projects / p5sagit/p5-mst-13.2.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | history | HEAD
perl5.004 hints file (maint and dev paths)
[p5sagit/p5-mst-13.2.git] / regcomp.c
CommitLineData
a0d0e21e 1/* regcomp.c
2 */
3
4/*
5 * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee
6 */
7
a687059c 8/* NOTE: this is derived from Henry Spencer's regexp code, and should not
9 * confused with the original package (see point 3 below). Thanks, Henry!
10 */
11
12/* Additional note: this code is very heavily munged from Henry's version
13 * in places. In some spots I've traded clarity for efficiency, so don't
14 * blame Henry for some of the lack of readability.
15 */
16
e50aee73 17/* The names of the functions have been changed from regcomp and
18 * regexec to pregcomp and pregexec in order to avoid conflicts
19 * with the POSIX routines of the same names.
20*/
21
f0fcb552 22/*SUPPRESS 112*/
a687059c 23/*
e50aee73 24 * pregcomp and pregexec -- regsub and regerror are not used in perl
a687059c 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 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 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 59
60#define REG_COMP_C
a687059c 61#include "regcomp.h"
62
d4cce5f1 63#ifdef op
11343788 64#undef op
d4cce5f1 65#endif /* op */
11343788 66
fe14fcc3 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 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 83#ifdef atarist
84#define PERL_META "^$.[()|?+*\\"
85#else
a687059c 86#define META "^$.[()|?+*\\"
2b69d0c2 87#endif
a687059c 88
35c8bce7 89#ifdef SPSTART
90#undef SPSTART /* dratted cpp namespace... */
91#endif
a687059c 92/*
93 * Flags to be passed up and down.
94 */
a687059c 95#define WORST 0 /* Worst case. */
a0d0e21e 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 100
101/*
e50aee73 102 * Forward declarations for pregcomp()'s friends.
a687059c 103 */
a0d0e21e 104
76e3520e 105#ifndef PERL_OBJECT
c277df42 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 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 123/* Length of a variant. */
124
76e3520e 125#ifndef PERL_OBJECT
c277df42 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 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 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 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 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 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 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) {
270 if (regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
271 OP(n) = OPTIMIZED;
272 NEXT_OFF(n) = 0;
273 }
274 n++;
275 }
276#endif
277
278 }
279 if (OP(scan) != CURLYX) {
280 int max = (reg_off_by_arg[OP(scan)] ? I32_MAX : U16_MAX);
281 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
282 int noff;
283 regnode *n = scan;
284
285 /* Skip NOTHING and LONGJMP. */
286 while ((n = regnext(n))
287 && ((regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
288 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
289 && off + noff < max)
290 off += noff;
291 if (reg_off_by_arg[OP(scan)])
292 ARG(scan) = off;
293 else
294 NEXT_OFF(scan) = off;
295 }
296 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
297 || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
298 next = regnext(scan);
299 code = OP(scan);
300
301 if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
302 I32 max1 = 0, min1 = I32_MAX, num = 0;
303
304 if (flags & SCF_DO_SUBSTR)
305 scan_commit(data);
306 while (OP(scan) == code) {
307 I32 deltanext, minnext;
308
309 num++;
310 data_fake.flags = 0;
311 next = regnext(scan);
312 scan = NEXTOPER(scan);
313 if (code != BRANCH)
314 scan = NEXTOPER(scan);
315 /* We suppose the run is continuous, last=next...*/
316 minnext = study_chunk(&scan, &deltanext, next,
317 &data_fake, 0);
318 if (min1 > minnext)
319 min1 = minnext;
320 if (max1 < minnext + deltanext)
321 max1 = minnext + deltanext;
322 if (deltanext == I32_MAX)
323 is_inf = 1;
324 scan = next;
325 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
326 pars++;
405ff068 327 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42 328 data->flags |= SF_HAS_EVAL;
329 if (code == SUSPEND)
330 break;
331 }
332 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
333 min1 = 0;
334 if (flags & SCF_DO_SUBSTR) {
335 data->pos_min += min1;
336 data->pos_delta += max1 - min1;
337 if (max1 != min1 || is_inf)
338 data->longest = &(data->longest_float);
339 }
340 min += min1;
341 delta += max1 - min1;
342 } else if (code == BRANCHJ) /* single branch is optimized. */
343 scan = NEXTOPER(NEXTOPER(scan));
344 else /* single branch is optimized. */
345 scan = NEXTOPER(scan);
346 continue;
347 } else if (OP(scan) == EXACT) {
348 min += *OPERAND(scan);
349 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
350 I32 l = *OPERAND(scan);
351
352 /* The code below prefers earlier match for fixed
353 offset, later match for variable offset. */
354 if (data->last_end == -1) { /* Update the start info. */
355 data->last_start_min = data->pos_min;
356 data->last_start_max = is_inf
357 ? I32_MAX : data->pos_min + data->pos_delta;
358 }
161b471a 359 sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), l);
c277df42 360 data->last_end = data->pos_min + l;
361 data->pos_min += l; /* As in the first entry. */
362 data->flags &= ~SF_BEFORE_EOL;
363 }
364 } else if (regkind[(U8)OP(scan)] == EXACT) {
365 if (flags & SCF_DO_SUBSTR)
366 scan_commit(data);
367 min += *OPERAND(scan);
368 if (data && (flags & SCF_DO_SUBSTR))
369 data->pos_min += *OPERAND(scan);
370 } else if (strchr(varies,OP(scan))) {
371 I32 mincount, maxcount, minnext, deltanext, pos_before, fl;
372 regnode *oscan = scan;
373
374 switch (regkind[(U8)OP(scan)]) {
375 case WHILEM:
376 scan = NEXTOPER(scan);
377 goto finish;
378 case PLUS:
379 if (flags & SCF_DO_SUBSTR) {
380 next = NEXTOPER(scan);
381 if (OP(next) == EXACT) {
382 mincount = 1;
383 maxcount = REG_INFTY;
384 next = regnext(scan);
385 scan = NEXTOPER(scan);
386 goto do_curly;
387 }
388 }
389 if (flags & SCF_DO_SUBSTR)
390 data->pos_min++;
391 min++;
392 /* Fall through. */
393 case STAR:
394 is_inf = 1;
395 scan = regnext(scan);
396 if (flags & SCF_DO_SUBSTR) {
397 scan_commit(data);
398 data->longest = &(data->longest_float);
399 }
400 goto optimize_curly_tail;
401 case CURLY:
402 mincount = ARG1(scan);
403 maxcount = ARG2(scan);
404 next = regnext(scan);
405 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
406 do_curly:
407 if (flags & SCF_DO_SUBSTR) {
408 if (mincount == 0) scan_commit(data);
409 pos_before = data->pos_min;
410 }
411 if (data) {
412 fl = data->flags;
413 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
414 if (is_inf)
415 data->flags |= SF_IS_INF;
416 }
417 /* This will finish on WHILEM, setting scan, or on NULL: */
418 minnext = study_chunk(&scan, &deltanext, last, data,
419 mincount == 0
420 ? (flags & ~SCF_DO_SUBSTR) : flags);
421 if (!scan) /* It was not CURLYX, but CURLY. */
422 scan = next;
423 if (dowarn && (minnext + deltanext == 0)
424 && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)))
425 warn("Strange *+?{} on zero-length expression");
426 min += minnext * mincount;
427 is_inf |= (maxcount == REG_INFTY && (minnext + deltanext) > 0
428 || deltanext == I32_MAX);
429 delta += (minnext + deltanext) * maxcount - minnext * mincount;
430
431 /* Try powerful optimization CURLYX => CURLYN. */
c277df42 432 if ( OP(oscan) == CURLYX && data
433 && data->flags & SF_IN_PAR
434 && !(data->flags & SF_HAS_EVAL)
435 && !deltanext && minnext == 1 ) {
436 /* Try to optimize to CURLYN. */
437 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
438 regnode *nxt1 = nxt, *nxt2;
439
440 /* Skip open. */
441 nxt = regnext(nxt);
442 if (!strchr(simple,OP(nxt))
443 && !(regkind[(U8)OP(nxt)] == EXACT
444 && *OPERAND(nxt) == 1))
445 goto nogo;
446 nxt2 = nxt;
447 nxt = regnext(nxt);
448 if (OP(nxt) != CLOSE)
449 goto nogo;
450 /* Now we know that nxt2 is the only contents: */
451 oscan->flags = ARG(nxt);
452 OP(oscan) = CURLYN;
453 OP(nxt1) = NOTHING; /* was OPEN. */
454#ifdef DEBUGGING
455 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
456 NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
457 NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
458 OP(nxt) = OPTIMIZED; /* was CLOSE. */
459 OP(nxt + 1) = OPTIMIZED; /* was count. */
460 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
461#endif
462 }
c277df42 463 nogo:
464
465 /* Try optimization CURLYX => CURLYM. */
466 if ( OP(oscan) == CURLYX && data
c277df42 467 && !(data->flags & SF_HAS_PAR)
c277df42 468 && !(data->flags & SF_HAS_EVAL)
469 && !deltanext ) {
470 /* XXXX How to optimize if data == 0? */
471 /* Optimize to a simpler form. */
472 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
473 regnode *nxt2;
474
475 OP(oscan) = CURLYM;
476 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
477 && (OP(nxt2) != WHILEM))
478 nxt = nxt2;
479 OP(nxt2) = SUCCEED; /* Whas WHILEM */
c277df42 480 /* Need to optimize away parenths. */
481 if (data->flags & SF_IN_PAR) {
482 /* Set the parenth number. */
483 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
484
485 if (OP(nxt) != CLOSE)
486 FAIL("panic opt close");
487 oscan->flags = ARG(nxt);
488 OP(nxt1) = OPTIMIZED; /* was OPEN. */
489 OP(nxt) = OPTIMIZED; /* was CLOSE. */
490#ifdef DEBUGGING
491 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
492 OP(nxt + 1) = OPTIMIZED; /* was count. */
493 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
494 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
495#endif
496#if 0
497 while ( nxt1 && (OP(nxt1) != WHILEM)) {
498 regnode *nnxt = regnext(nxt1);
499
500 if (nnxt == nxt) {
501 if (reg_off_by_arg[OP(nxt1)])
502 ARG_SET(nxt1, nxt2 - nxt1);
503 else if (nxt2 - nxt1 < U16_MAX)
504 NEXT_OFF(nxt1) = nxt2 - nxt1;
505 else
506 OP(nxt) = NOTHING; /* Cannot beautify */
507 }
508 nxt1 = nnxt;
509 }
510#endif
511 /* Optimize again: */
512 study_chunk(&nxt1, &deltanext, nxt, NULL, 0);
513 } else
514 oscan->flags = 0;
c277df42 515 }
516 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
517 pars++;
518 if (flags & SCF_DO_SUBSTR) {
519 SV *last_str = Nullsv;
520 int counted = mincount != 0;
521
522 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
523 I32 b = pos_before >= data->last_start_min
524 ? pos_before : data->last_start_min;
525 STRLEN l;
526 char *s = SvPV(data->last_found, l);
527
528 l -= b - data->last_start_min;
529 /* Get the added string: */
530 last_str = newSVpv(s + b - data->last_start_min, l);
531 if (deltanext == 0 && pos_before == b) {
532 /* What was added is a constant string */
533 if (mincount > 1) {
534 SvGROW(last_str, (mincount * l) + 1);
535 repeatcpy(SvPVX(last_str) + l,
536 SvPVX(last_str), l, mincount - 1);
537 SvCUR(last_str) *= mincount;
538 /* Add additional parts. */
539 SvCUR_set(data->last_found,
540 SvCUR(data->last_found) - l);
541 sv_catsv(data->last_found, last_str);
542 data->last_end += l * (mincount - 1);
543 }
544 }
545 }
546 /* It is counted once already... */
547 data->pos_min += minnext * (mincount - counted);
548 data->pos_delta += - counted * deltanext +
549 (minnext + deltanext) * maxcount - minnext * mincount;
550 if (mincount != maxcount) {
551 scan_commit(data);
552 if (mincount && last_str) {
553 sv_setsv(data->last_found, last_str);
554 data->last_end = data->pos_min;
555 data->last_start_min =
556 data->pos_min - SvCUR(last_str);
557 data->last_start_max = is_inf
558 ? I32_MAX
559 : data->pos_min + data->pos_delta
560 - SvCUR(last_str);
561 }
562 data->longest = &(data->longest_float);
563 }
564 }
405ff068 565 if (data && (fl & SF_HAS_EVAL))
c277df42 566 data->flags |= SF_HAS_EVAL;
567 optimize_curly_tail:
c277df42 568 if (OP(oscan) != CURLYX) {
569 while (regkind[(U8)OP(next = regnext(oscan))] == NOTHING
570 && NEXT_OFF(next))
571 NEXT_OFF(oscan) += NEXT_OFF(next);
572 }
c277df42 573 continue;
574 default: /* REF only? */
575 if (flags & SCF_DO_SUBSTR) {
576 scan_commit(data);
577 data->longest = &(data->longest_float);
578 }
579 is_inf = 1;
580 break;
581 }
582 } else if (strchr(simple,OP(scan))) {
583 if (flags & SCF_DO_SUBSTR) {
584 scan_commit(data);
585 data->pos_min++;
586 }
587 min++;
588 } else if (regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
589 data->flags |= (OP(scan) == MEOL
590 ? SF_BEFORE_MEOL
591 : SF_BEFORE_SEOL);
592 } else if (regkind[(U8)OP(scan)] == BRANCHJ
593 && (scan->flags || data)
594 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
595 I32 deltanext, minnext;
596 regnode *nscan;
597
598 data_fake.flags = 0;
599 next = regnext(scan);
600 nscan = NEXTOPER(NEXTOPER(scan));
601 minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0);
602 if (scan->flags) {
603 if (deltanext) {
604 FAIL("variable length lookbehind not implemented");
605 } else if (minnext > U8_MAX) {
606 FAIL2("lookbehind longer than %d not implemented", U8_MAX);
607 }
608 scan->flags = minnext;
609 }
610 if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
611 pars++;
405ff068 612 if (data && (data_fake.flags & SF_HAS_EVAL))
c277df42 613 data->flags |= SF_HAS_EVAL;
614 } else if (OP(scan) == OPEN) {
615 pars++;
616 } else if (OP(scan) == CLOSE && ARG(scan) == is_par) {
c277df42 617 next = regnext(scan);
618
619 if ( next && (OP(next) != WHILEM) && next < last)
c277df42 620 is_par = 0; /* Disable optimization */
621 } else if (OP(scan) == EVAL) {
622 if (data)
623 data->flags |= SF_HAS_EVAL;
624 }
625 /* Else: zero-length, ignore. */
626 scan = regnext(scan);
627 }
628
629 finish:
630 *scanp = scan;
631 *deltap = is_inf ? I32_MAX : delta;
632 if (flags & SCF_DO_SUBSTR && is_inf)
633 data->pos_delta = I32_MAX - data->pos_min;
634 if (is_par > U8_MAX)
635 is_par = 0;
636 if (is_par && pars==1 && data) {
637 data->flags |= SF_IN_PAR;
638 data->flags &= ~SF_HAS_PAR;
639 } else if (pars && data) {
640 data->flags |= SF_HAS_PAR;
641 data->flags &= ~SF_IN_PAR;
642 }
643 return min;
644}
645
76e3520e 646STATIC I32
c277df42 647add_data(I32 n, char *s)
648{
77d41b28 649 if (regcomp_rx->data) {
650 Renewc(regcomp_rx->data,
651 sizeof(*regcomp_rx->data) + sizeof(void*) * (regcomp_rx->data->count + n - 1),
c277df42 652 char, struct reg_data);
77d41b28 653 Renew(regcomp_rx->data->what, regcomp_rx->data->count + n, U8);
654 regcomp_rx->data->count += n;
c277df42 655 } else {
77d41b28 656 Newc(1207, regcomp_rx->data, sizeof(*regcomp_rx->data) + sizeof(void*) * (n - 1),
c277df42 657 char, struct reg_data);
77d41b28 658 New(1208, regcomp_rx->data->what, n, U8);
659 regcomp_rx->data->count = n;
c277df42 660 }
77d41b28 661 Copy(s, regcomp_rx->data->what + regcomp_rx->data->count - n, n, U8);
662 return regcomp_rx->data->count - n;
c277df42 663}
664
a687059c 665/*
e50aee73 666 - pregcomp - compile a regular expression into internal code
a687059c 667 *
668 * We can't allocate space until we know how big the compiled form will be,
669 * but we can't compile it (and thus know how big it is) until we've got a
670 * place to put the code. So we cheat: we compile it twice, once with code
671 * generation turned off and size counting turned on, and once "for real".
672 * This also means that we don't allocate space until we are sure that the
673 * thing really will compile successfully, and we never have to move the
674 * code and thus invalidate pointers into it. (Note that it has to be in
675 * one piece because free() must be able to free it all.) [NB: not true in perl]
676 *
677 * Beware that the optimization-preparation code in here knows about some
678 * of the structure of the compiled regexp. [I'll say.]
679 */
680regexp *
8ac85365 681pregcomp(char *exp, char *xend, PMOP *pm)
a687059c 682{
a0d0e21e 683 register regexp *r;
c277df42 684 regnode *scan;
685 SV **longest;
686 SV *longest_fixed;
687 SV *longest_float;
688 regnode *first;
a0d0e21e 689 I32 flags;
a0d0e21e 690 I32 minlen = 0;
691 I32 sawplus = 0;
692 I32 sawopen = 0;
693
694 if (exp == NULL)
c277df42 695 FAIL("NULL regexp argument");
a0d0e21e 696
bbce6d69 697 regprecomp = savepvn(exp, xend - exp);
c277df42 698 DEBUG_r(PerlIO_printf(Perl_debug_log, "compiling RE `%*s'\n",
699 xend - exp, regprecomp));
a0d0e21e 700 regflags = pm->op_pmflags;
bbce6d69 701 regsawback = 0;
702
c277df42 703 regseen = 0;
704 seen_zerolen = *exp == '^' ? -1 : 0;
705 extralen = 0;
706
bbce6d69 707 /* First pass: determine size, legality. */
77d41b28 708 regcomp_parse = exp;
a0d0e21e 709 regxend = xend;
a0d0e21e 710 regnaughty = 0;
a0d0e21e 711 regnpar = 1;
712 regsize = 0L;
713 regcode = &regdummy;
c277df42 714 regc((U8)MAGIC, (char*)regcode);
a0d0e21e 715 if (reg(0, &flags) == NULL) {
716 Safefree(regprecomp);
717 regprecomp = Nullch;
718 return(NULL);
719 }
c277df42 720 DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", regsize));
721
722 DEBUG_r(
723 if (!colorset) {
724 int i = 0;
76e3520e 725 char *s = PerlEnv_getenv("TERMCAP_COLORS");
c277df42 726
727 colorset = 1;
728 if (s) {
729 colors[0] = s = savepv(s);
730 while (++i < 4) {
731 s = strchr(s, '\t');
732 if (!s)
733 FAIL("Not enough TABs in TERMCAP_COLORS");
734 *s = '\0';
735 colors[i] = ++s;
736 }
737 } else {
738 while (i < 4)
739 colors[i++] = "";
740 }
741 /* Reset colors: */
742 PerlIO_printf(Perl_debug_log, "%s%s%s%s",
743 colors[0],colors[1],colors[2],colors[3]);
744 }
745 );
a0d0e21e 746
c277df42 747 /* Small enough for pointer-storage convention?
748 If extralen==0, this means that we will not need long jumps. */
c277df42 749 if (regsize >= 0x10000L && extralen)
750 regsize += extralen;
751 else
752 extralen = 0;
a0d0e21e 753
bbce6d69 754 /* Allocate space and initialize. */
c277df42 755 Newc(1001, r, sizeof(regexp) + (unsigned)regsize * sizeof(regnode),
756 char, regexp);
a0d0e21e 757 if (r == NULL)
758 FAIL("regexp out of space");
c277df42 759 r->refcnt = 1;
bbce6d69 760 r->prelen = xend - exp;
a0d0e21e 761 r->precomp = regprecomp;
762 r->subbeg = r->subbase = NULL;
77d41b28 763 regcomp_rx = r;
bbce6d69 764
765 /* Second pass: emit code. */
77d41b28 766 regcomp_parse = exp;
bbce6d69 767 regxend = xend;
768 regnaughty = 0;
a0d0e21e 769 regnpar = 1;
770 regcode = r->program;
c277df42 771 regc((U8)MAGIC, (char*) regcode++);
772 r->data = 0;
a0d0e21e 773 if (reg(0, &flags) == NULL)
774 return(NULL);
775
776 /* Dig out information for optimizations. */
777 pm->op_pmflags = regflags;
a0d0e21e 778 r->reganch = 0;
c277df42 779 r->regstclass = NULL;
a0d0e21e 780 r->naughty = regnaughty >= 10; /* Probably an expensive pattern. */
c277df42 781 scan = r->program + 1; /* First BRANCH. */
2779dcf1 782
783 /* XXXX To minimize changes to RE engine we always allocate
784 3-units-long substrs field. */
785 Newz(1004, r->substrs, 1, struct reg_substr_data);
786
c277df42 787 if (OP(scan) != BRANCH) { /* Only one top-level choice. */
788 scan_data_t data;
789 I32 fake;
c5254dd6 790 STRLEN longest_float_length, longest_fixed_length;
a0d0e21e 791
c277df42 792 StructCopy(&zero_scan_data, &data, scan_data_t);
a0d0e21e 793 first = scan;
c277df42 794 /* Skip introductions and multiplicators >= 1. */
a0d0e21e 795 while ((OP(first) == OPEN && (sawopen = 1)) ||
796 (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
797 (OP(first) == PLUS) ||
798 (OP(first) == MINMOD) ||
799 (regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
800 if (OP(first) == PLUS)
801 sawplus = 1;
802 else
803 first += regarglen[(U8)OP(first)];
804 first = NEXTOPER(first);
a687059c 805 }
806
a0d0e21e 807 /* Starting-point info. */
808 again:
c277df42 809 if (OP(first) == EXACT); /* Empty, get anchored substr later. */
a0d0e21e 810 else if (strchr(simple+2,OP(first)))
811 r->regstclass = first;
bbce6d69 812 else if (regkind[(U8)OP(first)] == BOUND ||
813 regkind[(U8)OP(first)] == NBOUND)
a0d0e21e 814 r->regstclass = first;
815 else if (regkind[(U8)OP(first)] == BOL) {
c277df42 816 r->reganch |= (OP(first) == MBOL ? ROPT_ANCH_MBOL: ROPT_ANCH_BOL);
a0d0e21e 817 first = NEXTOPER(first);
774d564b 818 goto again;
819 }
820 else if (OP(first) == GPOS) {
821 r->reganch |= ROPT_ANCH_GPOS;
822 first = NEXTOPER(first);
823 goto again;
a0d0e21e 824 }
825 else if ((OP(first) == STAR &&
826 regkind[(U8)OP(NEXTOPER(first))] == ANY) &&
827 !(r->reganch & ROPT_ANCH) )
828 {
829 /* turn .* into ^.* with an implied $*=1 */
774d564b 830 r->reganch |= ROPT_ANCH_BOL | ROPT_IMPLICIT;
a0d0e21e 831 first = NEXTOPER(first);
774d564b 832 goto again;
a0d0e21e 833 }
834 if (sawplus && (!sawopen || !regsawback))
835 r->reganch |= ROPT_SKIP; /* x+ must match 1st of run */
836
c277df42 837 /* Scan is after the zeroth branch, first is atomic matcher. */
838 DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n",
839 first - scan + 1));
a0d0e21e 840 /*
841 * If there's something expensive in the r.e., find the
842 * longest literal string that must appear and make it the
843 * regmust. Resolve ties in favor of later strings, since
844 * the regstart check works with the beginning of the r.e.
845 * and avoiding duplication strengthens checking. Not a
846 * strong reason, but sufficient in the absence of others.
847 * [Now we resolve ties in favor of the earlier string if
c277df42 848 * it happens that c_offset_min has been invalidated, since the
a0d0e21e 849 * earlier string may buy us something the later one won't.]
850 */
a0d0e21e 851 minlen = 0;
a687059c 852
c277df42 853 data.longest_fixed = newSVpv("",0);
854 data.longest_float = newSVpv("",0);
855 data.last_found = newSVpv("",0);
856 data.longest = &(data.longest_fixed);
857 first = scan;
858
859 minlen = study_chunk(&first, &fake, scan + regsize, /* Up to end */
860 &data, SCF_DO_SUBSTR);
861 if ( regnpar == 1 && data.longest == &(data.longest_fixed)
862 && data.last_start_min == 0 && data.last_end > 0
863 && !seen_zerolen
864 && (!(regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
865 r->reganch |= ROPT_CHECK_ALL;
866 scan_commit(&data);
867 SvREFCNT_dec(data.last_found);
868
c5254dd6 869 longest_float_length = SvCUR(data.longest_float);
870 if (longest_float_length
c277df42 871 || (data.flags & SF_FL_BEFORE_EOL
872 && (!(data.flags & SF_FL_BEFORE_MEOL)
873 || (regflags & PMf_MULTILINE)))) {
874 if (SvCUR(data.longest_fixed)
875 && data.offset_fixed == data.offset_float_min)
876 goto remove; /* Like in (a)+. */
877
878 r->float_substr = data.longest_float;
879 r->float_min_offset = data.offset_float_min;
880 r->float_max_offset = data.offset_float_max;
2779dcf1 881 fbm_compile(r->float_substr, 0);
c277df42 882 BmUSEFUL(r->float_substr) = 100;
883 if (data.flags & SF_FL_BEFORE_EOL /* Cannot have SEOL and MULTI */
884 && (!(data.flags & SF_FL_BEFORE_MEOL)
885 || (regflags & PMf_MULTILINE)))
886 SvTAIL_on(r->float_substr);
887 } else {
888 remove:
889 r->float_substr = Nullsv;
890 SvREFCNT_dec(data.longest_float);
c5254dd6 891 longest_float_length = 0;
a0d0e21e 892 }
c277df42 893
c5254dd6 894 longest_fixed_length = SvCUR(data.longest_fixed);
895 if (longest_fixed_length
c277df42 896 || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
897 && (!(data.flags & SF_FIX_BEFORE_MEOL)
898 || (regflags & PMf_MULTILINE)))) {
899 r->anchored_substr = data.longest_fixed;
900 r->anchored_offset = data.offset_fixed;
2779dcf1 901 fbm_compile(r->anchored_substr, 0);
c277df42 902 BmUSEFUL(r->anchored_substr) = 100;
903 if (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
904 && (!(data.flags & SF_FIX_BEFORE_MEOL)
905 || (regflags & PMf_MULTILINE)))
906 SvTAIL_on(r->anchored_substr);
907 } else {
908 r->anchored_substr = Nullsv;
909 SvREFCNT_dec(data.longest_fixed);
c5254dd6 910 longest_fixed_length = 0;
a0d0e21e 911 }
c277df42 912
913 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
c5254dd6 914 if (longest_fixed_length > longest_float_length) {
c277df42 915 r->check_substr = r->anchored_substr;
916 r->check_offset_min = r->check_offset_max = r->anchored_offset;
917 if (r->reganch & ROPT_ANCH_SINGLE)
918 r->reganch |= ROPT_NOSCAN;
919 } else {
920 r->check_substr = r->float_substr;
921 r->check_offset_min = data.offset_float_min;
922 r->check_offset_max = data.offset_float_max;
a0d0e21e 923 }
c277df42 924 } else {
925 /* Several toplevels. Best we can is to set minlen. */
926 I32 fake;
927
928 DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
929 scan = r->program + 1;
930 minlen = study_chunk(&scan, &fake, scan + regsize, NULL, 0);
931 r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
a0d0e21e 932 }
933
a0d0e21e 934 r->nparens = regnpar - 1;
935 r->minlen = minlen;
c277df42 936 if (regseen & REG_SEEN_GPOS)
937 r->reganch |= ROPT_GPOS_SEEN;
938 if (regseen & REG_SEEN_LOOKBEHIND)
939 r->reganch |= ROPT_LOOKBEHIND_SEEN;
ce862d02 940 if (regseen & REG_SEEN_EVAL)
941 r->reganch |= ROPT_EVAL_SEEN;
a0d0e21e 942 Newz(1002, r->startp, regnpar, char*);
943 Newz(1002, r->endp, regnpar, char*);
944 DEBUG_r(regdump(r));
945 return(r);
a687059c 946}
947
948/*
949 - reg - regular expression, i.e. main body or parenthesized thing
950 *
951 * Caller must absorb opening parenthesis.
952 *
953 * Combining parenthesis handling with the base level of regular expression
954 * is a trifle forced, but the need to tie the tails of the branches to what
955 * follows makes it hard to avoid.
956 */
76e3520e 957STATIC regnode *
8ac85365 958reg(I32 paren, I32 *flagp)
c277df42 959 /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
a687059c 960{
c277df42 961 register regnode *ret; /* Will be the head of the group. */
962 register regnode *br;
963 register regnode *lastbr;
964 register regnode *ender = 0;
a0d0e21e 965 register I32 parno = 0;
c277df42 966 I32 flags, oregflags = regflags, have_branch = 0, open = 0;
967 char c;
a0d0e21e 968
969 *flagp = HASWIDTH; /* Tentatively. */
970
971 /* Make an OPEN node, if parenthesized. */
972 if (paren) {
77d41b28 973 if (*regcomp_parse == '?') {
974 regcomp_parse++;
975 paren = *regcomp_parse++;
c277df42 976 ret = NULL; /* For look-ahead/behind. */
a0d0e21e 977 switch (paren) {
c277df42 978 case '<':
c277df42 979 regseen |= REG_SEEN_LOOKBEHIND;
77d41b28 980 if (*regcomp_parse == '!')
c277df42 981 paren = ',';
77d41b28 982 if (*regcomp_parse != '=' && *regcomp_parse != '!')
c277df42 983 goto unknown;
77d41b28 984 regcomp_parse++;
a0d0e21e 985 case '=':
986 case '!':
c277df42 987 seen_zerolen++;
988 case ':':
989 case '>':
a0d0e21e 990 break;
991 case '$':
992 case '@':
c277df42 993 FAIL2("Sequence (?%c...) not implemented", (int)paren);
a0d0e21e 994 break;
995 case '#':
77d41b28 996 while (*regcomp_parse && *regcomp_parse != ')')
997 regcomp_parse++;
998 if (*regcomp_parse != ')')
c277df42 999 FAIL("Sequence (?#... not terminated");
a0d0e21e 1000 nextchar();
1001 *flagp = TRYAGAIN;
1002 return NULL;
c277df42 1003 case '{':
1004 {
1005 dTHR;
1006 I32 count = 1, n = 0;
1007 char c;
77d41b28 1008 char *s = regcomp_parse;
c277df42 1009 SV *sv;
1010 OP_4tree *sop, *rop;
1011
1012 seen_zerolen++;
ce862d02 1013 regseen |= REG_SEEN_EVAL;
77d41b28 1014 while (count && (c = *regcomp_parse)) {
1015 if (c == '\\' && regcomp_parse[1])
1016 regcomp_parse++;
c277df42 1017 else if (c == '{')
1018 count++;
1019 else if (c == '}')
1020 count--;
77d41b28 1021 regcomp_parse++;
c277df42 1022 }
77d41b28 1023 if (*regcomp_parse != ')')
c277df42 1024 FAIL("Sequence (?{...}) not terminated or not {}-balanced");
1025 if (!SIZE_ONLY) {
1026 AV *av;
1027
77d41b28 1028 if (regcomp_parse - 1 - s)
1029 sv = newSVpv(s, regcomp_parse - 1 - s);
c277df42 1030 else
1031 sv = newSVpv("", 0);
1032
1033 rop = sv_compile_2op(sv, &sop, "re", &av);
1034
1035 n = add_data(3, "nso");
77d41b28 1036 regcomp_rx->data->data[n] = (void*)rop;
1037 regcomp_rx->data->data[n+1] = (void*)av;
1038 regcomp_rx->data->data[n+2] = (void*)sop;
c277df42 1039 SvREFCNT_dec(sv);
cc6b7395 1040 } else { /* First pass */
1041 if (tainted)
1042 FAIL("Eval-group in insecure regular expression");
c277df42 1043 }
1044
1045 nextchar();
c277df42 1046 return reganode(EVAL, n);
1047 }
1048 case '(':
1049 {
77d41b28 1050 if (regcomp_parse[0] == '?') {
1051 if (regcomp_parse[1] == '=' || regcomp_parse[1] == '!'
1052 || regcomp_parse[1] == '<'
1053 || regcomp_parse[1] == '{') { /* Lookahead or eval. */
c277df42 1054 I32 flag;
1055
1056 ret = reg_node(LOGICAL);
1057 regtail(ret, reg(1, &flag));
1058 goto insert_if;
1059 }
77d41b28 1060 } else if (regcomp_parse[0] >= '1' && regcomp_parse[0] <= '9' ) {
1061 parno = atoi(regcomp_parse++);
c277df42 1062
77d41b28 1063 while (isDIGIT(*regcomp_parse))
1064 regcomp_parse++;
c277df42 1065 ret = reganode(GROUPP, parno);
1066 if ((c = *nextchar()) != ')')
1067 FAIL2("Switch (?(number%c not recognized", c);
1068 insert_if:
1069 regtail(ret, reganode(IFTHEN, 0));
1070 br = regbranch(&flags, 1);
1071 if (br == NULL)
1072 br = reganode(LONGJMP, 0);
1073 else
1074 regtail(br, reganode(LONGJMP, 0));
1075 c = *nextchar();
1076 if (c == '|') {
1077 lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */
1078 regbranch(&flags, 1);
1079 regtail(ret, lastbr);
1080 c = *nextchar();
1081 } else
1082 lastbr = NULL;
1083 if (c != ')')
1084 FAIL("Switch (?(condition)... contains too many branches");
1085 ender = reg_node(TAIL);
1086 regtail(br, ender);
1087 if (lastbr) {
1088 regtail(lastbr, ender);
1089 regtail(NEXTOPER(NEXTOPER(lastbr)), ender);
1090 } else
1091 regtail(ret, ender);
1092 return ret;
1093 } else {
77d41b28 1094 FAIL2("Unknown condition for (?(%.2s", regcomp_parse);
c277df42 1095 }
1096 }
1b1626e4 1097 case 0:
c277df42 1098 FAIL("Sequence (? incomplete");
1b1626e4 1099 break;
a0d0e21e 1100 default:
77d41b28 1101 --regcomp_parse;
1102 while (*regcomp_parse && strchr("iogcmsx", *regcomp_parse)) {
1103 if (*regcomp_parse != 'o')
1104 pmflag(&regflags, *regcomp_parse);
1105 ++regcomp_parse;
48c036b1 1106 }
c277df42 1107 unknown:
77d41b28 1108 if (*regcomp_parse != ')')
1109 FAIL2("Sequence (?%c...) not recognized", *regcomp_parse);
a0d0e21e 1110 nextchar();
1111 *flagp = TRYAGAIN;
1112 return NULL;
1113 }
1114 }
1115 else {
1116 parno = regnpar;
1117 regnpar++;
1118 ret = reganode(OPEN, parno);
c277df42 1119 open = 1;
a0d0e21e 1120 }
1121 } else
1122 ret = NULL;
1123
1124 /* Pick up the branches, linking them together. */
c277df42 1125 br = regbranch(&flags, 1);
a0d0e21e 1126 if (br == NULL)
1127 return(NULL);
77d41b28 1128 if (*regcomp_parse == '|') {
c277df42 1129 if (!SIZE_ONLY && extralen) {
1130 reginsert(BRANCHJ, br);
1131 } else
1132 reginsert(BRANCH, br);
1133 have_branch = 1;
1134 if (SIZE_ONLY)
1135 extralen += 1; /* For BRANCHJ-BRANCH. */
1136 } else if (paren == ':') {
1137 *flagp |= flags&SIMPLE;
1138 }
1139 if (open) { /* Starts with OPEN. */
1140 regtail(ret, br); /* OPEN -> first. */
1141 } else if (paren != '?') /* Not Conditional */
a0d0e21e 1142 ret = br;
1143 if (!(flags&HASWIDTH))
1144 *flagp &= ~HASWIDTH;
1145 *flagp |= flags&SPSTART;
c277df42 1146 lastbr = br;
77d41b28 1147 while (*regcomp_parse == '|') {
c277df42 1148 if (!SIZE_ONLY && extralen) {
1149 ender = reganode(LONGJMP,0);
1150 regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
1151 }
1152 if (SIZE_ONLY)
1153 extralen += 2; /* Account for LONGJMP. */
a0d0e21e 1154 nextchar();
c277df42 1155 br = regbranch(&flags, 0);
a687059c 1156 if (br == NULL)
a0d0e21e 1157 return(NULL);
c277df42 1158 regtail(lastbr, br); /* BRANCH -> BRANCH. */
1159 lastbr = br;
a687059c 1160 if (!(flags&HASWIDTH))
a0d0e21e 1161 *flagp &= ~HASWIDTH;
a687059c 1162 *flagp |= flags&SPSTART;
a0d0e21e 1163 }
1164
c277df42 1165 if (have_branch || paren != ':') {
1166 /* Make a closing node, and hook it on the end. */
1167 switch (paren) {
1168 case ':':
1169 ender = reg_node(TAIL);
1170 break;
1171 case 1:
1172 ender = reganode(CLOSE, parno);
1173 break;
1174 case '<':
1175 case '>':
1176 case ',':
1177 case '=':
1178 case '!':
1179 ender = reg_node(SUCCEED);
1180 *flagp &= ~HASWIDTH;
1181 break;
1182 case 0:
1183 ender = reg_node(END);
1184 break;
1185 }
1186 regtail(lastbr, ender);
a0d0e21e 1187
c277df42 1188 if (have_branch) {
1189 /* Hook the tails of the branches to the closing node. */
1190 for (br = ret; br != NULL; br = regnext(br)) {
1191 regoptail(br, ender);
1192 }
1193 }
a0d0e21e 1194 }
c277df42 1195
1196 {
1197 char *p;
1198 static char parens[] = "=!<,>";
1199
1200 if (paren && (p = strchr(parens, paren))) {
1201 int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
1202 int flag = (p - parens) > 1;
1203
1204 if (paren == '>')
1205 node = SUSPEND, flag = 0;
1206 reginsert(node,ret);
c277df42 1207 ret->flags = flag;
c277df42 1208 regtail(ret, reg_node(TAIL));
1209 }
a0d0e21e 1210 }
1211
1212 /* Check for proper termination. */
77d41b28 1213 if (paren && (regcomp_parse >= regxend || *nextchar() != ')')) {
a0d0e21e 1214 FAIL("unmatched () in regexp");
77d41b28 1215 } else if (!paren && regcomp_parse < regxend) {
1216 if (*regcomp_parse == ')') {
a0d0e21e 1217 FAIL("unmatched () in regexp");
1218 } else
1219 FAIL("junk on end of regexp"); /* "Can't happen". */
1220 /* NOTREACHED */
1221 }
c277df42 1222 if (paren != 0) {
1223 regflags = oregflags;
1224 }
a687059c 1225
a0d0e21e 1226 return(ret);
a687059c 1227}
1228
1229/*
1230 - regbranch - one alternative of an | operator
1231 *
1232 * Implements the concatenation operator.
1233 */
76e3520e 1234STATIC regnode *
c277df42 1235regbranch(I32 *flagp, I32 first)
a687059c 1236{
c277df42 1237 register regnode *ret;
1238 register regnode *chain = NULL;
1239 register regnode *latest;
1240 I32 flags = 0, c = 0;
a0d0e21e 1241
c277df42 1242 if (first)
1243 ret = NULL;
1244 else {
1245 if (!SIZE_ONLY && extralen)
1246 ret = reganode(BRANCHJ,0);
1247 else
1248 ret = reg_node(BRANCH);
1249 }
1250
1251 if (!first && SIZE_ONLY)
1252 extralen += 1; /* BRANCHJ */
1253
1254 *flagp = WORST; /* Tentatively. */
a0d0e21e 1255
77d41b28 1256 regcomp_parse--;
a0d0e21e 1257 nextchar();
77d41b28 1258 while (regcomp_parse < regxend && *regcomp_parse != '|' && *regcomp_parse != ')') {
a0d0e21e 1259 flags &= ~TRYAGAIN;
1260 latest = regpiece(&flags);
1261 if (latest == NULL) {
1262 if (flags & TRYAGAIN)
1263 continue;
1264 return(NULL);
c277df42 1265 } else if (ret == NULL)
1266 ret = latest;
a0d0e21e 1267 *flagp |= flags&HASWIDTH;
c277df42 1268 if (chain == NULL) /* First piece. */
a0d0e21e 1269 *flagp |= flags&SPSTART;
1270 else {
1271 regnaughty++;
1272 regtail(chain, latest);
a687059c 1273 }
a0d0e21e 1274 chain = latest;
c277df42 1275 c++;
1276 }
1277 if (chain == NULL) { /* Loop ran zero times. */
1278 chain = reg_node(NOTHING);
1279 if (ret == NULL)
1280 ret = chain;
1281 }
1282 if (c == 1) {
1283 *flagp |= flags&SIMPLE;
a0d0e21e 1284 }
a687059c 1285
a0d0e21e 1286 return(ret);
a687059c 1287}
1288
1289/*
1290 - regpiece - something followed by possible [*+?]
1291 *
1292 * Note that the branching code sequences used for ? and the general cases
1293 * of * and + are somewhat optimized: they use the same NOTHING node as
1294 * both the endmarker for their branch list and the body of the last branch.
1295 * It might seem that this node could be dispensed with entirely, but the
1296 * endmarker role is not redundant.
1297 */
76e3520e 1298STATIC regnode *
8ac85365 1299regpiece(I32 *flagp)
a687059c 1300{
c277df42 1301 register regnode *ret;
a0d0e21e 1302 register char op;
1303 register char *next;
1304 I32 flags;
77d41b28 1305 char *origparse = regcomp_parse;
a0d0e21e 1306 char *maxpos;
1307 I32 min;
c277df42 1308 I32 max = REG_INFTY;
a0d0e21e 1309
1310 ret = regatom(&flags);
1311 if (ret == NULL) {
1312 if (flags & TRYAGAIN)
1313 *flagp |= TRYAGAIN;
1314 return(NULL);
1315 }
1316
77d41b28 1317 op = *regcomp_parse;
a0d0e21e 1318
77d41b28 1319 if (op == '{' && regcurly(regcomp_parse)) {
1320 next = regcomp_parse + 1;
a0d0e21e 1321 maxpos = Nullch;
1322 while (isDIGIT(*next) || *next == ',') {
1323 if (*next == ',') {
1324 if (maxpos)
1325 break;
1326 else
1327 maxpos = next;
a687059c 1328 }
a0d0e21e 1329 next++;
1330 }
1331 if (*next == '}') { /* got one */
1332 if (!maxpos)
1333 maxpos = next;
77d41b28 1334 regcomp_parse++;
1335 min = atoi(regcomp_parse);
a0d0e21e 1336 if (*maxpos == ',')
1337 maxpos++;
1338 else
77d41b28 1339 maxpos = regcomp_parse;
a0d0e21e 1340 max = atoi(maxpos);
1341 if (!max && *maxpos != '0')
c277df42 1342 max = REG_INFTY; /* meaning "infinity" */
1343 else if (max >= REG_INFTY)
1344 FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
77d41b28 1345 regcomp_parse = next;
a0d0e21e 1346 nextchar();
1347
1348 do_curly:
1349 if ((flags&SIMPLE)) {
1350 regnaughty += 2 + regnaughty / 2;
1351 reginsert(CURLY, ret);
1352 }
1353 else {
1354 regnaughty += 4 + regnaughty; /* compound interest */
c277df42 1355 regtail(ret, reg_node(WHILEM));
1356 if (!SIZE_ONLY && extralen) {
1357 reginsert(LONGJMP,ret);
1358 reginsert(NOTHING,ret);
1359 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
1360 }
a0d0e21e 1361 reginsert(CURLYX,ret);
c277df42 1362 if (!SIZE_ONLY && extralen)
1363 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
1364 regtail(ret, reg_node(NOTHING));
1365 if (SIZE_ONLY)
1366 extralen += 3;
a0d0e21e 1367 }
c277df42 1368 ret->flags = 0;
a0d0e21e 1369
1370 if (min > 0)
1371 *flagp = (WORST|HASWIDTH);
1372 if (max && max < min)
c277df42 1373 FAIL("Can't do {n,m} with n > m");
1374 if (!SIZE_ONLY) {
1375 ARG1_SET(ret, min);
1376 ARG2_SET(ret, max);
a687059c 1377 }
a687059c 1378
a0d0e21e 1379 goto nest_check;
a687059c 1380 }
a0d0e21e 1381 }
a687059c 1382
a0d0e21e 1383 if (!ISMULT1(op)) {
1384 *flagp = flags;
a687059c 1385 return(ret);
a0d0e21e 1386 }
bb20fd44 1387
c277df42 1388#if 0 /* Now runtime fix should be reliable. */
bb20fd44 1389 if (!(flags&HASWIDTH) && op != '?')
c277df42 1390 FAIL("regexp *+ operand could be empty");
1391#endif
bb20fd44 1392
a0d0e21e 1393 nextchar();
1394
1395 *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
1396
1397 if (op == '*' && (flags&SIMPLE)) {
1398 reginsert(STAR, ret);
c277df42 1399 ret->flags = 0;
a0d0e21e 1400 regnaughty += 4;
1401 }
1402 else if (op == '*') {
1403 min = 0;
1404 goto do_curly;
1405 } else if (op == '+' && (flags&SIMPLE)) {
1406 reginsert(PLUS, ret);
c277df42 1407 ret->flags = 0;
a0d0e21e 1408 regnaughty += 3;
1409 }
1410 else if (op == '+') {
1411 min = 1;
1412 goto do_curly;
1413 } else if (op == '?') {
1414 min = 0; max = 1;
1415 goto do_curly;
1416 }
1417 nest_check:
c277df42 1418 if (dowarn && !SIZE_ONLY && !(flags&HASWIDTH) && max > 10000) {
a0d0e21e 1419 warn("%.*s matches null string many times",
77d41b28 1420 regcomp_parse - origparse, origparse);
a0d0e21e 1421 }
1422
77d41b28 1423 if (*regcomp_parse == '?') {
a0d0e21e 1424 nextchar();
1425 reginsert(MINMOD, ret);
c277df42 1426 regtail(ret, ret + NODE_STEP_REGNODE);
a0d0e21e 1427 }
77d41b28 1428 if (ISMULT2(regcomp_parse))
a0d0e21e 1429 FAIL("nested *?+ in regexp");
1430
1431 return(ret);
a687059c 1432}
1433
1434/*
1435 - regatom - the lowest level
1436 *
1437 * Optimization: gobbles an entire sequence of ordinary characters so that
1438 * it can turn them into a single node, which is smaller to store and
1439 * faster to run. Backslashed characters are exceptions, each becoming a
1440 * separate node; the code is simpler that way and it's not worth fixing.
1441 *
1442 * [Yes, it is worth fixing, some scripts can run twice the speed.]
1443 */
76e3520e 1444STATIC regnode *
8ac85365 1445regatom(I32 *flagp)
a687059c 1446{
c277df42 1447 register regnode *ret = 0;
a0d0e21e 1448 I32 flags;
1449
1450 *flagp = WORST; /* Tentatively. */
1451
1452tryagain:
77d41b28 1453 switch (*regcomp_parse) {
a0d0e21e 1454 case '^':
c277df42 1455 seen_zerolen++;
a0d0e21e 1456 nextchar();
1457 if (regflags & PMf_MULTILINE)
c277df42 1458 ret = reg_node(MBOL);
a0d0e21e 1459 else if (regflags & PMf_SINGLELINE)
c277df42 1460 ret = reg_node(SBOL);
a0d0e21e 1461 else
c277df42 1462 ret = reg_node(BOL);
a0d0e21e 1463 break;
1464 case '$':
77d41b28 1465 if (regcomp_parse[1])
c277df42 1466 seen_zerolen++;
a0d0e21e 1467 nextchar();
1468 if (regflags & PMf_MULTILINE)
c277df42 1469 ret = reg_node(MEOL);
a0d0e21e 1470 else if (regflags & PMf_SINGLELINE)
c277df42 1471 ret = reg_node(SEOL);
a0d0e21e 1472 else
c277df42 1473 ret = reg_node(EOL);
a0d0e21e 1474 break;
1475 case '.':
1476 nextchar();
1477 if (regflags & PMf_SINGLELINE)
c277df42 1478 ret = reg_node(SANY);
a0d0e21e 1479 else
c277df42 1480 ret = reg_node(ANY);
a0d0e21e 1481 regnaughty++;
1482 *flagp |= HASWIDTH|SIMPLE;
1483 break;
1484 case '[':
77d41b28 1485 regcomp_parse++;
a0d0e21e 1486 ret = regclass();
1487 *flagp |= HASWIDTH|SIMPLE;
1488 break;
1489 case '(':
1490 nextchar();
1491 ret = reg(1, &flags);
1492 if (ret == NULL) {
1493 if (flags & TRYAGAIN)
1494 goto tryagain;
1495 return(NULL);
1496 }
c277df42 1497 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
a0d0e21e 1498 break;
1499 case '|':
1500 case ')':
1501 if (flags & TRYAGAIN) {
1502 *flagp |= TRYAGAIN;
1503 return NULL;
1504 }
77d41b28 1505 FAIL2("internal urp in regexp at /%s/", regcomp_parse);
a0d0e21e 1506 /* Supposed to be caught earlier. */
1507 break;
85afd4ae 1508 case '{':
77d41b28 1509 if (!regcurly(regcomp_parse)) {
1510 regcomp_parse++;
85afd4ae 1511 goto defchar;
1512 }
1513 /* FALL THROUGH */
a0d0e21e 1514 case '?':
1515 case '+':
1516 case '*':
3115e423 1517 FAIL("?+*{} follows nothing in regexp");
a0d0e21e 1518 break;
1519 case '\\':
77d41b28 1520 switch (*++regcomp_parse) {
a0d0e21e 1521 case 'A':
c277df42 1522 seen_zerolen++;
1523 ret = reg_node(SBOL);
a0d0e21e 1524 *flagp |= SIMPLE;
1525 nextchar();
1526 break;
1527 case 'G':
c277df42 1528 ret = reg_node(GPOS);
1529 regseen |= REG_SEEN_GPOS;
a0d0e21e 1530 *flagp |= SIMPLE;
1531 nextchar();
1532 break;
1533 case 'Z':
c277df42 1534 ret = reg_node(SEOL);
a0d0e21e 1535 *flagp |= SIMPLE;
1536 nextchar();
1537 break;
1538 case 'w':
c277df42 1539 ret = reg_node((regflags & PMf_LOCALE) ? ALNUML : ALNUM);
a0d0e21e 1540 *flagp |= HASWIDTH|SIMPLE;
1541 nextchar();
1542 break;
1543 case 'W':
c277df42 1544 ret = reg_node((regflags & PMf_LOCALE) ? NALNUML : NALNUM);
a0d0e21e 1545 *flagp |= HASWIDTH|SIMPLE;
1546 nextchar();
1547 break;
1548 case 'b':
c277df42 1549 seen_zerolen++;
1550 ret = reg_node((regflags & PMf_LOCALE) ? BOUNDL : BOUND);
a0d0e21e 1551 *flagp |= SIMPLE;
1552 nextchar();
1553 break;
1554 case 'B':
c277df42 1555 seen_zerolen++;
1556 ret = reg_node((regflags & PMf_LOCALE) ? NBOUNDL : NBOUND);
a0d0e21e 1557 *flagp |= SIMPLE;
1558 nextchar();
1559 break;
1560 case 's':
c277df42 1561 ret = reg_node((regflags & PMf_LOCALE) ? SPACEL : SPACE);
a0d0e21e 1562 *flagp |= HASWIDTH|SIMPLE;
1563 nextchar();
1564 break;
1565 case 'S':
c277df42 1566 ret = reg_node((regflags & PMf_LOCALE) ? NSPACEL : NSPACE);
a0d0e21e 1567 *flagp |= HASWIDTH|SIMPLE;
1568 nextchar();
1569 break;
1570 case 'd':
c277df42 1571 ret = reg_node(DIGIT);
a0d0e21e 1572 *flagp |= HASWIDTH|SIMPLE;
1573 nextchar();
1574 break;
1575 case 'D':
c277df42 1576 ret = reg_node(NDIGIT);
a0d0e21e 1577 *flagp |= HASWIDTH|SIMPLE;
1578 nextchar();
1579 break;
1580 case 'n':
1581 case 'r':
1582 case 't':
1583 case 'f':
1584 case 'e':
1585 case 'a':
1586 case 'x':
1587 case 'c':
1588 case '0':
1589 goto defchar;
1590 case '1': case '2': case '3': case '4':
1591 case '5': case '6': case '7': case '8': case '9':
1592 {
77d41b28 1593 I32 num = atoi(regcomp_parse);
a0d0e21e 1594
1595 if (num > 9 && num >= regnpar)
1596 goto defchar;
1597 else {
1598 regsawback = 1;
c8756f30 1599 ret = reganode((regflags & PMf_FOLD)
1600 ? ((regflags & PMf_LOCALE) ? REFFL : REFF)
1601 : REF, num);
a0d0e21e 1602 *flagp |= HASWIDTH;
77d41b28 1603 while (isDIGIT(*regcomp_parse))
1604 regcomp_parse++;
1605 regcomp_parse--;
a0d0e21e 1606 nextchar();
1607 }
1608 }
1609 break;
1610 case '\0':
77d41b28 1611 if (regcomp_parse >= regxend)
a0d0e21e 1612 FAIL("trailing \\ in regexp");
1613 /* FALL THROUGH */
1614 default:
1615 goto defchar;
1616 }
1617 break;
4633a7c4 1618
1619 case '#':
1620 if (regflags & PMf_EXTENDED) {
77d41b28 1621 while (regcomp_parse < regxend && *regcomp_parse != '\n') regcomp_parse++;
1622 if (regcomp_parse < regxend)
4633a7c4 1623 goto tryagain;
1624 }
1625 /* FALL THROUGH */
1626
a0d0e21e 1627 default: {
1628 register I32 len;
c277df42 1629 register U8 ender;
a0d0e21e 1630 register char *p;
c277df42 1631 char *oldp, *s;
a0d0e21e 1632 I32 numlen;
1633
77d41b28 1634 regcomp_parse++;
a0d0e21e 1635
1636 defchar:
c277df42 1637 ret = reg_node((regflags & PMf_FOLD)
bbce6d69 1638 ? ((regflags & PMf_LOCALE) ? EXACTFL : EXACTF)
1639 : EXACT);
161b471a 1640 s = (char *) OPERAND(ret);
c277df42 1641 regc(0, s++); /* save spot for len */
77d41b28 1642 for (len = 0, p = regcomp_parse - 1;
a0d0e21e 1643 len < 127 && p < regxend;
1644 len++)
1645 {
1646 oldp = p;
5b5a24f7 1647
1648 if (regflags & PMf_EXTENDED)
1649 p = regwhite(p, regxend);
a0d0e21e 1650 switch (*p) {
1651 case '^':
1652 case '$':
1653 case '.':
1654 case '[':
1655 case '(':
1656 case ')':
1657 case '|':
1658 goto loopdone;
1659 case '\\':
1660 switch (*++p) {
1661 case 'A':
1662 case 'G':
1663 case 'Z':
1664 case 'w':
1665 case 'W':
1666 case 'b':
1667 case 'B':
1668 case 's':
1669 case 'S':
1670 case 'd':
1671 case 'D':
1672 --p;
1673 goto loopdone;
1674 case 'n':
1675 ender = '\n';
1676 p++;
a687059c 1677 break;
a0d0e21e 1678 case 'r':
1679 ender = '\r';
1680 p++;
a687059c 1681 break;
a0d0e21e 1682 case 't':
1683 ender = '\t';
1684 p++;
a687059c 1685 break;
a0d0e21e 1686 case 'f':
1687 ender = '\f';
1688 p++;
a687059c 1689 break;
a0d0e21e 1690 case 'e':
1691 ender = '\033';
1692 p++;
a687059c 1693 break;
a0d0e21e 1694 case 'a':
1695 ender = '\007';
1696 p++;
a687059c 1697 break;
a0d0e21e 1698 case 'x':
1699 ender = scan_hex(++p, 2, &numlen);
1700 p += numlen;
a687059c 1701 break;
a0d0e21e 1702 case 'c':
1703 p++;
bbce6d69 1704 ender = UCHARAT(p++);
1705 ender = toCTRL(ender);
a687059c 1706 break;
a0d0e21e 1707 case '0': case '1': case '2': case '3':case '4':
1708 case '5': case '6': case '7': case '8':case '9':
1709 if (*p == '0' ||
1710 (isDIGIT(p[1]) && atoi(p) >= regnpar) ) {
1711 ender = scan_oct(p, 3, &numlen);
1712 p += numlen;
1713 }
1714 else {
1715 --p;
1716 goto loopdone;
a687059c 1717 }
1718 break;
a0d0e21e 1719 case '\0':
1720 if (p >= regxend)
a687059c 1721 FAIL("trailing \\ in regexp");
1722 /* FALL THROUGH */
a0d0e21e 1723 default:
1724 ender = *p++;
1725 break;
1726 }
1727 break;
a687059c 1728 default:
a0d0e21e 1729 ender = *p++;
1730 break;
a687059c 1731 }
5b5a24f7 1732 if (regflags & PMf_EXTENDED)
1733 p = regwhite(p, regxend);
a0d0e21e 1734 if (ISMULT2(p)) { /* Back off on ?+*. */
1735 if (len)
1736 p = oldp;
1737 else {
1738 len++;
c277df42 1739 regc(ender, s++);
a0d0e21e 1740 }
1741 break;
a687059c 1742 }
c277df42 1743 regc(ender, s++);
a0d0e21e 1744 }
1745 loopdone:
77d41b28 1746 regcomp_parse = p - 1;
a0d0e21e 1747 nextchar();
1748 if (len < 0)
1749 FAIL("internal disaster in regexp");
1750 if (len > 0)
1751 *flagp |= HASWIDTH;
1752 if (len == 1)
1753 *flagp |= SIMPLE;
c277df42 1754 if (!SIZE_ONLY)
a0d0e21e 1755 *OPERAND(ret) = len;
c277df42 1756 regc('\0', s++);
1757 if (SIZE_ONLY) {
c277df42 1758 regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
c277df42 1759 } else {
1760 regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode);
1761 }
a687059c 1762 }
a0d0e21e 1763 break;
1764 }
a687059c 1765
a0d0e21e 1766 return(ret);
a687059c 1767}
1768
873ef191 1769STATIC char *
8ac85365 1770regwhite(char *p, char *e)
5b5a24f7 1771{
1772 while (p < e) {
1773 if (isSPACE(*p))
1774 ++p;
1775 else if (*p == '#') {
1776 do {
1777 p++;
1778 } while (p < e && *p != '\n');
1779 }
1780 else
1781 break;
1782 }
1783 return p;
1784}
1785
76e3520e 1786STATIC regnode *
8ac85365 1787regclass(void)
a687059c 1788{
c277df42 1789 register char *opnd, *s;
8ac85365 1790 register I32 Class;
a0d0e21e 1791 register I32 lastclass = 1234;
1792 register I32 range = 0;
c277df42 1793 register regnode *ret;
a0d0e21e 1794 register I32 def;
1795 I32 numlen;
1796
161b471a 1797 s = opnd = (char *) OPERAND(regcode);
c277df42 1798 ret = reg_node(ANYOF);
8ac85365 1799 for (Class = 0; Class < 33; Class++)
c277df42 1800 regc(0, s++);
77d41b28 1801 if (*regcomp_parse == '^') { /* Complement of range. */
a0d0e21e 1802 regnaughty++;
77d41b28 1803 regcomp_parse++;
c277df42 1804 if (!SIZE_ONLY)
bbce6d69 1805 *opnd |= ANYOF_INVERT;
1806 }
c277df42 1807 if (!SIZE_ONLY) {
1808 regcode += ANY_SKIP;
bbce6d69 1809 if (regflags & PMf_FOLD)
1810 *opnd |= ANYOF_FOLD;
1811 if (regflags & PMf_LOCALE)
1812 *opnd |= ANYOF_LOCALE;
c277df42 1813 } else {
1814 regsize += ANY_SKIP;
a0d0e21e 1815 }
77d41b28 1816 if (*regcomp_parse == ']' || *regcomp_parse == '-')
a0d0e21e 1817 goto skipcond; /* allow 1st char to be ] or - */
77d41b28 1818 while (regcomp_parse < regxend && *regcomp_parse != ']') {
a0d0e21e 1819 skipcond:
77d41b28 1820 Class = UCHARAT(regcomp_parse++);
1821 if (Class == '[' && regcomp_parse + 1 < regxend &&
4599a1de 1822 /* I smell either [: or [= or [. -- POSIX has been here, right? */
77d41b28 1823 (*regcomp_parse == ':' || *regcomp_parse == '=' || *regcomp_parse == '.')) {
1824 char posixccc = *regcomp_parse;
1825 char* posixccs = regcomp_parse++;
4599a1de 1826
77d41b28 1827 while (regcomp_parse < regxend && *regcomp_parse != posixccc)
1828 regcomp_parse++;
1829 if (regcomp_parse == regxend)
4599a1de 1830 /* Grandfather lone [:, [=, [. */
77d41b28 1831 regcomp_parse = posixccs;
4599a1de 1832 else {
77d41b28 1833 regcomp_parse++; /* skip over the posixccc */
1834 if (*regcomp_parse == ']') {
4599a1de 1835 /* Not Implemented Yet.
1836 * (POSIX Extended Character Classes, that is)
1837 * The text between e.g. [: and :] would start
77d41b28 1838 * at posixccs + 1 and stop at regcomp_parse - 2. */
4599a1de 1839 if (dowarn && !SIZE_ONLY)
1840 warn("Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc);
77d41b28 1841 regcomp_parse++; /* skip over the ending ] */
4599a1de 1842 }
1843 }
1844 }
8ac85365 1845 if (Class == '\\') {
77d41b28 1846 Class = UCHARAT(regcomp_parse++);
8ac85365 1847 switch (Class) {
a0d0e21e 1848 case 'w':
ae5c130c 1849 if (!SIZE_ONLY) {
1850 if (regflags & PMf_LOCALE)
bbce6d69 1851 *opnd |= ANYOF_ALNUML;
ae5c130c 1852 else {
1853 for (Class = 0; Class < 256; Class++)
1854 if (isALNUM(Class))
1855 ANYOF_SET(opnd, Class);
1856 }
bbce6d69 1857 }
a0d0e21e 1858 lastclass = 1234;
1859 continue;
1860 case 'W':
ae5c130c 1861 if (!SIZE_ONLY) {
1862 if (regflags & PMf_LOCALE)
bbce6d69 1863 *opnd |= ANYOF_NALNUML;
ae5c130c 1864 else {
1865 for (Class = 0; Class < 256; Class++)
1866 if (!isALNUM(Class))
1867 ANYOF_SET(opnd, Class);
1868 }
bbce6d69 1869 }
a0d0e21e 1870 lastclass = 1234;
1871 continue;
1872 case 's':
ae5c130c 1873 if (!SIZE_ONLY) {
1874 if (regflags & PMf_LOCALE)
bbce6d69 1875 *opnd |= ANYOF_SPACEL;
ae5c130c 1876 else {
1877 for (Class = 0; Class < 256; Class++)
1878 if (isSPACE(Class))
1879 ANYOF_SET(opnd, Class);
1880 }
bbce6d69 1881 }
a0d0e21e 1882 lastclass = 1234;
1883 continue;
1884 case 'S':
ae5c130c 1885 if (!SIZE_ONLY) {
1886 if (regflags & PMf_LOCALE)
bbce6d69 1887 *opnd |= ANYOF_NSPACEL;
ae5c130c 1888 else {
1889 for (Class = 0; Class < 256; Class++)
1890 if (!isSPACE(Class))
1891 ANYOF_SET(opnd, Class);
1892 }
bbce6d69 1893 }
a0d0e21e 1894 lastclass = 1234;
1895 continue;
1896 case 'd':
ae5c130c 1897 if (!SIZE_ONLY) {
1898 for (Class = '0'; Class <= '9'; Class++)
1899 ANYOF_SET(opnd, Class);
1900 }
a0d0e21e 1901 lastclass = 1234;
1902 continue;
1903 case 'D':
ae5c130c 1904 if (!SIZE_ONLY) {
1905 for (Class = 0; Class < '0'; Class++)
1906 ANYOF_SET(opnd, Class);
1907 for (Class = '9' + 1; Class < 256; Class++)
1908 ANYOF_SET(opnd, Class);
1909 }
a0d0e21e 1910 lastclass = 1234;
1911 continue;
1912 case 'n':
8ac85365 1913 Class = '\n';
a0d0e21e 1914 break;
1915 case 'r':
8ac85365 1916 Class = '\r';
a0d0e21e 1917 break;
1918 case 't':
8ac85365 1919 Class = '\t';
a0d0e21e 1920 break;
1921 case 'f':
8ac85365 1922 Class = '\f';
a0d0e21e 1923 break;
1924 case 'b':
8ac85365 1925 Class = '\b';
a0d0e21e 1926 break;
1927 case 'e':
8ac85365 1928 Class = '\033';
a0d0e21e 1929 break;
1930 case 'a':
8ac85365 1931 Class = '\007';
a0d0e21e 1932 break;
1933 case 'x':
77d41b28 1934 Class = scan_hex(regcomp_parse, 2, &numlen);
1935 regcomp_parse += numlen;
a0d0e21e 1936 break;
1937 case 'c':
77d41b28 1938 Class = UCHARAT(regcomp_parse++);
8ac85365 1939 Class = toCTRL(Class);
a0d0e21e 1940 break;
1941 case '0': case '1': case '2': case '3': case '4':
1942 case '5': case '6': case '7': case '8': case '9':
77d41b28 1943 Class = scan_oct(--regcomp_parse, 3, &numlen);
1944 regcomp_parse += numlen;
a0d0e21e 1945 break;
1946 }
1947 }
1948 if (range) {
8ac85365 1949 if (lastclass > Class)
a0d0e21e 1950 FAIL("invalid [] range in regexp");
1951 range = 0;
1952 }
1953 else {
8ac85365 1954 lastclass = Class;
77d41b28 1955 if (*regcomp_parse == '-' && regcomp_parse+1 < regxend &&
1956 regcomp_parse[1] != ']') {
1957 regcomp_parse++;
a0d0e21e 1958 range = 1;
1959 continue; /* do it next time */
1960 }
a687059c 1961 }
ae5c130c 1962 if (!SIZE_ONLY) {
1963 for ( ; lastclass <= Class; lastclass++)
1964 ANYOF_SET(opnd, lastclass);
1965 }
8ac85365 1966 lastclass = Class;
a0d0e21e 1967 }
77d41b28 1968 if (*regcomp_parse != ']')
a0d0e21e 1969 FAIL("unmatched [] in regexp");
1970 nextchar();
ae5c130c 1971 /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
1972 if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) {
1973 for (Class = 0; Class < 256; ++Class) {
1974 if (ANYOF_TEST(opnd, Class)) {
1975 I32 cf = fold[Class];
1976 ANYOF_SET(opnd, cf);
1977 }
1978 }
1979 *opnd &= ~ANYOF_FOLD;
1980 }
1981 /* optimize inverted simple patterns (e.g. [^a-z]) */
1982 if (!SIZE_ONLY && (*opnd & 0xFF) == ANYOF_INVERT) {
1983 for (Class = 0; Class < 32; ++Class)
1984 opnd[1 + Class] ^= 0xFF;
1985 *opnd = 0;
1986 }
a0d0e21e 1987 return ret;
1988}
1989
76e3520e 1990STATIC char*
8ac85365 1991nextchar(void)
a0d0e21e 1992{
77d41b28 1993 char* retval = regcomp_parse++;
a0d0e21e 1994
4633a7c4 1995 for (;;) {
77d41b28 1996 if (*regcomp_parse == '(' && regcomp_parse[1] == '?' &&
1997 regcomp_parse[2] == '#') {
1998 while (*regcomp_parse && *regcomp_parse != ')')
1999 regcomp_parse++;
2000 regcomp_parse++;
4633a7c4 2001 continue;
2002 }
2003 if (regflags & PMf_EXTENDED) {
77d41b28 2004 if (isSPACE(*regcomp_parse)) {
2005 regcomp_parse++;
748a9306 2006 continue;
2007 }
77d41b28 2008 else if (*regcomp_parse == '#') {
2009 while (*regcomp_parse && *regcomp_parse != '\n')
2010 regcomp_parse++;
2011 regcomp_parse++;
748a9306 2012 continue;
2013 }
748a9306 2014 }
4633a7c4 2015 return retval;
a0d0e21e 2016 }
a687059c 2017}
2018
2019/*
c277df42 2020- reg_node - emit a node
a0d0e21e 2021*/
76e3520e 2022STATIC regnode * /* Location. */
c277df42 2023reg_node(U8 op)
a687059c 2024{
c277df42 2025 register regnode *ret;
2026 register regnode *ptr;
a687059c 2027
a0d0e21e 2028 ret = regcode;
c277df42 2029 if (SIZE_ONLY) {
2030 SIZE_ALIGN(regsize);
c277df42 2031 regsize += 1;
a0d0e21e 2032 return(ret);
2033 }
a687059c 2034
c277df42 2035 NODE_ALIGN_FILL(ret);
a0d0e21e 2036 ptr = ret;
c277df42 2037 FILL_ADVANCE_NODE(ptr, op);
a0d0e21e 2038 regcode = ptr;
a687059c 2039
a0d0e21e 2040 return(ret);
a687059c 2041}
2042
2043/*
a0d0e21e 2044- reganode - emit a node with an argument
2045*/
76e3520e 2046STATIC regnode * /* Location. */
c277df42 2047reganode(U8 op, U32 arg)
fe14fcc3 2048{
c277df42 2049 register regnode *ret;
2050 register regnode *ptr;
fe14fcc3 2051
a0d0e21e 2052 ret = regcode;
c277df42 2053 if (SIZE_ONLY) {
2054 SIZE_ALIGN(regsize);
c277df42 2055 regsize += 2;
a0d0e21e 2056 return(ret);
2057 }
fe14fcc3 2058
c277df42 2059 NODE_ALIGN_FILL(ret);
a0d0e21e 2060 ptr = ret;
c277df42 2061 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
a0d0e21e 2062 regcode = ptr;
fe14fcc3 2063
a0d0e21e 2064 return(ret);
fe14fcc3 2065}
2066
2067/*
a0d0e21e 2068- regc - emit (if appropriate) a byte of code
2069*/
76e3520e 2070STATIC void
c277df42 2071regc(U8 b, char* s)
a687059c 2072{
c277df42 2073 if (!SIZE_ONLY)
2074 *s = b;
a687059c 2075}
2076
2077/*
a0d0e21e 2078- reginsert - insert an operator in front of already-emitted operand
2079*
2080* Means relocating the operand.
2081*/
76e3520e 2082STATIC void
c277df42 2083reginsert(U8 op, regnode *opnd)
a687059c 2084{
c277df42 2085 register regnode *src;
2086 register regnode *dst;
2087 register regnode *place;
2088 register int offset = regarglen[(U8)op];
2089
2090/* (regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
2091
2092 if (SIZE_ONLY) {
2093 regsize += NODE_STEP_REGNODE + offset;
a0d0e21e 2094 return;
2095 }
a687059c 2096
a0d0e21e 2097 src = regcode;
c277df42 2098 regcode += NODE_STEP_REGNODE + offset;
a0d0e21e 2099 dst = regcode;
2100 while (src > opnd)
c277df42 2101 StructCopy(--src, --dst, regnode);
a0d0e21e 2102
2103 place = opnd; /* Op node, where operand used to be. */
c277df42 2104 src = NEXTOPER(place);
2105 FILL_ADVANCE_NODE(place, op);
2106 Zero(src, offset, regnode);
a687059c 2107}
2108
2109/*
c277df42 2110- regtail - set the next-pointer at the end of a node chain of p to val.
a0d0e21e 2111*/
76e3520e 2112STATIC void
c277df42 2113regtail(regnode *p, regnode *val)
a687059c 2114{
c277df42 2115 register regnode *scan;
2116 register regnode *temp;
a0d0e21e 2117 register I32 offset;
2118
c277df42 2119 if (SIZE_ONLY)
a0d0e21e 2120 return;
2121
2122 /* Find last node. */
2123 scan = p;
2124 for (;;) {
2125 temp = regnext(scan);
2126 if (temp == NULL)
2127 break;
2128 scan = temp;
2129 }
a687059c 2130
c277df42 2131 if (reg_off_by_arg[OP(scan)]) {
2132 ARG_SET(scan, val - scan);
2133 } else {
2134 NEXT_OFF(scan) = val - scan;
2135 }
a687059c 2136}
2137
2138/*
a0d0e21e 2139- regoptail - regtail on operand of first argument; nop if operandless
2140*/
76e3520e 2141STATIC void
c277df42 2142regoptail(regnode *p, regnode *val)
a687059c 2143{
a0d0e21e 2144 /* "Operandless" and "op != BRANCH" are synonymous in practice. */
c277df42 2145 if (p == NULL || SIZE_ONLY)
2146 return;
2147 if (regkind[(U8)OP(p)] == BRANCH) {
2148 regtail(NEXTOPER(p), val);
2149 } else if ( regkind[(U8)OP(p)] == BRANCHJ) {
2150 regtail(NEXTOPER(NEXTOPER(p)), val);
2151 } else
a0d0e21e 2152 return;
a687059c 2153}
2154
2155/*
2156 - regcurly - a little FSA that accepts {\d+,?\d*}
2157 */
79072805 2158STATIC I32
8ac85365 2159regcurly(register char *s)
a687059c 2160{
2161 if (*s++ != '{')
2162 return FALSE;
f0fcb552 2163 if (!isDIGIT(*s))
a687059c 2164 return FALSE;
f0fcb552 2165 while (isDIGIT(*s))
a687059c 2166 s++;
2167 if (*s == ',')
2168 s++;
f0fcb552 2169 while (isDIGIT(*s))
a687059c 2170 s++;
2171 if (*s != '}')
2172 return FALSE;
2173 return TRUE;
2174}
2175
a687059c 2176
76e3520e 2177STATIC regnode *
c277df42 2178dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
2179{
35ff7856 2180#ifdef DEBUGGING
c277df42 2181 register char op = EXACT; /* Arbitrary non-END op. */
2182 register regnode *next, *onode;
2183
2184 while (op != END && (!last || node < last)) {
2185 /* While that wasn't END last time... */
2186
2187 NODE_ALIGN(node);
2188 op = OP(node);
2189 if (op == CLOSE)
2190 l--;
2191 next = regnext(node);
2192 /* Where, what. */
2193 if (OP(node) == OPTIMIZED)
2194 goto after_print;
2195 regprop(sv, node);
2196 PerlIO_printf(Perl_debug_log, "%4d%*s%s", node - start,
2197 2*l + 1, "", SvPVX(sv));
2198 if (next == NULL) /* Next ptr. */
2199 PerlIO_printf(Perl_debug_log, "(0)");
2200 else
2201 PerlIO_printf(Perl_debug_log, "(%d)", next - start);
2202 (void)PerlIO_putc(Perl_debug_log, '\n');
2203 after_print:
2204 if (regkind[(U8)op] == BRANCHJ) {
2205 register regnode *nnode = (OP(next) == LONGJMP
2206 ? regnext(next)
2207 : next);
2208 if (last && nnode > last)
2209 nnode = last;
2210 node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
2211 } else if (regkind[(U8)op] == BRANCH) {
2212 node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
2213 } else if ( op == CURLY) { /* `next' might be very big: optimizer */
2214 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
2215 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
2216 } else if (regkind[(U8)op] == CURLY && op != CURLYX) {
2217 node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
2218 next, sv, l + 1);
2219 } else if ( op == PLUS || op == STAR) {
2220 node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
2221 } else if (op == ANYOF) {
2222 node = NEXTOPER(node);
2223 node += ANY_SKIP;
2224 } else if (regkind[(U8)op] == EXACT) {
2225 /* Literal string, where present. */
2226 node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode);
2227 node = NEXTOPER(node);
2228 } else {
2229 node = NEXTOPER(node);
2230 node += regarglen[(U8)op];
2231 }
2232 if (op == CURLYX || op == OPEN)
2233 l++;
2234 else if (op == WHILEM)
2235 l--;
2236 }
17c3b450 2237#endif /* DEBUGGING */
c277df42 2238 return node;
2239}
2240
a687059c 2241/*
fd181c75 2242 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
a687059c 2243 */
2244void
8ac85365 2245regdump(regexp *r)
a687059c 2246{
35ff7856 2247#ifdef DEBUGGING
46fc3d4c 2248 SV *sv = sv_newmortal();
a687059c 2249
c277df42 2250 (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
a0d0e21e 2251
2252 /* Header fields of interest. */
c277df42 2253 if (r->anchored_substr)
2254 PerlIO_printf(Perl_debug_log, "anchored `%s%s%s'%s at %d ",
2255 colors[0],
2256 SvPVX(r->anchored_substr),
2257 colors[1],
2258 SvTAIL(r->anchored_substr) ? "$" : "",
2259 r->anchored_offset);
2260 if (r->float_substr)
2261 PerlIO_printf(Perl_debug_log, "floating `%s%s%s'%s at %d..%u ",
2262 colors[0],
2263 SvPVX(r->float_substr),
2264 colors[1],
2265 SvTAIL(r->float_substr) ? "$" : "",
2266 r->float_min_offset, r->float_max_offset);
2267 if (r->check_substr)
2268 PerlIO_printf(Perl_debug_log,
2269 r->check_substr == r->float_substr
2270 ? "(checking floating" : "(checking anchored");
2271 if (r->reganch & ROPT_NOSCAN)
2272 PerlIO_printf(Perl_debug_log, " noscan");
2273 if (r->reganch & ROPT_CHECK_ALL)
2274 PerlIO_printf(Perl_debug_log, " isall");
2275 if (r->check_substr)
2276 PerlIO_printf(Perl_debug_log, ") ");
2277
46fc3d4c 2278 if (r->regstclass) {
2279 regprop(sv, r->regstclass);
2280 PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
2281 }
774d564b 2282 if (r->reganch & ROPT_ANCH) {
2283 PerlIO_printf(Perl_debug_log, "anchored");
2284 if (r->reganch & ROPT_ANCH_BOL)
2285 PerlIO_printf(Perl_debug_log, "(BOL)");
c277df42 2286 if (r->reganch & ROPT_ANCH_MBOL)
2287 PerlIO_printf(Perl_debug_log, "(MBOL)");
774d564b 2288 if (r->reganch & ROPT_ANCH_GPOS)
2289 PerlIO_printf(Perl_debug_log, "(GPOS)");
2290 PerlIO_putc(Perl_debug_log, ' ');
2291 }
c277df42 2292 if (r->reganch & ROPT_GPOS_SEEN)
2293 PerlIO_printf(Perl_debug_log, "GPOS ");
a0d0e21e 2294 if (r->reganch & ROPT_SKIP)
760ac839 2295 PerlIO_printf(Perl_debug_log, "plus ");
a0d0e21e 2296 if (r->reganch & ROPT_IMPLICIT)
760ac839 2297 PerlIO_printf(Perl_debug_log, "implicit ");
760ac839 2298 PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
ce862d02 2299 if (r->reganch & ROPT_EVAL_SEEN)
2300 PerlIO_printf(Perl_debug_log, "with eval ");
760ac839 2301 PerlIO_printf(Perl_debug_log, "\n");
17c3b450 2302#endif /* DEBUGGING */
a687059c 2303}
2304
2305/*
a0d0e21e 2306- regprop - printable representation of opcode
2307*/
46fc3d4c 2308void
c277df42 2309regprop(SV *sv, regnode *o)
a687059c 2310{
35ff7856 2311#ifdef DEBUGGING
a0d0e21e 2312 register char *p = 0;
2313
46fc3d4c 2314 sv_setpv(sv, ":");
11343788 2315 switch (OP(o)) {
a0d0e21e 2316 case BOL:
2317 p = "BOL";
2318 break;
2319 case MBOL:
2320 p = "MBOL";
2321 break;
2322 case SBOL:
2323 p = "SBOL";
2324 break;
2325 case EOL:
2326 p = "EOL";
2327 break;
2328 case MEOL:
2329 p = "MEOL";
2330 break;
2331 case SEOL:
2332 p = "SEOL";
2333 break;
2334 case ANY:
2335 p = "ANY";
2336 break;
2337 case SANY:
2338 p = "SANY";
2339 break;
2340 case ANYOF:
2341 p = "ANYOF";
2342 break;
2343 case BRANCH:
2344 p = "BRANCH";
2345 break;
bbce6d69 2346 case EXACT:
c277df42 2347 sv_catpvf(sv, "EXACT <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
bbce6d69 2348 break;
2349 case EXACTF:
c277df42 2350 sv_catpvf(sv, "EXACTF <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
bbce6d69 2351 break;
2352 case EXACTFL:
c277df42 2353 sv_catpvf(sv, "EXACTFL <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]);
a0d0e21e 2354 break;
2355 case NOTHING:
2356 p = "NOTHING";
2357 break;
c277df42 2358 case TAIL:
2359 p = "TAIL";
2360 break;
a0d0e21e 2361 case BACK:
2362 p = "BACK";
2363 break;
2364 case END:
2365 p = "END";
2366 break;
a0d0e21e 2367 case BOUND:
2368 p = "BOUND";
2369 break;
bbce6d69 2370 case BOUNDL:
2371 p = "BOUNDL";
2372 break;
a0d0e21e 2373 case NBOUND:
2374 p = "NBOUND";
2375 break;
bbce6d69 2376 case NBOUNDL:
2377 p = "NBOUNDL";
a0d0e21e 2378 break;
2379 case CURLY:
5dc0d613 2380 sv_catpvf(sv, "CURLY {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e 2381 break;
c277df42 2382 case CURLYM:
c277df42 2383 sv_catpvf(sv, "CURLYM[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
c277df42 2384 break;
2385 case CURLYN:
c277df42 2386 sv_catpvf(sv, "CURLYN[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
c277df42 2387 break;
a0d0e21e 2388 case CURLYX:
5dc0d613 2389 sv_catpvf(sv, "CURLYX {%d,%d}", ARG1(o), ARG2(o));
a0d0e21e 2390 break;
2391 case REF:
c277df42 2392 sv_catpvf(sv, "REF%d", ARG(o));
a0d0e21e 2393 break;
c8756f30 2394 case REFF:
c277df42 2395 sv_catpvf(sv, "REFF%d", ARG(o));
c8756f30 2396 break;
2397 case REFFL:
c277df42 2398 sv_catpvf(sv, "REFFL%d", ARG(o));
c8756f30 2399 break;
a0d0e21e 2400 case OPEN:
c277df42 2401 sv_catpvf(sv, "OPEN%d", ARG(o));
a0d0e21e 2402 break;
2403 case CLOSE:
c277df42 2404 sv_catpvf(sv, "CLOSE%d", ARG(o));
a0d0e21e 2405 p = NULL;
2406 break;
2407 case STAR:
2408 p = "STAR";
2409 break;
2410 case PLUS:
2411 p = "PLUS";
2412 break;
2413 case MINMOD:
2414 p = "MINMOD";
2415 break;
774d564b 2416 case GPOS:
2417 p = "GPOS";
a0d0e21e 2418 break;
2419 case UNLESSM:
c277df42 2420 sv_catpvf(sv, "UNLESSM[-%d]", o->flags);
a0d0e21e 2421 break;
2422 case IFMATCH:
c277df42 2423 sv_catpvf(sv, "IFMATCH[-%d]", o->flags);
a0d0e21e 2424 break;
2425 case SUCCEED:
2426 p = "SUCCEED";
2427 break;
2428 case WHILEM:
2429 p = "WHILEM";
2430 break;
bbce6d69 2431 case DIGIT:
2432 p = "DIGIT";
2433 break;
2434 case NDIGIT:
2435 p = "NDIGIT";
2436 break;
2437 case ALNUM:
2438 p = "ALNUM";
2439 break;
2440 case NALNUM:
2441 p = "NALNUM";
2442 break;
2443 case SPACE:
2444 p = "SPACE";
2445 break;
2446 case NSPACE:
2447 p = "NSPACE";
2448 break;
2449 case ALNUML:
2450 p = "ALNUML";
2451 break;
2452 case NALNUML:
2453 p = "NALNUML";
2454 break;
2455 case SPACEL:
2456 p = "SPACEL";
2457 break;
2458 case NSPACEL:
2459 p = "NSPACEL";
2460 break;
c277df42 2461 case EVAL:
2462 p = "EVAL";
2463 break;
2464 case LONGJMP:
2465 p = "LONGJMP";
2466 break;
2467 case BRANCHJ:
2468 p = "BRANCHJ";
2469 break;
2470 case IFTHEN:
2471 p = "IFTHEN";
2472 break;
2473 case GROUPP:
2474 sv_catpvf(sv, "GROUPP%d", ARG(o));
2475 break;
2476 case LOGICAL:
2477 p = "LOGICAL";
2478 break;
2479 case SUSPEND:
2480 p = "SUSPEND";
2481 break;
2482 case RENUM:
2483 p = "RENUM";
2484 break;
2485 case OPTIMIZED:
2486 p = "OPTIMIZED";
2487 break;
a0d0e21e 2488 default:
2489 FAIL("corrupted regexp opcode");
2490 }
46fc3d4c 2491 if (p)
2492 sv_catpv(sv, p);
17c3b450 2493#endif /* DEBUGGING */
35ff7856 2494}
a687059c 2495
2b69d0c2 2496void
8ac85365 2497pregfree(struct regexp *r)
a687059c 2498{
c277df42 2499 if (!r || (--r->refcnt > 0))
a0d0e21e 2500 return;
c277df42 2501 if (r->precomp)
a0d0e21e 2502 Safefree(r->precomp);
c277df42 2503 if (r->subbase)
a0d0e21e 2504 Safefree(r->subbase);
a193d654 2505 if (r->substrs) {
2506 if (r->anchored_substr)
2507 SvREFCNT_dec(r->anchored_substr);
2508 if (r->float_substr)
2509 SvREFCNT_dec(r->float_substr);
2779dcf1 2510 Safefree(r->substrs);
a193d654 2511 }
c277df42 2512 if (r->data) {
2513 int n = r->data->count;
2514 while (--n >= 0) {
2515 switch (r->data->what[n]) {
2516 case 's':
2517 SvREFCNT_dec((SV*)r->data->data[n]);
2518 break;
2519 case 'o':
2520 op_free((OP_4tree*)r->data->data[n]);
2521 break;
2522 case 'n':
2523 break;
2524 default:
2525 FAIL2("panic: regfree data code '%c'", r->data->what[n]);
2526 }
2527 }
2528 Safefree(r->data->what);
2529 Safefree(r->data);
a0d0e21e 2530 }
2531 Safefree(r->startp);
2532 Safefree(r->endp);
2533 Safefree(r);
a687059c 2534}
c277df42 2535
2536/*
2537 - regnext - dig the "next" pointer out of a node
2538 *
2539 * [Note, when REGALIGN is defined there are two places in regmatch()
2540 * that bypass this code for speed.]
2541 */
2542regnode *
2543regnext(register regnode *p)
2544{
2545 register I32 offset;
2546
2547 if (p == &regdummy)
2548 return(NULL);
2549
2550 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
2551 if (offset == 0)
2552 return(NULL);
2553
c277df42 2554 return(p+offset);
c277df42 2555}
2556
01f988be 2557STATIC void
c277df42 2558re_croak2(const char* pat1,const char* pat2,...)
c277df42 2559{
2560 va_list args;
2561 STRLEN l1 = strlen(pat1);
2562 STRLEN l2 = strlen(pat2);
2563 char buf[512];
2564 char *message;
2565
2566 if (l1 > 510)
2567 l1 = 510;
2568 if (l1 + l2 > 510)
2569 l2 = 510 - l1;
2570 Copy(pat1, buf, l1 , char);
2571 Copy(pat2, buf + l1, l2 , char);
3b818b81 2572 buf[l1 + l2] = '\n';
2573 buf[l1 + l2 + 1] = '\0';
c277df42 2574 va_start(args, pat2);
c277df42 2575 message = mess(buf, &args);
2576 va_end(args);
2577 l1 = strlen(message);
2578 if (l1 > 512)
2579 l1 = 512;
2580 Copy(message, buf, l1 , char);
2581 buf[l1] = '\0'; /* Overwrite \n */
2582 croak("%s", buf);
2583}
Unnamed repository; edit this file to name it for gitweb.