 256; value++)
2514 if (isLOWER(value))
2515 ANYOF_BITMAP_SET(opnd, value);
2516 }
a0d0e21e 2517 break;
b8c5462f 2518 case ANYOF_NLOWER:
2519 if (LOC)
2520 ANYOF_CLASS_SET(opnd, ANYOF_NLOWER);
2521 else {
2522 for (value = 0; value < 256; value++)
2523 if (!isLOWER(value))
2524 ANYOF_BITMAP_SET(opnd, value);
2525 }
2526 break;
2527 case ANYOF_PRINT:
2528 if (LOC)
2529 ANYOF_CLASS_SET(opnd, ANYOF_PRINT);
2530 else {
2531 for (value = 0; value < 256; value++)
2532 if (isPRINT(value))
2533 ANYOF_BITMAP_SET(opnd, value);
2534 }
2535 break;
2536 case ANYOF_NPRINT:
2537 if (LOC)
2538 ANYOF_CLASS_SET(opnd, ANYOF_NPRINT);
2539 else {
2540 for (value = 0; value < 256; value++)
2541 if (!isPRINT(value))
2542 ANYOF_BITMAP_SET(opnd, value);
2543 }
2544 break;
2545 case ANYOF_PUNCT:
2546 if (LOC)
2547 ANYOF_CLASS_SET(opnd, ANYOF_PUNCT);
2548 else {
2549 for (value = 0; value < 256; value++)
2550 if (isPUNCT(value))
2551 ANYOF_BITMAP_SET(opnd, value);
2552 }
2553 break;
2554 case ANYOF_NPUNCT:
2555 if (LOC)
2556 ANYOF_CLASS_SET(opnd, ANYOF_NPUNCT);
2557 else {
2558 for (value = 0; value < 256; value++)
2559 if (!isPUNCT(value))
2560 ANYOF_BITMAP_SET(opnd, value);
2561 }
2562 break;
2563 case ANYOF_UPPER:
2564 if (LOC)
2565 ANYOF_CLASS_SET(opnd, ANYOF_UPPER);
2566 else {
2567 for (value = 0; value < 256; value++)
2568 if (isUPPER(value))
2569 ANYOF_BITMAP_SET(opnd, value);
2570 }
2571 break;
2572 case ANYOF_NUPPER:
2573 if (LOC)
2574 ANYOF_CLASS_SET(opnd, ANYOF_NUPPER);
2575 else {
2576 for (value = 0; value < 256; value++)
2577 if (!isUPPER(value))
2578 ANYOF_BITMAP_SET(opnd, value);
2579 }
2580 break;
2581 case ANYOF_XDIGIT:
2582 if (LOC)
2583 ANYOF_CLASS_SET(opnd, ANYOF_XDIGIT);
2584 else {
2585 for (value = 0; value < 256; value++)
2586 if (isXDIGIT(value))
2587 ANYOF_BITMAP_SET(opnd, value);
2588 }
2589 break;
2590 case ANYOF_NXDIGIT:
2591 if (LOC)
2592 ANYOF_CLASS_SET(opnd, ANYOF_NXDIGIT);
2593 else {
2594 for (value = 0; value < 256; value++)
2595 if (!isXDIGIT(value))
2596 ANYOF_BITMAP_SET(opnd, value);
2597 }
2598 break;
2599 default:
2600 FAIL("invalid [::] class in regexp");
a0d0e21e 2601 break;
2602 }
b8c5462f 2603 if (LOC)
2604 ANYOF_FLAGS(opnd) |= ANYOF_CLASS;
2605 lastvalue = OOB_CHAR8;
a0d0e21e 2606 }
b8c5462f 2607 else
a0d0e21e 2608 if (range) {
a0ed51b3 2609 if (lastvalue > value)
a0d0e21e 2610 FAIL("invalid [] range in regexp");
2611 range = 0;
2612 }
2613 else {
a0ed51b3 2614 lastvalue = value;
3280af22 2615 if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend &&
2616 PL_regcomp_parse[1] != ']') {
2617 PL_regcomp_parse++;
a0d0e21e 2618 range = 1;
2619 continue; /* do it next time */
2620 }
a687059c 2621 }
ae5c130c 2622 if (!SIZE_ONLY) {
8ada0baa 2623#ifndef ASCIIish
2624 if ((isLOWER(lastvalue) && isLOWER(value)) ||
c8dba6f3 2625 (isUPPER(lastvalue) && isUPPER(value)))
2626 {
2627 I32 i;
8ada0baa 2628 if (isLOWER(lastvalue)) {
2629 for (i = lastvalue; i <= value; i++)
2630 if (isLOWER(i))
b8c5462f 2631 ANYOF_BITMAP_SET(opnd, i);
8ada0baa 2632 } else {
2633 for (i = lastvalue; i <= value; i++)
2634 if (isUPPER(i))
b8c5462f 2635 ANYOF_BITMAP_SET(opnd, i);
8ada0baa 2636 }
2637 }
2638 else
2639#endif
2640 for ( ; lastvalue <= value; lastvalue++)
b8c5462f 2641 ANYOF_BITMAP_SET(opnd, lastvalue);
8ada0baa 2642 }
a0ed51b3 2643 lastvalue = value;
a0d0e21e 2644 }
ae5c130c 2645 /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
b8c5462f 2646 if (!SIZE_ONLY &&
2647 (ANYOF_FLAGS(opnd) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD) {
a0ed51b3 2648 for (value = 0; value < 256; ++value) {
b8c5462f 2649 if (ANYOF_BITMAP_TEST(opnd, value)) {
22c35a8c 2650 I32 cf = PL_fold[value];
b8c5462f 2651 ANYOF_BITMAP_SET(opnd, cf);
ae5c130c 2652 }
2653 }
b8c5462f 2654 ANYOF_FLAGS(opnd) &= ~ANYOF_FOLD;
ae5c130c 2655 }
2656 /* optimize inverted simple patterns (e.g. [^a-z]) */
b8c5462f 2657 if (!SIZE_ONLY && (ANYOF_FLAGS(opnd) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
2658 for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
2659 opnd[ANYOF_BITMAP_OFFSET + value] ^= ANYOF_FLAGS_ALL;
2660 ANYOF_FLAGS(opnd) = 0;
ae5c130c 2661 }
a0d0e21e 2662 return ret;
2663}
2664
a0ed51b3 2665STATIC regnode *
cea2e8a9 2666S_regclassutf8(pTHX)
a0ed51b3 2667{
b8c5462f 2668 dTHR;
a0ed51b3 2669 register char *opnd, *e;
2670 register U32 value;
b8c5462f 2671 register U32 lastvalue = OOB_UTF8;
a0ed51b3 2672 register I32 range = 0;
2673 register regnode *ret;
2674 I32 numlen;
2675 I32 n;
2676 SV *listsv;
2677 U8 flags = 0;
b8c5462f 2678 I32 namedclass;
a0ed51b3 2679
2680 if (*PL_regcomp_parse == '^') { /* Complement of range. */
2681 PL_regnaughty++;
2682 PL_regcomp_parse++;
2683 if (!SIZE_ONLY)
2684 flags |= ANYOF_INVERT;
2685 }
2686 if (!SIZE_ONLY) {
2687 if (FOLD)
2688 flags |= ANYOF_FOLD;
2689 if (LOC)
2690 flags |= ANYOF_LOCALE;
79cb57f6 2691 listsv = newSVpvn("# comment\n",10);
a0ed51b3 2692 }
2693
b8c5462f 2694 checkposixcc();
2695
a0ed51b3 2696 if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-')
2697 goto skipcond; /* allow 1st char to be ] or - */
2698
2699 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') {
2700 skipcond:
b8c5462f 2701 namedclass = -1;
dfe13c55 2702 value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
a0ed51b3 2703 PL_regcomp_parse += numlen;
2704
620e46c5 2705 if (value == '[')
b8c5462f 2706 namedclass = regpposixcc(value);
620e46c5 2707 else if (value == '\\') {
dfe13c55 2708 value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
a0ed51b3 2709 PL_regcomp_parse += numlen;
2710 switch (value) {
b8c5462f 2711 case 'w': namedclass = ANYOF_ALNUM; break;
2712 case 'W': namedclass = ANYOF_NALNUM; break;
2713 case 's': namedclass = ANYOF_SPACE; break;
2714 case 'S': namedclass = ANYOF_NSPACE; break;
2715 case 'd': namedclass = ANYOF_DIGIT; break;
2716 case 'D': namedclass = ANYOF_NDIGIT; break;
a0ed51b3 2717 case 'p':
2718 case 'P':
2719 if (*PL_regcomp_parse == '{') {
2720 e = strchr(PL_regcomp_parse++, '}');
2721 if (!e)
2722 FAIL("Missing right brace on \\p{}");
2723 n = e - PL_regcomp_parse;
2724 }
2725 else {
2726 e = PL_regcomp_parse;
2727 n = 1;
2728 }
2729 if (!SIZE_ONLY) {
2730 if (value == 'p')
b8c5462f 2731 Perl_sv_catpvf(aTHX_ listsv,
2732 "+utf8::%.*s\n", n, PL_regcomp_parse);
a0ed51b3 2733 else
cea2e8a9 2734 Perl_sv_catpvf(aTHX_ listsv,
b8c5462f 2735 "!utf8::%.*s\n", n, PL_regcomp_parse);
a0ed51b3 2736 }
2737 PL_regcomp_parse = e + 1;
b8c5462f 2738 lastvalue = OOB_UTF8;
a0ed51b3 2739 continue;
b8c5462f 2740 case 'n': value = '\n'; break;
2741 case 'r': value = '\r'; break;
2742 case 't': value = '\t'; break;
2743 case 'f': value = '\f'; break;
2744 case 'b': value = '\b'; break;
2745 case 'e': value = '\033'; break;
2746 case 'a': value = '\007'; break;
a0ed51b3 2747 case 'x':
2748 if (*PL_regcomp_parse == '{') {
2749 e = strchr(PL_regcomp_parse++, '}');
2750 if (!e)
2751 FAIL("Missing right brace on \\x{}");
b8c5462f 2752 value = scan_hex(PL_regcomp_parse,
2753 e - PL_regcomp_parse,
2754 &numlen);
a0ed51b3 2755 PL_regcomp_parse = e + 1;
2756 }
2757 else {
2758 value = scan_hex(PL_regcomp_parse, 2, &numlen);
2759 PL_regcomp_parse += numlen;
2760 }
2761 break;
2762 case 'c':
2763 value = UCHARAT(PL_regcomp_parse++);
2764 value = toCTRL(value);
2765 break;
2766 case '0': case '1': case '2': case '3': case '4':
2767 case '5': case '6': case '7': case '8': case '9':
2768 value = scan_oct(--PL_regcomp_parse, 3, &numlen);
2769 PL_regcomp_parse += numlen;
2770 break;
2771 }
2772 }
b8c5462f 2773 if (!SIZE_ONLY && namedclass > -1) {
2774 switch (namedclass) {
2775 case ANYOF_ALNUM:
2776 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n"); break;
2777 case ANYOF_NALNUM:
2778 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n"); break;
2779 case ANYOF_ALNUMC:
2780 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n"); break;
2781 case ANYOF_NALNUMC:
2782 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n"); break;
2783 case ANYOF_ALPHA:
2784 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n"); break;
2785 case ANYOF_NALPHA:
2786 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n"); break;
2787 case ANYOF_ASCII:
2788 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n"); break;
2789 case ANYOF_NASCII:
2790 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n"); break;
2791 case ANYOF_CNTRL:
2792 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n"); break;
2793 case ANYOF_NCNTRL:
2794 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n"); break;
2795 case ANYOF_GRAPH:
2796 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n"); break;
2797 case ANYOF_NGRAPH:
2798 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n"); break;
2799 case ANYOF_DIGIT:
2800 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n"); break;
2801 case ANYOF_NDIGIT:
2802 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n"); break;
2803 case ANYOF_LOWER:
2804 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n"); break;
2805 case ANYOF_NLOWER:
2806 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n"); break;
2807 case ANYOF_PRINT:
2808 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n"); break;
2809 case ANYOF_NPRINT:
2810 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n"); break;
2811 case ANYOF_PUNCT:
2812 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n"); break;
2813 case ANYOF_NPUNCT:
2814 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n"); break;
2815 case ANYOF_SPACE:
2816 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n"); break;
2817 case ANYOF_NSPACE:
2818 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n"); break;
2819 case ANYOF_UPPER:
2820 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n"); break;
2821 case ANYOF_NUPPER:
2822 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n"); break;
2823 case ANYOF_XDIGIT:
2824 Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n"); break;
2825 case ANYOF_NXDIGIT:
2826 Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n"); break;
2827 }
2828 }
2829 else
2830 if (range) {
a0ed51b3 2831 if (lastvalue > value)
2832 FAIL("invalid [] range in regexp");
2833 if (!SIZE_ONLY)
cea2e8a9 2834 Perl_sv_catpvf(aTHX_ listsv, "%04x\t%04x\n", lastvalue, value);
a0ed51b3 2835 lastvalue = value;
2836 range = 0;
2837 }
2838 else {
2839 lastvalue = value;
2840 if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend &&
2841 PL_regcomp_parse[1] != ']') {
2842 PL_regcomp_parse++;
2843 range = 1;
2844 continue; /* do it next time */
2845 }
2846 if (!SIZE_ONLY)
cea2e8a9 2847 Perl_sv_catpvf(aTHX_ listsv, "%04x\n", value);
a0ed51b3 2848 }
2849 }
a0ed51b3 2850
2851 ret = reganode(ANYOFUTF8, 0);
2852
2853 if (!SIZE_ONLY) {
2854 SV *rv = swash_init("utf8", "", listsv, 1, 0);
2855 SvREFCNT_dec(listsv);
2856 n = add_data(1,"s");
2857 PL_regcomp_rx->data->data[n] = (void*)rv;
2858 ARG1_SET(ret, flags);
2859 ARG2_SET(ret, n);
2860 }
2861
2862 return ret;
2863}
2864
76e3520e 2865STATIC char*
cea2e8a9 2866S_nextchar(pTHX)
a0d0e21e 2867{
5c0ca799 2868 dTHR;
3280af22 2869 char* retval = PL_regcomp_parse++;
a0d0e21e 2870
4633a7c4 2871 for (;;) {
3280af22 2872 if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' &&
2873 PL_regcomp_parse[2] == '#') {
2874 while (*PL_regcomp_parse && *PL_regcomp_parse != ')')
2875 PL_regcomp_parse++;
2876 PL_regcomp_parse++;
4633a7c4 2877 continue;
2878 }
3280af22 2879 if (PL_regflags & PMf_EXTENDED) {
2880 if (isSPACE(*PL_regcomp_parse)) {
2881 PL_regcomp_parse++;
748a9306 2882 continue;
2883 }
3280af22 2884 else if (*PL_regcomp_parse == '#') {
2885 while (*PL_regcomp_parse && *PL_regcomp_parse != '\n')
2886 PL_regcomp_parse++;
2887 PL_regcomp_parse++;
748a9306 2888 continue;
2889 }
748a9306 2890 }
4633a7c4 2891 return retval;
a0d0e21e 2892 }
a687059c 2893}
2894
2895/*
c277df42 2896- reg_node - emit a node
a0d0e21e 2897*/
76e3520e 2898STATIC regnode * /* Location. */
cea2e8a9 2899S_reg_node(pTHX_ U8 op)
a687059c 2900{
5c0ca799 2901 dTHR;
c277df42 2902 register regnode *ret;
2903 register regnode *ptr;
a687059c 2904
3280af22 2905 ret = PL_regcode;
c277df42 2906 if (SIZE_ONLY) {
6b88bc9c 2907 SIZE_ALIGN(PL_regsize);
3280af22 2908 PL_regsize += 1;
a0d0e21e 2909 return(ret);
2910 }
a687059c 2911
c277df42 2912 NODE_ALIGN_FILL(ret);
a0d0e21e 2913 ptr = ret;
c277df42 2914 FILL_ADVANCE_NODE(ptr, op);
3280af22 2915 PL_regcode = ptr;
a687059c 2916
a0d0e21e 2917 return(ret);
a687059c 2918}
2919
2920/*
a0d0e21e 2921- reganode - emit a node with an argument
2922*/
76e3520e 2923STATIC regnode * /* Location. */
cea2e8a9 2924S_reganode(pTHX_ U8 op, U32 arg)
fe14fcc3 2925{
5c0ca799 2926 dTHR;
c277df42 2927 register regnode *ret;
2928 register regnode *ptr;
fe14fcc3 2929
3280af22 2930 ret = PL_regcode;
c277df42 2931 if (SIZE_ONLY) {
6b88bc9c 2932 SIZE_ALIGN(PL_regsize);
3280af22 2933 PL_regsize += 2;
a0d0e21e 2934 return(ret);
2935 }
fe14fcc3 2936
c277df42 2937 NODE_ALIGN_FILL(ret);
a0d0e21e 2938 ptr = ret;
c277df42 2939 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
3280af22 2940 PL_regcode = ptr;
fe14fcc3 2941
a0d0e21e 2942 return(ret);
fe14fcc3 2943}
2944
2945/*
a0ed51b3 2946- regc - emit (if appropriate) a Unicode character
2947*/
2948STATIC void
cea2e8a9 2949S_reguni(pTHX_ UV uv, char* s, I32* lenp)
a0ed51b3 2950{
c485e607 2951 dTHR;
a0ed51b3 2952 if (SIZE_ONLY) {
dfe13c55 2953 U8 tmpbuf[10];
a0ed51b3 2954 *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf;
2955 }
2956 else
dfe13c55 2957 *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s;
a0ed51b3 2958
2959}
2960
2961/*
a0d0e21e 2962- regc - emit (if appropriate) a byte of code
2963*/
76e3520e 2964STATIC void
cea2e8a9 2965S_regc(pTHX_ U8 b, char* s)
a687059c 2966{
5c0ca799 2967 dTHR;
c277df42 2968 if (!SIZE_ONLY)
2969 *s = b;
a687059c 2970}
2971
2972/*
a0d0e21e 2973- reginsert - insert an operator in front of already-emitted operand
2974*
2975* Means relocating the operand.
2976*/
76e3520e 2977STATIC void
cea2e8a9 2978S_reginsert(pTHX_ U8 op, regnode *opnd)
a687059c 2979{
5c0ca799 2980 dTHR;
c277df42 2981 register regnode *src;
2982 register regnode *dst;
2983 register regnode *place;
2984 register int offset = regarglen[(U8)op];
2985
22c35a8c 2986/* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
c277df42 2987
2988 if (SIZE_ONLY) {
3280af22 2989 PL_regsize += NODE_STEP_REGNODE + offset;
a0d0e21e 2990 return;
2991 }
a687059c 2992
3280af22 2993 src = PL_regcode;
2994 PL_regcode += NODE_STEP_REGNODE + offset;
2995 dst = PL_regcode;
a0d0e21e 2996 while (src > opnd)
c277df42 2997 StructCopy(--src, --dst, regnode);
a0d0e21e 2998
2999 place = opnd; /* Op node, where operand used to be. */
c277df42 3000 src = NEXTOPER(place);
3001 FILL_ADVANCE_NODE(place, op);
3002 Zero(src, offset, regnode);
a687059c 3003}
3004
3005/*
c277df42 3006- regtail - set the next-pointer at the end of a node chain of p to val.
a0d0e21e 3007*/
76e3520e 3008STATIC void
cea2e8a9 3009S_regtail(pTHX_ regnode *p, regnode *val)
a687059c 3010{
5c0ca799 3011 dTHR;
c277df42 3012 register regnode *scan;
3013 register regnode *temp;
a0d0e21e 3014 register I32 offset;
3015
c277df42 3016 if (SIZE_ONLY)
a0d0e21e 3017 return;
3018
3019 /* Find last node. */
3020 scan = p;
3021 for (;;) {
3022 temp = regnext(scan);
3023 if (temp == NULL)
3024 break;
3025 scan = temp;
3026 }
a687059c 3027
c277df42 3028 if (reg_off_by_arg[OP(scan)]) {
3029 ARG_SET(scan, val - scan);
a0ed51b3 3030 }
3031 else {
c277df42 3032 NEXT_OFF(scan) = val - scan;
3033 }
a687059c 3034}
3035
3036/*
a0d0e21e 3037- regoptail - regtail on operand of first argument; nop if operandless
3038*/
76e3520e 3039STATIC void
cea2e8a9 3040S_regoptail(pTHX_ regnode *p, regnode *val)
a687059c 3041{
5c0ca799 3042 dTHR;
a0d0e21e 3043 /* "Operandless" and "op != BRANCH" are synonymous in practice. */
c277df42 3044 if (p == NULL || SIZE_ONLY)
3045 return;
22c35a8c 3046 if (PL_regkind[(U8)OP(p)] == BRANCH) {
c277df42 3047 regtail(NEXTOPER(p), val);
a0ed51b3 3048 }
22c35a8c 3049 else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
c277df42 3050 regtail(NEXTOPER(NEXTOPER(p)), val);
a0ed51b3 3051 }
3052 else
a0d0e21e 3053 return;
a687059c 3054}
3055
3056/*
3057 - regcurly - a little FSA that accepts {\d+,?\d*}
3058 */
79072805 3059STATIC I32
cea2e8a9 3060S_regcurly(pTHX_ register char *s)
a687059c 3061{
3062 if (*s++ != '{')
3063 return FALSE;
f0fcb552 3064 if (!isDIGIT(*s))
a687059c 3065 return FALSE;
f0fcb552 3066 while (isDIGIT(*s))
a687059c 3067 s++;
3068 if (*s == ',')
3069 s++;
f0fcb552 3070 while (isDIGIT(*s))
a687059c 3071 s++;
3072 if (*s != '}')
3073 return FALSE;
3074 return TRUE;
3075}
3076
a687059c 3077
76e3520e 3078STATIC regnode *
cea2e8a9 3079S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
c277df42 3080{
35ff7856 3081#ifdef DEBUGGING
f248d071 3082 register U8 op = EXACT; /* Arbitrary non-END op. */
c277df42 3083 register regnode *next, *onode;
3084
3085 while (op != END && (!last || node < last)) {
3086 /* While that wasn't END last time... */
3087
3088 NODE_ALIGN(node);
3089 op = OP(node);
3090 if (op == CLOSE)
3091 l--;
3092 next = regnext(node);
3093 /* Where, what. */
3094 if (OP(node) == OPTIMIZED)
3095 goto after_print;
3096 regprop(sv, node);
54dc92de 3097 PerlIO_printf(Perl_debug_log, "%4d:%*s%s", node - start,
c277df42 3098 2*l + 1, "", SvPVX(sv));
3099 if (next == NULL) /* Next ptr. */
3100 PerlIO_printf(Perl_debug_log, "(0)");
3101 else
3102 PerlIO_printf(Perl_debug_log, "(%d)", next - start);
3103 (void)PerlIO_putc(Perl_debug_log, '\n');
3104 after_print:
22c35a8c 3105 if (PL_regkind[(U8)op] == BRANCHJ) {
c277df42 3106 register regnode *nnode = (OP(next) == LONGJMP
3107 ? regnext(next)
3108 : next);
3109 if (last && nnode > last)
3110 nnode = last;
3111 node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
a0ed51b3 3112 }
22c35a8c 3113 else if (PL_regkind[(U8)op] == BRANCH) {
c277df42 3114 node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
a0ed51b3 3115 }
3116 else if ( op == CURLY) { /* `next' might be very big: optimizer */
c277df42 3117 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
3118 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
a0ed51b3 3119 }
22c35a8c 3120 else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
c277df42 3121 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
3122 next, sv, l + 1);
a0ed51b3 3123 }
3124 else if ( op == PLUS || op == STAR) {
c277df42 3125 node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
a0ed51b3 3126 }
3127 else if (op == ANYOF) {
c277df42 3128 node = NEXTOPER(node);
3129 node += ANY_SKIP;
a0ed51b3 3130 }
22c35a8c 3131 else if (PL_regkind[(U8)op] == EXACT) {
c277df42 3132 /* Literal string, where present. */
3133 node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode);
3134 node = NEXTOPER(node);
a0ed51b3 3135 }
3136 else {
c277df42 3137 node = NEXTOPER(node);
3138 node += regarglen[(U8)op];
3139 }
3140 if (op == CURLYX || op == OPEN)
3141 l++;
3142 else if (op == WHILEM)
3143 l--;
3144 }
17c3b450 3145#endif /* DEBUGGING */
c277df42 3146 return node;
3147}
3148
a687059c 3149/*
fd181c75 3150 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
a687059c 3151 */
3152void
864dbfa3 3153Perl_regdump(pTHX_ regexp *r)
a687059c 3154{
35ff7856 3155#ifdef DEBUGGING
5c0ca799 3156 dTHR;
46fc3d4c 3157 SV *sv = sv_newmortal();
a687059c 3158
c277df42 3159 (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
a0d0e21e 3160
3161 /* Header fields of interest. */
c277df42 3162 if (r->anchored_substr)
82ba1be6 3163 PerlIO_printf(Perl_debug_log, "anchored `%s%.*s%s'%s at %d ",
3280af22 3164 PL_colors[0],
82ba1be6 3165 SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0),
c277df42 3166 SvPVX(r->anchored_substr),
3280af22 3167 PL_colors[1],
c277df42 3168 SvTAIL(r->anchored_substr) ? "$" : "",
3169 r->anchored_offset);
3170 if (r->float_substr)
82ba1be6 3171 PerlIO_printf(Perl_debug_log, "floating `%s%.*s%s'%s at %d..%u ",
3280af22 3172 PL_colors[0],
82ba1be6 3173 SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0),
3174 SvPVX(r->float_substr),
3280af22 3175 PL_colors[1],
c277df42 3176 SvTAIL(r->float_substr) ? "$" : "",
3177 r->float_min_offset, r->float_max_offset);
3178 if (r->check_substr)
3179 PerlIO_printf(Perl_debug_log,
3180 r->check_substr == r->float_substr
3181 ? "(checking floating" : "(checking anchored");
3182 if (r->reganch & ROPT_NOSCAN)
3183 PerlIO_printf(Perl_debug_log, " noscan");
3184 if (r->reganch & ROPT_CHECK_ALL)
3185 PerlIO_printf(Perl_debug_log, " isall");
3186 if (r->check_substr)
3187 PerlIO_printf(Perl_debug_log, ") ");
3188
46fc3d4c 3189 if (r->regstclass) {
3190 regprop(sv, r->regstclass);
3191 PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
3192 }
774d564b 3193 if (r->reganch & ROPT_ANCH) {
3194 PerlIO_printf(Perl_debug_log, "anchored");
3195 if (r->reganch & ROPT_ANCH_BOL)
3196 PerlIO_printf(Perl_debug_log, "(BOL)");
c277df42 3197 if (r->reganch & ROPT_ANCH_MBOL)
3198 PerlIO_printf(Perl_debug_log, "(MBOL)");
cad2e5aa 3199 if (r->reganch & ROPT_ANCH_SBOL)
3200 PerlIO_printf(Perl_debug_log, "(SBOL)");
774d564b 3201 if (r->reganch & ROPT_ANCH_GPOS)
3202 PerlIO_printf(Perl_debug_log, "(GPOS)");
3203 PerlIO_putc(Perl_debug_log, ' ');
3204 }
c277df42 3205 if (r->reganch & ROPT_GPOS_SEEN)
3206 PerlIO_printf(Perl_debug_log, "GPOS ");
a0d0e21e 3207 if (r->reganch & ROPT_SKIP)
760ac839 3208 PerlIO_printf(Perl_debug_log, "plus ");
a0d0e21e 3209 if (r->reganch & ROPT_IMPLICIT)
760ac839 3210 PerlIO_printf(Perl_debug_log, "implicit ");
760ac839 3211 PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
ce862d02 3212 if (r->reganch & ROPT_EVAL_SEEN)
3213 PerlIO_printf(Perl_debug_log, "with eval ");
760ac839 3214 PerlIO_printf(Perl_debug_log, "\n");
17c3b450 3215#endif /* DEBUGGING */
a687059c 3216}
3217
3218/*
a0d0e21e 3219- regprop - printable representation of opcode
3220*/
46fc3d4c 3221void
864dbfa3 3222Perl_regprop(pTHX_ SV *sv, regnode *o)
a687059c 3223{
35ff7856 3224#ifdef DEBUGGING
5c0ca799 3225 dTHR;
9b155405 3226 register int k;
a0d0e21e 3227
54dc92de 3228 sv_setpvn(sv, "", 0);
9b155405 3229 if (OP(o) >= reg_num) /* regnode.type is unsigned */
a0d0e21e 3230 FAIL("corrupted regexp opcode");
9b155405 3231 sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
3232
3233 k = PL_regkind[(U8)OP(o)];
3234
3235 if (k == EXACT)
cea2e8a9 3236 Perl_sv_catpvf(aTHX_ sv, " <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]);
9b155405 3237 else if (k == CURLY) {
3238 if (OP(o) == CURLYM || OP(o) == CURLYN)
cea2e8a9 3239 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
3240 Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e 3241 }
82ba1be6 3242 else if (k == WHILEM && o->flags) /* Ordinal/of */
3243 Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
9b155405 3244 else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
cea2e8a9 3245 Perl_sv_catpvf(aTHX_ sv, "%d", ARG(o)); /* Parenth number */
9b155405 3246 else if (k == LOGICAL)
cea2e8a9 3247 Perl_sv_catpvf(aTHX_ sv, "[%d]", ARG(o)); /* 2: embedded, otherwise 1 */
9b155405 3248 else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
cea2e8a9 3249 Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
17c3b450 3250#endif /* DEBUGGING */
35ff7856 3251}
a687059c 3252
cad2e5aa 3253SV *
3254Perl_re_intuit_string(pTHX_ regexp *prog)
3255{ /* Assume that RE_INTUIT is set */
3256 DEBUG_r(
3257 { STRLEN n_a;
3258 char *s = SvPV(prog->check_substr,n_a);
3259
3260 if (!PL_colorset) reginitcolors();
3261 PerlIO_printf(Perl_debug_log,
3262 "%sUsing REx substr:%s `%s%.60s%s%s'\n",
3263 PL_colors[4],PL_colors[5],PL_colors[0],
3264 s,
3265 PL_colors[1],
3266 (strlen(s) > 60 ? "..." : ""));
3267 } );
3268
3269 return prog->check_substr;
3270}
3271
2b69d0c2 3272void
864dbfa3 3273Perl_pregfree(pTHX_ struct regexp *r)
a687059c 3274{
5c0ca799 3275 dTHR;
adac82c7 3276 DEBUG_r(if (!PL_colorset) reginitcolors());
cad2e5aa 3277 DEBUG_r(PerlIO_printf(Perl_debug_log,
3278 "%sFreeing REx:%s `%s%.60s%s%s'\n",
3279 PL_colors[4],PL_colors[5],PL_colors[0],
3280 r->precomp,
3281 PL_colors[1],
3282 (strlen(r->precomp) > 60 ? "..." : "")));
3283
3284
c277df42 3285 if (!r || (--r->refcnt > 0))
a0d0e21e 3286 return;
c277df42 3287 if (r->precomp)
a0d0e21e 3288 Safefree(r->precomp);
cf93c79d 3289 if (RX_MATCH_COPIED(r))
3290 Safefree(r->subbeg);
a193d654 3291 if (r->substrs) {
3292 if (r->anchored_substr)
3293 SvREFCNT_dec(r->anchored_substr);
3294 if (r->float_substr)
3295 SvREFCNT_dec(r->float_substr);
2779dcf1 3296 Safefree(r->substrs);
a193d654 3297 }
c277df42 3298 if (r->data) {
3299 int n = r->data->count;
dfad63ad 3300 AV* new_comppad = NULL;
3301 AV* old_comppad;
3302 SV** old_curpad;
3303
c277df42 3304 while (--n >= 0) {
3305 switch (r->data->what[n]) {
3306 case 's':
3307 SvREFCNT_dec((SV*)r->data->data[n]);
3308 break;
dfad63ad 3309 case 'p':
3310 new_comppad = (AV*)r->data->data[n];
3311 break;
c277df42 3312 case 'o':
dfad63ad 3313 if (new_comppad == NULL)
cea2e8a9 3314 Perl_croak(aTHX_ "panic: pregfree comppad");
dfad63ad 3315 old_comppad = PL_comppad;
3316 old_curpad = PL_curpad;
3317 PL_comppad = new_comppad;
3318 PL_curpad = AvARRAY(new_comppad);
c277df42 3319 op_free((OP_4tree*)r->data->data[n]);
dfad63ad 3320 PL_comppad = old_comppad;
3321 PL_curpad = old_curpad;
3322 SvREFCNT_dec((SV*)new_comppad);
3323 new_comppad = NULL;
c277df42 3324 break;
3325 case 'n':
3326 break;
3327 default:
3328 FAIL2("panic: regfree data code '%c'", r->data->what[n]);
3329 }
3330 }
3331 Safefree(r->data->what);
3332 Safefree(r->data);
a0d0e21e 3333 }
3334 Safefree(r->startp);
3335 Safefree(r->endp);
3336 Safefree(r);
a687059c 3337}
c277df42 3338
3339/*
3340 - regnext - dig the "next" pointer out of a node
3341 *
3342 * [Note, when REGALIGN is defined there are two places in regmatch()
3343 * that bypass this code for speed.]
3344 */
3345regnode *
864dbfa3 3346Perl_regnext(pTHX_ register regnode *p)
c277df42 3347{
5c0ca799 3348 dTHR;
c277df42 3349 register I32 offset;
3350
3280af22 3351 if (p == &PL_regdummy)
c277df42 3352 return(NULL);
3353
3354 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
3355 if (offset == 0)
3356 return(NULL);
3357
c277df42 3358 return(p+offset);
c277df42 3359}
3360
01f988be 3361STATIC void
cea2e8a9 3362S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
c277df42 3363{
3364 va_list args;
3365 STRLEN l1 = strlen(pat1);
3366 STRLEN l2 = strlen(pat2);
3367 char buf[512];
06bf62c7 3368 SV *msv;
c277df42 3369 char *message;
3370
3371 if (l1 > 510)
3372 l1 = 510;
3373 if (l1 + l2 > 510)
3374 l2 = 510 - l1;
3375 Copy(pat1, buf, l1 , char);
3376 Copy(pat2, buf + l1, l2 , char);
3b818b81 3377 buf[l1 + l2] = '\n';
3378 buf[l1 + l2 + 1] = '\0';
8736538c 3379#ifdef I_STDARG
3380 /* ANSI variant takes additional second argument */
c277df42 3381 va_start(args, pat2);
8736538c 3382#else
3383 va_start(args);
3384#endif
06bf62c7 3385 msv = mess(buf, &args);
c277df42 3386 va_end(args);
06bf62c7 3387 message = SvPV(msv,l1);
c277df42 3388 if (l1 > 512)
3389 l1 = 512;
3390 Copy(message, buf, l1 , char);
3391 buf[l1] = '\0'; /* Overwrite \n */
cea2e8a9 3392 Perl_croak(aTHX_ "%s", buf);
c277df42 3393}
a0ed51b3 3394
3395/* XXX Here's a total kludge. But we need to re-enter for swash routines. */
3396
3397void
864dbfa3 3398Perl_save_re_context(pTHX)
c485e607 3399{
3400 dTHR;
a0ed51b3 3401 SAVEPPTR(PL_bostr);
3402 SAVEPPTR(PL_regprecomp); /* uncompiled string. */
3403 SAVEI32(PL_regnpar); /* () count. */
3404 SAVEI32(PL_regsize); /* Code size. */
3405 SAVEI16(PL_regflags); /* are we folding, multilining? */
3406 SAVEPPTR(PL_reginput); /* String-input pointer. */
3407 SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */
3408 SAVEPPTR(PL_regeol); /* End of input, for $ check. */
3409 SAVESPTR(PL_regstartp); /* Pointer to startp array. */
3410 SAVESPTR(PL_regendp); /* Ditto for endp. */
3411 SAVESPTR(PL_reglastparen); /* Similarly for lastparen. */
3412 SAVEPPTR(PL_regtill); /* How far we are required to go. */
3413 SAVEI32(PL_regprev); /* char before regbol, \n if none */
3414 SAVESPTR(PL_reg_start_tmp); /* from regexec.c */
3415 PL_reg_start_tmp = 0;
3416 SAVEFREEPV(PL_reg_start_tmp);
3417 SAVEI32(PL_reg_start_tmpl); /* from regexec.c */
3418 PL_reg_start_tmpl = 0;
3419 SAVESPTR(PL_regdata);
3420 SAVEI32(PL_reg_flags); /* from regexec.c */
3421 SAVEI32(PL_reg_eval_set); /* from regexec.c */
3422 SAVEI32(PL_regnarrate); /* from regexec.c */
3423 SAVESPTR(PL_regprogram); /* from regexec.c */
3424 SAVEINT(PL_regindent); /* from regexec.c */
3425 SAVESPTR(PL_regcc); /* from regexec.c */
3426 SAVESPTR(PL_curcop);
3427 SAVESPTR(PL_regcomp_rx); /* from regcomp.c */
3428 SAVEI32(PL_regseen); /* from regcomp.c */
3429 SAVEI32(PL_regsawback); /* Did we see \1, ...? */
3430 SAVEI32(PL_regnaughty); /* How bad is this pattern? */
3431 SAVESPTR(PL_regcode); /* Code-emit pointer; &regdummy = don't */
3432 SAVEPPTR(PL_regxend); /* End of input for compile */
3433 SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */
54b6e2fa 3434 SAVESPTR(PL_reg_call_cc); /* from regexec.c */
3435 SAVESPTR(PL_reg_re); /* from regexec.c */
3436 SAVEPPTR(PL_reg_ganch); /* from regexec.c */
3437 SAVESPTR(PL_reg_sv); /* from regexec.c */
3438 SAVESPTR(PL_reg_magic); /* from regexec.c */
3439 SAVEI32(PL_reg_oldpos); /* from regexec.c */
3440 SAVESPTR(PL_reg_oldcurpm); /* from regexec.c */
3441 SAVESPTR(PL_reg_curpm); /* from regexec.c */
3442#ifdef DEBUGGING
3443 SAVEPPTR(PL_reg_starttry); /* from regexec.c */
3444#endif
a0ed51b3 3445}
51371543 3446
3447#ifdef PERL_OBJECT
3448#define NO_XSLOCKS
3449#include "XSUB.h"
3450#undef this
3451#define this pPerl
3452#endif
3453
3454static void
3455clear_re(pTHXo_ void *r)
3456{
3457 ReREFCNT_dec((regexp *)r);
3458}
3459
Unnamed repository; edit this file to name it for gitweb.