This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Integrate post-beta tweaks to ansiperl
[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
56953603
IZ
41#endif
42
f0fcb552 43/*SUPPRESS 112*/
a687059c 44/*
e50aee73 45 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
46 *
47 * Copyright (c) 1986 by University of Toronto.
48 * Written by Henry Spencer. Not derived from licensed software.
49 *
50 * Permission is granted to anyone to use this software for any
51 * purpose on any computer system, and to redistribute it freely,
52 * subject to the following restrictions:
53 *
54 * 1. The author is not responsible for the consequences of use of
55 * this software, no matter how awful, even if they arise
56 * from defects in it.
57 *
58 * 2. The origin of this software must not be misrepresented, either
59 * by explicit claim or by omission.
60 *
61 * 3. Altered versions must be plainly marked as such, and must not
62 * be misrepresented as being the original software.
63 *
64 *
65 **** Alterations to Henry's code are...
66 ****
9607fc9c 67 **** Copyright (c) 1991-1997, Larry Wall
a687059c 68 ****
9ef589d8
LW
69 **** You may distribute under the terms of either the GNU General Public
70 **** License or the Artistic License, as specified in the README file.
71
a687059c
LW
72 *
73 * Beware that some of this code is subtly aware of the way operator
74 * precedence is structured in regular expressions. Serious changes in
75 * regular-expression syntax might require a total rethink.
76 */
77#include "EXTERN.h"
78#include "perl.h"
d06ea78c 79
b9d5759e 80#ifndef PERL_IN_XSUB_RE
d06ea78c
GS
81# include "INTERN.h"
82#endif
c277df42
IZ
83
84#define REG_COMP_C
a687059c
LW
85#include "regcomp.h"
86
d4cce5f1 87#ifdef op
11343788 88#undef op
d4cce5f1 89#endif /* op */
11343788 90
fe14fcc3
LW
91#ifdef MSDOS
92# if defined(BUGGY_MSC6)
93 /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
94 # pragma optimize("a",off)
95 /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
96 # pragma optimize("w",on )
97# endif /* BUGGY_MSC6 */
98#endif /* MSDOS */
99
a687059c
LW
100#ifndef STATIC
101#define STATIC static
102#endif
103
104#define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
105#define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
106 ((*s) == '{' && regcurly(s)))
2b69d0c2
LW
107#ifdef atarist
108#define PERL_META "^$.[()|?+*\\"
109#else
a687059c 110#define META "^$.[()|?+*\\"
2b69d0c2 111#endif
a687059c 112
35c8bce7
LW
113#ifdef SPSTART
114#undef SPSTART /* dratted cpp namespace... */
115#endif
a687059c
LW
116/*
117 * Flags to be passed up and down.
118 */
a687059c 119#define WORST 0 /* Worst case. */
821b33a5 120#define HASWIDTH 0x1 /* Known to match non-null strings. */
a0d0e21e
LW
121#define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
122#define SPSTART 0x4 /* Starts with * or +. */
123#define TRYAGAIN 0x8 /* Weeded out a declaration. */
a687059c
LW
124
125/*
e50aee73 126 * Forward declarations for pregcomp()'s friends.
a687059c 127 */
a0d0e21e 128
76e3520e 129#ifndef PERL_OBJECT
c277df42
IZ
130static regnode *reg _((I32, I32 *));
131static regnode *reganode _((U8, U32));
132static regnode *regatom _((I32 *));
133static regnode *regbranch _((I32 *, I32));
134static void regc _((U8, char *));
135static regnode *regclass _((void));
a0d0e21e 136STATIC I32 regcurly _((char *));
c277df42
IZ
137static regnode *reg_node _((U8));
138static regnode *regpiece _((I32 *));
139static void reginsert _((U8, regnode *));
140static void regoptail _((regnode *, regnode *));
c277df42 141static void regtail _((regnode *, regnode *));
873ef191 142static char* regwhite _((char *, char *));
a0d0e21e 143static char* nextchar _((void));
3bd495df 144static void re_croak2 _((const char* pat1,const char* pat2,...)) __attribute__((noreturn));
76e3520e 145#endif
a687059c 146
c277df42
IZ
147/* Length of a variant. */
148
76e3520e 149#ifndef PERL_OBJECT
c277df42
IZ
150typedef struct {
151 I32 len_min;
152 I32 len_delta;
153 I32 pos_min;
154 I32 pos_delta;
155 SV *last_found;
156 I32 last_end; /* min value, <0 unless valid. */
157 I32 last_start_min;
158 I32 last_start_max;
159 SV **longest; /* Either &l_fixed, or &l_float. */
160 SV *longest_fixed;
161 I32 offset_fixed;
162 SV *longest_float;
163 I32 offset_float_min;
164 I32 offset_float_max;
165 I32 flags;
166} scan_data_t;
76e3520e 167#endif
c277df42
IZ
168
169static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
170
171#define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
172#define SF_BEFORE_SEOL 0x1
173#define SF_BEFORE_MEOL 0x2
174#define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
175#define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
176
09b7f37c
CB
177#ifdef NO_UNARY_PLUS
178# define SF_FIX_SHIFT_EOL (0+2)
179# define SF_FL_SHIFT_EOL (0+4)
180#else
181# define SF_FIX_SHIFT_EOL (+2)
182# define SF_FL_SHIFT_EOL (+4)
183#endif
c277df42
IZ
184
185#define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
186#define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
187
188#define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
189#define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
190#define SF_IS_INF 0x40
191#define SF_HAS_PAR 0x80
192#define SF_IN_PAR 0x100
193#define SF_HAS_EVAL 0x200
4bfe0158 194#define SCF_DO_SUBSTR 0x400
c277df42 195
76e3520e 196STATIC void
c277df42
IZ
197scan_commit(scan_data_t *data)
198{
199 STRLEN l = SvCUR(data->last_found);
200 STRLEN old_l = SvCUR(*data->longest);
201
202 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
203 sv_setsv(*data->longest, data->last_found);
204 if (*data->longest == data->longest_fixed) {
205 data->offset_fixed = l ? data->last_start_min : data->pos_min;
206 if (data->flags & SF_BEFORE_EOL)
207 data->flags
208 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
209 else
210 data->flags &= ~SF_FIX_BEFORE_EOL;
211 } else {
212 data->offset_float_min = l ? data->last_start_min : data->pos_min;
213 data->offset_float_max = (l
214 ? data->last_start_max
215 : data->pos_min + data->pos_delta);
216 if (data->flags & SF_BEFORE_EOL)
217 data->flags
218 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
219 else
220 data->flags &= ~SF_FL_BEFORE_EOL;
221 }
222 }
223 SvCUR_set(data->last_found, 0);
224 data->last_end = -1;
225 data->flags &= ~SF_BEFORE_EOL;
226}
227
c277df42
IZ
228/* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
229 to the position after last scanned or to NULL. */
230
76e3520e 231STATIC I32
c277df42
IZ
232study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
233 /* scanp: Start here (read-write). */
234 /* deltap: Write maxlen-minlen here. */
235 /* last: Stop before this one. */
236{
5c0ca799 237 dTHR;
c277df42
IZ
238 I32 min = 0, pars = 0, code;
239 regnode *scan = *scanp, *next;
240 I32 delta = 0;
241 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
242 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
243 scan_data_t data_fake;
244
245 while (scan && OP(scan) != END && scan < last) {
246 /* Peephole optimizer: */
247
248 if (regkind[(U8)OP(scan)] == EXACT) {
249 regnode *n = regnext(scan);
250 U32 stringok = 1;
251#ifdef DEBUGGING
252 regnode *stop = scan;
253#endif
254
255 next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
256 /* Skip NOTHING, merge EXACT*. */
257 while (n &&
258 ( regkind[(U8)OP(n)] == NOTHING ||
259 (stringok && (OP(n) == OP(scan))))
260 && NEXT_OFF(n)
261 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
262 if (OP(n) == TAIL || n > next)
263 stringok = 0;
264 if (regkind[(U8)OP(n)] == NOTHING) {
265 NEXT_OFF(scan) += NEXT_OFF(n);
266 next = n + NODE_STEP_REGNODE;
267#ifdef DEBUGGING
268 if (stringok)
269 stop = n;
270#endif
271 n = regnext(n);
272 } else {
273 int oldl = *OPERAND(scan);
274 regnode *nnext = regnext(n);
275
276 if (oldl + *OPERAND(n) > U8_MAX)
277 break;
278 NEXT_OFF(scan) += NEXT_OFF(n);
279 *OPERAND(scan) += *OPERAND(n);
280 next = n + (*OPERAND(n) + 2 - 1)/sizeof(regnode) + 2;
281 /* Now we can overwrite *n : */
282 Move(OPERAND(n) + 1, OPERAND(scan) + oldl + 1,
283 *OPERAND(n) + 1, char);
284#ifdef DEBUGGING
285 if (stringok)
286 stop = next - 1;
287#endif
288 n = nnext;
289 }
290 }
291#ifdef DEBUGGING
292 /* Allow dumping */
293 n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
294 while (n <= stop) {
ca04da08
GS
295 /* Purify reports a benign UMR here sometimes, because we
296 * don't initialize the OP() slot of a node when that node
297 * is occupied by just the trailing null of the string in
298 * an EXACT node */
c277df42
IZ
299 if (regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
300 OP(n) = OPTIMIZED;
301 NEXT_OFF(n) = 0;
302 }
303 n++;
304 }
305#endif
306
307 }
308 if (OP(scan) != CURLYX) {
309 int max = (reg_off_by_arg[OP(scan)] ? I32_MAX : U16_MAX);
310 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
311 int noff;
312 regnode *n = scan;
313
314 /* Skip NOTHING and LONGJMP. */
315 while ((n = regnext(n))
316 && ((regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
317 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
318 && off + noff < max)
319 off += noff;
320 if (reg_off_by_arg[OP(scan)])
321 ARG(scan) = off;
322 else
323 NEXT_OFF(scan) = off;
324 }
325 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
326 || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
327 next = regnext(scan);
328 code = OP(scan);
329
330 if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
331 I32 max1 = 0, min1 = I32_MAX, num = 0;
332
333 if (flags & SCF_DO_SUBSTR)
334 scan_commit(data);
335 while (OP(scan) == code) {
336 I32 deltanext, minnext;
337
338 num++;
339 data_fake.flags = 0;
340 next = regnext(scan);
341 scan = NEXTOPER(scan);
342 if (code != BRANCH)
343 scan = NEXTOPER(scan);
344 /* We suppose the run is continuous, last=next...*/
345 minnext = study_chunk(&scan, &deltanext, next,
346 &data_fake, 0);
347 if (min1 > minnext)
348 min1 = minnext;
349 if (max1 < minnext + deltanext)
350 max1 = minnext + deltanext;
351 if (deltanext == I32_MAX)
352 is_inf = 1;
353 scan = next;
354 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
355 pars++;
405ff068 356 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42
IZ
357 data->flags |= SF_HAS_EVAL;
358 if (code == SUSPEND)
359 break;
360 }
361 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
362 min1 = 0;
363 if (flags & SCF_DO_SUBSTR) {
364 data->pos_min += min1;
365 data->pos_delta += max1 - min1;
366 if (max1 != min1 || is_inf)
367 data->longest = &(data->longest_float);
368 }
369 min += min1;
370 delta += max1 - min1;
371 } else if (code == BRANCHJ) /* single branch is optimized. */
372 scan = NEXTOPER(NEXTOPER(scan));
373 else /* single branch is optimized. */
374 scan = NEXTOPER(scan);
375 continue;
376 } else if (OP(scan) == EXACT) {
377 min += *OPERAND(scan);
378 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
379 I32 l = *OPERAND(scan);
380
381 /* The code below prefers earlier match for fixed
382 offset, later match for variable offset. */
383 if (data->last_end == -1) { /* Update the start info. */
384 data->last_start_min = data->pos_min;
385 data->last_start_max = is_inf
386 ? I32_MAX : data->pos_min + data->pos_delta;
387 }
161b471a 388 sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), l);
c277df42
IZ
389 data->last_end = data->pos_min + l;
390 data->pos_min += l; /* As in the first entry. */
391 data->flags &= ~SF_BEFORE_EOL;
392 }
393 } else if (regkind[(U8)OP(scan)] == EXACT) {
394 if (flags & SCF_DO_SUBSTR)
395 scan_commit(data);
396 min += *OPERAND(scan);
397 if (data && (flags & SCF_DO_SUBSTR))
398 data->pos_min += *OPERAND(scan);
399 } else if (strchr(varies,OP(scan))) {
400 I32 mincount, maxcount, minnext, deltanext, pos_before, fl;
401 regnode *oscan = scan;
402
403 switch (regkind[(U8)OP(scan)]) {
404 case WHILEM:
405 scan = NEXTOPER(scan);
406 goto finish;
407 case PLUS:
408 if (flags & SCF_DO_SUBSTR) {
409 next = NEXTOPER(scan);
410 if (OP(next) == EXACT) {
411 mincount = 1;
412 maxcount = REG_INFTY;
413 next = regnext(scan);
414 scan = NEXTOPER(scan);
415 goto do_curly;
416 }
417 }
418 if (flags & SCF_DO_SUBSTR)
419 data->pos_min++;
420 min++;
421 /* Fall through. */
422 case STAR:
423 is_inf = 1;
424 scan = regnext(scan);
425 if (flags & SCF_DO_SUBSTR) {
426 scan_commit(data);
427 data->longest = &(data->longest_float);
428 }
429 goto optimize_curly_tail;
430 case CURLY:
431 mincount = ARG1(scan);
432 maxcount = ARG2(scan);
433 next = regnext(scan);
434 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
435 do_curly:
436 if (flags & SCF_DO_SUBSTR) {
437 if (mincount == 0) scan_commit(data);
438 pos_before = data->pos_min;
439 }
440 if (data) {
441 fl = data->flags;
442 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
443 if (is_inf)
444 data->flags |= SF_IS_INF;
445 }
446 /* This will finish on WHILEM, setting scan, or on NULL: */
447 minnext = study_chunk(&scan, &deltanext, last, data,
448 mincount == 0
449 ? (flags & ~SCF_DO_SUBSTR) : flags);
450 if (!scan) /* It was not CURLYX, but CURLY. */
451 scan = next;
452 if (dowarn && (minnext + deltanext == 0)
821b33a5
IZ
453 && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
454 && maxcount <= 10000) /* Complement check for big count */
c277df42
IZ
455 warn("Strange *+?{} on zero-length expression");
456 min += minnext * mincount;
457 is_inf |= (maxcount == REG_INFTY && (minnext + deltanext) > 0
458 || deltanext == I32_MAX);
459 delta += (minnext + deltanext) * maxcount - minnext * mincount;
460
461 /* Try powerful optimization CURLYX => CURLYN. */
c277df42
IZ
462 if ( OP(oscan) == CURLYX && data
463 && data->flags & SF_IN_PAR
464 && !(data->flags & SF_HAS_EVAL)
465 && !deltanext && minnext == 1 ) {
466 /* Try to optimize to CURLYN. */
467 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
468 regnode *nxt1 = nxt, *nxt2;
469
470 /* Skip open. */
471 nxt = regnext(nxt);
472 if (!strchr(simple,OP(nxt))
473 && !(regkind[(U8)OP(nxt)] == EXACT
474 && *OPERAND(nxt) == 1))
475 goto nogo;
476 nxt2 = nxt;
477 nxt = regnext(nxt);
478 if (OP(nxt) != CLOSE)
479 goto nogo;
480 /* Now we know that nxt2 is the only contents: */
481 oscan->flags = ARG(nxt);
482 OP(oscan) = CURLYN;
483 OP(nxt1) = NOTHING; /* was OPEN. */
484#ifdef DEBUGGING
485 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
486 NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
487 NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
488 OP(nxt) = OPTIMIZED; /* was CLOSE. */
489 OP(nxt + 1) = OPTIMIZED; /* was count. */
490 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
491#endif
492 }
c277df42
IZ
493 nogo:
494
495 /* Try optimization CURLYX => CURLYM. */
496 if ( OP(oscan) == CURLYX && data
c277df42 497 && !(data->flags & SF_HAS_PAR)
c277df42
IZ
498 && !(data->flags & SF_HAS_EVAL)
499 && !deltanext ) {
500 /* XXXX How to optimize if data == 0? */
501 /* Optimize to a simpler form. */
502 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
503 regnode *nxt2;
504
505 OP(oscan) = CURLYM;
506 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
507 && (OP(nxt2) != WHILEM))
508 nxt = nxt2;
509 OP(nxt2) = SUCCEED; /* Whas WHILEM */
c277df42
IZ
510 /* Need to optimize away parenths. */
511 if (data->flags & SF_IN_PAR) {
512 /* Set the parenth number. */
513 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
514
515 if (OP(nxt) != CLOSE)
516 FAIL("panic opt close");
517 oscan->flags = ARG(nxt);
518 OP(nxt1) = OPTIMIZED; /* was OPEN. */
519 OP(nxt) = OPTIMIZED; /* was CLOSE. */
520#ifdef DEBUGGING
521 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
522 OP(nxt + 1) = OPTIMIZED; /* was count. */
523 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
524 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
525#endif
526#if 0
527 while ( nxt1 && (OP(nxt1) != WHILEM)) {
528 regnode *nnxt = regnext(nxt1);
529
530 if (nnxt == nxt) {
531 if (reg_off_by_arg[OP(nxt1)])
532 ARG_SET(nxt1, nxt2 - nxt1);
533 else if (nxt2 - nxt1 < U16_MAX)
534 NEXT_OFF(nxt1) = nxt2 - nxt1;
535 else
536 OP(nxt) = NOTHING; /* Cannot beautify */
537 }
538 nxt1 = nnxt;
539 }
540#endif
541 /* Optimize again: */
542 study_chunk(&nxt1, &deltanext, nxt, NULL, 0);
543 } else
544 oscan->flags = 0;
c277df42
IZ
545 }
546 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
547 pars++;
548 if (flags & SCF_DO_SUBSTR) {
549 SV *last_str = Nullsv;
550 int counted = mincount != 0;
551
552 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
553 I32 b = pos_before >= data->last_start_min
554 ? pos_before : data->last_start_min;
555 STRLEN l;
556 char *s = SvPV(data->last_found, l);
557
558 l -= b - data->last_start_min;
559 /* Get the added string: */
560 last_str = newSVpv(s + b - data->last_start_min, l);
561 if (deltanext == 0 && pos_before == b) {
562 /* What was added is a constant string */
563 if (mincount > 1) {
564 SvGROW(last_str, (mincount * l) + 1);
565 repeatcpy(SvPVX(last_str) + l,
566 SvPVX(last_str), l, mincount - 1);
567 SvCUR(last_str) *= mincount;
568 /* Add additional parts. */
569 SvCUR_set(data->last_found,
570 SvCUR(data->last_found) - l);
571 sv_catsv(data->last_found, last_str);
572 data->last_end += l * (mincount - 1);
573 }
574 }
575 }
576 /* It is counted once already... */
577 data->pos_min += minnext * (mincount - counted);
578 data->pos_delta += - counted * deltanext +
579 (minnext + deltanext) * maxcount - minnext * mincount;
580 if (mincount != maxcount) {
581 scan_commit(data);
582 if (mincount && last_str) {
583 sv_setsv(data->last_found, last_str);
584 data->last_end = data->pos_min;
585 data->last_start_min =
586 data->pos_min - SvCUR(last_str);
587 data->last_start_max = is_inf
588 ? I32_MAX
589 : data->pos_min + data->pos_delta
590 - SvCUR(last_str);
591 }
592 data->longest = &(data->longest_float);
593 }
594 }
405ff068 595 if (data && (fl & SF_HAS_EVAL))
c277df42
IZ
596 data->flags |= SF_HAS_EVAL;
597 optimize_curly_tail:
c277df42
IZ
598 if (OP(oscan) != CURLYX) {
599 while (regkind[(U8)OP(next = regnext(oscan))] == NOTHING
600 && NEXT_OFF(next))
601 NEXT_OFF(oscan) += NEXT_OFF(next);
602 }
c277df42
IZ
603 continue;
604 default: /* REF only? */
605 if (flags & SCF_DO_SUBSTR) {
606 scan_commit(data);
607 data->longest = &(data->longest_float);
608 }
609 is_inf = 1;
610 break;
611 }
612 } else if (strchr(simple,OP(scan))) {
613 if (flags & SCF_DO_SUBSTR) {
614 scan_commit(data);
615 data->pos_min++;
616 }
617 min++;
618 } else if (regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
619 data->flags |= (OP(scan) == MEOL
620 ? SF_BEFORE_MEOL
621 : SF_BEFORE_SEOL);
622 } else if (regkind[(U8)OP(scan)] == BRANCHJ
623 && (scan->flags || data)
624 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
625 I32 deltanext, minnext;
626 regnode *nscan;
627
628 data_fake.flags = 0;
629 next = regnext(scan);
630 nscan = NEXTOPER(NEXTOPER(scan));
631 minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0);
632 if (scan->flags) {
633 if (deltanext) {
634 FAIL("variable length lookbehind not implemented");
635 } else if (minnext > U8_MAX) {
636 FAIL2("lookbehind longer than %d not implemented", U8_MAX);
637 }
638 scan->flags = minnext;
639 }
640 if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
641 pars++;
405ff068 642 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42
IZ
643 data->flags |= SF_HAS_EVAL;
644 } else if (OP(scan) == OPEN) {
645 pars++;
646 } else if (OP(scan) == CLOSE && ARG(scan) == is_par) {
c277df42
IZ
647 next = regnext(scan);
648
649 if ( next && (OP(next) != WHILEM) && next < last)
c277df42
IZ
650 is_par = 0; /* Disable optimization */
651 } else if (OP(scan) == EVAL) {
652 if (data)
653 data->flags |= SF_HAS_EVAL;
654 }
655 /* Else: zero-length, ignore. */
656 scan = regnext(scan);
657 }
658
659 finish:
660 *scanp = scan;
661 *deltap = is_inf ? I32_MAX : delta;
662 if (flags & SCF_DO_SUBSTR && is_inf)
663 data->pos_delta = I32_MAX - data->pos_min;
664 if (is_par > U8_MAX)
665 is_par = 0;
666 if (is_par && pars==1 && data) {
667 data->flags |= SF_IN_PAR;
668 data->flags &= ~SF_HAS_PAR;
669 } else if (pars && data) {
670 data->flags |= SF_HAS_PAR;
671 data->flags &= ~SF_IN_PAR;
672 }
673 return min;
674}
675
76e3520e 676STATIC I32
c277df42
IZ
677add_data(I32 n, char *s)
678{
5c0ca799 679 dTHR;
77d41b28
GS
680 if (regcomp_rx->data) {
681 Renewc(regcomp_rx->data,
682 sizeof(*regcomp_rx->data) + sizeof(void*) * (regcomp_rx->data->count + n - 1),
c277df42 683 char, struct reg_data);
77d41b28
GS
684 Renew(regcomp_rx->data->what, regcomp_rx->data->count + n, U8);
685 regcomp_rx->data->count += n;
c277df42 686 } else {
77d41b28 687 Newc(1207, regcomp_rx->data, sizeof(*regcomp_rx->data) + sizeof(void*) * (n - 1),
c277df42 688 char, struct reg_data);
77d41b28
GS
689 New(1208, regcomp_rx->data->what, n, U8);
690 regcomp_rx->data->count = n;
c277df42 691 }
77d41b28
GS
692 Copy(s, regcomp_rx->data->what + regcomp_rx->data->count - n, n, U8);
693 return regcomp_rx->data->count - n;
c277df42
IZ
694}
695
a687059c 696/*
e50aee73 697 - pregcomp - compile a regular expression into internal code
a687059c
LW
698 *
699 * We can't allocate space until we know how big the compiled form will be,
700 * but we can't compile it (and thus know how big it is) until we've got a
701 * place to put the code. So we cheat: we compile it twice, once with code
702 * generation turned off and size counting turned on, and once "for real".
703 * This also means that we don't allocate space until we are sure that the
704 * thing really will compile successfully, and we never have to move the
705 * code and thus invalidate pointers into it. (Note that it has to be in
706 * one piece because free() must be able to free it all.) [NB: not true in perl]
707 *
708 * Beware that the optimization-preparation code in here knows about some
709 * of the structure of the compiled regexp. [I'll say.]
710 */
711regexp *
8ac85365 712pregcomp(char *exp, char *xend, PMOP *pm)
a687059c 713{
5c0ca799 714 dTHR;
a0d0e21e 715 register regexp *r;
c277df42
IZ
716 regnode *scan;
717 SV **longest;
718 SV *longest_fixed;
719 SV *longest_float;
720 regnode *first;
a0d0e21e 721 I32 flags;
a0d0e21e
LW
722 I32 minlen = 0;
723 I32 sawplus = 0;
724 I32 sawopen = 0;
725
726 if (exp == NULL)
c277df42 727 FAIL("NULL regexp argument");
a0d0e21e 728
bbce6d69 729 regprecomp = savepvn(exp, xend - exp);
c277df42
IZ
730 DEBUG_r(PerlIO_printf(Perl_debug_log, "compiling RE `%*s'\n",
731 xend - exp, regprecomp));
a0d0e21e 732 regflags = pm->op_pmflags;
bbce6d69
PP
733 regsawback = 0;
734
c277df42
IZ
735 regseen = 0;
736 seen_zerolen = *exp == '^' ? -1 : 0;
2cd61cdb 737 seen_evals = 0;
c277df42
IZ
738 extralen = 0;
739
bbce6d69 740 /* First pass: determine size, legality. */
77d41b28 741 regcomp_parse = exp;
a0d0e21e 742 regxend = xend;
a0d0e21e 743 regnaughty = 0;
a0d0e21e
LW
744 regnpar = 1;
745 regsize = 0L;
746 regcode = &regdummy;
c277df42 747 regc((U8)MAGIC, (char*)regcode);
a0d0e21e
LW
748 if (reg(0, &flags) == NULL) {
749 Safefree(regprecomp);
750 regprecomp = Nullch;
751 return(NULL);
752 }
c277df42
IZ
753 DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", regsize));
754
755 DEBUG_r(
756 if (!colorset) {
757 int i = 0;
76e3520e 758 char *s = PerlEnv_getenv("TERMCAP_COLORS");
c277df42
IZ
759
760 colorset = 1;
761 if (s) {
762 colors[0] = s = savepv(s);
763 while (++i < 4) {
764 s = strchr(s, '\t');
765 if (!s)
766 FAIL("Not enough TABs in TERMCAP_COLORS");
767 *s = '\0';
768 colors[i] = ++s;
769 }
770 } else {
771 while (i < 4)
772 colors[i++] = "";
773 }
774 /* Reset colors: */
775 PerlIO_printf(Perl_debug_log, "%s%s%s%s",
776 colors[0],colors[1],colors[2],colors[3]);
777 }
778 );
a0d0e21e 779
c277df42
IZ
780 /* Small enough for pointer-storage convention?
781 If extralen==0, this means that we will not need long jumps. */
c277df42
IZ
782 if (regsize >= 0x10000L && extralen)
783 regsize += extralen;
784 else
785 extralen = 0;
a0d0e21e 786
bbce6d69 787 /* Allocate space and initialize. */
c277df42
IZ
788 Newc(1001, r, sizeof(regexp) + (unsigned)regsize * sizeof(regnode),
789 char, regexp);
a0d0e21e
LW
790 if (r == NULL)
791 FAIL("regexp out of space");
c277df42 792 r->refcnt = 1;
bbce6d69 793 r->prelen = xend - exp;
a0d0e21e
LW
794 r->precomp = regprecomp;
795 r->subbeg = r->subbase = NULL;
ef64f398 796 r->nparens = regnpar - 1; /* set early to validate backrefs */
77d41b28 797 regcomp_rx = r;
bbce6d69
PP
798
799 /* Second pass: emit code. */
77d41b28 800 regcomp_parse = exp;
bbce6d69
PP
801 regxend = xend;
802 regnaughty = 0;
a0d0e21e
LW
803 regnpar = 1;
804 regcode = r->program;
2cd61cdb
IZ
805 /* Store the count of eval-groups for security checks: */
806 regcode->next_off = ((seen_evals > U16_MAX) ? U16_MAX : seen_evals);
c277df42
IZ
807 regc((U8)MAGIC, (char*) regcode++);
808 r->data = 0;
a0d0e21e
LW
809 if (reg(0, &flags) == NULL)
810 return(NULL);
811
812 /* Dig out information for optimizations. */
8782bef2 813 r->reganch = pm->op_pmflags & PMf_COMPILETIME;
a0d0e21e 814 pm->op_pmflags = regflags;
c277df42 815 r->regstclass = NULL;
a0d0e21e 816 r->naughty = regnaughty >= 10; /* Probably an expensive pattern. */
c277df42 817 scan = r->program + 1; /* First BRANCH. */
2779dcf1
IZ
818
819 /* XXXX To minimize changes to RE engine we always allocate
820 3-units-long substrs field. */
821 Newz(1004, r->substrs, 1, struct reg_substr_data);
822
c277df42
IZ
823 if (OP(scan) != BRANCH) { /* Only one top-level choice. */
824 scan_data_t data;
825 I32 fake;
c5254dd6 826 STRLEN longest_float_length, longest_fixed_length;
a0d0e21e 827
c277df42 828 StructCopy(&zero_scan_data, &data, scan_data_t);
a0d0e21e 829 first = scan;
c277df42 830 /* Skip introductions and multiplicators >= 1. */
a0d0e21e
LW
831 while ((OP(first) == OPEN && (sawopen = 1)) ||
832 (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
833 (OP(first) == PLUS) ||
834 (OP(first) == MINMOD) ||
835 (regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
836 if (OP(first) == PLUS)
837 sawplus = 1;
838 else
839 first += regarglen[(U8)OP(first)];
840 first = NEXTOPER(first);
a687059c
LW
841 }
842
a0d0e21e
LW
843 /* Starting-point info. */
844 again:
c277df42 845 if (OP(first) == EXACT); /* Empty, get anchored substr later. */
a0d0e21e
LW
846 else if (strchr(simple+2,OP(first)))
847 r->regstclass = first;
bbce6d69
PP
848 else if (regkind[(U8)OP(first)] == BOUND ||
849 regkind[(U8)OP(first)] == NBOUND)
a0d0e21e
LW
850 r->regstclass = first;
851 else if (regkind[(U8)OP(first)] == BOL) {
c277df42 852 r->reganch |= (OP(first) == MBOL ? ROPT_ANCH_MBOL: ROPT_ANCH_BOL);
a0d0e21e 853 first = NEXTOPER(first);
774d564b
PP
854 goto again;
855 }
856 else if (OP(first) == GPOS) {
857 r->reganch |= ROPT_ANCH_GPOS;
858 first = NEXTOPER(first);
859 goto again;
a0d0e21e
LW
860 }
861 else if ((OP(first) == STAR &&
862 regkind[(U8)OP(NEXTOPER(first))] == ANY) &&
863 !(r->reganch & ROPT_ANCH) )
864 {
865 /* turn .* into ^.* with an implied $*=1 */
774d564b 866 r->reganch |= ROPT_ANCH_BOL | ROPT_IMPLICIT;
a0d0e21e 867 first = NEXTOPER(first);
774d564b 868 goto again;
a0d0e21e
LW
869 }
870 if (sawplus && (!sawopen || !regsawback))
871 r->reganch |= ROPT_SKIP; /* x+ must match 1st of run */
872
c277df42
IZ
873 /* Scan is after the zeroth branch, first is atomic matcher. */
874 DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n",
875 first - scan + 1));
a0d0e21e
LW
876 /*
877 * If there's something expensive in the r.e., find the
878 * longest literal string that must appear and make it the
879 * regmust. Resolve ties in favor of later strings, since
880 * the regstart check works with the beginning of the r.e.
881 * and avoiding duplication strengthens checking. Not a
882 * strong reason, but sufficient in the absence of others.
883 * [Now we resolve ties in favor of the earlier string if
c277df42 884 * it happens that c_offset_min has been invalidated, since the
a0d0e21e
LW
885 * earlier string may buy us something the later one won't.]
886 */
a0d0e21e 887 minlen = 0;
a687059c 888
c277df42
IZ
889 data.longest_fixed = newSVpv("",0);
890 data.longest_float = newSVpv("",0);
891 data.last_found = newSVpv("",0);
892 data.longest = &(data.longest_fixed);
893 first = scan;
894
895 minlen = study_chunk(&first, &fake, scan + regsize, /* Up to end */
896 &data, SCF_DO_SUBSTR);
897 if ( regnpar == 1 && data.longest == &(data.longest_fixed)
898 && data.last_start_min == 0 && data.last_end > 0
899 && !seen_zerolen
900 && (!(regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
901 r->reganch |= ROPT_CHECK_ALL;
902 scan_commit(&data);
903 SvREFCNT_dec(data.last_found);
904
c5254dd6
MB
905 longest_float_length = SvCUR(data.longest_float);
906 if (longest_float_length
c277df42
IZ
907 || (data.flags & SF_FL_BEFORE_EOL
908 && (!(data.flags & SF_FL_BEFORE_MEOL)
909 || (regflags & PMf_MULTILINE)))) {
910 if (SvCUR(data.longest_fixed)
911 && data.offset_fixed == data.offset_float_min)
912 goto remove; /* Like in (a)+. */
913
914 r->float_substr = data.longest_float;
915 r->float_min_offset = data.offset_float_min;
916 r->float_max_offset = data.offset_float_max;
2779dcf1 917 fbm_compile(r->float_substr, 0);
c277df42
IZ
918 BmUSEFUL(r->float_substr) = 100;
919 if (data.flags & SF_FL_BEFORE_EOL /* Cannot have SEOL and MULTI */
920 && (!(data.flags & SF_FL_BEFORE_MEOL)
921 || (regflags & PMf_MULTILINE)))
922 SvTAIL_on(r->float_substr);
923 } else {
924 remove:
925 r->float_substr = Nullsv;
926 SvREFCNT_dec(data.longest_float);
c5254dd6 927 longest_float_length = 0;
a0d0e21e 928 }
c277df42 929
c5254dd6
MB
930 longest_fixed_length = SvCUR(data.longest_fixed);
931 if (longest_fixed_length
c277df42
IZ
932 || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
933 && (!(data.flags & SF_FIX_BEFORE_MEOL)
934 || (regflags & PMf_MULTILINE)))) {
935 r->anchored_substr = data.longest_fixed;
936 r->anchored_offset = data.offset_fixed;
2779dcf1 937 fbm_compile(r->anchored_substr, 0);
c277df42
IZ
938 BmUSEFUL(r->anchored_substr) = 100;
939 if (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
940 && (!(data.flags & SF_FIX_BEFORE_MEOL)
941 || (regflags & PMf_MULTILINE)))
942 SvTAIL_on(r->anchored_substr);
943 } else {
944 r->anchored_substr = Nullsv;
945 SvREFCNT_dec(data.longest_fixed);
c5254dd6 946 longest_fixed_length = 0;
a0d0e21e 947 }
c277df42
IZ
948
949 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
c5254dd6 950 if (longest_fixed_length > longest_float_length) {
c277df42
IZ
951 r->check_substr = r->anchored_substr;
952 r->check_offset_min = r->check_offset_max = r->anchored_offset;
953 if (r->reganch & ROPT_ANCH_SINGLE)
954 r->reganch |= ROPT_NOSCAN;
955 } else {
956 r->check_substr = r->float_substr;
957 r->check_offset_min = data.offset_float_min;
958 r->check_offset_max = data.offset_float_max;
a0d0e21e 959 }
c277df42
IZ
960 } else {
961 /* Several toplevels. Best we can is to set minlen. */
962 I32 fake;
963
964 DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
965 scan = r->program + 1;
966 minlen = study_chunk(&scan, &fake, scan + regsize, NULL, 0);
967 r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
a0d0e21e
LW
968 }
969
a0d0e21e 970 r->minlen = minlen;
c277df42
IZ
971 if (regseen & REG_SEEN_GPOS)
972 r->reganch |= ROPT_GPOS_SEEN;
973 if (regseen & REG_SEEN_LOOKBEHIND)
974 r->reganch |= ROPT_LOOKBEHIND_SEEN;
ce862d02
IZ
975 if (regseen & REG_SEEN_EVAL)
976 r->reganch |= ROPT_EVAL_SEEN;
a0d0e21e
LW
977 Newz(1002, r->startp, regnpar, char*);
978 Newz(1002, r->endp, regnpar, char*);
979 DEBUG_r(regdump(r));
980 return(r);
a687059c
LW
981}
982
983/*
984 - reg - regular expression, i.e. main body or parenthesized thing
985 *
986 * Caller must absorb opening parenthesis.
987 *
988 * Combining parenthesis handling with the base level of regular expression
989 * is a trifle forced, but the need to tie the tails of the branches to what
990 * follows makes it hard to avoid.
991 */
76e3520e 992STATIC regnode *
8ac85365 993reg(I32 paren, I32 *flagp)
c277df42 994 /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
a687059c 995{
5c0ca799 996 dTHR;
c277df42
IZ
997 register regnode *ret; /* Will be the head of the group. */
998 register regnode *br;
999 register regnode *lastbr;
1000 register regnode *ender = 0;
a0d0e21e 1001 register I32 parno = 0;
c277df42
IZ
1002 I32 flags, oregflags = regflags, have_branch = 0, open = 0;
1003 char c;
a0d0e21e 1004
821b33a5 1005 *flagp = 0; /* Tentatively. */
a0d0e21e
LW
1006
1007 /* Make an OPEN node, if parenthesized. */
1008 if (paren) {
77d41b28 1009 if (*regcomp_parse == '?') {
ca9dfc88
IZ
1010 U16 posflags = 0, negflags = 0;
1011 U16 *flagsp = &posflags;
1012
77d41b28
GS
1013 regcomp_parse++;
1014 paren = *regcomp_parse++;
c277df42 1015 ret = NULL; /* For look-ahead/behind. */
a0d0e21e 1016 switch (paren) {
c277df42 1017 case '<':
c277df42 1018 regseen |= REG_SEEN_LOOKBEHIND;
77d41b28 1019 if (*regcomp_parse == '!')
c277df42 1020 paren = ',';
77d41b28 1021 if (*regcomp_parse != '=' && *regcomp_parse != '!')
c277df42 1022 goto unknown;
77d41b28 1023 regcomp_parse++;
a0d0e21e
LW
1024 case '=':
1025 case '!':
c277df42
IZ
1026 seen_zerolen++;
1027 case ':':
1028 case '>':
a0d0e21e
LW
1029 break;
1030 case '$':
1031 case '@':
c277df42 1032 FAIL2("Sequence (?%c...) not implemented", (int)paren);
a0d0e21e
LW
1033 break;
1034 case '#':
77d41b28
GS
1035 while (*regcomp_parse && *regcomp_parse != ')')
1036 regcomp_parse++;
1037 if (*regcomp_parse != ')')
c277df42 1038 FAIL("Sequence (?#... not terminated");
a0d0e21e
LW
1039 nextchar();
1040 *flagp = TRYAGAIN;
1041 return NULL;
c277df42
IZ
1042 case '{':
1043 {
1044 dTHR;
1045 I32 count = 1, n = 0;
1046 char c;
77d41b28 1047 char *s = regcomp_parse;
c277df42
IZ
1048 SV *sv;
1049 OP_4tree *sop, *rop;
1050
1051 seen_zerolen++;
ce862d02 1052 regseen |= REG_SEEN_EVAL;
77d41b28
GS
1053 while (count && (c = *regcomp_parse)) {
1054 if (c == '\\' && regcomp_parse[1])
1055 regcomp_parse++;
c277df42
IZ
1056 else if (c == '{')
1057 count++;
1058 else if (c == '}')
1059 count--;
77d41b28 1060 regcomp_parse++;
c277df42 1061 }
77d41b28 1062 if (*regcomp_parse != ')')
c277df42
IZ
1063 FAIL("Sequence (?{...}) not terminated or not {}-balanced");
1064 if (!SIZE_ONLY) {
1065 AV *av;
1066
77d41b28
GS
1067 if (regcomp_parse - 1 - s)
1068 sv = newSVpv(s, regcomp_parse - 1 - s);
c277df42
IZ
1069 else
1070 sv = newSVpv("", 0);
1071
1072 rop = sv_compile_2op(sv, &sop, "re", &av);
1073
1074 n = add_data(3, "nso");
77d41b28
GS
1075 regcomp_rx->data->data[n] = (void*)rop;
1076 regcomp_rx->data->data[n+1] = (void*)av;
1077 regcomp_rx->data->data[n+2] = (void*)sop;
c277df42 1078 SvREFCNT_dec(sv);
cc6b7395 1079 } else { /* First pass */
2cd61cdb
IZ
1080 if (reginterp_cnt < ++seen_evals && curcop != &compiling)
1081 /* No compiled RE interpolated, has runtime
1082 components ===> unsafe. */
1083 FAIL("Eval-group not allowed at runtime, use re 'eval'");
cc6b7395
IZ
1084 if (tainted)
1085 FAIL("Eval-group in insecure regular expression");
c277df42
IZ
1086 }
1087
1088 nextchar();
c277df42
IZ
1089 return reganode(EVAL, n);
1090 }
1091 case '(':
1092 {
77d41b28
GS
1093 if (regcomp_parse[0] == '?') {
1094 if (regcomp_parse[1] == '=' || regcomp_parse[1] == '!'
1095 || regcomp_parse[1] == '<'
1096 || regcomp_parse[1] == '{') { /* Lookahead or eval. */
c277df42
IZ
1097 I32 flag;
1098
1099 ret = reg_node(LOGICAL);
1100 regtail(ret, reg(1, &flag));
1101 goto insert_if;
1102 }
77d41b28
GS
1103 } else if (regcomp_parse[0] >= '1' && regcomp_parse[0] <= '9' ) {
1104 parno = atoi(regcomp_parse++);
c277df42 1105
77d41b28
GS
1106 while (isDIGIT(*regcomp_parse))
1107 regcomp_parse++;
c277df42
IZ
1108 ret = reganode(GROUPP, parno);
1109 if ((c = *nextchar()) != ')')
1110 FAIL2("Switch (?(number%c not recognized", c);
1111 insert_if:
1112 regtail(ret, reganode(IFTHEN, 0));
1113 br = regbranch(&flags, 1);
1114 if (br == NULL)
1115 br = reganode(LONGJMP, 0);
1116 else
1117 regtail(br, reganode(LONGJMP, 0));
1118 c = *nextchar();
1119 if (c == '|') {
1120 lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */
1121 regbranch(&flags, 1);
1122 regtail(ret, lastbr);
1123 c = *nextchar();
1124 } else
1125 lastbr = NULL;
1126 if (c != ')')
1127 FAIL("Switch (?(condition)... contains too many branches");
1128 ender = reg_node(TAIL);
1129 regtail(br, ender);
1130 if (lastbr) {
1131 regtail(lastbr, ender);
1132 regtail(NEXTOPER(NEXTOPER(lastbr)), ender);
1133 } else
1134 regtail(ret, ender);
1135 return ret;
1136 } else {
77d41b28 1137 FAIL2("Unknown condition for (?(%.2s", regcomp_parse);
c277df42
IZ
1138 }
1139 }
1b1626e4 1140 case 0:
c277df42 1141 FAIL("Sequence (? incomplete");
1b1626e4 1142 break;
a0d0e21e 1143 default:
77d41b28 1144 --regcomp_parse;
ca9dfc88 1145 parse_flags:
77d41b28
GS
1146 while (*regcomp_parse && strchr("iogcmsx", *regcomp_parse)) {
1147 if (*regcomp_parse != 'o')
ca9dfc88
IZ
1148 pmflag(flagsp, *regcomp_parse);
1149 ++regcomp_parse;
1150 }
1151 if (*regcomp_parse == '-') {
1152 flagsp = &negflags;
77d41b28 1153 ++regcomp_parse;
ca9dfc88 1154 goto parse_flags;
48c036b1 1155 }
ca9dfc88
IZ
1156 regflags |= posflags;
1157 regflags &= ~negflags;
1158 if (*regcomp_parse == ':') {
1159 regcomp_parse++;
1160 paren = ':';
1161 break;
1162 }
c277df42 1163 unknown:
77d41b28
GS
1164 if (*regcomp_parse != ')')
1165 FAIL2("Sequence (?%c...) not recognized", *regcomp_parse);
a0d0e21e
LW
1166 nextchar();
1167 *flagp = TRYAGAIN;
1168 return NULL;
1169 }
1170 }
1171 else {
1172 parno = regnpar;
1173 regnpar++;
1174 ret = reganode(OPEN, parno);
c277df42 1175 open = 1;
a0d0e21e
LW
1176 }
1177 } else
1178 ret = NULL;
1179
1180 /* Pick up the branches, linking them together. */
c277df42 1181 br = regbranch(&flags, 1);
a0d0e21e
LW
1182 if (br == NULL)
1183 return(NULL);
77d41b28 1184 if (*regcomp_parse == '|') {
c277df42
IZ
1185 if (!SIZE_ONLY && extralen) {
1186 reginsert(BRANCHJ, br);
1187 } else
1188 reginsert(BRANCH, br);
1189 have_branch = 1;
1190 if (SIZE_ONLY)
1191 extralen += 1; /* For BRANCHJ-BRANCH. */
1192 } else if (paren == ':') {
1193 *flagp |= flags&SIMPLE;
1194 }
1195 if (open) { /* Starts with OPEN. */
1196 regtail(ret, br); /* OPEN -> first. */
1197 } else if (paren != '?') /* Not Conditional */
a0d0e21e 1198 ret = br;
821b33a5
IZ
1199 if (flags&HASWIDTH)
1200 *flagp |= HASWIDTH;
a0d0e21e 1201 *flagp |= flags&SPSTART;
c277df42 1202 lastbr = br;
77d41b28 1203 while (*regcomp_parse == '|') {
c277df42
IZ
1204 if (!SIZE_ONLY && extralen) {
1205 ender = reganode(LONGJMP,0);
1206 regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
1207 }
1208 if (SIZE_ONLY)
1209 extralen += 2; /* Account for LONGJMP. */
a0d0e21e 1210 nextchar();
c277df42 1211 br = regbranch(&flags, 0);
a687059c 1212 if (br == NULL)
a0d0e21e 1213 return(NULL);
c277df42
IZ
1214 regtail(lastbr, br); /* BRANCH -> BRANCH. */
1215 lastbr = br;
821b33a5
IZ
1216 if (flags&HASWIDTH)
1217 *flagp |= HASWIDTH;
a687059c 1218 *flagp |= flags&SPSTART;
a0d0e21e
LW
1219 }
1220
c277df42
IZ
1221 if (have_branch || paren != ':') {
1222 /* Make a closing node, and hook it on the end. */
1223 switch (paren) {
1224 case ':':
1225 ender = reg_node(TAIL);
1226 break;
1227 case 1:
1228 ender = reganode(CLOSE, parno);
1229 break;
1230 case '<':
c277df42
IZ
1231 case ',':
1232 case '=':
1233 case '!':
c277df42 1234 *flagp &= ~HASWIDTH;
821b33a5
IZ
1235 /* FALL THROUGH */
1236 case '>':
1237 ender = reg_node(SUCCEED);
c277df42
IZ
1238 break;
1239 case 0:
1240 ender = reg_node(END);
1241 break;
1242 }
1243 regtail(lastbr, ender);
a0d0e21e 1244
c277df42
IZ
1245 if (have_branch) {
1246 /* Hook the tails of the branches to the closing node. */
1247 for (br = ret; br != NULL; br = regnext(br)) {
1248 regoptail(br, ender);
1249 }
1250 }
a0d0e21e 1251 }
c277df42
IZ
1252
1253 {
1254 char *p;
1255 static char parens[] = "=!<,>";
1256
1257 if (paren && (p = strchr(parens, paren))) {
1258 int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
1259 int flag = (p - parens) > 1;
1260
1261 if (paren == '>')
1262 node = SUSPEND, flag = 0;
1263 reginsert(node,ret);
c277df42 1264 ret->flags = flag;
c277df42
IZ
1265 regtail(ret, reg_node(TAIL));
1266 }
a0d0e21e
LW
1267 }
1268
1269 /* Check for proper termination. */
77d41b28 1270 if (paren && (regcomp_parse >= regxend || *nextchar() != ')')) {
a0d0e21e 1271 FAIL("unmatched () in regexp");
77d41b28
GS
1272 } else if (!paren && regcomp_parse < regxend) {
1273 if (*regcomp_parse == ')') {
a0d0e21e
LW
1274 FAIL("unmatched () in regexp");
1275 } else
1276 FAIL("junk on end of regexp"); /* "Can't happen". */
1277 /* NOTREACHED */
1278 }
c277df42
IZ
1279 if (paren != 0) {
1280 regflags = oregflags;
1281 }
a687059c 1282
a0d0e21e 1283 return(ret);
a687059c
LW
1284}
1285
1286/*
1287 - regbranch - one alternative of an | operator
1288 *
1289 * Implements the concatenation operator.
1290 */
76e3520e 1291STATIC regnode *
c277df42 1292regbranch(I32 *flagp, I32 first)
a687059c 1293{
5c0ca799 1294 dTHR;
c277df42
IZ
1295 register regnode *ret;
1296 register regnode *chain = NULL;
1297 register regnode *latest;
1298 I32 flags = 0, c = 0;
a0d0e21e 1299
c277df42
IZ
1300 if (first)
1301 ret = NULL;
1302 else {
1303 if (!SIZE_ONLY && extralen)
1304 ret = reganode(BRANCHJ,0);
1305 else
1306 ret = reg_node(BRANCH);
1307 }
1308
1309 if (!first && SIZE_ONLY)
1310 extralen += 1; /* BRANCHJ */
1311
1312 *flagp = WORST; /* Tentatively. */
a0d0e21e 1313
77d41b28 1314 regcomp_parse--;
a0d0e21e 1315 nextchar();
77d41b28 1316 while (regcomp_parse < regxend && *regcomp_parse != '|' && *regcomp_parse != ')') {
a0d0e21e
LW
1317 flags &= ~TRYAGAIN;
1318 latest = regpiece(&flags);
1319 if (latest == NULL) {
1320 if (flags & TRYAGAIN)
1321 continue;
1322 return(NULL);
c277df42
IZ
1323 } else if (ret == NULL)
1324 ret = latest;
a0d0e21e 1325 *flagp |= flags&HASWIDTH;
c277df42 1326 if (chain == NULL) /* First piece. */
a0d0e21e
LW
1327 *flagp |= flags&SPSTART;
1328 else {
1329 regnaughty++;
1330 regtail(chain, latest);
a687059c 1331 }
a0d0e21e 1332 chain = latest;
c277df42
IZ
1333 c++;
1334 }
1335 if (chain == NULL) { /* Loop ran zero times. */
1336 chain = reg_node(NOTHING);
1337 if (ret == NULL)
1338 ret = chain;
1339 }
1340 if (c == 1) {
1341 *flagp |= flags&SIMPLE;
a0d0e21e 1342 }
a687059c 1343
a0d0e21e 1344 return(ret);
a687059c
LW
1345}
1346
1347/*
1348 - regpiece - something followed by possible [*+?]
1349 *
1350 * Note that the branching code sequences used for ? and the general cases
1351 * of * and + are somewhat optimized: they use the same NOTHING node as
1352 * both the endmarker for their branch list and the body of the last branch.
1353 * It might seem that this node could be dispensed with entirely, but the
1354 * endmarker role is not redundant.
1355 */
76e3520e 1356STATIC regnode *
8ac85365 1357regpiece(I32 *flagp)
a687059c 1358{
5c0ca799 1359 dTHR;
c277df42 1360 register regnode *ret;
a0d0e21e
LW
1361 register char op;
1362 register char *next;
1363 I32 flags;
77d41b28 1364 char *origparse = regcomp_parse;
a0d0e21e
LW
1365 char *maxpos;
1366 I32 min;
c277df42 1367 I32 max = REG_INFTY;
a0d0e21e
LW
1368
1369 ret = regatom(&flags);
1370 if (ret == NULL) {
1371 if (flags & TRYAGAIN)
1372 *flagp |= TRYAGAIN;
1373 return(NULL);
1374 }
1375
77d41b28 1376 op = *regcomp_parse;
a0d0e21e 1377
77d41b28
GS
1378 if (op == '{' && regcurly(regcomp_parse)) {
1379 next = regcomp_parse + 1;
a0d0e21e
LW
1380 maxpos = Nullch;
1381 while (isDIGIT(*next) || *next == ',') {
1382 if (*next == ',') {
1383 if (maxpos)
1384 break;
1385 else
1386 maxpos = next;
a687059c 1387 }
a0d0e21e
LW
1388 next++;
1389 }
1390 if (*next == '}') { /* got one */
1391 if (!maxpos)
1392 maxpos = next;
77d41b28
GS
1393 regcomp_parse++;
1394 min = atoi(regcomp_parse);
a0d0e21e
LW
1395 if (*maxpos == ',')
1396 maxpos++;
1397 else
77d41b28 1398 maxpos = regcomp_parse;
a0d0e21e
LW
1399 max = atoi(maxpos);
1400 if (!max && *maxpos != '0')
c277df42
IZ
1401 max = REG_INFTY; /* meaning "infinity" */
1402 else if (max >= REG_INFTY)
1403 FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
77d41b28 1404 regcomp_parse = next;
a0d0e21e
LW
1405 nextchar();
1406
1407 do_curly:
1408 if ((flags&SIMPLE)) {
1409 regnaughty += 2 + regnaughty / 2;
1410 reginsert(CURLY, ret);
1411 }
1412 else {
1413 regnaughty += 4 + regnaughty; /* compound interest */
c277df42
IZ
1414 regtail(ret, reg_node(WHILEM));
1415 if (!SIZE_ONLY && extralen) {
1416 reginsert(LONGJMP,ret);
1417 reginsert(NOTHING,ret);
1418 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
1419 }
a0d0e21e 1420 reginsert(CURLYX,ret);
c277df42
IZ
1421 if (!SIZE_ONLY && extralen)
1422 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
1423 regtail(ret, reg_node(NOTHING));
1424 if (SIZE_ONLY)
1425 extralen += 3;
a0d0e21e 1426 }
c277df42 1427 ret->flags = 0;
a0d0e21e
LW
1428
1429 if (min > 0)
821b33a5
IZ
1430 *flagp = WORST;
1431 if (max > 0)
1432 *flagp |= HASWIDTH;
a0d0e21e 1433 if (max && max < min)
c277df42
IZ
1434 FAIL("Can't do {n,m} with n > m");
1435 if (!SIZE_ONLY) {
1436 ARG1_SET(ret, min);
1437 ARG2_SET(ret, max);
a687059c 1438 }
a687059c 1439
a0d0e21e 1440 goto nest_check;
a687059c 1441 }
a0d0e21e 1442 }
a687059c 1443
a0d0e21e
LW
1444 if (!ISMULT1(op)) {
1445 *flagp = flags;
a687059c 1446 return(ret);
a0d0e21e 1447 }
bb20fd44 1448
c277df42 1449#if 0 /* Now runtime fix should be reliable. */
bb20fd44 1450 if (!(flags&HASWIDTH) && op != '?')
c277df42
IZ
1451 FAIL("regexp *+ operand could be empty");
1452#endif
bb20fd44 1453
a0d0e21e
LW
1454 nextchar();
1455
821b33a5 1456 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
a0d0e21e
LW
1457
1458 if (op == '*' && (flags&SIMPLE)) {
1459 reginsert(STAR, ret);
c277df42 1460 ret->flags = 0;
a0d0e21e
LW
1461 regnaughty += 4;
1462 }
1463 else if (op == '*') {
1464 min = 0;
1465 goto do_curly;
1466 } else if (op == '+' && (flags&SIMPLE)) {
1467 reginsert(PLUS, ret);
c277df42 1468 ret->flags = 0;
a0d0e21e
LW
1469 regnaughty += 3;
1470 }
1471 else if (op == '+') {
1472 min = 1;
1473 goto do_curly;
1474 } else if (op == '?') {
1475 min = 0; max = 1;
1476 goto do_curly;
1477 }
1478 nest_check:
c277df42 1479 if (dowarn && !SIZE_ONLY && !(flags&HASWIDTH) && max > 10000) {
a0d0e21e 1480 warn("%.*s matches null string many times",
77d41b28 1481 regcomp_parse - origparse, origparse);
a0d0e21e
LW
1482 }
1483
77d41b28 1484 if (*regcomp_parse == '?') {
a0d0e21e
LW
1485 nextchar();
1486 reginsert(MINMOD, ret);
c277df42 1487 regtail(ret, ret + NODE_STEP_REGNODE);
a0d0e21e 1488 }
77d41b28 1489 if (ISMULT2(regcomp_parse))
a0d0e21e
LW
1490 FAIL("nested *?+ in regexp");
1491
1492 return(ret);
a687059c
LW
1493}
1494
1495/*
1496 - regatom - the lowest level
1497 *
1498 * Optimization: gobbles an entire sequence of ordinary characters so that
1499 * it can turn them into a single node, which is smaller to store and
1500 * faster to run. Backslashed characters are exceptions, each becoming a
1501 * separate node; the code is simpler that way and it's not worth fixing.
1502 *
1503 * [Yes, it is worth fixing, some scripts can run twice the speed.]
1504 */
76e3520e 1505STATIC regnode *
8ac85365 1506regatom(I32 *flagp)
a687059c 1507{
5c0ca799 1508 dTHR;
c277df42 1509 register regnode *ret = 0;
a0d0e21e
LW
1510 I32 flags;
1511
1512 *flagp = WORST; /* Tentatively. */
1513
1514tryagain:
77d41b28 1515 switch (*regcomp_parse) {
a0d0e21e 1516 case '^':
c277df42 1517 seen_zerolen++;
a0d0e21e
LW
1518 nextchar();
1519 if (regflags & PMf_MULTILINE)
c277df42 1520 ret = reg_node(MBOL);
a0d0e21e 1521 else if (regflags & PMf_SINGLELINE)
c277df42 1522 ret = reg_node(SBOL);
a0d0e21e 1523 else
c277df42 1524 ret = reg_node(BOL);
a0d0e21e
LW
1525 break;
1526 case '$':
77d41b28 1527 if (regcomp_parse[1])
c277df42 1528 seen_zerolen++;
a0d0e21e
LW
1529 nextchar();
1530 if (regflags & PMf_MULTILINE)
c277df42 1531 ret = reg_node(MEOL);
a0d0e21e 1532 else if (regflags & PMf_SINGLELINE)
c277df42 1533 ret = reg_node(SEOL);
a0d0e21e 1534 else
c277df42 1535 ret = reg_node(EOL);
a0d0e21e
LW
1536 break;
1537 case '.':
1538 nextchar();
1539 if (regflags & PMf_SINGLELINE)
c277df42 1540 ret = reg_node(SANY);
a0d0e21e 1541 else
c277df42 1542 ret = reg_node(ANY);
a0d0e21e
LW
1543 regnaughty++;
1544 *flagp |= HASWIDTH|SIMPLE;
1545 break;
1546 case '[':
77d41b28 1547 regcomp_parse++;
a0d0e21e
LW
1548 ret = regclass();
1549 *flagp |= HASWIDTH|SIMPLE;
1550 break;
1551 case '(':
1552 nextchar();
1553 ret = reg(1, &flags);
1554 if (ret == NULL) {
1555 if (flags & TRYAGAIN)
1556 goto tryagain;
1557 return(NULL);
1558 }
c277df42 1559 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
a0d0e21e
LW
1560 break;
1561 case '|':
1562 case ')':
1563 if (flags & TRYAGAIN) {
1564 *flagp |= TRYAGAIN;
1565 return NULL;
1566 }
77d41b28 1567 FAIL2("internal urp in regexp at /%s/", regcomp_parse);
a0d0e21e
LW
1568 /* Supposed to be caught earlier. */
1569 break;
85afd4ae 1570 case '{':
77d41b28
GS
1571 if (!regcurly(regcomp_parse)) {
1572 regcomp_parse++;
85afd4ae
CS
1573 goto defchar;
1574 }
1575 /* FALL THROUGH */
a0d0e21e
LW
1576 case '?':
1577 case '+':
1578 case '*':
3115e423 1579 FAIL("?+*{} follows nothing in regexp");
a0d0e21e
LW
1580 break;
1581 case '\\':
77d41b28 1582 switch (*++regcomp_parse) {
a0d0e21e 1583 case 'A':
c277df42
IZ
1584 seen_zerolen++;
1585 ret = reg_node(SBOL);
a0d0e21e
LW
1586 *flagp |= SIMPLE;
1587 nextchar();
1588 break;
1589 case 'G':
c277df42
IZ
1590 ret = reg_node(GPOS);
1591 regseen |= REG_SEEN_GPOS;
a0d0e21e
LW
1592 *flagp |= SIMPLE;
1593 nextchar();
1594 break;
1595 case 'Z':
c277df42 1596 ret = reg_node(SEOL);
a0d0e21e
LW
1597 *flagp |= SIMPLE;
1598 nextchar();
1599 break;
b85d18e9
IZ
1600 case 'z':
1601 ret = reg_node(EOS);
1602 *flagp |= SIMPLE;
1603 seen_zerolen++; /* Do not optimize RE away */
1604 nextchar();
1605 break;
a0d0e21e 1606 case 'w':
c277df42 1607 ret = reg_node((regflags & PMf_LOCALE) ? ALNUML : ALNUM);
a0d0e21e
LW
1608 *flagp |= HASWIDTH|SIMPLE;
1609 nextchar();
1610 break;
1611 case 'W':
c277df42 1612 ret = reg_node((regflags & PMf_LOCALE) ? NALNUML : NALNUM);
a0d0e21e
LW
1613 *flagp |= HASWIDTH|SIMPLE;
1614 nextchar();
1615 break;
1616 case 'b':
c277df42
IZ
1617 seen_zerolen++;
1618 ret = reg_node((regflags & PMf_LOCALE) ? BOUNDL : BOUND);
a0d0e21e
LW
1619 *flagp |= SIMPLE;
1620 nextchar();
1621 break;
1622 case 'B':
c277df42
IZ
1623 seen_zerolen++;
1624 ret = reg_node((regflags & PMf_LOCALE) ? NBOUNDL : NBOUND);
a0d0e21e
LW
1625 *flagp |= SIMPLE;
1626 nextchar();
1627 break;
1628 case 's':
c277df42 1629 ret = reg_node((regflags & PMf_LOCALE) ? SPACEL : SPACE);
a0d0e21e
LW
1630 *flagp |= HASWIDTH|SIMPLE;
1631 nextchar();
1632 break;
1633 case 'S':
c277df42 1634 ret = reg_node((regflags & PMf_LOCALE) ? NSPACEL : NSPACE);
a0d0e21e
LW
1635 *flagp |= HASWIDTH|SIMPLE;
1636 nextchar();
1637 break;
1638 case 'd':
c277df42 1639 ret = reg_node(DIGIT);
a0d0e21e
LW
1640 *flagp |= HASWIDTH|SIMPLE;
1641 nextchar();
1642 break;
1643 case 'D':
c277df42 1644 ret = reg_node(NDIGIT);
a0d0e21e
LW
1645 *flagp |= HASWIDTH|SIMPLE;
1646 nextchar();
1647 break;
1648 case 'n':
1649 case 'r':
1650 case 't':
1651 case 'f':
1652 case 'e':
1653 case 'a':
1654 case 'x':
1655 case 'c':
1656 case '0':
1657 goto defchar;
1658 case '1': case '2': case '3': case '4':
1659 case '5': case '6': case '7': case '8': case '9':
1660 {
77d41b28 1661 I32 num = atoi(regcomp_parse);
a0d0e21e
LW
1662
1663 if (num > 9 && num >= regnpar)
1664 goto defchar;
1665 else {
ef64f398
HS
1666 if (!SIZE_ONLY && num > regcomp_rx->nparens)
1667 FAIL("reference to nonexistent group");
a0d0e21e 1668 regsawback = 1;
c8756f30
AK
1669 ret = reganode((regflags & PMf_FOLD)
1670 ? ((regflags & PMf_LOCALE) ? REFFL : REFF)
1671 : REF, num);
a0d0e21e 1672 *flagp |= HASWIDTH;
77d41b28
GS
1673 while (isDIGIT(*regcomp_parse))
1674 regcomp_parse++;
1675 regcomp_parse--;
a0d0e21e
LW
1676 nextchar();
1677 }
1678 }
1679 break;
1680 case '\0':
77d41b28 1681 if (regcomp_parse >= regxend)
a0d0e21e
LW
1682 FAIL("trailing \\ in regexp");
1683 /* FALL THROUGH */
1684 default:
1685 goto defchar;
1686 }
1687 break;
4633a7c4
LW
1688
1689 case '#':
1690 if (regflags & PMf_EXTENDED) {
77d41b28
GS
1691 while (regcomp_parse < regxend && *regcomp_parse != '\n') regcomp_parse++;
1692 if (regcomp_parse < regxend)
4633a7c4
LW
1693 goto tryagain;
1694 }
1695 /* FALL THROUGH */
1696
a0d0e21e
LW
1697 default: {
1698 register I32 len;
c277df42 1699 register U8 ender;
a0d0e21e 1700 register char *p;
c277df42 1701 char *oldp, *s;
a0d0e21e
LW
1702 I32 numlen;
1703
77d41b28 1704 regcomp_parse++;
a0d0e21e
LW
1705
1706 defchar:
c277df42 1707 ret = reg_node((regflags & PMf_FOLD)
bbce6d69
PP
1708 ? ((regflags & PMf_LOCALE) ? EXACTFL : EXACTF)
1709 : EXACT);
161b471a 1710 s = (char *) OPERAND(ret);
c277df42 1711 regc(0, s++); /* save spot for len */
77d41b28 1712 for (len = 0, p = regcomp_parse - 1;
a0d0e21e
LW
1713 len < 127 && p < regxend;
1714 len++)
1715 {
1716 oldp = p;
5b5a24f7
CS
1717
1718 if (regflags & PMf_EXTENDED)
1719 p = regwhite(p, regxend);
a0d0e21e
LW
1720 switch (*p) {
1721 case '^':
1722 case '$':
1723 case '.':
1724 case '[':
1725 case '(':
1726 case ')':
1727 case '|':
1728 goto loopdone;
1729 case '\\':
1730 switch (*++p) {
1731 case 'A':
1732 case 'G':
1733 case 'Z':
b85d18e9 1734 case 'z':
a0d0e21e
LW
1735 case 'w':
1736 case 'W':
1737 case 'b':
1738 case 'B':
1739 case 's':
1740 case 'S':
1741 case 'd':
1742 case 'D':
1743 --p;
1744 goto loopdone;
1745 case 'n':
1746 ender = '\n';
1747 p++;
a687059c 1748 break;
a0d0e21e
LW
1749 case 'r':
1750 ender = '\r';
1751 p++;
a687059c 1752 break;
a0d0e21e
LW
1753 case 't':
1754 ender = '\t';
1755 p++;
a687059c 1756 break;
a0d0e21e
LW
1757 case 'f':
1758 ender = '\f';
1759 p++;
a687059c 1760 break;
a0d0e21e
LW
1761 case 'e':
1762 ender = '\033';
1763 p++;
a687059c 1764 break;
a0d0e21e
LW
1765 case 'a':
1766 ender = '\007';
1767 p++;
a687059c 1768 break;
a0d0e21e
LW
1769 case 'x':
1770 ender = scan_hex(++p, 2, &numlen);
1771 p += numlen;
a687059c 1772 break;
a0d0e21e
LW
1773 case 'c':
1774 p++;
bbce6d69
PP
1775 ender = UCHARAT(p++);
1776 ender = toCTRL(ender);
a687059c 1777 break;
a0d0e21e
LW
1778 case '0': case '1': case '2': case '3':case '4':
1779 case '5': case '6': case '7': case '8':case '9':
1780 if (*p == '0' ||
1781 (isDIGIT(p[1]) && atoi(p) >= regnpar) ) {
1782 ender = scan_oct(p, 3, &numlen);
1783 p += numlen;
1784 }
1785 else {
1786 --p;
1787 goto loopdone;
a687059c
LW
1788 }
1789 break;
a0d0e21e
LW
1790 case '\0':
1791 if (p >= regxend)
a687059c
LW
1792 FAIL("trailing \\ in regexp");
1793 /* FALL THROUGH */
a0d0e21e
LW
1794 default:
1795 ender = *p++;
1796 break;
1797 }
1798 break;
a687059c 1799 default:
a0d0e21e
LW
1800 ender = *p++;
1801 break;
a687059c 1802 }
5b5a24f7
CS
1803 if (regflags & PMf_EXTENDED)
1804 p = regwhite(p, regxend);
a0d0e21e
LW
1805 if (ISMULT2(p)) { /* Back off on ?+*. */
1806 if (len)
1807 p = oldp;
1808 else {
1809 len++;
c277df42 1810 regc(ender, s++);
a0d0e21e
LW
1811 }
1812 break;
a687059c 1813 }
c277df42 1814 regc(ender, s++);
a0d0e21e
LW
1815 }
1816 loopdone:
77d41b28 1817 regcomp_parse = p - 1;
a0d0e21e
LW
1818 nextchar();
1819 if (len < 0)
1820 FAIL("internal disaster in regexp");
1821 if (len > 0)
1822 *flagp |= HASWIDTH;
1823 if (len == 1)
1824 *flagp |= SIMPLE;
c277df42 1825 if (!SIZE_ONLY)
a0d0e21e 1826 *OPERAND(ret) = len;
c277df42
IZ
1827 regc('\0', s++);
1828 if (SIZE_ONLY) {
c277df42 1829 regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
c277df42
IZ
1830 } else {
1831 regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
1832 }
a687059c 1833 }
a0d0e21e
LW
1834 break;
1835 }
a687059c 1836
a0d0e21e 1837 return(ret);
a687059c
LW
1838}
1839
873ef191 1840STATIC char *
8ac85365 1841regwhite(char *p, char *e)
5b5a24f7
CS
1842{
1843 while (p < e) {
1844 if (isSPACE(*p))
1845 ++p;
1846 else if (*p == '#') {
1847 do {
1848 p++;
1849 } while (p < e && *p != '\n');
1850 }
1851 else
1852 break;
1853 }
1854 return p;
1855}
1856
76e3520e 1857STATIC regnode *
8ac85365 1858regclass(void)
a687059c 1859{
5c0ca799 1860 dTHR;
c277df42 1861 register char *opnd, *s;
8ac85365 1862 register I32 Class;
a0d0e21e
LW
1863 register I32 lastclass = 1234;
1864 register I32 range = 0;
c277df42 1865 register regnode *ret;
a0d0e21e
LW
1866 register I32 def;
1867 I32 numlen;
1868
161b471a 1869 s = opnd = (char *) OPERAND(regcode);
c277df42 1870 ret = reg_node(ANYOF);
8ac85365 1871 for (Class = 0; Class < 33; Class++)
c277df42 1872 regc(0, s++);
77d41b28 1873 if (*regcomp_parse == '^') { /* Complement of range. */
a0d0e21e 1874 regnaughty++;
77d41b28 1875 regcomp_parse++;
c277df42 1876 if (!SIZE_ONLY)
bbce6d69
PP
1877 *opnd |= ANYOF_INVERT;
1878 }
c277df42
IZ
1879 if (!SIZE_ONLY) {
1880 regcode += ANY_SKIP;
bbce6d69
PP
1881 if (regflags & PMf_FOLD)
1882 *opnd |= ANYOF_FOLD;
1883 if (regflags & PMf_LOCALE)
1884 *opnd |= ANYOF_LOCALE;
c277df42
IZ
1885 } else {
1886 regsize += ANY_SKIP;
a0d0e21e 1887 }
77d41b28 1888 if (*regcomp_parse == ']' || *regcomp_parse == '-')
a0d0e21e 1889 goto skipcond; /* allow 1st char to be ] or - */
77d41b28 1890 while (regcomp_parse < regxend && *regcomp_parse != ']') {
a0d0e21e 1891 skipcond:
77d41b28
GS
1892 Class = UCHARAT(regcomp_parse++);
1893 if (Class == '[' && regcomp_parse + 1 < regxend &&
4599a1de 1894 /* I smell either [: or [= or [. -- POSIX has been here, right? */
77d41b28
GS
1895 (*regcomp_parse == ':' || *regcomp_parse == '=' || *regcomp_parse == '.')) {
1896 char posixccc = *regcomp_parse;
1897 char* posixccs = regcomp_parse++;
4599a1de 1898
77d41b28
GS
1899 while (regcomp_parse < regxend && *regcomp_parse != posixccc)
1900 regcomp_parse++;
1901 if (regcomp_parse == regxend)
4599a1de 1902 /* Grandfather lone [:, [=, [. */
77d41b28 1903 regcomp_parse = posixccs;
4599a1de 1904 else {
77d41b28
GS
1905 regcomp_parse++; /* skip over the posixccc */
1906 if (*regcomp_parse == ']') {
4599a1de
JH
1907 /* Not Implemented Yet.
1908 * (POSIX Extended Character Classes, that is)
1909 * The text between e.g. [: and :] would start
77d41b28 1910 * at posixccs + 1 and stop at regcomp_parse - 2. */
4599a1de
JH
1911 if (dowarn && !SIZE_ONLY)
1912 warn("Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc);
77d41b28 1913 regcomp_parse++; /* skip over the ending ] */
4599a1de
JH
1914 }
1915 }
1916 }
8ac85365 1917 if (Class == '\\') {
77d41b28 1918 Class = UCHARAT(regcomp_parse++);
8ac85365 1919 switch (Class) {
a0d0e21e 1920 case 'w':
ae5c130c
GS
1921 if (!SIZE_ONLY) {
1922 if (regflags & PMf_LOCALE)
bbce6d69 1923 *opnd |= ANYOF_ALNUML;
ae5c130c
GS
1924 else {
1925 for (Class = 0; Class < 256; Class++)
1926 if (isALNUM(Class))
1927 ANYOF_SET(opnd, Class);
1928 }
bbce6d69 1929 }
a0d0e21e
LW
1930 lastclass = 1234;
1931 continue;
1932 case 'W':
ae5c130c
GS
1933 if (!SIZE_ONLY) {
1934 if (regflags & PMf_LOCALE)
bbce6d69 1935 *opnd |= ANYOF_NALNUML;
ae5c130c
GS
1936 else {
1937 for (Class = 0; Class < 256; Class++)
1938 if (!isALNUM(Class))
1939 ANYOF_SET(opnd, Class);
1940 }
bbce6d69 1941 }
a0d0e21e
LW
1942 lastclass = 1234;
1943 continue;
1944 case 's':
ae5c130c
GS
1945 if (!SIZE_ONLY) {
1946 if (regflags & PMf_LOCALE)
bbce6d69 1947 *opnd |= ANYOF_SPACEL;
ae5c130c
GS
1948 else {
1949 for (Class = 0; Class < 256; Class++)
1950 if (isSPACE(Class))
1951 ANYOF_SET(opnd, Class);
1952 }
bbce6d69 1953 }
a0d0e21e
LW
1954 lastclass = 1234;
1955 continue;
1956 case 'S':
ae5c130c
GS
1957 if (!SIZE_ONLY) {
1958 if (regflags & PMf_LOCALE)
bbce6d69 1959 *opnd |= ANYOF_NSPACEL;
ae5c130c
GS
1960 else {
1961 for (Class = 0; Class < 256; Class++)
1962 if (!isSPACE(Class))
1963 ANYOF_SET(opnd, Class);
1964 }
bbce6d69 1965 }
a0d0e21e
LW
1966 lastclass = 1234;
1967 continue;
1968 case 'd':
ae5c130c
GS
1969 if (!SIZE_ONLY) {
1970 for (Class = '0'; Class <= '9'; Class++)
1971 ANYOF_SET(opnd, Class);
1972 }
a0d0e21e
LW
1973 lastclass = 1234;
1974 continue;
1975 case 'D':
ae5c130c
GS
1976 if (!SIZE_ONLY) {
1977 for (Class = 0; Class < '0'; Class++)
1978 ANYOF_SET(opnd, Class);
1979 for (Class = '9' + 1; Class < 256; Class++)
1980 ANYOF_SET(opnd, Class);
1981 }
a0d0e21e
LW
1982 lastclass = 1234;
1983 continue;
1984 case 'n':
8ac85365 1985 Class = '\n';
a0d0e21e
LW
1986 break;
1987 case 'r':
8ac85365 1988 Class = '\r';
a0d0e21e
LW
1989 break;
1990 case 't':
8ac85365 1991 Class = '\t';
a0d0e21e
LW
1992 break;
1993 case 'f':
8ac85365 1994 Class = '\f';
a0d0e21e
LW
1995 break;
1996 case 'b':
8ac85365 1997 Class = '\b';
a0d0e21e
LW
1998 break;
1999 case 'e':
8ac85365 2000 Class = '\033';
a0d0e21e
LW
2001 break;
2002 case 'a':
8ac85365 2003 Class = '\007';
a0d0e21e
LW
2004 break;
2005 case 'x':
77d41b28
GS
2006 Class = scan_hex(regcomp_parse, 2, &numlen);
2007 regcomp_parse += numlen;
a0d0e21e
LW
2008 break;
2009 case 'c':
77d41b28 2010 Class = UCHARAT(regcomp_parse++);
8ac85365 2011 Class = toCTRL(Class);
a0d0e21e
LW
2012 break;
2013 case '0': case '1': case '2': case '3': case '4':
2014 case '5': case '6': case '7': case '8': case '9':
77d41b28
GS
2015 Class = scan_oct(--regcomp_parse, 3, &numlen);
2016 regcomp_parse += numlen;
a0d0e21e
LW
2017 break;
2018 }
2019 }
2020 if (range) {
8ac85365 2021 if (lastclass > Class)
a0d0e21e
LW
2022 FAIL("invalid [] range in regexp");
2023 range = 0;
2024 }
2025 else {
8ac85365 2026 lastclass = Class;
77d41b28
GS
2027 if (*regcomp_parse == '-' && regcomp_parse+1 < regxend &&
2028 regcomp_parse[1] != ']') {
2029 regcomp_parse++;
a0d0e21e
LW
2030 range = 1;
2031 continue; /* do it next time */
2032 }
a687059c 2033 }
ae5c130c
GS
2034 if (!SIZE_ONLY) {
2035 for ( ; lastclass <= Class; lastclass++)
2036 ANYOF_SET(opnd, lastclass);
2037 }
8ac85365 2038 lastclass = Class;
a0d0e21e 2039 }
77d41b28 2040 if (*regcomp_parse != ']')
a0d0e21e
LW
2041 FAIL("unmatched [] in regexp");
2042 nextchar();
ae5c130c
GS
2043 /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
2044 if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) {
2045 for (Class = 0; Class < 256; ++Class) {
2046 if (ANYOF_TEST(opnd, Class)) {
2047 I32 cf = fold[Class];
2048 ANYOF_SET(opnd, cf);
2049 }
2050 }
2051 *opnd &= ~ANYOF_FOLD;
2052 }
2053 /* optimize inverted simple patterns (e.g. [^a-z]) */
2054 if (!SIZE_ONLY && (*opnd & 0xFF) == ANYOF_INVERT) {
2055 for (Class = 0; Class < 32; ++Class)
2056 opnd[1 + Class] ^= 0xFF;
2057 *opnd = 0;
2058 }
a0d0e21e
LW
2059 return ret;
2060}
2061
76e3520e 2062STATIC char*
8ac85365 2063nextchar(void)
a0d0e21e 2064{
5c0ca799 2065 dTHR;
77d41b28 2066 char* retval = regcomp_parse++;
a0d0e21e 2067
4633a7c4 2068 for (;;) {
77d41b28
GS
2069 if (*regcomp_parse == '(' && regcomp_parse[1] == '?' &&
2070 regcomp_parse[2] == '#') {
2071 while (*regcomp_parse && *regcomp_parse != ')')
2072 regcomp_parse++;
2073 regcomp_parse++;
4633a7c4
LW
2074 continue;
2075 }
2076 if (regflags & PMf_EXTENDED) {
77d41b28
GS
2077 if (isSPACE(*regcomp_parse)) {
2078 regcomp_parse++;
748a9306
LW
2079 continue;
2080 }
77d41b28
GS
2081 else if (*regcomp_parse == '#') {
2082 while (*regcomp_parse && *regcomp_parse != '\n')
2083 regcomp_parse++;
2084 regcomp_parse++;
748a9306
LW
2085 continue;
2086 }
748a9306 2087 }
4633a7c4 2088 return retval;
a0d0e21e 2089 }
a687059c
LW
2090}
2091
2092/*
c277df42 2093- reg_node - emit a node
a0d0e21e 2094*/
76e3520e 2095STATIC regnode * /* Location. */
c277df42 2096reg_node(U8 op)
a687059c 2097{
5c0ca799 2098 dTHR;
c277df42
IZ
2099 register regnode *ret;
2100 register regnode *ptr;
a687059c 2101
a0d0e21e 2102 ret = regcode;
c277df42
IZ
2103 if (SIZE_ONLY) {
2104 SIZE_ALIGN(regsize);
c277df42 2105 regsize += 1;
a0d0e21e
LW
2106 return(ret);
2107 }
a687059c 2108
c277df42 2109 NODE_ALIGN_FILL(ret);
a0d0e21e 2110 ptr = ret;
c277df42 2111 FILL_ADVANCE_NODE(ptr, op);
a0d0e21e 2112 regcode = ptr;
a687059c 2113
a0d0e21e 2114 return(ret);
a687059c
LW
2115}
2116
2117/*
a0d0e21e
LW
2118- reganode - emit a node with an argument
2119*/
76e3520e 2120STATIC regnode * /* Location. */
c277df42 2121reganode(U8 op, U32 arg)
fe14fcc3 2122{
5c0ca799 2123 dTHR;
c277df42
IZ
2124 register regnode *ret;
2125 register regnode *ptr;
fe14fcc3 2126
a0d0e21e 2127 ret = regcode;
c277df42
IZ
2128 if (SIZE_ONLY) {
2129 SIZE_ALIGN(regsize);
c277df42 2130 regsize += 2;
a0d0e21e
LW
2131 return(ret);
2132 }
fe14fcc3 2133
c277df42 2134 NODE_ALIGN_FILL(ret);
a0d0e21e 2135 ptr = ret;
c277df42 2136 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
a0d0e21e 2137 regcode = ptr;
fe14fcc3 2138
a0d0e21e 2139 return(ret);
fe14fcc3
LW
2140}
2141
2142/*
a0d0e21e
LW
2143- regc - emit (if appropriate) a byte of code
2144*/
76e3520e 2145STATIC void
c277df42 2146regc(U8 b, char* s)
a687059c 2147{
5c0ca799 2148 dTHR;
c277df42
IZ
2149 if (!SIZE_ONLY)
2150 *s = b;
a687059c
LW
2151}
2152
2153/*
a0d0e21e
LW
2154- reginsert - insert an operator in front of already-emitted operand
2155*
2156* Means relocating the operand.
2157*/
76e3520e 2158STATIC void
c277df42 2159reginsert(U8 op, regnode *opnd)
a687059c 2160{
5c0ca799 2161 dTHR;
c277df42
IZ
2162 register regnode *src;
2163 register regnode *dst;
2164 register regnode *place;
2165 register int offset = regarglen[(U8)op];
2166
2167/* (regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
2168
2169 if (SIZE_ONLY) {
2170 regsize += NODE_STEP_REGNODE + offset;
a0d0e21e
LW
2171 return;
2172 }
a687059c 2173
a0d0e21e 2174 src = regcode;
c277df42 2175 regcode += NODE_STEP_REGNODE + offset;
a0d0e21e
LW
2176 dst = regcode;
2177 while (src > opnd)
c277df42 2178 StructCopy(--src, --dst, regnode);
a0d0e21e
LW
2179
2180 place = opnd; /* Op node, where operand used to be. */
c277df42
IZ
2181 src = NEXTOPER(place);
2182 FILL_ADVANCE_NODE(place, op);
2183 Zero(src, offset, regnode);
a687059c
LW
2184}
2185
2186/*
c277df42 2187- regtail - set the next-pointer at the end of a node chain of p to val.
a0d0e21e 2188*/
76e3520e 2189STATIC void
c277df42 2190regtail(regnode *p, regnode *val)
a687059c 2191{
5c0ca799 2192 dTHR;
c277df42
IZ
2193 register regnode *scan;
2194 register regnode *temp;
a0d0e21e
LW
2195 register I32 offset;
2196
c277df42 2197 if (SIZE_ONLY)
a0d0e21e
LW
2198 return;
2199
2200 /* Find last node. */
2201 scan = p;
2202 for (;;) {
2203 temp = regnext(scan);
2204 if (temp == NULL)
2205 break;
2206 scan = temp;
2207 }
a687059c 2208
c277df42
IZ
2209 if (reg_off_by_arg[OP(scan)]) {
2210 ARG_SET(scan, val - scan);
2211 } else {
2212 NEXT_OFF(scan) = val - scan;
2213 }
a687059c
LW
2214}
2215
2216/*
a0d0e21e
LW
2217- regoptail - regtail on operand of first argument; nop if operandless
2218*/
76e3520e 2219STATIC void
c277df42 2220regoptail(regnode *p, regnode *val)
a687059c 2221{
5c0ca799 2222 dTHR;
a0d0e21e 2223 /* "Operandless" and "op != BRANCH" are synonymous in practice. */
c277df42
IZ
2224 if (p == NULL || SIZE_ONLY)
2225 return;
2226 if (regkind[(U8)OP(p)] == BRANCH) {
2227 regtail(NEXTOPER(p), val);
2228 } else if ( regkind[(U8)OP(p)] == BRANCHJ) {
2229 regtail(NEXTOPER(NEXTOPER(p)), val);
2230 } else
a0d0e21e 2231 return;
a687059c
LW
2232}
2233
2234/*
2235 - regcurly - a little FSA that accepts {\d+,?\d*}
2236 */
79072805 2237STATIC I32
8ac85365 2238regcurly(register char *s)
a687059c
LW
2239{
2240 if (*s++ != '{')
2241 return FALSE;
f0fcb552 2242 if (!isDIGIT(*s))
a687059c 2243 return FALSE;
f0fcb552 2244 while (isDIGIT(*s))
a687059c
LW
2245 s++;
2246 if (*s == ',')
2247 s++;
f0fcb552 2248 while (isDIGIT(*s))
a687059c
LW
2249 s++;
2250 if (*s != '}')
2251 return FALSE;
2252 return TRUE;
2253}
2254
a687059c 2255
76e3520e 2256STATIC regnode *
c277df42
IZ
2257dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
2258{
35ff7856 2259#ifdef DEBUGGING
c277df42
IZ
2260 register char op = EXACT; /* Arbitrary non-END op. */
2261 register regnode *next, *onode;
2262
2263 while (op != END && (!last || node < last)) {
2264 /* While that wasn't END last time... */
2265
2266 NODE_ALIGN(node);
2267 op = OP(node);
2268 if (op == CLOSE)
2269 l--;
2270 next = regnext(node);
2271 /* Where, what. */
2272 if (OP(node) == OPTIMIZED)
2273 goto after_print;
2274 regprop(sv, node);
54dc92de 2275 PerlIO_printf(Perl_debug_log, "%4d:%*s%s", node - start,
c277df42
IZ
2276 2*l + 1, "", SvPVX(sv));
2277 if (next == NULL) /* Next ptr. */
2278 PerlIO_printf(Perl_debug_log, "(0)");
2279 else
2280 PerlIO_printf(Perl_debug_log, "(%d)", next - start);
2281 (void)PerlIO_putc(Perl_debug_log, '\n');
2282 after_print:
2283 if (regkind[(U8)op] == BRANCHJ) {
2284 register regnode *nnode = (OP(next) == LONGJMP
2285 ? regnext(next)
2286 : next);
2287 if (last && nnode > last)
2288 nnode = last;
2289 node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
2290 } else if (regkind[(U8)op] == BRANCH) {
2291 node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
2292 } else if ( op == CURLY) { /* `next' might be very big: optimizer */
2293 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
2294 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
2295 } else if (regkind[(U8)op] == CURLY && op != CURLYX) {
2296 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
2297 next, sv, l + 1);
2298 } else if ( op == PLUS || op == STAR) {
2299 node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
2300 } else if (op == ANYOF) {
2301 node = NEXTOPER(node);
2302 node += ANY_SKIP;
2303 } else if (regkind[(U8)op] == EXACT) {
2304 /* Literal string, where present. */
2305 node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode);
2306 node = NEXTOPER(node);
2307 } else {
2308 node = NEXTOPER(node);
2309 node += regarglen[(U8)op];
2310 }
2311 if (op == CURLYX || op == OPEN)
2312 l++;
2313 else if (op == WHILEM)
2314 l--;
2315 }
17c3b450 2316#endif /* DEBUGGING */
c277df42
IZ
2317 return node;
2318}
2319
a687059c 2320/*
fd181c75 2321 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
a687059c
LW
2322 */
2323void
8ac85365 2324regdump(regexp *r)
a687059c 2325{
35ff7856 2326#ifdef DEBUGGING
5c0ca799 2327 dTHR;
46fc3d4c 2328 SV *sv = sv_newmortal();
a687059c 2329
c277df42 2330 (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
a0d0e21e
LW
2331
2332 /* Header fields of interest. */
c277df42
IZ
2333 if (r->anchored_substr)
2334 PerlIO_printf(Perl_debug_log, "anchored `%s%s%s'%s at %d ",
2335 colors[0],
2336 SvPVX(r->anchored_substr),
2337 colors[1],
2338 SvTAIL(r->anchored_substr) ? "$" : "",
2339 r->anchored_offset);
2340 if (r->float_substr)
2341 PerlIO_printf(Perl_debug_log, "floating `%s%s%s'%s at %d..%u ",
2342 colors[0],
2343 SvPVX(r->float_substr),
2344 colors[1],
2345 SvTAIL(r->float_substr) ? "$" : "",
2346 r->float_min_offset, r->float_max_offset);
2347 if (r->check_substr)
2348 PerlIO_printf(Perl_debug_log,
2349 r->check_substr == r->float_substr
2350 ? "(checking floating" : "(checking anchored");
2351 if (r->reganch & ROPT_NOSCAN)
2352 PerlIO_printf(Perl_debug_log, " noscan");
2353 if (r->reganch & ROPT_CHECK_ALL)
2354 PerlIO_printf(Perl_debug_log, " isall");
2355 if (r->check_substr)
2356 PerlIO_printf(Perl_debug_log, ") ");
2357
46fc3d4c
PP
2358 if (r->regstclass) {
2359 regprop(sv, r->regstclass);
2360 PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
2361 }
774d564b
PP
2362 if (r->reganch & ROPT_ANCH) {
2363 PerlIO_printf(Perl_debug_log, "anchored");
2364 if (r->reganch & ROPT_ANCH_BOL)
2365 PerlIO_printf(Perl_debug_log, "(BOL)");
c277df42
IZ
2366 if (r->reganch & ROPT_ANCH_MBOL)
2367 PerlIO_printf(Perl_debug_log, "(MBOL)");
774d564b
PP
2368 if (r->reganch & ROPT_ANCH_GPOS)
2369 PerlIO_printf(Perl_debug_log, "(GPOS)");
2370 PerlIO_putc(Perl_debug_log, ' ');
2371 }
c277df42
IZ
2372 if (r->reganch & ROPT_GPOS_SEEN)
2373 PerlIO_printf(Perl_debug_log, "GPOS ");
a0d0e21e 2374 if (r->reganch & ROPT_SKIP)
760ac839 2375 PerlIO_printf(Perl_debug_log, "plus ");
a0d0e21e 2376 if (r->reganch & ROPT_IMPLICIT)
760ac839 2377 PerlIO_printf(Perl_debug_log, "implicit ");
760ac839 2378 PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
ce862d02
IZ
2379 if (r->reganch & ROPT_EVAL_SEEN)
2380 PerlIO_printf(Perl_debug_log, "with eval ");
760ac839 2381 PerlIO_printf(Perl_debug_log, "\n");
17c3b450 2382#endif /* DEBUGGING */
a687059c
LW
2383}
2384
2385/*
a0d0e21e
LW
2386- regprop - printable representation of opcode
2387*/
46fc3d4c 2388void
c277df42 2389regprop(SV *sv, regnode *o)
a687059c 2390{
35ff7856 2391#ifdef DEBUGGING
5c0ca799 2392 dTHR;
a0d0e21e
LW
2393 register char *p = 0;
2394
54dc92de 2395 sv_setpvn(sv, "", 0);
11343788 2396 switch (OP(o)) {
a0d0e21e
LW
2397 case BOL:
2398 p = "BOL";
2399 break;
2400 case MBOL:
2401 p = "MBOL";
2402 break;
2403 case SBOL:
2404 p = "SBOL";
2405 break;
2406 case EOL:
2407 p = "EOL";
2408 break;
b85d18e9
IZ
2409 case EOS:
2410 p = "EOS";
2411 break;
a0d0e21e
LW
2412 case MEOL:
2413 p = "MEOL";
2414 break;
2415 case SEOL:
2416 p = "SEOL";
2417 break;
2418 case ANY:
2419 p = "ANY";
2420 break;
2421 case SANY:
2422 p = "SANY";
2423 break;
2424 case ANYOF:
2425 p = "ANYOF";
2426 break;
2427 case BRANCH:
2428 p = "BRANCH";
2429 break;
bbce6d69 2430 case EXACT:
c277df42 2431 sv_catpvf(sv, "EXACT <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
bbce6d69
PP
2432 break;
2433 case EXACTF:
c277df42 2434 sv_catpvf(sv, "EXACTF <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
bbce6d69
PP
2435 break;
2436 case EXACTFL:
c277df42 2437 sv_catpvf(sv, "EXACTFL <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
a0d0e21e
LW
2438 break;
2439 case NOTHING:
2440 p = "NOTHING";
2441 break;
c277df42
IZ
2442 case TAIL:
2443 p = "TAIL";
2444 break;
a0d0e21e
LW
2445 case BACK:
2446 p = "BACK";
2447 break;
2448 case END:
2449 p = "END";
2450 break;
a0d0e21e
LW
2451 case BOUND:
2452 p = "BOUND";
2453 break;
bbce6d69
PP
2454 case BOUNDL:
2455 p = "BOUNDL";
2456 break;
a0d0e21e
LW
2457 case NBOUND:
2458 p = "NBOUND";
2459 break;
bbce6d69
PP
2460 case NBOUNDL:
2461 p = "NBOUNDL";
a0d0e21e
LW
2462 break;
2463 case CURLY:
5dc0d613 2464 sv_catpvf(sv, "CURLY {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e 2465 break;
c277df42 2466 case CURLYM:
c277df42 2467 sv_catpvf(sv, "CURLYM[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
c277df42
IZ
2468 break;
2469 case CURLYN:
c277df42 2470 sv_catpvf(sv, "CURLYN[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
c277df42 2471 break;
a0d0e21e 2472 case CURLYX:
5dc0d613 2473 sv_catpvf(sv, "CURLYX {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e
LW
2474 break;
2475 case REF:
c277df42 2476 sv_catpvf(sv, "REF%d", ARG(o));
a0d0e21e 2477 break;
c8756f30 2478 case REFF:
c277df42 2479 sv_catpvf(sv, "REFF%d", ARG(o));
c8756f30
AK
2480 break;
2481 case REFFL:
c277df42 2482 sv_catpvf(sv, "REFFL%d", ARG(o));
c8756f30 2483 break;
a0d0e21e 2484 case OPEN:
c277df42 2485 sv_catpvf(sv, "OPEN%d", ARG(o));
a0d0e21e
LW
2486 break;
2487 case CLOSE:
c277df42 2488 sv_catpvf(sv, "CLOSE%d", ARG(o));
a0d0e21e
LW
2489 p = NULL;
2490 break;
2491 case STAR:
2492 p = "STAR";
2493 break;
2494 case PLUS:
2495 p = "PLUS";
2496 break;
2497 case MINMOD:
2498 p = "MINMOD";
2499 break;
774d564b
PP
2500 case GPOS:
2501 p = "GPOS";
a0d0e21e
LW
2502 break;
2503 case UNLESSM:
c277df42 2504 sv_catpvf(sv, "UNLESSM[-%d]", o->flags);
a0d0e21e
LW
2505 break;
2506 case IFMATCH:
c277df42 2507 sv_catpvf(sv, "IFMATCH[-%d]", o->flags);
a0d0e21e
LW
2508 break;
2509 case SUCCEED:
2510 p = "SUCCEED";
2511 break;
2512 case WHILEM:
2513 p = "WHILEM";
2514 break;
bbce6d69
PP
2515 case DIGIT:
2516 p = "DIGIT";
2517 break;
2518 case NDIGIT:
2519 p = "NDIGIT";
2520 break;
2521 case ALNUM:
2522 p = "ALNUM";
2523 break;
2524 case NALNUM:
2525 p = "NALNUM";
2526 break;
2527 case SPACE:
2528 p = "SPACE";
2529 break;
2530 case NSPACE:
2531 p = "NSPACE";
2532 break;
2533 case ALNUML:
2534 p = "ALNUML";
2535 break;
2536 case NALNUML:
2537 p = "NALNUML";
2538 break;
2539 case SPACEL:
2540 p = "SPACEL";
2541 break;
2542 case NSPACEL:
2543 p = "NSPACEL";
2544 break;
c277df42
IZ
2545 case EVAL:
2546 p = "EVAL";
2547 break;
2548 case LONGJMP:
2549 p = "LONGJMP";
2550 break;
2551 case BRANCHJ:
2552 p = "BRANCHJ";
2553 break;
2554 case IFTHEN:
2555 p = "IFTHEN";
2556 break;
2557 case GROUPP:
2558 sv_catpvf(sv, "GROUPP%d", ARG(o));
2559 break;
2560 case LOGICAL:
2561 p = "LOGICAL";
2562 break;
2563 case SUSPEND:
2564 p = "SUSPEND";
2565 break;
2566 case RENUM:
2567 p = "RENUM";
2568 break;
2569 case OPTIMIZED:
2570 p = "OPTIMIZED";
2571 break;
a0d0e21e
LW
2572 default:
2573 FAIL("corrupted regexp opcode");
2574 }
46fc3d4c
PP
2575 if (p)
2576 sv_catpv(sv, p);
17c3b450 2577#endif /* DEBUGGING */
35ff7856 2578}
a687059c 2579
2b69d0c2 2580void
8ac85365 2581pregfree(struct regexp *r)
a687059c 2582{
5c0ca799 2583 dTHR;
c277df42 2584 if (!r || (--r->refcnt > 0))
a0d0e21e 2585 return;
c277df42 2586 if (r->precomp)
a0d0e21e 2587 Safefree(r->precomp);
c277df42 2588 if (r->subbase)
a0d0e21e 2589 Safefree(r->subbase);
a193d654
GS
2590 if (r->substrs) {
2591 if (r->anchored_substr)
2592 SvREFCNT_dec(r->anchored_substr);
2593 if (r->float_substr)
2594 SvREFCNT_dec(r->float_substr);
2779dcf1 2595 Safefree(r->substrs);
a193d654 2596 }
c277df42
IZ
2597 if (r->data) {
2598 int n = r->data->count;
2599 while (--n >= 0) {
2600 switch (r->data->what[n]) {
2601 case 's':
2602 SvREFCNT_dec((SV*)r->data->data[n]);
2603 break;
2604 case 'o':
2605 op_free((OP_4tree*)r->data->data[n]);
2606 break;
2607 case 'n':
2608 break;
2609 default:
2610 FAIL2("panic: regfree data code '%c'", r->data->what[n]);
2611 }
2612 }
2613 Safefree(r->data->what);
2614 Safefree(r->data);
a0d0e21e
LW
2615 }
2616 Safefree(r->startp);
2617 Safefree(r->endp);
2618 Safefree(r);
a687059c 2619}
c277df42
IZ
2620
2621/*
2622 - regnext - dig the "next" pointer out of a node
2623 *
2624 * [Note, when REGALIGN is defined there are two places in regmatch()
2625 * that bypass this code for speed.]
2626 */
2627regnode *
2628regnext(register regnode *p)
2629{
5c0ca799 2630 dTHR;
c277df42
IZ
2631 register I32 offset;
2632
2633 if (p == &regdummy)
2634 return(NULL);
2635
2636 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
2637 if (offset == 0)
2638 return(NULL);
2639
c277df42 2640 return(p+offset);
c277df42
IZ
2641}
2642
01f988be 2643STATIC void
c277df42 2644re_croak2(const char* pat1,const char* pat2,...)
c277df42
IZ
2645{
2646 va_list args;
2647 STRLEN l1 = strlen(pat1);
2648 STRLEN l2 = strlen(pat2);
2649 char buf[512];
2650 char *message;
2651
2652 if (l1 > 510)
2653 l1 = 510;
2654 if (l1 + l2 > 510)
2655 l2 = 510 - l1;
2656 Copy(pat1, buf, l1 , char);
2657 Copy(pat2, buf + l1, l2 , char);
3b818b81
GS
2658 buf[l1 + l2] = '\n';
2659 buf[l1 + l2 + 1] = '\0';
c277df42 2660 va_start(args, pat2);
c277df42
IZ
2661 message = mess(buf, &args);
2662 va_end(args);
2663 l1 = strlen(message);
2664 if (l1 > 512)
2665 l1 = 512;
2666 Copy(message, buf, l1 , char);
2667 buf[l1] = '\0'; /* Overwrite \n */
2668 croak("%s", buf);
2669}