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) { | |
0f4b6630 JH |
705 | #ifdef UV_IS_QUAD |
706 | FAIL2("lookbehind longer than %" PERL_PRIu64 " not implemented", (UV)U8_MAX); | |
707 | #else | |
c277df42 | 708 | FAIL2("lookbehind longer than %d not implemented", U8_MAX); |
0f4b6630 | 709 | #endif |
c277df42 IZ |
710 | } |
711 | scan->flags = minnext; | |
712 | } | |
713 | if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
714 | pars++; | |
405ff068 | 715 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 | 716 | data->flags |= SF_HAS_EVAL; |
2c2d71f5 JH |
717 | if (data) |
718 | data->whilem_c = data_fake.whilem_c; | |
a0ed51b3 LW |
719 | } |
720 | else if (OP(scan) == OPEN) { | |
c277df42 | 721 | pars++; |
a0ed51b3 LW |
722 | } |
723 | else if (OP(scan) == CLOSE && ARG(scan) == is_par) { | |
c277df42 IZ |
724 | next = regnext(scan); |
725 | ||
726 | if ( next && (OP(next) != WHILEM) && next < last) | |
c277df42 | 727 | is_par = 0; /* Disable optimization */ |
a0ed51b3 LW |
728 | } |
729 | else if (OP(scan) == EVAL) { | |
c277df42 IZ |
730 | if (data) |
731 | data->flags |= SF_HAS_EVAL; | |
732 | } | |
0f5d15d6 IZ |
733 | else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */ |
734 | if (flags & SCF_DO_SUBSTR) { | |
735 | scan_commit(data); | |
736 | data->longest = &(data->longest_float); | |
737 | } | |
738 | is_inf = is_inf_internal = 1; | |
739 | } | |
c277df42 IZ |
740 | /* Else: zero-length, ignore. */ |
741 | scan = regnext(scan); | |
742 | } | |
743 | ||
744 | finish: | |
745 | *scanp = scan; | |
aca2d497 | 746 | *deltap = is_inf_internal ? I32_MAX : delta; |
c277df42 IZ |
747 | if (flags & SCF_DO_SUBSTR && is_inf) |
748 | data->pos_delta = I32_MAX - data->pos_min; | |
749 | if (is_par > U8_MAX) | |
750 | is_par = 0; | |
751 | if (is_par && pars==1 && data) { | |
752 | data->flags |= SF_IN_PAR; | |
753 | data->flags &= ~SF_HAS_PAR; | |
a0ed51b3 LW |
754 | } |
755 | else if (pars && data) { | |
c277df42 IZ |
756 | data->flags |= SF_HAS_PAR; |
757 | data->flags &= ~SF_IN_PAR; | |
758 | } | |
759 | return min; | |
760 | } | |
761 | ||
76e3520e | 762 | STATIC I32 |
cea2e8a9 | 763 | S_add_data(pTHX_ I32 n, char *s) |
c277df42 | 764 | { |
5c0ca799 | 765 | dTHR; |
3280af22 NIS |
766 | if (PL_regcomp_rx->data) { |
767 | Renewc(PL_regcomp_rx->data, | |
768 | sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1), | |
c277df42 | 769 | char, struct reg_data); |
3280af22 NIS |
770 | Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8); |
771 | PL_regcomp_rx->data->count += n; | |
a0ed51b3 LW |
772 | } |
773 | else { | |
3280af22 | 774 | Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1), |
c277df42 | 775 | char, struct reg_data); |
3280af22 NIS |
776 | New(1208, PL_regcomp_rx->data->what, n, U8); |
777 | PL_regcomp_rx->data->count = n; | |
c277df42 | 778 | } |
3280af22 NIS |
779 | Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8); |
780 | return PL_regcomp_rx->data->count - n; | |
c277df42 IZ |
781 | } |
782 | ||
d88dccdf | 783 | void |
864dbfa3 | 784 | Perl_reginitcolors(pTHX) |
d88dccdf IZ |
785 | { |
786 | dTHR; | |
787 | int i = 0; | |
788 | char *s = PerlEnv_getenv("PERL_RE_COLORS"); | |
789 | ||
790 | if (s) { | |
791 | PL_colors[0] = s = savepv(s); | |
792 | while (++i < 6) { | |
793 | s = strchr(s, '\t'); | |
794 | if (s) { | |
795 | *s = '\0'; | |
796 | PL_colors[i] = ++s; | |
797 | } | |
798 | else | |
c712d376 | 799 | PL_colors[i] = s = ""; |
d88dccdf IZ |
800 | } |
801 | } else { | |
802 | while (i < 6) | |
803 | PL_colors[i++] = ""; | |
804 | } | |
805 | PL_colorset = 1; | |
806 | } | |
807 | ||
a687059c | 808 | /* |
e50aee73 | 809 | - pregcomp - compile a regular expression into internal code |
a687059c LW |
810 | * |
811 | * We can't allocate space until we know how big the compiled form will be, | |
812 | * but we can't compile it (and thus know how big it is) until we've got a | |
813 | * place to put the code. So we cheat: we compile it twice, once with code | |
814 | * generation turned off and size counting turned on, and once "for real". | |
815 | * This also means that we don't allocate space until we are sure that the | |
816 | * thing really will compile successfully, and we never have to move the | |
817 | * code and thus invalidate pointers into it. (Note that it has to be in | |
818 | * one piece because free() must be able to free it all.) [NB: not true in perl] | |
819 | * | |
820 | * Beware that the optimization-preparation code in here knows about some | |
821 | * of the structure of the compiled regexp. [I'll say.] | |
822 | */ | |
823 | regexp * | |
864dbfa3 | 824 | Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm) |
a687059c | 825 | { |
5c0ca799 | 826 | dTHR; |
a0d0e21e | 827 | register regexp *r; |
c277df42 IZ |
828 | regnode *scan; |
829 | SV **longest; | |
830 | SV *longest_fixed; | |
831 | SV *longest_float; | |
832 | regnode *first; | |
a0d0e21e | 833 | I32 flags; |
a0d0e21e LW |
834 | I32 minlen = 0; |
835 | I32 sawplus = 0; | |
836 | I32 sawopen = 0; | |
2c2d71f5 | 837 | scan_data_t data; |
a0d0e21e LW |
838 | |
839 | if (exp == NULL) | |
c277df42 | 840 | FAIL("NULL regexp argument"); |
a0d0e21e | 841 | |
e24b16f9 | 842 | if (PL_curcop == &PL_compiling ? (PL_hints & HINT_UTF8) : IN_UTF8) |
a0ed51b3 LW |
843 | PL_reg_flags |= RF_utf8; |
844 | else | |
845 | PL_reg_flags = 0; | |
846 | ||
3280af22 | 847 | PL_regprecomp = savepvn(exp, xend - exp); |
35ef4773 | 848 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
2c2d71f5 | 849 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n", |
d88dccdf IZ |
850 | PL_colors[4],PL_colors[5],PL_colors[0], |
851 | xend - exp, PL_regprecomp, PL_colors[1])); | |
3280af22 NIS |
852 | PL_regflags = pm->op_pmflags; |
853 | PL_regsawback = 0; | |
bbce6d69 | 854 | |
3280af22 NIS |
855 | PL_regseen = 0; |
856 | PL_seen_zerolen = *exp == '^' ? -1 : 0; | |
857 | PL_seen_evals = 0; | |
858 | PL_extralen = 0; | |
c277df42 | 859 | |
bbce6d69 | 860 | /* First pass: determine size, legality. */ |
3280af22 NIS |
861 | PL_regcomp_parse = exp; |
862 | PL_regxend = xend; | |
863 | PL_regnaughty = 0; | |
864 | PL_regnpar = 1; | |
865 | PL_regsize = 0L; | |
866 | PL_regcode = &PL_regdummy; | |
2c2d71f5 | 867 | PL_reg_whilem_seen = 0; |
cd439c50 | 868 | REGC((U8)REG_MAGIC, (char*)PL_regcode); |
a0d0e21e | 869 | if (reg(0, &flags) == NULL) { |
3280af22 NIS |
870 | Safefree(PL_regprecomp); |
871 | PL_regprecomp = Nullch; | |
a0d0e21e LW |
872 | return(NULL); |
873 | } | |
3280af22 | 874 | DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", PL_regsize)); |
c277df42 | 875 | |
c277df42 IZ |
876 | /* Small enough for pointer-storage convention? |
877 | If extralen==0, this means that we will not need long jumps. */ | |
3280af22 NIS |
878 | if (PL_regsize >= 0x10000L && PL_extralen) |
879 | PL_regsize += PL_extralen; | |
c277df42 | 880 | else |
3280af22 | 881 | PL_extralen = 0; |
2c2d71f5 JH |
882 | if (PL_reg_whilem_seen > 15) |
883 | PL_reg_whilem_seen = 15; | |
a0d0e21e | 884 | |
bbce6d69 | 885 | /* Allocate space and initialize. */ |
3280af22 | 886 | Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), |
c277df42 | 887 | char, regexp); |
a0d0e21e LW |
888 | if (r == NULL) |
889 | FAIL("regexp out of space"); | |
c277df42 | 890 | r->refcnt = 1; |
bbce6d69 | 891 | r->prelen = xend - exp; |
3280af22 | 892 | r->precomp = PL_regprecomp; |
cf93c79d IZ |
893 | r->subbeg = NULL; |
894 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; | |
4327152a IZ |
895 | r->nparens = PL_regnpar - 1; /* set early to validate backrefs */ |
896 | ||
897 | r->substrs = 0; /* Useful during FAIL. */ | |
898 | r->startp = 0; /* Useful during FAIL. */ | |
899 | r->endp = 0; /* Useful during FAIL. */ | |
900 | ||
3280af22 | 901 | PL_regcomp_rx = r; |
bbce6d69 | 902 | |
903 | /* Second pass: emit code. */ | |
3280af22 NIS |
904 | PL_regcomp_parse = exp; |
905 | PL_regxend = xend; | |
906 | PL_regnaughty = 0; | |
907 | PL_regnpar = 1; | |
908 | PL_regcode = r->program; | |
2cd61cdb | 909 | /* Store the count of eval-groups for security checks: */ |
3280af22 | 910 | PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals); |
cd439c50 | 911 | REGC((U8)REG_MAGIC, (char*) PL_regcode++); |
c277df42 | 912 | r->data = 0; |
a0d0e21e LW |
913 | if (reg(0, &flags) == NULL) |
914 | return(NULL); | |
915 | ||
916 | /* Dig out information for optimizations. */ | |
cf93c79d | 917 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */ |
3280af22 | 918 | pm->op_pmflags = PL_regflags; |
a0ed51b3 LW |
919 | if (UTF) |
920 | r->reganch |= ROPT_UTF8; | |
c277df42 | 921 | r->regstclass = NULL; |
a0ed51b3 LW |
922 | if (PL_regnaughty >= 10) /* Probably an expensive pattern. */ |
923 | r->reganch |= ROPT_NAUGHTY; | |
c277df42 | 924 | scan = r->program + 1; /* First BRANCH. */ |
2779dcf1 IZ |
925 | |
926 | /* XXXX To minimize changes to RE engine we always allocate | |
927 | 3-units-long substrs field. */ | |
928 | Newz(1004, r->substrs, 1, struct reg_substr_data); | |
929 | ||
2c2d71f5 | 930 | StructCopy(&zero_scan_data, &data, scan_data_t); |
c277df42 | 931 | if (OP(scan) != BRANCH) { /* Only one top-level choice. */ |
c277df42 | 932 | I32 fake; |
c5254dd6 | 933 | STRLEN longest_float_length, longest_fixed_length; |
a0d0e21e LW |
934 | |
935 | first = scan; | |
c277df42 | 936 | /* Skip introductions and multiplicators >= 1. */ |
a0d0e21e LW |
937 | while ((OP(first) == OPEN && (sawopen = 1)) || |
938 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || | |
939 | (OP(first) == PLUS) || | |
940 | (OP(first) == MINMOD) || | |
22c35a8c | 941 | (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
a0d0e21e LW |
942 | if (OP(first) == PLUS) |
943 | sawplus = 1; | |
944 | else | |
945 | first += regarglen[(U8)OP(first)]; | |
946 | first = NEXTOPER(first); | |
a687059c LW |
947 | } |
948 | ||
a0d0e21e LW |
949 | /* Starting-point info. */ |
950 | again: | |
c277df42 | 951 | if (OP(first) == EXACT); /* Empty, get anchored substr later. */ |
4d61ec05 | 952 | else if (strchr((char*)PL_simple+4,OP(first))) |
a0d0e21e | 953 | r->regstclass = first; |
22c35a8c GS |
954 | else if (PL_regkind[(U8)OP(first)] == BOUND || |
955 | PL_regkind[(U8)OP(first)] == NBOUND) | |
a0d0e21e | 956 | r->regstclass = first; |
22c35a8c | 957 | else if (PL_regkind[(U8)OP(first)] == BOL) { |
cad2e5aa JH |
958 | r->reganch |= (OP(first) == MBOL |
959 | ? ROPT_ANCH_MBOL | |
960 | : (OP(first) == SBOL | |
961 | ? ROPT_ANCH_SBOL | |
962 | : ROPT_ANCH_BOL)); | |
a0d0e21e | 963 | first = NEXTOPER(first); |
774d564b | 964 | goto again; |
965 | } | |
966 | else if (OP(first) == GPOS) { | |
967 | r->reganch |= ROPT_ANCH_GPOS; | |
968 | first = NEXTOPER(first); | |
969 | goto again; | |
a0d0e21e LW |
970 | } |
971 | else if ((OP(first) == STAR && | |
22c35a8c | 972 | PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) && |
a0d0e21e LW |
973 | !(r->reganch & ROPT_ANCH) ) |
974 | { | |
975 | /* turn .* into ^.* with an implied $*=1 */ | |
cad2e5aa JH |
976 | int type = OP(NEXTOPER(first)); |
977 | ||
978 | if (type == REG_ANY || type == ANYUTF8) | |
979 | type = ROPT_ANCH_MBOL; | |
980 | else | |
981 | type = ROPT_ANCH_SBOL; | |
982 | ||
983 | r->reganch |= type | ROPT_IMPLICIT; | |
a0d0e21e | 984 | first = NEXTOPER(first); |
774d564b | 985 | goto again; |
a0d0e21e | 986 | } |
cad2e5aa JH |
987 | if (sawplus && (!sawopen || !PL_regsawback) |
988 | && !(PL_regseen & REG_SEEN_EVAL)) /* May examine pos and $& */ | |
989 | /* x+ must match at the 1st pos of run of x's */ | |
990 | r->reganch |= ROPT_SKIP; | |
a0d0e21e | 991 | |
c277df42 IZ |
992 | /* Scan is after the zeroth branch, first is atomic matcher. */ |
993 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n", | |
994 | first - scan + 1)); | |
a0d0e21e LW |
995 | /* |
996 | * If there's something expensive in the r.e., find the | |
997 | * longest literal string that must appear and make it the | |
998 | * regmust. Resolve ties in favor of later strings, since | |
999 | * the regstart check works with the beginning of the r.e. | |
1000 | * and avoiding duplication strengthens checking. Not a | |
1001 | * strong reason, but sufficient in the absence of others. | |
1002 | * [Now we resolve ties in favor of the earlier string if | |
c277df42 | 1003 | * it happens that c_offset_min has been invalidated, since the |
a0d0e21e LW |
1004 | * earlier string may buy us something the later one won't.] |
1005 | */ | |
a0d0e21e | 1006 | minlen = 0; |
a687059c | 1007 | |
79cb57f6 GS |
1008 | data.longest_fixed = newSVpvn("",0); |
1009 | data.longest_float = newSVpvn("",0); | |
1010 | data.last_found = newSVpvn("",0); | |
c277df42 IZ |
1011 | data.longest = &(data.longest_fixed); |
1012 | first = scan; | |
1013 | ||
3280af22 | 1014 | minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */ |
c277df42 | 1015 | &data, SCF_DO_SUBSTR); |
3280af22 | 1016 | if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed) |
c277df42 | 1017 | && data.last_start_min == 0 && data.last_end > 0 |
3280af22 NIS |
1018 | && !PL_seen_zerolen |
1019 | && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS))) | |
c277df42 IZ |
1020 | r->reganch |= ROPT_CHECK_ALL; |
1021 | scan_commit(&data); | |
1022 | SvREFCNT_dec(data.last_found); | |
1023 | ||
a0ed51b3 | 1024 | longest_float_length = CHR_SVLEN(data.longest_float); |
c5254dd6 | 1025 | if (longest_float_length |
c277df42 IZ |
1026 | || (data.flags & SF_FL_BEFORE_EOL |
1027 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
3280af22 | 1028 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1029 | int t; |
1030 | ||
a0ed51b3 | 1031 | if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */ |
aca2d497 IZ |
1032 | && data.offset_fixed == data.offset_float_min |
1033 | && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)) | |
1034 | goto remove_float; /* As in (a)+. */ | |
1035 | ||
c277df42 IZ |
1036 | r->float_substr = data.longest_float; |
1037 | r->float_min_offset = data.offset_float_min; | |
1038 | r->float_max_offset = data.offset_float_max; | |
cf93c79d IZ |
1039 | t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1040 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
1041 | || (PL_regflags & PMf_MULTILINE))); | |
1042 | fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1043 | } |
1044 | else { | |
aca2d497 | 1045 | remove_float: |
c277df42 IZ |
1046 | r->float_substr = Nullsv; |
1047 | SvREFCNT_dec(data.longest_float); | |
c5254dd6 | 1048 | longest_float_length = 0; |
a0d0e21e | 1049 | } |
c277df42 | 1050 | |
a0ed51b3 | 1051 | longest_fixed_length = CHR_SVLEN(data.longest_fixed); |
c5254dd6 | 1052 | if (longest_fixed_length |
c277df42 IZ |
1053 | || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
1054 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
3280af22 | 1055 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1056 | int t; |
1057 | ||
c277df42 IZ |
1058 | r->anchored_substr = data.longest_fixed; |
1059 | r->anchored_offset = data.offset_fixed; | |
cf93c79d IZ |
1060 | t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1061 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
1062 | || (PL_regflags & PMf_MULTILINE))); | |
1063 | fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1064 | } |
1065 | else { | |
c277df42 IZ |
1066 | r->anchored_substr = Nullsv; |
1067 | SvREFCNT_dec(data.longest_fixed); | |
c5254dd6 | 1068 | longest_fixed_length = 0; |
a0d0e21e | 1069 | } |
c277df42 IZ |
1070 | |
1071 | /* A temporary algorithm prefers floated substr to fixed one to dig more info. */ | |
c5254dd6 | 1072 | if (longest_fixed_length > longest_float_length) { |
c277df42 IZ |
1073 | r->check_substr = r->anchored_substr; |
1074 | r->check_offset_min = r->check_offset_max = r->anchored_offset; | |
1075 | if (r->reganch & ROPT_ANCH_SINGLE) | |
1076 | r->reganch |= ROPT_NOSCAN; | |
a0ed51b3 LW |
1077 | } |
1078 | else { | |
c277df42 IZ |
1079 | r->check_substr = r->float_substr; |
1080 | r->check_offset_min = data.offset_float_min; | |
1081 | r->check_offset_max = data.offset_float_max; | |
a0d0e21e | 1082 | } |
cad2e5aa JH |
1083 | if (r->check_substr) { |
1084 | r->reganch |= RE_USE_INTUIT; | |
1085 | if (SvTAIL(r->check_substr)) | |
1086 | r->reganch |= RE_INTUIT_TAIL; | |
1087 | } | |
a0ed51b3 LW |
1088 | } |
1089 | else { | |
c277df42 IZ |
1090 | /* Several toplevels. Best we can is to set minlen. */ |
1091 | I32 fake; | |
1092 | ||
1093 | DEBUG_r(PerlIO_printf(Perl_debug_log, "\n")); | |
1094 | scan = r->program + 1; | |
2c2d71f5 | 1095 | minlen = study_chunk(&scan, &fake, scan + PL_regsize, &data, 0); |
c277df42 | 1096 | r->check_substr = r->anchored_substr = r->float_substr = Nullsv; |
a0d0e21e LW |
1097 | } |
1098 | ||
a0d0e21e | 1099 | r->minlen = minlen; |
3280af22 | 1100 | if (PL_regseen & REG_SEEN_GPOS) |
c277df42 | 1101 | r->reganch |= ROPT_GPOS_SEEN; |
3280af22 | 1102 | if (PL_regseen & REG_SEEN_LOOKBEHIND) |
c277df42 | 1103 | r->reganch |= ROPT_LOOKBEHIND_SEEN; |
3280af22 | 1104 | if (PL_regseen & REG_SEEN_EVAL) |
ce862d02 | 1105 | r->reganch |= ROPT_EVAL_SEEN; |
cf93c79d IZ |
1106 | Newz(1002, r->startp, PL_regnpar, I32); |
1107 | Newz(1002, r->endp, PL_regnpar, I32); | |
a0d0e21e LW |
1108 | DEBUG_r(regdump(r)); |
1109 | return(r); | |
a687059c LW |
1110 | } |
1111 | ||
1112 | /* | |
1113 | - reg - regular expression, i.e. main body or parenthesized thing | |
1114 | * | |
1115 | * Caller must absorb opening parenthesis. | |
1116 | * | |
1117 | * Combining parenthesis handling with the base level of regular expression | |
1118 | * is a trifle forced, but the need to tie the tails of the branches to what | |
1119 | * follows makes it hard to avoid. | |
1120 | */ | |
76e3520e | 1121 | STATIC regnode * |
cea2e8a9 | 1122 | S_reg(pTHX_ I32 paren, I32 *flagp) |
c277df42 | 1123 | /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */ |
a687059c | 1124 | { |
5c0ca799 | 1125 | dTHR; |
c277df42 IZ |
1126 | register regnode *ret; /* Will be the head of the group. */ |
1127 | register regnode *br; | |
1128 | register regnode *lastbr; | |
1129 | register regnode *ender = 0; | |
a0d0e21e | 1130 | register I32 parno = 0; |
3280af22 | 1131 | I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0; |
c277df42 | 1132 | char c; |
a0d0e21e | 1133 | |
821b33a5 | 1134 | *flagp = 0; /* Tentatively. */ |
a0d0e21e LW |
1135 | |
1136 | /* Make an OPEN node, if parenthesized. */ | |
1137 | if (paren) { | |
3280af22 | 1138 | if (*PL_regcomp_parse == '?') { |
ca9dfc88 IZ |
1139 | U16 posflags = 0, negflags = 0; |
1140 | U16 *flagsp = &posflags; | |
0f5d15d6 | 1141 | int logical = 0; |
ca9dfc88 | 1142 | |
3280af22 NIS |
1143 | PL_regcomp_parse++; |
1144 | paren = *PL_regcomp_parse++; | |
c277df42 | 1145 | ret = NULL; /* For look-ahead/behind. */ |
a0d0e21e | 1146 | switch (paren) { |
c277df42 | 1147 | case '<': |
3280af22 NIS |
1148 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
1149 | if (*PL_regcomp_parse == '!') | |
c277df42 | 1150 | paren = ','; |
3280af22 | 1151 | if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!') |
c277df42 | 1152 | goto unknown; |
3280af22 | 1153 | PL_regcomp_parse++; |
a0d0e21e LW |
1154 | case '=': |
1155 | case '!': | |
3280af22 | 1156 | PL_seen_zerolen++; |
c277df42 IZ |
1157 | case ':': |
1158 | case '>': | |
a0d0e21e LW |
1159 | break; |
1160 | case '$': | |
1161 | case '@': | |
c277df42 | 1162 | FAIL2("Sequence (?%c...) not implemented", (int)paren); |
a0d0e21e LW |
1163 | break; |
1164 | case '#': | |
3280af22 NIS |
1165 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
1166 | PL_regcomp_parse++; | |
1167 | if (*PL_regcomp_parse != ')') | |
c277df42 | 1168 | FAIL("Sequence (?#... not terminated"); |
a0d0e21e LW |
1169 | nextchar(); |
1170 | *flagp = TRYAGAIN; | |
1171 | return NULL; | |
0f5d15d6 IZ |
1172 | case 'p': |
1173 | logical = 1; | |
1174 | paren = *PL_regcomp_parse++; | |
1175 | /* FALL THROUGH */ | |
c277df42 IZ |
1176 | case '{': |
1177 | { | |
1178 | dTHR; | |
1179 | I32 count = 1, n = 0; | |
1180 | char c; | |
3280af22 | 1181 | char *s = PL_regcomp_parse; |
c277df42 IZ |
1182 | SV *sv; |
1183 | OP_4tree *sop, *rop; | |
1184 | ||
3280af22 NIS |
1185 | PL_seen_zerolen++; |
1186 | PL_regseen |= REG_SEEN_EVAL; | |
1187 | while (count && (c = *PL_regcomp_parse)) { | |
1188 | if (c == '\\' && PL_regcomp_parse[1]) | |
1189 | PL_regcomp_parse++; | |
c277df42 IZ |
1190 | else if (c == '{') |
1191 | count++; | |
1192 | else if (c == '}') | |
1193 | count--; | |
3280af22 | 1194 | PL_regcomp_parse++; |
c277df42 | 1195 | } |
3280af22 | 1196 | if (*PL_regcomp_parse != ')') |
c277df42 IZ |
1197 | FAIL("Sequence (?{...}) not terminated or not {}-balanced"); |
1198 | if (!SIZE_ONLY) { | |
1199 | AV *av; | |
1200 | ||
3280af22 | 1201 | if (PL_regcomp_parse - 1 - s) |
79cb57f6 | 1202 | sv = newSVpvn(s, PL_regcomp_parse - 1 - s); |
c277df42 | 1203 | else |
79cb57f6 | 1204 | sv = newSVpvn("", 0); |
c277df42 IZ |
1205 | |
1206 | rop = sv_compile_2op(sv, &sop, "re", &av); | |
1207 | ||
dfad63ad | 1208 | n = add_data(3, "nop"); |
3280af22 | 1209 | PL_regcomp_rx->data->data[n] = (void*)rop; |
dfad63ad HS |
1210 | PL_regcomp_rx->data->data[n+1] = (void*)sop; |
1211 | PL_regcomp_rx->data->data[n+2] = (void*)av; | |
c277df42 | 1212 | SvREFCNT_dec(sv); |
a0ed51b3 | 1213 | } |
e24b16f9 GS |
1214 | else { /* First pass */ |
1215 | if (PL_reginterp_cnt < ++PL_seen_evals | |
1216 | && PL_curcop != &PL_compiling) | |
2cd61cdb IZ |
1217 | /* No compiled RE interpolated, has runtime |
1218 | components ===> unsafe. */ | |
1219 | FAIL("Eval-group not allowed at runtime, use re 'eval'"); | |
3280af22 | 1220 | if (PL_tainted) |
cc6b7395 | 1221 | FAIL("Eval-group in insecure regular expression"); |
c277df42 IZ |
1222 | } |
1223 | ||
1224 | nextchar(); | |
0f5d15d6 IZ |
1225 | if (logical) { |
1226 | ret = reg_node(LOGICAL); | |
1227 | if (!SIZE_ONLY) | |
1228 | ret->flags = 2; | |
1229 | regtail(ret, reganode(EVAL, n)); | |
1230 | return ret; | |
1231 | } | |
c277df42 IZ |
1232 | return reganode(EVAL, n); |
1233 | } | |
1234 | case '(': | |
1235 | { | |
3280af22 NIS |
1236 | if (PL_regcomp_parse[0] == '?') { |
1237 | if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!' | |
1238 | || PL_regcomp_parse[1] == '<' | |
1239 | || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */ | |
c277df42 IZ |
1240 | I32 flag; |
1241 | ||
1242 | ret = reg_node(LOGICAL); | |
0f5d15d6 IZ |
1243 | if (!SIZE_ONLY) |
1244 | ret->flags = 1; | |
c277df42 IZ |
1245 | regtail(ret, reg(1, &flag)); |
1246 | goto insert_if; | |
1247 | } | |
a0ed51b3 LW |
1248 | } |
1249 | else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) { | |
3280af22 | 1250 | parno = atoi(PL_regcomp_parse++); |
c277df42 | 1251 | |
3280af22 NIS |
1252 | while (isDIGIT(*PL_regcomp_parse)) |
1253 | PL_regcomp_parse++; | |
c277df42 IZ |
1254 | ret = reganode(GROUPP, parno); |
1255 | if ((c = *nextchar()) != ')') | |
1256 | FAIL2("Switch (?(number%c not recognized", c); | |
1257 | insert_if: | |
1258 | regtail(ret, reganode(IFTHEN, 0)); | |
1259 | br = regbranch(&flags, 1); | |
1260 | if (br == NULL) | |
1261 | br = reganode(LONGJMP, 0); | |
1262 | else | |
1263 | regtail(br, reganode(LONGJMP, 0)); | |
1264 | c = *nextchar(); | |
d1b80229 IZ |
1265 | if (flags&HASWIDTH) |
1266 | *flagp |= HASWIDTH; | |
c277df42 IZ |
1267 | if (c == '|') { |
1268 | lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */ | |
1269 | regbranch(&flags, 1); | |
1270 | regtail(ret, lastbr); | |
d1b80229 IZ |
1271 | if (flags&HASWIDTH) |
1272 | *flagp |= HASWIDTH; | |
c277df42 | 1273 | c = *nextchar(); |
a0ed51b3 LW |
1274 | } |
1275 | else | |
c277df42 IZ |
1276 | lastbr = NULL; |
1277 | if (c != ')') | |
1278 | FAIL("Switch (?(condition)... contains too many branches"); | |
1279 | ender = reg_node(TAIL); | |
1280 | regtail(br, ender); | |
1281 | if (lastbr) { | |
1282 | regtail(lastbr, ender); | |
1283 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); | |
a0ed51b3 LW |
1284 | } |
1285 | else | |
c277df42 IZ |
1286 | regtail(ret, ender); |
1287 | return ret; | |
a0ed51b3 LW |
1288 | } |
1289 | else { | |
3280af22 | 1290 | FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse); |
c277df42 IZ |
1291 | } |
1292 | } | |
1b1626e4 | 1293 | case 0: |
c277df42 | 1294 | FAIL("Sequence (? incomplete"); |
1b1626e4 | 1295 | break; |
a0d0e21e | 1296 | default: |
3280af22 | 1297 | --PL_regcomp_parse; |
ca9dfc88 | 1298 | parse_flags: |
3280af22 NIS |
1299 | while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) { |
1300 | if (*PL_regcomp_parse != 'o') | |
1301 | pmflag(flagsp, *PL_regcomp_parse); | |
1302 | ++PL_regcomp_parse; | |
ca9dfc88 | 1303 | } |
3280af22 | 1304 | if (*PL_regcomp_parse == '-') { |
ca9dfc88 | 1305 | flagsp = &negflags; |
3280af22 | 1306 | ++PL_regcomp_parse; |
ca9dfc88 | 1307 | goto parse_flags; |
48c036b1 | 1308 | } |
3280af22 NIS |
1309 | PL_regflags |= posflags; |
1310 | PL_regflags &= ~negflags; | |
1311 | if (*PL_regcomp_parse == ':') { | |
1312 | PL_regcomp_parse++; | |
ca9dfc88 IZ |
1313 | paren = ':'; |
1314 | break; | |
1315 | } | |
c277df42 | 1316 | unknown: |
3280af22 NIS |
1317 | if (*PL_regcomp_parse != ')') |
1318 | FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse); | |
a0d0e21e LW |
1319 | nextchar(); |
1320 | *flagp = TRYAGAIN; | |
1321 | return NULL; | |
1322 | } | |
1323 | } | |
1324 | else { | |
3280af22 NIS |
1325 | parno = PL_regnpar; |
1326 | PL_regnpar++; | |
a0d0e21e | 1327 | ret = reganode(OPEN, parno); |
c277df42 | 1328 | open = 1; |
a0d0e21e | 1329 | } |
a0ed51b3 LW |
1330 | } |
1331 | else | |
a0d0e21e LW |
1332 | ret = NULL; |
1333 | ||
1334 | /* Pick up the branches, linking them together. */ | |
c277df42 | 1335 | br = regbranch(&flags, 1); |
a0d0e21e LW |
1336 | if (br == NULL) |
1337 | return(NULL); | |
3280af22 NIS |
1338 | if (*PL_regcomp_parse == '|') { |
1339 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 | 1340 | reginsert(BRANCHJ, br); |
a0ed51b3 LW |
1341 | } |
1342 | else | |
c277df42 IZ |
1343 | reginsert(BRANCH, br); |
1344 | have_branch = 1; | |
1345 | if (SIZE_ONLY) | |
3280af22 | 1346 | PL_extralen += 1; /* For BRANCHJ-BRANCH. */ |
a0ed51b3 LW |
1347 | } |
1348 | else if (paren == ':') { | |
c277df42 IZ |
1349 | *flagp |= flags&SIMPLE; |
1350 | } | |
1351 | if (open) { /* Starts with OPEN. */ | |
1352 | regtail(ret, br); /* OPEN -> first. */ | |
a0ed51b3 LW |
1353 | } |
1354 | else if (paren != '?') /* Not Conditional */ | |
a0d0e21e | 1355 | ret = br; |
821b33a5 IZ |
1356 | if (flags&HASWIDTH) |
1357 | *flagp |= HASWIDTH; | |
a0d0e21e | 1358 | *flagp |= flags&SPSTART; |
c277df42 | 1359 | lastbr = br; |
3280af22 NIS |
1360 | while (*PL_regcomp_parse == '|') { |
1361 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 IZ |
1362 | ender = reganode(LONGJMP,0); |
1363 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ | |
1364 | } | |
1365 | if (SIZE_ONLY) | |
3280af22 | 1366 | PL_extralen += 2; /* Account for LONGJMP. */ |
a0d0e21e | 1367 | nextchar(); |
c277df42 | 1368 | br = regbranch(&flags, 0); |
a687059c | 1369 | if (br == NULL) |
a0d0e21e | 1370 | return(NULL); |
c277df42 IZ |
1371 | regtail(lastbr, br); /* BRANCH -> BRANCH. */ |
1372 | lastbr = br; | |
821b33a5 IZ |
1373 | if (flags&HASWIDTH) |
1374 | *flagp |= HASWIDTH; | |
a687059c | 1375 | *flagp |= flags&SPSTART; |
a0d0e21e LW |
1376 | } |
1377 | ||
c277df42 IZ |
1378 | if (have_branch || paren != ':') { |
1379 | /* Make a closing node, and hook it on the end. */ | |
1380 | switch (paren) { | |
1381 | case ':': | |
1382 | ender = reg_node(TAIL); | |
1383 | break; | |
1384 | case 1: | |
1385 | ender = reganode(CLOSE, parno); | |
1386 | break; | |
1387 | case '<': | |
c277df42 IZ |
1388 | case ',': |
1389 | case '=': | |
1390 | case '!': | |
c277df42 | 1391 | *flagp &= ~HASWIDTH; |
821b33a5 IZ |
1392 | /* FALL THROUGH */ |
1393 | case '>': | |
1394 | ender = reg_node(SUCCEED); | |
c277df42 IZ |
1395 | break; |
1396 | case 0: | |
1397 | ender = reg_node(END); | |
1398 | break; | |
1399 | } | |
1400 | regtail(lastbr, ender); | |
a0d0e21e | 1401 | |
c277df42 IZ |
1402 | if (have_branch) { |
1403 | /* Hook the tails of the branches to the closing node. */ | |
1404 | for (br = ret; br != NULL; br = regnext(br)) { | |
1405 | regoptail(br, ender); | |
1406 | } | |
1407 | } | |
a0d0e21e | 1408 | } |
c277df42 IZ |
1409 | |
1410 | { | |
1411 | char *p; | |
1412 | static char parens[] = "=!<,>"; | |
1413 | ||
1414 | if (paren && (p = strchr(parens, paren))) { | |
1415 | int node = ((p - parens) % 2) ? UNLESSM : IFMATCH; | |
1416 | int flag = (p - parens) > 1; | |
1417 | ||
1418 | if (paren == '>') | |
1419 | node = SUSPEND, flag = 0; | |
1420 | reginsert(node,ret); | |
c277df42 | 1421 | ret->flags = flag; |
c277df42 IZ |
1422 | regtail(ret, reg_node(TAIL)); |
1423 | } | |
a0d0e21e LW |
1424 | } |
1425 | ||
1426 | /* Check for proper termination. */ | |
ce3e6498 HS |
1427 | if (paren) { |
1428 | PL_regflags = oregflags; | |
1429 | if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') { | |
1430 | FAIL("unmatched () in regexp"); | |
1431 | } | |
a0ed51b3 LW |
1432 | } |
1433 | else if (!paren && PL_regcomp_parse < PL_regxend) { | |
3280af22 | 1434 | if (*PL_regcomp_parse == ')') { |
a0d0e21e | 1435 | FAIL("unmatched () in regexp"); |
a0ed51b3 LW |
1436 | } |
1437 | else | |
a0d0e21e LW |
1438 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
1439 | /* NOTREACHED */ | |
1440 | } | |
a687059c | 1441 | |
a0d0e21e | 1442 | return(ret); |
a687059c LW |
1443 | } |
1444 | ||
1445 | /* | |
1446 | - regbranch - one alternative of an | operator | |
1447 | * | |
1448 | * Implements the concatenation operator. | |
1449 | */ | |
76e3520e | 1450 | STATIC regnode * |
cea2e8a9 | 1451 | S_regbranch(pTHX_ I32 *flagp, I32 first) |
a687059c | 1452 | { |
5c0ca799 | 1453 | dTHR; |
c277df42 IZ |
1454 | register regnode *ret; |
1455 | register regnode *chain = NULL; | |
1456 | register regnode *latest; | |
1457 | I32 flags = 0, c = 0; | |
a0d0e21e | 1458 | |
c277df42 IZ |
1459 | if (first) |
1460 | ret = NULL; | |
1461 | else { | |
3280af22 | 1462 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
1463 | ret = reganode(BRANCHJ,0); |
1464 | else | |
1465 | ret = reg_node(BRANCH); | |
1466 | } | |
1467 | ||
1468 | if (!first && SIZE_ONLY) | |
3280af22 | 1469 | PL_extralen += 1; /* BRANCHJ */ |
c277df42 IZ |
1470 | |
1471 | *flagp = WORST; /* Tentatively. */ | |
a0d0e21e | 1472 | |
3280af22 | 1473 | PL_regcomp_parse--; |
a0d0e21e | 1474 | nextchar(); |
3280af22 | 1475 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') { |
a0d0e21e LW |
1476 | flags &= ~TRYAGAIN; |
1477 | latest = regpiece(&flags); | |
1478 | if (latest == NULL) { | |
1479 | if (flags & TRYAGAIN) | |
1480 | continue; | |
1481 | return(NULL); | |
a0ed51b3 LW |
1482 | } |
1483 | else if (ret == NULL) | |
c277df42 | 1484 | ret = latest; |
a0d0e21e | 1485 | *flagp |= flags&HASWIDTH; |
c277df42 | 1486 | if (chain == NULL) /* First piece. */ |
a0d0e21e LW |
1487 | *flagp |= flags&SPSTART; |
1488 | else { | |
3280af22 | 1489 | PL_regnaughty++; |
a0d0e21e | 1490 | regtail(chain, latest); |
a687059c | 1491 | } |
a0d0e21e | 1492 | chain = latest; |
c277df42 IZ |
1493 | c++; |
1494 | } | |
1495 | if (chain == NULL) { /* Loop ran zero times. */ | |
1496 | chain = reg_node(NOTHING); | |
1497 | if (ret == NULL) | |
1498 | ret = chain; | |
1499 | } | |
1500 | if (c == 1) { | |
1501 | *flagp |= flags&SIMPLE; | |
a0d0e21e | 1502 | } |
a687059c | 1503 | |
a0d0e21e | 1504 | return(ret); |
a687059c LW |
1505 | } |
1506 | ||
1507 | /* | |
1508 | - regpiece - something followed by possible [*+?] | |
1509 | * | |
1510 | * Note that the branching code sequences used for ? and the general cases | |
1511 | * of * and + are somewhat optimized: they use the same NOTHING node as | |
1512 | * both the endmarker for their branch list and the body of the last branch. | |
1513 | * It might seem that this node could be dispensed with entirely, but the | |
1514 | * endmarker role is not redundant. | |
1515 | */ | |
76e3520e | 1516 | STATIC regnode * |
cea2e8a9 | 1517 | S_regpiece(pTHX_ I32 *flagp) |
a687059c | 1518 | { |
5c0ca799 | 1519 | dTHR; |
c277df42 | 1520 | register regnode *ret; |
a0d0e21e LW |
1521 | register char op; |
1522 | register char *next; | |
1523 | I32 flags; | |
3280af22 | 1524 | char *origparse = PL_regcomp_parse; |
a0d0e21e LW |
1525 | char *maxpos; |
1526 | I32 min; | |
c277df42 | 1527 | I32 max = REG_INFTY; |
a0d0e21e LW |
1528 | |
1529 | ret = regatom(&flags); | |
1530 | if (ret == NULL) { | |
1531 | if (flags & TRYAGAIN) | |
1532 | *flagp |= TRYAGAIN; | |
1533 | return(NULL); | |
1534 | } | |
1535 | ||
3280af22 | 1536 | op = *PL_regcomp_parse; |
a0d0e21e | 1537 | |
3280af22 NIS |
1538 | if (op == '{' && regcurly(PL_regcomp_parse)) { |
1539 | next = PL_regcomp_parse + 1; | |
a0d0e21e LW |
1540 | maxpos = Nullch; |
1541 | while (isDIGIT(*next) || *next == ',') { | |
1542 | if (*next == ',') { | |
1543 | if (maxpos) | |
1544 | break; | |
1545 | else | |
1546 | maxpos = next; | |
a687059c | 1547 | } |
a0d0e21e LW |
1548 | next++; |
1549 | } | |
1550 | if (*next == '}') { /* got one */ | |
1551 | if (!maxpos) | |
1552 | maxpos = next; | |
3280af22 NIS |
1553 | PL_regcomp_parse++; |
1554 | min = atoi(PL_regcomp_parse); | |
a0d0e21e LW |
1555 | if (*maxpos == ',') |
1556 | maxpos++; | |
1557 | else | |
3280af22 | 1558 | maxpos = PL_regcomp_parse; |
a0d0e21e LW |
1559 | max = atoi(maxpos); |
1560 | if (!max && *maxpos != '0') | |
c277df42 IZ |
1561 | max = REG_INFTY; /* meaning "infinity" */ |
1562 | else if (max >= REG_INFTY) | |
1563 | FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); | |
3280af22 | 1564 | PL_regcomp_parse = next; |
a0d0e21e LW |
1565 | nextchar(); |
1566 | ||
1567 | do_curly: | |
1568 | if ((flags&SIMPLE)) { | |
3280af22 | 1569 | PL_regnaughty += 2 + PL_regnaughty / 2; |
a0d0e21e LW |
1570 | reginsert(CURLY, ret); |
1571 | } | |
1572 | else { | |
2c2d71f5 JH |
1573 | regnode *w = reg_node(WHILEM); |
1574 | ||
1575 | w->flags = 0; | |
1576 | regtail(ret, w); | |
3280af22 | 1577 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 IZ |
1578 | reginsert(LONGJMP,ret); |
1579 | reginsert(NOTHING,ret); | |
1580 | NEXT_OFF(ret) = 3; /* Go over LONGJMP. */ | |
1581 | } | |
a0d0e21e | 1582 | reginsert(CURLYX,ret); |
3280af22 | 1583 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
1584 | NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */ |
1585 | regtail(ret, reg_node(NOTHING)); | |
1586 | if (SIZE_ONLY) | |
2c2d71f5 JH |
1587 | PL_reg_whilem_seen++, PL_extralen += 3; |
1588 | PL_regnaughty += 4 + PL_regnaughty; /* compound interest */ | |
a0d0e21e | 1589 | } |
c277df42 | 1590 | ret->flags = 0; |
a0d0e21e LW |
1591 | |
1592 | if (min > 0) | |
821b33a5 IZ |
1593 | *flagp = WORST; |
1594 | if (max > 0) | |
1595 | *flagp |= HASWIDTH; | |
a0d0e21e | 1596 | if (max && max < min) |
c277df42 IZ |
1597 | FAIL("Can't do {n,m} with n > m"); |
1598 | if (!SIZE_ONLY) { | |
1599 | ARG1_SET(ret, min); | |
1600 | ARG2_SET(ret, max); | |
a687059c | 1601 | } |
a687059c | 1602 | |
a0d0e21e | 1603 | goto nest_check; |
a687059c | 1604 | } |
a0d0e21e | 1605 | } |
a687059c | 1606 | |
a0d0e21e LW |
1607 | if (!ISMULT1(op)) { |
1608 | *flagp = flags; | |
a687059c | 1609 | return(ret); |
a0d0e21e | 1610 | } |
bb20fd44 | 1611 | |
c277df42 | 1612 | #if 0 /* Now runtime fix should be reliable. */ |
bb20fd44 | 1613 | if (!(flags&HASWIDTH) && op != '?') |
c277df42 IZ |
1614 | FAIL("regexp *+ operand could be empty"); |
1615 | #endif | |
bb20fd44 | 1616 | |
a0d0e21e LW |
1617 | nextchar(); |
1618 | ||
821b33a5 | 1619 | *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); |
a0d0e21e LW |
1620 | |
1621 | if (op == '*' && (flags&SIMPLE)) { | |
1622 | reginsert(STAR, ret); | |
c277df42 | 1623 | ret->flags = 0; |
3280af22 | 1624 | PL_regnaughty += 4; |
a0d0e21e LW |
1625 | } |
1626 | else if (op == '*') { | |
1627 | min = 0; | |
1628 | goto do_curly; | |
a0ed51b3 LW |
1629 | } |
1630 | else if (op == '+' && (flags&SIMPLE)) { | |
a0d0e21e | 1631 | reginsert(PLUS, ret); |
c277df42 | 1632 | ret->flags = 0; |
3280af22 | 1633 | PL_regnaughty += 3; |
a0d0e21e LW |
1634 | } |
1635 | else if (op == '+') { | |
1636 | min = 1; | |
1637 | goto do_curly; | |
a0ed51b3 LW |
1638 | } |
1639 | else if (op == '?') { | |
a0d0e21e LW |
1640 | min = 0; max = 1; |
1641 | goto do_curly; | |
1642 | } | |
1643 | nest_check: | |
17feb5d5 | 1644 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) { |
cea2e8a9 | 1645 | Perl_warner(aTHX_ WARN_UNSAFE, "%.*s matches null string many times", |
3280af22 | 1646 | PL_regcomp_parse - origparse, origparse); |
a0d0e21e LW |
1647 | } |
1648 | ||
3280af22 | 1649 | if (*PL_regcomp_parse == '?') { |
a0d0e21e LW |
1650 | nextchar(); |
1651 | reginsert(MINMOD, ret); | |
c277df42 | 1652 | regtail(ret, ret + NODE_STEP_REGNODE); |
a0d0e21e | 1653 | } |
3280af22 | 1654 | if (ISMULT2(PL_regcomp_parse)) |
a0d0e21e LW |
1655 | FAIL("nested *?+ in regexp"); |
1656 | ||
1657 | return(ret); | |
a687059c LW |
1658 | } |
1659 | ||
1660 | /* | |
1661 | - regatom - the lowest level | |
1662 | * | |
1663 | * Optimization: gobbles an entire sequence of ordinary characters so that | |
1664 | * it can turn them into a single node, which is smaller to store and | |
1665 | * faster to run. Backslashed characters are exceptions, each becoming a | |
1666 | * separate node; the code is simpler that way and it's not worth fixing. | |
1667 | * | |
1668 | * [Yes, it is worth fixing, some scripts can run twice the speed.] | |
1669 | */ | |
76e3520e | 1670 | STATIC regnode * |
cea2e8a9 | 1671 | S_regatom(pTHX_ I32 *flagp) |
a687059c | 1672 | { |
5c0ca799 | 1673 | dTHR; |
c277df42 | 1674 | register regnode *ret = 0; |
a0d0e21e LW |
1675 | I32 flags; |
1676 | ||
1677 | *flagp = WORST; /* Tentatively. */ | |
1678 | ||
1679 | tryagain: | |
3280af22 | 1680 | switch (*PL_regcomp_parse) { |
a0d0e21e | 1681 | case '^': |
3280af22 | 1682 | PL_seen_zerolen++; |
a0d0e21e | 1683 | nextchar(); |
3280af22 | 1684 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 1685 | ret = reg_node(MBOL); |
3280af22 | 1686 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 1687 | ret = reg_node(SBOL); |
a0d0e21e | 1688 | else |
c277df42 | 1689 | ret = reg_node(BOL); |
a0d0e21e LW |
1690 | break; |
1691 | case '$': | |
3280af22 NIS |
1692 | if (PL_regcomp_parse[1]) |
1693 | PL_seen_zerolen++; | |
a0d0e21e | 1694 | nextchar(); |
3280af22 | 1695 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 1696 | ret = reg_node(MEOL); |
3280af22 | 1697 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 1698 | ret = reg_node(SEOL); |
a0d0e21e | 1699 | else |
c277df42 | 1700 | ret = reg_node(EOL); |
a0d0e21e LW |
1701 | break; |
1702 | case '.': | |
1703 | nextchar(); | |
a0ed51b3 LW |
1704 | if (UTF) { |
1705 | if (PL_regflags & PMf_SINGLELINE) | |
1706 | ret = reg_node(SANYUTF8); | |
1707 | else | |
1708 | ret = reg_node(ANYUTF8); | |
1709 | *flagp |= HASWIDTH; | |
1710 | } | |
1711 | else { | |
1712 | if (PL_regflags & PMf_SINGLELINE) | |
1713 | ret = reg_node(SANY); | |
1714 | else | |
22c35a8c | 1715 | ret = reg_node(REG_ANY); |
a0ed51b3 LW |
1716 | *flagp |= HASWIDTH|SIMPLE; |
1717 | } | |
3280af22 | 1718 | PL_regnaughty++; |
a0d0e21e LW |
1719 | break; |
1720 | case '[': | |
3280af22 | 1721 | PL_regcomp_parse++; |
a0ed51b3 | 1722 | ret = (UTF ? regclassutf8() : regclass()); |
a14b48bc LW |
1723 | if (*PL_regcomp_parse != ']') |
1724 | FAIL("unmatched [] in regexp"); | |
1725 | nextchar(); | |
a0d0e21e LW |
1726 | *flagp |= HASWIDTH|SIMPLE; |
1727 | break; | |
1728 | case '(': | |
1729 | nextchar(); | |
1730 | ret = reg(1, &flags); | |
1731 | if (ret == NULL) { | |
1732 | if (flags & TRYAGAIN) | |
1733 | goto tryagain; | |
1734 | return(NULL); | |
1735 | } | |
c277df42 | 1736 | *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE); |
a0d0e21e LW |
1737 | break; |
1738 | case '|': | |
1739 | case ')': | |
1740 | if (flags & TRYAGAIN) { | |
1741 | *flagp |= TRYAGAIN; | |
1742 | return NULL; | |
1743 | } | |
3280af22 | 1744 | FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse); |
a0d0e21e LW |
1745 | /* Supposed to be caught earlier. */ |
1746 | break; | |
85afd4ae | 1747 | case '{': |
3280af22 NIS |
1748 | if (!regcurly(PL_regcomp_parse)) { |
1749 | PL_regcomp_parse++; | |
85afd4ae CS |
1750 | goto defchar; |
1751 | } | |
1752 | /* FALL THROUGH */ | |
a0d0e21e LW |
1753 | case '?': |
1754 | case '+': | |
1755 | case '*': | |
3115e423 | 1756 | FAIL("?+*{} follows nothing in regexp"); |
a0d0e21e LW |
1757 | break; |
1758 | case '\\': | |
3280af22 | 1759 | switch (*++PL_regcomp_parse) { |
a0d0e21e | 1760 | case 'A': |
3280af22 | 1761 | PL_seen_zerolen++; |
c277df42 | 1762 | ret = reg_node(SBOL); |
a0d0e21e LW |
1763 | *flagp |= SIMPLE; |
1764 | nextchar(); | |
1765 | break; | |
1766 | case 'G': | |
c277df42 | 1767 | ret = reg_node(GPOS); |
3280af22 | 1768 | PL_regseen |= REG_SEEN_GPOS; |
a0d0e21e LW |
1769 | *flagp |= SIMPLE; |
1770 | nextchar(); | |
1771 | break; | |
1772 | case 'Z': | |
c277df42 | 1773 | ret = reg_node(SEOL); |
a0d0e21e LW |
1774 | *flagp |= SIMPLE; |
1775 | nextchar(); | |
1776 | break; | |
b85d18e9 IZ |
1777 | case 'z': |
1778 | ret = reg_node(EOS); | |
1779 | *flagp |= SIMPLE; | |
3280af22 | 1780 | PL_seen_zerolen++; /* Do not optimize RE away */ |
b85d18e9 IZ |
1781 | nextchar(); |
1782 | break; | |
423cee85 | 1783 | case 'O': |
a0ed51b3 LW |
1784 | ret = reg_node(SANY); |
1785 | *flagp |= HASWIDTH|SIMPLE; | |
1786 | nextchar(); | |
1787 | break; | |
1788 | case 'X': | |
1789 | ret = reg_node(CLUMP); | |
1790 | *flagp |= HASWIDTH; | |
1791 | nextchar(); | |
1792 | if (UTF && !PL_utf8_mark) | |
dfe13c55 | 1793 | is_utf8_mark((U8*)"~"); /* preload table */ |
a0ed51b3 | 1794 | break; |
a0d0e21e | 1795 | case 'w': |
a0ed51b3 LW |
1796 | ret = reg_node( |
1797 | UTF | |
1798 | ? (LOC ? ALNUMLUTF8 : ALNUMUTF8) | |
1799 | : (LOC ? ALNUML : ALNUM)); | |
a0d0e21e LW |
1800 | *flagp |= HASWIDTH|SIMPLE; |
1801 | nextchar(); | |
a0ed51b3 | 1802 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1803 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1804 | break; |
1805 | case 'W': | |
a0ed51b3 LW |
1806 | ret = reg_node( |
1807 | UTF | |
1808 | ? (LOC ? NALNUMLUTF8 : NALNUMUTF8) | |
1809 | : (LOC ? NALNUML : NALNUM)); | |
a0d0e21e LW |
1810 | *flagp |= HASWIDTH|SIMPLE; |
1811 | nextchar(); | |
a0ed51b3 | 1812 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1813 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1814 | break; |
1815 | case 'b': | |
3280af22 | 1816 | PL_seen_zerolen++; |
f5c9036e | 1817 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
1818 | ret = reg_node( |
1819 | UTF | |
1820 | ? (LOC ? BOUNDLUTF8 : BOUNDUTF8) | |
1821 | : (LOC ? BOUNDL : BOUND)); | |
a0d0e21e LW |
1822 | *flagp |= SIMPLE; |
1823 | nextchar(); | |
a0ed51b3 | 1824 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1825 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1826 | break; |
1827 | case 'B': | |
3280af22 | 1828 | PL_seen_zerolen++; |
f5c9036e | 1829 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
1830 | ret = reg_node( |
1831 | UTF | |
1832 | ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8) | |
1833 | : (LOC ? NBOUNDL : NBOUND)); | |
a0d0e21e LW |
1834 | *flagp |= SIMPLE; |
1835 | nextchar(); | |
a0ed51b3 | 1836 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 1837 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
1838 | break; |
1839 | case 's': | |
a0ed51b3 LW |
1840 | ret = reg_node( |
1841 | UTF | |
1842 | ? (LOC ? SPACELUTF8 : SPACEUTF8) | |
1843 | : (LOC ? SPACEL : SPACE)); | |
a0d0e21e LW |
1844 | *flagp |= HASWIDTH|SIMPLE; |
1845 | nextchar(); | |
a0ed51b3 | 1846 | if (UTF && !PL_utf8_space) |
dfe13c55 | 1847 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
1848 | break; |
1849 | case 'S': | |
a0ed51b3 LW |
1850 | ret = reg_node( |
1851 | UTF | |
1852 | ? (LOC ? NSPACELUTF8 : NSPACEUTF8) | |
1853 | : (LOC ? NSPACEL : NSPACE)); | |
a0d0e21e LW |
1854 | *flagp |= HASWIDTH|SIMPLE; |
1855 | nextchar(); | |
a0ed51b3 | 1856 | if (UTF && !PL_utf8_space) |
dfe13c55 | 1857 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
1858 | break; |
1859 | case 'd': | |
a0ed51b3 | 1860 | ret = reg_node(UTF ? DIGITUTF8 : DIGIT); |
a0d0e21e LW |
1861 | *flagp |= HASWIDTH|SIMPLE; |
1862 | nextchar(); | |
a0ed51b3 | 1863 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 1864 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e LW |
1865 | break; |
1866 | case 'D': | |
a0ed51b3 | 1867 | ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT); |
a0d0e21e LW |
1868 | *flagp |= HASWIDTH|SIMPLE; |
1869 | nextchar(); | |
a0ed51b3 | 1870 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 1871 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e | 1872 | break; |
a14b48bc LW |
1873 | case 'p': |
1874 | case 'P': | |
1875 | { /* a lovely hack--pretend we saw [\pX] instead */ | |
1876 | char* oldregxend = PL_regxend; | |
1877 | ||
1878 | if (PL_regcomp_parse[1] == '{') { | |
1879 | PL_regxend = strchr(PL_regcomp_parse, '}'); | |
1880 | if (!PL_regxend) | |
1881 | FAIL("Missing right brace on \\p{}"); | |
1882 | PL_regxend++; | |
1883 | } | |
1884 | else | |
1885 | PL_regxend = PL_regcomp_parse + 2; | |
1886 | PL_regcomp_parse--; | |
1887 | ||
1888 | ret = regclassutf8(); | |
1889 | ||
1890 | PL_regxend = oldregxend; | |
1891 | PL_regcomp_parse--; | |
1892 | nextchar(); | |
1893 | *flagp |= HASWIDTH|SIMPLE; | |
1894 | } | |
1895 | break; | |
a0d0e21e LW |
1896 | case 'n': |
1897 | case 'r': | |
1898 | case 't': | |
1899 | case 'f': | |
1900 | case 'e': | |
1901 | case 'a': | |
1902 | case 'x': | |
1903 | case 'c': | |
1904 | case '0': | |
1905 | goto defchar; | |
1906 | case '1': case '2': case '3': case '4': | |
1907 | case '5': case '6': case '7': case '8': case '9': | |
1908 | { | |
3280af22 | 1909 | I32 num = atoi(PL_regcomp_parse); |
a0d0e21e | 1910 | |
3280af22 | 1911 | if (num > 9 && num >= PL_regnpar) |
a0d0e21e LW |
1912 | goto defchar; |
1913 | else { | |
3280af22 | 1914 | if (!SIZE_ONLY && num > PL_regcomp_rx->nparens) |
ef64f398 | 1915 | FAIL("reference to nonexistent group"); |
3280af22 | 1916 | PL_regsawback = 1; |
a0ed51b3 LW |
1917 | ret = reganode(FOLD |
1918 | ? (LOC ? REFFL : REFF) | |
c8756f30 | 1919 | : REF, num); |
a0d0e21e | 1920 | *flagp |= HASWIDTH; |
3280af22 NIS |
1921 | while (isDIGIT(*PL_regcomp_parse)) |
1922 | PL_regcomp_parse++; | |
1923 | PL_regcomp_parse--; | |
a0d0e21e LW |
1924 | nextchar(); |
1925 | } | |
1926 | } | |
1927 | break; | |
1928 | case '\0': | |
3280af22 | 1929 | if (PL_regcomp_parse >= PL_regxend) |
a0d0e21e LW |
1930 | FAIL("trailing \\ in regexp"); |
1931 | /* FALL THROUGH */ | |
1932 | default: | |
c9f97d15 IZ |
1933 | /* Do not generate `unrecognized' warnings here, we fall |
1934 | back into the quick-grab loop below */ | |
a0d0e21e LW |
1935 | goto defchar; |
1936 | } | |
1937 | break; | |
4633a7c4 LW |
1938 | |
1939 | case '#': | |
3280af22 NIS |
1940 | if (PL_regflags & PMf_EXTENDED) { |
1941 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++; | |
1942 | if (PL_regcomp_parse < PL_regxend) | |
4633a7c4 LW |
1943 | goto tryagain; |
1944 | } | |
1945 | /* FALL THROUGH */ | |
1946 | ||
a0d0e21e LW |
1947 | default: { |
1948 | register I32 len; | |
a0ed51b3 | 1949 | register UV ender; |
a0d0e21e | 1950 | register char *p; |
c277df42 | 1951 | char *oldp, *s; |
a0d0e21e LW |
1952 | I32 numlen; |
1953 | ||
3280af22 | 1954 | PL_regcomp_parse++; |
a0d0e21e LW |
1955 | |
1956 | defchar: | |
a0ed51b3 LW |
1957 | ret = reg_node(FOLD |
1958 | ? (LOC ? EXACTFL : EXACTF) | |
bbce6d69 | 1959 | : EXACT); |
cd439c50 | 1960 | s = STRING(ret); |
3280af22 NIS |
1961 | for (len = 0, p = PL_regcomp_parse - 1; |
1962 | len < 127 && p < PL_regxend; | |
a0d0e21e LW |
1963 | len++) |
1964 | { | |
1965 | oldp = p; | |
5b5a24f7 | 1966 | |
3280af22 NIS |
1967 | if (PL_regflags & PMf_EXTENDED) |
1968 | p = regwhite(p, PL_regxend); | |
a0d0e21e LW |
1969 | switch (*p) { |
1970 | case '^': | |
1971 | case '$': | |
1972 | case '.': | |
1973 | case '[': | |
1974 | case '(': | |
1975 | case ')': | |
1976 | case '|': | |
1977 | goto loopdone; | |
1978 | case '\\': | |
1979 | switch (*++p) { | |
1980 | case 'A': | |
1981 | case 'G': | |
1982 | case 'Z': | |
b85d18e9 | 1983 | case 'z': |
a0d0e21e LW |
1984 | case 'w': |
1985 | case 'W': | |
1986 | case 'b': | |
1987 | case 'B': | |
1988 | case 's': | |
1989 | case 'S': | |
1990 | case 'd': | |
1991 | case 'D': | |
a14b48bc LW |
1992 | case 'p': |
1993 | case 'P': | |
a0d0e21e LW |
1994 | --p; |
1995 | goto loopdone; | |
1996 | case 'n': | |
1997 | ender = '\n'; | |
1998 | p++; | |
a687059c | 1999 | break; |
a0d0e21e LW |
2000 | case 'r': |
2001 | ender = '\r'; | |
2002 | p++; | |
a687059c | 2003 | break; |
a0d0e21e LW |
2004 | case 't': |
2005 | ender = '\t'; | |
2006 | p++; | |
a687059c | 2007 | break; |
a0d0e21e LW |
2008 | case 'f': |
2009 | ender = '\f'; | |
2010 | p++; | |
a687059c | 2011 | break; |
a0d0e21e LW |
2012 | case 'e': |
2013 | ender = '\033'; | |
2014 | p++; | |
a687059c | 2015 | break; |
a0d0e21e LW |
2016 | case 'a': |
2017 | ender = '\007'; | |
2018 | p++; | |
a687059c | 2019 | break; |
a0d0e21e | 2020 | case 'x': |
a0ed51b3 LW |
2021 | if (*++p == '{') { |
2022 | char* e = strchr(p, '}'); | |
2023 | ||
2024 | if (!e) | |
2025 | FAIL("Missing right brace on \\x{}"); | |
2026 | else if (UTF) { | |
2027 | ender = scan_hex(p + 1, e - p, &numlen); | |
2028 | if (numlen + len >= 127) { /* numlen is generous */ | |
2029 | p--; | |
2030 | goto loopdone; | |
2031 | } | |
2032 | p = e + 1; | |
2033 | } | |
2034 | else | |
2035 | FAIL("Can't use \\x{} without 'use utf8' declaration"); | |
2036 | } | |
2037 | else { | |
2038 | ender = scan_hex(p, 2, &numlen); | |
2039 | p += numlen; | |
2040 | } | |
a687059c | 2041 | break; |
a0d0e21e LW |
2042 | case 'c': |
2043 | p++; | |
bbce6d69 | 2044 | ender = UCHARAT(p++); |
2045 | ender = toCTRL(ender); | |
a687059c | 2046 | break; |
a0d0e21e LW |
2047 | case '0': case '1': case '2': case '3':case '4': |
2048 | case '5': case '6': case '7': case '8':case '9': | |
2049 | if (*p == '0' || | |
3280af22 | 2050 | (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) { |
a0d0e21e LW |
2051 | ender = scan_oct(p, 3, &numlen); |
2052 | p += numlen; | |
2053 | } | |
2054 | else { | |
2055 | --p; | |
2056 | goto loopdone; | |
a687059c LW |
2057 | } |
2058 | break; | |
a0d0e21e | 2059 | case '\0': |
3280af22 | 2060 | if (p >= PL_regxend) |
a687059c LW |
2061 | FAIL("trailing \\ in regexp"); |
2062 | /* FALL THROUGH */ | |
a0d0e21e | 2063 | default: |
c9f97d15 | 2064 | if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p)) |
cea2e8a9 | 2065 | Perl_warner(aTHX_ WARN_UNSAFE, |
c9f97d15 IZ |
2066 | "/%.127s/: Unrecognized escape \\%c passed through", |
2067 | PL_regprecomp, | |
2068 | *p); | |
a0ed51b3 | 2069 | goto normal_default; |
a0d0e21e LW |
2070 | } |
2071 | break; | |
a687059c | 2072 | default: |
a0ed51b3 LW |
2073 | normal_default: |
2074 | if ((*p & 0xc0) == 0xc0 && UTF) { | |
dfe13c55 | 2075 | ender = utf8_to_uv((U8*)p, &numlen); |
a0ed51b3 LW |
2076 | p += numlen; |
2077 | } | |
2078 | else | |
2079 | ender = *p++; | |
a0d0e21e | 2080 | break; |
a687059c | 2081 | } |
3280af22 NIS |
2082 | if (PL_regflags & PMf_EXTENDED) |
2083 | p = regwhite(p, PL_regxend); | |
a0ed51b3 LW |
2084 | if (UTF && FOLD) { |
2085 | if (LOC) | |
2086 | ender = toLOWER_LC_uni(ender); | |
2087 | else | |
2088 | ender = toLOWER_uni(ender); | |
2089 | } | |
a0d0e21e LW |
2090 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
2091 | if (len) | |
2092 | p = oldp; | |
a0ed51b3 LW |
2093 | else if (ender >= 0x80 && UTF) { |
2094 | reguni(ender, s, &numlen); | |
2095 | s += numlen; | |
2096 | len += numlen; | |
2097 | } | |
a0d0e21e LW |
2098 | else { |
2099 | len++; | |
cd439c50 | 2100 | REGC(ender, s++); |
a0d0e21e LW |
2101 | } |
2102 | break; | |
a687059c | 2103 | } |
a0ed51b3 LW |
2104 | if (ender >= 0x80 && UTF) { |
2105 | reguni(ender, s, &numlen); | |
2106 | s += numlen; | |
2107 | len += numlen - 1; | |
2108 | } | |
2109 | else | |
cd439c50 | 2110 | REGC(ender, s++); |
a0d0e21e LW |
2111 | } |
2112 | loopdone: | |
3280af22 | 2113 | PL_regcomp_parse = p - 1; |
a0d0e21e LW |
2114 | nextchar(); |
2115 | if (len < 0) | |
2116 | FAIL("internal disaster in regexp"); | |
2117 | if (len > 0) | |
2118 | *flagp |= HASWIDTH; | |
2119 | if (len == 1) | |
2120 | *flagp |= SIMPLE; | |
c277df42 | 2121 | if (!SIZE_ONLY) |
cd439c50 IZ |
2122 | STR_LEN(ret) = len; |
2123 | if (SIZE_ONLY) | |
2124 | PL_regsize += STR_SZ(len); | |
2125 | else | |
2126 | PL_regcode += STR_SZ(len); | |
a687059c | 2127 | } |
a0d0e21e LW |
2128 | break; |
2129 | } | |
a687059c | 2130 | |
a0d0e21e | 2131 | return(ret); |
a687059c LW |
2132 | } |
2133 | ||
873ef191 | 2134 | STATIC char * |
cea2e8a9 | 2135 | S_regwhite(pTHX_ char *p, char *e) |
5b5a24f7 CS |
2136 | { |
2137 | while (p < e) { | |
2138 | if (isSPACE(*p)) | |
2139 | ++p; | |
2140 | else if (*p == '#') { | |
2141 | do { | |
2142 | p++; | |
2143 | } while (p < e && *p != '\n'); | |
2144 | } | |
2145 | else | |
2146 | break; | |
2147 | } | |
2148 | return p; | |
2149 | } | |
2150 | ||
b8c5462f JH |
2151 | /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]]. |
2152 | Character classes ([:foo:]) can also be negated ([:^foo:]). | |
2153 | Returns a named class id (ANYOF_XXX) if successful, -1 otherwise. | |
2154 | Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed, | |
2155 | but trigger warnings because they are currently unimplemented. */ | |
2156 | STATIC I32 | |
cea2e8a9 | 2157 | S_regpposixcc(pTHX_ I32 value) |
620e46c5 | 2158 | { |
11b8faa4 | 2159 | dTHR; |
620e46c5 | 2160 | char *posixcc = 0; |
b8c5462f | 2161 | I32 namedclass = -1; |
620e46c5 JH |
2162 | |
2163 | if (value == '[' && PL_regcomp_parse + 1 < PL_regxend && | |
2164 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ | |
2165 | (*PL_regcomp_parse == ':' || | |
2166 | *PL_regcomp_parse == '=' || | |
2167 | *PL_regcomp_parse == '.')) { | |
2168 | char c = *PL_regcomp_parse; | |
2169 | char* s = PL_regcomp_parse++; | |
2170 | ||
2171 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c) | |
2172 | PL_regcomp_parse++; | |
2173 | if (PL_regcomp_parse == PL_regxend) | |
2174 | /* Grandfather lone [:, [=, [. */ | |
2175 | PL_regcomp_parse = s; | |
2176 | else { | |
b8c5462f JH |
2177 | char* t = PL_regcomp_parse++; /* skip over the c */ |
2178 | ||
2179 | if (*PL_regcomp_parse == ']') { | |
2180 | PL_regcomp_parse++; /* skip over the ending ] */ | |
2181 | posixcc = s + 1; | |
2182 | if (*s == ':') { | |
2183 | I32 complement = *posixcc == '^' ? *posixcc++ : 0; | |
2184 | I32 skip = 5; /* the most common skip */ | |
2185 | ||
2186 | switch (*posixcc) { | |
2187 | case 'a': | |
2188 | if (strnEQ(posixcc, "alnum", 5)) | |
2189 | namedclass = | |
2190 | complement ? ANYOF_NALNUMC : ANYOF_ALNUMC; | |
2191 | else if (strnEQ(posixcc, "alpha", 5)) | |
2192 | namedclass = | |
2193 | complement ? ANYOF_NALPHA : ANYOF_ALPHA; | |
2194 | else if (strnEQ(posixcc, "ascii", 5)) | |
2195 | namedclass = | |
2196 | complement ? ANYOF_NASCII : ANYOF_ASCII; | |
2197 | break; | |
2198 | case 'c': | |
2199 | if (strnEQ(posixcc, "cntrl", 5)) | |
2200 | namedclass = | |
2201 | complement ? ANYOF_NCNTRL : ANYOF_CNTRL; | |
2202 | break; | |
2203 | case 'd': | |
2204 | if (strnEQ(posixcc, "digit", 5)) | |
2205 | namedclass = | |
2206 | complement ? ANYOF_NDIGIT : ANYOF_DIGIT; | |
2207 | break; | |
2208 | case 'g': | |
2209 | if (strnEQ(posixcc, "graph", 5)) | |
2210 | namedclass = | |
2211 | complement ? ANYOF_NGRAPH : ANYOF_GRAPH; | |
2212 | break; | |
2213 | case 'l': | |
2214 | if (strnEQ(posixcc, "lower", 5)) | |
2215 | namedclass = | |
2216 | complement ? ANYOF_NLOWER : ANYOF_LOWER; | |
2217 | break; | |
2218 | case 'p': | |
2219 | if (strnEQ(posixcc, "print", 5)) | |
2220 | namedclass = | |
2221 | complement ? ANYOF_NPRINT : ANYOF_PRINT; | |
2222 | else if (strnEQ(posixcc, "punct", 5)) | |
2223 | namedclass = | |
2224 | complement ? ANYOF_NPUNCT : ANYOF_PUNCT; | |
2225 | break; | |
2226 | case 's': | |
2227 | if (strnEQ(posixcc, "space", 5)) | |
2228 | namedclass = | |
2229 | complement ? ANYOF_NSPACE : ANYOF_SPACE; | |
2230 | case 'u': | |
2231 | if (strnEQ(posixcc, "upper", 5)) | |
2232 | namedclass = | |
2233 | complement ? ANYOF_NUPPER : ANYOF_UPPER; | |
2234 | break; | |
2235 | case 'w': /* this is not POSIX, this is the Perl \w */ | |
2236 | if (strnEQ(posixcc, "word", 4)) { | |
2237 | namedclass = | |
2238 | complement ? ANYOF_NALNUM : ANYOF_ALNUM; | |
2239 | skip = 4; | |
2240 | } | |
2241 | break; | |
2242 | case 'x': | |
2243 | if (strnEQ(posixcc, "xdigit", 6)) { | |
2244 | namedclass = | |
2245 | complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT; | |
2246 | skip = 6; | |
2247 | } | |
2248 | break; | |
2249 | } | |
93733859 | 2250 | if ((namedclass == OOB_NAMEDCLASS || |
b8c5462f JH |
2251 | !(posixcc + skip + 2 < PL_regxend && |
2252 | (posixcc[skip] == ':' && | |
2253 | posixcc[skip + 1] == ']')))) | |
2254 | Perl_croak(aTHX_ "Character class [:%.*s:] unknown", | |
2255 | t - s - 1, s + 1); | |
2256 | } else if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY) | |
2257 | /* [[=foo=]] and [[.foo.]] are still future. */ | |
cea2e8a9 | 2258 | Perl_warner(aTHX_ WARN_UNSAFE, |
b8c5462f JH |
2259 | "Character class syntax [%c %c] is reserved for future extensions", c, c); |
2260 | } else { | |
2261 | /* Maternal grandfather: | |
2262 | * "[:" ending in ":" but not in ":]" */ | |
767d463e HS |
2263 | PL_regcomp_parse = s; |
2264 | } | |
620e46c5 JH |
2265 | } |
2266 | } | |
2267 | ||
b8c5462f JH |
2268 | return namedclass; |
2269 | } | |
2270 | ||
2271 | STATIC void | |
2272 | S_checkposixcc(pTHX) | |
2273 | { | |
2274 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && | |
2275 | (*PL_regcomp_parse == ':' || | |
2276 | *PL_regcomp_parse == '=' || | |
2277 | *PL_regcomp_parse == '.')) { | |
2278 | char *s = PL_regcomp_parse; | |
93733859 | 2279 | char c = *s++; |
b8c5462f JH |
2280 | |
2281 | while(*s && isALNUM(*s)) | |
2282 | s++; | |
2283 | if (*s && c == *s && s[1] == ']') { | |
2284 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2285 | "Character class syntax [%c %c] belongs inside character classes", c, c); | |
2286 | if (c == '=' || c == '.') | |
2287 | Perl_warner(aTHX_ WARN_UNSAFE, | |
2288 | "Character class syntax [%c %c] is reserved for future extensions", c, c); | |
2289 | } | |
2290 | } | |
620e46c5 JH |
2291 | } |
2292 | ||
76e3520e | 2293 | STATIC regnode * |
cea2e8a9 | 2294 | S_regclass(pTHX) |
a687059c | 2295 | { |
5c0ca799 | 2296 | dTHR; |
c277df42 | 2297 | register char *opnd, *s; |
a0ed51b3 | 2298 | register I32 value; |
b8c5462f | 2299 | register I32 lastvalue = OOB_CHAR8; |
a0d0e21e | 2300 | register I32 range = 0; |
c277df42 | 2301 | register regnode *ret; |
a0d0e21e LW |
2302 | register I32 def; |
2303 | I32 numlen; | |
b8c5462f | 2304 | I32 namedclass; |
a0d0e21e | 2305 | |
cd439c50 | 2306 | s = opnd = MASK(PL_regcode); |
c277df42 | 2307 | ret = reg_node(ANYOF); |
b8c5462f | 2308 | for (value = 0; value < ANYOF_SIZE; value++) |
cd439c50 | 2309 | REGC(0, s++); |
3280af22 NIS |
2310 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2311 | PL_regnaughty++; | |
2312 | PL_regcomp_parse++; | |
c277df42 | 2313 | if (!SIZE_ONLY) |
b8c5462f | 2314 | ANYOF_FLAGS(opnd) |= ANYOF_INVERT; |
bbce6d69 | 2315 | } |
c277df42 | 2316 | if (!SIZE_ONLY) { |
3280af22 | 2317 | PL_regcode += ANY_SKIP; |
a0ed51b3 | 2318 | if (FOLD) |
b8c5462f | 2319 | ANYOF_FLAGS(opnd) |= ANYOF_FOLD; |
a0ed51b3 | 2320 | if (LOC) |
b8c5462f | 2321 | ANYOF_FLAGS(opnd) |= ANYOF_LOCALE; |
a0ed51b3 LW |
2322 | } |
2323 | else { | |
3280af22 | 2324 | PL_regsize += ANY_SKIP; |
a0d0e21e | 2325 | } |
b8c5462f JH |
2326 | |
2327 | checkposixcc(); | |
2328 | ||
3280af22 | 2329 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
a0d0e21e | 2330 | goto skipcond; /* allow 1st char to be ] or - */ |
3280af22 | 2331 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
a0d0e21e | 2332 | skipcond: |
93733859 | 2333 | namedclass = OOB_NAMEDCLASS; |
a0ed51b3 | 2334 | value = UCHARAT(PL_regcomp_parse++); |
620e46c5 | 2335 | if (value == '[') |
b8c5462f | 2336 | namedclass = regpposixcc(value); |
620e46c5 | 2337 | else if (value == '\\') { |
a0ed51b3 LW |
2338 | value = UCHARAT(PL_regcomp_parse++); |
2339 | switch (value) { | |
b8c5462f JH |
2340 | case 'w': namedclass = ANYOF_ALNUM; break; |
2341 | case 'W': namedclass = ANYOF_NALNUM; break; | |
2342 | case 's': namedclass = ANYOF_SPACE; break; | |
2343 | case 'S': namedclass = ANYOF_NSPACE; break; | |
2344 | case 'd': namedclass = ANYOF_DIGIT; break; | |
2345 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
2346 | case 'n': value = '\n'; break; | |
2347 | case 'r': value = '\r'; break; | |
2348 | case 't': value = '\t'; break; | |
2349 | case 'f': value = '\f'; break; | |
2350 | case 'b': value = '\b'; break; | |
2351 | case 'e': value = '\033'; break; | |
2352 | case 'a': value = '\007'; break; | |
2353 | case 'x': | |
2354 | value = scan_hex(PL_regcomp_parse, 2, &numlen); | |
2355 | PL_regcomp_parse += numlen; | |
2356 | break; | |
2357 | case 'c': | |
2358 | value = UCHARAT(PL_regcomp_parse++); | |
2359 | value = toCTRL(value); | |
2360 | break; | |
2361 | case '0': case '1': case '2': case '3': case '4': | |
2362 | case '5': case '6': case '7': case '8': case '9': | |
2363 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); | |
2364 | PL_regcomp_parse += numlen; | |
2365 | break; | |
2366 | } | |
2367 | } | |
93733859 JH |
2368 | if (!SIZE_ONLY && namedclass > OOB_NAMEDCLASS) { |
2369 | if (range) | |
2370 | FAIL("invalid [] range in regexp"); /* [a-\w], [a-[:word:]] */ | |
b8c5462f JH |
2371 | switch (namedclass) { |
2372 | case ANYOF_ALNUM: | |
2373 | if (LOC) | |
2374 | ANYOF_CLASS_SET(opnd, ANYOF_ALNUM); | |
2375 | else { | |
2376 | for (value = 0; value < 256; value++) | |
2377 | if (isALNUM(value)) | |
2378 | ANYOF_BITMAP_SET(opnd, value); | |
bbce6d69 | 2379 | } |
b8c5462f JH |
2380 | break; |
2381 | case ANYOF_NALNUM: | |
2382 | if (LOC) | |
2383 | ANYOF_CLASS_SET(opnd, ANYOF_NALNUM); | |
2384 | else { | |
2385 | for (value = 0; value < 256; value++) | |
2386 | if (!isALNUM(value)) | |
2387 | ANYOF_BITMAP_SET(opnd, value); | |
bbce6d69 | 2388 | } |
b8c5462f JH |
2389 | break; |
2390 | case ANYOF_SPACE: | |
2391 | if (LOC) | |
2392 | ANYOF_CLASS_SET(opnd, ANYOF_SPACE); | |
2393 | else { | |
2394 | for (value = 0; value < 256; value++) | |
2395 | if (isSPACE(value)) | |
2396 | ANYOF_BITMAP_SET(opnd, value); | |
bbce6d69 | 2397 | } |
b8c5462f JH |
2398 | break; |
2399 | case ANYOF_NSPACE: | |
2400 | if (LOC) | |
2401 | ANYOF_CLASS_SET(opnd, ANYOF_NSPACE); | |
2402 | else { | |
2403 | for (value = 0; value < 256; value++) | |
2404 | if (!isSPACE(value)) | |
2405 | ANYOF_BITMAP_SET(opnd, value); | |
bbce6d69 | 2406 | } |
b8c5462f JH |
2407 | break; |
2408 | case ANYOF_DIGIT: | |
2409 | if (LOC) | |
2410 | ANYOF_CLASS_SET(opnd, ANYOF_DIGIT); | |
2411 | else { | |
a0ed51b3 | 2412 | for (value = '0'; value <= '9'; value++) |
b8c5462f | 2413 | ANYOF_BITMAP_SET(opnd, value); |
ae5c130c | 2414 | } |
b8c5462f JH |
2415 | break; |
2416 | case ANYOF_NDIGIT: | |
2417 | if (LOC) | |
2418 | ANYOF_CLASS_SET(opnd, ANYOF_NDIGIT); | |
2419 | else { | |
a0ed51b3 | 2420 | for (value = 0; value < '0'; value++) |
b8c5462f | 2421 | ANYOF_BITMAP_SET(opnd, value); |
a0ed51b3 | 2422 | for (value = '9' + 1; value < 256; value++) |
b8c5462f JH |
2423 | ANYOF_BITMAP_SET(opnd, value); |
2424 | } | |
2425 | break; | |
2426 | case ANYOF_NALNUMC: | |
2427 | if (LOC) | |
2428 | ANYOF_CLASS_SET(opnd, ANYOF_NALNUMC); | |
2429 | else { | |
2430 | for (value = 0; value < 256; value++) | |
2431 | if (!isALNUMC(value)) | |
2432 | ANYOF_BITMAP_SET(opnd, value); | |
2433 | } | |
2434 | break; | |
2435 | case ANYOF_ALNUMC: | |
2436 | if (LOC) | |
2437 | ANYOF_CLASS_SET(opnd, ANYOF_ALNUMC); | |
2438 | else { | |
2439 | for (value = 0; value < 256; value++) | |
2440 | if (isALNUMC(value)) | |
2441 | ANYOF_BITMAP_SET(opnd, value); | |
2442 | } | |
2443 | break; | |
2444 | case ANYOF_ALPHA: | |
2445 | if (LOC) | |
2446 | ANYOF_CLASS_SET(opnd, ANYOF_ALPHA); | |
2447 | else { | |
2448 | for (value = 0; value < 256; value++) | |
2449 | if (isALPHA(value)) | |
2450 | ANYOF_BITMAP_SET(opnd, value); | |
ae5c130c | 2451 | } |
a0d0e21e | 2452 | break; |
b8c5462f JH |
2453 | case ANYOF_NALPHA: |
2454 | if (LOC) | |
2455 | ANYOF_CLASS_SET(opnd, ANYOF_NALPHA); | |
2456 | else { | |
2457 | for (value = 0; value < 256; value++) | |
2458 | if (!isALPHA(value)) | |
2459 | ANYOF_BITMAP_SET(opnd, value); | |
2460 | } | |
a0d0e21e | 2461 | break; |
b8c5462f JH |
2462 | case ANYOF_ASCII: |
2463 | if (LOC) | |
2464 | ANYOF_CLASS_SET(opnd, ANYOF_ASCII); | |
2465 | else { | |
2466 | for (value = 0; value < 128; value++) | |
2467 | ANYOF_BITMAP_SET(opnd, value); | |
2468 | } | |
a0d0e21e | 2469 | break; |
b8c5462f JH |
2470 | case ANYOF_NASCII: |
2471 | if (LOC) | |
2472 | ANYOF_CLASS_SET(opnd, ANYOF_NASCII); | |
2473 | else { | |
2474 | for (value = 128; value < 256; value++) | |
2475 | ANYOF_BITMAP_SET(opnd, value); | |
2476 | } | |
a0d0e21e | 2477 | break; |
b8c5462f JH |
2478 | case ANYOF_CNTRL: |
2479 | if (LOC) | |
2480 | ANYOF_CLASS_SET(opnd, ANYOF_CNTRL); | |
2481 | else { | |
2482 | for (value = 0; value < 256; value++) | |
2483 | if (isCNTRL(value)) | |
2484 | ANYOF_BITMAP_SET(opnd, value); | |
2485 | } | |
2486 | lastvalue = OOB_CHAR8; | |
a0d0e21e | 2487 | break; |
b8c5462f JH |
2488 | case ANYOF_NCNTRL: |
2489 | if (LOC) | |
2490 | ANYOF_CLASS_SET(opnd, ANYOF_NCNTRL); | |
2491 | else { | |
2492 | for (value = 0; value < 256; value++) | |
2493 | if (!isCNTRL(value)) | |
2494 | ANYOF_BITMAP_SET(opnd, value); | |
2495 | } | |
a0d0e21e | 2496 | break; |
b8c5462f JH |
2497 | case ANYOF_GRAPH: |
2498 | if (LOC) | |
2499 | ANYOF_CLASS_SET(opnd, ANYOF_GRAPH); | |
2500 | else { | |
2501 | for (value = 0; value < 256; value++) | |
2502 | if (isGRAPH(value)) | |
2503 | ANYOF_BITMAP_SET(opnd, value); | |
2504 | } | |
a0d0e21e | 2505 | break; |
b8c5462f JH |
2506 | case ANYOF_NGRAPH: |
2507 | if (LOC) | |
2508 | ANYOF_CLASS_SET(opnd, ANYOF_NGRAPH); | |
2509 | else { | |
2510 | for (value = 0; value < 256; value++) | |
2511 | if (!isGRAPH(value)) | |
2512 | ANYOF_BITMAP_SET(opnd, value); | |
2513 | } | |
a0d0e21e | 2514 | break; |
b8c5462f JH |
2515 | case ANYOF_LOWER: |
2516 | if (LOC) | |
2517 | ANYOF_CLASS_SET(opnd, ANYOF_LOWER); | |
2518 | else { | |
2519 | for (value = 0; value < 256; value++) | |
2520 | if (isLOWER(value)) | |
2521 | ANYOF_BITMAP_SET(opnd, value); | |
2522 | } | |
a0d0e21e | 2523 | break; |
b8c5462f JH |
2524 | case ANYOF_NLOWER: |
2525 | if (LOC) | |
2526 | ANYOF_CLASS_SET(opnd, ANYOF_NLOWER); | |
2527 | else { | |
2528 | for (value = 0; value < 256; value++) | |
2529 | if (!isLOWER(value)) | |
2530 | ANYOF_BITMAP_SET(opnd, value); | |
2531 | } | |
2532 | break; | |
2533 | case ANYOF_PRINT: | |
2534 | if (LOC) | |
2535 | ANYOF_CLASS_SET(opnd, ANYOF_PRINT); | |
2536 | else { | |
2537 | for (value = 0; value < 256; value++) | |
2538 | if (isPRINT(value)) | |
2539 | ANYOF_BITMAP_SET(opnd, value); | |
2540 | } | |
2541 | break; | |
2542 | case ANYOF_NPRINT: | |
2543 | if (LOC) | |
2544 | ANYOF_CLASS_SET(opnd, ANYOF_NPRINT); | |
2545 | else { | |
2546 | for (value = 0; value < 256; value++) | |
2547 | if (!isPRINT(value)) | |
2548 | ANYOF_BITMAP_SET(opnd, value); | |
2549 | } | |
2550 | break; | |
2551 | case ANYOF_PUNCT: | |
2552 | if (LOC) | |
2553 | ANYOF_CLASS_SET(opnd, ANYOF_PUNCT); | |
2554 | else { | |
2555 | for (value = 0; value < 256; value++) | |
2556 | if (isPUNCT(value)) | |
2557 | ANYOF_BITMAP_SET(opnd, value); | |
2558 | } | |
2559 | break; | |
2560 | case ANYOF_NPUNCT: | |
2561 | if (LOC) | |
2562 | ANYOF_CLASS_SET(opnd, ANYOF_NPUNCT); | |
2563 | else { | |
2564 | for (value = 0; value < 256; value++) | |
2565 | if (!isPUNCT(value)) | |
2566 | ANYOF_BITMAP_SET(opnd, value); | |
2567 | } | |
2568 | break; | |
2569 | case ANYOF_UPPER: | |
2570 | if (LOC) | |
2571 | ANYOF_CLASS_SET(opnd, ANYOF_UPPER); | |
2572 | else { | |
2573 | for (value = 0; value < 256; value++) | |
2574 | if (isUPPER(value)) | |
2575 | ANYOF_BITMAP_SET(opnd, value); | |
2576 | } | |
2577 | break; | |
2578 | case ANYOF_NUPPER: | |
2579 | if (LOC) | |
2580 | ANYOF_CLASS_SET(opnd, ANYOF_NUPPER); | |
2581 | else { | |
2582 | for (value = 0; value < 256; value++) | |
2583 | if (!isUPPER(value)) | |
2584 | ANYOF_BITMAP_SET(opnd, value); | |
2585 | } | |
2586 | break; | |
2587 | case ANYOF_XDIGIT: | |
2588 | if (LOC) | |
2589 | ANYOF_CLASS_SET(opnd, ANYOF_XDIGIT); | |
2590 | else { | |
2591 | for (value = 0; value < 256; value++) | |
2592 | if (isXDIGIT(value)) | |
2593 | ANYOF_BITMAP_SET(opnd, value); | |
2594 | } | |
2595 | break; | |
2596 | case ANYOF_NXDIGIT: | |
2597 | if (LOC) | |
2598 | ANYOF_CLASS_SET(opnd, ANYOF_NXDIGIT); | |
2599 | else { | |
2600 | for (value = 0; value < 256; value++) | |
2601 | if (!isXDIGIT(value)) | |
2602 | ANYOF_BITMAP_SET(opnd, value); | |
2603 | } | |
2604 | break; | |
2605 | default: | |
2606 | FAIL("invalid [::] class in regexp"); | |
a0d0e21e LW |
2607 | break; |
2608 | } | |
b8c5462f JH |
2609 | if (LOC) |
2610 | ANYOF_FLAGS(opnd) |= ANYOF_CLASS; | |
93733859 | 2611 | continue; |
a0d0e21e LW |
2612 | } |
2613 | if (range) { | |
a0ed51b3 | 2614 | if (lastvalue > value) |
93733859 | 2615 | FAIL("invalid [] range in regexp"); /* [b-a] */ |
a0d0e21e LW |
2616 | range = 0; |
2617 | } | |
2618 | else { | |
a0ed51b3 | 2619 | lastvalue = value; |
3280af22 | 2620 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
93733859 JH |
2621 | PL_regcomp_parse[1] != ']') { |
2622 | if (namedclass > OOB_NAMEDCLASS) | |
2623 | FAIL("invalid [] range in regexp"); /* [\w-a] */ | |
3280af22 | 2624 | PL_regcomp_parse++; |
a0d0e21e LW |
2625 | range = 1; |
2626 | continue; /* do it next time */ | |
2627 | } | |
a687059c | 2628 | } |
93733859 | 2629 | /* now is the next time */ |
ae5c130c | 2630 | if (!SIZE_ONLY) { |
93733859 | 2631 | #ifndef ASCIIish /* EBCDIC, for example. */ |
8ada0baa | 2632 | if ((isLOWER(lastvalue) && isLOWER(value)) || |
c8dba6f3 GS |
2633 | (isUPPER(lastvalue) && isUPPER(value))) |
2634 | { | |
2635 | I32 i; | |
8ada0baa JH |
2636 | if (isLOWER(lastvalue)) { |
2637 | for (i = lastvalue; i <= value; i++) | |
2638 | if (isLOWER(i)) | |
b8c5462f | 2639 | ANYOF_BITMAP_SET(opnd, i); |
8ada0baa JH |
2640 | } else { |
2641 | for (i = lastvalue; i <= value; i++) | |
2642 | if (isUPPER(i)) | |
b8c5462f | 2643 | ANYOF_BITMAP_SET(opnd, i); |
8ada0baa JH |
2644 | } |
2645 | } | |
2646 | else | |
2647 | #endif | |
2648 | for ( ; lastvalue <= value; lastvalue++) | |
b8c5462f | 2649 | ANYOF_BITMAP_SET(opnd, lastvalue); |
8ada0baa | 2650 | } |
93733859 | 2651 | range = 0; |
a0d0e21e | 2652 | } |
ae5c130c | 2653 | /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ |
b8c5462f JH |
2654 | if (!SIZE_ONLY && |
2655 | (ANYOF_FLAGS(opnd) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD) { | |
a0ed51b3 | 2656 | for (value = 0; value < 256; ++value) { |
b8c5462f | 2657 | if (ANYOF_BITMAP_TEST(opnd, value)) { |
22c35a8c | 2658 | I32 cf = PL_fold[value]; |
b8c5462f | 2659 | ANYOF_BITMAP_SET(opnd, cf); |
ae5c130c GS |
2660 | } |
2661 | } | |
b8c5462f | 2662 | ANYOF_FLAGS(opnd) &= ~ANYOF_FOLD; |
ae5c130c GS |
2663 | } |
2664 | /* optimize inverted simple patterns (e.g. [^a-z]) */ | |
b8c5462f JH |
2665 | if (!SIZE_ONLY && (ANYOF_FLAGS(opnd) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) { |
2666 | for (value = 0; value < ANYOF_BITMAP_SIZE; ++value) | |
2667 | opnd[ANYOF_BITMAP_OFFSET + value] ^= ANYOF_FLAGS_ALL; | |
2668 | ANYOF_FLAGS(opnd) = 0; | |
ae5c130c | 2669 | } |
a0d0e21e LW |
2670 | return ret; |
2671 | } | |
2672 | ||
a0ed51b3 | 2673 | STATIC regnode * |
cea2e8a9 | 2674 | S_regclassutf8(pTHX) |
a0ed51b3 | 2675 | { |
b8c5462f | 2676 | dTHR; |
a0ed51b3 LW |
2677 | register char *opnd, *e; |
2678 | register U32 value; | |
b8c5462f | 2679 | register U32 lastvalue = OOB_UTF8; |
a0ed51b3 LW |
2680 | register I32 range = 0; |
2681 | register regnode *ret; | |
2682 | I32 numlen; | |
2683 | I32 n; | |
2684 | SV *listsv; | |
2685 | U8 flags = 0; | |
b8c5462f | 2686 | I32 namedclass; |
a0ed51b3 LW |
2687 | |
2688 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ | |
2689 | PL_regnaughty++; | |
2690 | PL_regcomp_parse++; | |
2691 | if (!SIZE_ONLY) | |
2692 | flags |= ANYOF_INVERT; | |
2693 | } | |
2694 | if (!SIZE_ONLY) { | |
2695 | if (FOLD) | |
2696 | flags |= ANYOF_FOLD; | |
2697 | if (LOC) | |
2698 | flags |= ANYOF_LOCALE; | |
79cb57f6 | 2699 | listsv = newSVpvn("# comment\n",10); |
a0ed51b3 LW |
2700 | } |
2701 | ||
b8c5462f JH |
2702 | checkposixcc(); |
2703 | ||
a0ed51b3 LW |
2704 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
2705 | goto skipcond; /* allow 1st char to be ] or - */ | |
2706 | ||
2707 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { | |
2708 | skipcond: | |
93733859 | 2709 | namedclass = OOB_NAMEDCLASS; |
dfe13c55 | 2710 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 LW |
2711 | PL_regcomp_parse += numlen; |
2712 | ||
620e46c5 | 2713 | if (value == '[') |
b8c5462f | 2714 | namedclass = regpposixcc(value); |
620e46c5 | 2715 | else if (value == '\\') { |
dfe13c55 | 2716 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 LW |
2717 | PL_regcomp_parse += numlen; |
2718 | switch (value) { | |
b8c5462f JH |
2719 | case 'w': namedclass = ANYOF_ALNUM; break; |
2720 | case 'W': namedclass = ANYOF_NALNUM; break; | |
2721 | case 's': namedclass = ANYOF_SPACE; break; | |
2722 | case 'S': namedclass = ANYOF_NSPACE; break; | |
2723 | case 'd': namedclass = ANYOF_DIGIT; break; | |
2724 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
a0ed51b3 LW |
2725 | case 'p': |
2726 | case 'P': | |
2727 | if (*PL_regcomp_parse == '{') { | |
2728 | e = strchr(PL_regcomp_parse++, '}'); | |
2729 | if (!e) | |
2730 | FAIL("Missing right brace on \\p{}"); | |
2731 | n = e - PL_regcomp_parse; | |
2732 | } | |
2733 | else { | |
2734 | e = PL_regcomp_parse; | |
2735 | n = 1; | |
2736 | } | |
2737 | if (!SIZE_ONLY) { | |
2738 | if (value == 'p') | |
b8c5462f JH |
2739 | Perl_sv_catpvf(aTHX_ listsv, |
2740 | "+utf8::%.*s\n", n, PL_regcomp_parse); | |
a0ed51b3 | 2741 | else |
cea2e8a9 | 2742 | Perl_sv_catpvf(aTHX_ listsv, |
b8c5462f | 2743 | "!utf8::%.*s\n", n, PL_regcomp_parse); |
a0ed51b3 LW |
2744 | } |
2745 | PL_regcomp_parse = e + 1; | |
b8c5462f | 2746 | lastvalue = OOB_UTF8; |
a0ed51b3 | 2747 | continue; |
b8c5462f JH |
2748 | case 'n': value = '\n'; break; |
2749 | case 'r': value = '\r'; break; | |
2750 | case 't': value = '\t'; break; | |
2751 | case 'f': value = '\f'; break; | |
2752 | case 'b': value = '\b'; break; | |
2753 | case 'e': value = '\033'; break; | |
2754 | case 'a': value = '\007'; break; | |
a0ed51b3 LW |
2755 | case 'x': |
2756 | if (*PL_regcomp_parse == '{') { | |
2757 | e = strchr(PL_regcomp_parse++, '}'); | |
2758 | if (!e) | |
2759 | FAIL("Missing right brace on \\x{}"); | |
b8c5462f JH |
2760 | value = scan_hex(PL_regcomp_parse, |
2761 | e - PL_regcomp_parse, | |
2762 | &numlen); | |
a0ed51b3 LW |
2763 | PL_regcomp_parse = e + 1; |
2764 | } | |
2765 | else { | |
2766 | value = scan_hex(PL_regcomp_parse, 2, &numlen); | |
2767 | PL_regcomp_parse += numlen; | |
2768 | } | |
2769 | break; | |
2770 | case 'c': | |
2771 | value = UCHARAT(PL_regcomp_parse++); | |
2772 | value = toCTRL(value); | |
2773 | break; | |
2774 | case '0': case '1': case '2': case '3': case '4': | |
2775 | case '5': case '6': case '7': case '8': case '9': | |
2776 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); | |
2777 | PL_regcomp_parse += numlen; | |
2778 | break; | |
2779 | } | |
2780 | } | |
93733859 JH |
2781 | if (!SIZE_ONLY && namedclass > OOB_NAMEDCLASS) { |
2782 | if (range) | |
2783 | FAIL("invalid [] range in regexp"); /* [a-\w], [a-[:word:]] */ | |
b8c5462f JH |
2784 | switch (namedclass) { |
2785 | case ANYOF_ALNUM: | |
2786 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n"); break; | |
2787 | case ANYOF_NALNUM: | |
2788 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n"); break; | |
2789 | case ANYOF_ALNUMC: | |
2790 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n"); break; | |
2791 | case ANYOF_NALNUMC: | |
2792 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n"); break; | |
2793 | case ANYOF_ALPHA: | |
2794 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n"); break; | |
2795 | case ANYOF_NALPHA: | |
2796 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n"); break; | |
2797 | case ANYOF_ASCII: | |
2798 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n"); break; | |
2799 | case ANYOF_NASCII: | |
2800 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n"); break; | |
2801 | case ANYOF_CNTRL: | |
2802 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n"); break; | |
2803 | case ANYOF_NCNTRL: | |
2804 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n"); break; | |
2805 | case ANYOF_GRAPH: | |
2806 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n"); break; | |
2807 | case ANYOF_NGRAPH: | |
2808 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n"); break; | |
2809 | case ANYOF_DIGIT: | |
2810 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n"); break; | |
2811 | case ANYOF_NDIGIT: | |
2812 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n"); break; | |
2813 | case ANYOF_LOWER: | |
2814 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n"); break; | |
2815 | case ANYOF_NLOWER: | |
2816 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n"); break; | |
2817 | case ANYOF_PRINT: | |
2818 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n"); break; | |
2819 | case ANYOF_NPRINT: | |
2820 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n"); break; | |
2821 | case ANYOF_PUNCT: | |
2822 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n"); break; | |
2823 | case ANYOF_NPUNCT: | |
2824 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n"); break; | |
2825 | case ANYOF_SPACE: | |
2826 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n"); break; | |
2827 | case ANYOF_NSPACE: | |
2828 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n"); break; | |
2829 | case ANYOF_UPPER: | |
2830 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n"); break; | |
2831 | case ANYOF_NUPPER: | |
2832 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n"); break; | |
2833 | case ANYOF_XDIGIT: | |
2834 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n"); break; | |
2835 | case ANYOF_NXDIGIT: | |
2836 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n"); break; | |
2837 | } | |
93733859 | 2838 | continue; |
b8c5462f | 2839 | } |
b8c5462f | 2840 | if (range) { |
a0ed51b3 | 2841 | if (lastvalue > value) |
93733859 | 2842 | FAIL("invalid [] range in regexp"); /* [b-a] */ |
f478a152 | 2843 | #ifdef UV_IS_QUAD |
b9c39e73 | 2844 | if (!SIZE_ONLY) |
f478a152 JH |
2845 | Perl_sv_catpvf(aTHX_ listsv, "%04" PERL_PRIx64 "\t%04" PERL_PRIx64 "\n", (UV)lastvalue, (UV)value); |
2846 | #else | |
b9c39e73 | 2847 | if (!SIZE_ONLY) |
f478a152 JH |
2848 | Perl_sv_catpvf(aTHX_ listsv, "%04x\t%04x\n", lastvalue, value); |
2849 | #endif | |
a0ed51b3 LW |
2850 | range = 0; |
2851 | } | |
2852 | else { | |
2853 | lastvalue = value; | |
2854 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && | |
93733859 JH |
2855 | PL_regcomp_parse[1] != ']') { |
2856 | if (namedclass > OOB_NAMEDCLASS) | |
2857 | FAIL("invalid [] range in regexp"); /* [\w-a] */ | |
a0ed51b3 LW |
2858 | PL_regcomp_parse++; |
2859 | range = 1; | |
2860 | continue; /* do it next time */ | |
2861 | } | |
93733859 JH |
2862 | } |
2863 | /* now is the next time */ | |
f478a152 | 2864 | #ifdef UV_IS_QUAD |
93733859 JH |
2865 | if (!SIZE_ONLY) |
2866 | Perl_sv_catpvf(aTHX_ listsv, "%04" PERL_PRIx64 "\n", (UV)value); | |
f478a152 | 2867 | #else |
93733859 JH |
2868 | if (!SIZE_ONLY) |
2869 | Perl_sv_catpvf(aTHX_ listsv, "%04x\n", value); | |
f478a152 | 2870 | #endif |
93733859 | 2871 | range = 0; |
a0ed51b3 | 2872 | } |
a0ed51b3 LW |
2873 | |
2874 | ret = reganode(ANYOFUTF8, 0); | |
2875 | ||
2876 | if (!SIZE_ONLY) { | |
2877 | SV *rv = swash_init("utf8", "", listsv, 1, 0); | |
2878 | SvREFCNT_dec(listsv); | |
2879 | n = add_data(1,"s"); | |
2880 | PL_regcomp_rx->data->data[n] = (void*)rv; | |
2881 | ARG1_SET(ret, flags); | |
2882 | ARG2_SET(ret, n); | |
2883 | } | |
2884 | ||
2885 | return ret; | |
2886 | } | |
2887 | ||
76e3520e | 2888 | STATIC char* |
cea2e8a9 | 2889 | S_nextchar(pTHX) |
a0d0e21e | 2890 | { |
5c0ca799 | 2891 | dTHR; |
3280af22 | 2892 | char* retval = PL_regcomp_parse++; |
a0d0e21e | 2893 | |
4633a7c4 | 2894 | for (;;) { |
3280af22 NIS |
2895 | if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' && |
2896 | PL_regcomp_parse[2] == '#') { | |
2897 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') | |
2898 | PL_regcomp_parse++; | |
2899 | PL_regcomp_parse++; | |
4633a7c4 LW |
2900 | continue; |
2901 | } | |
3280af22 NIS |
2902 | if (PL_regflags & PMf_EXTENDED) { |
2903 | if (isSPACE(*PL_regcomp_parse)) { | |
2904 | PL_regcomp_parse++; | |
748a9306 LW |
2905 | continue; |
2906 | } | |
3280af22 NIS |
2907 | else if (*PL_regcomp_parse == '#') { |
2908 | while (*PL_regcomp_parse && *PL_regcomp_parse != '\n') | |
2909 | PL_regcomp_parse++; | |
2910 | PL_regcomp_parse++; | |
748a9306 LW |
2911 | continue; |
2912 | } | |
748a9306 | 2913 | } |
4633a7c4 | 2914 | return retval; |
a0d0e21e | 2915 | } |
a687059c LW |
2916 | } |
2917 | ||
2918 | /* | |
c277df42 | 2919 | - reg_node - emit a node |
a0d0e21e | 2920 | */ |
76e3520e | 2921 | STATIC regnode * /* Location. */ |
cea2e8a9 | 2922 | S_reg_node(pTHX_ U8 op) |
a687059c | 2923 | { |
5c0ca799 | 2924 | dTHR; |
c277df42 IZ |
2925 | register regnode *ret; |
2926 | register regnode *ptr; | |
a687059c | 2927 | |
3280af22 | 2928 | ret = PL_regcode; |
c277df42 | 2929 | if (SIZE_ONLY) { |
6b88bc9c | 2930 | SIZE_ALIGN(PL_regsize); |
3280af22 | 2931 | PL_regsize += 1; |
a0d0e21e LW |
2932 | return(ret); |
2933 | } | |
a687059c | 2934 | |
c277df42 | 2935 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 2936 | ptr = ret; |
c277df42 | 2937 | FILL_ADVANCE_NODE(ptr, op); |
3280af22 | 2938 | PL_regcode = ptr; |
a687059c | 2939 | |
a0d0e21e | 2940 | return(ret); |
a687059c LW |
2941 | } |
2942 | ||
2943 | /* | |
a0d0e21e LW |
2944 | - reganode - emit a node with an argument |
2945 | */ | |
76e3520e | 2946 | STATIC regnode * /* Location. */ |
cea2e8a9 | 2947 | S_reganode(pTHX_ U8 op, U32 arg) |
fe14fcc3 | 2948 | { |
5c0ca799 | 2949 | dTHR; |
c277df42 IZ |
2950 | register regnode *ret; |
2951 | register regnode *ptr; | |
fe14fcc3 | 2952 | |
3280af22 | 2953 | ret = PL_regcode; |
c277df42 | 2954 | if (SIZE_ONLY) { |
6b88bc9c | 2955 | SIZE_ALIGN(PL_regsize); |
3280af22 | 2956 | PL_regsize += 2; |
a0d0e21e LW |
2957 | return(ret); |
2958 | } | |
fe14fcc3 | 2959 | |
c277df42 | 2960 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 2961 | ptr = ret; |
c277df42 | 2962 | FILL_ADVANCE_NODE_ARG(ptr, op, arg); |
3280af22 | 2963 | PL_regcode = ptr; |
fe14fcc3 | 2964 | |
a0d0e21e | 2965 | return(ret); |
fe14fcc3 LW |
2966 | } |
2967 | ||
2968 | /* | |
cd439c50 | 2969 | - reguni - emit (if appropriate) a Unicode character |
a0ed51b3 LW |
2970 | */ |
2971 | STATIC void | |
cea2e8a9 | 2972 | S_reguni(pTHX_ UV uv, char* s, I32* lenp) |
a0ed51b3 | 2973 | { |
c485e607 | 2974 | dTHR; |
a0ed51b3 | 2975 | if (SIZE_ONLY) { |
dfe13c55 | 2976 | U8 tmpbuf[10]; |
a0ed51b3 LW |
2977 | *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf; |
2978 | } | |
2979 | else | |
dfe13c55 | 2980 | *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s; |
a0ed51b3 LW |
2981 | |
2982 | } | |
2983 | ||
2984 | /* | |
a0d0e21e LW |
2985 | - reginsert - insert an operator in front of already-emitted operand |
2986 | * | |
2987 | * Means relocating the operand. | |
2988 | */ | |
76e3520e | 2989 | STATIC void |
cea2e8a9 | 2990 | S_reginsert(pTHX_ U8 op, regnode *opnd) |
a687059c | 2991 | { |
5c0ca799 | 2992 | dTHR; |
c277df42 IZ |
2993 | register regnode *src; |
2994 | register regnode *dst; | |
2995 | register regnode *place; | |
2996 | register int offset = regarglen[(U8)op]; | |
2997 | ||
22c35a8c | 2998 | /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */ |
c277df42 IZ |
2999 | |
3000 | if (SIZE_ONLY) { | |
3280af22 | 3001 | PL_regsize += NODE_STEP_REGNODE + offset; |
a0d0e21e LW |
3002 | return; |
3003 | } | |
a687059c | 3004 | |
3280af22 NIS |
3005 | src = PL_regcode; |
3006 | PL_regcode += NODE_STEP_REGNODE + offset; | |
3007 | dst = PL_regcode; | |
a0d0e21e | 3008 | while (src > opnd) |
c277df42 | 3009 | StructCopy(--src, --dst, regnode); |
a0d0e21e LW |
3010 | |
3011 | place = opnd; /* Op node, where operand used to be. */ | |
c277df42 IZ |
3012 | src = NEXTOPER(place); |
3013 | FILL_ADVANCE_NODE(place, op); | |
3014 | Zero(src, offset, regnode); | |
a687059c LW |
3015 | } |
3016 | ||
3017 | /* | |
c277df42 | 3018 | - regtail - set the next-pointer at the end of a node chain of p to val. |
a0d0e21e | 3019 | */ |
76e3520e | 3020 | STATIC void |
cea2e8a9 | 3021 | S_regtail(pTHX_ regnode *p, regnode *val) |
a687059c | 3022 | { |
5c0ca799 | 3023 | dTHR; |
c277df42 IZ |
3024 | register regnode *scan; |
3025 | register regnode *temp; | |
a0d0e21e LW |
3026 | register I32 offset; |
3027 | ||
c277df42 | 3028 | if (SIZE_ONLY) |
a0d0e21e LW |
3029 | return; |
3030 | ||
3031 | /* Find last node. */ | |
3032 | scan = p; | |
3033 | for (;;) { | |
3034 | temp = regnext(scan); | |
3035 | if (temp == NULL) | |
3036 | break; | |
3037 | scan = temp; | |
3038 | } | |
a687059c | 3039 | |
c277df42 IZ |
3040 | if (reg_off_by_arg[OP(scan)]) { |
3041 | ARG_SET(scan, val - scan); | |
a0ed51b3 LW |
3042 | } |
3043 | else { | |
c277df42 IZ |
3044 | NEXT_OFF(scan) = val - scan; |
3045 | } | |
a687059c LW |
3046 | } |
3047 | ||
3048 | /* | |
a0d0e21e LW |
3049 | - regoptail - regtail on operand of first argument; nop if operandless |
3050 | */ | |
76e3520e | 3051 | STATIC void |
cea2e8a9 | 3052 | S_regoptail(pTHX_ regnode *p, regnode *val) |
a687059c | 3053 | { |
5c0ca799 | 3054 | dTHR; |
a0d0e21e | 3055 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
c277df42 IZ |
3056 | if (p == NULL || SIZE_ONLY) |
3057 | return; | |
22c35a8c | 3058 | if (PL_regkind[(U8)OP(p)] == BRANCH) { |
c277df42 | 3059 | regtail(NEXTOPER(p), val); |
a0ed51b3 | 3060 | } |
22c35a8c | 3061 | else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) { |
c277df42 | 3062 | regtail(NEXTOPER(NEXTOPER(p)), val); |
a0ed51b3 LW |
3063 | } |
3064 | else | |
a0d0e21e | 3065 | return; |
a687059c LW |
3066 | } |
3067 | ||
3068 | /* | |
3069 | - regcurly - a little FSA that accepts {\d+,?\d*} | |
3070 | */ | |
79072805 | 3071 | STATIC I32 |
cea2e8a9 | 3072 | S_regcurly(pTHX_ register char *s) |
a687059c LW |
3073 | { |
3074 | if (*s++ != '{') | |
3075 | return FALSE; | |
f0fcb552 | 3076 | if (!isDIGIT(*s)) |
a687059c | 3077 | return FALSE; |
f0fcb552 | 3078 | while (isDIGIT(*s)) |
a687059c LW |
3079 | s++; |
3080 | if (*s == ',') | |
3081 | s++; | |
f0fcb552 | 3082 | while (isDIGIT(*s)) |
a687059c LW |
3083 | s++; |
3084 | if (*s != '}') | |
3085 | return FALSE; | |
3086 | return TRUE; | |
3087 | } | |
3088 | ||
a687059c | 3089 | |
76e3520e | 3090 | STATIC regnode * |
cea2e8a9 | 3091 | S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l) |
c277df42 | 3092 | { |
35ff7856 | 3093 | #ifdef DEBUGGING |
f248d071 | 3094 | register U8 op = EXACT; /* Arbitrary non-END op. */ |
c277df42 IZ |
3095 | register regnode *next, *onode; |
3096 | ||
3097 | while (op != END && (!last || node < last)) { | |
3098 | /* While that wasn't END last time... */ | |
3099 | ||
3100 | NODE_ALIGN(node); | |
3101 | op = OP(node); | |
3102 | if (op == CLOSE) | |
3103 | l--; | |
3104 | next = regnext(node); | |
3105 | /* Where, what. */ | |
3106 | if (OP(node) == OPTIMIZED) | |
3107 | goto after_print; | |
3108 | regprop(sv, node); | |
54dc92de | 3109 | PerlIO_printf(Perl_debug_log, "%4d:%*s%s", node - start, |
c277df42 IZ |
3110 | 2*l + 1, "", SvPVX(sv)); |
3111 | if (next == NULL) /* Next ptr. */ | |
3112 | PerlIO_printf(Perl_debug_log, "(0)"); | |
3113 | else | |
3114 | PerlIO_printf(Perl_debug_log, "(%d)", next - start); | |
3115 | (void)PerlIO_putc(Perl_debug_log, '\n'); | |
3116 | after_print: | |
22c35a8c | 3117 | if (PL_regkind[(U8)op] == BRANCHJ) { |
c277df42 IZ |
3118 | register regnode *nnode = (OP(next) == LONGJMP |
3119 | ? regnext(next) | |
3120 | : next); | |
3121 | if (last && nnode > last) | |
3122 | nnode = last; | |
3123 | node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1); | |
a0ed51b3 | 3124 | } |
22c35a8c | 3125 | else if (PL_regkind[(U8)op] == BRANCH) { |
c277df42 | 3126 | node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1); |
a0ed51b3 LW |
3127 | } |
3128 | else if ( op == CURLY) { /* `next' might be very big: optimizer */ | |
c277df42 IZ |
3129 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3130 | NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1); | |
a0ed51b3 | 3131 | } |
22c35a8c | 3132 | else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) { |
c277df42 IZ |
3133 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3134 | next, sv, l + 1); | |
a0ed51b3 LW |
3135 | } |
3136 | else if ( op == PLUS || op == STAR) { | |
c277df42 | 3137 | node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1); |
a0ed51b3 LW |
3138 | } |
3139 | else if (op == ANYOF) { | |
c277df42 IZ |
3140 | node = NEXTOPER(node); |
3141 | node += ANY_SKIP; | |
a0ed51b3 | 3142 | } |
22c35a8c | 3143 | else if (PL_regkind[(U8)op] == EXACT) { |
c277df42 | 3144 | /* Literal string, where present. */ |
cd439c50 | 3145 | node += NODE_SZ_STR(node) - 1; |
c277df42 | 3146 | node = NEXTOPER(node); |
a0ed51b3 LW |
3147 | } |
3148 | else { | |
c277df42 IZ |
3149 | node = NEXTOPER(node); |
3150 | node += regarglen[(U8)op]; | |
3151 | } | |
3152 | if (op == CURLYX || op == OPEN) | |
3153 | l++; | |
3154 | else if (op == WHILEM) | |
3155 | l--; | |
3156 | } | |
17c3b450 | 3157 | #endif /* DEBUGGING */ |
c277df42 IZ |
3158 | return node; |
3159 | } | |
3160 | ||
a687059c | 3161 | /* |
fd181c75 | 3162 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c LW |
3163 | */ |
3164 | void | |
864dbfa3 | 3165 | Perl_regdump(pTHX_ regexp *r) |
a687059c | 3166 | { |
35ff7856 | 3167 | #ifdef DEBUGGING |
5c0ca799 | 3168 | dTHR; |
46fc3d4c | 3169 | SV *sv = sv_newmortal(); |
a687059c | 3170 | |
c277df42 | 3171 | (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0); |
a0d0e21e LW |
3172 | |
3173 | /* Header fields of interest. */ | |
c277df42 | 3174 | if (r->anchored_substr) |
2c2d71f5 | 3175 | PerlIO_printf(Perl_debug_log, "anchored `%s%.*s%s'%s at %d ", |
3280af22 | 3176 | PL_colors[0], |
2c2d71f5 | 3177 | SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0), |
c277df42 | 3178 | SvPVX(r->anchored_substr), |
3280af22 | 3179 | PL_colors[1], |
c277df42 IZ |
3180 | SvTAIL(r->anchored_substr) ? "$" : "", |
3181 | r->anchored_offset); | |
3182 | if (r->float_substr) | |
2c2d71f5 | 3183 | PerlIO_printf(Perl_debug_log, "floating `%s%.*s%s'%s at %d..%u ", |
3280af22 | 3184 | PL_colors[0], |
2c2d71f5 JH |
3185 | SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0), |
3186 | SvPVX(r->float_substr), | |
3280af22 | 3187 | PL_colors[1], |
c277df42 IZ |
3188 | SvTAIL(r->float_substr) ? "$" : "", |
3189 | r->float_min_offset, r->float_max_offset); | |
3190 | if (r->check_substr) | |
3191 | PerlIO_printf(Perl_debug_log, | |
3192 | r->check_substr == r->float_substr | |
3193 | ? "(checking floating" : "(checking anchored"); | |
3194 | if (r->reganch & ROPT_NOSCAN) | |
3195 | PerlIO_printf(Perl_debug_log, " noscan"); | |
3196 | if (r->reganch & ROPT_CHECK_ALL) | |
3197 | PerlIO_printf(Perl_debug_log, " isall"); | |
3198 | if (r->check_substr) | |
3199 | PerlIO_printf(Perl_debug_log, ") "); | |
3200 | ||
46fc3d4c | 3201 | if (r->regstclass) { |
3202 | regprop(sv, r->regstclass); | |
3203 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv)); | |
3204 | } | |
774d564b | 3205 | if (r->reganch & ROPT_ANCH) { |
3206 | PerlIO_printf(Perl_debug_log, "anchored"); | |
3207 | if (r->reganch & ROPT_ANCH_BOL) | |
3208 | PerlIO_printf(Perl_debug_log, "(BOL)"); | |
c277df42 IZ |
3209 | if (r->reganch & ROPT_ANCH_MBOL) |
3210 | PerlIO_printf(Perl_debug_log, "(MBOL)"); | |
cad2e5aa JH |
3211 | if (r->reganch & ROPT_ANCH_SBOL) |
3212 | PerlIO_printf(Perl_debug_log, "(SBOL)"); | |
774d564b | 3213 | if (r->reganch & ROPT_ANCH_GPOS) |
3214 | PerlIO_printf(Perl_debug_log, "(GPOS)"); | |
3215 | PerlIO_putc(Perl_debug_log, ' '); | |
3216 | } | |
c277df42 IZ |
3217 | if (r->reganch & ROPT_GPOS_SEEN) |
3218 | PerlIO_printf(Perl_debug_log, "GPOS "); | |
a0d0e21e | 3219 | if (r->reganch & ROPT_SKIP) |
760ac839 | 3220 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e | 3221 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 | 3222 | PerlIO_printf(Perl_debug_log, "implicit "); |
760ac839 | 3223 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
ce862d02 IZ |
3224 | if (r->reganch & ROPT_EVAL_SEEN) |
3225 | PerlIO_printf(Perl_debug_log, "with eval "); | |
760ac839 | 3226 | PerlIO_printf(Perl_debug_log, "\n"); |
17c3b450 | 3227 | #endif /* DEBUGGING */ |
a687059c LW |
3228 | } |
3229 | ||
3230 | /* | |
a0d0e21e LW |
3231 | - regprop - printable representation of opcode |
3232 | */ | |
46fc3d4c | 3233 | void |
864dbfa3 | 3234 | Perl_regprop(pTHX_ SV *sv, regnode *o) |
a687059c | 3235 | { |
35ff7856 | 3236 | #ifdef DEBUGGING |
5c0ca799 | 3237 | dTHR; |
9b155405 | 3238 | register int k; |
a0d0e21e | 3239 | |
54dc92de | 3240 | sv_setpvn(sv, "", 0); |
9b155405 | 3241 | if (OP(o) >= reg_num) /* regnode.type is unsigned */ |
a0d0e21e | 3242 | FAIL("corrupted regexp opcode"); |
9b155405 IZ |
3243 | sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */ |
3244 | ||
3245 | k = PL_regkind[(U8)OP(o)]; | |
3246 | ||
3247 | if (k == EXACT) | |
cd439c50 IZ |
3248 | Perl_sv_catpvf(aTHX_ sv, " <%s%*s%s>", PL_colors[0], |
3249 | STR_LEN(o), STRING(o), PL_colors[1]); | |
9b155405 IZ |
3250 | else if (k == CURLY) { |
3251 | if (OP(o) == CURLYM || OP(o) == CURLYN) | |
cea2e8a9 GS |
3252 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */ |
3253 | Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o)); | |
a0d0e21e | 3254 | } |
2c2d71f5 JH |
3255 | else if (k == WHILEM && o->flags) /* Ordinal/of */ |
3256 | Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4); | |
9b155405 | 3257 | else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP ) |
cea2e8a9 | 3258 | Perl_sv_catpvf(aTHX_ sv, "%d", ARG(o)); /* Parenth number */ |
9b155405 | 3259 | else if (k == LOGICAL) |
cea2e8a9 | 3260 | Perl_sv_catpvf(aTHX_ sv, "[%d]", ARG(o)); /* 2: embedded, otherwise 1 */ |
9b155405 | 3261 | else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH)) |
cea2e8a9 | 3262 | Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags); |
17c3b450 | 3263 | #endif /* DEBUGGING */ |
35ff7856 | 3264 | } |
a687059c | 3265 | |
cad2e5aa JH |
3266 | SV * |
3267 | Perl_re_intuit_string(pTHX_ regexp *prog) | |
3268 | { /* Assume that RE_INTUIT is set */ | |
3269 | DEBUG_r( | |
3270 | { STRLEN n_a; | |
3271 | char *s = SvPV(prog->check_substr,n_a); | |
3272 | ||
3273 | if (!PL_colorset) reginitcolors(); | |
3274 | PerlIO_printf(Perl_debug_log, | |
3275 | "%sUsing REx substr:%s `%s%.60s%s%s'\n", | |
3276 | PL_colors[4],PL_colors[5],PL_colors[0], | |
3277 | s, | |
3278 | PL_colors[1], | |
3279 | (strlen(s) > 60 ? "..." : "")); | |
3280 | } ); | |
3281 | ||
3282 | return prog->check_substr; | |
3283 | } | |
3284 | ||
2b69d0c2 | 3285 | void |
864dbfa3 | 3286 | Perl_pregfree(pTHX_ struct regexp *r) |
a687059c | 3287 | { |
5c0ca799 | 3288 | dTHR; |
adac82c7 | 3289 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
cad2e5aa JH |
3290 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
3291 | "%sFreeing REx:%s `%s%.60s%s%s'\n", | |
3292 | PL_colors[4],PL_colors[5],PL_colors[0], | |
3293 | r->precomp, | |
3294 | PL_colors[1], | |
3295 | (strlen(r->precomp) > 60 ? "..." : ""))); | |
3296 | ||
3297 | ||
c277df42 | 3298 | if (!r || (--r->refcnt > 0)) |
a0d0e21e | 3299 | return; |
c277df42 | 3300 | if (r->precomp) |
a0d0e21e | 3301 | Safefree(r->precomp); |
cf93c79d IZ |
3302 | if (RX_MATCH_COPIED(r)) |
3303 | Safefree(r->subbeg); | |
a193d654 GS |
3304 | if (r->substrs) { |
3305 | if (r->anchored_substr) | |
3306 | SvREFCNT_dec(r->anchored_substr); | |
3307 | if (r->float_substr) | |
3308 | SvREFCNT_dec(r->float_substr); | |
2779dcf1 | 3309 | Safefree(r->substrs); |
a193d654 | 3310 | } |
c277df42 IZ |
3311 | if (r->data) { |
3312 | int n = r->data->count; | |
dfad63ad HS |
3313 | AV* new_comppad = NULL; |
3314 | AV* old_comppad; | |
3315 | SV** old_curpad; | |
3316 | ||
c277df42 IZ |
3317 | while (--n >= 0) { |
3318 | switch (r->data->what[n]) { | |
3319 | case 's': | |
3320 | SvREFCNT_dec((SV*)r->data->data[n]); | |
3321 | break; | |
dfad63ad HS |
3322 | case 'p': |
3323 | new_comppad = (AV*)r->data->data[n]; | |
3324 | break; | |
c277df42 | 3325 | case 'o': |
dfad63ad | 3326 | if (new_comppad == NULL) |
cea2e8a9 | 3327 | Perl_croak(aTHX_ "panic: pregfree comppad"); |
dfad63ad HS |
3328 | old_comppad = PL_comppad; |
3329 | old_curpad = PL_curpad; | |
3330 | PL_comppad = new_comppad; | |
3331 | PL_curpad = AvARRAY(new_comppad); | |
c277df42 | 3332 | op_free((OP_4tree*)r->data->data[n]); |
dfad63ad HS |
3333 | PL_comppad = old_comppad; |
3334 | PL_curpad = old_curpad; | |
3335 | SvREFCNT_dec((SV*)new_comppad); | |
3336 | new_comppad = NULL; | |
c277df42 IZ |
3337 | break; |
3338 | case 'n': | |
3339 | break; | |
3340 | default: | |
3341 | FAIL2("panic: regfree data code '%c'", r->data->what[n]); | |
3342 | } | |
3343 | } | |
3344 | Safefree(r->data->what); | |
3345 | Safefree(r->data); | |
a0d0e21e LW |
3346 | } |
3347 | Safefree(r->startp); | |
3348 | Safefree(r->endp); | |
3349 | Safefree(r); | |
a687059c | 3350 | } |
c277df42 IZ |
3351 | |
3352 | /* | |
3353 | - regnext - dig the "next" pointer out of a node | |
3354 | * | |
3355 | * [Note, when REGALIGN is defined there are two places in regmatch() | |
3356 | * that bypass this code for speed.] | |
3357 | */ | |
3358 | regnode * | |
864dbfa3 | 3359 | Perl_regnext(pTHX_ register regnode *p) |
c277df42 | 3360 | { |
5c0ca799 | 3361 | dTHR; |
c277df42 IZ |
3362 | register I32 offset; |
3363 | ||
3280af22 | 3364 | if (p == &PL_regdummy) |
c277df42 IZ |
3365 | return(NULL); |
3366 | ||
3367 | offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p)); | |
3368 | if (offset == 0) | |
3369 | return(NULL); | |
3370 | ||
c277df42 | 3371 | return(p+offset); |
c277df42 IZ |
3372 | } |
3373 | ||
01f988be | 3374 | STATIC void |
cea2e8a9 | 3375 | S_re_croak2(pTHX_ const char* pat1,const char* pat2,...) |
c277df42 IZ |
3376 | { |
3377 | va_list args; | |
3378 | STRLEN l1 = strlen(pat1); | |
3379 | STRLEN l2 = strlen(pat2); | |
3380 | char buf[512]; | |
06bf62c7 | 3381 | SV *msv; |
c277df42 IZ |
3382 | char *message; |
3383 | ||
3384 | if (l1 > 510) | |
3385 | l1 = 510; | |
3386 | if (l1 + l2 > 510) | |
3387 | l2 = 510 - l1; | |
3388 | Copy(pat1, buf, l1 , char); | |
3389 | Copy(pat2, buf + l1, l2 , char); | |
3b818b81 GS |
3390 | buf[l1 + l2] = '\n'; |
3391 | buf[l1 + l2 + 1] = '\0'; | |
8736538c AS |
3392 | #ifdef I_STDARG |
3393 | /* ANSI variant takes additional second argument */ | |
c277df42 | 3394 | va_start(args, pat2); |
8736538c AS |
3395 | #else |
3396 | va_start(args); | |
3397 | #endif | |
06bf62c7 | 3398 | msv = mess(buf, &args); |
c277df42 | 3399 | va_end(args); |
06bf62c7 | 3400 | message = SvPV(msv,l1); |
c277df42 IZ |
3401 | if (l1 > 512) |
3402 | l1 = 512; | |
3403 | Copy(message, buf, l1 , char); | |
3404 | buf[l1] = '\0'; /* Overwrite \n */ | |
cea2e8a9 | 3405 | Perl_croak(aTHX_ "%s", buf); |
c277df42 | 3406 | } |
a0ed51b3 LW |
3407 | |
3408 | /* XXX Here's a total kludge. But we need to re-enter for swash routines. */ | |
3409 | ||
3410 | void | |
864dbfa3 | 3411 | Perl_save_re_context(pTHX) |
c485e607 NIS |
3412 | { |
3413 | dTHR; | |
a0ed51b3 LW |
3414 | SAVEPPTR(PL_bostr); |
3415 | SAVEPPTR(PL_regprecomp); /* uncompiled string. */ | |
3416 | SAVEI32(PL_regnpar); /* () count. */ | |
3417 | SAVEI32(PL_regsize); /* Code size. */ | |
3418 | SAVEI16(PL_regflags); /* are we folding, multilining? */ | |
3419 | SAVEPPTR(PL_reginput); /* String-input pointer. */ | |
3420 | SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */ | |
3421 | SAVEPPTR(PL_regeol); /* End of input, for $ check. */ | |
3422 | SAVESPTR(PL_regstartp); /* Pointer to startp array. */ | |
3423 | SAVESPTR(PL_regendp); /* Ditto for endp. */ | |
3424 | SAVESPTR(PL_reglastparen); /* Similarly for lastparen. */ | |
3425 | SAVEPPTR(PL_regtill); /* How far we are required to go. */ | |
3426 | SAVEI32(PL_regprev); /* char before regbol, \n if none */ | |
3427 | SAVESPTR(PL_reg_start_tmp); /* from regexec.c */ | |
3428 | PL_reg_start_tmp = 0; | |
3429 | SAVEFREEPV(PL_reg_start_tmp); | |
3430 | SAVEI32(PL_reg_start_tmpl); /* from regexec.c */ | |
3431 | PL_reg_start_tmpl = 0; | |
3432 | SAVESPTR(PL_regdata); | |
3433 | SAVEI32(PL_reg_flags); /* from regexec.c */ | |
3434 | SAVEI32(PL_reg_eval_set); /* from regexec.c */ | |
3435 | SAVEI32(PL_regnarrate); /* from regexec.c */ | |
3436 | SAVESPTR(PL_regprogram); /* from regexec.c */ | |
3437 | SAVEINT(PL_regindent); /* from regexec.c */ | |
3438 | SAVESPTR(PL_regcc); /* from regexec.c */ | |
3439 | SAVESPTR(PL_curcop); | |
3440 | SAVESPTR(PL_regcomp_rx); /* from regcomp.c */ | |
3441 | SAVEI32(PL_regseen); /* from regcomp.c */ | |
3442 | SAVEI32(PL_regsawback); /* Did we see \1, ...? */ | |
3443 | SAVEI32(PL_regnaughty); /* How bad is this pattern? */ | |
3444 | SAVESPTR(PL_regcode); /* Code-emit pointer; ®dummy = don't */ | |
3445 | SAVEPPTR(PL_regxend); /* End of input for compile */ | |
3446 | SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */ | |
54b6e2fa IZ |
3447 | SAVESPTR(PL_reg_call_cc); /* from regexec.c */ |
3448 | SAVESPTR(PL_reg_re); /* from regexec.c */ | |
3449 | SAVEPPTR(PL_reg_ganch); /* from regexec.c */ | |
3450 | SAVESPTR(PL_reg_sv); /* from regexec.c */ | |
3451 | SAVESPTR(PL_reg_magic); /* from regexec.c */ | |
3452 | SAVEI32(PL_reg_oldpos); /* from regexec.c */ | |
3453 | SAVESPTR(PL_reg_oldcurpm); /* from regexec.c */ | |
3454 | SAVESPTR(PL_reg_curpm); /* from regexec.c */ | |
3455 | #ifdef DEBUGGING | |
3456 | SAVEPPTR(PL_reg_starttry); /* from regexec.c */ | |
3457 | #endif | |
a0ed51b3 | 3458 | } |
51371543 GS |
3459 | |
3460 | #ifdef PERL_OBJECT | |
3461 | #define NO_XSLOCKS | |
3462 | #include "XSUB.h" | |
3463 | #undef this | |
3464 | #define this pPerl | |
3465 | #endif | |
3466 | ||
3467 | static void | |
3468 | clear_re(pTHXo_ void *r) | |
3469 | { | |
3470 | ReREFCNT_dec((regexp *)r); | |
3471 | } | |
3472 |