This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
update repository copy of Asmdata.pm after `perl bytecode.pl`
[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
f0fcb552 22/*SUPPRESS 112*/
a687059c 23/*
e50aee73 24 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c
LW
25 *
26 * Copyright (c) 1986 by University of Toronto.
27 * Written by Henry Spencer. Not derived from licensed software.
28 *
29 * Permission is granted to anyone to use this software for any
30 * purpose on any computer system, and to redistribute it freely,
31 * subject to the following restrictions:
32 *
33 * 1. The author is not responsible for the consequences of use of
34 * this software, no matter how awful, even if they arise
35 * from defects in it.
36 *
37 * 2. The origin of this software must not be misrepresented, either
38 * by explicit claim or by omission.
39 *
40 * 3. Altered versions must be plainly marked as such, and must not
41 * be misrepresented as being the original software.
42 *
43 *
44 **** Alterations to Henry's code are...
45 ****
9607fc9c 46 **** Copyright (c) 1991-1997, Larry Wall
a687059c 47 ****
9ef589d8
LW
48 **** You may distribute under the terms of either the GNU General Public
49 **** License or the Artistic License, as specified in the README file.
50
a687059c
LW
51 *
52 * Beware that some of this code is subtly aware of the way operator
53 * precedence is structured in regular expressions. Serious changes in
54 * regular-expression syntax might require a total rethink.
55 */
56#include "EXTERN.h"
57#include "perl.h"
58#include "INTERN.h"
c277df42
IZ
59
60#define REG_COMP_C
a687059c
LW
61#include "regcomp.h"
62
d4cce5f1 63#ifdef op
11343788 64#undef op
d4cce5f1 65#endif /* op */
11343788 66
fe14fcc3
LW
67#ifdef MSDOS
68# if defined(BUGGY_MSC6)
69 /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
70 # pragma optimize("a",off)
71 /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
72 # pragma optimize("w",on )
73# endif /* BUGGY_MSC6 */
74#endif /* MSDOS */
75
a687059c
LW
76#ifndef STATIC
77#define STATIC static
78#endif
79
80#define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
81#define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
82 ((*s) == '{' && regcurly(s)))
2b69d0c2
LW
83#ifdef atarist
84#define PERL_META "^$.[()|?+*\\"
85#else
a687059c 86#define META "^$.[()|?+*\\"
2b69d0c2 87#endif
a687059c 88
35c8bce7
LW
89#ifdef SPSTART
90#undef SPSTART /* dratted cpp namespace... */
91#endif
a687059c
LW
92/*
93 * Flags to be passed up and down.
94 */
a687059c 95#define WORST 0 /* Worst case. */
a0d0e21e
LW
96#define HASWIDTH 0x1 /* Known never to match null string. */
97#define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
98#define SPSTART 0x4 /* Starts with * or +. */
99#define TRYAGAIN 0x8 /* Weeded out a declaration. */
a687059c
LW
100
101/*
e50aee73 102 * Forward declarations for pregcomp()'s friends.
a687059c 103 */
a0d0e21e 104
76e3520e 105#ifndef PERL_OBJECT
c277df42
IZ
106static regnode *reg _((I32, I32 *));
107static regnode *reganode _((U8, U32));
108static regnode *regatom _((I32 *));
109static regnode *regbranch _((I32 *, I32));
110static void regc _((U8, char *));
111static regnode *regclass _((void));
a0d0e21e 112STATIC I32 regcurly _((char *));
c277df42
IZ
113static regnode *reg_node _((U8));
114static regnode *regpiece _((I32 *));
115static void reginsert _((U8, regnode *));
116static void regoptail _((regnode *, regnode *));
c277df42 117static void regtail _((regnode *, regnode *));
873ef191 118static char* regwhite _((char *, char *));
a0d0e21e 119static char* nextchar _((void));
3bd495df 120static void re_croak2 _((const char* pat1,const char* pat2,...)) __attribute__((noreturn));
76e3520e 121#endif
a687059c 122
c277df42
IZ
123/* Length of a variant. */
124
76e3520e 125#ifndef PERL_OBJECT
c277df42
IZ
126typedef struct {
127 I32 len_min;
128 I32 len_delta;
129 I32 pos_min;
130 I32 pos_delta;
131 SV *last_found;
132 I32 last_end; /* min value, <0 unless valid. */
133 I32 last_start_min;
134 I32 last_start_max;
135 SV **longest; /* Either &l_fixed, or &l_float. */
136 SV *longest_fixed;
137 I32 offset_fixed;
138 SV *longest_float;
139 I32 offset_float_min;
140 I32 offset_float_max;
141 I32 flags;
142} scan_data_t;
76e3520e 143#endif
c277df42
IZ
144
145static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
146
147#define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
148#define SF_BEFORE_SEOL 0x1
149#define SF_BEFORE_MEOL 0x2
150#define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
151#define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
152
09b7f37c
CB
153#ifdef NO_UNARY_PLUS
154# define SF_FIX_SHIFT_EOL (0+2)
155# define SF_FL_SHIFT_EOL (0+4)
156#else
157# define SF_FIX_SHIFT_EOL (+2)
158# define SF_FL_SHIFT_EOL (+4)
159#endif
c277df42
IZ
160
161#define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
162#define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
163
164#define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
165#define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
166#define SF_IS_INF 0x40
167#define SF_HAS_PAR 0x80
168#define SF_IN_PAR 0x100
169#define SF_HAS_EVAL 0x200
4bfe0158 170#define SCF_DO_SUBSTR 0x400
c277df42 171
76e3520e 172STATIC void
c277df42
IZ
173scan_commit(scan_data_t *data)
174{
175 STRLEN l = SvCUR(data->last_found);
176 STRLEN old_l = SvCUR(*data->longest);
177
178 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
179 sv_setsv(*data->longest, data->last_found);
180 if (*data->longest == data->longest_fixed) {
181 data->offset_fixed = l ? data->last_start_min : data->pos_min;
182 if (data->flags & SF_BEFORE_EOL)
183 data->flags
184 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
185 else
186 data->flags &= ~SF_FIX_BEFORE_EOL;
187 } else {
188 data->offset_float_min = l ? data->last_start_min : data->pos_min;
189 data->offset_float_max = (l
190 ? data->last_start_max
191 : data->pos_min + data->pos_delta);
192 if (data->flags & SF_BEFORE_EOL)
193 data->flags
194 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
195 else
196 data->flags &= ~SF_FL_BEFORE_EOL;
197 }
198 }
199 SvCUR_set(data->last_found, 0);
200 data->last_end = -1;
201 data->flags &= ~SF_BEFORE_EOL;
202}
203
c277df42
IZ
204/* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
205 to the position after last scanned or to NULL. */
206
76e3520e 207STATIC I32
c277df42
IZ
208study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
209 /* scanp: Start here (read-write). */
210 /* deltap: Write maxlen-minlen here. */
211 /* last: Stop before this one. */
212{
213 I32 min = 0, pars = 0, code;
214 regnode *scan = *scanp, *next;
215 I32 delta = 0;
216 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
217 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
218 scan_data_t data_fake;
219
220 while (scan && OP(scan) != END && scan < last) {
221 /* Peephole optimizer: */
222
223 if (regkind[(U8)OP(scan)] == EXACT) {
224 regnode *n = regnext(scan);
225 U32 stringok = 1;
226#ifdef DEBUGGING
227 regnode *stop = scan;
228#endif
229
230 next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
231 /* Skip NOTHING, merge EXACT*. */
232 while (n &&
233 ( regkind[(U8)OP(n)] == NOTHING ||
234 (stringok && (OP(n) == OP(scan))))
235 && NEXT_OFF(n)
236 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
237 if (OP(n) == TAIL || n > next)
238 stringok = 0;
239 if (regkind[(U8)OP(n)] == NOTHING) {
240 NEXT_OFF(scan) += NEXT_OFF(n);
241 next = n + NODE_STEP_REGNODE;
242#ifdef DEBUGGING
243 if (stringok)
244 stop = n;
245#endif
246 n = regnext(n);
247 } else {
248 int oldl = *OPERAND(scan);
249 regnode *nnext = regnext(n);
250
251 if (oldl + *OPERAND(n) > U8_MAX)
252 break;
253 NEXT_OFF(scan) += NEXT_OFF(n);
254 *OPERAND(scan) += *OPERAND(n);
255 next = n + (*OPERAND(n) + 2 - 1)/sizeof(regnode) + 2;
256 /* Now we can overwrite *n : */
257 Move(OPERAND(n) + 1, OPERAND(scan) + oldl + 1,
258 *OPERAND(n) + 1, char);
259#ifdef DEBUGGING
260 if (stringok)
261 stop = next - 1;
262#endif
263 n = nnext;
264 }
265 }
266#ifdef DEBUGGING
267 /* Allow dumping */
268 n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
269 while (n <= stop) {
ca04da08
GS
270 /* Purify reports a benign UMR here sometimes, because we
271 * don't initialize the OP() slot of a node when that node
272 * is occupied by just the trailing null of the string in
273 * an EXACT node */
c277df42
IZ
274 if (regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
275 OP(n) = OPTIMIZED;
276 NEXT_OFF(n) = 0;
277 }
278 n++;
279 }
280#endif
281
282 }
283 if (OP(scan) != CURLYX) {
284 int max = (reg_off_by_arg[OP(scan)] ? I32_MAX : U16_MAX);
285 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
286 int noff;
287 regnode *n = scan;
288
289 /* Skip NOTHING and LONGJMP. */
290 while ((n = regnext(n))
291 && ((regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
292 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
293 && off + noff < max)
294 off += noff;
295 if (reg_off_by_arg[OP(scan)])
296 ARG(scan) = off;
297 else
298 NEXT_OFF(scan) = off;
299 }
300 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
301 || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
302 next = regnext(scan);
303 code = OP(scan);
304
305 if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
306 I32 max1 = 0, min1 = I32_MAX, num = 0;
307
308 if (flags & SCF_DO_SUBSTR)
309 scan_commit(data);
310 while (OP(scan) == code) {
311 I32 deltanext, minnext;
312
313 num++;
314 data_fake.flags = 0;
315 next = regnext(scan);
316 scan = NEXTOPER(scan);
317 if (code != BRANCH)
318 scan = NEXTOPER(scan);
319 /* We suppose the run is continuous, last=next...*/
320 minnext = study_chunk(&scan, &deltanext, next,
321 &data_fake, 0);
322 if (min1 > minnext)
323 min1 = minnext;
324 if (max1 < minnext + deltanext)
325 max1 = minnext + deltanext;
326 if (deltanext == I32_MAX)
327 is_inf = 1;
328 scan = next;
329 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
330 pars++;
405ff068 331 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42
IZ
332 data->flags |= SF_HAS_EVAL;
333 if (code == SUSPEND)
334 break;
335 }
336 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
337 min1 = 0;
338 if (flags & SCF_DO_SUBSTR) {
339 data->pos_min += min1;
340 data->pos_delta += max1 - min1;
341 if (max1 != min1 || is_inf)
342 data->longest = &(data->longest_float);
343 }
344 min += min1;
345 delta += max1 - min1;
346 } else if (code == BRANCHJ) /* single branch is optimized. */
347 scan = NEXTOPER(NEXTOPER(scan));
348 else /* single branch is optimized. */
349 scan = NEXTOPER(scan);
350 continue;
351 } else if (OP(scan) == EXACT) {
352 min += *OPERAND(scan);
353 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
354 I32 l = *OPERAND(scan);
355
356 /* The code below prefers earlier match for fixed
357 offset, later match for variable offset. */
358 if (data->last_end == -1) { /* Update the start info. */
359 data->last_start_min = data->pos_min;
360 data->last_start_max = is_inf
361 ? I32_MAX : data->pos_min + data->pos_delta;
362 }
161b471a 363 sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), l);
c277df42
IZ
364 data->last_end = data->pos_min + l;
365 data->pos_min += l; /* As in the first entry. */
366 data->flags &= ~SF_BEFORE_EOL;
367 }
368 } else if (regkind[(U8)OP(scan)] == EXACT) {
369 if (flags & SCF_DO_SUBSTR)
370 scan_commit(data);
371 min += *OPERAND(scan);
372 if (data && (flags & SCF_DO_SUBSTR))
373 data->pos_min += *OPERAND(scan);
374 } else if (strchr(varies,OP(scan))) {
375 I32 mincount, maxcount, minnext, deltanext, pos_before, fl;
376 regnode *oscan = scan;
377
378 switch (regkind[(U8)OP(scan)]) {
379 case WHILEM:
380 scan = NEXTOPER(scan);
381 goto finish;
382 case PLUS:
383 if (flags & SCF_DO_SUBSTR) {
384 next = NEXTOPER(scan);
385 if (OP(next) == EXACT) {
386 mincount = 1;
387 maxcount = REG_INFTY;
388 next = regnext(scan);
389 scan = NEXTOPER(scan);
390 goto do_curly;
391 }
392 }
393 if (flags & SCF_DO_SUBSTR)
394 data->pos_min++;
395 min++;
396 /* Fall through. */
397 case STAR:
398 is_inf = 1;
399 scan = regnext(scan);
400 if (flags & SCF_DO_SUBSTR) {
401 scan_commit(data);
402 data->longest = &(data->longest_float);
403 }
404 goto optimize_curly_tail;
405 case CURLY:
406 mincount = ARG1(scan);
407 maxcount = ARG2(scan);
408 next = regnext(scan);
409 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
410 do_curly:
411 if (flags & SCF_DO_SUBSTR) {
412 if (mincount == 0) scan_commit(data);
413 pos_before = data->pos_min;
414 }
415 if (data) {
416 fl = data->flags;
417 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
418 if (is_inf)
419 data->flags |= SF_IS_INF;
420 }
421 /* This will finish on WHILEM, setting scan, or on NULL: */
422 minnext = study_chunk(&scan, &deltanext, last, data,
423 mincount == 0
424 ? (flags & ~SCF_DO_SUBSTR) : flags);
425 if (!scan) /* It was not CURLYX, but CURLY. */
426 scan = next;
427 if (dowarn && (minnext + deltanext == 0)
428 && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)))
429 warn("Strange *+?{} on zero-length expression");
430 min += minnext * mincount;
431 is_inf |= (maxcount == REG_INFTY && (minnext + deltanext) > 0
432 || deltanext == I32_MAX);
433 delta += (minnext + deltanext) * maxcount - minnext * mincount;
434
435 /* Try powerful optimization CURLYX => CURLYN. */
c277df42
IZ
436 if ( OP(oscan) == CURLYX && data
437 && data->flags & SF_IN_PAR
438 && !(data->flags & SF_HAS_EVAL)
439 && !deltanext && minnext == 1 ) {
440 /* Try to optimize to CURLYN. */
441 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
442 regnode *nxt1 = nxt, *nxt2;
443
444 /* Skip open. */
445 nxt = regnext(nxt);
446 if (!strchr(simple,OP(nxt))
447 && !(regkind[(U8)OP(nxt)] == EXACT
448 && *OPERAND(nxt) == 1))
449 goto nogo;
450 nxt2 = nxt;
451 nxt = regnext(nxt);
452 if (OP(nxt) != CLOSE)
453 goto nogo;
454 /* Now we know that nxt2 is the only contents: */
455 oscan->flags = ARG(nxt);
456 OP(oscan) = CURLYN;
457 OP(nxt1) = NOTHING; /* was OPEN. */
458#ifdef DEBUGGING
459 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
460 NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
461 NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
462 OP(nxt) = OPTIMIZED; /* was CLOSE. */
463 OP(nxt + 1) = OPTIMIZED; /* was count. */
464 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
465#endif
466 }
c277df42
IZ
467 nogo:
468
469 /* Try optimization CURLYX => CURLYM. */
470 if ( OP(oscan) == CURLYX && data
c277df42 471 && !(data->flags & SF_HAS_PAR)
c277df42
IZ
472 && !(data->flags & SF_HAS_EVAL)
473 && !deltanext ) {
474 /* XXXX How to optimize if data == 0? */
475 /* Optimize to a simpler form. */
476 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
477 regnode *nxt2;
478
479 OP(oscan) = CURLYM;
480 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
481 && (OP(nxt2) != WHILEM))
482 nxt = nxt2;
483 OP(nxt2) = SUCCEED; /* Whas WHILEM */
c277df42
IZ
484 /* Need to optimize away parenths. */
485 if (data->flags & SF_IN_PAR) {
486 /* Set the parenth number. */
487 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
488
489 if (OP(nxt) != CLOSE)
490 FAIL("panic opt close");
491 oscan->flags = ARG(nxt);
492 OP(nxt1) = OPTIMIZED; /* was OPEN. */
493 OP(nxt) = OPTIMIZED; /* was CLOSE. */
494#ifdef DEBUGGING
495 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
496 OP(nxt + 1) = OPTIMIZED; /* was count. */
497 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
498 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
499#endif
500#if 0
501 while ( nxt1 && (OP(nxt1) != WHILEM)) {
502 regnode *nnxt = regnext(nxt1);
503
504 if (nnxt == nxt) {
505 if (reg_off_by_arg[OP(nxt1)])
506 ARG_SET(nxt1, nxt2 - nxt1);
507 else if (nxt2 - nxt1 < U16_MAX)
508 NEXT_OFF(nxt1) = nxt2 - nxt1;
509 else
510 OP(nxt) = NOTHING; /* Cannot beautify */
511 }
512 nxt1 = nnxt;
513 }
514#endif
515 /* Optimize again: */
516 study_chunk(&nxt1, &deltanext, nxt, NULL, 0);
517 } else
518 oscan->flags = 0;
c277df42
IZ
519 }
520 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
521 pars++;
522 if (flags & SCF_DO_SUBSTR) {
523 SV *last_str = Nullsv;
524 int counted = mincount != 0;
525
526 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
527 I32 b = pos_before >= data->last_start_min
528 ? pos_before : data->last_start_min;
529 STRLEN l;
530 char *s = SvPV(data->last_found, l);
531
532 l -= b - data->last_start_min;
533 /* Get the added string: */
534 last_str = newSVpv(s + b - data->last_start_min, l);
535 if (deltanext == 0 && pos_before == b) {
536 /* What was added is a constant string */
537 if (mincount > 1) {
538 SvGROW(last_str, (mincount * l) + 1);
539 repeatcpy(SvPVX(last_str) + l,
540 SvPVX(last_str), l, mincount - 1);
541 SvCUR(last_str) *= mincount;
542 /* Add additional parts. */
543 SvCUR_set(data->last_found,
544 SvCUR(data->last_found) - l);
545 sv_catsv(data->last_found, last_str);
546 data->last_end += l * (mincount - 1);
547 }
548 }
549 }
550 /* It is counted once already... */
551 data->pos_min += minnext * (mincount - counted);
552 data->pos_delta += - counted * deltanext +
553 (minnext + deltanext) * maxcount - minnext * mincount;
554 if (mincount != maxcount) {
555 scan_commit(data);
556 if (mincount && last_str) {
557 sv_setsv(data->last_found, last_str);
558 data->last_end = data->pos_min;
559 data->last_start_min =
560 data->pos_min - SvCUR(last_str);
561 data->last_start_max = is_inf
562 ? I32_MAX
563 : data->pos_min + data->pos_delta
564 - SvCUR(last_str);
565 }
566 data->longest = &(data->longest_float);
567 }
568 }
405ff068 569 if (data && (fl & SF_HAS_EVAL))
c277df42
IZ
570 data->flags |= SF_HAS_EVAL;
571 optimize_curly_tail:
c277df42
IZ
572 if (OP(oscan) != CURLYX) {
573 while (regkind[(U8)OP(next = regnext(oscan))] == NOTHING
574 && NEXT_OFF(next))
575 NEXT_OFF(oscan) += NEXT_OFF(next);
576 }
c277df42
IZ
577 continue;
578 default: /* REF only? */
579 if (flags & SCF_DO_SUBSTR) {
580 scan_commit(data);
581 data->longest = &(data->longest_float);
582 }
583 is_inf = 1;
584 break;
585 }
586 } else if (strchr(simple,OP(scan))) {
587 if (flags & SCF_DO_SUBSTR) {
588 scan_commit(data);
589 data->pos_min++;
590 }
591 min++;
592 } else if (regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
593 data->flags |= (OP(scan) == MEOL
594 ? SF_BEFORE_MEOL
595 : SF_BEFORE_SEOL);
596 } else if (regkind[(U8)OP(scan)] == BRANCHJ
597 && (scan->flags || data)
598 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
599 I32 deltanext, minnext;
600 regnode *nscan;
601
602 data_fake.flags = 0;
603 next = regnext(scan);
604 nscan = NEXTOPER(NEXTOPER(scan));
605 minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0);
606 if (scan->flags) {
607 if (deltanext) {
608 FAIL("variable length lookbehind not implemented");
609 } else if (minnext > U8_MAX) {
610 FAIL2("lookbehind longer than %d not implemented", U8_MAX);
611 }
612 scan->flags = minnext;
613 }
614 if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
615 pars++;
405ff068 616 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42
IZ
617 data->flags |= SF_HAS_EVAL;
618 } else if (OP(scan) == OPEN) {
619 pars++;
620 } else if (OP(scan) == CLOSE && ARG(scan) == is_par) {
c277df42
IZ
621 next = regnext(scan);
622
623 if ( next && (OP(next) != WHILEM) && next < last)
c277df42
IZ
624 is_par = 0; /* Disable optimization */
625 } else if (OP(scan) == EVAL) {
626 if (data)
627 data->flags |= SF_HAS_EVAL;
628 }
629 /* Else: zero-length, ignore. */
630 scan = regnext(scan);
631 }
632
633 finish:
634 *scanp = scan;
635 *deltap = is_inf ? I32_MAX : delta;
636 if (flags & SCF_DO_SUBSTR && is_inf)
637 data->pos_delta = I32_MAX - data->pos_min;
638 if (is_par > U8_MAX)
639 is_par = 0;
640 if (is_par && pars==1 && data) {
641 data->flags |= SF_IN_PAR;
642 data->flags &= ~SF_HAS_PAR;
643 } else if (pars && data) {
644 data->flags |= SF_HAS_PAR;
645 data->flags &= ~SF_IN_PAR;
646 }
647 return min;
648}
649
76e3520e 650STATIC I32
c277df42
IZ
651add_data(I32 n, char *s)
652{
77d41b28
GS
653 if (regcomp_rx->data) {
654 Renewc(regcomp_rx->data,
655 sizeof(*regcomp_rx->data) + sizeof(void*) * (regcomp_rx->data->count + n - 1),
c277df42 656 char, struct reg_data);
77d41b28
GS
657 Renew(regcomp_rx->data->what, regcomp_rx->data->count + n, U8);
658 regcomp_rx->data->count += n;
c277df42 659 } else {
77d41b28 660 Newc(1207, regcomp_rx->data, sizeof(*regcomp_rx->data) + sizeof(void*) * (n - 1),
c277df42 661 char, struct reg_data);
77d41b28
GS
662 New(1208, regcomp_rx->data->what, n, U8);
663 regcomp_rx->data->count = n;
c277df42 664 }
77d41b28
GS
665 Copy(s, regcomp_rx->data->what + regcomp_rx->data->count - n, n, U8);
666 return regcomp_rx->data->count - n;
c277df42
IZ
667}
668
a687059c 669/*
e50aee73 670 - pregcomp - compile a regular expression into internal code
a687059c
LW
671 *
672 * We can't allocate space until we know how big the compiled form will be,
673 * but we can't compile it (and thus know how big it is) until we've got a
674 * place to put the code. So we cheat: we compile it twice, once with code
675 * generation turned off and size counting turned on, and once "for real".
676 * This also means that we don't allocate space until we are sure that the
677 * thing really will compile successfully, and we never have to move the
678 * code and thus invalidate pointers into it. (Note that it has to be in
679 * one piece because free() must be able to free it all.) [NB: not true in perl]
680 *
681 * Beware that the optimization-preparation code in here knows about some
682 * of the structure of the compiled regexp. [I'll say.]
683 */
684regexp *
8ac85365 685pregcomp(char *exp, char *xend, PMOP *pm)
a687059c 686{
a0d0e21e 687 register regexp *r;
c277df42
IZ
688 regnode *scan;
689 SV **longest;
690 SV *longest_fixed;
691 SV *longest_float;
692 regnode *first;
a0d0e21e 693 I32 flags;
a0d0e21e
LW
694 I32 minlen = 0;
695 I32 sawplus = 0;
696 I32 sawopen = 0;
697
698 if (exp == NULL)
c277df42 699 FAIL("NULL regexp argument");
a0d0e21e 700
bbce6d69 701 regprecomp = savepvn(exp, xend - exp);
c277df42
IZ
702 DEBUG_r(PerlIO_printf(Perl_debug_log, "compiling RE `%*s'\n",
703 xend - exp, regprecomp));
a0d0e21e 704 regflags = pm->op_pmflags;
bbce6d69
PP
705 regsawback = 0;
706
c277df42
IZ
707 regseen = 0;
708 seen_zerolen = *exp == '^' ? -1 : 0;
709 extralen = 0;
710
bbce6d69 711 /* First pass: determine size, legality. */
77d41b28 712 regcomp_parse = exp;
a0d0e21e 713 regxend = xend;
a0d0e21e 714 regnaughty = 0;
a0d0e21e
LW
715 regnpar = 1;
716 regsize = 0L;
717 regcode = &regdummy;
c277df42 718 regc((U8)MAGIC, (char*)regcode);
a0d0e21e
LW
719 if (reg(0, &flags) == NULL) {
720 Safefree(regprecomp);
721 regprecomp = Nullch;
722 return(NULL);
723 }
c277df42
IZ
724 DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", regsize));
725
726 DEBUG_r(
727 if (!colorset) {
728 int i = 0;
76e3520e 729 char *s = PerlEnv_getenv("TERMCAP_COLORS");
c277df42
IZ
730
731 colorset = 1;
732 if (s) {
733 colors[0] = s = savepv(s);
734 while (++i < 4) {
735 s = strchr(s, '\t');
736 if (!s)
737 FAIL("Not enough TABs in TERMCAP_COLORS");
738 *s = '\0';
739 colors[i] = ++s;
740 }
741 } else {
742 while (i < 4)
743 colors[i++] = "";
744 }
745 /* Reset colors: */
746 PerlIO_printf(Perl_debug_log, "%s%s%s%s",
747 colors[0],colors[1],colors[2],colors[3]);
748 }
749 );
a0d0e21e 750
c277df42
IZ
751 /* Small enough for pointer-storage convention?
752 If extralen==0, this means that we will not need long jumps. */
c277df42
IZ
753 if (regsize >= 0x10000L && extralen)
754 regsize += extralen;
755 else
756 extralen = 0;
a0d0e21e 757
bbce6d69 758 /* Allocate space and initialize. */
c277df42
IZ
759 Newc(1001, r, sizeof(regexp) + (unsigned)regsize * sizeof(regnode),
760 char, regexp);
a0d0e21e
LW
761 if (r == NULL)
762 FAIL("regexp out of space");
c277df42 763 r->refcnt = 1;
bbce6d69 764 r->prelen = xend - exp;
a0d0e21e
LW
765 r->precomp = regprecomp;
766 r->subbeg = r->subbase = NULL;
77d41b28 767 regcomp_rx = r;
bbce6d69
PP
768
769 /* Second pass: emit code. */
77d41b28 770 regcomp_parse = exp;
bbce6d69
PP
771 regxend = xend;
772 regnaughty = 0;
a0d0e21e
LW
773 regnpar = 1;
774 regcode = r->program;
c277df42
IZ
775 regc((U8)MAGIC, (char*) regcode++);
776 r->data = 0;
a0d0e21e
LW
777 if (reg(0, &flags) == NULL)
778 return(NULL);
779
780 /* Dig out information for optimizations. */
781 pm->op_pmflags = regflags;
a0d0e21e 782 r->reganch = 0;
c277df42 783 r->regstclass = NULL;
a0d0e21e 784 r->naughty = regnaughty >= 10; /* Probably an expensive pattern. */
c277df42 785 scan = r->program + 1; /* First BRANCH. */
2779dcf1
IZ
786
787 /* XXXX To minimize changes to RE engine we always allocate
788 3-units-long substrs field. */
789 Newz(1004, r->substrs, 1, struct reg_substr_data);
790
c277df42
IZ
791 if (OP(scan) != BRANCH) { /* Only one top-level choice. */
792 scan_data_t data;
793 I32 fake;
c5254dd6 794 STRLEN longest_float_length, longest_fixed_length;
a0d0e21e 795
c277df42 796 StructCopy(&zero_scan_data, &data, scan_data_t);
a0d0e21e 797 first = scan;
c277df42 798 /* Skip introductions and multiplicators >= 1. */
a0d0e21e
LW
799 while ((OP(first) == OPEN && (sawopen = 1)) ||
800 (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
801 (OP(first) == PLUS) ||
802 (OP(first) == MINMOD) ||
803 (regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
804 if (OP(first) == PLUS)
805 sawplus = 1;
806 else
807 first += regarglen[(U8)OP(first)];
808 first = NEXTOPER(first);
a687059c
LW
809 }
810
a0d0e21e
LW
811 /* Starting-point info. */
812 again:
c277df42 813 if (OP(first) == EXACT); /* Empty, get anchored substr later. */
a0d0e21e
LW
814 else if (strchr(simple+2,OP(first)))
815 r->regstclass = first;
bbce6d69
PP
816 else if (regkind[(U8)OP(first)] == BOUND ||
817 regkind[(U8)OP(first)] == NBOUND)
a0d0e21e
LW
818 r->regstclass = first;
819 else if (regkind[(U8)OP(first)] == BOL) {
c277df42 820 r->reganch |= (OP(first) == MBOL ? ROPT_ANCH_MBOL: ROPT_ANCH_BOL);
a0d0e21e 821 first = NEXTOPER(first);
774d564b
PP
822 goto again;
823 }
824 else if (OP(first) == GPOS) {
825 r->reganch |= ROPT_ANCH_GPOS;
826 first = NEXTOPER(first);
827 goto again;
a0d0e21e
LW
828 }
829 else if ((OP(first) == STAR &&
830 regkind[(U8)OP(NEXTOPER(first))] == ANY) &&
831 !(r->reganch & ROPT_ANCH) )
832 {
833 /* turn .* into ^.* with an implied $*=1 */
774d564b 834 r->reganch |= ROPT_ANCH_BOL | ROPT_IMPLICIT;
a0d0e21e 835 first = NEXTOPER(first);
774d564b 836 goto again;
a0d0e21e
LW
837 }
838 if (sawplus && (!sawopen || !regsawback))
839 r->reganch |= ROPT_SKIP; /* x+ must match 1st of run */
840
c277df42
IZ
841 /* Scan is after the zeroth branch, first is atomic matcher. */
842 DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n",
843 first - scan + 1));
a0d0e21e
LW
844 /*
845 * If there's something expensive in the r.e., find the
846 * longest literal string that must appear and make it the
847 * regmust. Resolve ties in favor of later strings, since
848 * the regstart check works with the beginning of the r.e.
849 * and avoiding duplication strengthens checking. Not a
850 * strong reason, but sufficient in the absence of others.
851 * [Now we resolve ties in favor of the earlier string if
c277df42 852 * it happens that c_offset_min has been invalidated, since the
a0d0e21e
LW
853 * earlier string may buy us something the later one won't.]
854 */
a0d0e21e 855 minlen = 0;
a687059c 856
c277df42
IZ
857 data.longest_fixed = newSVpv("",0);
858 data.longest_float = newSVpv("",0);
859 data.last_found = newSVpv("",0);
860 data.longest = &(data.longest_fixed);
861 first = scan;
862
863 minlen = study_chunk(&first, &fake, scan + regsize, /* Up to end */
864 &data, SCF_DO_SUBSTR);
865 if ( regnpar == 1 && data.longest == &(data.longest_fixed)
866 && data.last_start_min == 0 && data.last_end > 0
867 && !seen_zerolen
868 && (!(regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
869 r->reganch |= ROPT_CHECK_ALL;
870 scan_commit(&data);
871 SvREFCNT_dec(data.last_found);
872
c5254dd6
MB
873 longest_float_length = SvCUR(data.longest_float);
874 if (longest_float_length
c277df42
IZ
875 || (data.flags & SF_FL_BEFORE_EOL
876 && (!(data.flags & SF_FL_BEFORE_MEOL)
877 || (regflags & PMf_MULTILINE)))) {
878 if (SvCUR(data.longest_fixed)
879 && data.offset_fixed == data.offset_float_min)
880 goto remove; /* Like in (a)+. */
881
882 r->float_substr = data.longest_float;
883 r->float_min_offset = data.offset_float_min;
884 r->float_max_offset = data.offset_float_max;
2779dcf1 885 fbm_compile(r->float_substr, 0);
c277df42
IZ
886 BmUSEFUL(r->float_substr) = 100;
887 if (data.flags & SF_FL_BEFORE_EOL /* Cannot have SEOL and MULTI */
888 && (!(data.flags & SF_FL_BEFORE_MEOL)
889 || (regflags & PMf_MULTILINE)))
890 SvTAIL_on(r->float_substr);
891 } else {
892 remove:
893 r->float_substr = Nullsv;
894 SvREFCNT_dec(data.longest_float);
c5254dd6 895 longest_float_length = 0;
a0d0e21e 896 }
c277df42 897
c5254dd6
MB
898 longest_fixed_length = SvCUR(data.longest_fixed);
899 if (longest_fixed_length
c277df42
IZ
900 || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
901 && (!(data.flags & SF_FIX_BEFORE_MEOL)
902 || (regflags & PMf_MULTILINE)))) {
903 r->anchored_substr = data.longest_fixed;
904 r->anchored_offset = data.offset_fixed;
2779dcf1 905 fbm_compile(r->anchored_substr, 0);
c277df42
IZ
906 BmUSEFUL(r->anchored_substr) = 100;
907 if (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
908 && (!(data.flags & SF_FIX_BEFORE_MEOL)
909 || (regflags & PMf_MULTILINE)))
910 SvTAIL_on(r->anchored_substr);
911 } else {
912 r->anchored_substr = Nullsv;
913 SvREFCNT_dec(data.longest_fixed);
c5254dd6 914 longest_fixed_length = 0;
a0d0e21e 915 }
c277df42
IZ
916
917 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
c5254dd6 918 if (longest_fixed_length > longest_float_length) {
c277df42
IZ
919 r->check_substr = r->anchored_substr;
920 r->check_offset_min = r->check_offset_max = r->anchored_offset;
921 if (r->reganch & ROPT_ANCH_SINGLE)
922 r->reganch |= ROPT_NOSCAN;
923 } else {
924 r->check_substr = r->float_substr;
925 r->check_offset_min = data.offset_float_min;
926 r->check_offset_max = data.offset_float_max;
a0d0e21e 927 }
c277df42
IZ
928 } else {
929 /* Several toplevels. Best we can is to set minlen. */
930 I32 fake;
931
932 DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
933 scan = r->program + 1;
934 minlen = study_chunk(&scan, &fake, scan + regsize, NULL, 0);
935 r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
a0d0e21e
LW
936 }
937
a0d0e21e
LW
938 r->nparens = regnpar - 1;
939 r->minlen = minlen;
c277df42
IZ
940 if (regseen & REG_SEEN_GPOS)
941 r->reganch |= ROPT_GPOS_SEEN;
942 if (regseen & REG_SEEN_LOOKBEHIND)
943 r->reganch |= ROPT_LOOKBEHIND_SEEN;
ce862d02
IZ
944 if (regseen & REG_SEEN_EVAL)
945 r->reganch |= ROPT_EVAL_SEEN;
a0d0e21e
LW
946 Newz(1002, r->startp, regnpar, char*);
947 Newz(1002, r->endp, regnpar, char*);
948 DEBUG_r(regdump(r));
949 return(r);
a687059c
LW
950}
951
952/*
953 - reg - regular expression, i.e. main body or parenthesized thing
954 *
955 * Caller must absorb opening parenthesis.
956 *
957 * Combining parenthesis handling with the base level of regular expression
958 * is a trifle forced, but the need to tie the tails of the branches to what
959 * follows makes it hard to avoid.
960 */
76e3520e 961STATIC regnode *
8ac85365 962reg(I32 paren, I32 *flagp)
c277df42 963 /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
a687059c 964{
c277df42
IZ
965 register regnode *ret; /* Will be the head of the group. */
966 register regnode *br;
967 register regnode *lastbr;
968 register regnode *ender = 0;
a0d0e21e 969 register I32 parno = 0;
c277df42
IZ
970 I32 flags, oregflags = regflags, have_branch = 0, open = 0;
971 char c;
a0d0e21e
LW
972
973 *flagp = HASWIDTH; /* Tentatively. */
974
975 /* Make an OPEN node, if parenthesized. */
976 if (paren) {
77d41b28
GS
977 if (*regcomp_parse == '?') {
978 regcomp_parse++;
979 paren = *regcomp_parse++;
c277df42 980 ret = NULL; /* For look-ahead/behind. */
a0d0e21e 981 switch (paren) {
c277df42 982 case '<':
c277df42 983 regseen |= REG_SEEN_LOOKBEHIND;
77d41b28 984 if (*regcomp_parse == '!')
c277df42 985 paren = ',';
77d41b28 986 if (*regcomp_parse != '=' && *regcomp_parse != '!')
c277df42 987 goto unknown;
77d41b28 988 regcomp_parse++;
a0d0e21e
LW
989 case '=':
990 case '!':
c277df42
IZ
991 seen_zerolen++;
992 case ':':
993 case '>':
a0d0e21e
LW
994 break;
995 case '$':
996 case '@':
c277df42 997 FAIL2("Sequence (?%c...) not implemented", (int)paren);
a0d0e21e
LW
998 break;
999 case '#':
77d41b28
GS
1000 while (*regcomp_parse && *regcomp_parse != ')')
1001 regcomp_parse++;
1002 if (*regcomp_parse != ')')
c277df42 1003 FAIL("Sequence (?#... not terminated");
a0d0e21e
LW
1004 nextchar();
1005 *flagp = TRYAGAIN;
1006 return NULL;
c277df42
IZ
1007 case '{':
1008 {
1009 dTHR;
1010 I32 count = 1, n = 0;
1011 char c;
77d41b28 1012 char *s = regcomp_parse;
c277df42
IZ
1013 SV *sv;
1014 OP_4tree *sop, *rop;
1015
1016 seen_zerolen++;
ce862d02 1017 regseen |= REG_SEEN_EVAL;
77d41b28
GS
1018 while (count && (c = *regcomp_parse)) {
1019 if (c == '\\' && regcomp_parse[1])
1020 regcomp_parse++;
c277df42
IZ
1021 else if (c == '{')
1022 count++;
1023 else if (c == '}')
1024 count--;
77d41b28 1025 regcomp_parse++;
c277df42 1026 }
77d41b28 1027 if (*regcomp_parse != ')')
c277df42
IZ
1028 FAIL("Sequence (?{...}) not terminated or not {}-balanced");
1029 if (!SIZE_ONLY) {
1030 AV *av;
1031
77d41b28
GS
1032 if (regcomp_parse - 1 - s)
1033 sv = newSVpv(s, regcomp_parse - 1 - s);
c277df42
IZ
1034 else
1035 sv = newSVpv("", 0);
1036
1037 rop = sv_compile_2op(sv, &sop, "re", &av);
1038
1039 n = add_data(3, "nso");
77d41b28
GS
1040 regcomp_rx->data->data[n] = (void*)rop;
1041 regcomp_rx->data->data[n+1] = (void*)av;
1042 regcomp_rx->data->data[n+2] = (void*)sop;
c277df42 1043 SvREFCNT_dec(sv);
cc6b7395
IZ
1044 } else { /* First pass */
1045 if (tainted)
1046 FAIL("Eval-group in insecure regular expression");
c277df42
IZ
1047 }
1048
1049 nextchar();
c277df42
IZ
1050 return reganode(EVAL, n);
1051 }
1052 case '(':
1053 {
77d41b28
GS
1054 if (regcomp_parse[0] == '?') {
1055 if (regcomp_parse[1] == '=' || regcomp_parse[1] == '!'
1056 || regcomp_parse[1] == '<'
1057 || regcomp_parse[1] == '{') { /* Lookahead or eval. */
c277df42
IZ
1058 I32 flag;
1059
1060 ret = reg_node(LOGICAL);
1061 regtail(ret, reg(1, &flag));
1062 goto insert_if;
1063 }
77d41b28
GS
1064 } else if (regcomp_parse[0] >= '1' && regcomp_parse[0] <= '9' ) {
1065 parno = atoi(regcomp_parse++);
c277df42 1066
77d41b28
GS
1067 while (isDIGIT(*regcomp_parse))
1068 regcomp_parse++;
c277df42
IZ
1069 ret = reganode(GROUPP, parno);
1070 if ((c = *nextchar()) != ')')
1071 FAIL2("Switch (?(number%c not recognized", c);
1072 insert_if:
1073 regtail(ret, reganode(IFTHEN, 0));
1074 br = regbranch(&flags, 1);
1075 if (br == NULL)
1076 br = reganode(LONGJMP, 0);
1077 else
1078 regtail(br, reganode(LONGJMP, 0));
1079 c = *nextchar();
1080 if (c == '|') {
1081 lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */
1082 regbranch(&flags, 1);
1083 regtail(ret, lastbr);
1084 c = *nextchar();
1085 } else
1086 lastbr = NULL;
1087 if (c != ')')
1088 FAIL("Switch (?(condition)... contains too many branches");
1089 ender = reg_node(TAIL);
1090 regtail(br, ender);
1091 if (lastbr) {
1092 regtail(lastbr, ender);
1093 regtail(NEXTOPER(NEXTOPER(lastbr)), ender);
1094 } else
1095 regtail(ret, ender);
1096 return ret;
1097 } else {
77d41b28 1098 FAIL2("Unknown condition for (?(%.2s", regcomp_parse);
c277df42
IZ
1099 }
1100 }
1b1626e4 1101 case 0:
c277df42 1102 FAIL("Sequence (? incomplete");
1b1626e4 1103 break;
a0d0e21e 1104 default:
77d41b28
GS
1105 --regcomp_parse;
1106 while (*regcomp_parse && strchr("iogcmsx", *regcomp_parse)) {
1107 if (*regcomp_parse != 'o')
1108 pmflag(&regflags, *regcomp_parse);
1109 ++regcomp_parse;
48c036b1 1110 }
c277df42 1111 unknown:
77d41b28
GS
1112 if (*regcomp_parse != ')')
1113 FAIL2("Sequence (?%c...) not recognized", *regcomp_parse);
a0d0e21e
LW
1114 nextchar();
1115 *flagp = TRYAGAIN;
1116 return NULL;
1117 }
1118 }
1119 else {
1120 parno = regnpar;
1121 regnpar++;
1122 ret = reganode(OPEN, parno);
c277df42 1123 open = 1;
a0d0e21e
LW
1124 }
1125 } else
1126 ret = NULL;
1127
1128 /* Pick up the branches, linking them together. */
c277df42 1129 br = regbranch(&flags, 1);
a0d0e21e
LW
1130 if (br == NULL)
1131 return(NULL);
77d41b28 1132 if (*regcomp_parse == '|') {
c277df42
IZ
1133 if (!SIZE_ONLY && extralen) {
1134 reginsert(BRANCHJ, br);
1135 } else
1136 reginsert(BRANCH, br);
1137 have_branch = 1;
1138 if (SIZE_ONLY)
1139 extralen += 1; /* For BRANCHJ-BRANCH. */
1140 } else if (paren == ':') {
1141 *flagp |= flags&SIMPLE;
1142 }
1143 if (open) { /* Starts with OPEN. */
1144 regtail(ret, br); /* OPEN -> first. */
1145 } else if (paren != '?') /* Not Conditional */
a0d0e21e
LW
1146 ret = br;
1147 if (!(flags&HASWIDTH))
1148 *flagp &= ~HASWIDTH;
1149 *flagp |= flags&SPSTART;
c277df42 1150 lastbr = br;
77d41b28 1151 while (*regcomp_parse == '|') {
c277df42
IZ
1152 if (!SIZE_ONLY && extralen) {
1153 ender = reganode(LONGJMP,0);
1154 regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
1155 }
1156 if (SIZE_ONLY)
1157 extralen += 2; /* Account for LONGJMP. */
a0d0e21e 1158 nextchar();
c277df42 1159 br = regbranch(&flags, 0);
a687059c 1160 if (br == NULL)
a0d0e21e 1161 return(NULL);
c277df42
IZ
1162 regtail(lastbr, br); /* BRANCH -> BRANCH. */
1163 lastbr = br;
a687059c 1164 if (!(flags&HASWIDTH))
a0d0e21e 1165 *flagp &= ~HASWIDTH;
a687059c 1166 *flagp |= flags&SPSTART;
a0d0e21e
LW
1167 }
1168
c277df42
IZ
1169 if (have_branch || paren != ':') {
1170 /* Make a closing node, and hook it on the end. */
1171 switch (paren) {
1172 case ':':
1173 ender = reg_node(TAIL);
1174 break;
1175 case 1:
1176 ender = reganode(CLOSE, parno);
1177 break;
1178 case '<':
1179 case '>':
1180 case ',':
1181 case '=':
1182 case '!':
1183 ender = reg_node(SUCCEED);
1184 *flagp &= ~HASWIDTH;
1185 break;
1186 case 0:
1187 ender = reg_node(END);
1188 break;
1189 }
1190 regtail(lastbr, ender);
a0d0e21e 1191
c277df42
IZ
1192 if (have_branch) {
1193 /* Hook the tails of the branches to the closing node. */
1194 for (br = ret; br != NULL; br = regnext(br)) {
1195 regoptail(br, ender);
1196 }
1197 }
a0d0e21e 1198 }
c277df42
IZ
1199
1200 {
1201 char *p;
1202 static char parens[] = "=!<,>";
1203
1204 if (paren && (p = strchr(parens, paren))) {
1205 int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
1206 int flag = (p - parens) > 1;
1207
1208 if (paren == '>')
1209 node = SUSPEND, flag = 0;
1210 reginsert(node,ret);
c277df42 1211 ret->flags = flag;
c277df42
IZ
1212 regtail(ret, reg_node(TAIL));
1213 }
a0d0e21e
LW
1214 }
1215
1216 /* Check for proper termination. */
77d41b28 1217 if (paren && (regcomp_parse >= regxend || *nextchar() != ')')) {
a0d0e21e 1218 FAIL("unmatched () in regexp");
77d41b28
GS
1219 } else if (!paren && regcomp_parse < regxend) {
1220 if (*regcomp_parse == ')') {
a0d0e21e
LW
1221 FAIL("unmatched () in regexp");
1222 } else
1223 FAIL("junk on end of regexp"); /* "Can't happen". */
1224 /* NOTREACHED */
1225 }
c277df42
IZ
1226 if (paren != 0) {
1227 regflags = oregflags;
1228 }
a687059c 1229
a0d0e21e 1230 return(ret);
a687059c
LW
1231}
1232
1233/*
1234 - regbranch - one alternative of an | operator
1235 *
1236 * Implements the concatenation operator.
1237 */
76e3520e 1238STATIC regnode *
c277df42 1239regbranch(I32 *flagp, I32 first)
a687059c 1240{
c277df42
IZ
1241 register regnode *ret;
1242 register regnode *chain = NULL;
1243 register regnode *latest;
1244 I32 flags = 0, c = 0;
a0d0e21e 1245
c277df42
IZ
1246 if (first)
1247 ret = NULL;
1248 else {
1249 if (!SIZE_ONLY && extralen)
1250 ret = reganode(BRANCHJ,0);
1251 else
1252 ret = reg_node(BRANCH);
1253 }
1254
1255 if (!first && SIZE_ONLY)
1256 extralen += 1; /* BRANCHJ */
1257
1258 *flagp = WORST; /* Tentatively. */
a0d0e21e 1259
77d41b28 1260 regcomp_parse--;
a0d0e21e 1261 nextchar();
77d41b28 1262 while (regcomp_parse < regxend && *regcomp_parse != '|' && *regcomp_parse != ')') {
a0d0e21e
LW
1263 flags &= ~TRYAGAIN;
1264 latest = regpiece(&flags);
1265 if (latest == NULL) {
1266 if (flags & TRYAGAIN)
1267 continue;
1268 return(NULL);
c277df42
IZ
1269 } else if (ret == NULL)
1270 ret = latest;
a0d0e21e 1271 *flagp |= flags&HASWIDTH;
c277df42 1272 if (chain == NULL) /* First piece. */
a0d0e21e
LW
1273 *flagp |= flags&SPSTART;
1274 else {
1275 regnaughty++;
1276 regtail(chain, latest);
a687059c 1277 }
a0d0e21e 1278 chain = latest;
c277df42
IZ
1279 c++;
1280 }
1281 if (chain == NULL) { /* Loop ran zero times. */
1282 chain = reg_node(NOTHING);
1283 if (ret == NULL)
1284 ret = chain;
1285 }
1286 if (c == 1) {
1287 *flagp |= flags&SIMPLE;
a0d0e21e 1288 }
a687059c 1289
a0d0e21e 1290 return(ret);
a687059c
LW
1291}
1292
1293/*
1294 - regpiece - something followed by possible [*+?]
1295 *
1296 * Note that the branching code sequences used for ? and the general cases
1297 * of * and + are somewhat optimized: they use the same NOTHING node as
1298 * both the endmarker for their branch list and the body of the last branch.
1299 * It might seem that this node could be dispensed with entirely, but the
1300 * endmarker role is not redundant.
1301 */
76e3520e 1302STATIC regnode *
8ac85365 1303regpiece(I32 *flagp)
a687059c 1304{
c277df42 1305 register regnode *ret;
a0d0e21e
LW
1306 register char op;
1307 register char *next;
1308 I32 flags;
77d41b28 1309 char *origparse = regcomp_parse;
a0d0e21e
LW
1310 char *maxpos;
1311 I32 min;
c277df42 1312 I32 max = REG_INFTY;
a0d0e21e
LW
1313
1314 ret = regatom(&flags);
1315 if (ret == NULL) {
1316 if (flags & TRYAGAIN)
1317 *flagp |= TRYAGAIN;
1318 return(NULL);
1319 }
1320
77d41b28 1321 op = *regcomp_parse;
a0d0e21e 1322
77d41b28
GS
1323 if (op == '{' && regcurly(regcomp_parse)) {
1324 next = regcomp_parse + 1;
a0d0e21e
LW
1325 maxpos = Nullch;
1326 while (isDIGIT(*next) || *next == ',') {
1327 if (*next == ',') {
1328 if (maxpos)
1329 break;
1330 else
1331 maxpos = next;
a687059c 1332 }
a0d0e21e
LW
1333 next++;
1334 }
1335 if (*next == '}') { /* got one */
1336 if (!maxpos)
1337 maxpos = next;
77d41b28
GS
1338 regcomp_parse++;
1339 min = atoi(regcomp_parse);
a0d0e21e
LW
1340 if (*maxpos == ',')
1341 maxpos++;
1342 else
77d41b28 1343 maxpos = regcomp_parse;
a0d0e21e
LW
1344 max = atoi(maxpos);
1345 if (!max && *maxpos != '0')
c277df42
IZ
1346 max = REG_INFTY; /* meaning "infinity" */
1347 else if (max >= REG_INFTY)
1348 FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
77d41b28 1349 regcomp_parse = next;
a0d0e21e
LW
1350 nextchar();
1351
1352 do_curly:
1353 if ((flags&SIMPLE)) {
1354 regnaughty += 2 + regnaughty / 2;
1355 reginsert(CURLY, ret);
1356 }
1357 else {
1358 regnaughty += 4 + regnaughty; /* compound interest */
c277df42
IZ
1359 regtail(ret, reg_node(WHILEM));
1360 if (!SIZE_ONLY && extralen) {
1361 reginsert(LONGJMP,ret);
1362 reginsert(NOTHING,ret);
1363 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
1364 }
a0d0e21e 1365 reginsert(CURLYX,ret);
c277df42
IZ
1366 if (!SIZE_ONLY && extralen)
1367 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
1368 regtail(ret, reg_node(NOTHING));
1369 if (SIZE_ONLY)
1370 extralen += 3;
a0d0e21e 1371 }
c277df42 1372 ret->flags = 0;
a0d0e21e
LW
1373
1374 if (min > 0)
1375 *flagp = (WORST|HASWIDTH);
1376 if (max && max < min)
c277df42
IZ
1377 FAIL("Can't do {n,m} with n > m");
1378 if (!SIZE_ONLY) {
1379 ARG1_SET(ret, min);
1380 ARG2_SET(ret, max);
a687059c 1381 }
a687059c 1382
a0d0e21e 1383 goto nest_check;
a687059c 1384 }
a0d0e21e 1385 }
a687059c 1386
a0d0e21e
LW
1387 if (!ISMULT1(op)) {
1388 *flagp = flags;
a687059c 1389 return(ret);
a0d0e21e 1390 }
bb20fd44 1391
c277df42 1392#if 0 /* Now runtime fix should be reliable. */
bb20fd44 1393 if (!(flags&HASWIDTH) && op != '?')
c277df42
IZ
1394 FAIL("regexp *+ operand could be empty");
1395#endif
bb20fd44 1396
a0d0e21e
LW
1397 nextchar();
1398
1399 *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
1400
1401 if (op == '*' && (flags&SIMPLE)) {
1402 reginsert(STAR, ret);
c277df42 1403 ret->flags = 0;
a0d0e21e
LW
1404 regnaughty += 4;
1405 }
1406 else if (op == '*') {
1407 min = 0;
1408 goto do_curly;
1409 } else if (op == '+' && (flags&SIMPLE)) {
1410 reginsert(PLUS, ret);
c277df42 1411 ret->flags = 0;
a0d0e21e
LW
1412 regnaughty += 3;
1413 }
1414 else if (op == '+') {
1415 min = 1;
1416 goto do_curly;
1417 } else if (op == '?') {
1418 min = 0; max = 1;
1419 goto do_curly;
1420 }
1421 nest_check:
c277df42 1422 if (dowarn && !SIZE_ONLY && !(flags&HASWIDTH) && max > 10000) {
a0d0e21e 1423 warn("%.*s matches null string many times",
77d41b28 1424 regcomp_parse - origparse, origparse);
a0d0e21e
LW
1425 }
1426
77d41b28 1427 if (*regcomp_parse == '?') {
a0d0e21e
LW
1428 nextchar();
1429 reginsert(MINMOD, ret);
c277df42 1430 regtail(ret, ret + NODE_STEP_REGNODE);
a0d0e21e 1431 }
77d41b28 1432 if (ISMULT2(regcomp_parse))
a0d0e21e
LW
1433 FAIL("nested *?+ in regexp");
1434
1435 return(ret);
a687059c
LW
1436}
1437
1438/*
1439 - regatom - the lowest level
1440 *
1441 * Optimization: gobbles an entire sequence of ordinary characters so that
1442 * it can turn them into a single node, which is smaller to store and
1443 * faster to run. Backslashed characters are exceptions, each becoming a
1444 * separate node; the code is simpler that way and it's not worth fixing.
1445 *
1446 * [Yes, it is worth fixing, some scripts can run twice the speed.]
1447 */
76e3520e 1448STATIC regnode *
8ac85365 1449regatom(I32 *flagp)
a687059c 1450{
c277df42 1451 register regnode *ret = 0;
a0d0e21e
LW
1452 I32 flags;
1453
1454 *flagp = WORST; /* Tentatively. */
1455
1456tryagain:
77d41b28 1457 switch (*regcomp_parse) {
a0d0e21e 1458 case '^':
c277df42 1459 seen_zerolen++;
a0d0e21e
LW
1460 nextchar();
1461 if (regflags & PMf_MULTILINE)
c277df42 1462 ret = reg_node(MBOL);
a0d0e21e 1463 else if (regflags & PMf_SINGLELINE)
c277df42 1464 ret = reg_node(SBOL);
a0d0e21e 1465 else
c277df42 1466 ret = reg_node(BOL);
a0d0e21e
LW
1467 break;
1468 case '$':
77d41b28 1469 if (regcomp_parse[1])
c277df42 1470 seen_zerolen++;
a0d0e21e
LW
1471 nextchar();
1472 if (regflags & PMf_MULTILINE)
c277df42 1473 ret = reg_node(MEOL);
a0d0e21e 1474 else if (regflags & PMf_SINGLELINE)
c277df42 1475 ret = reg_node(SEOL);
a0d0e21e 1476 else
c277df42 1477 ret = reg_node(EOL);
a0d0e21e
LW
1478 break;
1479 case '.':
1480 nextchar();
1481 if (regflags & PMf_SINGLELINE)
c277df42 1482 ret = reg_node(SANY);
a0d0e21e 1483 else
c277df42 1484 ret = reg_node(ANY);
a0d0e21e
LW
1485 regnaughty++;
1486 *flagp |= HASWIDTH|SIMPLE;
1487 break;
1488 case '[':
77d41b28 1489 regcomp_parse++;
a0d0e21e
LW
1490 ret = regclass();
1491 *flagp |= HASWIDTH|SIMPLE;
1492 break;
1493 case '(':
1494 nextchar();
1495 ret = reg(1, &flags);
1496 if (ret == NULL) {
1497 if (flags & TRYAGAIN)
1498 goto tryagain;
1499 return(NULL);
1500 }
c277df42 1501 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
a0d0e21e
LW
1502 break;
1503 case '|':
1504 case ')':
1505 if (flags & TRYAGAIN) {
1506 *flagp |= TRYAGAIN;
1507 return NULL;
1508 }
77d41b28 1509 FAIL2("internal urp in regexp at /%s/", regcomp_parse);
a0d0e21e
LW
1510 /* Supposed to be caught earlier. */
1511 break;
85afd4ae 1512 case '{':
77d41b28
GS
1513 if (!regcurly(regcomp_parse)) {
1514 regcomp_parse++;
85afd4ae
CS
1515 goto defchar;
1516 }
1517 /* FALL THROUGH */
a0d0e21e
LW
1518 case '?':
1519 case '+':
1520 case '*':
3115e423 1521 FAIL("?+*{} follows nothing in regexp");
a0d0e21e
LW
1522 break;
1523 case '\\':
77d41b28 1524 switch (*++regcomp_parse) {
a0d0e21e 1525 case 'A':
c277df42
IZ
1526 seen_zerolen++;
1527 ret = reg_node(SBOL);
a0d0e21e
LW
1528 *flagp |= SIMPLE;
1529 nextchar();
1530 break;
1531 case 'G':
c277df42
IZ
1532 ret = reg_node(GPOS);
1533 regseen |= REG_SEEN_GPOS;
a0d0e21e
LW
1534 *flagp |= SIMPLE;
1535 nextchar();
1536 break;
1537 case 'Z':
c277df42 1538 ret = reg_node(SEOL);
a0d0e21e
LW
1539 *flagp |= SIMPLE;
1540 nextchar();
1541 break;
1542 case 'w':
c277df42 1543 ret = reg_node((regflags & PMf_LOCALE) ? ALNUML : ALNUM);
a0d0e21e
LW
1544 *flagp |= HASWIDTH|SIMPLE;
1545 nextchar();
1546 break;
1547 case 'W':
c277df42 1548 ret = reg_node((regflags & PMf_LOCALE) ? NALNUML : NALNUM);
a0d0e21e
LW
1549 *flagp |= HASWIDTH|SIMPLE;
1550 nextchar();
1551 break;
1552 case 'b':
c277df42
IZ
1553 seen_zerolen++;
1554 ret = reg_node((regflags & PMf_LOCALE) ? BOUNDL : BOUND);
a0d0e21e
LW
1555 *flagp |= SIMPLE;
1556 nextchar();
1557 break;
1558 case 'B':
c277df42
IZ
1559 seen_zerolen++;
1560 ret = reg_node((regflags & PMf_LOCALE) ? NBOUNDL : NBOUND);
a0d0e21e
LW
1561 *flagp |= SIMPLE;
1562 nextchar();
1563 break;
1564 case 's':
c277df42 1565 ret = reg_node((regflags & PMf_LOCALE) ? SPACEL : SPACE);
a0d0e21e
LW
1566 *flagp |= HASWIDTH|SIMPLE;
1567 nextchar();
1568 break;
1569 case 'S':
c277df42 1570 ret = reg_node((regflags & PMf_LOCALE) ? NSPACEL : NSPACE);
a0d0e21e
LW
1571 *flagp |= HASWIDTH|SIMPLE;
1572 nextchar();
1573 break;
1574 case 'd':
c277df42 1575 ret = reg_node(DIGIT);
a0d0e21e
LW
1576 *flagp |= HASWIDTH|SIMPLE;
1577 nextchar();
1578 break;
1579 case 'D':
c277df42 1580 ret = reg_node(NDIGIT);
a0d0e21e
LW
1581 *flagp |= HASWIDTH|SIMPLE;
1582 nextchar();
1583 break;
1584 case 'n':
1585 case 'r':
1586 case 't':
1587 case 'f':
1588 case 'e':
1589 case 'a':
1590 case 'x':
1591 case 'c':
1592 case '0':
1593 goto defchar;
1594 case '1': case '2': case '3': case '4':
1595 case '5': case '6': case '7': case '8': case '9':
1596 {
77d41b28 1597 I32 num = atoi(regcomp_parse);
a0d0e21e
LW
1598
1599 if (num > 9 && num >= regnpar)
1600 goto defchar;
1601 else {
1602 regsawback = 1;
c8756f30
AK
1603 ret = reganode((regflags & PMf_FOLD)
1604 ? ((regflags & PMf_LOCALE) ? REFFL : REFF)
1605 : REF, num);
a0d0e21e 1606 *flagp |= HASWIDTH;
77d41b28
GS
1607 while (isDIGIT(*regcomp_parse))
1608 regcomp_parse++;
1609 regcomp_parse--;
a0d0e21e
LW
1610 nextchar();
1611 }
1612 }
1613 break;
1614 case '\0':
77d41b28 1615 if (regcomp_parse >= regxend)
a0d0e21e
LW
1616 FAIL("trailing \\ in regexp");
1617 /* FALL THROUGH */
1618 default:
1619 goto defchar;
1620 }
1621 break;
4633a7c4
LW
1622
1623 case '#':
1624 if (regflags & PMf_EXTENDED) {
77d41b28
GS
1625 while (regcomp_parse < regxend && *regcomp_parse != '\n') regcomp_parse++;
1626 if (regcomp_parse < regxend)
4633a7c4
LW
1627 goto tryagain;
1628 }
1629 /* FALL THROUGH */
1630
a0d0e21e
LW
1631 default: {
1632 register I32 len;
c277df42 1633 register U8 ender;
a0d0e21e 1634 register char *p;
c277df42 1635 char *oldp, *s;
a0d0e21e
LW
1636 I32 numlen;
1637
77d41b28 1638 regcomp_parse++;
a0d0e21e
LW
1639
1640 defchar:
c277df42 1641 ret = reg_node((regflags & PMf_FOLD)
bbce6d69
PP
1642 ? ((regflags & PMf_LOCALE) ? EXACTFL : EXACTF)
1643 : EXACT);
161b471a 1644 s = (char *) OPERAND(ret);
c277df42 1645 regc(0, s++); /* save spot for len */
77d41b28 1646 for (len = 0, p = regcomp_parse - 1;
a0d0e21e
LW
1647 len < 127 && p < regxend;
1648 len++)
1649 {
1650 oldp = p;
5b5a24f7
CS
1651
1652 if (regflags & PMf_EXTENDED)
1653 p = regwhite(p, regxend);
a0d0e21e
LW
1654 switch (*p) {
1655 case '^':
1656 case '$':
1657 case '.':
1658 case '[':
1659 case '(':
1660 case ')':
1661 case '|':
1662 goto loopdone;
1663 case '\\':
1664 switch (*++p) {
1665 case 'A':
1666 case 'G':
1667 case 'Z':
1668 case 'w':
1669 case 'W':
1670 case 'b':
1671 case 'B':
1672 case 's':
1673 case 'S':
1674 case 'd':
1675 case 'D':
1676 --p;
1677 goto loopdone;
1678 case 'n':
1679 ender = '\n';
1680 p++;
a687059c 1681 break;
a0d0e21e
LW
1682 case 'r':
1683 ender = '\r';
1684 p++;
a687059c 1685 break;
a0d0e21e
LW
1686 case 't':
1687 ender = '\t';
1688 p++;
a687059c 1689 break;
a0d0e21e
LW
1690 case 'f':
1691 ender = '\f';
1692 p++;
a687059c 1693 break;
a0d0e21e
LW
1694 case 'e':
1695 ender = '\033';
1696 p++;
a687059c 1697 break;
a0d0e21e
LW
1698 case 'a':
1699 ender = '\007';
1700 p++;
a687059c 1701 break;
a0d0e21e
LW
1702 case 'x':
1703 ender = scan_hex(++p, 2, &numlen);
1704 p += numlen;
a687059c 1705 break;
a0d0e21e
LW
1706 case 'c':
1707 p++;
bbce6d69
PP
1708 ender = UCHARAT(p++);
1709 ender = toCTRL(ender);
a687059c 1710 break;
a0d0e21e
LW
1711 case '0': case '1': case '2': case '3':case '4':
1712 case '5': case '6': case '7': case '8':case '9':
1713 if (*p == '0' ||
1714 (isDIGIT(p[1]) && atoi(p) >= regnpar) ) {
1715 ender = scan_oct(p, 3, &numlen);
1716 p += numlen;
1717 }
1718 else {
1719 --p;
1720 goto loopdone;
a687059c
LW
1721 }
1722 break;
a0d0e21e
LW
1723 case '\0':
1724 if (p >= regxend)
a687059c
LW
1725 FAIL("trailing \\ in regexp");
1726 /* FALL THROUGH */
a0d0e21e
LW
1727 default:
1728 ender = *p++;
1729 break;
1730 }
1731 break;
a687059c 1732 default:
a0d0e21e
LW
1733 ender = *p++;
1734 break;
a687059c 1735 }
5b5a24f7
CS
1736 if (regflags & PMf_EXTENDED)
1737 p = regwhite(p, regxend);
a0d0e21e
LW
1738 if (ISMULT2(p)) { /* Back off on ?+*. */
1739 if (len)
1740 p = oldp;
1741 else {
1742 len++;
c277df42 1743 regc(ender, s++);
a0d0e21e
LW
1744 }
1745 break;
a687059c 1746 }
c277df42 1747 regc(ender, s++);
a0d0e21e
LW
1748 }
1749 loopdone:
77d41b28 1750 regcomp_parse = p - 1;
a0d0e21e
LW
1751 nextchar();
1752 if (len < 0)
1753 FAIL("internal disaster in regexp");
1754 if (len > 0)
1755 *flagp |= HASWIDTH;
1756 if (len == 1)
1757 *flagp |= SIMPLE;
c277df42 1758 if (!SIZE_ONLY)
a0d0e21e 1759 *OPERAND(ret) = len;
c277df42
IZ
1760 regc('\0', s++);
1761 if (SIZE_ONLY) {
c277df42 1762 regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
c277df42
IZ
1763 } else {
1764 regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
1765 }
a687059c 1766 }
a0d0e21e
LW
1767 break;
1768 }
a687059c 1769
a0d0e21e 1770 return(ret);
a687059c
LW
1771}
1772
873ef191 1773STATIC char *
8ac85365 1774regwhite(char *p, char *e)
5b5a24f7
CS
1775{
1776 while (p < e) {
1777 if (isSPACE(*p))
1778 ++p;
1779 else if (*p == '#') {
1780 do {
1781 p++;
1782 } while (p < e && *p != '\n');
1783 }
1784 else
1785 break;
1786 }
1787 return p;
1788}
1789
76e3520e 1790STATIC regnode *
8ac85365 1791regclass(void)
a687059c 1792{
c277df42 1793 register char *opnd, *s;
8ac85365 1794 register I32 Class;
a0d0e21e
LW
1795 register I32 lastclass = 1234;
1796 register I32 range = 0;
c277df42 1797 register regnode *ret;
a0d0e21e
LW
1798 register I32 def;
1799 I32 numlen;
1800
161b471a 1801 s = opnd = (char *) OPERAND(regcode);
c277df42 1802 ret = reg_node(ANYOF);
8ac85365 1803 for (Class = 0; Class < 33; Class++)
c277df42 1804 regc(0, s++);
77d41b28 1805 if (*regcomp_parse == '^') { /* Complement of range. */
a0d0e21e 1806 regnaughty++;
77d41b28 1807 regcomp_parse++;
c277df42 1808 if (!SIZE_ONLY)
bbce6d69
PP
1809 *opnd |= ANYOF_INVERT;
1810 }
c277df42
IZ
1811 if (!SIZE_ONLY) {
1812 regcode += ANY_SKIP;
bbce6d69
PP
1813 if (regflags & PMf_FOLD)
1814 *opnd |= ANYOF_FOLD;
1815 if (regflags & PMf_LOCALE)
1816 *opnd |= ANYOF_LOCALE;
c277df42
IZ
1817 } else {
1818 regsize += ANY_SKIP;
a0d0e21e 1819 }
77d41b28 1820 if (*regcomp_parse == ']' || *regcomp_parse == '-')
a0d0e21e 1821 goto skipcond; /* allow 1st char to be ] or - */
77d41b28 1822 while (regcomp_parse < regxend && *regcomp_parse != ']') {
a0d0e21e 1823 skipcond:
77d41b28
GS
1824 Class = UCHARAT(regcomp_parse++);
1825 if (Class == '[' && regcomp_parse + 1 < regxend &&
4599a1de 1826 /* I smell either [: or [= or [. -- POSIX has been here, right? */
77d41b28
GS
1827 (*regcomp_parse == ':' || *regcomp_parse == '=' || *regcomp_parse == '.')) {
1828 char posixccc = *regcomp_parse;
1829 char* posixccs = regcomp_parse++;
4599a1de 1830
77d41b28
GS
1831 while (regcomp_parse < regxend && *regcomp_parse != posixccc)
1832 regcomp_parse++;
1833 if (regcomp_parse == regxend)
4599a1de 1834 /* Grandfather lone [:, [=, [. */
77d41b28 1835 regcomp_parse = posixccs;
4599a1de 1836 else {
77d41b28
GS
1837 regcomp_parse++; /* skip over the posixccc */
1838 if (*regcomp_parse == ']') {
4599a1de
JH
1839 /* Not Implemented Yet.
1840 * (POSIX Extended Character Classes, that is)
1841 * The text between e.g. [: and :] would start
77d41b28 1842 * at posixccs + 1 and stop at regcomp_parse - 2. */
4599a1de
JH
1843 if (dowarn && !SIZE_ONLY)
1844 warn("Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc);
77d41b28 1845 regcomp_parse++; /* skip over the ending ] */
4599a1de
JH
1846 }
1847 }
1848 }
8ac85365 1849 if (Class == '\\') {
77d41b28 1850 Class = UCHARAT(regcomp_parse++);
8ac85365 1851 switch (Class) {
a0d0e21e 1852 case 'w':
ae5c130c
GS
1853 if (!SIZE_ONLY) {
1854 if (regflags & PMf_LOCALE)
bbce6d69 1855 *opnd |= ANYOF_ALNUML;
ae5c130c
GS
1856 else {
1857 for (Class = 0; Class < 256; Class++)
1858 if (isALNUM(Class))
1859 ANYOF_SET(opnd, Class);
1860 }
bbce6d69 1861 }
a0d0e21e
LW
1862 lastclass = 1234;
1863 continue;
1864 case 'W':
ae5c130c
GS
1865 if (!SIZE_ONLY) {
1866 if (regflags & PMf_LOCALE)
bbce6d69 1867 *opnd |= ANYOF_NALNUML;
ae5c130c
GS
1868 else {
1869 for (Class = 0; Class < 256; Class++)
1870 if (!isALNUM(Class))
1871 ANYOF_SET(opnd, Class);
1872 }
bbce6d69 1873 }
a0d0e21e
LW
1874 lastclass = 1234;
1875 continue;
1876 case 's':
ae5c130c
GS
1877 if (!SIZE_ONLY) {
1878 if (regflags & PMf_LOCALE)
bbce6d69 1879 *opnd |= ANYOF_SPACEL;
ae5c130c
GS
1880 else {
1881 for (Class = 0; Class < 256; Class++)
1882 if (isSPACE(Class))
1883 ANYOF_SET(opnd, Class);
1884 }
bbce6d69 1885 }
a0d0e21e
LW
1886 lastclass = 1234;
1887 continue;
1888 case 'S':
ae5c130c
GS
1889 if (!SIZE_ONLY) {
1890 if (regflags & PMf_LOCALE)
bbce6d69 1891 *opnd |= ANYOF_NSPACEL;
ae5c130c
GS
1892 else {
1893 for (Class = 0; Class < 256; Class++)
1894 if (!isSPACE(Class))
1895 ANYOF_SET(opnd, Class);
1896 }
bbce6d69 1897 }
a0d0e21e
LW
1898 lastclass = 1234;
1899 continue;
1900 case 'd':
ae5c130c
GS
1901 if (!SIZE_ONLY) {
1902 for (Class = '0'; Class <= '9'; Class++)
1903 ANYOF_SET(opnd, Class);
1904 }
a0d0e21e
LW
1905 lastclass = 1234;
1906 continue;
1907 case 'D':
ae5c130c
GS
1908 if (!SIZE_ONLY) {
1909 for (Class = 0; Class < '0'; Class++)
1910 ANYOF_SET(opnd, Class);
1911 for (Class = '9' + 1; Class < 256; Class++)
1912 ANYOF_SET(opnd, Class);
1913 }
a0d0e21e
LW
1914 lastclass = 1234;
1915 continue;
1916 case 'n':
8ac85365 1917 Class = '\n';
a0d0e21e
LW
1918 break;
1919 case 'r':
8ac85365 1920 Class = '\r';
a0d0e21e
LW
1921 break;
1922 case 't':
8ac85365 1923 Class = '\t';
a0d0e21e
LW
1924 break;
1925 case 'f':
8ac85365 1926 Class = '\f';
a0d0e21e
LW
1927 break;
1928 case 'b':
8ac85365 1929 Class = '\b';
a0d0e21e
LW
1930 break;
1931 case 'e':
8ac85365 1932 Class = '\033';
a0d0e21e
LW
1933 break;
1934 case 'a':
8ac85365 1935 Class = '\007';
a0d0e21e
LW
1936 break;
1937 case 'x':
77d41b28
GS
1938 Class = scan_hex(regcomp_parse, 2, &numlen);
1939 regcomp_parse += numlen;
a0d0e21e
LW
1940 break;
1941 case 'c':
77d41b28 1942 Class = UCHARAT(regcomp_parse++);
8ac85365 1943 Class = toCTRL(Class);
a0d0e21e
LW
1944 break;
1945 case '0': case '1': case '2': case '3': case '4':
1946 case '5': case '6': case '7': case '8': case '9':
77d41b28
GS
1947 Class = scan_oct(--regcomp_parse, 3, &numlen);
1948 regcomp_parse += numlen;
a0d0e21e
LW
1949 break;
1950 }
1951 }
1952 if (range) {
8ac85365 1953 if (lastclass > Class)
a0d0e21e
LW
1954 FAIL("invalid [] range in regexp");
1955 range = 0;
1956 }
1957 else {
8ac85365 1958 lastclass = Class;
77d41b28
GS
1959 if (*regcomp_parse == '-' && regcomp_parse+1 < regxend &&
1960 regcomp_parse[1] != ']') {
1961 regcomp_parse++;
a0d0e21e
LW
1962 range = 1;
1963 continue; /* do it next time */
1964 }
a687059c 1965 }
ae5c130c
GS
1966 if (!SIZE_ONLY) {
1967 for ( ; lastclass <= Class; lastclass++)
1968 ANYOF_SET(opnd, lastclass);
1969 }
8ac85365 1970 lastclass = Class;
a0d0e21e 1971 }
77d41b28 1972 if (*regcomp_parse != ']')
a0d0e21e
LW
1973 FAIL("unmatched [] in regexp");
1974 nextchar();
ae5c130c
GS
1975 /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
1976 if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) {
1977 for (Class = 0; Class < 256; ++Class) {
1978 if (ANYOF_TEST(opnd, Class)) {
1979 I32 cf = fold[Class];
1980 ANYOF_SET(opnd, cf);
1981 }
1982 }
1983 *opnd &= ~ANYOF_FOLD;
1984 }
1985 /* optimize inverted simple patterns (e.g. [^a-z]) */
1986 if (!SIZE_ONLY && (*opnd & 0xFF) == ANYOF_INVERT) {
1987 for (Class = 0; Class < 32; ++Class)
1988 opnd[1 + Class] ^= 0xFF;
1989 *opnd = 0;
1990 }
a0d0e21e
LW
1991 return ret;
1992}
1993
76e3520e 1994STATIC char*
8ac85365 1995nextchar(void)
a0d0e21e 1996{
77d41b28 1997 char* retval = regcomp_parse++;
a0d0e21e 1998
4633a7c4 1999 for (;;) {
77d41b28
GS
2000 if (*regcomp_parse == '(' && regcomp_parse[1] == '?' &&
2001 regcomp_parse[2] == '#') {
2002 while (*regcomp_parse && *regcomp_parse != ')')
2003 regcomp_parse++;
2004 regcomp_parse++;
4633a7c4
LW
2005 continue;
2006 }
2007 if (regflags & PMf_EXTENDED) {
77d41b28
GS
2008 if (isSPACE(*regcomp_parse)) {
2009 regcomp_parse++;
748a9306
LW
2010 continue;
2011 }
77d41b28
GS
2012 else if (*regcomp_parse == '#') {
2013 while (*regcomp_parse && *regcomp_parse != '\n')
2014 regcomp_parse++;
2015 regcomp_parse++;
748a9306
LW
2016 continue;
2017 }
748a9306 2018 }
4633a7c4 2019 return retval;
a0d0e21e 2020 }
a687059c
LW
2021}
2022
2023/*
c277df42 2024- reg_node - emit a node
a0d0e21e 2025*/
76e3520e 2026STATIC regnode * /* Location. */
c277df42 2027reg_node(U8 op)
a687059c 2028{
c277df42
IZ
2029 register regnode *ret;
2030 register regnode *ptr;
a687059c 2031
a0d0e21e 2032 ret = regcode;
c277df42
IZ
2033 if (SIZE_ONLY) {
2034 SIZE_ALIGN(regsize);
c277df42 2035 regsize += 1;
a0d0e21e
LW
2036 return(ret);
2037 }
a687059c 2038
c277df42 2039 NODE_ALIGN_FILL(ret);
a0d0e21e 2040 ptr = ret;
c277df42 2041 FILL_ADVANCE_NODE(ptr, op);
a0d0e21e 2042 regcode = ptr;
a687059c 2043
a0d0e21e 2044 return(ret);
a687059c
LW
2045}
2046
2047/*
a0d0e21e
LW
2048- reganode - emit a node with an argument
2049*/
76e3520e 2050STATIC regnode * /* Location. */
c277df42 2051reganode(U8 op, U32 arg)
fe14fcc3 2052{
c277df42
IZ
2053 register regnode *ret;
2054 register regnode *ptr;
fe14fcc3 2055
a0d0e21e 2056 ret = regcode;
c277df42
IZ
2057 if (SIZE_ONLY) {
2058 SIZE_ALIGN(regsize);
c277df42 2059 regsize += 2;
a0d0e21e
LW
2060 return(ret);
2061 }
fe14fcc3 2062
c277df42 2063 NODE_ALIGN_FILL(ret);
a0d0e21e 2064 ptr = ret;
c277df42 2065 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
a0d0e21e 2066 regcode = ptr;
fe14fcc3 2067
a0d0e21e 2068 return(ret);
fe14fcc3
LW
2069}
2070
2071/*
a0d0e21e
LW
2072- regc - emit (if appropriate) a byte of code
2073*/
76e3520e 2074STATIC void
c277df42 2075regc(U8 b, char* s)
a687059c 2076{
c277df42
IZ
2077 if (!SIZE_ONLY)
2078 *s = b;
a687059c
LW
2079}
2080
2081/*
a0d0e21e
LW
2082- reginsert - insert an operator in front of already-emitted operand
2083*
2084* Means relocating the operand.
2085*/
76e3520e 2086STATIC void
c277df42 2087reginsert(U8 op, regnode *opnd)
a687059c 2088{
c277df42
IZ
2089 register regnode *src;
2090 register regnode *dst;
2091 register regnode *place;
2092 register int offset = regarglen[(U8)op];
2093
2094/* (regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
2095
2096 if (SIZE_ONLY) {
2097 regsize += NODE_STEP_REGNODE + offset;
a0d0e21e
LW
2098 return;
2099 }
a687059c 2100
a0d0e21e 2101 src = regcode;
c277df42 2102 regcode += NODE_STEP_REGNODE + offset;
a0d0e21e
LW
2103 dst = regcode;
2104 while (src > opnd)
c277df42 2105 StructCopy(--src, --dst, regnode);
a0d0e21e
LW
2106
2107 place = opnd; /* Op node, where operand used to be. */
c277df42
IZ
2108 src = NEXTOPER(place);
2109 FILL_ADVANCE_NODE(place, op);
2110 Zero(src, offset, regnode);
a687059c
LW
2111}
2112
2113/*
c277df42 2114- regtail - set the next-pointer at the end of a node chain of p to val.
a0d0e21e 2115*/
76e3520e 2116STATIC void
c277df42 2117regtail(regnode *p, regnode *val)
a687059c 2118{
c277df42
IZ
2119 register regnode *scan;
2120 register regnode *temp;
a0d0e21e
LW
2121 register I32 offset;
2122
c277df42 2123 if (SIZE_ONLY)
a0d0e21e
LW
2124 return;
2125
2126 /* Find last node. */
2127 scan = p;
2128 for (;;) {
2129 temp = regnext(scan);
2130 if (temp == NULL)
2131 break;
2132 scan = temp;
2133 }
a687059c 2134
c277df42
IZ
2135 if (reg_off_by_arg[OP(scan)]) {
2136 ARG_SET(scan, val - scan);
2137 } else {
2138 NEXT_OFF(scan) = val - scan;
2139 }
a687059c
LW
2140}
2141
2142/*
a0d0e21e
LW
2143- regoptail - regtail on operand of first argument; nop if operandless
2144*/
76e3520e 2145STATIC void
c277df42 2146regoptail(regnode *p, regnode *val)
a687059c 2147{
a0d0e21e 2148 /* "Operandless" and "op != BRANCH" are synonymous in practice. */
c277df42
IZ
2149 if (p == NULL || SIZE_ONLY)
2150 return;
2151 if (regkind[(U8)OP(p)] == BRANCH) {
2152 regtail(NEXTOPER(p), val);
2153 } else if ( regkind[(U8)OP(p)] == BRANCHJ) {
2154 regtail(NEXTOPER(NEXTOPER(p)), val);
2155 } else
a0d0e21e 2156 return;
a687059c
LW
2157}
2158
2159/*
2160 - regcurly - a little FSA that accepts {\d+,?\d*}
2161 */
79072805 2162STATIC I32
8ac85365 2163regcurly(register char *s)
a687059c
LW
2164{
2165 if (*s++ != '{')
2166 return FALSE;
f0fcb552 2167 if (!isDIGIT(*s))
a687059c 2168 return FALSE;
f0fcb552 2169 while (isDIGIT(*s))
a687059c
LW
2170 s++;
2171 if (*s == ',')
2172 s++;
f0fcb552 2173 while (isDIGIT(*s))
a687059c
LW
2174 s++;
2175 if (*s != '}')
2176 return FALSE;
2177 return TRUE;
2178}
2179
a687059c 2180
76e3520e 2181STATIC regnode *
c277df42
IZ
2182dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
2183{
35ff7856 2184#ifdef DEBUGGING
c277df42
IZ
2185 register char op = EXACT; /* Arbitrary non-END op. */
2186 register regnode *next, *onode;
2187
2188 while (op != END && (!last || node < last)) {
2189 /* While that wasn't END last time... */
2190
2191 NODE_ALIGN(node);
2192 op = OP(node);
2193 if (op == CLOSE)
2194 l--;
2195 next = regnext(node);
2196 /* Where, what. */
2197 if (OP(node) == OPTIMIZED)
2198 goto after_print;
2199 regprop(sv, node);
2200 PerlIO_printf(Perl_debug_log, "%4d%*s%s", node - start,
2201 2*l + 1, "", SvPVX(sv));
2202 if (next == NULL) /* Next ptr. */
2203 PerlIO_printf(Perl_debug_log, "(0)");
2204 else
2205 PerlIO_printf(Perl_debug_log, "(%d)", next - start);
2206 (void)PerlIO_putc(Perl_debug_log, '\n');
2207 after_print:
2208 if (regkind[(U8)op] == BRANCHJ) {
2209 register regnode *nnode = (OP(next) == LONGJMP
2210 ? regnext(next)
2211 : next);
2212 if (last && nnode > last)
2213 nnode = last;
2214 node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
2215 } else if (regkind[(U8)op] == BRANCH) {
2216 node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
2217 } else if ( op == CURLY) { /* `next' might be very big: optimizer */
2218 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
2219 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
2220 } else if (regkind[(U8)op] == CURLY && op != CURLYX) {
2221 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
2222 next, sv, l + 1);
2223 } else if ( op == PLUS || op == STAR) {
2224 node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
2225 } else if (op == ANYOF) {
2226 node = NEXTOPER(node);
2227 node += ANY_SKIP;
2228 } else if (regkind[(U8)op] == EXACT) {
2229 /* Literal string, where present. */
2230 node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode);
2231 node = NEXTOPER(node);
2232 } else {
2233 node = NEXTOPER(node);
2234 node += regarglen[(U8)op];
2235 }
2236 if (op == CURLYX || op == OPEN)
2237 l++;
2238 else if (op == WHILEM)
2239 l--;
2240 }
17c3b450 2241#endif /* DEBUGGING */
c277df42
IZ
2242 return node;
2243}
2244
a687059c 2245/*
fd181c75 2246 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
a687059c
LW
2247 */
2248void
8ac85365 2249regdump(regexp *r)
a687059c 2250{
35ff7856 2251#ifdef DEBUGGING
46fc3d4c 2252 SV *sv = sv_newmortal();
a687059c 2253
c277df42 2254 (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
a0d0e21e
LW
2255
2256 /* Header fields of interest. */
c277df42
IZ
2257 if (r->anchored_substr)
2258 PerlIO_printf(Perl_debug_log, "anchored `%s%s%s'%s at %d ",
2259 colors[0],
2260 SvPVX(r->anchored_substr),
2261 colors[1],
2262 SvTAIL(r->anchored_substr) ? "$" : "",
2263 r->anchored_offset);
2264 if (r->float_substr)
2265 PerlIO_printf(Perl_debug_log, "floating `%s%s%s'%s at %d..%u ",
2266 colors[0],
2267 SvPVX(r->float_substr),
2268 colors[1],
2269 SvTAIL(r->float_substr) ? "$" : "",
2270 r->float_min_offset, r->float_max_offset);
2271 if (r->check_substr)
2272 PerlIO_printf(Perl_debug_log,
2273 r->check_substr == r->float_substr
2274 ? "(checking floating" : "(checking anchored");
2275 if (r->reganch & ROPT_NOSCAN)
2276 PerlIO_printf(Perl_debug_log, " noscan");
2277 if (r->reganch & ROPT_CHECK_ALL)
2278 PerlIO_printf(Perl_debug_log, " isall");
2279 if (r->check_substr)
2280 PerlIO_printf(Perl_debug_log, ") ");
2281
46fc3d4c
PP
2282 if (r->regstclass) {
2283 regprop(sv, r->regstclass);
2284 PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
2285 }
774d564b
PP
2286 if (r->reganch & ROPT_ANCH) {
2287 PerlIO_printf(Perl_debug_log, "anchored");
2288 if (r->reganch & ROPT_ANCH_BOL)
2289 PerlIO_printf(Perl_debug_log, "(BOL)");
c277df42
IZ
2290 if (r->reganch & ROPT_ANCH_MBOL)
2291 PerlIO_printf(Perl_debug_log, "(MBOL)");
774d564b
PP
2292 if (r->reganch & ROPT_ANCH_GPOS)
2293 PerlIO_printf(Perl_debug_log, "(GPOS)");
2294 PerlIO_putc(Perl_debug_log, ' ');
2295 }
c277df42
IZ
2296 if (r->reganch & ROPT_GPOS_SEEN)
2297 PerlIO_printf(Perl_debug_log, "GPOS ");
a0d0e21e 2298 if (r->reganch & ROPT_SKIP)
760ac839 2299 PerlIO_printf(Perl_debug_log, "plus ");
a0d0e21e 2300 if (r->reganch & ROPT_IMPLICIT)
760ac839 2301 PerlIO_printf(Perl_debug_log, "implicit ");
760ac839 2302 PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
ce862d02
IZ
2303 if (r->reganch & ROPT_EVAL_SEEN)
2304 PerlIO_printf(Perl_debug_log, "with eval ");
760ac839 2305 PerlIO_printf(Perl_debug_log, "\n");
17c3b450 2306#endif /* DEBUGGING */
a687059c
LW
2307}
2308
2309/*
a0d0e21e
LW
2310- regprop - printable representation of opcode
2311*/
46fc3d4c 2312void
c277df42 2313regprop(SV *sv, regnode *o)
a687059c 2314{
35ff7856 2315#ifdef DEBUGGING
a0d0e21e
LW
2316 register char *p = 0;
2317
46fc3d4c 2318 sv_setpv(sv, ":");
11343788 2319 switch (OP(o)) {
a0d0e21e
LW
2320 case BOL:
2321 p = "BOL";
2322 break;
2323 case MBOL:
2324 p = "MBOL";
2325 break;
2326 case SBOL:
2327 p = "SBOL";
2328 break;
2329 case EOL:
2330 p = "EOL";
2331 break;
2332 case MEOL:
2333 p = "MEOL";
2334 break;
2335 case SEOL:
2336 p = "SEOL";
2337 break;
2338 case ANY:
2339 p = "ANY";
2340 break;
2341 case SANY:
2342 p = "SANY";
2343 break;
2344 case ANYOF:
2345 p = "ANYOF";
2346 break;
2347 case BRANCH:
2348 p = "BRANCH";
2349 break;
bbce6d69 2350 case EXACT:
c277df42 2351 sv_catpvf(sv, "EXACT <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
bbce6d69
PP
2352 break;
2353 case EXACTF:
c277df42 2354 sv_catpvf(sv, "EXACTF <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
bbce6d69
PP
2355 break;
2356 case EXACTFL:
c277df42 2357 sv_catpvf(sv, "EXACTFL <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
a0d0e21e
LW
2358 break;
2359 case NOTHING:
2360 p = "NOTHING";
2361 break;
c277df42
IZ
2362 case TAIL:
2363 p = "TAIL";
2364 break;
a0d0e21e
LW
2365 case BACK:
2366 p = "BACK";
2367 break;
2368 case END:
2369 p = "END";
2370 break;
a0d0e21e
LW
2371 case BOUND:
2372 p = "BOUND";
2373 break;
bbce6d69
PP
2374 case BOUNDL:
2375 p = "BOUNDL";
2376 break;
a0d0e21e
LW
2377 case NBOUND:
2378 p = "NBOUND";
2379 break;
bbce6d69
PP
2380 case NBOUNDL:
2381 p = "NBOUNDL";
a0d0e21e
LW
2382 break;
2383 case CURLY:
5dc0d613 2384 sv_catpvf(sv, "CURLY {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e 2385 break;
c277df42 2386 case CURLYM:
c277df42 2387 sv_catpvf(sv, "CURLYM[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
c277df42
IZ
2388 break;
2389 case CURLYN:
c277df42 2390 sv_catpvf(sv, "CURLYN[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
c277df42 2391 break;
a0d0e21e 2392 case CURLYX:
5dc0d613 2393 sv_catpvf(sv, "CURLYX {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e
LW
2394 break;
2395 case REF:
c277df42 2396 sv_catpvf(sv, "REF%d", ARG(o));
a0d0e21e 2397 break;
c8756f30 2398 case REFF:
c277df42 2399 sv_catpvf(sv, "REFF%d", ARG(o));
c8756f30
AK
2400 break;
2401 case REFFL:
c277df42 2402 sv_catpvf(sv, "REFFL%d", ARG(o));
c8756f30 2403 break;
a0d0e21e 2404 case OPEN:
c277df42 2405 sv_catpvf(sv, "OPEN%d", ARG(o));
a0d0e21e
LW
2406 break;
2407 case CLOSE:
c277df42 2408 sv_catpvf(sv, "CLOSE%d", ARG(o));
a0d0e21e
LW
2409 p = NULL;
2410 break;
2411 case STAR:
2412 p = "STAR";
2413 break;
2414 case PLUS:
2415 p = "PLUS";
2416 break;
2417 case MINMOD:
2418 p = "MINMOD";
2419 break;
774d564b
PP
2420 case GPOS:
2421 p = "GPOS";
a0d0e21e
LW
2422 break;
2423 case UNLESSM:
c277df42 2424 sv_catpvf(sv, "UNLESSM[-%d]", o->flags);
a0d0e21e
LW
2425 break;
2426 case IFMATCH:
c277df42 2427 sv_catpvf(sv, "IFMATCH[-%d]", o->flags);
a0d0e21e
LW
2428 break;
2429 case SUCCEED:
2430 p = "SUCCEED";
2431 break;
2432 case WHILEM:
2433 p = "WHILEM";
2434 break;
bbce6d69
PP
2435 case DIGIT:
2436 p = "DIGIT";
2437 break;
2438 case NDIGIT:
2439 p = "NDIGIT";
2440 break;
2441 case ALNUM:
2442 p = "ALNUM";
2443 break;
2444 case NALNUM:
2445 p = "NALNUM";
2446 break;
2447 case SPACE:
2448 p = "SPACE";
2449 break;
2450 case NSPACE:
2451 p = "NSPACE";
2452 break;
2453 case ALNUML:
2454 p = "ALNUML";
2455 break;
2456 case NALNUML:
2457 p = "NALNUML";
2458 break;
2459 case SPACEL:
2460 p = "SPACEL";
2461 break;
2462 case NSPACEL:
2463 p = "NSPACEL";
2464 break;
c277df42
IZ
2465 case EVAL:
2466 p = "EVAL";
2467 break;
2468 case LONGJMP:
2469 p = "LONGJMP";
2470 break;
2471 case BRANCHJ:
2472 p = "BRANCHJ";
2473 break;
2474 case IFTHEN:
2475 p = "IFTHEN";
2476 break;
2477 case GROUPP:
2478 sv_catpvf(sv, "GROUPP%d", ARG(o));
2479 break;
2480 case LOGICAL:
2481 p = "LOGICAL";
2482 break;
2483 case SUSPEND:
2484 p = "SUSPEND";
2485 break;
2486 case RENUM:
2487 p = "RENUM";
2488 break;
2489 case OPTIMIZED:
2490 p = "OPTIMIZED";
2491 break;
a0d0e21e
LW
2492 default:
2493 FAIL("corrupted regexp opcode");
2494 }
46fc3d4c
PP
2495 if (p)
2496 sv_catpv(sv, p);
17c3b450 2497#endif /* DEBUGGING */
35ff7856 2498}
a687059c 2499
2b69d0c2 2500void
8ac85365 2501pregfree(struct regexp *r)
a687059c 2502{
c277df42 2503 if (!r || (--r->refcnt > 0))
a0d0e21e 2504 return;
c277df42 2505 if (r->precomp)
a0d0e21e 2506 Safefree(r->precomp);
c277df42 2507 if (r->subbase)
a0d0e21e 2508 Safefree(r->subbase);
a193d654
GS
2509 if (r->substrs) {
2510 if (r->anchored_substr)
2511 SvREFCNT_dec(r->anchored_substr);
2512 if (r->float_substr)
2513 SvREFCNT_dec(r->float_substr);
2779dcf1 2514 Safefree(r->substrs);
a193d654 2515 }
c277df42
IZ
2516 if (r->data) {
2517 int n = r->data->count;
2518 while (--n >= 0) {
2519 switch (r->data->what[n]) {
2520 case 's':
2521 SvREFCNT_dec((SV*)r->data->data[n]);
2522 break;
2523 case 'o':
2524 op_free((OP_4tree*)r->data->data[n]);
2525 break;
2526 case 'n':
2527 break;
2528 default:
2529 FAIL2("panic: regfree data code '%c'", r->data->what[n]);
2530 }
2531 }
2532 Safefree(r->data->what);
2533 Safefree(r->data);
a0d0e21e
LW
2534 }
2535 Safefree(r->startp);
2536 Safefree(r->endp);
2537 Safefree(r);
a687059c 2538}
c277df42
IZ
2539
2540/*
2541 - regnext - dig the "next" pointer out of a node
2542 *
2543 * [Note, when REGALIGN is defined there are two places in regmatch()
2544 * that bypass this code for speed.]
2545 */
2546regnode *
2547regnext(register regnode *p)
2548{
2549 register I32 offset;
2550
2551 if (p == &regdummy)
2552 return(NULL);
2553
2554 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
2555 if (offset == 0)
2556 return(NULL);
2557
c277df42 2558 return(p+offset);
c277df42
IZ
2559}
2560
01f988be 2561STATIC void
c277df42 2562re_croak2(const char* pat1,const char* pat2,...)
c277df42
IZ
2563{
2564 va_list args;
2565 STRLEN l1 = strlen(pat1);
2566 STRLEN l2 = strlen(pat2);
2567 char buf[512];
2568 char *message;
2569
2570 if (l1 > 510)
2571 l1 = 510;
2572 if (l1 + l2 > 510)
2573 l2 = 510 - l1;
2574 Copy(pat1, buf, l1 , char);
2575 Copy(pat2, buf + l1, l2 , char);
3b818b81
GS
2576 buf[l1 + l2] = '\n';
2577 buf[l1 + l2 + 1] = '\0';
c277df42 2578 va_start(args, pat2);
c277df42
IZ
2579 message = mess(buf, &args);
2580 va_end(args);
2581 l1 = strlen(message);
2582 if (l1 > 512)
2583 l1 = 512;
2584 Copy(message, buf, l1 , char);
2585 buf[l1] = '\0'; /* Overwrite \n */
2586 croak("%s", buf);
2587}