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 | **** | |
3818b22b | 72 | **** Copyright (c) 1991-2000, 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; | |
653099ff | 154 | struct regnode_charclass_class *start_class; |
2c2d71f5 JH |
155 | } scan_data_t; |
156 | ||
a687059c | 157 | /* |
e50aee73 | 158 | * Forward declarations for pregcomp()'s friends. |
a687059c | 159 | */ |
a0d0e21e | 160 | |
cd488c12 | 161 | static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
653099ff | 162 | 0, 0, 0, 0, 0 }; |
c277df42 IZ |
163 | |
164 | #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL) | |
165 | #define SF_BEFORE_SEOL 0x1 | |
166 | #define SF_BEFORE_MEOL 0x2 | |
167 | #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL) | |
168 | #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL) | |
169 | ||
09b7f37c CB |
170 | #ifdef NO_UNARY_PLUS |
171 | # define SF_FIX_SHIFT_EOL (0+2) | |
172 | # define SF_FL_SHIFT_EOL (0+4) | |
173 | #else | |
174 | # define SF_FIX_SHIFT_EOL (+2) | |
175 | # define SF_FL_SHIFT_EOL (+4) | |
176 | #endif | |
c277df42 IZ |
177 | |
178 | #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL) | |
179 | #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL) | |
180 | ||
181 | #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL) | |
182 | #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */ | |
183 | #define SF_IS_INF 0x40 | |
184 | #define SF_HAS_PAR 0x80 | |
185 | #define SF_IN_PAR 0x100 | |
186 | #define SF_HAS_EVAL 0x200 | |
4bfe0158 | 187 | #define SCF_DO_SUBSTR 0x400 |
653099ff GS |
188 | #define SCF_DO_STCLASS_AND 0x0800 |
189 | #define SCF_DO_STCLASS_OR 0x1000 | |
190 | #define SCF_DO_STCLASS (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR) | |
c277df42 | 191 | |
a0ed51b3 LW |
192 | #define RF_utf8 8 |
193 | #define UTF (PL_reg_flags & RF_utf8) | |
194 | #define LOC (PL_regflags & PMf_LOCALE) | |
195 | #define FOLD (PL_regflags & PMf_FOLD) | |
196 | ||
b8c5462f JH |
197 | #define OOB_CHAR8 1234 |
198 | #define OOB_UTF8 123456 | |
93733859 | 199 | #define OOB_NAMEDCLASS -1 |
b8c5462f | 200 | |
a0ed51b3 LW |
201 | #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) |
202 | #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) | |
203 | ||
cd439c50 IZ |
204 | /* Allow for side effects in s */ |
205 | #define REGC(c,s) STMT_START { if (!SIZE_ONLY) *(s) = (c); else (s);} STMT_END | |
206 | ||
51371543 | 207 | static void clear_re(pTHXo_ void *r); |
4327152a | 208 | |
653099ff GS |
209 | /* Mark that we cannot extend a found fixed substring at this point. |
210 | Updata the longest found anchored substring and the longest found | |
211 | floating substrings if needed. */ | |
212 | ||
4327152a | 213 | STATIC void |
cea2e8a9 | 214 | S_scan_commit(pTHX_ scan_data_t *data) |
c277df42 | 215 | { |
c485e607 | 216 | dTHR; |
a0ed51b3 LW |
217 | STRLEN l = CHR_SVLEN(data->last_found); |
218 | STRLEN old_l = CHR_SVLEN(*data->longest); | |
c277df42 IZ |
219 | |
220 | if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) { | |
221 | sv_setsv(*data->longest, data->last_found); | |
222 | if (*data->longest == data->longest_fixed) { | |
223 | data->offset_fixed = l ? data->last_start_min : data->pos_min; | |
224 | if (data->flags & SF_BEFORE_EOL) | |
225 | data->flags | |
226 | |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL); | |
227 | else | |
228 | data->flags &= ~SF_FIX_BEFORE_EOL; | |
a0ed51b3 LW |
229 | } |
230 | else { | |
c277df42 IZ |
231 | data->offset_float_min = l ? data->last_start_min : data->pos_min; |
232 | data->offset_float_max = (l | |
233 | ? data->last_start_max | |
234 | : data->pos_min + data->pos_delta); | |
235 | if (data->flags & SF_BEFORE_EOL) | |
236 | data->flags | |
237 | |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL); | |
238 | else | |
239 | data->flags &= ~SF_FL_BEFORE_EOL; | |
240 | } | |
241 | } | |
242 | SvCUR_set(data->last_found, 0); | |
243 | data->last_end = -1; | |
244 | data->flags &= ~SF_BEFORE_EOL; | |
245 | } | |
246 | ||
653099ff GS |
247 | /* Can match anything (initialization) */ |
248 | STATIC void | |
249 | S_cl_anything(pTHX_ struct regnode_charclass_class *cl) | |
250 | { | |
251 | int value; | |
252 | ||
253 | ANYOF_CLASS_ZERO(cl); | |
254 | for (value = 0; value < 256; ++value) | |
255 | ANYOF_BITMAP_SET(cl, value); | |
256 | cl->flags = ANYOF_EOS; | |
257 | if (LOC) | |
258 | cl->flags |= ANYOF_LOCALE; | |
259 | } | |
260 | ||
261 | /* Can match anything (initialization) */ | |
262 | STATIC int | |
263 | S_cl_is_anything(pTHX_ struct regnode_charclass_class *cl) | |
264 | { | |
265 | int value; | |
266 | ||
267 | for (value = 0; value < ANYOF_MAX; value += 2) | |
268 | if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1)) | |
269 | return 1; | |
270 | for (value = 0; value < 256; ++value) | |
271 | if (!ANYOF_BITMAP_TEST(cl, value)) | |
272 | return 0; | |
273 | return 1; | |
274 | } | |
275 | ||
276 | /* Can match anything (initialization) */ | |
277 | STATIC void | |
278 | S_cl_init(pTHX_ struct regnode_charclass_class *cl) | |
279 | { | |
8ecf7187 | 280 | Zero(cl, 1, struct regnode_charclass_class); |
653099ff GS |
281 | cl->type = ANYOF; |
282 | cl_anything(cl); | |
283 | } | |
284 | ||
285 | STATIC void | |
286 | S_cl_init_zero(pTHX_ struct regnode_charclass_class *cl) | |
287 | { | |
8ecf7187 | 288 | Zero(cl, 1, struct regnode_charclass_class); |
653099ff GS |
289 | cl->type = ANYOF; |
290 | cl_anything(cl); | |
653099ff GS |
291 | if (LOC) |
292 | cl->flags |= ANYOF_LOCALE; | |
293 | } | |
294 | ||
295 | /* 'And' a given class with another one. Can create false positives */ | |
296 | /* We assume that cl is not inverted */ | |
297 | STATIC void | |
298 | S_cl_and(pTHX_ struct regnode_charclass_class *cl, | |
299 | struct regnode_charclass_class *and_with) | |
300 | { | |
301 | int value; | |
302 | ||
303 | if (!(and_with->flags & ANYOF_CLASS) | |
304 | && !(cl->flags & ANYOF_CLASS) | |
305 | && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE) | |
306 | && !(and_with->flags & ANYOF_FOLD) | |
307 | && !(cl->flags & ANYOF_FOLD)) { | |
308 | int i; | |
309 | ||
310 | if (and_with->flags & ANYOF_INVERT) | |
311 | for (i = 0; i < ANYOF_BITMAP_SIZE; i++) | |
312 | cl->bitmap[i] &= ~and_with->bitmap[i]; | |
313 | else | |
314 | for (i = 0; i < ANYOF_BITMAP_SIZE; i++) | |
315 | cl->bitmap[i] &= and_with->bitmap[i]; | |
316 | } /* XXXX: logic is complicated otherwise, leave it along for a moment. */ | |
317 | if (!(and_with->flags & ANYOF_EOS)) | |
318 | cl->flags &= ~ANYOF_EOS; | |
319 | } | |
320 | ||
321 | /* 'OR' a given class with another one. Can create false positives */ | |
322 | /* We assume that cl is not inverted */ | |
323 | STATIC void | |
324 | S_cl_or(pTHX_ struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with) | |
325 | { | |
326 | int value; | |
327 | ||
328 | if (or_with->flags & ANYOF_INVERT) { | |
329 | /* We do not use | |
330 | * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2)) | |
331 | * <= (B1 | !B2) | (CL1 | !CL2) | |
332 | * which is wasteful if CL2 is small, but we ignore CL2: | |
333 | * (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1 | |
334 | * XXXX Can we handle case-fold? Unclear: | |
335 | * (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) = | |
336 | * (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i')) | |
337 | */ | |
338 | if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE) | |
339 | && !(or_with->flags & ANYOF_FOLD) | |
340 | && !(cl->flags & ANYOF_FOLD) ) { | |
341 | int i; | |
342 | ||
343 | for (i = 0; i < ANYOF_BITMAP_SIZE; i++) | |
344 | cl->bitmap[i] |= ~or_with->bitmap[i]; | |
345 | } /* XXXX: logic is complicated otherwise */ | |
346 | else { | |
347 | cl_anything(cl); | |
348 | } | |
349 | } else { | |
350 | /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */ | |
351 | if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE) | |
352 | && (!(or_with->flags & ANYOF_FOLD) | |
353 | || (cl->flags & ANYOF_FOLD)) ) { | |
354 | int i; | |
355 | ||
356 | /* OR char bitmap and class bitmap separately */ | |
357 | for (i = 0; i < ANYOF_BITMAP_SIZE; i++) | |
358 | cl->bitmap[i] |= or_with->bitmap[i]; | |
359 | if (or_with->flags & ANYOF_CLASS) { | |
360 | for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++) | |
361 | cl->classflags[i] |= or_with->classflags[i]; | |
362 | cl->flags |= ANYOF_CLASS; | |
363 | } | |
364 | } | |
365 | else { /* XXXX: logic is complicated, leave it along for a moment. */ | |
366 | cl_anything(cl); | |
367 | } | |
368 | } | |
369 | if (or_with->flags & ANYOF_EOS) | |
370 | cl->flags |= ANYOF_EOS; | |
371 | } | |
372 | ||
373 | /* REx optimizer. Converts nodes into quickier variants "in place". | |
374 | Finds fixed substrings. */ | |
375 | ||
c277df42 IZ |
376 | /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set |
377 | to the position after last scanned or to NULL. */ | |
378 | ||
76e3520e | 379 | STATIC I32 |
cea2e8a9 | 380 | S_study_chunk(pTHX_ regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags) |
c277df42 IZ |
381 | /* scanp: Start here (read-write). */ |
382 | /* deltap: Write maxlen-minlen here. */ | |
383 | /* last: Stop before this one. */ | |
384 | { | |
5c0ca799 | 385 | dTHR; |
c277df42 IZ |
386 | I32 min = 0, pars = 0, code; |
387 | regnode *scan = *scanp, *next; | |
388 | I32 delta = 0; | |
389 | int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF); | |
aca2d497 | 390 | int is_inf_internal = 0; /* The studied chunk is infinite */ |
c277df42 IZ |
391 | I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0; |
392 | scan_data_t data_fake; | |
653099ff | 393 | struct regnode_charclass_class and_with; /* Valid if flags & SCF_DO_STCLASS_OR */ |
c277df42 IZ |
394 | |
395 | while (scan && OP(scan) != END && scan < last) { | |
396 | /* Peephole optimizer: */ | |
397 | ||
22c35a8c | 398 | if (PL_regkind[(U8)OP(scan)] == EXACT) { |
653099ff | 399 | /* Merge several consecutive EXACTish nodes into one. */ |
c277df42 IZ |
400 | regnode *n = regnext(scan); |
401 | U32 stringok = 1; | |
402 | #ifdef DEBUGGING | |
403 | regnode *stop = scan; | |
404 | #endif | |
405 | ||
cd439c50 | 406 | next = scan + NODE_SZ_STR(scan); |
c277df42 IZ |
407 | /* Skip NOTHING, merge EXACT*. */ |
408 | while (n && | |
22c35a8c | 409 | ( PL_regkind[(U8)OP(n)] == NOTHING || |
c277df42 IZ |
410 | (stringok && (OP(n) == OP(scan)))) |
411 | && NEXT_OFF(n) | |
412 | && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) { | |
413 | if (OP(n) == TAIL || n > next) | |
414 | stringok = 0; | |
22c35a8c | 415 | if (PL_regkind[(U8)OP(n)] == NOTHING) { |
c277df42 IZ |
416 | NEXT_OFF(scan) += NEXT_OFF(n); |
417 | next = n + NODE_STEP_REGNODE; | |
418 | #ifdef DEBUGGING | |
419 | if (stringok) | |
420 | stop = n; | |
421 | #endif | |
422 | n = regnext(n); | |
a0ed51b3 LW |
423 | } |
424 | else { | |
cd439c50 | 425 | int oldl = STR_LEN(scan); |
c277df42 IZ |
426 | regnode *nnext = regnext(n); |
427 | ||
cd439c50 | 428 | if (oldl + STR_LEN(n) > U8_MAX) |
c277df42 IZ |
429 | break; |
430 | NEXT_OFF(scan) += NEXT_OFF(n); | |
cd439c50 IZ |
431 | STR_LEN(scan) += STR_LEN(n); |
432 | next = n + NODE_SZ_STR(n); | |
c277df42 | 433 | /* Now we can overwrite *n : */ |
cd439c50 IZ |
434 | Move(STRING(n), STRING(scan) + oldl, |
435 | STR_LEN(n), char); | |
c277df42 IZ |
436 | #ifdef DEBUGGING |
437 | if (stringok) | |
438 | stop = next - 1; | |
439 | #endif | |
440 | n = nnext; | |
441 | } | |
442 | } | |
443 | #ifdef DEBUGGING | |
444 | /* Allow dumping */ | |
cd439c50 | 445 | n = scan + NODE_SZ_STR(scan); |
c277df42 | 446 | while (n <= stop) { |
22c35a8c | 447 | if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) { |
c277df42 IZ |
448 | OP(n) = OPTIMIZED; |
449 | NEXT_OFF(n) = 0; | |
450 | } | |
451 | n++; | |
452 | } | |
653099ff | 453 | #endif |
c277df42 | 454 | } |
653099ff GS |
455 | /* Follow the next-chain of the current node and optimize |
456 | away all the NOTHINGs from it. */ | |
c277df42 | 457 | if (OP(scan) != CURLYX) { |
048cfca1 GS |
458 | int max = (reg_off_by_arg[OP(scan)] |
459 | ? I32_MAX | |
460 | /* I32 may be smaller than U16 on CRAYs! */ | |
461 | : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX)); | |
c277df42 IZ |
462 | int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan)); |
463 | int noff; | |
464 | regnode *n = scan; | |
465 | ||
466 | /* Skip NOTHING and LONGJMP. */ | |
467 | while ((n = regnext(n)) | |
22c35a8c | 468 | && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n))) |
c277df42 IZ |
469 | || ((OP(n) == LONGJMP) && (noff = ARG(n)))) |
470 | && off + noff < max) | |
471 | off += noff; | |
472 | if (reg_off_by_arg[OP(scan)]) | |
473 | ARG(scan) = off; | |
474 | else | |
475 | NEXT_OFF(scan) = off; | |
476 | } | |
653099ff GS |
477 | /* The principal pseudo-switch. Cannot be a switch, since we |
478 | look into several different things. */ | |
c277df42 IZ |
479 | if (OP(scan) == BRANCH || OP(scan) == BRANCHJ |
480 | || OP(scan) == IFTHEN || OP(scan) == SUSPEND) { | |
481 | next = regnext(scan); | |
482 | code = OP(scan); | |
483 | ||
484 | if (OP(next) == code || code == IFTHEN || code == SUSPEND) { | |
485 | I32 max1 = 0, min1 = I32_MAX, num = 0; | |
653099ff | 486 | struct regnode_charclass_class accum; |
c277df42 | 487 | |
653099ff GS |
488 | if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */ |
489 | scan_commit(data); /* Cannot merge strings after this. */ | |
490 | if (flags & SCF_DO_STCLASS) | |
491 | cl_init_zero(&accum); | |
c277df42 | 492 | while (OP(scan) == code) { |
653099ff GS |
493 | I32 deltanext, minnext, f = 0; |
494 | struct regnode_charclass_class this_class; | |
c277df42 IZ |
495 | |
496 | num++; | |
497 | data_fake.flags = 0; | |
2c2d71f5 JH |
498 | if (data) |
499 | data_fake.whilem_c = data->whilem_c; | |
c277df42 IZ |
500 | next = regnext(scan); |
501 | scan = NEXTOPER(scan); | |
502 | if (code != BRANCH) | |
503 | scan = NEXTOPER(scan); | |
653099ff GS |
504 | if (flags & SCF_DO_STCLASS) { |
505 | cl_init(&this_class); | |
506 | data_fake.start_class = &this_class; | |
507 | f = SCF_DO_STCLASS_AND; | |
508 | } | |
509 | /* we suppose the run is continuous, last=next...*/ | |
c277df42 | 510 | minnext = study_chunk(&scan, &deltanext, next, |
653099ff | 511 | &data_fake, f); |
c277df42 IZ |
512 | if (min1 > minnext) |
513 | min1 = minnext; | |
514 | if (max1 < minnext + deltanext) | |
515 | max1 = minnext + deltanext; | |
516 | if (deltanext == I32_MAX) | |
aca2d497 | 517 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
518 | scan = next; |
519 | if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
520 | pars++; | |
405ff068 | 521 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 | 522 | data->flags |= SF_HAS_EVAL; |
2c2d71f5 JH |
523 | if (data) |
524 | data->whilem_c = data_fake.whilem_c; | |
653099ff GS |
525 | if (flags & SCF_DO_STCLASS) |
526 | cl_or(&accum, &this_class); | |
c277df42 IZ |
527 | if (code == SUSPEND) |
528 | break; | |
529 | } | |
530 | if (code == IFTHEN && num < 2) /* Empty ELSE branch */ | |
531 | min1 = 0; | |
532 | if (flags & SCF_DO_SUBSTR) { | |
533 | data->pos_min += min1; | |
534 | data->pos_delta += max1 - min1; | |
535 | if (max1 != min1 || is_inf) | |
536 | data->longest = &(data->longest_float); | |
537 | } | |
538 | min += min1; | |
539 | delta += max1 - min1; | |
653099ff GS |
540 | if (flags & SCF_DO_STCLASS_OR) { |
541 | cl_or(data->start_class, &accum); | |
542 | if (min1) { | |
543 | cl_and(data->start_class, &and_with); | |
544 | flags &= ~SCF_DO_STCLASS; | |
545 | } | |
546 | } | |
547 | else if (flags & SCF_DO_STCLASS_AND) { | |
de0c8cb8 GS |
548 | if (min1) { |
549 | cl_and(data->start_class, &accum); | |
653099ff | 550 | flags &= ~SCF_DO_STCLASS; |
de0c8cb8 GS |
551 | } |
552 | else { | |
553 | /* Switch to OR mode: cache the old value of | |
554 | * data->start_class */ | |
555 | StructCopy(data->start_class, &and_with, | |
556 | struct regnode_charclass_class); | |
557 | flags &= ~SCF_DO_STCLASS_AND; | |
558 | StructCopy(&accum, data->start_class, | |
559 | struct regnode_charclass_class); | |
560 | flags |= SCF_DO_STCLASS_OR; | |
561 | data->start_class->flags |= ANYOF_EOS; | |
562 | } | |
653099ff | 563 | } |
a0ed51b3 LW |
564 | } |
565 | else if (code == BRANCHJ) /* single branch is optimized. */ | |
c277df42 IZ |
566 | scan = NEXTOPER(NEXTOPER(scan)); |
567 | else /* single branch is optimized. */ | |
568 | scan = NEXTOPER(scan); | |
569 | continue; | |
a0ed51b3 LW |
570 | } |
571 | else if (OP(scan) == EXACT) { | |
cd439c50 | 572 | I32 l = STR_LEN(scan); |
a0ed51b3 | 573 | if (UTF) { |
cd439c50 | 574 | unsigned char *s = (unsigned char *)STRING(scan); |
a0ed51b3 LW |
575 | unsigned char *e = s + l; |
576 | I32 newl = 0; | |
577 | while (s < e) { | |
578 | newl++; | |
579 | s += UTF8SKIP(s); | |
580 | } | |
581 | l = newl; | |
582 | } | |
583 | min += l; | |
c277df42 | 584 | if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */ |
c277df42 IZ |
585 | /* The code below prefers earlier match for fixed |
586 | offset, later match for variable offset. */ | |
587 | if (data->last_end == -1) { /* Update the start info. */ | |
588 | data->last_start_min = data->pos_min; | |
589 | data->last_start_max = is_inf | |
590 | ? I32_MAX : data->pos_min + data->pos_delta; | |
591 | } | |
cd439c50 | 592 | sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan)); |
c277df42 IZ |
593 | data->last_end = data->pos_min + l; |
594 | data->pos_min += l; /* As in the first entry. */ | |
595 | data->flags &= ~SF_BEFORE_EOL; | |
596 | } | |
653099ff GS |
597 | if (flags & SCF_DO_STCLASS_AND) { |
598 | /* Check whether it is compatible with what we know already! */ | |
599 | int compat = 1; | |
600 | ||
601 | if (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE)) | |
602 | && !ANYOF_BITMAP_TEST(data->start_class, *STRING(scan)) | |
603 | && (!(data->start_class->flags & ANYOF_FOLD) | |
604 | || !ANYOF_BITMAP_TEST(data->start_class, | |
605 | PL_fold[*STRING(scan)]))) | |
606 | compat = 0; | |
607 | ANYOF_CLASS_ZERO(data->start_class); | |
608 | ANYOF_BITMAP_ZERO(data->start_class); | |
609 | if (compat) | |
610 | ANYOF_BITMAP_SET(data->start_class, *STRING(scan)); | |
611 | data->start_class->flags &= ~ANYOF_EOS; | |
612 | } | |
613 | else if (flags & SCF_DO_STCLASS_OR) { | |
614 | /* false positive possible if the class is case-folded */ | |
615 | ANYOF_BITMAP_SET(data->start_class, *STRING(scan)); | |
616 | data->start_class->flags &= ~ANYOF_EOS; | |
617 | cl_and(data->start_class, &and_with); | |
618 | } | |
619 | flags &= ~SCF_DO_STCLASS; | |
a0ed51b3 | 620 | } |
653099ff | 621 | else if (PL_regkind[(U8)OP(scan)] == EXACT) { /* But OP != EXACT! */ |
cd439c50 | 622 | I32 l = STR_LEN(scan); |
653099ff GS |
623 | |
624 | /* Search for fixed substrings supports EXACT only. */ | |
c277df42 IZ |
625 | if (flags & SCF_DO_SUBSTR) |
626 | scan_commit(data); | |
a0ed51b3 | 627 | if (UTF) { |
cd439c50 | 628 | unsigned char *s = (unsigned char *)STRING(scan); |
a0ed51b3 LW |
629 | unsigned char *e = s + l; |
630 | I32 newl = 0; | |
631 | while (s < e) { | |
632 | newl++; | |
633 | s += UTF8SKIP(s); | |
634 | } | |
635 | l = newl; | |
636 | } | |
637 | min += l; | |
c277df42 | 638 | if (data && (flags & SCF_DO_SUBSTR)) |
a0ed51b3 | 639 | data->pos_min += l; |
653099ff GS |
640 | if (flags & SCF_DO_STCLASS_AND) { |
641 | /* Check whether it is compatible with what we know already! */ | |
642 | int compat = 1; | |
643 | ||
644 | if (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE)) | |
645 | && !ANYOF_BITMAP_TEST(data->start_class, *STRING(scan)) | |
646 | && !ANYOF_BITMAP_TEST(data->start_class, | |
647 | PL_fold[*STRING(scan)])) | |
648 | compat = 0; | |
649 | ANYOF_CLASS_ZERO(data->start_class); | |
650 | ANYOF_BITMAP_ZERO(data->start_class); | |
651 | if (compat) { | |
652 | ANYOF_BITMAP_SET(data->start_class, *STRING(scan)); | |
653 | data->start_class->flags &= ~ANYOF_EOS; | |
654 | data->start_class->flags |= ANYOF_FOLD; | |
655 | if (OP(scan) == EXACTFL) | |
656 | data->start_class->flags |= ANYOF_LOCALE; | |
657 | } | |
658 | } | |
659 | else if (flags & SCF_DO_STCLASS_OR) { | |
660 | if (data->start_class->flags & ANYOF_FOLD) { | |
661 | /* false positive possible if the class is case-folded. | |
662 | Assume that the locale settings are the same... */ | |
663 | ANYOF_BITMAP_SET(data->start_class, *STRING(scan)); | |
664 | data->start_class->flags &= ~ANYOF_EOS; | |
665 | } | |
666 | cl_and(data->start_class, &and_with); | |
667 | } | |
668 | flags &= ~SCF_DO_STCLASS; | |
a0ed51b3 | 669 | } |
4d61ec05 | 670 | else if (strchr((char*)PL_varies,OP(scan))) { |
c277df42 | 671 | I32 mincount, maxcount, minnext, deltanext, pos_before, fl; |
653099ff | 672 | I32 f = flags; |
c277df42 | 673 | regnode *oscan = scan; |
653099ff GS |
674 | struct regnode_charclass_class this_class; |
675 | struct regnode_charclass_class *oclass = NULL; | |
676 | ||
22c35a8c | 677 | switch (PL_regkind[(U8)OP(scan)]) { |
653099ff | 678 | case WHILEM: /* End of (?:...)* . */ |
c277df42 IZ |
679 | scan = NEXTOPER(scan); |
680 | goto finish; | |
681 | case PLUS: | |
653099ff | 682 | if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) { |
c277df42 | 683 | next = NEXTOPER(scan); |
653099ff | 684 | if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) { |
c277df42 IZ |
685 | mincount = 1; |
686 | maxcount = REG_INFTY; | |
687 | next = regnext(scan); | |
688 | scan = NEXTOPER(scan); | |
689 | goto do_curly; | |
690 | } | |
691 | } | |
692 | if (flags & SCF_DO_SUBSTR) | |
693 | data->pos_min++; | |
694 | min++; | |
695 | /* Fall through. */ | |
696 | case STAR: | |
653099ff GS |
697 | if (flags & SCF_DO_STCLASS) { |
698 | mincount = 0; | |
699 | maxcount = REG_INFTY; | |
700 | next = regnext(scan); | |
701 | scan = NEXTOPER(scan); | |
702 | goto do_curly; | |
703 | } | |
aca2d497 | 704 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
705 | scan = regnext(scan); |
706 | if (flags & SCF_DO_SUBSTR) { | |
653099ff | 707 | scan_commit(data); /* Cannot extend fixed substrings */ |
c277df42 IZ |
708 | data->longest = &(data->longest_float); |
709 | } | |
710 | goto optimize_curly_tail; | |
711 | case CURLY: | |
712 | mincount = ARG1(scan); | |
713 | maxcount = ARG2(scan); | |
714 | next = regnext(scan); | |
715 | scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
716 | do_curly: | |
717 | if (flags & SCF_DO_SUBSTR) { | |
653099ff | 718 | if (mincount == 0) scan_commit(data); /* Cannot extend fixed substrings */ |
c277df42 IZ |
719 | pos_before = data->pos_min; |
720 | } | |
721 | if (data) { | |
722 | fl = data->flags; | |
723 | data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL); | |
724 | if (is_inf) | |
725 | data->flags |= SF_IS_INF; | |
726 | } | |
653099ff GS |
727 | if (flags & SCF_DO_STCLASS) { |
728 | cl_init(&this_class); | |
729 | oclass = data->start_class; | |
730 | data->start_class = &this_class; | |
731 | f |= SCF_DO_STCLASS_AND; | |
732 | f &= ~SCF_DO_STCLASS_OR; | |
733 | } | |
734 | ||
c277df42 IZ |
735 | /* This will finish on WHILEM, setting scan, or on NULL: */ |
736 | minnext = study_chunk(&scan, &deltanext, last, data, | |
737 | mincount == 0 | |
653099ff GS |
738 | ? (f & ~SCF_DO_SUBSTR) : f); |
739 | ||
740 | if (flags & SCF_DO_STCLASS) | |
741 | data->start_class = oclass; | |
742 | if (mincount == 0 || minnext == 0) { | |
743 | if (flags & SCF_DO_STCLASS_OR) { | |
744 | cl_or(data->start_class, &this_class); | |
745 | } | |
746 | else if (flags & SCF_DO_STCLASS_AND) { | |
747 | /* Switch to OR mode: cache the old value of | |
748 | * data->start_class */ | |
749 | StructCopy(data->start_class, &and_with, | |
750 | struct regnode_charclass_class); | |
751 | flags &= ~SCF_DO_STCLASS_AND; | |
752 | StructCopy(&this_class, data->start_class, | |
753 | struct regnode_charclass_class); | |
754 | flags |= SCF_DO_STCLASS_OR; | |
755 | data->start_class->flags |= ANYOF_EOS; | |
756 | } | |
757 | } else { /* Non-zero len */ | |
758 | if (flags & SCF_DO_STCLASS_OR) { | |
759 | cl_or(data->start_class, &this_class); | |
760 | cl_and(data->start_class, &and_with); | |
761 | } | |
762 | else if (flags & SCF_DO_STCLASS_AND) | |
763 | cl_and(data->start_class, &this_class); | |
764 | flags &= ~SCF_DO_STCLASS; | |
765 | } | |
c277df42 IZ |
766 | if (!scan) /* It was not CURLYX, but CURLY. */ |
767 | scan = next; | |
e476b1b5 | 768 | if (ckWARN(WARN_REGEXP) && (minnext + deltanext == 0) |
821b33a5 | 769 | && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) |
17feb5d5 | 770 | && maxcount <= REG_INFTY/3) /* Complement check for big count */ |
e476b1b5 | 771 | Perl_warner(aTHX_ WARN_REGEXP, |
c5be433b | 772 | "Strange *+?{} on zero-length expression"); |
c277df42 | 773 | min += minnext * mincount; |
aca2d497 IZ |
774 | is_inf_internal |= (maxcount == REG_INFTY |
775 | && (minnext + deltanext) > 0 | |
776 | || deltanext == I32_MAX); | |
777 | is_inf |= is_inf_internal; | |
c277df42 IZ |
778 | delta += (minnext + deltanext) * maxcount - minnext * mincount; |
779 | ||
780 | /* Try powerful optimization CURLYX => CURLYN. */ | |
c277df42 IZ |
781 | if ( OP(oscan) == CURLYX && data |
782 | && data->flags & SF_IN_PAR | |
783 | && !(data->flags & SF_HAS_EVAL) | |
784 | && !deltanext && minnext == 1 ) { | |
785 | /* Try to optimize to CURLYN. */ | |
786 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; | |
787 | regnode *nxt1 = nxt, *nxt2; | |
788 | ||
789 | /* Skip open. */ | |
790 | nxt = regnext(nxt); | |
4d61ec05 | 791 | if (!strchr((char*)PL_simple,OP(nxt)) |
22c35a8c | 792 | && !(PL_regkind[(U8)OP(nxt)] == EXACT |
cd439c50 | 793 | && STR_LEN(nxt) == 1)) |
c277df42 IZ |
794 | goto nogo; |
795 | nxt2 = nxt; | |
796 | nxt = regnext(nxt); | |
797 | if (OP(nxt) != CLOSE) | |
798 | goto nogo; | |
799 | /* Now we know that nxt2 is the only contents: */ | |
800 | oscan->flags = ARG(nxt); | |
801 | OP(oscan) = CURLYN; | |
802 | OP(nxt1) = NOTHING; /* was OPEN. */ | |
803 | #ifdef DEBUGGING | |
804 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ | |
805 | NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */ | |
806 | NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */ | |
807 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ | |
808 | OP(nxt + 1) = OPTIMIZED; /* was count. */ | |
809 | NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */ | |
810 | #endif | |
811 | } | |
c277df42 IZ |
812 | nogo: |
813 | ||
814 | /* Try optimization CURLYX => CURLYM. */ | |
815 | if ( OP(oscan) == CURLYX && data | |
c277df42 | 816 | && !(data->flags & SF_HAS_PAR) |
c277df42 IZ |
817 | && !(data->flags & SF_HAS_EVAL) |
818 | && !deltanext ) { | |
819 | /* XXXX How to optimize if data == 0? */ | |
820 | /* Optimize to a simpler form. */ | |
821 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */ | |
822 | regnode *nxt2; | |
823 | ||
824 | OP(oscan) = CURLYM; | |
825 | while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/ | |
826 | && (OP(nxt2) != WHILEM)) | |
827 | nxt = nxt2; | |
828 | OP(nxt2) = SUCCEED; /* Whas WHILEM */ | |
c277df42 IZ |
829 | /* Need to optimize away parenths. */ |
830 | if (data->flags & SF_IN_PAR) { | |
831 | /* Set the parenth number. */ | |
832 | regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/ | |
833 | ||
834 | if (OP(nxt) != CLOSE) | |
835 | FAIL("panic opt close"); | |
836 | oscan->flags = ARG(nxt); | |
837 | OP(nxt1) = OPTIMIZED; /* was OPEN. */ | |
838 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ | |
839 | #ifdef DEBUGGING | |
840 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ | |
841 | OP(nxt + 1) = OPTIMIZED; /* was count. */ | |
842 | NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */ | |
843 | NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */ | |
844 | #endif | |
845 | #if 0 | |
846 | while ( nxt1 && (OP(nxt1) != WHILEM)) { | |
847 | regnode *nnxt = regnext(nxt1); | |
848 | ||
849 | if (nnxt == nxt) { | |
850 | if (reg_off_by_arg[OP(nxt1)]) | |
851 | ARG_SET(nxt1, nxt2 - nxt1); | |
852 | else if (nxt2 - nxt1 < U16_MAX) | |
853 | NEXT_OFF(nxt1) = nxt2 - nxt1; | |
854 | else | |
855 | OP(nxt) = NOTHING; /* Cannot beautify */ | |
856 | } | |
857 | nxt1 = nnxt; | |
858 | } | |
859 | #endif | |
860 | /* Optimize again: */ | |
861 | study_chunk(&nxt1, &deltanext, nxt, NULL, 0); | |
a0ed51b3 LW |
862 | } |
863 | else | |
c277df42 | 864 | oscan->flags = 0; |
c277df42 | 865 | } |
2c2d71f5 JH |
866 | else if (OP(oscan) == CURLYX && data && ++data->whilem_c < 16) { |
867 | /* This stays as CURLYX, and can put the count/of pair. */ | |
868 | /* Find WHILEM (as in regexec.c) */ | |
869 | regnode *nxt = oscan + NEXT_OFF(oscan); | |
870 | ||
871 | if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */ | |
872 | nxt += ARG(nxt); | |
873 | PREVOPER(nxt)->flags = data->whilem_c | |
874 | | (PL_reg_whilem_seen << 4); /* On WHILEM */ | |
875 | } | |
c277df42 IZ |
876 | if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) |
877 | pars++; | |
878 | if (flags & SCF_DO_SUBSTR) { | |
879 | SV *last_str = Nullsv; | |
880 | int counted = mincount != 0; | |
881 | ||
882 | if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */ | |
883 | I32 b = pos_before >= data->last_start_min | |
884 | ? pos_before : data->last_start_min; | |
885 | STRLEN l; | |
886 | char *s = SvPV(data->last_found, l); | |
a0ed51b3 LW |
887 | I32 old = b - data->last_start_min; |
888 | ||
889 | if (UTF) | |
890 | old = utf8_hop((U8*)s, old) - (U8*)s; | |
c277df42 | 891 | |
a0ed51b3 | 892 | l -= old; |
c277df42 | 893 | /* Get the added string: */ |
79cb57f6 | 894 | last_str = newSVpvn(s + old, l); |
c277df42 IZ |
895 | if (deltanext == 0 && pos_before == b) { |
896 | /* What was added is a constant string */ | |
897 | if (mincount > 1) { | |
898 | SvGROW(last_str, (mincount * l) + 1); | |
899 | repeatcpy(SvPVX(last_str) + l, | |
900 | SvPVX(last_str), l, mincount - 1); | |
901 | SvCUR(last_str) *= mincount; | |
902 | /* Add additional parts. */ | |
903 | SvCUR_set(data->last_found, | |
904 | SvCUR(data->last_found) - l); | |
905 | sv_catsv(data->last_found, last_str); | |
906 | data->last_end += l * (mincount - 1); | |
907 | } | |
908 | } | |
909 | } | |
910 | /* It is counted once already... */ | |
911 | data->pos_min += minnext * (mincount - counted); | |
912 | data->pos_delta += - counted * deltanext + | |
913 | (minnext + deltanext) * maxcount - minnext * mincount; | |
914 | if (mincount != maxcount) { | |
653099ff GS |
915 | /* Cannot extend fixed substrings found inside |
916 | the group. */ | |
c277df42 IZ |
917 | scan_commit(data); |
918 | if (mincount && last_str) { | |
919 | sv_setsv(data->last_found, last_str); | |
920 | data->last_end = data->pos_min; | |
921 | data->last_start_min = | |
a0ed51b3 | 922 | data->pos_min - CHR_SVLEN(last_str); |
c277df42 IZ |
923 | data->last_start_max = is_inf |
924 | ? I32_MAX | |
925 | : data->pos_min + data->pos_delta | |
a0ed51b3 | 926 | - CHR_SVLEN(last_str); |
c277df42 IZ |
927 | } |
928 | data->longest = &(data->longest_float); | |
929 | } | |
aca2d497 | 930 | SvREFCNT_dec(last_str); |
c277df42 | 931 | } |
405ff068 | 932 | if (data && (fl & SF_HAS_EVAL)) |
c277df42 IZ |
933 | data->flags |= SF_HAS_EVAL; |
934 | optimize_curly_tail: | |
c277df42 | 935 | if (OP(oscan) != CURLYX) { |
22c35a8c | 936 | while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING |
c277df42 IZ |
937 | && NEXT_OFF(next)) |
938 | NEXT_OFF(oscan) += NEXT_OFF(next); | |
939 | } | |
c277df42 | 940 | continue; |
653099ff | 941 | default: /* REF and CLUMP only? */ |
c277df42 | 942 | if (flags & SCF_DO_SUBSTR) { |
653099ff | 943 | scan_commit(data); /* Cannot expect anything... */ |
c277df42 IZ |
944 | data->longest = &(data->longest_float); |
945 | } | |
aca2d497 | 946 | is_inf = is_inf_internal = 1; |
653099ff GS |
947 | if (flags & SCF_DO_STCLASS_OR) |
948 | cl_anything(data->start_class); | |
949 | flags &= ~SCF_DO_STCLASS; | |
c277df42 IZ |
950 | break; |
951 | } | |
a0ed51b3 | 952 | } |
4d61ec05 | 953 | else if (strchr((char*)PL_simple,OP(scan)) || PL_regkind[(U8)OP(scan)] == ANYUTF8) { |
653099ff GS |
954 | int value; |
955 | ||
c277df42 IZ |
956 | if (flags & SCF_DO_SUBSTR) { |
957 | scan_commit(data); | |
958 | data->pos_min++; | |
959 | } | |
960 | min++; | |
653099ff GS |
961 | if (flags & SCF_DO_STCLASS) { |
962 | data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */ | |
963 | ||
964 | /* Some of the logic below assumes that switching | |
965 | locale on will only add false positives. */ | |
966 | switch (PL_regkind[(U8)OP(scan)]) { | |
967 | case ANYUTF8: | |
968 | case SANY: | |
969 | case SANYUTF8: | |
970 | case ALNUMUTF8: | |
971 | case ANYOFUTF8: | |
972 | case ALNUMLUTF8: | |
973 | case NALNUMUTF8: | |
974 | case NALNUMLUTF8: | |
975 | case SPACEUTF8: | |
976 | case NSPACEUTF8: | |
977 | case SPACELUTF8: | |
978 | case NSPACELUTF8: | |
979 | case DIGITUTF8: | |
980 | case NDIGITUTF8: | |
981 | default: | |
982 | do_default: | |
983 | /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */ | |
984 | if (flags & SCF_DO_STCLASS_OR) /* Allow everything */ | |
985 | cl_anything(data->start_class); | |
986 | break; | |
987 | case REG_ANY: | |
988 | if (OP(scan) == SANY) | |
989 | goto do_default; | |
990 | if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */ | |
991 | value = (ANYOF_BITMAP_TEST(data->start_class,'\n') | |
992 | || (data->start_class->flags & ANYOF_CLASS)); | |
993 | cl_anything(data->start_class); | |
994 | } | |
995 | if (flags & SCF_DO_STCLASS_AND || !value) | |
996 | ANYOF_BITMAP_CLEAR(data->start_class,'\n'); | |
997 | break; | |
998 | case ANYOF: | |
999 | if (flags & SCF_DO_STCLASS_AND) | |
1000 | cl_and(data->start_class, | |
1001 | (struct regnode_charclass_class*)scan); | |
1002 | else | |
1003 | cl_or(data->start_class, | |
1004 | (struct regnode_charclass_class*)scan); | |
1005 | break; | |
1006 | case ALNUM: | |
1007 | if (flags & SCF_DO_STCLASS_AND) { | |
1008 | if (!(data->start_class->flags & ANYOF_LOCALE)) { | |
1009 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM); | |
1010 | for (value = 0; value < 256; value++) | |
1011 | if (!isALNUM(value)) | |
1012 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1013 | } | |
1014 | } | |
1015 | else { | |
1016 | if (data->start_class->flags & ANYOF_LOCALE) | |
1017 | ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM); | |
1018 | else { | |
1019 | for (value = 0; value < 256; value++) | |
1020 | if (isALNUM(value)) | |
1021 | ANYOF_BITMAP_SET(data->start_class, value); | |
1022 | } | |
1023 | } | |
1024 | break; | |
1025 | case ALNUML: | |
1026 | if (flags & SCF_DO_STCLASS_AND) { | |
1027 | if (data->start_class->flags & ANYOF_LOCALE) | |
1028 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM); | |
1029 | } | |
1030 | else { | |
1031 | ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM); | |
1032 | data->start_class->flags |= ANYOF_LOCALE; | |
1033 | } | |
1034 | break; | |
1035 | case NALNUM: | |
1036 | if (flags & SCF_DO_STCLASS_AND) { | |
1037 | if (!(data->start_class->flags & ANYOF_LOCALE)) { | |
1038 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM); | |
1039 | for (value = 0; value < 256; value++) | |
1040 | if (isALNUM(value)) | |
1041 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1042 | } | |
1043 | } | |
1044 | else { | |
1045 | if (data->start_class->flags & ANYOF_LOCALE) | |
1046 | ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM); | |
1047 | else { | |
1048 | for (value = 0; value < 256; value++) | |
1049 | if (!isALNUM(value)) | |
1050 | ANYOF_BITMAP_SET(data->start_class, value); | |
1051 | } | |
1052 | } | |
1053 | break; | |
1054 | case NALNUML: | |
1055 | if (flags & SCF_DO_STCLASS_AND) { | |
1056 | if (data->start_class->flags & ANYOF_LOCALE) | |
1057 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM); | |
1058 | } | |
1059 | else { | |
1060 | data->start_class->flags |= ANYOF_LOCALE; | |
1061 | ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM); | |
1062 | } | |
1063 | break; | |
1064 | case SPACE: | |
1065 | if (flags & SCF_DO_STCLASS_AND) { | |
1066 | if (!(data->start_class->flags & ANYOF_LOCALE)) { | |
1067 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE); | |
1068 | for (value = 0; value < 256; value++) | |
1069 | if (!isSPACE(value)) | |
1070 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1071 | } | |
1072 | } | |
1073 | else { | |
1074 | if (data->start_class->flags & ANYOF_LOCALE) | |
1075 | ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE); | |
1076 | else { | |
1077 | for (value = 0; value < 256; value++) | |
1078 | if (isSPACE(value)) | |
1079 | ANYOF_BITMAP_SET(data->start_class, value); | |
1080 | } | |
1081 | } | |
1082 | break; | |
1083 | case SPACEL: | |
1084 | if (flags & SCF_DO_STCLASS_AND) { | |
1085 | if (data->start_class->flags & ANYOF_LOCALE) | |
1086 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE); | |
1087 | } | |
1088 | else { | |
1089 | data->start_class->flags |= ANYOF_LOCALE; | |
1090 | ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE); | |
1091 | } | |
1092 | break; | |
1093 | case NSPACE: | |
1094 | if (flags & SCF_DO_STCLASS_AND) { | |
1095 | if (!(data->start_class->flags & ANYOF_LOCALE)) { | |
1096 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE); | |
1097 | for (value = 0; value < 256; value++) | |
1098 | if (isSPACE(value)) | |
1099 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1100 | } | |
1101 | } | |
1102 | else { | |
1103 | if (data->start_class->flags & ANYOF_LOCALE) | |
1104 | ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE); | |
1105 | else { | |
1106 | for (value = 0; value < 256; value++) | |
1107 | if (!isSPACE(value)) | |
1108 | ANYOF_BITMAP_SET(data->start_class, value); | |
1109 | } | |
1110 | } | |
1111 | break; | |
1112 | case NSPACEL: | |
1113 | if (flags & SCF_DO_STCLASS_AND) { | |
1114 | if (data->start_class->flags & ANYOF_LOCALE) { | |
1115 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE); | |
1116 | for (value = 0; value < 256; value++) | |
1117 | if (!isSPACE(value)) | |
1118 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1119 | } | |
1120 | } | |
1121 | else { | |
1122 | data->start_class->flags |= ANYOF_LOCALE; | |
1123 | ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE); | |
1124 | } | |
1125 | break; | |
1126 | case DIGIT: | |
1127 | if (flags & SCF_DO_STCLASS_AND) { | |
1128 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT); | |
1129 | for (value = 0; value < 256; value++) | |
1130 | if (!isDIGIT(value)) | |
1131 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1132 | } | |
1133 | else { | |
1134 | if (data->start_class->flags & ANYOF_LOCALE) | |
1135 | ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT); | |
1136 | else { | |
1137 | for (value = 0; value < 256; value++) | |
1138 | if (isDIGIT(value)) | |
1139 | ANYOF_BITMAP_SET(data->start_class, value); | |
1140 | } | |
1141 | } | |
1142 | break; | |
1143 | case NDIGIT: | |
1144 | if (flags & SCF_DO_STCLASS_AND) { | |
1145 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT); | |
1146 | for (value = 0; value < 256; value++) | |
1147 | if (isDIGIT(value)) | |
1148 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1149 | } | |
1150 | else { | |
1151 | if (data->start_class->flags & ANYOF_LOCALE) | |
1152 | ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT); | |
1153 | else { | |
1154 | for (value = 0; value < 256; value++) | |
1155 | if (!isDIGIT(value)) | |
1156 | ANYOF_BITMAP_SET(data->start_class, value); | |
1157 | } | |
1158 | } | |
1159 | break; | |
1160 | } | |
1161 | if (flags & SCF_DO_STCLASS_OR) | |
1162 | cl_and(data->start_class, &and_with); | |
1163 | flags &= ~SCF_DO_STCLASS; | |
1164 | } | |
a0ed51b3 | 1165 | } |
22c35a8c | 1166 | else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) { |
c277df42 IZ |
1167 | data->flags |= (OP(scan) == MEOL |
1168 | ? SF_BEFORE_MEOL | |
1169 | : SF_BEFORE_SEOL); | |
a0ed51b3 | 1170 | } |
653099ff GS |
1171 | else if ( PL_regkind[(U8)OP(scan)] == BRANCHJ |
1172 | /* Lookbehind, or need to calculate parens/evals/stclass: */ | |
1173 | && (scan->flags || data || (flags & SCF_DO_STCLASS)) | |
c277df42 | 1174 | && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) { |
653099ff | 1175 | /* Lookahead/lookbehind */ |
c277df42 IZ |
1176 | I32 deltanext, minnext; |
1177 | regnode *nscan; | |
653099ff GS |
1178 | struct regnode_charclass_class intrnl; |
1179 | int f = 0; | |
c277df42 IZ |
1180 | |
1181 | data_fake.flags = 0; | |
2c2d71f5 JH |
1182 | if (data) |
1183 | data_fake.whilem_c = data->whilem_c; | |
653099ff GS |
1184 | if ( flags & SCF_DO_STCLASS && !scan->flags |
1185 | && OP(scan) == IFMATCH ) { /* Lookahead */ | |
1186 | cl_init(&intrnl); | |
1187 | data_fake.start_class = &intrnl; | |
1188 | f = SCF_DO_STCLASS_AND; | |
1189 | } | |
c277df42 IZ |
1190 | next = regnext(scan); |
1191 | nscan = NEXTOPER(NEXTOPER(scan)); | |
653099ff | 1192 | minnext = study_chunk(&nscan, &deltanext, last, &data_fake, f); |
c277df42 IZ |
1193 | if (scan->flags) { |
1194 | if (deltanext) { | |
1195 | FAIL("variable length lookbehind not implemented"); | |
a0ed51b3 LW |
1196 | } |
1197 | else if (minnext > U8_MAX) { | |
57def98f | 1198 | FAIL2("lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX); |
c277df42 IZ |
1199 | } |
1200 | scan->flags = minnext; | |
1201 | } | |
1202 | if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
1203 | pars++; | |
405ff068 | 1204 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 | 1205 | data->flags |= SF_HAS_EVAL; |
2c2d71f5 JH |
1206 | if (data) |
1207 | data->whilem_c = data_fake.whilem_c; | |
653099ff GS |
1208 | if (f) { |
1209 | int was = (data->start_class->flags & ANYOF_EOS); | |
1210 | ||
1211 | cl_and(data->start_class, &intrnl); | |
1212 | if (was) | |
1213 | data->start_class->flags |= ANYOF_EOS; | |
1214 | } | |
a0ed51b3 LW |
1215 | } |
1216 | else if (OP(scan) == OPEN) { | |
c277df42 | 1217 | pars++; |
a0ed51b3 LW |
1218 | } |
1219 | else if (OP(scan) == CLOSE && ARG(scan) == is_par) { | |
c277df42 IZ |
1220 | next = regnext(scan); |
1221 | ||
1222 | if ( next && (OP(next) != WHILEM) && next < last) | |
c277df42 | 1223 | is_par = 0; /* Disable optimization */ |
a0ed51b3 LW |
1224 | } |
1225 | else if (OP(scan) == EVAL) { | |
c277df42 IZ |
1226 | if (data) |
1227 | data->flags |= SF_HAS_EVAL; | |
1228 | } | |
0f5d15d6 IZ |
1229 | else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */ |
1230 | if (flags & SCF_DO_SUBSTR) { | |
1231 | scan_commit(data); | |
1232 | data->longest = &(data->longest_float); | |
1233 | } | |
1234 | is_inf = is_inf_internal = 1; | |
653099ff GS |
1235 | if (flags & SCF_DO_STCLASS_OR) /* Allow everything */ |
1236 | cl_anything(data->start_class); | |
0f5d15d6 | 1237 | } |
c277df42 IZ |
1238 | /* Else: zero-length, ignore. */ |
1239 | scan = regnext(scan); | |
1240 | } | |
1241 | ||
1242 | finish: | |
1243 | *scanp = scan; | |
aca2d497 | 1244 | *deltap = is_inf_internal ? I32_MAX : delta; |
c277df42 IZ |
1245 | if (flags & SCF_DO_SUBSTR && is_inf) |
1246 | data->pos_delta = I32_MAX - data->pos_min; | |
1247 | if (is_par > U8_MAX) | |
1248 | is_par = 0; | |
1249 | if (is_par && pars==1 && data) { | |
1250 | data->flags |= SF_IN_PAR; | |
1251 | data->flags &= ~SF_HAS_PAR; | |
a0ed51b3 LW |
1252 | } |
1253 | else if (pars && data) { | |
c277df42 IZ |
1254 | data->flags |= SF_HAS_PAR; |
1255 | data->flags &= ~SF_IN_PAR; | |
1256 | } | |
653099ff GS |
1257 | if (flags & SCF_DO_STCLASS_OR) |
1258 | cl_and(data->start_class, &and_with); | |
c277df42 IZ |
1259 | return min; |
1260 | } | |
1261 | ||
76e3520e | 1262 | STATIC I32 |
cea2e8a9 | 1263 | S_add_data(pTHX_ I32 n, char *s) |
c277df42 | 1264 | { |
5c0ca799 | 1265 | dTHR; |
3280af22 NIS |
1266 | if (PL_regcomp_rx->data) { |
1267 | Renewc(PL_regcomp_rx->data, | |
1268 | sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1), | |
c277df42 | 1269 | char, struct reg_data); |
3280af22 NIS |
1270 | Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8); |
1271 | PL_regcomp_rx->data->count += n; | |
a0ed51b3 LW |
1272 | } |
1273 | else { | |
3280af22 | 1274 | Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1), |
c277df42 | 1275 | char, struct reg_data); |
3280af22 NIS |
1276 | New(1208, PL_regcomp_rx->data->what, n, U8); |
1277 | PL_regcomp_rx->data->count = n; | |
c277df42 | 1278 | } |
3280af22 NIS |
1279 | Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8); |
1280 | return PL_regcomp_rx->data->count - n; | |
c277df42 IZ |
1281 | } |
1282 | ||
d88dccdf | 1283 | void |
864dbfa3 | 1284 | Perl_reginitcolors(pTHX) |
d88dccdf IZ |
1285 | { |
1286 | dTHR; | |
1287 | int i = 0; | |
1288 | char *s = PerlEnv_getenv("PERL_RE_COLORS"); | |
1289 | ||
1290 | if (s) { | |
1291 | PL_colors[0] = s = savepv(s); | |
1292 | while (++i < 6) { | |
1293 | s = strchr(s, '\t'); | |
1294 | if (s) { | |
1295 | *s = '\0'; | |
1296 | PL_colors[i] = ++s; | |
1297 | } | |
1298 | else | |
c712d376 | 1299 | PL_colors[i] = s = ""; |
d88dccdf IZ |
1300 | } |
1301 | } else { | |
1302 | while (i < 6) | |
1303 | PL_colors[i++] = ""; | |
1304 | } | |
1305 | PL_colorset = 1; | |
1306 | } | |
1307 | ||
a687059c | 1308 | /* |
e50aee73 | 1309 | - pregcomp - compile a regular expression into internal code |
a687059c LW |
1310 | * |
1311 | * We can't allocate space until we know how big the compiled form will be, | |
1312 | * but we can't compile it (and thus know how big it is) until we've got a | |
1313 | * place to put the code. So we cheat: we compile it twice, once with code | |
1314 | * generation turned off and size counting turned on, and once "for real". | |
1315 | * This also means that we don't allocate space until we are sure that the | |
1316 | * thing really will compile successfully, and we never have to move the | |
1317 | * code and thus invalidate pointers into it. (Note that it has to be in | |
1318 | * one piece because free() must be able to free it all.) [NB: not true in perl] | |
1319 | * | |
1320 | * Beware that the optimization-preparation code in here knows about some | |
1321 | * of the structure of the compiled regexp. [I'll say.] | |
1322 | */ | |
1323 | regexp * | |
864dbfa3 | 1324 | Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm) |
a687059c | 1325 | { |
5c0ca799 | 1326 | dTHR; |
a0d0e21e | 1327 | register regexp *r; |
c277df42 IZ |
1328 | regnode *scan; |
1329 | SV **longest; | |
1330 | SV *longest_fixed; | |
1331 | SV *longest_float; | |
1332 | regnode *first; | |
a0d0e21e | 1333 | I32 flags; |
a0d0e21e LW |
1334 | I32 minlen = 0; |
1335 | I32 sawplus = 0; | |
1336 | I32 sawopen = 0; | |
2c2d71f5 | 1337 | scan_data_t data; |
a0d0e21e LW |
1338 | |
1339 | if (exp == NULL) | |
c277df42 | 1340 | FAIL("NULL regexp argument"); |
a0d0e21e | 1341 | |
393fec97 | 1342 | if (pm->op_pmdynflags & PMdf_UTF8) { |
a0ed51b3 | 1343 | PL_reg_flags |= RF_utf8; |
7e2040f0 | 1344 | } |
a0ed51b3 LW |
1345 | else |
1346 | PL_reg_flags = 0; | |
1347 | ||
3280af22 | 1348 | PL_regprecomp = savepvn(exp, xend - exp); |
35ef4773 | 1349 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
2c2d71f5 | 1350 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n", |
d88dccdf | 1351 | PL_colors[4],PL_colors[5],PL_colors[0], |
b900a521 | 1352 | (int)(xend - exp), PL_regprecomp, PL_colors[1])); |
3280af22 NIS |
1353 | PL_regflags = pm->op_pmflags; |
1354 | PL_regsawback = 0; | |
bbce6d69 | 1355 | |
3280af22 NIS |
1356 | PL_regseen = 0; |
1357 | PL_seen_zerolen = *exp == '^' ? -1 : 0; | |
1358 | PL_seen_evals = 0; | |
1359 | PL_extralen = 0; | |
c277df42 | 1360 | |
bbce6d69 | 1361 | /* First pass: determine size, legality. */ |
3280af22 NIS |
1362 | PL_regcomp_parse = exp; |
1363 | PL_regxend = xend; | |
1364 | PL_regnaughty = 0; | |
1365 | PL_regnpar = 1; | |
1366 | PL_regsize = 0L; | |
1367 | PL_regcode = &PL_regdummy; | |
2c2d71f5 | 1368 | PL_reg_whilem_seen = 0; |
cd439c50 | 1369 | REGC((U8)REG_MAGIC, (char*)PL_regcode); |
a0d0e21e | 1370 | if (reg(0, &flags) == NULL) { |
3280af22 NIS |
1371 | Safefree(PL_regprecomp); |
1372 | PL_regprecomp = Nullch; | |
a0d0e21e LW |
1373 | return(NULL); |
1374 | } | |
faccc32b | 1375 | DEBUG_r(PerlIO_printf(Perl_debug_log, "size %"IVdf" ", (IV)PL_regsize)); |
c277df42 | 1376 | |
c277df42 IZ |
1377 | /* Small enough for pointer-storage convention? |
1378 | If extralen==0, this means that we will not need long jumps. */ | |
3280af22 NIS |
1379 | if (PL_regsize >= 0x10000L && PL_extralen) |
1380 | PL_regsize += PL_extralen; | |
c277df42 | 1381 | else |
3280af22 | 1382 | PL_extralen = 0; |
2c2d71f5 JH |
1383 | if (PL_reg_whilem_seen > 15) |
1384 | PL_reg_whilem_seen = 15; | |
a0d0e21e | 1385 | |
bbce6d69 | 1386 | /* Allocate space and initialize. */ |
3280af22 | 1387 | Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), |
c277df42 | 1388 | char, regexp); |
a0d0e21e LW |
1389 | if (r == NULL) |
1390 | FAIL("regexp out of space"); | |
0f79a09d GS |
1391 | #ifdef DEBUGGING |
1392 | /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */ | |
1393 | Zero(r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), char); | |
1394 | #endif | |
c277df42 | 1395 | r->refcnt = 1; |
bbce6d69 | 1396 | r->prelen = xend - exp; |
3280af22 | 1397 | r->precomp = PL_regprecomp; |
cf93c79d IZ |
1398 | r->subbeg = NULL; |
1399 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; | |
4327152a IZ |
1400 | r->nparens = PL_regnpar - 1; /* set early to validate backrefs */ |
1401 | ||
1402 | r->substrs = 0; /* Useful during FAIL. */ | |
1403 | r->startp = 0; /* Useful during FAIL. */ | |
1404 | r->endp = 0; /* Useful during FAIL. */ | |
1405 | ||
3280af22 | 1406 | PL_regcomp_rx = r; |
bbce6d69 | 1407 | |
1408 | /* Second pass: emit code. */ | |
3280af22 NIS |
1409 | PL_regcomp_parse = exp; |
1410 | PL_regxend = xend; | |
1411 | PL_regnaughty = 0; | |
1412 | PL_regnpar = 1; | |
1413 | PL_regcode = r->program; | |
2cd61cdb | 1414 | /* Store the count of eval-groups for security checks: */ |
3280af22 | 1415 | PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals); |
cd439c50 | 1416 | REGC((U8)REG_MAGIC, (char*) PL_regcode++); |
c277df42 | 1417 | r->data = 0; |
a0d0e21e LW |
1418 | if (reg(0, &flags) == NULL) |
1419 | return(NULL); | |
1420 | ||
1421 | /* Dig out information for optimizations. */ | |
cf93c79d | 1422 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */ |
3280af22 | 1423 | pm->op_pmflags = PL_regflags; |
a0ed51b3 LW |
1424 | if (UTF) |
1425 | r->reganch |= ROPT_UTF8; | |
c277df42 | 1426 | r->regstclass = NULL; |
a0ed51b3 LW |
1427 | if (PL_regnaughty >= 10) /* Probably an expensive pattern. */ |
1428 | r->reganch |= ROPT_NAUGHTY; | |
c277df42 | 1429 | scan = r->program + 1; /* First BRANCH. */ |
2779dcf1 IZ |
1430 | |
1431 | /* XXXX To minimize changes to RE engine we always allocate | |
1432 | 3-units-long substrs field. */ | |
1433 | Newz(1004, r->substrs, 1, struct reg_substr_data); | |
1434 | ||
2c2d71f5 | 1435 | StructCopy(&zero_scan_data, &data, scan_data_t); |
653099ff | 1436 | /* XXXX Should not we check for something else? Usually it is OPEN1... */ |
c277df42 | 1437 | if (OP(scan) != BRANCH) { /* Only one top-level choice. */ |
c277df42 | 1438 | I32 fake; |
c5254dd6 | 1439 | STRLEN longest_float_length, longest_fixed_length; |
653099ff GS |
1440 | struct regnode_charclass_class ch_class; |
1441 | int stclass_flag; | |
a0d0e21e LW |
1442 | |
1443 | first = scan; | |
c277df42 | 1444 | /* Skip introductions and multiplicators >= 1. */ |
a0d0e21e | 1445 | while ((OP(first) == OPEN && (sawopen = 1)) || |
653099ff | 1446 | /* An OR of *one* alternative - should not happen now. */ |
a0d0e21e LW |
1447 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || |
1448 | (OP(first) == PLUS) || | |
1449 | (OP(first) == MINMOD) || | |
653099ff | 1450 | /* An {n,m} with n>0 */ |
22c35a8c | 1451 | (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
a0d0e21e LW |
1452 | if (OP(first) == PLUS) |
1453 | sawplus = 1; | |
1454 | else | |
1455 | first += regarglen[(U8)OP(first)]; | |
1456 | first = NEXTOPER(first); | |
a687059c LW |
1457 | } |
1458 | ||
a0d0e21e LW |
1459 | /* Starting-point info. */ |
1460 | again: | |
653099ff | 1461 | if (PL_regkind[(U8)OP(first)] == EXACT) { |
b3c9acc1 IZ |
1462 | if (OP(first) == EXACT); /* Empty, get anchored substr later. */ |
1463 | else if ((OP(first) == EXACTF || OP(first) == EXACTFL) | |
1464 | && !UTF) | |
1465 | r->regstclass = first; | |
1466 | } | |
653099ff | 1467 | else if (strchr((char*)PL_simple,OP(first))) |
a0d0e21e | 1468 | r->regstclass = first; |
22c35a8c GS |
1469 | else if (PL_regkind[(U8)OP(first)] == BOUND || |
1470 | PL_regkind[(U8)OP(first)] == NBOUND) | |
a0d0e21e | 1471 | r->regstclass = first; |
22c35a8c | 1472 | else if (PL_regkind[(U8)OP(first)] == BOL) { |
cad2e5aa JH |
1473 | r->reganch |= (OP(first) == MBOL |
1474 | ? ROPT_ANCH_MBOL | |
1475 | : (OP(first) == SBOL | |
1476 | ? ROPT_ANCH_SBOL | |
1477 | : ROPT_ANCH_BOL)); | |
a0d0e21e | 1478 | first = NEXTOPER(first); |
774d564b | 1479 | goto again; |
1480 | } | |
1481 | else if (OP(first) == GPOS) { | |
1482 | r->reganch |= ROPT_ANCH_GPOS; | |
1483 | first = NEXTOPER(first); | |
1484 | goto again; | |
a0d0e21e LW |
1485 | } |
1486 | else if ((OP(first) == STAR && | |
22c35a8c | 1487 | PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) && |
a0d0e21e LW |
1488 | !(r->reganch & ROPT_ANCH) ) |
1489 | { | |
1490 | /* turn .* into ^.* with an implied $*=1 */ | |
cad2e5aa JH |
1491 | int type = OP(NEXTOPER(first)); |
1492 | ||
1493 | if (type == REG_ANY || type == ANYUTF8) | |
1494 | type = ROPT_ANCH_MBOL; | |
1495 | else | |
1496 | type = ROPT_ANCH_SBOL; | |
1497 | ||
1498 | r->reganch |= type | ROPT_IMPLICIT; | |
a0d0e21e | 1499 | first = NEXTOPER(first); |
774d564b | 1500 | goto again; |
a0d0e21e | 1501 | } |
cad2e5aa JH |
1502 | if (sawplus && (!sawopen || !PL_regsawback) |
1503 | && !(PL_regseen & REG_SEEN_EVAL)) /* May examine pos and $& */ | |
1504 | /* x+ must match at the 1st pos of run of x's */ | |
1505 | r->reganch |= ROPT_SKIP; | |
a0d0e21e | 1506 | |
c277df42 | 1507 | /* Scan is after the zeroth branch, first is atomic matcher. */ |
b900a521 JH |
1508 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n", |
1509 | (IV)(first - scan + 1))); | |
a0d0e21e LW |
1510 | /* |
1511 | * If there's something expensive in the r.e., find the | |
1512 | * longest literal string that must appear and make it the | |
1513 | * regmust. Resolve ties in favor of later strings, since | |
1514 | * the regstart check works with the beginning of the r.e. | |
1515 | * and avoiding duplication strengthens checking. Not a | |
1516 | * strong reason, but sufficient in the absence of others. | |
1517 | * [Now we resolve ties in favor of the earlier string if | |
c277df42 | 1518 | * it happens that c_offset_min has been invalidated, since the |
a0d0e21e LW |
1519 | * earlier string may buy us something the later one won't.] |
1520 | */ | |
a0d0e21e | 1521 | minlen = 0; |
a687059c | 1522 | |
79cb57f6 GS |
1523 | data.longest_fixed = newSVpvn("",0); |
1524 | data.longest_float = newSVpvn("",0); | |
1525 | data.last_found = newSVpvn("",0); | |
c277df42 IZ |
1526 | data.longest = &(data.longest_fixed); |
1527 | first = scan; | |
653099ff GS |
1528 | if (!r->regstclass) { |
1529 | cl_init(&ch_class); | |
1530 | data.start_class = &ch_class; | |
1531 | stclass_flag = SCF_DO_STCLASS_AND; | |
1532 | } else /* XXXX Check for BOUND? */ | |
1533 | stclass_flag = 0; | |
1534 | ||
3280af22 | 1535 | minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */ |
653099ff | 1536 | &data, SCF_DO_SUBSTR | stclass_flag); |
3280af22 | 1537 | if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed) |
c277df42 | 1538 | && data.last_start_min == 0 && data.last_end > 0 |
3280af22 NIS |
1539 | && !PL_seen_zerolen |
1540 | && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS))) | |
c277df42 IZ |
1541 | r->reganch |= ROPT_CHECK_ALL; |
1542 | scan_commit(&data); | |
1543 | SvREFCNT_dec(data.last_found); | |
1544 | ||
a0ed51b3 | 1545 | longest_float_length = CHR_SVLEN(data.longest_float); |
c5254dd6 | 1546 | if (longest_float_length |
c277df42 IZ |
1547 | || (data.flags & SF_FL_BEFORE_EOL |
1548 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
3280af22 | 1549 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1550 | int t; |
1551 | ||
a0ed51b3 | 1552 | if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */ |
aca2d497 IZ |
1553 | && data.offset_fixed == data.offset_float_min |
1554 | && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)) | |
1555 | goto remove_float; /* As in (a)+. */ | |
1556 | ||
c277df42 IZ |
1557 | r->float_substr = data.longest_float; |
1558 | r->float_min_offset = data.offset_float_min; | |
1559 | r->float_max_offset = data.offset_float_max; | |
cf93c79d IZ |
1560 | t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1561 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
1562 | || (PL_regflags & PMf_MULTILINE))); | |
1563 | fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1564 | } |
1565 | else { | |
aca2d497 | 1566 | remove_float: |
c277df42 IZ |
1567 | r->float_substr = Nullsv; |
1568 | SvREFCNT_dec(data.longest_float); | |
c5254dd6 | 1569 | longest_float_length = 0; |
a0d0e21e | 1570 | } |
c277df42 | 1571 | |
a0ed51b3 | 1572 | longest_fixed_length = CHR_SVLEN(data.longest_fixed); |
c5254dd6 | 1573 | if (longest_fixed_length |
c277df42 IZ |
1574 | || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
1575 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
3280af22 | 1576 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1577 | int t; |
1578 | ||
c277df42 IZ |
1579 | r->anchored_substr = data.longest_fixed; |
1580 | r->anchored_offset = data.offset_fixed; | |
cf93c79d IZ |
1581 | t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1582 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
1583 | || (PL_regflags & PMf_MULTILINE))); | |
1584 | fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1585 | } |
1586 | else { | |
c277df42 IZ |
1587 | r->anchored_substr = Nullsv; |
1588 | SvREFCNT_dec(data.longest_fixed); | |
c5254dd6 | 1589 | longest_fixed_length = 0; |
a0d0e21e | 1590 | } |
653099ff GS |
1591 | if (r->regstclass |
1592 | && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == ANYUTF8 | |
1593 | || OP(r->regstclass) == SANYUTF8 || OP(r->regstclass) == SANY)) | |
1594 | r->regstclass = NULL; | |
1595 | if ((!r->anchored_substr || r->anchored_offset) && stclass_flag | |
1596 | && !(data.start_class->flags & ANYOF_EOS) | |
1597 | && !cl_is_anything(data.start_class)) { | |
1598 | SV *sv; | |
1599 | I32 n = add_data(1, "f"); | |
1600 | ||
1601 | New(1006, PL_regcomp_rx->data->data[n], 1, | |
1602 | struct regnode_charclass_class); | |
1603 | StructCopy(data.start_class, | |
1604 | (struct regnode_charclass_class*)PL_regcomp_rx->data->data[n], | |
1605 | struct regnode_charclass_class); | |
1606 | r->regstclass = (regnode*)PL_regcomp_rx->data->data[n]; | |
1607 | r->reganch &= ~ROPT_SKIP; /* Used in find_byclass(). */ | |
1608 | DEBUG_r((sv = sv_newmortal(), | |
1609 | regprop(sv, (regnode*)data.start_class), | |
894356b3 | 1610 | PerlIO_printf(Perl_debug_log, "synthetic stclass `%s'.\n", |
653099ff GS |
1611 | SvPVX(sv)))); |
1612 | } | |
c277df42 IZ |
1613 | |
1614 | /* A temporary algorithm prefers floated substr to fixed one to dig more info. */ | |
c5254dd6 | 1615 | if (longest_fixed_length > longest_float_length) { |
c277df42 IZ |
1616 | r->check_substr = r->anchored_substr; |
1617 | r->check_offset_min = r->check_offset_max = r->anchored_offset; | |
1618 | if (r->reganch & ROPT_ANCH_SINGLE) | |
1619 | r->reganch |= ROPT_NOSCAN; | |
a0ed51b3 LW |
1620 | } |
1621 | else { | |
c277df42 IZ |
1622 | r->check_substr = r->float_substr; |
1623 | r->check_offset_min = data.offset_float_min; | |
1624 | r->check_offset_max = data.offset_float_max; | |
a0d0e21e | 1625 | } |
30382c73 IZ |
1626 | /* XXXX Currently intuiting is not compatible with ANCH_GPOS. |
1627 | This should be changed ASAP! */ | |
1628 | if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) { | |
cad2e5aa JH |
1629 | r->reganch |= RE_USE_INTUIT; |
1630 | if (SvTAIL(r->check_substr)) | |
1631 | r->reganch |= RE_INTUIT_TAIL; | |
1632 | } | |
a0ed51b3 LW |
1633 | } |
1634 | else { | |
c277df42 IZ |
1635 | /* Several toplevels. Best we can is to set minlen. */ |
1636 | I32 fake; | |
653099ff | 1637 | struct regnode_charclass_class ch_class; |
c277df42 IZ |
1638 | |
1639 | DEBUG_r(PerlIO_printf(Perl_debug_log, "\n")); | |
1640 | scan = r->program + 1; | |
653099ff GS |
1641 | cl_init(&ch_class); |
1642 | data.start_class = &ch_class; | |
1643 | minlen = study_chunk(&scan, &fake, scan + PL_regsize, &data, SCF_DO_STCLASS_AND); | |
c277df42 | 1644 | r->check_substr = r->anchored_substr = r->float_substr = Nullsv; |
653099ff GS |
1645 | if (!(data.start_class->flags & ANYOF_EOS) |
1646 | && !cl_is_anything(data.start_class)) { | |
1647 | SV *sv; | |
1648 | I32 n = add_data(1, "f"); | |
1649 | ||
1650 | New(1006, PL_regcomp_rx->data->data[n], 1, | |
1651 | struct regnode_charclass_class); | |
1652 | StructCopy(data.start_class, | |
1653 | (struct regnode_charclass_class*)PL_regcomp_rx->data->data[n], | |
1654 | struct regnode_charclass_class); | |
1655 | r->regstclass = (regnode*)PL_regcomp_rx->data->data[n]; | |
1656 | r->reganch &= ~ROPT_SKIP; /* Used in find_byclass(). */ | |
1657 | DEBUG_r((sv = sv_newmortal(), | |
1658 | regprop(sv, (regnode*)data.start_class), | |
894356b3 | 1659 | PerlIO_printf(Perl_debug_log, "synthetic stclass `%s'.\n", |
653099ff GS |
1660 | SvPVX(sv)))); |
1661 | } | |
a0d0e21e LW |
1662 | } |
1663 | ||
a0d0e21e | 1664 | r->minlen = minlen; |
3280af22 | 1665 | if (PL_regseen & REG_SEEN_GPOS) |
c277df42 | 1666 | r->reganch |= ROPT_GPOS_SEEN; |
3280af22 | 1667 | if (PL_regseen & REG_SEEN_LOOKBEHIND) |
c277df42 | 1668 | r->reganch |= ROPT_LOOKBEHIND_SEEN; |
3280af22 | 1669 | if (PL_regseen & REG_SEEN_EVAL) |
ce862d02 | 1670 | r->reganch |= ROPT_EVAL_SEEN; |
cf93c79d IZ |
1671 | Newz(1002, r->startp, PL_regnpar, I32); |
1672 | Newz(1002, r->endp, PL_regnpar, I32); | |
a0d0e21e LW |
1673 | DEBUG_r(regdump(r)); |
1674 | return(r); | |
a687059c LW |
1675 | } |
1676 | ||
1677 | /* | |
1678 | - reg - regular expression, i.e. main body or parenthesized thing | |
1679 | * | |
1680 | * Caller must absorb opening parenthesis. | |
1681 | * | |
1682 | * Combining parenthesis handling with the base level of regular expression | |
1683 | * is a trifle forced, but the need to tie the tails of the branches to what | |
1684 | * follows makes it hard to avoid. | |
1685 | */ | |
76e3520e | 1686 | STATIC regnode * |
cea2e8a9 | 1687 | S_reg(pTHX_ I32 paren, I32 *flagp) |
c277df42 | 1688 | /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */ |
a687059c | 1689 | { |
5c0ca799 | 1690 | dTHR; |
c277df42 IZ |
1691 | register regnode *ret; /* Will be the head of the group. */ |
1692 | register regnode *br; | |
1693 | register regnode *lastbr; | |
1694 | register regnode *ender = 0; | |
a0d0e21e | 1695 | register I32 parno = 0; |
3280af22 | 1696 | I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0; |
c277df42 | 1697 | char c; |
a0d0e21e | 1698 | |
821b33a5 | 1699 | *flagp = 0; /* Tentatively. */ |
a0d0e21e LW |
1700 | |
1701 | /* Make an OPEN node, if parenthesized. */ | |
1702 | if (paren) { | |
3280af22 | 1703 | if (*PL_regcomp_parse == '?') { |
ca9dfc88 IZ |
1704 | U16 posflags = 0, negflags = 0; |
1705 | U16 *flagsp = &posflags; | |
0f5d15d6 | 1706 | int logical = 0; |
ca9dfc88 | 1707 | |
3280af22 NIS |
1708 | PL_regcomp_parse++; |
1709 | paren = *PL_regcomp_parse++; | |
c277df42 | 1710 | ret = NULL; /* For look-ahead/behind. */ |
a0d0e21e | 1711 | switch (paren) { |
c277df42 | 1712 | case '<': |
3280af22 NIS |
1713 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
1714 | if (*PL_regcomp_parse == '!') | |
c277df42 | 1715 | paren = ','; |
3280af22 | 1716 | if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!') |
c277df42 | 1717 | goto unknown; |
3280af22 | 1718 | PL_regcomp_parse++; |
a0d0e21e LW |
1719 | case '=': |
1720 | case '!': | |
3280af22 | 1721 | PL_seen_zerolen++; |
c277df42 IZ |
1722 | case ':': |
1723 | case '>': | |
a0d0e21e LW |
1724 | break; |
1725 | case '$': | |
1726 | case '@': | |
c277df42 | 1727 | FAIL2("Sequence (?%c...) not implemented", (int)paren); |
a0d0e21e LW |
1728 | break; |
1729 | case '#': | |
3280af22 NIS |
1730 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
1731 | PL_regcomp_parse++; | |
1732 | if (*PL_regcomp_parse != ')') | |
c277df42 | 1733 | FAIL("Sequence (?#... not terminated"); |
a0d0e21e LW |
1734 | nextchar(); |
1735 | *flagp = TRYAGAIN; | |
1736 | return NULL; | |
0f5d15d6 IZ |
1737 | case 'p': |
1738 | logical = 1; | |
1739 | paren = *PL_regcomp_parse++; | |
1740 | /* FALL THROUGH */ | |
c277df42 IZ |
1741 | case '{': |
1742 | { | |
1743 | dTHR; | |
1744 | I32 count = 1, n = 0; | |
1745 | char c; | |
3280af22 | 1746 | char *s = PL_regcomp_parse; |
c277df42 IZ |
1747 | SV *sv; |
1748 | OP_4tree *sop, *rop; | |
1749 | ||
3280af22 NIS |
1750 | PL_seen_zerolen++; |
1751 | PL_regseen |= REG_SEEN_EVAL; | |
1752 | while (count && (c = *PL_regcomp_parse)) { | |
1753 | if (c == '\\' && PL_regcomp_parse[1]) | |
1754 | PL_regcomp_parse++; | |
c277df42 IZ |
1755 | else if (c == '{') |
1756 | count++; | |
1757 | else if (c == '}') | |
1758 | count--; | |
3280af22 | 1759 | PL_regcomp_parse++; |
c277df42 | 1760 | } |
3280af22 | 1761 | if (*PL_regcomp_parse != ')') |
c277df42 IZ |
1762 | FAIL("Sequence (?{...}) not terminated or not {}-balanced"); |
1763 | if (!SIZE_ONLY) { | |
1764 | AV *av; | |
1765 | ||
3280af22 | 1766 | if (PL_regcomp_parse - 1 - s) |
79cb57f6 | 1767 | sv = newSVpvn(s, PL_regcomp_parse - 1 - s); |
c277df42 | 1768 | else |
79cb57f6 | 1769 | sv = newSVpvn("", 0); |
c277df42 IZ |
1770 | |
1771 | rop = sv_compile_2op(sv, &sop, "re", &av); | |
1772 | ||
dfad63ad | 1773 | n = add_data(3, "nop"); |
3280af22 | 1774 | PL_regcomp_rx->data->data[n] = (void*)rop; |
dfad63ad HS |
1775 | PL_regcomp_rx->data->data[n+1] = (void*)sop; |
1776 | PL_regcomp_rx->data->data[n+2] = (void*)av; | |
c277df42 | 1777 | SvREFCNT_dec(sv); |
a0ed51b3 | 1778 | } |
e24b16f9 GS |
1779 | else { /* First pass */ |
1780 | if (PL_reginterp_cnt < ++PL_seen_evals | |
1781 | && PL_curcop != &PL_compiling) | |
2cd61cdb IZ |
1782 | /* No compiled RE interpolated, has runtime |
1783 | components ===> unsafe. */ | |
1784 | FAIL("Eval-group not allowed at runtime, use re 'eval'"); | |
3280af22 | 1785 | if (PL_tainted) |
cc6b7395 | 1786 | FAIL("Eval-group in insecure regular expression"); |
c277df42 IZ |
1787 | } |
1788 | ||
1789 | nextchar(); | |
0f5d15d6 IZ |
1790 | if (logical) { |
1791 | ret = reg_node(LOGICAL); | |
1792 | if (!SIZE_ONLY) | |
1793 | ret->flags = 2; | |
1794 | regtail(ret, reganode(EVAL, n)); | |
1795 | return ret; | |
1796 | } | |
c277df42 IZ |
1797 | return reganode(EVAL, n); |
1798 | } | |
1799 | case '(': | |
1800 | { | |
3280af22 NIS |
1801 | if (PL_regcomp_parse[0] == '?') { |
1802 | if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!' | |
1803 | || PL_regcomp_parse[1] == '<' | |
1804 | || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */ | |
c277df42 IZ |
1805 | I32 flag; |
1806 | ||
1807 | ret = reg_node(LOGICAL); | |
0f5d15d6 IZ |
1808 | if (!SIZE_ONLY) |
1809 | ret->flags = 1; | |
c277df42 IZ |
1810 | regtail(ret, reg(1, &flag)); |
1811 | goto insert_if; | |
1812 | } | |
a0ed51b3 LW |
1813 | } |
1814 | else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) { | |
3280af22 | 1815 | parno = atoi(PL_regcomp_parse++); |
c277df42 | 1816 | |
3280af22 NIS |
1817 | while (isDIGIT(*PL_regcomp_parse)) |
1818 | PL_regcomp_parse++; | |
c277df42 IZ |
1819 | ret = reganode(GROUPP, parno); |
1820 | if ((c = *nextchar()) != ')') | |
1821 | FAIL2("Switch (?(number%c not recognized", c); | |
1822 | insert_if: | |
1823 | regtail(ret, reganode(IFTHEN, 0)); | |
1824 | br = regbranch(&flags, 1); | |
1825 | if (br == NULL) | |
1826 | br = reganode(LONGJMP, 0); | |
1827 | else | |
1828 | regtail(br, reganode(LONGJMP, 0)); | |
1829 | c = *nextchar(); | |
d1b80229 IZ |
1830 | if (flags&HASWIDTH) |
1831 | *flagp |= HASWIDTH; | |
c277df42 IZ |
1832 | if (c == '|') { |
1833 | lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */ | |
1834 | regbranch(&flags, 1); | |
1835 | regtail(ret, lastbr); | |
d1b80229 IZ |
1836 | if (flags&HASWIDTH) |
1837 | *flagp |= HASWIDTH; | |
c277df42 | 1838 | c = *nextchar(); |
a0ed51b3 LW |
1839 | } |
1840 | else | |
c277df42 IZ |
1841 | lastbr = NULL; |
1842 | if (c != ')') | |
1843 | FAIL("Switch (?(condition)... contains too many branches"); | |
1844 | ender = reg_node(TAIL); | |
1845 | regtail(br, ender); | |
1846 | if (lastbr) { | |
1847 | regtail(lastbr, ender); | |
1848 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); | |
a0ed51b3 LW |
1849 | } |
1850 | else | |
c277df42 IZ |
1851 | regtail(ret, ender); |
1852 | return ret; | |
a0ed51b3 LW |
1853 | } |
1854 | else { | |
3280af22 | 1855 | FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse); |
c277df42 IZ |
1856 | } |
1857 | } | |
1b1626e4 | 1858 | case 0: |
c277df42 | 1859 | FAIL("Sequence (? incomplete"); |
1b1626e4 | 1860 | break; |
a0d0e21e | 1861 | default: |
3280af22 | 1862 | --PL_regcomp_parse; |
ca9dfc88 | 1863 | parse_flags: |
3280af22 NIS |
1864 | while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) { |
1865 | if (*PL_regcomp_parse != 'o') | |
1866 | pmflag(flagsp, *PL_regcomp_parse); | |
1867 | ++PL_regcomp_parse; | |
ca9dfc88 | 1868 | } |
3280af22 | 1869 | if (*PL_regcomp_parse == '-') { |
ca9dfc88 | 1870 | flagsp = &negflags; |
3280af22 | 1871 | ++PL_regcomp_parse; |
ca9dfc88 | 1872 | goto parse_flags; |
48c036b1 | 1873 | } |
3280af22 NIS |
1874 | PL_regflags |= posflags; |
1875 | PL_regflags &= ~negflags; | |
1876 | if (*PL_regcomp_parse == ':') { | |
1877 | PL_regcomp_parse++; | |
ca9dfc88 IZ |
1878 | paren = ':'; |
1879 | break; | |
1880 | } | |
c277df42 | 1881 | unknown: |
3280af22 NIS |
1882 | if (*PL_regcomp_parse != ')') |
1883 | FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse); | |
a0d0e21e LW |
1884 | nextchar(); |
1885 | *flagp = TRYAGAIN; | |
1886 | return NULL; | |
1887 | } | |
1888 | } | |
1889 | else { | |
3280af22 NIS |
1890 | parno = PL_regnpar; |
1891 | PL_regnpar++; | |
a0d0e21e | 1892 | ret = reganode(OPEN, parno); |
c277df42 | 1893 | open = 1; |
a0d0e21e | 1894 | } |
a0ed51b3 LW |
1895 | } |
1896 | else | |
a0d0e21e LW |
1897 | ret = NULL; |
1898 | ||
1899 | /* Pick up the branches, linking them together. */ | |
c277df42 | 1900 | br = regbranch(&flags, 1); |
a0d0e21e LW |
1901 | if (br == NULL) |
1902 | return(NULL); | |
3280af22 NIS |
1903 | if (*PL_regcomp_parse == '|') { |
1904 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 | 1905 | reginsert(BRANCHJ, br); |
a0ed51b3 LW |
1906 | } |
1907 | else | |
c277df42 IZ |
1908 | reginsert(BRANCH, br); |
1909 | have_branch = 1; | |
1910 | if (SIZE_ONLY) | |
3280af22 | 1911 | PL_extralen += 1; /* For BRANCHJ-BRANCH. */ |
a0ed51b3 LW |
1912 | } |
1913 | else if (paren == ':') { | |
c277df42 IZ |
1914 | *flagp |= flags&SIMPLE; |
1915 | } | |
1916 | if (open) { /* Starts with OPEN. */ | |
1917 | regtail(ret, br); /* OPEN -> first. */ | |
a0ed51b3 LW |
1918 | } |
1919 | else if (paren != '?') /* Not Conditional */ | |
a0d0e21e | 1920 | ret = br; |
821b33a5 IZ |
1921 | if (flags&HASWIDTH) |
1922 | *flagp |= HASWIDTH; | |
a0d0e21e | 1923 | *flagp |= flags&SPSTART; |
c277df42 | 1924 | lastbr = br; |
3280af22 NIS |
1925 | while (*PL_regcomp_parse == '|') { |
1926 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 IZ |
1927 | ender = reganode(LONGJMP,0); |
1928 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ | |
1929 | } | |
1930 | if (SIZE_ONLY) | |
3280af22 | 1931 | PL_extralen += 2; /* Account for LONGJMP. */ |
a0d0e21e | 1932 | nextchar(); |
c277df42 | 1933 | br = regbranch(&flags, 0); |
a687059c | 1934 | if (br == NULL) |
a0d0e21e | 1935 | return(NULL); |
c277df42 IZ |
1936 | regtail(lastbr, br); /* BRANCH -> BRANCH. */ |
1937 | lastbr = br; | |
821b33a5 IZ |
1938 | if (flags&HASWIDTH) |
1939 | *flagp |= HASWIDTH; | |
a687059c | 1940 | *flagp |= flags&SPSTART; |
a0d0e21e LW |
1941 | } |
1942 | ||
c277df42 IZ |
1943 | if (have_branch || paren != ':') { |
1944 | /* Make a closing node, and hook it on the end. */ | |
1945 | switch (paren) { | |
1946 | case ':': | |
1947 | ender = reg_node(TAIL); | |
1948 | break; | |
1949 | case 1: | |
1950 | ender = reganode(CLOSE, parno); | |
1951 | break; | |
1952 | case '<': | |
c277df42 IZ |
1953 | case ',': |
1954 | case '=': | |
1955 | case '!': | |
c277df42 | 1956 | *flagp &= ~HASWIDTH; |
821b33a5 IZ |
1957 | /* FALL THROUGH */ |
1958 | case '>': | |
1959 | ender = reg_node(SUCCEED); | |
c277df42 IZ |
1960 | break; |
1961 | case 0: | |
1962 | ender = reg_node(END); | |
1963 | break; | |
1964 | } | |
1965 | regtail(lastbr, ender); | |
a0d0e21e | 1966 | |
c277df42 IZ |
1967 | if (have_branch) { |
1968 | /* Hook the tails of the branches to the closing node. */ | |
1969 | for (br = ret; br != NULL; br = regnext(br)) { | |
1970 | regoptail(br, ender); | |
1971 | } | |
1972 | } | |
a0d0e21e | 1973 | } |
c277df42 IZ |
1974 | |
1975 | { | |
1976 | char *p; | |
1977 | static char parens[] = "=!<,>"; | |
1978 | ||
1979 | if (paren && (p = strchr(parens, paren))) { | |
1980 | int node = ((p - parens) % 2) ? UNLESSM : IFMATCH; | |
1981 | int flag = (p - parens) > 1; | |
1982 | ||
1983 | if (paren == '>') | |
1984 | node = SUSPEND, flag = 0; | |
1985 | reginsert(node,ret); | |
c277df42 | 1986 | ret->flags = flag; |
c277df42 IZ |
1987 | regtail(ret, reg_node(TAIL)); |
1988 | } | |
a0d0e21e LW |
1989 | } |
1990 | ||
1991 | /* Check for proper termination. */ | |
ce3e6498 HS |
1992 | if (paren) { |
1993 | PL_regflags = oregflags; | |
1994 | if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') { | |
1995 | FAIL("unmatched () in regexp"); | |
1996 | } | |
a0ed51b3 LW |
1997 | } |
1998 | else if (!paren && PL_regcomp_parse < PL_regxend) { | |
3280af22 | 1999 | if (*PL_regcomp_parse == ')') { |
a0d0e21e | 2000 | FAIL("unmatched () in regexp"); |
a0ed51b3 LW |
2001 | } |
2002 | else | |
a0d0e21e LW |
2003 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
2004 | /* NOTREACHED */ | |
2005 | } | |
a687059c | 2006 | |
a0d0e21e | 2007 | return(ret); |
a687059c LW |
2008 | } |
2009 | ||
2010 | /* | |
2011 | - regbranch - one alternative of an | operator | |
2012 | * | |
2013 | * Implements the concatenation operator. | |
2014 | */ | |
76e3520e | 2015 | STATIC regnode * |
cea2e8a9 | 2016 | S_regbranch(pTHX_ I32 *flagp, I32 first) |
a687059c | 2017 | { |
5c0ca799 | 2018 | dTHR; |
c277df42 IZ |
2019 | register regnode *ret; |
2020 | register regnode *chain = NULL; | |
2021 | register regnode *latest; | |
2022 | I32 flags = 0, c = 0; | |
a0d0e21e | 2023 | |
c277df42 IZ |
2024 | if (first) |
2025 | ret = NULL; | |
2026 | else { | |
3280af22 | 2027 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
2028 | ret = reganode(BRANCHJ,0); |
2029 | else | |
2030 | ret = reg_node(BRANCH); | |
2031 | } | |
2032 | ||
2033 | if (!first && SIZE_ONLY) | |
3280af22 | 2034 | PL_extralen += 1; /* BRANCHJ */ |
c277df42 IZ |
2035 | |
2036 | *flagp = WORST; /* Tentatively. */ | |
a0d0e21e | 2037 | |
3280af22 | 2038 | PL_regcomp_parse--; |
a0d0e21e | 2039 | nextchar(); |
3280af22 | 2040 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') { |
a0d0e21e LW |
2041 | flags &= ~TRYAGAIN; |
2042 | latest = regpiece(&flags); | |
2043 | if (latest == NULL) { | |
2044 | if (flags & TRYAGAIN) | |
2045 | continue; | |
2046 | return(NULL); | |
a0ed51b3 LW |
2047 | } |
2048 | else if (ret == NULL) | |
c277df42 | 2049 | ret = latest; |
a0d0e21e | 2050 | *flagp |= flags&HASWIDTH; |
c277df42 | 2051 | if (chain == NULL) /* First piece. */ |
a0d0e21e LW |
2052 | *flagp |= flags&SPSTART; |
2053 | else { | |
3280af22 | 2054 | PL_regnaughty++; |
a0d0e21e | 2055 | regtail(chain, latest); |
a687059c | 2056 | } |
a0d0e21e | 2057 | chain = latest; |
c277df42 IZ |
2058 | c++; |
2059 | } | |
2060 | if (chain == NULL) { /* Loop ran zero times. */ | |
2061 | chain = reg_node(NOTHING); | |
2062 | if (ret == NULL) | |
2063 | ret = chain; | |
2064 | } | |
2065 | if (c == 1) { | |
2066 | *flagp |= flags&SIMPLE; | |
a0d0e21e | 2067 | } |
a687059c | 2068 | |
a0d0e21e | 2069 | return(ret); |
a687059c LW |
2070 | } |
2071 | ||
2072 | /* | |
2073 | - regpiece - something followed by possible [*+?] | |
2074 | * | |
2075 | * Note that the branching code sequences used for ? and the general cases | |
2076 | * of * and + are somewhat optimized: they use the same NOTHING node as | |
2077 | * both the endmarker for their branch list and the body of the last branch. | |
2078 | * It might seem that this node could be dispensed with entirely, but the | |
2079 | * endmarker role is not redundant. | |
2080 | */ | |
76e3520e | 2081 | STATIC regnode * |
cea2e8a9 | 2082 | S_regpiece(pTHX_ I32 *flagp) |
a687059c | 2083 | { |
5c0ca799 | 2084 | dTHR; |
c277df42 | 2085 | register regnode *ret; |
a0d0e21e LW |
2086 | register char op; |
2087 | register char *next; | |
2088 | I32 flags; | |
3280af22 | 2089 | char *origparse = PL_regcomp_parse; |
a0d0e21e LW |
2090 | char *maxpos; |
2091 | I32 min; | |
c277df42 | 2092 | I32 max = REG_INFTY; |
a0d0e21e LW |
2093 | |
2094 | ret = regatom(&flags); | |
2095 | if (ret == NULL) { | |
2096 | if (flags & TRYAGAIN) | |
2097 | *flagp |= TRYAGAIN; | |
2098 | return(NULL); | |
2099 | } | |
2100 | ||
3280af22 | 2101 | op = *PL_regcomp_parse; |
a0d0e21e | 2102 | |
3280af22 NIS |
2103 | if (op == '{' && regcurly(PL_regcomp_parse)) { |
2104 | next = PL_regcomp_parse + 1; | |
a0d0e21e LW |
2105 | maxpos = Nullch; |
2106 | while (isDIGIT(*next) || *next == ',') { | |
2107 | if (*next == ',') { | |
2108 | if (maxpos) | |
2109 | break; | |
2110 | else | |
2111 | maxpos = next; | |
a687059c | 2112 | } |
a0d0e21e LW |
2113 | next++; |
2114 | } | |
2115 | if (*next == '}') { /* got one */ | |
2116 | if (!maxpos) | |
2117 | maxpos = next; | |
3280af22 NIS |
2118 | PL_regcomp_parse++; |
2119 | min = atoi(PL_regcomp_parse); | |
a0d0e21e LW |
2120 | if (*maxpos == ',') |
2121 | maxpos++; | |
2122 | else | |
3280af22 | 2123 | maxpos = PL_regcomp_parse; |
a0d0e21e LW |
2124 | max = atoi(maxpos); |
2125 | if (!max && *maxpos != '0') | |
c277df42 IZ |
2126 | max = REG_INFTY; /* meaning "infinity" */ |
2127 | else if (max >= REG_INFTY) | |
2128 | FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); | |
3280af22 | 2129 | PL_regcomp_parse = next; |
a0d0e21e LW |
2130 | nextchar(); |
2131 | ||
2132 | do_curly: | |
2133 | if ((flags&SIMPLE)) { | |
3280af22 | 2134 | PL_regnaughty += 2 + PL_regnaughty / 2; |
a0d0e21e LW |
2135 | reginsert(CURLY, ret); |
2136 | } | |
2137 | else { | |
2c2d71f5 JH |
2138 | regnode *w = reg_node(WHILEM); |
2139 | ||
2140 | w->flags = 0; | |
2141 | regtail(ret, w); | |
3280af22 | 2142 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 IZ |
2143 | reginsert(LONGJMP,ret); |
2144 | reginsert(NOTHING,ret); | |
2145 | NEXT_OFF(ret) = 3; /* Go over LONGJMP. */ | |
2146 | } | |
a0d0e21e | 2147 | reginsert(CURLYX,ret); |
3280af22 | 2148 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
2149 | NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */ |
2150 | regtail(ret, reg_node(NOTHING)); | |
2151 | if (SIZE_ONLY) | |
2c2d71f5 JH |
2152 | PL_reg_whilem_seen++, PL_extralen += 3; |
2153 | PL_regnaughty += 4 + PL_regnaughty; /* compound interest */ | |
a0d0e21e | 2154 | } |
c277df42 | 2155 | ret->flags = 0; |
a0d0e21e LW |
2156 | |
2157 | if (min > 0) | |
821b33a5 IZ |
2158 | *flagp = WORST; |
2159 | if (max > 0) | |
2160 | *flagp |= HASWIDTH; | |
a0d0e21e | 2161 | if (max && max < min) |
c277df42 IZ |
2162 | FAIL("Can't do {n,m} with n > m"); |
2163 | if (!SIZE_ONLY) { | |
2164 | ARG1_SET(ret, min); | |
2165 | ARG2_SET(ret, max); | |
a687059c | 2166 | } |
a687059c | 2167 | |
a0d0e21e | 2168 | goto nest_check; |
a687059c | 2169 | } |
a0d0e21e | 2170 | } |
a687059c | 2171 | |
a0d0e21e LW |
2172 | if (!ISMULT1(op)) { |
2173 | *flagp = flags; | |
a687059c | 2174 | return(ret); |
a0d0e21e | 2175 | } |
bb20fd44 | 2176 | |
c277df42 | 2177 | #if 0 /* Now runtime fix should be reliable. */ |
bb20fd44 | 2178 | if (!(flags&HASWIDTH) && op != '?') |
c277df42 IZ |
2179 | FAIL("regexp *+ operand could be empty"); |
2180 | #endif | |
bb20fd44 | 2181 | |
a0d0e21e LW |
2182 | nextchar(); |
2183 | ||
821b33a5 | 2184 | *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); |
a0d0e21e LW |
2185 | |
2186 | if (op == '*' && (flags&SIMPLE)) { | |
2187 | reginsert(STAR, ret); | |
c277df42 | 2188 | ret->flags = 0; |
3280af22 | 2189 | PL_regnaughty += 4; |
a0d0e21e LW |
2190 | } |
2191 | else if (op == '*') { | |
2192 | min = 0; | |
2193 | goto do_curly; | |
a0ed51b3 LW |
2194 | } |
2195 | else if (op == '+' && (flags&SIMPLE)) { | |
a0d0e21e | 2196 | reginsert(PLUS, ret); |
c277df42 | 2197 | ret->flags = 0; |
3280af22 | 2198 | PL_regnaughty += 3; |
a0d0e21e LW |
2199 | } |
2200 | else if (op == '+') { | |
2201 | min = 1; | |
2202 | goto do_curly; | |
a0ed51b3 LW |
2203 | } |
2204 | else if (op == '?') { | |
a0d0e21e LW |
2205 | min = 0; max = 1; |
2206 | goto do_curly; | |
2207 | } | |
2208 | nest_check: | |
e476b1b5 GS |
2209 | if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) { |
2210 | Perl_warner(aTHX_ WARN_REGEXP, "%.*s matches null string many times", | |
3280af22 | 2211 | PL_regcomp_parse - origparse, origparse); |
a0d0e21e LW |
2212 | } |
2213 | ||
3280af22 | 2214 | if (*PL_regcomp_parse == '?') { |
a0d0e21e LW |
2215 | nextchar(); |
2216 | reginsert(MINMOD, ret); | |
c277df42 | 2217 | regtail(ret, ret + NODE_STEP_REGNODE); |
a0d0e21e | 2218 | } |
3280af22 | 2219 | if (ISMULT2(PL_regcomp_parse)) |
a0d0e21e LW |
2220 | FAIL("nested *?+ in regexp"); |
2221 | ||
2222 | return(ret); | |
a687059c LW |
2223 | } |
2224 | ||
2225 | /* | |
2226 | - regatom - the lowest level | |
2227 | * | |
2228 | * Optimization: gobbles an entire sequence of ordinary characters so that | |
2229 | * it can turn them into a single node, which is smaller to store and | |
2230 | * faster to run. Backslashed characters are exceptions, each becoming a | |
2231 | * separate node; the code is simpler that way and it's not worth fixing. | |
2232 | * | |
2233 | * [Yes, it is worth fixing, some scripts can run twice the speed.] | |
2234 | */ | |
76e3520e | 2235 | STATIC regnode * |
cea2e8a9 | 2236 | S_regatom(pTHX_ I32 *flagp) |
a687059c | 2237 | { |
5c0ca799 | 2238 | dTHR; |
c277df42 | 2239 | register regnode *ret = 0; |
a0d0e21e LW |
2240 | I32 flags; |
2241 | ||
2242 | *flagp = WORST; /* Tentatively. */ | |
2243 | ||
2244 | tryagain: | |
3280af22 | 2245 | switch (*PL_regcomp_parse) { |
a0d0e21e | 2246 | case '^': |
3280af22 | 2247 | PL_seen_zerolen++; |
a0d0e21e | 2248 | nextchar(); |
3280af22 | 2249 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 2250 | ret = reg_node(MBOL); |
3280af22 | 2251 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 2252 | ret = reg_node(SBOL); |
a0d0e21e | 2253 | else |
c277df42 | 2254 | ret = reg_node(BOL); |
a0d0e21e LW |
2255 | break; |
2256 | case '$': | |
3280af22 NIS |
2257 | if (PL_regcomp_parse[1]) |
2258 | PL_seen_zerolen++; | |
a0d0e21e | 2259 | nextchar(); |
3280af22 | 2260 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 2261 | ret = reg_node(MEOL); |
3280af22 | 2262 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 2263 | ret = reg_node(SEOL); |
a0d0e21e | 2264 | else |
c277df42 | 2265 | ret = reg_node(EOL); |
a0d0e21e LW |
2266 | break; |
2267 | case '.': | |
2268 | nextchar(); | |
a0ed51b3 LW |
2269 | if (UTF) { |
2270 | if (PL_regflags & PMf_SINGLELINE) | |
2271 | ret = reg_node(SANYUTF8); | |
2272 | else | |
2273 | ret = reg_node(ANYUTF8); | |
2274 | *flagp |= HASWIDTH; | |
2275 | } | |
2276 | else { | |
2277 | if (PL_regflags & PMf_SINGLELINE) | |
2278 | ret = reg_node(SANY); | |
2279 | else | |
22c35a8c | 2280 | ret = reg_node(REG_ANY); |
a0ed51b3 LW |
2281 | *flagp |= HASWIDTH|SIMPLE; |
2282 | } | |
3280af22 | 2283 | PL_regnaughty++; |
a0d0e21e LW |
2284 | break; |
2285 | case '[': | |
3280af22 | 2286 | PL_regcomp_parse++; |
a0ed51b3 | 2287 | ret = (UTF ? regclassutf8() : regclass()); |
a14b48bc LW |
2288 | if (*PL_regcomp_parse != ']') |
2289 | FAIL("unmatched [] in regexp"); | |
2290 | nextchar(); | |
a0d0e21e LW |
2291 | *flagp |= HASWIDTH|SIMPLE; |
2292 | break; | |
2293 | case '(': | |
2294 | nextchar(); | |
2295 | ret = reg(1, &flags); | |
2296 | if (ret == NULL) { | |
2297 | if (flags & TRYAGAIN) | |
2298 | goto tryagain; | |
2299 | return(NULL); | |
2300 | } | |
c277df42 | 2301 | *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE); |
a0d0e21e LW |
2302 | break; |
2303 | case '|': | |
2304 | case ')': | |
2305 | if (flags & TRYAGAIN) { | |
2306 | *flagp |= TRYAGAIN; | |
2307 | return NULL; | |
2308 | } | |
3280af22 | 2309 | FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse); |
a0d0e21e LW |
2310 | /* Supposed to be caught earlier. */ |
2311 | break; | |
85afd4ae | 2312 | case '{': |
3280af22 NIS |
2313 | if (!regcurly(PL_regcomp_parse)) { |
2314 | PL_regcomp_parse++; | |
85afd4ae CS |
2315 | goto defchar; |
2316 | } | |
2317 | /* FALL THROUGH */ | |
a0d0e21e LW |
2318 | case '?': |
2319 | case '+': | |
2320 | case '*': | |
3115e423 | 2321 | FAIL("?+*{} follows nothing in regexp"); |
a0d0e21e LW |
2322 | break; |
2323 | case '\\': | |
3280af22 | 2324 | switch (*++PL_regcomp_parse) { |
a0d0e21e | 2325 | case 'A': |
3280af22 | 2326 | PL_seen_zerolen++; |
c277df42 | 2327 | ret = reg_node(SBOL); |
a0d0e21e LW |
2328 | *flagp |= SIMPLE; |
2329 | nextchar(); | |
2330 | break; | |
2331 | case 'G': | |
c277df42 | 2332 | ret = reg_node(GPOS); |
3280af22 | 2333 | PL_regseen |= REG_SEEN_GPOS; |
a0d0e21e LW |
2334 | *flagp |= SIMPLE; |
2335 | nextchar(); | |
2336 | break; | |
2337 | case 'Z': | |
c277df42 | 2338 | ret = reg_node(SEOL); |
a0d0e21e LW |
2339 | *flagp |= SIMPLE; |
2340 | nextchar(); | |
2341 | break; | |
b85d18e9 IZ |
2342 | case 'z': |
2343 | ret = reg_node(EOS); | |
2344 | *flagp |= SIMPLE; | |
3280af22 | 2345 | PL_seen_zerolen++; /* Do not optimize RE away */ |
b85d18e9 IZ |
2346 | nextchar(); |
2347 | break; | |
4a2d328f | 2348 | case 'C': |
a0ed51b3 LW |
2349 | ret = reg_node(SANY); |
2350 | *flagp |= HASWIDTH|SIMPLE; | |
2351 | nextchar(); | |
2352 | break; | |
2353 | case 'X': | |
2354 | ret = reg_node(CLUMP); | |
2355 | *flagp |= HASWIDTH; | |
2356 | nextchar(); | |
2357 | if (UTF && !PL_utf8_mark) | |
dfe13c55 | 2358 | is_utf8_mark((U8*)"~"); /* preload table */ |
a0ed51b3 | 2359 | break; |
a0d0e21e | 2360 | case 'w': |
a0ed51b3 LW |
2361 | ret = reg_node( |
2362 | UTF | |
2363 | ? (LOC ? ALNUMLUTF8 : ALNUMUTF8) | |
2364 | : (LOC ? ALNUML : ALNUM)); | |
a0d0e21e LW |
2365 | *flagp |= HASWIDTH|SIMPLE; |
2366 | nextchar(); | |
a0ed51b3 | 2367 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 2368 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
2369 | break; |
2370 | case 'W': | |
a0ed51b3 LW |
2371 | ret = reg_node( |
2372 | UTF | |
2373 | ? (LOC ? NALNUMLUTF8 : NALNUMUTF8) | |
2374 | : (LOC ? NALNUML : NALNUM)); | |
a0d0e21e LW |
2375 | *flagp |= HASWIDTH|SIMPLE; |
2376 | nextchar(); | |
a0ed51b3 | 2377 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 2378 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
2379 | break; |
2380 | case 'b': | |
3280af22 | 2381 | PL_seen_zerolen++; |
f5c9036e | 2382 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
2383 | ret = reg_node( |
2384 | UTF | |
2385 | ? (LOC ? BOUNDLUTF8 : BOUNDUTF8) | |
2386 | : (LOC ? BOUNDL : BOUND)); | |
a0d0e21e LW |
2387 | *flagp |= SIMPLE; |
2388 | nextchar(); | |
a0ed51b3 | 2389 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 2390 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
2391 | break; |
2392 | case 'B': | |
3280af22 | 2393 | PL_seen_zerolen++; |
f5c9036e | 2394 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
2395 | ret = reg_node( |
2396 | UTF | |
2397 | ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8) | |
2398 | : (LOC ? NBOUNDL : NBOUND)); | |
a0d0e21e LW |
2399 | *flagp |= SIMPLE; |
2400 | nextchar(); | |
a0ed51b3 | 2401 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 2402 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
2403 | break; |
2404 | case 's': | |
a0ed51b3 LW |
2405 | ret = reg_node( |
2406 | UTF | |
2407 | ? (LOC ? SPACELUTF8 : SPACEUTF8) | |
2408 | : (LOC ? SPACEL : SPACE)); | |
a0d0e21e LW |
2409 | *flagp |= HASWIDTH|SIMPLE; |
2410 | nextchar(); | |
a0ed51b3 | 2411 | if (UTF && !PL_utf8_space) |
dfe13c55 | 2412 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
2413 | break; |
2414 | case 'S': | |
a0ed51b3 LW |
2415 | ret = reg_node( |
2416 | UTF | |
2417 | ? (LOC ? NSPACELUTF8 : NSPACEUTF8) | |
2418 | : (LOC ? NSPACEL : NSPACE)); | |
a0d0e21e LW |
2419 | *flagp |= HASWIDTH|SIMPLE; |
2420 | nextchar(); | |
a0ed51b3 | 2421 | if (UTF && !PL_utf8_space) |
dfe13c55 | 2422 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
2423 | break; |
2424 | case 'd': | |
a0ed51b3 | 2425 | ret = reg_node(UTF ? DIGITUTF8 : DIGIT); |
a0d0e21e LW |
2426 | *flagp |= HASWIDTH|SIMPLE; |
2427 | nextchar(); | |
a0ed51b3 | 2428 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 2429 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e LW |
2430 | break; |
2431 | case 'D': | |
a0ed51b3 | 2432 | ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT); |
a0d0e21e LW |
2433 | *flagp |= HASWIDTH|SIMPLE; |
2434 | nextchar(); | |
a0ed51b3 | 2435 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 2436 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e | 2437 | break; |
a14b48bc LW |
2438 | case 'p': |
2439 | case 'P': | |
2440 | { /* a lovely hack--pretend we saw [\pX] instead */ | |
2441 | char* oldregxend = PL_regxend; | |
2442 | ||
2443 | if (PL_regcomp_parse[1] == '{') { | |
2444 | PL_regxend = strchr(PL_regcomp_parse, '}'); | |
2445 | if (!PL_regxend) | |
2446 | FAIL("Missing right brace on \\p{}"); | |
2447 | PL_regxend++; | |
2448 | } | |
2449 | else | |
2450 | PL_regxend = PL_regcomp_parse + 2; | |
2451 | PL_regcomp_parse--; | |
2452 | ||
2453 | ret = regclassutf8(); | |
2454 | ||
2455 | PL_regxend = oldregxend; | |
2456 | PL_regcomp_parse--; | |
2457 | nextchar(); | |
2458 | *flagp |= HASWIDTH|SIMPLE; | |
2459 | } | |
2460 | break; | |
a0d0e21e LW |
2461 | case 'n': |
2462 | case 'r': | |
2463 | case 't': | |
2464 | case 'f': | |
2465 | case 'e': | |
2466 | case 'a': | |
2467 | case 'x': | |
2468 | case 'c': | |
2469 | case '0': | |
2470 | goto defchar; | |
2471 | case '1': case '2': case '3': case '4': | |
2472 | case '5': case '6': case '7': case '8': case '9': | |
2473 | { | |
3280af22 | 2474 | I32 num = atoi(PL_regcomp_parse); |
a0d0e21e | 2475 | |
3280af22 | 2476 | if (num > 9 && num >= PL_regnpar) |
a0d0e21e LW |
2477 | goto defchar; |
2478 | else { | |
3280af22 | 2479 | if (!SIZE_ONLY && num > PL_regcomp_rx->nparens) |
ef64f398 | 2480 | FAIL("reference to nonexistent group"); |
3280af22 | 2481 | PL_regsawback = 1; |
a0ed51b3 LW |
2482 | ret = reganode(FOLD |
2483 | ? (LOC ? REFFL : REFF) | |
c8756f30 | 2484 | : REF, num); |
a0d0e21e | 2485 | *flagp |= HASWIDTH; |
3280af22 NIS |
2486 | while (isDIGIT(*PL_regcomp_parse)) |
2487 | PL_regcomp_parse++; | |
2488 | PL_regcomp_parse--; | |
a0d0e21e LW |
2489 | nextchar(); |
2490 | } | |
2491 | } | |
2492 | break; | |
2493 | case '\0': | |
3280af22 | 2494 | if (PL_regcomp_parse >= PL_regxend) |
a0d0e21e LW |
2495 | FAIL("trailing \\ in regexp"); |
2496 | /* FALL THROUGH */ | |
2497 | default: | |
c9f97d15 IZ |
2498 | /* Do not generate `unrecognized' warnings here, we fall |
2499 | back into the quick-grab loop below */ | |
a0d0e21e LW |
2500 | goto defchar; |
2501 | } | |
2502 | break; | |
4633a7c4 LW |
2503 | |
2504 | case '#': | |
3280af22 NIS |
2505 | if (PL_regflags & PMf_EXTENDED) { |
2506 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++; | |
2507 | if (PL_regcomp_parse < PL_regxend) | |
4633a7c4 LW |
2508 | goto tryagain; |
2509 | } | |
2510 | /* FALL THROUGH */ | |
2511 | ||
a0d0e21e LW |
2512 | default: { |
2513 | register I32 len; | |
a0ed51b3 | 2514 | register UV ender; |
a0d0e21e | 2515 | register char *p; |
c277df42 | 2516 | char *oldp, *s; |
a0d0e21e LW |
2517 | I32 numlen; |
2518 | ||
3280af22 | 2519 | PL_regcomp_parse++; |
a0d0e21e LW |
2520 | |
2521 | defchar: | |
a0ed51b3 LW |
2522 | ret = reg_node(FOLD |
2523 | ? (LOC ? EXACTFL : EXACTF) | |
bbce6d69 | 2524 | : EXACT); |
cd439c50 | 2525 | s = STRING(ret); |
3280af22 NIS |
2526 | for (len = 0, p = PL_regcomp_parse - 1; |
2527 | len < 127 && p < PL_regxend; | |
a0d0e21e LW |
2528 | len++) |
2529 | { | |
2530 | oldp = p; | |
5b5a24f7 | 2531 | |
3280af22 NIS |
2532 | if (PL_regflags & PMf_EXTENDED) |
2533 | p = regwhite(p, PL_regxend); | |
a0d0e21e LW |
2534 | switch (*p) { |
2535 | case '^': | |
2536 | case '$': | |
2537 | case '.': | |
2538 | case '[': | |
2539 | case '(': | |
2540 | case ')': | |
2541 | case '|': | |
2542 | goto loopdone; | |
2543 | case '\\': | |
2544 | switch (*++p) { | |
2545 | case 'A': | |
2546 | case 'G': | |
2547 | case 'Z': | |
b85d18e9 | 2548 | case 'z': |
a0d0e21e LW |
2549 | case 'w': |
2550 | case 'W': | |
2551 | case 'b': | |
2552 | case 'B': | |
2553 | case 's': | |
2554 | case 'S': | |
2555 | case 'd': | |
2556 | case 'D': | |
a14b48bc LW |
2557 | case 'p': |
2558 | case 'P': | |
a0d0e21e LW |
2559 | --p; |
2560 | goto loopdone; | |
2561 | case 'n': | |
2562 | ender = '\n'; | |
2563 | p++; | |
a687059c | 2564 | break; |
a0d0e21e LW |
2565 | case 'r': |
2566 | ender = '\r'; | |
2567 | p++; | |
a687059c | 2568 | break; |
a0d0e21e LW |
2569 | case 't': |
2570 | ender = '\t'; | |
2571 | p++; | |
a687059c | 2572 | break; |
a0d0e21e LW |
2573 | case 'f': |
2574 | ender = '\f'; | |
2575 | p++; | |
a687059c | 2576 | break; |
a0d0e21e | 2577 | case 'e': |
0407a77b GS |
2578 | #ifdef ASCIIish |
2579 | ender = '\033'; | |
2580 | #else | |
2581 | ender = '\047'; | |
2582 | #endif | |
a0d0e21e | 2583 | p++; |
a687059c | 2584 | break; |
a0d0e21e | 2585 | case 'a': |
0407a77b GS |
2586 | #ifdef ASCIIish |
2587 | ender = '\007'; | |
2588 | #else | |
2589 | ender = '\057'; | |
2590 | #endif | |
a0d0e21e | 2591 | p++; |
a687059c | 2592 | break; |
a0d0e21e | 2593 | case 'x': |
a0ed51b3 LW |
2594 | if (*++p == '{') { |
2595 | char* e = strchr(p, '}'); | |
2596 | ||
2597 | if (!e) | |
2598 | FAIL("Missing right brace on \\x{}"); | |
2599 | else if (UTF) { | |
dff6d3cd | 2600 | ender = (UV)scan_hex(p + 1, e - p, &numlen); |
a0ed51b3 LW |
2601 | if (numlen + len >= 127) { /* numlen is generous */ |
2602 | p--; | |
2603 | goto loopdone; | |
2604 | } | |
2605 | p = e + 1; | |
2606 | } | |
2607 | else | |
2608 | FAIL("Can't use \\x{} without 'use utf8' declaration"); | |
2609 | } | |
2610 | else { | |
dff6d3cd | 2611 | ender = (UV)scan_hex(p, 2, &numlen); |
a0ed51b3 LW |
2612 | p += numlen; |
2613 | } | |
a687059c | 2614 | break; |
a0d0e21e LW |
2615 | case 'c': |
2616 | p++; | |
bbce6d69 | 2617 | ender = UCHARAT(p++); |
2618 | ender = toCTRL(ender); | |
a687059c | 2619 | break; |
a0d0e21e LW |
2620 | case '0': case '1': case '2': case '3':case '4': |
2621 | case '5': case '6': case '7': case '8':case '9': | |
2622 | if (*p == '0' || | |
3280af22 | 2623 | (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) { |
dff6d3cd | 2624 | ender = (UV)scan_oct(p, 3, &numlen); |
a0d0e21e LW |
2625 | p += numlen; |
2626 | } | |
2627 | else { | |
2628 | --p; | |
2629 | goto loopdone; | |
a687059c LW |
2630 | } |
2631 | break; | |
a0d0e21e | 2632 | case '\0': |
3280af22 | 2633 | if (p >= PL_regxend) |
a687059c LW |
2634 | FAIL("trailing \\ in regexp"); |
2635 | /* FALL THROUGH */ | |
a0d0e21e | 2636 | default: |
e476b1b5 GS |
2637 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p)) |
2638 | Perl_warner(aTHX_ WARN_REGEXP, | |
1028017a JH |
2639 | "/%.127s/: Unrecognized escape \\%c passed through", |
2640 | PL_regprecomp, | |
2641 | *p); | |
a0ed51b3 | 2642 | goto normal_default; |
a0d0e21e LW |
2643 | } |
2644 | break; | |
a687059c | 2645 | default: |
a0ed51b3 LW |
2646 | normal_default: |
2647 | if ((*p & 0xc0) == 0xc0 && UTF) { | |
dfe13c55 | 2648 | ender = utf8_to_uv((U8*)p, &numlen); |
a0ed51b3 LW |
2649 | p += numlen; |
2650 | } | |
2651 | else | |
2652 | ender = *p++; | |
a0d0e21e | 2653 | break; |
a687059c | 2654 | } |
3280af22 NIS |
2655 | if (PL_regflags & PMf_EXTENDED) |
2656 | p = regwhite(p, PL_regxend); | |
a0ed51b3 LW |
2657 | if (UTF && FOLD) { |
2658 | if (LOC) | |
2659 | ender = toLOWER_LC_uni(ender); | |
2660 | else | |
2661 | ender = toLOWER_uni(ender); | |
2662 | } | |
a0d0e21e LW |
2663 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
2664 | if (len) | |
2665 | p = oldp; | |
a0ed51b3 LW |
2666 | else if (ender >= 0x80 && UTF) { |
2667 | reguni(ender, s, &numlen); | |
2668 | s += numlen; | |
2669 | len += numlen; | |
2670 | } | |
a0d0e21e LW |
2671 | else { |
2672 | len++; | |
cd439c50 | 2673 | REGC(ender, s++); |
a0d0e21e LW |
2674 | } |
2675 | break; | |
a687059c | 2676 | } |
a0ed51b3 LW |
2677 | if (ender >= 0x80 && UTF) { |
2678 | reguni(ender, s, &numlen); | |
2679 | s += numlen; | |
2680 | len += numlen - 1; | |
2681 | } | |
2682 | else | |
cd439c50 | 2683 | REGC(ender, s++); |
a0d0e21e LW |
2684 | } |
2685 | loopdone: | |
3280af22 | 2686 | PL_regcomp_parse = p - 1; |
a0d0e21e LW |
2687 | nextchar(); |
2688 | if (len < 0) | |
2689 | FAIL("internal disaster in regexp"); | |
2690 | if (len > 0) | |
2691 | *flagp |= HASWIDTH; | |
2692 | if (len == 1) | |
2693 | *flagp |= SIMPLE; | |
c277df42 | 2694 | if (!SIZE_ONLY) |
cd439c50 IZ |
2695 | STR_LEN(ret) = len; |
2696 | if (SIZE_ONLY) | |
2697 | PL_regsize += STR_SZ(len); | |
2698 | else | |
2699 | PL_regcode += STR_SZ(len); | |
a687059c | 2700 | } |
a0d0e21e LW |
2701 | break; |
2702 | } | |
a687059c | 2703 | |
a0d0e21e | 2704 | return(ret); |
a687059c LW |
2705 | } |
2706 | ||
873ef191 | 2707 | STATIC char * |
cea2e8a9 | 2708 | S_regwhite(pTHX_ char *p, char *e) |
5b5a24f7 CS |
2709 | { |
2710 | while (p < e) { | |
2711 | if (isSPACE(*p)) | |
2712 | ++p; | |
2713 | else if (*p == '#') { | |
2714 | do { | |
2715 | p++; | |
2716 | } while (p < e && *p != '\n'); | |
2717 | } | |
2718 | else | |
2719 | break; | |
2720 | } | |
2721 | return p; | |
2722 | } | |
2723 | ||
b8c5462f JH |
2724 | /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]]. |
2725 | Character classes ([:foo:]) can also be negated ([:^foo:]). | |
2726 | Returns a named class id (ANYOF_XXX) if successful, -1 otherwise. | |
2727 | Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed, | |
2728 | but trigger warnings because they are currently unimplemented. */ | |
2729 | STATIC I32 | |
cea2e8a9 | 2730 | S_regpposixcc(pTHX_ I32 value) |
620e46c5 | 2731 | { |
11b8faa4 | 2732 | dTHR; |
620e46c5 | 2733 | char *posixcc = 0; |
936ed897 | 2734 | I32 namedclass = OOB_NAMEDCLASS; |
620e46c5 JH |
2735 | |
2736 | if (value == '[' && PL_regcomp_parse + 1 < PL_regxend && | |
2737 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ | |
2738 | (*PL_regcomp_parse == ':' || | |
2739 | *PL_regcomp_parse == '=' || | |
2740 | *PL_regcomp_parse == '.')) { | |
2741 | char c = *PL_regcomp_parse; | |
2742 | char* s = PL_regcomp_parse++; | |
2743 | ||
2744 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c) | |
2745 | PL_regcomp_parse++; | |
2746 | if (PL_regcomp_parse == PL_regxend) | |
2747 | /* Grandfather lone [:, [=, [. */ | |
2748 | PL_regcomp_parse = s; | |
2749 | else { | |
b8c5462f JH |
2750 | char* t = PL_regcomp_parse++; /* skip over the c */ |
2751 | ||
2752 | if (*PL_regcomp_parse == ']') { | |
2753 | PL_regcomp_parse++; /* skip over the ending ] */ | |
2754 | posixcc = s + 1; | |
2755 | if (*s == ':') { | |
2756 | I32 complement = *posixcc == '^' ? *posixcc++ : 0; | |
2757 | I32 skip = 5; /* the most common skip */ | |
2758 | ||
2759 | switch (*posixcc) { | |
2760 | case 'a': | |
2761 | if (strnEQ(posixcc, "alnum", 5)) | |
2762 | namedclass = | |
2763 | complement ? ANYOF_NALNUMC : ANYOF_ALNUMC; | |
2764 | else if (strnEQ(posixcc, "alpha", 5)) | |
2765 | namedclass = | |
2766 | complement ? ANYOF_NALPHA : ANYOF_ALPHA; | |
2767 | else if (strnEQ(posixcc, "ascii", 5)) | |
2768 | namedclass = | |
2769 | complement ? ANYOF_NASCII : ANYOF_ASCII; | |
2770 | break; | |
2771 | case 'c': | |
2772 | if (strnEQ(posixcc, "cntrl", 5)) | |
2773 | namedclass = | |
2774 | complement ? ANYOF_NCNTRL : ANYOF_CNTRL; | |
2775 | break; | |
2776 | case 'd': | |
2777 | if (strnEQ(posixcc, "digit", 5)) | |
2778 | namedclass = | |
2779 | complement ? ANYOF_NDIGIT : ANYOF_DIGIT; | |
2780 | break; | |
2781 | case 'g': | |
2782 | if (strnEQ(posixcc, "graph", 5)) | |
2783 | namedclass = | |
2784 | complement ? ANYOF_NGRAPH : ANYOF_GRAPH; | |
2785 | break; | |
2786 | case 'l': | |
2787 | if (strnEQ(posixcc, "lower", 5)) | |
2788 | namedclass = | |
2789 | complement ? ANYOF_NLOWER : ANYOF_LOWER; | |
2790 | break; | |
2791 | case 'p': | |
2792 | if (strnEQ(posixcc, "print", 5)) | |
2793 | namedclass = | |
2794 | complement ? ANYOF_NPRINT : ANYOF_PRINT; | |
2795 | else if (strnEQ(posixcc, "punct", 5)) | |
2796 | namedclass = | |
2797 | complement ? ANYOF_NPUNCT : ANYOF_PUNCT; | |
2798 | break; | |
2799 | case 's': | |
2800 | if (strnEQ(posixcc, "space", 5)) | |
2801 | namedclass = | |
2802 | complement ? ANYOF_NSPACE : ANYOF_SPACE; | |
2803 | case 'u': | |
2804 | if (strnEQ(posixcc, "upper", 5)) | |
2805 | namedclass = | |
2806 | complement ? ANYOF_NUPPER : ANYOF_UPPER; | |
2807 | break; | |
2808 | case 'w': /* this is not POSIX, this is the Perl \w */ | |
2809 | if (strnEQ(posixcc, "word", 4)) { | |
2810 | namedclass = | |
2811 | complement ? ANYOF_NALNUM : ANYOF_ALNUM; | |
2812 | skip = 4; | |
2813 | } | |
2814 | break; | |
2815 | case 'x': | |
2816 | if (strnEQ(posixcc, "xdigit", 6)) { | |
2817 | namedclass = | |
2818 | complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT; | |
2819 | skip = 6; | |
2820 | } | |
2821 | break; | |
2822 | } | |
93733859 | 2823 | if ((namedclass == OOB_NAMEDCLASS || |
b8c5462f JH |
2824 | !(posixcc + skip + 2 < PL_regxend && |
2825 | (posixcc[skip] == ':' && | |
2826 | posixcc[skip + 1] == ']')))) | |
2827 | Perl_croak(aTHX_ "Character class [:%.*s:] unknown", | |
2828 | t - s - 1, s + 1); | |
e476b1b5 | 2829 | } else if (ckWARN(WARN_REGEXP) && !SIZE_ONLY) |
b8c5462f | 2830 | /* [[=foo=]] and [[.foo.]] are still future. */ |
e476b1b5 | 2831 | Perl_warner(aTHX_ WARN_REGEXP, |
b8c5462f JH |
2832 | "Character class syntax [%c %c] is reserved for future extensions", c, c); |
2833 | } else { | |
2834 | /* Maternal grandfather: | |
2835 | * "[:" ending in ":" but not in ":]" */ | |
767d463e HS |
2836 | PL_regcomp_parse = s; |
2837 | } | |
620e46c5 JH |
2838 | } |
2839 | } | |
2840 | ||
b8c5462f JH |
2841 | return namedclass; |
2842 | } | |
2843 | ||
2844 | STATIC void | |
2845 | S_checkposixcc(pTHX) | |
2846 | { | |
e476b1b5 | 2847 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && |
b8c5462f JH |
2848 | (*PL_regcomp_parse == ':' || |
2849 | *PL_regcomp_parse == '=' || | |
2850 | *PL_regcomp_parse == '.')) { | |
2851 | char *s = PL_regcomp_parse; | |
93733859 | 2852 | char c = *s++; |
b8c5462f JH |
2853 | |
2854 | while(*s && isALNUM(*s)) | |
2855 | s++; | |
2856 | if (*s && c == *s && s[1] == ']') { | |
e476b1b5 | 2857 | Perl_warner(aTHX_ WARN_REGEXP, |
b8c5462f JH |
2858 | "Character class syntax [%c %c] belongs inside character classes", c, c); |
2859 | if (c == '=' || c == '.') | |
e476b1b5 | 2860 | Perl_warner(aTHX_ WARN_REGEXP, |
b8c5462f JH |
2861 | "Character class syntax [%c %c] is reserved for future extensions", c, c); |
2862 | } | |
2863 | } | |
620e46c5 JH |
2864 | } |
2865 | ||
76e3520e | 2866 | STATIC regnode * |
cea2e8a9 | 2867 | S_regclass(pTHX) |
a687059c | 2868 | { |
5c0ca799 | 2869 | dTHR; |
dff6d3cd | 2870 | register UV value; |
b8c5462f | 2871 | register I32 lastvalue = OOB_CHAR8; |
a0d0e21e | 2872 | register I32 range = 0; |
c277df42 | 2873 | register regnode *ret; |
a0d0e21e LW |
2874 | register I32 def; |
2875 | I32 numlen; | |
b8c5462f | 2876 | I32 namedclass; |
73b437c8 | 2877 | char *rangebegin; |
936ed897 | 2878 | bool need_class = 0; |
a0d0e21e | 2879 | |
c277df42 | 2880 | ret = reg_node(ANYOF); |
936ed897 IZ |
2881 | if (SIZE_ONLY) |
2882 | PL_regsize += ANYOF_SKIP; | |
2883 | else { | |
2884 | ret->flags = 0; | |
2885 | ANYOF_BITMAP_ZERO(ret); | |
2886 | PL_regcode += ANYOF_SKIP; | |
2887 | if (FOLD) | |
2888 | ANYOF_FLAGS(ret) |= ANYOF_FOLD; | |
2889 | if (LOC) | |
2890 | ANYOF_FLAGS(ret) |= ANYOF_LOCALE; | |
2891 | } | |
3280af22 NIS |
2892 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2893 | PL_regnaughty++; | |
2894 | PL_regcomp_parse++; | |
c277df42 | 2895 | if (!SIZE_ONLY) |
936ed897 | 2896 | ANYOF_FLAGS(ret) |= ANYOF_INVERT; |
a0d0e21e | 2897 | } |
b8c5462f | 2898 | |
e476b1b5 | 2899 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP)) |
936ed897 | 2900 | checkposixcc(); |
b8c5462f | 2901 | |
3280af22 | 2902 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
a0d0e21e | 2903 | goto skipcond; /* allow 1st char to be ] or - */ |
3280af22 | 2904 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
a0d0e21e | 2905 | skipcond: |
93733859 | 2906 | namedclass = OOB_NAMEDCLASS; |
73b437c8 JH |
2907 | if (!range) |
2908 | rangebegin = PL_regcomp_parse; | |
a0ed51b3 | 2909 | value = UCHARAT(PL_regcomp_parse++); |
620e46c5 | 2910 | if (value == '[') |
b8c5462f | 2911 | namedclass = regpposixcc(value); |
620e46c5 | 2912 | else if (value == '\\') { |
a0ed51b3 LW |
2913 | value = UCHARAT(PL_regcomp_parse++); |
2914 | switch (value) { | |
b8c5462f JH |
2915 | case 'w': namedclass = ANYOF_ALNUM; break; |
2916 | case 'W': namedclass = ANYOF_NALNUM; break; | |
2917 | case 's': namedclass = ANYOF_SPACE; break; | |
2918 | case 'S': namedclass = ANYOF_NSPACE; break; | |
2919 | case 'd': namedclass = ANYOF_DIGIT; break; | |
2920 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
2921 | case 'n': value = '\n'; break; | |
2922 | case 'r': value = '\r'; break; | |
2923 | case 't': value = '\t'; break; | |
2924 | case 'f': value = '\f'; break; | |
2925 | case 'b': value = '\b'; break; | |
0407a77b | 2926 | #ifdef ASCIIish |
b8c5462f JH |
2927 | case 'e': value = '\033'; break; |
2928 | case 'a': value = '\007'; break; | |
0407a77b GS |
2929 | #else |
2930 | case 'e': value = '\047'; break; | |
2931 | case 'a': value = '\057'; break; | |
2932 | #endif | |
b8c5462f | 2933 | case 'x': |
dff6d3cd | 2934 | value = (UV)scan_hex(PL_regcomp_parse, 2, &numlen); |
b8c5462f JH |
2935 | PL_regcomp_parse += numlen; |
2936 | break; | |
2937 | case 'c': | |
2938 | value = UCHARAT(PL_regcomp_parse++); | |
2939 | value = toCTRL(value); | |
2940 | break; | |
2941 | case '0': case '1': case '2': case '3': case '4': | |
2942 | case '5': case '6': case '7': case '8': case '9': | |
dff6d3cd | 2943 | value = (UV)scan_oct(--PL_regcomp_parse, 3, &numlen); |
b8c5462f JH |
2944 | PL_regcomp_parse += numlen; |
2945 | break; | |
1028017a | 2946 | default: |
e476b1b5 GS |
2947 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value)) |
2948 | Perl_warner(aTHX_ WARN_REGEXP, | |
1028017a JH |
2949 | "/%.127s/: Unrecognized escape \\%c in character class passed through", |
2950 | PL_regprecomp, | |
447ff955 | 2951 | (int)value); |
1028017a | 2952 | break; |
b8c5462f JH |
2953 | } |
2954 | } | |
73b437c8 | 2955 | if (namedclass > OOB_NAMEDCLASS) { |
936ed897 IZ |
2956 | if (!need_class && !SIZE_ONLY) |
2957 | ANYOF_CLASS_ZERO(ret); | |
2958 | need_class = 1; | |
73b437c8 JH |
2959 | if (range) { /* a-\d, a-[:digit:] */ |
2960 | if (!SIZE_ONLY) { | |
e476b1b5 GS |
2961 | if (ckWARN(WARN_REGEXP)) |
2962 | Perl_warner(aTHX_ WARN_REGEXP, | |
73b437c8 JH |
2963 | "/%.127s/: false [] range \"%*.*s\" in regexp", |
2964 | PL_regprecomp, | |
2965 | PL_regcomp_parse - rangebegin, | |
2966 | PL_regcomp_parse - rangebegin, | |
2967 | rangebegin); | |
936ed897 IZ |
2968 | ANYOF_BITMAP_SET(ret, lastvalue); |
2969 | ANYOF_BITMAP_SET(ret, '-'); | |
b8c5462f | 2970 | } |
73b437c8 JH |
2971 | range = 0; /* this is not a true range */ |
2972 | } | |
2973 | if (!SIZE_ONLY) { | |
2974 | switch (namedclass) { | |
2975 | case ANYOF_ALNUM: | |
2976 | if (LOC) | |
936ed897 | 2977 | ANYOF_CLASS_SET(ret, ANYOF_ALNUM); |
73b437c8 JH |
2978 | else { |
2979 | for (value = 0; value < 256; value++) | |
2980 | if (isALNUM(value)) | |
936ed897 | 2981 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
2982 | } |
2983 | break; | |
2984 | case ANYOF_NALNUM: | |
2985 | if (LOC) | |
936ed897 | 2986 | ANYOF_CLASS_SET(ret, ANYOF_NALNUM); |
73b437c8 JH |
2987 | else { |
2988 | for (value = 0; value < 256; value++) | |
2989 | if (!isALNUM(value)) | |
936ed897 | 2990 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
2991 | } |
2992 | break; | |
2993 | case ANYOF_SPACE: | |
2994 | if (LOC) | |
936ed897 | 2995 | ANYOF_CLASS_SET(ret, ANYOF_SPACE); |
73b437c8 JH |
2996 | else { |
2997 | for (value = 0; value < 256; value++) | |
2998 | if (isSPACE(value)) | |
936ed897 | 2999 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3000 | } |
3001 | break; | |
3002 | case ANYOF_NSPACE: | |
3003 | if (LOC) | |
936ed897 | 3004 | ANYOF_CLASS_SET(ret, ANYOF_NSPACE); |
73b437c8 JH |
3005 | else { |
3006 | for (value = 0; value < 256; value++) | |
3007 | if (!isSPACE(value)) | |
936ed897 | 3008 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3009 | } |
3010 | break; | |
3011 | case ANYOF_DIGIT: | |
3012 | if (LOC) | |
936ed897 | 3013 | ANYOF_CLASS_SET(ret, ANYOF_DIGIT); |
73b437c8 JH |
3014 | else { |
3015 | for (value = '0'; value <= '9'; value++) | |
936ed897 | 3016 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3017 | } |
3018 | break; | |
3019 | case ANYOF_NDIGIT: | |
3020 | if (LOC) | |
936ed897 | 3021 | ANYOF_CLASS_SET(ret, ANYOF_NDIGIT); |
73b437c8 JH |
3022 | else { |
3023 | for (value = 0; value < '0'; value++) | |
936ed897 | 3024 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 | 3025 | for (value = '9' + 1; value < 256; value++) |
936ed897 | 3026 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3027 | } |
3028 | break; | |
3029 | case ANYOF_NALNUMC: | |
3030 | if (LOC) | |
936ed897 | 3031 | ANYOF_CLASS_SET(ret, ANYOF_NALNUMC); |
73b437c8 JH |
3032 | else { |
3033 | for (value = 0; value < 256; value++) | |
3034 | if (!isALNUMC(value)) | |
936ed897 | 3035 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3036 | } |
3037 | break; | |
3038 | case ANYOF_ALNUMC: | |
3039 | if (LOC) | |
936ed897 | 3040 | ANYOF_CLASS_SET(ret, ANYOF_ALNUMC); |
73b437c8 JH |
3041 | else { |
3042 | for (value = 0; value < 256; value++) | |
3043 | if (isALNUMC(value)) | |
936ed897 | 3044 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3045 | } |
3046 | break; | |
3047 | case ANYOF_ALPHA: | |
3048 | if (LOC) | |
936ed897 | 3049 | ANYOF_CLASS_SET(ret, ANYOF_ALPHA); |
73b437c8 JH |
3050 | else { |
3051 | for (value = 0; value < 256; value++) | |
3052 | if (isALPHA(value)) | |
936ed897 | 3053 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3054 | } |
3055 | break; | |
3056 | case ANYOF_NALPHA: | |
3057 | if (LOC) | |
936ed897 | 3058 | ANYOF_CLASS_SET(ret, ANYOF_NALPHA); |
73b437c8 JH |
3059 | else { |
3060 | for (value = 0; value < 256; value++) | |
3061 | if (!isALPHA(value)) | |
936ed897 | 3062 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3063 | } |
3064 | break; | |
3065 | case ANYOF_ASCII: | |
3066 | if (LOC) | |
936ed897 | 3067 | ANYOF_CLASS_SET(ret, ANYOF_ASCII); |
73b437c8 | 3068 | else { |
a5993822 | 3069 | #ifdef ASCIIish |
73b437c8 | 3070 | for (value = 0; value < 128; value++) |
936ed897 | 3071 | ANYOF_BITMAP_SET(ret, value); |
a5993822 PP |
3072 | #else /* EBCDIC */ |
3073 | for (value = 0; value < 256; value++) | |
3074 | if (isASCII(value)) | |
3075 | ANYOF_BITMAP_SET(ret, value); | |
3076 | #endif /* EBCDIC */ | |
73b437c8 JH |
3077 | } |
3078 | break; | |
3079 | case ANYOF_NASCII: | |
3080 | if (LOC) | |
936ed897 | 3081 | ANYOF_CLASS_SET(ret, ANYOF_NASCII); |
73b437c8 | 3082 | else { |
a5993822 | 3083 | #ifdef ASCIIish |
73b437c8 | 3084 | for (value = 128; value < 256; value++) |
936ed897 | 3085 | ANYOF_BITMAP_SET(ret, value); |
a5993822 PP |
3086 | #else /* EBCDIC */ |
3087 | for (value = 0; value < 256; value++) | |
3088 | if (!isASCII(value)) | |
3089 | ANYOF_BITMAP_SET(ret, value); | |
3090 | #endif /* EBCDIC */ | |
73b437c8 JH |
3091 | } |
3092 | break; | |
3093 | case ANYOF_CNTRL: | |
3094 | if (LOC) | |
936ed897 | 3095 | ANYOF_CLASS_SET(ret, ANYOF_CNTRL); |
73b437c8 JH |
3096 | else { |
3097 | for (value = 0; value < 256; value++) | |
3098 | if (isCNTRL(value)) | |
936ed897 | 3099 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3100 | } |
3101 | lastvalue = OOB_CHAR8; | |
3102 | break; | |
3103 | case ANYOF_NCNTRL: | |
3104 | if (LOC) | |
936ed897 | 3105 | ANYOF_CLASS_SET(ret, ANYOF_NCNTRL); |
73b437c8 JH |
3106 | else { |
3107 | for (value = 0; value < 256; value++) | |
3108 | if (!isCNTRL(value)) | |
936ed897 | 3109 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3110 | } |
3111 | break; | |
3112 | case ANYOF_GRAPH: | |
3113 | if (LOC) | |
936ed897 | 3114 | ANYOF_CLASS_SET(ret, ANYOF_GRAPH); |
73b437c8 JH |
3115 | else { |
3116 | for (value = 0; value < 256; value++) | |
3117 | if (isGRAPH(value)) | |
936ed897 | 3118 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3119 | } |
3120 | break; | |
3121 | case ANYOF_NGRAPH: | |
3122 | if (LOC) | |
936ed897 | 3123 | ANYOF_CLASS_SET(ret, ANYOF_NGRAPH); |
73b437c8 JH |
3124 | else { |
3125 | for (value = 0; value < 256; value++) | |
3126 | if (!isGRAPH(value)) | |
936ed897 | 3127 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3128 | } |
3129 | break; | |
3130 | case ANYOF_LOWER: | |
3131 | if (LOC) | |
936ed897 | 3132 | ANYOF_CLASS_SET(ret, ANYOF_LOWER); |
73b437c8 JH |
3133 | else { |
3134 | for (value = 0; value < 256; value++) | |
3135 | if (isLOWER(value)) | |
936ed897 | 3136 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3137 | } |
3138 | break; | |
3139 | case ANYOF_NLOWER: | |
3140 | if (LOC) | |
936ed897 | 3141 | ANYOF_CLASS_SET(ret, ANYOF_NLOWER); |
73b437c8 JH |
3142 | else { |
3143 | for (value = 0; value < 256; value++) | |
3144 | if (!isLOWER(value)) | |
936ed897 | 3145 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3146 | } |
3147 | break; | |
3148 | case ANYOF_PRINT: | |
3149 | if (LOC) | |
936ed897 | 3150 | ANYOF_CLASS_SET(ret, ANYOF_PRINT); |
73b437c8 JH |
3151 | else { |
3152 | for (value = 0; value < 256; value++) | |
3153 | if (isPRINT(value)) | |
936ed897 | 3154 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3155 | } |
3156 | break; | |
3157 | case ANYOF_NPRINT: | |
3158 | if (LOC) | |
936ed897 | 3159 | ANYOF_CLASS_SET(ret, ANYOF_NPRINT); |
73b437c8 JH |
3160 | else { |
3161 | for (value = 0; value < 256; value++) | |
3162 | if (!isPRINT(value)) | |
936ed897 | 3163 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3164 | } |
3165 | break; | |
3166 | case ANYOF_PUNCT: | |
3167 | if (LOC) | |
936ed897 | 3168 | ANYOF_CLASS_SET(ret, ANYOF_PUNCT); |
73b437c8 JH |
3169 | else { |
3170 | for (value = 0; value < 256; value++) | |
3171 | if (isPUNCT(value)) | |
936ed897 | 3172 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3173 | } |
3174 | break; | |
3175 | case ANYOF_NPUNCT: | |
3176 | if (LOC) | |
936ed897 | 3177 | ANYOF_CLASS_SET(ret, ANYOF_NPUNCT); |
73b437c8 JH |
3178 | else { |
3179 | for (value = 0; value < 256; value++) | |
3180 | if (!isPUNCT(value)) | |
936ed897 | 3181 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3182 | } |
3183 | break; | |
3184 | case ANYOF_UPPER: | |
3185 | if (LOC) | |
936ed897 | 3186 | ANYOF_CLASS_SET(ret, ANYOF_UPPER); |
73b437c8 JH |
3187 | else { |
3188 | for (value = 0; value < 256; value++) | |
3189 | if (isUPPER(value)) | |
936ed897 | 3190 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3191 | } |
3192 | break; | |
3193 | case ANYOF_NUPPER: | |
3194 | if (LOC) | |
936ed897 | 3195 | ANYOF_CLASS_SET(ret, ANYOF_NUPPER); |
73b437c8 JH |
3196 | else { |
3197 | for (value = 0; value < 256; value++) | |
3198 | if (!isUPPER(value)) | |
936ed897 | 3199 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3200 | } |
3201 | break; | |
3202 | case ANYOF_XDIGIT: | |
3203 | if (LOC) | |
936ed897 | 3204 | ANYOF_CLASS_SET(ret, ANYOF_XDIGIT); |
73b437c8 JH |
3205 | else { |
3206 | for (value = 0; value < 256; value++) | |
3207 | if (isXDIGIT(value)) | |
936ed897 | 3208 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3209 | } |
3210 | break; | |
3211 | case ANYOF_NXDIGIT: | |
3212 | if (LOC) | |
936ed897 | 3213 | ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT); |
73b437c8 JH |
3214 | else { |
3215 | for (value = 0; value < 256; value++) | |
3216 | if (!isXDIGIT(value)) | |
936ed897 | 3217 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3218 | } |
3219 | break; | |
3220 | default: | |
3221 | FAIL("invalid [::] class in regexp"); | |
3222 | break; | |
b8c5462f | 3223 | } |
b8c5462f | 3224 | if (LOC) |
936ed897 | 3225 | ANYOF_FLAGS(ret) |= ANYOF_CLASS; |
73b437c8 | 3226 | continue; |
a0d0e21e LW |
3227 | } |
3228 | } | |
3229 | if (range) { | |
73b437c8 JH |
3230 | if (lastvalue > value) /* b-a */ { |
3231 | Perl_croak(aTHX_ | |
3232 | "/%.127s/: invalid [] range \"%*.*s\" in regexp", | |
3233 | PL_regprecomp, | |
3234 | PL_regcomp_parse - rangebegin, | |
3235 | PL_regcomp_parse - rangebegin, | |
3236 | rangebegin); | |
3237 | } | |
a0d0e21e LW |
3238 | range = 0; |
3239 | } | |
3240 | else { | |
a0ed51b3 | 3241 | lastvalue = value; |
3280af22 | 3242 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
93733859 | 3243 | PL_regcomp_parse[1] != ']') { |
3280af22 | 3244 | PL_regcomp_parse++; |
73b437c8 | 3245 | if (namedclass > OOB_NAMEDCLASS) { /* \w-, [:word:]- */ |
e476b1b5 GS |
3246 | if (ckWARN(WARN_REGEXP)) |
3247 | Perl_warner(aTHX_ WARN_REGEXP, | |
73b437c8 JH |
3248 | "/%.127s/: false [] range \"%*.*s\" in regexp", |
3249 | PL_regprecomp, | |
3250 | PL_regcomp_parse - rangebegin, | |
3251 | PL_regcomp_parse - rangebegin, | |
3252 | rangebegin); | |
3253 | if (!SIZE_ONLY) | |
936ed897 | 3254 | ANYOF_BITMAP_SET(ret, '-'); |
73b437c8 JH |
3255 | } else |
3256 | range = 1; | |
a0d0e21e LW |
3257 | continue; /* do it next time */ |
3258 | } | |
a687059c | 3259 | } |
93733859 | 3260 | /* now is the next time */ |
ae5c130c | 3261 | if (!SIZE_ONLY) { |
93733859 | 3262 | #ifndef ASCIIish /* EBCDIC, for example. */ |
8ada0baa | 3263 | if ((isLOWER(lastvalue) && isLOWER(value)) || |
c8dba6f3 GS |
3264 | (isUPPER(lastvalue) && isUPPER(value))) |
3265 | { | |
3266 | I32 i; | |
8ada0baa JH |
3267 | if (isLOWER(lastvalue)) { |
3268 | for (i = lastvalue; i <= value; i++) | |
3269 | if (isLOWER(i)) | |
936ed897 | 3270 | ANYOF_BITMAP_SET(ret, i); |
8ada0baa JH |
3271 | } else { |
3272 | for (i = lastvalue; i <= value; i++) | |
3273 | if (isUPPER(i)) | |
936ed897 | 3274 | ANYOF_BITMAP_SET(ret, i); |
8ada0baa JH |
3275 | } |
3276 | } | |
3277 | else | |
3278 | #endif | |
3279 | for ( ; lastvalue <= value; lastvalue++) | |
936ed897 | 3280 | ANYOF_BITMAP_SET(ret, lastvalue); |
8ada0baa | 3281 | } |
93733859 | 3282 | range = 0; |
a0d0e21e | 3283 | } |
936ed897 IZ |
3284 | if (need_class) { |
3285 | if (SIZE_ONLY) | |
3286 | PL_regsize += ANYOF_CLASS_ADD_SKIP; | |
3287 | else | |
3288 | PL_regcode += ANYOF_CLASS_ADD_SKIP; | |
3289 | } | |
ae5c130c | 3290 | /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ |
b8c5462f | 3291 | if (!SIZE_ONLY && |
936ed897 | 3292 | (ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD) { |
a0ed51b3 | 3293 | for (value = 0; value < 256; ++value) { |
936ed897 | 3294 | if (ANYOF_BITMAP_TEST(ret, value)) { |
22c35a8c | 3295 | I32 cf = PL_fold[value]; |
936ed897 | 3296 | ANYOF_BITMAP_SET(ret, cf); |
ae5c130c GS |
3297 | } |
3298 | } | |
936ed897 | 3299 | ANYOF_FLAGS(ret) &= ~ANYOF_FOLD; |
ae5c130c GS |
3300 | } |
3301 | /* optimize inverted simple patterns (e.g. [^a-z]) */ | |
936ed897 | 3302 | if (!SIZE_ONLY && (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) { |
b8c5462f | 3303 | for (value = 0; value < ANYOF_BITMAP_SIZE; ++value) |
936ed897 IZ |
3304 | ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL; |
3305 | ANYOF_FLAGS(ret) = 0; | |
ae5c130c | 3306 | } |
a0d0e21e LW |
3307 | return ret; |
3308 | } | |
3309 | ||
a0ed51b3 | 3310 | STATIC regnode * |
cea2e8a9 | 3311 | S_regclassutf8(pTHX) |
a0ed51b3 | 3312 | { |
b8c5462f | 3313 | dTHR; |
936ed897 | 3314 | register char *e; |
dff6d3cd | 3315 | register UV value; |
b8c5462f | 3316 | register U32 lastvalue = OOB_UTF8; |
a0ed51b3 LW |
3317 | register I32 range = 0; |
3318 | register regnode *ret; | |
3319 | I32 numlen; | |
3320 | I32 n; | |
3321 | SV *listsv; | |
3322 | U8 flags = 0; | |
b8c5462f | 3323 | I32 namedclass; |
73b437c8 | 3324 | char *rangebegin; |
a0ed51b3 LW |
3325 | |
3326 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ | |
3327 | PL_regnaughty++; | |
3328 | PL_regcomp_parse++; | |
3329 | if (!SIZE_ONLY) | |
3330 | flags |= ANYOF_INVERT; | |
3331 | } | |
3332 | if (!SIZE_ONLY) { | |
3333 | if (FOLD) | |
3334 | flags |= ANYOF_FOLD; | |
3335 | if (LOC) | |
3336 | flags |= ANYOF_LOCALE; | |
79cb57f6 | 3337 | listsv = newSVpvn("# comment\n",10); |
a0ed51b3 LW |
3338 | } |
3339 | ||
e476b1b5 | 3340 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP)) |
936ed897 | 3341 | checkposixcc(); |
b8c5462f | 3342 | |
a0ed51b3 LW |
3343 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
3344 | goto skipcond; /* allow 1st char to be ] or - */ | |
3345 | ||
3346 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { | |
3347 | skipcond: | |
93733859 | 3348 | namedclass = OOB_NAMEDCLASS; |
73b437c8 JH |
3349 | if (!range) |
3350 | rangebegin = PL_regcomp_parse; | |
dfe13c55 | 3351 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 | 3352 | PL_regcomp_parse += numlen; |
620e46c5 | 3353 | if (value == '[') |
b8c5462f | 3354 | namedclass = regpposixcc(value); |
620e46c5 | 3355 | else if (value == '\\') { |
dfe13c55 | 3356 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 LW |
3357 | PL_regcomp_parse += numlen; |
3358 | switch (value) { | |
b8c5462f JH |
3359 | case 'w': namedclass = ANYOF_ALNUM; break; |
3360 | case 'W': namedclass = ANYOF_NALNUM; break; | |
3361 | case 's': namedclass = ANYOF_SPACE; break; | |
3362 | case 'S': namedclass = ANYOF_NSPACE; break; | |
3363 | case 'd': namedclass = ANYOF_DIGIT; break; | |
3364 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
a0ed51b3 LW |
3365 | case 'p': |
3366 | case 'P': | |
3367 | if (*PL_regcomp_parse == '{') { | |
3368 | e = strchr(PL_regcomp_parse++, '}'); | |
3369 | if (!e) | |
3370 | FAIL("Missing right brace on \\p{}"); | |
3371 | n = e - PL_regcomp_parse; | |
3372 | } | |
3373 | else { | |
3374 | e = PL_regcomp_parse; | |
3375 | n = 1; | |
3376 | } | |
3377 | if (!SIZE_ONLY) { | |
3378 | if (value == 'p') | |
b8c5462f | 3379 | Perl_sv_catpvf(aTHX_ listsv, |
894356b3 | 3380 | "+utf8::%.*s\n", (int)n, PL_regcomp_parse); |
a0ed51b3 | 3381 | else |
cea2e8a9 | 3382 | Perl_sv_catpvf(aTHX_ listsv, |
894356b3 | 3383 | "!utf8::%.*s\n", (int)n, PL_regcomp_parse); |
a0ed51b3 LW |
3384 | } |
3385 | PL_regcomp_parse = e + 1; | |
b8c5462f | 3386 | lastvalue = OOB_UTF8; |
a0ed51b3 | 3387 | continue; |
b8c5462f JH |
3388 | case 'n': value = '\n'; break; |
3389 | case 'r': value = '\r'; break; | |
3390 | case 't': value = '\t'; break; | |
3391 | case 'f': value = '\f'; break; | |
3392 | case 'b': value = '\b'; break; | |
0407a77b | 3393 | #ifdef ASCIIish |
b8c5462f JH |
3394 | case 'e': value = '\033'; break; |
3395 | case 'a': value = '\007'; break; | |
0407a77b GS |
3396 | #else |
3397 | case 'e': value = '\047'; break; | |
3398 | case 'a': value = '\057'; break; | |
3399 | #endif | |
a0ed51b3 LW |
3400 | case 'x': |
3401 | if (*PL_regcomp_parse == '{') { | |
3402 | e = strchr(PL_regcomp_parse++, '}'); | |
3403 | if (!e) | |
3404 | FAIL("Missing right brace on \\x{}"); | |
dff6d3cd | 3405 | value = (UV)scan_hex(PL_regcomp_parse, |
b8c5462f JH |
3406 | e - PL_regcomp_parse, |
3407 | &numlen); | |
a0ed51b3 LW |
3408 | PL_regcomp_parse = e + 1; |
3409 | } | |
3410 | else { | |
dff6d3cd | 3411 | value = (UV)scan_hex(PL_regcomp_parse, 2, &numlen); |
a0ed51b3 LW |
3412 | PL_regcomp_parse += numlen; |
3413 | } | |
3414 | break; | |
3415 | case 'c': | |
3416 | value = UCHARAT(PL_regcomp_parse++); | |
3417 | value = toCTRL(value); | |
3418 | break; | |
3419 | case '0': case '1': case '2': case '3': case '4': | |
3420 | case '5': case '6': case '7': case '8': case '9': | |
dff6d3cd | 3421 | value = (UV)scan_oct(--PL_regcomp_parse, 3, &numlen); |
a0ed51b3 LW |
3422 | PL_regcomp_parse += numlen; |
3423 | break; | |
1028017a | 3424 | default: |
e476b1b5 GS |
3425 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value)) |
3426 | Perl_warner(aTHX_ WARN_REGEXP, | |
1028017a JH |
3427 | "/%.127s/: Unrecognized escape \\%c in character class passed through", |
3428 | PL_regprecomp, | |
447ff955 | 3429 | (int)value); |
1028017a | 3430 | break; |
a0ed51b3 LW |
3431 | } |
3432 | } | |
73b437c8 JH |
3433 | if (namedclass > OOB_NAMEDCLASS) { |
3434 | if (range) { /* a-\d, a-[:digit:] */ | |
3435 | if (!SIZE_ONLY) { | |
e476b1b5 GS |
3436 | if (ckWARN(WARN_REGEXP)) |
3437 | Perl_warner(aTHX_ WARN_REGEXP, | |
73b437c8 JH |
3438 | "/%.127s/: false [] range \"%*.*s\" in regexp", |
3439 | PL_regprecomp, | |
3440 | PL_regcomp_parse - rangebegin, | |
3441 | PL_regcomp_parse - rangebegin, | |
3442 | rangebegin); | |
3443 | Perl_sv_catpvf(aTHX_ listsv, | |
3444 | /* 0x002D is Unicode for '-' */ | |
3445 | "%04"UVxf"\n002D\n", (UV)lastvalue); | |
3446 | } | |
3447 | range = 0; | |
3448 | } | |
3449 | if (!SIZE_ONLY) { | |
3450 | switch (namedclass) { | |
3451 | case ANYOF_ALNUM: | |
3452 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n"); break; | |
3453 | case ANYOF_NALNUM: | |
3454 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n"); break; | |
3455 | case ANYOF_ALNUMC: | |
3456 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n"); break; | |
3457 | case ANYOF_NALNUMC: | |
3458 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n"); break; | |
3459 | case ANYOF_ALPHA: | |
3460 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n"); break; | |
3461 | case ANYOF_NALPHA: | |
3462 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n"); break; | |
3463 | case ANYOF_ASCII: | |
3464 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n"); break; | |
3465 | case ANYOF_NASCII: | |
3466 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n"); break; | |
3467 | case ANYOF_CNTRL: | |
3468 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n"); break; | |
3469 | case ANYOF_NCNTRL: | |
3470 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n"); break; | |
3471 | case ANYOF_GRAPH: | |
3472 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n"); break; | |
3473 | case ANYOF_NGRAPH: | |
3474 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n"); break; | |
3475 | case ANYOF_DIGIT: | |
3476 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n"); break; | |
3477 | case ANYOF_NDIGIT: | |
3478 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n"); break; | |
3479 | case ANYOF_LOWER: | |
3480 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n"); break; | |
3481 | case ANYOF_NLOWER: | |
3482 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n"); break; | |
3483 | case ANYOF_PRINT: | |
3484 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n"); break; | |
3485 | case ANYOF_NPRINT: | |
3486 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n"); break; | |
3487 | case ANYOF_PUNCT: | |
3488 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n"); break; | |
3489 | case ANYOF_NPUNCT: | |
3490 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n"); break; | |
3491 | case ANYOF_SPACE: | |
3492 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n"); break; | |
3493 | case ANYOF_NSPACE: | |
3494 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n"); break; | |
3495 | case ANYOF_UPPER: | |
3496 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n"); break; | |
3497 | case ANYOF_NUPPER: | |
3498 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n"); break; | |
3499 | case ANYOF_XDIGIT: | |
3500 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n"); break; | |
3501 | case ANYOF_NXDIGIT: | |
3502 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n"); break; | |
3503 | } | |
3504 | continue; | |
b8c5462f JH |
3505 | } |
3506 | } | |
b8c5462f | 3507 | if (range) { |
73b437c8 JH |
3508 | if (lastvalue > value) { /* b-a */ |
3509 | Perl_croak(aTHX_ | |
3510 | "/%.127s/: invalid [] range \"%*.*s\" in regexp", | |
3511 | PL_regprecomp, | |
3512 | PL_regcomp_parse - rangebegin, | |
3513 | PL_regcomp_parse - rangebegin, | |
3514 | rangebegin); | |
3515 | } | |
a0ed51b3 LW |
3516 | range = 0; |
3517 | } | |
3518 | else { | |
3519 | lastvalue = value; | |
3520 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && | |
93733859 | 3521 | PL_regcomp_parse[1] != ']') { |
a0ed51b3 | 3522 | PL_regcomp_parse++; |
73b437c8 | 3523 | if (namedclass > OOB_NAMEDCLASS) { /* \w-, [:word:]- */ |
e476b1b5 GS |
3524 | if (ckWARN(WARN_REGEXP)) |
3525 | Perl_warner(aTHX_ WARN_REGEXP, | |
73b437c8 JH |
3526 | "/%.127s/: false [] range \"%*.*s\" in regexp", |
3527 | PL_regprecomp, | |
3528 | PL_regcomp_parse - rangebegin, | |
3529 | PL_regcomp_parse - rangebegin, | |
3530 | rangebegin); | |
3531 | if (!SIZE_ONLY) | |
3532 | Perl_sv_catpvf(aTHX_ listsv, | |
3533 | /* 0x002D is Unicode for '-' */ | |
3534 | "002D\n"); | |
3535 | } else | |
3536 | range = 1; | |
a0ed51b3 LW |
3537 | continue; /* do it next time */ |
3538 | } | |
93733859 JH |
3539 | } |
3540 | /* now is the next time */ | |
93733859 | 3541 | if (!SIZE_ONLY) |
73b437c8 JH |
3542 | Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n", |
3543 | (UV)lastvalue, (UV)value); | |
93733859 | 3544 | range = 0; |
a0ed51b3 | 3545 | } |
a0ed51b3 LW |
3546 | |
3547 | ret = reganode(ANYOFUTF8, 0); | |
3548 | ||
3549 | if (!SIZE_ONLY) { | |
3550 | SV *rv = swash_init("utf8", "", listsv, 1, 0); | |
3551 | SvREFCNT_dec(listsv); | |
3552 | n = add_data(1,"s"); | |
3553 | PL_regcomp_rx->data->data[n] = (void*)rv; | |
3554 | ARG1_SET(ret, flags); | |
3555 | ARG2_SET(ret, n); | |
3556 | } | |
3557 | ||
3558 | return ret; | |
3559 | } | |
3560 | ||
76e3520e | 3561 | STATIC char* |
cea2e8a9 | 3562 | S_nextchar(pTHX) |
a0d0e21e | 3563 | { |
5c0ca799 | 3564 | dTHR; |
3280af22 | 3565 | char* retval = PL_regcomp_parse++; |
a0d0e21e | 3566 | |
4633a7c4 | 3567 | for (;;) { |
3280af22 NIS |
3568 | if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' && |
3569 | PL_regcomp_parse[2] == '#') { | |
3570 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') | |
3571 | PL_regcomp_parse++; | |
3572 | PL_regcomp_parse++; | |
4633a7c4 LW |
3573 | continue; |
3574 | } | |
3280af22 NIS |
3575 | if (PL_regflags & PMf_EXTENDED) { |
3576 | if (isSPACE(*PL_regcomp_parse)) { | |
3577 | PL_regcomp_parse++; | |
748a9306 LW |
3578 | continue; |
3579 | } | |
3280af22 NIS |
3580 | else if (*PL_regcomp_parse == '#') { |
3581 | while (*PL_regcomp_parse && *PL_regcomp_parse != '\n') | |
3582 | PL_regcomp_parse++; | |
3583 | PL_regcomp_parse++; | |
748a9306 LW |
3584 | continue; |
3585 | } | |
748a9306 | 3586 | } |
4633a7c4 | 3587 | return retval; |
a0d0e21e | 3588 | } |
a687059c LW |
3589 | } |
3590 | ||
3591 | /* | |
c277df42 | 3592 | - reg_node - emit a node |
a0d0e21e | 3593 | */ |
76e3520e | 3594 | STATIC regnode * /* Location. */ |
cea2e8a9 | 3595 | S_reg_node(pTHX_ U8 op) |
a687059c | 3596 | { |
5c0ca799 | 3597 | dTHR; |
c277df42 IZ |
3598 | register regnode *ret; |
3599 | register regnode *ptr; | |
a687059c | 3600 | |
3280af22 | 3601 | ret = PL_regcode; |
c277df42 | 3602 | if (SIZE_ONLY) { |
6b88bc9c | 3603 | SIZE_ALIGN(PL_regsize); |
3280af22 | 3604 | PL_regsize += 1; |
a0d0e21e LW |
3605 | return(ret); |
3606 | } | |
a687059c | 3607 | |
c277df42 | 3608 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 3609 | ptr = ret; |
c277df42 | 3610 | FILL_ADVANCE_NODE(ptr, op); |
3280af22 | 3611 | PL_regcode = ptr; |
a687059c | 3612 | |
a0d0e21e | 3613 | return(ret); |
a687059c LW |
3614 | } |
3615 | ||
3616 | /* | |
a0d0e21e LW |
3617 | - reganode - emit a node with an argument |
3618 | */ | |
76e3520e | 3619 | STATIC regnode * /* Location. */ |
cea2e8a9 | 3620 | S_reganode(pTHX_ U8 op, U32 arg) |
fe14fcc3 | 3621 | { |
5c0ca799 | 3622 | dTHR; |
c277df42 IZ |
3623 | register regnode *ret; |
3624 | register regnode *ptr; | |
fe14fcc3 | 3625 | |
3280af22 | 3626 | ret = PL_regcode; |
c277df42 | 3627 | if (SIZE_ONLY) { |
6b88bc9c | 3628 | SIZE_ALIGN(PL_regsize); |
3280af22 | 3629 | PL_regsize += 2; |
a0d0e21e LW |
3630 | return(ret); |
3631 | } | |
fe14fcc3 | 3632 | |
c277df42 | 3633 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 3634 | ptr = ret; |
c277df42 | 3635 | FILL_ADVANCE_NODE_ARG(ptr, op, arg); |
3280af22 | 3636 | PL_regcode = ptr; |
fe14fcc3 | 3637 | |
a0d0e21e | 3638 | return(ret); |
fe14fcc3 LW |
3639 | } |
3640 | ||
3641 | /* | |
cd439c50 | 3642 | - reguni - emit (if appropriate) a Unicode character |
a0ed51b3 LW |
3643 | */ |
3644 | STATIC void | |
cea2e8a9 | 3645 | S_reguni(pTHX_ UV uv, char* s, I32* lenp) |
a0ed51b3 | 3646 | { |
c485e607 | 3647 | dTHR; |
a0ed51b3 | 3648 | if (SIZE_ONLY) { |
dfe13c55 | 3649 | U8 tmpbuf[10]; |
a0ed51b3 LW |
3650 | *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf; |
3651 | } | |
3652 | else | |
dfe13c55 | 3653 | *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s; |
a0ed51b3 LW |
3654 | |
3655 | } | |
3656 | ||
3657 | /* | |
a0d0e21e LW |
3658 | - reginsert - insert an operator in front of already-emitted operand |
3659 | * | |
3660 | * Means relocating the operand. | |
3661 | */ | |
76e3520e | 3662 | STATIC void |
cea2e8a9 | 3663 | S_reginsert(pTHX_ U8 op, regnode *opnd) |
a687059c | 3664 | { |
5c0ca799 | 3665 | dTHR; |
c277df42 IZ |
3666 | register regnode *src; |
3667 | register regnode *dst; | |
3668 | register regnode *place; | |
3669 | register int offset = regarglen[(U8)op]; | |
3670 | ||
22c35a8c | 3671 | /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */ |
c277df42 IZ |
3672 | |
3673 | if (SIZE_ONLY) { | |
3280af22 | 3674 | PL_regsize += NODE_STEP_REGNODE + offset; |
a0d0e21e LW |
3675 | return; |
3676 | } | |
a687059c | 3677 | |
3280af22 NIS |
3678 | src = PL_regcode; |
3679 | PL_regcode += NODE_STEP_REGNODE + offset; | |
3680 | dst = PL_regcode; | |
a0d0e21e | 3681 | while (src > opnd) |
c277df42 | 3682 | StructCopy(--src, --dst, regnode); |
a0d0e21e LW |
3683 | |
3684 | place = opnd; /* Op node, where operand used to be. */ | |
c277df42 IZ |
3685 | src = NEXTOPER(place); |
3686 | FILL_ADVANCE_NODE(place, op); | |
3687 | Zero(src, offset, regnode); | |
a687059c LW |
3688 | } |
3689 | ||
3690 | /* | |
c277df42 | 3691 | - regtail - set the next-pointer at the end of a node chain of p to val. |
a0d0e21e | 3692 | */ |
76e3520e | 3693 | STATIC void |
cea2e8a9 | 3694 | S_regtail(pTHX_ regnode *p, regnode *val) |
a687059c | 3695 | { |
5c0ca799 | 3696 | dTHR; |
c277df42 IZ |
3697 | register regnode *scan; |
3698 | register regnode *temp; | |
a0d0e21e LW |
3699 | register I32 offset; |
3700 | ||
c277df42 | 3701 | if (SIZE_ONLY) |
a0d0e21e LW |
3702 | return; |
3703 | ||
3704 | /* Find last node. */ | |
3705 | scan = p; | |
3706 | for (;;) { | |
3707 | temp = regnext(scan); | |
3708 | if (temp == NULL) | |
3709 | break; | |
3710 | scan = temp; | |
3711 | } | |
a687059c | 3712 | |
c277df42 IZ |
3713 | if (reg_off_by_arg[OP(scan)]) { |
3714 | ARG_SET(scan, val - scan); | |
a0ed51b3 LW |
3715 | } |
3716 | else { | |
c277df42 IZ |
3717 | NEXT_OFF(scan) = val - scan; |
3718 | } | |
a687059c LW |
3719 | } |
3720 | ||
3721 | /* | |
a0d0e21e LW |
3722 | - regoptail - regtail on operand of first argument; nop if operandless |
3723 | */ | |
76e3520e | 3724 | STATIC void |
cea2e8a9 | 3725 | S_regoptail(pTHX_ regnode *p, regnode *val) |
a687059c | 3726 | { |
5c0ca799 | 3727 | dTHR; |
a0d0e21e | 3728 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
c277df42 IZ |
3729 | if (p == NULL || SIZE_ONLY) |
3730 | return; | |
22c35a8c | 3731 | if (PL_regkind[(U8)OP(p)] == BRANCH) { |
c277df42 | 3732 | regtail(NEXTOPER(p), val); |
a0ed51b3 | 3733 | } |
22c35a8c | 3734 | else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) { |
c277df42 | 3735 | regtail(NEXTOPER(NEXTOPER(p)), val); |
a0ed51b3 LW |
3736 | } |
3737 | else | |
a0d0e21e | 3738 | return; |
a687059c LW |
3739 | } |
3740 | ||
3741 | /* | |
3742 | - regcurly - a little FSA that accepts {\d+,?\d*} | |
3743 | */ | |
79072805 | 3744 | STATIC I32 |
cea2e8a9 | 3745 | S_regcurly(pTHX_ register char *s) |
a687059c LW |
3746 | { |
3747 | if (*s++ != '{') | |
3748 | return FALSE; | |
f0fcb552 | 3749 | if (!isDIGIT(*s)) |
a687059c | 3750 | return FALSE; |
f0fcb552 | 3751 | while (isDIGIT(*s)) |
a687059c LW |
3752 | s++; |
3753 | if (*s == ',') | |
3754 | s++; | |
f0fcb552 | 3755 | while (isDIGIT(*s)) |
a687059c LW |
3756 | s++; |
3757 | if (*s != '}') | |
3758 | return FALSE; | |
3759 | return TRUE; | |
3760 | } | |
3761 | ||
a687059c | 3762 | |
76e3520e | 3763 | STATIC regnode * |
cea2e8a9 | 3764 | S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l) |
c277df42 | 3765 | { |
35ff7856 | 3766 | #ifdef DEBUGGING |
f248d071 | 3767 | register U8 op = EXACT; /* Arbitrary non-END op. */ |
c277df42 IZ |
3768 | register regnode *next, *onode; |
3769 | ||
3770 | while (op != END && (!last || node < last)) { | |
3771 | /* While that wasn't END last time... */ | |
3772 | ||
3773 | NODE_ALIGN(node); | |
3774 | op = OP(node); | |
3775 | if (op == CLOSE) | |
3776 | l--; | |
3777 | next = regnext(node); | |
3778 | /* Where, what. */ | |
3779 | if (OP(node) == OPTIMIZED) | |
3780 | goto after_print; | |
3781 | regprop(sv, node); | |
b900a521 | 3782 | PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start), |
f1dbda3d | 3783 | (int)(2*l + 1), "", SvPVX(sv)); |
c277df42 IZ |
3784 | if (next == NULL) /* Next ptr. */ |
3785 | PerlIO_printf(Perl_debug_log, "(0)"); | |
3786 | else | |
b900a521 | 3787 | PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start)); |
c277df42 IZ |
3788 | (void)PerlIO_putc(Perl_debug_log, '\n'); |
3789 | after_print: | |
22c35a8c | 3790 | if (PL_regkind[(U8)op] == BRANCHJ) { |
c277df42 IZ |
3791 | register regnode *nnode = (OP(next) == LONGJMP |
3792 | ? regnext(next) | |
3793 | : next); | |
3794 | if (last && nnode > last) | |
3795 | nnode = last; | |
3796 | node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1); | |
a0ed51b3 | 3797 | } |
22c35a8c | 3798 | else if (PL_regkind[(U8)op] == BRANCH) { |
c277df42 | 3799 | node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1); |
a0ed51b3 LW |
3800 | } |
3801 | else if ( op == CURLY) { /* `next' might be very big: optimizer */ | |
c277df42 IZ |
3802 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3803 | NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1); | |
a0ed51b3 | 3804 | } |
22c35a8c | 3805 | else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) { |
c277df42 IZ |
3806 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3807 | next, sv, l + 1); | |
a0ed51b3 LW |
3808 | } |
3809 | else if ( op == PLUS || op == STAR) { | |
c277df42 | 3810 | node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1); |
a0ed51b3 LW |
3811 | } |
3812 | else if (op == ANYOF) { | |
c277df42 | 3813 | node = NEXTOPER(node); |
936ed897 | 3814 | node += ANYOF_SKIP; |
a0ed51b3 | 3815 | } |
22c35a8c | 3816 | else if (PL_regkind[(U8)op] == EXACT) { |
c277df42 | 3817 | /* Literal string, where present. */ |
cd439c50 | 3818 | node += NODE_SZ_STR(node) - 1; |
c277df42 | 3819 | node = NEXTOPER(node); |
a0ed51b3 LW |
3820 | } |
3821 | else { | |
c277df42 IZ |
3822 | node = NEXTOPER(node); |
3823 | node += regarglen[(U8)op]; | |
3824 | } | |
3825 | if (op == CURLYX || op == OPEN) | |
3826 | l++; | |
3827 | else if (op == WHILEM) | |
3828 | l--; | |
3829 | } | |
17c3b450 | 3830 | #endif /* DEBUGGING */ |
c277df42 IZ |
3831 | return node; |
3832 | } | |
3833 | ||
a687059c | 3834 | /* |
fd181c75 | 3835 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c LW |
3836 | */ |
3837 | void | |
864dbfa3 | 3838 | Perl_regdump(pTHX_ regexp *r) |
a687059c | 3839 | { |
35ff7856 | 3840 | #ifdef DEBUGGING |
5c0ca799 | 3841 | dTHR; |
46fc3d4c | 3842 | SV *sv = sv_newmortal(); |
a687059c | 3843 | |
c277df42 | 3844 | (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0); |
a0d0e21e LW |
3845 | |
3846 | /* Header fields of interest. */ | |
c277df42 | 3847 | if (r->anchored_substr) |
7b0972df JH |
3848 | PerlIO_printf(Perl_debug_log, |
3849 | "anchored `%s%.*s%s'%s at %"IVdf" ", | |
3280af22 | 3850 | PL_colors[0], |
7b0972df | 3851 | (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)), |
c277df42 | 3852 | SvPVX(r->anchored_substr), |
3280af22 | 3853 | PL_colors[1], |
c277df42 | 3854 | SvTAIL(r->anchored_substr) ? "$" : "", |
7b0972df | 3855 | (IV)r->anchored_offset); |
c277df42 | 3856 | if (r->float_substr) |
7b0972df JH |
3857 | PerlIO_printf(Perl_debug_log, |
3858 | "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ", | |
3280af22 | 3859 | PL_colors[0], |
7b0972df | 3860 | (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)), |
2c2d71f5 | 3861 | SvPVX(r->float_substr), |
3280af22 | 3862 | PL_colors[1], |
c277df42 | 3863 | SvTAIL(r->float_substr) ? "$" : "", |
7b0972df | 3864 | (IV)r->float_min_offset, (UV)r->float_max_offset); |
c277df42 IZ |
3865 | if (r->check_substr) |
3866 | PerlIO_printf(Perl_debug_log, | |
3867 | r->check_substr == r->float_substr | |
3868 | ? "(checking floating" : "(checking anchored"); | |
3869 | if (r->reganch & ROPT_NOSCAN) | |
3870 | PerlIO_printf(Perl_debug_log, " noscan"); | |
3871 | if (r->reganch & ROPT_CHECK_ALL) | |
3872 | PerlIO_printf(Perl_debug_log, " isall"); | |
3873 | if (r->check_substr) | |
3874 | PerlIO_printf(Perl_debug_log, ") "); | |
3875 | ||
46fc3d4c | 3876 | if (r->regstclass) { |
3877 | regprop(sv, r->regstclass); | |
3878 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv)); | |
3879 | } | |
774d564b | 3880 | if (r->reganch & ROPT_ANCH) { |
3881 | PerlIO_printf(Perl_debug_log, "anchored"); | |
3882 | if (r->reganch & ROPT_ANCH_BOL) | |
3883 | PerlIO_printf(Perl_debug_log, "(BOL)"); | |
c277df42 IZ |
3884 | if (r->reganch & ROPT_ANCH_MBOL) |
3885 | PerlIO_printf(Perl_debug_log, "(MBOL)"); | |
cad2e5aa JH |
3886 | if (r->reganch & ROPT_ANCH_SBOL) |
3887 | PerlIO_printf(Perl_debug_log, "(SBOL)"); | |
774d564b | 3888 | if (r->reganch & ROPT_ANCH_GPOS) |
3889 | PerlIO_printf(Perl_debug_log, "(GPOS)"); | |
3890 | PerlIO_putc(Perl_debug_log, ' '); | |
3891 | } | |
c277df42 IZ |
3892 | if (r->reganch & ROPT_GPOS_SEEN) |
3893 | PerlIO_printf(Perl_debug_log, "GPOS "); | |
a0d0e21e | 3894 | if (r->reganch & ROPT_SKIP) |
760ac839 | 3895 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e | 3896 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 | 3897 | PerlIO_printf(Perl_debug_log, "implicit "); |
760ac839 | 3898 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
ce862d02 IZ |
3899 | if (r->reganch & ROPT_EVAL_SEEN) |
3900 | PerlIO_printf(Perl_debug_log, "with eval "); | |
760ac839 | 3901 | PerlIO_printf(Perl_debug_log, "\n"); |
17c3b450 | 3902 | #endif /* DEBUGGING */ |
a687059c LW |
3903 | } |
3904 | ||
653099ff GS |
3905 | STATIC void |
3906 | S_put_byte(pTHX_ SV *sv, int c) | |
3907 | { | |
3908 | if (c <= ' ' || c == 127 || c == 255) | |
3909 | Perl_sv_catpvf(aTHX_ sv, "\\%o", c); | |
3910 | else if (c == '-' || c == ']' || c == '\\' || c == '^') | |
3911 | Perl_sv_catpvf(aTHX_ sv, "\\%c", c); | |
3912 | else | |
3913 | Perl_sv_catpvf(aTHX_ sv, "%c", c); | |
3914 | } | |
3915 | ||
a687059c | 3916 | /* |
a0d0e21e LW |
3917 | - regprop - printable representation of opcode |
3918 | */ | |
46fc3d4c | 3919 | void |
864dbfa3 | 3920 | Perl_regprop(pTHX_ SV *sv, regnode *o) |
a687059c | 3921 | { |
35ff7856 | 3922 | #ifdef DEBUGGING |
5c0ca799 | 3923 | dTHR; |
9b155405 | 3924 | register int k; |
a0d0e21e | 3925 | |
54dc92de | 3926 | sv_setpvn(sv, "", 0); |
9b155405 | 3927 | if (OP(o) >= reg_num) /* regnode.type is unsigned */ |
a0d0e21e | 3928 | FAIL("corrupted regexp opcode"); |
9b155405 IZ |
3929 | sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */ |
3930 | ||
3931 | k = PL_regkind[(U8)OP(o)]; | |
3932 | ||
3933 | if (k == EXACT) | |
7821416a | 3934 | Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0], |
cd439c50 | 3935 | STR_LEN(o), STRING(o), PL_colors[1]); |
9b155405 IZ |
3936 | else if (k == CURLY) { |
3937 | if (OP(o) == CURLYM || OP(o) == CURLYN) | |
cea2e8a9 GS |
3938 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */ |
3939 | Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o)); | |
a0d0e21e | 3940 | } |
2c2d71f5 JH |
3941 | else if (k == WHILEM && o->flags) /* Ordinal/of */ |
3942 | Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4); | |
9b155405 | 3943 | else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP ) |
894356b3 | 3944 | Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o)); /* Parenth number */ |
9b155405 | 3945 | else if (k == LOGICAL) |
04ebc1ab | 3946 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* 2: embedded, otherwise 1 */ |
653099ff GS |
3947 | else if (k == ANYOF) { |
3948 | int i, rangestart = -1; | |
3949 | const char * const out[] = { /* Should be syncronized with | |
3950 | a table in regcomp.h */ | |
3951 | "\\w", | |
3952 | "\\W", | |
3953 | "\\s", | |
3954 | "\\S", | |
3955 | "\\d", | |
3956 | "\\D", | |
3957 | "[:alnum:]", | |
3958 | "[:^alnum:]", | |
3959 | "[:alpha:]", | |
3960 | "[:^alpha:]", | |
3961 | "[:ascii:]", | |
3962 | "[:^ascii:]", | |
3963 | "[:ctrl:]", | |
3964 | "[:^ctrl:]", | |
3965 | "[:graph:]", | |
3966 | "[:^graph:]", | |
3967 | "[:lower:]", | |
3968 | "[:^lower:]", | |
3969 | "[:print:]", | |
3970 | "[:^print:]", | |
3971 | "[:punct:]", | |
3972 | "[:^punct:]", | |
3973 | "[:upper:]", | |
3974 | "[:!upper:]", | |
3975 | "[:xdigit:]", | |
3976 | "[:^xdigit:]" | |
3977 | }; | |
3978 | ||
3979 | if (o->flags & ANYOF_LOCALE) | |
3980 | sv_catpv(sv, "{loc}"); | |
3981 | if (o->flags & ANYOF_FOLD) | |
3982 | sv_catpv(sv, "{i}"); | |
3983 | Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]); | |
3984 | if (o->flags & ANYOF_INVERT) | |
3985 | sv_catpv(sv, "^"); | |
3986 | for (i = 0; i <= 256; i++) { | |
3987 | if (i < 256 && ANYOF_BITMAP_TEST(o,i)) { | |
3988 | if (rangestart == -1) | |
3989 | rangestart = i; | |
3990 | } else if (rangestart != -1) { | |
3991 | if (i <= rangestart + 3) | |
3992 | for (; rangestart < i; rangestart++) | |
3993 | put_byte(sv, rangestart); | |
3994 | else { | |
3995 | put_byte(sv, rangestart); | |
3996 | sv_catpv(sv, "-"); | |
3997 | put_byte(sv, i - 1); | |
3998 | } | |
3999 | rangestart = -1; | |
4000 | } | |
4001 | } | |
4002 | if (o->flags & ANYOF_CLASS) | |
4003 | for (i = 0; i < sizeof(out)/sizeof(char*); i++) | |
4004 | if (ANYOF_CLASS_TEST(o,i)) | |
4005 | sv_catpv(sv, out[i]); | |
4006 | Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]); | |
4007 | } | |
9b155405 | 4008 | else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH)) |
cea2e8a9 | 4009 | Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags); |
17c3b450 | 4010 | #endif /* DEBUGGING */ |
35ff7856 | 4011 | } |
a687059c | 4012 | |
cad2e5aa JH |
4013 | SV * |
4014 | Perl_re_intuit_string(pTHX_ regexp *prog) | |
4015 | { /* Assume that RE_INTUIT is set */ | |
4016 | DEBUG_r( | |
4017 | { STRLEN n_a; | |
4018 | char *s = SvPV(prog->check_substr,n_a); | |
4019 | ||
4020 | if (!PL_colorset) reginitcolors(); | |
4021 | PerlIO_printf(Perl_debug_log, | |
4022 | "%sUsing REx substr:%s `%s%.60s%s%s'\n", | |
4023 | PL_colors[4],PL_colors[5],PL_colors[0], | |
4024 | s, | |
4025 | PL_colors[1], | |
4026 | (strlen(s) > 60 ? "..." : "")); | |
4027 | } ); | |
4028 | ||
4029 | return prog->check_substr; | |
4030 | } | |
4031 | ||
2b69d0c2 | 4032 | void |
864dbfa3 | 4033 | Perl_pregfree(pTHX_ struct regexp *r) |
a687059c | 4034 | { |
5c0ca799 | 4035 | dTHR; |
adac82c7 | 4036 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
7821416a IZ |
4037 | |
4038 | if (!r || (--r->refcnt > 0)) | |
4039 | return; | |
cad2e5aa JH |
4040 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
4041 | "%sFreeing REx:%s `%s%.60s%s%s'\n", | |
4042 | PL_colors[4],PL_colors[5],PL_colors[0], | |
4043 | r->precomp, | |
4044 | PL_colors[1], | |
4045 | (strlen(r->precomp) > 60 ? "..." : ""))); | |
4046 | ||
c277df42 | 4047 | if (r->precomp) |
a0d0e21e | 4048 | Safefree(r->precomp); |
cf93c79d IZ |
4049 | if (RX_MATCH_COPIED(r)) |
4050 | Safefree(r->subbeg); | |
a193d654 GS |
4051 | if (r->substrs) { |
4052 | if (r->anchored_substr) | |
4053 | SvREFCNT_dec(r->anchored_substr); | |
4054 | if (r->float_substr) | |
4055 | SvREFCNT_dec(r->float_substr); | |
2779dcf1 | 4056 | Safefree(r->substrs); |
a193d654 | 4057 | } |
c277df42 IZ |
4058 | if (r->data) { |
4059 | int n = r->data->count; | |
dfad63ad HS |
4060 | AV* new_comppad = NULL; |
4061 | AV* old_comppad; | |
4062 | SV** old_curpad; | |
4063 | ||
c277df42 IZ |
4064 | while (--n >= 0) { |
4065 | switch (r->data->what[n]) { | |
4066 | case 's': | |
4067 | SvREFCNT_dec((SV*)r->data->data[n]); | |
4068 | break; | |
653099ff GS |
4069 | case 'f': |
4070 | Safefree(r->data->data[n]); | |
4071 | break; | |
dfad63ad HS |
4072 | case 'p': |
4073 | new_comppad = (AV*)r->data->data[n]; | |
4074 | break; | |
c277df42 | 4075 | case 'o': |
dfad63ad | 4076 | if (new_comppad == NULL) |
cea2e8a9 | 4077 | Perl_croak(aTHX_ "panic: pregfree comppad"); |
dfad63ad HS |
4078 | old_comppad = PL_comppad; |
4079 | old_curpad = PL_curpad; | |
4080 | PL_comppad = new_comppad; | |
4081 | PL_curpad = AvARRAY(new_comppad); | |
c277df42 | 4082 | op_free((OP_4tree*)r->data->data[n]); |
dfad63ad HS |
4083 | PL_comppad = old_comppad; |
4084 | PL_curpad = old_curpad; | |
4085 | SvREFCNT_dec((SV*)new_comppad); | |
4086 | new_comppad = NULL; | |
c277df42 IZ |
4087 | break; |
4088 | case 'n': | |
4089 | break; | |
4090 | default: | |
4091 | FAIL2("panic: regfree data code '%c'", r->data->what[n]); | |
4092 | } | |
4093 | } | |
4094 | Safefree(r->data->what); | |
4095 | Safefree(r->data); | |
a0d0e21e LW |
4096 | } |
4097 | Safefree(r->startp); | |
4098 | Safefree(r->endp); | |
4099 | Safefree(r); | |
a687059c | 4100 | } |
c277df42 IZ |
4101 | |
4102 | /* | |
4103 | - regnext - dig the "next" pointer out of a node | |
4104 | * | |
4105 | * [Note, when REGALIGN is defined there are two places in regmatch() | |
4106 | * that bypass this code for speed.] | |
4107 | */ | |
4108 | regnode * | |
864dbfa3 | 4109 | Perl_regnext(pTHX_ register regnode *p) |
c277df42 | 4110 | { |
5c0ca799 | 4111 | dTHR; |
c277df42 IZ |
4112 | register I32 offset; |
4113 | ||
3280af22 | 4114 | if (p == &PL_regdummy) |
c277df42 IZ |
4115 | return(NULL); |
4116 | ||
4117 | offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p)); | |
4118 | if (offset == 0) | |
4119 | return(NULL); | |
4120 | ||
c277df42 | 4121 | return(p+offset); |
c277df42 IZ |
4122 | } |
4123 | ||
01f988be | 4124 | STATIC void |
cea2e8a9 | 4125 | S_re_croak2(pTHX_ const char* pat1,const char* pat2,...) |
c277df42 IZ |
4126 | { |
4127 | va_list args; | |
4128 | STRLEN l1 = strlen(pat1); | |
4129 | STRLEN l2 = strlen(pat2); | |
4130 | char buf[512]; | |
06bf62c7 | 4131 | SV *msv; |
c277df42 IZ |
4132 | char *message; |
4133 | ||
4134 | if (l1 > 510) | |
4135 | l1 = 510; | |
4136 | if (l1 + l2 > 510) | |
4137 | l2 = 510 - l1; | |
4138 | Copy(pat1, buf, l1 , char); | |
4139 | Copy(pat2, buf + l1, l2 , char); | |
3b818b81 GS |
4140 | buf[l1 + l2] = '\n'; |
4141 | buf[l1 + l2 + 1] = '\0'; | |
8736538c AS |
4142 | #ifdef I_STDARG |
4143 | /* ANSI variant takes additional second argument */ | |
c277df42 | 4144 | va_start(args, pat2); |
8736538c AS |
4145 | #else |
4146 | va_start(args); | |
4147 | #endif | |
5a844595 | 4148 | msv = vmess(buf, &args); |
c277df42 | 4149 | va_end(args); |
06bf62c7 | 4150 | message = SvPV(msv,l1); |
c277df42 IZ |
4151 | if (l1 > 512) |
4152 | l1 = 512; | |
4153 | Copy(message, buf, l1 , char); | |
4154 | buf[l1] = '\0'; /* Overwrite \n */ | |
cea2e8a9 | 4155 | Perl_croak(aTHX_ "%s", buf); |
c277df42 | 4156 | } |
a0ed51b3 LW |
4157 | |
4158 | /* XXX Here's a total kludge. But we need to re-enter for swash routines. */ | |
4159 | ||
4160 | void | |
864dbfa3 | 4161 | Perl_save_re_context(pTHX) |
c485e607 NIS |
4162 | { |
4163 | dTHR; | |
a0ed51b3 LW |
4164 | SAVEPPTR(PL_bostr); |
4165 | SAVEPPTR(PL_regprecomp); /* uncompiled string. */ | |
4166 | SAVEI32(PL_regnpar); /* () count. */ | |
4167 | SAVEI32(PL_regsize); /* Code size. */ | |
4168 | SAVEI16(PL_regflags); /* are we folding, multilining? */ | |
4169 | SAVEPPTR(PL_reginput); /* String-input pointer. */ | |
4170 | SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */ | |
4171 | SAVEPPTR(PL_regeol); /* End of input, for $ check. */ | |
7766f137 GS |
4172 | SAVEVPTR(PL_regstartp); /* Pointer to startp array. */ |
4173 | SAVEVPTR(PL_regendp); /* Ditto for endp. */ | |
4174 | SAVEVPTR(PL_reglastparen); /* Similarly for lastparen. */ | |
a0ed51b3 | 4175 | SAVEPPTR(PL_regtill); /* How far we are required to go. */ |
e8347627 | 4176 | SAVEI8(PL_regprev); /* char before regbol, \n if none */ |
7766f137 | 4177 | SAVEVPTR(PL_reg_start_tmp); /* from regexec.c */ |
a0ed51b3 LW |
4178 | PL_reg_start_tmp = 0; |
4179 | SAVEFREEPV(PL_reg_start_tmp); | |
4180 | SAVEI32(PL_reg_start_tmpl); /* from regexec.c */ | |
4181 | PL_reg_start_tmpl = 0; | |
7766f137 | 4182 | SAVEVPTR(PL_regdata); |
a0ed51b3 LW |
4183 | SAVEI32(PL_reg_flags); /* from regexec.c */ |
4184 | SAVEI32(PL_reg_eval_set); /* from regexec.c */ | |
4185 | SAVEI32(PL_regnarrate); /* from regexec.c */ | |
7766f137 | 4186 | SAVEVPTR(PL_regprogram); /* from regexec.c */ |
a0ed51b3 | 4187 | SAVEINT(PL_regindent); /* from regexec.c */ |
7766f137 GS |
4188 | SAVEVPTR(PL_regcc); /* from regexec.c */ |
4189 | SAVEVPTR(PL_curcop); | |
4190 | SAVEVPTR(PL_regcomp_rx); /* from regcomp.c */ | |
a0ed51b3 LW |
4191 | SAVEI32(PL_regseen); /* from regcomp.c */ |
4192 | SAVEI32(PL_regsawback); /* Did we see \1, ...? */ | |
4193 | SAVEI32(PL_regnaughty); /* How bad is this pattern? */ | |
7766f137 | 4194 | SAVEVPTR(PL_regcode); /* Code-emit pointer; ®dummy = don't */ |
a0ed51b3 LW |
4195 | SAVEPPTR(PL_regxend); /* End of input for compile */ |
4196 | SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */ | |
7766f137 GS |
4197 | SAVEVPTR(PL_reg_call_cc); /* from regexec.c */ |
4198 | SAVEVPTR(PL_reg_re); /* from regexec.c */ | |
54b6e2fa IZ |
4199 | SAVEPPTR(PL_reg_ganch); /* from regexec.c */ |
4200 | SAVESPTR(PL_reg_sv); /* from regexec.c */ | |
7766f137 | 4201 | SAVEVPTR(PL_reg_magic); /* from regexec.c */ |
54b6e2fa | 4202 | SAVEI32(PL_reg_oldpos); /* from regexec.c */ |
7766f137 GS |
4203 | SAVEVPTR(PL_reg_oldcurpm); /* from regexec.c */ |
4204 | SAVEVPTR(PL_reg_curpm); /* from regexec.c */ | |
54b6e2fa IZ |
4205 | #ifdef DEBUGGING |
4206 | SAVEPPTR(PL_reg_starttry); /* from regexec.c */ | |
4207 | #endif | |
a0ed51b3 | 4208 | } |
51371543 GS |
4209 | |
4210 | #ifdef PERL_OBJECT | |
51371543 GS |
4211 | #include "XSUB.h" |
4212 | #undef this | |
4213 | #define this pPerl | |
4214 | #endif | |
4215 | ||
4216 | static void | |
4217 | clear_re(pTHXo_ void *r) | |
4218 | { | |
4219 | ReREFCNT_dec((regexp *)r); | |
4220 | } | |
4221 |