Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | /* regcomp.c |
2 | */ | |
3 | ||
4 | /* | |
5 | * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee | |
6 | */ | |
7 | ||
a687059c LW |
8 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
9 | * confused with the original package (see point 3 below). Thanks, Henry! | |
10 | */ | |
11 | ||
12 | /* Additional note: this code is very heavily munged from Henry's version | |
13 | * in places. In some spots I've traded clarity for efficiency, so don't | |
14 | * blame Henry for some of the lack of readability. | |
15 | */ | |
16 | ||
e50aee73 AD |
17 | /* The names of the functions have been changed from regcomp and |
18 | * regexec to pregcomp and pregexec in order to avoid conflicts | |
19 | * with the POSIX routines of the same names. | |
20 | */ | |
21 | ||
b9d5759e AD |
22 | #ifdef PERL_EXT_RE_BUILD |
23 | /* need to replace pregcomp et al, so enable that */ | |
24 | # ifndef PERL_IN_XSUB_RE | |
25 | # define PERL_IN_XSUB_RE | |
26 | # endif | |
27 | /* need access to debugger hooks */ | |
cad2e5aa | 28 | # if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING) |
b9d5759e AD |
29 | # define DEBUGGING |
30 | # endif | |
31 | #endif | |
32 | ||
33 | #ifdef PERL_IN_XSUB_RE | |
d06ea78c | 34 | /* We *really* need to overwrite these symbols: */ |
56953603 IZ |
35 | # define Perl_pregcomp my_regcomp |
36 | # define Perl_regdump my_regdump | |
37 | # define Perl_regprop my_regprop | |
d06ea78c | 38 | # define Perl_pregfree my_regfree |
cad2e5aa JH |
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 |
c5be433b GS |
44 | |
45 | # define PERL_NO_GET_CONTEXT | |
56953603 IZ |
46 | #endif |
47 | ||
f0fcb552 | 48 | /*SUPPRESS 112*/ |
a687059c | 49 | /* |
e50aee73 | 50 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c LW |
51 | * |
52 | * Copyright (c) 1986 by University of Toronto. | |
53 | * Written by Henry Spencer. Not derived from licensed software. | |
54 | * | |
55 | * Permission is granted to anyone to use this software for any | |
56 | * purpose on any computer system, and to redistribute it freely, | |
57 | * subject to the following restrictions: | |
58 | * | |
59 | * 1. The author is not responsible for the consequences of use of | |
60 | * this software, no matter how awful, even if they arise | |
61 | * from defects in it. | |
62 | * | |
63 | * 2. The origin of this software must not be misrepresented, either | |
64 | * by explicit claim or by omission. | |
65 | * | |
66 | * 3. Altered versions must be plainly marked as such, and must not | |
67 | * be misrepresented as being the original software. | |
68 | * | |
69 | * | |
70 | **** Alterations to Henry's code are... | |
71 | **** | |
4eb8286e | 72 | **** Copyright (c) 1991-1999, Larry Wall |
a687059c | 73 | **** |
9ef589d8 LW |
74 | **** You may distribute under the terms of either the GNU General Public |
75 | **** License or the Artistic License, as specified in the README file. | |
76 | ||
a687059c LW |
77 | * |
78 | * Beware that some of this code is subtly aware of the way operator | |
79 | * precedence is structured in regular expressions. Serious changes in | |
80 | * regular-expression syntax might require a total rethink. | |
81 | */ | |
82 | #include "EXTERN.h" | |
864dbfa3 | 83 | #define PERL_IN_REGCOMP_C |
a687059c | 84 | #include "perl.h" |
d06ea78c | 85 | |
c5be433b GS |
86 | #ifdef PERL_IN_XSUB_RE |
87 | # if defined(PERL_CAPI) || defined(PERL_OBJECT) | |
88 | # include "XSUB.h" | |
89 | # endif | |
90 | #else | |
d06ea78c GS |
91 | # include "INTERN.h" |
92 | #endif | |
c277df42 IZ |
93 | |
94 | #define REG_COMP_C | |
a687059c LW |
95 | #include "regcomp.h" |
96 | ||
d4cce5f1 | 97 | #ifdef op |
11343788 | 98 | #undef op |
d4cce5f1 | 99 | #endif /* op */ |
11343788 | 100 | |
fe14fcc3 LW |
101 | #ifdef MSDOS |
102 | # if defined(BUGGY_MSC6) | |
103 | /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */ | |
104 | # pragma optimize("a",off) | |
105 | /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/ | |
106 | # pragma optimize("w",on ) | |
107 | # endif /* BUGGY_MSC6 */ | |
108 | #endif /* MSDOS */ | |
109 | ||
a687059c LW |
110 | #ifndef STATIC |
111 | #define STATIC static | |
112 | #endif | |
113 | ||
114 | #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?') | |
115 | #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \ | |
116 | ((*s) == '{' && regcurly(s))) | |
2b69d0c2 LW |
117 | #ifdef atarist |
118 | #define PERL_META "^$.[()|?+*\\" | |
119 | #else | |
a687059c | 120 | #define META "^$.[()|?+*\\" |
2b69d0c2 | 121 | #endif |
a687059c | 122 | |
35c8bce7 LW |
123 | #ifdef SPSTART |
124 | #undef SPSTART /* dratted cpp namespace... */ | |
125 | #endif | |
a687059c LW |
126 | /* |
127 | * Flags to be passed up and down. | |
128 | */ | |
a687059c | 129 | #define WORST 0 /* Worst case. */ |
821b33a5 | 130 | #define HASWIDTH 0x1 /* Known to match non-null strings. */ |
a0d0e21e LW |
131 | #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ |
132 | #define SPSTART 0x4 /* Starts with * or +. */ | |
133 | #define TRYAGAIN 0x8 /* Weeded out a declaration. */ | |
a687059c LW |
134 | |
135 | /* | |
e50aee73 | 136 | * Forward declarations for pregcomp()'s friends. |
a687059c | 137 | */ |
a0d0e21e | 138 | |
cd488c12 JB |
139 | static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
140 | 0, 0, 0 }; | |
c277df42 IZ |
141 | |
142 | #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL) | |
143 | #define SF_BEFORE_SEOL 0x1 | |
144 | #define SF_BEFORE_MEOL 0x2 | |
145 | #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL) | |
146 | #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL) | |
147 | ||
09b7f37c CB |
148 | #ifdef NO_UNARY_PLUS |
149 | # define SF_FIX_SHIFT_EOL (0+2) | |
150 | # define SF_FL_SHIFT_EOL (0+4) | |
151 | #else | |
152 | # define SF_FIX_SHIFT_EOL (+2) | |
153 | # define SF_FL_SHIFT_EOL (+4) | |
154 | #endif | |
c277df42 IZ |
155 | |
156 | #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL) | |
157 | #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL) | |
158 | ||
159 | #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL) | |
160 | #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */ | |
161 | #define SF_IS_INF 0x40 | |
162 | #define SF_HAS_PAR 0x80 | |
163 | #define SF_IN_PAR 0x100 | |
164 | #define SF_HAS_EVAL 0x200 | |
4bfe0158 | 165 | #define SCF_DO_SUBSTR 0x400 |
c277df42 | 166 | |
a0ed51b3 LW |
167 | #define RF_utf8 8 |
168 | #define UTF (PL_reg_flags & RF_utf8) | |
169 | #define LOC (PL_regflags & PMf_LOCALE) | |
170 | #define FOLD (PL_regflags & PMf_FOLD) | |
171 | ||
b8c5462f JH |
172 | #define OOB_CHAR8 1234 |
173 | #define OOB_UTF8 123456 | |
174 | ||
a0ed51b3 LW |
175 | #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) |
176 | #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) | |
177 | ||
51371543 | 178 | static void clear_re(pTHXo_ void *r); |
4327152a IZ |
179 | |
180 | STATIC void | |
cea2e8a9 | 181 | S_scan_commit(pTHX_ scan_data_t *data) |
c277df42 | 182 | { |
c485e607 | 183 | dTHR; |
a0ed51b3 LW |
184 | STRLEN l = CHR_SVLEN(data->last_found); |
185 | STRLEN old_l = CHR_SVLEN(*data->longest); | |
c277df42 IZ |
186 | |
187 | if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) { | |
188 | sv_setsv(*data->longest, data->last_found); | |
189 | if (*data->longest == data->longest_fixed) { | |
190 | data->offset_fixed = l ? data->last_start_min : data->pos_min; | |
191 | if (data->flags & SF_BEFORE_EOL) | |
192 | data->flags | |
193 | |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL); | |
194 | else | |
195 | data->flags &= ~SF_FIX_BEFORE_EOL; | |
a0ed51b3 LW |
196 | } |
197 | else { | |
c277df42 IZ |
198 | data->offset_float_min = l ? data->last_start_min : data->pos_min; |
199 | data->offset_float_max = (l | |
200 | ? data->last_start_max | |
201 | : data->pos_min + data->pos_delta); | |
202 | if (data->flags & SF_BEFORE_EOL) | |
203 | data->flags | |
204 | |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL); | |
205 | else | |
206 | data->flags &= ~SF_FL_BEFORE_EOL; | |
207 | } | |
208 | } | |
209 | SvCUR_set(data->last_found, 0); | |
210 | data->last_end = -1; | |
211 | data->flags &= ~SF_BEFORE_EOL; | |
212 | } | |
213 | ||
c277df42 IZ |
214 | /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set |
215 | to the position after last scanned or to NULL. */ | |
216 | ||
76e3520e | 217 | STATIC I32 |
cea2e8a9 | 218 | S_study_chunk(pTHX_ regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags) |
c277df42 IZ |
219 | /* scanp: Start here (read-write). */ |
220 | /* deltap: Write maxlen-minlen here. */ | |
221 | /* last: Stop before this one. */ | |
222 | { | |
5c0ca799 | 223 | dTHR; |
c277df42 IZ |
224 | I32 min = 0, pars = 0, code; |
225 | regnode *scan = *scanp, *next; | |
226 | I32 delta = 0; | |
227 | int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF); | |
aca2d497 | 228 | int is_inf_internal = 0; /* The studied chunk is infinite */ |
c277df42 IZ |
229 | I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0; |
230 | scan_data_t data_fake; | |
231 | ||
232 | while (scan && OP(scan) != END && scan < last) { | |
233 | /* Peephole optimizer: */ | |
234 | ||
22c35a8c | 235 | if (PL_regkind[(U8)OP(scan)] == EXACT) { |
c277df42 IZ |
236 | regnode *n = regnext(scan); |
237 | U32 stringok = 1; | |
238 | #ifdef DEBUGGING | |
239 | regnode *stop = scan; | |
240 | #endif | |
241 | ||
242 | next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; | |
243 | /* Skip NOTHING, merge EXACT*. */ | |
244 | while (n && | |
22c35a8c | 245 | ( PL_regkind[(U8)OP(n)] == NOTHING || |
c277df42 IZ |
246 | (stringok && (OP(n) == OP(scan)))) |
247 | && NEXT_OFF(n) | |
248 | && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) { | |
249 | if (OP(n) == TAIL || n > next) | |
250 | stringok = 0; | |
22c35a8c | 251 | if (PL_regkind[(U8)OP(n)] == NOTHING) { |
c277df42 IZ |
252 | NEXT_OFF(scan) += NEXT_OFF(n); |
253 | next = n + NODE_STEP_REGNODE; | |
254 | #ifdef DEBUGGING | |
255 | if (stringok) | |
256 | stop = n; | |
257 | #endif | |
258 | n = regnext(n); | |
a0ed51b3 LW |
259 | } |
260 | else { | |
c277df42 IZ |
261 | int oldl = *OPERAND(scan); |
262 | regnode *nnext = regnext(n); | |
263 | ||
264 | if (oldl + *OPERAND(n) > U8_MAX) | |
265 | break; | |
266 | NEXT_OFF(scan) += NEXT_OFF(n); | |
267 | *OPERAND(scan) += *OPERAND(n); | |
268 | next = n + (*OPERAND(n) + 2 - 1)/sizeof(regnode) + 2; | |
269 | /* Now we can overwrite *n : */ | |
270 | Move(OPERAND(n) + 1, OPERAND(scan) + oldl + 1, | |
271 | *OPERAND(n) + 1, char); | |
272 | #ifdef DEBUGGING | |
273 | if (stringok) | |
274 | stop = next - 1; | |
275 | #endif | |
276 | n = nnext; | |
277 | } | |
278 | } | |
279 | #ifdef DEBUGGING | |
280 | /* Allow dumping */ | |
281 | n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; | |
282 | while (n <= stop) { | |
ca04da08 GS |
283 | /* Purify reports a benign UMR here sometimes, because we |
284 | * don't initialize the OP() slot of a node when that node | |
285 | * is occupied by just the trailing null of the string in | |
286 | * an EXACT node */ | |
22c35a8c | 287 | if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) { |
c277df42 IZ |
288 | OP(n) = OPTIMIZED; |
289 | NEXT_OFF(n) = 0; | |
290 | } | |
291 | n++; | |
292 | } | |
293 | #endif | |
294 | ||
295 | } | |
296 | if (OP(scan) != CURLYX) { | |
048cfca1 GS |
297 | int max = (reg_off_by_arg[OP(scan)] |
298 | ? I32_MAX | |
299 | /* I32 may be smaller than U16 on CRAYs! */ | |
300 | : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX)); | |
c277df42 IZ |
301 | int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan)); |
302 | int noff; | |
303 | regnode *n = scan; | |
304 | ||
305 | /* Skip NOTHING and LONGJMP. */ | |
306 | while ((n = regnext(n)) | |
22c35a8c | 307 | && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n))) |
c277df42 IZ |
308 | || ((OP(n) == LONGJMP) && (noff = ARG(n)))) |
309 | && off + noff < max) | |
310 | off += noff; | |
311 | if (reg_off_by_arg[OP(scan)]) | |
312 | ARG(scan) = off; | |
313 | else | |
314 | NEXT_OFF(scan) = off; | |
315 | } | |
316 | if (OP(scan) == BRANCH || OP(scan) == BRANCHJ | |
317 | || OP(scan) == IFTHEN || OP(scan) == SUSPEND) { | |
318 | next = regnext(scan); | |
319 | code = OP(scan); | |
320 | ||
321 | if (OP(next) == code || code == IFTHEN || code == SUSPEND) { | |
322 | I32 max1 = 0, min1 = I32_MAX, num = 0; | |
323 | ||
324 | if (flags & SCF_DO_SUBSTR) | |
325 | scan_commit(data); | |
326 | while (OP(scan) == code) { | |
327 | I32 deltanext, minnext; | |
328 | ||
329 | num++; | |
330 | data_fake.flags = 0; | |
331 | next = regnext(scan); | |
332 | scan = NEXTOPER(scan); | |
333 | if (code != BRANCH) | |
334 | scan = NEXTOPER(scan); | |
335 | /* We suppose the run is continuous, last=next...*/ | |
336 | minnext = study_chunk(&scan, &deltanext, next, | |
337 | &data_fake, 0); | |
338 | if (min1 > minnext) | |
339 | min1 = minnext; | |
340 | if (max1 < minnext + deltanext) | |
341 | max1 = minnext + deltanext; | |
342 | if (deltanext == I32_MAX) | |
aca2d497 | 343 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
344 | scan = next; |
345 | if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
346 | pars++; | |
405ff068 | 347 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 IZ |
348 | data->flags |= SF_HAS_EVAL; |
349 | if (code == SUSPEND) | |
350 | break; | |
351 | } | |
352 | if (code == IFTHEN && num < 2) /* Empty ELSE branch */ | |
353 | min1 = 0; | |
354 | if (flags & SCF_DO_SUBSTR) { | |
355 | data->pos_min += min1; | |
356 | data->pos_delta += max1 - min1; | |
357 | if (max1 != min1 || is_inf) | |
358 | data->longest = &(data->longest_float); | |
359 | } | |
360 | min += min1; | |
361 | delta += max1 - min1; | |
a0ed51b3 LW |
362 | } |
363 | else if (code == BRANCHJ) /* single branch is optimized. */ | |
c277df42 IZ |
364 | scan = NEXTOPER(NEXTOPER(scan)); |
365 | else /* single branch is optimized. */ | |
366 | scan = NEXTOPER(scan); | |
367 | continue; | |
a0ed51b3 LW |
368 | } |
369 | else if (OP(scan) == EXACT) { | |
370 | I32 l = *OPERAND(scan); | |
371 | if (UTF) { | |
372 | unsigned char *s = (unsigned char *)(OPERAND(scan)+1); | |
373 | unsigned char *e = s + l; | |
374 | I32 newl = 0; | |
375 | while (s < e) { | |
376 | newl++; | |
377 | s += UTF8SKIP(s); | |
378 | } | |
379 | l = newl; | |
380 | } | |
381 | min += l; | |
c277df42 | 382 | if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */ |
c277df42 IZ |
383 | /* The code below prefers earlier match for fixed |
384 | offset, later match for variable offset. */ | |
385 | if (data->last_end == -1) { /* Update the start info. */ | |
386 | data->last_start_min = data->pos_min; | |
387 | data->last_start_max = is_inf | |
388 | ? I32_MAX : data->pos_min + data->pos_delta; | |
389 | } | |
a0ed51b3 | 390 | sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), *OPERAND(scan)); |
c277df42 IZ |
391 | data->last_end = data->pos_min + l; |
392 | data->pos_min += l; /* As in the first entry. */ | |
393 | data->flags &= ~SF_BEFORE_EOL; | |
394 | } | |
a0ed51b3 | 395 | } |
22c35a8c | 396 | else if (PL_regkind[(U8)OP(scan)] == EXACT) { |
a0ed51b3 | 397 | I32 l = *OPERAND(scan); |
c277df42 IZ |
398 | if (flags & SCF_DO_SUBSTR) |
399 | scan_commit(data); | |
a0ed51b3 LW |
400 | if (UTF) { |
401 | unsigned char *s = (unsigned char *)(OPERAND(scan)+1); | |
402 | unsigned char *e = s + l; | |
403 | I32 newl = 0; | |
404 | while (s < e) { | |
405 | newl++; | |
406 | s += UTF8SKIP(s); | |
407 | } | |
408 | l = newl; | |
409 | } | |
410 | min += l; | |
c277df42 | 411 | if (data && (flags & SCF_DO_SUBSTR)) |
a0ed51b3 LW |
412 | data->pos_min += l; |
413 | } | |
4d61ec05 | 414 | else if (strchr((char*)PL_varies,OP(scan))) { |
c277df42 IZ |
415 | I32 mincount, maxcount, minnext, deltanext, pos_before, fl; |
416 | regnode *oscan = scan; | |
417 | ||
22c35a8c | 418 | switch (PL_regkind[(U8)OP(scan)]) { |
c277df42 IZ |
419 | case WHILEM: |
420 | scan = NEXTOPER(scan); | |
421 | goto finish; | |
422 | case PLUS: | |
423 | if (flags & SCF_DO_SUBSTR) { | |
424 | next = NEXTOPER(scan); | |
425 | if (OP(next) == EXACT) { | |
426 | mincount = 1; | |
427 | maxcount = REG_INFTY; | |
428 | next = regnext(scan); | |
429 | scan = NEXTOPER(scan); | |
430 | goto do_curly; | |
431 | } | |
432 | } | |
433 | if (flags & SCF_DO_SUBSTR) | |
434 | data->pos_min++; | |
435 | min++; | |
436 | /* Fall through. */ | |
437 | case STAR: | |
aca2d497 | 438 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
439 | scan = regnext(scan); |
440 | if (flags & SCF_DO_SUBSTR) { | |
441 | scan_commit(data); | |
442 | data->longest = &(data->longest_float); | |
443 | } | |
444 | goto optimize_curly_tail; | |
445 | case CURLY: | |
446 | mincount = ARG1(scan); | |
447 | maxcount = ARG2(scan); | |
448 | next = regnext(scan); | |
449 | scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
450 | do_curly: | |
451 | if (flags & SCF_DO_SUBSTR) { | |
452 | if (mincount == 0) scan_commit(data); | |
453 | pos_before = data->pos_min; | |
454 | } | |
455 | if (data) { | |
456 | fl = data->flags; | |
457 | data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL); | |
458 | if (is_inf) | |
459 | data->flags |= SF_IS_INF; | |
460 | } | |
461 | /* This will finish on WHILEM, setting scan, or on NULL: */ | |
462 | minnext = study_chunk(&scan, &deltanext, last, data, | |
463 | mincount == 0 | |
464 | ? (flags & ~SCF_DO_SUBSTR) : flags); | |
465 | if (!scan) /* It was not CURLYX, but CURLY. */ | |
466 | scan = next; | |
599cee73 | 467 | if (ckWARN(WARN_UNSAFE) && (minnext + deltanext == 0) |
821b33a5 | 468 | && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) |
17feb5d5 | 469 | && maxcount <= REG_INFTY/3) /* Complement check for big count */ |
c5be433b GS |
470 | Perl_warner(aTHX_ WARN_UNSAFE, |
471 | "Strange *+?{} on zero-length expression"); | |
c277df42 | 472 | min += minnext * mincount; |
aca2d497 IZ |
473 | is_inf_internal |= (maxcount == REG_INFTY |
474 | && (minnext + deltanext) > 0 | |
475 | || deltanext == I32_MAX); | |
476 | is_inf |= is_inf_internal; | |
c277df42 IZ |
477 | delta += (minnext + deltanext) * maxcount - minnext * mincount; |
478 | ||
479 | /* Try powerful optimization CURLYX => CURLYN. */ | |
c277df42 IZ |
480 | if ( OP(oscan) == CURLYX && data |
481 | && data->flags & SF_IN_PAR | |
482 | && !(data->flags & SF_HAS_EVAL) | |
483 | && !deltanext && minnext == 1 ) { | |
484 | /* Try to optimize to CURLYN. */ | |
485 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; | |
486 | regnode *nxt1 = nxt, *nxt2; | |
487 | ||
488 | /* Skip open. */ | |
489 | nxt = regnext(nxt); | |
4d61ec05 | 490 | if (!strchr((char*)PL_simple,OP(nxt)) |
22c35a8c | 491 | && !(PL_regkind[(U8)OP(nxt)] == EXACT |
c277df42 IZ |
492 | && *OPERAND(nxt) == 1)) |
493 | goto nogo; | |
494 | nxt2 = nxt; | |
495 | nxt = regnext(nxt); | |
496 | if (OP(nxt) != CLOSE) | |
497 | goto nogo; | |
498 | /* Now we know that nxt2 is the only contents: */ | |
499 | oscan->flags = ARG(nxt); | |
500 | OP(oscan) = CURLYN; | |
501 | OP(nxt1) = NOTHING; /* was OPEN. */ | |
502 | #ifdef DEBUGGING | |
503 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ | |
504 | NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */ | |
505 | NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */ | |
506 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ | |
507 | OP(nxt + 1) = OPTIMIZED; /* was count. */ | |
508 | NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */ | |
509 | #endif | |
510 | } | |
c277df42 IZ |
511 | nogo: |
512 | ||
513 | /* Try optimization CURLYX => CURLYM. */ | |
514 | if ( OP(oscan) == CURLYX && data | |
c277df42 | 515 | && !(data->flags & SF_HAS_PAR) |
c277df42 IZ |
516 | && !(data->flags & SF_HAS_EVAL) |
517 | && !deltanext ) { | |
518 | /* XXXX How to optimize if data == 0? */ | |
519 | /* Optimize to a simpler form. */ | |
520 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */ | |
521 | regnode *nxt2; | |
522 | ||
523 | OP(oscan) = CURLYM; | |
524 | while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/ | |
525 | && (OP(nxt2) != WHILEM)) | |
526 | nxt = nxt2; | |
527 | OP(nxt2) = SUCCEED; /* Whas WHILEM */ | |
c277df42 IZ |
528 | /* Need to optimize away parenths. */ |
529 | if (data->flags & SF_IN_PAR) { | |
530 | /* Set the parenth number. */ | |
531 | regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/ | |
532 | ||
533 | if (OP(nxt) != CLOSE) | |
534 | FAIL("panic opt close"); | |
535 | oscan->flags = ARG(nxt); | |
536 | OP(nxt1) = OPTIMIZED; /* was OPEN. */ | |
537 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ | |
538 | #ifdef DEBUGGING | |
539 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ | |
540 | OP(nxt + 1) = OPTIMIZED; /* was count. */ | |
541 | NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */ | |
542 | NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */ | |
543 | #endif | |
544 | #if 0 | |
545 | while ( nxt1 && (OP(nxt1) != WHILEM)) { | |
546 | regnode *nnxt = regnext(nxt1); | |
547 | ||
548 | if (nnxt == nxt) { | |
549 | if (reg_off_by_arg[OP(nxt1)]) | |
550 | ARG_SET(nxt1, nxt2 - nxt1); | |
551 | else if (nxt2 - nxt1 < U16_MAX) | |
552 | NEXT_OFF(nxt1) = nxt2 - nxt1; | |
553 | else | |
554 | OP(nxt) = NOTHING; /* Cannot beautify */ | |
555 | } | |
556 | nxt1 = nnxt; | |
557 | } | |
558 | #endif | |
559 | /* Optimize again: */ | |
560 | study_chunk(&nxt1, &deltanext, nxt, NULL, 0); | |
a0ed51b3 LW |
561 | } |
562 | else | |
c277df42 | 563 | oscan->flags = 0; |
c277df42 IZ |
564 | } |
565 | if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) | |
566 | pars++; | |
567 | if (flags & SCF_DO_SUBSTR) { | |
568 | SV *last_str = Nullsv; | |
569 | int counted = mincount != 0; | |
570 | ||
571 | if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */ | |
572 | I32 b = pos_before >= data->last_start_min | |
573 | ? pos_before : data->last_start_min; | |
574 | STRLEN l; | |
575 | char *s = SvPV(data->last_found, l); | |
a0ed51b3 LW |
576 | I32 old = b - data->last_start_min; |
577 | ||
578 | if (UTF) | |
579 | old = utf8_hop((U8*)s, old) - (U8*)s; | |
c277df42 | 580 | |
a0ed51b3 | 581 | l -= old; |
c277df42 | 582 | /* Get the added string: */ |
79cb57f6 | 583 | last_str = newSVpvn(s + old, l); |
c277df42 IZ |
584 | if (deltanext == 0 && pos_before == b) { |
585 | /* What was added is a constant string */ | |
586 | if (mincount > 1) { | |
587 | SvGROW(last_str, (mincount * l) + 1); | |
588 | repeatcpy(SvPVX(last_str) + l, | |
589 | SvPVX(last_str), l, mincount - 1); | |
590 | SvCUR(last_str) *= mincount; | |
591 | /* Add additional parts. */ | |
592 | SvCUR_set(data->last_found, | |
593 | SvCUR(data->last_found) - l); | |
594 | sv_catsv(data->last_found, last_str); | |
595 | data->last_end += l * (mincount - 1); | |
596 | } | |
597 | } | |
598 | } | |
599 | /* It is counted once already... */ | |
600 | data->pos_min += minnext * (mincount - counted); | |
601 | data->pos_delta += - counted * deltanext + | |
602 | (minnext + deltanext) * maxcount - minnext * mincount; | |
603 | if (mincount != maxcount) { | |
604 | scan_commit(data); | |
605 | if (mincount && last_str) { | |
606 | sv_setsv(data->last_found, last_str); | |
607 | data->last_end = data->pos_min; | |
608 | data->last_start_min = | |
a0ed51b3 | 609 | data->pos_min - CHR_SVLEN(last_str); |
c277df42 IZ |
610 | data->last_start_max = is_inf |
611 | ? I32_MAX | |
612 | : data->pos_min + data->pos_delta | |
a0ed51b3 | 613 | - CHR_SVLEN(last_str); |
c277df42 IZ |
614 | } |
615 | data->longest = &(data->longest_float); | |
616 | } | |
aca2d497 | 617 | SvREFCNT_dec(last_str); |
c277df42 | 618 | } |
405ff068 | 619 | if (data && (fl & SF_HAS_EVAL)) |
c277df42 IZ |
620 | data->flags |= SF_HAS_EVAL; |
621 | optimize_curly_tail: | |
c277df42 | 622 | if (OP(oscan) != CURLYX) { |
22c35a8c | 623 | while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING |
c277df42 IZ |
624 | && NEXT_OFF(next)) |
625 | NEXT_OFF(oscan) += NEXT_OFF(next); | |
626 | } | |
c277df42 IZ |
627 | continue; |
628 | default: /* REF only? */ | |
629 | if (flags & SCF_DO_SUBSTR) { | |
630 | scan_commit(data); | |
631 | data->longest = &(data->longest_float); | |
632 | } | |
aca2d497 | 633 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
634 | break; |
635 | } | |
a0ed51b3 | 636 | } |
4d61ec05 | 637 | else if (strchr((char*)PL_simple,OP(scan)) || PL_regkind[(U8)OP(scan)] == ANYUTF8) { |
c277df42 IZ |
638 | if (flags & SCF_DO_SUBSTR) { |
639 | scan_commit(data); | |
640 | data->pos_min++; | |
641 | } | |
642 | min++; | |
a0ed51b3 | 643 | } |
22c35a8c | 644 | else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) { |
c277df42 IZ |
645 | data->flags |= (OP(scan) == MEOL |
646 | ? SF_BEFORE_MEOL | |
647 | : SF_BEFORE_SEOL); | |
a0ed51b3 | 648 | } |
22c35a8c | 649 | else if (PL_regkind[(U8)OP(scan)] == BRANCHJ |
c277df42 IZ |
650 | && (scan->flags || data) |
651 | && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) { | |
652 | I32 deltanext, minnext; | |
653 | regnode *nscan; | |
654 | ||
655 | data_fake.flags = 0; | |
656 | next = regnext(scan); | |
657 | nscan = NEXTOPER(NEXTOPER(scan)); | |
658 | minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0); | |
659 | if (scan->flags) { | |
660 | if (deltanext) { | |
661 | FAIL("variable length lookbehind not implemented"); | |
a0ed51b3 LW |
662 | } |
663 | else if (minnext > U8_MAX) { | |
c277df42 IZ |
664 | FAIL2("lookbehind longer than %d not implemented", U8_MAX); |
665 | } | |
666 | scan->flags = minnext; | |
667 | } | |
668 | if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
669 | pars++; | |
405ff068 | 670 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 | 671 | data->flags |= SF_HAS_EVAL; |
a0ed51b3 LW |
672 | } |
673 | else if (OP(scan) == OPEN) { | |
c277df42 | 674 | pars++; |
a0ed51b3 LW |
675 | } |
676 | else if (OP(scan) == CLOSE && ARG(scan) == is_par) { | |
c277df42 IZ |
677 | next = regnext(scan); |
678 | ||
679 | if ( next && (OP(next) != WHILEM) && next < last) | |
c277df42 | 680 | is_par = 0; /* Disable optimization */ |
a0ed51b3 LW |
681 | } |
682 | else if (OP(scan) == EVAL) { | |
c277df42 IZ |
683 | if (data) |
684 | data->flags |= SF_HAS_EVAL; | |
685 | } | |
0f5d15d6 IZ |
686 | else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */ |
687 | if (flags & SCF_DO_SUBSTR) { | |
688 | scan_commit(data); | |
689 | data->longest = &(data->longest_float); | |
690 | } | |
691 | is_inf = is_inf_internal = 1; | |
692 | } | |
c277df42 IZ |
693 | /* Else: zero-length, ignore. */ |
694 | scan = regnext(scan); | |
695 | } | |
696 | ||
697 | finish: | |
698 | *scanp = scan; | |
aca2d497 | 699 | *deltap = is_inf_internal ? I32_MAX : delta; |
c277df42 IZ |
700 | if (flags & SCF_DO_SUBSTR && is_inf) |
701 | data->pos_delta = I32_MAX - data->pos_min; | |
702 | if (is_par > U8_MAX) | |
703 | is_par = 0; | |
704 | if (is_par && pars==1 && data) { | |
705 | data->flags |= SF_IN_PAR; | |
706 | data->flags &= ~SF_HAS_PAR; | |
a0ed51b3 LW |
707 | } |
708 | else if (pars && data) { | |
c277df42 IZ |
709 | data->flags |= SF_HAS_PAR; |
710 | data->flags &= ~SF_IN_PAR; | |
711 | } | |
712 | return min; | |
713 | } | |
714 | ||
76e3520e | 715 | STATIC I32 |
cea2e8a9 | 716 | S_add_data(pTHX_ I32 n, char *s) |
c277df42 | 717 | { |
5c0ca799 | 718 | dTHR; |
3280af22 NIS |
719 | if (PL_regcomp_rx->data) { |
720 | Renewc(PL_regcomp_rx->data, | |
721 | sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1), | |
c277df42 | 722 | char, struct reg_data); |
3280af22 NIS |
723 | Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8); |
724 | PL_regcomp_rx->data->count += n; | |
a0ed51b3 LW |
725 | } |
726 | else { | |
3280af22 | 727 | Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1), |
c277df42 | 728 | char, struct reg_data); |
3280af22 NIS |
729 | New(1208, PL_regcomp_rx->data->what, n, U8); |
730 | PL_regcomp_rx->data->count = n; | |
c277df42 | 731 | } |
3280af22 NIS |
732 | Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8); |
733 | return PL_regcomp_rx->data->count - n; | |
c277df42 IZ |
734 | } |
735 | ||
d88dccdf | 736 | void |
864dbfa3 | 737 | Perl_reginitcolors(pTHX) |
d88dccdf IZ |
738 | { |
739 | dTHR; | |
740 | int i = 0; | |
741 | char *s = PerlEnv_getenv("PERL_RE_COLORS"); | |
742 | ||
743 | if (s) { | |
744 | PL_colors[0] = s = savepv(s); | |
745 | while (++i < 6) { | |
746 | s = strchr(s, '\t'); | |
747 | if (s) { | |
748 | *s = '\0'; | |
749 | PL_colors[i] = ++s; | |
750 | } | |
751 | else | |
c712d376 | 752 | PL_colors[i] = s = ""; |
d88dccdf IZ |
753 | } |
754 | } else { | |
755 | while (i < 6) | |
756 | PL_colors[i++] = ""; | |
757 | } | |
758 | PL_colorset = 1; | |
759 | } | |
760 | ||
a687059c | 761 | /* |
e50aee73 | 762 | - pregcomp - compile a regular expression into internal code |
a687059c LW |
763 | * |
764 | * We can't allocate space until we know how big the compiled form will be, | |
765 | * but we can't compile it (and thus know how big it is) until we've got a | |
766 | * place to put the code. So we cheat: we compile it twice, once with code | |
767 | * generation turned off and size counting turned on, and once "for real". | |
768 | * This also means that we don't allocate space until we are sure that the | |
769 | * thing really will compile successfully, and we never have to move the | |
770 | * code and thus invalidate pointers into it. (Note that it has to be in | |
771 | * one piece because free() must be able to free it all.) [NB: not true in perl] | |
772 | * | |
773 | * Beware that the optimization-preparation code in here knows about some | |
774 | * of the structure of the compiled regexp. [I'll say.] | |
775 | */ | |
776 | regexp * | |
864dbfa3 | 777 | Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm) |
a687059c | 778 | { |
5c0ca799 | 779 | dTHR; |
a0d0e21e | 780 | register regexp *r; |
c277df42 IZ |
781 | regnode *scan; |
782 | SV **longest; | |
783 | SV *longest_fixed; | |
784 | SV *longest_float; | |
785 | regnode *first; | |
a0d0e21e | 786 | I32 flags; |
a0d0e21e LW |
787 | I32 minlen = 0; |
788 | I32 sawplus = 0; | |
789 | I32 sawopen = 0; | |
790 | ||
791 | if (exp == NULL) | |
c277df42 | 792 | FAIL("NULL regexp argument"); |
a0d0e21e | 793 | |
e24b16f9 | 794 | if (PL_curcop == &PL_compiling ? (PL_hints & HINT_UTF8) : IN_UTF8) |
a0ed51b3 LW |
795 | PL_reg_flags |= RF_utf8; |
796 | else | |
797 | PL_reg_flags = 0; | |
798 | ||
3280af22 | 799 | PL_regprecomp = savepvn(exp, xend - exp); |
35ef4773 GS |
800 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
801 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling%s RE `%s%*s%s'\n", | |
d88dccdf IZ |
802 | PL_colors[4],PL_colors[5],PL_colors[0], |
803 | xend - exp, PL_regprecomp, PL_colors[1])); | |
3280af22 NIS |
804 | PL_regflags = pm->op_pmflags; |
805 | PL_regsawback = 0; | |
bbce6d69 | 806 | |
3280af22 NIS |
807 | PL_regseen = 0; |
808 | PL_seen_zerolen = *exp == '^' ? -1 : 0; | |
809 | PL_seen_evals = 0; | |
810 | PL_extralen = 0; | |
c277df42 | 811 | |
bbce6d69 | 812 | /* First pass: determine size, legality. */ |
3280af22 NIS |
813 | PL_regcomp_parse = exp; |
814 | PL_regxend = xend; | |
815 | PL_regnaughty = 0; | |
816 | PL_regnpar = 1; | |
817 | PL_regsize = 0L; | |
818 | PL_regcode = &PL_regdummy; | |
22c35a8c | 819 | regc((U8)REG_MAGIC, (char*)PL_regcode); |
a0d0e21e | 820 | if (reg(0, &flags) == NULL) { |
3280af22 NIS |
821 | Safefree(PL_regprecomp); |
822 | PL_regprecomp = Nullch; | |
a0d0e21e LW |
823 | return(NULL); |
824 | } | |
3280af22 | 825 | DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", PL_regsize)); |
c277df42 | 826 | |
c277df42 IZ |
827 | /* Small enough for pointer-storage convention? |
828 | If extralen==0, this means that we will not need long jumps. */ | |
3280af22 NIS |
829 | if (PL_regsize >= 0x10000L && PL_extralen) |
830 | PL_regsize += PL_extralen; | |
c277df42 | 831 | else |
3280af22 | 832 | PL_extralen = 0; |
a0d0e21e | 833 | |
bbce6d69 | 834 | /* Allocate space and initialize. */ |
3280af22 | 835 | Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), |
c277df42 | 836 | char, regexp); |
a0d0e21e LW |
837 | if (r == NULL) |
838 | FAIL("regexp out of space"); | |
c277df42 | 839 | r->refcnt = 1; |
bbce6d69 | 840 | r->prelen = xend - exp; |
3280af22 | 841 | r->precomp = PL_regprecomp; |
cf93c79d IZ |
842 | r->subbeg = NULL; |
843 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; | |
4327152a IZ |
844 | r->nparens = PL_regnpar - 1; /* set early to validate backrefs */ |
845 | ||
846 | r->substrs = 0; /* Useful during FAIL. */ | |
847 | r->startp = 0; /* Useful during FAIL. */ | |
848 | r->endp = 0; /* Useful during FAIL. */ | |
849 | ||
3280af22 | 850 | PL_regcomp_rx = r; |
bbce6d69 | 851 | |
852 | /* Second pass: emit code. */ | |
3280af22 NIS |
853 | PL_regcomp_parse = exp; |
854 | PL_regxend = xend; | |
855 | PL_regnaughty = 0; | |
856 | PL_regnpar = 1; | |
857 | PL_regcode = r->program; | |
2cd61cdb | 858 | /* Store the count of eval-groups for security checks: */ |
3280af22 | 859 | PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals); |
22c35a8c | 860 | regc((U8)REG_MAGIC, (char*) PL_regcode++); |
c277df42 | 861 | r->data = 0; |
a0d0e21e LW |
862 | if (reg(0, &flags) == NULL) |
863 | return(NULL); | |
864 | ||
865 | /* Dig out information for optimizations. */ | |
cf93c79d | 866 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */ |
3280af22 | 867 | pm->op_pmflags = PL_regflags; |
a0ed51b3 LW |
868 | if (UTF) |
869 | r->reganch |= ROPT_UTF8; | |
c277df42 | 870 | r->regstclass = NULL; |
a0ed51b3 LW |
871 | if (PL_regnaughty >= 10) /* Probably an expensive pattern. */ |
872 | r->reganch |= ROPT_NAUGHTY; | |
c277df42 | 873 | scan = r->program + 1; /* First BRANCH. */ |
2779dcf1 IZ |
874 | |
875 | /* XXXX To minimize changes to RE engine we always allocate | |
876 | 3-units-long substrs field. */ | |
877 | Newz(1004, r->substrs, 1, struct reg_substr_data); | |
878 | ||
c277df42 IZ |
879 | if (OP(scan) != BRANCH) { /* Only one top-level choice. */ |
880 | scan_data_t data; | |
881 | I32 fake; | |
c5254dd6 | 882 | STRLEN longest_float_length, longest_fixed_length; |
a0d0e21e | 883 | |
c277df42 | 884 | StructCopy(&zero_scan_data, &data, scan_data_t); |
a0d0e21e | 885 | first = scan; |
c277df42 | 886 | /* Skip introductions and multiplicators >= 1. */ |
a0d0e21e LW |
887 | while ((OP(first) == OPEN && (sawopen = 1)) || |
888 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || | |
889 | (OP(first) == PLUS) || | |
890 | (OP(first) == MINMOD) || | |
22c35a8c | 891 | (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
a0d0e21e LW |
892 | if (OP(first) == PLUS) |
893 | sawplus = 1; | |
894 | else | |
895 | first += regarglen[(U8)OP(first)]; | |
896 | first = NEXTOPER(first); | |
a687059c LW |
897 | } |
898 | ||
a0d0e21e LW |
899 | /* Starting-point info. */ |
900 | again: | |
c277df42 | 901 | if (OP(first) == EXACT); /* Empty, get anchored substr later. */ |
4d61ec05 | 902 | else if (strchr((char*)PL_simple+4,OP(first))) |
a0d0e21e | 903 | r->regstclass = first; |
22c35a8c GS |
904 | else if (PL_regkind[(U8)OP(first)] == BOUND || |
905 | PL_regkind[(U8)OP(first)] == NBOUND) | |
a0d0e21e | 906 | r->regstclass = first; |
22c35a8c | 907 | else if (PL_regkind[(U8)OP(first)] == BOL) { |
cad2e5aa JH |
908 | r->reganch |= (OP(first) == MBOL |
909 | ? ROPT_ANCH_MBOL | |
910 | : (OP(first) == SBOL | |
911 | ? ROPT_ANCH_SBOL | |
912 | : ROPT_ANCH_BOL)); | |
a0d0e21e | 913 | first = NEXTOPER(first); |
774d564b | 914 | goto again; |
915 | } | |
916 | else if (OP(first) == GPOS) { | |
917 | r->reganch |= ROPT_ANCH_GPOS; | |
918 | first = NEXTOPER(first); | |
919 | goto again; | |
a0d0e21e LW |
920 | } |
921 | else if ((OP(first) == STAR && | |
22c35a8c | 922 | PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) && |
a0d0e21e LW |
923 | !(r->reganch & ROPT_ANCH) ) |
924 | { | |
925 | /* turn .* into ^.* with an implied $*=1 */ | |
cad2e5aa JH |
926 | int type = OP(NEXTOPER(first)); |
927 | ||
928 | if (type == REG_ANY || type == ANYUTF8) | |
929 | type = ROPT_ANCH_MBOL; | |
930 | else | |
931 | type = ROPT_ANCH_SBOL; | |
932 | ||
933 | r->reganch |= type | ROPT_IMPLICIT; | |
a0d0e21e | 934 | first = NEXTOPER(first); |
774d564b | 935 | goto again; |
a0d0e21e | 936 | } |
cad2e5aa JH |
937 | if (sawplus && (!sawopen || !PL_regsawback) |
938 | && !(PL_regseen & REG_SEEN_EVAL)) /* May examine pos and $& */ | |
939 | /* x+ must match at the 1st pos of run of x's */ | |
940 | r->reganch |= ROPT_SKIP; | |
a0d0e21e | 941 | |
c277df42 IZ |
942 | /* Scan is after the zeroth branch, first is atomic matcher. */ |
943 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n", | |
944 | first - scan + 1)); | |
a0d0e21e LW |
945 | /* |
946 | * If there's something expensive in the r.e., find the | |
947 | * longest literal string that must appear and make it the | |
948 | * regmust. Resolve ties in favor of later strings, since | |
949 | * the regstart check works with the beginning of the r.e. | |
950 | * and avoiding duplication strengthens checking. Not a | |
951 | * strong reason, but sufficient in the absence of others. | |
952 | * [Now we resolve ties in favor of the earlier string if | |
c277df42 | 953 | * it happens that c_offset_min has been invalidated, since the |
a0d0e21e LW |
954 | * earlier string may buy us something the later one won't.] |
955 | */ | |
a0d0e21e | 956 | minlen = 0; |
a687059c | 957 | |
79cb57f6 GS |
958 | data.longest_fixed = newSVpvn("",0); |
959 | data.longest_float = newSVpvn("",0); | |
960 | data.last_found = newSVpvn("",0); | |
c277df42 IZ |
961 | data.longest = &(data.longest_fixed); |
962 | first = scan; | |
963 | ||
3280af22 | 964 | minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */ |
c277df42 | 965 | &data, SCF_DO_SUBSTR); |
3280af22 | 966 | if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed) |
c277df42 | 967 | && data.last_start_min == 0 && data.last_end > 0 |
3280af22 NIS |
968 | && !PL_seen_zerolen |
969 | && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS))) | |
c277df42 IZ |
970 | r->reganch |= ROPT_CHECK_ALL; |
971 | scan_commit(&data); | |
972 | SvREFCNT_dec(data.last_found); | |
973 | ||
a0ed51b3 | 974 | longest_float_length = CHR_SVLEN(data.longest_float); |
c5254dd6 | 975 | if (longest_float_length |
c277df42 IZ |
976 | || (data.flags & SF_FL_BEFORE_EOL |
977 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
3280af22 | 978 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
979 | int t; |
980 | ||
a0ed51b3 | 981 | if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */ |
aca2d497 IZ |
982 | && data.offset_fixed == data.offset_float_min |
983 | && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)) | |
984 | goto remove_float; /* As in (a)+. */ | |
985 | ||
c277df42 IZ |
986 | r->float_substr = data.longest_float; |
987 | r->float_min_offset = data.offset_float_min; | |
988 | r->float_max_offset = data.offset_float_max; | |
cf93c79d IZ |
989 | t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */ |
990 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
991 | || (PL_regflags & PMf_MULTILINE))); | |
992 | fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
993 | } |
994 | else { | |
aca2d497 | 995 | remove_float: |
c277df42 IZ |
996 | r->float_substr = Nullsv; |
997 | SvREFCNT_dec(data.longest_float); | |
c5254dd6 | 998 | longest_float_length = 0; |
a0d0e21e | 999 | } |
c277df42 | 1000 | |
a0ed51b3 | 1001 | longest_fixed_length = CHR_SVLEN(data.longest_fixed); |
c5254dd6 | 1002 | if (longest_fixed_length |
c277df42 IZ |
1003 | || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
1004 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
3280af22 | 1005 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1006 | int t; |
1007 | ||
c277df42 IZ |
1008 | r->anchored_substr = data.longest_fixed; |
1009 | r->anchored_offset = data.offset_fixed; | |
cf93c79d IZ |
1010 | t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1011 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
1012 | || (PL_regflags & PMf_MULTILINE))); | |
1013 | fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1014 | } |
1015 | else { | |
c277df42 IZ |
1016 | r->anchored_substr = Nullsv; |
1017 | SvREFCNT_dec(data.longest_fixed); | |
c5254dd6 | 1018 | longest_fixed_length = 0; |
a0d0e21e | 1019 | } |
c277df42 IZ |
1020 | |
1021 | /* A temporary algorithm prefers floated substr to fixed one to dig more info. */ | |
c5254dd6 | 1022 | if (longest_fixed_length > longest_float_length) { |
c277df42 IZ |
1023 | r->check_substr = r->anchored_substr; |
1024 | r->check_offset_min = r->check_offset_max = r->anchored_offset; | |
1025 | if (r->reganch & ROPT_ANCH_SINGLE) | |
1026 | r->reganch |= ROPT_NOSCAN; | |
a0ed51b3 LW |
1027 | } |
1028 | else { | |
c277df42 IZ |
1029 | r->check_substr = r->float_substr; |
1030 | r->check_offset_min = data.offset_float_min; | |
1031 | r->check_offset_max = data.offset_float_max; | |
a0d0e21e | 1032 | } |
cad2e5aa JH |
1033 | if (r->check_substr) { |
1034 | r->reganch |= RE_USE_INTUIT; | |
1035 | if (SvTAIL(r->check_substr)) | |
1036 | r->reganch |= RE_INTUIT_TAIL; | |
1037 | } | |
a0ed51b3 LW |
1038 | } |
1039 | else { | |
c277df42 IZ |
1040 | /* Several toplevels. Best we can is to set minlen. */ |
1041 | I32 fake; | |
1042 | ||
1043 | DEBUG_r(PerlIO_printf(Perl_debug_log, "\n")); | |
1044 | scan = r->program + 1; | |
3280af22 | 1045 | minlen = study_chunk(&scan, &fake, scan + PL_regsize, NULL, 0); |
c277df42 | 1046 | r->check_substr = r->anchored_substr = r->float_substr = Nullsv; |
a0d0e21e LW |
1047 | } |
1048 | ||
a0d0e21e | 1049 | r->minlen = minlen; |
3280af22 | 1050 | if (PL_regseen & REG_SEEN_GPOS) |
c277df42 | 1051 | r->reganch |= ROPT_GPOS_SEEN; |
3280af22 | 1052 | if (PL_regseen & REG_SEEN_LOOKBEHIND) |
c277df42 | 1053 | r->reganch |= ROPT_LOOKBEHIND_SEEN; |
3280af22 | 1054 | if (PL_regseen & REG_SEEN_EVAL) |
ce862d02 | 1055 | r->reganch |= ROPT_EVAL_SEEN; |
cf93c79d IZ |
1056 | Newz(1002, r->startp, PL_regnpar, I32); |
1057 | Newz(1002, r->endp, PL_regnpar, I32); | |
a0d0e21e LW |
1058 | DEBUG_r(regdump(r)); |
1059 | return(r); | |
a687059c LW |
1060 | } |
1061 | ||
1062 | /* | |
1063 | - reg - regular expression, i.e. main body or parenthesized thing | |
1064 | * | |
1065 | * Caller must absorb opening parenthesis. | |
1066 | * | |
1067 | * Combining parenthesis handling with the base level of regular expression | |
1068 | * is a trifle forced, but the need to tie the tails of the branches to what | |
1069 | * follows makes it hard to avoid. | |
1070 | */ | |
76e3520e | 1071 | STATIC regnode * |
cea2e8a9 | 1072 | S_reg(pTHX_ I32 paren, I32 *flagp) |
c277df42 | 1073 | /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */ |
a687059c | 1074 | { |
5c0ca799 | 1075 | dTHR; |
c277df42 IZ |
1076 | register regnode *ret; /* Will be the head of the group. */ |
1077 | register regnode *br; | |
1078 | register regnode *lastbr; | |
1079 | register regnode *ender = 0; | |
a0d0e21e | 1080 | register I32 parno = 0; |
3280af22 | 1081 | I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0; |
c277df42 | 1082 | char c; |
a0d0e21e | 1083 | |
821b33a5 | 1084 | *flagp = 0; /* Tentatively. */ |
a0d0e21e LW |
1085 | |
1086 | /* Make an OPEN node, if parenthesized. */ | |
1087 | if (paren) { | |
3280af22 | 1088 | if (*PL_regcomp_parse == '?') { |
ca9dfc88 IZ |
1089 | U16 posflags = 0, negflags = 0; |
1090 | U16 *flagsp = &posflags; | |
0f5d15d6 | 1091 | int logical = 0; |
ca9dfc88 | 1092 | |
3280af22 NIS |
1093 | PL_regcomp_parse++; |
1094 | paren = *PL_regcomp_parse++; | |
c277df42 | 1095 | ret = NULL; /* For look-ahead/behind. */ |
a0d0e21e | 1096 | switch (paren) { |
c277df42 | 1097 | case '<': |
3280af22 NIS |
1098 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
1099 | if (*PL_regcomp_parse == '!') | |
c277df42 | 1100 | paren = ','; |
3280af22 | 1101 | if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!') |
c277df42 | 1102 | goto unknown; |
3280af22 | 1103 | PL_regcomp_parse++; |
a0d0e21e LW |
1104 | case '=': |
1105 | case '!': | |
3280af22 | 1106 | PL_seen_zerolen++; |
c277df42 IZ |
1107 | case ':': |
1108 | case '>': | |
a0d0e21e LW |
1109 | break; |
1110 | case '$': | |
1111 | case '@': | |
c277df42 | 1112 | FAIL2("Sequence (?%c...) not implemented", (int)paren); |
a0d0e21e LW |
1113 | break; |
1114 | case '#': | |
3280af22 NIS |
1115 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
1116 | PL_regcomp_parse++; | |
1117 | if (*PL_regcomp_parse != ')') | |
c277df42 | 1118 | FAIL("Sequence (?#... not terminated"); |
a0d0e21e LW |
1119 | nextchar(); |
1120 | *flagp = TRYAGAIN; | |
1121 | return NULL; | |
0f5d15d6 IZ |
1122 | case 'p': |
1123 | logical = 1; | |
1124 | paren = *PL_regcomp_parse++; | |
1125 | /* FALL THROUGH */ | |
c277df42 IZ |
1126 | case '{': |
1127 | { | |
1128 | dTHR; | |
1129 | I32 count = 1, n = 0; | |
1130 | char c; | |
3280af22 | 1131 | char *s = PL_regcomp_parse; |
c277df42 IZ |
1132 | SV *sv; |
1133 | OP_4tree *sop, *rop; | |
1134 | ||
3280af22 NIS |
1135 | PL_seen_zerolen++; |
1136 | PL_regseen |= REG_SEEN_EVAL; | |
1137 | while (count && (c = *PL_regcomp_parse)) { | |
1138 | if (c == '\\' && PL_regcomp_parse[1]) | |
1139 | PL_regcomp_parse++; | |
c277df42 IZ |
1140 | else if (c == '{') |
1141 | count++; | |
1142 | else if (c == '}') | |
1143 | count--; | |
3280af22 | 1144 | PL_regcomp_parse++; |
c277df42 | 1145 | } |
3280af22 | 1146 | if (*PL_regcomp_parse != ')') |
c277df42 IZ |
1147 | FAIL("Sequence (?{...}) not terminated or not {}-balanced"); |
1148 | if (!SIZE_ONLY) { | |
1149 | AV *av; | |
1150 | ||
3280af22 | 1151 | if (PL_regcomp_parse - 1 - s) |
79cb57f6 | 1152 | sv = newSVpvn(s, PL_regcomp_parse - 1 - s); |
c277df42 | 1153 | else |
79cb57f6 | 1154 | sv = newSVpvn("", 0); |
c277df42 IZ |
1155 | |
1156 | rop = sv_compile_2op(sv, &sop, "re", &av); | |
1157 | ||
dfad63ad | 1158 | n = add_data(3, "nop"); |
3280af22 | 1159 | PL_regcomp_rx->data->data[n] = (void*)rop; |
dfad63ad HS |
1160 | PL_regcomp_rx->data->data[n+1] = (void*)sop; |
1161 | PL_regcomp_rx->data->data[n+2] = (void*)av; | |
c277df42 | 1162 | SvREFCNT_dec(sv); |
a0ed51b3 | 1163 | } |
e24b16f9 GS |
1164 | else { /* First pass */ |
1165 | if (PL_reginterp_cnt < ++PL_seen_evals | |
1166 | && PL_curcop != &PL_compiling) | |
2cd61cdb IZ |
1167 | /* No compiled RE interpolated, has runtime |
1168 | components ===> unsafe. */ | |
1169 | FAIL("Eval-group not allowed at runtime, use re 'eval'"); | |
3280af22 | 1170 | if (PL_tainted) |
cc6b7395 | 1171 | FAIL("Eval-group in insecure regular expression"); |
c277df42 IZ |
1172 | } |
1173 | ||
1174 | nextchar(); | |
0f5d15d6 IZ |
1175 | if (logical) { |
1176 | ret = reg_node(LOGICAL); | |
1177 | if (!SIZE_ONLY) | |
1178 | ret->flags = 2; | |
1179 | regtail(ret, reganode(EVAL, n)); | |
1180 | return ret; | |
1181 | } | |
c277df42 IZ |
1182 | return reganode(EVAL, n); |
1183 | } | |
1184 | case '(': | |
1185 | { | |
3280af22 NIS |
1186 | if (PL_regcomp_parse[0] == '?') { |
1187 | if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!' | |
1188 | || PL_regcomp_parse[1] == '<' | |
1189 | || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */ | |
c277df42 IZ |
1190 | I32 flag; |
1191 | ||
1192 | ret = reg_node(LOGICAL); | |
0f5d15d6 IZ |
1193 | if (!SIZE_ONLY) |
1194 | ret->flags = 1; | |
c277df42 IZ |
1195 | regtail(ret, reg(1, &flag)); |
1196 | goto insert_if; | |
1197 | } | |
a0ed51b3 LW |
1198 | } |
1199 | else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) { | |
3280af22 | 1200 | parno = atoi(PL_regcomp_parse++); |
c277df42 | 1201 | |
3280af22 NIS |
1202 | while (isDIGIT(*PL_regcomp_parse)) |
1203 | PL_regcomp_parse++; | |
c277df42 IZ |
1204 | ret = reganode(GROUPP, parno); |
1205 | if ((c = *nextchar()) != ')') | |
1206 | FAIL2("Switch (?(number%c not recognized", c); | |
1207 | insert_if: | |
1208 | regtail(ret, reganode(IFTHEN, 0)); | |
1209 | br = regbranch(&flags, 1); | |
1210 | if (br == NULL) | |
1211 | br = reganode(LONGJMP, 0); | |
1212 | else | |
1213 | regtail(br, reganode(LONGJMP, 0)); | |
1214 | c = *nextchar(); | |
d1b80229 IZ |
1215 | if (flags&HASWIDTH) |
1216 | *flagp |= HASWIDTH; | |
c277df42 IZ |
1217 | if (c == '|') { |
1218 | lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */ | |
1219 | regbranch(&flags, 1); | |
1220 | regtail(ret, lastbr); | |
d1b80229 IZ |
1221 | if (flags&HASWIDTH) |
1222 | *flagp |= HASWIDTH; | |
c277df42 | 1223 | c = *nextchar(); |
a0ed51b3 LW |
1224 | } |
1225 | else | |
c277df42 IZ |
1226 | lastbr = NULL; |
1227 | if (c != ')') | |
1228 | FAIL("Switch (?(condition)... contains too many branches"); | |
1229 | ender = reg_node(TAIL); | |
1230 | regtail(br, ender); | |
1231 | if (lastbr) { | |
1232 | regtail(lastbr, ender); | |
1233 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); | |
a0ed51b3 LW |
1234 | } |
1235 | else | |
c277df42 IZ |
1236 | regtail(ret, ender); |
1237 | return ret; | |
a0ed51b3 LW |
1238 | } |
1239 | else { | |
3280af22 | 1240 | FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse); |
c277df42 IZ |
1241 | } |
1242 | } | |
1b1626e4 | 1243 | case 0: |
c277df42 | 1244 | FAIL("Sequence (? incomplete"); |
1b1626e4 | 1245 | break; |
a0d0e21e | 1246 | default: |
3280af22 | 1247 | --PL_regcomp_parse; |
ca9dfc88 | 1248 | parse_flags: |
3280af22 NIS |
1249 | while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) { |
1250 | if (*PL_regcomp_parse != 'o') | |
1251 | pmflag(flagsp, *PL_regcomp_parse); | |
1252 | ++PL_regcomp_parse; | |
ca9dfc88 | 1253 | } |
3280af22 | 1254 | if (*PL_regcomp_parse == '-') { |
ca9dfc88 | 1255 | flagsp = &negflags; |
3280af22 | 1256 | ++PL_regcomp_parse; |
ca9dfc88 | 1257 | goto parse_flags; |
48c036b1 | 1258 | } |
3280af22 NIS |
1259 | PL_regflags |= posflags; |
1260 | PL_regflags &= ~negflags; | |
1261 | if (*PL_regcomp_parse == ':') { | |
1262 | PL_regcomp_parse++; | |
ca9dfc88 IZ |
1263 | paren = ':'; |
1264 | break; | |
1265 | } | |
c277df42 | 1266 | unknown: |
3280af22 NIS |
1267 | if (*PL_regcomp_parse != ')') |
1268 | FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse); | |
a0d0e21e LW |
1269 | nextchar(); |
1270 | *flagp = TRYAGAIN; | |
1271 | return NULL; | |
1272 | } | |
1273 | } | |
1274 | else { | |
3280af22 NIS |
1275 | parno = PL_regnpar; |
1276 | PL_regnpar++; | |
a0d0e21e | 1277 | ret = reganode(OPEN, parno); |
c277df42 | 1278 | open = 1; |
a0d0e21e | 1279 | } |
a0ed51b3 LW |
1280 | } |
1281 | else | |
a0d0e21e LW |
1282 | ret = NULL; |
1283 | ||
1284 | /* Pick up the branches, linking them together. */ | |
c277df42 | 1285 | br = regbranch(&flags, 1); |
a0d0e21e LW |
1286 | if (br == NULL) |
1287 | return(NULL); | |
3280af22 NIS |
1288 | if (*PL_regcomp_parse == '|') { |
1289 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 | 1290 | reginsert(BRANCHJ, br); |
a0ed51b3 LW |
1291 | } |
1292 | else | |
c277df42 IZ |
1293 | reginsert(BRANCH, br); |
1294 | have_branch = 1; | |
1295 | if (SIZE_ONLY) | |
3280af22 | 1296 | PL_extralen += 1; /* For BRANCHJ-BRANCH. */ |
a0ed51b3 LW |
1297 | } |
1298 | else if (paren == ':') { | |
c277df42 IZ |
1299 | *flagp |= flags&SIMPLE; |
1300 | } | |
1301 | if (open) { /* Starts with OPEN. */ | |
1302 | regtail(ret, br); /* OPEN -> first. */ | |
a0ed51b3 LW |
1303 | } |
1304 | else if (paren != '?') /* Not Conditional */ | |
a0d0e21e | 1305 | ret = br; |
821b33a5 IZ |
1306 | if (flags&HASWIDTH) |
1307 | *flagp |= HASWIDTH; | |
a0d0e21e | 1308 | *flagp |= flags&SPSTART; |
c277df42 | 1309 | lastbr = br; |
3280af22 NIS |
1310 | while (*PL_regcomp_parse == '|') { |
1311 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 IZ |
1312 | ender = reganode(LONGJMP,0); |
1313 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ | |
1314 | } | |
1315 | if (SIZE_ONLY) | |
3280af22 | 1316 | PL_extralen += 2; /* Account for LONGJMP. */ |
a0d0e21e | 1317 | nextchar(); |
c277df42 | 1318 | br = regbranch(&flags, 0); |
a687059c | 1319 | if (br == NULL) |
a0d0e21e | 1320 | return(NULL); |
c277df42 IZ |
1321 | regtail(lastbr, br); /* BRANCH -> BRANCH. */ |
1322 | lastbr = br; | |
821b33a5 IZ |
1323 | if (flags&HASWIDTH) |
1324 | *flagp |= HASWIDTH; | |
a687059c | 1325 | *flagp |= flags&SPSTART; |
a0d0e21e LW |
1326 | } |
1327 | ||
c277df42 IZ |
1328 | if (have_branch || paren != ':') { |
1329 | /* Make a closing node, and hook it on the end. */ | |
1330 | switch (paren) { | |
1331 | case ':': | |
1332 | ender = reg_node(TAIL); | |
1333 | break; | |
1334 | case 1: | |
1335 | ender = reganode(CLOSE, parno); | |
1336 | break; | |
1337 | case '<': | |
c277df42 IZ |
1338 | case ',': |
1339 | case '=': | |
1340 | case '!': | |
c277df42 | 1341 | *flagp &= ~HASWIDTH; |
821b33a5 IZ |
1342 | /* FALL THROUGH */ |
1343 | case '>': | |
1344 | ender = reg_node(SUCCEED); | |
c277df42 IZ |
1345 | break; |
1346 | case 0: | |
1347 | ender = reg_node(END); | |
1348 | break; | |
1349 | } | |
1350 | regtail(lastbr, ender); | |
a0d0e21e | 1351 | |
c277df42 IZ |
1352 | if (have_branch) { |
1353 | /* Hook the tails of the branches to the closing node. */ | |
1354 | for (br = ret; br != NULL; br = regnext(br)) { | |
1355 | regoptail(br, ender); | |
1356 | } | |
1357 | } | |
a0d0e21e | 1358 | } |
c277df42 IZ |
1359 | |
1360 | { | |
1361 | char *p; | |
1362 | static char parens[] = "=!<,>"; | |
1363 | ||
1364 | if (paren && (p = strchr(parens, paren))) { | |
1365 | int node = ((p - parens) % 2) ? UNLESSM : IFMATCH; | |
1366 | int flag = (p - parens) > 1; | |
1367 | ||
1368 | if (paren == '>') | |
1369 | node = SUSPEND, flag = 0; | |
1370 | reginsert(node,ret); | |
c277df42 | 1371 | ret->flags = flag; |
c277df42 IZ |
1372 | regtail(ret, reg_node(TAIL)); |
1373 | } | |
a0d0e21e LW |
1374 | } |
1375 | ||
1376 | /* Check for proper termination. */ | |
ce3e6498 HS |
1377 | if (paren) { |
1378 | PL_regflags = oregflags; | |
1379 | if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') { | |
1380 | FAIL("unmatched () in regexp"); | |
1381 | } | |
a0ed51b3 LW |
1382 | } |
1383 | else if (!paren && PL_regcomp_parse < PL_regxend) { | |
3280af22 | 1384 | if (*PL_regcomp_parse == ')') { |
a0d0e21e | 1385 | FAIL("unmatched () in regexp"); |
a0ed51b3 LW |
1386 | } |
1387 | else | |
a0d0e21e LW |
1388 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
1389 | /* NOTREACHED */ | |
1390 | } | |
a687059c | 1391 | |
a0d0e21e | 1392 | return(ret); |
a687059c LW |
1393 | } |
1394 | ||
1395 | /* | |
1396 | - regbranch - one alternative of an | operator | |
1397 | * | |
1398 | * Implements the concatenation operator. | |
1399 | */ | |
76e3520e | 1400 | STATIC regnode * |
cea2e8a9 | 1401 | S_regbranch(pTHX_ I32 *flagp, I32 first) |
a687059c | 1402 | { |
5c0ca799 | 1403 | dTHR; |
c277df42 IZ |
1404 | register regnode *ret; |
1405 | register regnode *chain = NULL; | |
1406 | register regnode *latest; | |
1407 | I32 flags = 0, c = 0; | |
a0d0e21e | 1408 | |
c277df42 IZ |
1409 | if (first) |
1410 | ret = NULL; | |
1411 | else { | |
3280af22 | 1412 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
1413 | ret = reganode(BRANCHJ,0); |
1414 | else | |
1415 | ret = reg_node(BRANCH); | |
1416 | } | |
1417 | ||
1418 | if (!first && SIZE_ONLY) | |
3280af22 | 1419 | PL_extralen += 1; /* BRANCHJ */ |
c277df42 IZ |
1420 | |
1421 | *flagp = WORST; /* Tentatively. */ | |
a0d0e21e | 1422 | |
3280af22 | 1423 | PL_regcomp_parse--; |
a0d0e21e | 1424 | nextchar(); |
3280af22 | 1425 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') { |
a0d0e21e LW |
1426 | flags &= ~TRYAGAIN; |
1427 | latest = regpiece(&flags); | |
1428 | if (latest == NULL) { | |
1429 | if (flags & TRYAGAIN) | |
1430 | continue; | |
1431 | return(NULL); | |
a0ed51b3 LW |
1432 | } |
1433 | else if (ret == NULL) | |
c277df42 | 1434 | ret = latest; |
a0d0e21e | 1435 | *flagp |= flags&HASWIDTH; |
c277df42 | 1436 | if (chain == NULL) /* First piece. */ |
a0d0e21e LW |
1437 | *flagp |= flags&SPSTART; |
1438 | else { | |
3280af22 | 1439 | PL_regnaughty++; |
a0d0e21e | 1440 | regtail(chain, latest); |
a687059c | 1441 | } |
a0d0e21e | 1442 | chain = latest; |
c277df42 IZ |
1443 | c++; |
1444 | } | |
1445 | if (chain == NULL) { /* Loop ran zero times. */ | |
1446 | chain = reg_node(NOTHING); | |
1447 | if (ret == NULL) | |
1448 | ret = chain; | |
1449 | } | |
1450 | if (c == 1) { | |
1451 | *flagp |= flags&SIMPLE; | |
a0d0e21e | 1452 | } |
a687059c | 1453 | |
a0d0e21e | 1454 | return(ret); |
a687059c LW |
1455 | } |
1456 | ||
1457 | /* | |
1458 | - regpiece - something followed by possible [*+?] | |
1459 | * | |
1460 | * Note that the branching code sequences used for ? and the general cases | |
1461 | * of * and + are somewhat optimized: they use the same NOTHING node as | |
1462 | * both the endmarker for their branch list and the body of the last branch. | |
1463 | * It might seem that this node could be dispensed with entirely, but the | |
1464 | * endmarker role is not redundant. | |
1465 | */ | |
76e3520e | 1466 | STATIC regnode * |
cea2e8a9 | 1467 | S_regpiece(pTHX_ I32 *flagp) |
a687059c | 1468 | { |
5c0ca799 | 1469 | dTHR; |
c277df42 | 1470 | register regnode *ret; |
a0d0e21e LW |
1471 | register char op; |
1472 | register char *next; | |
1473 | I32 flags; | |
3280af22 | 1474 | char *origparse = PL_regcomp_parse; |
a0d0e21e LW |
1475 | char *maxpos; |
1476 | I32 min; | |
c277df42 | 1477 | I32 max = REG_INFTY; |
a0d0e21e LW |
1478 | |
1479 | ret = regatom(&flags); | |
1480 | if (ret == NULL) { | |
1481 | if (flags & TRYAGAIN) | |
1482 | *flagp |= TRYAGAIN; | |
1483 | return(NULL); | |
1484 | } | |
1485 | ||
3280af22 | 1486 | op = *PL_regcomp_parse; |
a0d0e21e | 1487 | |
3280af22 NIS |
1488 | if (op == '{' && regcurly(PL_regcomp_parse)) { |
1489 | next = PL_regcomp_parse + 1; | |
a0d0e21e LW |
1490 | maxpos = Nullch; |
1491 | while (isDIGIT(*next) || *next == ',') { | |
1492 | if (*next == ',') { | |
1493 | if (maxpos) | |
1494 | break; | |
1495 | else | |
1496 | maxpos = next; | |
a687059c | 1497 | } |
a0d0e21e LW |
1498 | next++; |
1499 | } | |
1500 | if (*next == '}') { /* got one */ | |
1501 | if (!maxpos) | |
1502 | maxpos = next; | |
3280af22 NIS |
1503 | PL_regcomp_parse++; |
1504 | min = atoi(PL_regcomp_parse); | |
a0d0e21e LW |
1505 | if (*maxpos == ',') |
1506 | maxpos++; | |
1507 | else | |
3280af22 | 1508 | maxpos = PL_regcomp_parse; |
a0d0e21e LW |
1509 | max = atoi(maxpos); |
1510 | if (!max && *maxpos != '0') | |
c277df42 IZ |
1511 | max = REG_INFTY; /* meaning "infinity" */ |
1512 | else if (max >= REG_INFTY) | |
1513 | FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); | |
3280af22 | 1514 | PL_regcomp_parse = next; |
a0d0e21e LW |
1515 | nextchar(); |
1516 | ||
1517 | do_curly: | |
1518 | if ((flags&SIMPLE)) { | |
3280af22 | 1519 | PL_regnaughty += 2 + PL_regnaughty / 2; |
a0d0e21e LW |
1520 | reginsert(CURLY, ret); |
1521 | } | |
1522 | else { | |
3280af22 | 1523 | PL_regnaughty += 4 + PL_regnaughty; /* compound interest */ |
c277df42 | 1524 | regtail(ret, reg_node(WHILEM)); |
3280af22 | 1525 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 IZ |
1526 | reginsert(LONGJMP,ret); |
1527 | reginsert(NOTHING,ret); | |
1528 | NEXT_OFF(ret) = 3; /* Go over LONGJMP. */ | |
1529 | } | |
a0d0e21e | 1530 | reginsert(CURLYX,ret); |
3280af22 | 1531 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
1532 | NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */ |
1533 | regtail(ret, reg_node(NOTHING)); | |
1534 | if (SIZE_ONLY) | |
3280af22 | 1535 | PL_extralen += 3; |
a0d0e21e | 1536 | } |
c277df42 | 1537 | ret->flags = 0; |
a0d0e21e LW |
1538 | |
1539 | if (min > 0) | |
821b33a5 IZ |
1540 | *flagp = WORST; |
1541 | if (max > 0) | |
1542 | *flagp |= HASWIDTH; | |
a0d0e21e | 1543 | if (max && max < min) |
c277df42 IZ |
1544 | FAIL("Can't do {n,m} with n > m"); |
1545 | if (!SIZE_ONLY) { | |
1546 | ARG1_SET(ret, min); | |
1547 | ARG2_SET(ret, max); | |
a687059c | 1548 | } |
a687059c | 1549 | |
a0d0e21e | 1550 | goto nest_check; |
a687059c | 1551 | } |
a0d0e21e | 1552 | } |
a687059c | 1553 | |
a0d0e21e LW |
1554 | if (!ISMULT1(op)) { |
1555 | *flagp = flags; | |
a687059c | 1556 | return(ret); |
a0d0e21e | 1557 | } |
bb20fd44 | 1558 | |
c277df42 | 1559 | #if 0 /* Now runtime fix should be reliable. */ |
bb20fd44 | 1560 | if (!(flags&HASWIDTH) && op != '?') |
c277df42 IZ |
1561 | FAIL("regexp *+ operand could be empty"); |
1562 | #endif | |
bb20fd44 | 1563 | |
a0d0e21e LW |
1564 | nextchar(); |
1565 | ||
821b33a5 | 1566 | *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); |
a0d0e21e LW |
1567 | |
1568 | if (op == '*' && (flags&SIMPLE)) { | |
1569 | reginsert(STAR, ret); | |
c277df42 | 1570 | ret->flags = 0; |
3280af22 | 1571 | PL_regnaughty += 4; |
a0d0e21e LW |
1572 | } |
1573 | else if (op == '*') { | |
1574 | min = 0; | |
1575 | goto do_curly; | |
a0ed51b3 LW |
1576 | } |
1577 | else if (op == '+' && (flags&SIMPLE)) { | |
a0d0e21e | 1578 | reginsert(PLUS, ret); |
c277df42 | 1579 | ret->flags = 0; |
3280af22 | 1580 | PL_regnaughty += 3; |
a0d0e21e LW |
1581 | } |
1582 | else if (op == '+') { | |
1583 | min = 1; | |
1584 | goto do_curly; | |
a0ed51b3 LW |
1585 | } |
1586 | else if (op == '?') { | |
a0d0e21e LW |
1587 | min = 0; max = 1; |
1588 | goto do_curly; | |
1589 | } | |
1590 | nest_check: | |
17feb5d5 | 1591 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) { |
cea2e8a9 | 1592 | Perl_warner(aTHX_ WARN_UNSAFE, "%.*s matches null string many times", |
3280af22 | 1593 | PL_regcomp_parse - origparse, origparse); |
a0d0e21e LW |
1594 | } |
1595 | ||
3280af22 | 1596 | if (*PL_regcomp_parse == '?') { |
a0d0e21e LW |
1597 | nextchar(); |
1598 | reginsert(MINMOD, ret); | |
c277df42 | 1599 | regtail(ret, ret + NODE_STEP_REGNODE); |
a0d0e21e | 1600 | } |
3280af22 | 1601 | if (ISMULT2(PL_regcomp_parse)) |
a0d0e21e LW |
1602 | FAIL("nested *?+ in regexp"); |
1603 | ||
1604 | return(ret); | |
a687059c LW |
1605 | } |
1606 | ||
1607 | /* | |
1608 | - regatom - the lowest level | |
1609 | * | |
1610 | * Optimization: gobbles an entire sequence of ordinary characters so that | |
1611 | * it can turn them into a single node, which is smaller to store and | |
1612 | * faster to run. Backslashed characters are exceptions, each becoming a | |
1613 | * separate node; the code is simpler that way and it's not worth fixing. | |
1614 | * | |
1615 | * [Yes, it is worth fixing, some scripts can run twice the speed.] | |
1616 | */ | |
76e3520e | 1617 | STATIC regnode * |
cea2e8a9 | 1618 | S_regatom(pTHX_ I32 *flagp) |
a687059c | 1619 | { |
5c0ca799 | 1620 | dTHR; |
c277df42 | 1621 | register regnode *ret = 0; |
a0d0e21e LW |
1622 | I32 flags; |
1623 | ||
1624 | *flagp = WORST; /* Tentatively. */ | |
1625 | ||
1626 | tryagain: | |
3280af22 | 1627 | switch (*PL_regcomp_parse) { |
a0d0e21e | 1628 | case '^': |
3280af22 | 1629 | PL_seen_zerolen++; |
a0d0e21e | 1630 | nextchar(); |
3280af22 | 1631 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 1632 | ret = reg_node(MBOL); |
3280af22 | 1633 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 1634 | ret = reg_node(SBOL); |
a0d0e21e | 1635 | else |
c277df42 | 1636 | ret = reg_node(BOL); |
a0d0e21e LW |
1637 | break; |
1638 | case '$': | |
3280af22 NIS |
1639 | if (PL_regcomp_parse[1]) |
1640 | PL_seen_zerolen++; | |
a0d0e21e | 1641 | nextchar(); |
3280af22 | 1642 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 1643 | ret = reg_node(MEOL); |
3280af22 | 1644 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 1645 | ret = reg_node(SEOL); |
a0d0e21e | 1646 | else |
c277df42 | 1647 | ret = reg_node(EOL); |
a0d0e21e LW |
1648 | break; |
1649 | case '.': | |
1650 | nextchar(); | |
a0ed51b3 LW |
1651 | if (UTF) { |
1652 | if (PL_regflags & PMf_SINGLELINE) | |
1653 | ret = reg_node(SANYUTF8); | |
1654 | else | |
1655 | ret = reg_node(ANYUTF8); | |
1656 | *flagp |= HASWIDTH; | |
1657 | } | |
1658 | else { | |
1659 | if (PL_regflags & PMf_SINGLELINE) | |
1660 | ret = reg_node(SANY); | |
1661 | else | |
22c35a8c | 1662 | ret = reg_node(REG_ANY); |
a0ed51b3 LW |
1663 | *flagp |= HASWIDTH|SIMPLE; |
1664 | } | |
3280af22 | 1665 | PL_regnaughty++; |
a0d0e21e LW |
1666 | break; |
1667 | case '[': | |
3280af22 | 1668 | PL_regcomp_parse++; |
a0ed51b3 | 1669 | ret = (UTF ? regclassutf8() : regclass()); |
a14b48bc LW |
1670 | if (*PL_regcomp_parse != ']') |
1671 | FAIL("unmatched [] in regexp"); | |
1672 | nextchar(); | |
a0d0e21e LW |
1673 | *flagp |= HASWIDTH|SIMPLE; |
1674 | break; | |
1675 | case '(': | |
1676 | nextchar(); | |
1677 | ret = reg(1, &flags); | |
1678 | if (ret == NULL) { | |
1679 | if (flags & TRYAGAIN) | |
1680 | goto tryagain; | |
1681 | return(NULL); | |
1682 | } | |
c277df42 | 1683 | *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE); |
a0d0e21e LW |
1684 | break; |
1685 | case '|': | |
1686 | case ')': | |
1687 | if (flags & TRYAGAIN) { | |
1688 | *flagp |= TRYAGAIN; | |
1689 | return NULL; | |
1690 | } | |
3280af22 | 1691 | FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse); |
a0d0e21e LW |
1692 | /* Supposed to be caught earlier. */ |
1693 | break; | |
85afd4ae | 1694 | case '{': |
3280af22 NIS |
1695 | if (!regcurly(PL_regcomp_parse)) { |
1696 | PL_regcomp_parse++; | |
85afd4ae CS |
1697 | goto defchar; |
1698 | } | |
1699 | /* FALL THROUGH */ | |
a0d0e21e LW |
1700 | case '?': |
1701 | case '+': | |
1702 | case '*': | |
3115e423 | 1703 | FAIL("?+*{} follows nothing in regexp"); |
a0d0e21e LW |
1704 | break; |
1705 | case '\\': | |
3280af22 | 1706 | switch (*++PL_regcomp_parse) { |
a0d0e21e | 1707 | case 'A': |
3280af22 | 1708 | PL_seen_zerolen++; |
c277df42 | 1709 | ret = reg_node(SBOL); |
a0d0e21e LW |
1710 | *flagp |= SIMPLE; |
1711 | nextchar(); | |
1712 | break; | |
1713 | case 'G': | |
c277df42 | 1714 | ret = reg_node(GPOS); |
3280af22 | 1715 | PL_regseen |= REG_SEEN_GPOS; |
a0d0e21e LW |
1716 | *flagp |= SIMPLE; |
1717 | nextchar(); | |
1718 | break; | |
1719 | case 'Z': | |
c277df42 | 1720 | ret = reg_node(SEOL); |
a0d0e21e LW |
1721 | *flagp |= SIMPLE; |
1722 | nextchar(); | |
1723 | break; | |
b85d18e9 IZ |
1724 | case 'z': |
1725 | ret = reg_node(EOS); | |
1726 | *flagp |= SIMPLE; | |
3280af22 | 1727 | PL_seen_zerolen++; /* Do not optimize RE away */ |
b85d18e9 IZ |
1728 | nextchar(); |
1729 | break; | |
a0ed51b3 LW |
1730 | case 'C': |
1731 | ret = reg_node(SANY); | |
1732 | *flagp |= HASWIDTH|SIMPLE; | |
1733 | nextchar(); | |
1734 | break; | |
1735 | case 'X': | |
1736 | ret = reg_node(CLUMP); | |
1737 | *flagp |= HASWIDTH; | |
1738 | nextchar(); | |
1739 | if (UTF && !PL_utf8_mark) | |
dfe13c55 | 1740 | is_utf8_mark((U8*)"~"); /* preload table */ |
a0ed51b3 | 1741 | break; |
a0d0e21e | 1742 | case 'w': |
a0ed51b3 LW |
1743 | ret = reg_node( |
1744 | UTF | |
1745 | ? (LOC ? ALNUMLUTF8 : ALNUMUTF8) | |
1746 | : (LOC ? ALNUML : ALNUM)); | |
a0d0e21e LW |
1747 | *flagp |= HASWIDTH|SIMPLE; |
1748 | nextchar(); | |
a0ed51b3 | 1749 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1750 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1751 | break; |
1752 | case 'W': | |
a0ed51b3 LW |
1753 | ret = reg_node( |
1754 | UTF | |
1755 | ? (LOC ? NALNUMLUTF8 : NALNUMUTF8) | |
1756 | : (LOC ? NALNUML : NALNUM)); | |
a0d0e21e LW |
1757 | *flagp |= HASWIDTH|SIMPLE; |
1758 | nextchar(); | |
a0ed51b3 | 1759 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1760 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1761 | break; |
1762 | case 'b': | |
3280af22 | 1763 | PL_seen_zerolen++; |
f5c9036e | 1764 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
1765 | ret = reg_node( |
1766 | UTF | |
1767 | ? (LOC ? BOUNDLUTF8 : BOUNDUTF8) | |
1768 | : (LOC ? BOUNDL : BOUND)); | |
a0d0e21e LW |
1769 | *flagp |= SIMPLE; |
1770 | nextchar(); | |
a0ed51b3 | 1771 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1772 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1773 | break; |
1774 | case 'B': | |
3280af22 | 1775 | PL_seen_zerolen++; |
f5c9036e | 1776 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
1777 | ret = reg_node( |
1778 | UTF | |
1779 | ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8) | |
1780 | : (LOC ? NBOUNDL : NBOUND)); | |
a0d0e21e LW |
1781 | *flagp |= SIMPLE; |
1782 | nextchar(); | |
a0ed51b3 | 1783 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1784 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1785 | break; |
1786 | case 's': | |
a0ed51b3 LW |
1787 | ret = reg_node( |
1788 | UTF | |
1789 | ? (LOC ? SPACELUTF8 : SPACEUTF8) | |
1790 | : (LOC ? SPACEL : SPACE)); | |
a0d0e21e LW |
1791 | *flagp |= HASWIDTH|SIMPLE; |
1792 | nextchar(); | |
a0ed51b3 | 1793 | if (UTF && !PL_utf8_space) |
dfe13c55 | 1794 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
1795 | break; |
1796 | case 'S': | |
a0ed51b3 LW |
1797 | ret = reg_node( |
1798 | UTF | |
1799 | ? (LOC ? NSPACELUTF8 : NSPACEUTF8) | |
1800 | : (LOC ? NSPACEL : NSPACE)); | |
a0d0e21e LW |
1801 | *flagp |= HASWIDTH|SIMPLE; |
1802 | nextchar(); | |
a0ed51b3 | 1803 | if (UTF && !PL_utf8_space) |
dfe13c55 | 1804 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
1805 | break; |
1806 | case 'd': | |
a0ed51b3 | 1807 | ret = reg_node(UTF ? DIGITUTF8 : DIGIT); |
a0d0e21e LW |
1808 | *flagp |= HASWIDTH|SIMPLE; |
1809 | nextchar(); | |
a0ed51b3 | 1810 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 1811 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e LW |
1812 | break; |
1813 | case 'D': | |
a0ed51b3 | 1814 | ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT); |
a0d0e21e LW |
1815 | *flagp |= HASWIDTH|SIMPLE; |
1816 | nextchar(); | |
a0ed51b3 | 1817 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 1818 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e | 1819 | break; |
a14b48bc LW |
1820 | case 'p': |
1821 | case 'P': | |
1822 | { /* a lovely hack--pretend we saw [\pX] instead */ | |
1823 | char* oldregxend = PL_regxend; | |
1824 | ||
1825 | if (PL_regcomp_parse[1] == '{') { | |
1826 | PL_regxend = strchr(PL_regcomp_parse, '}'); | |
1827 | if (!PL_regxend) | |
1828 | FAIL("Missing right brace on \\p{}"); | |
1829 | PL_regxend++; | |
1830 | } | |
1831 | else | |
1832 | PL_regxend = PL_regcomp_parse + 2; | |
1833 | PL_regcomp_parse--; | |
1834 | ||
1835 | ret = regclassutf8(); | |
1836 | ||
1837 | PL_regxend = oldregxend; | |
1838 | PL_regcomp_parse--; | |
1839 | nextchar(); | |
1840 | *flagp |= HASWIDTH|SIMPLE; | |
1841 | } | |
1842 | break; | |
a0d0e21e LW |
1843 | case 'n': |
1844 | case 'r': | |
1845 | case 't': | |
1846 | case 'f': | |
1847 | case 'e': | |
1848 | case 'a': | |
1849 | case 'x': | |
1850 | case 'c': | |
1851 | case '0': | |
1852 | goto defchar; | |
1853 | case '1': case '2': case '3': case '4': | |
1854 | case '5': case '6': case '7': case '8': case '9': | |
1855 | { | |
3280af22 | 1856 | I32 num = atoi(PL_regcomp_parse); |
a0d0e21e | 1857 | |
3280af22 | 1858 | if (num > 9 && num >= PL_regnpar) |
a0d0e21e LW |
1859 | goto defchar; |
1860 | else { | |
3280af22 | 1861 | if (!SIZE_ONLY && num > PL_regcomp_rx->nparens) |
ef64f398 | 1862 | FAIL("reference to nonexistent group"); |
3280af22 | 1863 | PL_regsawback = 1; |
a0ed51b3 LW |
1864 | ret = reganode(FOLD |
1865 | ? (LOC ? REFFL : REFF) | |
c8756f30 | 1866 | : REF, num); |
a0d0e21e | 1867 | *flagp |= HASWIDTH; |
3280af22 NIS |
1868 | while (isDIGIT(*PL_regcomp_parse)) |
1869 | PL_regcomp_parse++; | |
1870 | PL_regcomp_parse--; | |
a0d0e21e LW |
1871 | nextchar(); |
1872 | } | |
1873 | } | |
1874 | break; | |
1875 | case '\0': | |
3280af22 | 1876 | if (PL_regcomp_parse >= PL_regxend) |
a0d0e21e LW |
1877 | FAIL("trailing \\ in regexp"); |
1878 | /* FALL THROUGH */ | |
1879 | default: | |
c9f97d15 IZ |
1880 | /* Do not generate `unrecognized' warnings here, we fall |
1881 | back into the quick-grab loop below */ | |
a0d0e21e LW |
1882 | goto defchar; |
1883 | } | |
1884 | break; | |
4633a7c4 LW |
1885 | |
1886 | case '#': | |
3280af22 NIS |
1887 | if (PL_regflags & PMf_EXTENDED) { |
1888 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++; | |
1889 | if (PL_regcomp_parse < PL_regxend) | |
4633a7c4 LW |
1890 | goto tryagain; |
1891 | } | |
1892 | /* FALL THROUGH */ | |
1893 | ||
a0d0e21e LW |
1894 | default: { |
1895 | register I32 len; | |
a0ed51b3 | 1896 | register UV ender; |
a0d0e21e | 1897 | register char *p; |
c277df42 | 1898 | char *oldp, *s; |
a0d0e21e LW |
1899 | I32 numlen; |
1900 | ||
3280af22 | 1901 | PL_regcomp_parse++; |
a0d0e21e LW |
1902 | |
1903 | defchar: | |
a0ed51b3 LW |
1904 | ret = reg_node(FOLD |
1905 | ? (LOC ? EXACTFL : EXACTF) | |
bbce6d69 | 1906 | : EXACT); |
161b471a | 1907 | s = (char *) OPERAND(ret); |
c277df42 | 1908 | regc(0, s++); /* save spot for len */ |
3280af22 NIS |
1909 | for (len = 0, p = PL_regcomp_parse - 1; |
1910 | len < 127 && p < PL_regxend; | |
a0d0e21e LW |
1911 | len++) |
1912 | { | |
1913 | oldp = p; | |
5b5a24f7 | 1914 | |
3280af22 NIS |
1915 | if (PL_regflags & PMf_EXTENDED) |
1916 | p = regwhite(p, PL_regxend); | |
a0d0e21e LW |
1917 | switch (*p) { |
1918 | case '^': | |
1919 | case '$': | |
1920 | case '.': | |
1921 | case '[': | |
1922 | case '(': | |
1923 | case ')': | |
1924 | case '|': | |
1925 | goto loopdone; | |
1926 | case '\\': | |
1927 | switch (*++p) { | |
1928 | case 'A': | |
1929 | case 'G': | |
1930 | case 'Z': | |
b85d18e9 | 1931 | case 'z': |
a0d0e21e LW |
1932 | case 'w': |
1933 | case 'W': | |
1934 | case 'b': | |
1935 | case 'B': | |
1936 | case 's': | |
1937 | case 'S': | |
1938 | case 'd': | |
1939 | case 'D': | |
a14b48bc LW |
1940 | case 'p': |
1941 | case 'P': | |
a0d0e21e LW |
1942 | --p; |
1943 | goto loopdone; | |
1944 | case 'n': | |
1945 | ender = '\n'; | |
1946 | p++; | |
a687059c | 1947 | break; |
a0d0e21e LW |
1948 | case 'r': |
1949 | ender = '\r'; | |
1950 | p++; | |
a687059c | 1951 | break; |
a0d0e21e LW |
1952 | case 't': |
1953 | ender = '\t'; | |
1954 | p++; | |
a687059c | 1955 | break; |
a0d0e21e LW |
1956 | case 'f': |
1957 | ender = '\f'; | |
1958 | p++; | |
a687059c | 1959 | break; |
a0d0e21e LW |
1960 | case 'e': |
1961 | ender = '\033'; | |
1962 | p++; | |
a687059c | 1963 | break; |
a0d0e21e LW |
1964 | case 'a': |
1965 | ender = '\007'; | |
1966 | p++; | |
a687059c | 1967 | break; |
a0d0e21e | 1968 | case 'x': |
a0ed51b3 LW |
1969 | if (*++p == '{') { |
1970 | char* e = strchr(p, '}'); | |
1971 | ||
1972 | if (!e) | |
1973 | FAIL("Missing right brace on \\x{}"); | |
1974 | else if (UTF) { | |
1975 | ender = scan_hex(p + 1, e - p, &numlen); | |
1976 | if (numlen + len >= 127) { /* numlen is generous */ | |
1977 | p--; | |
1978 | goto loopdone; | |
1979 | } | |
1980 | p = e + 1; | |
1981 | } | |
1982 | else | |
1983 | FAIL("Can't use \\x{} without 'use utf8' declaration"); | |
1984 | } | |
1985 | else { | |
1986 | ender = scan_hex(p, 2, &numlen); | |
1987 | p += numlen; | |
1988 | } | |
a687059c | 1989 | break; |
a0d0e21e LW |
1990 | case 'c': |
1991 | p++; | |
bbce6d69 | 1992 | ender = UCHARAT(p++); |
1993 | ender = toCTRL(ender); | |
a687059c | 1994 | break; |
a0d0e21e LW |
1995 | case '0': case '1': case '2': case '3':case '4': |
1996 | case '5': case '6': case '7': case '8':case '9': | |
1997 | if (*p == '0' || | |
3280af22 | 1998 | (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) { |
a0d0e21e LW |
1999 | ender = scan_oct(p, 3, &numlen); |
2000 | p += numlen; | |
2001 | } | |
2002 | else { | |
2003 | --p; | |
2004 | goto loopdone; | |
a687059c LW |
2005 | } |
2006 | break; | |
a0d0e21e | 2007 | case '\0': |
3280af22 | 2008 | if (p >= PL_regxend) |
a687059c LW |
2009 | FAIL("trailing \\ in regexp"); |
2010 | /* FALL THROUGH */ | |
a0d0e21e | 2011 | default: |
c9f97d15 | 2012 | if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p)) |
cea2e8a9 | 2013 | Perl_warner(aTHX_ WARN_UNSAFE, |
c9f97d15 IZ |
2014 | "/%.127s/: Unrecognized escape \\%c passed through", |
2015 | PL_regprecomp, | |
2016 | *p); | |
a0ed51b3 | 2017 | goto normal_default; |
a0d0e21e LW |
2018 | } |
2019 | break; | |
a687059c | 2020 | default: |
a0ed51b3 LW |
2021 | normal_default: |
2022 | if ((*p & 0xc0) == 0xc0 && UTF) { | |
dfe13c55 | 2023 | ender = utf8_to_uv((U8*)p, &numlen); |
a0ed51b3 LW |
2024 | p += numlen; |
2025 | } | |
2026 | else | |
2027 | ender = *p++; | |
a0d0e21e | 2028 | break; |
a687059c | 2029 | } |
3280af22 NIS |
2030 | if (PL_regflags & PMf_EXTENDED) |
2031 | p = regwhite(p, PL_regxend); | |
a0ed51b3 LW |
2032 | if (UTF && FOLD) { |
2033 | if (LOC) | |
2034 | ender = toLOWER_LC_uni(ender); | |
2035 | else | |
2036 | ender = toLOWER_uni(ender); | |
2037 | } | |
a0d0e21e LW |
2038 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
2039 | if (len) | |
2040 | p = oldp; | |
a0ed51b3 LW |
2041 | else if (ender >= 0x80 && UTF) { |
2042 | reguni(ender, s, &numlen); | |
2043 | s += numlen; | |
2044 | len += numlen; | |
2045 | } | |
a0d0e21e LW |
2046 | else { |
2047 | len++; | |
c277df42 | 2048 | regc(ender, s++); |
a0d0e21e LW |
2049 | } |
2050 | break; | |
a687059c | 2051 | } |
a0ed51b3 LW |
2052 | if (ender >= 0x80 && UTF) { |
2053 | reguni(ender, s, &numlen); | |
2054 | s += numlen; | |
2055 | len += numlen - 1; | |
2056 | } | |
2057 | else | |
2058 | regc(ender, s++); | |
a0d0e21e LW |
2059 | } |
2060 | loopdone: | |
3280af22 | 2061 | PL_regcomp_parse = p - 1; |
a0d0e21e LW |
2062 | nextchar(); |
2063 | if (len < 0) | |
2064 | FAIL("internal disaster in regexp"); | |
2065 | if (len > 0) | |
2066 | *flagp |= HASWIDTH; | |
2067 | if (len == 1) | |
2068 | *flagp |= SIMPLE; | |
c277df42 | 2069 | if (!SIZE_ONLY) |
a0d0e21e | 2070 | *OPERAND(ret) = len; |
c277df42 IZ |
2071 | regc('\0', s++); |
2072 | if (SIZE_ONLY) { | |
3280af22 | 2073 | PL_regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
a0ed51b3 LW |
2074 | } |
2075 | else { | |
3280af22 | 2076 | PL_regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
c277df42 | 2077 | } |
a687059c | 2078 | } |
a0d0e21e LW |
2079 | break; |
2080 | } | |
a687059c | 2081 | |
a0d0e21e | 2082 | return(ret); |
a687059c LW |
2083 | } |
2084 | ||
873ef191 | 2085 | STATIC char * |
cea2e8a9 | 2086 | S_regwhite(pTHX_ char *p, char *e) |
5b5a24f7 CS |
2087 | { |
2088 | while (p < e) { | |
2089 | if (isSPACE(*p)) | |
2090 | ++p; | |
2091 | else if (*p == '#') { | |
2092 | do { | |
2093 | p++; | |
2094 | } while (p < e && *p != '\n'); | |
2095 | } | |
2096 | else | |
2097 | break; | |
2098 | } | |
2099 | return p; | |
2100 | } | |
2101 | ||
b8c5462f JH |
2102 | /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]]. |
2103 | Character classes ([:foo:]) can also be negated ([:^foo:]). | |
2104 | Returns a named class id (ANYOF_XXX) if successful, -1 otherwise. | |
2105 | Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed, | |
2106 | but trigger warnings because they are currently unimplemented. */ | |
2107 | STATIC I32 | |
cea2e8a9 | 2108 | S_regpposixcc(pTHX_ I32 value) |
620e46c5 | 2109 | { |
11b8faa4 | 2110 | dTHR; |
620e46c5 | 2111 | char *posixcc = 0; |
b8c5462f | 2112 | I32 namedclass = -1; |
620e46c5 JH |
2113 | |
2114 | if (value == '[' && PL_regcomp_parse + 1 < PL_regxend && | |
2115 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ | |
2116 | (*PL_regcomp_parse == ':' || | |
2117 | *PL_regcomp_parse == '=' || | |
2118 | *PL_regcomp_parse == '.')) { | |
2119 | char c = *PL_regcomp_parse; | |
2120 | char* s = PL_regcomp_parse++; | |
2121 | ||
2122 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c) | |
2123 | PL_regcomp_parse++; | |
2124 | if (PL_regcomp_parse == PL_regxend) | |
2125 | /* Grandfather lone [:, [=, [. */ | |
2126 | PL_regcomp_parse = s; | |
2127 | else { | |
b8c5462f JH |
2128 | char* t = PL_regcomp_parse++; /* skip over the c */ |
2129 | ||
2130 | if (*PL_regcomp_parse == ']') { | |
2131 | PL_regcomp_parse++; /* skip over the ending ] */ | |
2132 | posixcc = s + 1; | |
2133 | if (*s == ':') { | |
2134 | I32 complement = *posixcc == '^' ? *posixcc++ : 0; | |
2135 | I32 skip = 5; /* the most common skip */ | |
2136 | ||
2137 | switch (*posixcc) { | |
2138 | case 'a': | |
2139 | if (strnEQ(posixcc, "alnum", 5)) | |
2140 | namedclass = | |
2141 | complement ? ANYOF_NALNUMC : ANYOF_ALNUMC; | |
2142 | else if (strnEQ(posixcc, "alpha", 5)) | |
2143 | namedclass = | |
2144 | complement ? ANYOF_NALPHA : ANYOF_ALPHA; | |
2145 | else if (strnEQ(posixcc, "ascii", 5)) | |
2146 | namedclass = | |
2147 | complement ? ANYOF_NASCII : ANYOF_ASCII; | |
2148 | break; | |
2149 | case 'c': | |
2150 | if (strnEQ(posixcc, "cntrl", 5)) | |
2151 | namedclass = | |
2152 | complement ? ANYOF_NCNTRL : ANYOF_CNTRL; | |
2153 | break; | |
2154 | case 'd': | |
2155 | if (strnEQ(posixcc, "digit", 5)) | |
2156 | namedclass = | |
2157 | complement ? ANYOF_NDIGIT : ANYOF_DIGIT; | |
2158 | break; | |
2159 | case 'g': | |
2160 | if (strnEQ(posixcc, "graph", 5)) | |
2161 | namedclass = | |
2162 | complement ? ANYOF_NGRAPH : ANYOF_GRAPH; | |
2163 | break; | |
2164 | case 'l': | |
2165 | if (strnEQ(posixcc, "lower", 5)) | |
2166 | namedclass = | |
2167 | complement ? ANYOF_NLOWER : ANYOF_LOWER; | |
2168 | break; | |
2169 | case 'p': | |
2170 | if (strnEQ(posixcc, "print", 5)) | |
2171 | namedclass = | |
2172 | complement ? ANYOF_NPRINT : ANYOF_PRINT; | |
2173 | else if (strnEQ(posixcc, "punct", 5)) | |
2174 | namedclass = | |
2175 | complement ? ANYOF_NPUNCT : ANYOF_PUNCT; | |
2176 | break; | |
2177 | case 's': | |
2178 | if (strnEQ(posixcc, "space", 5)) | |
2179 | namedclass = | |
2180 | complement ? ANYOF_NSPACE : ANYOF_SPACE; | |
2181 | case 'u': | |
2182 | if (strnEQ(posixcc, "upper", 5)) | |
2183 | namedclass = | |
2184 | complement ? ANYOF_NUPPER : ANYOF_UPPER; | |
2185 | break; | |
2186 | case 'w': /* this is not POSIX, this is the Perl \w */ | |
2187 | if (strnEQ(posixcc, "word", 4)) { | |
2188 | namedclass = | |
2189 | complement ? ANYOF_NALNUM : ANYOF_ALNUM; | |
2190 | skip = 4; | |
2191 | } | |
2192 | break; | |
2193 | case 'x': | |
2194 | if (strnEQ(posixcc, "xdigit", 6)) { | |
2195 | namedclass = | |
2196 | complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT; | |
2197 | skip = 6; | |
2198 | } | |
2199 | break; | |
2200 | } | |
2201 | if ((namedclass == -1 || | |
2202 | !(posixcc + skip + 2 < PL_regxend && | |
2203 | (posixcc[skip] == ':' && | |
2204 | posixcc[skip + 1] == ']')))) | |
2205 | Perl_croak(aTHX_ "Character class [:%.*s:] unknown", | |
2206 | t - s - 1, s + 1); | |
2207 | } else if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY) | |
2208 | /* [[=foo=]] and [[.foo.]] are still future. */ | |
cea2e8a9 | 2209 | Perl_warner(aTHX_ WARN_UNSAFE, |
b8c5462f JH |
2210 | "Character class syntax [%c %c] is reserved for future extensions", c, c); |
2211 | } else { | |
2212 | /* Maternal grandfather: | |
2213 | * "[:" ending in ":" but not in ":]" */ | |
767d463e HS |
2214 | PL_regcomp_parse = s; |
2215 | } | |
620e46c5 JH |
2216 | } |
2217 | } | |
2218 | ||
b8c5462f JH |
2219 | return namedclass; |
2220 | } | |
2221 | ||
2222 | STATIC void | |
2223 | S_checkposixcc(pTHX) | |
2224 | { | |
2225 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && | |
2226 | (*PL_regcomp_parse == ':' || | |
2227 | *PL_regcomp_parse == '=' || | |
2228 | *PL_regcomp_parse == '.')) { | |
2229 | char *s = PL_regcomp_parse; | |
2230 | char c = *s++; | |
2231 | ||
2232 | while(*s && isALNUM(*s)) | |
2233 | s++; | |
2234 | if (*s && c == *s && s[1] == ']') { | |
2235 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2236 | "Character class syntax [%c %c] belongs inside character classes", c, c); | |
2237 | if (c == '=' || c == '.') | |
2238 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2239 | "Character class syntax [%c %c] is reserved for future extensions", c, c); | |
2240 | } | |
2241 | } | |
620e46c5 JH |
2242 | } |
2243 | ||
76e3520e | 2244 | STATIC regnode * |
cea2e8a9 | 2245 | S_regclass(pTHX) |
a687059c | 2246 | { |
5c0ca799 | 2247 | dTHR; |
c277df42 | 2248 | register char *opnd, *s; |
a0ed51b3 | 2249 | register I32 value; |
b8c5462f | 2250 | register I32 lastvalue = OOB_CHAR8; |
a0d0e21e | 2251 | register I32 range = 0; |
c277df42 | 2252 | register regnode *ret; |
a0d0e21e LW |
2253 | register I32 def; |
2254 | I32 numlen; | |
b8c5462f | 2255 | I32 namedclass; |
a0d0e21e | 2256 | |
3280af22 | 2257 | s = opnd = (char *) OPERAND(PL_regcode); |
c277df42 | 2258 | ret = reg_node(ANYOF); |
b8c5462f | 2259 | for (value = 0; value < ANYOF_SIZE; value++) |
c277df42 | 2260 | regc(0, s++); |
3280af22 NIS |
2261 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2262 | PL_regnaughty++; | |
2263 | PL_regcomp_parse++; | |
c277df42 | 2264 | if (!SIZE_ONLY) |
b8c5462f | 2265 | ANYOF_FLAGS(opnd) |= ANYOF_INVERT; |
bbce6d69 | 2266 | } |
c277df42 | 2267 | if (!SIZE_ONLY) { |
3280af22 | 2268 | PL_regcode += ANY_SKIP; |
a0ed51b3 | 2269 | if (FOLD) |
b8c5462f | 2270 | ANYOF_FLAGS(opnd) |= ANYOF_FOLD; |
a0ed51b3 | 2271 | if (LOC) |
b8c5462f | 2272 | ANYOF_FLAGS(opnd) |= ANYOF_LOCALE; |
a0ed51b3 LW |
2273 | } |
2274 | else { | |
3280af22 | 2275 | PL_regsize += ANY_SKIP; |
a0d0e21e | 2276 | } |
b8c5462f JH |
2277 | |
2278 | checkposixcc(); | |
2279 | ||
3280af22 | 2280 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
a0d0e21e | 2281 | goto skipcond; /* allow 1st char to be ] or - */ |
3280af22 | 2282 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
a0d0e21e | 2283 | skipcond: |
b8c5462f | 2284 | namedclass = -1; |
a0ed51b3 | 2285 | value = UCHARAT(PL_regcomp_parse++); |
620e46c5 | 2286 | if (value == '[') |
b8c5462f | 2287 | namedclass = regpposixcc(value); |
620e46c5 | 2288 | else if (value == '\\') { |
a0ed51b3 LW |
2289 | value = UCHARAT(PL_regcomp_parse++); |
2290 | switch (value) { | |
b8c5462f JH |
2291 | case 'w': namedclass = ANYOF_ALNUM; break; |
2292 | case 'W': namedclass = ANYOF_NALNUM; break; | |
2293 | case 's': namedclass = ANYOF_SPACE; break; | |
2294 | case 'S': namedclass = ANYOF_NSPACE; break; | |
2295 | case 'd': namedclass = ANYOF_DIGIT; break; | |
2296 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
2297 | case 'n': value = '\n'; break; | |
2298 | case 'r': value = '\r'; break; | |
2299 | case 't': value = '\t'; break; | |
2300 | case 'f': value = '\f'; break; | |
2301 | case 'b': value = '\b'; break; | |
2302 | case 'e': value = '\033'; break; | |
2303 | case 'a': value = '\007'; break; | |
2304 | case 'x': | |
2305 | value = scan_hex(PL_regcomp_parse, 2, &numlen); | |
2306 | PL_regcomp_parse += numlen; | |
2307 | break; | |
2308 | case 'c': | |
2309 | value = UCHARAT(PL_regcomp_parse++); | |
2310 | value = toCTRL(value); | |
2311 | break; | |
2312 | case '0': case '1': case '2': case '3': case '4': | |
2313 | case '5': case '6': case '7': case '8': case '9': | |
2314 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); | |
2315 | PL_regcomp_parse += numlen; | |
2316 | break; | |
2317 | } | |
2318 | } | |
2319 | if (!SIZE_ONLY && namedclass > -1) { | |
2320 | switch (namedclass) { | |
2321 | case ANYOF_ALNUM: | |
2322 | if (LOC) | |
2323 | ANYOF_CLASS_SET(opnd, ANYOF_ALNUM); | |
2324 | else { | |
2325 | for (value = 0; value < 256; value++) | |
2326 | if (isALNUM(value)) | |
2327 | ANYOF_BITMAP_SET(opnd, value); | |
bbce6d69 | 2328 | } |
b8c5462f JH |
2329 | break; |
2330 | case ANYOF_NALNUM: | |
2331 | if (LOC) | |
2332 | ANYOF_CLASS_SET(opnd, ANYOF_NALNUM); | |
2333 | else { | |
2334 | for (value = 0; value < 256; value++) | |
2335 | if (!isALNUM(value)) | |
2336 | ANYOF_BITMAP_SET(opnd, value); | |
bbce6d69 | 2337 | } |
b8c5462f JH |
2338 | break; |
2339 | case ANYOF_SPACE: | |
2340 | if (LOC) | |
2341 | ANYOF_CLASS_SET(opnd, ANYOF_SPACE); | |
2342 | else { | |
2343 | for (value = 0; value < 256; value++) | |
2344 | if (isSPACE(value)) | |
2345 | ANYOF_BITMAP_SET(opnd, value); | |
bbce6d69 | 2346 | } |
b8c5462f JH |
2347 | break; |
2348 | case ANYOF_NSPACE: | |
2349 | if (LOC) | |
2350 | ANYOF_CLASS_SET(opnd, ANYOF_NSPACE); | |
2351 | else { | |
2352 | for (value = 0; value < 256; value++) | |
2353 | if (!isSPACE(value)) | |
2354 | ANYOF_BITMAP_SET(opnd, value); | |
bbce6d69 | 2355 | } |
b8c5462f JH |
2356 | break; |
2357 | case ANYOF_DIGIT: | |
2358 | if (LOC) | |
2359 | ANYOF_CLASS_SET(opnd, ANYOF_DIGIT); | |
2360 | else { | |
a0ed51b3 | 2361 | for (value = '0'; value <= '9'; value++) |
b8c5462f | 2362 | ANYOF_BITMAP_SET(opnd, value); |
ae5c130c | 2363 | } |
b8c5462f JH |
2364 | break; |
2365 | case ANYOF_NDIGIT: | |
2366 | if (LOC) | |
2367 | ANYOF_CLASS_SET(opnd, ANYOF_NDIGIT); | |
2368 | else { | |
a0ed51b3 | 2369 | for (value = 0; value < '0'; value++) |
b8c5462f | 2370 | ANYOF_BITMAP_SET(opnd, value); |
a0ed51b3 | 2371 | for (value = '9' + 1; value < 256; value++) |
b8c5462f JH |
2372 | ANYOF_BITMAP_SET(opnd, value); |
2373 | } | |
2374 | break; | |
2375 | case ANYOF_NALNUMC: | |
2376 | if (LOC) | |
2377 | ANYOF_CLASS_SET(opnd, ANYOF_NALNUMC); | |
2378 | else { | |
2379 | for (value = 0; value < 256; value++) | |
2380 | if (!isALNUMC(value)) | |
2381 | ANYOF_BITMAP_SET(opnd, value); | |
2382 | } | |
2383 | break; | |
2384 | case ANYOF_ALNUMC: | |
2385 | if (LOC) | |
2386 | ANYOF_CLASS_SET(opnd, ANYOF_ALNUMC); | |
2387 | else { | |
2388 | for (value = 0; value < 256; value++) | |
2389 | if (isALNUMC(value)) | |
2390 | ANYOF_BITMAP_SET(opnd, value); | |
2391 | } | |
2392 | break; | |
2393 | case ANYOF_ALPHA: | |
2394 | if (LOC) | |
2395 | ANYOF_CLASS_SET(opnd, ANYOF_ALPHA); | |
2396 | else { | |
2397 | for (value = 0; value < 256; value++) | |
2398 | if (isALPHA(value)) | |
2399 | ANYOF_BITMAP_SET(opnd, value); | |
ae5c130c | 2400 | } |
a0d0e21e | 2401 | break; |
b8c5462f JH |
2402 | case ANYOF_NALPHA: |
2403 | if (LOC) | |
2404 | ANYOF_CLASS_SET(opnd, ANYOF_NALPHA); | |
2405 | else { | |
2406 | for (value = 0; value < 256; value++) | |
2407 | if (!isALPHA(value)) | |
2408 | ANYOF_BITMAP_SET(opnd, value); | |
2409 | } | |
a0d0e21e | 2410 | break; |
b8c5462f JH |
2411 | case ANYOF_ASCII: |
2412 | if (LOC) | |
2413 | ANYOF_CLASS_SET(opnd, ANYOF_ASCII); | |
2414 | else { | |
2415 | for (value = 0; value < 128; value++) | |
2416 | ANYOF_BITMAP_SET(opnd, value); | |
2417 | } | |
a0d0e21e | 2418 | break; |
b8c5462f JH |
2419 | case ANYOF_NASCII: |
2420 | if (LOC) | |
2421 | ANYOF_CLASS_SET(opnd, ANYOF_NASCII); | |
2422 | else { | |
2423 | for (value = 128; value < 256; value++) | |
2424 | ANYOF_BITMAP_SET(opnd, value); | |
2425 | } | |
a0d0e21e | 2426 | break; |
b8c5462f JH |
2427 | case ANYOF_CNTRL: |
2428 | if (LOC) | |
2429 | ANYOF_CLASS_SET(opnd, ANYOF_CNTRL); | |
2430 | else { | |
2431 | for (value = 0; value < 256; value++) | |
2432 | if (isCNTRL(value)) | |
2433 | ANYOF_BITMAP_SET(opnd, value); | |
2434 | } | |
2435 | lastvalue = OOB_CHAR8; | |
a0d0e21e | 2436 | break; |
b8c5462f JH |
2437 | case ANYOF_NCNTRL: |
2438 | if (LOC) | |
2439 | ANYOF_CLASS_SET(opnd, ANYOF_NCNTRL); | |
2440 | else { | |
2441 | for (value = 0; value < 256; value++) | |
2442 | if (!isCNTRL(value)) | |
2443 | ANYOF_BITMAP_SET(opnd, value); | |
2444 | } | |
a0d0e21e | 2445 | break; |
b8c5462f JH |
2446 | case ANYOF_GRAPH: |
2447 | if (LOC) | |
2448 | ANYOF_CLASS_SET(opnd, ANYOF_GRAPH); | |
2449 | else { | |
2450 | for (value = 0; value < 256; value++) | |
2451 | if (isGRAPH(value)) | |
2452 | ANYOF_BITMAP_SET(opnd, value); | |
2453 | } | |
a0d0e21e | 2454 | break; |
b8c5462f JH |
2455 | case ANYOF_NGRAPH: |
2456 | if (LOC) | |
2457 | ANYOF_CLASS_SET(opnd, ANYOF_NGRAPH); | |
2458 | else { | |
2459 | for (value = 0; value < 256; value++) | |
2460 | if (!isGRAPH(value)) | |
2461 | ANYOF_BITMAP_SET(opnd, value); | |
2462 | } | |
a0d0e21e | 2463 | break; |
b8c5462f JH |
2464 | case ANYOF_LOWER: |
2465 | if (LOC) | |
2466 | ANYOF_CLASS_SET(opnd, ANYOF_LOWER); | |
2467 | else { | |
2468 | for (value = 0; value < 256; value++) | |
2469 | if (isLOWER(value)) | |
2470 | ANYOF_BITMAP_SET(opnd, value); | |
2471 | } | |
a0d0e21e | 2472 | break; |
b8c5462f JH |
2473 | case ANYOF_NLOWER: |
2474 | if (LOC) | |
2475 | ANYOF_CLASS_SET(opnd, ANYOF_NLOWER); | |
2476 | else { | |
2477 | for (value = 0; value < 256; value++) | |
2478 | if (!isLOWER(value)) | |
2479 | ANYOF_BITMAP_SET(opnd, value); | |
2480 | } | |
2481 | break; | |
2482 | case ANYOF_PRINT: | |
2483 | if (LOC) | |
2484 | ANYOF_CLASS_SET(opnd, ANYOF_PRINT); | |
2485 | else { | |
2486 | for (value = 0; value < 256; value++) | |
2487 | if (isPRINT(value)) | |
2488 | ANYOF_BITMAP_SET(opnd, value); | |
2489 | } | |
2490 | break; | |
2491 | case ANYOF_NPRINT: | |
2492 | if (LOC) | |
2493 | ANYOF_CLASS_SET(opnd, ANYOF_NPRINT); | |
2494 | else { | |
2495 | for (value = 0; value < 256; value++) | |
2496 | if (!isPRINT(value)) | |
2497 | ANYOF_BITMAP_SET(opnd, value); | |
2498 | } | |
2499 | break; | |
2500 | case ANYOF_PUNCT: | |
2501 | if (LOC) | |
2502 | ANYOF_CLASS_SET(opnd, ANYOF_PUNCT); | |
2503 | else { | |
2504 | for (value = 0; value < 256; value++) | |
2505 | if (isPUNCT(value)) | |
2506 | ANYOF_BITMAP_SET(opnd, value); | |
2507 | } | |
2508 | break; | |
2509 | case ANYOF_NPUNCT: | |
2510 | if (LOC) | |
2511 | ANYOF_CLASS_SET(opnd, ANYOF_NPUNCT); | |
2512 | else { | |
2513 | for (value = 0; value < 256; value++) | |
2514 | if (!isPUNCT(value)) | |
2515 | ANYOF_BITMAP_SET(opnd, value); | |
2516 | } | |
2517 | break; | |
2518 | case ANYOF_UPPER: | |
2519 | if (LOC) | |
2520 | ANYOF_CLASS_SET(opnd, ANYOF_UPPER); | |
2521 | else { | |
2522 | for (value = 0; value < 256; value++) | |
2523 | if (isUPPER(value)) | |
2524 | ANYOF_BITMAP_SET(opnd, value); | |
2525 | } | |
2526 | break; | |
2527 | case ANYOF_NUPPER: | |
2528 | if (LOC) | |
2529 | ANYOF_CLASS_SET(opnd, ANYOF_NUPPER); | |
2530 | else { | |
2531 | for (value = 0; value < 256; value++) | |
2532 | if (!isUPPER(value)) | |
2533 | ANYOF_BITMAP_SET(opnd, value); | |
2534 | } | |
2535 | break; | |
2536 | case ANYOF_XDIGIT: | |
2537 | if (LOC) | |
2538 | ANYOF_CLASS_SET(opnd, ANYOF_XDIGIT); | |
2539 | else { | |
2540 | for (value = 0; value < 256; value++) | |
2541 | if (isXDIGIT(value)) | |
2542 | ANYOF_BITMAP_SET(opnd, value); | |
2543 | } | |
2544 | break; | |
2545 | case ANYOF_NXDIGIT: | |
2546 | if (LOC) | |
2547 | ANYOF_CLASS_SET(opnd, ANYOF_NXDIGIT); | |
2548 | else { | |
2549 | for (value = 0; value < 256; value++) | |
2550 | if (!isXDIGIT(value)) | |
2551 | ANYOF_BITMAP_SET(opnd, value); | |
2552 | } | |
2553 | break; | |
2554 | default: | |
2555 | FAIL("invalid [::] class in regexp"); | |
a0d0e21e LW |
2556 | break; |
2557 | } | |
b8c5462f JH |
2558 | if (LOC) |
2559 | ANYOF_FLAGS(opnd) |= ANYOF_CLASS; | |
2560 | lastvalue = OOB_CHAR8; | |
a0d0e21e | 2561 | } |
b8c5462f | 2562 | else |
a0d0e21e | 2563 | if (range) { |
a0ed51b3 | 2564 | if (lastvalue > value) |
a0d0e21e LW |
2565 | FAIL("invalid [] range in regexp"); |
2566 | range = 0; | |
2567 | } | |
2568 | else { | |
a0ed51b3 | 2569 | lastvalue = value; |
3280af22 NIS |
2570 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
2571 | PL_regcomp_parse[1] != ']') { | |
2572 | PL_regcomp_parse++; | |
a0d0e21e LW |
2573 | range = 1; |
2574 | continue; /* do it next time */ | |
2575 | } | |
a687059c | 2576 | } |
ae5c130c | 2577 | if (!SIZE_ONLY) { |
8ada0baa JH |
2578 | #ifndef ASCIIish |
2579 | if ((isLOWER(lastvalue) && isLOWER(value)) || | |
c8dba6f3 GS |
2580 | (isUPPER(lastvalue) && isUPPER(value))) |
2581 | { | |
2582 | I32 i; | |
8ada0baa JH |
2583 | if (isLOWER(lastvalue)) { |
2584 | for (i = lastvalue; i <= value; i++) | |
2585 | if (isLOWER(i)) | |
b8c5462f | 2586 | ANYOF_BITMAP_SET(opnd, i); |
8ada0baa JH |
2587 | } else { |
2588 | for (i = lastvalue; i <= value; i++) | |
2589 | if (isUPPER(i)) | |
b8c5462f | 2590 | ANYOF_BITMAP_SET(opnd, i); |
8ada0baa JH |
2591 | } |
2592 | } | |
2593 | else | |
2594 | #endif | |
2595 | for ( ; lastvalue <= value; lastvalue++) | |
b8c5462f | 2596 | ANYOF_BITMAP_SET(opnd, lastvalue); |
8ada0baa | 2597 | } |
a0ed51b3 | 2598 | lastvalue = value; |
a0d0e21e | 2599 | } |
ae5c130c | 2600 | /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ |
b8c5462f JH |
2601 | if (!SIZE_ONLY && |
2602 | (ANYOF_FLAGS(opnd) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD) { | |
a0ed51b3 | 2603 | for (value = 0; value < 256; ++value) { |
b8c5462f | 2604 | if (ANYOF_BITMAP_TEST(opnd, value)) { |
22c35a8c | 2605 | I32 cf = PL_fold[value]; |
b8c5462f | 2606 | ANYOF_BITMAP_SET(opnd, cf); |
ae5c130c GS |
2607 | } |
2608 | } | |
b8c5462f | 2609 | ANYOF_FLAGS(opnd) &= ~ANYOF_FOLD; |
ae5c130c GS |
2610 | } |
2611 | /* optimize inverted simple patterns (e.g. [^a-z]) */ | |
b8c5462f JH |
2612 | if (!SIZE_ONLY && (ANYOF_FLAGS(opnd) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) { |
2613 | for (value = 0; value < ANYOF_BITMAP_SIZE; ++value) | |
2614 | opnd[ANYOF_BITMAP_OFFSET + value] ^= ANYOF_FLAGS_ALL; | |
2615 | ANYOF_FLAGS(opnd) = 0; | |
ae5c130c | 2616 | } |
a0d0e21e LW |
2617 | return ret; |
2618 | } | |
2619 | ||
a0ed51b3 | 2620 | STATIC regnode * |
cea2e8a9 | 2621 | S_regclassutf8(pTHX) |
a0ed51b3 | 2622 | { |
b8c5462f | 2623 | dTHR; |
a0ed51b3 LW |
2624 | register char *opnd, *e; |
2625 | register U32 value; | |
b8c5462f | 2626 | register U32 lastvalue = OOB_UTF8; |
a0ed51b3 LW |
2627 | register I32 range = 0; |
2628 | register regnode *ret; | |
2629 | I32 numlen; | |
2630 | I32 n; | |
2631 | SV *listsv; | |
2632 | U8 flags = 0; | |
b8c5462f | 2633 | I32 namedclass; |
a0ed51b3 LW |
2634 | |
2635 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ | |
2636 | PL_regnaughty++; | |
2637 | PL_regcomp_parse++; | |
2638 | if (!SIZE_ONLY) | |
2639 | flags |= ANYOF_INVERT; | |
2640 | } | |
2641 | if (!SIZE_ONLY) { | |
2642 | if (FOLD) | |
2643 | flags |= ANYOF_FOLD; | |
2644 | if (LOC) | |
2645 | flags |= ANYOF_LOCALE; | |
79cb57f6 | 2646 | listsv = newSVpvn("# comment\n",10); |
a0ed51b3 LW |
2647 | } |
2648 | ||
b8c5462f JH |
2649 | checkposixcc(); |
2650 | ||
a0ed51b3 LW |
2651 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
2652 | goto skipcond; /* allow 1st char to be ] or - */ | |
2653 | ||
2654 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { | |
2655 | skipcond: | |
b8c5462f | 2656 | namedclass = -1; |
dfe13c55 | 2657 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 LW |
2658 | PL_regcomp_parse += numlen; |
2659 | ||
620e46c5 | 2660 | if (value == '[') |
b8c5462f | 2661 | namedclass = regpposixcc(value); |
620e46c5 | 2662 | else if (value == '\\') { |
dfe13c55 | 2663 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 LW |
2664 | PL_regcomp_parse += numlen; |
2665 | switch (value) { | |
b8c5462f JH |
2666 | case 'w': namedclass = ANYOF_ALNUM; break; |
2667 | case 'W': namedclass = ANYOF_NALNUM; break; | |
2668 | case 's': namedclass = ANYOF_SPACE; break; | |
2669 | case 'S': namedclass = ANYOF_NSPACE; break; | |
2670 | case 'd': namedclass = ANYOF_DIGIT; break; | |
2671 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
a0ed51b3 LW |
2672 | case 'p': |
2673 | case 'P': | |
2674 | if (*PL_regcomp_parse == '{') { | |
2675 | e = strchr(PL_regcomp_parse++, '}'); | |
2676 | if (!e) | |
2677 | FAIL("Missing right brace on \\p{}"); | |
2678 | n = e - PL_regcomp_parse; | |
2679 | } | |
2680 | else { | |
2681 | e = PL_regcomp_parse; | |
2682 | n = 1; | |
2683 | } | |
2684 | if (!SIZE_ONLY) { | |
2685 | if (value == 'p') | |
b8c5462f JH |
2686 | Perl_sv_catpvf(aTHX_ listsv, |
2687 | "+utf8::%.*s\n", n, PL_regcomp_parse); | |
a0ed51b3 | 2688 | else |
cea2e8a9 | 2689 | Perl_sv_catpvf(aTHX_ listsv, |
b8c5462f | 2690 | "!utf8::%.*s\n", n, PL_regcomp_parse); |
a0ed51b3 LW |
2691 | } |
2692 | PL_regcomp_parse = e + 1; | |
b8c5462f | 2693 | lastvalue = OOB_UTF8; |
a0ed51b3 | 2694 | continue; |
b8c5462f JH |
2695 | case 'n': value = '\n'; break; |
2696 | case 'r': value = '\r'; break; | |
2697 | case 't': value = '\t'; break; | |
2698 | case 'f': value = '\f'; break; | |
2699 | case 'b': value = '\b'; break; | |
2700 | case 'e': value = '\033'; break; | |
2701 | case 'a': value = '\007'; break; | |
a0ed51b3 LW |
2702 | case 'x': |
2703 | if (*PL_regcomp_parse == '{') { | |
2704 | e = strchr(PL_regcomp_parse++, '}'); | |
2705 | if (!e) | |
2706 | FAIL("Missing right brace on \\x{}"); | |
b8c5462f JH |
2707 | value = scan_hex(PL_regcomp_parse, |
2708 | e - PL_regcomp_parse, | |
2709 | &numlen); | |
a0ed51b3 LW |
2710 | PL_regcomp_parse = e + 1; |
2711 | } | |
2712 | else { | |
2713 | value = scan_hex(PL_regcomp_parse, 2, &numlen); | |
2714 | PL_regcomp_parse += numlen; | |
2715 | } | |
2716 | break; | |
2717 | case 'c': | |
2718 | value = UCHARAT(PL_regcomp_parse++); | |
2719 | value = toCTRL(value); | |
2720 | break; | |
2721 | case '0': case '1': case '2': case '3': case '4': | |
2722 | case '5': case '6': case '7': case '8': case '9': | |
2723 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); | |
2724 | PL_regcomp_parse += numlen; | |
2725 | break; | |
2726 | } | |
2727 | } | |
b8c5462f JH |
2728 | if (!SIZE_ONLY && namedclass > -1) { |
2729 | switch (namedclass) { | |
2730 | case ANYOF_ALNUM: | |
2731 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n"); break; | |
2732 | case ANYOF_NALNUM: | |
2733 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n"); break; | |
2734 | case ANYOF_ALNUMC: | |
2735 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n"); break; | |
2736 | case ANYOF_NALNUMC: | |
2737 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n"); break; | |
2738 | case ANYOF_ALPHA: | |
2739 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n"); break; | |
2740 | case ANYOF_NALPHA: | |
2741 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n"); break; | |
2742 | case ANYOF_ASCII: | |
2743 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n"); break; | |
2744 | case ANYOF_NASCII: | |
2745 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n"); break; | |
2746 | case ANYOF_CNTRL: | |
2747 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n"); break; | |
2748 | case ANYOF_NCNTRL: | |
2749 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n"); break; | |
2750 | case ANYOF_GRAPH: | |
2751 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n"); break; | |
2752 | case ANYOF_NGRAPH: | |
2753 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n"); break; | |
2754 | case ANYOF_DIGIT: | |
2755 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n"); break; | |
2756 | case ANYOF_NDIGIT: | |
2757 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n"); break; | |
2758 | case ANYOF_LOWER: | |
2759 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n"); break; | |
2760 | case ANYOF_NLOWER: | |
2761 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n"); break; | |
2762 | case ANYOF_PRINT: | |
2763 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n"); break; | |
2764 | case ANYOF_NPRINT: | |
2765 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n"); break; | |
2766 | case ANYOF_PUNCT: | |
2767 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n"); break; | |
2768 | case ANYOF_NPUNCT: | |
2769 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n"); break; | |
2770 | case ANYOF_SPACE: | |
2771 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n"); break; | |
2772 | case ANYOF_NSPACE: | |
2773 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n"); break; | |
2774 | case ANYOF_UPPER: | |
2775 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n"); break; | |
2776 | case ANYOF_NUPPER: | |
2777 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n"); break; | |
2778 | case ANYOF_XDIGIT: | |
2779 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n"); break; | |
2780 | case ANYOF_NXDIGIT: | |
2781 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n"); break; | |
2782 | } | |
2783 | } | |
2784 | else | |
2785 | if (range) { | |
a0ed51b3 LW |
2786 | if (lastvalue > value) |
2787 | FAIL("invalid [] range in regexp"); | |
2788 | if (!SIZE_ONLY) | |
cea2e8a9 | 2789 | Perl_sv_catpvf(aTHX_ listsv, "%04x\t%04x\n", lastvalue, value); |
a0ed51b3 LW |
2790 | lastvalue = value; |
2791 | range = 0; | |
2792 | } | |
2793 | else { | |
2794 | lastvalue = value; | |
2795 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && | |
2796 | PL_regcomp_parse[1] != ']') { | |
2797 | PL_regcomp_parse++; | |
2798 | range = 1; | |
2799 | continue; /* do it next time */ | |
2800 | } | |
2801 | if (!SIZE_ONLY) | |
cea2e8a9 | 2802 | Perl_sv_catpvf(aTHX_ listsv, "%04x\n", value); |
a0ed51b3 LW |
2803 | } |
2804 | } | |
a0ed51b3 LW |
2805 | |
2806 | ret = reganode(ANYOFUTF8, 0); | |
2807 | ||
2808 | if (!SIZE_ONLY) { | |
2809 | SV *rv = swash_init("utf8", "", listsv, 1, 0); | |
2810 | SvREFCNT_dec(listsv); | |
2811 | n = add_data(1,"s"); | |
2812 | PL_regcomp_rx->data->data[n] = (void*)rv; | |
2813 | ARG1_SET(ret, flags); | |
2814 | ARG2_SET(ret, n); | |
2815 | } | |
2816 | ||
2817 | return ret; | |
2818 | } | |
2819 | ||
76e3520e | 2820 | STATIC char* |
cea2e8a9 | 2821 | S_nextchar(pTHX) |
a0d0e21e | 2822 | { |
5c0ca799 | 2823 | dTHR; |
3280af22 | 2824 | char* retval = PL_regcomp_parse++; |
a0d0e21e | 2825 | |
4633a7c4 | 2826 | for (;;) { |
3280af22 NIS |
2827 | if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' && |
2828 | PL_regcomp_parse[2] == '#') { | |
2829 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') | |
2830 | PL_regcomp_parse++; | |
2831 | PL_regcomp_parse++; | |
4633a7c4 LW |
2832 | continue; |
2833 | } | |
3280af22 NIS |
2834 | if (PL_regflags & PMf_EXTENDED) { |
2835 | if (isSPACE(*PL_regcomp_parse)) { | |
2836 | PL_regcomp_parse++; | |
748a9306 LW |
2837 | continue; |
2838 | } | |
3280af22 NIS |
2839 | else if (*PL_regcomp_parse == '#') { |
2840 | while (*PL_regcomp_parse && *PL_regcomp_parse != '\n') | |
2841 | PL_regcomp_parse++; | |
2842 | PL_regcomp_parse++; | |
748a9306 LW |
2843 | continue; |
2844 | } | |
748a9306 | 2845 | } |
4633a7c4 | 2846 | return retval; |
a0d0e21e | 2847 | } |
a687059c LW |
2848 | } |
2849 | ||
2850 | /* | |
c277df42 | 2851 | - reg_node - emit a node |
a0d0e21e | 2852 | */ |
76e3520e | 2853 | STATIC regnode * /* Location. */ |
cea2e8a9 | 2854 | S_reg_node(pTHX_ U8 op) |
a687059c | 2855 | { |
5c0ca799 | 2856 | dTHR; |
c277df42 IZ |
2857 | register regnode *ret; |
2858 | register regnode *ptr; | |
a687059c | 2859 | |
3280af22 | 2860 | ret = PL_regcode; |
c277df42 | 2861 | if (SIZE_ONLY) { |
6b88bc9c | 2862 | SIZE_ALIGN(PL_regsize); |
3280af22 | 2863 | PL_regsize += 1; |
a0d0e21e LW |
2864 | return(ret); |
2865 | } | |
a687059c | 2866 | |
c277df42 | 2867 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 2868 | ptr = ret; |
c277df42 | 2869 | FILL_ADVANCE_NODE(ptr, op); |
3280af22 | 2870 | PL_regcode = ptr; |
a687059c | 2871 | |
a0d0e21e | 2872 | return(ret); |
a687059c LW |
2873 | } |
2874 | ||
2875 | /* | |
a0d0e21e LW |
2876 | - reganode - emit a node with an argument |
2877 | */ | |
76e3520e | 2878 | STATIC regnode * /* Location. */ |
cea2e8a9 | 2879 | S_reganode(pTHX_ U8 op, U32 arg) |
fe14fcc3 | 2880 | { |
5c0ca799 | 2881 | dTHR; |
c277df42 IZ |
2882 | register regnode *ret; |
2883 | register regnode *ptr; | |
fe14fcc3 | 2884 | |
3280af22 | 2885 | ret = PL_regcode; |
c277df42 | 2886 | if (SIZE_ONLY) { |
6b88bc9c | 2887 | SIZE_ALIGN(PL_regsize); |
3280af22 | 2888 | PL_regsize += 2; |
a0d0e21e LW |
2889 | return(ret); |
2890 | } | |
fe14fcc3 | 2891 | |
c277df42 | 2892 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 2893 | ptr = ret; |
c277df42 | 2894 | FILL_ADVANCE_NODE_ARG(ptr, op, arg); |
3280af22 | 2895 | PL_regcode = ptr; |
fe14fcc3 | 2896 | |
a0d0e21e | 2897 | return(ret); |
fe14fcc3 LW |
2898 | } |
2899 | ||
2900 | /* | |
a0ed51b3 LW |
2901 | - regc - emit (if appropriate) a Unicode character |
2902 | */ | |
2903 | STATIC void | |
cea2e8a9 | 2904 | S_reguni(pTHX_ UV uv, char* s, I32* lenp) |
a0ed51b3 | 2905 | { |
c485e607 | 2906 | dTHR; |
a0ed51b3 | 2907 | if (SIZE_ONLY) { |
dfe13c55 | 2908 | U8 tmpbuf[10]; |
a0ed51b3 LW |
2909 | *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf; |
2910 | } | |
2911 | else | |
dfe13c55 | 2912 | *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s; |
a0ed51b3 LW |
2913 | |
2914 | } | |
2915 | ||
2916 | /* | |
a0d0e21e LW |
2917 | - regc - emit (if appropriate) a byte of code |
2918 | */ | |
76e3520e | 2919 | STATIC void |
cea2e8a9 | 2920 | S_regc(pTHX_ U8 b, char* s) |
a687059c | 2921 | { |
5c0ca799 | 2922 | dTHR; |
c277df42 IZ |
2923 | if (!SIZE_ONLY) |
2924 | *s = b; | |
a687059c LW |
2925 | } |
2926 | ||
2927 | /* | |
a0d0e21e LW |
2928 | - reginsert - insert an operator in front of already-emitted operand |
2929 | * | |
2930 | * Means relocating the operand. | |
2931 | */ | |
76e3520e | 2932 | STATIC void |
cea2e8a9 | 2933 | S_reginsert(pTHX_ U8 op, regnode *opnd) |
a687059c | 2934 | { |
5c0ca799 | 2935 | dTHR; |
c277df42 IZ |
2936 | register regnode *src; |
2937 | register regnode *dst; | |
2938 | register regnode *place; | |
2939 | register int offset = regarglen[(U8)op]; | |
2940 | ||
22c35a8c | 2941 | /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */ |
c277df42 IZ |
2942 | |
2943 | if (SIZE_ONLY) { | |
3280af22 | 2944 | PL_regsize += NODE_STEP_REGNODE + offset; |
a0d0e21e LW |
2945 | return; |
2946 | } | |
a687059c | 2947 | |
3280af22 NIS |
2948 | src = PL_regcode; |
2949 | PL_regcode += NODE_STEP_REGNODE + offset; | |
2950 | dst = PL_regcode; | |
a0d0e21e | 2951 | while (src > opnd) |
c277df42 | 2952 | StructCopy(--src, --dst, regnode); |
a0d0e21e LW |
2953 | |
2954 | place = opnd; /* Op node, where operand used to be. */ | |
c277df42 IZ |
2955 | src = NEXTOPER(place); |
2956 | FILL_ADVANCE_NODE(place, op); | |
2957 | Zero(src, offset, regnode); | |
a687059c LW |
2958 | } |
2959 | ||
2960 | /* | |
c277df42 | 2961 | - regtail - set the next-pointer at the end of a node chain of p to val. |
a0d0e21e | 2962 | */ |
76e3520e | 2963 | STATIC void |
cea2e8a9 | 2964 | S_regtail(pTHX_ regnode *p, regnode *val) |
a687059c | 2965 | { |
5c0ca799 | 2966 | dTHR; |
c277df42 IZ |
2967 | register regnode *scan; |
2968 | register regnode *temp; | |
a0d0e21e LW |
2969 | register I32 offset; |
2970 | ||
c277df42 | 2971 | if (SIZE_ONLY) |
a0d0e21e LW |
2972 | return; |
2973 | ||
2974 | /* Find last node. */ | |
2975 | scan = p; | |
2976 | for (;;) { | |
2977 | temp = regnext(scan); | |
2978 | if (temp == NULL) | |
2979 | break; | |
2980 | scan = temp; | |
2981 | } | |
a687059c | 2982 | |
c277df42 IZ |
2983 | if (reg_off_by_arg[OP(scan)]) { |
2984 | ARG_SET(scan, val - scan); | |
a0ed51b3 LW |
2985 | } |
2986 | else { | |
c277df42 IZ |
2987 | NEXT_OFF(scan) = val - scan; |
2988 | } | |
a687059c LW |
2989 | } |
2990 | ||
2991 | /* | |
a0d0e21e LW |
2992 | - regoptail - regtail on operand of first argument; nop if operandless |
2993 | */ | |
76e3520e | 2994 | STATIC void |
cea2e8a9 | 2995 | S_regoptail(pTHX_ regnode *p, regnode *val) |
a687059c | 2996 | { |
5c0ca799 | 2997 | dTHR; |
a0d0e21e | 2998 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
c277df42 IZ |
2999 | if (p == NULL || SIZE_ONLY) |
3000 | return; | |
22c35a8c | 3001 | if (PL_regkind[(U8)OP(p)] == BRANCH) { |
c277df42 | 3002 | regtail(NEXTOPER(p), val); |
a0ed51b3 | 3003 | } |
22c35a8c | 3004 | else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) { |
c277df42 | 3005 | regtail(NEXTOPER(NEXTOPER(p)), val); |
a0ed51b3 LW |
3006 | } |
3007 | else | |
a0d0e21e | 3008 | return; |
a687059c LW |
3009 | } |
3010 | ||
3011 | /* | |
3012 | - regcurly - a little FSA that accepts {\d+,?\d*} | |
3013 | */ | |
79072805 | 3014 | STATIC I32 |
cea2e8a9 | 3015 | S_regcurly(pTHX_ register char *s) |
a687059c LW |
3016 | { |
3017 | if (*s++ != '{') | |
3018 | return FALSE; | |
f0fcb552 | 3019 | if (!isDIGIT(*s)) |
a687059c | 3020 | return FALSE; |
f0fcb552 | 3021 | while (isDIGIT(*s)) |
a687059c LW |
3022 | s++; |
3023 | if (*s == ',') | |
3024 | s++; | |
f0fcb552 | 3025 | while (isDIGIT(*s)) |
a687059c LW |
3026 | s++; |
3027 | if (*s != '}') | |
3028 | return FALSE; | |
3029 | return TRUE; | |
3030 | } | |
3031 | ||
a687059c | 3032 | |
76e3520e | 3033 | STATIC regnode * |
cea2e8a9 | 3034 | S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l) |
c277df42 | 3035 | { |
35ff7856 | 3036 | #ifdef DEBUGGING |
f248d071 | 3037 | register U8 op = EXACT; /* Arbitrary non-END op. */ |
c277df42 IZ |
3038 | register regnode *next, *onode; |
3039 | ||
3040 | while (op != END && (!last || node < last)) { | |
3041 | /* While that wasn't END last time... */ | |
3042 | ||
3043 | NODE_ALIGN(node); | |
3044 | op = OP(node); | |
3045 | if (op == CLOSE) | |
3046 | l--; | |
3047 | next = regnext(node); | |
3048 | /* Where, what. */ | |
3049 | if (OP(node) == OPTIMIZED) | |
3050 | goto after_print; | |
3051 | regprop(sv, node); | |
54dc92de | 3052 | PerlIO_printf(Perl_debug_log, "%4d:%*s%s", node - start, |
c277df42 IZ |
3053 | 2*l + 1, "", SvPVX(sv)); |
3054 | if (next == NULL) /* Next ptr. */ | |
3055 | PerlIO_printf(Perl_debug_log, "(0)"); | |
3056 | else | |
3057 | PerlIO_printf(Perl_debug_log, "(%d)", next - start); | |
3058 | (void)PerlIO_putc(Perl_debug_log, '\n'); | |
3059 | after_print: | |
22c35a8c | 3060 | if (PL_regkind[(U8)op] == BRANCHJ) { |
c277df42 IZ |
3061 | register regnode *nnode = (OP(next) == LONGJMP |
3062 | ? regnext(next) | |
3063 | : next); | |
3064 | if (last && nnode > last) | |
3065 | nnode = last; | |
3066 | node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1); | |
a0ed51b3 | 3067 | } |
22c35a8c | 3068 | else if (PL_regkind[(U8)op] == BRANCH) { |
c277df42 | 3069 | node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1); |
a0ed51b3 LW |
3070 | } |
3071 | else if ( op == CURLY) { /* `next' might be very big: optimizer */ | |
c277df42 IZ |
3072 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3073 | NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1); | |
a0ed51b3 | 3074 | } |
22c35a8c | 3075 | else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) { |
c277df42 IZ |
3076 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3077 | next, sv, l + 1); | |
a0ed51b3 LW |
3078 | } |
3079 | else if ( op == PLUS || op == STAR) { | |
c277df42 | 3080 | node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1); |
a0ed51b3 LW |
3081 | } |
3082 | else if (op == ANYOF) { | |
c277df42 IZ |
3083 | node = NEXTOPER(node); |
3084 | node += ANY_SKIP; | |
a0ed51b3 | 3085 | } |
22c35a8c | 3086 | else if (PL_regkind[(U8)op] == EXACT) { |
c277df42 IZ |
3087 | /* Literal string, where present. */ |
3088 | node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode); | |
3089 | node = NEXTOPER(node); | |
a0ed51b3 LW |
3090 | } |
3091 | else { | |
c277df42 IZ |
3092 | node = NEXTOPER(node); |
3093 | node += regarglen[(U8)op]; | |
3094 | } | |
3095 | if (op == CURLYX || op == OPEN) | |
3096 | l++; | |
3097 | else if (op == WHILEM) | |
3098 | l--; | |
3099 | } | |
17c3b450 | 3100 | #endif /* DEBUGGING */ |
c277df42 IZ |
3101 | return node; |
3102 | } | |
3103 | ||
a687059c | 3104 | /* |
fd181c75 | 3105 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c LW |
3106 | */ |
3107 | void | |
864dbfa3 | 3108 | Perl_regdump(pTHX_ regexp *r) |
a687059c | 3109 | { |
35ff7856 | 3110 | #ifdef DEBUGGING |
5c0ca799 | 3111 | dTHR; |
46fc3d4c | 3112 | SV *sv = sv_newmortal(); |
a687059c | 3113 | |
c277df42 | 3114 | (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0); |
a0d0e21e LW |
3115 | |
3116 | /* Header fields of interest. */ | |
c277df42 IZ |
3117 | if (r->anchored_substr) |
3118 | PerlIO_printf(Perl_debug_log, "anchored `%s%s%s'%s at %d ", | |
3280af22 | 3119 | PL_colors[0], |
c277df42 | 3120 | SvPVX(r->anchored_substr), |
3280af22 | 3121 | PL_colors[1], |
c277df42 IZ |
3122 | SvTAIL(r->anchored_substr) ? "$" : "", |
3123 | r->anchored_offset); | |
3124 | if (r->float_substr) | |
3125 | PerlIO_printf(Perl_debug_log, "floating `%s%s%s'%s at %d..%u ", | |
3280af22 | 3126 | PL_colors[0], |
c277df42 | 3127 | SvPVX(r->float_substr), |
3280af22 | 3128 | PL_colors[1], |
c277df42 IZ |
3129 | SvTAIL(r->float_substr) ? "$" : "", |
3130 | r->float_min_offset, r->float_max_offset); | |
3131 | if (r->check_substr) | |
3132 | PerlIO_printf(Perl_debug_log, | |
3133 | r->check_substr == r->float_substr | |
3134 | ? "(checking floating" : "(checking anchored"); | |
3135 | if (r->reganch & ROPT_NOSCAN) | |
3136 | PerlIO_printf(Perl_debug_log, " noscan"); | |
3137 | if (r->reganch & ROPT_CHECK_ALL) | |
3138 | PerlIO_printf(Perl_debug_log, " isall"); | |
3139 | if (r->check_substr) | |
3140 | PerlIO_printf(Perl_debug_log, ") "); | |
3141 | ||
46fc3d4c | 3142 | if (r->regstclass) { |
3143 | regprop(sv, r->regstclass); | |
3144 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv)); | |
3145 | } | |
774d564b | 3146 | if (r->reganch & ROPT_ANCH) { |
3147 | PerlIO_printf(Perl_debug_log, "anchored"); | |
3148 | if (r->reganch & ROPT_ANCH_BOL) | |
3149 | PerlIO_printf(Perl_debug_log, "(BOL)"); | |
c277df42 IZ |
3150 | if (r->reganch & ROPT_ANCH_MBOL) |
3151 | PerlIO_printf(Perl_debug_log, "(MBOL)"); | |
cad2e5aa JH |
3152 | if (r->reganch & ROPT_ANCH_SBOL) |
3153 | PerlIO_printf(Perl_debug_log, "(SBOL)"); | |
774d564b | 3154 | if (r->reganch & ROPT_ANCH_GPOS) |
3155 | PerlIO_printf(Perl_debug_log, "(GPOS)"); | |
3156 | PerlIO_putc(Perl_debug_log, ' '); | |
3157 | } | |
c277df42 IZ |
3158 | if (r->reganch & ROPT_GPOS_SEEN) |
3159 | PerlIO_printf(Perl_debug_log, "GPOS "); | |
a0d0e21e | 3160 | if (r->reganch & ROPT_SKIP) |
760ac839 | 3161 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e | 3162 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 | 3163 | PerlIO_printf(Perl_debug_log, "implicit "); |
760ac839 | 3164 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
ce862d02 IZ |
3165 | if (r->reganch & ROPT_EVAL_SEEN) |
3166 | PerlIO_printf(Perl_debug_log, "with eval "); | |
760ac839 | 3167 | PerlIO_printf(Perl_debug_log, "\n"); |
17c3b450 | 3168 | #endif /* DEBUGGING */ |
a687059c LW |
3169 | } |
3170 | ||
3171 | /* | |
a0d0e21e LW |
3172 | - regprop - printable representation of opcode |
3173 | */ | |
46fc3d4c | 3174 | void |
864dbfa3 | 3175 | Perl_regprop(pTHX_ SV *sv, regnode *o) |
a687059c | 3176 | { |
35ff7856 | 3177 | #ifdef DEBUGGING |
5c0ca799 | 3178 | dTHR; |
9b155405 | 3179 | register int k; |
a0d0e21e | 3180 | |
54dc92de | 3181 | sv_setpvn(sv, "", 0); |
9b155405 | 3182 | if (OP(o) >= reg_num) /* regnode.type is unsigned */ |
a0d0e21e | 3183 | FAIL("corrupted regexp opcode"); |
9b155405 IZ |
3184 | sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */ |
3185 | ||
3186 | k = PL_regkind[(U8)OP(o)]; | |
3187 | ||
3188 | if (k == EXACT) | |
cea2e8a9 | 3189 | Perl_sv_catpvf(aTHX_ sv, " <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]); |
9b155405 IZ |
3190 | else if (k == CURLY) { |
3191 | if (OP(o) == CURLYM || OP(o) == CURLYN) | |
cea2e8a9 GS |
3192 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */ |
3193 | Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o)); | |
a0d0e21e | 3194 | } |
9b155405 | 3195 | else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP ) |
cea2e8a9 | 3196 | Perl_sv_catpvf(aTHX_ sv, "%d", ARG(o)); /* Parenth number */ |
9b155405 | 3197 | else if (k == LOGICAL) |
cea2e8a9 | 3198 | Perl_sv_catpvf(aTHX_ sv, "[%d]", ARG(o)); /* 2: embedded, otherwise 1 */ |
9b155405 | 3199 | else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH)) |
cea2e8a9 | 3200 | Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags); |
17c3b450 | 3201 | #endif /* DEBUGGING */ |
35ff7856 | 3202 | } |
a687059c | 3203 | |
cad2e5aa JH |
3204 | SV * |
3205 | Perl_re_intuit_string(pTHX_ regexp *prog) | |
3206 | { /* Assume that RE_INTUIT is set */ | |
3207 | DEBUG_r( | |
3208 | { STRLEN n_a; | |
3209 | char *s = SvPV(prog->check_substr,n_a); | |
3210 | ||
3211 | if (!PL_colorset) reginitcolors(); | |
3212 | PerlIO_printf(Perl_debug_log, | |
3213 | "%sUsing REx substr:%s `%s%.60s%s%s'\n", | |
3214 | PL_colors[4],PL_colors[5],PL_colors[0], | |
3215 | s, | |
3216 | PL_colors[1], | |
3217 | (strlen(s) > 60 ? "..." : "")); | |
3218 | } ); | |
3219 | ||
3220 | return prog->check_substr; | |
3221 | } | |
3222 | ||
2b69d0c2 | 3223 | void |
864dbfa3 | 3224 | Perl_pregfree(pTHX_ struct regexp *r) |
a687059c | 3225 | { |
5c0ca799 | 3226 | dTHR; |
adac82c7 | 3227 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
cad2e5aa JH |
3228 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
3229 | "%sFreeing REx:%s `%s%.60s%s%s'\n", | |
3230 | PL_colors[4],PL_colors[5],PL_colors[0], | |
3231 | r->precomp, | |
3232 | PL_colors[1], | |
3233 | (strlen(r->precomp) > 60 ? "..." : ""))); | |
3234 | ||
3235 | ||
c277df42 | 3236 | if (!r || (--r->refcnt > 0)) |
a0d0e21e | 3237 | return; |
c277df42 | 3238 | if (r->precomp) |
a0d0e21e | 3239 | Safefree(r->precomp); |
cf93c79d IZ |
3240 | if (RX_MATCH_COPIED(r)) |
3241 | Safefree(r->subbeg); | |
a193d654 GS |
3242 | if (r->substrs) { |
3243 | if (r->anchored_substr) | |
3244 | SvREFCNT_dec(r->anchored_substr); | |
3245 | if (r->float_substr) | |
3246 | SvREFCNT_dec(r->float_substr); | |
2779dcf1 | 3247 | Safefree(r->substrs); |
a193d654 | 3248 | } |
c277df42 IZ |
3249 | if (r->data) { |
3250 | int n = r->data->count; | |
dfad63ad HS |
3251 | AV* new_comppad = NULL; |
3252 | AV* old_comppad; | |
3253 | SV** old_curpad; | |
3254 | ||
c277df42 IZ |
3255 | while (--n >= 0) { |
3256 | switch (r->data->what[n]) { | |
3257 | case 's': | |
3258 | SvREFCNT_dec((SV*)r->data->data[n]); | |
3259 | break; | |
dfad63ad HS |
3260 | case 'p': |
3261 | new_comppad = (AV*)r->data->data[n]; | |
3262 | break; | |
c277df42 | 3263 | case 'o': |
dfad63ad | 3264 | if (new_comppad == NULL) |
cea2e8a9 | 3265 | Perl_croak(aTHX_ "panic: pregfree comppad"); |
dfad63ad HS |
3266 | old_comppad = PL_comppad; |
3267 | old_curpad = PL_curpad; | |
3268 | PL_comppad = new_comppad; | |
3269 | PL_curpad = AvARRAY(new_comppad); | |
c277df42 | 3270 | op_free((OP_4tree*)r->data->data[n]); |
dfad63ad HS |
3271 | PL_comppad = old_comppad; |
3272 | PL_curpad = old_curpad; | |
3273 | SvREFCNT_dec((SV*)new_comppad); | |
3274 | new_comppad = NULL; | |
c277df42 IZ |
3275 | break; |
3276 | case 'n': | |
3277 | break; | |
3278 | default: | |
3279 | FAIL2("panic: regfree data code '%c'", r->data->what[n]); | |
3280 | } | |
3281 | } | |
3282 | Safefree(r->data->what); | |
3283 | Safefree(r->data); | |
a0d0e21e LW |
3284 | } |
3285 | Safefree(r->startp); | |
3286 | Safefree(r->endp); | |
3287 | Safefree(r); | |
a687059c | 3288 | } |
c277df42 IZ |
3289 | |
3290 | /* | |
3291 | - regnext - dig the "next" pointer out of a node | |
3292 | * | |
3293 | * [Note, when REGALIGN is defined there are two places in regmatch() | |
3294 | * that bypass this code for speed.] | |
3295 | */ | |
3296 | regnode * | |
864dbfa3 | 3297 | Perl_regnext(pTHX_ register regnode *p) |
c277df42 | 3298 | { |
5c0ca799 | 3299 | dTHR; |
c277df42 IZ |
3300 | register I32 offset; |
3301 | ||
3280af22 | 3302 | if (p == &PL_regdummy) |
c277df42 IZ |
3303 | return(NULL); |
3304 | ||
3305 | offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p)); | |
3306 | if (offset == 0) | |
3307 | return(NULL); | |
3308 | ||
c277df42 | 3309 | return(p+offset); |
c277df42 IZ |
3310 | } |
3311 | ||
01f988be | 3312 | STATIC void |
cea2e8a9 | 3313 | S_re_croak2(pTHX_ const char* pat1,const char* pat2,...) |
c277df42 IZ |
3314 | { |
3315 | va_list args; | |
3316 | STRLEN l1 = strlen(pat1); | |
3317 | STRLEN l2 = strlen(pat2); | |
3318 | char buf[512]; | |
06bf62c7 | 3319 | SV *msv; |
c277df42 IZ |
3320 | char *message; |
3321 | ||
3322 | if (l1 > 510) | |
3323 | l1 = 510; | |
3324 | if (l1 + l2 > 510) | |
3325 | l2 = 510 - l1; | |
3326 | Copy(pat1, buf, l1 , char); | |
3327 | Copy(pat2, buf + l1, l2 , char); | |
3b818b81 GS |
3328 | buf[l1 + l2] = '\n'; |
3329 | buf[l1 + l2 + 1] = '\0'; | |
8736538c AS |
3330 | #ifdef I_STDARG |
3331 | /* ANSI variant takes additional second argument */ | |
c277df42 | 3332 | va_start(args, pat2); |
8736538c AS |
3333 | #else |
3334 | va_start(args); | |
3335 | #endif | |
06bf62c7 | 3336 | msv = mess(buf, &args); |
c277df42 | 3337 | va_end(args); |
06bf62c7 | 3338 | message = SvPV(msv,l1); |
c277df42 IZ |
3339 | if (l1 > 512) |
3340 | l1 = 512; | |
3341 | Copy(message, buf, l1 , char); | |
3342 | buf[l1] = '\0'; /* Overwrite \n */ | |
cea2e8a9 | 3343 | Perl_croak(aTHX_ "%s", buf); |
c277df42 | 3344 | } |
a0ed51b3 LW |
3345 | |
3346 | /* XXX Here's a total kludge. But we need to re-enter for swash routines. */ | |
3347 | ||
3348 | void | |
864dbfa3 | 3349 | Perl_save_re_context(pTHX) |
c485e607 NIS |
3350 | { |
3351 | dTHR; | |
a0ed51b3 LW |
3352 | SAVEPPTR(PL_bostr); |
3353 | SAVEPPTR(PL_regprecomp); /* uncompiled string. */ | |
3354 | SAVEI32(PL_regnpar); /* () count. */ | |
3355 | SAVEI32(PL_regsize); /* Code size. */ | |
3356 | SAVEI16(PL_regflags); /* are we folding, multilining? */ | |
3357 | SAVEPPTR(PL_reginput); /* String-input pointer. */ | |
3358 | SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */ | |
3359 | SAVEPPTR(PL_regeol); /* End of input, for $ check. */ | |
3360 | SAVESPTR(PL_regstartp); /* Pointer to startp array. */ | |
3361 | SAVESPTR(PL_regendp); /* Ditto for endp. */ | |
3362 | SAVESPTR(PL_reglastparen); /* Similarly for lastparen. */ | |
3363 | SAVEPPTR(PL_regtill); /* How far we are required to go. */ | |
3364 | SAVEI32(PL_regprev); /* char before regbol, \n if none */ | |
3365 | SAVESPTR(PL_reg_start_tmp); /* from regexec.c */ | |
3366 | PL_reg_start_tmp = 0; | |
3367 | SAVEFREEPV(PL_reg_start_tmp); | |
3368 | SAVEI32(PL_reg_start_tmpl); /* from regexec.c */ | |
3369 | PL_reg_start_tmpl = 0; | |
3370 | SAVESPTR(PL_regdata); | |
3371 | SAVEI32(PL_reg_flags); /* from regexec.c */ | |
3372 | SAVEI32(PL_reg_eval_set); /* from regexec.c */ | |
3373 | SAVEI32(PL_regnarrate); /* from regexec.c */ | |
3374 | SAVESPTR(PL_regprogram); /* from regexec.c */ | |
3375 | SAVEINT(PL_regindent); /* from regexec.c */ | |
3376 | SAVESPTR(PL_regcc); /* from regexec.c */ | |
3377 | SAVESPTR(PL_curcop); | |
3378 | SAVESPTR(PL_regcomp_rx); /* from regcomp.c */ | |
3379 | SAVEI32(PL_regseen); /* from regcomp.c */ | |
3380 | SAVEI32(PL_regsawback); /* Did we see \1, ...? */ | |
3381 | SAVEI32(PL_regnaughty); /* How bad is this pattern? */ | |
3382 | SAVESPTR(PL_regcode); /* Code-emit pointer; ®dummy = don't */ | |
3383 | SAVEPPTR(PL_regxend); /* End of input for compile */ | |
3384 | SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */ | |
54b6e2fa IZ |
3385 | SAVESPTR(PL_reg_call_cc); /* from regexec.c */ |
3386 | SAVESPTR(PL_reg_re); /* from regexec.c */ | |
3387 | SAVEPPTR(PL_reg_ganch); /* from regexec.c */ | |
3388 | SAVESPTR(PL_reg_sv); /* from regexec.c */ | |
3389 | SAVESPTR(PL_reg_magic); /* from regexec.c */ | |
3390 | SAVEI32(PL_reg_oldpos); /* from regexec.c */ | |
3391 | SAVESPTR(PL_reg_oldcurpm); /* from regexec.c */ | |
3392 | SAVESPTR(PL_reg_curpm); /* from regexec.c */ | |
3393 | #ifdef DEBUGGING | |
3394 | SAVEPPTR(PL_reg_starttry); /* from regexec.c */ | |
3395 | #endif | |
a0ed51b3 | 3396 | } |
51371543 GS |
3397 | |
3398 | #ifdef PERL_OBJECT | |
3399 | #define NO_XSLOCKS | |
3400 | #include "XSUB.h" | |
3401 | #undef this | |
3402 | #define this pPerl | |
3403 | #endif | |
3404 | ||
3405 | static void | |
3406 | clear_re(pTHXo_ void *r) | |
3407 | { | |
3408 | ReREFCNT_dec((regexp *)r); | |
3409 | } | |
3410 |