This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
correct places that said newSVpv() when they meant newSVpvn()
[perl5.git] / regcomp.c
CommitLineData
a0d0e21e
LW
1/* regcomp.c
2 */
3
4/*
5 * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee
6 */
7
a687059c
LW
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
AD
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
AD
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 */
28# ifndef DEBUGGING
29# define DEBUGGING
30# endif
31#endif
32
33#ifdef PERL_IN_XSUB_RE
d06ea78c 34/* We *really* need to overwrite these symbols: */
56953603
IZ
35# define Perl_pregcomp my_regcomp
36# define Perl_regdump my_regdump
37# define Perl_regprop my_regprop
d06ea78c
GS
38/* *These* symbols are masked to allow static link. */
39# define Perl_pregfree my_regfree
40# define Perl_regnext my_regnext
f0b8d043 41# define Perl_save_re_context my_save_re_context
d88dccdf 42# define Perl_reginitcolors my_reginitcolors
56953603
IZ
43#endif
44
f0fcb552 45/*SUPPRESS 112*/
a687059c 46/*
e50aee73 47 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
48 *
49 * Copyright (c) 1986 by University of Toronto.
50 * Written by Henry Spencer. Not derived from licensed software.
51 *
52 * Permission is granted to anyone to use this software for any
53 * purpose on any computer system, and to redistribute it freely,
54 * subject to the following restrictions:
55 *
56 * 1. The author is not responsible for the consequences of use of
57 * this software, no matter how awful, even if they arise
58 * from defects in it.
59 *
60 * 2. The origin of this software must not be misrepresented, either
61 * by explicit claim or by omission.
62 *
63 * 3. Altered versions must be plainly marked as such, and must not
64 * be misrepresented as being the original software.
65 *
66 *
67 **** Alterations to Henry's code are...
68 ****
4eb8286e 69 **** Copyright (c) 1991-1999, Larry Wall
a687059c 70 ****
9ef589d8
LW
71 **** You may distribute under the terms of either the GNU General Public
72 **** License or the Artistic License, as specified in the README file.
73
a687059c
LW
74 *
75 * Beware that some of this code is subtly aware of the way operator
76 * precedence is structured in regular expressions. Serious changes in
77 * regular-expression syntax might require a total rethink.
78 */
79#include "EXTERN.h"
80#include "perl.h"
d06ea78c 81
b9d5759e 82#ifndef PERL_IN_XSUB_RE
d06ea78c
GS
83# include "INTERN.h"
84#endif
c277df42
IZ
85
86#define REG_COMP_C
a687059c
LW
87#include "regcomp.h"
88
d4cce5f1 89#ifdef op
11343788 90#undef op
d4cce5f1 91#endif /* op */
11343788 92
fe14fcc3
LW
93#ifdef MSDOS
94# if defined(BUGGY_MSC6)
95 /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
96 # pragma optimize("a",off)
97 /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
98 # pragma optimize("w",on )
99# endif /* BUGGY_MSC6 */
100#endif /* MSDOS */
101
a687059c
LW
102#ifndef STATIC
103#define STATIC static
104#endif
105
106#define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
107#define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
108 ((*s) == '{' && regcurly(s)))
2b69d0c2
LW
109#ifdef atarist
110#define PERL_META "^$.[()|?+*\\"
111#else
a687059c 112#define META "^$.[()|?+*\\"
2b69d0c2 113#endif
a687059c 114
35c8bce7
LW
115#ifdef SPSTART
116#undef SPSTART /* dratted cpp namespace... */
117#endif
a687059c
LW
118/*
119 * Flags to be passed up and down.
120 */
a687059c 121#define WORST 0 /* Worst case. */
821b33a5 122#define HASWIDTH 0x1 /* Known to match non-null strings. */
a0d0e21e
LW
123#define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
124#define SPSTART 0x4 /* Starts with * or +. */
125#define TRYAGAIN 0x8 /* Weeded out a declaration. */
a687059c
LW
126
127/*
e50aee73 128 * Forward declarations for pregcomp()'s friends.
a687059c 129 */
a0d0e21e 130
76e3520e 131#ifndef PERL_OBJECT
c277df42
IZ
132static regnode *reg _((I32, I32 *));
133static regnode *reganode _((U8, U32));
134static regnode *regatom _((I32 *));
135static regnode *regbranch _((I32 *, I32));
136static void regc _((U8, char *));
a0ed51b3 137static void reguni _((UV, char *, I32*));
c277df42 138static regnode *regclass _((void));
a0ed51b3 139static regnode *regclassutf8 _((void));
a0d0e21e 140STATIC I32 regcurly _((char *));
c277df42
IZ
141static regnode *reg_node _((U8));
142static regnode *regpiece _((I32 *));
143static void reginsert _((U8, regnode *));
144static void regoptail _((regnode *, regnode *));
c277df42 145static void regtail _((regnode *, regnode *));
873ef191 146static char* regwhite _((char *, char *));
a0d0e21e 147static char* nextchar _((void));
3bd495df 148static void re_croak2 _((const char* pat1,const char* pat2,...)) __attribute__((noreturn));
eb1ab10a 149static char* regpposixcc _((I32 value));
4327152a 150static void clear_re _((void *r));
76e3520e 151#endif
a687059c 152
c277df42
IZ
153/* Length of a variant. */
154
76e3520e 155#ifndef PERL_OBJECT
c277df42
IZ
156typedef struct {
157 I32 len_min;
158 I32 len_delta;
a0ed51b3
LW
159 I32 pos_min; /* CC */
160 I32 pos_delta; /* CC */
c277df42
IZ
161 SV *last_found;
162 I32 last_end; /* min value, <0 unless valid. */
a0ed51b3
LW
163 I32 last_start_min; /* CC */
164 I32 last_start_max; /* CC */
c277df42
IZ
165 SV **longest; /* Either &l_fixed, or &l_float. */
166 SV *longest_fixed;
a0ed51b3 167 I32 offset_fixed; /* CC */
c277df42 168 SV *longest_float;
a0ed51b3
LW
169 I32 offset_float_min; /* CC */
170 I32 offset_float_max; /* CC */
c277df42
IZ
171 I32 flags;
172} scan_data_t;
76e3520e 173#endif
c277df42 174
cd488c12
JB
175static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
176 0, 0, 0 };
c277df42
IZ
177
178#define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
179#define SF_BEFORE_SEOL 0x1
180#define SF_BEFORE_MEOL 0x2
181#define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
182#define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
183
09b7f37c
CB
184#ifdef NO_UNARY_PLUS
185# define SF_FIX_SHIFT_EOL (0+2)
186# define SF_FL_SHIFT_EOL (0+4)
187#else
188# define SF_FIX_SHIFT_EOL (+2)
189# define SF_FL_SHIFT_EOL (+4)
190#endif
c277df42
IZ
191
192#define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
193#define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
194
195#define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
196#define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
197#define SF_IS_INF 0x40
198#define SF_HAS_PAR 0x80
199#define SF_IN_PAR 0x100
200#define SF_HAS_EVAL 0x200
4bfe0158 201#define SCF_DO_SUBSTR 0x400
c277df42 202
a0ed51b3
LW
203#define RF_utf8 8
204#define UTF (PL_reg_flags & RF_utf8)
205#define LOC (PL_regflags & PMf_LOCALE)
206#define FOLD (PL_regflags & PMf_FOLD)
207
208#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
209#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
210
76e3520e 211STATIC void
4327152a
IZ
212clear_re(void *r)
213{
214 ReREFCNT_dec((regexp *)r);
215}
216
217STATIC void
c277df42
IZ
218scan_commit(scan_data_t *data)
219{
c485e607 220 dTHR;
a0ed51b3
LW
221 STRLEN l = CHR_SVLEN(data->last_found);
222 STRLEN old_l = CHR_SVLEN(*data->longest);
c277df42
IZ
223
224 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
225 sv_setsv(*data->longest, data->last_found);
226 if (*data->longest == data->longest_fixed) {
227 data->offset_fixed = l ? data->last_start_min : data->pos_min;
228 if (data->flags & SF_BEFORE_EOL)
229 data->flags
230 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
231 else
232 data->flags &= ~SF_FIX_BEFORE_EOL;
a0ed51b3
LW
233 }
234 else {
c277df42
IZ
235 data->offset_float_min = l ? data->last_start_min : data->pos_min;
236 data->offset_float_max = (l
237 ? data->last_start_max
238 : data->pos_min + data->pos_delta);
239 if (data->flags & SF_BEFORE_EOL)
240 data->flags
241 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
242 else
243 data->flags &= ~SF_FL_BEFORE_EOL;
244 }
245 }
246 SvCUR_set(data->last_found, 0);
247 data->last_end = -1;
248 data->flags &= ~SF_BEFORE_EOL;
249}
250
c277df42
IZ
251/* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
252 to the position after last scanned or to NULL. */
253
76e3520e 254STATIC I32
c277df42
IZ
255study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
256 /* scanp: Start here (read-write). */
257 /* deltap: Write maxlen-minlen here. */
258 /* last: Stop before this one. */
259{
5c0ca799 260 dTHR;
c277df42
IZ
261 I32 min = 0, pars = 0, code;
262 regnode *scan = *scanp, *next;
263 I32 delta = 0;
264 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
aca2d497 265 int is_inf_internal = 0; /* The studied chunk is infinite */
c277df42
IZ
266 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
267 scan_data_t data_fake;
268
269 while (scan && OP(scan) != END && scan < last) {
270 /* Peephole optimizer: */
271
22c35a8c 272 if (PL_regkind[(U8)OP(scan)] == EXACT) {
c277df42
IZ
273 regnode *n = regnext(scan);
274 U32 stringok = 1;
275#ifdef DEBUGGING
276 regnode *stop = scan;
277#endif
278
279 next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
280 /* Skip NOTHING, merge EXACT*. */
281 while (n &&
22c35a8c 282 ( PL_regkind[(U8)OP(n)] == NOTHING ||
c277df42
IZ
283 (stringok && (OP(n) == OP(scan))))
284 && NEXT_OFF(n)
285 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
286 if (OP(n) == TAIL || n > next)
287 stringok = 0;
22c35a8c 288 if (PL_regkind[(U8)OP(n)] == NOTHING) {
c277df42
IZ
289 NEXT_OFF(scan) += NEXT_OFF(n);
290 next = n + NODE_STEP_REGNODE;
291#ifdef DEBUGGING
292 if (stringok)
293 stop = n;
294#endif
295 n = regnext(n);
a0ed51b3
LW
296 }
297 else {
c277df42
IZ
298 int oldl = *OPERAND(scan);
299 regnode *nnext = regnext(n);
300
301 if (oldl + *OPERAND(n) > U8_MAX)
302 break;
303 NEXT_OFF(scan) += NEXT_OFF(n);
304 *OPERAND(scan) += *OPERAND(n);
305 next = n + (*OPERAND(n) + 2 - 1)/sizeof(regnode) + 2;
306 /* Now we can overwrite *n : */
307 Move(OPERAND(n) + 1, OPERAND(scan) + oldl + 1,
308 *OPERAND(n) + 1, char);
309#ifdef DEBUGGING
310 if (stringok)
311 stop = next - 1;
312#endif
313 n = nnext;
314 }
315 }
316#ifdef DEBUGGING
317 /* Allow dumping */
318 n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
319 while (n <= stop) {
ca04da08
GS
320 /* Purify reports a benign UMR here sometimes, because we
321 * don't initialize the OP() slot of a node when that node
322 * is occupied by just the trailing null of the string in
323 * an EXACT node */
22c35a8c 324 if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
c277df42
IZ
325 OP(n) = OPTIMIZED;
326 NEXT_OFF(n) = 0;
327 }
328 n++;
329 }
330#endif
331
332 }
333 if (OP(scan) != CURLYX) {
048cfca1
GS
334 int max = (reg_off_by_arg[OP(scan)]
335 ? I32_MAX
336 /* I32 may be smaller than U16 on CRAYs! */
337 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
c277df42
IZ
338 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
339 int noff;
340 regnode *n = scan;
341
342 /* Skip NOTHING and LONGJMP. */
343 while ((n = regnext(n))
22c35a8c 344 && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
c277df42
IZ
345 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
346 && off + noff < max)
347 off += noff;
348 if (reg_off_by_arg[OP(scan)])
349 ARG(scan) = off;
350 else
351 NEXT_OFF(scan) = off;
352 }
353 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
354 || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
355 next = regnext(scan);
356 code = OP(scan);
357
358 if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
359 I32 max1 = 0, min1 = I32_MAX, num = 0;
360
361 if (flags & SCF_DO_SUBSTR)
362 scan_commit(data);
363 while (OP(scan) == code) {
364 I32 deltanext, minnext;
365
366 num++;
367 data_fake.flags = 0;
368 next = regnext(scan);
369 scan = NEXTOPER(scan);
370 if (code != BRANCH)
371 scan = NEXTOPER(scan);
372 /* We suppose the run is continuous, last=next...*/
373 minnext = study_chunk(&scan, &deltanext, next,
374 &data_fake, 0);
375 if (min1 > minnext)
376 min1 = minnext;
377 if (max1 < minnext + deltanext)
378 max1 = minnext + deltanext;
379 if (deltanext == I32_MAX)
aca2d497 380 is_inf = is_inf_internal = 1;
c277df42
IZ
381 scan = next;
382 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
383 pars++;
405ff068 384 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42
IZ
385 data->flags |= SF_HAS_EVAL;
386 if (code == SUSPEND)
387 break;
388 }
389 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
390 min1 = 0;
391 if (flags & SCF_DO_SUBSTR) {
392 data->pos_min += min1;
393 data->pos_delta += max1 - min1;
394 if (max1 != min1 || is_inf)
395 data->longest = &(data->longest_float);
396 }
397 min += min1;
398 delta += max1 - min1;
a0ed51b3
LW
399 }
400 else if (code == BRANCHJ) /* single branch is optimized. */
c277df42
IZ
401 scan = NEXTOPER(NEXTOPER(scan));
402 else /* single branch is optimized. */
403 scan = NEXTOPER(scan);
404 continue;
a0ed51b3
LW
405 }
406 else if (OP(scan) == EXACT) {
407 I32 l = *OPERAND(scan);
408 if (UTF) {
409 unsigned char *s = (unsigned char *)(OPERAND(scan)+1);
410 unsigned char *e = s + l;
411 I32 newl = 0;
412 while (s < e) {
413 newl++;
414 s += UTF8SKIP(s);
415 }
416 l = newl;
417 }
418 min += l;
c277df42 419 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
c277df42
IZ
420 /* The code below prefers earlier match for fixed
421 offset, later match for variable offset. */
422 if (data->last_end == -1) { /* Update the start info. */
423 data->last_start_min = data->pos_min;
424 data->last_start_max = is_inf
425 ? I32_MAX : data->pos_min + data->pos_delta;
426 }
a0ed51b3 427 sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), *OPERAND(scan));
c277df42
IZ
428 data->last_end = data->pos_min + l;
429 data->pos_min += l; /* As in the first entry. */
430 data->flags &= ~SF_BEFORE_EOL;
431 }
a0ed51b3 432 }
22c35a8c 433 else if (PL_regkind[(U8)OP(scan)] == EXACT) {
a0ed51b3 434 I32 l = *OPERAND(scan);
c277df42
IZ
435 if (flags & SCF_DO_SUBSTR)
436 scan_commit(data);
a0ed51b3
LW
437 if (UTF) {
438 unsigned char *s = (unsigned char *)(OPERAND(scan)+1);
439 unsigned char *e = s + l;
440 I32 newl = 0;
441 while (s < e) {
442 newl++;
443 s += UTF8SKIP(s);
444 }
445 l = newl;
446 }
447 min += l;
c277df42 448 if (data && (flags & SCF_DO_SUBSTR))
a0ed51b3
LW
449 data->pos_min += l;
450 }
22c35a8c 451 else if (strchr(PL_varies,OP(scan))) {
c277df42
IZ
452 I32 mincount, maxcount, minnext, deltanext, pos_before, fl;
453 regnode *oscan = scan;
454
22c35a8c 455 switch (PL_regkind[(U8)OP(scan)]) {
c277df42
IZ
456 case WHILEM:
457 scan = NEXTOPER(scan);
458 goto finish;
459 case PLUS:
460 if (flags & SCF_DO_SUBSTR) {
461 next = NEXTOPER(scan);
462 if (OP(next) == EXACT) {
463 mincount = 1;
464 maxcount = REG_INFTY;
465 next = regnext(scan);
466 scan = NEXTOPER(scan);
467 goto do_curly;
468 }
469 }
470 if (flags & SCF_DO_SUBSTR)
471 data->pos_min++;
472 min++;
473 /* Fall through. */
474 case STAR:
aca2d497 475 is_inf = is_inf_internal = 1;
c277df42
IZ
476 scan = regnext(scan);
477 if (flags & SCF_DO_SUBSTR) {
478 scan_commit(data);
479 data->longest = &(data->longest_float);
480 }
481 goto optimize_curly_tail;
482 case CURLY:
483 mincount = ARG1(scan);
484 maxcount = ARG2(scan);
485 next = regnext(scan);
486 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
487 do_curly:
488 if (flags & SCF_DO_SUBSTR) {
489 if (mincount == 0) scan_commit(data);
490 pos_before = data->pos_min;
491 }
492 if (data) {
493 fl = data->flags;
494 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
495 if (is_inf)
496 data->flags |= SF_IS_INF;
497 }
498 /* This will finish on WHILEM, setting scan, or on NULL: */
499 minnext = study_chunk(&scan, &deltanext, last, data,
500 mincount == 0
501 ? (flags & ~SCF_DO_SUBSTR) : flags);
502 if (!scan) /* It was not CURLYX, but CURLY. */
503 scan = next;
599cee73 504 if (ckWARN(WARN_UNSAFE) && (minnext + deltanext == 0)
821b33a5 505 && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
17feb5d5 506 && maxcount <= REG_INFTY/3) /* Complement check for big count */
599cee73 507 warner(WARN_UNSAFE, "Strange *+?{} on zero-length expression");
c277df42 508 min += minnext * mincount;
aca2d497
IZ
509 is_inf_internal |= (maxcount == REG_INFTY
510 && (minnext + deltanext) > 0
511 || deltanext == I32_MAX);
512 is_inf |= is_inf_internal;
c277df42
IZ
513 delta += (minnext + deltanext) * maxcount - minnext * mincount;
514
515 /* Try powerful optimization CURLYX => CURLYN. */
c277df42
IZ
516 if ( OP(oscan) == CURLYX && data
517 && data->flags & SF_IN_PAR
518 && !(data->flags & SF_HAS_EVAL)
519 && !deltanext && minnext == 1 ) {
520 /* Try to optimize to CURLYN. */
521 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
522 regnode *nxt1 = nxt, *nxt2;
523
524 /* Skip open. */
525 nxt = regnext(nxt);
22c35a8c
GS
526 if (!strchr(PL_simple,OP(nxt))
527 && !(PL_regkind[(U8)OP(nxt)] == EXACT
c277df42
IZ
528 && *OPERAND(nxt) == 1))
529 goto nogo;
530 nxt2 = nxt;
531 nxt = regnext(nxt);
532 if (OP(nxt) != CLOSE)
533 goto nogo;
534 /* Now we know that nxt2 is the only contents: */
535 oscan->flags = ARG(nxt);
536 OP(oscan) = CURLYN;
537 OP(nxt1) = NOTHING; /* was OPEN. */
538#ifdef DEBUGGING
539 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
540 NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
541 NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
542 OP(nxt) = OPTIMIZED; /* was CLOSE. */
543 OP(nxt + 1) = OPTIMIZED; /* was count. */
544 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
545#endif
546 }
c277df42
IZ
547 nogo:
548
549 /* Try optimization CURLYX => CURLYM. */
550 if ( OP(oscan) == CURLYX && data
c277df42 551 && !(data->flags & SF_HAS_PAR)
c277df42
IZ
552 && !(data->flags & SF_HAS_EVAL)
553 && !deltanext ) {
554 /* XXXX How to optimize if data == 0? */
555 /* Optimize to a simpler form. */
556 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
557 regnode *nxt2;
558
559 OP(oscan) = CURLYM;
560 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
561 && (OP(nxt2) != WHILEM))
562 nxt = nxt2;
563 OP(nxt2) = SUCCEED; /* Whas WHILEM */
c277df42
IZ
564 /* Need to optimize away parenths. */
565 if (data->flags & SF_IN_PAR) {
566 /* Set the parenth number. */
567 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
568
569 if (OP(nxt) != CLOSE)
570 FAIL("panic opt close");
571 oscan->flags = ARG(nxt);
572 OP(nxt1) = OPTIMIZED; /* was OPEN. */
573 OP(nxt) = OPTIMIZED; /* was CLOSE. */
574#ifdef DEBUGGING
575 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
576 OP(nxt + 1) = OPTIMIZED; /* was count. */
577 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
578 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
579#endif
580#if 0
581 while ( nxt1 && (OP(nxt1) != WHILEM)) {
582 regnode *nnxt = regnext(nxt1);
583
584 if (nnxt == nxt) {
585 if (reg_off_by_arg[OP(nxt1)])
586 ARG_SET(nxt1, nxt2 - nxt1);
587 else if (nxt2 - nxt1 < U16_MAX)
588 NEXT_OFF(nxt1) = nxt2 - nxt1;
589 else
590 OP(nxt) = NOTHING; /* Cannot beautify */
591 }
592 nxt1 = nnxt;
593 }
594#endif
595 /* Optimize again: */
596 study_chunk(&nxt1, &deltanext, nxt, NULL, 0);
a0ed51b3
LW
597 }
598 else
c277df42 599 oscan->flags = 0;
c277df42
IZ
600 }
601 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
602 pars++;
603 if (flags & SCF_DO_SUBSTR) {
604 SV *last_str = Nullsv;
605 int counted = mincount != 0;
606
607 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
608 I32 b = pos_before >= data->last_start_min
609 ? pos_before : data->last_start_min;
610 STRLEN l;
611 char *s = SvPV(data->last_found, l);
a0ed51b3
LW
612 I32 old = b - data->last_start_min;
613
614 if (UTF)
615 old = utf8_hop((U8*)s, old) - (U8*)s;
c277df42 616
a0ed51b3 617 l -= old;
c277df42 618 /* Get the added string: */
79cb57f6 619 last_str = newSVpvn(s + old, l);
c277df42
IZ
620 if (deltanext == 0 && pos_before == b) {
621 /* What was added is a constant string */
622 if (mincount > 1) {
623 SvGROW(last_str, (mincount * l) + 1);
624 repeatcpy(SvPVX(last_str) + l,
625 SvPVX(last_str), l, mincount - 1);
626 SvCUR(last_str) *= mincount;
627 /* Add additional parts. */
628 SvCUR_set(data->last_found,
629 SvCUR(data->last_found) - l);
630 sv_catsv(data->last_found, last_str);
631 data->last_end += l * (mincount - 1);
632 }
633 }
634 }
635 /* It is counted once already... */
636 data->pos_min += minnext * (mincount - counted);
637 data->pos_delta += - counted * deltanext +
638 (minnext + deltanext) * maxcount - minnext * mincount;
639 if (mincount != maxcount) {
640 scan_commit(data);
641 if (mincount && last_str) {
642 sv_setsv(data->last_found, last_str);
643 data->last_end = data->pos_min;
644 data->last_start_min =
a0ed51b3 645 data->pos_min - CHR_SVLEN(last_str);
c277df42
IZ
646 data->last_start_max = is_inf
647 ? I32_MAX
648 : data->pos_min + data->pos_delta
a0ed51b3 649 - CHR_SVLEN(last_str);
c277df42
IZ
650 }
651 data->longest = &(data->longest_float);
652 }
aca2d497 653 SvREFCNT_dec(last_str);
c277df42 654 }
405ff068 655 if (data && (fl & SF_HAS_EVAL))
c277df42
IZ
656 data->flags |= SF_HAS_EVAL;
657 optimize_curly_tail:
c277df42 658 if (OP(oscan) != CURLYX) {
22c35a8c 659 while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
c277df42
IZ
660 && NEXT_OFF(next))
661 NEXT_OFF(oscan) += NEXT_OFF(next);
662 }
c277df42
IZ
663 continue;
664 default: /* REF only? */
665 if (flags & SCF_DO_SUBSTR) {
666 scan_commit(data);
667 data->longest = &(data->longest_float);
668 }
aca2d497 669 is_inf = is_inf_internal = 1;
c277df42
IZ
670 break;
671 }
a0ed51b3 672 }
22c35a8c 673 else if (strchr(PL_simple,OP(scan)) || PL_regkind[(U8)OP(scan)] == ANYUTF8) {
c277df42
IZ
674 if (flags & SCF_DO_SUBSTR) {
675 scan_commit(data);
676 data->pos_min++;
677 }
678 min++;
a0ed51b3 679 }
22c35a8c 680 else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
c277df42
IZ
681 data->flags |= (OP(scan) == MEOL
682 ? SF_BEFORE_MEOL
683 : SF_BEFORE_SEOL);
a0ed51b3 684 }
22c35a8c 685 else if (PL_regkind[(U8)OP(scan)] == BRANCHJ
c277df42
IZ
686 && (scan->flags || data)
687 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
688 I32 deltanext, minnext;
689 regnode *nscan;
690
691 data_fake.flags = 0;
692 next = regnext(scan);
693 nscan = NEXTOPER(NEXTOPER(scan));
694 minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0);
695 if (scan->flags) {
696 if (deltanext) {
697 FAIL("variable length lookbehind not implemented");
a0ed51b3
LW
698 }
699 else if (minnext > U8_MAX) {
c277df42
IZ
700 FAIL2("lookbehind longer than %d not implemented", U8_MAX);
701 }
702 scan->flags = minnext;
703 }
704 if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
705 pars++;
405ff068 706 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42 707 data->flags |= SF_HAS_EVAL;
a0ed51b3
LW
708 }
709 else if (OP(scan) == OPEN) {
c277df42 710 pars++;
a0ed51b3
LW
711 }
712 else if (OP(scan) == CLOSE && ARG(scan) == is_par) {
c277df42
IZ
713 next = regnext(scan);
714
715 if ( next && (OP(next) != WHILEM) && next < last)
c277df42 716 is_par = 0; /* Disable optimization */
a0ed51b3
LW
717 }
718 else if (OP(scan) == EVAL) {
c277df42
IZ
719 if (data)
720 data->flags |= SF_HAS_EVAL;
721 }
0f5d15d6
IZ
722 else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */
723 if (flags & SCF_DO_SUBSTR) {
724 scan_commit(data);
725 data->longest = &(data->longest_float);
726 }
727 is_inf = is_inf_internal = 1;
728 }
c277df42
IZ
729 /* Else: zero-length, ignore. */
730 scan = regnext(scan);
731 }
732
733 finish:
734 *scanp = scan;
aca2d497 735 *deltap = is_inf_internal ? I32_MAX : delta;
c277df42
IZ
736 if (flags & SCF_DO_SUBSTR && is_inf)
737 data->pos_delta = I32_MAX - data->pos_min;
738 if (is_par > U8_MAX)
739 is_par = 0;
740 if (is_par && pars==1 && data) {
741 data->flags |= SF_IN_PAR;
742 data->flags &= ~SF_HAS_PAR;
a0ed51b3
LW
743 }
744 else if (pars && data) {
c277df42
IZ
745 data->flags |= SF_HAS_PAR;
746 data->flags &= ~SF_IN_PAR;
747 }
748 return min;
749}
750
76e3520e 751STATIC I32
c277df42
IZ
752add_data(I32 n, char *s)
753{
5c0ca799 754 dTHR;
3280af22
NIS
755 if (PL_regcomp_rx->data) {
756 Renewc(PL_regcomp_rx->data,
757 sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1),
c277df42 758 char, struct reg_data);
3280af22
NIS
759 Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8);
760 PL_regcomp_rx->data->count += n;
a0ed51b3
LW
761 }
762 else {
3280af22 763 Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1),
c277df42 764 char, struct reg_data);
3280af22
NIS
765 New(1208, PL_regcomp_rx->data->what, n, U8);
766 PL_regcomp_rx->data->count = n;
c277df42 767 }
3280af22
NIS
768 Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8);
769 return PL_regcomp_rx->data->count - n;
c277df42
IZ
770}
771
d88dccdf
IZ
772void
773reginitcolors(void)
774{
775 dTHR;
776 int i = 0;
777 char *s = PerlEnv_getenv("PERL_RE_COLORS");
778
779 if (s) {
780 PL_colors[0] = s = savepv(s);
781 while (++i < 6) {
782 s = strchr(s, '\t');
783 if (s) {
784 *s = '\0';
785 PL_colors[i] = ++s;
786 }
787 else
c712d376 788 PL_colors[i] = s = "";
d88dccdf
IZ
789 }
790 } else {
791 while (i < 6)
792 PL_colors[i++] = "";
793 }
794 PL_colorset = 1;
795}
796
a687059c 797/*
e50aee73 798 - pregcomp - compile a regular expression into internal code
a687059c
LW
799 *
800 * We can't allocate space until we know how big the compiled form will be,
801 * but we can't compile it (and thus know how big it is) until we've got a
802 * place to put the code. So we cheat: we compile it twice, once with code
803 * generation turned off and size counting turned on, and once "for real".
804 * This also means that we don't allocate space until we are sure that the
805 * thing really will compile successfully, and we never have to move the
806 * code and thus invalidate pointers into it. (Note that it has to be in
807 * one piece because free() must be able to free it all.) [NB: not true in perl]
808 *
809 * Beware that the optimization-preparation code in here knows about some
810 * of the structure of the compiled regexp. [I'll say.]
811 */
812regexp *
8ac85365 813pregcomp(char *exp, char *xend, PMOP *pm)
a687059c 814{
5c0ca799 815 dTHR;
a0d0e21e 816 register regexp *r;
c277df42
IZ
817 regnode *scan;
818 SV **longest;
819 SV *longest_fixed;
820 SV *longest_float;
821 regnode *first;
a0d0e21e 822 I32 flags;
a0d0e21e
LW
823 I32 minlen = 0;
824 I32 sawplus = 0;
825 I32 sawopen = 0;
826
827 if (exp == NULL)
c277df42 828 FAIL("NULL regexp argument");
a0d0e21e 829
e24b16f9 830 if (PL_curcop == &PL_compiling ? (PL_hints & HINT_UTF8) : IN_UTF8)
a0ed51b3
LW
831 PL_reg_flags |= RF_utf8;
832 else
833 PL_reg_flags = 0;
834
3280af22 835 PL_regprecomp = savepvn(exp, xend - exp);
35ef4773
GS
836 DEBUG_r(if (!PL_colorset) reginitcolors());
837 DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling%s RE `%s%*s%s'\n",
d88dccdf
IZ
838 PL_colors[4],PL_colors[5],PL_colors[0],
839 xend - exp, PL_regprecomp, PL_colors[1]));
3280af22
NIS
840 PL_regflags = pm->op_pmflags;
841 PL_regsawback = 0;
bbce6d69 842
3280af22
NIS
843 PL_regseen = 0;
844 PL_seen_zerolen = *exp == '^' ? -1 : 0;
845 PL_seen_evals = 0;
846 PL_extralen = 0;
c277df42 847
bbce6d69 848 /* First pass: determine size, legality. */
3280af22
NIS
849 PL_regcomp_parse = exp;
850 PL_regxend = xend;
851 PL_regnaughty = 0;
852 PL_regnpar = 1;
853 PL_regsize = 0L;
854 PL_regcode = &PL_regdummy;
22c35a8c 855 regc((U8)REG_MAGIC, (char*)PL_regcode);
a0d0e21e 856 if (reg(0, &flags) == NULL) {
3280af22
NIS
857 Safefree(PL_regprecomp);
858 PL_regprecomp = Nullch;
a0d0e21e
LW
859 return(NULL);
860 }
3280af22 861 DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", PL_regsize));
c277df42 862
c277df42
IZ
863 /* Small enough for pointer-storage convention?
864 If extralen==0, this means that we will not need long jumps. */
3280af22
NIS
865 if (PL_regsize >= 0x10000L && PL_extralen)
866 PL_regsize += PL_extralen;
c277df42 867 else
3280af22 868 PL_extralen = 0;
a0d0e21e 869
bbce6d69 870 /* Allocate space and initialize. */
3280af22 871 Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode),
c277df42 872 char, regexp);
a0d0e21e
LW
873 if (r == NULL)
874 FAIL("regexp out of space");
c277df42 875 r->refcnt = 1;
bbce6d69 876 r->prelen = xend - exp;
3280af22 877 r->precomp = PL_regprecomp;
a0d0e21e 878 r->subbeg = r->subbase = NULL;
4327152a
IZ
879 r->nparens = PL_regnpar - 1; /* set early to validate backrefs */
880
881 r->substrs = 0; /* Useful during FAIL. */
882 r->startp = 0; /* Useful during FAIL. */
883 r->endp = 0; /* Useful during FAIL. */
884
3280af22 885 PL_regcomp_rx = r;
bbce6d69 886
887 /* Second pass: emit code. */
3280af22
NIS
888 PL_regcomp_parse = exp;
889 PL_regxend = xend;
890 PL_regnaughty = 0;
891 PL_regnpar = 1;
892 PL_regcode = r->program;
2cd61cdb 893 /* Store the count of eval-groups for security checks: */
3280af22 894 PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals);
22c35a8c 895 regc((U8)REG_MAGIC, (char*) PL_regcode++);
c277df42 896 r->data = 0;
a0d0e21e
LW
897 if (reg(0, &flags) == NULL)
898 return(NULL);
899
900 /* Dig out information for optimizations. */
8782bef2 901 r->reganch = pm->op_pmflags & PMf_COMPILETIME;
3280af22 902 pm->op_pmflags = PL_regflags;
a0ed51b3
LW
903 if (UTF)
904 r->reganch |= ROPT_UTF8;
c277df42 905 r->regstclass = NULL;
a0ed51b3
LW
906 if (PL_regnaughty >= 10) /* Probably an expensive pattern. */
907 r->reganch |= ROPT_NAUGHTY;
c277df42 908 scan = r->program + 1; /* First BRANCH. */
2779dcf1
IZ
909
910 /* XXXX To minimize changes to RE engine we always allocate
911 3-units-long substrs field. */
912 Newz(1004, r->substrs, 1, struct reg_substr_data);
913
c277df42
IZ
914 if (OP(scan) != BRANCH) { /* Only one top-level choice. */
915 scan_data_t data;
916 I32 fake;
c5254dd6 917 STRLEN longest_float_length, longest_fixed_length;
a0d0e21e 918
c277df42 919 StructCopy(&zero_scan_data, &data, scan_data_t);
a0d0e21e 920 first = scan;
c277df42 921 /* Skip introductions and multiplicators >= 1. */
a0d0e21e
LW
922 while ((OP(first) == OPEN && (sawopen = 1)) ||
923 (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
924 (OP(first) == PLUS) ||
925 (OP(first) == MINMOD) ||
22c35a8c 926 (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
a0d0e21e
LW
927 if (OP(first) == PLUS)
928 sawplus = 1;
929 else
930 first += regarglen[(U8)OP(first)];
931 first = NEXTOPER(first);
a687059c
LW
932 }
933
a0d0e21e
LW
934 /* Starting-point info. */
935 again:
c277df42 936 if (OP(first) == EXACT); /* Empty, get anchored substr later. */
22c35a8c 937 else if (strchr(PL_simple+4,OP(first)))
a0d0e21e 938 r->regstclass = first;
22c35a8c
GS
939 else if (PL_regkind[(U8)OP(first)] == BOUND ||
940 PL_regkind[(U8)OP(first)] == NBOUND)
a0d0e21e 941 r->regstclass = first;
22c35a8c 942 else if (PL_regkind[(U8)OP(first)] == BOL) {
c277df42 943 r->reganch |= (OP(first) == MBOL ? ROPT_ANCH_MBOL: ROPT_ANCH_BOL);
a0d0e21e 944 first = NEXTOPER(first);
774d564b 945 goto again;
946 }
947 else if (OP(first) == GPOS) {
948 r->reganch |= ROPT_ANCH_GPOS;
949 first = NEXTOPER(first);
950 goto again;
a0d0e21e
LW
951 }
952 else if ((OP(first) == STAR &&
22c35a8c 953 PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
a0d0e21e
LW
954 !(r->reganch & ROPT_ANCH) )
955 {
956 /* turn .* into ^.* with an implied $*=1 */
774d564b 957 r->reganch |= ROPT_ANCH_BOL | ROPT_IMPLICIT;
a0d0e21e 958 first = NEXTOPER(first);
774d564b 959 goto again;
a0d0e21e 960 }
3280af22 961 if (sawplus && (!sawopen || !PL_regsawback))
a0d0e21e
LW
962 r->reganch |= ROPT_SKIP; /* x+ must match 1st of run */
963
c277df42
IZ
964 /* Scan is after the zeroth branch, first is atomic matcher. */
965 DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n",
966 first - scan + 1));
a0d0e21e
LW
967 /*
968 * If there's something expensive in the r.e., find the
969 * longest literal string that must appear and make it the
970 * regmust. Resolve ties in favor of later strings, since
971 * the regstart check works with the beginning of the r.e.
972 * and avoiding duplication strengthens checking. Not a
973 * strong reason, but sufficient in the absence of others.
974 * [Now we resolve ties in favor of the earlier string if
c277df42 975 * it happens that c_offset_min has been invalidated, since the
a0d0e21e
LW
976 * earlier string may buy us something the later one won't.]
977 */
a0d0e21e 978 minlen = 0;
a687059c 979
79cb57f6
GS
980 data.longest_fixed = newSVpvn("",0);
981 data.longest_float = newSVpvn("",0);
982 data.last_found = newSVpvn("",0);
c277df42
IZ
983 data.longest = &(data.longest_fixed);
984 first = scan;
985
3280af22 986 minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */
c277df42 987 &data, SCF_DO_SUBSTR);
3280af22 988 if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed)
c277df42 989 && data.last_start_min == 0 && data.last_end > 0
3280af22
NIS
990 && !PL_seen_zerolen
991 && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
c277df42
IZ
992 r->reganch |= ROPT_CHECK_ALL;
993 scan_commit(&data);
994 SvREFCNT_dec(data.last_found);
995
a0ed51b3 996 longest_float_length = CHR_SVLEN(data.longest_float);
c5254dd6 997 if (longest_float_length
c277df42
IZ
998 || (data.flags & SF_FL_BEFORE_EOL
999 && (!(data.flags & SF_FL_BEFORE_MEOL)
3280af22 1000 || (PL_regflags & PMf_MULTILINE)))) {
a0ed51b3 1001 if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */
aca2d497
IZ
1002 && data.offset_fixed == data.offset_float_min
1003 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
1004 goto remove_float; /* As in (a)+. */
1005
c277df42
IZ
1006 r->float_substr = data.longest_float;
1007 r->float_min_offset = data.offset_float_min;
1008 r->float_max_offset = data.offset_float_max;
2779dcf1 1009 fbm_compile(r->float_substr, 0);
c277df42
IZ
1010 BmUSEFUL(r->float_substr) = 100;
1011 if (data.flags & SF_FL_BEFORE_EOL /* Cannot have SEOL and MULTI */
1012 && (!(data.flags & SF_FL_BEFORE_MEOL)
3280af22 1013 || (PL_regflags & PMf_MULTILINE)))
c277df42 1014 SvTAIL_on(r->float_substr);
a0ed51b3
LW
1015 }
1016 else {
aca2d497 1017 remove_float:
c277df42
IZ
1018 r->float_substr = Nullsv;
1019 SvREFCNT_dec(data.longest_float);
c5254dd6 1020 longest_float_length = 0;
a0d0e21e 1021 }
c277df42 1022
a0ed51b3 1023 longest_fixed_length = CHR_SVLEN(data.longest_fixed);
c5254dd6 1024 if (longest_fixed_length
c277df42
IZ
1025 || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1026 && (!(data.flags & SF_FIX_BEFORE_MEOL)
3280af22 1027 || (PL_regflags & PMf_MULTILINE)))) {
c277df42
IZ
1028 r->anchored_substr = data.longest_fixed;
1029 r->anchored_offset = data.offset_fixed;
2779dcf1 1030 fbm_compile(r->anchored_substr, 0);
c277df42
IZ
1031 BmUSEFUL(r->anchored_substr) = 100;
1032 if (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1033 && (!(data.flags & SF_FIX_BEFORE_MEOL)
3280af22 1034 || (PL_regflags & PMf_MULTILINE)))
c277df42 1035 SvTAIL_on(r->anchored_substr);
a0ed51b3
LW
1036 }
1037 else {
c277df42
IZ
1038 r->anchored_substr = Nullsv;
1039 SvREFCNT_dec(data.longest_fixed);
c5254dd6 1040 longest_fixed_length = 0;
a0d0e21e 1041 }
c277df42
IZ
1042
1043 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
c5254dd6 1044 if (longest_fixed_length > longest_float_length) {
c277df42
IZ
1045 r->check_substr = r->anchored_substr;
1046 r->check_offset_min = r->check_offset_max = r->anchored_offset;
1047 if (r->reganch & ROPT_ANCH_SINGLE)
1048 r->reganch |= ROPT_NOSCAN;
a0ed51b3
LW
1049 }
1050 else {
c277df42
IZ
1051 r->check_substr = r->float_substr;
1052 r->check_offset_min = data.offset_float_min;
1053 r->check_offset_max = data.offset_float_max;
a0d0e21e 1054 }
a0ed51b3
LW
1055 }
1056 else {
c277df42
IZ
1057 /* Several toplevels. Best we can is to set minlen. */
1058 I32 fake;
1059
1060 DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
1061 scan = r->program + 1;
3280af22 1062 minlen = study_chunk(&scan, &fake, scan + PL_regsize, NULL, 0);
c277df42 1063 r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
a0d0e21e
LW
1064 }
1065
a0d0e21e 1066 r->minlen = minlen;
3280af22 1067 if (PL_regseen & REG_SEEN_GPOS)
c277df42 1068 r->reganch |= ROPT_GPOS_SEEN;
3280af22 1069 if (PL_regseen & REG_SEEN_LOOKBEHIND)
c277df42 1070 r->reganch |= ROPT_LOOKBEHIND_SEEN;
3280af22 1071 if (PL_regseen & REG_SEEN_EVAL)
ce862d02 1072 r->reganch |= ROPT_EVAL_SEEN;
3280af22
NIS
1073 Newz(1002, r->startp, PL_regnpar, char*);
1074 Newz(1002, r->endp, PL_regnpar, char*);
a0d0e21e
LW
1075 DEBUG_r(regdump(r));
1076 return(r);
a687059c
LW
1077}
1078
1079/*
1080 - reg - regular expression, i.e. main body or parenthesized thing
1081 *
1082 * Caller must absorb opening parenthesis.
1083 *
1084 * Combining parenthesis handling with the base level of regular expression
1085 * is a trifle forced, but the need to tie the tails of the branches to what
1086 * follows makes it hard to avoid.
1087 */
76e3520e 1088STATIC regnode *
8ac85365 1089reg(I32 paren, I32 *flagp)
c277df42 1090 /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
a687059c 1091{
5c0ca799 1092 dTHR;
c277df42
IZ
1093 register regnode *ret; /* Will be the head of the group. */
1094 register regnode *br;
1095 register regnode *lastbr;
1096 register regnode *ender = 0;
a0d0e21e 1097 register I32 parno = 0;
3280af22 1098 I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0;
c277df42 1099 char c;
a0d0e21e 1100
821b33a5 1101 *flagp = 0; /* Tentatively. */
a0d0e21e
LW
1102
1103 /* Make an OPEN node, if parenthesized. */
1104 if (paren) {
3280af22 1105 if (*PL_regcomp_parse == '?') {
ca9dfc88
IZ
1106 U16 posflags = 0, negflags = 0;
1107 U16 *flagsp = &posflags;
0f5d15d6 1108 int logical = 0;
ca9dfc88 1109
3280af22
NIS
1110 PL_regcomp_parse++;
1111 paren = *PL_regcomp_parse++;
c277df42 1112 ret = NULL; /* For look-ahead/behind. */
a0d0e21e 1113 switch (paren) {
c277df42 1114 case '<':
3280af22
NIS
1115 PL_regseen |= REG_SEEN_LOOKBEHIND;
1116 if (*PL_regcomp_parse == '!')
c277df42 1117 paren = ',';
3280af22 1118 if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!')
c277df42 1119 goto unknown;
3280af22 1120 PL_regcomp_parse++;
a0d0e21e
LW
1121 case '=':
1122 case '!':
3280af22 1123 PL_seen_zerolen++;
c277df42
IZ
1124 case ':':
1125 case '>':
a0d0e21e
LW
1126 break;
1127 case '$':
1128 case '@':
c277df42 1129 FAIL2("Sequence (?%c...) not implemented", (int)paren);
a0d0e21e
LW
1130 break;
1131 case '#':
3280af22
NIS
1132 while (*PL_regcomp_parse && *PL_regcomp_parse != ')')
1133 PL_regcomp_parse++;
1134 if (*PL_regcomp_parse != ')')
c277df42 1135 FAIL("Sequence (?#... not terminated");
a0d0e21e
LW
1136 nextchar();
1137 *flagp = TRYAGAIN;
1138 return NULL;
0f5d15d6
IZ
1139 case 'p':
1140 logical = 1;
1141 paren = *PL_regcomp_parse++;
1142 /* FALL THROUGH */
c277df42
IZ
1143 case '{':
1144 {
1145 dTHR;
1146 I32 count = 1, n = 0;
1147 char c;
3280af22 1148 char *s = PL_regcomp_parse;
c277df42
IZ
1149 SV *sv;
1150 OP_4tree *sop, *rop;
1151
3280af22
NIS
1152 PL_seen_zerolen++;
1153 PL_regseen |= REG_SEEN_EVAL;
1154 while (count && (c = *PL_regcomp_parse)) {
1155 if (c == '\\' && PL_regcomp_parse[1])
1156 PL_regcomp_parse++;
c277df42
IZ
1157 else if (c == '{')
1158 count++;
1159 else if (c == '}')
1160 count--;
3280af22 1161 PL_regcomp_parse++;
c277df42 1162 }
3280af22 1163 if (*PL_regcomp_parse != ')')
c277df42
IZ
1164 FAIL("Sequence (?{...}) not terminated or not {}-balanced");
1165 if (!SIZE_ONLY) {
1166 AV *av;
1167
3280af22 1168 if (PL_regcomp_parse - 1 - s)
79cb57f6 1169 sv = newSVpvn(s, PL_regcomp_parse - 1 - s);
c277df42 1170 else
79cb57f6 1171 sv = newSVpvn("", 0);
c277df42
IZ
1172
1173 rop = sv_compile_2op(sv, &sop, "re", &av);
1174
dfad63ad 1175 n = add_data(3, "nop");
3280af22 1176 PL_regcomp_rx->data->data[n] = (void*)rop;
dfad63ad
HS
1177 PL_regcomp_rx->data->data[n+1] = (void*)sop;
1178 PL_regcomp_rx->data->data[n+2] = (void*)av;
c277df42 1179 SvREFCNT_dec(sv);
a0ed51b3 1180 }
e24b16f9
GS
1181 else { /* First pass */
1182 if (PL_reginterp_cnt < ++PL_seen_evals
1183 && PL_curcop != &PL_compiling)
2cd61cdb
IZ
1184 /* No compiled RE interpolated, has runtime
1185 components ===> unsafe. */
1186 FAIL("Eval-group not allowed at runtime, use re 'eval'");
3280af22 1187 if (PL_tainted)
cc6b7395 1188 FAIL("Eval-group in insecure regular expression");
c277df42
IZ
1189 }
1190
1191 nextchar();
0f5d15d6
IZ
1192 if (logical) {
1193 ret = reg_node(LOGICAL);
1194 if (!SIZE_ONLY)
1195 ret->flags = 2;
1196 regtail(ret, reganode(EVAL, n));
1197 return ret;
1198 }
c277df42
IZ
1199 return reganode(EVAL, n);
1200 }
1201 case '(':
1202 {
3280af22
NIS
1203 if (PL_regcomp_parse[0] == '?') {
1204 if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!'
1205 || PL_regcomp_parse[1] == '<'
1206 || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */
c277df42
IZ
1207 I32 flag;
1208
1209 ret = reg_node(LOGICAL);
0f5d15d6
IZ
1210 if (!SIZE_ONLY)
1211 ret->flags = 1;
c277df42
IZ
1212 regtail(ret, reg(1, &flag));
1213 goto insert_if;
1214 }
a0ed51b3
LW
1215 }
1216 else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) {
3280af22 1217 parno = atoi(PL_regcomp_parse++);
c277df42 1218
3280af22
NIS
1219 while (isDIGIT(*PL_regcomp_parse))
1220 PL_regcomp_parse++;
c277df42
IZ
1221 ret = reganode(GROUPP, parno);
1222 if ((c = *nextchar()) != ')')
1223 FAIL2("Switch (?(number%c not recognized", c);
1224 insert_if:
1225 regtail(ret, reganode(IFTHEN, 0));
1226 br = regbranch(&flags, 1);
1227 if (br == NULL)
1228 br = reganode(LONGJMP, 0);
1229 else
1230 regtail(br, reganode(LONGJMP, 0));
1231 c = *nextchar();
d1b80229
IZ
1232 if (flags&HASWIDTH)
1233 *flagp |= HASWIDTH;
c277df42
IZ
1234 if (c == '|') {
1235 lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */
1236 regbranch(&flags, 1);
1237 regtail(ret, lastbr);
d1b80229
IZ
1238 if (flags&HASWIDTH)
1239 *flagp |= HASWIDTH;
c277df42 1240 c = *nextchar();
a0ed51b3
LW
1241 }
1242 else
c277df42
IZ
1243 lastbr = NULL;
1244 if (c != ')')
1245 FAIL("Switch (?(condition)... contains too many branches");
1246 ender = reg_node(TAIL);
1247 regtail(br, ender);
1248 if (lastbr) {
1249 regtail(lastbr, ender);
1250 regtail(NEXTOPER(NEXTOPER(lastbr)), ender);
a0ed51b3
LW
1251 }
1252 else
c277df42
IZ
1253 regtail(ret, ender);
1254 return ret;
a0ed51b3
LW
1255 }
1256 else {
3280af22 1257 FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse);
c277df42
IZ
1258 }
1259 }
1b1626e4 1260 case 0:
c277df42 1261 FAIL("Sequence (? incomplete");
1b1626e4 1262 break;
a0d0e21e 1263 default:
3280af22 1264 --PL_regcomp_parse;
ca9dfc88 1265 parse_flags:
3280af22
NIS
1266 while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) {
1267 if (*PL_regcomp_parse != 'o')
1268 pmflag(flagsp, *PL_regcomp_parse);
1269 ++PL_regcomp_parse;
ca9dfc88 1270 }
3280af22 1271 if (*PL_regcomp_parse == '-') {
ca9dfc88 1272 flagsp = &negflags;
3280af22 1273 ++PL_regcomp_parse;
ca9dfc88 1274 goto parse_flags;
48c036b1 1275 }
3280af22
NIS
1276 PL_regflags |= posflags;
1277 PL_regflags &= ~negflags;
1278 if (*PL_regcomp_parse == ':') {
1279 PL_regcomp_parse++;
ca9dfc88
IZ
1280 paren = ':';
1281 break;
1282 }
c277df42 1283 unknown:
3280af22
NIS
1284 if (*PL_regcomp_parse != ')')
1285 FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse);
a0d0e21e
LW
1286 nextchar();
1287 *flagp = TRYAGAIN;
1288 return NULL;
1289 }
1290 }
1291 else {
3280af22
NIS
1292 parno = PL_regnpar;
1293 PL_regnpar++;
a0d0e21e 1294 ret = reganode(OPEN, parno);
c277df42 1295 open = 1;
a0d0e21e 1296 }
a0ed51b3
LW
1297 }
1298 else
a0d0e21e
LW
1299 ret = NULL;
1300
1301 /* Pick up the branches, linking them together. */
c277df42 1302 br = regbranch(&flags, 1);
a0d0e21e
LW
1303 if (br == NULL)
1304 return(NULL);
3280af22
NIS
1305 if (*PL_regcomp_parse == '|') {
1306 if (!SIZE_ONLY && PL_extralen) {
c277df42 1307 reginsert(BRANCHJ, br);
a0ed51b3
LW
1308 }
1309 else
c277df42
IZ
1310 reginsert(BRANCH, br);
1311 have_branch = 1;
1312 if (SIZE_ONLY)
3280af22 1313 PL_extralen += 1; /* For BRANCHJ-BRANCH. */
a0ed51b3
LW
1314 }
1315 else if (paren == ':') {
c277df42
IZ
1316 *flagp |= flags&SIMPLE;
1317 }
1318 if (open) { /* Starts with OPEN. */
1319 regtail(ret, br); /* OPEN -> first. */
a0ed51b3
LW
1320 }
1321 else if (paren != '?') /* Not Conditional */
a0d0e21e 1322 ret = br;
821b33a5
IZ
1323 if (flags&HASWIDTH)
1324 *flagp |= HASWIDTH;
a0d0e21e 1325 *flagp |= flags&SPSTART;
c277df42 1326 lastbr = br;
3280af22
NIS
1327 while (*PL_regcomp_parse == '|') {
1328 if (!SIZE_ONLY && PL_extralen) {
c277df42
IZ
1329 ender = reganode(LONGJMP,0);
1330 regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
1331 }
1332 if (SIZE_ONLY)
3280af22 1333 PL_extralen += 2; /* Account for LONGJMP. */
a0d0e21e 1334 nextchar();
c277df42 1335 br = regbranch(&flags, 0);
a687059c 1336 if (br == NULL)
a0d0e21e 1337 return(NULL);
c277df42
IZ
1338 regtail(lastbr, br); /* BRANCH -> BRANCH. */
1339 lastbr = br;
821b33a5
IZ
1340 if (flags&HASWIDTH)
1341 *flagp |= HASWIDTH;
a687059c 1342 *flagp |= flags&SPSTART;
a0d0e21e
LW
1343 }
1344
c277df42
IZ
1345 if (have_branch || paren != ':') {
1346 /* Make a closing node, and hook it on the end. */
1347 switch (paren) {
1348 case ':':
1349 ender = reg_node(TAIL);
1350 break;
1351 case 1:
1352 ender = reganode(CLOSE, parno);
1353 break;
1354 case '<':
c277df42
IZ
1355 case ',':
1356 case '=':
1357 case '!':
c277df42 1358 *flagp &= ~HASWIDTH;
821b33a5
IZ
1359 /* FALL THROUGH */
1360 case '>':
1361 ender = reg_node(SUCCEED);
c277df42
IZ
1362 break;
1363 case 0:
1364 ender = reg_node(END);
1365 break;
1366 }
1367 regtail(lastbr, ender);
a0d0e21e 1368
c277df42
IZ
1369 if (have_branch) {
1370 /* Hook the tails of the branches to the closing node. */
1371 for (br = ret; br != NULL; br = regnext(br)) {
1372 regoptail(br, ender);
1373 }
1374 }
a0d0e21e 1375 }
c277df42
IZ
1376
1377 {
1378 char *p;
1379 static char parens[] = "=!<,>";
1380
1381 if (paren && (p = strchr(parens, paren))) {
1382 int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
1383 int flag = (p - parens) > 1;
1384
1385 if (paren == '>')
1386 node = SUSPEND, flag = 0;
1387 reginsert(node,ret);
c277df42 1388 ret->flags = flag;
c277df42
IZ
1389 regtail(ret, reg_node(TAIL));
1390 }
a0d0e21e
LW
1391 }
1392
1393 /* Check for proper termination. */
ce3e6498
HS
1394 if (paren) {
1395 PL_regflags = oregflags;
1396 if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') {
1397 FAIL("unmatched () in regexp");
1398 }
a0ed51b3
LW
1399 }
1400 else if (!paren && PL_regcomp_parse < PL_regxend) {
3280af22 1401 if (*PL_regcomp_parse == ')') {
a0d0e21e 1402 FAIL("unmatched () in regexp");
a0ed51b3
LW
1403 }
1404 else
a0d0e21e
LW
1405 FAIL("junk on end of regexp"); /* "Can't happen". */
1406 /* NOTREACHED */
1407 }
a687059c 1408
a0d0e21e 1409 return(ret);
a687059c
LW
1410}
1411
1412/*
1413 - regbranch - one alternative of an | operator
1414 *
1415 * Implements the concatenation operator.
1416 */
76e3520e 1417STATIC regnode *
c277df42 1418regbranch(I32 *flagp, I32 first)
a687059c 1419{
5c0ca799 1420 dTHR;
c277df42
IZ
1421 register regnode *ret;
1422 register regnode *chain = NULL;
1423 register regnode *latest;
1424 I32 flags = 0, c = 0;
a0d0e21e 1425
c277df42
IZ
1426 if (first)
1427 ret = NULL;
1428 else {
3280af22 1429 if (!SIZE_ONLY && PL_extralen)
c277df42
IZ
1430 ret = reganode(BRANCHJ,0);
1431 else
1432 ret = reg_node(BRANCH);
1433 }
1434
1435 if (!first && SIZE_ONLY)
3280af22 1436 PL_extralen += 1; /* BRANCHJ */
c277df42
IZ
1437
1438 *flagp = WORST; /* Tentatively. */
a0d0e21e 1439
3280af22 1440 PL_regcomp_parse--;
a0d0e21e 1441 nextchar();
3280af22 1442 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') {
a0d0e21e
LW
1443 flags &= ~TRYAGAIN;
1444 latest = regpiece(&flags);
1445 if (latest == NULL) {
1446 if (flags & TRYAGAIN)
1447 continue;
1448 return(NULL);
a0ed51b3
LW
1449 }
1450 else if (ret == NULL)
c277df42 1451 ret = latest;
a0d0e21e 1452 *flagp |= flags&HASWIDTH;
c277df42 1453 if (chain == NULL) /* First piece. */
a0d0e21e
LW
1454 *flagp |= flags&SPSTART;
1455 else {
3280af22 1456 PL_regnaughty++;
a0d0e21e 1457 regtail(chain, latest);
a687059c 1458 }
a0d0e21e 1459 chain = latest;
c277df42
IZ
1460 c++;
1461 }
1462 if (chain == NULL) { /* Loop ran zero times. */
1463 chain = reg_node(NOTHING);
1464 if (ret == NULL)
1465 ret = chain;
1466 }
1467 if (c == 1) {
1468 *flagp |= flags&SIMPLE;
a0d0e21e 1469 }
a687059c 1470
a0d0e21e 1471 return(ret);
a687059c
LW
1472}
1473
1474/*
1475 - regpiece - something followed by possible [*+?]
1476 *
1477 * Note that the branching code sequences used for ? and the general cases
1478 * of * and + are somewhat optimized: they use the same NOTHING node as
1479 * both the endmarker for their branch list and the body of the last branch.
1480 * It might seem that this node could be dispensed with entirely, but the
1481 * endmarker role is not redundant.
1482 */
76e3520e 1483STATIC regnode *
8ac85365 1484regpiece(I32 *flagp)
a687059c 1485{
5c0ca799 1486 dTHR;
c277df42 1487 register regnode *ret;
a0d0e21e
LW
1488 register char op;
1489 register char *next;
1490 I32 flags;
3280af22 1491 char *origparse = PL_regcomp_parse;
a0d0e21e
LW
1492 char *maxpos;
1493 I32 min;
c277df42 1494 I32 max = REG_INFTY;
a0d0e21e
LW
1495
1496 ret = regatom(&flags);
1497 if (ret == NULL) {
1498 if (flags & TRYAGAIN)
1499 *flagp |= TRYAGAIN;
1500 return(NULL);
1501 }
1502
3280af22 1503 op = *PL_regcomp_parse;
a0d0e21e 1504
3280af22
NIS
1505 if (op == '{' && regcurly(PL_regcomp_parse)) {
1506 next = PL_regcomp_parse + 1;
a0d0e21e
LW
1507 maxpos = Nullch;
1508 while (isDIGIT(*next) || *next == ',') {
1509 if (*next == ',') {
1510 if (maxpos)
1511 break;
1512 else
1513 maxpos = next;
a687059c 1514 }
a0d0e21e
LW
1515 next++;
1516 }
1517 if (*next == '}') { /* got one */
1518 if (!maxpos)
1519 maxpos = next;
3280af22
NIS
1520 PL_regcomp_parse++;
1521 min = atoi(PL_regcomp_parse);
a0d0e21e
LW
1522 if (*maxpos == ',')
1523 maxpos++;
1524 else
3280af22 1525 maxpos = PL_regcomp_parse;
a0d0e21e
LW
1526 max = atoi(maxpos);
1527 if (!max && *maxpos != '0')
c277df42
IZ
1528 max = REG_INFTY; /* meaning "infinity" */
1529 else if (max >= REG_INFTY)
1530 FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
3280af22 1531 PL_regcomp_parse = next;
a0d0e21e
LW
1532 nextchar();
1533
1534 do_curly:
1535 if ((flags&SIMPLE)) {
3280af22 1536 PL_regnaughty += 2 + PL_regnaughty / 2;
a0d0e21e
LW
1537 reginsert(CURLY, ret);
1538 }
1539 else {
3280af22 1540 PL_regnaughty += 4 + PL_regnaughty; /* compound interest */
c277df42 1541 regtail(ret, reg_node(WHILEM));
3280af22 1542 if (!SIZE_ONLY && PL_extralen) {
c277df42
IZ
1543 reginsert(LONGJMP,ret);
1544 reginsert(NOTHING,ret);
1545 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
1546 }
a0d0e21e 1547 reginsert(CURLYX,ret);
3280af22 1548 if (!SIZE_ONLY && PL_extralen)
c277df42
IZ
1549 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
1550 regtail(ret, reg_node(NOTHING));
1551 if (SIZE_ONLY)
3280af22 1552 PL_extralen += 3;
a0d0e21e 1553 }
c277df42 1554 ret->flags = 0;
a0d0e21e
LW
1555
1556 if (min > 0)
821b33a5
IZ
1557 *flagp = WORST;
1558 if (max > 0)
1559 *flagp |= HASWIDTH;
a0d0e21e 1560 if (max && max < min)
c277df42
IZ
1561 FAIL("Can't do {n,m} with n > m");
1562 if (!SIZE_ONLY) {
1563 ARG1_SET(ret, min);
1564 ARG2_SET(ret, max);
a687059c 1565 }
a687059c 1566
a0d0e21e 1567 goto nest_check;
a687059c 1568 }
a0d0e21e 1569 }
a687059c 1570
a0d0e21e
LW
1571 if (!ISMULT1(op)) {
1572 *flagp = flags;
a687059c 1573 return(ret);
a0d0e21e 1574 }
bb20fd44 1575
c277df42 1576#if 0 /* Now runtime fix should be reliable. */
bb20fd44 1577 if (!(flags&HASWIDTH) && op != '?')
c277df42
IZ
1578 FAIL("regexp *+ operand could be empty");
1579#endif
bb20fd44 1580
a0d0e21e
LW
1581 nextchar();
1582
821b33a5 1583 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
a0d0e21e
LW
1584
1585 if (op == '*' && (flags&SIMPLE)) {
1586 reginsert(STAR, ret);
c277df42 1587 ret->flags = 0;
3280af22 1588 PL_regnaughty += 4;
a0d0e21e
LW
1589 }
1590 else if (op == '*') {
1591 min = 0;
1592 goto do_curly;
a0ed51b3
LW
1593 }
1594 else if (op == '+' && (flags&SIMPLE)) {
a0d0e21e 1595 reginsert(PLUS, ret);
c277df42 1596 ret->flags = 0;
3280af22 1597 PL_regnaughty += 3;
a0d0e21e
LW
1598 }
1599 else if (op == '+') {
1600 min = 1;
1601 goto do_curly;
a0ed51b3
LW
1602 }
1603 else if (op == '?') {
a0d0e21e
LW
1604 min = 0; max = 1;
1605 goto do_curly;
1606 }
1607 nest_check:
17feb5d5 1608 if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
599cee73 1609 warner(WARN_UNSAFE, "%.*s matches null string many times",
3280af22 1610 PL_regcomp_parse - origparse, origparse);
a0d0e21e
LW
1611 }
1612
3280af22 1613 if (*PL_regcomp_parse == '?') {
a0d0e21e
LW
1614 nextchar();
1615 reginsert(MINMOD, ret);
c277df42 1616 regtail(ret, ret + NODE_STEP_REGNODE);
a0d0e21e 1617 }
3280af22 1618 if (ISMULT2(PL_regcomp_parse))
a0d0e21e
LW
1619 FAIL("nested *?+ in regexp");
1620
1621 return(ret);
a687059c
LW
1622}
1623
1624/*
1625 - regatom - the lowest level
1626 *
1627 * Optimization: gobbles an entire sequence of ordinary characters so that
1628 * it can turn them into a single node, which is smaller to store and
1629 * faster to run. Backslashed characters are exceptions, each becoming a
1630 * separate node; the code is simpler that way and it's not worth fixing.
1631 *
1632 * [Yes, it is worth fixing, some scripts can run twice the speed.]
1633 */
76e3520e 1634STATIC regnode *
8ac85365 1635regatom(I32 *flagp)
a687059c 1636{
5c0ca799 1637 dTHR;
c277df42 1638 register regnode *ret = 0;
a0d0e21e
LW
1639 I32 flags;
1640
1641 *flagp = WORST; /* Tentatively. */
1642
1643tryagain:
3280af22 1644 switch (*PL_regcomp_parse) {
a0d0e21e 1645 case '^':
3280af22 1646 PL_seen_zerolen++;
a0d0e21e 1647 nextchar();
3280af22 1648 if (PL_regflags & PMf_MULTILINE)
c277df42 1649 ret = reg_node(MBOL);
3280af22 1650 else if (PL_regflags & PMf_SINGLELINE)
c277df42 1651 ret = reg_node(SBOL);
a0d0e21e 1652 else
c277df42 1653 ret = reg_node(BOL);
a0d0e21e
LW
1654 break;
1655 case '$':
3280af22
NIS
1656 if (PL_regcomp_parse[1])
1657 PL_seen_zerolen++;
a0d0e21e 1658 nextchar();
3280af22 1659 if (PL_regflags & PMf_MULTILINE)
c277df42 1660 ret = reg_node(MEOL);
3280af22 1661 else if (PL_regflags & PMf_SINGLELINE)
c277df42 1662 ret = reg_node(SEOL);
a0d0e21e 1663 else
c277df42 1664 ret = reg_node(EOL);
a0d0e21e
LW
1665 break;
1666 case '.':
1667 nextchar();
a0ed51b3
LW
1668 if (UTF) {
1669 if (PL_regflags & PMf_SINGLELINE)
1670 ret = reg_node(SANYUTF8);
1671 else
1672 ret = reg_node(ANYUTF8);
1673 *flagp |= HASWIDTH;
1674 }
1675 else {
1676 if (PL_regflags & PMf_SINGLELINE)
1677 ret = reg_node(SANY);
1678 else
22c35a8c 1679 ret = reg_node(REG_ANY);
a0ed51b3
LW
1680 *flagp |= HASWIDTH|SIMPLE;
1681 }
3280af22 1682 PL_regnaughty++;
a0d0e21e
LW
1683 break;
1684 case '[':
3280af22 1685 PL_regcomp_parse++;
a0ed51b3 1686 ret = (UTF ? regclassutf8() : regclass());
a14b48bc
LW
1687 if (*PL_regcomp_parse != ']')
1688 FAIL("unmatched [] in regexp");
1689 nextchar();
a0d0e21e
LW
1690 *flagp |= HASWIDTH|SIMPLE;
1691 break;
1692 case '(':
1693 nextchar();
1694 ret = reg(1, &flags);
1695 if (ret == NULL) {
1696 if (flags & TRYAGAIN)
1697 goto tryagain;
1698 return(NULL);
1699 }
c277df42 1700 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
a0d0e21e
LW
1701 break;
1702 case '|':
1703 case ')':
1704 if (flags & TRYAGAIN) {
1705 *flagp |= TRYAGAIN;
1706 return NULL;
1707 }
3280af22 1708 FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse);
a0d0e21e
LW
1709 /* Supposed to be caught earlier. */
1710 break;
85afd4ae 1711 case '{':
3280af22
NIS
1712 if (!regcurly(PL_regcomp_parse)) {
1713 PL_regcomp_parse++;
85afd4ae
CS
1714 goto defchar;
1715 }
1716 /* FALL THROUGH */
a0d0e21e
LW
1717 case '?':
1718 case '+':
1719 case '*':
3115e423 1720 FAIL("?+*{} follows nothing in regexp");
a0d0e21e
LW
1721 break;
1722 case '\\':
3280af22 1723 switch (*++PL_regcomp_parse) {
a0d0e21e 1724 case 'A':
3280af22 1725 PL_seen_zerolen++;
c277df42 1726 ret = reg_node(SBOL);
a0d0e21e
LW
1727 *flagp |= SIMPLE;
1728 nextchar();
1729 break;
1730 case 'G':
c277df42 1731 ret = reg_node(GPOS);
3280af22 1732 PL_regseen |= REG_SEEN_GPOS;
a0d0e21e
LW
1733 *flagp |= SIMPLE;
1734 nextchar();
1735 break;
1736 case 'Z':
c277df42 1737 ret = reg_node(SEOL);
a0d0e21e
LW
1738 *flagp |= SIMPLE;
1739 nextchar();
1740 break;
b85d18e9
IZ
1741 case 'z':
1742 ret = reg_node(EOS);
1743 *flagp |= SIMPLE;
3280af22 1744 PL_seen_zerolen++; /* Do not optimize RE away */
b85d18e9
IZ
1745 nextchar();
1746 break;
a0ed51b3
LW
1747 case 'C':
1748 ret = reg_node(SANY);
1749 *flagp |= HASWIDTH|SIMPLE;
1750 nextchar();
1751 break;
1752 case 'X':
1753 ret = reg_node(CLUMP);
1754 *flagp |= HASWIDTH;
1755 nextchar();
1756 if (UTF && !PL_utf8_mark)
dfe13c55 1757 is_utf8_mark((U8*)"~"); /* preload table */
a0ed51b3 1758 break;
a0d0e21e 1759 case 'w':
a0ed51b3
LW
1760 ret = reg_node(
1761 UTF
1762 ? (LOC ? ALNUMLUTF8 : ALNUMUTF8)
1763 : (LOC ? ALNUML : ALNUM));
a0d0e21e
LW
1764 *flagp |= HASWIDTH|SIMPLE;
1765 nextchar();
a0ed51b3 1766 if (UTF && !PL_utf8_alnum)
dfe13c55 1767 is_utf8_alnum((U8*)"a"); /* preload table */
a0d0e21e
LW
1768 break;
1769 case 'W':
a0ed51b3
LW
1770 ret = reg_node(
1771 UTF
1772 ? (LOC ? NALNUMLUTF8 : NALNUMUTF8)
1773 : (LOC ? NALNUML : NALNUM));
a0d0e21e
LW
1774 *flagp |= HASWIDTH|SIMPLE;
1775 nextchar();
a0ed51b3 1776 if (UTF && !PL_utf8_alnum)
dfe13c55 1777 is_utf8_alnum((U8*)"a"); /* preload table */
a0d0e21e
LW
1778 break;
1779 case 'b':
3280af22 1780 PL_seen_zerolen++;
f5c9036e 1781 PL_regseen |= REG_SEEN_LOOKBEHIND;
a0ed51b3
LW
1782 ret = reg_node(
1783 UTF
1784 ? (LOC ? BOUNDLUTF8 : BOUNDUTF8)
1785 : (LOC ? BOUNDL : BOUND));
a0d0e21e
LW
1786 *flagp |= SIMPLE;
1787 nextchar();
a0ed51b3 1788 if (UTF && !PL_utf8_alnum)
dfe13c55 1789 is_utf8_alnum((U8*)"a"); /* preload table */
a0d0e21e
LW
1790 break;
1791 case 'B':
3280af22 1792 PL_seen_zerolen++;
f5c9036e 1793 PL_regseen |= REG_SEEN_LOOKBEHIND;
a0ed51b3
LW
1794 ret = reg_node(
1795 UTF
1796 ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8)
1797 : (LOC ? NBOUNDL : NBOUND));
a0d0e21e
LW
1798 *flagp |= SIMPLE;
1799 nextchar();
a0ed51b3 1800 if (UTF && !PL_utf8_alnum)
dfe13c55 1801 is_utf8_alnum((U8*)"a"); /* preload table */
a0d0e21e
LW
1802 break;
1803 case 's':
a0ed51b3
LW
1804 ret = reg_node(
1805 UTF
1806 ? (LOC ? SPACELUTF8 : SPACEUTF8)
1807 : (LOC ? SPACEL : SPACE));
a0d0e21e
LW
1808 *flagp |= HASWIDTH|SIMPLE;
1809 nextchar();
a0ed51b3 1810 if (UTF && !PL_utf8_space)
dfe13c55 1811 is_utf8_space((U8*)" "); /* preload table */
a0d0e21e
LW
1812 break;
1813 case 'S':
a0ed51b3
LW
1814 ret = reg_node(
1815 UTF
1816 ? (LOC ? NSPACELUTF8 : NSPACEUTF8)
1817 : (LOC ? NSPACEL : NSPACE));
a0d0e21e
LW
1818 *flagp |= HASWIDTH|SIMPLE;
1819 nextchar();
a0ed51b3 1820 if (UTF && !PL_utf8_space)
dfe13c55 1821 is_utf8_space((U8*)" "); /* preload table */
a0d0e21e
LW
1822 break;
1823 case 'd':
a0ed51b3 1824 ret = reg_node(UTF ? DIGITUTF8 : DIGIT);
a0d0e21e
LW
1825 *flagp |= HASWIDTH|SIMPLE;
1826 nextchar();
a0ed51b3 1827 if (UTF && !PL_utf8_digit)
dfe13c55 1828 is_utf8_digit((U8*)"1"); /* preload table */
a0d0e21e
LW
1829 break;
1830 case 'D':
a0ed51b3 1831 ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT);
a0d0e21e
LW
1832 *flagp |= HASWIDTH|SIMPLE;
1833 nextchar();
a0ed51b3 1834 if (UTF && !PL_utf8_digit)
dfe13c55 1835 is_utf8_digit((U8*)"1"); /* preload table */
a0d0e21e 1836 break;
a14b48bc
LW
1837 case 'p':
1838 case 'P':
1839 { /* a lovely hack--pretend we saw [\pX] instead */
1840 char* oldregxend = PL_regxend;
1841
1842 if (PL_regcomp_parse[1] == '{') {
1843 PL_regxend = strchr(PL_regcomp_parse, '}');
1844 if (!PL_regxend)
1845 FAIL("Missing right brace on \\p{}");
1846 PL_regxend++;
1847 }
1848 else
1849 PL_regxend = PL_regcomp_parse + 2;
1850 PL_regcomp_parse--;
1851
1852 ret = regclassutf8();
1853
1854 PL_regxend = oldregxend;
1855 PL_regcomp_parse--;
1856 nextchar();
1857 *flagp |= HASWIDTH|SIMPLE;
1858 }
1859 break;
a0d0e21e
LW
1860 case 'n':
1861 case 'r':
1862 case 't':
1863 case 'f':
1864 case 'e':
1865 case 'a':
1866 case 'x':
1867 case 'c':
1868 case '0':
1869 goto defchar;
1870 case '1': case '2': case '3': case '4':
1871 case '5': case '6': case '7': case '8': case '9':
1872 {
3280af22 1873 I32 num = atoi(PL_regcomp_parse);
a0d0e21e 1874
3280af22 1875 if (num > 9 && num >= PL_regnpar)
a0d0e21e
LW
1876 goto defchar;
1877 else {
3280af22 1878 if (!SIZE_ONLY && num > PL_regcomp_rx->nparens)
ef64f398 1879 FAIL("reference to nonexistent group");
3280af22 1880 PL_regsawback = 1;
a0ed51b3
LW
1881 ret = reganode(FOLD
1882 ? (LOC ? REFFL : REFF)
c8756f30 1883 : REF, num);
a0d0e21e 1884 *flagp |= HASWIDTH;
3280af22
NIS
1885 while (isDIGIT(*PL_regcomp_parse))
1886 PL_regcomp_parse++;
1887 PL_regcomp_parse--;
a0d0e21e
LW
1888 nextchar();
1889 }
1890 }
1891 break;
1892 case '\0':
3280af22 1893 if (PL_regcomp_parse >= PL_regxend)
a0d0e21e
LW
1894 FAIL("trailing \\ in regexp");
1895 /* FALL THROUGH */
1896 default:
c9f97d15
IZ
1897 /* Do not generate `unrecognized' warnings here, we fall
1898 back into the quick-grab loop below */
a0d0e21e
LW
1899 goto defchar;
1900 }
1901 break;
4633a7c4
LW
1902
1903 case '#':
3280af22
NIS
1904 if (PL_regflags & PMf_EXTENDED) {
1905 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++;
1906 if (PL_regcomp_parse < PL_regxend)
4633a7c4
LW
1907 goto tryagain;
1908 }
1909 /* FALL THROUGH */
1910
a0d0e21e
LW
1911 default: {
1912 register I32 len;
a0ed51b3 1913 register UV ender;
a0d0e21e 1914 register char *p;
c277df42 1915 char *oldp, *s;
a0d0e21e
LW
1916 I32 numlen;
1917
3280af22 1918 PL_regcomp_parse++;
a0d0e21e
LW
1919
1920 defchar:
a0ed51b3
LW
1921 ret = reg_node(FOLD
1922 ? (LOC ? EXACTFL : EXACTF)
bbce6d69 1923 : EXACT);
161b471a 1924 s = (char *) OPERAND(ret);
c277df42 1925 regc(0, s++); /* save spot for len */
3280af22
NIS
1926 for (len = 0, p = PL_regcomp_parse - 1;
1927 len < 127 && p < PL_regxend;
a0d0e21e
LW
1928 len++)
1929 {
1930 oldp = p;
5b5a24f7 1931
3280af22
NIS
1932 if (PL_regflags & PMf_EXTENDED)
1933 p = regwhite(p, PL_regxend);
a0d0e21e
LW
1934 switch (*p) {
1935 case '^':
1936 case '$':
1937 case '.':
1938 case '[':
1939 case '(':
1940 case ')':
1941 case '|':
1942 goto loopdone;
1943 case '\\':
1944 switch (*++p) {
1945 case 'A':
1946 case 'G':
1947 case 'Z':
b85d18e9 1948 case 'z':
a0d0e21e
LW
1949 case 'w':
1950 case 'W':
1951 case 'b':
1952 case 'B':
1953 case 's':
1954 case 'S':
1955 case 'd':
1956 case 'D':
a14b48bc
LW
1957 case 'p':
1958 case 'P':
a0d0e21e
LW
1959 --p;
1960 goto loopdone;
1961 case 'n':
1962 ender = '\n';
1963 p++;
a687059c 1964 break;
a0d0e21e
LW
1965 case 'r':
1966 ender = '\r';
1967 p++;
a687059c 1968 break;
a0d0e21e
LW
1969 case 't':
1970 ender = '\t';
1971 p++;
a687059c 1972 break;
a0d0e21e
LW
1973 case 'f':
1974 ender = '\f';
1975 p++;
a687059c 1976 break;
a0d0e21e
LW
1977 case 'e':
1978 ender = '\033';
1979 p++;
a687059c 1980 break;
a0d0e21e
LW
1981 case 'a':
1982 ender = '\007';
1983 p++;
a687059c 1984 break;
a0d0e21e 1985 case 'x':
a0ed51b3
LW
1986 if (*++p == '{') {
1987 char* e = strchr(p, '}');
1988
1989 if (!e)
1990 FAIL("Missing right brace on \\x{}");
1991 else if (UTF) {
1992 ender = scan_hex(p + 1, e - p, &numlen);
1993 if (numlen + len >= 127) { /* numlen is generous */
1994 p--;
1995 goto loopdone;
1996 }
1997 p = e + 1;
1998 }
1999 else
2000 FAIL("Can't use \\x{} without 'use utf8' declaration");
2001 }
2002 else {
2003 ender = scan_hex(p, 2, &numlen);
2004 p += numlen;
2005 }
a687059c 2006 break;
a0d0e21e
LW
2007 case 'c':
2008 p++;
bbce6d69 2009 ender = UCHARAT(p++);
2010 ender = toCTRL(ender);
a687059c 2011 break;
a0d0e21e
LW
2012 case '0': case '1': case '2': case '3':case '4':
2013 case '5': case '6': case '7': case '8':case '9':
2014 if (*p == '0' ||
3280af22 2015 (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) {
a0d0e21e
LW
2016 ender = scan_oct(p, 3, &numlen);
2017 p += numlen;
2018 }
2019 else {
2020 --p;
2021 goto loopdone;
a687059c
LW
2022 }
2023 break;
a0d0e21e 2024 case '\0':
3280af22 2025 if (p >= PL_regxend)
a687059c
LW
2026 FAIL("trailing \\ in regexp");
2027 /* FALL THROUGH */
a0d0e21e 2028 default:
c9f97d15
IZ
2029 if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p))
2030 warner(WARN_UNSAFE,
2031 "/%.127s/: Unrecognized escape \\%c passed through",
2032 PL_regprecomp,
2033 *p);
a0ed51b3 2034 goto normal_default;
a0d0e21e
LW
2035 }
2036 break;
a687059c 2037 default:
a0ed51b3
LW
2038 normal_default:
2039 if ((*p & 0xc0) == 0xc0 && UTF) {
dfe13c55 2040 ender = utf8_to_uv((U8*)p, &numlen);
a0ed51b3
LW
2041 p += numlen;
2042 }
2043 else
2044 ender = *p++;
a0d0e21e 2045 break;
a687059c 2046 }
3280af22
NIS
2047 if (PL_regflags & PMf_EXTENDED)
2048 p = regwhite(p, PL_regxend);
a0ed51b3
LW
2049 if (UTF && FOLD) {
2050 if (LOC)
2051 ender = toLOWER_LC_uni(ender);
2052 else
2053 ender = toLOWER_uni(ender);
2054 }
a0d0e21e
LW
2055 if (ISMULT2(p)) { /* Back off on ?+*. */
2056 if (len)
2057 p = oldp;
a0ed51b3
LW
2058 else if (ender >= 0x80 && UTF) {
2059 reguni(ender, s, &numlen);
2060 s += numlen;
2061 len += numlen;
2062 }
a0d0e21e
LW
2063 else {
2064 len++;
c277df42 2065 regc(ender, s++);
a0d0e21e
LW
2066 }
2067 break;
a687059c 2068 }
a0ed51b3
LW
2069 if (ender >= 0x80 && UTF) {
2070 reguni(ender, s, &numlen);
2071 s += numlen;
2072 len += numlen - 1;
2073 }
2074 else
2075 regc(ender, s++);
a0d0e21e
LW
2076 }
2077 loopdone:
3280af22 2078 PL_regcomp_parse = p - 1;
a0d0e21e
LW
2079 nextchar();
2080 if (len < 0)
2081 FAIL("internal disaster in regexp");
2082 if (len > 0)
2083 *flagp |= HASWIDTH;
2084 if (len == 1)
2085 *flagp |= SIMPLE;
c277df42 2086 if (!SIZE_ONLY)
a0d0e21e 2087 *OPERAND(ret) = len;
c277df42
IZ
2088 regc('\0', s++);
2089 if (SIZE_ONLY) {
3280af22 2090 PL_regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
a0ed51b3
LW
2091 }
2092 else {
3280af22 2093 PL_regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
c277df42 2094 }
a687059c 2095 }
a0d0e21e
LW
2096 break;
2097 }
a687059c 2098
a0d0e21e 2099 return(ret);
a687059c
LW
2100}
2101
873ef191 2102STATIC char *
8ac85365 2103regwhite(char *p, char *e)
5b5a24f7
CS
2104{
2105 while (p < e) {
2106 if (isSPACE(*p))
2107 ++p;
2108 else if (*p == '#') {
2109 do {
2110 p++;
2111 } while (p < e && *p != '\n');
2112 }
2113 else
2114 break;
2115 }
2116 return p;
2117}
2118
620e46c5
JH
2119/* parse POSIX character classes like [[:foo:]] */
2120STATIC char*
2121regpposixcc(I32 value)
2122{
11b8faa4 2123 dTHR;
620e46c5
JH
2124 char *posixcc = 0;
2125
2126 if (value == '[' && PL_regcomp_parse + 1 < PL_regxend &&
2127 /* I smell either [: or [= or [. -- POSIX has been here, right? */
2128 (*PL_regcomp_parse == ':' ||
2129 *PL_regcomp_parse == '=' ||
2130 *PL_regcomp_parse == '.')) {
2131 char c = *PL_regcomp_parse;
2132 char* s = PL_regcomp_parse++;
2133
2134 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c)
2135 PL_regcomp_parse++;
2136 if (PL_regcomp_parse == PL_regxend)
2137 /* Grandfather lone [:, [=, [. */
2138 PL_regcomp_parse = s;
2139 else {
2140 PL_regcomp_parse++; /* skip over the c */
2141 if (*PL_regcomp_parse == ']') {
2142 /* Not Implemented Yet.
2143 * (POSIX Extended Character Classes, that is)
2144 * The text between e.g. [: and :] would start
2145 * at s + 1 and stop at regcomp_parse - 2. */
2146 if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
2147 warner(WARN_UNSAFE,
2148 "Character class syntax [%c %c] is reserved for future extensions", c, c);
2149 PL_regcomp_parse++; /* skip over the ending ] */
2150 posixcc = s + 1;
2151 }
2152 }
2153 }
2154
2155 return posixcc;
2156}
2157
76e3520e 2158STATIC regnode *
8ac85365 2159regclass(void)
a687059c 2160{
5c0ca799 2161 dTHR;
c277df42 2162 register char *opnd, *s;
a0ed51b3
LW
2163 register I32 value;
2164 register I32 lastvalue = 1234;
a0d0e21e 2165 register I32 range = 0;
c277df42 2166 register regnode *ret;
a0d0e21e
LW
2167 register I32 def;
2168 I32 numlen;
2169
3280af22 2170 s = opnd = (char *) OPERAND(PL_regcode);
c277df42 2171 ret = reg_node(ANYOF);
a0ed51b3 2172 for (value = 0; value < 33; value++)
c277df42 2173 regc(0, s++);
3280af22
NIS
2174 if (*PL_regcomp_parse == '^') { /* Complement of range. */
2175 PL_regnaughty++;
2176 PL_regcomp_parse++;
c277df42 2177 if (!SIZE_ONLY)
bbce6d69 2178 *opnd |= ANYOF_INVERT;
2179 }
c277df42 2180 if (!SIZE_ONLY) {
3280af22 2181 PL_regcode += ANY_SKIP;
a0ed51b3 2182 if (FOLD)
bbce6d69 2183 *opnd |= ANYOF_FOLD;
a0ed51b3 2184 if (LOC)
bbce6d69 2185 *opnd |= ANYOF_LOCALE;
a0ed51b3
LW
2186 }
2187 else {
3280af22 2188 PL_regsize += ANY_SKIP;
a0d0e21e 2189 }
3280af22 2190 if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-')
a0d0e21e 2191 goto skipcond; /* allow 1st char to be ] or - */
3280af22 2192 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') {
a0d0e21e 2193 skipcond:
a0ed51b3 2194 value = UCHARAT(PL_regcomp_parse++);
620e46c5
JH
2195 if (value == '[')
2196 (void)regpposixcc(value); /* ignore the return value for now */
2197 else if (value == '\\') {
a0ed51b3
LW
2198 value = UCHARAT(PL_regcomp_parse++);
2199 switch (value) {
a0d0e21e 2200 case 'w':
ae5c130c 2201 if (!SIZE_ONLY) {
a0ed51b3 2202 if (LOC)
bbce6d69 2203 *opnd |= ANYOF_ALNUML;
ae5c130c 2204 else {
a0ed51b3
LW
2205 for (value = 0; value < 256; value++)
2206 if (isALNUM(value))
2207 ANYOF_SET(opnd, value);
ae5c130c 2208 }
bbce6d69 2209 }
a0ed51b3 2210 lastvalue = 1234;
a0d0e21e
LW
2211 continue;
2212 case 'W':
ae5c130c 2213 if (!SIZE_ONLY) {
a0ed51b3 2214 if (LOC)
bbce6d69 2215 *opnd |= ANYOF_NALNUML;
ae5c130c 2216 else {
a0ed51b3
LW
2217 for (value = 0; value < 256; value++)
2218 if (!isALNUM(value))
2219 ANYOF_SET(opnd, value);
ae5c130c 2220 }
bbce6d69 2221 }
a0ed51b3 2222 lastvalue = 1234;
a0d0e21e
LW
2223 continue;
2224 case 's':
ae5c130c 2225 if (!SIZE_ONLY) {
a0ed51b3 2226 if (LOC)
bbce6d69 2227 *opnd |= ANYOF_SPACEL;
ae5c130c 2228 else {
a0ed51b3
LW
2229 for (value = 0; value < 256; value++)
2230 if (isSPACE(value))
2231 ANYOF_SET(opnd, value);
ae5c130c 2232 }
bbce6d69 2233 }
a0ed51b3 2234 lastvalue = 1234;
a0d0e21e
LW
2235 continue;
2236 case 'S':
ae5c130c 2237 if (!SIZE_ONLY) {
a0ed51b3 2238 if (LOC)
bbce6d69 2239 *opnd |= ANYOF_NSPACEL;
ae5c130c 2240 else {
a0ed51b3
LW
2241 for (value = 0; value < 256; value++)
2242 if (!isSPACE(value))
2243 ANYOF_SET(opnd, value);
ae5c130c 2244 }
bbce6d69 2245 }
a0ed51b3 2246 lastvalue = 1234;
a0d0e21e
LW
2247 continue;
2248 case 'd':
ae5c130c 2249 if (!SIZE_ONLY) {
a0ed51b3
LW
2250 for (value = '0'; value <= '9'; value++)
2251 ANYOF_SET(opnd, value);
ae5c130c 2252 }
a0ed51b3 2253 lastvalue = 1234;
a0d0e21e
LW
2254 continue;
2255 case 'D':
ae5c130c 2256 if (!SIZE_ONLY) {
a0ed51b3
LW
2257 for (value = 0; value < '0'; value++)
2258 ANYOF_SET(opnd, value);
2259 for (value = '9' + 1; value < 256; value++)
2260 ANYOF_SET(opnd, value);
ae5c130c 2261 }
a0ed51b3 2262 lastvalue = 1234;
a0d0e21e
LW
2263 continue;
2264 case 'n':
a0ed51b3 2265 value = '\n';
a0d0e21e
LW
2266 break;
2267 case 'r':
a0ed51b3 2268 value = '\r';
a0d0e21e
LW
2269 break;
2270 case 't':
a0ed51b3 2271 value = '\t';
a0d0e21e
LW
2272 break;
2273 case 'f':
a0ed51b3 2274 value = '\f';
a0d0e21e
LW
2275 break;
2276 case 'b':
a0ed51b3 2277 value = '\b';
a0d0e21e
LW
2278 break;
2279 case 'e':
a0ed51b3 2280 value = '\033';
a0d0e21e
LW
2281 break;
2282 case 'a':
a0ed51b3 2283 value = '\007';
a0d0e21e
LW
2284 break;
2285 case 'x':
a0ed51b3 2286 value = scan_hex(PL_regcomp_parse, 2, &numlen);
3280af22 2287 PL_regcomp_parse += numlen;
a0d0e21e
LW
2288 break;
2289 case 'c':
a0ed51b3
LW
2290 value = UCHARAT(PL_regcomp_parse++);
2291 value = toCTRL(value);
a0d0e21e
LW
2292 break;
2293 case '0': case '1': case '2': case '3': case '4':
2294 case '5': case '6': case '7': case '8': case '9':
a0ed51b3 2295 value = scan_oct(--PL_regcomp_parse, 3, &numlen);
3280af22 2296 PL_regcomp_parse += numlen;
a0d0e21e
LW
2297 break;
2298 }
2299 }
2300 if (range) {
a0ed51b3 2301 if (lastvalue > value)
a0d0e21e
LW
2302 FAIL("invalid [] range in regexp");
2303 range = 0;
2304 }
2305 else {
a0ed51b3 2306 lastvalue = value;
3280af22
NIS
2307 if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend &&
2308 PL_regcomp_parse[1] != ']') {
2309 PL_regcomp_parse++;
a0d0e21e
LW
2310 range = 1;
2311 continue; /* do it next time */
2312 }
a687059c 2313 }
ae5c130c 2314 if (!SIZE_ONLY) {
8ada0baa
JH
2315#ifndef ASCIIish
2316 if ((isLOWER(lastvalue) && isLOWER(value)) ||
c8dba6f3
GS
2317 (isUPPER(lastvalue) && isUPPER(value)))
2318 {
2319 I32 i;
8ada0baa
JH
2320 if (isLOWER(lastvalue)) {
2321 for (i = lastvalue; i <= value; i++)
2322 if (isLOWER(i))
2323 ANYOF_SET(opnd, i);
2324 } else {
2325 for (i = lastvalue; i <= value; i++)
2326 if (isUPPER(i))
2327 ANYOF_SET(opnd, i);
2328 }
2329 }
2330 else
2331#endif
2332 for ( ; lastvalue <= value; lastvalue++)
2333 ANYOF_SET(opnd, lastvalue);
2334 }
a0ed51b3 2335 lastvalue = value;
a0d0e21e 2336 }
ae5c130c
GS
2337 /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
2338 if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) {
a0ed51b3
LW
2339 for (value = 0; value < 256; ++value) {
2340 if (ANYOF_TEST(opnd, value)) {
22c35a8c 2341 I32 cf = PL_fold[value];
ae5c130c
GS
2342 ANYOF_SET(opnd, cf);
2343 }
2344 }
2345 *opnd &= ~ANYOF_FOLD;
2346 }
2347 /* optimize inverted simple patterns (e.g. [^a-z]) */
2348 if (!SIZE_ONLY && (*opnd & 0xFF) == ANYOF_INVERT) {
a0ed51b3
LW
2349 for (value = 0; value < 32; ++value)
2350 opnd[1 + value] ^= 0xFF;
ae5c130c
GS
2351 *opnd = 0;
2352 }
a0d0e21e
LW
2353 return ret;
2354}
2355
a0ed51b3
LW
2356STATIC regnode *
2357regclassutf8(void)
2358{
2359 register char *opnd, *e;
2360 register U32 value;
2361 register U32 lastvalue = 123456;
2362 register I32 range = 0;
2363 register regnode *ret;
2364 I32 numlen;
2365 I32 n;
2366 SV *listsv;
2367 U8 flags = 0;
c485e607 2368 dTHR;
a0ed51b3
LW
2369
2370 if (*PL_regcomp_parse == '^') { /* Complement of range. */
2371 PL_regnaughty++;
2372 PL_regcomp_parse++;
2373 if (!SIZE_ONLY)
2374 flags |= ANYOF_INVERT;
2375 }
2376 if (!SIZE_ONLY) {
2377 if (FOLD)
2378 flags |= ANYOF_FOLD;
2379 if (LOC)
2380 flags |= ANYOF_LOCALE;
79cb57f6 2381 listsv = newSVpvn("# comment\n",10);
a0ed51b3
LW
2382 }
2383
2384 if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-')
2385 goto skipcond; /* allow 1st char to be ] or - */
2386
2387 while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') {
2388 skipcond:
dfe13c55 2389 value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
a0ed51b3
LW
2390 PL_regcomp_parse += numlen;
2391
620e46c5
JH
2392 if (value == '[')
2393 (void)regpposixcc(value); /* ignore the return value for now */
2394 else if (value == '\\') {
dfe13c55 2395 value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
a0ed51b3
LW
2396 PL_regcomp_parse += numlen;
2397 switch (value) {
2398 case 'w':
2399 if (!SIZE_ONLY) {
2400 if (LOC)
2401 flags |= ANYOF_ALNUML;
2402
2403 sv_catpvf(listsv, "+utf8::IsAlnum\n");
2404 }
2405 lastvalue = 123456;
2406 continue;
2407 case 'W':
2408 if (!SIZE_ONLY) {
2409 if (LOC)
2410 flags |= ANYOF_NALNUML;
2411
2412 sv_catpvf(listsv,
2413 "-utf8::IsAlpha\n-utf8::IsDigit\n0000\t%04x\n%04x\tffff\n",
2414 '_' - 1,
2415 '_' + 1);
2416 }
2417 lastvalue = 123456;
2418 continue;
2419 case 's':
2420 if (!SIZE_ONLY) {
2421 if (LOC)
2422 flags |= ANYOF_SPACEL;
2423 sv_catpvf(listsv, "+utf8::IsSpace\n");
2424 if (!PL_utf8_space)
dfe13c55 2425 is_utf8_space((U8*)" ");
a0ed51b3
LW
2426 }
2427 lastvalue = 123456;
2428 continue;
2429 case 'S':
2430 if (!SIZE_ONLY) {
2431 if (LOC)
2432 flags |= ANYOF_NSPACEL;
2433 sv_catpvf(listsv,
2434 "!utf8::IsSpace\n");
2435 if (!PL_utf8_space)
dfe13c55 2436 is_utf8_space((U8*)" ");
a0ed51b3
LW
2437 }
2438 lastvalue = 123456;
2439 continue;
2440 case 'd':
2441 if (!SIZE_ONLY) {
2442 sv_catpvf(listsv, "+utf8::IsDigit\n");
2443 }
2444 lastvalue = 123456;
2445 continue;
2446 case 'D':
2447 if (!SIZE_ONLY) {
2448 sv_catpvf(listsv,
2449 "!utf8::IsDigit\n");
2450 }
2451 lastvalue = 123456;
2452 continue;
2453 case 'p':
2454 case 'P':
2455 if (*PL_regcomp_parse == '{') {
2456 e = strchr(PL_regcomp_parse++, '}');
2457 if (!e)
2458 FAIL("Missing right brace on \\p{}");
2459 n = e - PL_regcomp_parse;
2460 }
2461 else {
2462 e = PL_regcomp_parse;
2463 n = 1;
2464 }
2465 if (!SIZE_ONLY) {
2466 if (value == 'p')
2467 sv_catpvf(listsv, "+utf8::%.*s\n", n, PL_regcomp_parse);
2468 else
2469 sv_catpvf(listsv,
2470 "!utf8::%.*s\n", n, PL_regcomp_parse);
2471 }
2472 PL_regcomp_parse = e + 1;
2473 lastvalue = 123456;
2474 continue;
2475 case 'n':
2476 value = '\n';
2477 break;
2478 case 'r':
2479 value = '\r';
2480 break;
2481 case 't':
2482 value = '\t';
2483 break;
2484 case 'f':
2485 value = '\f';
2486 break;
2487 case 'b':
2488 value = '\b';
2489 break;
2490 case 'e':
2491 value = '\033';
2492 break;
2493 case 'a':
2494 value = '\007';
2495 break;
2496 case 'x':
2497 if (*PL_regcomp_parse == '{') {
2498 e = strchr(PL_regcomp_parse++, '}');
2499 if (!e)
2500 FAIL("Missing right brace on \\x{}");
f96ec2a2 2501 value = scan_hex(PL_regcomp_parse, e - PL_regcomp_parse, &numlen);
a0ed51b3
LW
2502 PL_regcomp_parse = e + 1;
2503 }
2504 else {
2505 value = scan_hex(PL_regcomp_parse, 2, &numlen);
2506 PL_regcomp_parse += numlen;
2507 }
2508 break;
2509 case 'c':
2510 value = UCHARAT(PL_regcomp_parse++);
2511 value = toCTRL(value);
2512 break;
2513 case '0': case '1': case '2': case '3': case '4':
2514 case '5': case '6': case '7': case '8': case '9':
2515 value = scan_oct(--PL_regcomp_parse, 3, &numlen);
2516 PL_regcomp_parse += numlen;
2517 break;
2518 }
2519 }
2520 if (range) {
2521 if (lastvalue > value)
2522 FAIL("invalid [] range in regexp");
2523 if (!SIZE_ONLY)
2524 sv_catpvf(listsv, "%04x\t%04x\n", lastvalue, value);
2525 lastvalue = value;
2526 range = 0;
2527 }
2528 else {
2529 lastvalue = value;
2530 if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend &&
2531 PL_regcomp_parse[1] != ']') {
2532 PL_regcomp_parse++;
2533 range = 1;
2534 continue; /* do it next time */
2535 }
2536 if (!SIZE_ONLY)
2537 sv_catpvf(listsv, "%04x\n", value);
2538 }
2539 }
a0ed51b3
LW
2540
2541 ret = reganode(ANYOFUTF8, 0);
2542
2543 if (!SIZE_ONLY) {
2544 SV *rv = swash_init("utf8", "", listsv, 1, 0);
2545 SvREFCNT_dec(listsv);
2546 n = add_data(1,"s");
2547 PL_regcomp_rx->data->data[n] = (void*)rv;
2548 ARG1_SET(ret, flags);
2549 ARG2_SET(ret, n);
2550 }
2551
2552 return ret;
2553}
2554
76e3520e 2555STATIC char*
8ac85365 2556nextchar(void)
a0d0e21e 2557{
5c0ca799 2558 dTHR;
3280af22 2559 char* retval = PL_regcomp_parse++;
a0d0e21e 2560
4633a7c4 2561 for (;;) {
3280af22
NIS
2562 if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' &&
2563 PL_regcomp_parse[2] == '#') {
2564 while (*PL_regcomp_parse && *PL_regcomp_parse != ')')
2565 PL_regcomp_parse++;
2566 PL_regcomp_parse++;
4633a7c4
LW
2567 continue;
2568 }
3280af22
NIS
2569 if (PL_regflags & PMf_EXTENDED) {
2570 if (isSPACE(*PL_regcomp_parse)) {
2571 PL_regcomp_parse++;
748a9306
LW
2572 continue;
2573 }
3280af22
NIS
2574 else if (*PL_regcomp_parse == '#') {
2575 while (*PL_regcomp_parse && *PL_regcomp_parse != '\n')
2576 PL_regcomp_parse++;
2577 PL_regcomp_parse++;
748a9306
LW
2578 continue;
2579 }
748a9306 2580 }
4633a7c4 2581 return retval;
a0d0e21e 2582 }
a687059c
LW
2583}
2584
2585/*
c277df42 2586- reg_node - emit a node
a0d0e21e 2587*/
76e3520e 2588STATIC regnode * /* Location. */
c277df42 2589reg_node(U8 op)
a687059c 2590{
5c0ca799 2591 dTHR;
c277df42
IZ
2592 register regnode *ret;
2593 register regnode *ptr;
a687059c 2594
3280af22 2595 ret = PL_regcode;
c277df42 2596 if (SIZE_ONLY) {
6b88bc9c 2597 SIZE_ALIGN(PL_regsize);
3280af22 2598 PL_regsize += 1;
a0d0e21e
LW
2599 return(ret);
2600 }
a687059c 2601
c277df42 2602 NODE_ALIGN_FILL(ret);
a0d0e21e 2603 ptr = ret;
c277df42 2604 FILL_ADVANCE_NODE(ptr, op);
3280af22 2605 PL_regcode = ptr;
a687059c 2606
a0d0e21e 2607 return(ret);
a687059c
LW
2608}
2609
2610/*
a0d0e21e
LW
2611- reganode - emit a node with an argument
2612*/
76e3520e 2613STATIC regnode * /* Location. */
c277df42 2614reganode(U8 op, U32 arg)
fe14fcc3 2615{
5c0ca799 2616 dTHR;
c277df42
IZ
2617 register regnode *ret;
2618 register regnode *ptr;
fe14fcc3 2619
3280af22 2620 ret = PL_regcode;
c277df42 2621 if (SIZE_ONLY) {
6b88bc9c 2622 SIZE_ALIGN(PL_regsize);
3280af22 2623 PL_regsize += 2;
a0d0e21e
LW
2624 return(ret);
2625 }
fe14fcc3 2626
c277df42 2627 NODE_ALIGN_FILL(ret);
a0d0e21e 2628 ptr = ret;
c277df42 2629 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
3280af22 2630 PL_regcode = ptr;
fe14fcc3 2631
a0d0e21e 2632 return(ret);
fe14fcc3
LW
2633}
2634
2635/*
a0ed51b3
LW
2636- regc - emit (if appropriate) a Unicode character
2637*/
2638STATIC void
2639reguni(UV uv, char* s, I32* lenp)
2640{
c485e607 2641 dTHR;
a0ed51b3 2642 if (SIZE_ONLY) {
dfe13c55 2643 U8 tmpbuf[10];
a0ed51b3
LW
2644 *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf;
2645 }
2646 else
dfe13c55 2647 *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s;
a0ed51b3
LW
2648
2649}
2650
2651/*
a0d0e21e
LW
2652- regc - emit (if appropriate) a byte of code
2653*/
76e3520e 2654STATIC void
c277df42 2655regc(U8 b, char* s)
a687059c 2656{
5c0ca799 2657 dTHR;
c277df42
IZ
2658 if (!SIZE_ONLY)
2659 *s = b;
a687059c
LW
2660}
2661
2662/*
a0d0e21e
LW
2663- reginsert - insert an operator in front of already-emitted operand
2664*
2665* Means relocating the operand.
2666*/
76e3520e 2667STATIC void
c277df42 2668reginsert(U8 op, regnode *opnd)
a687059c 2669{
5c0ca799 2670 dTHR;
c277df42
IZ
2671 register regnode *src;
2672 register regnode *dst;
2673 register regnode *place;
2674 register int offset = regarglen[(U8)op];
2675
22c35a8c 2676/* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
c277df42
IZ
2677
2678 if (SIZE_ONLY) {
3280af22 2679 PL_regsize += NODE_STEP_REGNODE + offset;
a0d0e21e
LW
2680 return;
2681 }
a687059c 2682
3280af22
NIS
2683 src = PL_regcode;
2684 PL_regcode += NODE_STEP_REGNODE + offset;
2685 dst = PL_regcode;
a0d0e21e 2686 while (src > opnd)
c277df42 2687 StructCopy(--src, --dst, regnode);
a0d0e21e
LW
2688
2689 place = opnd; /* Op node, where operand used to be. */
c277df42
IZ
2690 src = NEXTOPER(place);
2691 FILL_ADVANCE_NODE(place, op);
2692 Zero(src, offset, regnode);
a687059c
LW
2693}
2694
2695/*
c277df42 2696- regtail - set the next-pointer at the end of a node chain of p to val.
a0d0e21e 2697*/
76e3520e 2698STATIC void
c277df42 2699regtail(regnode *p, regnode *val)
a687059c 2700{
5c0ca799 2701 dTHR;
c277df42
IZ
2702 register regnode *scan;
2703 register regnode *temp;
a0d0e21e
LW
2704 register I32 offset;
2705
c277df42 2706 if (SIZE_ONLY)
a0d0e21e
LW
2707 return;
2708
2709 /* Find last node. */
2710 scan = p;
2711 for (;;) {
2712 temp = regnext(scan);
2713 if (temp == NULL)
2714 break;
2715 scan = temp;
2716 }
a687059c 2717
c277df42
IZ
2718 if (reg_off_by_arg[OP(scan)]) {
2719 ARG_SET(scan, val - scan);
a0ed51b3
LW
2720 }
2721 else {
c277df42
IZ
2722 NEXT_OFF(scan) = val - scan;
2723 }
a687059c
LW
2724}
2725
2726/*
a0d0e21e
LW
2727- regoptail - regtail on operand of first argument; nop if operandless
2728*/
76e3520e 2729STATIC void
c277df42 2730regoptail(regnode *p, regnode *val)
a687059c 2731{
5c0ca799 2732 dTHR;
a0d0e21e 2733 /* "Operandless" and "op != BRANCH" are synonymous in practice. */
c277df42
IZ
2734 if (p == NULL || SIZE_ONLY)
2735 return;
22c35a8c 2736 if (PL_regkind[(U8)OP(p)] == BRANCH) {
c277df42 2737 regtail(NEXTOPER(p), val);
a0ed51b3 2738 }
22c35a8c 2739 else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
c277df42 2740 regtail(NEXTOPER(NEXTOPER(p)), val);
a0ed51b3
LW
2741 }
2742 else
a0d0e21e 2743 return;
a687059c
LW
2744}
2745
2746/*
2747 - regcurly - a little FSA that accepts {\d+,?\d*}
2748 */
79072805 2749STATIC I32
8ac85365 2750regcurly(register char *s)
a687059c
LW
2751{
2752 if (*s++ != '{')
2753 return FALSE;
f0fcb552 2754 if (!isDIGIT(*s))
a687059c 2755 return FALSE;
f0fcb552 2756 while (isDIGIT(*s))
a687059c
LW
2757 s++;
2758 if (*s == ',')
2759 s++;
f0fcb552 2760 while (isDIGIT(*s))
a687059c
LW
2761 s++;
2762 if (*s != '}')
2763 return FALSE;
2764 return TRUE;
2765}
2766
a687059c 2767
76e3520e 2768STATIC regnode *
c277df42
IZ
2769dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
2770{
35ff7856 2771#ifdef DEBUGGING
c277df42
IZ
2772 register char op = EXACT; /* Arbitrary non-END op. */
2773 register regnode *next, *onode;
2774
2775 while (op != END && (!last || node < last)) {
2776 /* While that wasn't END last time... */
2777
2778 NODE_ALIGN(node);
2779 op = OP(node);
2780 if (op == CLOSE)
2781 l--;
2782 next = regnext(node);
2783 /* Where, what. */
2784 if (OP(node) == OPTIMIZED)
2785 goto after_print;
2786 regprop(sv, node);
54dc92de 2787 PerlIO_printf(Perl_debug_log, "%4d:%*s%s", node - start,
c277df42
IZ
2788 2*l + 1, "", SvPVX(sv));
2789 if (next == NULL) /* Next ptr. */
2790 PerlIO_printf(Perl_debug_log, "(0)");
2791 else
2792 PerlIO_printf(Perl_debug_log, "(%d)", next - start);
2793 (void)PerlIO_putc(Perl_debug_log, '\n');
2794 after_print:
22c35a8c 2795 if (PL_regkind[(U8)op] == BRANCHJ) {
c277df42
IZ
2796 register regnode *nnode = (OP(next) == LONGJMP
2797 ? regnext(next)
2798 : next);
2799 if (last && nnode > last)
2800 nnode = last;
2801 node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
a0ed51b3 2802 }
22c35a8c 2803 else if (PL_regkind[(U8)op] == BRANCH) {
c277df42 2804 node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
a0ed51b3
LW
2805 }
2806 else if ( op == CURLY) { /* `next' might be very big: optimizer */
c277df42
IZ
2807 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
2808 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
a0ed51b3 2809 }
22c35a8c 2810 else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
c277df42
IZ
2811 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
2812 next, sv, l + 1);
a0ed51b3
LW
2813 }
2814 else if ( op == PLUS || op == STAR) {
c277df42 2815 node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
a0ed51b3
LW
2816 }
2817 else if (op == ANYOF) {
c277df42
IZ
2818 node = NEXTOPER(node);
2819 node += ANY_SKIP;
a0ed51b3 2820 }
22c35a8c 2821 else if (PL_regkind[(U8)op] == EXACT) {
c277df42
IZ
2822 /* Literal string, where present. */
2823 node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode);
2824 node = NEXTOPER(node);
a0ed51b3
LW
2825 }
2826 else {
c277df42
IZ
2827 node = NEXTOPER(node);
2828 node += regarglen[(U8)op];
2829 }
2830 if (op == CURLYX || op == OPEN)
2831 l++;
2832 else if (op == WHILEM)
2833 l--;
2834 }
17c3b450 2835#endif /* DEBUGGING */
c277df42
IZ
2836 return node;
2837}
2838
a687059c 2839/*
fd181c75 2840 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
a687059c
LW
2841 */
2842void
8ac85365 2843regdump(regexp *r)
a687059c 2844{
35ff7856 2845#ifdef DEBUGGING
5c0ca799 2846 dTHR;
46fc3d4c 2847 SV *sv = sv_newmortal();
a687059c 2848
c277df42 2849 (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
a0d0e21e
LW
2850
2851 /* Header fields of interest. */
c277df42
IZ
2852 if (r->anchored_substr)
2853 PerlIO_printf(Perl_debug_log, "anchored `%s%s%s'%s at %d ",
3280af22 2854 PL_colors[0],
c277df42 2855 SvPVX(r->anchored_substr),
3280af22 2856 PL_colors[1],
c277df42
IZ
2857 SvTAIL(r->anchored_substr) ? "$" : "",
2858 r->anchored_offset);
2859 if (r->float_substr)
2860 PerlIO_printf(Perl_debug_log, "floating `%s%s%s'%s at %d..%u ",
3280af22 2861 PL_colors[0],
c277df42 2862 SvPVX(r->float_substr),
3280af22 2863 PL_colors[1],
c277df42
IZ
2864 SvTAIL(r->float_substr) ? "$" : "",
2865 r->float_min_offset, r->float_max_offset);
2866 if (r->check_substr)
2867 PerlIO_printf(Perl_debug_log,
2868 r->check_substr == r->float_substr
2869 ? "(checking floating" : "(checking anchored");
2870 if (r->reganch & ROPT_NOSCAN)
2871 PerlIO_printf(Perl_debug_log, " noscan");
2872 if (r->reganch & ROPT_CHECK_ALL)
2873 PerlIO_printf(Perl_debug_log, " isall");
2874 if (r->check_substr)
2875 PerlIO_printf(Perl_debug_log, ") ");
2876
46fc3d4c 2877 if (r->regstclass) {
2878 regprop(sv, r->regstclass);
2879 PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
2880 }
774d564b 2881 if (r->reganch & ROPT_ANCH) {
2882 PerlIO_printf(Perl_debug_log, "anchored");
2883 if (r->reganch & ROPT_ANCH_BOL)
2884 PerlIO_printf(Perl_debug_log, "(BOL)");
c277df42
IZ
2885 if (r->reganch & ROPT_ANCH_MBOL)
2886 PerlIO_printf(Perl_debug_log, "(MBOL)");
774d564b 2887 if (r->reganch & ROPT_ANCH_GPOS)
2888 PerlIO_printf(Perl_debug_log, "(GPOS)");
2889 PerlIO_putc(Perl_debug_log, ' ');
2890 }
c277df42
IZ
2891 if (r->reganch & ROPT_GPOS_SEEN)
2892 PerlIO_printf(Perl_debug_log, "GPOS ");
a0d0e21e 2893 if (r->reganch & ROPT_SKIP)
760ac839 2894 PerlIO_printf(Perl_debug_log, "plus ");
a0d0e21e 2895 if (r->reganch & ROPT_IMPLICIT)
760ac839 2896 PerlIO_printf(Perl_debug_log, "implicit ");
760ac839 2897 PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
ce862d02
IZ
2898 if (r->reganch & ROPT_EVAL_SEEN)
2899 PerlIO_printf(Perl_debug_log, "with eval ");
760ac839 2900 PerlIO_printf(Perl_debug_log, "\n");
17c3b450 2901#endif /* DEBUGGING */
a687059c
LW
2902}
2903
2904/*
a0d0e21e
LW
2905- regprop - printable representation of opcode
2906*/
46fc3d4c 2907void
c277df42 2908regprop(SV *sv, regnode *o)
a687059c 2909{
35ff7856 2910#ifdef DEBUGGING
5c0ca799 2911 dTHR;
a0d0e21e
LW
2912 register char *p = 0;
2913
54dc92de 2914 sv_setpvn(sv, "", 0);
11343788 2915 switch (OP(o)) {
a0d0e21e
LW
2916 case BOL:
2917 p = "BOL";
2918 break;
2919 case MBOL:
2920 p = "MBOL";
2921 break;
2922 case SBOL:
2923 p = "SBOL";
2924 break;
2925 case EOL:
2926 p = "EOL";
2927 break;
b85d18e9
IZ
2928 case EOS:
2929 p = "EOS";
2930 break;
a0d0e21e
LW
2931 case MEOL:
2932 p = "MEOL";
2933 break;
2934 case SEOL:
2935 p = "SEOL";
2936 break;
22c35a8c 2937 case REG_ANY:
a0d0e21e
LW
2938 p = "ANY";
2939 break;
2940 case SANY:
2941 p = "SANY";
2942 break;
a0ed51b3
LW
2943 case ANYUTF8:
2944 p = "ANYUTF8";
2945 break;
2946 case SANYUTF8:
2947 p = "SANYUTF8";
2948 break;
2949 case ANYOFUTF8:
2950 p = "ANYOFUTF8";
2951 break;
a0d0e21e
LW
2952 case ANYOF:
2953 p = "ANYOF";
2954 break;
2955 case BRANCH:
2956 p = "BRANCH";
2957 break;
bbce6d69 2958 case EXACT:
3280af22 2959 sv_catpvf(sv, "EXACT <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]);
bbce6d69 2960 break;
2961 case EXACTF:
3280af22 2962 sv_catpvf(sv, "EXACTF <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]);
bbce6d69 2963 break;
2964 case EXACTFL:
3280af22 2965 sv_catpvf(sv, "EXACTFL <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]);
a0d0e21e
LW
2966 break;
2967 case NOTHING:
2968 p = "NOTHING";
2969 break;
c277df42
IZ
2970 case TAIL:
2971 p = "TAIL";
2972 break;
a0d0e21e
LW
2973 case BACK:
2974 p = "BACK";
2975 break;
2976 case END:
2977 p = "END";
2978 break;
a0d0e21e
LW
2979 case BOUND:
2980 p = "BOUND";
2981 break;
bbce6d69 2982 case BOUNDL:
2983 p = "BOUNDL";
2984 break;
a0d0e21e
LW
2985 case NBOUND:
2986 p = "NBOUND";
2987 break;
bbce6d69 2988 case NBOUNDL:
2989 p = "NBOUNDL";
a0d0e21e
LW
2990 break;
2991 case CURLY:
5dc0d613 2992 sv_catpvf(sv, "CURLY {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e 2993 break;
c277df42 2994 case CURLYM:
c277df42 2995 sv_catpvf(sv, "CURLYM[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
c277df42
IZ
2996 break;
2997 case CURLYN:
c277df42 2998 sv_catpvf(sv, "CURLYN[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
c277df42 2999 break;
a0d0e21e 3000 case CURLYX:
5dc0d613 3001 sv_catpvf(sv, "CURLYX {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e
LW
3002 break;
3003 case REF:
c277df42 3004 sv_catpvf(sv, "REF%d", ARG(o));
a0d0e21e 3005 break;
c8756f30 3006 case REFF:
c277df42 3007 sv_catpvf(sv, "REFF%d", ARG(o));
c8756f30
AK
3008 break;
3009 case REFFL:
c277df42 3010 sv_catpvf(sv, "REFFL%d", ARG(o));
c8756f30 3011 break;
a0d0e21e 3012 case OPEN:
c277df42 3013 sv_catpvf(sv, "OPEN%d", ARG(o));
a0d0e21e
LW
3014 break;
3015 case CLOSE:
c277df42 3016 sv_catpvf(sv, "CLOSE%d", ARG(o));
a0d0e21e
LW
3017 p = NULL;
3018 break;
3019 case STAR:
3020 p = "STAR";
3021 break;
3022 case PLUS:
3023 p = "PLUS";
3024 break;
3025 case MINMOD:
3026 p = "MINMOD";
3027 break;
774d564b 3028 case GPOS:
3029 p = "GPOS";
a0d0e21e
LW
3030 break;
3031 case UNLESSM:
c277df42 3032 sv_catpvf(sv, "UNLESSM[-%d]", o->flags);
a0d0e21e
LW
3033 break;
3034 case IFMATCH:
c277df42 3035 sv_catpvf(sv, "IFMATCH[-%d]", o->flags);
a0d0e21e
LW
3036 break;
3037 case SUCCEED:
3038 p = "SUCCEED";
3039 break;
3040 case WHILEM:
3041 p = "WHILEM";
3042 break;
bbce6d69 3043 case DIGIT:
3044 p = "DIGIT";
3045 break;
3046 case NDIGIT:
3047 p = "NDIGIT";
3048 break;
3049 case ALNUM:
3050 p = "ALNUM";
3051 break;
3052 case NALNUM:
3053 p = "NALNUM";
3054 break;
3055 case SPACE:
3056 p = "SPACE";
3057 break;
3058 case NSPACE:
3059 p = "NSPACE";
3060 break;
3061 case ALNUML:
3062 p = "ALNUML";
3063 break;
3064 case NALNUML:
3065 p = "NALNUML";
3066 break;
3067 case SPACEL:
3068 p = "SPACEL";
3069 break;
3070 case NSPACEL:
3071 p = "NSPACEL";
3072 break;
c277df42
IZ
3073 case EVAL:
3074 p = "EVAL";
3075 break;
3076 case LONGJMP:
3077 p = "LONGJMP";
3078 break;
3079 case BRANCHJ:
3080 p = "BRANCHJ";
3081 break;
3082 case IFTHEN:
3083 p = "IFTHEN";
3084 break;
3085 case GROUPP:
3086 sv_catpvf(sv, "GROUPP%d", ARG(o));
3087 break;
3088 case LOGICAL:
0f5d15d6 3089 sv_catpvf(sv, "LOGICAL[%d]", o->flags);
c277df42
IZ
3090 break;
3091 case SUSPEND:
3092 p = "SUSPEND";
3093 break;
3094 case RENUM:
3095 p = "RENUM";
3096 break;
3097 case OPTIMIZED:
3098 p = "OPTIMIZED";
3099 break;
a0d0e21e
LW
3100 default:
3101 FAIL("corrupted regexp opcode");
3102 }
46fc3d4c 3103 if (p)
3104 sv_catpv(sv, p);
17c3b450 3105#endif /* DEBUGGING */
35ff7856 3106}
a687059c 3107
2b69d0c2 3108void
8ac85365 3109pregfree(struct regexp *r)
a687059c 3110{
5c0ca799 3111 dTHR;
c277df42 3112 if (!r || (--r->refcnt > 0))
a0d0e21e 3113 return;
c277df42 3114 if (r->precomp)
a0d0e21e 3115 Safefree(r->precomp);
c277df42 3116 if (r->subbase)
a0d0e21e 3117 Safefree(r->subbase);
a193d654
GS
3118 if (r->substrs) {
3119 if (r->anchored_substr)
3120 SvREFCNT_dec(r->anchored_substr);
3121 if (r->float_substr)
3122 SvREFCNT_dec(r->float_substr);
2779dcf1 3123 Safefree(r->substrs);
a193d654 3124 }
c277df42
IZ
3125 if (r->data) {
3126 int n = r->data->count;
dfad63ad
HS
3127 AV* new_comppad = NULL;
3128 AV* old_comppad;
3129 SV** old_curpad;
3130
c277df42
IZ
3131 while (--n >= 0) {
3132 switch (r->data->what[n]) {
3133 case 's':
3134 SvREFCNT_dec((SV*)r->data->data[n]);
3135 break;
dfad63ad
HS
3136 case 'p':
3137 new_comppad = (AV*)r->data->data[n];
3138 break;
c277df42 3139 case 'o':
dfad63ad
HS
3140 if (new_comppad == NULL)
3141 croak("panic: pregfree comppad");
3142 old_comppad = PL_comppad;
3143 old_curpad = PL_curpad;
3144 PL_comppad = new_comppad;
3145 PL_curpad = AvARRAY(new_comppad);
c277df42 3146 op_free((OP_4tree*)r->data->data[n]);
dfad63ad
HS
3147 PL_comppad = old_comppad;
3148 PL_curpad = old_curpad;
3149 SvREFCNT_dec((SV*)new_comppad);
3150 new_comppad = NULL;
c277df42
IZ
3151 break;
3152 case 'n':
3153 break;
3154 default:
3155 FAIL2("panic: regfree data code '%c'", r->data->what[n]);
3156 }
3157 }
3158 Safefree(r->data->what);
3159 Safefree(r->data);
a0d0e21e
LW
3160 }
3161 Safefree(r->startp);
3162 Safefree(r->endp);
3163 Safefree(r);
a687059c 3164}
c277df42
IZ
3165
3166/*
3167 - regnext - dig the "next" pointer out of a node
3168 *
3169 * [Note, when REGALIGN is defined there are two places in regmatch()
3170 * that bypass this code for speed.]
3171 */
3172regnode *
3173regnext(register regnode *p)
3174{
5c0ca799 3175 dTHR;
c277df42
IZ
3176 register I32 offset;
3177
3280af22 3178 if (p == &PL_regdummy)
c277df42
IZ
3179 return(NULL);
3180
3181 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
3182 if (offset == 0)
3183 return(NULL);
3184
c277df42 3185 return(p+offset);
c277df42
IZ
3186}
3187
01f988be 3188STATIC void
c277df42 3189re_croak2(const char* pat1,const char* pat2,...)
c277df42
IZ
3190{
3191 va_list args;
3192 STRLEN l1 = strlen(pat1);
3193 STRLEN l2 = strlen(pat2);
3194 char buf[512];
3195 char *message;
3196
3197 if (l1 > 510)
3198 l1 = 510;
3199 if (l1 + l2 > 510)
3200 l2 = 510 - l1;
3201 Copy(pat1, buf, l1 , char);
3202 Copy(pat2, buf + l1, l2 , char);
3b818b81
GS
3203 buf[l1 + l2] = '\n';
3204 buf[l1 + l2 + 1] = '\0';
c277df42 3205 va_start(args, pat2);
c277df42
IZ
3206 message = mess(buf, &args);
3207 va_end(args);
3208 l1 = strlen(message);
3209 if (l1 > 512)
3210 l1 = 512;
3211 Copy(message, buf, l1 , char);
3212 buf[l1] = '\0'; /* Overwrite \n */
3213 croak("%s", buf);
3214}
a0ed51b3
LW
3215
3216/* XXX Here's a total kludge. But we need to re-enter for swash routines. */
3217
3218void
3219save_re_context(void)
c485e607
NIS
3220{
3221 dTHR;
a0ed51b3
LW
3222 SAVEPPTR(PL_bostr);
3223 SAVEPPTR(PL_regprecomp); /* uncompiled string. */
3224 SAVEI32(PL_regnpar); /* () count. */
3225 SAVEI32(PL_regsize); /* Code size. */
3226 SAVEI16(PL_regflags); /* are we folding, multilining? */
3227 SAVEPPTR(PL_reginput); /* String-input pointer. */
3228 SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */
3229 SAVEPPTR(PL_regeol); /* End of input, for $ check. */
3230 SAVESPTR(PL_regstartp); /* Pointer to startp array. */
3231 SAVESPTR(PL_regendp); /* Ditto for endp. */
3232 SAVESPTR(PL_reglastparen); /* Similarly for lastparen. */
3233 SAVEPPTR(PL_regtill); /* How far we are required to go. */
3234 SAVEI32(PL_regprev); /* char before regbol, \n if none */
3235 SAVESPTR(PL_reg_start_tmp); /* from regexec.c */
3236 PL_reg_start_tmp = 0;
3237 SAVEFREEPV(PL_reg_start_tmp);
3238 SAVEI32(PL_reg_start_tmpl); /* from regexec.c */
3239 PL_reg_start_tmpl = 0;
3240 SAVESPTR(PL_regdata);
3241 SAVEI32(PL_reg_flags); /* from regexec.c */
3242 SAVEI32(PL_reg_eval_set); /* from regexec.c */
3243 SAVEI32(PL_regnarrate); /* from regexec.c */
3244 SAVESPTR(PL_regprogram); /* from regexec.c */
3245 SAVEINT(PL_regindent); /* from regexec.c */
3246 SAVESPTR(PL_regcc); /* from regexec.c */
3247 SAVESPTR(PL_curcop);
3248 SAVESPTR(PL_regcomp_rx); /* from regcomp.c */
3249 SAVEI32(PL_regseen); /* from regcomp.c */
3250 SAVEI32(PL_regsawback); /* Did we see \1, ...? */
3251 SAVEI32(PL_regnaughty); /* How bad is this pattern? */
3252 SAVESPTR(PL_regcode); /* Code-emit pointer; &regdummy = don't */
3253 SAVEPPTR(PL_regxend); /* End of input for compile */
3254 SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */
54b6e2fa
IZ
3255 SAVESPTR(PL_reg_call_cc); /* from regexec.c */
3256 SAVESPTR(PL_reg_re); /* from regexec.c */
3257 SAVEPPTR(PL_reg_ganch); /* from regexec.c */
3258 SAVESPTR(PL_reg_sv); /* from regexec.c */
3259 SAVESPTR(PL_reg_magic); /* from regexec.c */
3260 SAVEI32(PL_reg_oldpos); /* from regexec.c */
3261 SAVESPTR(PL_reg_oldcurpm); /* from regexec.c */
3262 SAVESPTR(PL_reg_curpm); /* from regexec.c */
3263#ifdef DEBUGGING
3264 SAVEPPTR(PL_reg_starttry); /* from regexec.c */
3265#endif
a0ed51b3 3266}