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 | 134 | |
2c2d71f5 JH |
135 | /* Length of a variant. */ |
136 | ||
137 | typedef struct scan_data_t { | |
138 | I32 len_min; | |
139 | I32 len_delta; | |
140 | I32 pos_min; | |
141 | I32 pos_delta; | |
142 | SV *last_found; | |
143 | I32 last_end; /* min value, <0 unless valid. */ | |
144 | I32 last_start_min; | |
145 | I32 last_start_max; | |
146 | SV **longest; /* Either &l_fixed, or &l_float. */ | |
147 | SV *longest_fixed; | |
148 | I32 offset_fixed; | |
149 | SV *longest_float; | |
150 | I32 offset_float_min; | |
151 | I32 offset_float_max; | |
152 | I32 flags; | |
153 | I32 whilem_c; | |
154 | } scan_data_t; | |
155 | ||
a687059c | 156 | /* |
e50aee73 | 157 | * Forward declarations for pregcomp()'s friends. |
a687059c | 158 | */ |
a0d0e21e | 159 | |
cd488c12 | 160 | static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
2c2d71f5 | 161 | 0, 0, 0, 0 }; |
c277df42 IZ |
162 | |
163 | #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL) | |
164 | #define SF_BEFORE_SEOL 0x1 | |
165 | #define SF_BEFORE_MEOL 0x2 | |
166 | #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL) | |
167 | #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL) | |
168 | ||
09b7f37c CB |
169 | #ifdef NO_UNARY_PLUS |
170 | # define SF_FIX_SHIFT_EOL (0+2) | |
171 | # define SF_FL_SHIFT_EOL (0+4) | |
172 | #else | |
173 | # define SF_FIX_SHIFT_EOL (+2) | |
174 | # define SF_FL_SHIFT_EOL (+4) | |
175 | #endif | |
c277df42 IZ |
176 | |
177 | #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL) | |
178 | #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL) | |
179 | ||
180 | #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL) | |
181 | #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */ | |
182 | #define SF_IS_INF 0x40 | |
183 | #define SF_HAS_PAR 0x80 | |
184 | #define SF_IN_PAR 0x100 | |
185 | #define SF_HAS_EVAL 0x200 | |
4bfe0158 | 186 | #define SCF_DO_SUBSTR 0x400 |
c277df42 | 187 | |
a0ed51b3 LW |
188 | #define RF_utf8 8 |
189 | #define UTF (PL_reg_flags & RF_utf8) | |
190 | #define LOC (PL_regflags & PMf_LOCALE) | |
191 | #define FOLD (PL_regflags & PMf_FOLD) | |
192 | ||
b8c5462f JH |
193 | #define OOB_CHAR8 1234 |
194 | #define OOB_UTF8 123456 | |
93733859 | 195 | #define OOB_NAMEDCLASS -1 |
b8c5462f | 196 | |
a0ed51b3 LW |
197 | #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) |
198 | #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) | |
199 | ||
cd439c50 IZ |
200 | /* Allow for side effects in s */ |
201 | #define REGC(c,s) STMT_START { if (!SIZE_ONLY) *(s) = (c); else (s);} STMT_END | |
202 | ||
51371543 | 203 | static void clear_re(pTHXo_ void *r); |
4327152a IZ |
204 | |
205 | STATIC void | |
cea2e8a9 | 206 | S_scan_commit(pTHX_ scan_data_t *data) |
c277df42 | 207 | { |
c485e607 | 208 | dTHR; |
a0ed51b3 LW |
209 | STRLEN l = CHR_SVLEN(data->last_found); |
210 | STRLEN old_l = CHR_SVLEN(*data->longest); | |
c277df42 IZ |
211 | |
212 | if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) { | |
213 | sv_setsv(*data->longest, data->last_found); | |
214 | if (*data->longest == data->longest_fixed) { | |
215 | data->offset_fixed = l ? data->last_start_min : data->pos_min; | |
216 | if (data->flags & SF_BEFORE_EOL) | |
217 | data->flags | |
218 | |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL); | |
219 | else | |
220 | data->flags &= ~SF_FIX_BEFORE_EOL; | |
a0ed51b3 LW |
221 | } |
222 | else { | |
c277df42 IZ |
223 | data->offset_float_min = l ? data->last_start_min : data->pos_min; |
224 | data->offset_float_max = (l | |
225 | ? data->last_start_max | |
226 | : data->pos_min + data->pos_delta); | |
227 | if (data->flags & SF_BEFORE_EOL) | |
228 | data->flags | |
229 | |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL); | |
230 | else | |
231 | data->flags &= ~SF_FL_BEFORE_EOL; | |
232 | } | |
233 | } | |
234 | SvCUR_set(data->last_found, 0); | |
235 | data->last_end = -1; | |
236 | data->flags &= ~SF_BEFORE_EOL; | |
237 | } | |
238 | ||
c277df42 IZ |
239 | /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set |
240 | to the position after last scanned or to NULL. */ | |
241 | ||
76e3520e | 242 | STATIC I32 |
cea2e8a9 | 243 | S_study_chunk(pTHX_ regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags) |
c277df42 IZ |
244 | /* scanp: Start here (read-write). */ |
245 | /* deltap: Write maxlen-minlen here. */ | |
246 | /* last: Stop before this one. */ | |
247 | { | |
5c0ca799 | 248 | dTHR; |
c277df42 IZ |
249 | I32 min = 0, pars = 0, code; |
250 | regnode *scan = *scanp, *next; | |
251 | I32 delta = 0; | |
252 | int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF); | |
aca2d497 | 253 | int is_inf_internal = 0; /* The studied chunk is infinite */ |
c277df42 IZ |
254 | I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0; |
255 | scan_data_t data_fake; | |
256 | ||
257 | while (scan && OP(scan) != END && scan < last) { | |
258 | /* Peephole optimizer: */ | |
259 | ||
22c35a8c | 260 | if (PL_regkind[(U8)OP(scan)] == EXACT) { |
c277df42 IZ |
261 | regnode *n = regnext(scan); |
262 | U32 stringok = 1; | |
263 | #ifdef DEBUGGING | |
264 | regnode *stop = scan; | |
265 | #endif | |
266 | ||
cd439c50 | 267 | next = scan + NODE_SZ_STR(scan); |
c277df42 IZ |
268 | /* Skip NOTHING, merge EXACT*. */ |
269 | while (n && | |
22c35a8c | 270 | ( PL_regkind[(U8)OP(n)] == NOTHING || |
c277df42 IZ |
271 | (stringok && (OP(n) == OP(scan)))) |
272 | && NEXT_OFF(n) | |
273 | && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) { | |
274 | if (OP(n) == TAIL || n > next) | |
275 | stringok = 0; | |
22c35a8c | 276 | if (PL_regkind[(U8)OP(n)] == NOTHING) { |
c277df42 IZ |
277 | NEXT_OFF(scan) += NEXT_OFF(n); |
278 | next = n + NODE_STEP_REGNODE; | |
279 | #ifdef DEBUGGING | |
280 | if (stringok) | |
281 | stop = n; | |
282 | #endif | |
283 | n = regnext(n); | |
a0ed51b3 LW |
284 | } |
285 | else { | |
cd439c50 | 286 | int oldl = STR_LEN(scan); |
c277df42 IZ |
287 | regnode *nnext = regnext(n); |
288 | ||
cd439c50 | 289 | if (oldl + STR_LEN(n) > U8_MAX) |
c277df42 IZ |
290 | break; |
291 | NEXT_OFF(scan) += NEXT_OFF(n); | |
cd439c50 IZ |
292 | STR_LEN(scan) += STR_LEN(n); |
293 | next = n + NODE_SZ_STR(n); | |
c277df42 | 294 | /* Now we can overwrite *n : */ |
cd439c50 IZ |
295 | Move(STRING(n), STRING(scan) + oldl, |
296 | STR_LEN(n), char); | |
c277df42 IZ |
297 | #ifdef DEBUGGING |
298 | if (stringok) | |
299 | stop = next - 1; | |
300 | #endif | |
301 | n = nnext; | |
302 | } | |
303 | } | |
304 | #ifdef DEBUGGING | |
305 | /* Allow dumping */ | |
cd439c50 | 306 | n = scan + NODE_SZ_STR(scan); |
c277df42 | 307 | while (n <= stop) { |
ca04da08 GS |
308 | /* Purify reports a benign UMR here sometimes, because we |
309 | * don't initialize the OP() slot of a node when that node | |
310 | * is occupied by just the trailing null of the string in | |
311 | * an EXACT node */ | |
22c35a8c | 312 | if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) { |
c277df42 IZ |
313 | OP(n) = OPTIMIZED; |
314 | NEXT_OFF(n) = 0; | |
315 | } | |
316 | n++; | |
317 | } | |
318 | #endif | |
319 | ||
320 | } | |
321 | if (OP(scan) != CURLYX) { | |
048cfca1 GS |
322 | int max = (reg_off_by_arg[OP(scan)] |
323 | ? I32_MAX | |
324 | /* I32 may be smaller than U16 on CRAYs! */ | |
325 | : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX)); | |
c277df42 IZ |
326 | int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan)); |
327 | int noff; | |
328 | regnode *n = scan; | |
329 | ||
330 | /* Skip NOTHING and LONGJMP. */ | |
331 | while ((n = regnext(n)) | |
22c35a8c | 332 | && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n))) |
c277df42 IZ |
333 | || ((OP(n) == LONGJMP) && (noff = ARG(n)))) |
334 | && off + noff < max) | |
335 | off += noff; | |
336 | if (reg_off_by_arg[OP(scan)]) | |
337 | ARG(scan) = off; | |
338 | else | |
339 | NEXT_OFF(scan) = off; | |
340 | } | |
341 | if (OP(scan) == BRANCH || OP(scan) == BRANCHJ | |
342 | || OP(scan) == IFTHEN || OP(scan) == SUSPEND) { | |
343 | next = regnext(scan); | |
344 | code = OP(scan); | |
345 | ||
346 | if (OP(next) == code || code == IFTHEN || code == SUSPEND) { | |
347 | I32 max1 = 0, min1 = I32_MAX, num = 0; | |
348 | ||
349 | if (flags & SCF_DO_SUBSTR) | |
350 | scan_commit(data); | |
351 | while (OP(scan) == code) { | |
352 | I32 deltanext, minnext; | |
353 | ||
354 | num++; | |
355 | data_fake.flags = 0; | |
2c2d71f5 JH |
356 | if (data) |
357 | data_fake.whilem_c = data->whilem_c; | |
c277df42 IZ |
358 | next = regnext(scan); |
359 | scan = NEXTOPER(scan); | |
360 | if (code != BRANCH) | |
361 | scan = NEXTOPER(scan); | |
362 | /* We suppose the run is continuous, last=next...*/ | |
363 | minnext = study_chunk(&scan, &deltanext, next, | |
364 | &data_fake, 0); | |
365 | if (min1 > minnext) | |
366 | min1 = minnext; | |
367 | if (max1 < minnext + deltanext) | |
368 | max1 = minnext + deltanext; | |
369 | if (deltanext == I32_MAX) | |
aca2d497 | 370 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
371 | scan = next; |
372 | if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
373 | pars++; | |
405ff068 | 374 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 | 375 | data->flags |= SF_HAS_EVAL; |
2c2d71f5 JH |
376 | if (data) |
377 | data->whilem_c = data_fake.whilem_c; | |
c277df42 IZ |
378 | if (code == SUSPEND) |
379 | break; | |
380 | } | |
381 | if (code == IFTHEN && num < 2) /* Empty ELSE branch */ | |
382 | min1 = 0; | |
383 | if (flags & SCF_DO_SUBSTR) { | |
384 | data->pos_min += min1; | |
385 | data->pos_delta += max1 - min1; | |
386 | if (max1 != min1 || is_inf) | |
387 | data->longest = &(data->longest_float); | |
388 | } | |
389 | min += min1; | |
390 | delta += max1 - min1; | |
a0ed51b3 LW |
391 | } |
392 | else if (code == BRANCHJ) /* single branch is optimized. */ | |
c277df42 IZ |
393 | scan = NEXTOPER(NEXTOPER(scan)); |
394 | else /* single branch is optimized. */ | |
395 | scan = NEXTOPER(scan); | |
396 | continue; | |
a0ed51b3 LW |
397 | } |
398 | else if (OP(scan) == EXACT) { | |
cd439c50 | 399 | I32 l = STR_LEN(scan); |
a0ed51b3 | 400 | if (UTF) { |
cd439c50 | 401 | unsigned char *s = (unsigned char *)STRING(scan); |
a0ed51b3 LW |
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 (flags & SCF_DO_SUBSTR) { /* Update longest substr. */ |
c277df42 IZ |
412 | /* The code below prefers earlier match for fixed |
413 | offset, later match for variable offset. */ | |
414 | if (data->last_end == -1) { /* Update the start info. */ | |
415 | data->last_start_min = data->pos_min; | |
416 | data->last_start_max = is_inf | |
417 | ? I32_MAX : data->pos_min + data->pos_delta; | |
418 | } | |
cd439c50 | 419 | sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan)); |
c277df42 IZ |
420 | data->last_end = data->pos_min + l; |
421 | data->pos_min += l; /* As in the first entry. */ | |
422 | data->flags &= ~SF_BEFORE_EOL; | |
423 | } | |
a0ed51b3 | 424 | } |
22c35a8c | 425 | else if (PL_regkind[(U8)OP(scan)] == EXACT) { |
cd439c50 | 426 | I32 l = STR_LEN(scan); |
c277df42 IZ |
427 | if (flags & SCF_DO_SUBSTR) |
428 | scan_commit(data); | |
a0ed51b3 | 429 | if (UTF) { |
cd439c50 | 430 | unsigned char *s = (unsigned char *)STRING(scan); |
a0ed51b3 LW |
431 | unsigned char *e = s + l; |
432 | I32 newl = 0; | |
433 | while (s < e) { | |
434 | newl++; | |
435 | s += UTF8SKIP(s); | |
436 | } | |
437 | l = newl; | |
438 | } | |
439 | min += l; | |
c277df42 | 440 | if (data && (flags & SCF_DO_SUBSTR)) |
a0ed51b3 LW |
441 | data->pos_min += l; |
442 | } | |
4d61ec05 | 443 | else if (strchr((char*)PL_varies,OP(scan))) { |
c277df42 IZ |
444 | I32 mincount, maxcount, minnext, deltanext, pos_before, fl; |
445 | regnode *oscan = scan; | |
446 | ||
22c35a8c | 447 | switch (PL_regkind[(U8)OP(scan)]) { |
c277df42 IZ |
448 | case WHILEM: |
449 | scan = NEXTOPER(scan); | |
450 | goto finish; | |
451 | case PLUS: | |
452 | if (flags & SCF_DO_SUBSTR) { | |
453 | next = NEXTOPER(scan); | |
454 | if (OP(next) == EXACT) { | |
455 | mincount = 1; | |
456 | maxcount = REG_INFTY; | |
457 | next = regnext(scan); | |
458 | scan = NEXTOPER(scan); | |
459 | goto do_curly; | |
460 | } | |
461 | } | |
462 | if (flags & SCF_DO_SUBSTR) | |
463 | data->pos_min++; | |
464 | min++; | |
465 | /* Fall through. */ | |
466 | case STAR: | |
aca2d497 | 467 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
468 | scan = regnext(scan); |
469 | if (flags & SCF_DO_SUBSTR) { | |
470 | scan_commit(data); | |
471 | data->longest = &(data->longest_float); | |
472 | } | |
473 | goto optimize_curly_tail; | |
474 | case CURLY: | |
475 | mincount = ARG1(scan); | |
476 | maxcount = ARG2(scan); | |
477 | next = regnext(scan); | |
478 | scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
479 | do_curly: | |
480 | if (flags & SCF_DO_SUBSTR) { | |
481 | if (mincount == 0) scan_commit(data); | |
482 | pos_before = data->pos_min; | |
483 | } | |
484 | if (data) { | |
485 | fl = data->flags; | |
486 | data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL); | |
487 | if (is_inf) | |
488 | data->flags |= SF_IS_INF; | |
489 | } | |
490 | /* This will finish on WHILEM, setting scan, or on NULL: */ | |
491 | minnext = study_chunk(&scan, &deltanext, last, data, | |
492 | mincount == 0 | |
493 | ? (flags & ~SCF_DO_SUBSTR) : flags); | |
494 | if (!scan) /* It was not CURLYX, but CURLY. */ | |
495 | scan = next; | |
599cee73 | 496 | if (ckWARN(WARN_UNSAFE) && (minnext + deltanext == 0) |
821b33a5 | 497 | && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) |
17feb5d5 | 498 | && maxcount <= REG_INFTY/3) /* Complement check for big count */ |
c5be433b GS |
499 | Perl_warner(aTHX_ WARN_UNSAFE, |
500 | "Strange *+?{} on zero-length expression"); | |
c277df42 | 501 | min += minnext * mincount; |
aca2d497 IZ |
502 | is_inf_internal |= (maxcount == REG_INFTY |
503 | && (minnext + deltanext) > 0 | |
504 | || deltanext == I32_MAX); | |
505 | is_inf |= is_inf_internal; | |
c277df42 IZ |
506 | delta += (minnext + deltanext) * maxcount - minnext * mincount; |
507 | ||
508 | /* Try powerful optimization CURLYX => CURLYN. */ | |
c277df42 IZ |
509 | if ( OP(oscan) == CURLYX && data |
510 | && data->flags & SF_IN_PAR | |
511 | && !(data->flags & SF_HAS_EVAL) | |
512 | && !deltanext && minnext == 1 ) { | |
513 | /* Try to optimize to CURLYN. */ | |
514 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; | |
515 | regnode *nxt1 = nxt, *nxt2; | |
516 | ||
517 | /* Skip open. */ | |
518 | nxt = regnext(nxt); | |
4d61ec05 | 519 | if (!strchr((char*)PL_simple,OP(nxt)) |
22c35a8c | 520 | && !(PL_regkind[(U8)OP(nxt)] == EXACT |
cd439c50 | 521 | && STR_LEN(nxt) == 1)) |
c277df42 IZ |
522 | goto nogo; |
523 | nxt2 = nxt; | |
524 | nxt = regnext(nxt); | |
525 | if (OP(nxt) != CLOSE) | |
526 | goto nogo; | |
527 | /* Now we know that nxt2 is the only contents: */ | |
528 | oscan->flags = ARG(nxt); | |
529 | OP(oscan) = CURLYN; | |
530 | OP(nxt1) = NOTHING; /* was OPEN. */ | |
531 | #ifdef DEBUGGING | |
532 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ | |
533 | NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */ | |
534 | NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */ | |
535 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ | |
536 | OP(nxt + 1) = OPTIMIZED; /* was count. */ | |
537 | NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */ | |
538 | #endif | |
539 | } | |
c277df42 IZ |
540 | nogo: |
541 | ||
542 | /* Try optimization CURLYX => CURLYM. */ | |
543 | if ( OP(oscan) == CURLYX && data | |
c277df42 | 544 | && !(data->flags & SF_HAS_PAR) |
c277df42 IZ |
545 | && !(data->flags & SF_HAS_EVAL) |
546 | && !deltanext ) { | |
547 | /* XXXX How to optimize if data == 0? */ | |
548 | /* Optimize to a simpler form. */ | |
549 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */ | |
550 | regnode *nxt2; | |
551 | ||
552 | OP(oscan) = CURLYM; | |
553 | while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/ | |
554 | && (OP(nxt2) != WHILEM)) | |
555 | nxt = nxt2; | |
556 | OP(nxt2) = SUCCEED; /* Whas WHILEM */ | |
c277df42 IZ |
557 | /* Need to optimize away parenths. */ |
558 | if (data->flags & SF_IN_PAR) { | |
559 | /* Set the parenth number. */ | |
560 | regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/ | |
561 | ||
562 | if (OP(nxt) != CLOSE) | |
563 | FAIL("panic opt close"); | |
564 | oscan->flags = ARG(nxt); | |
565 | OP(nxt1) = OPTIMIZED; /* was OPEN. */ | |
566 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ | |
567 | #ifdef DEBUGGING | |
568 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ | |
569 | OP(nxt + 1) = OPTIMIZED; /* was count. */ | |
570 | NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */ | |
571 | NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */ | |
572 | #endif | |
573 | #if 0 | |
574 | while ( nxt1 && (OP(nxt1) != WHILEM)) { | |
575 | regnode *nnxt = regnext(nxt1); | |
576 | ||
577 | if (nnxt == nxt) { | |
578 | if (reg_off_by_arg[OP(nxt1)]) | |
579 | ARG_SET(nxt1, nxt2 - nxt1); | |
580 | else if (nxt2 - nxt1 < U16_MAX) | |
581 | NEXT_OFF(nxt1) = nxt2 - nxt1; | |
582 | else | |
583 | OP(nxt) = NOTHING; /* Cannot beautify */ | |
584 | } | |
585 | nxt1 = nnxt; | |
586 | } | |
587 | #endif | |
588 | /* Optimize again: */ | |
589 | study_chunk(&nxt1, &deltanext, nxt, NULL, 0); | |
a0ed51b3 LW |
590 | } |
591 | else | |
c277df42 | 592 | oscan->flags = 0; |
c277df42 | 593 | } |
2c2d71f5 JH |
594 | else if (OP(oscan) == CURLYX && data && ++data->whilem_c < 16) { |
595 | /* This stays as CURLYX, and can put the count/of pair. */ | |
596 | /* Find WHILEM (as in regexec.c) */ | |
597 | regnode *nxt = oscan + NEXT_OFF(oscan); | |
598 | ||
599 | if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */ | |
600 | nxt += ARG(nxt); | |
601 | PREVOPER(nxt)->flags = data->whilem_c | |
602 | | (PL_reg_whilem_seen << 4); /* On WHILEM */ | |
603 | } | |
c277df42 IZ |
604 | if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) |
605 | pars++; | |
606 | if (flags & SCF_DO_SUBSTR) { | |
607 | SV *last_str = Nullsv; | |
608 | int counted = mincount != 0; | |
609 | ||
610 | if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */ | |
611 | I32 b = pos_before >= data->last_start_min | |
612 | ? pos_before : data->last_start_min; | |
613 | STRLEN l; | |
614 | char *s = SvPV(data->last_found, l); | |
a0ed51b3 LW |
615 | I32 old = b - data->last_start_min; |
616 | ||
617 | if (UTF) | |
618 | old = utf8_hop((U8*)s, old) - (U8*)s; | |
c277df42 | 619 | |
a0ed51b3 | 620 | l -= old; |
c277df42 | 621 | /* Get the added string: */ |
79cb57f6 | 622 | last_str = newSVpvn(s + old, l); |
c277df42 IZ |
623 | if (deltanext == 0 && pos_before == b) { |
624 | /* What was added is a constant string */ | |
625 | if (mincount > 1) { | |
626 | SvGROW(last_str, (mincount * l) + 1); | |
627 | repeatcpy(SvPVX(last_str) + l, | |
628 | SvPVX(last_str), l, mincount - 1); | |
629 | SvCUR(last_str) *= mincount; | |
630 | /* Add additional parts. */ | |
631 | SvCUR_set(data->last_found, | |
632 | SvCUR(data->last_found) - l); | |
633 | sv_catsv(data->last_found, last_str); | |
634 | data->last_end += l * (mincount - 1); | |
635 | } | |
636 | } | |
637 | } | |
638 | /* It is counted once already... */ | |
639 | data->pos_min += minnext * (mincount - counted); | |
640 | data->pos_delta += - counted * deltanext + | |
641 | (minnext + deltanext) * maxcount - minnext * mincount; | |
642 | if (mincount != maxcount) { | |
643 | scan_commit(data); | |
644 | if (mincount && last_str) { | |
645 | sv_setsv(data->last_found, last_str); | |
646 | data->last_end = data->pos_min; | |
647 | data->last_start_min = | |
a0ed51b3 | 648 | data->pos_min - CHR_SVLEN(last_str); |
c277df42 IZ |
649 | data->last_start_max = is_inf |
650 | ? I32_MAX | |
651 | : data->pos_min + data->pos_delta | |
a0ed51b3 | 652 | - CHR_SVLEN(last_str); |
c277df42 IZ |
653 | } |
654 | data->longest = &(data->longest_float); | |
655 | } | |
aca2d497 | 656 | SvREFCNT_dec(last_str); |
c277df42 | 657 | } |
405ff068 | 658 | if (data && (fl & SF_HAS_EVAL)) |
c277df42 IZ |
659 | data->flags |= SF_HAS_EVAL; |
660 | optimize_curly_tail: | |
c277df42 | 661 | if (OP(oscan) != CURLYX) { |
22c35a8c | 662 | while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING |
c277df42 IZ |
663 | && NEXT_OFF(next)) |
664 | NEXT_OFF(oscan) += NEXT_OFF(next); | |
665 | } | |
c277df42 IZ |
666 | continue; |
667 | default: /* REF only? */ | |
668 | if (flags & SCF_DO_SUBSTR) { | |
669 | scan_commit(data); | |
670 | data->longest = &(data->longest_float); | |
671 | } | |
aca2d497 | 672 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
673 | break; |
674 | } | |
a0ed51b3 | 675 | } |
4d61ec05 | 676 | else if (strchr((char*)PL_simple,OP(scan)) || PL_regkind[(U8)OP(scan)] == ANYUTF8) { |
c277df42 IZ |
677 | if (flags & SCF_DO_SUBSTR) { |
678 | scan_commit(data); | |
679 | data->pos_min++; | |
680 | } | |
681 | min++; | |
a0ed51b3 | 682 | } |
22c35a8c | 683 | else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) { |
c277df42 IZ |
684 | data->flags |= (OP(scan) == MEOL |
685 | ? SF_BEFORE_MEOL | |
686 | : SF_BEFORE_SEOL); | |
a0ed51b3 | 687 | } |
22c35a8c | 688 | else if (PL_regkind[(U8)OP(scan)] == BRANCHJ |
c277df42 IZ |
689 | && (scan->flags || data) |
690 | && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) { | |
691 | I32 deltanext, minnext; | |
692 | regnode *nscan; | |
693 | ||
694 | data_fake.flags = 0; | |
2c2d71f5 JH |
695 | if (data) |
696 | data_fake.whilem_c = data->whilem_c; | |
c277df42 IZ |
697 | next = regnext(scan); |
698 | nscan = NEXTOPER(NEXTOPER(scan)); | |
699 | minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0); | |
700 | if (scan->flags) { | |
701 | if (deltanext) { | |
702 | FAIL("variable length lookbehind not implemented"); | |
a0ed51b3 LW |
703 | } |
704 | else if (minnext > U8_MAX) { | |
57def98f | 705 | FAIL2("lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX); |
c277df42 IZ |
706 | } |
707 | scan->flags = minnext; | |
708 | } | |
709 | if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
710 | pars++; | |
405ff068 | 711 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 | 712 | data->flags |= SF_HAS_EVAL; |
2c2d71f5 JH |
713 | if (data) |
714 | data->whilem_c = data_fake.whilem_c; | |
a0ed51b3 LW |
715 | } |
716 | else if (OP(scan) == OPEN) { | |
c277df42 | 717 | pars++; |
a0ed51b3 LW |
718 | } |
719 | else if (OP(scan) == CLOSE && ARG(scan) == is_par) { | |
c277df42 IZ |
720 | next = regnext(scan); |
721 | ||
722 | if ( next && (OP(next) != WHILEM) && next < last) | |
c277df42 | 723 | is_par = 0; /* Disable optimization */ |
a0ed51b3 LW |
724 | } |
725 | else if (OP(scan) == EVAL) { | |
c277df42 IZ |
726 | if (data) |
727 | data->flags |= SF_HAS_EVAL; | |
728 | } | |
0f5d15d6 IZ |
729 | else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */ |
730 | if (flags & SCF_DO_SUBSTR) { | |
731 | scan_commit(data); | |
732 | data->longest = &(data->longest_float); | |
733 | } | |
734 | is_inf = is_inf_internal = 1; | |
735 | } | |
c277df42 IZ |
736 | /* Else: zero-length, ignore. */ |
737 | scan = regnext(scan); | |
738 | } | |
739 | ||
740 | finish: | |
741 | *scanp = scan; | |
aca2d497 | 742 | *deltap = is_inf_internal ? I32_MAX : delta; |
c277df42 IZ |
743 | if (flags & SCF_DO_SUBSTR && is_inf) |
744 | data->pos_delta = I32_MAX - data->pos_min; | |
745 | if (is_par > U8_MAX) | |
746 | is_par = 0; | |
747 | if (is_par && pars==1 && data) { | |
748 | data->flags |= SF_IN_PAR; | |
749 | data->flags &= ~SF_HAS_PAR; | |
a0ed51b3 LW |
750 | } |
751 | else if (pars && data) { | |
c277df42 IZ |
752 | data->flags |= SF_HAS_PAR; |
753 | data->flags &= ~SF_IN_PAR; | |
754 | } | |
755 | return min; | |
756 | } | |
757 | ||
76e3520e | 758 | STATIC I32 |
cea2e8a9 | 759 | S_add_data(pTHX_ I32 n, char *s) |
c277df42 | 760 | { |
5c0ca799 | 761 | dTHR; |
3280af22 NIS |
762 | if (PL_regcomp_rx->data) { |
763 | Renewc(PL_regcomp_rx->data, | |
764 | sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1), | |
c277df42 | 765 | char, struct reg_data); |
3280af22 NIS |
766 | Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8); |
767 | PL_regcomp_rx->data->count += n; | |
a0ed51b3 LW |
768 | } |
769 | else { | |
3280af22 | 770 | Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1), |
c277df42 | 771 | char, struct reg_data); |
3280af22 NIS |
772 | New(1208, PL_regcomp_rx->data->what, n, U8); |
773 | PL_regcomp_rx->data->count = n; | |
c277df42 | 774 | } |
3280af22 NIS |
775 | Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8); |
776 | return PL_regcomp_rx->data->count - n; | |
c277df42 IZ |
777 | } |
778 | ||
d88dccdf | 779 | void |
864dbfa3 | 780 | Perl_reginitcolors(pTHX) |
d88dccdf IZ |
781 | { |
782 | dTHR; | |
783 | int i = 0; | |
784 | char *s = PerlEnv_getenv("PERL_RE_COLORS"); | |
785 | ||
786 | if (s) { | |
787 | PL_colors[0] = s = savepv(s); | |
788 | while (++i < 6) { | |
789 | s = strchr(s, '\t'); | |
790 | if (s) { | |
791 | *s = '\0'; | |
792 | PL_colors[i] = ++s; | |
793 | } | |
794 | else | |
c712d376 | 795 | PL_colors[i] = s = ""; |
d88dccdf IZ |
796 | } |
797 | } else { | |
798 | while (i < 6) | |
799 | PL_colors[i++] = ""; | |
800 | } | |
801 | PL_colorset = 1; | |
802 | } | |
803 | ||
a687059c | 804 | /* |
e50aee73 | 805 | - pregcomp - compile a regular expression into internal code |
a687059c LW |
806 | * |
807 | * We can't allocate space until we know how big the compiled form will be, | |
808 | * but we can't compile it (and thus know how big it is) until we've got a | |
809 | * place to put the code. So we cheat: we compile it twice, once with code | |
810 | * generation turned off and size counting turned on, and once "for real". | |
811 | * This also means that we don't allocate space until we are sure that the | |
812 | * thing really will compile successfully, and we never have to move the | |
813 | * code and thus invalidate pointers into it. (Note that it has to be in | |
814 | * one piece because free() must be able to free it all.) [NB: not true in perl] | |
815 | * | |
816 | * Beware that the optimization-preparation code in here knows about some | |
817 | * of the structure of the compiled regexp. [I'll say.] | |
818 | */ | |
819 | regexp * | |
864dbfa3 | 820 | Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm) |
a687059c | 821 | { |
5c0ca799 | 822 | dTHR; |
a0d0e21e | 823 | register regexp *r; |
c277df42 IZ |
824 | regnode *scan; |
825 | SV **longest; | |
826 | SV *longest_fixed; | |
827 | SV *longest_float; | |
828 | regnode *first; | |
a0d0e21e | 829 | I32 flags; |
a0d0e21e LW |
830 | I32 minlen = 0; |
831 | I32 sawplus = 0; | |
832 | I32 sawopen = 0; | |
2c2d71f5 | 833 | scan_data_t data; |
a0d0e21e LW |
834 | |
835 | if (exp == NULL) | |
c277df42 | 836 | FAIL("NULL regexp argument"); |
a0d0e21e | 837 | |
e24b16f9 | 838 | if (PL_curcop == &PL_compiling ? (PL_hints & HINT_UTF8) : IN_UTF8) |
a0ed51b3 LW |
839 | PL_reg_flags |= RF_utf8; |
840 | else | |
841 | PL_reg_flags = 0; | |
842 | ||
3280af22 | 843 | PL_regprecomp = savepvn(exp, xend - exp); |
35ef4773 | 844 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
2c2d71f5 | 845 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n", |
d88dccdf IZ |
846 | PL_colors[4],PL_colors[5],PL_colors[0], |
847 | xend - exp, PL_regprecomp, PL_colors[1])); | |
3280af22 NIS |
848 | PL_regflags = pm->op_pmflags; |
849 | PL_regsawback = 0; | |
bbce6d69 | 850 | |
3280af22 NIS |
851 | PL_regseen = 0; |
852 | PL_seen_zerolen = *exp == '^' ? -1 : 0; | |
853 | PL_seen_evals = 0; | |
854 | PL_extralen = 0; | |
c277df42 | 855 | |
bbce6d69 | 856 | /* First pass: determine size, legality. */ |
3280af22 NIS |
857 | PL_regcomp_parse = exp; |
858 | PL_regxend = xend; | |
859 | PL_regnaughty = 0; | |
860 | PL_regnpar = 1; | |
861 | PL_regsize = 0L; | |
862 | PL_regcode = &PL_regdummy; | |
2c2d71f5 | 863 | PL_reg_whilem_seen = 0; |
cd439c50 | 864 | REGC((U8)REG_MAGIC, (char*)PL_regcode); |
a0d0e21e | 865 | if (reg(0, &flags) == NULL) { |
3280af22 NIS |
866 | Safefree(PL_regprecomp); |
867 | PL_regprecomp = Nullch; | |
a0d0e21e LW |
868 | return(NULL); |
869 | } | |
3280af22 | 870 | DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", PL_regsize)); |
c277df42 | 871 | |
c277df42 IZ |
872 | /* Small enough for pointer-storage convention? |
873 | If extralen==0, this means that we will not need long jumps. */ | |
3280af22 NIS |
874 | if (PL_regsize >= 0x10000L && PL_extralen) |
875 | PL_regsize += PL_extralen; | |
c277df42 | 876 | else |
3280af22 | 877 | PL_extralen = 0; |
2c2d71f5 JH |
878 | if (PL_reg_whilem_seen > 15) |
879 | PL_reg_whilem_seen = 15; | |
a0d0e21e | 880 | |
bbce6d69 | 881 | /* Allocate space and initialize. */ |
3280af22 | 882 | Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), |
c277df42 | 883 | char, regexp); |
a0d0e21e LW |
884 | if (r == NULL) |
885 | FAIL("regexp out of space"); | |
c277df42 | 886 | r->refcnt = 1; |
bbce6d69 | 887 | r->prelen = xend - exp; |
3280af22 | 888 | r->precomp = PL_regprecomp; |
cf93c79d IZ |
889 | r->subbeg = NULL; |
890 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; | |
4327152a IZ |
891 | r->nparens = PL_regnpar - 1; /* set early to validate backrefs */ |
892 | ||
893 | r->substrs = 0; /* Useful during FAIL. */ | |
894 | r->startp = 0; /* Useful during FAIL. */ | |
895 | r->endp = 0; /* Useful during FAIL. */ | |
896 | ||
3280af22 | 897 | PL_regcomp_rx = r; |
bbce6d69 | 898 | |
899 | /* Second pass: emit code. */ | |
3280af22 NIS |
900 | PL_regcomp_parse = exp; |
901 | PL_regxend = xend; | |
902 | PL_regnaughty = 0; | |
903 | PL_regnpar = 1; | |
904 | PL_regcode = r->program; | |
2cd61cdb | 905 | /* Store the count of eval-groups for security checks: */ |
3280af22 | 906 | PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals); |
cd439c50 | 907 | REGC((U8)REG_MAGIC, (char*) PL_regcode++); |
c277df42 | 908 | r->data = 0; |
a0d0e21e LW |
909 | if (reg(0, &flags) == NULL) |
910 | return(NULL); | |
911 | ||
912 | /* Dig out information for optimizations. */ | |
cf93c79d | 913 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */ |
3280af22 | 914 | pm->op_pmflags = PL_regflags; |
a0ed51b3 LW |
915 | if (UTF) |
916 | r->reganch |= ROPT_UTF8; | |
c277df42 | 917 | r->regstclass = NULL; |
a0ed51b3 LW |
918 | if (PL_regnaughty >= 10) /* Probably an expensive pattern. */ |
919 | r->reganch |= ROPT_NAUGHTY; | |
c277df42 | 920 | scan = r->program + 1; /* First BRANCH. */ |
2779dcf1 IZ |
921 | |
922 | /* XXXX To minimize changes to RE engine we always allocate | |
923 | 3-units-long substrs field. */ | |
924 | Newz(1004, r->substrs, 1, struct reg_substr_data); | |
925 | ||
2c2d71f5 | 926 | StructCopy(&zero_scan_data, &data, scan_data_t); |
c277df42 | 927 | if (OP(scan) != BRANCH) { /* Only one top-level choice. */ |
c277df42 | 928 | I32 fake; |
c5254dd6 | 929 | STRLEN longest_float_length, longest_fixed_length; |
a0d0e21e LW |
930 | |
931 | first = scan; | |
c277df42 | 932 | /* Skip introductions and multiplicators >= 1. */ |
a0d0e21e LW |
933 | while ((OP(first) == OPEN && (sawopen = 1)) || |
934 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || | |
935 | (OP(first) == PLUS) || | |
936 | (OP(first) == MINMOD) || | |
22c35a8c | 937 | (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
a0d0e21e LW |
938 | if (OP(first) == PLUS) |
939 | sawplus = 1; | |
940 | else | |
941 | first += regarglen[(U8)OP(first)]; | |
942 | first = NEXTOPER(first); | |
a687059c LW |
943 | } |
944 | ||
a0d0e21e LW |
945 | /* Starting-point info. */ |
946 | again: | |
c277df42 | 947 | if (OP(first) == EXACT); /* Empty, get anchored substr later. */ |
4d61ec05 | 948 | else if (strchr((char*)PL_simple+4,OP(first))) |
a0d0e21e | 949 | r->regstclass = first; |
22c35a8c GS |
950 | else if (PL_regkind[(U8)OP(first)] == BOUND || |
951 | PL_regkind[(U8)OP(first)] == NBOUND) | |
a0d0e21e | 952 | r->regstclass = first; |
22c35a8c | 953 | else if (PL_regkind[(U8)OP(first)] == BOL) { |
cad2e5aa JH |
954 | r->reganch |= (OP(first) == MBOL |
955 | ? ROPT_ANCH_MBOL | |
956 | : (OP(first) == SBOL | |
957 | ? ROPT_ANCH_SBOL | |
958 | : ROPT_ANCH_BOL)); | |
a0d0e21e | 959 | first = NEXTOPER(first); |
774d564b | 960 | goto again; |
961 | } | |
962 | else if (OP(first) == GPOS) { | |
963 | r->reganch |= ROPT_ANCH_GPOS; | |
964 | first = NEXTOPER(first); | |
965 | goto again; | |
a0d0e21e LW |
966 | } |
967 | else if ((OP(first) == STAR && | |
22c35a8c | 968 | PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) && |
a0d0e21e LW |
969 | !(r->reganch & ROPT_ANCH) ) |
970 | { | |
971 | /* turn .* into ^.* with an implied $*=1 */ | |
cad2e5aa JH |
972 | int type = OP(NEXTOPER(first)); |
973 | ||
974 | if (type == REG_ANY || type == ANYUTF8) | |
975 | type = ROPT_ANCH_MBOL; | |
976 | else | |
977 | type = ROPT_ANCH_SBOL; | |
978 | ||
979 | r->reganch |= type | ROPT_IMPLICIT; | |
a0d0e21e | 980 | first = NEXTOPER(first); |
774d564b | 981 | goto again; |
a0d0e21e | 982 | } |
cad2e5aa JH |
983 | if (sawplus && (!sawopen || !PL_regsawback) |
984 | && !(PL_regseen & REG_SEEN_EVAL)) /* May examine pos and $& */ | |
985 | /* x+ must match at the 1st pos of run of x's */ | |
986 | r->reganch |= ROPT_SKIP; | |
a0d0e21e | 987 | |
c277df42 IZ |
988 | /* Scan is after the zeroth branch, first is atomic matcher. */ |
989 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n", | |
990 | first - scan + 1)); | |
a0d0e21e LW |
991 | /* |
992 | * If there's something expensive in the r.e., find the | |
993 | * longest literal string that must appear and make it the | |
994 | * regmust. Resolve ties in favor of later strings, since | |
995 | * the regstart check works with the beginning of the r.e. | |
996 | * and avoiding duplication strengthens checking. Not a | |
997 | * strong reason, but sufficient in the absence of others. | |
998 | * [Now we resolve ties in favor of the earlier string if | |
c277df42 | 999 | * it happens that c_offset_min has been invalidated, since the |
a0d0e21e LW |
1000 | * earlier string may buy us something the later one won't.] |
1001 | */ | |
a0d0e21e | 1002 | minlen = 0; |
a687059c | 1003 | |
79cb57f6 GS |
1004 | data.longest_fixed = newSVpvn("",0); |
1005 | data.longest_float = newSVpvn("",0); | |
1006 | data.last_found = newSVpvn("",0); | |
c277df42 IZ |
1007 | data.longest = &(data.longest_fixed); |
1008 | first = scan; | |
1009 | ||
3280af22 | 1010 | minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */ |
c277df42 | 1011 | &data, SCF_DO_SUBSTR); |
3280af22 | 1012 | if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed) |
c277df42 | 1013 | && data.last_start_min == 0 && data.last_end > 0 |
3280af22 NIS |
1014 | && !PL_seen_zerolen |
1015 | && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS))) | |
c277df42 IZ |
1016 | r->reganch |= ROPT_CHECK_ALL; |
1017 | scan_commit(&data); | |
1018 | SvREFCNT_dec(data.last_found); | |
1019 | ||
a0ed51b3 | 1020 | longest_float_length = CHR_SVLEN(data.longest_float); |
c5254dd6 | 1021 | if (longest_float_length |
c277df42 IZ |
1022 | || (data.flags & SF_FL_BEFORE_EOL |
1023 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
3280af22 | 1024 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1025 | int t; |
1026 | ||
a0ed51b3 | 1027 | if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */ |
aca2d497 IZ |
1028 | && data.offset_fixed == data.offset_float_min |
1029 | && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)) | |
1030 | goto remove_float; /* As in (a)+. */ | |
1031 | ||
c277df42 IZ |
1032 | r->float_substr = data.longest_float; |
1033 | r->float_min_offset = data.offset_float_min; | |
1034 | r->float_max_offset = data.offset_float_max; | |
cf93c79d IZ |
1035 | t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1036 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
1037 | || (PL_regflags & PMf_MULTILINE))); | |
1038 | fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1039 | } |
1040 | else { | |
aca2d497 | 1041 | remove_float: |
c277df42 IZ |
1042 | r->float_substr = Nullsv; |
1043 | SvREFCNT_dec(data.longest_float); | |
c5254dd6 | 1044 | longest_float_length = 0; |
a0d0e21e | 1045 | } |
c277df42 | 1046 | |
a0ed51b3 | 1047 | longest_fixed_length = CHR_SVLEN(data.longest_fixed); |
c5254dd6 | 1048 | if (longest_fixed_length |
c277df42 IZ |
1049 | || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
1050 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
3280af22 | 1051 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1052 | int t; |
1053 | ||
c277df42 IZ |
1054 | r->anchored_substr = data.longest_fixed; |
1055 | r->anchored_offset = data.offset_fixed; | |
cf93c79d IZ |
1056 | t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1057 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
1058 | || (PL_regflags & PMf_MULTILINE))); | |
1059 | fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1060 | } |
1061 | else { | |
c277df42 IZ |
1062 | r->anchored_substr = Nullsv; |
1063 | SvREFCNT_dec(data.longest_fixed); | |
c5254dd6 | 1064 | longest_fixed_length = 0; |
a0d0e21e | 1065 | } |
c277df42 IZ |
1066 | |
1067 | /* A temporary algorithm prefers floated substr to fixed one to dig more info. */ | |
c5254dd6 | 1068 | if (longest_fixed_length > longest_float_length) { |
c277df42 IZ |
1069 | r->check_substr = r->anchored_substr; |
1070 | r->check_offset_min = r->check_offset_max = r->anchored_offset; | |
1071 | if (r->reganch & ROPT_ANCH_SINGLE) | |
1072 | r->reganch |= ROPT_NOSCAN; | |
a0ed51b3 LW |
1073 | } |
1074 | else { | |
c277df42 IZ |
1075 | r->check_substr = r->float_substr; |
1076 | r->check_offset_min = data.offset_float_min; | |
1077 | r->check_offset_max = data.offset_float_max; | |
a0d0e21e | 1078 | } |
30382c73 IZ |
1079 | /* XXXX Currently intuiting is not compatible with ANCH_GPOS. |
1080 | This should be changed ASAP! */ | |
1081 | if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) { | |
cad2e5aa JH |
1082 | r->reganch |= RE_USE_INTUIT; |
1083 | if (SvTAIL(r->check_substr)) | |
1084 | r->reganch |= RE_INTUIT_TAIL; | |
1085 | } | |
a0ed51b3 LW |
1086 | } |
1087 | else { | |
c277df42 IZ |
1088 | /* Several toplevels. Best we can is to set minlen. */ |
1089 | I32 fake; | |
1090 | ||
1091 | DEBUG_r(PerlIO_printf(Perl_debug_log, "\n")); | |
1092 | scan = r->program + 1; | |
2c2d71f5 | 1093 | minlen = study_chunk(&scan, &fake, scan + PL_regsize, &data, 0); |
c277df42 | 1094 | r->check_substr = r->anchored_substr = r->float_substr = Nullsv; |
a0d0e21e LW |
1095 | } |
1096 | ||
a0d0e21e | 1097 | r->minlen = minlen; |
3280af22 | 1098 | if (PL_regseen & REG_SEEN_GPOS) |
c277df42 | 1099 | r->reganch |= ROPT_GPOS_SEEN; |
3280af22 | 1100 | if (PL_regseen & REG_SEEN_LOOKBEHIND) |
c277df42 | 1101 | r->reganch |= ROPT_LOOKBEHIND_SEEN; |
3280af22 | 1102 | if (PL_regseen & REG_SEEN_EVAL) |
ce862d02 | 1103 | r->reganch |= ROPT_EVAL_SEEN; |
cf93c79d IZ |
1104 | Newz(1002, r->startp, PL_regnpar, I32); |
1105 | Newz(1002, r->endp, PL_regnpar, I32); | |
a0d0e21e LW |
1106 | DEBUG_r(regdump(r)); |
1107 | return(r); | |
a687059c LW |
1108 | } |
1109 | ||
1110 | /* | |
1111 | - reg - regular expression, i.e. main body or parenthesized thing | |
1112 | * | |
1113 | * Caller must absorb opening parenthesis. | |
1114 | * | |
1115 | * Combining parenthesis handling with the base level of regular expression | |
1116 | * is a trifle forced, but the need to tie the tails of the branches to what | |
1117 | * follows makes it hard to avoid. | |
1118 | */ | |
76e3520e | 1119 | STATIC regnode * |
cea2e8a9 | 1120 | S_reg(pTHX_ I32 paren, I32 *flagp) |
c277df42 | 1121 | /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */ |
a687059c | 1122 | { |
5c0ca799 | 1123 | dTHR; |
c277df42 IZ |
1124 | register regnode *ret; /* Will be the head of the group. */ |
1125 | register regnode *br; | |
1126 | register regnode *lastbr; | |
1127 | register regnode *ender = 0; | |
a0d0e21e | 1128 | register I32 parno = 0; |
3280af22 | 1129 | I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0; |
c277df42 | 1130 | char c; |
a0d0e21e | 1131 | |
821b33a5 | 1132 | *flagp = 0; /* Tentatively. */ |
a0d0e21e LW |
1133 | |
1134 | /* Make an OPEN node, if parenthesized. */ | |
1135 | if (paren) { | |
3280af22 | 1136 | if (*PL_regcomp_parse == '?') { |
ca9dfc88 IZ |
1137 | U16 posflags = 0, negflags = 0; |
1138 | U16 *flagsp = &posflags; | |
0f5d15d6 | 1139 | int logical = 0; |
ca9dfc88 | 1140 | |
3280af22 NIS |
1141 | PL_regcomp_parse++; |
1142 | paren = *PL_regcomp_parse++; | |
c277df42 | 1143 | ret = NULL; /* For look-ahead/behind. */ |
a0d0e21e | 1144 | switch (paren) { |
c277df42 | 1145 | case '<': |
3280af22 NIS |
1146 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
1147 | if (*PL_regcomp_parse == '!') | |
c277df42 | 1148 | paren = ','; |
3280af22 | 1149 | if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!') |
c277df42 | 1150 | goto unknown; |
3280af22 | 1151 | PL_regcomp_parse++; |
a0d0e21e LW |
1152 | case '=': |
1153 | case '!': | |
3280af22 | 1154 | PL_seen_zerolen++; |
c277df42 IZ |
1155 | case ':': |
1156 | case '>': | |
a0d0e21e LW |
1157 | break; |
1158 | case '$': | |
1159 | case '@': | |
c277df42 | 1160 | FAIL2("Sequence (?%c...) not implemented", (int)paren); |
a0d0e21e LW |
1161 | break; |
1162 | case '#': | |
3280af22 NIS |
1163 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
1164 | PL_regcomp_parse++; | |
1165 | if (*PL_regcomp_parse != ')') | |
c277df42 | 1166 | FAIL("Sequence (?#... not terminated"); |
a0d0e21e LW |
1167 | nextchar(); |
1168 | *flagp = TRYAGAIN; | |
1169 | return NULL; | |
0f5d15d6 IZ |
1170 | case 'p': |
1171 | logical = 1; | |
1172 | paren = *PL_regcomp_parse++; | |
1173 | /* FALL THROUGH */ | |
c277df42 IZ |
1174 | case '{': |
1175 | { | |
1176 | dTHR; | |
1177 | I32 count = 1, n = 0; | |
1178 | char c; | |
3280af22 | 1179 | char *s = PL_regcomp_parse; |
c277df42 IZ |
1180 | SV *sv; |
1181 | OP_4tree *sop, *rop; | |
1182 | ||
3280af22 NIS |
1183 | PL_seen_zerolen++; |
1184 | PL_regseen |= REG_SEEN_EVAL; | |
1185 | while (count && (c = *PL_regcomp_parse)) { | |
1186 | if (c == '\\' && PL_regcomp_parse[1]) | |
1187 | PL_regcomp_parse++; | |
c277df42 IZ |
1188 | else if (c == '{') |
1189 | count++; | |
1190 | else if (c == '}') | |
1191 | count--; | |
3280af22 | 1192 | PL_regcomp_parse++; |
c277df42 | 1193 | } |
3280af22 | 1194 | if (*PL_regcomp_parse != ')') |
c277df42 IZ |
1195 | FAIL("Sequence (?{...}) not terminated or not {}-balanced"); |
1196 | if (!SIZE_ONLY) { | |
1197 | AV *av; | |
1198 | ||
3280af22 | 1199 | if (PL_regcomp_parse - 1 - s) |
79cb57f6 | 1200 | sv = newSVpvn(s, PL_regcomp_parse - 1 - s); |
c277df42 | 1201 | else |
79cb57f6 | 1202 | sv = newSVpvn("", 0); |
c277df42 IZ |
1203 | |
1204 | rop = sv_compile_2op(sv, &sop, "re", &av); | |
1205 | ||
dfad63ad | 1206 | n = add_data(3, "nop"); |
3280af22 | 1207 | PL_regcomp_rx->data->data[n] = (void*)rop; |
dfad63ad HS |
1208 | PL_regcomp_rx->data->data[n+1] = (void*)sop; |
1209 | PL_regcomp_rx->data->data[n+2] = (void*)av; | |
c277df42 | 1210 | SvREFCNT_dec(sv); |
a0ed51b3 | 1211 | } |
e24b16f9 GS |
1212 | else { /* First pass */ |
1213 | if (PL_reginterp_cnt < ++PL_seen_evals | |
1214 | && PL_curcop != &PL_compiling) | |
2cd61cdb IZ |
1215 | /* No compiled RE interpolated, has runtime |
1216 | components ===> unsafe. */ | |
1217 | FAIL("Eval-group not allowed at runtime, use re 'eval'"); | |
3280af22 | 1218 | if (PL_tainted) |
cc6b7395 | 1219 | FAIL("Eval-group in insecure regular expression"); |
c277df42 IZ |
1220 | } |
1221 | ||
1222 | nextchar(); | |
0f5d15d6 IZ |
1223 | if (logical) { |
1224 | ret = reg_node(LOGICAL); | |
1225 | if (!SIZE_ONLY) | |
1226 | ret->flags = 2; | |
1227 | regtail(ret, reganode(EVAL, n)); | |
1228 | return ret; | |
1229 | } | |
c277df42 IZ |
1230 | return reganode(EVAL, n); |
1231 | } | |
1232 | case '(': | |
1233 | { | |
3280af22 NIS |
1234 | if (PL_regcomp_parse[0] == '?') { |
1235 | if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!' | |
1236 | || PL_regcomp_parse[1] == '<' | |
1237 | || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */ | |
c277df42 IZ |
1238 | I32 flag; |
1239 | ||
1240 | ret = reg_node(LOGICAL); | |
0f5d15d6 IZ |
1241 | if (!SIZE_ONLY) |
1242 | ret->flags = 1; | |
c277df42 IZ |
1243 | regtail(ret, reg(1, &flag)); |
1244 | goto insert_if; | |
1245 | } | |
a0ed51b3 LW |
1246 | } |
1247 | else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) { | |
3280af22 | 1248 | parno = atoi(PL_regcomp_parse++); |
c277df42 | 1249 | |
3280af22 NIS |
1250 | while (isDIGIT(*PL_regcomp_parse)) |
1251 | PL_regcomp_parse++; | |
c277df42 IZ |
1252 | ret = reganode(GROUPP, parno); |
1253 | if ((c = *nextchar()) != ')') | |
1254 | FAIL2("Switch (?(number%c not recognized", c); | |
1255 | insert_if: | |
1256 | regtail(ret, reganode(IFTHEN, 0)); | |
1257 | br = regbranch(&flags, 1); | |
1258 | if (br == NULL) | |
1259 | br = reganode(LONGJMP, 0); | |
1260 | else | |
1261 | regtail(br, reganode(LONGJMP, 0)); | |
1262 | c = *nextchar(); | |
d1b80229 IZ |
1263 | if (flags&HASWIDTH) |
1264 | *flagp |= HASWIDTH; | |
c277df42 IZ |
1265 | if (c == '|') { |
1266 | lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */ | |
1267 | regbranch(&flags, 1); | |
1268 | regtail(ret, lastbr); | |
d1b80229 IZ |
1269 | if (flags&HASWIDTH) |
1270 | *flagp |= HASWIDTH; | |
c277df42 | 1271 | c = *nextchar(); |
a0ed51b3 LW |
1272 | } |
1273 | else | |
c277df42 IZ |
1274 | lastbr = NULL; |
1275 | if (c != ')') | |
1276 | FAIL("Switch (?(condition)... contains too many branches"); | |
1277 | ender = reg_node(TAIL); | |
1278 | regtail(br, ender); | |
1279 | if (lastbr) { | |
1280 | regtail(lastbr, ender); | |
1281 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); | |
a0ed51b3 LW |
1282 | } |
1283 | else | |
c277df42 IZ |
1284 | regtail(ret, ender); |
1285 | return ret; | |
a0ed51b3 LW |
1286 | } |
1287 | else { | |
3280af22 | 1288 | FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse); |
c277df42 IZ |
1289 | } |
1290 | } | |
1b1626e4 | 1291 | case 0: |
c277df42 | 1292 | FAIL("Sequence (? incomplete"); |
1b1626e4 | 1293 | break; |
a0d0e21e | 1294 | default: |
3280af22 | 1295 | --PL_regcomp_parse; |
ca9dfc88 | 1296 | parse_flags: |
3280af22 NIS |
1297 | while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) { |
1298 | if (*PL_regcomp_parse != 'o') | |
1299 | pmflag(flagsp, *PL_regcomp_parse); | |
1300 | ++PL_regcomp_parse; | |
ca9dfc88 | 1301 | } |
3280af22 | 1302 | if (*PL_regcomp_parse == '-') { |
ca9dfc88 | 1303 | flagsp = &negflags; |
3280af22 | 1304 | ++PL_regcomp_parse; |
ca9dfc88 | 1305 | goto parse_flags; |
48c036b1 | 1306 | } |
3280af22 NIS |
1307 | PL_regflags |= posflags; |
1308 | PL_regflags &= ~negflags; | |
1309 | if (*PL_regcomp_parse == ':') { | |
1310 | PL_regcomp_parse++; | |
ca9dfc88 IZ |
1311 | paren = ':'; |
1312 | break; | |
1313 | } | |
c277df42 | 1314 | unknown: |
3280af22 NIS |
1315 | if (*PL_regcomp_parse != ')') |
1316 | FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse); | |
a0d0e21e LW |
1317 | nextchar(); |
1318 | *flagp = TRYAGAIN; | |
1319 | return NULL; | |
1320 | } | |
1321 | } | |
1322 | else { | |
3280af22 NIS |
1323 | parno = PL_regnpar; |
1324 | PL_regnpar++; | |
a0d0e21e | 1325 | ret = reganode(OPEN, parno); |
c277df42 | 1326 | open = 1; |
a0d0e21e | 1327 | } |
a0ed51b3 LW |
1328 | } |
1329 | else | |
a0d0e21e LW |
1330 | ret = NULL; |
1331 | ||
1332 | /* Pick up the branches, linking them together. */ | |
c277df42 | 1333 | br = regbranch(&flags, 1); |
a0d0e21e LW |
1334 | if (br == NULL) |
1335 | return(NULL); | |
3280af22 NIS |
1336 | if (*PL_regcomp_parse == '|') { |
1337 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 | 1338 | reginsert(BRANCHJ, br); |
a0ed51b3 LW |
1339 | } |
1340 | else | |
c277df42 IZ |
1341 | reginsert(BRANCH, br); |
1342 | have_branch = 1; | |
1343 | if (SIZE_ONLY) | |
3280af22 | 1344 | PL_extralen += 1; /* For BRANCHJ-BRANCH. */ |
a0ed51b3 LW |
1345 | } |
1346 | else if (paren == ':') { | |
c277df42 IZ |
1347 | *flagp |= flags&SIMPLE; |
1348 | } | |
1349 | if (open) { /* Starts with OPEN. */ | |
1350 | regtail(ret, br); /* OPEN -> first. */ | |
a0ed51b3 LW |
1351 | } |
1352 | else if (paren != '?') /* Not Conditional */ | |
a0d0e21e | 1353 | ret = br; |
821b33a5 IZ |
1354 | if (flags&HASWIDTH) |
1355 | *flagp |= HASWIDTH; | |
a0d0e21e | 1356 | *flagp |= flags&SPSTART; |
c277df42 | 1357 | lastbr = br; |
3280af22 NIS |
1358 | while (*PL_regcomp_parse == '|') { |
1359 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 IZ |
1360 | ender = reganode(LONGJMP,0); |
1361 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ | |
1362 | } | |
1363 | if (SIZE_ONLY) | |
3280af22 | 1364 | PL_extralen += 2; /* Account for LONGJMP. */ |
a0d0e21e | 1365 | nextchar(); |
c277df42 | 1366 | br = regbranch(&flags, 0); |
a687059c | 1367 | if (br == NULL) |
a0d0e21e | 1368 | return(NULL); |
c277df42 IZ |
1369 | regtail(lastbr, br); /* BRANCH -> BRANCH. */ |
1370 | lastbr = br; | |
821b33a5 IZ |
1371 | if (flags&HASWIDTH) |
1372 | *flagp |= HASWIDTH; | |
a687059c | 1373 | *flagp |= flags&SPSTART; |
a0d0e21e LW |
1374 | } |
1375 | ||
c277df42 IZ |
1376 | if (have_branch || paren != ':') { |
1377 | /* Make a closing node, and hook it on the end. */ | |
1378 | switch (paren) { | |
1379 | case ':': | |
1380 | ender = reg_node(TAIL); | |
1381 | break; | |
1382 | case 1: | |
1383 | ender = reganode(CLOSE, parno); | |
1384 | break; | |
1385 | case '<': | |
c277df42 IZ |
1386 | case ',': |
1387 | case '=': | |
1388 | case '!': | |
c277df42 | 1389 | *flagp &= ~HASWIDTH; |
821b33a5 IZ |
1390 | /* FALL THROUGH */ |
1391 | case '>': | |
1392 | ender = reg_node(SUCCEED); | |
c277df42 IZ |
1393 | break; |
1394 | case 0: | |
1395 | ender = reg_node(END); | |
1396 | break; | |
1397 | } | |
1398 | regtail(lastbr, ender); | |
a0d0e21e | 1399 | |
c277df42 IZ |
1400 | if (have_branch) { |
1401 | /* Hook the tails of the branches to the closing node. */ | |
1402 | for (br = ret; br != NULL; br = regnext(br)) { | |
1403 | regoptail(br, ender); | |
1404 | } | |
1405 | } | |
a0d0e21e | 1406 | } |
c277df42 IZ |
1407 | |
1408 | { | |
1409 | char *p; | |
1410 | static char parens[] = "=!<,>"; | |
1411 | ||
1412 | if (paren && (p = strchr(parens, paren))) { | |
1413 | int node = ((p - parens) % 2) ? UNLESSM : IFMATCH; | |
1414 | int flag = (p - parens) > 1; | |
1415 | ||
1416 | if (paren == '>') | |
1417 | node = SUSPEND, flag = 0; | |
1418 | reginsert(node,ret); | |
c277df42 | 1419 | ret->flags = flag; |
c277df42 IZ |
1420 | regtail(ret, reg_node(TAIL)); |
1421 | } | |
a0d0e21e LW |
1422 | } |
1423 | ||
1424 | /* Check for proper termination. */ | |
ce3e6498 HS |
1425 | if (paren) { |
1426 | PL_regflags = oregflags; | |
1427 | if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') { | |
1428 | FAIL("unmatched () in regexp"); | |
1429 | } | |
a0ed51b3 LW |
1430 | } |
1431 | else if (!paren && PL_regcomp_parse < PL_regxend) { | |
3280af22 | 1432 | if (*PL_regcomp_parse == ')') { |
a0d0e21e | 1433 | FAIL("unmatched () in regexp"); |
a0ed51b3 LW |
1434 | } |
1435 | else | |
a0d0e21e LW |
1436 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
1437 | /* NOTREACHED */ | |
1438 | } | |
a687059c | 1439 | |
a0d0e21e | 1440 | return(ret); |
a687059c LW |
1441 | } |
1442 | ||
1443 | /* | |
1444 | - regbranch - one alternative of an | operator | |
1445 | * | |
1446 | * Implements the concatenation operator. | |
1447 | */ | |
76e3520e | 1448 | STATIC regnode * |
cea2e8a9 | 1449 | S_regbranch(pTHX_ I32 *flagp, I32 first) |
a687059c | 1450 | { |
5c0ca799 | 1451 | dTHR; |
c277df42 IZ |
1452 | register regnode *ret; |
1453 | register regnode *chain = NULL; | |
1454 | register regnode *latest; | |
1455 | I32 flags = 0, c = 0; | |
a0d0e21e | 1456 | |
c277df42 IZ |
1457 | if (first) |
1458 | ret = NULL; | |
1459 | else { | |
3280af22 | 1460 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
1461 | ret = reganode(BRANCHJ,0); |
1462 | else | |
1463 | ret = reg_node(BRANCH); | |
1464 | } | |
1465 | ||
1466 | if (!first && SIZE_ONLY) | |
3280af22 | 1467 | PL_extralen += 1; /* BRANCHJ */ |
c277df42 IZ |
1468 | |
1469 | *flagp = WORST; /* Tentatively. */ | |
a0d0e21e | 1470 | |
3280af22 | 1471 | PL_regcomp_parse--; |
a0d0e21e | 1472 | nextchar(); |
3280af22 | 1473 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') { |
a0d0e21e LW |
1474 | flags &= ~TRYAGAIN; |
1475 | latest = regpiece(&flags); | |
1476 | if (latest == NULL) { | |
1477 | if (flags & TRYAGAIN) | |
1478 | continue; | |
1479 | return(NULL); | |
a0ed51b3 LW |
1480 | } |
1481 | else if (ret == NULL) | |
c277df42 | 1482 | ret = latest; |
a0d0e21e | 1483 | *flagp |= flags&HASWIDTH; |
c277df42 | 1484 | if (chain == NULL) /* First piece. */ |
a0d0e21e LW |
1485 | *flagp |= flags&SPSTART; |
1486 | else { | |
3280af22 | 1487 | PL_regnaughty++; |
a0d0e21e | 1488 | regtail(chain, latest); |
a687059c | 1489 | } |
a0d0e21e | 1490 | chain = latest; |
c277df42 IZ |
1491 | c++; |
1492 | } | |
1493 | if (chain == NULL) { /* Loop ran zero times. */ | |
1494 | chain = reg_node(NOTHING); | |
1495 | if (ret == NULL) | |
1496 | ret = chain; | |
1497 | } | |
1498 | if (c == 1) { | |
1499 | *flagp |= flags&SIMPLE; | |
a0d0e21e | 1500 | } |
a687059c | 1501 | |
a0d0e21e | 1502 | return(ret); |
a687059c LW |
1503 | } |
1504 | ||
1505 | /* | |
1506 | - regpiece - something followed by possible [*+?] | |
1507 | * | |
1508 | * Note that the branching code sequences used for ? and the general cases | |
1509 | * of * and + are somewhat optimized: they use the same NOTHING node as | |
1510 | * both the endmarker for their branch list and the body of the last branch. | |
1511 | * It might seem that this node could be dispensed with entirely, but the | |
1512 | * endmarker role is not redundant. | |
1513 | */ | |
76e3520e | 1514 | STATIC regnode * |
cea2e8a9 | 1515 | S_regpiece(pTHX_ I32 *flagp) |
a687059c | 1516 | { |
5c0ca799 | 1517 | dTHR; |
c277df42 | 1518 | register regnode *ret; |
a0d0e21e LW |
1519 | register char op; |
1520 | register char *next; | |
1521 | I32 flags; | |
3280af22 | 1522 | char *origparse = PL_regcomp_parse; |
a0d0e21e LW |
1523 | char *maxpos; |
1524 | I32 min; | |
c277df42 | 1525 | I32 max = REG_INFTY; |
a0d0e21e LW |
1526 | |
1527 | ret = regatom(&flags); | |
1528 | if (ret == NULL) { | |
1529 | if (flags & TRYAGAIN) | |
1530 | *flagp |= TRYAGAIN; | |
1531 | return(NULL); | |
1532 | } | |
1533 | ||
3280af22 | 1534 | op = *PL_regcomp_parse; |
a0d0e21e | 1535 | |
3280af22 NIS |
1536 | if (op == '{' && regcurly(PL_regcomp_parse)) { |
1537 | next = PL_regcomp_parse + 1; | |
a0d0e21e LW |
1538 | maxpos = Nullch; |
1539 | while (isDIGIT(*next) || *next == ',') { | |
1540 | if (*next == ',') { | |
1541 | if (maxpos) | |
1542 | break; | |
1543 | else | |
1544 | maxpos = next; | |
a687059c | 1545 | } |
a0d0e21e LW |
1546 | next++; |
1547 | } | |
1548 | if (*next == '}') { /* got one */ | |
1549 | if (!maxpos) | |
1550 | maxpos = next; | |
3280af22 NIS |
1551 | PL_regcomp_parse++; |
1552 | min = atoi(PL_regcomp_parse); | |
a0d0e21e LW |
1553 | if (*maxpos == ',') |
1554 | maxpos++; | |
1555 | else | |
3280af22 | 1556 | maxpos = PL_regcomp_parse; |
a0d0e21e LW |
1557 | max = atoi(maxpos); |
1558 | if (!max && *maxpos != '0') | |
c277df42 IZ |
1559 | max = REG_INFTY; /* meaning "infinity" */ |
1560 | else if (max >= REG_INFTY) | |
1561 | FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); | |
3280af22 | 1562 | PL_regcomp_parse = next; |
a0d0e21e LW |
1563 | nextchar(); |
1564 | ||
1565 | do_curly: | |
1566 | if ((flags&SIMPLE)) { | |
3280af22 | 1567 | PL_regnaughty += 2 + PL_regnaughty / 2; |
a0d0e21e LW |
1568 | reginsert(CURLY, ret); |
1569 | } | |
1570 | else { | |
2c2d71f5 JH |
1571 | regnode *w = reg_node(WHILEM); |
1572 | ||
1573 | w->flags = 0; | |
1574 | regtail(ret, w); | |
3280af22 | 1575 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 IZ |
1576 | reginsert(LONGJMP,ret); |
1577 | reginsert(NOTHING,ret); | |
1578 | NEXT_OFF(ret) = 3; /* Go over LONGJMP. */ | |
1579 | } | |
a0d0e21e | 1580 | reginsert(CURLYX,ret); |
3280af22 | 1581 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
1582 | NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */ |
1583 | regtail(ret, reg_node(NOTHING)); | |
1584 | if (SIZE_ONLY) | |
2c2d71f5 JH |
1585 | PL_reg_whilem_seen++, PL_extralen += 3; |
1586 | PL_regnaughty += 4 + PL_regnaughty; /* compound interest */ | |
a0d0e21e | 1587 | } |
c277df42 | 1588 | ret->flags = 0; |
a0d0e21e LW |
1589 | |
1590 | if (min > 0) | |
821b33a5 IZ |
1591 | *flagp = WORST; |
1592 | if (max > 0) | |
1593 | *flagp |= HASWIDTH; | |
a0d0e21e | 1594 | if (max && max < min) |
c277df42 IZ |
1595 | FAIL("Can't do {n,m} with n > m"); |
1596 | if (!SIZE_ONLY) { | |
1597 | ARG1_SET(ret, min); | |
1598 | ARG2_SET(ret, max); | |
a687059c | 1599 | } |
a687059c | 1600 | |
a0d0e21e | 1601 | goto nest_check; |
a687059c | 1602 | } |
a0d0e21e | 1603 | } |
a687059c | 1604 | |
a0d0e21e LW |
1605 | if (!ISMULT1(op)) { |
1606 | *flagp = flags; | |
a687059c | 1607 | return(ret); |
a0d0e21e | 1608 | } |
bb20fd44 | 1609 | |
c277df42 | 1610 | #if 0 /* Now runtime fix should be reliable. */ |
bb20fd44 | 1611 | if (!(flags&HASWIDTH) && op != '?') |
c277df42 IZ |
1612 | FAIL("regexp *+ operand could be empty"); |
1613 | #endif | |
bb20fd44 | 1614 | |
a0d0e21e LW |
1615 | nextchar(); |
1616 | ||
821b33a5 | 1617 | *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); |
a0d0e21e LW |
1618 | |
1619 | if (op == '*' && (flags&SIMPLE)) { | |
1620 | reginsert(STAR, ret); | |
c277df42 | 1621 | ret->flags = 0; |
3280af22 | 1622 | PL_regnaughty += 4; |
a0d0e21e LW |
1623 | } |
1624 | else if (op == '*') { | |
1625 | min = 0; | |
1626 | goto do_curly; | |
a0ed51b3 LW |
1627 | } |
1628 | else if (op == '+' && (flags&SIMPLE)) { | |
a0d0e21e | 1629 | reginsert(PLUS, ret); |
c277df42 | 1630 | ret->flags = 0; |
3280af22 | 1631 | PL_regnaughty += 3; |
a0d0e21e LW |
1632 | } |
1633 | else if (op == '+') { | |
1634 | min = 1; | |
1635 | goto do_curly; | |
a0ed51b3 LW |
1636 | } |
1637 | else if (op == '?') { | |
a0d0e21e LW |
1638 | min = 0; max = 1; |
1639 | goto do_curly; | |
1640 | } | |
1641 | nest_check: | |
17feb5d5 | 1642 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) { |
cea2e8a9 | 1643 | Perl_warner(aTHX_ WARN_UNSAFE, "%.*s matches null string many times", |
3280af22 | 1644 | PL_regcomp_parse - origparse, origparse); |
a0d0e21e LW |
1645 | } |
1646 | ||
3280af22 | 1647 | if (*PL_regcomp_parse == '?') { |
a0d0e21e LW |
1648 | nextchar(); |
1649 | reginsert(MINMOD, ret); | |
c277df42 | 1650 | regtail(ret, ret + NODE_STEP_REGNODE); |
a0d0e21e | 1651 | } |
3280af22 | 1652 | if (ISMULT2(PL_regcomp_parse)) |
a0d0e21e LW |
1653 | FAIL("nested *?+ in regexp"); |
1654 | ||
1655 | return(ret); | |
a687059c LW |
1656 | } |
1657 | ||
1658 | /* | |
1659 | - regatom - the lowest level | |
1660 | * | |
1661 | * Optimization: gobbles an entire sequence of ordinary characters so that | |
1662 | * it can turn them into a single node, which is smaller to store and | |
1663 | * faster to run. Backslashed characters are exceptions, each becoming a | |
1664 | * separate node; the code is simpler that way and it's not worth fixing. | |
1665 | * | |
1666 | * [Yes, it is worth fixing, some scripts can run twice the speed.] | |
1667 | */ | |
76e3520e | 1668 | STATIC regnode * |
cea2e8a9 | 1669 | S_regatom(pTHX_ I32 *flagp) |
a687059c | 1670 | { |
5c0ca799 | 1671 | dTHR; |
c277df42 | 1672 | register regnode *ret = 0; |
a0d0e21e LW |
1673 | I32 flags; |
1674 | ||
1675 | *flagp = WORST; /* Tentatively. */ | |
1676 | ||
1677 | tryagain: | |
3280af22 | 1678 | switch (*PL_regcomp_parse) { |
a0d0e21e | 1679 | case '^': |
3280af22 | 1680 | PL_seen_zerolen++; |
a0d0e21e | 1681 | nextchar(); |
3280af22 | 1682 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 1683 | ret = reg_node(MBOL); |
3280af22 | 1684 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 1685 | ret = reg_node(SBOL); |
a0d0e21e | 1686 | else |
c277df42 | 1687 | ret = reg_node(BOL); |
a0d0e21e LW |
1688 | break; |
1689 | case '$': | |
3280af22 NIS |
1690 | if (PL_regcomp_parse[1]) |
1691 | PL_seen_zerolen++; | |
a0d0e21e | 1692 | nextchar(); |
3280af22 | 1693 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 1694 | ret = reg_node(MEOL); |
3280af22 | 1695 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 1696 | ret = reg_node(SEOL); |
a0d0e21e | 1697 | else |
c277df42 | 1698 | ret = reg_node(EOL); |
a0d0e21e LW |
1699 | break; |
1700 | case '.': | |
1701 | nextchar(); | |
a0ed51b3 LW |
1702 | if (UTF) { |
1703 | if (PL_regflags & PMf_SINGLELINE) | |
1704 | ret = reg_node(SANYUTF8); | |
1705 | else | |
1706 | ret = reg_node(ANYUTF8); | |
1707 | *flagp |= HASWIDTH; | |
1708 | } | |
1709 | else { | |
1710 | if (PL_regflags & PMf_SINGLELINE) | |
1711 | ret = reg_node(SANY); | |
1712 | else | |
22c35a8c | 1713 | ret = reg_node(REG_ANY); |
a0ed51b3 LW |
1714 | *flagp |= HASWIDTH|SIMPLE; |
1715 | } | |
3280af22 | 1716 | PL_regnaughty++; |
a0d0e21e LW |
1717 | break; |
1718 | case '[': | |
3280af22 | 1719 | PL_regcomp_parse++; |
a0ed51b3 | 1720 | ret = (UTF ? regclassutf8() : regclass()); |
a14b48bc LW |
1721 | if (*PL_regcomp_parse != ']') |
1722 | FAIL("unmatched [] in regexp"); | |
1723 | nextchar(); | |
a0d0e21e LW |
1724 | *flagp |= HASWIDTH|SIMPLE; |
1725 | break; | |
1726 | case '(': | |
1727 | nextchar(); | |
1728 | ret = reg(1, &flags); | |
1729 | if (ret == NULL) { | |
1730 | if (flags & TRYAGAIN) | |
1731 | goto tryagain; | |
1732 | return(NULL); | |
1733 | } | |
c277df42 | 1734 | *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE); |
a0d0e21e LW |
1735 | break; |
1736 | case '|': | |
1737 | case ')': | |
1738 | if (flags & TRYAGAIN) { | |
1739 | *flagp |= TRYAGAIN; | |
1740 | return NULL; | |
1741 | } | |
3280af22 | 1742 | FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse); |
a0d0e21e LW |
1743 | /* Supposed to be caught earlier. */ |
1744 | break; | |
85afd4ae | 1745 | case '{': |
3280af22 NIS |
1746 | if (!regcurly(PL_regcomp_parse)) { |
1747 | PL_regcomp_parse++; | |
85afd4ae CS |
1748 | goto defchar; |
1749 | } | |
1750 | /* FALL THROUGH */ | |
a0d0e21e LW |
1751 | case '?': |
1752 | case '+': | |
1753 | case '*': | |
3115e423 | 1754 | FAIL("?+*{} follows nothing in regexp"); |
a0d0e21e LW |
1755 | break; |
1756 | case '\\': | |
3280af22 | 1757 | switch (*++PL_regcomp_parse) { |
a0d0e21e | 1758 | case 'A': |
3280af22 | 1759 | PL_seen_zerolen++; |
c277df42 | 1760 | ret = reg_node(SBOL); |
a0d0e21e LW |
1761 | *flagp |= SIMPLE; |
1762 | nextchar(); | |
1763 | break; | |
1764 | case 'G': | |
c277df42 | 1765 | ret = reg_node(GPOS); |
3280af22 | 1766 | PL_regseen |= REG_SEEN_GPOS; |
a0d0e21e LW |
1767 | *flagp |= SIMPLE; |
1768 | nextchar(); | |
1769 | break; | |
1770 | case 'Z': | |
c277df42 | 1771 | ret = reg_node(SEOL); |
a0d0e21e LW |
1772 | *flagp |= SIMPLE; |
1773 | nextchar(); | |
1774 | break; | |
b85d18e9 IZ |
1775 | case 'z': |
1776 | ret = reg_node(EOS); | |
1777 | *flagp |= SIMPLE; | |
3280af22 | 1778 | PL_seen_zerolen++; /* Do not optimize RE away */ |
b85d18e9 IZ |
1779 | nextchar(); |
1780 | break; | |
4a2d328f | 1781 | case 'C': |
a0ed51b3 LW |
1782 | ret = reg_node(SANY); |
1783 | *flagp |= HASWIDTH|SIMPLE; | |
1784 | nextchar(); | |
1785 | break; | |
1786 | case 'X': | |
1787 | ret = reg_node(CLUMP); | |
1788 | *flagp |= HASWIDTH; | |
1789 | nextchar(); | |
1790 | if (UTF && !PL_utf8_mark) | |
dfe13c55 | 1791 | is_utf8_mark((U8*)"~"); /* preload table */ |
a0ed51b3 | 1792 | break; |
a0d0e21e | 1793 | case 'w': |
a0ed51b3 LW |
1794 | ret = reg_node( |
1795 | UTF | |
1796 | ? (LOC ? ALNUMLUTF8 : ALNUMUTF8) | |
1797 | : (LOC ? ALNUML : ALNUM)); | |
a0d0e21e LW |
1798 | *flagp |= HASWIDTH|SIMPLE; |
1799 | nextchar(); | |
a0ed51b3 | 1800 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1801 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1802 | break; |
1803 | case 'W': | |
a0ed51b3 LW |
1804 | ret = reg_node( |
1805 | UTF | |
1806 | ? (LOC ? NALNUMLUTF8 : NALNUMUTF8) | |
1807 | : (LOC ? NALNUML : NALNUM)); | |
a0d0e21e LW |
1808 | *flagp |= HASWIDTH|SIMPLE; |
1809 | nextchar(); | |
a0ed51b3 | 1810 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1811 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1812 | break; |
1813 | case 'b': | |
3280af22 | 1814 | PL_seen_zerolen++; |
f5c9036e | 1815 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
1816 | ret = reg_node( |
1817 | UTF | |
1818 | ? (LOC ? BOUNDLUTF8 : BOUNDUTF8) | |
1819 | : (LOC ? BOUNDL : BOUND)); | |
a0d0e21e LW |
1820 | *flagp |= SIMPLE; |
1821 | nextchar(); | |
a0ed51b3 | 1822 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1823 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1824 | break; |
1825 | case 'B': | |
3280af22 | 1826 | PL_seen_zerolen++; |
f5c9036e | 1827 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
1828 | ret = reg_node( |
1829 | UTF | |
1830 | ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8) | |
1831 | : (LOC ? NBOUNDL : NBOUND)); | |
a0d0e21e LW |
1832 | *flagp |= SIMPLE; |
1833 | nextchar(); | |
a0ed51b3 | 1834 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1835 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1836 | break; |
1837 | case 's': | |
a0ed51b3 LW |
1838 | ret = reg_node( |
1839 | UTF | |
1840 | ? (LOC ? SPACELUTF8 : SPACEUTF8) | |
1841 | : (LOC ? SPACEL : SPACE)); | |
a0d0e21e LW |
1842 | *flagp |= HASWIDTH|SIMPLE; |
1843 | nextchar(); | |
a0ed51b3 | 1844 | if (UTF && !PL_utf8_space) |
dfe13c55 | 1845 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
1846 | break; |
1847 | case 'S': | |
a0ed51b3 LW |
1848 | ret = reg_node( |
1849 | UTF | |
1850 | ? (LOC ? NSPACELUTF8 : NSPACEUTF8) | |
1851 | : (LOC ? NSPACEL : NSPACE)); | |
a0d0e21e LW |
1852 | *flagp |= HASWIDTH|SIMPLE; |
1853 | nextchar(); | |
a0ed51b3 | 1854 | if (UTF && !PL_utf8_space) |
dfe13c55 | 1855 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
1856 | break; |
1857 | case 'd': | |
a0ed51b3 | 1858 | ret = reg_node(UTF ? DIGITUTF8 : DIGIT); |
a0d0e21e LW |
1859 | *flagp |= HASWIDTH|SIMPLE; |
1860 | nextchar(); | |
a0ed51b3 | 1861 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 1862 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e LW |
1863 | break; |
1864 | case 'D': | |
a0ed51b3 | 1865 | ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT); |
a0d0e21e LW |
1866 | *flagp |= HASWIDTH|SIMPLE; |
1867 | nextchar(); | |
a0ed51b3 | 1868 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 1869 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e | 1870 | break; |
a14b48bc LW |
1871 | case 'p': |
1872 | case 'P': | |
1873 | { /* a lovely hack--pretend we saw [\pX] instead */ | |
1874 | char* oldregxend = PL_regxend; | |
1875 | ||
1876 | if (PL_regcomp_parse[1] == '{') { | |
1877 | PL_regxend = strchr(PL_regcomp_parse, '}'); | |
1878 | if (!PL_regxend) | |
1879 | FAIL("Missing right brace on \\p{}"); | |
1880 | PL_regxend++; | |
1881 | } | |
1882 | else | |
1883 | PL_regxend = PL_regcomp_parse + 2; | |
1884 | PL_regcomp_parse--; | |
1885 | ||
1886 | ret = regclassutf8(); | |
1887 | ||
1888 | PL_regxend = oldregxend; | |
1889 | PL_regcomp_parse--; | |
1890 | nextchar(); | |
1891 | *flagp |= HASWIDTH|SIMPLE; | |
1892 | } | |
1893 | break; | |
a0d0e21e LW |
1894 | case 'n': |
1895 | case 'r': | |
1896 | case 't': | |
1897 | case 'f': | |
1898 | case 'e': | |
1899 | case 'a': | |
1900 | case 'x': | |
1901 | case 'c': | |
1902 | case '0': | |
1903 | goto defchar; | |
1904 | case '1': case '2': case '3': case '4': | |
1905 | case '5': case '6': case '7': case '8': case '9': | |
1906 | { | |
3280af22 | 1907 | I32 num = atoi(PL_regcomp_parse); |
a0d0e21e | 1908 | |
3280af22 | 1909 | if (num > 9 && num >= PL_regnpar) |
a0d0e21e LW |
1910 | goto defchar; |
1911 | else { | |
3280af22 | 1912 | if (!SIZE_ONLY && num > PL_regcomp_rx->nparens) |
ef64f398 | 1913 | FAIL("reference to nonexistent group"); |
3280af22 | 1914 | PL_regsawback = 1; |
a0ed51b3 LW |
1915 | ret = reganode(FOLD |
1916 | ? (LOC ? REFFL : REFF) | |
c8756f30 | 1917 | : REF, num); |
a0d0e21e | 1918 | *flagp |= HASWIDTH; |
3280af22 NIS |
1919 | while (isDIGIT(*PL_regcomp_parse)) |
1920 | PL_regcomp_parse++; | |
1921 | PL_regcomp_parse--; | |
a0d0e21e LW |
1922 | nextchar(); |
1923 | } | |
1924 | } | |
1925 | break; | |
1926 | case '\0': | |
3280af22 | 1927 | if (PL_regcomp_parse >= PL_regxend) |
a0d0e21e LW |
1928 | FAIL("trailing \\ in regexp"); |
1929 | /* FALL THROUGH */ | |
1930 | default: | |
c9f97d15 IZ |
1931 | /* Do not generate `unrecognized' warnings here, we fall |
1932 | back into the quick-grab loop below */ | |
a0d0e21e LW |
1933 | goto defchar; |
1934 | } | |
1935 | break; | |
4633a7c4 LW |
1936 | |
1937 | case '#': | |
3280af22 NIS |
1938 | if (PL_regflags & PMf_EXTENDED) { |
1939 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++; | |
1940 | if (PL_regcomp_parse < PL_regxend) | |
4633a7c4 LW |
1941 | goto tryagain; |
1942 | } | |
1943 | /* FALL THROUGH */ | |
1944 | ||
a0d0e21e LW |
1945 | default: { |
1946 | register I32 len; | |
a0ed51b3 | 1947 | register UV ender; |
a0d0e21e | 1948 | register char *p; |
c277df42 | 1949 | char *oldp, *s; |
a0d0e21e LW |
1950 | I32 numlen; |
1951 | ||
3280af22 | 1952 | PL_regcomp_parse++; |
a0d0e21e LW |
1953 | |
1954 | defchar: | |
a0ed51b3 LW |
1955 | ret = reg_node(FOLD |
1956 | ? (LOC ? EXACTFL : EXACTF) | |
bbce6d69 | 1957 | : EXACT); |
cd439c50 | 1958 | s = STRING(ret); |
3280af22 NIS |
1959 | for (len = 0, p = PL_regcomp_parse - 1; |
1960 | len < 127 && p < PL_regxend; | |
a0d0e21e LW |
1961 | len++) |
1962 | { | |
1963 | oldp = p; | |
5b5a24f7 | 1964 | |
3280af22 NIS |
1965 | if (PL_regflags & PMf_EXTENDED) |
1966 | p = regwhite(p, PL_regxend); | |
a0d0e21e LW |
1967 | switch (*p) { |
1968 | case '^': | |
1969 | case '$': | |
1970 | case '.': | |
1971 | case '[': | |
1972 | case '(': | |
1973 | case ')': | |
1974 | case '|': | |
1975 | goto loopdone; | |
1976 | case '\\': | |
1977 | switch (*++p) { | |
1978 | case 'A': | |
1979 | case 'G': | |
1980 | case 'Z': | |
b85d18e9 | 1981 | case 'z': |
a0d0e21e LW |
1982 | case 'w': |
1983 | case 'W': | |
1984 | case 'b': | |
1985 | case 'B': | |
1986 | case 's': | |
1987 | case 'S': | |
1988 | case 'd': | |
1989 | case 'D': | |
a14b48bc LW |
1990 | case 'p': |
1991 | case 'P': | |
a0d0e21e LW |
1992 | --p; |
1993 | goto loopdone; | |
1994 | case 'n': | |
1995 | ender = '\n'; | |
1996 | p++; | |
a687059c | 1997 | break; |
a0d0e21e LW |
1998 | case 'r': |
1999 | ender = '\r'; | |
2000 | p++; | |
a687059c | 2001 | break; |
a0d0e21e LW |
2002 | case 't': |
2003 | ender = '\t'; | |
2004 | p++; | |
a687059c | 2005 | break; |
a0d0e21e LW |
2006 | case 'f': |
2007 | ender = '\f'; | |
2008 | p++; | |
a687059c | 2009 | break; |
a0d0e21e LW |
2010 | case 'e': |
2011 | ender = '\033'; | |
2012 | p++; | |
a687059c | 2013 | break; |
a0d0e21e LW |
2014 | case 'a': |
2015 | ender = '\007'; | |
2016 | p++; | |
a687059c | 2017 | break; |
a0d0e21e | 2018 | case 'x': |
a0ed51b3 LW |
2019 | if (*++p == '{') { |
2020 | char* e = strchr(p, '}'); | |
2021 | ||
2022 | if (!e) | |
2023 | FAIL("Missing right brace on \\x{}"); | |
2024 | else if (UTF) { | |
2025 | ender = scan_hex(p + 1, e - p, &numlen); | |
2026 | if (numlen + len >= 127) { /* numlen is generous */ | |
2027 | p--; | |
2028 | goto loopdone; | |
2029 | } | |
2030 | p = e + 1; | |
2031 | } | |
2032 | else | |
2033 | FAIL("Can't use \\x{} without 'use utf8' declaration"); | |
2034 | } | |
2035 | else { | |
2036 | ender = scan_hex(p, 2, &numlen); | |
2037 | p += numlen; | |
2038 | } | |
a687059c | 2039 | break; |
a0d0e21e LW |
2040 | case 'c': |
2041 | p++; | |
bbce6d69 | 2042 | ender = UCHARAT(p++); |
2043 | ender = toCTRL(ender); | |
a687059c | 2044 | break; |
a0d0e21e LW |
2045 | case '0': case '1': case '2': case '3':case '4': |
2046 | case '5': case '6': case '7': case '8':case '9': | |
2047 | if (*p == '0' || | |
3280af22 | 2048 | (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) { |
a0d0e21e LW |
2049 | ender = scan_oct(p, 3, &numlen); |
2050 | p += numlen; | |
2051 | } | |
2052 | else { | |
2053 | --p; | |
2054 | goto loopdone; | |
a687059c LW |
2055 | } |
2056 | break; | |
a0d0e21e | 2057 | case '\0': |
3280af22 | 2058 | if (p >= PL_regxend) |
a687059c LW |
2059 | FAIL("trailing \\ in regexp"); |
2060 | /* FALL THROUGH */ | |
a0d0e21e | 2061 | default: |
c9f97d15 | 2062 | if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p)) |
cea2e8a9 | 2063 | Perl_warner(aTHX_ WARN_UNSAFE, |
1028017a JH |
2064 | "/%.127s/: Unrecognized escape \\%c passed through", |
2065 | PL_regprecomp, | |
2066 | *p); | |
a0ed51b3 | 2067 | goto normal_default; |
a0d0e21e LW |
2068 | } |
2069 | break; | |
a687059c | 2070 | default: |
a0ed51b3 LW |
2071 | normal_default: |
2072 | if ((*p & 0xc0) == 0xc0 && UTF) { | |
dfe13c55 | 2073 | ender = utf8_to_uv((U8*)p, &numlen); |
a0ed51b3 LW |
2074 | p += numlen; |
2075 | } | |
2076 | else | |
2077 | ender = *p++; | |
a0d0e21e | 2078 | break; |
a687059c | 2079 | } |
3280af22 NIS |
2080 | if (PL_regflags & PMf_EXTENDED) |
2081 | p = regwhite(p, PL_regxend); | |
a0ed51b3 LW |
2082 | if (UTF && FOLD) { |
2083 | if (LOC) | |
2084 | ender = toLOWER_LC_uni(ender); | |
2085 | else | |
2086 | ender = toLOWER_uni(ender); | |
2087 | } | |
a0d0e21e LW |
2088 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
2089 | if (len) | |
2090 | p = oldp; | |
a0ed51b3 LW |
2091 | else if (ender >= 0x80 && UTF) { |
2092 | reguni(ender, s, &numlen); | |
2093 | s += numlen; | |
2094 | len += numlen; | |
2095 | } | |
a0d0e21e LW |
2096 | else { |
2097 | len++; | |
cd439c50 | 2098 | REGC(ender, s++); |
a0d0e21e LW |
2099 | } |
2100 | break; | |
a687059c | 2101 | } |
a0ed51b3 LW |
2102 | if (ender >= 0x80 && UTF) { |
2103 | reguni(ender, s, &numlen); | |
2104 | s += numlen; | |
2105 | len += numlen - 1; | |
2106 | } | |
2107 | else | |
cd439c50 | 2108 | REGC(ender, s++); |
a0d0e21e LW |
2109 | } |
2110 | loopdone: | |
3280af22 | 2111 | PL_regcomp_parse = p - 1; |
a0d0e21e LW |
2112 | nextchar(); |
2113 | if (len < 0) | |
2114 | FAIL("internal disaster in regexp"); | |
2115 | if (len > 0) | |
2116 | *flagp |= HASWIDTH; | |
2117 | if (len == 1) | |
2118 | *flagp |= SIMPLE; | |
c277df42 | 2119 | if (!SIZE_ONLY) |
cd439c50 IZ |
2120 | STR_LEN(ret) = len; |
2121 | if (SIZE_ONLY) | |
2122 | PL_regsize += STR_SZ(len); | |
2123 | else | |
2124 | PL_regcode += STR_SZ(len); | |
a687059c | 2125 | } |
a0d0e21e LW |
2126 | break; |
2127 | } | |
a687059c | 2128 | |
a0d0e21e | 2129 | return(ret); |
a687059c LW |
2130 | } |
2131 | ||
873ef191 | 2132 | STATIC char * |
cea2e8a9 | 2133 | S_regwhite(pTHX_ char *p, char *e) |
5b5a24f7 CS |
2134 | { |
2135 | while (p < e) { | |
2136 | if (isSPACE(*p)) | |
2137 | ++p; | |
2138 | else if (*p == '#') { | |
2139 | do { | |
2140 | p++; | |
2141 | } while (p < e && *p != '\n'); | |
2142 | } | |
2143 | else | |
2144 | break; | |
2145 | } | |
2146 | return p; | |
2147 | } | |
2148 | ||
b8c5462f JH |
2149 | /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]]. |
2150 | Character classes ([:foo:]) can also be negated ([:^foo:]). | |
2151 | Returns a named class id (ANYOF_XXX) if successful, -1 otherwise. | |
2152 | Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed, | |
2153 | but trigger warnings because they are currently unimplemented. */ | |
2154 | STATIC I32 | |
cea2e8a9 | 2155 | S_regpposixcc(pTHX_ I32 value) |
620e46c5 | 2156 | { |
11b8faa4 | 2157 | dTHR; |
620e46c5 | 2158 | char *posixcc = 0; |
b8c5462f | 2159 | I32 namedclass = -1; |
620e46c5 JH |
2160 | |
2161 | if (value == '[' && PL_regcomp_parse + 1 < PL_regxend && | |
2162 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ | |
2163 | (*PL_regcomp_parse == ':' || | |
2164 | *PL_regcomp_parse == '=' || | |
2165 | *PL_regcomp_parse == '.')) { | |
2166 | char c = *PL_regcomp_parse; | |
2167 | char* s = PL_regcomp_parse++; | |
2168 | ||
2169 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c) | |
2170 | PL_regcomp_parse++; | |
2171 | if (PL_regcomp_parse == PL_regxend) | |
2172 | /* Grandfather lone [:, [=, [. */ | |
2173 | PL_regcomp_parse = s; | |
2174 | else { | |
b8c5462f JH |
2175 | char* t = PL_regcomp_parse++; /* skip over the c */ |
2176 | ||
2177 | if (*PL_regcomp_parse == ']') { | |
2178 | PL_regcomp_parse++; /* skip over the ending ] */ | |
2179 | posixcc = s + 1; | |
2180 | if (*s == ':') { | |
2181 | I32 complement = *posixcc == '^' ? *posixcc++ : 0; | |
2182 | I32 skip = 5; /* the most common skip */ | |
2183 | ||
2184 | switch (*posixcc) { | |
2185 | case 'a': | |
2186 | if (strnEQ(posixcc, "alnum", 5)) | |
2187 | namedclass = | |
2188 | complement ? ANYOF_NALNUMC : ANYOF_ALNUMC; | |
2189 | else if (strnEQ(posixcc, "alpha", 5)) | |
2190 | namedclass = | |
2191 | complement ? ANYOF_NALPHA : ANYOF_ALPHA; | |
2192 | else if (strnEQ(posixcc, "ascii", 5)) | |
2193 | namedclass = | |
2194 | complement ? ANYOF_NASCII : ANYOF_ASCII; | |
2195 | break; | |
2196 | case 'c': | |
2197 | if (strnEQ(posixcc, "cntrl", 5)) | |
2198 | namedclass = | |
2199 | complement ? ANYOF_NCNTRL : ANYOF_CNTRL; | |
2200 | break; | |
2201 | case 'd': | |
2202 | if (strnEQ(posixcc, "digit", 5)) | |
2203 | namedclass = | |
2204 | complement ? ANYOF_NDIGIT : ANYOF_DIGIT; | |
2205 | break; | |
2206 | case 'g': | |
2207 | if (strnEQ(posixcc, "graph", 5)) | |
2208 | namedclass = | |
2209 | complement ? ANYOF_NGRAPH : ANYOF_GRAPH; | |
2210 | break; | |
2211 | case 'l': | |
2212 | if (strnEQ(posixcc, "lower", 5)) | |
2213 | namedclass = | |
2214 | complement ? ANYOF_NLOWER : ANYOF_LOWER; | |
2215 | break; | |
2216 | case 'p': | |
2217 | if (strnEQ(posixcc, "print", 5)) | |
2218 | namedclass = | |
2219 | complement ? ANYOF_NPRINT : ANYOF_PRINT; | |
2220 | else if (strnEQ(posixcc, "punct", 5)) | |
2221 | namedclass = | |
2222 | complement ? ANYOF_NPUNCT : ANYOF_PUNCT; | |
2223 | break; | |
2224 | case 's': | |
2225 | if (strnEQ(posixcc, "space", 5)) | |
2226 | namedclass = | |
2227 | complement ? ANYOF_NSPACE : ANYOF_SPACE; | |
2228 | case 'u': | |
2229 | if (strnEQ(posixcc, "upper", 5)) | |
2230 | namedclass = | |
2231 | complement ? ANYOF_NUPPER : ANYOF_UPPER; | |
2232 | break; | |
2233 | case 'w': /* this is not POSIX, this is the Perl \w */ | |
2234 | if (strnEQ(posixcc, "word", 4)) { | |
2235 | namedclass = | |
2236 | complement ? ANYOF_NALNUM : ANYOF_ALNUM; | |
2237 | skip = 4; | |
2238 | } | |
2239 | break; | |
2240 | case 'x': | |
2241 | if (strnEQ(posixcc, "xdigit", 6)) { | |
2242 | namedclass = | |
2243 | complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT; | |
2244 | skip = 6; | |
2245 | } | |
2246 | break; | |
2247 | } | |
93733859 | 2248 | if ((namedclass == OOB_NAMEDCLASS || |
b8c5462f JH |
2249 | !(posixcc + skip + 2 < PL_regxend && |
2250 | (posixcc[skip] == ':' && | |
2251 | posixcc[skip + 1] == ']')))) | |
2252 | Perl_croak(aTHX_ "Character class [:%.*s:] unknown", | |
2253 | t - s - 1, s + 1); | |
2254 | } else if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY) | |
2255 | /* [[=foo=]] and [[.foo.]] are still future. */ | |
cea2e8a9 | 2256 | Perl_warner(aTHX_ WARN_UNSAFE, |
b8c5462f JH |
2257 | "Character class syntax [%c %c] is reserved for future extensions", c, c); |
2258 | } else { | |
2259 | /* Maternal grandfather: | |
2260 | * "[:" ending in ":" but not in ":]" */ | |
767d463e HS |
2261 | PL_regcomp_parse = s; |
2262 | } | |
620e46c5 JH |
2263 | } |
2264 | } | |
2265 | ||
b8c5462f JH |
2266 | return namedclass; |
2267 | } | |
2268 | ||
2269 | STATIC void | |
2270 | S_checkposixcc(pTHX) | |
2271 | { | |
2272 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && | |
2273 | (*PL_regcomp_parse == ':' || | |
2274 | *PL_regcomp_parse == '=' || | |
2275 | *PL_regcomp_parse == '.')) { | |
2276 | char *s = PL_regcomp_parse; | |
93733859 | 2277 | char c = *s++; |
b8c5462f JH |
2278 | |
2279 | while(*s && isALNUM(*s)) | |
2280 | s++; | |
2281 | if (*s && c == *s && s[1] == ']') { | |
2282 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2283 | "Character class syntax [%c %c] belongs inside character classes", c, c); | |
2284 | if (c == '=' || c == '.') | |
2285 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2286 | "Character class syntax [%c %c] is reserved for future extensions", c, c); | |
2287 | } | |
2288 | } | |
620e46c5 JH |
2289 | } |
2290 | ||
76e3520e | 2291 | STATIC regnode * |
cea2e8a9 | 2292 | S_regclass(pTHX) |
a687059c | 2293 | { |
5c0ca799 | 2294 | dTHR; |
c277df42 | 2295 | register char *opnd, *s; |
a0ed51b3 | 2296 | register I32 value; |
b8c5462f | 2297 | register I32 lastvalue = OOB_CHAR8; |
a0d0e21e | 2298 | register I32 range = 0; |
c277df42 | 2299 | register regnode *ret; |
a0d0e21e LW |
2300 | register I32 def; |
2301 | I32 numlen; | |
b8c5462f | 2302 | I32 namedclass; |
73b437c8 | 2303 | char *rangebegin; |
a0d0e21e | 2304 | |
cd439c50 | 2305 | s = opnd = MASK(PL_regcode); |
c277df42 | 2306 | ret = reg_node(ANYOF); |
b8c5462f | 2307 | for (value = 0; value < ANYOF_SIZE; value++) |
cd439c50 | 2308 | REGC(0, s++); |
3280af22 NIS |
2309 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2310 | PL_regnaughty++; | |
2311 | PL_regcomp_parse++; | |
c277df42 | 2312 | if (!SIZE_ONLY) |
b8c5462f | 2313 | ANYOF_FLAGS(opnd) |= ANYOF_INVERT; |
bbce6d69 | 2314 | } |
c277df42 | 2315 | if (!SIZE_ONLY) { |
3280af22 | 2316 | PL_regcode += ANY_SKIP; |
a0ed51b3 | 2317 | if (FOLD) |
b8c5462f | 2318 | ANYOF_FLAGS(opnd) |= ANYOF_FOLD; |
a0ed51b3 | 2319 | if (LOC) |
b8c5462f | 2320 | ANYOF_FLAGS(opnd) |= ANYOF_LOCALE; |
a0ed51b3 LW |
2321 | } |
2322 | else { | |
3280af22 | 2323 | PL_regsize += ANY_SKIP; |
a0d0e21e | 2324 | } |
b8c5462f JH |
2325 | |
2326 | checkposixcc(); | |
2327 | ||
3280af22 | 2328 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
a0d0e21e | 2329 | goto skipcond; /* allow 1st char to be ] or - */ |
3280af22 | 2330 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
a0d0e21e | 2331 | skipcond: |
93733859 | 2332 | namedclass = OOB_NAMEDCLASS; |
73b437c8 JH |
2333 | if (!range) |
2334 | rangebegin = PL_regcomp_parse; | |
a0ed51b3 | 2335 | value = UCHARAT(PL_regcomp_parse++); |
620e46c5 | 2336 | if (value == '[') |
b8c5462f | 2337 | namedclass = regpposixcc(value); |
620e46c5 | 2338 | else if (value == '\\') { |
a0ed51b3 LW |
2339 | value = UCHARAT(PL_regcomp_parse++); |
2340 | switch (value) { | |
b8c5462f JH |
2341 | case 'w': namedclass = ANYOF_ALNUM; break; |
2342 | case 'W': namedclass = ANYOF_NALNUM; break; | |
2343 | case 's': namedclass = ANYOF_SPACE; break; | |
2344 | case 'S': namedclass = ANYOF_NSPACE; break; | |
2345 | case 'd': namedclass = ANYOF_DIGIT; break; | |
2346 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
2347 | case 'n': value = '\n'; break; | |
2348 | case 'r': value = '\r'; break; | |
2349 | case 't': value = '\t'; break; | |
2350 | case 'f': value = '\f'; break; | |
2351 | case 'b': value = '\b'; break; | |
2352 | case 'e': value = '\033'; break; | |
2353 | case 'a': value = '\007'; break; | |
2354 | case 'x': | |
2355 | value = scan_hex(PL_regcomp_parse, 2, &numlen); | |
2356 | PL_regcomp_parse += numlen; | |
2357 | break; | |
2358 | case 'c': | |
2359 | value = UCHARAT(PL_regcomp_parse++); | |
2360 | value = toCTRL(value); | |
2361 | break; | |
2362 | case '0': case '1': case '2': case '3': case '4': | |
2363 | case '5': case '6': case '7': case '8': case '9': | |
2364 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); | |
2365 | PL_regcomp_parse += numlen; | |
2366 | break; | |
1028017a JH |
2367 | default: |
2368 | if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(value)) | |
2369 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2370 | "/%.127s/: Unrecognized escape \\%c in character class passed through", | |
2371 | PL_regprecomp, | |
2372 | value); | |
2373 | break; | |
b8c5462f JH |
2374 | } |
2375 | } | |
73b437c8 JH |
2376 | if (namedclass > OOB_NAMEDCLASS) { |
2377 | if (range) { /* a-\d, a-[:digit:] */ | |
2378 | if (!SIZE_ONLY) { | |
2379 | if (ckWARN(WARN_UNSAFE)) | |
2380 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2381 | "/%.127s/: false [] range \"%*.*s\" in regexp", | |
2382 | PL_regprecomp, | |
2383 | PL_regcomp_parse - rangebegin, | |
2384 | PL_regcomp_parse - rangebegin, | |
2385 | rangebegin); | |
2386 | ANYOF_BITMAP_SET(opnd, lastvalue); | |
2387 | ANYOF_BITMAP_SET(opnd, '-'); | |
b8c5462f | 2388 | } |
73b437c8 JH |
2389 | range = 0; /* this is not a true range */ |
2390 | } | |
2391 | if (!SIZE_ONLY) { | |
2392 | switch (namedclass) { | |
2393 | case ANYOF_ALNUM: | |
2394 | if (LOC) | |
2395 | ANYOF_CLASS_SET(opnd, ANYOF_ALNUM); | |
2396 | else { | |
2397 | for (value = 0; value < 256; value++) | |
2398 | if (isALNUM(value)) | |
2399 | ANYOF_BITMAP_SET(opnd, value); | |
2400 | } | |
2401 | break; | |
2402 | case ANYOF_NALNUM: | |
2403 | if (LOC) | |
2404 | ANYOF_CLASS_SET(opnd, ANYOF_NALNUM); | |
2405 | else { | |
2406 | for (value = 0; value < 256; value++) | |
2407 | if (!isALNUM(value)) | |
2408 | ANYOF_BITMAP_SET(opnd, value); | |
2409 | } | |
2410 | break; | |
2411 | case ANYOF_SPACE: | |
2412 | if (LOC) | |
2413 | ANYOF_CLASS_SET(opnd, ANYOF_SPACE); | |
2414 | else { | |
2415 | for (value = 0; value < 256; value++) | |
2416 | if (isSPACE(value)) | |
2417 | ANYOF_BITMAP_SET(opnd, value); | |
2418 | } | |
2419 | break; | |
2420 | case ANYOF_NSPACE: | |
2421 | if (LOC) | |
2422 | ANYOF_CLASS_SET(opnd, ANYOF_NSPACE); | |
2423 | else { | |
2424 | for (value = 0; value < 256; value++) | |
2425 | if (!isSPACE(value)) | |
2426 | ANYOF_BITMAP_SET(opnd, value); | |
2427 | } | |
2428 | break; | |
2429 | case ANYOF_DIGIT: | |
2430 | if (LOC) | |
2431 | ANYOF_CLASS_SET(opnd, ANYOF_DIGIT); | |
2432 | else { | |
2433 | for (value = '0'; value <= '9'; value++) | |
b8c5462f | 2434 | ANYOF_BITMAP_SET(opnd, value); |
73b437c8 JH |
2435 | } |
2436 | break; | |
2437 | case ANYOF_NDIGIT: | |
2438 | if (LOC) | |
2439 | ANYOF_CLASS_SET(opnd, ANYOF_NDIGIT); | |
2440 | else { | |
2441 | for (value = 0; value < '0'; value++) | |
b8c5462f | 2442 | ANYOF_BITMAP_SET(opnd, value); |
73b437c8 | 2443 | for (value = '9' + 1; value < 256; value++) |
b8c5462f | 2444 | ANYOF_BITMAP_SET(opnd, value); |
73b437c8 JH |
2445 | } |
2446 | break; | |
2447 | case ANYOF_NALNUMC: | |
2448 | if (LOC) | |
2449 | ANYOF_CLASS_SET(opnd, ANYOF_NALNUMC); | |
2450 | else { | |
2451 | for (value = 0; value < 256; value++) | |
2452 | if (!isALNUMC(value)) | |
2453 | ANYOF_BITMAP_SET(opnd, value); | |
2454 | } | |
2455 | break; | |
2456 | case ANYOF_ALNUMC: | |
2457 | if (LOC) | |
2458 | ANYOF_CLASS_SET(opnd, ANYOF_ALNUMC); | |
2459 | else { | |
2460 | for (value = 0; value < 256; value++) | |
2461 | if (isALNUMC(value)) | |
2462 | ANYOF_BITMAP_SET(opnd, value); | |
2463 | } | |
2464 | break; | |
2465 | case ANYOF_ALPHA: | |
2466 | if (LOC) | |
2467 | ANYOF_CLASS_SET(opnd, ANYOF_ALPHA); | |
2468 | else { | |
2469 | for (value = 0; value < 256; value++) | |
2470 | if (isALPHA(value)) | |
2471 | ANYOF_BITMAP_SET(opnd, value); | |
2472 | } | |
2473 | break; | |
2474 | case ANYOF_NALPHA: | |
2475 | if (LOC) | |
2476 | ANYOF_CLASS_SET(opnd, ANYOF_NALPHA); | |
2477 | else { | |
2478 | for (value = 0; value < 256; value++) | |
2479 | if (!isALPHA(value)) | |
2480 | ANYOF_BITMAP_SET(opnd, value); | |
2481 | } | |
2482 | break; | |
2483 | case ANYOF_ASCII: | |
2484 | if (LOC) | |
2485 | ANYOF_CLASS_SET(opnd, ANYOF_ASCII); | |
2486 | else { | |
2487 | for (value = 0; value < 128; value++) | |
b8c5462f | 2488 | ANYOF_BITMAP_SET(opnd, value); |
73b437c8 JH |
2489 | } |
2490 | break; | |
2491 | case ANYOF_NASCII: | |
2492 | if (LOC) | |
2493 | ANYOF_CLASS_SET(opnd, ANYOF_NASCII); | |
2494 | else { | |
2495 | for (value = 128; value < 256; value++) | |
b8c5462f | 2496 | ANYOF_BITMAP_SET(opnd, value); |
73b437c8 JH |
2497 | } |
2498 | break; | |
2499 | case ANYOF_CNTRL: | |
2500 | if (LOC) | |
2501 | ANYOF_CLASS_SET(opnd, ANYOF_CNTRL); | |
2502 | else { | |
2503 | for (value = 0; value < 256; value++) | |
2504 | if (isCNTRL(value)) | |
2505 | ANYOF_BITMAP_SET(opnd, value); | |
2506 | } | |
2507 | lastvalue = OOB_CHAR8; | |
2508 | break; | |
2509 | case ANYOF_NCNTRL: | |
2510 | if (LOC) | |
2511 | ANYOF_CLASS_SET(opnd, ANYOF_NCNTRL); | |
2512 | else { | |
2513 | for (value = 0; value < 256; value++) | |
2514 | if (!isCNTRL(value)) | |
2515 | ANYOF_BITMAP_SET(opnd, value); | |
2516 | } | |
2517 | break; | |
2518 | case ANYOF_GRAPH: | |
2519 | if (LOC) | |
2520 | ANYOF_CLASS_SET(opnd, ANYOF_GRAPH); | |
2521 | else { | |
2522 | for (value = 0; value < 256; value++) | |
2523 | if (isGRAPH(value)) | |
2524 | ANYOF_BITMAP_SET(opnd, value); | |
2525 | } | |
2526 | break; | |
2527 | case ANYOF_NGRAPH: | |
2528 | if (LOC) | |
2529 | ANYOF_CLASS_SET(opnd, ANYOF_NGRAPH); | |
2530 | else { | |
2531 | for (value = 0; value < 256; value++) | |
2532 | if (!isGRAPH(value)) | |
2533 | ANYOF_BITMAP_SET(opnd, value); | |
2534 | } | |
2535 | break; | |
2536 | case ANYOF_LOWER: | |
2537 | if (LOC) | |
2538 | ANYOF_CLASS_SET(opnd, ANYOF_LOWER); | |
2539 | else { | |
2540 | for (value = 0; value < 256; value++) | |
2541 | if (isLOWER(value)) | |
2542 | ANYOF_BITMAP_SET(opnd, value); | |
2543 | } | |
2544 | break; | |
2545 | case ANYOF_NLOWER: | |
2546 | if (LOC) | |
2547 | ANYOF_CLASS_SET(opnd, ANYOF_NLOWER); | |
2548 | else { | |
2549 | for (value = 0; value < 256; value++) | |
2550 | if (!isLOWER(value)) | |
2551 | ANYOF_BITMAP_SET(opnd, value); | |
2552 | } | |
2553 | break; | |
2554 | case ANYOF_PRINT: | |
2555 | if (LOC) | |
2556 | ANYOF_CLASS_SET(opnd, ANYOF_PRINT); | |
2557 | else { | |
2558 | for (value = 0; value < 256; value++) | |
2559 | if (isPRINT(value)) | |
2560 | ANYOF_BITMAP_SET(opnd, value); | |
2561 | } | |
2562 | break; | |
2563 | case ANYOF_NPRINT: | |
2564 | if (LOC) | |
2565 | ANYOF_CLASS_SET(opnd, ANYOF_NPRINT); | |
2566 | else { | |
2567 | for (value = 0; value < 256; value++) | |
2568 | if (!isPRINT(value)) | |
2569 | ANYOF_BITMAP_SET(opnd, value); | |
2570 | } | |
2571 | break; | |
2572 | case ANYOF_PUNCT: | |
2573 | if (LOC) | |
2574 | ANYOF_CLASS_SET(opnd, ANYOF_PUNCT); | |
2575 | else { | |
2576 | for (value = 0; value < 256; value++) | |
2577 | if (isPUNCT(value)) | |
2578 | ANYOF_BITMAP_SET(opnd, value); | |
2579 | } | |
2580 | break; | |
2581 | case ANYOF_NPUNCT: | |
2582 | if (LOC) | |
2583 | ANYOF_CLASS_SET(opnd, ANYOF_NPUNCT); | |
2584 | else { | |
2585 | for (value = 0; value < 256; value++) | |
2586 | if (!isPUNCT(value)) | |
2587 | ANYOF_BITMAP_SET(opnd, value); | |
2588 | } | |
2589 | break; | |
2590 | case ANYOF_UPPER: | |
2591 | if (LOC) | |
2592 | ANYOF_CLASS_SET(opnd, ANYOF_UPPER); | |
2593 | else { | |
2594 | for (value = 0; value < 256; value++) | |
2595 | if (isUPPER(value)) | |
2596 | ANYOF_BITMAP_SET(opnd, value); | |
2597 | } | |
2598 | break; | |
2599 | case ANYOF_NUPPER: | |
2600 | if (LOC) | |
2601 | ANYOF_CLASS_SET(opnd, ANYOF_NUPPER); | |
2602 | else { | |
2603 | for (value = 0; value < 256; value++) | |
2604 | if (!isUPPER(value)) | |
2605 | ANYOF_BITMAP_SET(opnd, value); | |
2606 | } | |
2607 | break; | |
2608 | case ANYOF_XDIGIT: | |
2609 | if (LOC) | |
2610 | ANYOF_CLASS_SET(opnd, ANYOF_XDIGIT); | |
2611 | else { | |
2612 | for (value = 0; value < 256; value++) | |
2613 | if (isXDIGIT(value)) | |
2614 | ANYOF_BITMAP_SET(opnd, value); | |
2615 | } | |
2616 | break; | |
2617 | case ANYOF_NXDIGIT: | |
2618 | if (LOC) | |
2619 | ANYOF_CLASS_SET(opnd, ANYOF_NXDIGIT); | |
2620 | else { | |
2621 | for (value = 0; value < 256; value++) | |
2622 | if (!isXDIGIT(value)) | |
2623 | ANYOF_BITMAP_SET(opnd, value); | |
2624 | } | |
2625 | break; | |
2626 | default: | |
2627 | FAIL("invalid [::] class in regexp"); | |
2628 | break; | |
b8c5462f | 2629 | } |
b8c5462f | 2630 | if (LOC) |
73b437c8 JH |
2631 | ANYOF_FLAGS(opnd) |= ANYOF_CLASS; |
2632 | continue; | |
a0d0e21e LW |
2633 | } |
2634 | } | |
2635 | if (range) { | |
73b437c8 JH |
2636 | if (lastvalue > value) /* b-a */ { |
2637 | Perl_croak(aTHX_ | |
2638 | "/%.127s/: invalid [] range \"%*.*s\" in regexp", | |
2639 | PL_regprecomp, | |
2640 | PL_regcomp_parse - rangebegin, | |
2641 | PL_regcomp_parse - rangebegin, | |
2642 | rangebegin); | |
2643 | } | |
a0d0e21e LW |
2644 | range = 0; |
2645 | } | |
2646 | else { | |
a0ed51b3 | 2647 | lastvalue = value; |
3280af22 | 2648 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
93733859 | 2649 | PL_regcomp_parse[1] != ']') { |
3280af22 | 2650 | PL_regcomp_parse++; |
73b437c8 JH |
2651 | if (namedclass > OOB_NAMEDCLASS) { /* \w-, [:word:]- */ |
2652 | if (ckWARN(WARN_UNSAFE)) | |
2653 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2654 | "/%.127s/: false [] range \"%*.*s\" in regexp", | |
2655 | PL_regprecomp, | |
2656 | PL_regcomp_parse - rangebegin, | |
2657 | PL_regcomp_parse - rangebegin, | |
2658 | rangebegin); | |
2659 | if (!SIZE_ONLY) | |
2660 | ANYOF_BITMAP_SET(opnd, '-'); | |
2661 | } else | |
2662 | range = 1; | |
a0d0e21e LW |
2663 | continue; /* do it next time */ |
2664 | } | |
a687059c | 2665 | } |
93733859 | 2666 | /* now is the next time */ |
ae5c130c | 2667 | if (!SIZE_ONLY) { |
93733859 | 2668 | #ifndef ASCIIish /* EBCDIC, for example. */ |
8ada0baa | 2669 | if ((isLOWER(lastvalue) && isLOWER(value)) || |
c8dba6f3 GS |
2670 | (isUPPER(lastvalue) && isUPPER(value))) |
2671 | { | |
2672 | I32 i; | |
8ada0baa JH |
2673 | if (isLOWER(lastvalue)) { |
2674 | for (i = lastvalue; i <= value; i++) | |
2675 | if (isLOWER(i)) | |
b8c5462f | 2676 | ANYOF_BITMAP_SET(opnd, i); |
8ada0baa JH |
2677 | } else { |
2678 | for (i = lastvalue; i <= value; i++) | |
2679 | if (isUPPER(i)) | |
b8c5462f | 2680 | ANYOF_BITMAP_SET(opnd, i); |
8ada0baa JH |
2681 | } |
2682 | } | |
2683 | else | |
2684 | #endif | |
2685 | for ( ; lastvalue <= value; lastvalue++) | |
b8c5462f | 2686 | ANYOF_BITMAP_SET(opnd, lastvalue); |
8ada0baa | 2687 | } |
93733859 | 2688 | range = 0; |
a0d0e21e | 2689 | } |
ae5c130c | 2690 | /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ |
b8c5462f JH |
2691 | if (!SIZE_ONLY && |
2692 | (ANYOF_FLAGS(opnd) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD) { | |
a0ed51b3 | 2693 | for (value = 0; value < 256; ++value) { |
b8c5462f | 2694 | if (ANYOF_BITMAP_TEST(opnd, value)) { |
22c35a8c | 2695 | I32 cf = PL_fold[value]; |
b8c5462f | 2696 | ANYOF_BITMAP_SET(opnd, cf); |
ae5c130c GS |
2697 | } |
2698 | } | |
b8c5462f | 2699 | ANYOF_FLAGS(opnd) &= ~ANYOF_FOLD; |
ae5c130c GS |
2700 | } |
2701 | /* optimize inverted simple patterns (e.g. [^a-z]) */ | |
b8c5462f JH |
2702 | if (!SIZE_ONLY && (ANYOF_FLAGS(opnd) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) { |
2703 | for (value = 0; value < ANYOF_BITMAP_SIZE; ++value) | |
2704 | opnd[ANYOF_BITMAP_OFFSET + value] ^= ANYOF_FLAGS_ALL; | |
2705 | ANYOF_FLAGS(opnd) = 0; | |
ae5c130c | 2706 | } |
a0d0e21e LW |
2707 | return ret; |
2708 | } | |
2709 | ||
a0ed51b3 | 2710 | STATIC regnode * |
cea2e8a9 | 2711 | S_regclassutf8(pTHX) |
a0ed51b3 | 2712 | { |
b8c5462f | 2713 | dTHR; |
a0ed51b3 LW |
2714 | register char *opnd, *e; |
2715 | register U32 value; | |
b8c5462f | 2716 | register U32 lastvalue = OOB_UTF8; |
a0ed51b3 LW |
2717 | register I32 range = 0; |
2718 | register regnode *ret; | |
2719 | I32 numlen; | |
2720 | I32 n; | |
2721 | SV *listsv; | |
2722 | U8 flags = 0; | |
b8c5462f | 2723 | I32 namedclass; |
73b437c8 | 2724 | char *rangebegin; |
a0ed51b3 LW |
2725 | |
2726 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ | |
2727 | PL_regnaughty++; | |
2728 | PL_regcomp_parse++; | |
2729 | if (!SIZE_ONLY) | |
2730 | flags |= ANYOF_INVERT; | |
2731 | } | |
2732 | if (!SIZE_ONLY) { | |
2733 | if (FOLD) | |
2734 | flags |= ANYOF_FOLD; | |
2735 | if (LOC) | |
2736 | flags |= ANYOF_LOCALE; | |
79cb57f6 | 2737 | listsv = newSVpvn("# comment\n",10); |
a0ed51b3 LW |
2738 | } |
2739 | ||
b8c5462f JH |
2740 | checkposixcc(); |
2741 | ||
a0ed51b3 LW |
2742 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
2743 | goto skipcond; /* allow 1st char to be ] or - */ | |
2744 | ||
2745 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { | |
2746 | skipcond: | |
93733859 | 2747 | namedclass = OOB_NAMEDCLASS; |
73b437c8 JH |
2748 | if (!range) |
2749 | rangebegin = PL_regcomp_parse; | |
dfe13c55 | 2750 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 | 2751 | PL_regcomp_parse += numlen; |
620e46c5 | 2752 | if (value == '[') |
b8c5462f | 2753 | namedclass = regpposixcc(value); |
620e46c5 | 2754 | else if (value == '\\') { |
dfe13c55 | 2755 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 LW |
2756 | PL_regcomp_parse += numlen; |
2757 | switch (value) { | |
b8c5462f JH |
2758 | case 'w': namedclass = ANYOF_ALNUM; break; |
2759 | case 'W': namedclass = ANYOF_NALNUM; break; | |
2760 | case 's': namedclass = ANYOF_SPACE; break; | |
2761 | case 'S': namedclass = ANYOF_NSPACE; break; | |
2762 | case 'd': namedclass = ANYOF_DIGIT; break; | |
2763 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
a0ed51b3 LW |
2764 | case 'p': |
2765 | case 'P': | |
2766 | if (*PL_regcomp_parse == '{') { | |
2767 | e = strchr(PL_regcomp_parse++, '}'); | |
2768 | if (!e) | |
2769 | FAIL("Missing right brace on \\p{}"); | |
2770 | n = e - PL_regcomp_parse; | |
2771 | } | |
2772 | else { | |
2773 | e = PL_regcomp_parse; | |
2774 | n = 1; | |
2775 | } | |
2776 | if (!SIZE_ONLY) { | |
2777 | if (value == 'p') | |
b8c5462f JH |
2778 | Perl_sv_catpvf(aTHX_ listsv, |
2779 | "+utf8::%.*s\n", n, PL_regcomp_parse); | |
a0ed51b3 | 2780 | else |
cea2e8a9 | 2781 | Perl_sv_catpvf(aTHX_ listsv, |
b8c5462f | 2782 | "!utf8::%.*s\n", n, PL_regcomp_parse); |
a0ed51b3 LW |
2783 | } |
2784 | PL_regcomp_parse = e + 1; | |
b8c5462f | 2785 | lastvalue = OOB_UTF8; |
a0ed51b3 | 2786 | continue; |
b8c5462f JH |
2787 | case 'n': value = '\n'; break; |
2788 | case 'r': value = '\r'; break; | |
2789 | case 't': value = '\t'; break; | |
2790 | case 'f': value = '\f'; break; | |
2791 | case 'b': value = '\b'; break; | |
2792 | case 'e': value = '\033'; break; | |
2793 | case 'a': value = '\007'; break; | |
a0ed51b3 LW |
2794 | case 'x': |
2795 | if (*PL_regcomp_parse == '{') { | |
2796 | e = strchr(PL_regcomp_parse++, '}'); | |
2797 | if (!e) | |
2798 | FAIL("Missing right brace on \\x{}"); | |
b8c5462f JH |
2799 | value = scan_hex(PL_regcomp_parse, |
2800 | e - PL_regcomp_parse, | |
2801 | &numlen); | |
a0ed51b3 LW |
2802 | PL_regcomp_parse = e + 1; |
2803 | } | |
2804 | else { | |
2805 | value = scan_hex(PL_regcomp_parse, 2, &numlen); | |
2806 | PL_regcomp_parse += numlen; | |
2807 | } | |
2808 | break; | |
2809 | case 'c': | |
2810 | value = UCHARAT(PL_regcomp_parse++); | |
2811 | value = toCTRL(value); | |
2812 | break; | |
2813 | case '0': case '1': case '2': case '3': case '4': | |
2814 | case '5': case '6': case '7': case '8': case '9': | |
2815 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); | |
2816 | PL_regcomp_parse += numlen; | |
2817 | break; | |
1028017a JH |
2818 | default: |
2819 | if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(value)) | |
2820 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2821 | "/%.127s/: Unrecognized escape \\%c in character class passed through", | |
2822 | PL_regprecomp, | |
2823 | value); | |
2824 | break; | |
a0ed51b3 LW |
2825 | } |
2826 | } | |
73b437c8 JH |
2827 | if (namedclass > OOB_NAMEDCLASS) { |
2828 | if (range) { /* a-\d, a-[:digit:] */ | |
2829 | if (!SIZE_ONLY) { | |
2830 | if (ckWARN(WARN_UNSAFE)) | |
2831 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2832 | "/%.127s/: false [] range \"%*.*s\" in regexp", | |
2833 | PL_regprecomp, | |
2834 | PL_regcomp_parse - rangebegin, | |
2835 | PL_regcomp_parse - rangebegin, | |
2836 | rangebegin); | |
2837 | Perl_sv_catpvf(aTHX_ listsv, | |
2838 | /* 0x002D is Unicode for '-' */ | |
2839 | "%04"UVxf"\n002D\n", (UV)lastvalue); | |
2840 | } | |
2841 | range = 0; | |
2842 | } | |
2843 | if (!SIZE_ONLY) { | |
2844 | switch (namedclass) { | |
2845 | case ANYOF_ALNUM: | |
2846 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n"); break; | |
2847 | case ANYOF_NALNUM: | |
2848 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n"); break; | |
2849 | case ANYOF_ALNUMC: | |
2850 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n"); break; | |
2851 | case ANYOF_NALNUMC: | |
2852 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n"); break; | |
2853 | case ANYOF_ALPHA: | |
2854 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n"); break; | |
2855 | case ANYOF_NALPHA: | |
2856 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n"); break; | |
2857 | case ANYOF_ASCII: | |
2858 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n"); break; | |
2859 | case ANYOF_NASCII: | |
2860 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n"); break; | |
2861 | case ANYOF_CNTRL: | |
2862 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n"); break; | |
2863 | case ANYOF_NCNTRL: | |
2864 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n"); break; | |
2865 | case ANYOF_GRAPH: | |
2866 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n"); break; | |
2867 | case ANYOF_NGRAPH: | |
2868 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n"); break; | |
2869 | case ANYOF_DIGIT: | |
2870 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n"); break; | |
2871 | case ANYOF_NDIGIT: | |
2872 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n"); break; | |
2873 | case ANYOF_LOWER: | |
2874 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n"); break; | |
2875 | case ANYOF_NLOWER: | |
2876 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n"); break; | |
2877 | case ANYOF_PRINT: | |
2878 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n"); break; | |
2879 | case ANYOF_NPRINT: | |
2880 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n"); break; | |
2881 | case ANYOF_PUNCT: | |
2882 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n"); break; | |
2883 | case ANYOF_NPUNCT: | |
2884 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n"); break; | |
2885 | case ANYOF_SPACE: | |
2886 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n"); break; | |
2887 | case ANYOF_NSPACE: | |
2888 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n"); break; | |
2889 | case ANYOF_UPPER: | |
2890 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n"); break; | |
2891 | case ANYOF_NUPPER: | |
2892 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n"); break; | |
2893 | case ANYOF_XDIGIT: | |
2894 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n"); break; | |
2895 | case ANYOF_NXDIGIT: | |
2896 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n"); break; | |
2897 | } | |
2898 | continue; | |
b8c5462f JH |
2899 | } |
2900 | } | |
b8c5462f | 2901 | if (range) { |
73b437c8 JH |
2902 | if (lastvalue > value) { /* b-a */ |
2903 | Perl_croak(aTHX_ | |
2904 | "/%.127s/: invalid [] range \"%*.*s\" in regexp", | |
2905 | PL_regprecomp, | |
2906 | PL_regcomp_parse - rangebegin, | |
2907 | PL_regcomp_parse - rangebegin, | |
2908 | rangebegin); | |
2909 | } | |
a0ed51b3 LW |
2910 | range = 0; |
2911 | } | |
2912 | else { | |
2913 | lastvalue = value; | |
2914 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && | |
93733859 | 2915 | PL_regcomp_parse[1] != ']') { |
a0ed51b3 | 2916 | PL_regcomp_parse++; |
73b437c8 JH |
2917 | if (namedclass > OOB_NAMEDCLASS) { /* \w-, [:word:]- */ |
2918 | if (ckWARN(WARN_UNSAFE)) | |
2919 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2920 | "/%.127s/: false [] range \"%*.*s\" in regexp", | |
2921 | PL_regprecomp, | |
2922 | PL_regcomp_parse - rangebegin, | |
2923 | PL_regcomp_parse - rangebegin, | |
2924 | rangebegin); | |
2925 | if (!SIZE_ONLY) | |
2926 | Perl_sv_catpvf(aTHX_ listsv, | |
2927 | /* 0x002D is Unicode for '-' */ | |
2928 | "002D\n"); | |
2929 | } else | |
2930 | range = 1; | |
a0ed51b3 LW |
2931 | continue; /* do it next time */ |
2932 | } | |
93733859 JH |
2933 | } |
2934 | /* now is the next time */ | |
93733859 | 2935 | if (!SIZE_ONLY) |
73b437c8 JH |
2936 | Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n", |
2937 | (UV)lastvalue, (UV)value); | |
93733859 | 2938 | range = 0; |
a0ed51b3 | 2939 | } |
a0ed51b3 LW |
2940 | |
2941 | ret = reganode(ANYOFUTF8, 0); | |
2942 | ||
2943 | if (!SIZE_ONLY) { | |
2944 | SV *rv = swash_init("utf8", "", listsv, 1, 0); | |
2945 | SvREFCNT_dec(listsv); | |
2946 | n = add_data(1,"s"); | |
2947 | PL_regcomp_rx->data->data[n] = (void*)rv; | |
2948 | ARG1_SET(ret, flags); | |
2949 | ARG2_SET(ret, n); | |
2950 | } | |
2951 | ||
2952 | return ret; | |
2953 | } | |
2954 | ||
76e3520e | 2955 | STATIC char* |
cea2e8a9 | 2956 | S_nextchar(pTHX) |
a0d0e21e | 2957 | { |
5c0ca799 | 2958 | dTHR; |
3280af22 | 2959 | char* retval = PL_regcomp_parse++; |
a0d0e21e | 2960 | |
4633a7c4 | 2961 | for (;;) { |
3280af22 NIS |
2962 | if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' && |
2963 | PL_regcomp_parse[2] == '#') { | |
2964 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') | |
2965 | PL_regcomp_parse++; | |
2966 | PL_regcomp_parse++; | |
4633a7c4 LW |
2967 | continue; |
2968 | } | |
3280af22 NIS |
2969 | if (PL_regflags & PMf_EXTENDED) { |
2970 | if (isSPACE(*PL_regcomp_parse)) { | |
2971 | PL_regcomp_parse++; | |
748a9306 LW |
2972 | continue; |
2973 | } | |
3280af22 NIS |
2974 | else if (*PL_regcomp_parse == '#') { |
2975 | while (*PL_regcomp_parse && *PL_regcomp_parse != '\n') | |
2976 | PL_regcomp_parse++; | |
2977 | PL_regcomp_parse++; | |
748a9306 LW |
2978 | continue; |
2979 | } | |
748a9306 | 2980 | } |
4633a7c4 | 2981 | return retval; |
a0d0e21e | 2982 | } |
a687059c LW |
2983 | } |
2984 | ||
2985 | /* | |
c277df42 | 2986 | - reg_node - emit a node |
a0d0e21e | 2987 | */ |
76e3520e | 2988 | STATIC regnode * /* Location. */ |
cea2e8a9 | 2989 | S_reg_node(pTHX_ U8 op) |
a687059c | 2990 | { |
5c0ca799 | 2991 | dTHR; |
c277df42 IZ |
2992 | register regnode *ret; |
2993 | register regnode *ptr; | |
a687059c | 2994 | |
3280af22 | 2995 | ret = PL_regcode; |
c277df42 | 2996 | if (SIZE_ONLY) { |
6b88bc9c | 2997 | SIZE_ALIGN(PL_regsize); |
3280af22 | 2998 | PL_regsize += 1; |
a0d0e21e LW |
2999 | return(ret); |
3000 | } | |
a687059c | 3001 | |
c277df42 | 3002 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 3003 | ptr = ret; |
c277df42 | 3004 | FILL_ADVANCE_NODE(ptr, op); |
3280af22 | 3005 | PL_regcode = ptr; |
a687059c | 3006 | |
a0d0e21e | 3007 | return(ret); |
a687059c LW |
3008 | } |
3009 | ||
3010 | /* | |
a0d0e21e LW |
3011 | - reganode - emit a node with an argument |
3012 | */ | |
76e3520e | 3013 | STATIC regnode * /* Location. */ |
cea2e8a9 | 3014 | S_reganode(pTHX_ U8 op, U32 arg) |
fe14fcc3 | 3015 | { |
5c0ca799 | 3016 | dTHR; |
c277df42 IZ |
3017 | register regnode *ret; |
3018 | register regnode *ptr; | |
fe14fcc3 | 3019 | |
3280af22 | 3020 | ret = PL_regcode; |
c277df42 | 3021 | if (SIZE_ONLY) { |
6b88bc9c | 3022 | SIZE_ALIGN(PL_regsize); |
3280af22 | 3023 | PL_regsize += 2; |
a0d0e21e LW |
3024 | return(ret); |
3025 | } | |
fe14fcc3 | 3026 | |
c277df42 | 3027 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 3028 | ptr = ret; |
c277df42 | 3029 | FILL_ADVANCE_NODE_ARG(ptr, op, arg); |
3280af22 | 3030 | PL_regcode = ptr; |
fe14fcc3 | 3031 | |
a0d0e21e | 3032 | return(ret); |
fe14fcc3 LW |
3033 | } |
3034 | ||
3035 | /* | |
cd439c50 | 3036 | - reguni - emit (if appropriate) a Unicode character |
a0ed51b3 LW |
3037 | */ |
3038 | STATIC void | |
cea2e8a9 | 3039 | S_reguni(pTHX_ UV uv, char* s, I32* lenp) |
a0ed51b3 | 3040 | { |
c485e607 | 3041 | dTHR; |
a0ed51b3 | 3042 | if (SIZE_ONLY) { |
dfe13c55 | 3043 | U8 tmpbuf[10]; |
a0ed51b3 LW |
3044 | *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf; |
3045 | } | |
3046 | else | |
dfe13c55 | 3047 | *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s; |
a0ed51b3 LW |
3048 | |
3049 | } | |
3050 | ||
3051 | /* | |
a0d0e21e LW |
3052 | - reginsert - insert an operator in front of already-emitted operand |
3053 | * | |
3054 | * Means relocating the operand. | |
3055 | */ | |
76e3520e | 3056 | STATIC void |
cea2e8a9 | 3057 | S_reginsert(pTHX_ U8 op, regnode *opnd) |
a687059c | 3058 | { |
5c0ca799 | 3059 | dTHR; |
c277df42 IZ |
3060 | register regnode *src; |
3061 | register regnode *dst; | |
3062 | register regnode *place; | |
3063 | register int offset = regarglen[(U8)op]; | |
3064 | ||
22c35a8c | 3065 | /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */ |
c277df42 IZ |
3066 | |
3067 | if (SIZE_ONLY) { | |
3280af22 | 3068 | PL_regsize += NODE_STEP_REGNODE + offset; |
a0d0e21e LW |
3069 | return; |
3070 | } | |
a687059c | 3071 | |
3280af22 NIS |
3072 | src = PL_regcode; |
3073 | PL_regcode += NODE_STEP_REGNODE + offset; | |
3074 | dst = PL_regcode; | |
a0d0e21e | 3075 | while (src > opnd) |
c277df42 | 3076 | StructCopy(--src, --dst, regnode); |
a0d0e21e LW |
3077 | |
3078 | place = opnd; /* Op node, where operand used to be. */ | |
c277df42 IZ |
3079 | src = NEXTOPER(place); |
3080 | FILL_ADVANCE_NODE(place, op); | |
3081 | Zero(src, offset, regnode); | |
a687059c LW |
3082 | } |
3083 | ||
3084 | /* | |
c277df42 | 3085 | - regtail - set the next-pointer at the end of a node chain of p to val. |
a0d0e21e | 3086 | */ |
76e3520e | 3087 | STATIC void |
cea2e8a9 | 3088 | S_regtail(pTHX_ regnode *p, regnode *val) |
a687059c | 3089 | { |
5c0ca799 | 3090 | dTHR; |
c277df42 IZ |
3091 | register regnode *scan; |
3092 | register regnode *temp; | |
a0d0e21e LW |
3093 | register I32 offset; |
3094 | ||
c277df42 | 3095 | if (SIZE_ONLY) |
a0d0e21e LW |
3096 | return; |
3097 | ||
3098 | /* Find last node. */ | |
3099 | scan = p; | |
3100 | for (;;) { | |
3101 | temp = regnext(scan); | |
3102 | if (temp == NULL) | |
3103 | break; | |
3104 | scan = temp; | |
3105 | } | |
a687059c | 3106 | |
c277df42 IZ |
3107 | if (reg_off_by_arg[OP(scan)]) { |
3108 | ARG_SET(scan, val - scan); | |
a0ed51b3 LW |
3109 | } |
3110 | else { | |
c277df42 IZ |
3111 | NEXT_OFF(scan) = val - scan; |
3112 | } | |
a687059c LW |
3113 | } |
3114 | ||
3115 | /* | |
a0d0e21e LW |
3116 | - regoptail - regtail on operand of first argument; nop if operandless |
3117 | */ | |
76e3520e | 3118 | STATIC void |
cea2e8a9 | 3119 | S_regoptail(pTHX_ regnode *p, regnode *val) |
a687059c | 3120 | { |
5c0ca799 | 3121 | dTHR; |
a0d0e21e | 3122 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
c277df42 IZ |
3123 | if (p == NULL || SIZE_ONLY) |
3124 | return; | |
22c35a8c | 3125 | if (PL_regkind[(U8)OP(p)] == BRANCH) { |
c277df42 | 3126 | regtail(NEXTOPER(p), val); |
a0ed51b3 | 3127 | } |
22c35a8c | 3128 | else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) { |
c277df42 | 3129 | regtail(NEXTOPER(NEXTOPER(p)), val); |
a0ed51b3 LW |
3130 | } |
3131 | else | |
a0d0e21e | 3132 | return; |
a687059c LW |
3133 | } |
3134 | ||
3135 | /* | |
3136 | - regcurly - a little FSA that accepts {\d+,?\d*} | |
3137 | */ | |
79072805 | 3138 | STATIC I32 |
cea2e8a9 | 3139 | S_regcurly(pTHX_ register char *s) |
a687059c LW |
3140 | { |
3141 | if (*s++ != '{') | |
3142 | return FALSE; | |
f0fcb552 | 3143 | if (!isDIGIT(*s)) |
a687059c | 3144 | return FALSE; |
f0fcb552 | 3145 | while (isDIGIT(*s)) |
a687059c LW |
3146 | s++; |
3147 | if (*s == ',') | |
3148 | s++; | |
f0fcb552 | 3149 | while (isDIGIT(*s)) |
a687059c LW |
3150 | s++; |
3151 | if (*s != '}') | |
3152 | return FALSE; | |
3153 | return TRUE; | |
3154 | } | |
3155 | ||
a687059c | 3156 | |
76e3520e | 3157 | STATIC regnode * |
cea2e8a9 | 3158 | S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l) |
c277df42 | 3159 | { |
35ff7856 | 3160 | #ifdef DEBUGGING |
f248d071 | 3161 | register U8 op = EXACT; /* Arbitrary non-END op. */ |
c277df42 IZ |
3162 | register regnode *next, *onode; |
3163 | ||
3164 | while (op != END && (!last || node < last)) { | |
3165 | /* While that wasn't END last time... */ | |
3166 | ||
3167 | NODE_ALIGN(node); | |
3168 | op = OP(node); | |
3169 | if (op == CLOSE) | |
3170 | l--; | |
3171 | next = regnext(node); | |
3172 | /* Where, what. */ | |
3173 | if (OP(node) == OPTIMIZED) | |
3174 | goto after_print; | |
3175 | regprop(sv, node); | |
54dc92de | 3176 | PerlIO_printf(Perl_debug_log, "%4d:%*s%s", node - start, |
c277df42 IZ |
3177 | 2*l + 1, "", SvPVX(sv)); |
3178 | if (next == NULL) /* Next ptr. */ | |
3179 | PerlIO_printf(Perl_debug_log, "(0)"); | |
3180 | else | |
3181 | PerlIO_printf(Perl_debug_log, "(%d)", next - start); | |
3182 | (void)PerlIO_putc(Perl_debug_log, '\n'); | |
3183 | after_print: | |
22c35a8c | 3184 | if (PL_regkind[(U8)op] == BRANCHJ) { |
c277df42 IZ |
3185 | register regnode *nnode = (OP(next) == LONGJMP |
3186 | ? regnext(next) | |
3187 | : next); | |
3188 | if (last && nnode > last) | |
3189 | nnode = last; | |
3190 | node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1); | |
a0ed51b3 | 3191 | } |
22c35a8c | 3192 | else if (PL_regkind[(U8)op] == BRANCH) { |
c277df42 | 3193 | node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1); |
a0ed51b3 LW |
3194 | } |
3195 | else if ( op == CURLY) { /* `next' might be very big: optimizer */ | |
c277df42 IZ |
3196 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3197 | NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1); | |
a0ed51b3 | 3198 | } |
22c35a8c | 3199 | else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) { |
c277df42 IZ |
3200 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3201 | next, sv, l + 1); | |
a0ed51b3 LW |
3202 | } |
3203 | else if ( op == PLUS || op == STAR) { | |
c277df42 | 3204 | node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1); |
a0ed51b3 LW |
3205 | } |
3206 | else if (op == ANYOF) { | |
c277df42 IZ |
3207 | node = NEXTOPER(node); |
3208 | node += ANY_SKIP; | |
a0ed51b3 | 3209 | } |
22c35a8c | 3210 | else if (PL_regkind[(U8)op] == EXACT) { |
c277df42 | 3211 | /* Literal string, where present. */ |
cd439c50 | 3212 | node += NODE_SZ_STR(node) - 1; |
c277df42 | 3213 | node = NEXTOPER(node); |
a0ed51b3 LW |
3214 | } |
3215 | else { | |
c277df42 IZ |
3216 | node = NEXTOPER(node); |
3217 | node += regarglen[(U8)op]; | |
3218 | } | |
3219 | if (op == CURLYX || op == OPEN) | |
3220 | l++; | |
3221 | else if (op == WHILEM) | |
3222 | l--; | |
3223 | } | |
17c3b450 | 3224 | #endif /* DEBUGGING */ |
c277df42 IZ |
3225 | return node; |
3226 | } | |
3227 | ||
a687059c | 3228 | /* |
fd181c75 | 3229 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c LW |
3230 | */ |
3231 | void | |
864dbfa3 | 3232 | Perl_regdump(pTHX_ regexp *r) |
a687059c | 3233 | { |
35ff7856 | 3234 | #ifdef DEBUGGING |
5c0ca799 | 3235 | dTHR; |
46fc3d4c | 3236 | SV *sv = sv_newmortal(); |
a687059c | 3237 | |
c277df42 | 3238 | (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0); |
a0d0e21e LW |
3239 | |
3240 | /* Header fields of interest. */ | |
c277df42 | 3241 | if (r->anchored_substr) |
2c2d71f5 | 3242 | PerlIO_printf(Perl_debug_log, "anchored `%s%.*s%s'%s at %d ", |
3280af22 | 3243 | PL_colors[0], |
2c2d71f5 | 3244 | SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0), |
c277df42 | 3245 | SvPVX(r->anchored_substr), |
3280af22 | 3246 | PL_colors[1], |
c277df42 IZ |
3247 | SvTAIL(r->anchored_substr) ? "$" : "", |
3248 | r->anchored_offset); | |
3249 | if (r->float_substr) | |
2c2d71f5 | 3250 | PerlIO_printf(Perl_debug_log, "floating `%s%.*s%s'%s at %d..%u ", |
3280af22 | 3251 | PL_colors[0], |
2c2d71f5 JH |
3252 | SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0), |
3253 | SvPVX(r->float_substr), | |
3280af22 | 3254 | PL_colors[1], |
c277df42 IZ |
3255 | SvTAIL(r->float_substr) ? "$" : "", |
3256 | r->float_min_offset, r->float_max_offset); | |
3257 | if (r->check_substr) | |
3258 | PerlIO_printf(Perl_debug_log, | |
3259 | r->check_substr == r->float_substr | |
3260 | ? "(checking floating" : "(checking anchored"); | |
3261 | if (r->reganch & ROPT_NOSCAN) | |
3262 | PerlIO_printf(Perl_debug_log, " noscan"); | |
3263 | if (r->reganch & ROPT_CHECK_ALL) | |
3264 | PerlIO_printf(Perl_debug_log, " isall"); | |
3265 | if (r->check_substr) | |
3266 | PerlIO_printf(Perl_debug_log, ") "); | |
3267 | ||
46fc3d4c | 3268 | if (r->regstclass) { |
3269 | regprop(sv, r->regstclass); | |
3270 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv)); | |
3271 | } | |
774d564b | 3272 | if (r->reganch & ROPT_ANCH) { |
3273 | PerlIO_printf(Perl_debug_log, "anchored"); | |
3274 | if (r->reganch & ROPT_ANCH_BOL) | |
3275 | PerlIO_printf(Perl_debug_log, "(BOL)"); | |
c277df42 IZ |
3276 | if (r->reganch & ROPT_ANCH_MBOL) |
3277 | PerlIO_printf(Perl_debug_log, "(MBOL)"); | |
cad2e5aa JH |
3278 | if (r->reganch & ROPT_ANCH_SBOL) |
3279 | PerlIO_printf(Perl_debug_log, "(SBOL)"); | |
774d564b | 3280 | if (r->reganch & ROPT_ANCH_GPOS) |
3281 | PerlIO_printf(Perl_debug_log, "(GPOS)"); | |
3282 | PerlIO_putc(Perl_debug_log, ' '); | |
3283 | } | |
c277df42 IZ |
3284 | if (r->reganch & ROPT_GPOS_SEEN) |
3285 | PerlIO_printf(Perl_debug_log, "GPOS "); | |
a0d0e21e | 3286 | if (r->reganch & ROPT_SKIP) |
760ac839 | 3287 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e | 3288 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 | 3289 | PerlIO_printf(Perl_debug_log, "implicit "); |
760ac839 | 3290 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
ce862d02 IZ |
3291 | if (r->reganch & ROPT_EVAL_SEEN) |
3292 | PerlIO_printf(Perl_debug_log, "with eval "); | |
760ac839 | 3293 | PerlIO_printf(Perl_debug_log, "\n"); |
17c3b450 | 3294 | #endif /* DEBUGGING */ |
a687059c LW |
3295 | } |
3296 | ||
3297 | /* | |
a0d0e21e LW |
3298 | - regprop - printable representation of opcode |
3299 | */ | |
46fc3d4c | 3300 | void |
864dbfa3 | 3301 | Perl_regprop(pTHX_ SV *sv, regnode *o) |
a687059c | 3302 | { |
35ff7856 | 3303 | #ifdef DEBUGGING |
5c0ca799 | 3304 | dTHR; |
9b155405 | 3305 | register int k; |
a0d0e21e | 3306 | |
54dc92de | 3307 | sv_setpvn(sv, "", 0); |
9b155405 | 3308 | if (OP(o) >= reg_num) /* regnode.type is unsigned */ |
a0d0e21e | 3309 | FAIL("corrupted regexp opcode"); |
9b155405 IZ |
3310 | sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */ |
3311 | ||
3312 | k = PL_regkind[(U8)OP(o)]; | |
3313 | ||
3314 | if (k == EXACT) | |
7821416a | 3315 | Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0], |
cd439c50 | 3316 | STR_LEN(o), STRING(o), PL_colors[1]); |
9b155405 IZ |
3317 | else if (k == CURLY) { |
3318 | if (OP(o) == CURLYM || OP(o) == CURLYN) | |
cea2e8a9 GS |
3319 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */ |
3320 | Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o)); | |
a0d0e21e | 3321 | } |
2c2d71f5 JH |
3322 | else if (k == WHILEM && o->flags) /* Ordinal/of */ |
3323 | Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4); | |
9b155405 | 3324 | else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP ) |
cea2e8a9 | 3325 | Perl_sv_catpvf(aTHX_ sv, "%d", ARG(o)); /* Parenth number */ |
9b155405 | 3326 | else if (k == LOGICAL) |
04ebc1ab | 3327 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* 2: embedded, otherwise 1 */ |
9b155405 | 3328 | else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH)) |
cea2e8a9 | 3329 | Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags); |
17c3b450 | 3330 | #endif /* DEBUGGING */ |
35ff7856 | 3331 | } |
a687059c | 3332 | |
cad2e5aa JH |
3333 | SV * |
3334 | Perl_re_intuit_string(pTHX_ regexp *prog) | |
3335 | { /* Assume that RE_INTUIT is set */ | |
3336 | DEBUG_r( | |
3337 | { STRLEN n_a; | |
3338 | char *s = SvPV(prog->check_substr,n_a); | |
3339 | ||
3340 | if (!PL_colorset) reginitcolors(); | |
3341 | PerlIO_printf(Perl_debug_log, | |
3342 | "%sUsing REx substr:%s `%s%.60s%s%s'\n", | |
3343 | PL_colors[4],PL_colors[5],PL_colors[0], | |
3344 | s, | |
3345 | PL_colors[1], | |
3346 | (strlen(s) > 60 ? "..." : "")); | |
3347 | } ); | |
3348 | ||
3349 | return prog->check_substr; | |
3350 | } | |
3351 | ||
2b69d0c2 | 3352 | void |
864dbfa3 | 3353 | Perl_pregfree(pTHX_ struct regexp *r) |
a687059c | 3354 | { |
5c0ca799 | 3355 | dTHR; |
adac82c7 | 3356 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
7821416a IZ |
3357 | |
3358 | if (!r || (--r->refcnt > 0)) | |
3359 | return; | |
cad2e5aa JH |
3360 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
3361 | "%sFreeing REx:%s `%s%.60s%s%s'\n", | |
3362 | PL_colors[4],PL_colors[5],PL_colors[0], | |
3363 | r->precomp, | |
3364 | PL_colors[1], | |
3365 | (strlen(r->precomp) > 60 ? "..." : ""))); | |
3366 | ||
c277df42 | 3367 | if (r->precomp) |
a0d0e21e | 3368 | Safefree(r->precomp); |
cf93c79d IZ |
3369 | if (RX_MATCH_COPIED(r)) |
3370 | Safefree(r->subbeg); | |
a193d654 GS |
3371 | if (r->substrs) { |
3372 | if (r->anchored_substr) | |
3373 | SvREFCNT_dec(r->anchored_substr); | |
3374 | if (r->float_substr) | |
3375 | SvREFCNT_dec(r->float_substr); | |
2779dcf1 | 3376 | Safefree(r->substrs); |
a193d654 | 3377 | } |
c277df42 IZ |
3378 | if (r->data) { |
3379 | int n = r->data->count; | |
dfad63ad HS |
3380 | AV* new_comppad = NULL; |
3381 | AV* old_comppad; | |
3382 | SV** old_curpad; | |
3383 | ||
c277df42 IZ |
3384 | while (--n >= 0) { |
3385 | switch (r->data->what[n]) { | |
3386 | case 's': | |
3387 | SvREFCNT_dec((SV*)r->data->data[n]); | |
3388 | break; | |
dfad63ad HS |
3389 | case 'p': |
3390 | new_comppad = (AV*)r->data->data[n]; | |
3391 | break; | |
c277df42 | 3392 | case 'o': |
dfad63ad | 3393 | if (new_comppad == NULL) |
cea2e8a9 | 3394 | Perl_croak(aTHX_ "panic: pregfree comppad"); |
dfad63ad HS |
3395 | old_comppad = PL_comppad; |
3396 | old_curpad = PL_curpad; | |
3397 | PL_comppad = new_comppad; | |
3398 | PL_curpad = AvARRAY(new_comppad); | |
c277df42 | 3399 | op_free((OP_4tree*)r->data->data[n]); |
dfad63ad HS |
3400 | PL_comppad = old_comppad; |
3401 | PL_curpad = old_curpad; | |
3402 | SvREFCNT_dec((SV*)new_comppad); | |
3403 | new_comppad = NULL; | |
c277df42 IZ |
3404 | break; |
3405 | case 'n': | |
3406 | break; | |
3407 | default: | |
3408 | FAIL2("panic: regfree data code '%c'", r->data->what[n]); | |
3409 | } | |
3410 | } | |
3411 | Safefree(r->data->what); | |
3412 | Safefree(r->data); | |
a0d0e21e LW |
3413 | } |
3414 | Safefree(r->startp); | |
3415 | Safefree(r->endp); | |
3416 | Safefree(r); | |
a687059c | 3417 | } |
c277df42 IZ |
3418 | |
3419 | /* | |
3420 | - regnext - dig the "next" pointer out of a node | |
3421 | * | |
3422 | * [Note, when REGALIGN is defined there are two places in regmatch() | |
3423 | * that bypass this code for speed.] | |
3424 | */ | |
3425 | regnode * | |
864dbfa3 | 3426 | Perl_regnext(pTHX_ register regnode *p) |
c277df42 | 3427 | { |
5c0ca799 | 3428 | dTHR; |
c277df42 IZ |
3429 | register I32 offset; |
3430 | ||
3280af22 | 3431 | if (p == &PL_regdummy) |
c277df42 IZ |
3432 | return(NULL); |
3433 | ||
3434 | offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p)); | |
3435 | if (offset == 0) | |
3436 | return(NULL); | |
3437 | ||
c277df42 | 3438 | return(p+offset); |
c277df42 IZ |
3439 | } |
3440 | ||
01f988be | 3441 | STATIC void |
cea2e8a9 | 3442 | S_re_croak2(pTHX_ const char* pat1,const char* pat2,...) |
c277df42 IZ |
3443 | { |
3444 | va_list args; | |
3445 | STRLEN l1 = strlen(pat1); | |
3446 | STRLEN l2 = strlen(pat2); | |
3447 | char buf[512]; | |
06bf62c7 | 3448 | SV *msv; |
c277df42 IZ |
3449 | char *message; |
3450 | ||
3451 | if (l1 > 510) | |
3452 | l1 = 510; | |
3453 | if (l1 + l2 > 510) | |
3454 | l2 = 510 - l1; | |
3455 | Copy(pat1, buf, l1 , char); | |
3456 | Copy(pat2, buf + l1, l2 , char); | |
3b818b81 GS |
3457 | buf[l1 + l2] = '\n'; |
3458 | buf[l1 + l2 + 1] = '\0'; | |
8736538c AS |
3459 | #ifdef I_STDARG |
3460 | /* ANSI variant takes additional second argument */ | |
c277df42 | 3461 | va_start(args, pat2); |
8736538c AS |
3462 | #else |
3463 | va_start(args); | |
3464 | #endif | |
5a844595 | 3465 | msv = vmess(buf, &args); |
c277df42 | 3466 | va_end(args); |
06bf62c7 | 3467 | message = SvPV(msv,l1); |
c277df42 IZ |
3468 | if (l1 > 512) |
3469 | l1 = 512; | |
3470 | Copy(message, buf, l1 , char); | |
3471 | buf[l1] = '\0'; /* Overwrite \n */ | |
cea2e8a9 | 3472 | Perl_croak(aTHX_ "%s", buf); |
c277df42 | 3473 | } |
a0ed51b3 LW |
3474 | |
3475 | /* XXX Here's a total kludge. But we need to re-enter for swash routines. */ | |
3476 | ||
3477 | void | |
864dbfa3 | 3478 | Perl_save_re_context(pTHX) |
c485e607 NIS |
3479 | { |
3480 | dTHR; | |
a0ed51b3 LW |
3481 | SAVEPPTR(PL_bostr); |
3482 | SAVEPPTR(PL_regprecomp); /* uncompiled string. */ | |
3483 | SAVEI32(PL_regnpar); /* () count. */ | |
3484 | SAVEI32(PL_regsize); /* Code size. */ | |
3485 | SAVEI16(PL_regflags); /* are we folding, multilining? */ | |
3486 | SAVEPPTR(PL_reginput); /* String-input pointer. */ | |
3487 | SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */ | |
3488 | SAVEPPTR(PL_regeol); /* End of input, for $ check. */ | |
3489 | SAVESPTR(PL_regstartp); /* Pointer to startp array. */ | |
3490 | SAVESPTR(PL_regendp); /* Ditto for endp. */ | |
3491 | SAVESPTR(PL_reglastparen); /* Similarly for lastparen. */ | |
3492 | SAVEPPTR(PL_regtill); /* How far we are required to go. */ | |
3493 | SAVEI32(PL_regprev); /* char before regbol, \n if none */ | |
3494 | SAVESPTR(PL_reg_start_tmp); /* from regexec.c */ | |
3495 | PL_reg_start_tmp = 0; | |
3496 | SAVEFREEPV(PL_reg_start_tmp); | |
3497 | SAVEI32(PL_reg_start_tmpl); /* from regexec.c */ | |
3498 | PL_reg_start_tmpl = 0; | |
3499 | SAVESPTR(PL_regdata); | |
3500 | SAVEI32(PL_reg_flags); /* from regexec.c */ | |
3501 | SAVEI32(PL_reg_eval_set); /* from regexec.c */ | |
3502 | SAVEI32(PL_regnarrate); /* from regexec.c */ | |
3503 | SAVESPTR(PL_regprogram); /* from regexec.c */ | |
3504 | SAVEINT(PL_regindent); /* from regexec.c */ | |
3505 | SAVESPTR(PL_regcc); /* from regexec.c */ | |
3506 | SAVESPTR(PL_curcop); | |
3507 | SAVESPTR(PL_regcomp_rx); /* from regcomp.c */ | |
3508 | SAVEI32(PL_regseen); /* from regcomp.c */ | |
3509 | SAVEI32(PL_regsawback); /* Did we see \1, ...? */ | |
3510 | SAVEI32(PL_regnaughty); /* How bad is this pattern? */ | |
3511 | SAVESPTR(PL_regcode); /* Code-emit pointer; ®dummy = don't */ | |
3512 | SAVEPPTR(PL_regxend); /* End of input for compile */ | |
3513 | SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */ | |
54b6e2fa IZ |
3514 | SAVESPTR(PL_reg_call_cc); /* from regexec.c */ |
3515 | SAVESPTR(PL_reg_re); /* from regexec.c */ | |
3516 | SAVEPPTR(PL_reg_ganch); /* from regexec.c */ | |
3517 | SAVESPTR(PL_reg_sv); /* from regexec.c */ | |
3518 | SAVESPTR(PL_reg_magic); /* from regexec.c */ | |
3519 | SAVEI32(PL_reg_oldpos); /* from regexec.c */ | |
3520 | SAVESPTR(PL_reg_oldcurpm); /* from regexec.c */ | |
3521 | SAVESPTR(PL_reg_curpm); /* from regexec.c */ | |
3522 | #ifdef DEBUGGING | |
3523 | SAVEPPTR(PL_reg_starttry); /* from regexec.c */ | |
3524 | #endif | |
a0ed51b3 | 3525 | } |
51371543 GS |
3526 | |
3527 | #ifdef PERL_OBJECT | |
3528 | #define NO_XSLOCKS | |
3529 | #include "XSUB.h" | |
3530 | #undef this | |
3531 | #define this pPerl | |
3532 | #endif | |
3533 | ||
3534 | static void | |
3535 | clear_re(pTHXo_ void *r) | |
3536 | { | |
3537 | ReREFCNT_dec((regexp *)r); | |
3538 | } | |
3539 |