Commit | Line | Data |
a0d0e21e |
1 | /* regcomp.c |
2 | */ |
3 | |
4 | /* |
5 | * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee |
6 | */ |
7 | |
a687059c |
8 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
9 | * confused with the original package (see point 3 below). Thanks, Henry! |
10 | */ |
11 | |
12 | /* Additional note: this code is very heavily munged from Henry's version |
13 | * in places. In some spots I've traded clarity for efficiency, so don't |
14 | * blame Henry for some of the lack of readability. |
15 | */ |
16 | |
e50aee73 |
17 | /* The names of the functions have been changed from regcomp and |
18 | * regexec to pregcomp and pregexec in order to avoid conflicts |
19 | * with the POSIX routines of the same names. |
20 | */ |
21 | |
b9d5759e |
22 | #ifdef PERL_EXT_RE_BUILD |
23 | /* need to replace pregcomp et al, so enable that */ |
24 | # ifndef PERL_IN_XSUB_RE |
25 | # define PERL_IN_XSUB_RE |
26 | # endif |
27 | /* need access to debugger hooks */ |
f722798b |
28 | # if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING) |
b9d5759e |
29 | # define DEBUGGING |
30 | # endif |
31 | #endif |
32 | |
33 | #ifdef PERL_IN_XSUB_RE |
d06ea78c |
34 | /* We *really* need to overwrite these symbols: */ |
56953603 |
35 | # define Perl_pregcomp my_regcomp |
36 | # define Perl_regdump my_regdump |
37 | # define Perl_regprop my_regprop |
d06ea78c |
38 | # define Perl_pregfree my_regfree |
f722798b |
39 | # define Perl_re_intuit_string my_re_intuit_string |
40 | /* *These* symbols are masked to allow static link. */ |
d06ea78c |
41 | # define Perl_regnext my_regnext |
f0b8d043 |
42 | # define Perl_save_re_context my_save_re_context |
d88dccdf |
43 | # define Perl_reginitcolors my_reginitcolors |
56953603 |
44 | #endif |
45 | |
f0fcb552 |
46 | /*SUPPRESS 112*/ |
a687059c |
47 | /* |
e50aee73 |
48 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c |
49 | * |
50 | * Copyright (c) 1986 by University of Toronto. |
51 | * Written by Henry Spencer. Not derived from licensed software. |
52 | * |
53 | * Permission is granted to anyone to use this software for any |
54 | * purpose on any computer system, and to redistribute it freely, |
55 | * subject to the following restrictions: |
56 | * |
57 | * 1. The author is not responsible for the consequences of use of |
58 | * this software, no matter how awful, even if they arise |
59 | * from defects in it. |
60 | * |
61 | * 2. The origin of this software must not be misrepresented, either |
62 | * by explicit claim or by omission. |
63 | * |
64 | * 3. Altered versions must be plainly marked as such, and must not |
65 | * be misrepresented as being the original software. |
66 | * |
67 | * |
68 | **** Alterations to Henry's code are... |
69 | **** |
4eb8286e |
70 | **** Copyright (c) 1991-1999, Larry Wall |
a687059c |
71 | **** |
9ef589d8 |
72 | **** You may distribute under the terms of either the GNU General Public |
73 | **** License or the Artistic License, as specified in the README file. |
74 | |
a687059c |
75 | * |
76 | * Beware that some of this code is subtly aware of the way operator |
77 | * precedence is structured in regular expressions. Serious changes in |
78 | * regular-expression syntax might require a total rethink. |
79 | */ |
80 | #include "EXTERN.h" |
864dbfa3 |
81 | #define PERL_IN_REGCOMP_C |
a687059c |
82 | #include "perl.h" |
d06ea78c |
83 | |
b9d5759e |
84 | #ifndef PERL_IN_XSUB_RE |
d06ea78c |
85 | # include "INTERN.h" |
86 | #endif |
c277df42 |
87 | |
88 | #define REG_COMP_C |
a687059c |
89 | #include "regcomp.h" |
90 | |
d4cce5f1 |
91 | #ifdef op |
11343788 |
92 | #undef op |
d4cce5f1 |
93 | #endif /* op */ |
11343788 |
94 | |
fe14fcc3 |
95 | #ifdef MSDOS |
96 | # if defined(BUGGY_MSC6) |
97 | /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */ |
98 | # pragma optimize("a",off) |
99 | /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/ |
100 | # pragma optimize("w",on ) |
101 | # endif /* BUGGY_MSC6 */ |
102 | #endif /* MSDOS */ |
103 | |
a687059c |
104 | #ifndef STATIC |
105 | #define STATIC static |
106 | #endif |
107 | |
108 | #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?') |
109 | #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \ |
110 | ((*s) == '{' && regcurly(s))) |
2b69d0c2 |
111 | #ifdef atarist |
112 | #define PERL_META "^$.[()|?+*\\" |
113 | #else |
a687059c |
114 | #define META "^$.[()|?+*\\" |
2b69d0c2 |
115 | #endif |
a687059c |
116 | |
35c8bce7 |
117 | #ifdef SPSTART |
118 | #undef SPSTART /* dratted cpp namespace... */ |
119 | #endif |
a687059c |
120 | /* |
121 | * Flags to be passed up and down. |
122 | */ |
a687059c |
123 | #define WORST 0 /* Worst case. */ |
821b33a5 |
124 | #define HASWIDTH 0x1 /* Known to match non-null strings. */ |
a0d0e21e |
125 | #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ |
126 | #define SPSTART 0x4 /* Starts with * or +. */ |
127 | #define TRYAGAIN 0x8 /* Weeded out a declaration. */ |
a687059c |
128 | |
129 | /* |
e50aee73 |
130 | * Forward declarations for pregcomp()'s friends. |
a687059c |
131 | */ |
a0d0e21e |
132 | |
cd488c12 |
133 | static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
134 | 0, 0, 0 }; |
c277df42 |
135 | |
136 | #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL) |
137 | #define SF_BEFORE_SEOL 0x1 |
138 | #define SF_BEFORE_MEOL 0x2 |
139 | #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL) |
140 | #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL) |
141 | |
09b7f37c |
142 | #ifdef NO_UNARY_PLUS |
143 | # define SF_FIX_SHIFT_EOL (0+2) |
144 | # define SF_FL_SHIFT_EOL (0+4) |
145 | #else |
146 | # define SF_FIX_SHIFT_EOL (+2) |
147 | # define SF_FL_SHIFT_EOL (+4) |
148 | #endif |
c277df42 |
149 | |
150 | #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL) |
151 | #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL) |
152 | |
153 | #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL) |
154 | #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */ |
155 | #define SF_IS_INF 0x40 |
156 | #define SF_HAS_PAR 0x80 |
157 | #define SF_IN_PAR 0x100 |
158 | #define SF_HAS_EVAL 0x200 |
4bfe0158 |
159 | #define SCF_DO_SUBSTR 0x400 |
c277df42 |
160 | |
a0ed51b3 |
161 | #define RF_utf8 8 |
162 | #define UTF (PL_reg_flags & RF_utf8) |
163 | #define LOC (PL_regflags & PMf_LOCALE) |
164 | #define FOLD (PL_regflags & PMf_FOLD) |
165 | |
166 | #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) |
167 | #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) |
168 | |
76e3520e |
169 | STATIC void |
cea2e8a9 |
170 | S_clear_re(pTHX_ void *r) |
4327152a |
171 | { |
172 | ReREFCNT_dec((regexp *)r); |
173 | } |
174 | |
175 | STATIC void |
cea2e8a9 |
176 | S_scan_commit(pTHX_ scan_data_t *data) |
c277df42 |
177 | { |
c485e607 |
178 | dTHR; |
a0ed51b3 |
179 | STRLEN l = CHR_SVLEN(data->last_found); |
180 | STRLEN old_l = CHR_SVLEN(*data->longest); |
c277df42 |
181 | |
182 | if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) { |
183 | sv_setsv(*data->longest, data->last_found); |
184 | if (*data->longest == data->longest_fixed) { |
185 | data->offset_fixed = l ? data->last_start_min : data->pos_min; |
186 | if (data->flags & SF_BEFORE_EOL) |
187 | data->flags |
188 | |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL); |
189 | else |
190 | data->flags &= ~SF_FIX_BEFORE_EOL; |
a0ed51b3 |
191 | } |
192 | else { |
c277df42 |
193 | data->offset_float_min = l ? data->last_start_min : data->pos_min; |
194 | data->offset_float_max = (l |
195 | ? data->last_start_max |
196 | : data->pos_min + data->pos_delta); |
197 | if (data->flags & SF_BEFORE_EOL) |
198 | data->flags |
199 | |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL); |
200 | else |
201 | data->flags &= ~SF_FL_BEFORE_EOL; |
202 | } |
203 | } |
204 | SvCUR_set(data->last_found, 0); |
205 | data->last_end = -1; |
206 | data->flags &= ~SF_BEFORE_EOL; |
207 | } |
208 | |
c277df42 |
209 | /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set |
210 | to the position after last scanned or to NULL. */ |
211 | |
76e3520e |
212 | STATIC I32 |
cea2e8a9 |
213 | S_study_chunk(pTHX_ regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags) |
c277df42 |
214 | /* scanp: Start here (read-write). */ |
215 | /* deltap: Write maxlen-minlen here. */ |
216 | /* last: Stop before this one. */ |
217 | { |
5c0ca799 |
218 | dTHR; |
c277df42 |
219 | I32 min = 0, pars = 0, code; |
220 | regnode *scan = *scanp, *next; |
221 | I32 delta = 0; |
222 | int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF); |
aca2d497 |
223 | int is_inf_internal = 0; /* The studied chunk is infinite */ |
c277df42 |
224 | I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0; |
225 | scan_data_t data_fake; |
226 | |
227 | while (scan && OP(scan) != END && scan < last) { |
228 | /* Peephole optimizer: */ |
229 | |
22c35a8c |
230 | if (PL_regkind[(U8)OP(scan)] == EXACT) { |
c277df42 |
231 | regnode *n = regnext(scan); |
232 | U32 stringok = 1; |
233 | #ifdef DEBUGGING |
234 | regnode *stop = scan; |
235 | #endif |
236 | |
237 | next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; |
238 | /* Skip NOTHING, merge EXACT*. */ |
239 | while (n && |
22c35a8c |
240 | ( PL_regkind[(U8)OP(n)] == NOTHING || |
c277df42 |
241 | (stringok && (OP(n) == OP(scan)))) |
242 | && NEXT_OFF(n) |
243 | && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) { |
244 | if (OP(n) == TAIL || n > next) |
245 | stringok = 0; |
22c35a8c |
246 | if (PL_regkind[(U8)OP(n)] == NOTHING) { |
c277df42 |
247 | NEXT_OFF(scan) += NEXT_OFF(n); |
248 | next = n + NODE_STEP_REGNODE; |
249 | #ifdef DEBUGGING |
250 | if (stringok) |
251 | stop = n; |
252 | #endif |
253 | n = regnext(n); |
a0ed51b3 |
254 | } |
255 | else { |
c277df42 |
256 | int oldl = *OPERAND(scan); |
257 | regnode *nnext = regnext(n); |
258 | |
259 | if (oldl + *OPERAND(n) > U8_MAX) |
260 | break; |
261 | NEXT_OFF(scan) += NEXT_OFF(n); |
262 | *OPERAND(scan) += *OPERAND(n); |
263 | next = n + (*OPERAND(n) + 2 - 1)/sizeof(regnode) + 2; |
264 | /* Now we can overwrite *n : */ |
265 | Move(OPERAND(n) + 1, OPERAND(scan) + oldl + 1, |
266 | *OPERAND(n) + 1, char); |
267 | #ifdef DEBUGGING |
268 | if (stringok) |
269 | stop = next - 1; |
270 | #endif |
271 | n = nnext; |
272 | } |
273 | } |
274 | #ifdef DEBUGGING |
275 | /* Allow dumping */ |
276 | n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; |
277 | while (n <= stop) { |
ca04da08 |
278 | /* Purify reports a benign UMR here sometimes, because we |
279 | * don't initialize the OP() slot of a node when that node |
280 | * is occupied by just the trailing null of the string in |
281 | * an EXACT node */ |
22c35a8c |
282 | if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) { |
c277df42 |
283 | OP(n) = OPTIMIZED; |
284 | NEXT_OFF(n) = 0; |
285 | } |
286 | n++; |
287 | } |
288 | #endif |
289 | |
290 | } |
291 | if (OP(scan) != CURLYX) { |
048cfca1 |
292 | int max = (reg_off_by_arg[OP(scan)] |
293 | ? I32_MAX |
294 | /* I32 may be smaller than U16 on CRAYs! */ |
295 | : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX)); |
c277df42 |
296 | int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan)); |
297 | int noff; |
298 | regnode *n = scan; |
299 | |
300 | /* Skip NOTHING and LONGJMP. */ |
301 | while ((n = regnext(n)) |
22c35a8c |
302 | && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n))) |
c277df42 |
303 | || ((OP(n) == LONGJMP) && (noff = ARG(n)))) |
304 | && off + noff < max) |
305 | off += noff; |
306 | if (reg_off_by_arg[OP(scan)]) |
307 | ARG(scan) = off; |
308 | else |
309 | NEXT_OFF(scan) = off; |
310 | } |
311 | if (OP(scan) == BRANCH || OP(scan) == BRANCHJ |
312 | || OP(scan) == IFTHEN || OP(scan) == SUSPEND) { |
313 | next = regnext(scan); |
314 | code = OP(scan); |
315 | |
316 | if (OP(next) == code || code == IFTHEN || code == SUSPEND) { |
317 | I32 max1 = 0, min1 = I32_MAX, num = 0; |
318 | |
319 | if (flags & SCF_DO_SUBSTR) |
320 | scan_commit(data); |
321 | while (OP(scan) == code) { |
322 | I32 deltanext, minnext; |
323 | |
324 | num++; |
325 | data_fake.flags = 0; |
326 | next = regnext(scan); |
327 | scan = NEXTOPER(scan); |
328 | if (code != BRANCH) |
329 | scan = NEXTOPER(scan); |
330 | /* We suppose the run is continuous, last=next...*/ |
331 | minnext = study_chunk(&scan, &deltanext, next, |
332 | &data_fake, 0); |
333 | if (min1 > minnext) |
334 | min1 = minnext; |
335 | if (max1 < minnext + deltanext) |
336 | max1 = minnext + deltanext; |
337 | if (deltanext == I32_MAX) |
aca2d497 |
338 | is_inf = is_inf_internal = 1; |
c277df42 |
339 | scan = next; |
340 | if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) |
341 | pars++; |
405ff068 |
342 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 |
343 | data->flags |= SF_HAS_EVAL; |
344 | if (code == SUSPEND) |
345 | break; |
346 | } |
347 | if (code == IFTHEN && num < 2) /* Empty ELSE branch */ |
348 | min1 = 0; |
349 | if (flags & SCF_DO_SUBSTR) { |
350 | data->pos_min += min1; |
351 | data->pos_delta += max1 - min1; |
352 | if (max1 != min1 || is_inf) |
353 | data->longest = &(data->longest_float); |
354 | } |
355 | min += min1; |
356 | delta += max1 - min1; |
a0ed51b3 |
357 | } |
358 | else if (code == BRANCHJ) /* single branch is optimized. */ |
c277df42 |
359 | scan = NEXTOPER(NEXTOPER(scan)); |
360 | else /* single branch is optimized. */ |
361 | scan = NEXTOPER(scan); |
362 | continue; |
a0ed51b3 |
363 | } |
364 | else if (OP(scan) == EXACT) { |
365 | I32 l = *OPERAND(scan); |
366 | if (UTF) { |
367 | unsigned char *s = (unsigned char *)(OPERAND(scan)+1); |
368 | unsigned char *e = s + l; |
369 | I32 newl = 0; |
370 | while (s < e) { |
371 | newl++; |
372 | s += UTF8SKIP(s); |
373 | } |
374 | l = newl; |
375 | } |
376 | min += l; |
c277df42 |
377 | if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */ |
c277df42 |
378 | /* The code below prefers earlier match for fixed |
379 | offset, later match for variable offset. */ |
380 | if (data->last_end == -1) { /* Update the start info. */ |
381 | data->last_start_min = data->pos_min; |
382 | data->last_start_max = is_inf |
383 | ? I32_MAX : data->pos_min + data->pos_delta; |
384 | } |
a0ed51b3 |
385 | sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), *OPERAND(scan)); |
c277df42 |
386 | data->last_end = data->pos_min + l; |
387 | data->pos_min += l; /* As in the first entry. */ |
388 | data->flags &= ~SF_BEFORE_EOL; |
389 | } |
a0ed51b3 |
390 | } |
22c35a8c |
391 | else if (PL_regkind[(U8)OP(scan)] == EXACT) { |
a0ed51b3 |
392 | I32 l = *OPERAND(scan); |
c277df42 |
393 | if (flags & SCF_DO_SUBSTR) |
394 | scan_commit(data); |
a0ed51b3 |
395 | if (UTF) { |
396 | unsigned char *s = (unsigned char *)(OPERAND(scan)+1); |
397 | unsigned char *e = s + l; |
398 | I32 newl = 0; |
399 | while (s < e) { |
400 | newl++; |
401 | s += UTF8SKIP(s); |
402 | } |
403 | l = newl; |
404 | } |
405 | min += l; |
c277df42 |
406 | if (data && (flags & SCF_DO_SUBSTR)) |
a0ed51b3 |
407 | data->pos_min += l; |
408 | } |
22c35a8c |
409 | else if (strchr(PL_varies,OP(scan))) { |
c277df42 |
410 | I32 mincount, maxcount, minnext, deltanext, pos_before, fl; |
411 | regnode *oscan = scan; |
412 | |
22c35a8c |
413 | switch (PL_regkind[(U8)OP(scan)]) { |
c277df42 |
414 | case WHILEM: |
415 | scan = NEXTOPER(scan); |
416 | goto finish; |
417 | case PLUS: |
418 | if (flags & SCF_DO_SUBSTR) { |
419 | next = NEXTOPER(scan); |
420 | if (OP(next) == EXACT) { |
421 | mincount = 1; |
422 | maxcount = REG_INFTY; |
423 | next = regnext(scan); |
424 | scan = NEXTOPER(scan); |
425 | goto do_curly; |
426 | } |
427 | } |
428 | if (flags & SCF_DO_SUBSTR) |
429 | data->pos_min++; |
430 | min++; |
431 | /* Fall through. */ |
432 | case STAR: |
aca2d497 |
433 | is_inf = is_inf_internal = 1; |
c277df42 |
434 | scan = regnext(scan); |
435 | if (flags & SCF_DO_SUBSTR) { |
436 | scan_commit(data); |
437 | data->longest = &(data->longest_float); |
438 | } |
439 | goto optimize_curly_tail; |
440 | case CURLY: |
441 | mincount = ARG1(scan); |
442 | maxcount = ARG2(scan); |
443 | next = regnext(scan); |
444 | scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; |
445 | do_curly: |
446 | if (flags & SCF_DO_SUBSTR) { |
447 | if (mincount == 0) scan_commit(data); |
448 | pos_before = data->pos_min; |
449 | } |
450 | if (data) { |
451 | fl = data->flags; |
452 | data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL); |
453 | if (is_inf) |
454 | data->flags |= SF_IS_INF; |
455 | } |
456 | /* This will finish on WHILEM, setting scan, or on NULL: */ |
457 | minnext = study_chunk(&scan, &deltanext, last, data, |
458 | mincount == 0 |
459 | ? (flags & ~SCF_DO_SUBSTR) : flags); |
460 | if (!scan) /* It was not CURLYX, but CURLY. */ |
461 | scan = next; |
599cee73 |
462 | if (ckWARN(WARN_UNSAFE) && (minnext + deltanext == 0) |
821b33a5 |
463 | && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) |
17feb5d5 |
464 | && maxcount <= REG_INFTY/3) /* Complement check for big count */ |
cea2e8a9 |
465 | Perl_warner(aTHX_ WARN_UNSAFE, "Strange *+?{} on zero-length expression"); |
c277df42 |
466 | min += minnext * mincount; |
aca2d497 |
467 | is_inf_internal |= (maxcount == REG_INFTY |
468 | && (minnext + deltanext) > 0 |
469 | || deltanext == I32_MAX); |
470 | is_inf |= is_inf_internal; |
c277df42 |
471 | delta += (minnext + deltanext) * maxcount - minnext * mincount; |
472 | |
473 | /* Try powerful optimization CURLYX => CURLYN. */ |
c277df42 |
474 | if ( OP(oscan) == CURLYX && data |
475 | && data->flags & SF_IN_PAR |
476 | && !(data->flags & SF_HAS_EVAL) |
477 | && !deltanext && minnext == 1 ) { |
478 | /* Try to optimize to CURLYN. */ |
479 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; |
480 | regnode *nxt1 = nxt, *nxt2; |
481 | |
482 | /* Skip open. */ |
483 | nxt = regnext(nxt); |
22c35a8c |
484 | if (!strchr(PL_simple,OP(nxt)) |
485 | && !(PL_regkind[(U8)OP(nxt)] == EXACT |
c277df42 |
486 | && *OPERAND(nxt) == 1)) |
487 | goto nogo; |
488 | nxt2 = nxt; |
489 | nxt = regnext(nxt); |
490 | if (OP(nxt) != CLOSE) |
491 | goto nogo; |
492 | /* Now we know that nxt2 is the only contents: */ |
493 | oscan->flags = ARG(nxt); |
494 | OP(oscan) = CURLYN; |
495 | OP(nxt1) = NOTHING; /* was OPEN. */ |
496 | #ifdef DEBUGGING |
497 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ |
498 | NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */ |
499 | NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */ |
500 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ |
501 | OP(nxt + 1) = OPTIMIZED; /* was count. */ |
502 | NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */ |
503 | #endif |
504 | } |
c277df42 |
505 | nogo: |
506 | |
507 | /* Try optimization CURLYX => CURLYM. */ |
508 | if ( OP(oscan) == CURLYX && data |
c277df42 |
509 | && !(data->flags & SF_HAS_PAR) |
c277df42 |
510 | && !(data->flags & SF_HAS_EVAL) |
511 | && !deltanext ) { |
512 | /* XXXX How to optimize if data == 0? */ |
513 | /* Optimize to a simpler form. */ |
514 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */ |
515 | regnode *nxt2; |
516 | |
517 | OP(oscan) = CURLYM; |
518 | while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/ |
519 | && (OP(nxt2) != WHILEM)) |
520 | nxt = nxt2; |
521 | OP(nxt2) = SUCCEED; /* Whas WHILEM */ |
c277df42 |
522 | /* Need to optimize away parenths. */ |
523 | if (data->flags & SF_IN_PAR) { |
524 | /* Set the parenth number. */ |
525 | regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/ |
526 | |
527 | if (OP(nxt) != CLOSE) |
528 | FAIL("panic opt close"); |
529 | oscan->flags = ARG(nxt); |
530 | OP(nxt1) = OPTIMIZED; /* was OPEN. */ |
531 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ |
532 | #ifdef DEBUGGING |
533 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ |
534 | OP(nxt + 1) = OPTIMIZED; /* was count. */ |
535 | NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */ |
536 | NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */ |
537 | #endif |
538 | #if 0 |
539 | while ( nxt1 && (OP(nxt1) != WHILEM)) { |
540 | regnode *nnxt = regnext(nxt1); |
541 | |
542 | if (nnxt == nxt) { |
543 | if (reg_off_by_arg[OP(nxt1)]) |
544 | ARG_SET(nxt1, nxt2 - nxt1); |
545 | else if (nxt2 - nxt1 < U16_MAX) |
546 | NEXT_OFF(nxt1) = nxt2 - nxt1; |
547 | else |
548 | OP(nxt) = NOTHING; /* Cannot beautify */ |
549 | } |
550 | nxt1 = nnxt; |
551 | } |
552 | #endif |
553 | /* Optimize again: */ |
554 | study_chunk(&nxt1, &deltanext, nxt, NULL, 0); |
a0ed51b3 |
555 | } |
556 | else |
c277df42 |
557 | oscan->flags = 0; |
c277df42 |
558 | } |
559 | if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) |
560 | pars++; |
561 | if (flags & SCF_DO_SUBSTR) { |
562 | SV *last_str = Nullsv; |
563 | int counted = mincount != 0; |
564 | |
565 | if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */ |
566 | I32 b = pos_before >= data->last_start_min |
567 | ? pos_before : data->last_start_min; |
568 | STRLEN l; |
569 | char *s = SvPV(data->last_found, l); |
a0ed51b3 |
570 | I32 old = b - data->last_start_min; |
571 | |
572 | if (UTF) |
573 | old = utf8_hop((U8*)s, old) - (U8*)s; |
c277df42 |
574 | |
a0ed51b3 |
575 | l -= old; |
c277df42 |
576 | /* Get the added string: */ |
79cb57f6 |
577 | last_str = newSVpvn(s + old, l); |
c277df42 |
578 | if (deltanext == 0 && pos_before == b) { |
579 | /* What was added is a constant string */ |
580 | if (mincount > 1) { |
581 | SvGROW(last_str, (mincount * l) + 1); |
582 | repeatcpy(SvPVX(last_str) + l, |
583 | SvPVX(last_str), l, mincount - 1); |
584 | SvCUR(last_str) *= mincount; |
585 | /* Add additional parts. */ |
586 | SvCUR_set(data->last_found, |
587 | SvCUR(data->last_found) - l); |
588 | sv_catsv(data->last_found, last_str); |
589 | data->last_end += l * (mincount - 1); |
590 | } |
591 | } |
592 | } |
593 | /* It is counted once already... */ |
594 | data->pos_min += minnext * (mincount - counted); |
595 | data->pos_delta += - counted * deltanext + |
596 | (minnext + deltanext) * maxcount - minnext * mincount; |
597 | if (mincount != maxcount) { |
598 | scan_commit(data); |
599 | if (mincount && last_str) { |
600 | sv_setsv(data->last_found, last_str); |
601 | data->last_end = data->pos_min; |
602 | data->last_start_min = |
a0ed51b3 |
603 | data->pos_min - CHR_SVLEN(last_str); |
c277df42 |
604 | data->last_start_max = is_inf |
605 | ? I32_MAX |
606 | : data->pos_min + data->pos_delta |
a0ed51b3 |
607 | - CHR_SVLEN(last_str); |
c277df42 |
608 | } |
609 | data->longest = &(data->longest_float); |
610 | } |
aca2d497 |
611 | SvREFCNT_dec(last_str); |
c277df42 |
612 | } |
405ff068 |
613 | if (data && (fl & SF_HAS_EVAL)) |
c277df42 |
614 | data->flags |= SF_HAS_EVAL; |
615 | optimize_curly_tail: |
c277df42 |
616 | if (OP(oscan) != CURLYX) { |
22c35a8c |
617 | while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING |
c277df42 |
618 | && NEXT_OFF(next)) |
619 | NEXT_OFF(oscan) += NEXT_OFF(next); |
620 | } |
c277df42 |
621 | continue; |
622 | default: /* REF only? */ |
623 | if (flags & SCF_DO_SUBSTR) { |
624 | scan_commit(data); |
625 | data->longest = &(data->longest_float); |
626 | } |
aca2d497 |
627 | is_inf = is_inf_internal = 1; |
c277df42 |
628 | break; |
629 | } |
a0ed51b3 |
630 | } |
22c35a8c |
631 | else if (strchr(PL_simple,OP(scan)) || PL_regkind[(U8)OP(scan)] == ANYUTF8) { |
c277df42 |
632 | if (flags & SCF_DO_SUBSTR) { |
633 | scan_commit(data); |
634 | data->pos_min++; |
635 | } |
636 | min++; |
a0ed51b3 |
637 | } |
22c35a8c |
638 | else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) { |
c277df42 |
639 | data->flags |= (OP(scan) == MEOL |
640 | ? SF_BEFORE_MEOL |
641 | : SF_BEFORE_SEOL); |
a0ed51b3 |
642 | } |
22c35a8c |
643 | else if (PL_regkind[(U8)OP(scan)] == BRANCHJ |
c277df42 |
644 | && (scan->flags || data) |
645 | && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) { |
646 | I32 deltanext, minnext; |
647 | regnode *nscan; |
648 | |
649 | data_fake.flags = 0; |
650 | next = regnext(scan); |
651 | nscan = NEXTOPER(NEXTOPER(scan)); |
652 | minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0); |
653 | if (scan->flags) { |
654 | if (deltanext) { |
655 | FAIL("variable length lookbehind not implemented"); |
a0ed51b3 |
656 | } |
657 | else if (minnext > U8_MAX) { |
c277df42 |
658 | FAIL2("lookbehind longer than %d not implemented", U8_MAX); |
659 | } |
660 | scan->flags = minnext; |
661 | } |
662 | if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) |
663 | pars++; |
405ff068 |
664 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 |
665 | data->flags |= SF_HAS_EVAL; |
a0ed51b3 |
666 | } |
667 | else if (OP(scan) == OPEN) { |
c277df42 |
668 | pars++; |
a0ed51b3 |
669 | } |
670 | else if (OP(scan) == CLOSE && ARG(scan) == is_par) { |
c277df42 |
671 | next = regnext(scan); |
672 | |
673 | if ( next && (OP(next) != WHILEM) && next < last) |
c277df42 |
674 | is_par = 0; /* Disable optimization */ |
a0ed51b3 |
675 | } |
676 | else if (OP(scan) == EVAL) { |
c277df42 |
677 | if (data) |
678 | data->flags |= SF_HAS_EVAL; |
679 | } |
0f5d15d6 |
680 | else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */ |
681 | if (flags & SCF_DO_SUBSTR) { |
682 | scan_commit(data); |
683 | data->longest = &(data->longest_float); |
684 | } |
685 | is_inf = is_inf_internal = 1; |
686 | } |
c277df42 |
687 | /* Else: zero-length, ignore. */ |
688 | scan = regnext(scan); |
689 | } |
690 | |
691 | finish: |
692 | *scanp = scan; |
aca2d497 |
693 | *deltap = is_inf_internal ? I32_MAX : delta; |
c277df42 |
694 | if (flags & SCF_DO_SUBSTR && is_inf) |
695 | data->pos_delta = I32_MAX - data->pos_min; |
696 | if (is_par > U8_MAX) |
697 | is_par = 0; |
698 | if (is_par && pars==1 && data) { |
699 | data->flags |= SF_IN_PAR; |
700 | data->flags &= ~SF_HAS_PAR; |
a0ed51b3 |
701 | } |
702 | else if (pars && data) { |
c277df42 |
703 | data->flags |= SF_HAS_PAR; |
704 | data->flags &= ~SF_IN_PAR; |
705 | } |
706 | return min; |
707 | } |
708 | |
76e3520e |
709 | STATIC I32 |
cea2e8a9 |
710 | S_add_data(pTHX_ I32 n, char *s) |
c277df42 |
711 | { |
5c0ca799 |
712 | dTHR; |
3280af22 |
713 | if (PL_regcomp_rx->data) { |
714 | Renewc(PL_regcomp_rx->data, |
715 | sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1), |
c277df42 |
716 | char, struct reg_data); |
3280af22 |
717 | Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8); |
718 | PL_regcomp_rx->data->count += n; |
a0ed51b3 |
719 | } |
720 | else { |
3280af22 |
721 | Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1), |
c277df42 |
722 | char, struct reg_data); |
3280af22 |
723 | New(1208, PL_regcomp_rx->data->what, n, U8); |
724 | PL_regcomp_rx->data->count = n; |
c277df42 |
725 | } |
3280af22 |
726 | Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8); |
727 | return PL_regcomp_rx->data->count - n; |
c277df42 |
728 | } |
729 | |
d88dccdf |
730 | void |
864dbfa3 |
731 | Perl_reginitcolors(pTHX) |
d88dccdf |
732 | { |
733 | dTHR; |
734 | int i = 0; |
735 | char *s = PerlEnv_getenv("PERL_RE_COLORS"); |
736 | |
737 | if (s) { |
738 | PL_colors[0] = s = savepv(s); |
739 | while (++i < 6) { |
740 | s = strchr(s, '\t'); |
741 | if (s) { |
742 | *s = '\0'; |
743 | PL_colors[i] = ++s; |
744 | } |
745 | else |
c712d376 |
746 | PL_colors[i] = s = ""; |
d88dccdf |
747 | } |
748 | } else { |
749 | while (i < 6) |
750 | PL_colors[i++] = ""; |
751 | } |
752 | PL_colorset = 1; |
753 | } |
754 | |
a687059c |
755 | /* |
e50aee73 |
756 | - pregcomp - compile a regular expression into internal code |
a687059c |
757 | * |
758 | * We can't allocate space until we know how big the compiled form will be, |
759 | * but we can't compile it (and thus know how big it is) until we've got a |
760 | * place to put the code. So we cheat: we compile it twice, once with code |
761 | * generation turned off and size counting turned on, and once "for real". |
762 | * This also means that we don't allocate space until we are sure that the |
763 | * thing really will compile successfully, and we never have to move the |
764 | * code and thus invalidate pointers into it. (Note that it has to be in |
765 | * one piece because free() must be able to free it all.) [NB: not true in perl] |
766 | * |
767 | * Beware that the optimization-preparation code in here knows about some |
768 | * of the structure of the compiled regexp. [I'll say.] |
769 | */ |
770 | regexp * |
864dbfa3 |
771 | Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm) |
a687059c |
772 | { |
5c0ca799 |
773 | dTHR; |
a0d0e21e |
774 | register regexp *r; |
c277df42 |
775 | regnode *scan; |
776 | SV **longest; |
777 | SV *longest_fixed; |
778 | SV *longest_float; |
779 | regnode *first; |
a0d0e21e |
780 | I32 flags; |
a0d0e21e |
781 | I32 minlen = 0; |
782 | I32 sawplus = 0; |
783 | I32 sawopen = 0; |
784 | |
785 | if (exp == NULL) |
c277df42 |
786 | FAIL("NULL regexp argument"); |
a0d0e21e |
787 | |
e24b16f9 |
788 | if (PL_curcop == &PL_compiling ? (PL_hints & HINT_UTF8) : IN_UTF8) |
a0ed51b3 |
789 | PL_reg_flags |= RF_utf8; |
790 | else |
791 | PL_reg_flags = 0; |
792 | |
3280af22 |
793 | PL_regprecomp = savepvn(exp, xend - exp); |
35ef4773 |
794 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
795 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling%s RE `%s%*s%s'\n", |
d88dccdf |
796 | PL_colors[4],PL_colors[5],PL_colors[0], |
797 | xend - exp, PL_regprecomp, PL_colors[1])); |
3280af22 |
798 | PL_regflags = pm->op_pmflags; |
799 | PL_regsawback = 0; |
bbce6d69 |
800 | |
3280af22 |
801 | PL_regseen = 0; |
802 | PL_seen_zerolen = *exp == '^' ? -1 : 0; |
803 | PL_seen_evals = 0; |
804 | PL_extralen = 0; |
c277df42 |
805 | |
bbce6d69 |
806 | /* First pass: determine size, legality. */ |
3280af22 |
807 | PL_regcomp_parse = exp; |
808 | PL_regxend = xend; |
809 | PL_regnaughty = 0; |
810 | PL_regnpar = 1; |
811 | PL_regsize = 0L; |
812 | PL_regcode = &PL_regdummy; |
22c35a8c |
813 | regc((U8)REG_MAGIC, (char*)PL_regcode); |
a0d0e21e |
814 | if (reg(0, &flags) == NULL) { |
3280af22 |
815 | Safefree(PL_regprecomp); |
816 | PL_regprecomp = Nullch; |
a0d0e21e |
817 | return(NULL); |
818 | } |
3280af22 |
819 | DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", PL_regsize)); |
c277df42 |
820 | |
c277df42 |
821 | /* Small enough for pointer-storage convention? |
822 | If extralen==0, this means that we will not need long jumps. */ |
3280af22 |
823 | if (PL_regsize >= 0x10000L && PL_extralen) |
824 | PL_regsize += PL_extralen; |
c277df42 |
825 | else |
3280af22 |
826 | PL_extralen = 0; |
a0d0e21e |
827 | |
bbce6d69 |
828 | /* Allocate space and initialize. */ |
3280af22 |
829 | Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), |
c277df42 |
830 | char, regexp); |
a0d0e21e |
831 | if (r == NULL) |
832 | FAIL("regexp out of space"); |
c277df42 |
833 | r->refcnt = 1; |
bbce6d69 |
834 | r->prelen = xend - exp; |
3280af22 |
835 | r->precomp = PL_regprecomp; |
cf93c79d |
836 | r->subbeg = NULL; |
837 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; |
4327152a |
838 | r->nparens = PL_regnpar - 1; /* set early to validate backrefs */ |
839 | |
840 | r->substrs = 0; /* Useful during FAIL. */ |
841 | r->startp = 0; /* Useful during FAIL. */ |
842 | r->endp = 0; /* Useful during FAIL. */ |
843 | |
3280af22 |
844 | PL_regcomp_rx = r; |
bbce6d69 |
845 | |
846 | /* Second pass: emit code. */ |
3280af22 |
847 | PL_regcomp_parse = exp; |
848 | PL_regxend = xend; |
849 | PL_regnaughty = 0; |
850 | PL_regnpar = 1; |
851 | PL_regcode = r->program; |
2cd61cdb |
852 | /* Store the count of eval-groups for security checks: */ |
3280af22 |
853 | PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals); |
22c35a8c |
854 | regc((U8)REG_MAGIC, (char*) PL_regcode++); |
c277df42 |
855 | r->data = 0; |
a0d0e21e |
856 | if (reg(0, &flags) == NULL) |
857 | return(NULL); |
858 | |
859 | /* Dig out information for optimizations. */ |
cf93c79d |
860 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */ |
3280af22 |
861 | pm->op_pmflags = PL_regflags; |
a0ed51b3 |
862 | if (UTF) |
863 | r->reganch |= ROPT_UTF8; |
c277df42 |
864 | r->regstclass = NULL; |
a0ed51b3 |
865 | if (PL_regnaughty >= 10) /* Probably an expensive pattern. */ |
866 | r->reganch |= ROPT_NAUGHTY; |
c277df42 |
867 | scan = r->program + 1; /* First BRANCH. */ |
2779dcf1 |
868 | |
869 | /* XXXX To minimize changes to RE engine we always allocate |
870 | 3-units-long substrs field. */ |
871 | Newz(1004, r->substrs, 1, struct reg_substr_data); |
872 | |
c277df42 |
873 | if (OP(scan) != BRANCH) { /* Only one top-level choice. */ |
874 | scan_data_t data; |
875 | I32 fake; |
c5254dd6 |
876 | STRLEN longest_float_length, longest_fixed_length; |
a0d0e21e |
877 | |
c277df42 |
878 | StructCopy(&zero_scan_data, &data, scan_data_t); |
a0d0e21e |
879 | first = scan; |
c277df42 |
880 | /* Skip introductions and multiplicators >= 1. */ |
a0d0e21e |
881 | while ((OP(first) == OPEN && (sawopen = 1)) || |
882 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || |
883 | (OP(first) == PLUS) || |
884 | (OP(first) == MINMOD) || |
22c35a8c |
885 | (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
a0d0e21e |
886 | if (OP(first) == PLUS) |
887 | sawplus = 1; |
888 | else |
889 | first += regarglen[(U8)OP(first)]; |
890 | first = NEXTOPER(first); |
a687059c |
891 | } |
892 | |
a0d0e21e |
893 | /* Starting-point info. */ |
894 | again: |
c277df42 |
895 | if (OP(first) == EXACT); /* Empty, get anchored substr later. */ |
22c35a8c |
896 | else if (strchr(PL_simple+4,OP(first))) |
a0d0e21e |
897 | r->regstclass = first; |
22c35a8c |
898 | else if (PL_regkind[(U8)OP(first)] == BOUND || |
899 | PL_regkind[(U8)OP(first)] == NBOUND) |
a0d0e21e |
900 | r->regstclass = first; |
22c35a8c |
901 | else if (PL_regkind[(U8)OP(first)] == BOL) { |
f722798b |
902 | r->reganch |= (OP(first) == MBOL |
903 | ? ROPT_ANCH_MBOL |
904 | : (OP(first) == SBOL |
905 | ? ROPT_ANCH_SBOL |
906 | : ROPT_ANCH_BOL)); |
a0d0e21e |
907 | first = NEXTOPER(first); |
774d564b |
908 | goto again; |
909 | } |
910 | else if (OP(first) == GPOS) { |
911 | r->reganch |= ROPT_ANCH_GPOS; |
912 | first = NEXTOPER(first); |
913 | goto again; |
a0d0e21e |
914 | } |
915 | else if ((OP(first) == STAR && |
22c35a8c |
916 | PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) && |
a0d0e21e |
917 | !(r->reganch & ROPT_ANCH) ) |
918 | { |
919 | /* turn .* into ^.* with an implied $*=1 */ |
f722798b |
920 | int type = OP(NEXTOPER(first)); |
921 | |
922 | if (type == REG_ANY || type == ANYUTF8) |
923 | type = ROPT_ANCH_MBOL; |
924 | else |
925 | type = ROPT_ANCH_SBOL; |
926 | |
927 | r->reganch |= type | ROPT_IMPLICIT; |
a0d0e21e |
928 | first = NEXTOPER(first); |
774d564b |
929 | goto again; |
a0d0e21e |
930 | } |
f722798b |
931 | if (sawplus && (!sawopen || !PL_regsawback) |
932 | && !(PL_regseen & REG_SEEN_EVAL)) /* May examine pos and $& */ |
933 | /* x+ must match at the 1st pos of run of x's */ |
934 | r->reganch |= ROPT_SKIP; |
a0d0e21e |
935 | |
c277df42 |
936 | /* Scan is after the zeroth branch, first is atomic matcher. */ |
937 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n", |
938 | first - scan + 1)); |
a0d0e21e |
939 | /* |
940 | * If there's something expensive in the r.e., find the |
941 | * longest literal string that must appear and make it the |
942 | * regmust. Resolve ties in favor of later strings, since |
943 | * the regstart check works with the beginning of the r.e. |
944 | * and avoiding duplication strengthens checking. Not a |
945 | * strong reason, but sufficient in the absence of others. |
946 | * [Now we resolve ties in favor of the earlier string if |
c277df42 |
947 | * it happens that c_offset_min has been invalidated, since the |
a0d0e21e |
948 | * earlier string may buy us something the later one won't.] |
949 | */ |
a0d0e21e |
950 | minlen = 0; |
a687059c |
951 | |
79cb57f6 |
952 | data.longest_fixed = newSVpvn("",0); |
953 | data.longest_float = newSVpvn("",0); |
954 | data.last_found = newSVpvn("",0); |
c277df42 |
955 | data.longest = &(data.longest_fixed); |
956 | first = scan; |
957 | |
3280af22 |
958 | minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */ |
c277df42 |
959 | &data, SCF_DO_SUBSTR); |
3280af22 |
960 | if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed) |
c277df42 |
961 | && data.last_start_min == 0 && data.last_end > 0 |
3280af22 |
962 | && !PL_seen_zerolen |
963 | && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS))) |
c277df42 |
964 | r->reganch |= ROPT_CHECK_ALL; |
965 | scan_commit(&data); |
966 | SvREFCNT_dec(data.last_found); |
967 | |
a0ed51b3 |
968 | longest_float_length = CHR_SVLEN(data.longest_float); |
c5254dd6 |
969 | if (longest_float_length |
c277df42 |
970 | || (data.flags & SF_FL_BEFORE_EOL |
971 | && (!(data.flags & SF_FL_BEFORE_MEOL) |
3280af22 |
972 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d |
973 | int t; |
974 | |
a0ed51b3 |
975 | if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */ |
aca2d497 |
976 | && data.offset_fixed == data.offset_float_min |
977 | && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)) |
978 | goto remove_float; /* As in (a)+. */ |
979 | |
c277df42 |
980 | r->float_substr = data.longest_float; |
981 | r->float_min_offset = data.offset_float_min; |
982 | r->float_max_offset = data.offset_float_max; |
cf93c79d |
983 | t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */ |
984 | && (!(data.flags & SF_FL_BEFORE_MEOL) |
985 | || (PL_regflags & PMf_MULTILINE))); |
986 | fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0); |
a0ed51b3 |
987 | } |
988 | else { |
aca2d497 |
989 | remove_float: |
c277df42 |
990 | r->float_substr = Nullsv; |
991 | SvREFCNT_dec(data.longest_float); |
c5254dd6 |
992 | longest_float_length = 0; |
a0d0e21e |
993 | } |
c277df42 |
994 | |
a0ed51b3 |
995 | longest_fixed_length = CHR_SVLEN(data.longest_fixed); |
c5254dd6 |
996 | if (longest_fixed_length |
c277df42 |
997 | || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
998 | && (!(data.flags & SF_FIX_BEFORE_MEOL) |
3280af22 |
999 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d |
1000 | int t; |
1001 | |
c277df42 |
1002 | r->anchored_substr = data.longest_fixed; |
1003 | r->anchored_offset = data.offset_fixed; |
cf93c79d |
1004 | t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1005 | && (!(data.flags & SF_FIX_BEFORE_MEOL) |
1006 | || (PL_regflags & PMf_MULTILINE))); |
1007 | fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0); |
a0ed51b3 |
1008 | } |
1009 | else { |
c277df42 |
1010 | r->anchored_substr = Nullsv; |
1011 | SvREFCNT_dec(data.longest_fixed); |
c5254dd6 |
1012 | longest_fixed_length = 0; |
a0d0e21e |
1013 | } |
c277df42 |
1014 | |
1015 | /* A temporary algorithm prefers floated substr to fixed one to dig more info. */ |
c5254dd6 |
1016 | if (longest_fixed_length > longest_float_length) { |
c277df42 |
1017 | r->check_substr = r->anchored_substr; |
1018 | r->check_offset_min = r->check_offset_max = r->anchored_offset; |
1019 | if (r->reganch & ROPT_ANCH_SINGLE) |
1020 | r->reganch |= ROPT_NOSCAN; |
a0ed51b3 |
1021 | } |
1022 | else { |
c277df42 |
1023 | r->check_substr = r->float_substr; |
1024 | r->check_offset_min = data.offset_float_min; |
1025 | r->check_offset_max = data.offset_float_max; |
a0d0e21e |
1026 | } |
f722798b |
1027 | if (r->check_substr) { |
1028 | r->reganch |= RE_USE_INTUIT; |
1029 | if (SvTAIL(r->check_substr)) |
1030 | r->reganch |= RE_INTUIT_TAIL; |
1031 | } |
a0ed51b3 |
1032 | } |
1033 | else { |
c277df42 |
1034 | /* Several toplevels. Best we can is to set minlen. */ |
1035 | I32 fake; |
1036 | |
1037 | DEBUG_r(PerlIO_printf(Perl_debug_log, "\n")); |
1038 | scan = r->program + 1; |
3280af22 |
1039 | minlen = study_chunk(&scan, &fake, scan + PL_regsize, NULL, 0); |
c277df42 |
1040 | r->check_substr = r->anchored_substr = r->float_substr = Nullsv; |
a0d0e21e |
1041 | } |
1042 | |
a0d0e21e |
1043 | r->minlen = minlen; |
3280af22 |
1044 | if (PL_regseen & REG_SEEN_GPOS) |
c277df42 |
1045 | r->reganch |= ROPT_GPOS_SEEN; |
3280af22 |
1046 | if (PL_regseen & REG_SEEN_LOOKBEHIND) |
c277df42 |
1047 | r->reganch |= ROPT_LOOKBEHIND_SEEN; |
3280af22 |
1048 | if (PL_regseen & REG_SEEN_EVAL) |
ce862d02 |
1049 | r->reganch |= ROPT_EVAL_SEEN; |
cf93c79d |
1050 | Newz(1002, r->startp, PL_regnpar, I32); |
1051 | Newz(1002, r->endp, PL_regnpar, I32); |
a0d0e21e |
1052 | DEBUG_r(regdump(r)); |
1053 | return(r); |
a687059c |
1054 | } |
1055 | |
1056 | /* |
1057 | - reg - regular expression, i.e. main body or parenthesized thing |
1058 | * |
1059 | * Caller must absorb opening parenthesis. |
1060 | * |
1061 | * Combining parenthesis handling with the base level of regular expression |
1062 | * is a trifle forced, but the need to tie the tails of the branches to what |
1063 | * follows makes it hard to avoid. |
1064 | */ |
76e3520e |
1065 | STATIC regnode * |
cea2e8a9 |
1066 | S_reg(pTHX_ I32 paren, I32 *flagp) |
c277df42 |
1067 | /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */ |
a687059c |
1068 | { |
5c0ca799 |
1069 | dTHR; |
c277df42 |
1070 | register regnode *ret; /* Will be the head of the group. */ |
1071 | register regnode *br; |
1072 | register regnode *lastbr; |
1073 | register regnode *ender = 0; |
a0d0e21e |
1074 | register I32 parno = 0; |
3280af22 |
1075 | I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0; |
c277df42 |
1076 | char c; |
a0d0e21e |
1077 | |
821b33a5 |
1078 | *flagp = 0; /* Tentatively. */ |
a0d0e21e |
1079 | |
1080 | /* Make an OPEN node, if parenthesized. */ |
1081 | if (paren) { |
3280af22 |
1082 | if (*PL_regcomp_parse == '?') { |
ca9dfc88 |
1083 | U16 posflags = 0, negflags = 0; |
1084 | U16 *flagsp = &posflags; |
0f5d15d6 |
1085 | int logical = 0; |
ca9dfc88 |
1086 | |
3280af22 |
1087 | PL_regcomp_parse++; |
1088 | paren = *PL_regcomp_parse++; |
c277df42 |
1089 | ret = NULL; /* For look-ahead/behind. */ |
a0d0e21e |
1090 | switch (paren) { |
c277df42 |
1091 | case '<': |
3280af22 |
1092 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
1093 | if (*PL_regcomp_parse == '!') |
c277df42 |
1094 | paren = ','; |
3280af22 |
1095 | if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!') |
c277df42 |
1096 | goto unknown; |
3280af22 |
1097 | PL_regcomp_parse++; |
a0d0e21e |
1098 | case '=': |
1099 | case '!': |
3280af22 |
1100 | PL_seen_zerolen++; |
c277df42 |
1101 | case ':': |
1102 | case '>': |
a0d0e21e |
1103 | break; |
1104 | case '$': |
1105 | case '@': |
c277df42 |
1106 | FAIL2("Sequence (?%c...) not implemented", (int)paren); |
a0d0e21e |
1107 | break; |
1108 | case '#': |
3280af22 |
1109 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
1110 | PL_regcomp_parse++; |
1111 | if (*PL_regcomp_parse != ')') |
c277df42 |
1112 | FAIL("Sequence (?#... not terminated"); |
a0d0e21e |
1113 | nextchar(); |
1114 | *flagp = TRYAGAIN; |
1115 | return NULL; |
0f5d15d6 |
1116 | case 'p': |
1117 | logical = 1; |
1118 | paren = *PL_regcomp_parse++; |
1119 | /* FALL THROUGH */ |
c277df42 |
1120 | case '{': |
1121 | { |
1122 | dTHR; |
1123 | I32 count = 1, n = 0; |
1124 | char c; |
3280af22 |
1125 | char *s = PL_regcomp_parse; |
c277df42 |
1126 | SV *sv; |
1127 | OP_4tree *sop, *rop; |
1128 | |
3280af22 |
1129 | PL_seen_zerolen++; |
1130 | PL_regseen |= REG_SEEN_EVAL; |
1131 | while (count && (c = *PL_regcomp_parse)) { |
1132 | if (c == '\\' && PL_regcomp_parse[1]) |
1133 | PL_regcomp_parse++; |
c277df42 |
1134 | else if (c == '{') |
1135 | count++; |
1136 | else if (c == '}') |
1137 | count--; |
3280af22 |
1138 | PL_regcomp_parse++; |
c277df42 |
1139 | } |
3280af22 |
1140 | if (*PL_regcomp_parse != ')') |
c277df42 |
1141 | FAIL("Sequence (?{...}) not terminated or not {}-balanced"); |
1142 | if (!SIZE_ONLY) { |
1143 | AV *av; |
1144 | |
3280af22 |
1145 | if (PL_regcomp_parse - 1 - s) |
79cb57f6 |
1146 | sv = newSVpvn(s, PL_regcomp_parse - 1 - s); |
c277df42 |
1147 | else |
79cb57f6 |
1148 | sv = newSVpvn("", 0); |
c277df42 |
1149 | |
1150 | rop = sv_compile_2op(sv, &sop, "re", &av); |
1151 | |
dfad63ad |
1152 | n = add_data(3, "nop"); |
3280af22 |
1153 | PL_regcomp_rx->data->data[n] = (void*)rop; |
dfad63ad |
1154 | PL_regcomp_rx->data->data[n+1] = (void*)sop; |
1155 | PL_regcomp_rx->data->data[n+2] = (void*)av; |
c277df42 |
1156 | SvREFCNT_dec(sv); |
a0ed51b3 |
1157 | } |
e24b16f9 |
1158 | else { /* First pass */ |
1159 | if (PL_reginterp_cnt < ++PL_seen_evals |
1160 | && PL_curcop != &PL_compiling) |
2cd61cdb |
1161 | /* No compiled RE interpolated, has runtime |
1162 | components ===> unsafe. */ |
1163 | FAIL("Eval-group not allowed at runtime, use re 'eval'"); |
3280af22 |
1164 | if (PL_tainted) |
cc6b7395 |
1165 | FAIL("Eval-group in insecure regular expression"); |
c277df42 |
1166 | } |
1167 | |
1168 | nextchar(); |
0f5d15d6 |
1169 | if (logical) { |
1170 | ret = reg_node(LOGICAL); |
1171 | if (!SIZE_ONLY) |
1172 | ret->flags = 2; |
1173 | regtail(ret, reganode(EVAL, n)); |
1174 | return ret; |
1175 | } |
c277df42 |
1176 | return reganode(EVAL, n); |
1177 | } |
1178 | case '(': |
1179 | { |
3280af22 |
1180 | if (PL_regcomp_parse[0] == '?') { |
1181 | if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!' |
1182 | || PL_regcomp_parse[1] == '<' |
1183 | || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */ |
c277df42 |
1184 | I32 flag; |
1185 | |
1186 | ret = reg_node(LOGICAL); |
0f5d15d6 |
1187 | if (!SIZE_ONLY) |
1188 | ret->flags = 1; |
c277df42 |
1189 | regtail(ret, reg(1, &flag)); |
1190 | goto insert_if; |
1191 | } |
a0ed51b3 |
1192 | } |
1193 | else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) { |
3280af22 |
1194 | parno = atoi(PL_regcomp_parse++); |
c277df42 |
1195 | |
3280af22 |
1196 | while (isDIGIT(*PL_regcomp_parse)) |
1197 | PL_regcomp_parse++; |
c277df42 |
1198 | ret = reganode(GROUPP, parno); |
1199 | if ((c = *nextchar()) != ')') |
1200 | FAIL2("Switch (?(number%c not recognized", c); |
1201 | insert_if: |
1202 | regtail(ret, reganode(IFTHEN, 0)); |
1203 | br = regbranch(&flags, 1); |
1204 | if (br == NULL) |
1205 | br = reganode(LONGJMP, 0); |
1206 | else |
1207 | regtail(br, reganode(LONGJMP, 0)); |
1208 | c = *nextchar(); |
d1b80229 |
1209 | if (flags&HASWIDTH) |
1210 | *flagp |= HASWIDTH; |
c277df42 |
1211 | if (c == '|') { |
1212 | lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */ |
1213 | regbranch(&flags, 1); |
1214 | regtail(ret, lastbr); |
d1b80229 |
1215 | if (flags&HASWIDTH) |
1216 | *flagp |= HASWIDTH; |
c277df42 |
1217 | c = *nextchar(); |
a0ed51b3 |
1218 | } |
1219 | else |
c277df42 |
1220 | lastbr = NULL; |
1221 | if (c != ')') |
1222 | FAIL("Switch (?(condition)... contains too many branches"); |
1223 | ender = reg_node(TAIL); |
1224 | regtail(br, ender); |
1225 | if (lastbr) { |
1226 | regtail(lastbr, ender); |
1227 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); |
a0ed51b3 |
1228 | } |
1229 | else |
c277df42 |
1230 | regtail(ret, ender); |
1231 | return ret; |
a0ed51b3 |
1232 | } |
1233 | else { |
3280af22 |
1234 | FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse); |
c277df42 |
1235 | } |
1236 | } |
1b1626e4 |
1237 | case 0: |
c277df42 |
1238 | FAIL("Sequence (? incomplete"); |
1b1626e4 |
1239 | break; |
a0d0e21e |
1240 | default: |
3280af22 |
1241 | --PL_regcomp_parse; |
ca9dfc88 |
1242 | parse_flags: |
3280af22 |
1243 | while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) { |
1244 | if (*PL_regcomp_parse != 'o') |
1245 | pmflag(flagsp, *PL_regcomp_parse); |
1246 | ++PL_regcomp_parse; |
ca9dfc88 |
1247 | } |
3280af22 |
1248 | if (*PL_regcomp_parse == '-') { |
ca9dfc88 |
1249 | flagsp = &negflags; |
3280af22 |
1250 | ++PL_regcomp_parse; |
ca9dfc88 |
1251 | goto parse_flags; |
48c036b1 |
1252 | } |
3280af22 |
1253 | PL_regflags |= posflags; |
1254 | PL_regflags &= ~negflags; |
1255 | if (*PL_regcomp_parse == ':') { |
1256 | PL_regcomp_parse++; |
ca9dfc88 |
1257 | paren = ':'; |
1258 | break; |
1259 | } |
c277df42 |
1260 | unknown: |
3280af22 |
1261 | if (*PL_regcomp_parse != ')') |
1262 | FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse); |
a0d0e21e |
1263 | nextchar(); |
1264 | *flagp = TRYAGAIN; |
1265 | return NULL; |
1266 | } |
1267 | } |
1268 | else { |
3280af22 |
1269 | parno = PL_regnpar; |
1270 | PL_regnpar++; |
a0d0e21e |
1271 | ret = reganode(OPEN, parno); |
c277df42 |
1272 | open = 1; |
a0d0e21e |
1273 | } |
a0ed51b3 |
1274 | } |
1275 | else |
a0d0e21e |
1276 | ret = NULL; |
1277 | |
1278 | /* Pick up the branches, linking them together. */ |
c277df42 |
1279 | br = regbranch(&flags, 1); |
a0d0e21e |
1280 | if (br == NULL) |
1281 | return(NULL); |
3280af22 |
1282 | if (*PL_regcomp_parse == '|') { |
1283 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 |
1284 | reginsert(BRANCHJ, br); |
a0ed51b3 |
1285 | } |
1286 | else |
c277df42 |
1287 | reginsert(BRANCH, br); |
1288 | have_branch = 1; |
1289 | if (SIZE_ONLY) |
3280af22 |
1290 | PL_extralen += 1; /* For BRANCHJ-BRANCH. */ |
a0ed51b3 |
1291 | } |
1292 | else if (paren == ':') { |
c277df42 |
1293 | *flagp |= flags&SIMPLE; |
1294 | } |
1295 | if (open) { /* Starts with OPEN. */ |
1296 | regtail(ret, br); /* OPEN -> first. */ |
a0ed51b3 |
1297 | } |
1298 | else if (paren != '?') /* Not Conditional */ |
a0d0e21e |
1299 | ret = br; |
821b33a5 |
1300 | if (flags&HASWIDTH) |
1301 | *flagp |= HASWIDTH; |
a0d0e21e |
1302 | *flagp |= flags&SPSTART; |
c277df42 |
1303 | lastbr = br; |
3280af22 |
1304 | while (*PL_regcomp_parse == '|') { |
1305 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 |
1306 | ender = reganode(LONGJMP,0); |
1307 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ |
1308 | } |
1309 | if (SIZE_ONLY) |
3280af22 |
1310 | PL_extralen += 2; /* Account for LONGJMP. */ |
a0d0e21e |
1311 | nextchar(); |
c277df42 |
1312 | br = regbranch(&flags, 0); |
a687059c |
1313 | if (br == NULL) |
a0d0e21e |
1314 | return(NULL); |
c277df42 |
1315 | regtail(lastbr, br); /* BRANCH -> BRANCH. */ |
1316 | lastbr = br; |
821b33a5 |
1317 | if (flags&HASWIDTH) |
1318 | *flagp |= HASWIDTH; |
a687059c |
1319 | *flagp |= flags&SPSTART; |
a0d0e21e |
1320 | } |
1321 | |
c277df42 |
1322 | if (have_branch || paren != ':') { |
1323 | /* Make a closing node, and hook it on the end. */ |
1324 | switch (paren) { |
1325 | case ':': |
1326 | ender = reg_node(TAIL); |
1327 | break; |
1328 | case 1: |
1329 | ender = reganode(CLOSE, parno); |
1330 | break; |
1331 | case '<': |
c277df42 |
1332 | case ',': |
1333 | case '=': |
1334 | case '!': |
c277df42 |
1335 | *flagp &= ~HASWIDTH; |
821b33a5 |
1336 | /* FALL THROUGH */ |
1337 | case '>': |
1338 | ender = reg_node(SUCCEED); |
c277df42 |
1339 | break; |
1340 | case 0: |
1341 | ender = reg_node(END); |
1342 | break; |
1343 | } |
1344 | regtail(lastbr, ender); |
a0d0e21e |
1345 | |
c277df42 |
1346 | if (have_branch) { |
1347 | /* Hook the tails of the branches to the closing node. */ |
1348 | for (br = ret; br != NULL; br = regnext(br)) { |
1349 | regoptail(br, ender); |
1350 | } |
1351 | } |
a0d0e21e |
1352 | } |
c277df42 |
1353 | |
1354 | { |
1355 | char *p; |
1356 | static char parens[] = "=!<,>"; |
1357 | |
1358 | if (paren && (p = strchr(parens, paren))) { |
1359 | int node = ((p - parens) % 2) ? UNLESSM : IFMATCH; |
1360 | int flag = (p - parens) > 1; |
1361 | |
1362 | if (paren == '>') |
1363 | node = SUSPEND, flag = 0; |
1364 | reginsert(node,ret); |
c277df42 |
1365 | ret->flags = flag; |
c277df42 |
1366 | regtail(ret, reg_node(TAIL)); |
1367 | } |
a0d0e21e |
1368 | } |
1369 | |
1370 | /* Check for proper termination. */ |
ce3e6498 |
1371 | if (paren) { |
1372 | PL_regflags = oregflags; |
1373 | if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') { |
1374 | FAIL("unmatched () in regexp"); |
1375 | } |
a0ed51b3 |
1376 | } |
1377 | else if (!paren && PL_regcomp_parse < PL_regxend) { |
3280af22 |
1378 | if (*PL_regcomp_parse == ')') { |
a0d0e21e |
1379 | FAIL("unmatched () in regexp"); |
a0ed51b3 |
1380 | } |
1381 | else |
a0d0e21e |
1382 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
1383 | /* NOTREACHED */ |
1384 | } |
a687059c |
1385 | |
a0d0e21e |
1386 | return(ret); |
a687059c |
1387 | } |
1388 | |
1389 | /* |
1390 | - regbranch - one alternative of an | operator |
1391 | * |
1392 | * Implements the concatenation operator. |
1393 | */ |
76e3520e |
1394 | STATIC regnode * |
cea2e8a9 |
1395 | S_regbranch(pTHX_ I32 *flagp, I32 first) |
a687059c |
1396 | { |
5c0ca799 |
1397 | dTHR; |
c277df42 |
1398 | register regnode *ret; |
1399 | register regnode *chain = NULL; |
1400 | register regnode *latest; |
1401 | I32 flags = 0, c = 0; |
a0d0e21e |
1402 | |
c277df42 |
1403 | if (first) |
1404 | ret = NULL; |
1405 | else { |
3280af22 |
1406 | if (!SIZE_ONLY && PL_extralen) |
c277df42 |
1407 | ret = reganode(BRANCHJ,0); |
1408 | else |
1409 | ret = reg_node(BRANCH); |
1410 | } |
1411 | |
1412 | if (!first && SIZE_ONLY) |
3280af22 |
1413 | PL_extralen += 1; /* BRANCHJ */ |
c277df42 |
1414 | |
1415 | *flagp = WORST; /* Tentatively. */ |
a0d0e21e |
1416 | |
3280af22 |
1417 | PL_regcomp_parse--; |
a0d0e21e |
1418 | nextchar(); |
3280af22 |
1419 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') { |
a0d0e21e |
1420 | flags &= ~TRYAGAIN; |
1421 | latest = regpiece(&flags); |
1422 | if (latest == NULL) { |
1423 | if (flags & TRYAGAIN) |
1424 | continue; |
1425 | return(NULL); |
a0ed51b3 |
1426 | } |
1427 | else if (ret == NULL) |
c277df42 |
1428 | ret = latest; |
a0d0e21e |
1429 | *flagp |= flags&HASWIDTH; |
c277df42 |
1430 | if (chain == NULL) /* First piece. */ |
a0d0e21e |
1431 | *flagp |= flags&SPSTART; |
1432 | else { |
3280af22 |
1433 | PL_regnaughty++; |
a0d0e21e |
1434 | regtail(chain, latest); |
a687059c |
1435 | } |
a0d0e21e |
1436 | chain = latest; |
c277df42 |
1437 | c++; |
1438 | } |
1439 | if (chain == NULL) { /* Loop ran zero times. */ |
1440 | chain = reg_node(NOTHING); |
1441 | if (ret == NULL) |
1442 | ret = chain; |
1443 | } |
1444 | if (c == 1) { |
1445 | *flagp |= flags&SIMPLE; |
a0d0e21e |
1446 | } |
a687059c |
1447 | |
a0d0e21e |
1448 | return(ret); |
a687059c |
1449 | } |
1450 | |
1451 | /* |
1452 | - regpiece - something followed by possible [*+?] |
1453 | * |
1454 | * Note that the branching code sequences used for ? and the general cases |
1455 | * of * and + are somewhat optimized: they use the same NOTHING node as |
1456 | * both the endmarker for their branch list and the body of the last branch. |
1457 | * It might seem that this node could be dispensed with entirely, but the |
1458 | * endmarker role is not redundant. |
1459 | */ |
76e3520e |
1460 | STATIC regnode * |
cea2e8a9 |
1461 | S_regpiece(pTHX_ I32 *flagp) |
a687059c |
1462 | { |
5c0ca799 |
1463 | dTHR; |
c277df42 |
1464 | register regnode *ret; |
a0d0e21e |
1465 | register char op; |
1466 | register char *next; |
1467 | I32 flags; |
3280af22 |
1468 | char *origparse = PL_regcomp_parse; |
a0d0e21e |
1469 | char *maxpos; |
1470 | I32 min; |
c277df42 |
1471 | I32 max = REG_INFTY; |
a0d0e21e |
1472 | |
1473 | ret = regatom(&flags); |
1474 | if (ret == NULL) { |
1475 | if (flags & TRYAGAIN) |
1476 | *flagp |= TRYAGAIN; |
1477 | return(NULL); |
1478 | } |
1479 | |
3280af22 |
1480 | op = *PL_regcomp_parse; |
a0d0e21e |
1481 | |
3280af22 |
1482 | if (op == '{' && regcurly(PL_regcomp_parse)) { |
1483 | next = PL_regcomp_parse + 1; |
a0d0e21e |
1484 | maxpos = Nullch; |
1485 | while (isDIGIT(*next) || *next == ',') { |
1486 | if (*next == ',') { |
1487 | if (maxpos) |
1488 | break; |
1489 | else |
1490 | maxpos = next; |
a687059c |
1491 | } |
a0d0e21e |
1492 | next++; |
1493 | } |
1494 | if (*next == '}') { /* got one */ |
1495 | if (!maxpos) |
1496 | maxpos = next; |
3280af22 |
1497 | PL_regcomp_parse++; |
1498 | min = atoi(PL_regcomp_parse); |
a0d0e21e |
1499 | if (*maxpos == ',') |
1500 | maxpos++; |
1501 | else |
3280af22 |
1502 | maxpos = PL_regcomp_parse; |
a0d0e21e |
1503 | max = atoi(maxpos); |
1504 | if (!max && *maxpos != '0') |
c277df42 |
1505 | max = REG_INFTY; /* meaning "infinity" */ |
1506 | else if (max >= REG_INFTY) |
1507 | FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); |
3280af22 |
1508 | PL_regcomp_parse = next; |
a0d0e21e |
1509 | nextchar(); |
1510 | |
1511 | do_curly: |
1512 | if ((flags&SIMPLE)) { |
3280af22 |
1513 | PL_regnaughty += 2 + PL_regnaughty / 2; |
a0d0e21e |
1514 | reginsert(CURLY, ret); |
1515 | } |
1516 | else { |
3280af22 |
1517 | PL_regnaughty += 4 + PL_regnaughty; /* compound interest */ |
c277df42 |
1518 | regtail(ret, reg_node(WHILEM)); |
3280af22 |
1519 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 |
1520 | reginsert(LONGJMP,ret); |
1521 | reginsert(NOTHING,ret); |
1522 | NEXT_OFF(ret) = 3; /* Go over LONGJMP. */ |
1523 | } |
a0d0e21e |
1524 | reginsert(CURLYX,ret); |
3280af22 |
1525 | if (!SIZE_ONLY && PL_extralen) |
c277df42 |
1526 | NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */ |
1527 | regtail(ret, reg_node(NOTHING)); |
1528 | if (SIZE_ONLY) |
3280af22 |
1529 | PL_extralen += 3; |
a0d0e21e |
1530 | } |
c277df42 |
1531 | ret->flags = 0; |
a0d0e21e |
1532 | |
1533 | if (min > 0) |
821b33a5 |
1534 | *flagp = WORST; |
1535 | if (max > 0) |
1536 | *flagp |= HASWIDTH; |
a0d0e21e |
1537 | if (max && max < min) |
c277df42 |
1538 | FAIL("Can't do {n,m} with n > m"); |
1539 | if (!SIZE_ONLY) { |
1540 | ARG1_SET(ret, min); |
1541 | ARG2_SET(ret, max); |
a687059c |
1542 | } |
a687059c |
1543 | |
a0d0e21e |
1544 | goto nest_check; |
a687059c |
1545 | } |
a0d0e21e |
1546 | } |
a687059c |
1547 | |
a0d0e21e |
1548 | if (!ISMULT1(op)) { |
1549 | *flagp = flags; |
a687059c |
1550 | return(ret); |
a0d0e21e |
1551 | } |
bb20fd44 |
1552 | |
c277df42 |
1553 | #if 0 /* Now runtime fix should be reliable. */ |
bb20fd44 |
1554 | if (!(flags&HASWIDTH) && op != '?') |
c277df42 |
1555 | FAIL("regexp *+ operand could be empty"); |
1556 | #endif |
bb20fd44 |
1557 | |
a0d0e21e |
1558 | nextchar(); |
1559 | |
821b33a5 |
1560 | *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); |
a0d0e21e |
1561 | |
1562 | if (op == '*' && (flags&SIMPLE)) { |
1563 | reginsert(STAR, ret); |
c277df42 |
1564 | ret->flags = 0; |
3280af22 |
1565 | PL_regnaughty += 4; |
a0d0e21e |
1566 | } |
1567 | else if (op == '*') { |
1568 | min = 0; |
1569 | goto do_curly; |
a0ed51b3 |
1570 | } |
1571 | else if (op == '+' && (flags&SIMPLE)) { |
a0d0e21e |
1572 | reginsert(PLUS, ret); |
c277df42 |
1573 | ret->flags = 0; |
3280af22 |
1574 | PL_regnaughty += 3; |
a0d0e21e |
1575 | } |
1576 | else if (op == '+') { |
1577 | min = 1; |
1578 | goto do_curly; |
a0ed51b3 |
1579 | } |
1580 | else if (op == '?') { |
a0d0e21e |
1581 | min = 0; max = 1; |
1582 | goto do_curly; |
1583 | } |
1584 | nest_check: |
17feb5d5 |
1585 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) { |
cea2e8a9 |
1586 | Perl_warner(aTHX_ WARN_UNSAFE, "%.*s matches null string many times", |
3280af22 |
1587 | PL_regcomp_parse - origparse, origparse); |
a0d0e21e |
1588 | } |
1589 | |
3280af22 |
1590 | if (*PL_regcomp_parse == '?') { |
a0d0e21e |
1591 | nextchar(); |
1592 | reginsert(MINMOD, ret); |
c277df42 |
1593 | regtail(ret, ret + NODE_STEP_REGNODE); |
a0d0e21e |
1594 | } |
3280af22 |
1595 | if (ISMULT2(PL_regcomp_parse)) |
a0d0e21e |
1596 | FAIL("nested *?+ in regexp"); |
1597 | |
1598 | return(ret); |
a687059c |
1599 | } |
1600 | |
1601 | /* |
1602 | - regatom - the lowest level |
1603 | * |
1604 | * Optimization: gobbles an entire sequence of ordinary characters so that |
1605 | * it can turn them into a single node, which is smaller to store and |
1606 | * faster to run. Backslashed characters are exceptions, each becoming a |
1607 | * separate node; the code is simpler that way and it's not worth fixing. |
1608 | * |
1609 | * [Yes, it is worth fixing, some scripts can run twice the speed.] |
1610 | */ |
76e3520e |
1611 | STATIC regnode * |
cea2e8a9 |
1612 | S_regatom(pTHX_ I32 *flagp) |
a687059c |
1613 | { |
5c0ca799 |
1614 | dTHR; |
c277df42 |
1615 | register regnode *ret = 0; |
a0d0e21e |
1616 | I32 flags; |
1617 | |
1618 | *flagp = WORST; /* Tentatively. */ |
1619 | |
1620 | tryagain: |
3280af22 |
1621 | switch (*PL_regcomp_parse) { |
a0d0e21e |
1622 | case '^': |
3280af22 |
1623 | PL_seen_zerolen++; |
a0d0e21e |
1624 | nextchar(); |
3280af22 |
1625 | if (PL_regflags & PMf_MULTILINE) |
c277df42 |
1626 | ret = reg_node(MBOL); |
3280af22 |
1627 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 |
1628 | ret = reg_node(SBOL); |
a0d0e21e |
1629 | else |
c277df42 |
1630 | ret = reg_node(BOL); |
a0d0e21e |
1631 | break; |
1632 | case '$': |
3280af22 |
1633 | if (PL_regcomp_parse[1]) |
1634 | PL_seen_zerolen++; |
a0d0e21e |
1635 | nextchar(); |
3280af22 |
1636 | if (PL_regflags & PMf_MULTILINE) |
c277df42 |
1637 | ret = reg_node(MEOL); |
3280af22 |
1638 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 |
1639 | ret = reg_node(SEOL); |
a0d0e21e |
1640 | else |
c277df42 |
1641 | ret = reg_node(EOL); |
a0d0e21e |
1642 | break; |
1643 | case '.': |
1644 | nextchar(); |
a0ed51b3 |
1645 | if (UTF) { |
1646 | if (PL_regflags & PMf_SINGLELINE) |
1647 | ret = reg_node(SANYUTF8); |
1648 | else |
1649 | ret = reg_node(ANYUTF8); |
1650 | *flagp |= HASWIDTH; |
1651 | } |
1652 | else { |
1653 | if (PL_regflags & PMf_SINGLELINE) |
1654 | ret = reg_node(SANY); |
1655 | else |
22c35a8c |
1656 | ret = reg_node(REG_ANY); |
a0ed51b3 |
1657 | *flagp |= HASWIDTH|SIMPLE; |
1658 | } |
3280af22 |
1659 | PL_regnaughty++; |
a0d0e21e |
1660 | break; |
1661 | case '[': |
3280af22 |
1662 | PL_regcomp_parse++; |
a0ed51b3 |
1663 | ret = (UTF ? regclassutf8() : regclass()); |
a14b48bc |
1664 | if (*PL_regcomp_parse != ']') |
1665 | FAIL("unmatched [] in regexp"); |
1666 | nextchar(); |
a0d0e21e |
1667 | *flagp |= HASWIDTH|SIMPLE; |
1668 | break; |
1669 | case '(': |
1670 | nextchar(); |
1671 | ret = reg(1, &flags); |
1672 | if (ret == NULL) { |
1673 | if (flags & TRYAGAIN) |
1674 | goto tryagain; |
1675 | return(NULL); |
1676 | } |
c277df42 |
1677 | *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE); |
a0d0e21e |
1678 | break; |
1679 | case '|': |
1680 | case ')': |
1681 | if (flags & TRYAGAIN) { |
1682 | *flagp |= TRYAGAIN; |
1683 | return NULL; |
1684 | } |
3280af22 |
1685 | FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse); |
a0d0e21e |
1686 | /* Supposed to be caught earlier. */ |
1687 | break; |
85afd4ae |
1688 | case '{': |
3280af22 |
1689 | if (!regcurly(PL_regcomp_parse)) { |
1690 | PL_regcomp_parse++; |
85afd4ae |
1691 | goto defchar; |
1692 | } |
1693 | /* FALL THROUGH */ |
a0d0e21e |
1694 | case '?': |
1695 | case '+': |
1696 | case '*': |
3115e423 |
1697 | FAIL("?+*{} follows nothing in regexp"); |
a0d0e21e |
1698 | break; |
1699 | case '\\': |
3280af22 |
1700 | switch (*++PL_regcomp_parse) { |
a0d0e21e |
1701 | case 'A': |
3280af22 |
1702 | PL_seen_zerolen++; |
c277df42 |
1703 | ret = reg_node(SBOL); |
a0d0e21e |
1704 | *flagp |= SIMPLE; |
1705 | nextchar(); |
1706 | break; |
1707 | case 'G': |
c277df42 |
1708 | ret = reg_node(GPOS); |
3280af22 |
1709 | PL_regseen |= REG_SEEN_GPOS; |
a0d0e21e |
1710 | *flagp |= SIMPLE; |
1711 | nextchar(); |
1712 | break; |
1713 | case 'Z': |
c277df42 |
1714 | ret = reg_node(SEOL); |
a0d0e21e |
1715 | *flagp |= SIMPLE; |
1716 | nextchar(); |
1717 | break; |
b85d18e9 |
1718 | case 'z': |
1719 | ret = reg_node(EOS); |
1720 | *flagp |= SIMPLE; |
3280af22 |
1721 | PL_seen_zerolen++; /* Do not optimize RE away */ |
b85d18e9 |
1722 | nextchar(); |
1723 | break; |
a0ed51b3 |
1724 | case 'C': |
1725 | ret = reg_node(SANY); |
1726 | *flagp |= HASWIDTH|SIMPLE; |
1727 | nextchar(); |
1728 | break; |
1729 | case 'X': |
1730 | ret = reg_node(CLUMP); |
1731 | *flagp |= HASWIDTH; |
1732 | nextchar(); |
1733 | if (UTF && !PL_utf8_mark) |
dfe13c55 |
1734 | is_utf8_mark((U8*)"~"); /* preload table */ |
a0ed51b3 |
1735 | break; |
a0d0e21e |
1736 | case 'w': |
a0ed51b3 |
1737 | ret = reg_node( |
1738 | UTF |
1739 | ? (LOC ? ALNUMLUTF8 : ALNUMUTF8) |
1740 | : (LOC ? ALNUML : ALNUM)); |
a0d0e21e |
1741 | *flagp |= HASWIDTH|SIMPLE; |
1742 | nextchar(); |
a0ed51b3 |
1743 | if (UTF && !PL_utf8_alnum) |
dfe13c55 |
1744 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e |
1745 | break; |
1746 | case 'W': |
a0ed51b3 |
1747 | ret = reg_node( |
1748 | UTF |
1749 | ? (LOC ? NALNUMLUTF8 : NALNUMUTF8) |
1750 | : (LOC ? NALNUML : NALNUM)); |
a0d0e21e |
1751 | *flagp |= HASWIDTH|SIMPLE; |
1752 | nextchar(); |
a0ed51b3 |
1753 | if (UTF && !PL_utf8_alnum) |
dfe13c55 |
1754 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e |
1755 | break; |
1756 | case 'b': |
3280af22 |
1757 | PL_seen_zerolen++; |
f5c9036e |
1758 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 |
1759 | ret = reg_node( |
1760 | UTF |
1761 | ? (LOC ? BOUNDLUTF8 : BOUNDUTF8) |
1762 | : (LOC ? BOUNDL : BOUND)); |
a0d0e21e |
1763 | *flagp |= SIMPLE; |
1764 | nextchar(); |
a0ed51b3 |
1765 | if (UTF && !PL_utf8_alnum) |
dfe13c55 |
1766 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e |
1767 | break; |
1768 | case 'B': |
3280af22 |
1769 | PL_seen_zerolen++; |
f5c9036e |
1770 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 |
1771 | ret = reg_node( |
1772 | UTF |
1773 | ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8) |
1774 | : (LOC ? NBOUNDL : NBOUND)); |
a0d0e21e |
1775 | *flagp |= SIMPLE; |
1776 | nextchar(); |
a0ed51b3 |
1777 | if (UTF && !PL_utf8_alnum) |
dfe13c55 |
1778 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e |
1779 | break; |
1780 | case 's': |
a0ed51b3 |
1781 | ret = reg_node( |
1782 | UTF |
1783 | ? (LOC ? SPACELUTF8 : SPACEUTF8) |
1784 | : (LOC ? SPACEL : SPACE)); |
a0d0e21e |
1785 | *flagp |= HASWIDTH|SIMPLE; |
1786 | nextchar(); |
a0ed51b3 |
1787 | if (UTF && !PL_utf8_space) |
dfe13c55 |
1788 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e |
1789 | break; |
1790 | case 'S': |
a0ed51b3 |
1791 | ret = reg_node( |
1792 | UTF |
1793 | ? (LOC ? NSPACELUTF8 : NSPACEUTF8) |
1794 | : (LOC ? NSPACEL : NSPACE)); |
a0d0e21e |
1795 | *flagp |= HASWIDTH|SIMPLE; |
1796 | nextchar(); |
a0ed51b3 |
1797 | if (UTF && !PL_utf8_space) |
dfe13c55 |
1798 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e |
1799 | break; |
1800 | case 'd': |
a0ed51b3 |
1801 | ret = reg_node(UTF ? DIGITUTF8 : DIGIT); |
a0d0e21e |
1802 | *flagp |= HASWIDTH|SIMPLE; |
1803 | nextchar(); |
a0ed51b3 |
1804 | if (UTF && !PL_utf8_digit) |
dfe13c55 |
1805 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e |
1806 | break; |
1807 | case 'D': |
a0ed51b3 |
1808 | ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT); |
a0d0e21e |
1809 | *flagp |= HASWIDTH|SIMPLE; |
1810 | nextchar(); |
a0ed51b3 |
1811 | if (UTF && !PL_utf8_digit) |
dfe13c55 |
1812 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e |
1813 | break; |
a14b48bc |
1814 | case 'p': |
1815 | case 'P': |
1816 | { /* a lovely hack--pretend we saw [\pX] instead */ |
1817 | char* oldregxend = PL_regxend; |
1818 | |
1819 | if (PL_regcomp_parse[1] == '{') { |
1820 | PL_regxend = strchr(PL_regcomp_parse, '}'); |
1821 | if (!PL_regxend) |
1822 | FAIL("Missing right brace on \\p{}"); |
1823 | PL_regxend++; |
1824 | } |
1825 | else |
1826 | PL_regxend = PL_regcomp_parse + 2; |
1827 | PL_regcomp_parse--; |
1828 | |
1829 | ret = regclassutf8(); |
1830 | |
1831 | PL_regxend = oldregxend; |
1832 | PL_regcomp_parse--; |
1833 | nextchar(); |
1834 | *flagp |= HASWIDTH|SIMPLE; |
1835 | } |
1836 | break; |
a0d0e21e |
1837 | case 'n': |
1838 | case 'r': |
1839 | case 't': |
1840 | case 'f': |
1841 | case 'e': |
1842 | case 'a': |
1843 | case 'x': |
1844 | case 'c': |
1845 | case '0': |
1846 | goto defchar; |
1847 | case '1': case '2': case '3': case '4': |
1848 | case '5': case '6': case '7': case '8': case '9': |
1849 | { |
3280af22 |
1850 | I32 num = atoi(PL_regcomp_parse); |
a0d0e21e |
1851 | |
3280af22 |
1852 | if (num > 9 && num >= PL_regnpar) |
a0d0e21e |
1853 | goto defchar; |
1854 | else { |
3280af22 |
1855 | if (!SIZE_ONLY && num > PL_regcomp_rx->nparens) |
ef64f398 |
1856 | FAIL("reference to nonexistent group"); |
3280af22 |
1857 | PL_regsawback = 1; |
a0ed51b3 |
1858 | ret = reganode(FOLD |
1859 | ? (LOC ? REFFL : REFF) |
c8756f30 |
1860 | : REF, num); |
a0d0e21e |
1861 | *flagp |= HASWIDTH; |
3280af22 |
1862 | while (isDIGIT(*PL_regcomp_parse)) |
1863 | PL_regcomp_parse++; |
1864 | PL_regcomp_parse--; |
a0d0e21e |
1865 | nextchar(); |
1866 | } |
1867 | } |
1868 | break; |
1869 | case '\0': |
3280af22 |
1870 | if (PL_regcomp_parse >= PL_regxend) |
a0d0e21e |
1871 | FAIL("trailing \\ in regexp"); |
1872 | /* FALL THROUGH */ |
1873 | default: |
c9f97d15 |
1874 | /* Do not generate `unrecognized' warnings here, we fall |
1875 | back into the quick-grab loop below */ |
a0d0e21e |
1876 | goto defchar; |
1877 | } |
1878 | break; |
4633a7c4 |
1879 | |
1880 | case '#': |
3280af22 |
1881 | if (PL_regflags & PMf_EXTENDED) { |
1882 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++; |
1883 | if (PL_regcomp_parse < PL_regxend) |
4633a7c4 |
1884 | goto tryagain; |
1885 | } |
1886 | /* FALL THROUGH */ |
1887 | |
a0d0e21e |
1888 | default: { |
1889 | register I32 len; |
a0ed51b3 |
1890 | register UV ender; |
a0d0e21e |
1891 | register char *p; |
c277df42 |
1892 | char *oldp, *s; |
a0d0e21e |
1893 | I32 numlen; |
1894 | |
3280af22 |
1895 | PL_regcomp_parse++; |
a0d0e21e |
1896 | |
1897 | defchar: |
a0ed51b3 |
1898 | ret = reg_node(FOLD |
1899 | ? (LOC ? EXACTFL : EXACTF) |
bbce6d69 |
1900 | : EXACT); |
161b471a |
1901 | s = (char *) OPERAND(ret); |
c277df42 |
1902 | regc(0, s++); /* save spot for len */ |
3280af22 |
1903 | for (len = 0, p = PL_regcomp_parse - 1; |
1904 | len < 127 && p < PL_regxend; |
a0d0e21e |
1905 | len++) |
1906 | { |
1907 | oldp = p; |
5b5a24f7 |
1908 | |
3280af22 |
1909 | if (PL_regflags & PMf_EXTENDED) |
1910 | p = regwhite(p, PL_regxend); |
a0d0e21e |
1911 | switch (*p) { |
1912 | case '^': |
1913 | case '$': |
1914 | case '.': |
1915 | case '[': |
1916 | case '(': |
1917 | case ')': |
1918 | case '|': |
1919 | goto loopdone; |
1920 | case '\\': |
1921 | switch (*++p) { |
1922 | case 'A': |
1923 | case 'G': |
1924 | case 'Z': |
b85d18e9 |
1925 | case 'z': |
a0d0e21e |
1926 | case 'w': |
1927 | case 'W': |
1928 | case 'b': |
1929 | case 'B': |
1930 | case 's': |
1931 | case 'S': |
1932 | case 'd': |
1933 | case 'D': |
a14b48bc |
1934 | case 'p': |
1935 | case 'P': |
a0d0e21e |
1936 | --p; |
1937 | goto loopdone; |
1938 | case 'n': |
1939 | ender = '\n'; |
1940 | p++; |
a687059c |
1941 | break; |
a0d0e21e |
1942 | case 'r': |
1943 | ender = '\r'; |
1944 | p++; |
a687059c |
1945 | break; |
a0d0e21e |
1946 | case 't': |
1947 | ender = '\t'; |
1948 | p++; |
a687059c |
1949 | break; |
a0d0e21e |
1950 | case 'f': |
1951 | ender = '\f'; |
1952 | p++; |
a687059c |
1953 | break; |
a0d0e21e |
1954 | case 'e': |
1955 | ender = '\033'; |
1956 | p++; |
a687059c |
1957 | break; |
a0d0e21e |
1958 | case 'a': |
1959 | ender = '\007'; |
1960 | p++; |
a687059c |
1961 | break; |
a0d0e21e |
1962 | case 'x': |
a0ed51b3 |
1963 | if (*++p == '{') { |
1964 | char* e = strchr(p, '}'); |
1965 | |
1966 | if (!e) |
1967 | FAIL("Missing right brace on \\x{}"); |
1968 | else if (UTF) { |
1969 | ender = scan_hex(p + 1, e - p, &numlen); |
1970 | if (numlen + len >= 127) { /* numlen is generous */ |
1971 | p--; |
1972 | goto loopdone; |
1973 | } |
1974 | p = e + 1; |
1975 | } |
1976 | else |
1977 | FAIL("Can't use \\x{} without 'use utf8' declaration"); |
1978 | } |
1979 | else { |
1980 | ender = scan_hex(p, 2, &numlen); |
1981 | p += numlen; |
1982 | } |
a687059c |
1983 | break; |
a0d0e21e |
1984 | case 'c': |
1985 | p++; |
bbce6d69 |
1986 | ender = UCHARAT(p++); |
1987 | ender = toCTRL(ender); |
a687059c |
1988 | break; |
a0d0e21e |
1989 | case '0': case '1': case '2': case '3':case '4': |
1990 | case '5': case '6': case '7': case '8':case '9': |
1991 | if (*p == '0' || |
3280af22 |
1992 | (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) { |
a0d0e21e |
1993 | ender = scan_oct(p, 3, &numlen); |
1994 | p += numlen; |
1995 | } |
1996 | else { |
1997 | --p; |
1998 | goto loopdone; |
a687059c |
1999 | } |
2000 | break; |
a0d0e21e |
2001 | case '\0': |
3280af22 |
2002 | if (p >= PL_regxend) |
a687059c |
2003 | FAIL("trailing \\ in regexp"); |
2004 | /* FALL THROUGH */ |
a0d0e21e |
2005 | default: |
c9f97d15 |
2006 | if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p)) |
cea2e8a9 |
2007 | Perl_warner(aTHX_ WARN_UNSAFE, |
c9f97d15 |
2008 | "/%.127s/: Unrecognized escape \\%c passed through", |
2009 | PL_regprecomp, |
2010 | *p); |
a0ed51b3 |
2011 | goto normal_default; |
a0d0e21e |
2012 | } |
2013 | break; |
a687059c |
2014 | default: |
a0ed51b3 |
2015 | normal_default: |
2016 | if ((*p & 0xc0) == 0xc0 && UTF) { |
dfe13c55 |
2017 | ender = utf8_to_uv((U8*)p, &numlen); |
a0ed51b3 |
2018 | p += numlen; |
2019 | } |
2020 | else |
2021 | ender = *p++; |
a0d0e21e |
2022 | break; |
a687059c |
2023 | } |
3280af22 |
2024 | if (PL_regflags & PMf_EXTENDED) |
2025 | p = regwhite(p, PL_regxend); |
a0ed51b3 |
2026 | if (UTF && FOLD) { |
2027 | if (LOC) |
2028 | ender = toLOWER_LC_uni(ender); |
2029 | else |
2030 | ender = toLOWER_uni(ender); |
2031 | } |
a0d0e21e |
2032 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
2033 | if (len) |
2034 | p = oldp; |
a0ed51b3 |
2035 | else if (ender >= 0x80 && UTF) { |
2036 | reguni(ender, s, &numlen); |
2037 | s += numlen; |
2038 | len += numlen; |
2039 | } |
a0d0e21e |
2040 | else { |
2041 | len++; |
c277df42 |
2042 | regc(ender, s++); |
a0d0e21e |
2043 | } |
2044 | break; |
a687059c |
2045 | } |
a0ed51b3 |
2046 | if (ender >= 0x80 && UTF) { |
2047 | reguni(ender, s, &numlen); |
2048 | s += numlen; |
2049 | len += numlen - 1; |
2050 | } |
2051 | else |
2052 | regc(ender, s++); |
a0d0e21e |
2053 | } |
2054 | loopdone: |
3280af22 |
2055 | PL_regcomp_parse = p - 1; |
a0d0e21e |
2056 | nextchar(); |
2057 | if (len < 0) |
2058 | FAIL("internal disaster in regexp"); |
2059 | if (len > 0) |
2060 | *flagp |= HASWIDTH; |
2061 | if (len == 1) |
2062 | *flagp |= SIMPLE; |
c277df42 |
2063 | if (!SIZE_ONLY) |
a0d0e21e |
2064 | *OPERAND(ret) = len; |
c277df42 |
2065 | regc('\0', s++); |
2066 | if (SIZE_ONLY) { |
3280af22 |
2067 | PL_regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
a0ed51b3 |
2068 | } |
2069 | else { |
3280af22 |
2070 | PL_regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
c277df42 |
2071 | } |
a687059c |
2072 | } |
a0d0e21e |
2073 | break; |
2074 | } |
a687059c |
2075 | |
a0d0e21e |
2076 | return(ret); |
a687059c |
2077 | } |
2078 | |
873ef191 |
2079 | STATIC char * |
cea2e8a9 |
2080 | S_regwhite(pTHX_ char *p, char *e) |
5b5a24f7 |
2081 | { |
2082 | while (p < e) { |
2083 | if (isSPACE(*p)) |
2084 | ++p; |
2085 | else if (*p == '#') { |
2086 | do { |
2087 | p++; |
2088 | } while (p < e && *p != '\n'); |
2089 | } |
2090 | else |
2091 | break; |
2092 | } |
2093 | return p; |
2094 | } |
2095 | |
620e46c5 |
2096 | /* parse POSIX character classes like [[:foo:]] */ |
2097 | STATIC char* |
cea2e8a9 |
2098 | S_regpposixcc(pTHX_ I32 value) |
620e46c5 |
2099 | { |
11b8faa4 |
2100 | dTHR; |
620e46c5 |
2101 | char *posixcc = 0; |
2102 | |
2103 | if (value == '[' && PL_regcomp_parse + 1 < PL_regxend && |
2104 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ |
2105 | (*PL_regcomp_parse == ':' || |
2106 | *PL_regcomp_parse == '=' || |
2107 | *PL_regcomp_parse == '.')) { |
2108 | char c = *PL_regcomp_parse; |
2109 | char* s = PL_regcomp_parse++; |
2110 | |
2111 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c) |
2112 | PL_regcomp_parse++; |
2113 | if (PL_regcomp_parse == PL_regxend) |
2114 | /* Grandfather lone [:, [=, [. */ |
2115 | PL_regcomp_parse = s; |
2116 | else { |
2117 | PL_regcomp_parse++; /* skip over the c */ |
2118 | if (*PL_regcomp_parse == ']') { |
2119 | /* Not Implemented Yet. |
2120 | * (POSIX Extended Character Classes, that is) |
2121 | * The text between e.g. [: and :] would start |
2122 | * at s + 1 and stop at regcomp_parse - 2. */ |
2123 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY) |
cea2e8a9 |
2124 | Perl_warner(aTHX_ WARN_UNSAFE, |
620e46c5 |
2125 | "Character class syntax [%c %c] is reserved for future extensions", c, c); |
2126 | PL_regcomp_parse++; /* skip over the ending ] */ |
2127 | posixcc = s + 1; |
2128 | } |
767d463e |
2129 | else { |
2130 | /* maternal grandfather */ |
2131 | PL_regcomp_parse = s; |
2132 | } |
620e46c5 |
2133 | } |
2134 | } |
2135 | |
2136 | return posixcc; |
2137 | } |
2138 | |
76e3520e |
2139 | STATIC regnode * |
cea2e8a9 |
2140 | S_regclass(pTHX) |
a687059c |
2141 | { |
5c0ca799 |
2142 | dTHR; |
c277df42 |
2143 | register char *opnd, *s; |
a0ed51b3 |
2144 | register I32 value; |
2145 | register I32 lastvalue = 1234; |
a0d0e21e |
2146 | register I32 range = 0; |
c277df42 |
2147 | register regnode *ret; |
a0d0e21e |
2148 | register I32 def; |
2149 | I32 numlen; |
2150 | |
3280af22 |
2151 | s = opnd = (char *) OPERAND(PL_regcode); |
c277df42 |
2152 | ret = reg_node(ANYOF); |
a0ed51b3 |
2153 | for (value = 0; value < 33; value++) |
c277df42 |
2154 | regc(0, s++); |
3280af22 |
2155 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2156 | PL_regnaughty++; |
2157 | PL_regcomp_parse++; |
c277df42 |
2158 | if (!SIZE_ONLY) |
bbce6d69 |
2159 | *opnd |= ANYOF_INVERT; |
2160 | } |
c277df42 |
2161 | if (!SIZE_ONLY) { |
3280af22 |
2162 | PL_regcode += ANY_SKIP; |
a0ed51b3 |
2163 | if (FOLD) |
bbce6d69 |
2164 | *opnd |= ANYOF_FOLD; |
a0ed51b3 |
2165 | if (LOC) |
bbce6d69 |
2166 | *opnd |= ANYOF_LOCALE; |
a0ed51b3 |
2167 | } |
2168 | else { |
3280af22 |
2169 | PL_regsize += ANY_SKIP; |
a0d0e21e |
2170 | } |
3280af22 |
2171 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
a0d0e21e |
2172 | goto skipcond; /* allow 1st char to be ] or - */ |
3280af22 |
2173 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
a0d0e21e |
2174 | skipcond: |
a0ed51b3 |
2175 | value = UCHARAT(PL_regcomp_parse++); |
620e46c5 |
2176 | if (value == '[') |
2177 | (void)regpposixcc(value); /* ignore the return value for now */ |
2178 | else if (value == '\\') { |
a0ed51b3 |
2179 | value = UCHARAT(PL_regcomp_parse++); |
2180 | switch (value) { |
a0d0e21e |
2181 | case 'w': |
ae5c130c |
2182 | if (!SIZE_ONLY) { |
a0ed51b3 |
2183 | if (LOC) |
bbce6d69 |
2184 | *opnd |= ANYOF_ALNUML; |
ae5c130c |
2185 | else { |
a0ed51b3 |
2186 | for (value = 0; value < 256; value++) |
2187 | if (isALNUM(value)) |
2188 | ANYOF_SET(opnd, value); |
ae5c130c |
2189 | } |
bbce6d69 |
2190 | } |
a0ed51b3 |
2191 | lastvalue = 1234; |
a0d0e21e |
2192 | continue; |
2193 | case 'W': |
ae5c130c |
2194 | if (!SIZE_ONLY) { |
a0ed51b3 |
2195 | if (LOC) |
bbce6d69 |
2196 | *opnd |= ANYOF_NALNUML; |
ae5c130c |
2197 | else { |
a0ed51b3 |
2198 | for (value = 0; value < 256; value++) |
2199 | if (!isALNUM(value)) |
2200 | ANYOF_SET(opnd, value); |
ae5c130c |
2201 | } |
bbce6d69 |
2202 | } |
a0ed51b3 |
2203 | lastvalue = 1234; |
a0d0e21e |
2204 | continue; |
2205 | case 's': |
ae5c130c |
2206 | if (!SIZE_ONLY) { |
a0ed51b3 |
2207 | if (LOC) |
bbce6d69 |
2208 | *opnd |= ANYOF_SPACEL; |
ae5c130c |
2209 | else { |
a0ed51b3 |
2210 | for (value = 0; value < 256; value++) |
2211 | if (isSPACE(value)) |
2212 | ANYOF_SET(opnd, value); |
ae5c130c |
2213 | } |
bbce6d69 |
2214 | } |
a0ed51b3 |
2215 | lastvalue = 1234; |
a0d0e21e |
2216 | continue; |
2217 | case 'S': |
ae5c130c |
2218 | if (!SIZE_ONLY) { |
a0ed51b3 |
2219 | if (LOC) |
bbce6d69 |
2220 | *opnd |= ANYOF_NSPACEL; |
ae5c130c |
2221 | else { |
a0ed51b3 |
2222 | for (value = 0; value < 256; value++) |
2223 | if (!isSPACE(value)) |
2224 | ANYOF_SET(opnd, value); |
ae5c130c |
2225 | } |
bbce6d69 |
2226 | } |
a0ed51b3 |
2227 | lastvalue = 1234; |
a0d0e21e |
2228 | continue; |
2229 | case 'd': |
ae5c130c |
2230 | if (!SIZE_ONLY) { |
a0ed51b3 |
2231 | for (value = '0'; value <= '9'; value++) |
2232 | ANYOF_SET(opnd, value); |
ae5c130c |
2233 | } |
a0ed51b3 |
2234 | lastvalue = 1234; |
a0d0e21e |
2235 | continue; |
2236 | case 'D': |
ae5c130c |
2237 | if (!SIZE_ONLY) { |
a0ed51b3 |
2238 | for (value = 0; value < '0'; value++) |
2239 | ANYOF_SET(opnd, value); |
2240 | for (value = '9' + 1; value < 256; value++) |
2241 | ANYOF_SET(opnd, value); |
ae5c130c |
2242 | } |
a0ed51b3 |
2243 | lastvalue = 1234; |
a0d0e21e |
2244 | continue; |
2245 | case 'n': |
a0ed51b3 |
2246 | value = '\n'; |
a0d0e21e |
2247 | break; |
2248 | case 'r': |
a0ed51b3 |
2249 | value = '\r'; |
a0d0e21e |
2250 | break; |
2251 | case 't': |
a0ed51b3 |
2252 | value = '\t'; |
a0d0e21e |
2253 | break; |
2254 | case 'f': |
a0ed51b3 |
2255 | value = '\f'; |
a0d0e21e |
2256 | break; |
2257 | case 'b': |
a0ed51b3 |
2258 | value = '\b'; |
a0d0e21e |
2259 | break; |
2260 | case 'e': |
a0ed51b3 |
2261 | value = '\033'; |
a0d0e21e |
2262 | break; |
2263 | case 'a': |
a0ed51b3 |
2264 | value = '\007'; |
a0d0e21e |
2265 | break; |
2266 | case 'x': |
a0ed51b3 |
2267 | value = scan_hex(PL_regcomp_parse, 2, &numlen); |
3280af22 |
2268 | PL_regcomp_parse += numlen; |
a0d0e21e |
2269 | break; |
2270 | case 'c': |
a0ed51b3 |
2271 | value = UCHARAT(PL_regcomp_parse++); |
2272 | value = toCTRL(value); |
a0d0e21e |
2273 | break; |
2274 | case '0': case '1': case '2': case '3': case '4': |
2275 | case '5': case '6': case '7': case '8': case '9': |
a0ed51b3 |
2276 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); |
3280af22 |
2277 | PL_regcomp_parse += numlen; |
a0d0e21e |
2278 | break; |
2279 | } |
2280 | } |
2281 | if (range) { |
a0ed51b3 |
2282 | if (lastvalue > value) |
a0d0e21e |
2283 | FAIL("invalid [] range in regexp"); |
2284 | range = 0; |
2285 | } |
2286 | else { |
a0ed51b3 |
2287 | lastvalue = value; |
3280af22 |
2288 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
2289 | PL_regcomp_parse[1] != ']') { |
2290 | PL_regcomp_parse++; |
a0d0e21e |
2291 | range = 1; |
2292 | continue; /* do it next time */ |
2293 | } |
a687059c |
2294 | } |
ae5c130c |
2295 | if (!SIZE_ONLY) { |
8ada0baa |
2296 | #ifndef ASCIIish |
2297 | if ((isLOWER(lastvalue) && isLOWER(value)) || |
c8dba6f3 |
2298 | (isUPPER(lastvalue) && isUPPER(value))) |
2299 | { |
2300 | I32 i; |
8ada0baa |
2301 | if (isLOWER(lastvalue)) { |
2302 | for (i = lastvalue; i <= value; i++) |
2303 | if (isLOWER(i)) |
2304 | ANYOF_SET(opnd, i); |
2305 | } else { |
2306 | for (i = lastvalue; i <= value; i++) |
2307 | if (isUPPER(i)) |
2308 | ANYOF_SET(opnd, i); |
2309 | } |
2310 | } |
2311 | else |
2312 | #endif |
2313 | for ( ; lastvalue <= value; lastvalue++) |
2314 | ANYOF_SET(opnd, lastvalue); |
2315 | } |
a0ed51b3 |
2316 | lastvalue = value; |
a0d0e21e |
2317 | } |
ae5c130c |
2318 | /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ |
2319 | if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) { |
a0ed51b3 |
2320 | for (value = 0; value < 256; ++value) { |
2321 | if (ANYOF_TEST(opnd, value)) { |
22c35a8c |
2322 | I32 cf = PL_fold[value]; |
ae5c130c |
2323 | ANYOF_SET(opnd, cf); |
2324 | } |
2325 | } |
2326 | *opnd &= ~ANYOF_FOLD; |
2327 | } |
2328 | /* optimize inverted simple patterns (e.g. [^a-z]) */ |
2329 | if (!SIZE_ONLY && (*opnd & 0xFF) == ANYOF_INVERT) { |
a0ed51b3 |
2330 | for (value = 0; value < 32; ++value) |
2331 | opnd[1 + value] ^= 0xFF; |
ae5c130c |
2332 | *opnd = 0; |
2333 | } |
a0d0e21e |
2334 | return ret; |
2335 | } |
2336 | |
a0ed51b3 |
2337 | STATIC regnode * |
cea2e8a9 |
2338 | S_regclassutf8(pTHX) |
a0ed51b3 |
2339 | { |
2340 | register char *opnd, *e; |
2341 | register U32 value; |
2342 | register U32 lastvalue = 123456; |
2343 | register I32 range = 0; |
2344 | register regnode *ret; |
2345 | I32 numlen; |
2346 | I32 n; |
2347 | SV *listsv; |
2348 | U8 flags = 0; |
c485e607 |
2349 | dTHR; |
a0ed51b3 |
2350 | |
2351 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2352 | PL_regnaughty++; |
2353 | PL_regcomp_parse++; |
2354 | if (!SIZE_ONLY) |
2355 | flags |= ANYOF_INVERT; |
2356 | } |
2357 | if (!SIZE_ONLY) { |
2358 | if (FOLD) |
2359 | flags |= ANYOF_FOLD; |
2360 | if (LOC) |
2361 | flags |= ANYOF_LOCALE; |
79cb57f6 |
2362 | listsv = newSVpvn("# comment\n",10); |
a0ed51b3 |
2363 | } |
2364 | |
2365 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
2366 | goto skipcond; /* allow 1st char to be ] or - */ |
2367 | |
2368 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
2369 | skipcond: |
dfe13c55 |
2370 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 |
2371 | PL_regcomp_parse += numlen; |
2372 | |
620e46c5 |
2373 | if (value == '[') |
2374 | (void)regpposixcc(value); /* ignore the return value for now */ |
2375 | else if (value == '\\') { |
dfe13c55 |
2376 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 |
2377 | PL_regcomp_parse += numlen; |
2378 | switch (value) { |
2379 | case 'w': |
2380 | if (!SIZE_ONLY) { |
2381 | if (LOC) |
2382 | flags |= ANYOF_ALNUML; |
2383 | |
cea2e8a9 |
2384 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n"); |
a0ed51b3 |
2385 | } |
2386 | lastvalue = 123456; |
2387 | continue; |
2388 | case 'W': |
2389 | if (!SIZE_ONLY) { |
2390 | if (LOC) |
2391 | flags |= ANYOF_NALNUML; |
2392 | |
cea2e8a9 |
2393 | Perl_sv_catpvf(aTHX_ listsv, |
a0ed51b3 |
2394 | "-utf8::IsAlpha\n-utf8::IsDigit\n0000\t%04x\n%04x\tffff\n", |
2395 | '_' - 1, |
2396 | '_' + 1); |
2397 | } |
2398 | lastvalue = 123456; |
2399 | continue; |
2400 | case 's': |
2401 | if (!SIZE_ONLY) { |
2402 | if (LOC) |
2403 | flags |= ANYOF_SPACEL; |
cea2e8a9 |
2404 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n"); |
a0ed51b3 |
2405 | if (!PL_utf8_space) |
dfe13c55 |
2406 | is_utf8_space((U8*)" "); |
a0ed51b3 |
2407 | } |
2408 | lastvalue = 123456; |
2409 | continue; |
2410 | case 'S': |
2411 | if (!SIZE_ONLY) { |
2412 | if (LOC) |
2413 | flags |= ANYOF_NSPACEL; |
cea2e8a9 |
2414 | Perl_sv_catpvf(aTHX_ listsv, |
a0ed51b3 |
2415 | "!utf8::IsSpace\n"); |
2416 | if (!PL_utf8_space) |
dfe13c55 |
2417 | is_utf8_space((U8*)" "); |
a0ed51b3 |
2418 | } |
2419 | lastvalue = 123456; |
2420 | continue; |
2421 | case 'd': |
2422 | if (!SIZE_ONLY) { |
cea2e8a9 |
2423 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n"); |
a0ed51b3 |
2424 | } |
2425 | lastvalue = 123456; |
2426 | continue; |
2427 | case 'D': |
2428 | if (!SIZE_ONLY) { |
cea2e8a9 |
2429 | Perl_sv_catpvf(aTHX_ listsv, |
a0ed51b3 |
2430 | "!utf8::IsDigit\n"); |
2431 | } |
2432 | lastvalue = 123456; |
2433 | continue; |
2434 | case 'p': |
2435 | case 'P': |
2436 | if (*PL_regcomp_parse == '{') { |
2437 | e = strchr(PL_regcomp_parse++, '}'); |
2438 | if (!e) |
2439 | FAIL("Missing right brace on \\p{}"); |
2440 | n = e - PL_regcomp_parse; |
2441 | } |
2442 | else { |
2443 | e = PL_regcomp_parse; |
2444 | n = 1; |
2445 | } |
2446 | if (!SIZE_ONLY) { |
2447 | if (value == 'p') |
cea2e8a9 |
2448 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::%.*s\n", n, PL_regcomp_parse); |
a0ed51b3 |
2449 | else |
cea2e8a9 |
2450 | Perl_sv_catpvf(aTHX_ listsv, |
a0ed51b3 |
2451 | "!utf8::%.*s\n", n, PL_regcomp_parse); |
2452 | } |
2453 | PL_regcomp_parse = e + 1; |
2454 | lastvalue = 123456; |
2455 | continue; |
2456 | case 'n': |
2457 | value = '\n'; |
2458 | break; |
2459 | case 'r': |
2460 | value = '\r'; |
2461 | break; |
2462 | case 't': |
2463 | value = '\t'; |
2464 | break; |
2465 | case 'f': |
2466 | value = '\f'; |
2467 | break; |
2468 | case 'b': |
2469 | value = '\b'; |
2470 | break; |
2471 | case 'e': |
2472 | value = '\033'; |
2473 | break; |
2474 | case 'a': |
2475 | value = '\007'; |
2476 | break; |
2477 | case 'x': |
2478 | if (*PL_regcomp_parse == '{') { |
2479 | e = strchr(PL_regcomp_parse++, '}'); |
2480 | if (!e) |
2481 | FAIL("Missing right brace on \\x{}"); |
f96ec2a2 |
2482 | value = scan_hex(PL_regcomp_parse, e - PL_regcomp_parse, &numlen); |
a0ed51b3 |
2483 | PL_regcomp_parse = e + 1; |
2484 | } |
2485 | else { |
2486 | value = scan_hex(PL_regcomp_parse, 2, &numlen); |
2487 | PL_regcomp_parse += numlen; |
2488 | } |
2489 | break; |
2490 | case 'c': |
2491 | value = UCHARAT(PL_regcomp_parse++); |
2492 | value = toCTRL(value); |
2493 | break; |
2494 | case '0': case '1': case '2': case '3': case '4': |
2495 | case '5': case '6': case '7': case '8': case '9': |
2496 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); |
2497 | PL_regcomp_parse += numlen; |
2498 | break; |
2499 | } |
2500 | } |
2501 | if (range) { |
2502 | if (lastvalue > value) |
2503 | FAIL("invalid [] range in regexp"); |
2504 | if (!SIZE_ONLY) |
cea2e8a9 |
2505 | Perl_sv_catpvf(aTHX_ listsv, "%04x\t%04x\n", lastvalue, value); |
a0ed51b3 |
2506 | lastvalue = value; |
2507 | range = 0; |
2508 | } |
2509 | else { |
2510 | lastvalue = value; |
2511 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
2512 | PL_regcomp_parse[1] != ']') { |
2513 | PL_regcomp_parse++; |
2514 | range = 1; |
2515 | continue; /* do it next time */ |
2516 | } |
2517 | if (!SIZE_ONLY) |
cea2e8a9 |
2518 | Perl_sv_catpvf(aTHX_ listsv, "%04x\n", value); |
a0ed51b3 |
2519 | } |
2520 | } |
a0ed51b3 |
2521 | |
2522 | ret = reganode(ANYOFUTF8, 0); |
2523 | |
2524 | if (!SIZE_ONLY) { |
2525 | SV *rv = swash_init("utf8", "", listsv, 1, 0); |
2526 | SvREFCNT_dec(listsv); |
2527 | n = add_data(1,"s"); |
2528 | PL_regcomp_rx->data->data[n] = (void*)rv; |
2529 | ARG1_SET(ret, flags); |
2530 | ARG2_SET(ret, n); |
2531 | } |
2532 | |
2533 | return ret; |
2534 | } |
2535 | |
76e3520e |
2536 | STATIC char* |
cea2e8a9 |
2537 | S_nextchar(pTHX) |
a0d0e21e |
2538 | { |
5c0ca799 |
2539 | dTHR; |
3280af22 |
2540 | char* retval = PL_regcomp_parse++; |
a0d0e21e |
2541 | |
4633a7c4 |
2542 | for (;;) { |
3280af22 |
2543 | if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' && |
2544 | PL_regcomp_parse[2] == '#') { |
2545 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
2546 | PL_regcomp_parse++; |
2547 | PL_regcomp_parse++; |
4633a7c4 |
2548 | continue; |
2549 | } |
3280af22 |
2550 | if (PL_regflags & PMf_EXTENDED) { |
2551 | if (isSPACE(*PL_regcomp_parse)) { |
2552 | PL_regcomp_parse++; |
748a9306 |
2553 | continue; |
2554 | } |
3280af22 |
2555 | else if (*PL_regcomp_parse == '#') { |
2556 | while (*PL_regcomp_parse && *PL_regcomp_parse != '\n') |
2557 | PL_regcomp_parse++; |
2558 | PL_regcomp_parse++; |
748a9306 |
2559 | continue; |
2560 | } |
748a9306 |
2561 | } |
4633a7c4 |
2562 | return retval; |
a0d0e21e |
2563 | } |
a687059c |
2564 | } |
2565 | |
2566 | /* |
c277df42 |
2567 | - reg_node - emit a node |
a0d0e21e |
2568 | */ |
76e3520e |
2569 | STATIC regnode * /* Location. */ |
cea2e8a9 |
2570 | S_reg_node(pTHX_ U8 op) |
a687059c |
2571 | { |
5c0ca799 |
2572 | dTHR; |
c277df42 |
2573 | register regnode *ret; |
2574 | register regnode *ptr; |
a687059c |
2575 | |
3280af22 |
2576 | ret = PL_regcode; |
c277df42 |
2577 | if (SIZE_ONLY) { |
6b88bc9c |
2578 | SIZE_ALIGN(PL_regsize); |
3280af22 |
2579 | PL_regsize += 1; |
a0d0e21e |
2580 | return(ret); |
2581 | } |
a687059c |
2582 | |
c277df42 |
2583 | NODE_ALIGN_FILL(ret); |
a0d0e21e |
2584 | ptr = ret; |
c277df42 |
2585 | FILL_ADVANCE_NODE(ptr, op); |
3280af22 |
2586 | PL_regcode = ptr; |
a687059c |
2587 | |
a0d0e21e |
2588 | return(ret); |
a687059c |
2589 | } |
2590 | |
2591 | /* |
a0d0e21e |
2592 | - reganode - emit a node with an argument |
2593 | */ |
76e3520e |
2594 | STATIC regnode * /* Location. */ |
cea2e8a9 |
2595 | S_reganode(pTHX_ U8 op, U32 arg) |
fe14fcc3 |
2596 | { |
5c0ca799 |
2597 | dTHR; |
c277df42 |
2598 | register regnode *ret; |
2599 | register regnode *ptr; |
fe14fcc3 |
2600 | |
3280af22 |
2601 | ret = PL_regcode; |
c277df42 |
2602 | if (SIZE_ONLY) { |
6b88bc9c |
2603 | SIZE_ALIGN(PL_regsize); |
3280af22 |
2604 | PL_regsize += 2; |
a0d0e21e |
2605 | return(ret); |
2606 | } |
fe14fcc3 |
2607 | |
c277df42 |
2608 | NODE_ALIGN_FILL(ret); |
a0d0e21e |
2609 | ptr = ret; |
c277df42 |
2610 | FILL_ADVANCE_NODE_ARG(ptr, op, arg); |
3280af22 |
2611 | PL_regcode = ptr; |
fe14fcc3 |
2612 | |
a0d0e21e |
2613 | return(ret); |
fe14fcc3 |
2614 | } |
2615 | |
2616 | /* |
a0ed51b3 |
2617 | - regc - emit (if appropriate) a Unicode character |
2618 | */ |
2619 | STATIC void |
cea2e8a9 |
2620 | S_reguni(pTHX_ UV uv, char* s, I32* lenp) |
a0ed51b3 |
2621 | { |
c485e607 |
2622 | dTHR; |
a0ed51b3 |
2623 | if (SIZE_ONLY) { |
dfe13c55 |
2624 | U8 tmpbuf[10]; |
a0ed51b3 |
2625 | *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf; |
2626 | } |
2627 | else |
dfe13c55 |
2628 | *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s; |
a0ed51b3 |
2629 | |
2630 | } |
2631 | |
2632 | /* |
a0d0e21e |
2633 | - regc - emit (if appropriate) a byte of code |
2634 | */ |
76e3520e |
2635 | STATIC void |
cea2e8a9 |
2636 | S_regc(pTHX_ U8 b, char* s) |
a687059c |
2637 | { |
5c0ca799 |
2638 | dTHR; |
c277df42 |
2639 | if (!SIZE_ONLY) |
2640 | *s = b; |
a687059c |
2641 | } |
2642 | |
2643 | /* |
a0d0e21e |
2644 | - reginsert - insert an operator in front of already-emitted operand |
2645 | * |
2646 | * Means relocating the operand. |
2647 | */ |
76e3520e |
2648 | STATIC void |
cea2e8a9 |
2649 | S_reginsert(pTHX_ U8 op, regnode *opnd) |
a687059c |
2650 | { |
5c0ca799 |
2651 | dTHR; |
c277df42 |
2652 | register regnode *src; |
2653 | register regnode *dst; |
2654 | register regnode *place; |
2655 | register int offset = regarglen[(U8)op]; |
2656 | |
22c35a8c |
2657 | /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */ |
c277df42 |
2658 | |
2659 | if (SIZE_ONLY) { |
3280af22 |
2660 | PL_regsize += NODE_STEP_REGNODE + offset; |
a0d0e21e |
2661 | return; |
2662 | } |
a687059c |
2663 | |
3280af22 |
2664 | src = PL_regcode; |
2665 | PL_regcode += NODE_STEP_REGNODE + offset; |
2666 | dst = PL_regcode; |
a0d0e21e |
2667 | while (src > opnd) |
c277df42 |
2668 | StructCopy(--src, --dst, regnode); |
a0d0e21e |
2669 | |
2670 | place = opnd; /* Op node, where operand used to be. */ |
c277df42 |
2671 | src = NEXTOPER(place); |
2672 | FILL_ADVANCE_NODE(place, op); |
2673 | Zero(src, offset, regnode); |
a687059c |
2674 | } |
2675 | |
2676 | /* |
c277df42 |
2677 | - regtail - set the next-pointer at the end of a node chain of p to val. |
a0d0e21e |
2678 | */ |
76e3520e |
2679 | STATIC void |
cea2e8a9 |
2680 | S_regtail(pTHX_ regnode *p, regnode *val) |
a687059c |
2681 | { |
5c0ca799 |
2682 | dTHR; |
c277df42 |
2683 | register regnode *scan; |
2684 | register regnode *temp; |
a0d0e21e |
2685 | register I32 offset; |
2686 | |
c277df42 |
2687 | if (SIZE_ONLY) |
a0d0e21e |
2688 | return; |
2689 | |
2690 | /* Find last node. */ |
2691 | scan = p; |
2692 | for (;;) { |
2693 | temp = regnext(scan); |
2694 | if (temp == NULL) |
2695 | break; |
2696 | scan = temp; |
2697 | } |
a687059c |
2698 | |
c277df42 |
2699 | if (reg_off_by_arg[OP(scan)]) { |
2700 | ARG_SET(scan, val - scan); |
a0ed51b3 |
2701 | } |
2702 | else { |
c277df42 |
2703 | NEXT_OFF(scan) = val - scan; |
2704 | } |
a687059c |
2705 | } |
2706 | |
2707 | /* |
a0d0e21e |
2708 | - regoptail - regtail on operand of first argument; nop if operandless |
2709 | */ |
76e3520e |
2710 | STATIC void |
cea2e8a9 |
2711 | S_regoptail(pTHX_ regnode *p, regnode *val) |
a687059c |
2712 | { |
5c0ca799 |
2713 | dTHR; |
a0d0e21e |
2714 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
c277df42 |
2715 | if (p == NULL || SIZE_ONLY) |
2716 | return; |
22c35a8c |
2717 | if (PL_regkind[(U8)OP(p)] == BRANCH) { |
c277df42 |
2718 | regtail(NEXTOPER(p), val); |
a0ed51b3 |
2719 | } |
22c35a8c |
2720 | else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) { |
c277df42 |
2721 | regtail(NEXTOPER(NEXTOPER(p)), val); |
a0ed51b3 |
2722 | } |
2723 | else |
a0d0e21e |
2724 | return; |
a687059c |
2725 | } |
2726 | |
2727 | /* |
2728 | - regcurly - a little FSA that accepts {\d+,?\d*} |
2729 | */ |
79072805 |
2730 | STATIC I32 |
cea2e8a9 |
2731 | S_regcurly(pTHX_ register char *s) |
a687059c |
2732 | { |
2733 | if (*s++ != '{') |
2734 | return FALSE; |
f0fcb552 |
2735 | if (!isDIGIT(*s)) |
a687059c |
2736 | return FALSE; |
f0fcb552 |
2737 | while (isDIGIT(*s)) |
a687059c |
2738 | s++; |
2739 | if (*s == ',') |
2740 | s++; |
f0fcb552 |
2741 | while (isDIGIT(*s)) |
a687059c |
2742 | s++; |
2743 | if (*s != '}') |
2744 | return FALSE; |
2745 | return TRUE; |
2746 | } |
2747 | |
a687059c |
2748 | |
76e3520e |
2749 | STATIC regnode * |
cea2e8a9 |
2750 | S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l) |
c277df42 |
2751 | { |
35ff7856 |
2752 | #ifdef DEBUGGING |
c277df42 |
2753 | register char op = EXACT; /* Arbitrary non-END op. */ |
2754 | register regnode *next, *onode; |
2755 | |
2756 | while (op != END && (!last || node < last)) { |
2757 | /* While that wasn't END last time... */ |
2758 | |
2759 | NODE_ALIGN(node); |
2760 | op = OP(node); |
2761 | if (op == CLOSE) |
2762 | l--; |
2763 | next = regnext(node); |
2764 | /* Where, what. */ |
2765 | if (OP(node) == OPTIMIZED) |
2766 | goto after_print; |
2767 | regprop(sv, node); |
54dc92de |
2768 | PerlIO_printf(Perl_debug_log, "%4d:%*s%s", node - start, |
c277df42 |
2769 | 2*l + 1, "", SvPVX(sv)); |
2770 | if (next == NULL) /* Next ptr. */ |
2771 | PerlIO_printf(Perl_debug_log, "(0)"); |
2772 | else |
2773 | PerlIO_printf(Perl_debug_log, "(%d)", next - start); |
2774 | (void)PerlIO_putc(Perl_debug_log, '\n'); |
2775 | after_print: |
22c35a8c |
2776 | if (PL_regkind[(U8)op] == BRANCHJ) { |
c277df42 |
2777 | register regnode *nnode = (OP(next) == LONGJMP |
2778 | ? regnext(next) |
2779 | : next); |
2780 | if (last && nnode > last) |
2781 | nnode = last; |
2782 | node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1); |
a0ed51b3 |
2783 | } |
22c35a8c |
2784 | else if (PL_regkind[(U8)op] == BRANCH) { |
c277df42 |
2785 | node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1); |
a0ed51b3 |
2786 | } |
2787 | else if ( op == CURLY) { /* `next' might be very big: optimizer */ |
c277df42 |
2788 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
2789 | NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1); |
a0ed51b3 |
2790 | } |
22c35a8c |
2791 | else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) { |
c277df42 |
2792 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
2793 | next, sv, l + 1); |
a0ed51b3 |
2794 | } |
2795 | else if ( op == PLUS || op == STAR) { |
c277df42 |
2796 | node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1); |
a0ed51b3 |
2797 | } |
2798 | else if (op == ANYOF) { |
c277df42 |
2799 | node = NEXTOPER(node); |
2800 | node += ANY_SKIP; |
a0ed51b3 |
2801 | } |
22c35a8c |
2802 | else if (PL_regkind[(U8)op] == EXACT) { |
c277df42 |
2803 | /* Literal string, where present. */ |
2804 | node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
2805 | node = NEXTOPER(node); |
a0ed51b3 |
2806 | } |
2807 | else { |
c277df42 |
2808 | node = NEXTOPER(node); |
2809 | node += regarglen[(U8)op]; |
2810 | } |
2811 | if (op == CURLYX || op == OPEN) |
2812 | l++; |
2813 | else if (op == WHILEM) |
2814 | l--; |
2815 | } |
17c3b450 |
2816 | #endif /* DEBUGGING */ |
c277df42 |
2817 | return node; |
2818 | } |
2819 | |
a687059c |
2820 | /* |
fd181c75 |
2821 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c |
2822 | */ |
2823 | void |
864dbfa3 |
2824 | Perl_regdump(pTHX_ regexp *r) |
a687059c |
2825 | { |
35ff7856 |
2826 | #ifdef DEBUGGING |
5c0ca799 |
2827 | dTHR; |
46fc3d4c |
2828 | SV *sv = sv_newmortal(); |
a687059c |
2829 | |
c277df42 |
2830 | (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0); |
a0d0e21e |
2831 | |
2832 | /* Header fields of interest. */ |
c277df42 |
2833 | if (r->anchored_substr) |
2834 | PerlIO_printf(Perl_debug_log, "anchored `%s%s%s'%s at %d ", |
3280af22 |
2835 | PL_colors[0], |
c277df42 |
2836 | SvPVX(r->anchored_substr), |
3280af22 |
2837 | PL_colors[1], |
c277df42 |
2838 | SvTAIL(r->anchored_substr) ? "$" : "", |
2839 | r->anchored_offset); |
2840 | if (r->float_substr) |
2841 | PerlIO_printf(Perl_debug_log, "floating `%s%s%s'%s at %d..%u ", |
3280af22 |
2842 | PL_colors[0], |
c277df42 |
2843 | SvPVX(r->float_substr), |
3280af22 |
2844 | PL_colors[1], |
c277df42 |
2845 | SvTAIL(r->float_substr) ? "$" : "", |
2846 | r->float_min_offset, r->float_max_offset); |
2847 | if (r->check_substr) |
2848 | PerlIO_printf(Perl_debug_log, |
2849 | r->check_substr == r->float_substr |
2850 | ? "(checking floating" : "(checking anchored"); |
2851 | if (r->reganch & ROPT_NOSCAN) |
2852 | PerlIO_printf(Perl_debug_log, " noscan"); |
2853 | if (r->reganch & ROPT_CHECK_ALL) |
2854 | PerlIO_printf(Perl_debug_log, " isall"); |
2855 | if (r->check_substr) |
2856 | PerlIO_printf(Perl_debug_log, ") "); |
2857 | |
46fc3d4c |
2858 | if (r->regstclass) { |
2859 | regprop(sv, r->regstclass); |
2860 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv)); |
2861 | } |
774d564b |
2862 | if (r->reganch & ROPT_ANCH) { |
2863 | PerlIO_printf(Perl_debug_log, "anchored"); |
2864 | if (r->reganch & ROPT_ANCH_BOL) |
2865 | PerlIO_printf(Perl_debug_log, "(BOL)"); |
c277df42 |
2866 | if (r->reganch & ROPT_ANCH_MBOL) |
2867 | PerlIO_printf(Perl_debug_log, "(MBOL)"); |
f722798b |
2868 | if (r->reganch & ROPT_ANCH_SBOL) |
2869 | PerlIO_printf(Perl_debug_log, "(SBOL)"); |
774d564b |
2870 | if (r->reganch & ROPT_ANCH_GPOS) |
2871 | PerlIO_printf(Perl_debug_log, "(GPOS)"); |
2872 | PerlIO_putc(Perl_debug_log, ' '); |
2873 | } |
c277df42 |
2874 | if (r->reganch & ROPT_GPOS_SEEN) |
2875 | PerlIO_printf(Perl_debug_log, "GPOS "); |
a0d0e21e |
2876 | if (r->reganch & ROPT_SKIP) |
760ac839 |
2877 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e |
2878 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 |
2879 | PerlIO_printf(Perl_debug_log, "implicit "); |
760ac839 |
2880 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
ce862d02 |
2881 | if (r->reganch & ROPT_EVAL_SEEN) |
2882 | PerlIO_printf(Perl_debug_log, "with eval "); |
760ac839 |
2883 | PerlIO_printf(Perl_debug_log, "\n"); |
17c3b450 |
2884 | #endif /* DEBUGGING */ |
a687059c |
2885 | } |
2886 | |
2887 | /* |
a0d0e21e |
2888 | - regprop - printable representation of opcode |
2889 | */ |
46fc3d4c |
2890 | void |
864dbfa3 |
2891 | Perl_regprop(pTHX_ SV *sv, regnode *o) |
a687059c |
2892 | { |
35ff7856 |
2893 | #ifdef DEBUGGING |
5c0ca799 |
2894 | dTHR; |
9b155405 |
2895 | register int k; |
a0d0e21e |
2896 | |
54dc92de |
2897 | sv_setpvn(sv, "", 0); |
9b155405 |
2898 | if (OP(o) >= reg_num) /* regnode.type is unsigned */ |
a0d0e21e |
2899 | FAIL("corrupted regexp opcode"); |
9b155405 |
2900 | sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */ |
2901 | |
2902 | k = PL_regkind[(U8)OP(o)]; |
2903 | |
2904 | if (k == EXACT) |
cea2e8a9 |
2905 | Perl_sv_catpvf(aTHX_ sv, " <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]); |
9b155405 |
2906 | else if (k == CURLY) { |
2907 | if (OP(o) == CURLYM || OP(o) == CURLYN) |
cea2e8a9 |
2908 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */ |
2909 | Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o)); |
a0d0e21e |
2910 | } |
9b155405 |
2911 | else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP ) |
cea2e8a9 |
2912 | Perl_sv_catpvf(aTHX_ sv, "%d", ARG(o)); /* Parenth number */ |
9b155405 |
2913 | else if (k == LOGICAL) |
cea2e8a9 |
2914 | Perl_sv_catpvf(aTHX_ sv, "[%d]", ARG(o)); /* 2: embedded, otherwise 1 */ |
9b155405 |
2915 | else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH)) |
cea2e8a9 |
2916 | Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags); |
17c3b450 |
2917 | #endif /* DEBUGGING */ |
35ff7856 |
2918 | } |
a687059c |
2919 | |
f722798b |
2920 | SV * |
2921 | Perl_re_intuit_string(pTHX_ regexp *prog) |
2922 | { /* Assume that RE_INTUIT is set */ |
2923 | DEBUG_r( |
2924 | { STRLEN n_a; |
2925 | char *s = SvPV(prog->check_substr,n_a); |
2926 | |
2927 | if (!PL_colorset) reginitcolors(); |
2928 | PerlIO_printf(Perl_debug_log, |
2929 | "%sUsing REx substr:%s `%s%.60s%s%s'\n", |
2930 | PL_colors[4],PL_colors[5],PL_colors[0], |
2931 | s, |
2932 | PL_colors[1], |
2933 | (strlen(s) > 60 ? "..." : "")); |
2934 | } ); |
2935 | |
2936 | return prog->check_substr; |
2937 | } |
2938 | |
2b69d0c2 |
2939 | void |
864dbfa3 |
2940 | Perl_pregfree(pTHX_ struct regexp *r) |
a687059c |
2941 | { |
5c0ca799 |
2942 | dTHR; |
f722798b |
2943 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
2944 | "%sFreeing REx:%s `%s%.60s%s%s'\n", |
2945 | PL_colors[4],PL_colors[5],PL_colors[0], |
2946 | r->precomp, |
2947 | PL_colors[1], |
2948 | (strlen(r->precomp) > 60 ? "..." : ""))); |
2949 | |
2950 | |
c277df42 |
2951 | if (!r || (--r->refcnt > 0)) |
a0d0e21e |
2952 | return; |
c277df42 |
2953 | if (r->precomp) |
a0d0e21e |
2954 | Safefree(r->precomp); |
cf93c79d |
2955 | if (RX_MATCH_COPIED(r)) |
2956 | Safefree(r->subbeg); |
a193d654 |
2957 | if (r->substrs) { |
2958 | if (r->anchored_substr) |
2959 | SvREFCNT_dec(r->anchored_substr); |
2960 | if (r->float_substr) |
2961 | SvREFCNT_dec(r->float_substr); |
2779dcf1 |
2962 | Safefree(r->substrs); |
a193d654 |
2963 | } |
c277df42 |
2964 | if (r->data) { |
2965 | int n = r->data->count; |
dfad63ad |
2966 | AV* new_comppad = NULL; |
2967 | AV* old_comppad; |
2968 | SV** old_curpad; |
2969 | |
c277df42 |
2970 | while (--n >= 0) { |
2971 | switch (r->data->what[n]) { |
2972 | case 's': |
2973 | SvREFCNT_dec((SV*)r->data->data[n]); |
2974 | break; |
dfad63ad |
2975 | case 'p': |
2976 | new_comppad = (AV*)r->data->data[n]; |
2977 | break; |
c277df42 |
2978 | case 'o': |
dfad63ad |
2979 | if (new_comppad == NULL) |
cea2e8a9 |
2980 | Perl_croak(aTHX_ "panic: pregfree comppad"); |
dfad63ad |
2981 | old_comppad = PL_comppad; |
2982 | old_curpad = PL_curpad; |
2983 | PL_comppad = new_comppad; |
2984 | PL_curpad = AvARRAY(new_comppad); |
c277df42 |
2985 | op_free((OP_4tree*)r->data->data[n]); |
dfad63ad |
2986 | PL_comppad = old_comppad; |
2987 | PL_curpad = old_curpad; |
2988 | SvREFCNT_dec((SV*)new_comppad); |
2989 | new_comppad = NULL; |
c277df42 |
2990 | break; |
2991 | case 'n': |
2992 | break; |
2993 | default: |
2994 | FAIL2("panic: regfree data code '%c'", r->data->what[n]); |
2995 | } |
2996 | } |
2997 | Safefree(r->data->what); |
2998 | Safefree(r->data); |
a0d0e21e |
2999 | } |
3000 | Safefree(r->startp); |
3001 | Safefree(r->endp); |
3002 | Safefree(r); |
a687059c |
3003 | } |
c277df42 |
3004 | |
3005 | /* |
3006 | - regnext - dig the "next" pointer out of a node |
3007 | * |
3008 | * [Note, when REGALIGN is defined there are two places in regmatch() |
3009 | * that bypass this code for speed.] |
3010 | */ |
3011 | regnode * |
864dbfa3 |
3012 | Perl_regnext(pTHX_ register regnode *p) |
c277df42 |
3013 | { |
5c0ca799 |
3014 | dTHR; |
c277df42 |
3015 | register I32 offset; |
3016 | |
3280af22 |
3017 | if (p == &PL_regdummy) |
c277df42 |
3018 | return(NULL); |
3019 | |
3020 | offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p)); |
3021 | if (offset == 0) |
3022 | return(NULL); |
3023 | |
c277df42 |
3024 | return(p+offset); |
c277df42 |
3025 | } |
3026 | |
01f988be |
3027 | STATIC void |
cea2e8a9 |
3028 | S_re_croak2(pTHX_ const char* pat1,const char* pat2,...) |
c277df42 |
3029 | { |
3030 | va_list args; |
3031 | STRLEN l1 = strlen(pat1); |
3032 | STRLEN l2 = strlen(pat2); |
3033 | char buf[512]; |
06bf62c7 |
3034 | SV *msv; |
c277df42 |
3035 | char *message; |
3036 | |
3037 | if (l1 > 510) |
3038 | l1 = 510; |
3039 | if (l1 + l2 > 510) |
3040 | l2 = 510 - l1; |
3041 | Copy(pat1, buf, l1 , char); |
3042 | Copy(pat2, buf + l1, l2 , char); |
3b818b81 |
3043 | buf[l1 + l2] = '\n'; |
3044 | buf[l1 + l2 + 1] = '\0'; |
8736538c |
3045 | #ifdef I_STDARG |
3046 | /* ANSI variant takes additional second argument */ |
c277df42 |
3047 | va_start(args, pat2); |
8736538c |
3048 | #else |
3049 | va_start(args); |
3050 | #endif |
06bf62c7 |
3051 | msv = mess(buf, &args); |
c277df42 |
3052 | va_end(args); |
06bf62c7 |
3053 | message = SvPV(msv,l1); |
c277df42 |
3054 | if (l1 > 512) |
3055 | l1 = 512; |
3056 | Copy(message, buf, l1 , char); |
3057 | buf[l1] = '\0'; /* Overwrite \n */ |
cea2e8a9 |
3058 | Perl_croak(aTHX_ "%s", buf); |
c277df42 |
3059 | } |
a0ed51b3 |
3060 | |
3061 | /* XXX Here's a total kludge. But we need to re-enter for swash routines. */ |
3062 | |
3063 | void |
864dbfa3 |
3064 | Perl_save_re_context(pTHX) |
c485e607 |
3065 | { |
3066 | dTHR; |
a0ed51b3 |
3067 | SAVEPPTR(PL_bostr); |
3068 | SAVEPPTR(PL_regprecomp); /* uncompiled string. */ |
3069 | SAVEI32(PL_regnpar); /* () count. */ |
3070 | SAVEI32(PL_regsize); /* Code size. */ |
3071 | SAVEI16(PL_regflags); /* are we folding, multilining? */ |
3072 | SAVEPPTR(PL_reginput); /* String-input pointer. */ |
3073 | SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */ |
3074 | SAVEPPTR(PL_regeol); /* End of input, for $ check. */ |
3075 | SAVESPTR(PL_regstartp); /* Pointer to startp array. */ |
3076 | SAVESPTR(PL_regendp); /* Ditto for endp. */ |
3077 | SAVESPTR(PL_reglastparen); /* Similarly for lastparen. */ |
3078 | SAVEPPTR(PL_regtill); /* How far we are required to go. */ |
3079 | SAVEI32(PL_regprev); /* char before regbol, \n if none */ |
3080 | SAVESPTR(PL_reg_start_tmp); /* from regexec.c */ |
3081 | PL_reg_start_tmp = 0; |
3082 | SAVEFREEPV(PL_reg_start_tmp); |
3083 | SAVEI32(PL_reg_start_tmpl); /* from regexec.c */ |
3084 | PL_reg_start_tmpl = 0; |
3085 | SAVESPTR(PL_regdata); |
3086 | SAVEI32(PL_reg_flags); /* from regexec.c */ |
3087 | SAVEI32(PL_reg_eval_set); /* from regexec.c */ |
3088 | SAVEI32(PL_regnarrate); /* from regexec.c */ |
3089 | SAVESPTR(PL_regprogram); /* from regexec.c */ |
3090 | SAVEINT(PL_regindent); /* from regexec.c */ |
3091 | SAVESPTR(PL_regcc); /* from regexec.c */ |
3092 | SAVESPTR(PL_curcop); |
3093 | SAVESPTR(PL_regcomp_rx); /* from regcomp.c */ |
3094 | SAVEI32(PL_regseen); /* from regcomp.c */ |
3095 | SAVEI32(PL_regsawback); /* Did we see \1, ...? */ |
3096 | SAVEI32(PL_regnaughty); /* How bad is this pattern? */ |
3097 | SAVESPTR(PL_regcode); /* Code-emit pointer; ®dummy = don't */ |
3098 | SAVEPPTR(PL_regxend); /* End of input for compile */ |
3099 | SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */ |
54b6e2fa |
3100 | SAVESPTR(PL_reg_call_cc); /* from regexec.c */ |
3101 | SAVESPTR(PL_reg_re); /* from regexec.c */ |
3102 | SAVEPPTR(PL_reg_ganch); /* from regexec.c */ |
3103 | SAVESPTR(PL_reg_sv); /* from regexec.c */ |
3104 | SAVESPTR(PL_reg_magic); /* from regexec.c */ |
3105 | SAVEI32(PL_reg_oldpos); /* from regexec.c */ |
3106 | SAVESPTR(PL_reg_oldcurpm); /* from regexec.c */ |
3107 | SAVESPTR(PL_reg_curpm); /* from regexec.c */ |
3108 | #ifdef DEBUGGING |
3109 | SAVEPPTR(PL_reg_starttry); /* from regexec.c */ |
3110 | #endif |
a0ed51b3 |
3111 | } |