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 | { | |
653099ff GS |
301 | if (!(and_with->flags & ANYOF_CLASS) |
302 | && !(cl->flags & ANYOF_CLASS) | |
303 | && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE) | |
304 | && !(and_with->flags & ANYOF_FOLD) | |
305 | && !(cl->flags & ANYOF_FOLD)) { | |
306 | int i; | |
307 | ||
308 | if (and_with->flags & ANYOF_INVERT) | |
309 | for (i = 0; i < ANYOF_BITMAP_SIZE; i++) | |
310 | cl->bitmap[i] &= ~and_with->bitmap[i]; | |
311 | else | |
312 | for (i = 0; i < ANYOF_BITMAP_SIZE; i++) | |
313 | cl->bitmap[i] &= and_with->bitmap[i]; | |
314 | } /* XXXX: logic is complicated otherwise, leave it along for a moment. */ | |
315 | if (!(and_with->flags & ANYOF_EOS)) | |
316 | cl->flags &= ~ANYOF_EOS; | |
317 | } | |
318 | ||
319 | /* 'OR' a given class with another one. Can create false positives */ | |
320 | /* We assume that cl is not inverted */ | |
321 | STATIC void | |
322 | S_cl_or(pTHX_ struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with) | |
323 | { | |
653099ff GS |
324 | if (or_with->flags & ANYOF_INVERT) { |
325 | /* We do not use | |
326 | * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2)) | |
327 | * <= (B1 | !B2) | (CL1 | !CL2) | |
328 | * which is wasteful if CL2 is small, but we ignore CL2: | |
329 | * (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1 | |
330 | * XXXX Can we handle case-fold? Unclear: | |
331 | * (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) = | |
332 | * (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i')) | |
333 | */ | |
334 | if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE) | |
335 | && !(or_with->flags & ANYOF_FOLD) | |
336 | && !(cl->flags & ANYOF_FOLD) ) { | |
337 | int i; | |
338 | ||
339 | for (i = 0; i < ANYOF_BITMAP_SIZE; i++) | |
340 | cl->bitmap[i] |= ~or_with->bitmap[i]; | |
341 | } /* XXXX: logic is complicated otherwise */ | |
342 | else { | |
343 | cl_anything(cl); | |
344 | } | |
345 | } else { | |
346 | /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */ | |
347 | if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE) | |
348 | && (!(or_with->flags & ANYOF_FOLD) | |
349 | || (cl->flags & ANYOF_FOLD)) ) { | |
350 | int i; | |
351 | ||
352 | /* OR char bitmap and class bitmap separately */ | |
353 | for (i = 0; i < ANYOF_BITMAP_SIZE; i++) | |
354 | cl->bitmap[i] |= or_with->bitmap[i]; | |
355 | if (or_with->flags & ANYOF_CLASS) { | |
356 | for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++) | |
357 | cl->classflags[i] |= or_with->classflags[i]; | |
358 | cl->flags |= ANYOF_CLASS; | |
359 | } | |
360 | } | |
361 | else { /* XXXX: logic is complicated, leave it along for a moment. */ | |
362 | cl_anything(cl); | |
363 | } | |
364 | } | |
365 | if (or_with->flags & ANYOF_EOS) | |
366 | cl->flags |= ANYOF_EOS; | |
367 | } | |
368 | ||
369 | /* REx optimizer. Converts nodes into quickier variants "in place". | |
370 | Finds fixed substrings. */ | |
371 | ||
c277df42 IZ |
372 | /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set |
373 | to the position after last scanned or to NULL. */ | |
374 | ||
76e3520e | 375 | STATIC I32 |
cea2e8a9 | 376 | S_study_chunk(pTHX_ regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags) |
c277df42 IZ |
377 | /* scanp: Start here (read-write). */ |
378 | /* deltap: Write maxlen-minlen here. */ | |
379 | /* last: Stop before this one. */ | |
380 | { | |
5c0ca799 | 381 | dTHR; |
c277df42 IZ |
382 | I32 min = 0, pars = 0, code; |
383 | regnode *scan = *scanp, *next; | |
384 | I32 delta = 0; | |
385 | int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF); | |
aca2d497 | 386 | int is_inf_internal = 0; /* The studied chunk is infinite */ |
c277df42 IZ |
387 | I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0; |
388 | scan_data_t data_fake; | |
653099ff | 389 | struct regnode_charclass_class and_with; /* Valid if flags & SCF_DO_STCLASS_OR */ |
c277df42 IZ |
390 | |
391 | while (scan && OP(scan) != END && scan < last) { | |
392 | /* Peephole optimizer: */ | |
393 | ||
22c35a8c | 394 | if (PL_regkind[(U8)OP(scan)] == EXACT) { |
653099ff | 395 | /* Merge several consecutive EXACTish nodes into one. */ |
c277df42 IZ |
396 | regnode *n = regnext(scan); |
397 | U32 stringok = 1; | |
398 | #ifdef DEBUGGING | |
399 | regnode *stop = scan; | |
400 | #endif | |
401 | ||
cd439c50 | 402 | next = scan + NODE_SZ_STR(scan); |
c277df42 IZ |
403 | /* Skip NOTHING, merge EXACT*. */ |
404 | while (n && | |
22c35a8c | 405 | ( PL_regkind[(U8)OP(n)] == NOTHING || |
c277df42 IZ |
406 | (stringok && (OP(n) == OP(scan)))) |
407 | && NEXT_OFF(n) | |
408 | && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) { | |
409 | if (OP(n) == TAIL || n > next) | |
410 | stringok = 0; | |
22c35a8c | 411 | if (PL_regkind[(U8)OP(n)] == NOTHING) { |
c277df42 IZ |
412 | NEXT_OFF(scan) += NEXT_OFF(n); |
413 | next = n + NODE_STEP_REGNODE; | |
414 | #ifdef DEBUGGING | |
415 | if (stringok) | |
416 | stop = n; | |
417 | #endif | |
418 | n = regnext(n); | |
a0ed51b3 LW |
419 | } |
420 | else { | |
cd439c50 | 421 | int oldl = STR_LEN(scan); |
c277df42 IZ |
422 | regnode *nnext = regnext(n); |
423 | ||
cd439c50 | 424 | if (oldl + STR_LEN(n) > U8_MAX) |
c277df42 IZ |
425 | break; |
426 | NEXT_OFF(scan) += NEXT_OFF(n); | |
cd439c50 IZ |
427 | STR_LEN(scan) += STR_LEN(n); |
428 | next = n + NODE_SZ_STR(n); | |
c277df42 | 429 | /* Now we can overwrite *n : */ |
cd439c50 IZ |
430 | Move(STRING(n), STRING(scan) + oldl, |
431 | STR_LEN(n), char); | |
c277df42 IZ |
432 | #ifdef DEBUGGING |
433 | if (stringok) | |
434 | stop = next - 1; | |
435 | #endif | |
436 | n = nnext; | |
437 | } | |
438 | } | |
439 | #ifdef DEBUGGING | |
440 | /* Allow dumping */ | |
cd439c50 | 441 | n = scan + NODE_SZ_STR(scan); |
c277df42 | 442 | while (n <= stop) { |
22c35a8c | 443 | if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) { |
c277df42 IZ |
444 | OP(n) = OPTIMIZED; |
445 | NEXT_OFF(n) = 0; | |
446 | } | |
447 | n++; | |
448 | } | |
653099ff | 449 | #endif |
c277df42 | 450 | } |
653099ff GS |
451 | /* Follow the next-chain of the current node and optimize |
452 | away all the NOTHINGs from it. */ | |
c277df42 | 453 | if (OP(scan) != CURLYX) { |
048cfca1 GS |
454 | int max = (reg_off_by_arg[OP(scan)] |
455 | ? I32_MAX | |
456 | /* I32 may be smaller than U16 on CRAYs! */ | |
457 | : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX)); | |
c277df42 IZ |
458 | int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan)); |
459 | int noff; | |
460 | regnode *n = scan; | |
461 | ||
462 | /* Skip NOTHING and LONGJMP. */ | |
463 | while ((n = regnext(n)) | |
22c35a8c | 464 | && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n))) |
c277df42 IZ |
465 | || ((OP(n) == LONGJMP) && (noff = ARG(n)))) |
466 | && off + noff < max) | |
467 | off += noff; | |
468 | if (reg_off_by_arg[OP(scan)]) | |
469 | ARG(scan) = off; | |
470 | else | |
471 | NEXT_OFF(scan) = off; | |
472 | } | |
653099ff GS |
473 | /* The principal pseudo-switch. Cannot be a switch, since we |
474 | look into several different things. */ | |
c277df42 IZ |
475 | if (OP(scan) == BRANCH || OP(scan) == BRANCHJ |
476 | || OP(scan) == IFTHEN || OP(scan) == SUSPEND) { | |
477 | next = regnext(scan); | |
478 | code = OP(scan); | |
479 | ||
480 | if (OP(next) == code || code == IFTHEN || code == SUSPEND) { | |
481 | I32 max1 = 0, min1 = I32_MAX, num = 0; | |
653099ff | 482 | struct regnode_charclass_class accum; |
c277df42 | 483 | |
653099ff GS |
484 | if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */ |
485 | scan_commit(data); /* Cannot merge strings after this. */ | |
486 | if (flags & SCF_DO_STCLASS) | |
487 | cl_init_zero(&accum); | |
c277df42 | 488 | while (OP(scan) == code) { |
653099ff GS |
489 | I32 deltanext, minnext, f = 0; |
490 | struct regnode_charclass_class this_class; | |
c277df42 IZ |
491 | |
492 | num++; | |
493 | data_fake.flags = 0; | |
2c2d71f5 JH |
494 | if (data) |
495 | data_fake.whilem_c = data->whilem_c; | |
c277df42 IZ |
496 | next = regnext(scan); |
497 | scan = NEXTOPER(scan); | |
498 | if (code != BRANCH) | |
499 | scan = NEXTOPER(scan); | |
653099ff GS |
500 | if (flags & SCF_DO_STCLASS) { |
501 | cl_init(&this_class); | |
502 | data_fake.start_class = &this_class; | |
503 | f = SCF_DO_STCLASS_AND; | |
504 | } | |
505 | /* we suppose the run is continuous, last=next...*/ | |
c277df42 | 506 | minnext = study_chunk(&scan, &deltanext, next, |
653099ff | 507 | &data_fake, f); |
c277df42 IZ |
508 | if (min1 > minnext) |
509 | min1 = minnext; | |
510 | if (max1 < minnext + deltanext) | |
511 | max1 = minnext + deltanext; | |
512 | if (deltanext == I32_MAX) | |
aca2d497 | 513 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
514 | scan = next; |
515 | if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
516 | pars++; | |
405ff068 | 517 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 | 518 | data->flags |= SF_HAS_EVAL; |
2c2d71f5 JH |
519 | if (data) |
520 | data->whilem_c = data_fake.whilem_c; | |
653099ff GS |
521 | if (flags & SCF_DO_STCLASS) |
522 | cl_or(&accum, &this_class); | |
c277df42 IZ |
523 | if (code == SUSPEND) |
524 | break; | |
525 | } | |
526 | if (code == IFTHEN && num < 2) /* Empty ELSE branch */ | |
527 | min1 = 0; | |
528 | if (flags & SCF_DO_SUBSTR) { | |
529 | data->pos_min += min1; | |
530 | data->pos_delta += max1 - min1; | |
531 | if (max1 != min1 || is_inf) | |
532 | data->longest = &(data->longest_float); | |
533 | } | |
534 | min += min1; | |
535 | delta += max1 - min1; | |
653099ff GS |
536 | if (flags & SCF_DO_STCLASS_OR) { |
537 | cl_or(data->start_class, &accum); | |
538 | if (min1) { | |
539 | cl_and(data->start_class, &and_with); | |
540 | flags &= ~SCF_DO_STCLASS; | |
541 | } | |
542 | } | |
543 | else if (flags & SCF_DO_STCLASS_AND) { | |
de0c8cb8 GS |
544 | if (min1) { |
545 | cl_and(data->start_class, &accum); | |
653099ff | 546 | flags &= ~SCF_DO_STCLASS; |
de0c8cb8 GS |
547 | } |
548 | else { | |
549 | /* Switch to OR mode: cache the old value of | |
550 | * data->start_class */ | |
551 | StructCopy(data->start_class, &and_with, | |
552 | struct regnode_charclass_class); | |
553 | flags &= ~SCF_DO_STCLASS_AND; | |
554 | StructCopy(&accum, data->start_class, | |
555 | struct regnode_charclass_class); | |
556 | flags |= SCF_DO_STCLASS_OR; | |
557 | data->start_class->flags |= ANYOF_EOS; | |
558 | } | |
653099ff | 559 | } |
a0ed51b3 LW |
560 | } |
561 | else if (code == BRANCHJ) /* single branch is optimized. */ | |
c277df42 IZ |
562 | scan = NEXTOPER(NEXTOPER(scan)); |
563 | else /* single branch is optimized. */ | |
564 | scan = NEXTOPER(scan); | |
565 | continue; | |
a0ed51b3 LW |
566 | } |
567 | else if (OP(scan) == EXACT) { | |
cd439c50 | 568 | I32 l = STR_LEN(scan); |
a0ed51b3 | 569 | if (UTF) { |
cd439c50 | 570 | unsigned char *s = (unsigned char *)STRING(scan); |
a0ed51b3 LW |
571 | unsigned char *e = s + l; |
572 | I32 newl = 0; | |
573 | while (s < e) { | |
574 | newl++; | |
575 | s += UTF8SKIP(s); | |
576 | } | |
577 | l = newl; | |
578 | } | |
579 | min += l; | |
c277df42 | 580 | if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */ |
c277df42 IZ |
581 | /* The code below prefers earlier match for fixed |
582 | offset, later match for variable offset. */ | |
583 | if (data->last_end == -1) { /* Update the start info. */ | |
584 | data->last_start_min = data->pos_min; | |
585 | data->last_start_max = is_inf | |
586 | ? I32_MAX : data->pos_min + data->pos_delta; | |
587 | } | |
cd439c50 | 588 | sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan)); |
c277df42 IZ |
589 | data->last_end = data->pos_min + l; |
590 | data->pos_min += l; /* As in the first entry. */ | |
591 | data->flags &= ~SF_BEFORE_EOL; | |
592 | } | |
653099ff GS |
593 | if (flags & SCF_DO_STCLASS_AND) { |
594 | /* Check whether it is compatible with what we know already! */ | |
595 | int compat = 1; | |
596 | ||
597 | if (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE)) | |
598 | && !ANYOF_BITMAP_TEST(data->start_class, *STRING(scan)) | |
599 | && (!(data->start_class->flags & ANYOF_FOLD) | |
600 | || !ANYOF_BITMAP_TEST(data->start_class, | |
155aba94 | 601 | PL_fold[*(U8*)STRING(scan)]))) |
653099ff GS |
602 | compat = 0; |
603 | ANYOF_CLASS_ZERO(data->start_class); | |
604 | ANYOF_BITMAP_ZERO(data->start_class); | |
605 | if (compat) | |
606 | ANYOF_BITMAP_SET(data->start_class, *STRING(scan)); | |
607 | data->start_class->flags &= ~ANYOF_EOS; | |
608 | } | |
609 | else if (flags & SCF_DO_STCLASS_OR) { | |
610 | /* false positive possible if the class is case-folded */ | |
611 | ANYOF_BITMAP_SET(data->start_class, *STRING(scan)); | |
612 | data->start_class->flags &= ~ANYOF_EOS; | |
613 | cl_and(data->start_class, &and_with); | |
614 | } | |
615 | flags &= ~SCF_DO_STCLASS; | |
a0ed51b3 | 616 | } |
653099ff | 617 | else if (PL_regkind[(U8)OP(scan)] == EXACT) { /* But OP != EXACT! */ |
cd439c50 | 618 | I32 l = STR_LEN(scan); |
653099ff GS |
619 | |
620 | /* Search for fixed substrings supports EXACT only. */ | |
c277df42 IZ |
621 | if (flags & SCF_DO_SUBSTR) |
622 | scan_commit(data); | |
a0ed51b3 | 623 | if (UTF) { |
cd439c50 | 624 | unsigned char *s = (unsigned char *)STRING(scan); |
a0ed51b3 LW |
625 | unsigned char *e = s + l; |
626 | I32 newl = 0; | |
627 | while (s < e) { | |
628 | newl++; | |
629 | s += UTF8SKIP(s); | |
630 | } | |
631 | l = newl; | |
632 | } | |
633 | min += l; | |
c277df42 | 634 | if (data && (flags & SCF_DO_SUBSTR)) |
a0ed51b3 | 635 | data->pos_min += l; |
653099ff GS |
636 | if (flags & SCF_DO_STCLASS_AND) { |
637 | /* Check whether it is compatible with what we know already! */ | |
638 | int compat = 1; | |
639 | ||
640 | if (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE)) | |
641 | && !ANYOF_BITMAP_TEST(data->start_class, *STRING(scan)) | |
642 | && !ANYOF_BITMAP_TEST(data->start_class, | |
155aba94 | 643 | PL_fold[*(U8*)STRING(scan)])) |
653099ff GS |
644 | compat = 0; |
645 | ANYOF_CLASS_ZERO(data->start_class); | |
646 | ANYOF_BITMAP_ZERO(data->start_class); | |
647 | if (compat) { | |
648 | ANYOF_BITMAP_SET(data->start_class, *STRING(scan)); | |
649 | data->start_class->flags &= ~ANYOF_EOS; | |
650 | data->start_class->flags |= ANYOF_FOLD; | |
651 | if (OP(scan) == EXACTFL) | |
652 | data->start_class->flags |= ANYOF_LOCALE; | |
653 | } | |
654 | } | |
655 | else if (flags & SCF_DO_STCLASS_OR) { | |
656 | if (data->start_class->flags & ANYOF_FOLD) { | |
657 | /* false positive possible if the class is case-folded. | |
658 | Assume that the locale settings are the same... */ | |
659 | ANYOF_BITMAP_SET(data->start_class, *STRING(scan)); | |
660 | data->start_class->flags &= ~ANYOF_EOS; | |
661 | } | |
662 | cl_and(data->start_class, &and_with); | |
663 | } | |
664 | flags &= ~SCF_DO_STCLASS; | |
a0ed51b3 | 665 | } |
4d61ec05 | 666 | else if (strchr((char*)PL_varies,OP(scan))) { |
c277df42 | 667 | I32 mincount, maxcount, minnext, deltanext, pos_before, fl; |
653099ff | 668 | I32 f = flags; |
c277df42 | 669 | regnode *oscan = scan; |
653099ff GS |
670 | struct regnode_charclass_class this_class; |
671 | struct regnode_charclass_class *oclass = NULL; | |
672 | ||
22c35a8c | 673 | switch (PL_regkind[(U8)OP(scan)]) { |
653099ff | 674 | case WHILEM: /* End of (?:...)* . */ |
c277df42 IZ |
675 | scan = NEXTOPER(scan); |
676 | goto finish; | |
677 | case PLUS: | |
653099ff | 678 | if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) { |
c277df42 | 679 | next = NEXTOPER(scan); |
653099ff | 680 | if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) { |
c277df42 IZ |
681 | mincount = 1; |
682 | maxcount = REG_INFTY; | |
683 | next = regnext(scan); | |
684 | scan = NEXTOPER(scan); | |
685 | goto do_curly; | |
686 | } | |
687 | } | |
688 | if (flags & SCF_DO_SUBSTR) | |
689 | data->pos_min++; | |
690 | min++; | |
691 | /* Fall through. */ | |
692 | case STAR: | |
653099ff GS |
693 | if (flags & SCF_DO_STCLASS) { |
694 | mincount = 0; | |
695 | maxcount = REG_INFTY; | |
696 | next = regnext(scan); | |
697 | scan = NEXTOPER(scan); | |
698 | goto do_curly; | |
699 | } | |
aca2d497 | 700 | is_inf = is_inf_internal = 1; |
c277df42 IZ |
701 | scan = regnext(scan); |
702 | if (flags & SCF_DO_SUBSTR) { | |
653099ff | 703 | scan_commit(data); /* Cannot extend fixed substrings */ |
c277df42 IZ |
704 | data->longest = &(data->longest_float); |
705 | } | |
706 | goto optimize_curly_tail; | |
707 | case CURLY: | |
708 | mincount = ARG1(scan); | |
709 | maxcount = ARG2(scan); | |
710 | next = regnext(scan); | |
711 | scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
712 | do_curly: | |
713 | if (flags & SCF_DO_SUBSTR) { | |
653099ff | 714 | if (mincount == 0) scan_commit(data); /* Cannot extend fixed substrings */ |
c277df42 IZ |
715 | pos_before = data->pos_min; |
716 | } | |
717 | if (data) { | |
718 | fl = data->flags; | |
719 | data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL); | |
720 | if (is_inf) | |
721 | data->flags |= SF_IS_INF; | |
722 | } | |
653099ff GS |
723 | if (flags & SCF_DO_STCLASS) { |
724 | cl_init(&this_class); | |
725 | oclass = data->start_class; | |
726 | data->start_class = &this_class; | |
727 | f |= SCF_DO_STCLASS_AND; | |
728 | f &= ~SCF_DO_STCLASS_OR; | |
729 | } | |
730 | ||
c277df42 IZ |
731 | /* This will finish on WHILEM, setting scan, or on NULL: */ |
732 | minnext = study_chunk(&scan, &deltanext, last, data, | |
733 | mincount == 0 | |
653099ff GS |
734 | ? (f & ~SCF_DO_SUBSTR) : f); |
735 | ||
736 | if (flags & SCF_DO_STCLASS) | |
737 | data->start_class = oclass; | |
738 | if (mincount == 0 || minnext == 0) { | |
739 | if (flags & SCF_DO_STCLASS_OR) { | |
740 | cl_or(data->start_class, &this_class); | |
741 | } | |
742 | else if (flags & SCF_DO_STCLASS_AND) { | |
743 | /* Switch to OR mode: cache the old value of | |
744 | * data->start_class */ | |
745 | StructCopy(data->start_class, &and_with, | |
746 | struct regnode_charclass_class); | |
747 | flags &= ~SCF_DO_STCLASS_AND; | |
748 | StructCopy(&this_class, data->start_class, | |
749 | struct regnode_charclass_class); | |
750 | flags |= SCF_DO_STCLASS_OR; | |
751 | data->start_class->flags |= ANYOF_EOS; | |
752 | } | |
753 | } else { /* Non-zero len */ | |
754 | if (flags & SCF_DO_STCLASS_OR) { | |
755 | cl_or(data->start_class, &this_class); | |
756 | cl_and(data->start_class, &and_with); | |
757 | } | |
758 | else if (flags & SCF_DO_STCLASS_AND) | |
759 | cl_and(data->start_class, &this_class); | |
760 | flags &= ~SCF_DO_STCLASS; | |
761 | } | |
c277df42 IZ |
762 | if (!scan) /* It was not CURLYX, but CURLY. */ |
763 | scan = next; | |
e476b1b5 | 764 | if (ckWARN(WARN_REGEXP) && (minnext + deltanext == 0) |
821b33a5 | 765 | && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) |
17feb5d5 | 766 | && maxcount <= REG_INFTY/3) /* Complement check for big count */ |
e476b1b5 | 767 | Perl_warner(aTHX_ WARN_REGEXP, |
c5be433b | 768 | "Strange *+?{} on zero-length expression"); |
c277df42 | 769 | min += minnext * mincount; |
155aba94 GS |
770 | is_inf_internal |= ((maxcount == REG_INFTY |
771 | && (minnext + deltanext) > 0) | |
772 | || deltanext == I32_MAX); | |
aca2d497 | 773 | is_inf |= is_inf_internal; |
c277df42 IZ |
774 | delta += (minnext + deltanext) * maxcount - minnext * mincount; |
775 | ||
776 | /* Try powerful optimization CURLYX => CURLYN. */ | |
c277df42 IZ |
777 | if ( OP(oscan) == CURLYX && data |
778 | && data->flags & SF_IN_PAR | |
779 | && !(data->flags & SF_HAS_EVAL) | |
780 | && !deltanext && minnext == 1 ) { | |
781 | /* Try to optimize to CURLYN. */ | |
782 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; | |
783 | regnode *nxt1 = nxt, *nxt2; | |
784 | ||
785 | /* Skip open. */ | |
786 | nxt = regnext(nxt); | |
4d61ec05 | 787 | if (!strchr((char*)PL_simple,OP(nxt)) |
22c35a8c | 788 | && !(PL_regkind[(U8)OP(nxt)] == EXACT |
cd439c50 | 789 | && STR_LEN(nxt) == 1)) |
c277df42 IZ |
790 | goto nogo; |
791 | nxt2 = nxt; | |
792 | nxt = regnext(nxt); | |
793 | if (OP(nxt) != CLOSE) | |
794 | goto nogo; | |
795 | /* Now we know that nxt2 is the only contents: */ | |
796 | oscan->flags = ARG(nxt); | |
797 | OP(oscan) = CURLYN; | |
798 | OP(nxt1) = NOTHING; /* was OPEN. */ | |
799 | #ifdef DEBUGGING | |
800 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ | |
801 | NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */ | |
802 | NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */ | |
803 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ | |
804 | OP(nxt + 1) = OPTIMIZED; /* was count. */ | |
805 | NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */ | |
806 | #endif | |
807 | } | |
c277df42 IZ |
808 | nogo: |
809 | ||
810 | /* Try optimization CURLYX => CURLYM. */ | |
811 | if ( OP(oscan) == CURLYX && data | |
c277df42 | 812 | && !(data->flags & SF_HAS_PAR) |
c277df42 IZ |
813 | && !(data->flags & SF_HAS_EVAL) |
814 | && !deltanext ) { | |
815 | /* XXXX How to optimize if data == 0? */ | |
816 | /* Optimize to a simpler form. */ | |
817 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */ | |
818 | regnode *nxt2; | |
819 | ||
820 | OP(oscan) = CURLYM; | |
821 | while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/ | |
822 | && (OP(nxt2) != WHILEM)) | |
823 | nxt = nxt2; | |
824 | OP(nxt2) = SUCCEED; /* Whas WHILEM */ | |
c277df42 IZ |
825 | /* Need to optimize away parenths. */ |
826 | if (data->flags & SF_IN_PAR) { | |
827 | /* Set the parenth number. */ | |
828 | regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/ | |
829 | ||
830 | if (OP(nxt) != CLOSE) | |
831 | FAIL("panic opt close"); | |
832 | oscan->flags = ARG(nxt); | |
833 | OP(nxt1) = OPTIMIZED; /* was OPEN. */ | |
834 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ | |
835 | #ifdef DEBUGGING | |
836 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ | |
837 | OP(nxt + 1) = OPTIMIZED; /* was count. */ | |
838 | NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */ | |
839 | NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */ | |
840 | #endif | |
841 | #if 0 | |
842 | while ( nxt1 && (OP(nxt1) != WHILEM)) { | |
843 | regnode *nnxt = regnext(nxt1); | |
844 | ||
845 | if (nnxt == nxt) { | |
846 | if (reg_off_by_arg[OP(nxt1)]) | |
847 | ARG_SET(nxt1, nxt2 - nxt1); | |
848 | else if (nxt2 - nxt1 < U16_MAX) | |
849 | NEXT_OFF(nxt1) = nxt2 - nxt1; | |
850 | else | |
851 | OP(nxt) = NOTHING; /* Cannot beautify */ | |
852 | } | |
853 | nxt1 = nnxt; | |
854 | } | |
855 | #endif | |
856 | /* Optimize again: */ | |
857 | study_chunk(&nxt1, &deltanext, nxt, NULL, 0); | |
a0ed51b3 LW |
858 | } |
859 | else | |
c277df42 | 860 | oscan->flags = 0; |
c277df42 | 861 | } |
2c2d71f5 JH |
862 | else if (OP(oscan) == CURLYX && data && ++data->whilem_c < 16) { |
863 | /* This stays as CURLYX, and can put the count/of pair. */ | |
864 | /* Find WHILEM (as in regexec.c) */ | |
865 | regnode *nxt = oscan + NEXT_OFF(oscan); | |
866 | ||
867 | if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */ | |
868 | nxt += ARG(nxt); | |
869 | PREVOPER(nxt)->flags = data->whilem_c | |
870 | | (PL_reg_whilem_seen << 4); /* On WHILEM */ | |
871 | } | |
c277df42 IZ |
872 | if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) |
873 | pars++; | |
874 | if (flags & SCF_DO_SUBSTR) { | |
875 | SV *last_str = Nullsv; | |
876 | int counted = mincount != 0; | |
877 | ||
878 | if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */ | |
879 | I32 b = pos_before >= data->last_start_min | |
880 | ? pos_before : data->last_start_min; | |
881 | STRLEN l; | |
882 | char *s = SvPV(data->last_found, l); | |
a0ed51b3 LW |
883 | I32 old = b - data->last_start_min; |
884 | ||
885 | if (UTF) | |
886 | old = utf8_hop((U8*)s, old) - (U8*)s; | |
c277df42 | 887 | |
a0ed51b3 | 888 | l -= old; |
c277df42 | 889 | /* Get the added string: */ |
79cb57f6 | 890 | last_str = newSVpvn(s + old, l); |
c277df42 IZ |
891 | if (deltanext == 0 && pos_before == b) { |
892 | /* What was added is a constant string */ | |
893 | if (mincount > 1) { | |
894 | SvGROW(last_str, (mincount * l) + 1); | |
895 | repeatcpy(SvPVX(last_str) + l, | |
896 | SvPVX(last_str), l, mincount - 1); | |
897 | SvCUR(last_str) *= mincount; | |
898 | /* Add additional parts. */ | |
899 | SvCUR_set(data->last_found, | |
900 | SvCUR(data->last_found) - l); | |
901 | sv_catsv(data->last_found, last_str); | |
902 | data->last_end += l * (mincount - 1); | |
903 | } | |
2a8d9689 HS |
904 | } else { |
905 | /* start offset must point into the last copy */ | |
906 | data->last_start_min += minnext * (mincount - 1); | |
4b2cff9a HS |
907 | data->last_start_max += is_inf ? 0 : (maxcount - 1) |
908 | * (minnext + data->pos_delta); | |
c277df42 IZ |
909 | } |
910 | } | |
911 | /* It is counted once already... */ | |
912 | data->pos_min += minnext * (mincount - counted); | |
913 | data->pos_delta += - counted * deltanext + | |
914 | (minnext + deltanext) * maxcount - minnext * mincount; | |
915 | if (mincount != maxcount) { | |
653099ff GS |
916 | /* Cannot extend fixed substrings found inside |
917 | the group. */ | |
c277df42 IZ |
918 | scan_commit(data); |
919 | if (mincount && last_str) { | |
920 | sv_setsv(data->last_found, last_str); | |
921 | data->last_end = data->pos_min; | |
922 | data->last_start_min = | |
a0ed51b3 | 923 | data->pos_min - CHR_SVLEN(last_str); |
c277df42 IZ |
924 | data->last_start_max = is_inf |
925 | ? I32_MAX | |
926 | : data->pos_min + data->pos_delta | |
a0ed51b3 | 927 | - CHR_SVLEN(last_str); |
c277df42 IZ |
928 | } |
929 | data->longest = &(data->longest_float); | |
930 | } | |
aca2d497 | 931 | SvREFCNT_dec(last_str); |
c277df42 | 932 | } |
405ff068 | 933 | if (data && (fl & SF_HAS_EVAL)) |
c277df42 IZ |
934 | data->flags |= SF_HAS_EVAL; |
935 | optimize_curly_tail: | |
c277df42 | 936 | if (OP(oscan) != CURLYX) { |
22c35a8c | 937 | while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING |
c277df42 IZ |
938 | && NEXT_OFF(next)) |
939 | NEXT_OFF(oscan) += NEXT_OFF(next); | |
940 | } | |
c277df42 | 941 | continue; |
653099ff | 942 | default: /* REF and CLUMP only? */ |
c277df42 | 943 | if (flags & SCF_DO_SUBSTR) { |
653099ff | 944 | scan_commit(data); /* Cannot expect anything... */ |
c277df42 IZ |
945 | data->longest = &(data->longest_float); |
946 | } | |
aca2d497 | 947 | is_inf = is_inf_internal = 1; |
653099ff GS |
948 | if (flags & SCF_DO_STCLASS_OR) |
949 | cl_anything(data->start_class); | |
950 | flags &= ~SCF_DO_STCLASS; | |
c277df42 IZ |
951 | break; |
952 | } | |
a0ed51b3 | 953 | } |
4d61ec05 | 954 | else if (strchr((char*)PL_simple,OP(scan)) || PL_regkind[(U8)OP(scan)] == ANYUTF8) { |
653099ff GS |
955 | int value; |
956 | ||
c277df42 IZ |
957 | if (flags & SCF_DO_SUBSTR) { |
958 | scan_commit(data); | |
959 | data->pos_min++; | |
960 | } | |
961 | min++; | |
653099ff GS |
962 | if (flags & SCF_DO_STCLASS) { |
963 | data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */ | |
964 | ||
965 | /* Some of the logic below assumes that switching | |
966 | locale on will only add false positives. */ | |
967 | switch (PL_regkind[(U8)OP(scan)]) { | |
968 | case ANYUTF8: | |
969 | case SANY: | |
970 | case SANYUTF8: | |
971 | case ALNUMUTF8: | |
972 | case ANYOFUTF8: | |
973 | case ALNUMLUTF8: | |
974 | case NALNUMUTF8: | |
975 | case NALNUMLUTF8: | |
976 | case SPACEUTF8: | |
977 | case NSPACEUTF8: | |
978 | case SPACELUTF8: | |
979 | case NSPACELUTF8: | |
980 | case DIGITUTF8: | |
981 | case NDIGITUTF8: | |
982 | default: | |
983 | do_default: | |
984 | /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */ | |
985 | if (flags & SCF_DO_STCLASS_OR) /* Allow everything */ | |
986 | cl_anything(data->start_class); | |
987 | break; | |
988 | case REG_ANY: | |
989 | if (OP(scan) == SANY) | |
990 | goto do_default; | |
991 | if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */ | |
992 | value = (ANYOF_BITMAP_TEST(data->start_class,'\n') | |
993 | || (data->start_class->flags & ANYOF_CLASS)); | |
994 | cl_anything(data->start_class); | |
995 | } | |
996 | if (flags & SCF_DO_STCLASS_AND || !value) | |
997 | ANYOF_BITMAP_CLEAR(data->start_class,'\n'); | |
998 | break; | |
999 | case ANYOF: | |
1000 | if (flags & SCF_DO_STCLASS_AND) | |
1001 | cl_and(data->start_class, | |
1002 | (struct regnode_charclass_class*)scan); | |
1003 | else | |
1004 | cl_or(data->start_class, | |
1005 | (struct regnode_charclass_class*)scan); | |
1006 | break; | |
1007 | case ALNUM: | |
1008 | if (flags & SCF_DO_STCLASS_AND) { | |
1009 | if (!(data->start_class->flags & ANYOF_LOCALE)) { | |
1010 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM); | |
1011 | for (value = 0; value < 256; value++) | |
1012 | if (!isALNUM(value)) | |
1013 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1014 | } | |
1015 | } | |
1016 | else { | |
1017 | if (data->start_class->flags & ANYOF_LOCALE) | |
1018 | ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM); | |
1019 | else { | |
1020 | for (value = 0; value < 256; value++) | |
1021 | if (isALNUM(value)) | |
1022 | ANYOF_BITMAP_SET(data->start_class, value); | |
1023 | } | |
1024 | } | |
1025 | break; | |
1026 | case ALNUML: | |
1027 | if (flags & SCF_DO_STCLASS_AND) { | |
1028 | if (data->start_class->flags & ANYOF_LOCALE) | |
1029 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM); | |
1030 | } | |
1031 | else { | |
1032 | ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM); | |
1033 | data->start_class->flags |= ANYOF_LOCALE; | |
1034 | } | |
1035 | break; | |
1036 | case NALNUM: | |
1037 | if (flags & SCF_DO_STCLASS_AND) { | |
1038 | if (!(data->start_class->flags & ANYOF_LOCALE)) { | |
1039 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM); | |
1040 | for (value = 0; value < 256; value++) | |
1041 | if (isALNUM(value)) | |
1042 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1043 | } | |
1044 | } | |
1045 | else { | |
1046 | if (data->start_class->flags & ANYOF_LOCALE) | |
1047 | ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM); | |
1048 | else { | |
1049 | for (value = 0; value < 256; value++) | |
1050 | if (!isALNUM(value)) | |
1051 | ANYOF_BITMAP_SET(data->start_class, value); | |
1052 | } | |
1053 | } | |
1054 | break; | |
1055 | case NALNUML: | |
1056 | if (flags & SCF_DO_STCLASS_AND) { | |
1057 | if (data->start_class->flags & ANYOF_LOCALE) | |
1058 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM); | |
1059 | } | |
1060 | else { | |
1061 | data->start_class->flags |= ANYOF_LOCALE; | |
1062 | ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM); | |
1063 | } | |
1064 | break; | |
1065 | case SPACE: | |
1066 | if (flags & SCF_DO_STCLASS_AND) { | |
1067 | if (!(data->start_class->flags & ANYOF_LOCALE)) { | |
1068 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE); | |
1069 | for (value = 0; value < 256; value++) | |
1070 | if (!isSPACE(value)) | |
1071 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1072 | } | |
1073 | } | |
1074 | else { | |
1075 | if (data->start_class->flags & ANYOF_LOCALE) | |
1076 | ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE); | |
1077 | else { | |
1078 | for (value = 0; value < 256; value++) | |
1079 | if (isSPACE(value)) | |
1080 | ANYOF_BITMAP_SET(data->start_class, value); | |
1081 | } | |
1082 | } | |
1083 | break; | |
1084 | case SPACEL: | |
1085 | if (flags & SCF_DO_STCLASS_AND) { | |
1086 | if (data->start_class->flags & ANYOF_LOCALE) | |
1087 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE); | |
1088 | } | |
1089 | else { | |
1090 | data->start_class->flags |= ANYOF_LOCALE; | |
1091 | ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE); | |
1092 | } | |
1093 | break; | |
1094 | case NSPACE: | |
1095 | if (flags & SCF_DO_STCLASS_AND) { | |
1096 | if (!(data->start_class->flags & ANYOF_LOCALE)) { | |
1097 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE); | |
1098 | for (value = 0; value < 256; value++) | |
1099 | if (isSPACE(value)) | |
1100 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1101 | } | |
1102 | } | |
1103 | else { | |
1104 | if (data->start_class->flags & ANYOF_LOCALE) | |
1105 | ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE); | |
1106 | else { | |
1107 | for (value = 0; value < 256; value++) | |
1108 | if (!isSPACE(value)) | |
1109 | ANYOF_BITMAP_SET(data->start_class, value); | |
1110 | } | |
1111 | } | |
1112 | break; | |
1113 | case NSPACEL: | |
1114 | if (flags & SCF_DO_STCLASS_AND) { | |
1115 | if (data->start_class->flags & ANYOF_LOCALE) { | |
1116 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE); | |
1117 | for (value = 0; value < 256; value++) | |
1118 | if (!isSPACE(value)) | |
1119 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1120 | } | |
1121 | } | |
1122 | else { | |
1123 | data->start_class->flags |= ANYOF_LOCALE; | |
1124 | ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE); | |
1125 | } | |
1126 | break; | |
1127 | case DIGIT: | |
1128 | if (flags & SCF_DO_STCLASS_AND) { | |
1129 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT); | |
1130 | for (value = 0; value < 256; value++) | |
1131 | if (!isDIGIT(value)) | |
1132 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1133 | } | |
1134 | else { | |
1135 | if (data->start_class->flags & ANYOF_LOCALE) | |
1136 | ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT); | |
1137 | else { | |
1138 | for (value = 0; value < 256; value++) | |
1139 | if (isDIGIT(value)) | |
1140 | ANYOF_BITMAP_SET(data->start_class, value); | |
1141 | } | |
1142 | } | |
1143 | break; | |
1144 | case NDIGIT: | |
1145 | if (flags & SCF_DO_STCLASS_AND) { | |
1146 | ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT); | |
1147 | for (value = 0; value < 256; value++) | |
1148 | if (isDIGIT(value)) | |
1149 | ANYOF_BITMAP_CLEAR(data->start_class, value); | |
1150 | } | |
1151 | else { | |
1152 | if (data->start_class->flags & ANYOF_LOCALE) | |
1153 | ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT); | |
1154 | else { | |
1155 | for (value = 0; value < 256; value++) | |
1156 | if (!isDIGIT(value)) | |
1157 | ANYOF_BITMAP_SET(data->start_class, value); | |
1158 | } | |
1159 | } | |
1160 | break; | |
1161 | } | |
1162 | if (flags & SCF_DO_STCLASS_OR) | |
1163 | cl_and(data->start_class, &and_with); | |
1164 | flags &= ~SCF_DO_STCLASS; | |
1165 | } | |
a0ed51b3 | 1166 | } |
22c35a8c | 1167 | else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) { |
c277df42 IZ |
1168 | data->flags |= (OP(scan) == MEOL |
1169 | ? SF_BEFORE_MEOL | |
1170 | : SF_BEFORE_SEOL); | |
a0ed51b3 | 1171 | } |
653099ff GS |
1172 | else if ( PL_regkind[(U8)OP(scan)] == BRANCHJ |
1173 | /* Lookbehind, or need to calculate parens/evals/stclass: */ | |
1174 | && (scan->flags || data || (flags & SCF_DO_STCLASS)) | |
c277df42 | 1175 | && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) { |
653099ff | 1176 | /* Lookahead/lookbehind */ |
c277df42 IZ |
1177 | I32 deltanext, minnext; |
1178 | regnode *nscan; | |
653099ff GS |
1179 | struct regnode_charclass_class intrnl; |
1180 | int f = 0; | |
c277df42 IZ |
1181 | |
1182 | data_fake.flags = 0; | |
2c2d71f5 JH |
1183 | if (data) |
1184 | data_fake.whilem_c = data->whilem_c; | |
653099ff GS |
1185 | if ( flags & SCF_DO_STCLASS && !scan->flags |
1186 | && OP(scan) == IFMATCH ) { /* Lookahead */ | |
1187 | cl_init(&intrnl); | |
1188 | data_fake.start_class = &intrnl; | |
1189 | f = SCF_DO_STCLASS_AND; | |
1190 | } | |
c277df42 IZ |
1191 | next = regnext(scan); |
1192 | nscan = NEXTOPER(NEXTOPER(scan)); | |
653099ff | 1193 | minnext = study_chunk(&nscan, &deltanext, last, &data_fake, f); |
c277df42 IZ |
1194 | if (scan->flags) { |
1195 | if (deltanext) { | |
1196 | FAIL("variable length lookbehind not implemented"); | |
a0ed51b3 LW |
1197 | } |
1198 | else if (minnext > U8_MAX) { | |
57def98f | 1199 | FAIL2("lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX); |
c277df42 IZ |
1200 | } |
1201 | scan->flags = minnext; | |
1202 | } | |
1203 | if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) | |
1204 | pars++; | |
405ff068 | 1205 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 | 1206 | data->flags |= SF_HAS_EVAL; |
2c2d71f5 JH |
1207 | if (data) |
1208 | data->whilem_c = data_fake.whilem_c; | |
653099ff GS |
1209 | if (f) { |
1210 | int was = (data->start_class->flags & ANYOF_EOS); | |
1211 | ||
1212 | cl_and(data->start_class, &intrnl); | |
1213 | if (was) | |
1214 | data->start_class->flags |= ANYOF_EOS; | |
1215 | } | |
a0ed51b3 LW |
1216 | } |
1217 | else if (OP(scan) == OPEN) { | |
c277df42 | 1218 | pars++; |
a0ed51b3 LW |
1219 | } |
1220 | else if (OP(scan) == CLOSE && ARG(scan) == is_par) { | |
c277df42 IZ |
1221 | next = regnext(scan); |
1222 | ||
1223 | if ( next && (OP(next) != WHILEM) && next < last) | |
c277df42 | 1224 | is_par = 0; /* Disable optimization */ |
a0ed51b3 LW |
1225 | } |
1226 | else if (OP(scan) == EVAL) { | |
c277df42 IZ |
1227 | if (data) |
1228 | data->flags |= SF_HAS_EVAL; | |
1229 | } | |
96776eda | 1230 | else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded follows */ |
0f5d15d6 IZ |
1231 | if (flags & SCF_DO_SUBSTR) { |
1232 | scan_commit(data); | |
1233 | data->longest = &(data->longest_float); | |
1234 | } | |
1235 | is_inf = is_inf_internal = 1; | |
653099ff GS |
1236 | if (flags & SCF_DO_STCLASS_OR) /* Allow everything */ |
1237 | cl_anything(data->start_class); | |
96776eda | 1238 | flags &= ~SCF_DO_STCLASS; |
0f5d15d6 | 1239 | } |
c277df42 IZ |
1240 | /* Else: zero-length, ignore. */ |
1241 | scan = regnext(scan); | |
1242 | } | |
1243 | ||
1244 | finish: | |
1245 | *scanp = scan; | |
aca2d497 | 1246 | *deltap = is_inf_internal ? I32_MAX : delta; |
c277df42 IZ |
1247 | if (flags & SCF_DO_SUBSTR && is_inf) |
1248 | data->pos_delta = I32_MAX - data->pos_min; | |
1249 | if (is_par > U8_MAX) | |
1250 | is_par = 0; | |
1251 | if (is_par && pars==1 && data) { | |
1252 | data->flags |= SF_IN_PAR; | |
1253 | data->flags &= ~SF_HAS_PAR; | |
a0ed51b3 LW |
1254 | } |
1255 | else if (pars && data) { | |
c277df42 IZ |
1256 | data->flags |= SF_HAS_PAR; |
1257 | data->flags &= ~SF_IN_PAR; | |
1258 | } | |
653099ff GS |
1259 | if (flags & SCF_DO_STCLASS_OR) |
1260 | cl_and(data->start_class, &and_with); | |
c277df42 IZ |
1261 | return min; |
1262 | } | |
1263 | ||
76e3520e | 1264 | STATIC I32 |
cea2e8a9 | 1265 | S_add_data(pTHX_ I32 n, char *s) |
c277df42 | 1266 | { |
5c0ca799 | 1267 | dTHR; |
3280af22 NIS |
1268 | if (PL_regcomp_rx->data) { |
1269 | Renewc(PL_regcomp_rx->data, | |
1270 | sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1), | |
c277df42 | 1271 | char, struct reg_data); |
3280af22 NIS |
1272 | Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8); |
1273 | PL_regcomp_rx->data->count += n; | |
a0ed51b3 LW |
1274 | } |
1275 | else { | |
3280af22 | 1276 | Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1), |
c277df42 | 1277 | char, struct reg_data); |
3280af22 NIS |
1278 | New(1208, PL_regcomp_rx->data->what, n, U8); |
1279 | PL_regcomp_rx->data->count = n; | |
c277df42 | 1280 | } |
3280af22 NIS |
1281 | Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8); |
1282 | return PL_regcomp_rx->data->count - n; | |
c277df42 IZ |
1283 | } |
1284 | ||
d88dccdf | 1285 | void |
864dbfa3 | 1286 | Perl_reginitcolors(pTHX) |
d88dccdf IZ |
1287 | { |
1288 | dTHR; | |
1289 | int i = 0; | |
1290 | char *s = PerlEnv_getenv("PERL_RE_COLORS"); | |
1291 | ||
1292 | if (s) { | |
1293 | PL_colors[0] = s = savepv(s); | |
1294 | while (++i < 6) { | |
1295 | s = strchr(s, '\t'); | |
1296 | if (s) { | |
1297 | *s = '\0'; | |
1298 | PL_colors[i] = ++s; | |
1299 | } | |
1300 | else | |
c712d376 | 1301 | PL_colors[i] = s = ""; |
d88dccdf IZ |
1302 | } |
1303 | } else { | |
1304 | while (i < 6) | |
1305 | PL_colors[i++] = ""; | |
1306 | } | |
1307 | PL_colorset = 1; | |
1308 | } | |
1309 | ||
a687059c | 1310 | /* |
e50aee73 | 1311 | - pregcomp - compile a regular expression into internal code |
a687059c LW |
1312 | * |
1313 | * We can't allocate space until we know how big the compiled form will be, | |
1314 | * but we can't compile it (and thus know how big it is) until we've got a | |
1315 | * place to put the code. So we cheat: we compile it twice, once with code | |
1316 | * generation turned off and size counting turned on, and once "for real". | |
1317 | * This also means that we don't allocate space until we are sure that the | |
1318 | * thing really will compile successfully, and we never have to move the | |
1319 | * code and thus invalidate pointers into it. (Note that it has to be in | |
1320 | * one piece because free() must be able to free it all.) [NB: not true in perl] | |
1321 | * | |
1322 | * Beware that the optimization-preparation code in here knows about some | |
1323 | * of the structure of the compiled regexp. [I'll say.] | |
1324 | */ | |
1325 | regexp * | |
864dbfa3 | 1326 | Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm) |
a687059c | 1327 | { |
5c0ca799 | 1328 | dTHR; |
a0d0e21e | 1329 | register regexp *r; |
c277df42 | 1330 | regnode *scan; |
c277df42 | 1331 | regnode *first; |
a0d0e21e | 1332 | I32 flags; |
a0d0e21e LW |
1333 | I32 minlen = 0; |
1334 | I32 sawplus = 0; | |
1335 | I32 sawopen = 0; | |
2c2d71f5 | 1336 | scan_data_t data; |
a0d0e21e LW |
1337 | |
1338 | if (exp == NULL) | |
c277df42 | 1339 | FAIL("NULL regexp argument"); |
a0d0e21e | 1340 | |
393fec97 | 1341 | if (pm->op_pmdynflags & PMdf_UTF8) { |
a0ed51b3 | 1342 | PL_reg_flags |= RF_utf8; |
7e2040f0 | 1343 | } |
a0ed51b3 LW |
1344 | else |
1345 | PL_reg_flags = 0; | |
1346 | ||
3280af22 | 1347 | PL_regprecomp = savepvn(exp, xend - exp); |
35ef4773 | 1348 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
2c2d71f5 | 1349 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n", |
d88dccdf | 1350 | PL_colors[4],PL_colors[5],PL_colors[0], |
b900a521 | 1351 | (int)(xend - exp), PL_regprecomp, PL_colors[1])); |
3280af22 NIS |
1352 | PL_regflags = pm->op_pmflags; |
1353 | PL_regsawback = 0; | |
bbce6d69 | 1354 | |
3280af22 NIS |
1355 | PL_regseen = 0; |
1356 | PL_seen_zerolen = *exp == '^' ? -1 : 0; | |
1357 | PL_seen_evals = 0; | |
1358 | PL_extralen = 0; | |
c277df42 | 1359 | |
bbce6d69 | 1360 | /* First pass: determine size, legality. */ |
3280af22 NIS |
1361 | PL_regcomp_parse = exp; |
1362 | PL_regxend = xend; | |
1363 | PL_regnaughty = 0; | |
1364 | PL_regnpar = 1; | |
1365 | PL_regsize = 0L; | |
1366 | PL_regcode = &PL_regdummy; | |
2c2d71f5 | 1367 | PL_reg_whilem_seen = 0; |
85ddcde9 JH |
1368 | #if 0 /* REGC() is (currently) a NOP at the first pass. |
1369 | * Clever compilers notice this and complain. --jhi */ | |
cd439c50 | 1370 | REGC((U8)REG_MAGIC, (char*)PL_regcode); |
85ddcde9 | 1371 | #endif |
a0d0e21e | 1372 | if (reg(0, &flags) == NULL) { |
3280af22 NIS |
1373 | Safefree(PL_regprecomp); |
1374 | PL_regprecomp = Nullch; | |
a0d0e21e LW |
1375 | return(NULL); |
1376 | } | |
faccc32b | 1377 | DEBUG_r(PerlIO_printf(Perl_debug_log, "size %"IVdf" ", (IV)PL_regsize)); |
c277df42 | 1378 | |
c277df42 IZ |
1379 | /* Small enough for pointer-storage convention? |
1380 | If extralen==0, this means that we will not need long jumps. */ | |
3280af22 NIS |
1381 | if (PL_regsize >= 0x10000L && PL_extralen) |
1382 | PL_regsize += PL_extralen; | |
c277df42 | 1383 | else |
3280af22 | 1384 | PL_extralen = 0; |
2c2d71f5 JH |
1385 | if (PL_reg_whilem_seen > 15) |
1386 | PL_reg_whilem_seen = 15; | |
a0d0e21e | 1387 | |
bbce6d69 | 1388 | /* Allocate space and initialize. */ |
3280af22 | 1389 | Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), |
c277df42 | 1390 | char, regexp); |
a0d0e21e LW |
1391 | if (r == NULL) |
1392 | FAIL("regexp out of space"); | |
0f79a09d GS |
1393 | #ifdef DEBUGGING |
1394 | /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */ | |
1395 | Zero(r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), char); | |
1396 | #endif | |
c277df42 | 1397 | r->refcnt = 1; |
bbce6d69 | 1398 | r->prelen = xend - exp; |
3280af22 | 1399 | r->precomp = PL_regprecomp; |
cf93c79d IZ |
1400 | r->subbeg = NULL; |
1401 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; | |
4327152a IZ |
1402 | r->nparens = PL_regnpar - 1; /* set early to validate backrefs */ |
1403 | ||
1404 | r->substrs = 0; /* Useful during FAIL. */ | |
1405 | r->startp = 0; /* Useful during FAIL. */ | |
1406 | r->endp = 0; /* Useful during FAIL. */ | |
1407 | ||
3280af22 | 1408 | PL_regcomp_rx = r; |
bbce6d69 | 1409 | |
1410 | /* Second pass: emit code. */ | |
3280af22 NIS |
1411 | PL_regcomp_parse = exp; |
1412 | PL_regxend = xend; | |
1413 | PL_regnaughty = 0; | |
1414 | PL_regnpar = 1; | |
1415 | PL_regcode = r->program; | |
2cd61cdb | 1416 | /* Store the count of eval-groups for security checks: */ |
3280af22 | 1417 | PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals); |
cd439c50 | 1418 | REGC((U8)REG_MAGIC, (char*) PL_regcode++); |
c277df42 | 1419 | r->data = 0; |
a0d0e21e LW |
1420 | if (reg(0, &flags) == NULL) |
1421 | return(NULL); | |
1422 | ||
1423 | /* Dig out information for optimizations. */ | |
cf93c79d | 1424 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */ |
3280af22 | 1425 | pm->op_pmflags = PL_regflags; |
a0ed51b3 LW |
1426 | if (UTF) |
1427 | r->reganch |= ROPT_UTF8; | |
c277df42 | 1428 | r->regstclass = NULL; |
a0ed51b3 LW |
1429 | if (PL_regnaughty >= 10) /* Probably an expensive pattern. */ |
1430 | r->reganch |= ROPT_NAUGHTY; | |
c277df42 | 1431 | scan = r->program + 1; /* First BRANCH. */ |
2779dcf1 IZ |
1432 | |
1433 | /* XXXX To minimize changes to RE engine we always allocate | |
1434 | 3-units-long substrs field. */ | |
1435 | Newz(1004, r->substrs, 1, struct reg_substr_data); | |
1436 | ||
2c2d71f5 | 1437 | StructCopy(&zero_scan_data, &data, scan_data_t); |
653099ff | 1438 | /* XXXX Should not we check for something else? Usually it is OPEN1... */ |
c277df42 | 1439 | if (OP(scan) != BRANCH) { /* Only one top-level choice. */ |
c277df42 | 1440 | I32 fake; |
c5254dd6 | 1441 | STRLEN longest_float_length, longest_fixed_length; |
653099ff GS |
1442 | struct regnode_charclass_class ch_class; |
1443 | int stclass_flag; | |
a0d0e21e LW |
1444 | |
1445 | first = scan; | |
c277df42 | 1446 | /* Skip introductions and multiplicators >= 1. */ |
a0d0e21e | 1447 | while ((OP(first) == OPEN && (sawopen = 1)) || |
653099ff | 1448 | /* An OR of *one* alternative - should not happen now. */ |
a0d0e21e LW |
1449 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || |
1450 | (OP(first) == PLUS) || | |
1451 | (OP(first) == MINMOD) || | |
653099ff | 1452 | /* An {n,m} with n>0 */ |
22c35a8c | 1453 | (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
a0d0e21e LW |
1454 | if (OP(first) == PLUS) |
1455 | sawplus = 1; | |
1456 | else | |
1457 | first += regarglen[(U8)OP(first)]; | |
1458 | first = NEXTOPER(first); | |
a687059c LW |
1459 | } |
1460 | ||
a0d0e21e LW |
1461 | /* Starting-point info. */ |
1462 | again: | |
653099ff | 1463 | if (PL_regkind[(U8)OP(first)] == EXACT) { |
b3c9acc1 IZ |
1464 | if (OP(first) == EXACT); /* Empty, get anchored substr later. */ |
1465 | else if ((OP(first) == EXACTF || OP(first) == EXACTFL) | |
1466 | && !UTF) | |
1467 | r->regstclass = first; | |
1468 | } | |
653099ff | 1469 | else if (strchr((char*)PL_simple,OP(first))) |
a0d0e21e | 1470 | r->regstclass = first; |
22c35a8c GS |
1471 | else if (PL_regkind[(U8)OP(first)] == BOUND || |
1472 | PL_regkind[(U8)OP(first)] == NBOUND) | |
a0d0e21e | 1473 | r->regstclass = first; |
22c35a8c | 1474 | else if (PL_regkind[(U8)OP(first)] == BOL) { |
cad2e5aa JH |
1475 | r->reganch |= (OP(first) == MBOL |
1476 | ? ROPT_ANCH_MBOL | |
1477 | : (OP(first) == SBOL | |
1478 | ? ROPT_ANCH_SBOL | |
1479 | : ROPT_ANCH_BOL)); | |
a0d0e21e | 1480 | first = NEXTOPER(first); |
774d564b | 1481 | goto again; |
1482 | } | |
1483 | else if (OP(first) == GPOS) { | |
1484 | r->reganch |= ROPT_ANCH_GPOS; | |
1485 | first = NEXTOPER(first); | |
1486 | goto again; | |
a0d0e21e LW |
1487 | } |
1488 | else if ((OP(first) == STAR && | |
22c35a8c | 1489 | PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) && |
a0d0e21e LW |
1490 | !(r->reganch & ROPT_ANCH) ) |
1491 | { | |
1492 | /* turn .* into ^.* with an implied $*=1 */ | |
cad2e5aa JH |
1493 | int type = OP(NEXTOPER(first)); |
1494 | ||
1495 | if (type == REG_ANY || type == ANYUTF8) | |
1496 | type = ROPT_ANCH_MBOL; | |
1497 | else | |
1498 | type = ROPT_ANCH_SBOL; | |
1499 | ||
1500 | r->reganch |= type | ROPT_IMPLICIT; | |
a0d0e21e | 1501 | first = NEXTOPER(first); |
774d564b | 1502 | goto again; |
a0d0e21e | 1503 | } |
cad2e5aa JH |
1504 | if (sawplus && (!sawopen || !PL_regsawback) |
1505 | && !(PL_regseen & REG_SEEN_EVAL)) /* May examine pos and $& */ | |
1506 | /* x+ must match at the 1st pos of run of x's */ | |
1507 | r->reganch |= ROPT_SKIP; | |
a0d0e21e | 1508 | |
c277df42 | 1509 | /* Scan is after the zeroth branch, first is atomic matcher. */ |
b900a521 JH |
1510 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n", |
1511 | (IV)(first - scan + 1))); | |
a0d0e21e LW |
1512 | /* |
1513 | * If there's something expensive in the r.e., find the | |
1514 | * longest literal string that must appear and make it the | |
1515 | * regmust. Resolve ties in favor of later strings, since | |
1516 | * the regstart check works with the beginning of the r.e. | |
1517 | * and avoiding duplication strengthens checking. Not a | |
1518 | * strong reason, but sufficient in the absence of others. | |
1519 | * [Now we resolve ties in favor of the earlier string if | |
c277df42 | 1520 | * it happens that c_offset_min has been invalidated, since the |
a0d0e21e LW |
1521 | * earlier string may buy us something the later one won't.] |
1522 | */ | |
a0d0e21e | 1523 | minlen = 0; |
a687059c | 1524 | |
79cb57f6 GS |
1525 | data.longest_fixed = newSVpvn("",0); |
1526 | data.longest_float = newSVpvn("",0); | |
1527 | data.last_found = newSVpvn("",0); | |
c277df42 IZ |
1528 | data.longest = &(data.longest_fixed); |
1529 | first = scan; | |
653099ff GS |
1530 | if (!r->regstclass) { |
1531 | cl_init(&ch_class); | |
1532 | data.start_class = &ch_class; | |
1533 | stclass_flag = SCF_DO_STCLASS_AND; | |
1534 | } else /* XXXX Check for BOUND? */ | |
1535 | stclass_flag = 0; | |
1536 | ||
3280af22 | 1537 | minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */ |
653099ff | 1538 | &data, SCF_DO_SUBSTR | stclass_flag); |
3280af22 | 1539 | if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed) |
c277df42 | 1540 | && data.last_start_min == 0 && data.last_end > 0 |
3280af22 NIS |
1541 | && !PL_seen_zerolen |
1542 | && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS))) | |
c277df42 IZ |
1543 | r->reganch |= ROPT_CHECK_ALL; |
1544 | scan_commit(&data); | |
1545 | SvREFCNT_dec(data.last_found); | |
1546 | ||
a0ed51b3 | 1547 | longest_float_length = CHR_SVLEN(data.longest_float); |
c5254dd6 | 1548 | if (longest_float_length |
c277df42 IZ |
1549 | || (data.flags & SF_FL_BEFORE_EOL |
1550 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
3280af22 | 1551 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1552 | int t; |
1553 | ||
a0ed51b3 | 1554 | if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */ |
aca2d497 IZ |
1555 | && data.offset_fixed == data.offset_float_min |
1556 | && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)) | |
1557 | goto remove_float; /* As in (a)+. */ | |
1558 | ||
c277df42 IZ |
1559 | r->float_substr = data.longest_float; |
1560 | r->float_min_offset = data.offset_float_min; | |
1561 | r->float_max_offset = data.offset_float_max; | |
cf93c79d IZ |
1562 | t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1563 | && (!(data.flags & SF_FL_BEFORE_MEOL) | |
1564 | || (PL_regflags & PMf_MULTILINE))); | |
1565 | fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1566 | } |
1567 | else { | |
aca2d497 | 1568 | remove_float: |
c277df42 IZ |
1569 | r->float_substr = Nullsv; |
1570 | SvREFCNT_dec(data.longest_float); | |
c5254dd6 | 1571 | longest_float_length = 0; |
a0d0e21e | 1572 | } |
c277df42 | 1573 | |
a0ed51b3 | 1574 | longest_fixed_length = CHR_SVLEN(data.longest_fixed); |
c5254dd6 | 1575 | if (longest_fixed_length |
c277df42 IZ |
1576 | || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
1577 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
3280af22 | 1578 | || (PL_regflags & PMf_MULTILINE)))) { |
cf93c79d IZ |
1579 | int t; |
1580 | ||
c277df42 IZ |
1581 | r->anchored_substr = data.longest_fixed; |
1582 | r->anchored_offset = data.offset_fixed; | |
cf93c79d IZ |
1583 | t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */ |
1584 | && (!(data.flags & SF_FIX_BEFORE_MEOL) | |
1585 | || (PL_regflags & PMf_MULTILINE))); | |
1586 | fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0); | |
a0ed51b3 LW |
1587 | } |
1588 | else { | |
c277df42 IZ |
1589 | r->anchored_substr = Nullsv; |
1590 | SvREFCNT_dec(data.longest_fixed); | |
c5254dd6 | 1591 | longest_fixed_length = 0; |
a0d0e21e | 1592 | } |
653099ff GS |
1593 | if (r->regstclass |
1594 | && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == ANYUTF8 | |
1595 | || OP(r->regstclass) == SANYUTF8 || OP(r->regstclass) == SANY)) | |
1596 | r->regstclass = NULL; | |
1597 | if ((!r->anchored_substr || r->anchored_offset) && stclass_flag | |
1598 | && !(data.start_class->flags & ANYOF_EOS) | |
1599 | && !cl_is_anything(data.start_class)) { | |
1600 | SV *sv; | |
1601 | I32 n = add_data(1, "f"); | |
1602 | ||
1603 | New(1006, PL_regcomp_rx->data->data[n], 1, | |
1604 | struct regnode_charclass_class); | |
1605 | StructCopy(data.start_class, | |
1606 | (struct regnode_charclass_class*)PL_regcomp_rx->data->data[n], | |
1607 | struct regnode_charclass_class); | |
1608 | r->regstclass = (regnode*)PL_regcomp_rx->data->data[n]; | |
1609 | r->reganch &= ~ROPT_SKIP; /* Used in find_byclass(). */ | |
1610 | DEBUG_r((sv = sv_newmortal(), | |
1611 | regprop(sv, (regnode*)data.start_class), | |
894356b3 | 1612 | PerlIO_printf(Perl_debug_log, "synthetic stclass `%s'.\n", |
653099ff GS |
1613 | SvPVX(sv)))); |
1614 | } | |
c277df42 IZ |
1615 | |
1616 | /* A temporary algorithm prefers floated substr to fixed one to dig more info. */ | |
c5254dd6 | 1617 | if (longest_fixed_length > longest_float_length) { |
c277df42 IZ |
1618 | r->check_substr = r->anchored_substr; |
1619 | r->check_offset_min = r->check_offset_max = r->anchored_offset; | |
1620 | if (r->reganch & ROPT_ANCH_SINGLE) | |
1621 | r->reganch |= ROPT_NOSCAN; | |
a0ed51b3 LW |
1622 | } |
1623 | else { | |
c277df42 IZ |
1624 | r->check_substr = r->float_substr; |
1625 | r->check_offset_min = data.offset_float_min; | |
1626 | r->check_offset_max = data.offset_float_max; | |
a0d0e21e | 1627 | } |
30382c73 IZ |
1628 | /* XXXX Currently intuiting is not compatible with ANCH_GPOS. |
1629 | This should be changed ASAP! */ | |
1630 | if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) { | |
cad2e5aa JH |
1631 | r->reganch |= RE_USE_INTUIT; |
1632 | if (SvTAIL(r->check_substr)) | |
1633 | r->reganch |= RE_INTUIT_TAIL; | |
1634 | } | |
a0ed51b3 LW |
1635 | } |
1636 | else { | |
c277df42 IZ |
1637 | /* Several toplevels. Best we can is to set minlen. */ |
1638 | I32 fake; | |
653099ff | 1639 | struct regnode_charclass_class ch_class; |
c277df42 IZ |
1640 | |
1641 | DEBUG_r(PerlIO_printf(Perl_debug_log, "\n")); | |
1642 | scan = r->program + 1; | |
653099ff GS |
1643 | cl_init(&ch_class); |
1644 | data.start_class = &ch_class; | |
1645 | minlen = study_chunk(&scan, &fake, scan + PL_regsize, &data, SCF_DO_STCLASS_AND); | |
c277df42 | 1646 | r->check_substr = r->anchored_substr = r->float_substr = Nullsv; |
653099ff GS |
1647 | if (!(data.start_class->flags & ANYOF_EOS) |
1648 | && !cl_is_anything(data.start_class)) { | |
1649 | SV *sv; | |
1650 | I32 n = add_data(1, "f"); | |
1651 | ||
1652 | New(1006, PL_regcomp_rx->data->data[n], 1, | |
1653 | struct regnode_charclass_class); | |
1654 | StructCopy(data.start_class, | |
1655 | (struct regnode_charclass_class*)PL_regcomp_rx->data->data[n], | |
1656 | struct regnode_charclass_class); | |
1657 | r->regstclass = (regnode*)PL_regcomp_rx->data->data[n]; | |
1658 | r->reganch &= ~ROPT_SKIP; /* Used in find_byclass(). */ | |
1659 | DEBUG_r((sv = sv_newmortal(), | |
1660 | regprop(sv, (regnode*)data.start_class), | |
894356b3 | 1661 | PerlIO_printf(Perl_debug_log, "synthetic stclass `%s'.\n", |
653099ff GS |
1662 | SvPVX(sv)))); |
1663 | } | |
a0d0e21e LW |
1664 | } |
1665 | ||
a0d0e21e | 1666 | r->minlen = minlen; |
3280af22 | 1667 | if (PL_regseen & REG_SEEN_GPOS) |
c277df42 | 1668 | r->reganch |= ROPT_GPOS_SEEN; |
3280af22 | 1669 | if (PL_regseen & REG_SEEN_LOOKBEHIND) |
c277df42 | 1670 | r->reganch |= ROPT_LOOKBEHIND_SEEN; |
3280af22 | 1671 | if (PL_regseen & REG_SEEN_EVAL) |
ce862d02 | 1672 | r->reganch |= ROPT_EVAL_SEEN; |
cf93c79d IZ |
1673 | Newz(1002, r->startp, PL_regnpar, I32); |
1674 | Newz(1002, r->endp, PL_regnpar, I32); | |
a0d0e21e LW |
1675 | DEBUG_r(regdump(r)); |
1676 | return(r); | |
a687059c LW |
1677 | } |
1678 | ||
1679 | /* | |
1680 | - reg - regular expression, i.e. main body or parenthesized thing | |
1681 | * | |
1682 | * Caller must absorb opening parenthesis. | |
1683 | * | |
1684 | * Combining parenthesis handling with the base level of regular expression | |
1685 | * is a trifle forced, but the need to tie the tails of the branches to what | |
1686 | * follows makes it hard to avoid. | |
1687 | */ | |
76e3520e | 1688 | STATIC regnode * |
cea2e8a9 | 1689 | S_reg(pTHX_ I32 paren, I32 *flagp) |
c277df42 | 1690 | /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */ |
a687059c | 1691 | { |
5c0ca799 | 1692 | dTHR; |
c277df42 IZ |
1693 | register regnode *ret; /* Will be the head of the group. */ |
1694 | register regnode *br; | |
1695 | register regnode *lastbr; | |
1696 | register regnode *ender = 0; | |
a0d0e21e | 1697 | register I32 parno = 0; |
3280af22 | 1698 | I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0; |
c277df42 | 1699 | char c; |
a0d0e21e | 1700 | |
821b33a5 | 1701 | *flagp = 0; /* Tentatively. */ |
a0d0e21e LW |
1702 | |
1703 | /* Make an OPEN node, if parenthesized. */ | |
1704 | if (paren) { | |
3280af22 | 1705 | if (*PL_regcomp_parse == '?') { |
ca9dfc88 IZ |
1706 | U16 posflags = 0, negflags = 0; |
1707 | U16 *flagsp = &posflags; | |
0f5d15d6 | 1708 | int logical = 0; |
ca9dfc88 | 1709 | |
3280af22 NIS |
1710 | PL_regcomp_parse++; |
1711 | paren = *PL_regcomp_parse++; | |
c277df42 | 1712 | ret = NULL; /* For look-ahead/behind. */ |
a0d0e21e | 1713 | switch (paren) { |
c277df42 | 1714 | case '<': |
3280af22 NIS |
1715 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
1716 | if (*PL_regcomp_parse == '!') | |
c277df42 | 1717 | paren = ','; |
3280af22 | 1718 | if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!') |
c277df42 | 1719 | goto unknown; |
3280af22 | 1720 | PL_regcomp_parse++; |
a0d0e21e LW |
1721 | case '=': |
1722 | case '!': | |
3280af22 | 1723 | PL_seen_zerolen++; |
c277df42 IZ |
1724 | case ':': |
1725 | case '>': | |
a0d0e21e LW |
1726 | break; |
1727 | case '$': | |
1728 | case '@': | |
c277df42 | 1729 | FAIL2("Sequence (?%c...) not implemented", (int)paren); |
a0d0e21e LW |
1730 | break; |
1731 | case '#': | |
3280af22 NIS |
1732 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
1733 | PL_regcomp_parse++; | |
1734 | if (*PL_regcomp_parse != ')') | |
c277df42 | 1735 | FAIL("Sequence (?#... not terminated"); |
a0d0e21e LW |
1736 | nextchar(); |
1737 | *flagp = TRYAGAIN; | |
1738 | return NULL; | |
8c8ad484 | 1739 | case 'p': |
d7e9e385 | 1740 | if (SIZE_ONLY) |
9228e8ca GS |
1741 | Perl_warner(aTHX_ WARN_REGEXP, |
1742 | "(?p{}) is deprecated - use (??{})"); | |
8c8ad484 | 1743 | /* FALL THROUGH*/ |
14455d6c | 1744 | case '?': |
0f5d15d6 IZ |
1745 | logical = 1; |
1746 | paren = *PL_regcomp_parse++; | |
1747 | /* FALL THROUGH */ | |
c277df42 IZ |
1748 | case '{': |
1749 | { | |
1750 | dTHR; | |
1751 | I32 count = 1, n = 0; | |
1752 | char c; | |
3280af22 | 1753 | char *s = PL_regcomp_parse; |
c277df42 IZ |
1754 | SV *sv; |
1755 | OP_4tree *sop, *rop; | |
1756 | ||
3280af22 NIS |
1757 | PL_seen_zerolen++; |
1758 | PL_regseen |= REG_SEEN_EVAL; | |
1759 | while (count && (c = *PL_regcomp_parse)) { | |
1760 | if (c == '\\' && PL_regcomp_parse[1]) | |
1761 | PL_regcomp_parse++; | |
c277df42 IZ |
1762 | else if (c == '{') |
1763 | count++; | |
1764 | else if (c == '}') | |
1765 | count--; | |
3280af22 | 1766 | PL_regcomp_parse++; |
c277df42 | 1767 | } |
3280af22 | 1768 | if (*PL_regcomp_parse != ')') |
c277df42 IZ |
1769 | FAIL("Sequence (?{...}) not terminated or not {}-balanced"); |
1770 | if (!SIZE_ONLY) { | |
1771 | AV *av; | |
1772 | ||
3280af22 | 1773 | if (PL_regcomp_parse - 1 - s) |
79cb57f6 | 1774 | sv = newSVpvn(s, PL_regcomp_parse - 1 - s); |
c277df42 | 1775 | else |
79cb57f6 | 1776 | sv = newSVpvn("", 0); |
c277df42 IZ |
1777 | |
1778 | rop = sv_compile_2op(sv, &sop, "re", &av); | |
1779 | ||
dfad63ad | 1780 | n = add_data(3, "nop"); |
3280af22 | 1781 | PL_regcomp_rx->data->data[n] = (void*)rop; |
dfad63ad HS |
1782 | PL_regcomp_rx->data->data[n+1] = (void*)sop; |
1783 | PL_regcomp_rx->data->data[n+2] = (void*)av; | |
c277df42 | 1784 | SvREFCNT_dec(sv); |
a0ed51b3 | 1785 | } |
e24b16f9 GS |
1786 | else { /* First pass */ |
1787 | if (PL_reginterp_cnt < ++PL_seen_evals | |
1788 | && PL_curcop != &PL_compiling) | |
2cd61cdb IZ |
1789 | /* No compiled RE interpolated, has runtime |
1790 | components ===> unsafe. */ | |
1791 | FAIL("Eval-group not allowed at runtime, use re 'eval'"); | |
3280af22 | 1792 | if (PL_tainted) |
cc6b7395 | 1793 | FAIL("Eval-group in insecure regular expression"); |
c277df42 IZ |
1794 | } |
1795 | ||
1796 | nextchar(); | |
0f5d15d6 IZ |
1797 | if (logical) { |
1798 | ret = reg_node(LOGICAL); | |
1799 | if (!SIZE_ONLY) | |
1800 | ret->flags = 2; | |
1801 | regtail(ret, reganode(EVAL, n)); | |
1802 | return ret; | |
1803 | } | |
c277df42 IZ |
1804 | return reganode(EVAL, n); |
1805 | } | |
1806 | case '(': | |
1807 | { | |
3280af22 NIS |
1808 | if (PL_regcomp_parse[0] == '?') { |
1809 | if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!' | |
1810 | || PL_regcomp_parse[1] == '<' | |
1811 | || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */ | |
c277df42 IZ |
1812 | I32 flag; |
1813 | ||
1814 | ret = reg_node(LOGICAL); | |
0f5d15d6 IZ |
1815 | if (!SIZE_ONLY) |
1816 | ret->flags = 1; | |
c277df42 IZ |
1817 | regtail(ret, reg(1, &flag)); |
1818 | goto insert_if; | |
1819 | } | |
a0ed51b3 LW |
1820 | } |
1821 | else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) { | |
3280af22 | 1822 | parno = atoi(PL_regcomp_parse++); |
c277df42 | 1823 | |
3280af22 NIS |
1824 | while (isDIGIT(*PL_regcomp_parse)) |
1825 | PL_regcomp_parse++; | |
c277df42 IZ |
1826 | ret = reganode(GROUPP, parno); |
1827 | if ((c = *nextchar()) != ')') | |
1828 | FAIL2("Switch (?(number%c not recognized", c); | |
1829 | insert_if: | |
1830 | regtail(ret, reganode(IFTHEN, 0)); | |
1831 | br = regbranch(&flags, 1); | |
1832 | if (br == NULL) | |
1833 | br = reganode(LONGJMP, 0); | |
1834 | else | |
1835 | regtail(br, reganode(LONGJMP, 0)); | |
1836 | c = *nextchar(); | |
d1b80229 IZ |
1837 | if (flags&HASWIDTH) |
1838 | *flagp |= HASWIDTH; | |
c277df42 IZ |
1839 | if (c == '|') { |
1840 | lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */ | |
1841 | regbranch(&flags, 1); | |
1842 | regtail(ret, lastbr); | |
d1b80229 IZ |
1843 | if (flags&HASWIDTH) |
1844 | *flagp |= HASWIDTH; | |
c277df42 | 1845 | c = *nextchar(); |
a0ed51b3 LW |
1846 | } |
1847 | else | |
c277df42 IZ |
1848 | lastbr = NULL; |
1849 | if (c != ')') | |
1850 | FAIL("Switch (?(condition)... contains too many branches"); | |
1851 | ender = reg_node(TAIL); | |
1852 | regtail(br, ender); | |
1853 | if (lastbr) { | |
1854 | regtail(lastbr, ender); | |
1855 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); | |
a0ed51b3 LW |
1856 | } |
1857 | else | |
c277df42 IZ |
1858 | regtail(ret, ender); |
1859 | return ret; | |
a0ed51b3 LW |
1860 | } |
1861 | else { | |
3280af22 | 1862 | FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse); |
c277df42 IZ |
1863 | } |
1864 | } | |
1b1626e4 | 1865 | case 0: |
c277df42 | 1866 | FAIL("Sequence (? incomplete"); |
1b1626e4 | 1867 | break; |
a0d0e21e | 1868 | default: |
3280af22 | 1869 | --PL_regcomp_parse; |
ca9dfc88 | 1870 | parse_flags: |
3280af22 NIS |
1871 | while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) { |
1872 | if (*PL_regcomp_parse != 'o') | |
1873 | pmflag(flagsp, *PL_regcomp_parse); | |
1874 | ++PL_regcomp_parse; | |
ca9dfc88 | 1875 | } |
3280af22 | 1876 | if (*PL_regcomp_parse == '-') { |
ca9dfc88 | 1877 | flagsp = &negflags; |
3280af22 | 1878 | ++PL_regcomp_parse; |
ca9dfc88 | 1879 | goto parse_flags; |
48c036b1 | 1880 | } |
3280af22 NIS |
1881 | PL_regflags |= posflags; |
1882 | PL_regflags &= ~negflags; | |
1883 | if (*PL_regcomp_parse == ':') { | |
1884 | PL_regcomp_parse++; | |
ca9dfc88 IZ |
1885 | paren = ':'; |
1886 | break; | |
1887 | } | |
c277df42 | 1888 | unknown: |
3280af22 NIS |
1889 | if (*PL_regcomp_parse != ')') |
1890 | FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse); | |
a0d0e21e LW |
1891 | nextchar(); |
1892 | *flagp = TRYAGAIN; | |
1893 | return NULL; | |
1894 | } | |
1895 | } | |
1896 | else { | |
3280af22 NIS |
1897 | parno = PL_regnpar; |
1898 | PL_regnpar++; | |
a0d0e21e | 1899 | ret = reganode(OPEN, parno); |
c277df42 | 1900 | open = 1; |
a0d0e21e | 1901 | } |
a0ed51b3 LW |
1902 | } |
1903 | else | |
a0d0e21e LW |
1904 | ret = NULL; |
1905 | ||
1906 | /* Pick up the branches, linking them together. */ | |
c277df42 | 1907 | br = regbranch(&flags, 1); |
a0d0e21e LW |
1908 | if (br == NULL) |
1909 | return(NULL); | |
3280af22 NIS |
1910 | if (*PL_regcomp_parse == '|') { |
1911 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 | 1912 | reginsert(BRANCHJ, br); |
a0ed51b3 LW |
1913 | } |
1914 | else | |
c277df42 IZ |
1915 | reginsert(BRANCH, br); |
1916 | have_branch = 1; | |
1917 | if (SIZE_ONLY) | |
3280af22 | 1918 | PL_extralen += 1; /* For BRANCHJ-BRANCH. */ |
a0ed51b3 LW |
1919 | } |
1920 | else if (paren == ':') { | |
c277df42 IZ |
1921 | *flagp |= flags&SIMPLE; |
1922 | } | |
1923 | if (open) { /* Starts with OPEN. */ | |
1924 | regtail(ret, br); /* OPEN -> first. */ | |
a0ed51b3 LW |
1925 | } |
1926 | else if (paren != '?') /* Not Conditional */ | |
a0d0e21e | 1927 | ret = br; |
821b33a5 IZ |
1928 | if (flags&HASWIDTH) |
1929 | *flagp |= HASWIDTH; | |
a0d0e21e | 1930 | *flagp |= flags&SPSTART; |
c277df42 | 1931 | lastbr = br; |
3280af22 NIS |
1932 | while (*PL_regcomp_parse == '|') { |
1933 | if (!SIZE_ONLY && PL_extralen) { | |
c277df42 IZ |
1934 | ender = reganode(LONGJMP,0); |
1935 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ | |
1936 | } | |
1937 | if (SIZE_ONLY) | |
3280af22 | 1938 | PL_extralen += 2; /* Account for LONGJMP. */ |
a0d0e21e | 1939 | nextchar(); |
c277df42 | 1940 | br = regbranch(&flags, 0); |
a687059c | 1941 | if (br == NULL) |
a0d0e21e | 1942 | return(NULL); |
c277df42 IZ |
1943 | regtail(lastbr, br); /* BRANCH -> BRANCH. */ |
1944 | lastbr = br; | |
821b33a5 IZ |
1945 | if (flags&HASWIDTH) |
1946 | *flagp |= HASWIDTH; | |
a687059c | 1947 | *flagp |= flags&SPSTART; |
a0d0e21e LW |
1948 | } |
1949 | ||
c277df42 IZ |
1950 | if (have_branch || paren != ':') { |
1951 | /* Make a closing node, and hook it on the end. */ | |
1952 | switch (paren) { | |
1953 | case ':': | |
1954 | ender = reg_node(TAIL); | |
1955 | break; | |
1956 | case 1: | |
1957 | ender = reganode(CLOSE, parno); | |
1958 | break; | |
1959 | case '<': | |
c277df42 IZ |
1960 | case ',': |
1961 | case '=': | |
1962 | case '!': | |
c277df42 | 1963 | *flagp &= ~HASWIDTH; |
821b33a5 IZ |
1964 | /* FALL THROUGH */ |
1965 | case '>': | |
1966 | ender = reg_node(SUCCEED); | |
c277df42 IZ |
1967 | break; |
1968 | case 0: | |
1969 | ender = reg_node(END); | |
1970 | break; | |
1971 | } | |
1972 | regtail(lastbr, ender); | |
a0d0e21e | 1973 | |
c277df42 IZ |
1974 | if (have_branch) { |
1975 | /* Hook the tails of the branches to the closing node. */ | |
1976 | for (br = ret; br != NULL; br = regnext(br)) { | |
1977 | regoptail(br, ender); | |
1978 | } | |
1979 | } | |
a0d0e21e | 1980 | } |
c277df42 IZ |
1981 | |
1982 | { | |
1983 | char *p; | |
1984 | static char parens[] = "=!<,>"; | |
1985 | ||
1986 | if (paren && (p = strchr(parens, paren))) { | |
1987 | int node = ((p - parens) % 2) ? UNLESSM : IFMATCH; | |
1988 | int flag = (p - parens) > 1; | |
1989 | ||
1990 | if (paren == '>') | |
1991 | node = SUSPEND, flag = 0; | |
1992 | reginsert(node,ret); | |
c277df42 | 1993 | ret->flags = flag; |
c277df42 IZ |
1994 | regtail(ret, reg_node(TAIL)); |
1995 | } | |
a0d0e21e LW |
1996 | } |
1997 | ||
1998 | /* Check for proper termination. */ | |
ce3e6498 HS |
1999 | if (paren) { |
2000 | PL_regflags = oregflags; | |
2001 | if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') { | |
2002 | FAIL("unmatched () in regexp"); | |
2003 | } | |
a0ed51b3 LW |
2004 | } |
2005 | else if (!paren && PL_regcomp_parse < PL_regxend) { | |
3280af22 | 2006 | if (*PL_regcomp_parse == ')') { |
a0d0e21e | 2007 | FAIL("unmatched () in regexp"); |
a0ed51b3 LW |
2008 | } |
2009 | else | |
a0d0e21e LW |
2010 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
2011 | /* NOTREACHED */ | |
2012 | } | |
a687059c | 2013 | |
a0d0e21e | 2014 | return(ret); |
a687059c LW |
2015 | } |
2016 | ||
2017 | /* | |
2018 | - regbranch - one alternative of an | operator | |
2019 | * | |
2020 | * Implements the concatenation operator. | |
2021 | */ | |
76e3520e | 2022 | STATIC regnode * |
cea2e8a9 | 2023 | S_regbranch(pTHX_ I32 *flagp, I32 first) |
a687059c | 2024 | { |
5c0ca799 | 2025 | dTHR; |
c277df42 IZ |
2026 | register regnode *ret; |
2027 | register regnode *chain = NULL; | |
2028 | register regnode *latest; | |
2029 | I32 flags = 0, c = 0; | |
a0d0e21e | 2030 | |
c277df42 IZ |
2031 | if (first) |
2032 | ret = NULL; | |
2033 | else { | |
3280af22 | 2034 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
2035 | ret = reganode(BRANCHJ,0); |
2036 | else | |
2037 | ret = reg_node(BRANCH); | |
2038 | } | |
2039 | ||
2040 | if (!first && SIZE_ONLY) | |
3280af22 | 2041 | PL_extralen += 1; /* BRANCHJ */ |
c277df42 IZ |
2042 | |
2043 | *flagp = WORST; /* Tentatively. */ | |
a0d0e21e | 2044 | |
3280af22 | 2045 | PL_regcomp_parse--; |
a0d0e21e | 2046 | nextchar(); |
3280af22 | 2047 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') { |
a0d0e21e LW |
2048 | flags &= ~TRYAGAIN; |
2049 | latest = regpiece(&flags); | |
2050 | if (latest == NULL) { | |
2051 | if (flags & TRYAGAIN) | |
2052 | continue; | |
2053 | return(NULL); | |
a0ed51b3 LW |
2054 | } |
2055 | else if (ret == NULL) | |
c277df42 | 2056 | ret = latest; |
a0d0e21e | 2057 | *flagp |= flags&HASWIDTH; |
c277df42 | 2058 | if (chain == NULL) /* First piece. */ |
a0d0e21e LW |
2059 | *flagp |= flags&SPSTART; |
2060 | else { | |
3280af22 | 2061 | PL_regnaughty++; |
a0d0e21e | 2062 | regtail(chain, latest); |
a687059c | 2063 | } |
a0d0e21e | 2064 | chain = latest; |
c277df42 IZ |
2065 | c++; |
2066 | } | |
2067 | if (chain == NULL) { /* Loop ran zero times. */ | |
2068 | chain = reg_node(NOTHING); | |
2069 | if (ret == NULL) | |
2070 | ret = chain; | |
2071 | } | |
2072 | if (c == 1) { | |
2073 | *flagp |= flags&SIMPLE; | |
a0d0e21e | 2074 | } |
a687059c | 2075 | |
a0d0e21e | 2076 | return(ret); |
a687059c LW |
2077 | } |
2078 | ||
2079 | /* | |
2080 | - regpiece - something followed by possible [*+?] | |
2081 | * | |
2082 | * Note that the branching code sequences used for ? and the general cases | |
2083 | * of * and + are somewhat optimized: they use the same NOTHING node as | |
2084 | * both the endmarker for their branch list and the body of the last branch. | |
2085 | * It might seem that this node could be dispensed with entirely, but the | |
2086 | * endmarker role is not redundant. | |
2087 | */ | |
76e3520e | 2088 | STATIC regnode * |
cea2e8a9 | 2089 | S_regpiece(pTHX_ I32 *flagp) |
a687059c | 2090 | { |
5c0ca799 | 2091 | dTHR; |
c277df42 | 2092 | register regnode *ret; |
a0d0e21e LW |
2093 | register char op; |
2094 | register char *next; | |
2095 | I32 flags; | |
3280af22 | 2096 | char *origparse = PL_regcomp_parse; |
a0d0e21e LW |
2097 | char *maxpos; |
2098 | I32 min; | |
c277df42 | 2099 | I32 max = REG_INFTY; |
a0d0e21e LW |
2100 | |
2101 | ret = regatom(&flags); | |
2102 | if (ret == NULL) { | |
2103 | if (flags & TRYAGAIN) | |
2104 | *flagp |= TRYAGAIN; | |
2105 | return(NULL); | |
2106 | } | |
2107 | ||
3280af22 | 2108 | op = *PL_regcomp_parse; |
a0d0e21e | 2109 | |
3280af22 NIS |
2110 | if (op == '{' && regcurly(PL_regcomp_parse)) { |
2111 | next = PL_regcomp_parse + 1; | |
a0d0e21e LW |
2112 | maxpos = Nullch; |
2113 | while (isDIGIT(*next) || *next == ',') { | |
2114 | if (*next == ',') { | |
2115 | if (maxpos) | |
2116 | break; | |
2117 | else | |
2118 | maxpos = next; | |
a687059c | 2119 | } |
a0d0e21e LW |
2120 | next++; |
2121 | } | |
2122 | if (*next == '}') { /* got one */ | |
2123 | if (!maxpos) | |
2124 | maxpos = next; | |
3280af22 NIS |
2125 | PL_regcomp_parse++; |
2126 | min = atoi(PL_regcomp_parse); | |
a0d0e21e LW |
2127 | if (*maxpos == ',') |
2128 | maxpos++; | |
2129 | else | |
3280af22 | 2130 | maxpos = PL_regcomp_parse; |
a0d0e21e LW |
2131 | max = atoi(maxpos); |
2132 | if (!max && *maxpos != '0') | |
c277df42 IZ |
2133 | max = REG_INFTY; /* meaning "infinity" */ |
2134 | else if (max >= REG_INFTY) | |
2135 | FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); | |
3280af22 | 2136 | PL_regcomp_parse = next; |
a0d0e21e LW |
2137 | nextchar(); |
2138 | ||
2139 | do_curly: | |
2140 | if ((flags&SIMPLE)) { | |
3280af22 | 2141 | PL_regnaughty += 2 + PL_regnaughty / 2; |
a0d0e21e LW |
2142 | reginsert(CURLY, ret); |
2143 | } | |
2144 | else { | |
2c2d71f5 JH |
2145 | regnode *w = reg_node(WHILEM); |
2146 | ||
2147 | w->flags = 0; | |
2148 | regtail(ret, w); | |
3280af22 | 2149 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 IZ |
2150 | reginsert(LONGJMP,ret); |
2151 | reginsert(NOTHING,ret); | |
2152 | NEXT_OFF(ret) = 3; /* Go over LONGJMP. */ | |
2153 | } | |
a0d0e21e | 2154 | reginsert(CURLYX,ret); |
3280af22 | 2155 | if (!SIZE_ONLY && PL_extralen) |
c277df42 IZ |
2156 | NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */ |
2157 | regtail(ret, reg_node(NOTHING)); | |
2158 | if (SIZE_ONLY) | |
2c2d71f5 JH |
2159 | PL_reg_whilem_seen++, PL_extralen += 3; |
2160 | PL_regnaughty += 4 + PL_regnaughty; /* compound interest */ | |
a0d0e21e | 2161 | } |
c277df42 | 2162 | ret->flags = 0; |
a0d0e21e LW |
2163 | |
2164 | if (min > 0) | |
821b33a5 IZ |
2165 | *flagp = WORST; |
2166 | if (max > 0) | |
2167 | *flagp |= HASWIDTH; | |
a0d0e21e | 2168 | if (max && max < min) |
c277df42 IZ |
2169 | FAIL("Can't do {n,m} with n > m"); |
2170 | if (!SIZE_ONLY) { | |
2171 | ARG1_SET(ret, min); | |
2172 | ARG2_SET(ret, max); | |
a687059c | 2173 | } |
a687059c | 2174 | |
a0d0e21e | 2175 | goto nest_check; |
a687059c | 2176 | } |
a0d0e21e | 2177 | } |
a687059c | 2178 | |
a0d0e21e LW |
2179 | if (!ISMULT1(op)) { |
2180 | *flagp = flags; | |
a687059c | 2181 | return(ret); |
a0d0e21e | 2182 | } |
bb20fd44 | 2183 | |
c277df42 | 2184 | #if 0 /* Now runtime fix should be reliable. */ |
bb20fd44 | 2185 | if (!(flags&HASWIDTH) && op != '?') |
c277df42 IZ |
2186 | FAIL("regexp *+ operand could be empty"); |
2187 | #endif | |
bb20fd44 | 2188 | |
a0d0e21e LW |
2189 | nextchar(); |
2190 | ||
821b33a5 | 2191 | *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); |
a0d0e21e LW |
2192 | |
2193 | if (op == '*' && (flags&SIMPLE)) { | |
2194 | reginsert(STAR, ret); | |
c277df42 | 2195 | ret->flags = 0; |
3280af22 | 2196 | PL_regnaughty += 4; |
a0d0e21e LW |
2197 | } |
2198 | else if (op == '*') { | |
2199 | min = 0; | |
2200 | goto do_curly; | |
a0ed51b3 LW |
2201 | } |
2202 | else if (op == '+' && (flags&SIMPLE)) { | |
a0d0e21e | 2203 | reginsert(PLUS, ret); |
c277df42 | 2204 | ret->flags = 0; |
3280af22 | 2205 | PL_regnaughty += 3; |
a0d0e21e LW |
2206 | } |
2207 | else if (op == '+') { | |
2208 | min = 1; | |
2209 | goto do_curly; | |
a0ed51b3 LW |
2210 | } |
2211 | else if (op == '?') { | |
a0d0e21e LW |
2212 | min = 0; max = 1; |
2213 | goto do_curly; | |
2214 | } | |
2215 | nest_check: | |
e476b1b5 GS |
2216 | if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) { |
2217 | Perl_warner(aTHX_ WARN_REGEXP, "%.*s matches null string many times", | |
3280af22 | 2218 | PL_regcomp_parse - origparse, origparse); |
a0d0e21e LW |
2219 | } |
2220 | ||
3280af22 | 2221 | if (*PL_regcomp_parse == '?') { |
a0d0e21e LW |
2222 | nextchar(); |
2223 | reginsert(MINMOD, ret); | |
c277df42 | 2224 | regtail(ret, ret + NODE_STEP_REGNODE); |
a0d0e21e | 2225 | } |
3280af22 | 2226 | if (ISMULT2(PL_regcomp_parse)) |
a0d0e21e LW |
2227 | FAIL("nested *?+ in regexp"); |
2228 | ||
2229 | return(ret); | |
a687059c LW |
2230 | } |
2231 | ||
2232 | /* | |
2233 | - regatom - the lowest level | |
2234 | * | |
2235 | * Optimization: gobbles an entire sequence of ordinary characters so that | |
2236 | * it can turn them into a single node, which is smaller to store and | |
2237 | * faster to run. Backslashed characters are exceptions, each becoming a | |
2238 | * separate node; the code is simpler that way and it's not worth fixing. | |
2239 | * | |
2240 | * [Yes, it is worth fixing, some scripts can run twice the speed.] | |
2241 | */ | |
76e3520e | 2242 | STATIC regnode * |
cea2e8a9 | 2243 | S_regatom(pTHX_ I32 *flagp) |
a687059c | 2244 | { |
5c0ca799 | 2245 | dTHR; |
c277df42 | 2246 | register regnode *ret = 0; |
a0d0e21e LW |
2247 | I32 flags; |
2248 | ||
2249 | *flagp = WORST; /* Tentatively. */ | |
2250 | ||
2251 | tryagain: | |
3280af22 | 2252 | switch (*PL_regcomp_parse) { |
a0d0e21e | 2253 | case '^': |
3280af22 | 2254 | PL_seen_zerolen++; |
a0d0e21e | 2255 | nextchar(); |
3280af22 | 2256 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 2257 | ret = reg_node(MBOL); |
3280af22 | 2258 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 2259 | ret = reg_node(SBOL); |
a0d0e21e | 2260 | else |
c277df42 | 2261 | ret = reg_node(BOL); |
a0d0e21e LW |
2262 | break; |
2263 | case '$': | |
3280af22 NIS |
2264 | if (PL_regcomp_parse[1]) |
2265 | PL_seen_zerolen++; | |
a0d0e21e | 2266 | nextchar(); |
3280af22 | 2267 | if (PL_regflags & PMf_MULTILINE) |
c277df42 | 2268 | ret = reg_node(MEOL); |
3280af22 | 2269 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 | 2270 | ret = reg_node(SEOL); |
a0d0e21e | 2271 | else |
c277df42 | 2272 | ret = reg_node(EOL); |
a0d0e21e LW |
2273 | break; |
2274 | case '.': | |
2275 | nextchar(); | |
a0ed51b3 LW |
2276 | if (UTF) { |
2277 | if (PL_regflags & PMf_SINGLELINE) | |
2278 | ret = reg_node(SANYUTF8); | |
2279 | else | |
2280 | ret = reg_node(ANYUTF8); | |
2281 | *flagp |= HASWIDTH; | |
2282 | } | |
2283 | else { | |
2284 | if (PL_regflags & PMf_SINGLELINE) | |
2285 | ret = reg_node(SANY); | |
2286 | else | |
22c35a8c | 2287 | ret = reg_node(REG_ANY); |
a0ed51b3 LW |
2288 | *flagp |= HASWIDTH|SIMPLE; |
2289 | } | |
3280af22 | 2290 | PL_regnaughty++; |
a0d0e21e LW |
2291 | break; |
2292 | case '[': | |
3280af22 | 2293 | PL_regcomp_parse++; |
a0ed51b3 | 2294 | ret = (UTF ? regclassutf8() : regclass()); |
a14b48bc LW |
2295 | if (*PL_regcomp_parse != ']') |
2296 | FAIL("unmatched [] in regexp"); | |
2297 | nextchar(); | |
a0d0e21e LW |
2298 | *flagp |= HASWIDTH|SIMPLE; |
2299 | break; | |
2300 | case '(': | |
2301 | nextchar(); | |
2302 | ret = reg(1, &flags); | |
2303 | if (ret == NULL) { | |
bf93d4cc GS |
2304 | if (flags & TRYAGAIN) { |
2305 | if (PL_regcomp_parse == PL_regxend) { | |
2306 | /* Make parent create an empty node if needed. */ | |
2307 | *flagp |= TRYAGAIN; | |
2308 | return(NULL); | |
2309 | } | |
a0d0e21e | 2310 | goto tryagain; |
bf93d4cc | 2311 | } |
a0d0e21e LW |
2312 | return(NULL); |
2313 | } | |
c277df42 | 2314 | *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE); |
a0d0e21e LW |
2315 | break; |
2316 | case '|': | |
2317 | case ')': | |
2318 | if (flags & TRYAGAIN) { | |
2319 | *flagp |= TRYAGAIN; | |
2320 | return NULL; | |
2321 | } | |
3280af22 | 2322 | FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse); |
a0d0e21e LW |
2323 | /* Supposed to be caught earlier. */ |
2324 | break; | |
85afd4ae | 2325 | case '{': |
3280af22 NIS |
2326 | if (!regcurly(PL_regcomp_parse)) { |
2327 | PL_regcomp_parse++; | |
85afd4ae CS |
2328 | goto defchar; |
2329 | } | |
2330 | /* FALL THROUGH */ | |
a0d0e21e LW |
2331 | case '?': |
2332 | case '+': | |
2333 | case '*': | |
3115e423 | 2334 | FAIL("?+*{} follows nothing in regexp"); |
a0d0e21e LW |
2335 | break; |
2336 | case '\\': | |
3280af22 | 2337 | switch (*++PL_regcomp_parse) { |
a0d0e21e | 2338 | case 'A': |
3280af22 | 2339 | PL_seen_zerolen++; |
c277df42 | 2340 | ret = reg_node(SBOL); |
a0d0e21e LW |
2341 | *flagp |= SIMPLE; |
2342 | nextchar(); | |
2343 | break; | |
2344 | case 'G': | |
c277df42 | 2345 | ret = reg_node(GPOS); |
3280af22 | 2346 | PL_regseen |= REG_SEEN_GPOS; |
a0d0e21e LW |
2347 | *flagp |= SIMPLE; |
2348 | nextchar(); | |
2349 | break; | |
2350 | case 'Z': | |
c277df42 | 2351 | ret = reg_node(SEOL); |
a0d0e21e LW |
2352 | *flagp |= SIMPLE; |
2353 | nextchar(); | |
2354 | break; | |
b85d18e9 IZ |
2355 | case 'z': |
2356 | ret = reg_node(EOS); | |
2357 | *flagp |= SIMPLE; | |
3280af22 | 2358 | PL_seen_zerolen++; /* Do not optimize RE away */ |
b85d18e9 IZ |
2359 | nextchar(); |
2360 | break; | |
4a2d328f | 2361 | case 'C': |
a0ed51b3 LW |
2362 | ret = reg_node(SANY); |
2363 | *flagp |= HASWIDTH|SIMPLE; | |
2364 | nextchar(); | |
2365 | break; | |
2366 | case 'X': | |
2367 | ret = reg_node(CLUMP); | |
2368 | *flagp |= HASWIDTH; | |
2369 | nextchar(); | |
2370 | if (UTF && !PL_utf8_mark) | |
dfe13c55 | 2371 | is_utf8_mark((U8*)"~"); /* preload table */ |
a0ed51b3 | 2372 | break; |
a0d0e21e | 2373 | case 'w': |
a0ed51b3 LW |
2374 | ret = reg_node( |
2375 | UTF | |
2376 | ? (LOC ? ALNUMLUTF8 : ALNUMUTF8) | |
2377 | : (LOC ? ALNUML : ALNUM)); | |
a0d0e21e LW |
2378 | *flagp |= HASWIDTH|SIMPLE; |
2379 | nextchar(); | |
a0ed51b3 | 2380 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 2381 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
2382 | break; |
2383 | case 'W': | |
a0ed51b3 LW |
2384 | ret = reg_node( |
2385 | UTF | |
2386 | ? (LOC ? NALNUMLUTF8 : NALNUMUTF8) | |
2387 | : (LOC ? NALNUML : NALNUM)); | |
a0d0e21e LW |
2388 | *flagp |= HASWIDTH|SIMPLE; |
2389 | nextchar(); | |
a0ed51b3 | 2390 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 2391 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
2392 | break; |
2393 | case 'b': | |
3280af22 | 2394 | PL_seen_zerolen++; |
f5c9036e | 2395 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
2396 | ret = reg_node( |
2397 | UTF | |
2398 | ? (LOC ? BOUNDLUTF8 : BOUNDUTF8) | |
2399 | : (LOC ? BOUNDL : BOUND)); | |
a0d0e21e LW |
2400 | *flagp |= SIMPLE; |
2401 | nextchar(); | |
a0ed51b3 | 2402 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 2403 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
2404 | break; |
2405 | case 'B': | |
3280af22 | 2406 | PL_seen_zerolen++; |
f5c9036e | 2407 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
a0ed51b3 LW |
2408 | ret = reg_node( |
2409 | UTF | |
2410 | ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8) | |
2411 | : (LOC ? NBOUNDL : NBOUND)); | |
a0d0e21e LW |
2412 | *flagp |= SIMPLE; |
2413 | nextchar(); | |
a0ed51b3 | 2414 | if (UTF && !PL_utf8_alnum) |
dfe13c55 | 2415 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e LW |
2416 | break; |
2417 | case 's': | |
a0ed51b3 LW |
2418 | ret = reg_node( |
2419 | UTF | |
2420 | ? (LOC ? SPACELUTF8 : SPACEUTF8) | |
2421 | : (LOC ? SPACEL : SPACE)); | |
a0d0e21e LW |
2422 | *flagp |= HASWIDTH|SIMPLE; |
2423 | nextchar(); | |
a0ed51b3 | 2424 | if (UTF && !PL_utf8_space) |
dfe13c55 | 2425 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
2426 | break; |
2427 | case 'S': | |
a0ed51b3 LW |
2428 | ret = reg_node( |
2429 | UTF | |
2430 | ? (LOC ? NSPACELUTF8 : NSPACEUTF8) | |
2431 | : (LOC ? NSPACEL : NSPACE)); | |
a0d0e21e LW |
2432 | *flagp |= HASWIDTH|SIMPLE; |
2433 | nextchar(); | |
a0ed51b3 | 2434 | if (UTF && !PL_utf8_space) |
dfe13c55 | 2435 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e LW |
2436 | break; |
2437 | case 'd': | |
a0ed51b3 | 2438 | ret = reg_node(UTF ? DIGITUTF8 : DIGIT); |
a0d0e21e LW |
2439 | *flagp |= HASWIDTH|SIMPLE; |
2440 | nextchar(); | |
a0ed51b3 | 2441 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 2442 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e LW |
2443 | break; |
2444 | case 'D': | |
a0ed51b3 | 2445 | ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT); |
a0d0e21e LW |
2446 | *flagp |= HASWIDTH|SIMPLE; |
2447 | nextchar(); | |
a0ed51b3 | 2448 | if (UTF && !PL_utf8_digit) |
dfe13c55 | 2449 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e | 2450 | break; |
a14b48bc LW |
2451 | case 'p': |
2452 | case 'P': | |
2453 | { /* a lovely hack--pretend we saw [\pX] instead */ | |
2454 | char* oldregxend = PL_regxend; | |
2455 | ||
2456 | if (PL_regcomp_parse[1] == '{') { | |
2457 | PL_regxend = strchr(PL_regcomp_parse, '}'); | |
2458 | if (!PL_regxend) | |
2459 | FAIL("Missing right brace on \\p{}"); | |
2460 | PL_regxend++; | |
2461 | } | |
2462 | else | |
2463 | PL_regxend = PL_regcomp_parse + 2; | |
2464 | PL_regcomp_parse--; | |
2465 | ||
2466 | ret = regclassutf8(); | |
2467 | ||
2468 | PL_regxend = oldregxend; | |
2469 | PL_regcomp_parse--; | |
2470 | nextchar(); | |
2471 | *flagp |= HASWIDTH|SIMPLE; | |
2472 | } | |
2473 | break; | |
a0d0e21e LW |
2474 | case 'n': |
2475 | case 'r': | |
2476 | case 't': | |
2477 | case 'f': | |
2478 | case 'e': | |
2479 | case 'a': | |
2480 | case 'x': | |
2481 | case 'c': | |
2482 | case '0': | |
2483 | goto defchar; | |
2484 | case '1': case '2': case '3': case '4': | |
2485 | case '5': case '6': case '7': case '8': case '9': | |
2486 | { | |
3280af22 | 2487 | I32 num = atoi(PL_regcomp_parse); |
a0d0e21e | 2488 | |
3280af22 | 2489 | if (num > 9 && num >= PL_regnpar) |
a0d0e21e LW |
2490 | goto defchar; |
2491 | else { | |
3280af22 | 2492 | if (!SIZE_ONLY && num > PL_regcomp_rx->nparens) |
ef64f398 | 2493 | FAIL("reference to nonexistent group"); |
3280af22 | 2494 | PL_regsawback = 1; |
a0ed51b3 LW |
2495 | ret = reganode(FOLD |
2496 | ? (LOC ? REFFL : REFF) | |
c8756f30 | 2497 | : REF, num); |
a0d0e21e | 2498 | *flagp |= HASWIDTH; |
3280af22 NIS |
2499 | while (isDIGIT(*PL_regcomp_parse)) |
2500 | PL_regcomp_parse++; | |
2501 | PL_regcomp_parse--; | |
a0d0e21e LW |
2502 | nextchar(); |
2503 | } | |
2504 | } | |
2505 | break; | |
2506 | case '\0': | |
3280af22 | 2507 | if (PL_regcomp_parse >= PL_regxend) |
a0d0e21e LW |
2508 | FAIL("trailing \\ in regexp"); |
2509 | /* FALL THROUGH */ | |
2510 | default: | |
c9f97d15 IZ |
2511 | /* Do not generate `unrecognized' warnings here, we fall |
2512 | back into the quick-grab loop below */ | |
a0d0e21e LW |
2513 | goto defchar; |
2514 | } | |
2515 | break; | |
4633a7c4 LW |
2516 | |
2517 | case '#': | |
3280af22 NIS |
2518 | if (PL_regflags & PMf_EXTENDED) { |
2519 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++; | |
2520 | if (PL_regcomp_parse < PL_regxend) | |
4633a7c4 LW |
2521 | goto tryagain; |
2522 | } | |
2523 | /* FALL THROUGH */ | |
2524 | ||
a0d0e21e LW |
2525 | default: { |
2526 | register I32 len; | |
a0ed51b3 | 2527 | register UV ender; |
a0d0e21e | 2528 | register char *p; |
c277df42 | 2529 | char *oldp, *s; |
a0d0e21e LW |
2530 | I32 numlen; |
2531 | ||
3280af22 | 2532 | PL_regcomp_parse++; |
a0d0e21e LW |
2533 | |
2534 | defchar: | |
a0ed51b3 LW |
2535 | ret = reg_node(FOLD |
2536 | ? (LOC ? EXACTFL : EXACTF) | |
bbce6d69 | 2537 | : EXACT); |
cd439c50 | 2538 | s = STRING(ret); |
3280af22 NIS |
2539 | for (len = 0, p = PL_regcomp_parse - 1; |
2540 | len < 127 && p < PL_regxend; | |
a0d0e21e LW |
2541 | len++) |
2542 | { | |
2543 | oldp = p; | |
5b5a24f7 | 2544 | |
3280af22 NIS |
2545 | if (PL_regflags & PMf_EXTENDED) |
2546 | p = regwhite(p, PL_regxend); | |
a0d0e21e LW |
2547 | switch (*p) { |
2548 | case '^': | |
2549 | case '$': | |
2550 | case '.': | |
2551 | case '[': | |
2552 | case '(': | |
2553 | case ')': | |
2554 | case '|': | |
2555 | goto loopdone; | |
2556 | case '\\': | |
2557 | switch (*++p) { | |
2558 | case 'A': | |
2559 | case 'G': | |
2560 | case 'Z': | |
b85d18e9 | 2561 | case 'z': |
a0d0e21e LW |
2562 | case 'w': |
2563 | case 'W': | |
2564 | case 'b': | |
2565 | case 'B': | |
2566 | case 's': | |
2567 | case 'S': | |
2568 | case 'd': | |
2569 | case 'D': | |
a14b48bc LW |
2570 | case 'p': |
2571 | case 'P': | |
a0d0e21e LW |
2572 | --p; |
2573 | goto loopdone; | |
2574 | case 'n': | |
2575 | ender = '\n'; | |
2576 | p++; | |
a687059c | 2577 | break; |
a0d0e21e LW |
2578 | case 'r': |
2579 | ender = '\r'; | |
2580 | p++; | |
a687059c | 2581 | break; |
a0d0e21e LW |
2582 | case 't': |
2583 | ender = '\t'; | |
2584 | p++; | |
a687059c | 2585 | break; |
a0d0e21e LW |
2586 | case 'f': |
2587 | ender = '\f'; | |
2588 | p++; | |
a687059c | 2589 | break; |
a0d0e21e | 2590 | case 'e': |
0407a77b GS |
2591 | #ifdef ASCIIish |
2592 | ender = '\033'; | |
2593 | #else | |
2594 | ender = '\047'; | |
2595 | #endif | |
a0d0e21e | 2596 | p++; |
a687059c | 2597 | break; |
a0d0e21e | 2598 | case 'a': |
0407a77b GS |
2599 | #ifdef ASCIIish |
2600 | ender = '\007'; | |
2601 | #else | |
2602 | ender = '\057'; | |
2603 | #endif | |
a0d0e21e | 2604 | p++; |
a687059c | 2605 | break; |
a0d0e21e | 2606 | case 'x': |
a0ed51b3 LW |
2607 | if (*++p == '{') { |
2608 | char* e = strchr(p, '}'); | |
2609 | ||
2610 | if (!e) | |
2611 | FAIL("Missing right brace on \\x{}"); | |
2612 | else if (UTF) { | |
b21ed0a9 | 2613 | numlen = 1; /* allow underscores */ |
de35ba6f | 2614 | ender = (UV)scan_hex(p + 1, e - p - 1, &numlen); |
b21ed0a9 GS |
2615 | /* numlen is generous */ |
2616 | if (numlen + len >= 127) { | |
a0ed51b3 LW |
2617 | p--; |
2618 | goto loopdone; | |
2619 | } | |
2620 | p = e + 1; | |
2621 | } | |
2622 | else | |
2623 | FAIL("Can't use \\x{} without 'use utf8' declaration"); | |
2624 | } | |
2625 | else { | |
b21ed0a9 | 2626 | numlen = 0; /* disallow underscores */ |
dff6d3cd | 2627 | ender = (UV)scan_hex(p, 2, &numlen); |
a0ed51b3 LW |
2628 | p += numlen; |
2629 | } | |
a687059c | 2630 | break; |
a0d0e21e LW |
2631 | case 'c': |
2632 | p++; | |
bbce6d69 | 2633 | ender = UCHARAT(p++); |
2634 | ender = toCTRL(ender); | |
a687059c | 2635 | break; |
a0d0e21e LW |
2636 | case '0': case '1': case '2': case '3':case '4': |
2637 | case '5': case '6': case '7': case '8':case '9': | |
2638 | if (*p == '0' || | |
3280af22 | 2639 | (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) { |
b21ed0a9 | 2640 | numlen = 0; /* disallow underscores */ |
dff6d3cd | 2641 | ender = (UV)scan_oct(p, 3, &numlen); |
a0d0e21e LW |
2642 | p += numlen; |
2643 | } | |
2644 | else { | |
2645 | --p; | |
2646 | goto loopdone; | |
a687059c LW |
2647 | } |
2648 | break; | |
a0d0e21e | 2649 | case '\0': |
3280af22 | 2650 | if (p >= PL_regxend) |
a687059c LW |
2651 | FAIL("trailing \\ in regexp"); |
2652 | /* FALL THROUGH */ | |
a0d0e21e | 2653 | default: |
e476b1b5 GS |
2654 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p)) |
2655 | Perl_warner(aTHX_ WARN_REGEXP, | |
1028017a JH |
2656 | "/%.127s/: Unrecognized escape \\%c passed through", |
2657 | PL_regprecomp, | |
2658 | *p); | |
a0ed51b3 | 2659 | goto normal_default; |
a0d0e21e LW |
2660 | } |
2661 | break; | |
a687059c | 2662 | default: |
a0ed51b3 LW |
2663 | normal_default: |
2664 | if ((*p & 0xc0) == 0xc0 && UTF) { | |
dfe13c55 | 2665 | ender = utf8_to_uv((U8*)p, &numlen); |
a0ed51b3 LW |
2666 | p += numlen; |
2667 | } | |
2668 | else | |
2669 | ender = *p++; | |
a0d0e21e | 2670 | break; |
a687059c | 2671 | } |
3280af22 NIS |
2672 | if (PL_regflags & PMf_EXTENDED) |
2673 | p = regwhite(p, PL_regxend); | |
a0ed51b3 LW |
2674 | if (UTF && FOLD) { |
2675 | if (LOC) | |
2676 | ender = toLOWER_LC_uni(ender); | |
2677 | else | |
2678 | ender = toLOWER_uni(ender); | |
2679 | } | |
a0d0e21e LW |
2680 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
2681 | if (len) | |
2682 | p = oldp; | |
a0ed51b3 LW |
2683 | else if (ender >= 0x80 && UTF) { |
2684 | reguni(ender, s, &numlen); | |
2685 | s += numlen; | |
2686 | len += numlen; | |
2687 | } | |
a0d0e21e LW |
2688 | else { |
2689 | len++; | |
cd439c50 | 2690 | REGC(ender, s++); |
a0d0e21e LW |
2691 | } |
2692 | break; | |
a687059c | 2693 | } |
a0ed51b3 LW |
2694 | if (ender >= 0x80 && UTF) { |
2695 | reguni(ender, s, &numlen); | |
2696 | s += numlen; | |
2697 | len += numlen - 1; | |
2698 | } | |
2699 | else | |
cd439c50 | 2700 | REGC(ender, s++); |
a0d0e21e LW |
2701 | } |
2702 | loopdone: | |
3280af22 | 2703 | PL_regcomp_parse = p - 1; |
a0d0e21e LW |
2704 | nextchar(); |
2705 | if (len < 0) | |
2706 | FAIL("internal disaster in regexp"); | |
2707 | if (len > 0) | |
2708 | *flagp |= HASWIDTH; | |
2709 | if (len == 1) | |
2710 | *flagp |= SIMPLE; | |
c277df42 | 2711 | if (!SIZE_ONLY) |
cd439c50 IZ |
2712 | STR_LEN(ret) = len; |
2713 | if (SIZE_ONLY) | |
2714 | PL_regsize += STR_SZ(len); | |
2715 | else | |
2716 | PL_regcode += STR_SZ(len); | |
a687059c | 2717 | } |
a0d0e21e LW |
2718 | break; |
2719 | } | |
a687059c | 2720 | |
a0d0e21e | 2721 | return(ret); |
a687059c LW |
2722 | } |
2723 | ||
873ef191 | 2724 | STATIC char * |
cea2e8a9 | 2725 | S_regwhite(pTHX_ char *p, char *e) |
5b5a24f7 CS |
2726 | { |
2727 | while (p < e) { | |
2728 | if (isSPACE(*p)) | |
2729 | ++p; | |
2730 | else if (*p == '#') { | |
2731 | do { | |
2732 | p++; | |
2733 | } while (p < e && *p != '\n'); | |
2734 | } | |
2735 | else | |
2736 | break; | |
2737 | } | |
2738 | return p; | |
2739 | } | |
2740 | ||
b8c5462f JH |
2741 | /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]]. |
2742 | Character classes ([:foo:]) can also be negated ([:^foo:]). | |
2743 | Returns a named class id (ANYOF_XXX) if successful, -1 otherwise. | |
2744 | Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed, | |
2745 | but trigger warnings because they are currently unimplemented. */ | |
2746 | STATIC I32 | |
cea2e8a9 | 2747 | S_regpposixcc(pTHX_ I32 value) |
620e46c5 | 2748 | { |
11b8faa4 | 2749 | dTHR; |
620e46c5 | 2750 | char *posixcc = 0; |
936ed897 | 2751 | I32 namedclass = OOB_NAMEDCLASS; |
620e46c5 JH |
2752 | |
2753 | if (value == '[' && PL_regcomp_parse + 1 < PL_regxend && | |
2754 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ | |
2755 | (*PL_regcomp_parse == ':' || | |
2756 | *PL_regcomp_parse == '=' || | |
2757 | *PL_regcomp_parse == '.')) { | |
2758 | char c = *PL_regcomp_parse; | |
2759 | char* s = PL_regcomp_parse++; | |
2760 | ||
2761 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c) | |
2762 | PL_regcomp_parse++; | |
2763 | if (PL_regcomp_parse == PL_regxend) | |
2764 | /* Grandfather lone [:, [=, [. */ | |
2765 | PL_regcomp_parse = s; | |
2766 | else { | |
b8c5462f JH |
2767 | char* t = PL_regcomp_parse++; /* skip over the c */ |
2768 | ||
2769 | if (*PL_regcomp_parse == ']') { | |
2770 | PL_regcomp_parse++; /* skip over the ending ] */ | |
2771 | posixcc = s + 1; | |
2772 | if (*s == ':') { | |
2773 | I32 complement = *posixcc == '^' ? *posixcc++ : 0; | |
2774 | I32 skip = 5; /* the most common skip */ | |
2775 | ||
2776 | switch (*posixcc) { | |
2777 | case 'a': | |
2778 | if (strnEQ(posixcc, "alnum", 5)) | |
2779 | namedclass = | |
2780 | complement ? ANYOF_NALNUMC : ANYOF_ALNUMC; | |
2781 | else if (strnEQ(posixcc, "alpha", 5)) | |
2782 | namedclass = | |
2783 | complement ? ANYOF_NALPHA : ANYOF_ALPHA; | |
2784 | else if (strnEQ(posixcc, "ascii", 5)) | |
2785 | namedclass = | |
2786 | complement ? ANYOF_NASCII : ANYOF_ASCII; | |
2787 | break; | |
2788 | case 'c': | |
2789 | if (strnEQ(posixcc, "cntrl", 5)) | |
2790 | namedclass = | |
2791 | complement ? ANYOF_NCNTRL : ANYOF_CNTRL; | |
2792 | break; | |
2793 | case 'd': | |
2794 | if (strnEQ(posixcc, "digit", 5)) | |
2795 | namedclass = | |
2796 | complement ? ANYOF_NDIGIT : ANYOF_DIGIT; | |
2797 | break; | |
2798 | case 'g': | |
2799 | if (strnEQ(posixcc, "graph", 5)) | |
2800 | namedclass = | |
2801 | complement ? ANYOF_NGRAPH : ANYOF_GRAPH; | |
2802 | break; | |
2803 | case 'l': | |
2804 | if (strnEQ(posixcc, "lower", 5)) | |
2805 | namedclass = | |
2806 | complement ? ANYOF_NLOWER : ANYOF_LOWER; | |
2807 | break; | |
2808 | case 'p': | |
2809 | if (strnEQ(posixcc, "print", 5)) | |
2810 | namedclass = | |
2811 | complement ? ANYOF_NPRINT : ANYOF_PRINT; | |
2812 | else if (strnEQ(posixcc, "punct", 5)) | |
2813 | namedclass = | |
2814 | complement ? ANYOF_NPUNCT : ANYOF_PUNCT; | |
2815 | break; | |
2816 | case 's': | |
2817 | if (strnEQ(posixcc, "space", 5)) | |
2818 | namedclass = | |
2819 | complement ? ANYOF_NSPACE : ANYOF_SPACE; | |
cc4319de | 2820 | break; |
b8c5462f JH |
2821 | case 'u': |
2822 | if (strnEQ(posixcc, "upper", 5)) | |
2823 | namedclass = | |
2824 | complement ? ANYOF_NUPPER : ANYOF_UPPER; | |
2825 | break; | |
2826 | case 'w': /* this is not POSIX, this is the Perl \w */ | |
2827 | if (strnEQ(posixcc, "word", 4)) { | |
2828 | namedclass = | |
2829 | complement ? ANYOF_NALNUM : ANYOF_ALNUM; | |
2830 | skip = 4; | |
2831 | } | |
2832 | break; | |
2833 | case 'x': | |
2834 | if (strnEQ(posixcc, "xdigit", 6)) { | |
2835 | namedclass = | |
2836 | complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT; | |
2837 | skip = 6; | |
2838 | } | |
2839 | break; | |
2840 | } | |
ac561586 JH |
2841 | if (namedclass == OOB_NAMEDCLASS || |
2842 | posixcc[skip] != ':' || | |
2843 | posixcc[skip+1] != ']') | |
2844 | Perl_croak(aTHX_ | |
2845 | "Character class [:%.*s:] unknown", | |
2846 | t - s - 1, s + 1); | |
3802429d | 2847 | } else if (!SIZE_ONLY) |
b8c5462f | 2848 | /* [[=foo=]] and [[.foo.]] are still future. */ |
3802429d JF |
2849 | Perl_croak(aTHX_ |
2850 | "Character class syntax [%c %c] is reserved for future extensions", c, c); | |
b8c5462f JH |
2851 | } else { |
2852 | /* Maternal grandfather: | |
2853 | * "[:" ending in ":" but not in ":]" */ | |
767d463e HS |
2854 | PL_regcomp_parse = s; |
2855 | } | |
620e46c5 JH |
2856 | } |
2857 | } | |
2858 | ||
b8c5462f JH |
2859 | return namedclass; |
2860 | } | |
2861 | ||
2862 | STATIC void | |
2863 | S_checkposixcc(pTHX) | |
2864 | { | |
e476b1b5 | 2865 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && |
b8c5462f JH |
2866 | (*PL_regcomp_parse == ':' || |
2867 | *PL_regcomp_parse == '=' || | |
2868 | *PL_regcomp_parse == '.')) { | |
2869 | char *s = PL_regcomp_parse; | |
93733859 | 2870 | char c = *s++; |
b8c5462f JH |
2871 | |
2872 | while(*s && isALNUM(*s)) | |
2873 | s++; | |
2874 | if (*s && c == *s && s[1] == ']') { | |
e476b1b5 | 2875 | Perl_warner(aTHX_ WARN_REGEXP, |
b8c5462f JH |
2876 | "Character class syntax [%c %c] belongs inside character classes", c, c); |
2877 | if (c == '=' || c == '.') | |
3802429d JF |
2878 | Perl_croak(aTHX_ |
2879 | "Character class syntax [%c %c] is reserved for future extensions", c, c); | |
b8c5462f JH |
2880 | } |
2881 | } | |
620e46c5 JH |
2882 | } |
2883 | ||
76e3520e | 2884 | STATIC regnode * |
cea2e8a9 | 2885 | S_regclass(pTHX) |
a687059c | 2886 | { |
5c0ca799 | 2887 | dTHR; |
470c3474 | 2888 | register U32 value; |
b8c5462f | 2889 | register I32 lastvalue = OOB_CHAR8; |
a0d0e21e | 2890 | register I32 range = 0; |
c277df42 | 2891 | register regnode *ret; |
a0d0e21e | 2892 | I32 numlen; |
b8c5462f | 2893 | I32 namedclass; |
73b437c8 | 2894 | char *rangebegin; |
936ed897 | 2895 | bool need_class = 0; |
a0d0e21e | 2896 | |
c277df42 | 2897 | ret = reg_node(ANYOF); |
936ed897 IZ |
2898 | if (SIZE_ONLY) |
2899 | PL_regsize += ANYOF_SKIP; | |
2900 | else { | |
2901 | ret->flags = 0; | |
2902 | ANYOF_BITMAP_ZERO(ret); | |
2903 | PL_regcode += ANYOF_SKIP; | |
2904 | if (FOLD) | |
2905 | ANYOF_FLAGS(ret) |= ANYOF_FOLD; | |
2906 | if (LOC) | |
2907 | ANYOF_FLAGS(ret) |= ANYOF_LOCALE; | |
2908 | } | |
3280af22 NIS |
2909 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2910 | PL_regnaughty++; | |
2911 | PL_regcomp_parse++; | |
c277df42 | 2912 | if (!SIZE_ONLY) |
936ed897 | 2913 | ANYOF_FLAGS(ret) |= ANYOF_INVERT; |
a0d0e21e | 2914 | } |
b8c5462f | 2915 | |
e476b1b5 | 2916 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP)) |
936ed897 | 2917 | checkposixcc(); |
b8c5462f | 2918 | |
3280af22 | 2919 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
a0d0e21e | 2920 | goto skipcond; /* allow 1st char to be ] or - */ |
3280af22 | 2921 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
a0d0e21e | 2922 | skipcond: |
93733859 | 2923 | namedclass = OOB_NAMEDCLASS; |
73b437c8 JH |
2924 | if (!range) |
2925 | rangebegin = PL_regcomp_parse; | |
a0ed51b3 | 2926 | value = UCHARAT(PL_regcomp_parse++); |
620e46c5 | 2927 | if (value == '[') |
b8c5462f | 2928 | namedclass = regpposixcc(value); |
620e46c5 | 2929 | else if (value == '\\') { |
a0ed51b3 | 2930 | value = UCHARAT(PL_regcomp_parse++); |
470c3474 JH |
2931 | /* Some compilers cannot handle switching on 64-bit integer |
2932 | * values, therefore value cannot be an UV. --jhi */ | |
a0ed51b3 | 2933 | switch (value) { |
b8c5462f JH |
2934 | case 'w': namedclass = ANYOF_ALNUM; break; |
2935 | case 'W': namedclass = ANYOF_NALNUM; break; | |
2936 | case 's': namedclass = ANYOF_SPACE; break; | |
2937 | case 'S': namedclass = ANYOF_NSPACE; break; | |
2938 | case 'd': namedclass = ANYOF_DIGIT; break; | |
2939 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
2940 | case 'n': value = '\n'; break; | |
2941 | case 'r': value = '\r'; break; | |
2942 | case 't': value = '\t'; break; | |
2943 | case 'f': value = '\f'; break; | |
2944 | case 'b': value = '\b'; break; | |
0407a77b | 2945 | #ifdef ASCIIish |
b8c5462f JH |
2946 | case 'e': value = '\033'; break; |
2947 | case 'a': value = '\007'; break; | |
0407a77b GS |
2948 | #else |
2949 | case 'e': value = '\047'; break; | |
2950 | case 'a': value = '\057'; break; | |
2951 | #endif | |
b8c5462f | 2952 | case 'x': |
b21ed0a9 | 2953 | numlen = 0; /* disallow underscores */ |
dff6d3cd | 2954 | value = (UV)scan_hex(PL_regcomp_parse, 2, &numlen); |
b8c5462f JH |
2955 | PL_regcomp_parse += numlen; |
2956 | break; | |
2957 | case 'c': | |
2958 | value = UCHARAT(PL_regcomp_parse++); | |
2959 | value = toCTRL(value); | |
2960 | break; | |
2961 | case '0': case '1': case '2': case '3': case '4': | |
2962 | case '5': case '6': case '7': case '8': case '9': | |
b21ed0a9 | 2963 | numlen = 0; /* disallow underscores */ |
dff6d3cd | 2964 | value = (UV)scan_oct(--PL_regcomp_parse, 3, &numlen); |
b8c5462f JH |
2965 | PL_regcomp_parse += numlen; |
2966 | break; | |
1028017a | 2967 | default: |
e476b1b5 GS |
2968 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value)) |
2969 | Perl_warner(aTHX_ WARN_REGEXP, | |
1028017a JH |
2970 | "/%.127s/: Unrecognized escape \\%c in character class passed through", |
2971 | PL_regprecomp, | |
447ff955 | 2972 | (int)value); |
1028017a | 2973 | break; |
b8c5462f JH |
2974 | } |
2975 | } | |
73b437c8 | 2976 | if (namedclass > OOB_NAMEDCLASS) { |
936ed897 IZ |
2977 | if (!need_class && !SIZE_ONLY) |
2978 | ANYOF_CLASS_ZERO(ret); | |
2979 | need_class = 1; | |
73b437c8 JH |
2980 | if (range) { /* a-\d, a-[:digit:] */ |
2981 | if (!SIZE_ONLY) { | |
e476b1b5 GS |
2982 | if (ckWARN(WARN_REGEXP)) |
2983 | Perl_warner(aTHX_ WARN_REGEXP, | |
73b437c8 JH |
2984 | "/%.127s/: false [] range \"%*.*s\" in regexp", |
2985 | PL_regprecomp, | |
2986 | PL_regcomp_parse - rangebegin, | |
2987 | PL_regcomp_parse - rangebegin, | |
2988 | rangebegin); | |
936ed897 IZ |
2989 | ANYOF_BITMAP_SET(ret, lastvalue); |
2990 | ANYOF_BITMAP_SET(ret, '-'); | |
b8c5462f | 2991 | } |
73b437c8 JH |
2992 | range = 0; /* this is not a true range */ |
2993 | } | |
2994 | if (!SIZE_ONLY) { | |
2995 | switch (namedclass) { | |
2996 | case ANYOF_ALNUM: | |
2997 | if (LOC) | |
936ed897 | 2998 | ANYOF_CLASS_SET(ret, ANYOF_ALNUM); |
73b437c8 JH |
2999 | else { |
3000 | for (value = 0; value < 256; value++) | |
3001 | if (isALNUM(value)) | |
936ed897 | 3002 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3003 | } |
3004 | break; | |
3005 | case ANYOF_NALNUM: | |
3006 | if (LOC) | |
936ed897 | 3007 | ANYOF_CLASS_SET(ret, ANYOF_NALNUM); |
73b437c8 JH |
3008 | else { |
3009 | for (value = 0; value < 256; value++) | |
3010 | if (!isALNUM(value)) | |
936ed897 | 3011 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3012 | } |
3013 | break; | |
3014 | case ANYOF_SPACE: | |
3015 | if (LOC) | |
936ed897 | 3016 | ANYOF_CLASS_SET(ret, ANYOF_SPACE); |
73b437c8 JH |
3017 | else { |
3018 | for (value = 0; value < 256; value++) | |
3019 | if (isSPACE(value)) | |
936ed897 | 3020 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3021 | } |
3022 | break; | |
3023 | case ANYOF_NSPACE: | |
3024 | if (LOC) | |
936ed897 | 3025 | ANYOF_CLASS_SET(ret, ANYOF_NSPACE); |
73b437c8 JH |
3026 | else { |
3027 | for (value = 0; value < 256; value++) | |
3028 | if (!isSPACE(value)) | |
936ed897 | 3029 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3030 | } |
3031 | break; | |
3032 | case ANYOF_DIGIT: | |
3033 | if (LOC) | |
936ed897 | 3034 | ANYOF_CLASS_SET(ret, ANYOF_DIGIT); |
73b437c8 JH |
3035 | else { |
3036 | for (value = '0'; value <= '9'; value++) | |
936ed897 | 3037 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3038 | } |
3039 | break; | |
3040 | case ANYOF_NDIGIT: | |
3041 | if (LOC) | |
936ed897 | 3042 | ANYOF_CLASS_SET(ret, ANYOF_NDIGIT); |
73b437c8 JH |
3043 | else { |
3044 | for (value = 0; value < '0'; value++) | |
936ed897 | 3045 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 | 3046 | for (value = '9' + 1; value < 256; value++) |
936ed897 | 3047 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3048 | } |
3049 | break; | |
3050 | case ANYOF_NALNUMC: | |
3051 | if (LOC) | |
936ed897 | 3052 | ANYOF_CLASS_SET(ret, ANYOF_NALNUMC); |
73b437c8 JH |
3053 | else { |
3054 | for (value = 0; value < 256; value++) | |
3055 | if (!isALNUMC(value)) | |
936ed897 | 3056 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3057 | } |
3058 | break; | |
3059 | case ANYOF_ALNUMC: | |
3060 | if (LOC) | |
936ed897 | 3061 | ANYOF_CLASS_SET(ret, ANYOF_ALNUMC); |
73b437c8 JH |
3062 | else { |
3063 | for (value = 0; value < 256; value++) | |
3064 | if (isALNUMC(value)) | |
936ed897 | 3065 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3066 | } |
3067 | break; | |
3068 | case ANYOF_ALPHA: | |
3069 | if (LOC) | |
936ed897 | 3070 | ANYOF_CLASS_SET(ret, ANYOF_ALPHA); |
73b437c8 JH |
3071 | else { |
3072 | for (value = 0; value < 256; value++) | |
3073 | if (isALPHA(value)) | |
936ed897 | 3074 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3075 | } |
3076 | break; | |
3077 | case ANYOF_NALPHA: | |
3078 | if (LOC) | |
936ed897 | 3079 | ANYOF_CLASS_SET(ret, ANYOF_NALPHA); |
73b437c8 JH |
3080 | else { |
3081 | for (value = 0; value < 256; value++) | |
3082 | if (!isALPHA(value)) | |
936ed897 | 3083 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3084 | } |
3085 | break; | |
3086 | case ANYOF_ASCII: | |
3087 | if (LOC) | |
936ed897 | 3088 | ANYOF_CLASS_SET(ret, ANYOF_ASCII); |
73b437c8 | 3089 | else { |
a5993822 | 3090 | #ifdef ASCIIish |
73b437c8 | 3091 | for (value = 0; value < 128; value++) |
936ed897 | 3092 | ANYOF_BITMAP_SET(ret, value); |
a5993822 PP |
3093 | #else /* EBCDIC */ |
3094 | for (value = 0; value < 256; value++) | |
3095 | if (isASCII(value)) | |
3096 | ANYOF_BITMAP_SET(ret, value); | |
3097 | #endif /* EBCDIC */ | |
73b437c8 JH |
3098 | } |
3099 | break; | |
3100 | case ANYOF_NASCII: | |
3101 | if (LOC) | |
936ed897 | 3102 | ANYOF_CLASS_SET(ret, ANYOF_NASCII); |
73b437c8 | 3103 | else { |
a5993822 | 3104 | #ifdef ASCIIish |
73b437c8 | 3105 | for (value = 128; value < 256; value++) |
936ed897 | 3106 | ANYOF_BITMAP_SET(ret, value); |
a5993822 PP |
3107 | #else /* EBCDIC */ |
3108 | for (value = 0; value < 256; value++) | |
3109 | if (!isASCII(value)) | |
3110 | ANYOF_BITMAP_SET(ret, value); | |
3111 | #endif /* EBCDIC */ | |
73b437c8 JH |
3112 | } |
3113 | break; | |
3114 | case ANYOF_CNTRL: | |
3115 | if (LOC) | |
936ed897 | 3116 | ANYOF_CLASS_SET(ret, ANYOF_CNTRL); |
73b437c8 JH |
3117 | else { |
3118 | for (value = 0; value < 256; value++) | |
3119 | if (isCNTRL(value)) | |
936ed897 | 3120 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3121 | } |
3122 | lastvalue = OOB_CHAR8; | |
3123 | break; | |
3124 | case ANYOF_NCNTRL: | |
3125 | if (LOC) | |
936ed897 | 3126 | ANYOF_CLASS_SET(ret, ANYOF_NCNTRL); |
73b437c8 JH |
3127 | else { |
3128 | for (value = 0; value < 256; value++) | |
3129 | if (!isCNTRL(value)) | |
936ed897 | 3130 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3131 | } |
3132 | break; | |
3133 | case ANYOF_GRAPH: | |
3134 | if (LOC) | |
936ed897 | 3135 | ANYOF_CLASS_SET(ret, ANYOF_GRAPH); |
73b437c8 JH |
3136 | else { |
3137 | for (value = 0; value < 256; value++) | |
3138 | if (isGRAPH(value)) | |
936ed897 | 3139 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3140 | } |
3141 | break; | |
3142 | case ANYOF_NGRAPH: | |
3143 | if (LOC) | |
936ed897 | 3144 | ANYOF_CLASS_SET(ret, ANYOF_NGRAPH); |
73b437c8 JH |
3145 | else { |
3146 | for (value = 0; value < 256; value++) | |
3147 | if (!isGRAPH(value)) | |
936ed897 | 3148 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3149 | } |
3150 | break; | |
3151 | case ANYOF_LOWER: | |
3152 | if (LOC) | |
936ed897 | 3153 | ANYOF_CLASS_SET(ret, ANYOF_LOWER); |
73b437c8 JH |
3154 | else { |
3155 | for (value = 0; value < 256; value++) | |
3156 | if (isLOWER(value)) | |
936ed897 | 3157 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3158 | } |
3159 | break; | |
3160 | case ANYOF_NLOWER: | |
3161 | if (LOC) | |
936ed897 | 3162 | ANYOF_CLASS_SET(ret, ANYOF_NLOWER); |
73b437c8 JH |
3163 | else { |
3164 | for (value = 0; value < 256; value++) | |
3165 | if (!isLOWER(value)) | |
936ed897 | 3166 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3167 | } |
3168 | break; | |
3169 | case ANYOF_PRINT: | |
3170 | if (LOC) | |
936ed897 | 3171 | ANYOF_CLASS_SET(ret, ANYOF_PRINT); |
73b437c8 JH |
3172 | else { |
3173 | for (value = 0; value < 256; value++) | |
3174 | if (isPRINT(value)) | |
936ed897 | 3175 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3176 | } |
3177 | break; | |
3178 | case ANYOF_NPRINT: | |
3179 | if (LOC) | |
936ed897 | 3180 | ANYOF_CLASS_SET(ret, ANYOF_NPRINT); |
73b437c8 JH |
3181 | else { |
3182 | for (value = 0; value < 256; value++) | |
3183 | if (!isPRINT(value)) | |
936ed897 | 3184 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3185 | } |
3186 | break; | |
3187 | case ANYOF_PUNCT: | |
3188 | if (LOC) | |
936ed897 | 3189 | ANYOF_CLASS_SET(ret, ANYOF_PUNCT); |
73b437c8 JH |
3190 | else { |
3191 | for (value = 0; value < 256; value++) | |
3192 | if (isPUNCT(value)) | |
936ed897 | 3193 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3194 | } |
3195 | break; | |
3196 | case ANYOF_NPUNCT: | |
3197 | if (LOC) | |
936ed897 | 3198 | ANYOF_CLASS_SET(ret, ANYOF_NPUNCT); |
73b437c8 JH |
3199 | else { |
3200 | for (value = 0; value < 256; value++) | |
3201 | if (!isPUNCT(value)) | |
936ed897 | 3202 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3203 | } |
3204 | break; | |
3205 | case ANYOF_UPPER: | |
3206 | if (LOC) | |
936ed897 | 3207 | ANYOF_CLASS_SET(ret, ANYOF_UPPER); |
73b437c8 JH |
3208 | else { |
3209 | for (value = 0; value < 256; value++) | |
3210 | if (isUPPER(value)) | |
936ed897 | 3211 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3212 | } |
3213 | break; | |
3214 | case ANYOF_NUPPER: | |
3215 | if (LOC) | |
936ed897 | 3216 | ANYOF_CLASS_SET(ret, ANYOF_NUPPER); |
73b437c8 JH |
3217 | else { |
3218 | for (value = 0; value < 256; value++) | |
3219 | if (!isUPPER(value)) | |
936ed897 | 3220 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3221 | } |
3222 | break; | |
3223 | case ANYOF_XDIGIT: | |
3224 | if (LOC) | |
936ed897 | 3225 | ANYOF_CLASS_SET(ret, ANYOF_XDIGIT); |
73b437c8 JH |
3226 | else { |
3227 | for (value = 0; value < 256; value++) | |
3228 | if (isXDIGIT(value)) | |
936ed897 | 3229 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3230 | } |
3231 | break; | |
3232 | case ANYOF_NXDIGIT: | |
3233 | if (LOC) | |
936ed897 | 3234 | ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT); |
73b437c8 JH |
3235 | else { |
3236 | for (value = 0; value < 256; value++) | |
3237 | if (!isXDIGIT(value)) | |
936ed897 | 3238 | ANYOF_BITMAP_SET(ret, value); |
73b437c8 JH |
3239 | } |
3240 | break; | |
3241 | default: | |
3242 | FAIL("invalid [::] class in regexp"); | |
3243 | break; | |
b8c5462f | 3244 | } |
b8c5462f | 3245 | if (LOC) |
936ed897 | 3246 | ANYOF_FLAGS(ret) |= ANYOF_CLASS; |
73b437c8 | 3247 | continue; |
a0d0e21e LW |
3248 | } |
3249 | } | |
3250 | if (range) { | |
73b437c8 JH |
3251 | if (lastvalue > value) /* b-a */ { |
3252 | Perl_croak(aTHX_ | |
3253 | "/%.127s/: invalid [] range \"%*.*s\" in regexp", | |
3254 | PL_regprecomp, | |
3255 | PL_regcomp_parse - rangebegin, | |
3256 | PL_regcomp_parse - rangebegin, | |
3257 | rangebegin); | |
3258 | } | |
a0d0e21e LW |
3259 | range = 0; |
3260 | } | |
3261 | else { | |
a0ed51b3 | 3262 | lastvalue = value; |
3280af22 | 3263 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
93733859 | 3264 | PL_regcomp_parse[1] != ']') { |
3280af22 | 3265 | PL_regcomp_parse++; |
73b437c8 | 3266 | if (namedclass > OOB_NAMEDCLASS) { /* \w-, [:word:]- */ |
e476b1b5 GS |
3267 | if (ckWARN(WARN_REGEXP)) |
3268 | Perl_warner(aTHX_ WARN_REGEXP, | |
73b437c8 JH |
3269 | "/%.127s/: false [] range \"%*.*s\" in regexp", |
3270 | PL_regprecomp, | |
3271 | PL_regcomp_parse - rangebegin, | |
3272 | PL_regcomp_parse - rangebegin, | |
3273 | rangebegin); | |
3274 | if (!SIZE_ONLY) | |
936ed897 | 3275 | ANYOF_BITMAP_SET(ret, '-'); |
73b437c8 JH |
3276 | } else |
3277 | range = 1; | |
a0d0e21e LW |
3278 | continue; /* do it next time */ |
3279 | } | |
a687059c | 3280 | } |
93733859 | 3281 | /* now is the next time */ |
ae5c130c | 3282 | if (!SIZE_ONLY) { |
93733859 | 3283 | #ifndef ASCIIish /* EBCDIC, for example. */ |
8ada0baa | 3284 | if ((isLOWER(lastvalue) && isLOWER(value)) || |
c8dba6f3 GS |
3285 | (isUPPER(lastvalue) && isUPPER(value))) |
3286 | { | |
3287 | I32 i; | |
8ada0baa JH |
3288 | if (isLOWER(lastvalue)) { |
3289 | for (i = lastvalue; i <= value; i++) | |
3290 | if (isLOWER(i)) | |
936ed897 | 3291 | ANYOF_BITMAP_SET(ret, i); |
8ada0baa JH |
3292 | } else { |
3293 | for (i = lastvalue; i <= value; i++) | |
3294 | if (isUPPER(i)) | |
936ed897 | 3295 | ANYOF_BITMAP_SET(ret, i); |
8ada0baa JH |
3296 | } |
3297 | } | |
3298 | else | |
3299 | #endif | |
3300 | for ( ; lastvalue <= value; lastvalue++) | |
936ed897 | 3301 | ANYOF_BITMAP_SET(ret, lastvalue); |
8ada0baa | 3302 | } |
93733859 | 3303 | range = 0; |
a0d0e21e | 3304 | } |
936ed897 IZ |
3305 | if (need_class) { |
3306 | if (SIZE_ONLY) | |
3307 | PL_regsize += ANYOF_CLASS_ADD_SKIP; | |
3308 | else | |
3309 | PL_regcode += ANYOF_CLASS_ADD_SKIP; | |
3310 | } | |
ae5c130c | 3311 | /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ |
b8c5462f | 3312 | if (!SIZE_ONLY && |
936ed897 | 3313 | (ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD) { |
a0ed51b3 | 3314 | for (value = 0; value < 256; ++value) { |
936ed897 | 3315 | if (ANYOF_BITMAP_TEST(ret, value)) { |
22c35a8c | 3316 | I32 cf = PL_fold[value]; |
936ed897 | 3317 | ANYOF_BITMAP_SET(ret, cf); |
ae5c130c GS |
3318 | } |
3319 | } | |
936ed897 | 3320 | ANYOF_FLAGS(ret) &= ~ANYOF_FOLD; |
ae5c130c GS |
3321 | } |
3322 | /* optimize inverted simple patterns (e.g. [^a-z]) */ | |
936ed897 | 3323 | if (!SIZE_ONLY && (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) { |
b8c5462f | 3324 | for (value = 0; value < ANYOF_BITMAP_SIZE; ++value) |
936ed897 IZ |
3325 | ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL; |
3326 | ANYOF_FLAGS(ret) = 0; | |
ae5c130c | 3327 | } |
a0d0e21e LW |
3328 | return ret; |
3329 | } | |
3330 | ||
a0ed51b3 | 3331 | STATIC regnode * |
cea2e8a9 | 3332 | S_regclassutf8(pTHX) |
a0ed51b3 | 3333 | { |
b8c5462f | 3334 | dTHR; |
936ed897 | 3335 | register char *e; |
470c3474 | 3336 | register U32 value; |
b8c5462f | 3337 | register U32 lastvalue = OOB_UTF8; |
a0ed51b3 LW |
3338 | register I32 range = 0; |
3339 | register regnode *ret; | |
3340 | I32 numlen; | |
3341 | I32 n; | |
3342 | SV *listsv; | |
3343 | U8 flags = 0; | |
b8c5462f | 3344 | I32 namedclass; |
73b437c8 | 3345 | char *rangebegin; |
a0ed51b3 LW |
3346 | |
3347 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ | |
3348 | PL_regnaughty++; | |
3349 | PL_regcomp_parse++; | |
3350 | if (!SIZE_ONLY) | |
3351 | flags |= ANYOF_INVERT; | |
3352 | } | |
3353 | if (!SIZE_ONLY) { | |
3354 | if (FOLD) | |
3355 | flags |= ANYOF_FOLD; | |
3356 | if (LOC) | |
3357 | flags |= ANYOF_LOCALE; | |
79cb57f6 | 3358 | listsv = newSVpvn("# comment\n",10); |
a0ed51b3 LW |
3359 | } |
3360 | ||
e476b1b5 | 3361 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP)) |
936ed897 | 3362 | checkposixcc(); |
b8c5462f | 3363 | |
a0ed51b3 LW |
3364 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
3365 | goto skipcond; /* allow 1st char to be ] or - */ | |
3366 | ||
3367 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { | |
3368 | skipcond: | |
93733859 | 3369 | namedclass = OOB_NAMEDCLASS; |
73b437c8 JH |
3370 | if (!range) |
3371 | rangebegin = PL_regcomp_parse; | |
dfe13c55 | 3372 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 | 3373 | PL_regcomp_parse += numlen; |
620e46c5 | 3374 | if (value == '[') |
b8c5462f | 3375 | namedclass = regpposixcc(value); |
620e46c5 | 3376 | else if (value == '\\') { |
470c3474 | 3377 | value = (U32)utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 | 3378 | PL_regcomp_parse += numlen; |
470c3474 JH |
3379 | /* Some compilers cannot handle switching on 64-bit integer |
3380 | * values, therefore value cannot be an UV. Yes, this will | |
3381 | * be a problem later if we want switch on Unicode. --jhi */ | |
a0ed51b3 | 3382 | switch (value) { |
b8c5462f JH |
3383 | case 'w': namedclass = ANYOF_ALNUM; break; |
3384 | case 'W': namedclass = ANYOF_NALNUM; break; | |
3385 | case 's': namedclass = ANYOF_SPACE; break; | |
3386 | case 'S': namedclass = ANYOF_NSPACE; break; | |
3387 | case 'd': namedclass = ANYOF_DIGIT; break; | |
3388 | case 'D': namedclass = ANYOF_NDIGIT; break; | |
a0ed51b3 LW |
3389 | case 'p': |
3390 | case 'P': | |
3391 | if (*PL_regcomp_parse == '{') { | |
3392 | e = strchr(PL_regcomp_parse++, '}'); | |
3393 | if (!e) | |
3394 | FAIL("Missing right brace on \\p{}"); | |
3395 | n = e - PL_regcomp_parse; | |
3396 | } | |
3397 | else { | |
3398 | e = PL_regcomp_parse; | |
3399 | n = 1; | |
3400 | } | |
3401 | if (!SIZE_ONLY) { | |
3402 | if (value == 'p') | |
b8c5462f | 3403 | Perl_sv_catpvf(aTHX_ listsv, |
894356b3 | 3404 | "+utf8::%.*s\n", (int)n, PL_regcomp_parse); |
a0ed51b3 | 3405 | else |
cea2e8a9 | 3406 | Perl_sv_catpvf(aTHX_ listsv, |
894356b3 | 3407 | "!utf8::%.*s\n", (int)n, PL_regcomp_parse); |
a0ed51b3 LW |
3408 | } |
3409 | PL_regcomp_parse = e + 1; | |
b8c5462f | 3410 | lastvalue = OOB_UTF8; |
a0ed51b3 | 3411 | continue; |
b8c5462f JH |
3412 | case 'n': value = '\n'; break; |
3413 | case 'r': value = '\r'; break; | |
3414 | case 't': value = '\t'; break; | |
3415 | case 'f': value = '\f'; break; | |
3416 | case 'b': value = '\b'; break; | |
0407a77b | 3417 | #ifdef ASCIIish |
b8c5462f JH |
3418 | case 'e': value = '\033'; break; |
3419 | case 'a': value = '\007'; break; | |
0407a77b GS |
3420 | #else |
3421 | case 'e': value = '\047'; break; | |
3422 | case 'a': value = '\057'; break; | |
3423 | #endif | |
a0ed51b3 LW |
3424 | case 'x': |
3425 | if (*PL_regcomp_parse == '{') { | |
3426 | e = strchr(PL_regcomp_parse++, '}'); | |
3427 | if (!e) | |
3428 | FAIL("Missing right brace on \\x{}"); | |
b21ed0a9 | 3429 | numlen = 1; /* allow underscores */ |
dff6d3cd | 3430 | value = (UV)scan_hex(PL_regcomp_parse, |
b8c5462f JH |
3431 | e - PL_regcomp_parse, |
3432 | &numlen); | |
a0ed51b3 LW |
3433 | PL_regcomp_parse = e + 1; |
3434 | } | |
3435 | else { | |
b21ed0a9 | 3436 | numlen = 0; /* disallow underscores */ |
dff6d3cd | 3437 | value = (UV)scan_hex(PL_regcomp_parse, 2, &numlen); |
a0ed51b3 LW |
3438 | PL_regcomp_parse += numlen; |
3439 | } | |
3440 | break; | |
3441 | case 'c': | |
3442 | value = UCHARAT(PL_regcomp_parse++); | |
3443 | value = toCTRL(value); | |
3444 | break; | |
3445 | case '0': case '1': case '2': case '3': case '4': | |
3446 | case '5': case '6': case '7': case '8': case '9': | |
b21ed0a9 | 3447 | numlen = 0; /* disallow underscores */ |
dff6d3cd | 3448 | value = (UV)scan_oct(--PL_regcomp_parse, 3, &numlen); |
a0ed51b3 LW |
3449 | PL_regcomp_parse += numlen; |
3450 | break; | |
1028017a | 3451 | default: |
e476b1b5 GS |
3452 | if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value)) |
3453 | Perl_warner(aTHX_ WARN_REGEXP, | |
1028017a JH |
3454 | "/%.127s/: Unrecognized escape \\%c in character class passed through", |
3455 | PL_regprecomp, | |
447ff955 | 3456 | (int)value); |
1028017a | 3457 | break; |
a0ed51b3 LW |
3458 | } |
3459 | } | |
73b437c8 JH |
3460 | if (namedclass > OOB_NAMEDCLASS) { |
3461 | if (range) { /* a-\d, a-[:digit:] */ | |
3462 | if (!SIZE_ONLY) { | |
e476b1b5 GS |
3463 | if (ckWARN(WARN_REGEXP)) |
3464 | Perl_warner(aTHX_ WARN_REGEXP, | |
73b437c8 JH |
3465 | "/%.127s/: false [] range \"%*.*s\" in regexp", |
3466 | PL_regprecomp, | |
3467 | PL_regcomp_parse - rangebegin, | |
3468 | PL_regcomp_parse - rangebegin, | |
3469 | rangebegin); | |
3470 | Perl_sv_catpvf(aTHX_ listsv, | |
3471 | /* 0x002D is Unicode for '-' */ | |
3472 | "%04"UVxf"\n002D\n", (UV)lastvalue); | |
3473 | } | |
3474 | range = 0; | |
3475 | } | |
3476 | if (!SIZE_ONLY) { | |
3477 | switch (namedclass) { | |
3478 | case ANYOF_ALNUM: | |
3479 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n"); break; | |
3480 | case ANYOF_NALNUM: | |
3481 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n"); break; | |
3482 | case ANYOF_ALNUMC: | |
3483 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n"); break; | |
3484 | case ANYOF_NALNUMC: | |
3485 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n"); break; | |
3486 | case ANYOF_ALPHA: | |
3487 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n"); break; | |
3488 | case ANYOF_NALPHA: | |
3489 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n"); break; | |
3490 | case ANYOF_ASCII: | |
3491 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n"); break; | |
3492 | case ANYOF_NASCII: | |
3493 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n"); break; | |
3494 | case ANYOF_CNTRL: | |
3495 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n"); break; | |
3496 | case ANYOF_NCNTRL: | |
3497 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n"); break; | |
3498 | case ANYOF_GRAPH: | |
3499 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n"); break; | |
3500 | case ANYOF_NGRAPH: | |
3501 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n"); break; | |
3502 | case ANYOF_DIGIT: | |
3503 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n"); break; | |
3504 | case ANYOF_NDIGIT: | |
3505 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n"); break; | |
3506 | case ANYOF_LOWER: | |
3507 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n"); break; | |
3508 | case ANYOF_NLOWER: | |
3509 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n"); break; | |
3510 | case ANYOF_PRINT: | |
3511 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n"); break; | |
3512 | case ANYOF_NPRINT: | |
3513 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n"); break; | |
3514 | case ANYOF_PUNCT: | |
3515 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n"); break; | |
3516 | case ANYOF_NPUNCT: | |
3517 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n"); break; | |
3518 | case ANYOF_SPACE: | |
3519 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n"); break; | |
3520 | case ANYOF_NSPACE: | |
3521 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n"); break; | |
3522 | case ANYOF_UPPER: | |
3523 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n"); break; | |
3524 | case ANYOF_NUPPER: | |
3525 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n"); break; | |
3526 | case ANYOF_XDIGIT: | |
3527 | Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n"); break; | |
3528 | case ANYOF_NXDIGIT: | |
3529 | Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n"); break; | |
3530 | } | |
3531 | continue; | |
b8c5462f JH |
3532 | } |
3533 | } | |
b8c5462f | 3534 | if (range) { |
73b437c8 JH |
3535 | if (lastvalue > value) { /* b-a */ |
3536 | Perl_croak(aTHX_ | |
3537 | "/%.127s/: invalid [] range \"%*.*s\" in regexp", | |
3538 | PL_regprecomp, | |
3539 | PL_regcomp_parse - rangebegin, | |
3540 | PL_regcomp_parse - rangebegin, | |
3541 | rangebegin); | |
3542 | } | |
a0ed51b3 LW |
3543 | range = 0; |
3544 | } | |
3545 | else { | |
3546 | lastvalue = value; | |
3547 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && | |
93733859 | 3548 | PL_regcomp_parse[1] != ']') { |
a0ed51b3 | 3549 | PL_regcomp_parse++; |
73b437c8 | 3550 | if (namedclass > OOB_NAMEDCLASS) { /* \w-, [:word:]- */ |
e476b1b5 GS |
3551 | if (ckWARN(WARN_REGEXP)) |
3552 | Perl_warner(aTHX_ WARN_REGEXP, | |
73b437c8 JH |
3553 | "/%.127s/: false [] range \"%*.*s\" in regexp", |
3554 | PL_regprecomp, | |
3555 | PL_regcomp_parse - rangebegin, | |
3556 | PL_regcomp_parse - rangebegin, | |
3557 | rangebegin); | |
3558 | if (!SIZE_ONLY) | |
3559 | Perl_sv_catpvf(aTHX_ listsv, | |
3560 | /* 0x002D is Unicode for '-' */ | |
3561 | "002D\n"); | |
3562 | } else | |
3563 | range = 1; | |
a0ed51b3 LW |
3564 | continue; /* do it next time */ |
3565 | } | |
93733859 JH |
3566 | } |
3567 | /* now is the next time */ | |
93733859 | 3568 | if (!SIZE_ONLY) |
73b437c8 JH |
3569 | Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n", |
3570 | (UV)lastvalue, (UV)value); | |
93733859 | 3571 | range = 0; |
a0ed51b3 | 3572 | } |
a0ed51b3 LW |
3573 | |
3574 | ret = reganode(ANYOFUTF8, 0); | |
3575 | ||
3576 | if (!SIZE_ONLY) { | |
3577 | SV *rv = swash_init("utf8", "", listsv, 1, 0); | |
3578 | SvREFCNT_dec(listsv); | |
3579 | n = add_data(1,"s"); | |
3580 | PL_regcomp_rx->data->data[n] = (void*)rv; | |
3581 | ARG1_SET(ret, flags); | |
3582 | ARG2_SET(ret, n); | |
3583 | } | |
3584 | ||
3585 | return ret; | |
3586 | } | |
3587 | ||
76e3520e | 3588 | STATIC char* |
cea2e8a9 | 3589 | S_nextchar(pTHX) |
a0d0e21e | 3590 | { |
5c0ca799 | 3591 | dTHR; |
3280af22 | 3592 | char* retval = PL_regcomp_parse++; |
a0d0e21e | 3593 | |
4633a7c4 | 3594 | for (;;) { |
3280af22 NIS |
3595 | if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' && |
3596 | PL_regcomp_parse[2] == '#') { | |
3597 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') | |
3598 | PL_regcomp_parse++; | |
3599 | PL_regcomp_parse++; | |
4633a7c4 LW |
3600 | continue; |
3601 | } | |
3280af22 NIS |
3602 | if (PL_regflags & PMf_EXTENDED) { |
3603 | if (isSPACE(*PL_regcomp_parse)) { | |
3604 | PL_regcomp_parse++; | |
748a9306 LW |
3605 | continue; |
3606 | } | |
3280af22 NIS |
3607 | else if (*PL_regcomp_parse == '#') { |
3608 | while (*PL_regcomp_parse && *PL_regcomp_parse != '\n') | |
3609 | PL_regcomp_parse++; | |
3610 | PL_regcomp_parse++; | |
748a9306 LW |
3611 | continue; |
3612 | } | |
748a9306 | 3613 | } |
4633a7c4 | 3614 | return retval; |
a0d0e21e | 3615 | } |
a687059c LW |
3616 | } |
3617 | ||
3618 | /* | |
c277df42 | 3619 | - reg_node - emit a node |
a0d0e21e | 3620 | */ |
76e3520e | 3621 | STATIC regnode * /* Location. */ |
cea2e8a9 | 3622 | S_reg_node(pTHX_ U8 op) |
a687059c | 3623 | { |
5c0ca799 | 3624 | dTHR; |
c277df42 IZ |
3625 | register regnode *ret; |
3626 | register regnode *ptr; | |
a687059c | 3627 | |
3280af22 | 3628 | ret = PL_regcode; |
c277df42 | 3629 | if (SIZE_ONLY) { |
6b88bc9c | 3630 | SIZE_ALIGN(PL_regsize); |
3280af22 | 3631 | PL_regsize += 1; |
a0d0e21e LW |
3632 | return(ret); |
3633 | } | |
a687059c | 3634 | |
c277df42 | 3635 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 3636 | ptr = ret; |
c277df42 | 3637 | FILL_ADVANCE_NODE(ptr, op); |
3280af22 | 3638 | PL_regcode = ptr; |
a687059c | 3639 | |
a0d0e21e | 3640 | return(ret); |
a687059c LW |
3641 | } |
3642 | ||
3643 | /* | |
a0d0e21e LW |
3644 | - reganode - emit a node with an argument |
3645 | */ | |
76e3520e | 3646 | STATIC regnode * /* Location. */ |
cea2e8a9 | 3647 | S_reganode(pTHX_ U8 op, U32 arg) |
fe14fcc3 | 3648 | { |
5c0ca799 | 3649 | dTHR; |
c277df42 IZ |
3650 | register regnode *ret; |
3651 | register regnode *ptr; | |
fe14fcc3 | 3652 | |
3280af22 | 3653 | ret = PL_regcode; |
c277df42 | 3654 | if (SIZE_ONLY) { |
6b88bc9c | 3655 | SIZE_ALIGN(PL_regsize); |
3280af22 | 3656 | PL_regsize += 2; |
a0d0e21e LW |
3657 | return(ret); |
3658 | } | |
fe14fcc3 | 3659 | |
c277df42 | 3660 | NODE_ALIGN_FILL(ret); |
a0d0e21e | 3661 | ptr = ret; |
c277df42 | 3662 | FILL_ADVANCE_NODE_ARG(ptr, op, arg); |
3280af22 | 3663 | PL_regcode = ptr; |
fe14fcc3 | 3664 | |
a0d0e21e | 3665 | return(ret); |
fe14fcc3 LW |
3666 | } |
3667 | ||
3668 | /* | |
cd439c50 | 3669 | - reguni - emit (if appropriate) a Unicode character |
a0ed51b3 LW |
3670 | */ |
3671 | STATIC void | |
cea2e8a9 | 3672 | S_reguni(pTHX_ UV uv, char* s, I32* lenp) |
a0ed51b3 | 3673 | { |
c485e607 | 3674 | dTHR; |
a0ed51b3 | 3675 | if (SIZE_ONLY) { |
806e7201 | 3676 | U8 tmpbuf[UTF8_MAXLEN]; |
a0ed51b3 LW |
3677 | *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf; |
3678 | } | |
3679 | else | |
dfe13c55 | 3680 | *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s; |
a0ed51b3 LW |
3681 | |
3682 | } | |
3683 | ||
3684 | /* | |
a0d0e21e LW |
3685 | - reginsert - insert an operator in front of already-emitted operand |
3686 | * | |
3687 | * Means relocating the operand. | |
3688 | */ | |
76e3520e | 3689 | STATIC void |
cea2e8a9 | 3690 | S_reginsert(pTHX_ U8 op, regnode *opnd) |
a687059c | 3691 | { |
5c0ca799 | 3692 | dTHR; |
c277df42 IZ |
3693 | register regnode *src; |
3694 | register regnode *dst; | |
3695 | register regnode *place; | |
3696 | register int offset = regarglen[(U8)op]; | |
3697 | ||
22c35a8c | 3698 | /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */ |
c277df42 IZ |
3699 | |
3700 | if (SIZE_ONLY) { | |
3280af22 | 3701 | PL_regsize += NODE_STEP_REGNODE + offset; |
a0d0e21e LW |
3702 | return; |
3703 | } | |
a687059c | 3704 | |
3280af22 NIS |
3705 | src = PL_regcode; |
3706 | PL_regcode += NODE_STEP_REGNODE + offset; | |
3707 | dst = PL_regcode; | |
a0d0e21e | 3708 | while (src > opnd) |
c277df42 | 3709 | StructCopy(--src, --dst, regnode); |
a0d0e21e LW |
3710 | |
3711 | place = opnd; /* Op node, where operand used to be. */ | |
c277df42 IZ |
3712 | src = NEXTOPER(place); |
3713 | FILL_ADVANCE_NODE(place, op); | |
3714 | Zero(src, offset, regnode); | |
a687059c LW |
3715 | } |
3716 | ||
3717 | /* | |
c277df42 | 3718 | - regtail - set the next-pointer at the end of a node chain of p to val. |
a0d0e21e | 3719 | */ |
76e3520e | 3720 | STATIC void |
cea2e8a9 | 3721 | S_regtail(pTHX_ regnode *p, regnode *val) |
a687059c | 3722 | { |
5c0ca799 | 3723 | dTHR; |
c277df42 IZ |
3724 | register regnode *scan; |
3725 | register regnode *temp; | |
a0d0e21e | 3726 | |
c277df42 | 3727 | if (SIZE_ONLY) |
a0d0e21e LW |
3728 | return; |
3729 | ||
3730 | /* Find last node. */ | |
3731 | scan = p; | |
3732 | for (;;) { | |
3733 | temp = regnext(scan); | |
3734 | if (temp == NULL) | |
3735 | break; | |
3736 | scan = temp; | |
3737 | } | |
a687059c | 3738 | |
c277df42 IZ |
3739 | if (reg_off_by_arg[OP(scan)]) { |
3740 | ARG_SET(scan, val - scan); | |
a0ed51b3 LW |
3741 | } |
3742 | else { | |
c277df42 IZ |
3743 | NEXT_OFF(scan) = val - scan; |
3744 | } | |
a687059c LW |
3745 | } |
3746 | ||
3747 | /* | |
a0d0e21e LW |
3748 | - regoptail - regtail on operand of first argument; nop if operandless |
3749 | */ | |
76e3520e | 3750 | STATIC void |
cea2e8a9 | 3751 | S_regoptail(pTHX_ regnode *p, regnode *val) |
a687059c | 3752 | { |
5c0ca799 | 3753 | dTHR; |
a0d0e21e | 3754 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
c277df42 IZ |
3755 | if (p == NULL || SIZE_ONLY) |
3756 | return; | |
22c35a8c | 3757 | if (PL_regkind[(U8)OP(p)] == BRANCH) { |
c277df42 | 3758 | regtail(NEXTOPER(p), val); |
a0ed51b3 | 3759 | } |
22c35a8c | 3760 | else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) { |
c277df42 | 3761 | regtail(NEXTOPER(NEXTOPER(p)), val); |
a0ed51b3 LW |
3762 | } |
3763 | else | |
a0d0e21e | 3764 | return; |
a687059c LW |
3765 | } |
3766 | ||
3767 | /* | |
3768 | - regcurly - a little FSA that accepts {\d+,?\d*} | |
3769 | */ | |
79072805 | 3770 | STATIC I32 |
cea2e8a9 | 3771 | S_regcurly(pTHX_ register char *s) |
a687059c LW |
3772 | { |
3773 | if (*s++ != '{') | |
3774 | return FALSE; | |
f0fcb552 | 3775 | if (!isDIGIT(*s)) |
a687059c | 3776 | return FALSE; |
f0fcb552 | 3777 | while (isDIGIT(*s)) |
a687059c LW |
3778 | s++; |
3779 | if (*s == ',') | |
3780 | s++; | |
f0fcb552 | 3781 | while (isDIGIT(*s)) |
a687059c LW |
3782 | s++; |
3783 | if (*s != '}') | |
3784 | return FALSE; | |
3785 | return TRUE; | |
3786 | } | |
3787 | ||
a687059c | 3788 | |
76e3520e | 3789 | STATIC regnode * |
cea2e8a9 | 3790 | S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l) |
c277df42 | 3791 | { |
35ff7856 | 3792 | #ifdef DEBUGGING |
f248d071 | 3793 | register U8 op = EXACT; /* Arbitrary non-END op. */ |
155aba94 | 3794 | register regnode *next; |
c277df42 IZ |
3795 | |
3796 | while (op != END && (!last || node < last)) { | |
3797 | /* While that wasn't END last time... */ | |
3798 | ||
3799 | NODE_ALIGN(node); | |
3800 | op = OP(node); | |
3801 | if (op == CLOSE) | |
3802 | l--; | |
3803 | next = regnext(node); | |
3804 | /* Where, what. */ | |
3805 | if (OP(node) == OPTIMIZED) | |
3806 | goto after_print; | |
3807 | regprop(sv, node); | |
b900a521 | 3808 | PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start), |
f1dbda3d | 3809 | (int)(2*l + 1), "", SvPVX(sv)); |
c277df42 IZ |
3810 | if (next == NULL) /* Next ptr. */ |
3811 | PerlIO_printf(Perl_debug_log, "(0)"); | |
3812 | else | |
b900a521 | 3813 | PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start)); |
c277df42 IZ |
3814 | (void)PerlIO_putc(Perl_debug_log, '\n'); |
3815 | after_print: | |
22c35a8c | 3816 | if (PL_regkind[(U8)op] == BRANCHJ) { |
c277df42 IZ |
3817 | register regnode *nnode = (OP(next) == LONGJMP |
3818 | ? regnext(next) | |
3819 | : next); | |
3820 | if (last && nnode > last) | |
3821 | nnode = last; | |
3822 | node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1); | |
a0ed51b3 | 3823 | } |
22c35a8c | 3824 | else if (PL_regkind[(U8)op] == BRANCH) { |
c277df42 | 3825 | node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1); |
a0ed51b3 LW |
3826 | } |
3827 | else if ( op == CURLY) { /* `next' might be very big: optimizer */ | |
c277df42 IZ |
3828 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3829 | NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1); | |
a0ed51b3 | 3830 | } |
22c35a8c | 3831 | else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) { |
c277df42 IZ |
3832 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
3833 | next, sv, l + 1); | |
a0ed51b3 LW |
3834 | } |
3835 | else if ( op == PLUS || op == STAR) { | |
c277df42 | 3836 | node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1); |
a0ed51b3 LW |
3837 | } |
3838 | else if (op == ANYOF) { | |
c277df42 | 3839 | node = NEXTOPER(node); |
936ed897 | 3840 | node += ANYOF_SKIP; |
a0ed51b3 | 3841 | } |
22c35a8c | 3842 | else if (PL_regkind[(U8)op] == EXACT) { |
c277df42 | 3843 | /* Literal string, where present. */ |
cd439c50 | 3844 | node += NODE_SZ_STR(node) - 1; |
c277df42 | 3845 | node = NEXTOPER(node); |
a0ed51b3 LW |
3846 | } |
3847 | else { | |
c277df42 IZ |
3848 | node = NEXTOPER(node); |
3849 | node += regarglen[(U8)op]; | |
3850 | } | |
3851 | if (op == CURLYX || op == OPEN) | |
3852 | l++; | |
3853 | else if (op == WHILEM) | |
3854 | l--; | |
3855 | } | |
17c3b450 | 3856 | #endif /* DEBUGGING */ |
c277df42 IZ |
3857 | return node; |
3858 | } | |
3859 | ||
a687059c | 3860 | /* |
fd181c75 | 3861 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c LW |
3862 | */ |
3863 | void | |
864dbfa3 | 3864 | Perl_regdump(pTHX_ regexp *r) |
a687059c | 3865 | { |
35ff7856 | 3866 | #ifdef DEBUGGING |
5c0ca799 | 3867 | dTHR; |
46fc3d4c | 3868 | SV *sv = sv_newmortal(); |
a687059c | 3869 | |
c277df42 | 3870 | (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0); |
a0d0e21e LW |
3871 | |
3872 | /* Header fields of interest. */ | |
c277df42 | 3873 | if (r->anchored_substr) |
7b0972df JH |
3874 | PerlIO_printf(Perl_debug_log, |
3875 | "anchored `%s%.*s%s'%s at %"IVdf" ", | |
3280af22 | 3876 | PL_colors[0], |
7b0972df | 3877 | (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)), |
c277df42 | 3878 | SvPVX(r->anchored_substr), |
3280af22 | 3879 | PL_colors[1], |
c277df42 | 3880 | SvTAIL(r->anchored_substr) ? "$" : "", |
7b0972df | 3881 | (IV)r->anchored_offset); |
c277df42 | 3882 | if (r->float_substr) |
7b0972df JH |
3883 | PerlIO_printf(Perl_debug_log, |
3884 | "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ", | |
3280af22 | 3885 | PL_colors[0], |
7b0972df | 3886 | (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)), |
2c2d71f5 | 3887 | SvPVX(r->float_substr), |
3280af22 | 3888 | PL_colors[1], |
c277df42 | 3889 | SvTAIL(r->float_substr) ? "$" : "", |
7b0972df | 3890 | (IV)r->float_min_offset, (UV)r->float_max_offset); |
c277df42 IZ |
3891 | if (r->check_substr) |
3892 | PerlIO_printf(Perl_debug_log, | |
3893 | r->check_substr == r->float_substr | |
3894 | ? "(checking floating" : "(checking anchored"); | |
3895 | if (r->reganch & ROPT_NOSCAN) | |
3896 | PerlIO_printf(Perl_debug_log, " noscan"); | |
3897 | if (r->reganch & ROPT_CHECK_ALL) | |
3898 | PerlIO_printf(Perl_debug_log, " isall"); | |
3899 | if (r->check_substr) | |
3900 | PerlIO_printf(Perl_debug_log, ") "); | |
3901 | ||
46fc3d4c | 3902 | if (r->regstclass) { |
3903 | regprop(sv, r->regstclass); | |
3904 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv)); | |
3905 | } | |
774d564b | 3906 | if (r->reganch & ROPT_ANCH) { |
3907 | PerlIO_printf(Perl_debug_log, "anchored"); | |
3908 | if (r->reganch & ROPT_ANCH_BOL) | |
3909 | PerlIO_printf(Perl_debug_log, "(BOL)"); | |
c277df42 IZ |
3910 | if (r->reganch & ROPT_ANCH_MBOL) |
3911 | PerlIO_printf(Perl_debug_log, "(MBOL)"); | |
cad2e5aa JH |
3912 | if (r->reganch & ROPT_ANCH_SBOL) |
3913 | PerlIO_printf(Perl_debug_log, "(SBOL)"); | |
774d564b | 3914 | if (r->reganch & ROPT_ANCH_GPOS) |
3915 | PerlIO_printf(Perl_debug_log, "(GPOS)"); | |
3916 | PerlIO_putc(Perl_debug_log, ' '); | |
3917 | } | |
c277df42 IZ |
3918 | if (r->reganch & ROPT_GPOS_SEEN) |
3919 | PerlIO_printf(Perl_debug_log, "GPOS "); | |
a0d0e21e | 3920 | if (r->reganch & ROPT_SKIP) |
760ac839 | 3921 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e | 3922 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 | 3923 | PerlIO_printf(Perl_debug_log, "implicit "); |
760ac839 | 3924 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
ce862d02 IZ |
3925 | if (r->reganch & ROPT_EVAL_SEEN) |
3926 | PerlIO_printf(Perl_debug_log, "with eval "); | |
760ac839 | 3927 | PerlIO_printf(Perl_debug_log, "\n"); |
17c3b450 | 3928 | #endif /* DEBUGGING */ |
a687059c LW |
3929 | } |
3930 | ||
653099ff GS |
3931 | STATIC void |
3932 | S_put_byte(pTHX_ SV *sv, int c) | |
3933 | { | |
3934 | if (c <= ' ' || c == 127 || c == 255) | |
3935 | Perl_sv_catpvf(aTHX_ sv, "\\%o", c); | |
3936 | else if (c == '-' || c == ']' || c == '\\' || c == '^') | |
3937 | Perl_sv_catpvf(aTHX_ sv, "\\%c", c); | |
3938 | else | |
3939 | Perl_sv_catpvf(aTHX_ sv, "%c", c); | |
3940 | } | |
3941 | ||
a687059c | 3942 | /* |
a0d0e21e LW |
3943 | - regprop - printable representation of opcode |
3944 | */ | |
46fc3d4c | 3945 | void |
864dbfa3 | 3946 | Perl_regprop(pTHX_ SV *sv, regnode *o) |
a687059c | 3947 | { |
35ff7856 | 3948 | #ifdef DEBUGGING |
5c0ca799 | 3949 | dTHR; |
9b155405 | 3950 | register int k; |
a0d0e21e | 3951 | |
54dc92de | 3952 | sv_setpvn(sv, "", 0); |
9b155405 | 3953 | if (OP(o) >= reg_num) /* regnode.type is unsigned */ |
a0d0e21e | 3954 | FAIL("corrupted regexp opcode"); |
9b155405 IZ |
3955 | sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */ |
3956 | ||
3957 | k = PL_regkind[(U8)OP(o)]; | |
3958 | ||
3959 | if (k == EXACT) | |
7821416a | 3960 | Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0], |
cd439c50 | 3961 | STR_LEN(o), STRING(o), PL_colors[1]); |
9b155405 IZ |
3962 | else if (k == CURLY) { |
3963 | if (OP(o) == CURLYM || OP(o) == CURLYN) | |
cea2e8a9 GS |
3964 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */ |
3965 | Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o)); | |
a0d0e21e | 3966 | } |
2c2d71f5 JH |
3967 | else if (k == WHILEM && o->flags) /* Ordinal/of */ |
3968 | Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4); | |
9b155405 | 3969 | else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP ) |
894356b3 | 3970 | Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o)); /* Parenth number */ |
9b155405 | 3971 | else if (k == LOGICAL) |
04ebc1ab | 3972 | Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* 2: embedded, otherwise 1 */ |
653099ff GS |
3973 | else if (k == ANYOF) { |
3974 | int i, rangestart = -1; | |
3975 | const char * const out[] = { /* Should be syncronized with | |
3976 | a table in regcomp.h */ | |
3977 | "\\w", | |
3978 | "\\W", | |
3979 | "\\s", | |
3980 | "\\S", | |
3981 | "\\d", | |
3982 | "\\D", | |
3983 | "[:alnum:]", | |
3984 | "[:^alnum:]", | |
3985 | "[:alpha:]", | |
3986 | "[:^alpha:]", | |
3987 | "[:ascii:]", | |
3988 | "[:^ascii:]", | |
3989 | "[:ctrl:]", | |
3990 | "[:^ctrl:]", | |
3991 | "[:graph:]", | |
3992 | "[:^graph:]", | |
3993 | "[:lower:]", | |
3994 | "[:^lower:]", | |
3995 | "[:print:]", | |
3996 | "[:^print:]", | |
3997 | "[:punct:]", | |
3998 | "[:^punct:]", | |
3999 | "[:upper:]", | |
4000 | "[:!upper:]", | |
4001 | "[:xdigit:]", | |
4002 | "[:^xdigit:]" | |
4003 | }; | |
4004 | ||
4005 | if (o->flags & ANYOF_LOCALE) | |
4006 | sv_catpv(sv, "{loc}"); | |
4007 | if (o->flags & ANYOF_FOLD) | |
4008 | sv_catpv(sv, "{i}"); | |
4009 | Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]); | |
4010 | if (o->flags & ANYOF_INVERT) | |
4011 | sv_catpv(sv, "^"); | |
4012 | for (i = 0; i <= 256; i++) { | |
4013 | if (i < 256 && ANYOF_BITMAP_TEST(o,i)) { | |
4014 | if (rangestart == -1) | |
4015 | rangestart = i; | |
4016 | } else if (rangestart != -1) { | |
4017 | if (i <= rangestart + 3) | |
4018 | for (; rangestart < i; rangestart++) | |
4019 | put_byte(sv, rangestart); | |
4020 | else { | |
4021 | put_byte(sv, rangestart); | |
4022 | sv_catpv(sv, "-"); | |
4023 | put_byte(sv, i - 1); | |
4024 | } | |
4025 | rangestart = -1; | |
4026 | } | |
4027 | } | |
4028 | if (o->flags & ANYOF_CLASS) | |
4029 | for (i = 0; i < sizeof(out)/sizeof(char*); i++) | |
4030 | if (ANYOF_CLASS_TEST(o,i)) | |
4031 | sv_catpv(sv, out[i]); | |
4032 | Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]); | |
4033 | } | |
9b155405 | 4034 | else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH)) |
cea2e8a9 | 4035 | Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags); |
17c3b450 | 4036 | #endif /* DEBUGGING */ |
35ff7856 | 4037 | } |
a687059c | 4038 | |
cad2e5aa JH |
4039 | SV * |
4040 | Perl_re_intuit_string(pTHX_ regexp *prog) | |
4041 | { /* Assume that RE_INTUIT is set */ | |
4042 | DEBUG_r( | |
4043 | { STRLEN n_a; | |
4044 | char *s = SvPV(prog->check_substr,n_a); | |
4045 | ||
4046 | if (!PL_colorset) reginitcolors(); | |
4047 | PerlIO_printf(Perl_debug_log, | |
4048 | "%sUsing REx substr:%s `%s%.60s%s%s'\n", | |
4049 | PL_colors[4],PL_colors[5],PL_colors[0], | |
4050 | s, | |
4051 | PL_colors[1], | |
4052 | (strlen(s) > 60 ? "..." : "")); | |
4053 | } ); | |
4054 | ||
4055 | return prog->check_substr; | |
4056 | } | |
4057 | ||
2b69d0c2 | 4058 | void |
864dbfa3 | 4059 | Perl_pregfree(pTHX_ struct regexp *r) |
a687059c | 4060 | { |
5c0ca799 | 4061 | dTHR; |
adac82c7 | 4062 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
7821416a IZ |
4063 | |
4064 | if (!r || (--r->refcnt > 0)) | |
4065 | return; | |
cad2e5aa JH |
4066 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
4067 | "%sFreeing REx:%s `%s%.60s%s%s'\n", | |
4068 | PL_colors[4],PL_colors[5],PL_colors[0], | |
4069 | r->precomp, | |
4070 | PL_colors[1], | |
4071 | (strlen(r->precomp) > 60 ? "..." : ""))); | |
4072 | ||
c277df42 | 4073 | if (r->precomp) |
a0d0e21e | 4074 | Safefree(r->precomp); |
cf93c79d IZ |
4075 | if (RX_MATCH_COPIED(r)) |
4076 | Safefree(r->subbeg); | |
a193d654 GS |
4077 | if (r->substrs) { |
4078 | if (r->anchored_substr) | |
4079 | SvREFCNT_dec(r->anchored_substr); | |
4080 | if (r->float_substr) | |
4081 | SvREFCNT_dec(r->float_substr); | |
2779dcf1 | 4082 | Safefree(r->substrs); |
a193d654 | 4083 | } |
c277df42 IZ |
4084 | if (r->data) { |
4085 | int n = r->data->count; | |
dfad63ad HS |
4086 | AV* new_comppad = NULL; |
4087 | AV* old_comppad; | |
4088 | SV** old_curpad; | |
4089 | ||
c277df42 IZ |
4090 | while (--n >= 0) { |
4091 | switch (r->data->what[n]) { | |
4092 | case 's': | |
4093 | SvREFCNT_dec((SV*)r->data->data[n]); | |
4094 | break; | |
653099ff GS |
4095 | case 'f': |
4096 | Safefree(r->data->data[n]); | |
4097 | break; | |
dfad63ad HS |
4098 | case 'p': |
4099 | new_comppad = (AV*)r->data->data[n]; | |
4100 | break; | |
c277df42 | 4101 | case 'o': |
dfad63ad | 4102 | if (new_comppad == NULL) |
cea2e8a9 | 4103 | Perl_croak(aTHX_ "panic: pregfree comppad"); |
dfad63ad HS |
4104 | old_comppad = PL_comppad; |
4105 | old_curpad = PL_curpad; | |
4106 | PL_comppad = new_comppad; | |
4107 | PL_curpad = AvARRAY(new_comppad); | |
c277df42 | 4108 | op_free((OP_4tree*)r->data->data[n]); |
dfad63ad HS |
4109 | PL_comppad = old_comppad; |
4110 | PL_curpad = old_curpad; | |
4111 | SvREFCNT_dec((SV*)new_comppad); | |
4112 | new_comppad = NULL; | |
c277df42 IZ |
4113 | break; |
4114 | case 'n': | |
4115 | break; | |
4116 | default: | |
4117 | FAIL2("panic: regfree data code '%c'", r->data->what[n]); | |
4118 | } | |
4119 | } | |
4120 | Safefree(r->data->what); | |
4121 | Safefree(r->data); | |
a0d0e21e LW |
4122 | } |
4123 | Safefree(r->startp); | |
4124 | Safefree(r->endp); | |
4125 | Safefree(r); | |
a687059c | 4126 | } |
c277df42 IZ |
4127 | |
4128 | /* | |
4129 | - regnext - dig the "next" pointer out of a node | |
4130 | * | |
4131 | * [Note, when REGALIGN is defined there are two places in regmatch() | |
4132 | * that bypass this code for speed.] | |
4133 | */ | |
4134 | regnode * | |
864dbfa3 | 4135 | Perl_regnext(pTHX_ register regnode *p) |
c277df42 | 4136 | { |
5c0ca799 | 4137 | dTHR; |
c277df42 IZ |
4138 | register I32 offset; |
4139 | ||
3280af22 | 4140 | if (p == &PL_regdummy) |
c277df42 IZ |
4141 | return(NULL); |
4142 | ||
4143 | offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p)); | |
4144 | if (offset == 0) | |
4145 | return(NULL); | |
4146 | ||
c277df42 | 4147 | return(p+offset); |
c277df42 IZ |
4148 | } |
4149 | ||
01f988be | 4150 | STATIC void |
cea2e8a9 | 4151 | S_re_croak2(pTHX_ const char* pat1,const char* pat2,...) |
c277df42 IZ |
4152 | { |
4153 | va_list args; | |
4154 | STRLEN l1 = strlen(pat1); | |
4155 | STRLEN l2 = strlen(pat2); | |
4156 | char buf[512]; | |
06bf62c7 | 4157 | SV *msv; |
c277df42 IZ |
4158 | char *message; |
4159 | ||
4160 | if (l1 > 510) | |
4161 | l1 = 510; | |
4162 | if (l1 + l2 > 510) | |
4163 | l2 = 510 - l1; | |
4164 | Copy(pat1, buf, l1 , char); | |
4165 | Copy(pat2, buf + l1, l2 , char); | |
3b818b81 GS |
4166 | buf[l1 + l2] = '\n'; |
4167 | buf[l1 + l2 + 1] = '\0'; | |
8736538c AS |
4168 | #ifdef I_STDARG |
4169 | /* ANSI variant takes additional second argument */ | |
c277df42 | 4170 | va_start(args, pat2); |
8736538c AS |
4171 | #else |
4172 | va_start(args); | |
4173 | #endif | |
5a844595 | 4174 | msv = vmess(buf, &args); |
c277df42 | 4175 | va_end(args); |
06bf62c7 | 4176 | message = SvPV(msv,l1); |
c277df42 IZ |
4177 | if (l1 > 512) |
4178 | l1 = 512; | |
4179 | Copy(message, buf, l1 , char); | |
4180 | buf[l1] = '\0'; /* Overwrite \n */ | |
cea2e8a9 | 4181 | Perl_croak(aTHX_ "%s", buf); |
c277df42 | 4182 | } |
a0ed51b3 LW |
4183 | |
4184 | /* XXX Here's a total kludge. But we need to re-enter for swash routines. */ | |
4185 | ||
4186 | void | |
864dbfa3 | 4187 | Perl_save_re_context(pTHX) |
c485e607 NIS |
4188 | { |
4189 | dTHR; | |
a0ed51b3 LW |
4190 | SAVEPPTR(PL_bostr); |
4191 | SAVEPPTR(PL_regprecomp); /* uncompiled string. */ | |
4192 | SAVEI32(PL_regnpar); /* () count. */ | |
4193 | SAVEI32(PL_regsize); /* Code size. */ | |
4194 | SAVEI16(PL_regflags); /* are we folding, multilining? */ | |
4195 | SAVEPPTR(PL_reginput); /* String-input pointer. */ | |
4196 | SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */ | |
4197 | SAVEPPTR(PL_regeol); /* End of input, for $ check. */ | |
7766f137 GS |
4198 | SAVEVPTR(PL_regstartp); /* Pointer to startp array. */ |
4199 | SAVEVPTR(PL_regendp); /* Ditto for endp. */ | |
4200 | SAVEVPTR(PL_reglastparen); /* Similarly for lastparen. */ | |
a0ed51b3 | 4201 | SAVEPPTR(PL_regtill); /* How far we are required to go. */ |
e8347627 | 4202 | SAVEI8(PL_regprev); /* char before regbol, \n if none */ |
7766f137 | 4203 | SAVEVPTR(PL_reg_start_tmp); /* from regexec.c */ |
a0ed51b3 LW |
4204 | PL_reg_start_tmp = 0; |
4205 | SAVEFREEPV(PL_reg_start_tmp); | |
4206 | SAVEI32(PL_reg_start_tmpl); /* from regexec.c */ | |
4207 | PL_reg_start_tmpl = 0; | |
7766f137 | 4208 | SAVEVPTR(PL_regdata); |
a0ed51b3 LW |
4209 | SAVEI32(PL_reg_flags); /* from regexec.c */ |
4210 | SAVEI32(PL_reg_eval_set); /* from regexec.c */ | |
4211 | SAVEI32(PL_regnarrate); /* from regexec.c */ | |
7766f137 | 4212 | SAVEVPTR(PL_regprogram); /* from regexec.c */ |
a0ed51b3 | 4213 | SAVEINT(PL_regindent); /* from regexec.c */ |
7766f137 GS |
4214 | SAVEVPTR(PL_regcc); /* from regexec.c */ |
4215 | SAVEVPTR(PL_curcop); | |
4216 | SAVEVPTR(PL_regcomp_rx); /* from regcomp.c */ | |
a0ed51b3 LW |
4217 | SAVEI32(PL_regseen); /* from regcomp.c */ |
4218 | SAVEI32(PL_regsawback); /* Did we see \1, ...? */ | |
4219 | SAVEI32(PL_regnaughty); /* How bad is this pattern? */ | |
7766f137 | 4220 | SAVEVPTR(PL_regcode); /* Code-emit pointer; ®dummy = don't */ |
a0ed51b3 LW |
4221 | SAVEPPTR(PL_regxend); /* End of input for compile */ |
4222 | SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */ | |
7766f137 GS |
4223 | SAVEVPTR(PL_reg_call_cc); /* from regexec.c */ |
4224 | SAVEVPTR(PL_reg_re); /* from regexec.c */ | |
54b6e2fa IZ |
4225 | SAVEPPTR(PL_reg_ganch); /* from regexec.c */ |
4226 | SAVESPTR(PL_reg_sv); /* from regexec.c */ | |
7766f137 | 4227 | SAVEVPTR(PL_reg_magic); /* from regexec.c */ |
54b6e2fa | 4228 | SAVEI32(PL_reg_oldpos); /* from regexec.c */ |
7766f137 GS |
4229 | SAVEVPTR(PL_reg_oldcurpm); /* from regexec.c */ |
4230 | SAVEVPTR(PL_reg_curpm); /* from regexec.c */ | |
54b6e2fa IZ |
4231 | #ifdef DEBUGGING |
4232 | SAVEPPTR(PL_reg_starttry); /* from regexec.c */ | |
4233 | #endif | |
a0ed51b3 | 4234 | } |
51371543 GS |
4235 | |
4236 | #ifdef PERL_OBJECT | |
51371543 GS |
4237 | #include "XSUB.h" |
4238 | #undef this | |
4239 | #define this pPerl | |
4240 | #endif | |
4241 | ||
4242 | static void | |
4243 | clear_re(pTHXo_ void *r) | |
4244 | { | |
4245 | ReREFCNT_dec((regexp *)r); | |
4246 | } | |
4247 |