Commit | Line | Data |
a0d0e21e |
1 | /* regcomp.c |
2 | */ |
3 | |
4 | /* |
5 | * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee |
6 | */ |
7 | |
a687059c |
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 |
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 |
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 */ |
28 | # ifndef DEBUGGING |
29 | # define DEBUGGING |
30 | # endif |
31 | #endif |
32 | |
33 | #ifdef PERL_IN_XSUB_RE |
d06ea78c |
34 | /* We *really* need to overwrite these symbols: */ |
56953603 |
35 | # define Perl_pregcomp my_regcomp |
36 | # define Perl_regdump my_regdump |
37 | # define Perl_regprop my_regprop |
d06ea78c |
38 | /* *These* symbols are masked to allow static link. */ |
39 | # define Perl_pregfree my_regfree |
40 | # define Perl_regnext my_regnext |
f0b8d043 |
41 | # define Perl_save_re_context my_save_re_context |
56953603 |
42 | #endif |
43 | |
f0fcb552 |
44 | /*SUPPRESS 112*/ |
a687059c |
45 | /* |
e50aee73 |
46 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c |
47 | * |
48 | * Copyright (c) 1986 by University of Toronto. |
49 | * Written by Henry Spencer. Not derived from licensed software. |
50 | * |
51 | * Permission is granted to anyone to use this software for any |
52 | * purpose on any computer system, and to redistribute it freely, |
53 | * subject to the following restrictions: |
54 | * |
55 | * 1. The author is not responsible for the consequences of use of |
56 | * this software, no matter how awful, even if they arise |
57 | * from defects in it. |
58 | * |
59 | * 2. The origin of this software must not be misrepresented, either |
60 | * by explicit claim or by omission. |
61 | * |
62 | * 3. Altered versions must be plainly marked as such, and must not |
63 | * be misrepresented as being the original software. |
64 | * |
65 | * |
66 | **** Alterations to Henry's code are... |
67 | **** |
a0ed51b3 |
68 | **** Copyright (c) 1991-1998, Larry Wall |
a687059c |
69 | **** |
9ef589d8 |
70 | **** You may distribute under the terms of either the GNU General Public |
71 | **** License or the Artistic License, as specified in the README file. |
72 | |
a687059c |
73 | * |
74 | * Beware that some of this code is subtly aware of the way operator |
75 | * precedence is structured in regular expressions. Serious changes in |
76 | * regular-expression syntax might require a total rethink. |
77 | */ |
78 | #include "EXTERN.h" |
79 | #include "perl.h" |
d06ea78c |
80 | |
b9d5759e |
81 | #ifndef PERL_IN_XSUB_RE |
d06ea78c |
82 | # include "INTERN.h" |
83 | #endif |
c277df42 |
84 | |
85 | #define REG_COMP_C |
a687059c |
86 | #include "regcomp.h" |
87 | |
d4cce5f1 |
88 | #ifdef op |
11343788 |
89 | #undef op |
d4cce5f1 |
90 | #endif /* op */ |
11343788 |
91 | |
fe14fcc3 |
92 | #ifdef MSDOS |
93 | # if defined(BUGGY_MSC6) |
94 | /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */ |
95 | # pragma optimize("a",off) |
96 | /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/ |
97 | # pragma optimize("w",on ) |
98 | # endif /* BUGGY_MSC6 */ |
99 | #endif /* MSDOS */ |
100 | |
a687059c |
101 | #ifndef STATIC |
102 | #define STATIC static |
103 | #endif |
104 | |
105 | #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?') |
106 | #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \ |
107 | ((*s) == '{' && regcurly(s))) |
2b69d0c2 |
108 | #ifdef atarist |
109 | #define PERL_META "^$.[()|?+*\\" |
110 | #else |
a687059c |
111 | #define META "^$.[()|?+*\\" |
2b69d0c2 |
112 | #endif |
a687059c |
113 | |
35c8bce7 |
114 | #ifdef SPSTART |
115 | #undef SPSTART /* dratted cpp namespace... */ |
116 | #endif |
a687059c |
117 | /* |
118 | * Flags to be passed up and down. |
119 | */ |
a687059c |
120 | #define WORST 0 /* Worst case. */ |
821b33a5 |
121 | #define HASWIDTH 0x1 /* Known to match non-null strings. */ |
a0d0e21e |
122 | #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ |
123 | #define SPSTART 0x4 /* Starts with * or +. */ |
124 | #define TRYAGAIN 0x8 /* Weeded out a declaration. */ |
a687059c |
125 | |
126 | /* |
e50aee73 |
127 | * Forward declarations for pregcomp()'s friends. |
a687059c |
128 | */ |
a0d0e21e |
129 | |
76e3520e |
130 | #ifndef PERL_OBJECT |
c277df42 |
131 | static regnode *reg _((I32, I32 *)); |
132 | static regnode *reganode _((U8, U32)); |
133 | static regnode *regatom _((I32 *)); |
134 | static regnode *regbranch _((I32 *, I32)); |
135 | static void regc _((U8, char *)); |
a0ed51b3 |
136 | static void reguni _((UV, char *, I32*)); |
c277df42 |
137 | static regnode *regclass _((void)); |
a0ed51b3 |
138 | static regnode *regclassutf8 _((void)); |
a0d0e21e |
139 | STATIC I32 regcurly _((char *)); |
c277df42 |
140 | static regnode *reg_node _((U8)); |
141 | static regnode *regpiece _((I32 *)); |
142 | static void reginsert _((U8, regnode *)); |
143 | static void regoptail _((regnode *, regnode *)); |
c277df42 |
144 | static void regtail _((regnode *, regnode *)); |
873ef191 |
145 | static char* regwhite _((char *, char *)); |
a0d0e21e |
146 | static char* nextchar _((void)); |
3bd495df |
147 | static void re_croak2 _((const char* pat1,const char* pat2,...)) __attribute__((noreturn)); |
76e3520e |
148 | #endif |
a687059c |
149 | |
c277df42 |
150 | /* Length of a variant. */ |
151 | |
76e3520e |
152 | #ifndef PERL_OBJECT |
c277df42 |
153 | typedef struct { |
154 | I32 len_min; |
155 | I32 len_delta; |
a0ed51b3 |
156 | I32 pos_min; /* CC */ |
157 | I32 pos_delta; /* CC */ |
c277df42 |
158 | SV *last_found; |
159 | I32 last_end; /* min value, <0 unless valid. */ |
a0ed51b3 |
160 | I32 last_start_min; /* CC */ |
161 | I32 last_start_max; /* CC */ |
c277df42 |
162 | SV **longest; /* Either &l_fixed, or &l_float. */ |
163 | SV *longest_fixed; |
a0ed51b3 |
164 | I32 offset_fixed; /* CC */ |
c277df42 |
165 | SV *longest_float; |
a0ed51b3 |
166 | I32 offset_float_min; /* CC */ |
167 | I32 offset_float_max; /* CC */ |
c277df42 |
168 | I32 flags; |
169 | } scan_data_t; |
76e3520e |
170 | #endif |
c277df42 |
171 | |
172 | static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; |
173 | |
174 | #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL) |
175 | #define SF_BEFORE_SEOL 0x1 |
176 | #define SF_BEFORE_MEOL 0x2 |
177 | #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL) |
178 | #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL) |
179 | |
09b7f37c |
180 | #ifdef NO_UNARY_PLUS |
181 | # define SF_FIX_SHIFT_EOL (0+2) |
182 | # define SF_FL_SHIFT_EOL (0+4) |
183 | #else |
184 | # define SF_FIX_SHIFT_EOL (+2) |
185 | # define SF_FL_SHIFT_EOL (+4) |
186 | #endif |
c277df42 |
187 | |
188 | #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL) |
189 | #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL) |
190 | |
191 | #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL) |
192 | #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */ |
193 | #define SF_IS_INF 0x40 |
194 | #define SF_HAS_PAR 0x80 |
195 | #define SF_IN_PAR 0x100 |
196 | #define SF_HAS_EVAL 0x200 |
4bfe0158 |
197 | #define SCF_DO_SUBSTR 0x400 |
c277df42 |
198 | |
a0ed51b3 |
199 | #define RF_utf8 8 |
200 | #define UTF (PL_reg_flags & RF_utf8) |
201 | #define LOC (PL_regflags & PMf_LOCALE) |
202 | #define FOLD (PL_regflags & PMf_FOLD) |
203 | |
204 | #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) |
205 | #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) |
206 | |
76e3520e |
207 | STATIC void |
c277df42 |
208 | scan_commit(scan_data_t *data) |
209 | { |
c485e607 |
210 | dTHR; |
a0ed51b3 |
211 | STRLEN l = CHR_SVLEN(data->last_found); |
212 | STRLEN old_l = CHR_SVLEN(*data->longest); |
c277df42 |
213 | |
214 | if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) { |
215 | sv_setsv(*data->longest, data->last_found); |
216 | if (*data->longest == data->longest_fixed) { |
217 | data->offset_fixed = l ? data->last_start_min : data->pos_min; |
218 | if (data->flags & SF_BEFORE_EOL) |
219 | data->flags |
220 | |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL); |
221 | else |
222 | data->flags &= ~SF_FIX_BEFORE_EOL; |
a0ed51b3 |
223 | } |
224 | else { |
c277df42 |
225 | data->offset_float_min = l ? data->last_start_min : data->pos_min; |
226 | data->offset_float_max = (l |
227 | ? data->last_start_max |
228 | : data->pos_min + data->pos_delta); |
229 | if (data->flags & SF_BEFORE_EOL) |
230 | data->flags |
231 | |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL); |
232 | else |
233 | data->flags &= ~SF_FL_BEFORE_EOL; |
234 | } |
235 | } |
236 | SvCUR_set(data->last_found, 0); |
237 | data->last_end = -1; |
238 | data->flags &= ~SF_BEFORE_EOL; |
239 | } |
240 | |
c277df42 |
241 | /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set |
242 | to the position after last scanned or to NULL. */ |
243 | |
76e3520e |
244 | STATIC I32 |
c277df42 |
245 | study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags) |
246 | /* scanp: Start here (read-write). */ |
247 | /* deltap: Write maxlen-minlen here. */ |
248 | /* last: Stop before this one. */ |
249 | { |
5c0ca799 |
250 | dTHR; |
c277df42 |
251 | I32 min = 0, pars = 0, code; |
252 | regnode *scan = *scanp, *next; |
253 | I32 delta = 0; |
254 | int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF); |
aca2d497 |
255 | int is_inf_internal = 0; /* The studied chunk is infinite */ |
c277df42 |
256 | I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0; |
257 | scan_data_t data_fake; |
258 | |
259 | while (scan && OP(scan) != END && scan < last) { |
260 | /* Peephole optimizer: */ |
261 | |
262 | if (regkind[(U8)OP(scan)] == EXACT) { |
263 | regnode *n = regnext(scan); |
264 | U32 stringok = 1; |
265 | #ifdef DEBUGGING |
266 | regnode *stop = scan; |
267 | #endif |
268 | |
269 | next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; |
270 | /* Skip NOTHING, merge EXACT*. */ |
271 | while (n && |
272 | ( regkind[(U8)OP(n)] == NOTHING || |
273 | (stringok && (OP(n) == OP(scan)))) |
274 | && NEXT_OFF(n) |
275 | && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) { |
276 | if (OP(n) == TAIL || n > next) |
277 | stringok = 0; |
278 | if (regkind[(U8)OP(n)] == NOTHING) { |
279 | NEXT_OFF(scan) += NEXT_OFF(n); |
280 | next = n + NODE_STEP_REGNODE; |
281 | #ifdef DEBUGGING |
282 | if (stringok) |
283 | stop = n; |
284 | #endif |
285 | n = regnext(n); |
a0ed51b3 |
286 | } |
287 | else { |
c277df42 |
288 | int oldl = *OPERAND(scan); |
289 | regnode *nnext = regnext(n); |
290 | |
291 | if (oldl + *OPERAND(n) > U8_MAX) |
292 | break; |
293 | NEXT_OFF(scan) += NEXT_OFF(n); |
294 | *OPERAND(scan) += *OPERAND(n); |
295 | next = n + (*OPERAND(n) + 2 - 1)/sizeof(regnode) + 2; |
296 | /* Now we can overwrite *n : */ |
297 | Move(OPERAND(n) + 1, OPERAND(scan) + oldl + 1, |
298 | *OPERAND(n) + 1, char); |
299 | #ifdef DEBUGGING |
300 | if (stringok) |
301 | stop = next - 1; |
302 | #endif |
303 | n = nnext; |
304 | } |
305 | } |
306 | #ifdef DEBUGGING |
307 | /* Allow dumping */ |
308 | n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; |
309 | while (n <= stop) { |
ca04da08 |
310 | /* Purify reports a benign UMR here sometimes, because we |
311 | * don't initialize the OP() slot of a node when that node |
312 | * is occupied by just the trailing null of the string in |
313 | * an EXACT node */ |
c277df42 |
314 | if (regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) { |
315 | OP(n) = OPTIMIZED; |
316 | NEXT_OFF(n) = 0; |
317 | } |
318 | n++; |
319 | } |
320 | #endif |
321 | |
322 | } |
323 | if (OP(scan) != CURLYX) { |
048cfca1 |
324 | int max = (reg_off_by_arg[OP(scan)] |
325 | ? I32_MAX |
326 | /* I32 may be smaller than U16 on CRAYs! */ |
327 | : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX)); |
c277df42 |
328 | int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan)); |
329 | int noff; |
330 | regnode *n = scan; |
331 | |
332 | /* Skip NOTHING and LONGJMP. */ |
333 | while ((n = regnext(n)) |
334 | && ((regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n))) |
335 | || ((OP(n) == LONGJMP) && (noff = ARG(n)))) |
336 | && off + noff < max) |
337 | off += noff; |
338 | if (reg_off_by_arg[OP(scan)]) |
339 | ARG(scan) = off; |
340 | else |
341 | NEXT_OFF(scan) = off; |
342 | } |
343 | if (OP(scan) == BRANCH || OP(scan) == BRANCHJ |
344 | || OP(scan) == IFTHEN || OP(scan) == SUSPEND) { |
345 | next = regnext(scan); |
346 | code = OP(scan); |
347 | |
348 | if (OP(next) == code || code == IFTHEN || code == SUSPEND) { |
349 | I32 max1 = 0, min1 = I32_MAX, num = 0; |
350 | |
351 | if (flags & SCF_DO_SUBSTR) |
352 | scan_commit(data); |
353 | while (OP(scan) == code) { |
354 | I32 deltanext, minnext; |
355 | |
356 | num++; |
357 | data_fake.flags = 0; |
358 | next = regnext(scan); |
359 | scan = NEXTOPER(scan); |
360 | if (code != BRANCH) |
361 | scan = NEXTOPER(scan); |
362 | /* We suppose the run is continuous, last=next...*/ |
363 | minnext = study_chunk(&scan, &deltanext, next, |
364 | &data_fake, 0); |
365 | if (min1 > minnext) |
366 | min1 = minnext; |
367 | if (max1 < minnext + deltanext) |
368 | max1 = minnext + deltanext; |
369 | if (deltanext == I32_MAX) |
aca2d497 |
370 | is_inf = is_inf_internal = 1; |
c277df42 |
371 | scan = next; |
372 | if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) |
373 | pars++; |
405ff068 |
374 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 |
375 | data->flags |= SF_HAS_EVAL; |
376 | if (code == SUSPEND) |
377 | break; |
378 | } |
379 | if (code == IFTHEN && num < 2) /* Empty ELSE branch */ |
380 | min1 = 0; |
381 | if (flags & SCF_DO_SUBSTR) { |
382 | data->pos_min += min1; |
383 | data->pos_delta += max1 - min1; |
384 | if (max1 != min1 || is_inf) |
385 | data->longest = &(data->longest_float); |
386 | } |
387 | min += min1; |
388 | delta += max1 - min1; |
a0ed51b3 |
389 | } |
390 | else if (code == BRANCHJ) /* single branch is optimized. */ |
c277df42 |
391 | scan = NEXTOPER(NEXTOPER(scan)); |
392 | else /* single branch is optimized. */ |
393 | scan = NEXTOPER(scan); |
394 | continue; |
a0ed51b3 |
395 | } |
396 | else if (OP(scan) == EXACT) { |
397 | I32 l = *OPERAND(scan); |
398 | if (UTF) { |
399 | unsigned char *s = (unsigned char *)(OPERAND(scan)+1); |
400 | unsigned char *e = s + l; |
401 | I32 newl = 0; |
402 | while (s < e) { |
403 | newl++; |
404 | s += UTF8SKIP(s); |
405 | } |
406 | l = newl; |
407 | } |
408 | min += l; |
c277df42 |
409 | if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */ |
c277df42 |
410 | /* The code below prefers earlier match for fixed |
411 | offset, later match for variable offset. */ |
412 | if (data->last_end == -1) { /* Update the start info. */ |
413 | data->last_start_min = data->pos_min; |
414 | data->last_start_max = is_inf |
415 | ? I32_MAX : data->pos_min + data->pos_delta; |
416 | } |
a0ed51b3 |
417 | sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), *OPERAND(scan)); |
c277df42 |
418 | data->last_end = data->pos_min + l; |
419 | data->pos_min += l; /* As in the first entry. */ |
420 | data->flags &= ~SF_BEFORE_EOL; |
421 | } |
a0ed51b3 |
422 | } |
423 | else if (regkind[(U8)OP(scan)] == EXACT) { |
424 | I32 l = *OPERAND(scan); |
c277df42 |
425 | if (flags & SCF_DO_SUBSTR) |
426 | scan_commit(data); |
a0ed51b3 |
427 | if (UTF) { |
428 | unsigned char *s = (unsigned char *)(OPERAND(scan)+1); |
429 | unsigned char *e = s + l; |
430 | I32 newl = 0; |
431 | while (s < e) { |
432 | newl++; |
433 | s += UTF8SKIP(s); |
434 | } |
435 | l = newl; |
436 | } |
437 | min += l; |
c277df42 |
438 | if (data && (flags & SCF_DO_SUBSTR)) |
a0ed51b3 |
439 | data->pos_min += l; |
440 | } |
441 | else if (strchr(varies,OP(scan))) { |
c277df42 |
442 | I32 mincount, maxcount, minnext, deltanext, pos_before, fl; |
443 | regnode *oscan = scan; |
444 | |
445 | switch (regkind[(U8)OP(scan)]) { |
446 | case WHILEM: |
447 | scan = NEXTOPER(scan); |
448 | goto finish; |
449 | case PLUS: |
450 | if (flags & SCF_DO_SUBSTR) { |
451 | next = NEXTOPER(scan); |
452 | if (OP(next) == EXACT) { |
453 | mincount = 1; |
454 | maxcount = REG_INFTY; |
455 | next = regnext(scan); |
456 | scan = NEXTOPER(scan); |
457 | goto do_curly; |
458 | } |
459 | } |
460 | if (flags & SCF_DO_SUBSTR) |
461 | data->pos_min++; |
462 | min++; |
463 | /* Fall through. */ |
464 | case STAR: |
aca2d497 |
465 | is_inf = is_inf_internal = 1; |
c277df42 |
466 | scan = regnext(scan); |
467 | if (flags & SCF_DO_SUBSTR) { |
468 | scan_commit(data); |
469 | data->longest = &(data->longest_float); |
470 | } |
471 | goto optimize_curly_tail; |
472 | case CURLY: |
473 | mincount = ARG1(scan); |
474 | maxcount = ARG2(scan); |
475 | next = regnext(scan); |
476 | scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; |
477 | do_curly: |
478 | if (flags & SCF_DO_SUBSTR) { |
479 | if (mincount == 0) scan_commit(data); |
480 | pos_before = data->pos_min; |
481 | } |
482 | if (data) { |
483 | fl = data->flags; |
484 | data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL); |
485 | if (is_inf) |
486 | data->flags |= SF_IS_INF; |
487 | } |
488 | /* This will finish on WHILEM, setting scan, or on NULL: */ |
489 | minnext = study_chunk(&scan, &deltanext, last, data, |
490 | mincount == 0 |
491 | ? (flags & ~SCF_DO_SUBSTR) : flags); |
492 | if (!scan) /* It was not CURLYX, but CURLY. */ |
493 | scan = next; |
599cee73 |
494 | if (ckWARN(WARN_UNSAFE) && (minnext + deltanext == 0) |
821b33a5 |
495 | && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) |
496 | && maxcount <= 10000) /* Complement check for big count */ |
599cee73 |
497 | warner(WARN_UNSAFE, "Strange *+?{} on zero-length expression"); |
c277df42 |
498 | min += minnext * mincount; |
aca2d497 |
499 | is_inf_internal |= (maxcount == REG_INFTY |
500 | && (minnext + deltanext) > 0 |
501 | || deltanext == I32_MAX); |
502 | is_inf |= is_inf_internal; |
c277df42 |
503 | delta += (minnext + deltanext) * maxcount - minnext * mincount; |
504 | |
505 | /* Try powerful optimization CURLYX => CURLYN. */ |
c277df42 |
506 | if ( OP(oscan) == CURLYX && data |
507 | && data->flags & SF_IN_PAR |
508 | && !(data->flags & SF_HAS_EVAL) |
509 | && !deltanext && minnext == 1 ) { |
510 | /* Try to optimize to CURLYN. */ |
511 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; |
512 | regnode *nxt1 = nxt, *nxt2; |
513 | |
514 | /* Skip open. */ |
515 | nxt = regnext(nxt); |
516 | if (!strchr(simple,OP(nxt)) |
517 | && !(regkind[(U8)OP(nxt)] == EXACT |
518 | && *OPERAND(nxt) == 1)) |
519 | goto nogo; |
520 | nxt2 = nxt; |
521 | nxt = regnext(nxt); |
522 | if (OP(nxt) != CLOSE) |
523 | goto nogo; |
524 | /* Now we know that nxt2 is the only contents: */ |
525 | oscan->flags = ARG(nxt); |
526 | OP(oscan) = CURLYN; |
527 | OP(nxt1) = NOTHING; /* was OPEN. */ |
528 | #ifdef DEBUGGING |
529 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ |
530 | NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */ |
531 | NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */ |
532 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ |
533 | OP(nxt + 1) = OPTIMIZED; /* was count. */ |
534 | NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */ |
535 | #endif |
536 | } |
c277df42 |
537 | nogo: |
538 | |
539 | /* Try optimization CURLYX => CURLYM. */ |
540 | if ( OP(oscan) == CURLYX && data |
c277df42 |
541 | && !(data->flags & SF_HAS_PAR) |
c277df42 |
542 | && !(data->flags & SF_HAS_EVAL) |
543 | && !deltanext ) { |
544 | /* XXXX How to optimize if data == 0? */ |
545 | /* Optimize to a simpler form. */ |
546 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */ |
547 | regnode *nxt2; |
548 | |
549 | OP(oscan) = CURLYM; |
550 | while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/ |
551 | && (OP(nxt2) != WHILEM)) |
552 | nxt = nxt2; |
553 | OP(nxt2) = SUCCEED; /* Whas WHILEM */ |
c277df42 |
554 | /* Need to optimize away parenths. */ |
555 | if (data->flags & SF_IN_PAR) { |
556 | /* Set the parenth number. */ |
557 | regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/ |
558 | |
559 | if (OP(nxt) != CLOSE) |
560 | FAIL("panic opt close"); |
561 | oscan->flags = ARG(nxt); |
562 | OP(nxt1) = OPTIMIZED; /* was OPEN. */ |
563 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ |
564 | #ifdef DEBUGGING |
565 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ |
566 | OP(nxt + 1) = OPTIMIZED; /* was count. */ |
567 | NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */ |
568 | NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */ |
569 | #endif |
570 | #if 0 |
571 | while ( nxt1 && (OP(nxt1) != WHILEM)) { |
572 | regnode *nnxt = regnext(nxt1); |
573 | |
574 | if (nnxt == nxt) { |
575 | if (reg_off_by_arg[OP(nxt1)]) |
576 | ARG_SET(nxt1, nxt2 - nxt1); |
577 | else if (nxt2 - nxt1 < U16_MAX) |
578 | NEXT_OFF(nxt1) = nxt2 - nxt1; |
579 | else |
580 | OP(nxt) = NOTHING; /* Cannot beautify */ |
581 | } |
582 | nxt1 = nnxt; |
583 | } |
584 | #endif |
585 | /* Optimize again: */ |
586 | study_chunk(&nxt1, &deltanext, nxt, NULL, 0); |
a0ed51b3 |
587 | } |
588 | else |
c277df42 |
589 | oscan->flags = 0; |
c277df42 |
590 | } |
591 | if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) |
592 | pars++; |
593 | if (flags & SCF_DO_SUBSTR) { |
594 | SV *last_str = Nullsv; |
595 | int counted = mincount != 0; |
596 | |
597 | if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */ |
598 | I32 b = pos_before >= data->last_start_min |
599 | ? pos_before : data->last_start_min; |
600 | STRLEN l; |
601 | char *s = SvPV(data->last_found, l); |
a0ed51b3 |
602 | I32 old = b - data->last_start_min; |
603 | |
604 | if (UTF) |
605 | old = utf8_hop((U8*)s, old) - (U8*)s; |
c277df42 |
606 | |
a0ed51b3 |
607 | l -= old; |
c277df42 |
608 | /* Get the added string: */ |
a0ed51b3 |
609 | last_str = newSVpv(s + old, l); |
c277df42 |
610 | if (deltanext == 0 && pos_before == b) { |
611 | /* What was added is a constant string */ |
612 | if (mincount > 1) { |
613 | SvGROW(last_str, (mincount * l) + 1); |
614 | repeatcpy(SvPVX(last_str) + l, |
615 | SvPVX(last_str), l, mincount - 1); |
616 | SvCUR(last_str) *= mincount; |
617 | /* Add additional parts. */ |
618 | SvCUR_set(data->last_found, |
619 | SvCUR(data->last_found) - l); |
620 | sv_catsv(data->last_found, last_str); |
621 | data->last_end += l * (mincount - 1); |
622 | } |
623 | } |
624 | } |
625 | /* It is counted once already... */ |
626 | data->pos_min += minnext * (mincount - counted); |
627 | data->pos_delta += - counted * deltanext + |
628 | (minnext + deltanext) * maxcount - minnext * mincount; |
629 | if (mincount != maxcount) { |
630 | scan_commit(data); |
631 | if (mincount && last_str) { |
632 | sv_setsv(data->last_found, last_str); |
633 | data->last_end = data->pos_min; |
634 | data->last_start_min = |
a0ed51b3 |
635 | data->pos_min - CHR_SVLEN(last_str); |
c277df42 |
636 | data->last_start_max = is_inf |
637 | ? I32_MAX |
638 | : data->pos_min + data->pos_delta |
a0ed51b3 |
639 | - CHR_SVLEN(last_str); |
c277df42 |
640 | } |
641 | data->longest = &(data->longest_float); |
642 | } |
aca2d497 |
643 | SvREFCNT_dec(last_str); |
c277df42 |
644 | } |
405ff068 |
645 | if (data && (fl & SF_HAS_EVAL)) |
c277df42 |
646 | data->flags |= SF_HAS_EVAL; |
647 | optimize_curly_tail: |
c277df42 |
648 | if (OP(oscan) != CURLYX) { |
649 | while (regkind[(U8)OP(next = regnext(oscan))] == NOTHING |
650 | && NEXT_OFF(next)) |
651 | NEXT_OFF(oscan) += NEXT_OFF(next); |
652 | } |
c277df42 |
653 | continue; |
654 | default: /* REF only? */ |
655 | if (flags & SCF_DO_SUBSTR) { |
656 | scan_commit(data); |
657 | data->longest = &(data->longest_float); |
658 | } |
aca2d497 |
659 | is_inf = is_inf_internal = 1; |
c277df42 |
660 | break; |
661 | } |
a0ed51b3 |
662 | } |
663 | else if (strchr(simple,OP(scan)) || regkind[(U8)OP(scan)] == ANYUTF8) { |
c277df42 |
664 | if (flags & SCF_DO_SUBSTR) { |
665 | scan_commit(data); |
666 | data->pos_min++; |
667 | } |
668 | min++; |
a0ed51b3 |
669 | } |
670 | else if (regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) { |
c277df42 |
671 | data->flags |= (OP(scan) == MEOL |
672 | ? SF_BEFORE_MEOL |
673 | : SF_BEFORE_SEOL); |
a0ed51b3 |
674 | } |
675 | else if (regkind[(U8)OP(scan)] == BRANCHJ |
c277df42 |
676 | && (scan->flags || data) |
677 | && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) { |
678 | I32 deltanext, minnext; |
679 | regnode *nscan; |
680 | |
681 | data_fake.flags = 0; |
682 | next = regnext(scan); |
683 | nscan = NEXTOPER(NEXTOPER(scan)); |
684 | minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0); |
685 | if (scan->flags) { |
686 | if (deltanext) { |
687 | FAIL("variable length lookbehind not implemented"); |
a0ed51b3 |
688 | } |
689 | else if (minnext > U8_MAX) { |
c277df42 |
690 | FAIL2("lookbehind longer than %d not implemented", U8_MAX); |
691 | } |
692 | scan->flags = minnext; |
693 | } |
694 | if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) |
695 | pars++; |
405ff068 |
696 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 |
697 | data->flags |= SF_HAS_EVAL; |
a0ed51b3 |
698 | } |
699 | else if (OP(scan) == OPEN) { |
c277df42 |
700 | pars++; |
a0ed51b3 |
701 | } |
702 | else if (OP(scan) == CLOSE && ARG(scan) == is_par) { |
c277df42 |
703 | next = regnext(scan); |
704 | |
705 | if ( next && (OP(next) != WHILEM) && next < last) |
c277df42 |
706 | is_par = 0; /* Disable optimization */ |
a0ed51b3 |
707 | } |
708 | else if (OP(scan) == EVAL) { |
c277df42 |
709 | if (data) |
710 | data->flags |= SF_HAS_EVAL; |
711 | } |
0f5d15d6 |
712 | else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */ |
713 | if (flags & SCF_DO_SUBSTR) { |
714 | scan_commit(data); |
715 | data->longest = &(data->longest_float); |
716 | } |
717 | is_inf = is_inf_internal = 1; |
718 | } |
c277df42 |
719 | /* Else: zero-length, ignore. */ |
720 | scan = regnext(scan); |
721 | } |
722 | |
723 | finish: |
724 | *scanp = scan; |
aca2d497 |
725 | *deltap = is_inf_internal ? I32_MAX : delta; |
c277df42 |
726 | if (flags & SCF_DO_SUBSTR && is_inf) |
727 | data->pos_delta = I32_MAX - data->pos_min; |
728 | if (is_par > U8_MAX) |
729 | is_par = 0; |
730 | if (is_par && pars==1 && data) { |
731 | data->flags |= SF_IN_PAR; |
732 | data->flags &= ~SF_HAS_PAR; |
a0ed51b3 |
733 | } |
734 | else if (pars && data) { |
c277df42 |
735 | data->flags |= SF_HAS_PAR; |
736 | data->flags &= ~SF_IN_PAR; |
737 | } |
738 | return min; |
739 | } |
740 | |
76e3520e |
741 | STATIC I32 |
c277df42 |
742 | add_data(I32 n, char *s) |
743 | { |
5c0ca799 |
744 | dTHR; |
3280af22 |
745 | if (PL_regcomp_rx->data) { |
746 | Renewc(PL_regcomp_rx->data, |
747 | sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (PL_regcomp_rx->data->count + n - 1), |
c277df42 |
748 | char, struct reg_data); |
3280af22 |
749 | Renew(PL_regcomp_rx->data->what, PL_regcomp_rx->data->count + n, U8); |
750 | PL_regcomp_rx->data->count += n; |
a0ed51b3 |
751 | } |
752 | else { |
3280af22 |
753 | Newc(1207, PL_regcomp_rx->data, sizeof(*PL_regcomp_rx->data) + sizeof(void*) * (n - 1), |
c277df42 |
754 | char, struct reg_data); |
3280af22 |
755 | New(1208, PL_regcomp_rx->data->what, n, U8); |
756 | PL_regcomp_rx->data->count = n; |
c277df42 |
757 | } |
3280af22 |
758 | Copy(s, PL_regcomp_rx->data->what + PL_regcomp_rx->data->count - n, n, U8); |
759 | return PL_regcomp_rx->data->count - n; |
c277df42 |
760 | } |
761 | |
a687059c |
762 | /* |
e50aee73 |
763 | - pregcomp - compile a regular expression into internal code |
a687059c |
764 | * |
765 | * We can't allocate space until we know how big the compiled form will be, |
766 | * but we can't compile it (and thus know how big it is) until we've got a |
767 | * place to put the code. So we cheat: we compile it twice, once with code |
768 | * generation turned off and size counting turned on, and once "for real". |
769 | * This also means that we don't allocate space until we are sure that the |
770 | * thing really will compile successfully, and we never have to move the |
771 | * code and thus invalidate pointers into it. (Note that it has to be in |
772 | * one piece because free() must be able to free it all.) [NB: not true in perl] |
773 | * |
774 | * Beware that the optimization-preparation code in here knows about some |
775 | * of the structure of the compiled regexp. [I'll say.] |
776 | */ |
777 | regexp * |
8ac85365 |
778 | pregcomp(char *exp, char *xend, PMOP *pm) |
a687059c |
779 | { |
5c0ca799 |
780 | dTHR; |
a0d0e21e |
781 | register regexp *r; |
c277df42 |
782 | regnode *scan; |
783 | SV **longest; |
784 | SV *longest_fixed; |
785 | SV *longest_float; |
786 | regnode *first; |
a0d0e21e |
787 | I32 flags; |
a0d0e21e |
788 | I32 minlen = 0; |
789 | I32 sawplus = 0; |
790 | I32 sawopen = 0; |
791 | |
792 | if (exp == NULL) |
c277df42 |
793 | FAIL("NULL regexp argument"); |
a0d0e21e |
794 | |
a0ed51b3 |
795 | if (PL_curcop == &compiling ? (PL_hints & HINT_UTF8) : IN_UTF8) |
796 | PL_reg_flags |= RF_utf8; |
797 | else |
798 | PL_reg_flags = 0; |
799 | |
3280af22 |
800 | PL_regprecomp = savepvn(exp, xend - exp); |
8d300b32 |
801 | DEBUG_r( |
802 | if (!PL_colorset) { |
803 | int i = 0; |
804 | char *s = PerlEnv_getenv("PERL_RE_COLORS"); |
805 | |
806 | if (s) { |
807 | PL_colors[0] = s = savepv(s); |
808 | while (++i < 6) { |
809 | s = strchr(s, '\t'); |
810 | if (s) { |
811 | *s = '\0'; |
812 | PL_colors[i] = ++s; |
813 | } |
814 | else |
815 | PL_colors[i] = ""; |
816 | } |
817 | } else { |
818 | while (i < 6) |
819 | PL_colors[i++] = ""; |
820 | } |
821 | PL_colorset = 1; |
822 | } |
823 | ); |
824 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling%s RE `%s%*s%s'\n", |
825 | PL_colors[4],PL_colors[5],PL_colors[0], |
826 | xend - exp, PL_regprecomp, PL_colors[1])); |
3280af22 |
827 | PL_regflags = pm->op_pmflags; |
828 | PL_regsawback = 0; |
bbce6d69 |
829 | |
3280af22 |
830 | PL_regseen = 0; |
831 | PL_seen_zerolen = *exp == '^' ? -1 : 0; |
832 | PL_seen_evals = 0; |
833 | PL_extralen = 0; |
c277df42 |
834 | |
bbce6d69 |
835 | /* First pass: determine size, legality. */ |
3280af22 |
836 | PL_regcomp_parse = exp; |
837 | PL_regxend = xend; |
838 | PL_regnaughty = 0; |
839 | PL_regnpar = 1; |
840 | PL_regsize = 0L; |
841 | PL_regcode = &PL_regdummy; |
842 | regc((U8)MAGIC, (char*)PL_regcode); |
a0d0e21e |
843 | if (reg(0, &flags) == NULL) { |
3280af22 |
844 | Safefree(PL_regprecomp); |
845 | PL_regprecomp = Nullch; |
a0d0e21e |
846 | return(NULL); |
847 | } |
3280af22 |
848 | DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", PL_regsize)); |
c277df42 |
849 | |
c277df42 |
850 | /* Small enough for pointer-storage convention? |
851 | If extralen==0, this means that we will not need long jumps. */ |
3280af22 |
852 | if (PL_regsize >= 0x10000L && PL_extralen) |
853 | PL_regsize += PL_extralen; |
c277df42 |
854 | else |
3280af22 |
855 | PL_extralen = 0; |
a0d0e21e |
856 | |
bbce6d69 |
857 | /* Allocate space and initialize. */ |
3280af22 |
858 | Newc(1001, r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), |
c277df42 |
859 | char, regexp); |
a0d0e21e |
860 | if (r == NULL) |
861 | FAIL("regexp out of space"); |
c277df42 |
862 | r->refcnt = 1; |
bbce6d69 |
863 | r->prelen = xend - exp; |
3280af22 |
864 | r->precomp = PL_regprecomp; |
a0d0e21e |
865 | r->subbeg = r->subbase = NULL; |
3280af22 |
866 | r->nparens = PL_regnpar - 1; /* set early to validate backrefs */ |
867 | PL_regcomp_rx = r; |
bbce6d69 |
868 | |
869 | /* Second pass: emit code. */ |
3280af22 |
870 | PL_regcomp_parse = exp; |
871 | PL_regxend = xend; |
872 | PL_regnaughty = 0; |
873 | PL_regnpar = 1; |
874 | PL_regcode = r->program; |
2cd61cdb |
875 | /* Store the count of eval-groups for security checks: */ |
3280af22 |
876 | PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals); |
877 | regc((U8)MAGIC, (char*) PL_regcode++); |
c277df42 |
878 | r->data = 0; |
a0d0e21e |
879 | if (reg(0, &flags) == NULL) |
880 | return(NULL); |
881 | |
882 | /* Dig out information for optimizations. */ |
8782bef2 |
883 | r->reganch = pm->op_pmflags & PMf_COMPILETIME; |
3280af22 |
884 | pm->op_pmflags = PL_regflags; |
a0ed51b3 |
885 | if (UTF) |
886 | r->reganch |= ROPT_UTF8; |
c277df42 |
887 | r->regstclass = NULL; |
a0ed51b3 |
888 | if (PL_regnaughty >= 10) /* Probably an expensive pattern. */ |
889 | r->reganch |= ROPT_NAUGHTY; |
c277df42 |
890 | scan = r->program + 1; /* First BRANCH. */ |
2779dcf1 |
891 | |
892 | /* XXXX To minimize changes to RE engine we always allocate |
893 | 3-units-long substrs field. */ |
894 | Newz(1004, r->substrs, 1, struct reg_substr_data); |
895 | |
c277df42 |
896 | if (OP(scan) != BRANCH) { /* Only one top-level choice. */ |
897 | scan_data_t data; |
898 | I32 fake; |
c5254dd6 |
899 | STRLEN longest_float_length, longest_fixed_length; |
a0d0e21e |
900 | |
c277df42 |
901 | StructCopy(&zero_scan_data, &data, scan_data_t); |
a0d0e21e |
902 | first = scan; |
c277df42 |
903 | /* Skip introductions and multiplicators >= 1. */ |
a0d0e21e |
904 | while ((OP(first) == OPEN && (sawopen = 1)) || |
905 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || |
906 | (OP(first) == PLUS) || |
907 | (OP(first) == MINMOD) || |
908 | (regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
909 | if (OP(first) == PLUS) |
910 | sawplus = 1; |
911 | else |
912 | first += regarglen[(U8)OP(first)]; |
913 | first = NEXTOPER(first); |
a687059c |
914 | } |
915 | |
a0d0e21e |
916 | /* Starting-point info. */ |
917 | again: |
c277df42 |
918 | if (OP(first) == EXACT); /* Empty, get anchored substr later. */ |
a0ed51b3 |
919 | else if (strchr(simple+4,OP(first))) |
a0d0e21e |
920 | r->regstclass = first; |
bbce6d69 |
921 | else if (regkind[(U8)OP(first)] == BOUND || |
922 | regkind[(U8)OP(first)] == NBOUND) |
a0d0e21e |
923 | r->regstclass = first; |
924 | else if (regkind[(U8)OP(first)] == BOL) { |
c277df42 |
925 | r->reganch |= (OP(first) == MBOL ? ROPT_ANCH_MBOL: ROPT_ANCH_BOL); |
a0d0e21e |
926 | first = NEXTOPER(first); |
774d564b |
927 | goto again; |
928 | } |
929 | else if (OP(first) == GPOS) { |
930 | r->reganch |= ROPT_ANCH_GPOS; |
931 | first = NEXTOPER(first); |
932 | goto again; |
a0d0e21e |
933 | } |
934 | else if ((OP(first) == STAR && |
935 | regkind[(U8)OP(NEXTOPER(first))] == ANY) && |
936 | !(r->reganch & ROPT_ANCH) ) |
937 | { |
938 | /* turn .* into ^.* with an implied $*=1 */ |
774d564b |
939 | r->reganch |= ROPT_ANCH_BOL | ROPT_IMPLICIT; |
a0d0e21e |
940 | first = NEXTOPER(first); |
774d564b |
941 | goto again; |
a0d0e21e |
942 | } |
3280af22 |
943 | if (sawplus && (!sawopen || !PL_regsawback)) |
a0d0e21e |
944 | r->reganch |= ROPT_SKIP; /* x+ must match 1st of run */ |
945 | |
c277df42 |
946 | /* Scan is after the zeroth branch, first is atomic matcher. */ |
947 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n", |
948 | first - scan + 1)); |
a0d0e21e |
949 | /* |
950 | * If there's something expensive in the r.e., find the |
951 | * longest literal string that must appear and make it the |
952 | * regmust. Resolve ties in favor of later strings, since |
953 | * the regstart check works with the beginning of the r.e. |
954 | * and avoiding duplication strengthens checking. Not a |
955 | * strong reason, but sufficient in the absence of others. |
956 | * [Now we resolve ties in favor of the earlier string if |
c277df42 |
957 | * it happens that c_offset_min has been invalidated, since the |
a0d0e21e |
958 | * earlier string may buy us something the later one won't.] |
959 | */ |
a0d0e21e |
960 | minlen = 0; |
a687059c |
961 | |
c277df42 |
962 | data.longest_fixed = newSVpv("",0); |
963 | data.longest_float = newSVpv("",0); |
964 | data.last_found = newSVpv("",0); |
965 | data.longest = &(data.longest_fixed); |
966 | first = scan; |
967 | |
3280af22 |
968 | minlen = study_chunk(&first, &fake, scan + PL_regsize, /* Up to end */ |
c277df42 |
969 | &data, SCF_DO_SUBSTR); |
3280af22 |
970 | if ( PL_regnpar == 1 && data.longest == &(data.longest_fixed) |
c277df42 |
971 | && data.last_start_min == 0 && data.last_end > 0 |
3280af22 |
972 | && !PL_seen_zerolen |
973 | && (!(PL_regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS))) |
c277df42 |
974 | r->reganch |= ROPT_CHECK_ALL; |
975 | scan_commit(&data); |
976 | SvREFCNT_dec(data.last_found); |
977 | |
a0ed51b3 |
978 | longest_float_length = CHR_SVLEN(data.longest_float); |
c5254dd6 |
979 | if (longest_float_length |
c277df42 |
980 | || (data.flags & SF_FL_BEFORE_EOL |
981 | && (!(data.flags & SF_FL_BEFORE_MEOL) |
3280af22 |
982 | || (PL_regflags & PMf_MULTILINE)))) { |
a0ed51b3 |
983 | if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */ |
aca2d497 |
984 | && data.offset_fixed == data.offset_float_min |
985 | && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)) |
986 | goto remove_float; /* As in (a)+. */ |
987 | |
c277df42 |
988 | r->float_substr = data.longest_float; |
989 | r->float_min_offset = data.offset_float_min; |
990 | r->float_max_offset = data.offset_float_max; |
2779dcf1 |
991 | fbm_compile(r->float_substr, 0); |
c277df42 |
992 | BmUSEFUL(r->float_substr) = 100; |
993 | if (data.flags & SF_FL_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
994 | && (!(data.flags & SF_FL_BEFORE_MEOL) |
3280af22 |
995 | || (PL_regflags & PMf_MULTILINE))) |
c277df42 |
996 | SvTAIL_on(r->float_substr); |
a0ed51b3 |
997 | } |
998 | else { |
aca2d497 |
999 | remove_float: |
c277df42 |
1000 | r->float_substr = Nullsv; |
1001 | SvREFCNT_dec(data.longest_float); |
c5254dd6 |
1002 | longest_float_length = 0; |
a0d0e21e |
1003 | } |
c277df42 |
1004 | |
a0ed51b3 |
1005 | longest_fixed_length = CHR_SVLEN(data.longest_fixed); |
c5254dd6 |
1006 | if (longest_fixed_length |
c277df42 |
1007 | || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
1008 | && (!(data.flags & SF_FIX_BEFORE_MEOL) |
3280af22 |
1009 | || (PL_regflags & PMf_MULTILINE)))) { |
c277df42 |
1010 | r->anchored_substr = data.longest_fixed; |
1011 | r->anchored_offset = data.offset_fixed; |
2779dcf1 |
1012 | fbm_compile(r->anchored_substr, 0); |
c277df42 |
1013 | BmUSEFUL(r->anchored_substr) = 100; |
1014 | if (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
1015 | && (!(data.flags & SF_FIX_BEFORE_MEOL) |
3280af22 |
1016 | || (PL_regflags & PMf_MULTILINE))) |
c277df42 |
1017 | SvTAIL_on(r->anchored_substr); |
a0ed51b3 |
1018 | } |
1019 | else { |
c277df42 |
1020 | r->anchored_substr = Nullsv; |
1021 | SvREFCNT_dec(data.longest_fixed); |
c5254dd6 |
1022 | longest_fixed_length = 0; |
a0d0e21e |
1023 | } |
c277df42 |
1024 | |
1025 | /* A temporary algorithm prefers floated substr to fixed one to dig more info. */ |
c5254dd6 |
1026 | if (longest_fixed_length > longest_float_length) { |
c277df42 |
1027 | r->check_substr = r->anchored_substr; |
1028 | r->check_offset_min = r->check_offset_max = r->anchored_offset; |
1029 | if (r->reganch & ROPT_ANCH_SINGLE) |
1030 | r->reganch |= ROPT_NOSCAN; |
a0ed51b3 |
1031 | } |
1032 | else { |
c277df42 |
1033 | r->check_substr = r->float_substr; |
1034 | r->check_offset_min = data.offset_float_min; |
1035 | r->check_offset_max = data.offset_float_max; |
a0d0e21e |
1036 | } |
a0ed51b3 |
1037 | } |
1038 | else { |
c277df42 |
1039 | /* Several toplevels. Best we can is to set minlen. */ |
1040 | I32 fake; |
1041 | |
1042 | DEBUG_r(PerlIO_printf(Perl_debug_log, "\n")); |
1043 | scan = r->program + 1; |
3280af22 |
1044 | minlen = study_chunk(&scan, &fake, scan + PL_regsize, NULL, 0); |
c277df42 |
1045 | r->check_substr = r->anchored_substr = r->float_substr = Nullsv; |
a0d0e21e |
1046 | } |
1047 | |
a0d0e21e |
1048 | r->minlen = minlen; |
3280af22 |
1049 | if (PL_regseen & REG_SEEN_GPOS) |
c277df42 |
1050 | r->reganch |= ROPT_GPOS_SEEN; |
3280af22 |
1051 | if (PL_regseen & REG_SEEN_LOOKBEHIND) |
c277df42 |
1052 | r->reganch |= ROPT_LOOKBEHIND_SEEN; |
3280af22 |
1053 | if (PL_regseen & REG_SEEN_EVAL) |
ce862d02 |
1054 | r->reganch |= ROPT_EVAL_SEEN; |
3280af22 |
1055 | Newz(1002, r->startp, PL_regnpar, char*); |
1056 | Newz(1002, r->endp, PL_regnpar, char*); |
a0d0e21e |
1057 | DEBUG_r(regdump(r)); |
1058 | return(r); |
a687059c |
1059 | } |
1060 | |
1061 | /* |
1062 | - reg - regular expression, i.e. main body or parenthesized thing |
1063 | * |
1064 | * Caller must absorb opening parenthesis. |
1065 | * |
1066 | * Combining parenthesis handling with the base level of regular expression |
1067 | * is a trifle forced, but the need to tie the tails of the branches to what |
1068 | * follows makes it hard to avoid. |
1069 | */ |
76e3520e |
1070 | STATIC regnode * |
8ac85365 |
1071 | reg(I32 paren, I32 *flagp) |
c277df42 |
1072 | /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */ |
a687059c |
1073 | { |
5c0ca799 |
1074 | dTHR; |
c277df42 |
1075 | register regnode *ret; /* Will be the head of the group. */ |
1076 | register regnode *br; |
1077 | register regnode *lastbr; |
1078 | register regnode *ender = 0; |
a0d0e21e |
1079 | register I32 parno = 0; |
3280af22 |
1080 | I32 flags, oregflags = PL_regflags, have_branch = 0, open = 0; |
c277df42 |
1081 | char c; |
a0d0e21e |
1082 | |
821b33a5 |
1083 | *flagp = 0; /* Tentatively. */ |
a0d0e21e |
1084 | |
1085 | /* Make an OPEN node, if parenthesized. */ |
1086 | if (paren) { |
3280af22 |
1087 | if (*PL_regcomp_parse == '?') { |
ca9dfc88 |
1088 | U16 posflags = 0, negflags = 0; |
1089 | U16 *flagsp = &posflags; |
0f5d15d6 |
1090 | int logical = 0; |
ca9dfc88 |
1091 | |
3280af22 |
1092 | PL_regcomp_parse++; |
1093 | paren = *PL_regcomp_parse++; |
c277df42 |
1094 | ret = NULL; /* For look-ahead/behind. */ |
a0d0e21e |
1095 | switch (paren) { |
c277df42 |
1096 | case '<': |
3280af22 |
1097 | PL_regseen |= REG_SEEN_LOOKBEHIND; |
1098 | if (*PL_regcomp_parse == '!') |
c277df42 |
1099 | paren = ','; |
3280af22 |
1100 | if (*PL_regcomp_parse != '=' && *PL_regcomp_parse != '!') |
c277df42 |
1101 | goto unknown; |
3280af22 |
1102 | PL_regcomp_parse++; |
a0d0e21e |
1103 | case '=': |
1104 | case '!': |
3280af22 |
1105 | PL_seen_zerolen++; |
c277df42 |
1106 | case ':': |
1107 | case '>': |
a0d0e21e |
1108 | break; |
1109 | case '$': |
1110 | case '@': |
c277df42 |
1111 | FAIL2("Sequence (?%c...) not implemented", (int)paren); |
a0d0e21e |
1112 | break; |
1113 | case '#': |
3280af22 |
1114 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
1115 | PL_regcomp_parse++; |
1116 | if (*PL_regcomp_parse != ')') |
c277df42 |
1117 | FAIL("Sequence (?#... not terminated"); |
a0d0e21e |
1118 | nextchar(); |
1119 | *flagp = TRYAGAIN; |
1120 | return NULL; |
0f5d15d6 |
1121 | case 'p': |
1122 | logical = 1; |
1123 | paren = *PL_regcomp_parse++; |
1124 | /* FALL THROUGH */ |
c277df42 |
1125 | case '{': |
1126 | { |
1127 | dTHR; |
1128 | I32 count = 1, n = 0; |
1129 | char c; |
3280af22 |
1130 | char *s = PL_regcomp_parse; |
c277df42 |
1131 | SV *sv; |
1132 | OP_4tree *sop, *rop; |
1133 | |
3280af22 |
1134 | PL_seen_zerolen++; |
1135 | PL_regseen |= REG_SEEN_EVAL; |
1136 | while (count && (c = *PL_regcomp_parse)) { |
1137 | if (c == '\\' && PL_regcomp_parse[1]) |
1138 | PL_regcomp_parse++; |
c277df42 |
1139 | else if (c == '{') |
1140 | count++; |
1141 | else if (c == '}') |
1142 | count--; |
3280af22 |
1143 | PL_regcomp_parse++; |
c277df42 |
1144 | } |
3280af22 |
1145 | if (*PL_regcomp_parse != ')') |
c277df42 |
1146 | FAIL("Sequence (?{...}) not terminated or not {}-balanced"); |
1147 | if (!SIZE_ONLY) { |
1148 | AV *av; |
1149 | |
3280af22 |
1150 | if (PL_regcomp_parse - 1 - s) |
1151 | sv = newSVpv(s, PL_regcomp_parse - 1 - s); |
c277df42 |
1152 | else |
1153 | sv = newSVpv("", 0); |
1154 | |
1155 | rop = sv_compile_2op(sv, &sop, "re", &av); |
1156 | |
1157 | n = add_data(3, "nso"); |
3280af22 |
1158 | PL_regcomp_rx->data->data[n] = (void*)rop; |
1159 | PL_regcomp_rx->data->data[n+1] = (void*)av; |
1160 | PL_regcomp_rx->data->data[n+2] = (void*)sop; |
c277df42 |
1161 | SvREFCNT_dec(sv); |
a0ed51b3 |
1162 | } |
1163 | else { /* First pass */ |
1164 | if (PL_reginterp_cnt < ++PL_seen_evals && PL_curcop != &compiling) |
2cd61cdb |
1165 | /* No compiled RE interpolated, has runtime |
1166 | components ===> unsafe. */ |
1167 | FAIL("Eval-group not allowed at runtime, use re 'eval'"); |
3280af22 |
1168 | if (PL_tainted) |
cc6b7395 |
1169 | FAIL("Eval-group in insecure regular expression"); |
c277df42 |
1170 | } |
1171 | |
1172 | nextchar(); |
0f5d15d6 |
1173 | if (logical) { |
1174 | ret = reg_node(LOGICAL); |
1175 | if (!SIZE_ONLY) |
1176 | ret->flags = 2; |
1177 | regtail(ret, reganode(EVAL, n)); |
1178 | return ret; |
1179 | } |
c277df42 |
1180 | return reganode(EVAL, n); |
1181 | } |
1182 | case '(': |
1183 | { |
3280af22 |
1184 | if (PL_regcomp_parse[0] == '?') { |
1185 | if (PL_regcomp_parse[1] == '=' || PL_regcomp_parse[1] == '!' |
1186 | || PL_regcomp_parse[1] == '<' |
1187 | || PL_regcomp_parse[1] == '{') { /* Lookahead or eval. */ |
c277df42 |
1188 | I32 flag; |
1189 | |
1190 | ret = reg_node(LOGICAL); |
0f5d15d6 |
1191 | if (!SIZE_ONLY) |
1192 | ret->flags = 1; |
c277df42 |
1193 | regtail(ret, reg(1, &flag)); |
1194 | goto insert_if; |
1195 | } |
a0ed51b3 |
1196 | } |
1197 | else if (PL_regcomp_parse[0] >= '1' && PL_regcomp_parse[0] <= '9' ) { |
3280af22 |
1198 | parno = atoi(PL_regcomp_parse++); |
c277df42 |
1199 | |
3280af22 |
1200 | while (isDIGIT(*PL_regcomp_parse)) |
1201 | PL_regcomp_parse++; |
c277df42 |
1202 | ret = reganode(GROUPP, parno); |
1203 | if ((c = *nextchar()) != ')') |
1204 | FAIL2("Switch (?(number%c not recognized", c); |
1205 | insert_if: |
1206 | regtail(ret, reganode(IFTHEN, 0)); |
1207 | br = regbranch(&flags, 1); |
1208 | if (br == NULL) |
1209 | br = reganode(LONGJMP, 0); |
1210 | else |
1211 | regtail(br, reganode(LONGJMP, 0)); |
1212 | c = *nextchar(); |
d1b80229 |
1213 | if (flags&HASWIDTH) |
1214 | *flagp |= HASWIDTH; |
c277df42 |
1215 | if (c == '|') { |
1216 | lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */ |
1217 | regbranch(&flags, 1); |
1218 | regtail(ret, lastbr); |
d1b80229 |
1219 | if (flags&HASWIDTH) |
1220 | *flagp |= HASWIDTH; |
c277df42 |
1221 | c = *nextchar(); |
a0ed51b3 |
1222 | } |
1223 | else |
c277df42 |
1224 | lastbr = NULL; |
1225 | if (c != ')') |
1226 | FAIL("Switch (?(condition)... contains too many branches"); |
1227 | ender = reg_node(TAIL); |
1228 | regtail(br, ender); |
1229 | if (lastbr) { |
1230 | regtail(lastbr, ender); |
1231 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); |
a0ed51b3 |
1232 | } |
1233 | else |
c277df42 |
1234 | regtail(ret, ender); |
1235 | return ret; |
a0ed51b3 |
1236 | } |
1237 | else { |
3280af22 |
1238 | FAIL2("Unknown condition for (?(%.2s", PL_regcomp_parse); |
c277df42 |
1239 | } |
1240 | } |
1b1626e4 |
1241 | case 0: |
c277df42 |
1242 | FAIL("Sequence (? incomplete"); |
1b1626e4 |
1243 | break; |
a0d0e21e |
1244 | default: |
3280af22 |
1245 | --PL_regcomp_parse; |
ca9dfc88 |
1246 | parse_flags: |
3280af22 |
1247 | while (*PL_regcomp_parse && strchr("iogcmsx", *PL_regcomp_parse)) { |
1248 | if (*PL_regcomp_parse != 'o') |
1249 | pmflag(flagsp, *PL_regcomp_parse); |
1250 | ++PL_regcomp_parse; |
ca9dfc88 |
1251 | } |
3280af22 |
1252 | if (*PL_regcomp_parse == '-') { |
ca9dfc88 |
1253 | flagsp = &negflags; |
3280af22 |
1254 | ++PL_regcomp_parse; |
ca9dfc88 |
1255 | goto parse_flags; |
48c036b1 |
1256 | } |
3280af22 |
1257 | PL_regflags |= posflags; |
1258 | PL_regflags &= ~negflags; |
1259 | if (*PL_regcomp_parse == ':') { |
1260 | PL_regcomp_parse++; |
ca9dfc88 |
1261 | paren = ':'; |
1262 | break; |
1263 | } |
c277df42 |
1264 | unknown: |
3280af22 |
1265 | if (*PL_regcomp_parse != ')') |
1266 | FAIL2("Sequence (?%c...) not recognized", *PL_regcomp_parse); |
a0d0e21e |
1267 | nextchar(); |
1268 | *flagp = TRYAGAIN; |
1269 | return NULL; |
1270 | } |
1271 | } |
1272 | else { |
3280af22 |
1273 | parno = PL_regnpar; |
1274 | PL_regnpar++; |
a0d0e21e |
1275 | ret = reganode(OPEN, parno); |
c277df42 |
1276 | open = 1; |
a0d0e21e |
1277 | } |
a0ed51b3 |
1278 | } |
1279 | else |
a0d0e21e |
1280 | ret = NULL; |
1281 | |
1282 | /* Pick up the branches, linking them together. */ |
c277df42 |
1283 | br = regbranch(&flags, 1); |
a0d0e21e |
1284 | if (br == NULL) |
1285 | return(NULL); |
3280af22 |
1286 | if (*PL_regcomp_parse == '|') { |
1287 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 |
1288 | reginsert(BRANCHJ, br); |
a0ed51b3 |
1289 | } |
1290 | else |
c277df42 |
1291 | reginsert(BRANCH, br); |
1292 | have_branch = 1; |
1293 | if (SIZE_ONLY) |
3280af22 |
1294 | PL_extralen += 1; /* For BRANCHJ-BRANCH. */ |
a0ed51b3 |
1295 | } |
1296 | else if (paren == ':') { |
c277df42 |
1297 | *flagp |= flags&SIMPLE; |
1298 | } |
1299 | if (open) { /* Starts with OPEN. */ |
1300 | regtail(ret, br); /* OPEN -> first. */ |
a0ed51b3 |
1301 | } |
1302 | else if (paren != '?') /* Not Conditional */ |
a0d0e21e |
1303 | ret = br; |
821b33a5 |
1304 | if (flags&HASWIDTH) |
1305 | *flagp |= HASWIDTH; |
a0d0e21e |
1306 | *flagp |= flags&SPSTART; |
c277df42 |
1307 | lastbr = br; |
3280af22 |
1308 | while (*PL_regcomp_parse == '|') { |
1309 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 |
1310 | ender = reganode(LONGJMP,0); |
1311 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ |
1312 | } |
1313 | if (SIZE_ONLY) |
3280af22 |
1314 | PL_extralen += 2; /* Account for LONGJMP. */ |
a0d0e21e |
1315 | nextchar(); |
c277df42 |
1316 | br = regbranch(&flags, 0); |
a687059c |
1317 | if (br == NULL) |
a0d0e21e |
1318 | return(NULL); |
c277df42 |
1319 | regtail(lastbr, br); /* BRANCH -> BRANCH. */ |
1320 | lastbr = br; |
821b33a5 |
1321 | if (flags&HASWIDTH) |
1322 | *flagp |= HASWIDTH; |
a687059c |
1323 | *flagp |= flags&SPSTART; |
a0d0e21e |
1324 | } |
1325 | |
c277df42 |
1326 | if (have_branch || paren != ':') { |
1327 | /* Make a closing node, and hook it on the end. */ |
1328 | switch (paren) { |
1329 | case ':': |
1330 | ender = reg_node(TAIL); |
1331 | break; |
1332 | case 1: |
1333 | ender = reganode(CLOSE, parno); |
1334 | break; |
1335 | case '<': |
c277df42 |
1336 | case ',': |
1337 | case '=': |
1338 | case '!': |
c277df42 |
1339 | *flagp &= ~HASWIDTH; |
821b33a5 |
1340 | /* FALL THROUGH */ |
1341 | case '>': |
1342 | ender = reg_node(SUCCEED); |
c277df42 |
1343 | break; |
1344 | case 0: |
1345 | ender = reg_node(END); |
1346 | break; |
1347 | } |
1348 | regtail(lastbr, ender); |
a0d0e21e |
1349 | |
c277df42 |
1350 | if (have_branch) { |
1351 | /* Hook the tails of the branches to the closing node. */ |
1352 | for (br = ret; br != NULL; br = regnext(br)) { |
1353 | regoptail(br, ender); |
1354 | } |
1355 | } |
a0d0e21e |
1356 | } |
c277df42 |
1357 | |
1358 | { |
1359 | char *p; |
1360 | static char parens[] = "=!<,>"; |
1361 | |
1362 | if (paren && (p = strchr(parens, paren))) { |
1363 | int node = ((p - parens) % 2) ? UNLESSM : IFMATCH; |
1364 | int flag = (p - parens) > 1; |
1365 | |
1366 | if (paren == '>') |
1367 | node = SUSPEND, flag = 0; |
1368 | reginsert(node,ret); |
c277df42 |
1369 | ret->flags = flag; |
c277df42 |
1370 | regtail(ret, reg_node(TAIL)); |
1371 | } |
a0d0e21e |
1372 | } |
1373 | |
1374 | /* Check for proper termination. */ |
3280af22 |
1375 | if (paren && (PL_regcomp_parse >= PL_regxend || *nextchar() != ')')) { |
a0d0e21e |
1376 | FAIL("unmatched () in regexp"); |
a0ed51b3 |
1377 | } |
1378 | else if (!paren && PL_regcomp_parse < PL_regxend) { |
3280af22 |
1379 | if (*PL_regcomp_parse == ')') { |
a0d0e21e |
1380 | FAIL("unmatched () in regexp"); |
a0ed51b3 |
1381 | } |
1382 | else |
a0d0e21e |
1383 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
1384 | /* NOTREACHED */ |
1385 | } |
c277df42 |
1386 | if (paren != 0) { |
3280af22 |
1387 | PL_regflags = oregflags; |
c277df42 |
1388 | } |
a687059c |
1389 | |
a0d0e21e |
1390 | return(ret); |
a687059c |
1391 | } |
1392 | |
1393 | /* |
1394 | - regbranch - one alternative of an | operator |
1395 | * |
1396 | * Implements the concatenation operator. |
1397 | */ |
76e3520e |
1398 | STATIC regnode * |
c277df42 |
1399 | regbranch(I32 *flagp, I32 first) |
a687059c |
1400 | { |
5c0ca799 |
1401 | dTHR; |
c277df42 |
1402 | register regnode *ret; |
1403 | register regnode *chain = NULL; |
1404 | register regnode *latest; |
1405 | I32 flags = 0, c = 0; |
a0d0e21e |
1406 | |
c277df42 |
1407 | if (first) |
1408 | ret = NULL; |
1409 | else { |
3280af22 |
1410 | if (!SIZE_ONLY && PL_extralen) |
c277df42 |
1411 | ret = reganode(BRANCHJ,0); |
1412 | else |
1413 | ret = reg_node(BRANCH); |
1414 | } |
1415 | |
1416 | if (!first && SIZE_ONLY) |
3280af22 |
1417 | PL_extralen += 1; /* BRANCHJ */ |
c277df42 |
1418 | |
1419 | *flagp = WORST; /* Tentatively. */ |
a0d0e21e |
1420 | |
3280af22 |
1421 | PL_regcomp_parse--; |
a0d0e21e |
1422 | nextchar(); |
3280af22 |
1423 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '|' && *PL_regcomp_parse != ')') { |
a0d0e21e |
1424 | flags &= ~TRYAGAIN; |
1425 | latest = regpiece(&flags); |
1426 | if (latest == NULL) { |
1427 | if (flags & TRYAGAIN) |
1428 | continue; |
1429 | return(NULL); |
a0ed51b3 |
1430 | } |
1431 | else if (ret == NULL) |
c277df42 |
1432 | ret = latest; |
a0d0e21e |
1433 | *flagp |= flags&HASWIDTH; |
c277df42 |
1434 | if (chain == NULL) /* First piece. */ |
a0d0e21e |
1435 | *flagp |= flags&SPSTART; |
1436 | else { |
3280af22 |
1437 | PL_regnaughty++; |
a0d0e21e |
1438 | regtail(chain, latest); |
a687059c |
1439 | } |
a0d0e21e |
1440 | chain = latest; |
c277df42 |
1441 | c++; |
1442 | } |
1443 | if (chain == NULL) { /* Loop ran zero times. */ |
1444 | chain = reg_node(NOTHING); |
1445 | if (ret == NULL) |
1446 | ret = chain; |
1447 | } |
1448 | if (c == 1) { |
1449 | *flagp |= flags&SIMPLE; |
a0d0e21e |
1450 | } |
a687059c |
1451 | |
a0d0e21e |
1452 | return(ret); |
a687059c |
1453 | } |
1454 | |
1455 | /* |
1456 | - regpiece - something followed by possible [*+?] |
1457 | * |
1458 | * Note that the branching code sequences used for ? and the general cases |
1459 | * of * and + are somewhat optimized: they use the same NOTHING node as |
1460 | * both the endmarker for their branch list and the body of the last branch. |
1461 | * It might seem that this node could be dispensed with entirely, but the |
1462 | * endmarker role is not redundant. |
1463 | */ |
76e3520e |
1464 | STATIC regnode * |
8ac85365 |
1465 | regpiece(I32 *flagp) |
a687059c |
1466 | { |
5c0ca799 |
1467 | dTHR; |
c277df42 |
1468 | register regnode *ret; |
a0d0e21e |
1469 | register char op; |
1470 | register char *next; |
1471 | I32 flags; |
3280af22 |
1472 | char *origparse = PL_regcomp_parse; |
a0d0e21e |
1473 | char *maxpos; |
1474 | I32 min; |
c277df42 |
1475 | I32 max = REG_INFTY; |
a0d0e21e |
1476 | |
1477 | ret = regatom(&flags); |
1478 | if (ret == NULL) { |
1479 | if (flags & TRYAGAIN) |
1480 | *flagp |= TRYAGAIN; |
1481 | return(NULL); |
1482 | } |
1483 | |
3280af22 |
1484 | op = *PL_regcomp_parse; |
a0d0e21e |
1485 | |
3280af22 |
1486 | if (op == '{' && regcurly(PL_regcomp_parse)) { |
1487 | next = PL_regcomp_parse + 1; |
a0d0e21e |
1488 | maxpos = Nullch; |
1489 | while (isDIGIT(*next) || *next == ',') { |
1490 | if (*next == ',') { |
1491 | if (maxpos) |
1492 | break; |
1493 | else |
1494 | maxpos = next; |
a687059c |
1495 | } |
a0d0e21e |
1496 | next++; |
1497 | } |
1498 | if (*next == '}') { /* got one */ |
1499 | if (!maxpos) |
1500 | maxpos = next; |
3280af22 |
1501 | PL_regcomp_parse++; |
1502 | min = atoi(PL_regcomp_parse); |
a0d0e21e |
1503 | if (*maxpos == ',') |
1504 | maxpos++; |
1505 | else |
3280af22 |
1506 | maxpos = PL_regcomp_parse; |
a0d0e21e |
1507 | max = atoi(maxpos); |
1508 | if (!max && *maxpos != '0') |
c277df42 |
1509 | max = REG_INFTY; /* meaning "infinity" */ |
1510 | else if (max >= REG_INFTY) |
1511 | FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); |
3280af22 |
1512 | PL_regcomp_parse = next; |
a0d0e21e |
1513 | nextchar(); |
1514 | |
1515 | do_curly: |
1516 | if ((flags&SIMPLE)) { |
3280af22 |
1517 | PL_regnaughty += 2 + PL_regnaughty / 2; |
a0d0e21e |
1518 | reginsert(CURLY, ret); |
1519 | } |
1520 | else { |
3280af22 |
1521 | PL_regnaughty += 4 + PL_regnaughty; /* compound interest */ |
c277df42 |
1522 | regtail(ret, reg_node(WHILEM)); |
3280af22 |
1523 | if (!SIZE_ONLY && PL_extralen) { |
c277df42 |
1524 | reginsert(LONGJMP,ret); |
1525 | reginsert(NOTHING,ret); |
1526 | NEXT_OFF(ret) = 3; /* Go over LONGJMP. */ |
1527 | } |
a0d0e21e |
1528 | reginsert(CURLYX,ret); |
3280af22 |
1529 | if (!SIZE_ONLY && PL_extralen) |
c277df42 |
1530 | NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */ |
1531 | regtail(ret, reg_node(NOTHING)); |
1532 | if (SIZE_ONLY) |
3280af22 |
1533 | PL_extralen += 3; |
a0d0e21e |
1534 | } |
c277df42 |
1535 | ret->flags = 0; |
a0d0e21e |
1536 | |
1537 | if (min > 0) |
821b33a5 |
1538 | *flagp = WORST; |
1539 | if (max > 0) |
1540 | *flagp |= HASWIDTH; |
a0d0e21e |
1541 | if (max && max < min) |
c277df42 |
1542 | FAIL("Can't do {n,m} with n > m"); |
1543 | if (!SIZE_ONLY) { |
1544 | ARG1_SET(ret, min); |
1545 | ARG2_SET(ret, max); |
a687059c |
1546 | } |
a687059c |
1547 | |
a0d0e21e |
1548 | goto nest_check; |
a687059c |
1549 | } |
a0d0e21e |
1550 | } |
a687059c |
1551 | |
a0d0e21e |
1552 | if (!ISMULT1(op)) { |
1553 | *flagp = flags; |
a687059c |
1554 | return(ret); |
a0d0e21e |
1555 | } |
bb20fd44 |
1556 | |
c277df42 |
1557 | #if 0 /* Now runtime fix should be reliable. */ |
bb20fd44 |
1558 | if (!(flags&HASWIDTH) && op != '?') |
c277df42 |
1559 | FAIL("regexp *+ operand could be empty"); |
1560 | #endif |
bb20fd44 |
1561 | |
a0d0e21e |
1562 | nextchar(); |
1563 | |
821b33a5 |
1564 | *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); |
a0d0e21e |
1565 | |
1566 | if (op == '*' && (flags&SIMPLE)) { |
1567 | reginsert(STAR, ret); |
c277df42 |
1568 | ret->flags = 0; |
3280af22 |
1569 | PL_regnaughty += 4; |
a0d0e21e |
1570 | } |
1571 | else if (op == '*') { |
1572 | min = 0; |
1573 | goto do_curly; |
a0ed51b3 |
1574 | } |
1575 | else if (op == '+' && (flags&SIMPLE)) { |
a0d0e21e |
1576 | reginsert(PLUS, ret); |
c277df42 |
1577 | ret->flags = 0; |
3280af22 |
1578 | PL_regnaughty += 3; |
a0d0e21e |
1579 | } |
1580 | else if (op == '+') { |
1581 | min = 1; |
1582 | goto do_curly; |
a0ed51b3 |
1583 | } |
1584 | else if (op == '?') { |
a0d0e21e |
1585 | min = 0; max = 1; |
1586 | goto do_curly; |
1587 | } |
1588 | nest_check: |
599cee73 |
1589 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > 10000) { |
1590 | warner(WARN_UNSAFE, "%.*s matches null string many times", |
3280af22 |
1591 | PL_regcomp_parse - origparse, origparse); |
a0d0e21e |
1592 | } |
1593 | |
3280af22 |
1594 | if (*PL_regcomp_parse == '?') { |
a0d0e21e |
1595 | nextchar(); |
1596 | reginsert(MINMOD, ret); |
c277df42 |
1597 | regtail(ret, ret + NODE_STEP_REGNODE); |
a0d0e21e |
1598 | } |
3280af22 |
1599 | if (ISMULT2(PL_regcomp_parse)) |
a0d0e21e |
1600 | FAIL("nested *?+ in regexp"); |
1601 | |
1602 | return(ret); |
a687059c |
1603 | } |
1604 | |
1605 | /* |
1606 | - regatom - the lowest level |
1607 | * |
1608 | * Optimization: gobbles an entire sequence of ordinary characters so that |
1609 | * it can turn them into a single node, which is smaller to store and |
1610 | * faster to run. Backslashed characters are exceptions, each becoming a |
1611 | * separate node; the code is simpler that way and it's not worth fixing. |
1612 | * |
1613 | * [Yes, it is worth fixing, some scripts can run twice the speed.] |
1614 | */ |
76e3520e |
1615 | STATIC regnode * |
8ac85365 |
1616 | regatom(I32 *flagp) |
a687059c |
1617 | { |
5c0ca799 |
1618 | dTHR; |
c277df42 |
1619 | register regnode *ret = 0; |
a0d0e21e |
1620 | I32 flags; |
1621 | |
1622 | *flagp = WORST; /* Tentatively. */ |
1623 | |
1624 | tryagain: |
3280af22 |
1625 | switch (*PL_regcomp_parse) { |
a0d0e21e |
1626 | case '^': |
3280af22 |
1627 | PL_seen_zerolen++; |
a0d0e21e |
1628 | nextchar(); |
3280af22 |
1629 | if (PL_regflags & PMf_MULTILINE) |
c277df42 |
1630 | ret = reg_node(MBOL); |
3280af22 |
1631 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 |
1632 | ret = reg_node(SBOL); |
a0d0e21e |
1633 | else |
c277df42 |
1634 | ret = reg_node(BOL); |
a0d0e21e |
1635 | break; |
1636 | case '$': |
3280af22 |
1637 | if (PL_regcomp_parse[1]) |
1638 | PL_seen_zerolen++; |
a0d0e21e |
1639 | nextchar(); |
3280af22 |
1640 | if (PL_regflags & PMf_MULTILINE) |
c277df42 |
1641 | ret = reg_node(MEOL); |
3280af22 |
1642 | else if (PL_regflags & PMf_SINGLELINE) |
c277df42 |
1643 | ret = reg_node(SEOL); |
a0d0e21e |
1644 | else |
c277df42 |
1645 | ret = reg_node(EOL); |
a0d0e21e |
1646 | break; |
1647 | case '.': |
1648 | nextchar(); |
a0ed51b3 |
1649 | if (UTF) { |
1650 | if (PL_regflags & PMf_SINGLELINE) |
1651 | ret = reg_node(SANYUTF8); |
1652 | else |
1653 | ret = reg_node(ANYUTF8); |
1654 | *flagp |= HASWIDTH; |
1655 | } |
1656 | else { |
1657 | if (PL_regflags & PMf_SINGLELINE) |
1658 | ret = reg_node(SANY); |
1659 | else |
1660 | ret = reg_node(ANY); |
1661 | *flagp |= HASWIDTH|SIMPLE; |
1662 | } |
3280af22 |
1663 | PL_regnaughty++; |
a0d0e21e |
1664 | break; |
1665 | case '[': |
3280af22 |
1666 | PL_regcomp_parse++; |
a0ed51b3 |
1667 | ret = (UTF ? regclassutf8() : regclass()); |
a14b48bc |
1668 | if (*PL_regcomp_parse != ']') |
1669 | FAIL("unmatched [] in regexp"); |
1670 | nextchar(); |
a0d0e21e |
1671 | *flagp |= HASWIDTH|SIMPLE; |
1672 | break; |
1673 | case '(': |
1674 | nextchar(); |
1675 | ret = reg(1, &flags); |
1676 | if (ret == NULL) { |
1677 | if (flags & TRYAGAIN) |
1678 | goto tryagain; |
1679 | return(NULL); |
1680 | } |
c277df42 |
1681 | *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE); |
a0d0e21e |
1682 | break; |
1683 | case '|': |
1684 | case ')': |
1685 | if (flags & TRYAGAIN) { |
1686 | *flagp |= TRYAGAIN; |
1687 | return NULL; |
1688 | } |
3280af22 |
1689 | FAIL2("internal urp in regexp at /%s/", PL_regcomp_parse); |
a0d0e21e |
1690 | /* Supposed to be caught earlier. */ |
1691 | break; |
85afd4ae |
1692 | case '{': |
3280af22 |
1693 | if (!regcurly(PL_regcomp_parse)) { |
1694 | PL_regcomp_parse++; |
85afd4ae |
1695 | goto defchar; |
1696 | } |
1697 | /* FALL THROUGH */ |
a0d0e21e |
1698 | case '?': |
1699 | case '+': |
1700 | case '*': |
3115e423 |
1701 | FAIL("?+*{} follows nothing in regexp"); |
a0d0e21e |
1702 | break; |
1703 | case '\\': |
3280af22 |
1704 | switch (*++PL_regcomp_parse) { |
a0d0e21e |
1705 | case 'A': |
3280af22 |
1706 | PL_seen_zerolen++; |
c277df42 |
1707 | ret = reg_node(SBOL); |
a0d0e21e |
1708 | *flagp |= SIMPLE; |
1709 | nextchar(); |
1710 | break; |
1711 | case 'G': |
c277df42 |
1712 | ret = reg_node(GPOS); |
3280af22 |
1713 | PL_regseen |= REG_SEEN_GPOS; |
a0d0e21e |
1714 | *flagp |= SIMPLE; |
1715 | nextchar(); |
1716 | break; |
1717 | case 'Z': |
c277df42 |
1718 | ret = reg_node(SEOL); |
a0d0e21e |
1719 | *flagp |= SIMPLE; |
1720 | nextchar(); |
1721 | break; |
b85d18e9 |
1722 | case 'z': |
1723 | ret = reg_node(EOS); |
1724 | *flagp |= SIMPLE; |
3280af22 |
1725 | PL_seen_zerolen++; /* Do not optimize RE away */ |
b85d18e9 |
1726 | nextchar(); |
1727 | break; |
a0ed51b3 |
1728 | case 'C': |
1729 | ret = reg_node(SANY); |
1730 | *flagp |= HASWIDTH|SIMPLE; |
1731 | nextchar(); |
1732 | break; |
1733 | case 'X': |
1734 | ret = reg_node(CLUMP); |
1735 | *flagp |= HASWIDTH; |
1736 | nextchar(); |
1737 | if (UTF && !PL_utf8_mark) |
dfe13c55 |
1738 | is_utf8_mark((U8*)"~"); /* preload table */ |
a0ed51b3 |
1739 | break; |
a0d0e21e |
1740 | case 'w': |
a0ed51b3 |
1741 | ret = reg_node( |
1742 | UTF |
1743 | ? (LOC ? ALNUMLUTF8 : ALNUMUTF8) |
1744 | : (LOC ? ALNUML : ALNUM)); |
a0d0e21e |
1745 | *flagp |= HASWIDTH|SIMPLE; |
1746 | nextchar(); |
a0ed51b3 |
1747 | if (UTF && !PL_utf8_alnum) |
dfe13c55 |
1748 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e |
1749 | break; |
1750 | case 'W': |
a0ed51b3 |
1751 | ret = reg_node( |
1752 | UTF |
1753 | ? (LOC ? NALNUMLUTF8 : NALNUMUTF8) |
1754 | : (LOC ? NALNUML : NALNUM)); |
a0d0e21e |
1755 | *flagp |= HASWIDTH|SIMPLE; |
1756 | nextchar(); |
a0ed51b3 |
1757 | if (UTF && !PL_utf8_alnum) |
dfe13c55 |
1758 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e |
1759 | break; |
1760 | case 'b': |
3280af22 |
1761 | PL_seen_zerolen++; |
a0ed51b3 |
1762 | ret = reg_node( |
1763 | UTF |
1764 | ? (LOC ? BOUNDLUTF8 : BOUNDUTF8) |
1765 | : (LOC ? BOUNDL : BOUND)); |
a0d0e21e |
1766 | *flagp |= SIMPLE; |
1767 | nextchar(); |
a0ed51b3 |
1768 | if (UTF && !PL_utf8_alnum) |
dfe13c55 |
1769 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e |
1770 | break; |
1771 | case 'B': |
3280af22 |
1772 | PL_seen_zerolen++; |
a0ed51b3 |
1773 | ret = reg_node( |
1774 | UTF |
1775 | ? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8) |
1776 | : (LOC ? NBOUNDL : NBOUND)); |
a0d0e21e |
1777 | *flagp |= SIMPLE; |
1778 | nextchar(); |
a0ed51b3 |
1779 | if (UTF && !PL_utf8_alnum) |
dfe13c55 |
1780 | is_utf8_alnum((U8*)"a"); /* preload table */ |
a0d0e21e |
1781 | break; |
1782 | case 's': |
a0ed51b3 |
1783 | ret = reg_node( |
1784 | UTF |
1785 | ? (LOC ? SPACELUTF8 : SPACEUTF8) |
1786 | : (LOC ? SPACEL : SPACE)); |
a0d0e21e |
1787 | *flagp |= HASWIDTH|SIMPLE; |
1788 | nextchar(); |
a0ed51b3 |
1789 | if (UTF && !PL_utf8_space) |
dfe13c55 |
1790 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e |
1791 | break; |
1792 | case 'S': |
a0ed51b3 |
1793 | ret = reg_node( |
1794 | UTF |
1795 | ? (LOC ? NSPACELUTF8 : NSPACEUTF8) |
1796 | : (LOC ? NSPACEL : NSPACE)); |
a0d0e21e |
1797 | *flagp |= HASWIDTH|SIMPLE; |
1798 | nextchar(); |
a0ed51b3 |
1799 | if (UTF && !PL_utf8_space) |
dfe13c55 |
1800 | is_utf8_space((U8*)" "); /* preload table */ |
a0d0e21e |
1801 | break; |
1802 | case 'd': |
a0ed51b3 |
1803 | ret = reg_node(UTF ? DIGITUTF8 : DIGIT); |
a0d0e21e |
1804 | *flagp |= HASWIDTH|SIMPLE; |
1805 | nextchar(); |
a0ed51b3 |
1806 | if (UTF && !PL_utf8_digit) |
dfe13c55 |
1807 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e |
1808 | break; |
1809 | case 'D': |
a0ed51b3 |
1810 | ret = reg_node(UTF ? NDIGITUTF8 : NDIGIT); |
a0d0e21e |
1811 | *flagp |= HASWIDTH|SIMPLE; |
1812 | nextchar(); |
a0ed51b3 |
1813 | if (UTF && !PL_utf8_digit) |
dfe13c55 |
1814 | is_utf8_digit((U8*)"1"); /* preload table */ |
a0d0e21e |
1815 | break; |
a14b48bc |
1816 | case 'p': |
1817 | case 'P': |
1818 | { /* a lovely hack--pretend we saw [\pX] instead */ |
1819 | char* oldregxend = PL_regxend; |
1820 | |
1821 | if (PL_regcomp_parse[1] == '{') { |
1822 | PL_regxend = strchr(PL_regcomp_parse, '}'); |
1823 | if (!PL_regxend) |
1824 | FAIL("Missing right brace on \\p{}"); |
1825 | PL_regxend++; |
1826 | } |
1827 | else |
1828 | PL_regxend = PL_regcomp_parse + 2; |
1829 | PL_regcomp_parse--; |
1830 | |
1831 | ret = regclassutf8(); |
1832 | |
1833 | PL_regxend = oldregxend; |
1834 | PL_regcomp_parse--; |
1835 | nextchar(); |
1836 | *flagp |= HASWIDTH|SIMPLE; |
1837 | } |
1838 | break; |
a0d0e21e |
1839 | case 'n': |
1840 | case 'r': |
1841 | case 't': |
1842 | case 'f': |
1843 | case 'e': |
1844 | case 'a': |
1845 | case 'x': |
1846 | case 'c': |
1847 | case '0': |
1848 | goto defchar; |
1849 | case '1': case '2': case '3': case '4': |
1850 | case '5': case '6': case '7': case '8': case '9': |
1851 | { |
3280af22 |
1852 | I32 num = atoi(PL_regcomp_parse); |
a0d0e21e |
1853 | |
3280af22 |
1854 | if (num > 9 && num >= PL_regnpar) |
a0d0e21e |
1855 | goto defchar; |
1856 | else { |
3280af22 |
1857 | if (!SIZE_ONLY && num > PL_regcomp_rx->nparens) |
ef64f398 |
1858 | FAIL("reference to nonexistent group"); |
3280af22 |
1859 | PL_regsawback = 1; |
a0ed51b3 |
1860 | ret = reganode(FOLD |
1861 | ? (LOC ? REFFL : REFF) |
c8756f30 |
1862 | : REF, num); |
a0d0e21e |
1863 | *flagp |= HASWIDTH; |
3280af22 |
1864 | while (isDIGIT(*PL_regcomp_parse)) |
1865 | PL_regcomp_parse++; |
1866 | PL_regcomp_parse--; |
a0d0e21e |
1867 | nextchar(); |
1868 | } |
1869 | } |
1870 | break; |
1871 | case '\0': |
3280af22 |
1872 | if (PL_regcomp_parse >= PL_regxend) |
a0d0e21e |
1873 | FAIL("trailing \\ in regexp"); |
1874 | /* FALL THROUGH */ |
1875 | default: |
1876 | goto defchar; |
1877 | } |
1878 | break; |
4633a7c4 |
1879 | |
1880 | case '#': |
3280af22 |
1881 | if (PL_regflags & PMf_EXTENDED) { |
1882 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != '\n') PL_regcomp_parse++; |
1883 | if (PL_regcomp_parse < PL_regxend) |
4633a7c4 |
1884 | goto tryagain; |
1885 | } |
1886 | /* FALL THROUGH */ |
1887 | |
a0d0e21e |
1888 | default: { |
1889 | register I32 len; |
a0ed51b3 |
1890 | register UV ender; |
a0d0e21e |
1891 | register char *p; |
c277df42 |
1892 | char *oldp, *s; |
a0d0e21e |
1893 | I32 numlen; |
1894 | |
3280af22 |
1895 | PL_regcomp_parse++; |
a0d0e21e |
1896 | |
1897 | defchar: |
a0ed51b3 |
1898 | ret = reg_node(FOLD |
1899 | ? (LOC ? EXACTFL : EXACTF) |
bbce6d69 |
1900 | : EXACT); |
161b471a |
1901 | s = (char *) OPERAND(ret); |
c277df42 |
1902 | regc(0, s++); /* save spot for len */ |
3280af22 |
1903 | for (len = 0, p = PL_regcomp_parse - 1; |
1904 | len < 127 && p < PL_regxend; |
a0d0e21e |
1905 | len++) |
1906 | { |
1907 | oldp = p; |
5b5a24f7 |
1908 | |
3280af22 |
1909 | if (PL_regflags & PMf_EXTENDED) |
1910 | p = regwhite(p, PL_regxend); |
a0d0e21e |
1911 | switch (*p) { |
1912 | case '^': |
1913 | case '$': |
1914 | case '.': |
1915 | case '[': |
1916 | case '(': |
1917 | case ')': |
1918 | case '|': |
1919 | goto loopdone; |
1920 | case '\\': |
1921 | switch (*++p) { |
1922 | case 'A': |
1923 | case 'G': |
1924 | case 'Z': |
b85d18e9 |
1925 | case 'z': |
a0d0e21e |
1926 | case 'w': |
1927 | case 'W': |
1928 | case 'b': |
1929 | case 'B': |
1930 | case 's': |
1931 | case 'S': |
1932 | case 'd': |
1933 | case 'D': |
a14b48bc |
1934 | case 'p': |
1935 | case 'P': |
a0d0e21e |
1936 | --p; |
1937 | goto loopdone; |
1938 | case 'n': |
1939 | ender = '\n'; |
1940 | p++; |
a687059c |
1941 | break; |
a0d0e21e |
1942 | case 'r': |
1943 | ender = '\r'; |
1944 | p++; |
a687059c |
1945 | break; |
a0d0e21e |
1946 | case 't': |
1947 | ender = '\t'; |
1948 | p++; |
a687059c |
1949 | break; |
a0d0e21e |
1950 | case 'f': |
1951 | ender = '\f'; |
1952 | p++; |
a687059c |
1953 | break; |
a0d0e21e |
1954 | case 'e': |
1955 | ender = '\033'; |
1956 | p++; |
a687059c |
1957 | break; |
a0d0e21e |
1958 | case 'a': |
1959 | ender = '\007'; |
1960 | p++; |
a687059c |
1961 | break; |
a0d0e21e |
1962 | case 'x': |
a0ed51b3 |
1963 | if (*++p == '{') { |
1964 | char* e = strchr(p, '}'); |
1965 | |
1966 | if (!e) |
1967 | FAIL("Missing right brace on \\x{}"); |
1968 | else if (UTF) { |
1969 | ender = scan_hex(p + 1, e - p, &numlen); |
1970 | if (numlen + len >= 127) { /* numlen is generous */ |
1971 | p--; |
1972 | goto loopdone; |
1973 | } |
1974 | p = e + 1; |
1975 | } |
1976 | else |
1977 | FAIL("Can't use \\x{} without 'use utf8' declaration"); |
1978 | } |
1979 | else { |
1980 | ender = scan_hex(p, 2, &numlen); |
1981 | p += numlen; |
1982 | } |
a687059c |
1983 | break; |
a0d0e21e |
1984 | case 'c': |
1985 | p++; |
bbce6d69 |
1986 | ender = UCHARAT(p++); |
1987 | ender = toCTRL(ender); |
a687059c |
1988 | break; |
a0d0e21e |
1989 | case '0': case '1': case '2': case '3':case '4': |
1990 | case '5': case '6': case '7': case '8':case '9': |
1991 | if (*p == '0' || |
3280af22 |
1992 | (isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) { |
a0d0e21e |
1993 | ender = scan_oct(p, 3, &numlen); |
1994 | p += numlen; |
1995 | } |
1996 | else { |
1997 | --p; |
1998 | goto loopdone; |
a687059c |
1999 | } |
2000 | break; |
a0d0e21e |
2001 | case '\0': |
3280af22 |
2002 | if (p >= PL_regxend) |
a687059c |
2003 | FAIL("trailing \\ in regexp"); |
2004 | /* FALL THROUGH */ |
a0d0e21e |
2005 | default: |
a0ed51b3 |
2006 | goto normal_default; |
a0d0e21e |
2007 | } |
2008 | break; |
a687059c |
2009 | default: |
a0ed51b3 |
2010 | normal_default: |
2011 | if ((*p & 0xc0) == 0xc0 && UTF) { |
dfe13c55 |
2012 | ender = utf8_to_uv((U8*)p, &numlen); |
a0ed51b3 |
2013 | p += numlen; |
2014 | } |
2015 | else |
2016 | ender = *p++; |
a0d0e21e |
2017 | break; |
a687059c |
2018 | } |
3280af22 |
2019 | if (PL_regflags & PMf_EXTENDED) |
2020 | p = regwhite(p, PL_regxend); |
a0ed51b3 |
2021 | if (UTF && FOLD) { |
2022 | if (LOC) |
2023 | ender = toLOWER_LC_uni(ender); |
2024 | else |
2025 | ender = toLOWER_uni(ender); |
2026 | } |
a0d0e21e |
2027 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
2028 | if (len) |
2029 | p = oldp; |
a0ed51b3 |
2030 | else if (ender >= 0x80 && UTF) { |
2031 | reguni(ender, s, &numlen); |
2032 | s += numlen; |
2033 | len += numlen; |
2034 | } |
a0d0e21e |
2035 | else { |
2036 | len++; |
c277df42 |
2037 | regc(ender, s++); |
a0d0e21e |
2038 | } |
2039 | break; |
a687059c |
2040 | } |
a0ed51b3 |
2041 | if (ender >= 0x80 && UTF) { |
2042 | reguni(ender, s, &numlen); |
2043 | s += numlen; |
2044 | len += numlen - 1; |
2045 | } |
2046 | else |
2047 | regc(ender, s++); |
a0d0e21e |
2048 | } |
2049 | loopdone: |
3280af22 |
2050 | PL_regcomp_parse = p - 1; |
a0d0e21e |
2051 | nextchar(); |
2052 | if (len < 0) |
2053 | FAIL("internal disaster in regexp"); |
2054 | if (len > 0) |
2055 | *flagp |= HASWIDTH; |
2056 | if (len == 1) |
2057 | *flagp |= SIMPLE; |
c277df42 |
2058 | if (!SIZE_ONLY) |
a0d0e21e |
2059 | *OPERAND(ret) = len; |
c277df42 |
2060 | regc('\0', s++); |
2061 | if (SIZE_ONLY) { |
3280af22 |
2062 | PL_regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
a0ed51b3 |
2063 | } |
2064 | else { |
3280af22 |
2065 | PL_regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
c277df42 |
2066 | } |
a687059c |
2067 | } |
a0d0e21e |
2068 | break; |
2069 | } |
a687059c |
2070 | |
a0d0e21e |
2071 | return(ret); |
a687059c |
2072 | } |
2073 | |
873ef191 |
2074 | STATIC char * |
8ac85365 |
2075 | regwhite(char *p, char *e) |
5b5a24f7 |
2076 | { |
2077 | while (p < e) { |
2078 | if (isSPACE(*p)) |
2079 | ++p; |
2080 | else if (*p == '#') { |
2081 | do { |
2082 | p++; |
2083 | } while (p < e && *p != '\n'); |
2084 | } |
2085 | else |
2086 | break; |
2087 | } |
2088 | return p; |
2089 | } |
2090 | |
76e3520e |
2091 | STATIC regnode * |
8ac85365 |
2092 | regclass(void) |
a687059c |
2093 | { |
5c0ca799 |
2094 | dTHR; |
c277df42 |
2095 | register char *opnd, *s; |
a0ed51b3 |
2096 | register I32 value; |
2097 | register I32 lastvalue = 1234; |
a0d0e21e |
2098 | register I32 range = 0; |
c277df42 |
2099 | register regnode *ret; |
a0d0e21e |
2100 | register I32 def; |
2101 | I32 numlen; |
2102 | |
3280af22 |
2103 | s = opnd = (char *) OPERAND(PL_regcode); |
c277df42 |
2104 | ret = reg_node(ANYOF); |
a0ed51b3 |
2105 | for (value = 0; value < 33; value++) |
c277df42 |
2106 | regc(0, s++); |
3280af22 |
2107 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2108 | PL_regnaughty++; |
2109 | PL_regcomp_parse++; |
c277df42 |
2110 | if (!SIZE_ONLY) |
bbce6d69 |
2111 | *opnd |= ANYOF_INVERT; |
2112 | } |
c277df42 |
2113 | if (!SIZE_ONLY) { |
3280af22 |
2114 | PL_regcode += ANY_SKIP; |
a0ed51b3 |
2115 | if (FOLD) |
bbce6d69 |
2116 | *opnd |= ANYOF_FOLD; |
a0ed51b3 |
2117 | if (LOC) |
bbce6d69 |
2118 | *opnd |= ANYOF_LOCALE; |
a0ed51b3 |
2119 | } |
2120 | else { |
3280af22 |
2121 | PL_regsize += ANY_SKIP; |
a0d0e21e |
2122 | } |
3280af22 |
2123 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
a0d0e21e |
2124 | goto skipcond; /* allow 1st char to be ] or - */ |
3280af22 |
2125 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
a0d0e21e |
2126 | skipcond: |
a0ed51b3 |
2127 | value = UCHARAT(PL_regcomp_parse++); |
2128 | if (value == '[' && PL_regcomp_parse + 1 < PL_regxend && |
4599a1de |
2129 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ |
3280af22 |
2130 | (*PL_regcomp_parse == ':' || *PL_regcomp_parse == '=' || *PL_regcomp_parse == '.')) { |
2131 | char posixccc = *PL_regcomp_parse; |
2132 | char* posixccs = PL_regcomp_parse++; |
4599a1de |
2133 | |
3280af22 |
2134 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != posixccc) |
2135 | PL_regcomp_parse++; |
2136 | if (PL_regcomp_parse == PL_regxend) |
4599a1de |
2137 | /* Grandfather lone [:, [=, [. */ |
3280af22 |
2138 | PL_regcomp_parse = posixccs; |
4599a1de |
2139 | else { |
3280af22 |
2140 | PL_regcomp_parse++; /* skip over the posixccc */ |
2141 | if (*PL_regcomp_parse == ']') { |
4599a1de |
2142 | /* Not Implemented Yet. |
2143 | * (POSIX Extended Character Classes, that is) |
2144 | * The text between e.g. [: and :] would start |
77d41b28 |
2145 | * at posixccs + 1 and stop at regcomp_parse - 2. */ |
599cee73 |
2146 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY) |
2147 | warner(WARN_UNSAFE, |
2148 | "Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc); |
3280af22 |
2149 | PL_regcomp_parse++; /* skip over the ending ] */ |
4599a1de |
2150 | } |
2151 | } |
2152 | } |
a0ed51b3 |
2153 | if (value == '\\') { |
2154 | value = UCHARAT(PL_regcomp_parse++); |
2155 | switch (value) { |
a0d0e21e |
2156 | case 'w': |
ae5c130c |
2157 | if (!SIZE_ONLY) { |
a0ed51b3 |
2158 | if (LOC) |
bbce6d69 |
2159 | *opnd |= ANYOF_ALNUML; |
ae5c130c |
2160 | else { |
a0ed51b3 |
2161 | for (value = 0; value < 256; value++) |
2162 | if (isALNUM(value)) |
2163 | ANYOF_SET(opnd, value); |
ae5c130c |
2164 | } |
bbce6d69 |
2165 | } |
a0ed51b3 |
2166 | lastvalue = 1234; |
a0d0e21e |
2167 | continue; |
2168 | case 'W': |
ae5c130c |
2169 | if (!SIZE_ONLY) { |
a0ed51b3 |
2170 | if (LOC) |
bbce6d69 |
2171 | *opnd |= ANYOF_NALNUML; |
ae5c130c |
2172 | else { |
a0ed51b3 |
2173 | for (value = 0; value < 256; value++) |
2174 | if (!isALNUM(value)) |
2175 | ANYOF_SET(opnd, value); |
ae5c130c |
2176 | } |
bbce6d69 |
2177 | } |
a0ed51b3 |
2178 | lastvalue = 1234; |
a0d0e21e |
2179 | continue; |
2180 | case 's': |
ae5c130c |
2181 | if (!SIZE_ONLY) { |
a0ed51b3 |
2182 | if (LOC) |
bbce6d69 |
2183 | *opnd |= ANYOF_SPACEL; |
ae5c130c |
2184 | else { |
a0ed51b3 |
2185 | for (value = 0; value < 256; value++) |
2186 | if (isSPACE(value)) |
2187 | ANYOF_SET(opnd, value); |
ae5c130c |
2188 | } |
bbce6d69 |
2189 | } |
a0ed51b3 |
2190 | lastvalue = 1234; |
a0d0e21e |
2191 | continue; |
2192 | case 'S': |
ae5c130c |
2193 | if (!SIZE_ONLY) { |
a0ed51b3 |
2194 | if (LOC) |
bbce6d69 |
2195 | *opnd |= ANYOF_NSPACEL; |
ae5c130c |
2196 | else { |
a0ed51b3 |
2197 | for (value = 0; value < 256; value++) |
2198 | if (!isSPACE(value)) |
2199 | ANYOF_SET(opnd, value); |
ae5c130c |
2200 | } |
bbce6d69 |
2201 | } |
a0ed51b3 |
2202 | lastvalue = 1234; |
a0d0e21e |
2203 | continue; |
2204 | case 'd': |
ae5c130c |
2205 | if (!SIZE_ONLY) { |
a0ed51b3 |
2206 | for (value = '0'; value <= '9'; value++) |
2207 | ANYOF_SET(opnd, value); |
ae5c130c |
2208 | } |
a0ed51b3 |
2209 | lastvalue = 1234; |
a0d0e21e |
2210 | continue; |
2211 | case 'D': |
ae5c130c |
2212 | if (!SIZE_ONLY) { |
a0ed51b3 |
2213 | for (value = 0; value < '0'; value++) |
2214 | ANYOF_SET(opnd, value); |
2215 | for (value = '9' + 1; value < 256; value++) |
2216 | ANYOF_SET(opnd, value); |
ae5c130c |
2217 | } |
a0ed51b3 |
2218 | lastvalue = 1234; |
a0d0e21e |
2219 | continue; |
2220 | case 'n': |
a0ed51b3 |
2221 | value = '\n'; |
a0d0e21e |
2222 | break; |
2223 | case 'r': |
a0ed51b3 |
2224 | value = '\r'; |
a0d0e21e |
2225 | break; |
2226 | case 't': |
a0ed51b3 |
2227 | value = '\t'; |
a0d0e21e |
2228 | break; |
2229 | case 'f': |
a0ed51b3 |
2230 | value = '\f'; |
a0d0e21e |
2231 | break; |
2232 | case 'b': |
a0ed51b3 |
2233 | value = '\b'; |
a0d0e21e |
2234 | break; |
2235 | case 'e': |
a0ed51b3 |
2236 | value = '\033'; |
a0d0e21e |
2237 | break; |
2238 | case 'a': |
a0ed51b3 |
2239 | value = '\007'; |
a0d0e21e |
2240 | break; |
2241 | case 'x': |
a0ed51b3 |
2242 | value = scan_hex(PL_regcomp_parse, 2, &numlen); |
3280af22 |
2243 | PL_regcomp_parse += numlen; |
a0d0e21e |
2244 | break; |
2245 | case 'c': |
a0ed51b3 |
2246 | value = UCHARAT(PL_regcomp_parse++); |
2247 | value = toCTRL(value); |
a0d0e21e |
2248 | break; |
2249 | case '0': case '1': case '2': case '3': case '4': |
2250 | case '5': case '6': case '7': case '8': case '9': |
a0ed51b3 |
2251 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); |
3280af22 |
2252 | PL_regcomp_parse += numlen; |
a0d0e21e |
2253 | break; |
2254 | } |
2255 | } |
2256 | if (range) { |
a0ed51b3 |
2257 | if (lastvalue > value) |
a0d0e21e |
2258 | FAIL("invalid [] range in regexp"); |
2259 | range = 0; |
2260 | } |
2261 | else { |
a0ed51b3 |
2262 | lastvalue = value; |
3280af22 |
2263 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
2264 | PL_regcomp_parse[1] != ']') { |
2265 | PL_regcomp_parse++; |
a0d0e21e |
2266 | range = 1; |
2267 | continue; /* do it next time */ |
2268 | } |
a687059c |
2269 | } |
ae5c130c |
2270 | if (!SIZE_ONLY) { |
8ada0baa |
2271 | #ifndef ASCIIish |
2272 | if ((isLOWER(lastvalue) && isLOWER(value)) || |
2273 | (isUPPER(lastvalue) && isUPPER(value))) { |
2274 | if (isLOWER(lastvalue)) { |
2275 | for (i = lastvalue; i <= value; i++) |
2276 | if (isLOWER(i)) |
2277 | ANYOF_SET(opnd, i); |
2278 | } else { |
2279 | for (i = lastvalue; i <= value; i++) |
2280 | if (isUPPER(i)) |
2281 | ANYOF_SET(opnd, i); |
2282 | } |
2283 | } |
2284 | else |
2285 | #endif |
2286 | for ( ; lastvalue <= value; lastvalue++) |
2287 | ANYOF_SET(opnd, lastvalue); |
2288 | } |
a0ed51b3 |
2289 | lastvalue = value; |
a0d0e21e |
2290 | } |
ae5c130c |
2291 | /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ |
2292 | if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) { |
a0ed51b3 |
2293 | for (value = 0; value < 256; ++value) { |
2294 | if (ANYOF_TEST(opnd, value)) { |
2295 | I32 cf = fold[value]; |
ae5c130c |
2296 | ANYOF_SET(opnd, cf); |
2297 | } |
2298 | } |
2299 | *opnd &= ~ANYOF_FOLD; |
2300 | } |
2301 | /* optimize inverted simple patterns (e.g. [^a-z]) */ |
2302 | if (!SIZE_ONLY && (*opnd & 0xFF) == ANYOF_INVERT) { |
a0ed51b3 |
2303 | for (value = 0; value < 32; ++value) |
2304 | opnd[1 + value] ^= 0xFF; |
ae5c130c |
2305 | *opnd = 0; |
2306 | } |
a0d0e21e |
2307 | return ret; |
2308 | } |
2309 | |
a0ed51b3 |
2310 | STATIC regnode * |
2311 | regclassutf8(void) |
2312 | { |
2313 | register char *opnd, *e; |
2314 | register U32 value; |
2315 | register U32 lastvalue = 123456; |
2316 | register I32 range = 0; |
2317 | register regnode *ret; |
2318 | I32 numlen; |
2319 | I32 n; |
2320 | SV *listsv; |
2321 | U8 flags = 0; |
c485e607 |
2322 | dTHR; |
a0ed51b3 |
2323 | |
2324 | if (*PL_regcomp_parse == '^') { /* Complement of range. */ |
2325 | PL_regnaughty++; |
2326 | PL_regcomp_parse++; |
2327 | if (!SIZE_ONLY) |
2328 | flags |= ANYOF_INVERT; |
2329 | } |
2330 | if (!SIZE_ONLY) { |
2331 | if (FOLD) |
2332 | flags |= ANYOF_FOLD; |
2333 | if (LOC) |
2334 | flags |= ANYOF_LOCALE; |
2335 | listsv = newSVpv("# comment\n",0); |
2336 | } |
2337 | |
2338 | if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-') |
2339 | goto skipcond; /* allow 1st char to be ] or - */ |
2340 | |
2341 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') { |
2342 | skipcond: |
dfe13c55 |
2343 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 |
2344 | PL_regcomp_parse += numlen; |
2345 | |
2346 | if (value == '[' && PL_regcomp_parse + 1 < PL_regxend && |
2347 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ |
2348 | (*PL_regcomp_parse == ':' || *PL_regcomp_parse == '=' || *PL_regcomp_parse == '.')) { |
2349 | char posixccc = *PL_regcomp_parse; |
2350 | char* posixccs = PL_regcomp_parse++; |
2351 | |
2352 | while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != posixccc) |
2353 | PL_regcomp_parse++; |
2354 | if (PL_regcomp_parse == PL_regxend) |
2355 | /* Grandfather lone [:, [=, [. */ |
2356 | PL_regcomp_parse = posixccs; |
2357 | else { |
2358 | PL_regcomp_parse++; /* skip over the posixccc */ |
2359 | if (*PL_regcomp_parse == ']') { |
2360 | /* Not Implemented Yet. |
2361 | * (POSIX Extended Character Classes, that is) |
2362 | * The text between e.g. [: and :] would start |
2363 | * at posixccs + 1 and stop at regcomp_parse - 2. */ |
599cee73 |
2364 | if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY) |
2365 | warner(WARN_UNSAFE, |
2366 | "Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc); |
a0ed51b3 |
2367 | PL_regcomp_parse++; /* skip over the ending ] */ |
2368 | } |
2369 | } |
2370 | } |
2371 | |
2372 | if (value == '\\') { |
dfe13c55 |
2373 | value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen); |
a0ed51b3 |
2374 | PL_regcomp_parse += numlen; |
2375 | switch (value) { |
2376 | case 'w': |
2377 | if (!SIZE_ONLY) { |
2378 | if (LOC) |
2379 | flags |= ANYOF_ALNUML; |
2380 | |
2381 | sv_catpvf(listsv, "+utf8::IsAlnum\n"); |
2382 | } |
2383 | lastvalue = 123456; |
2384 | continue; |
2385 | case 'W': |
2386 | if (!SIZE_ONLY) { |
2387 | if (LOC) |
2388 | flags |= ANYOF_NALNUML; |
2389 | |
2390 | sv_catpvf(listsv, |
2391 | "-utf8::IsAlpha\n-utf8::IsDigit\n0000\t%04x\n%04x\tffff\n", |
2392 | '_' - 1, |
2393 | '_' + 1); |
2394 | } |
2395 | lastvalue = 123456; |
2396 | continue; |
2397 | case 's': |
2398 | if (!SIZE_ONLY) { |
2399 | if (LOC) |
2400 | flags |= ANYOF_SPACEL; |
2401 | sv_catpvf(listsv, "+utf8::IsSpace\n"); |
2402 | if (!PL_utf8_space) |
dfe13c55 |
2403 | is_utf8_space((U8*)" "); |
a0ed51b3 |
2404 | } |
2405 | lastvalue = 123456; |
2406 | continue; |
2407 | case 'S': |
2408 | if (!SIZE_ONLY) { |
2409 | if (LOC) |
2410 | flags |= ANYOF_NSPACEL; |
2411 | sv_catpvf(listsv, |
2412 | "!utf8::IsSpace\n"); |
2413 | if (!PL_utf8_space) |
dfe13c55 |
2414 | is_utf8_space((U8*)" "); |
a0ed51b3 |
2415 | } |
2416 | lastvalue = 123456; |
2417 | continue; |
2418 | case 'd': |
2419 | if (!SIZE_ONLY) { |
2420 | sv_catpvf(listsv, "+utf8::IsDigit\n"); |
2421 | } |
2422 | lastvalue = 123456; |
2423 | continue; |
2424 | case 'D': |
2425 | if (!SIZE_ONLY) { |
2426 | sv_catpvf(listsv, |
2427 | "!utf8::IsDigit\n"); |
2428 | } |
2429 | lastvalue = 123456; |
2430 | continue; |
2431 | case 'p': |
2432 | case 'P': |
2433 | if (*PL_regcomp_parse == '{') { |
2434 | e = strchr(PL_regcomp_parse++, '}'); |
2435 | if (!e) |
2436 | FAIL("Missing right brace on \\p{}"); |
2437 | n = e - PL_regcomp_parse; |
2438 | } |
2439 | else { |
2440 | e = PL_regcomp_parse; |
2441 | n = 1; |
2442 | } |
2443 | if (!SIZE_ONLY) { |
2444 | if (value == 'p') |
2445 | sv_catpvf(listsv, "+utf8::%.*s\n", n, PL_regcomp_parse); |
2446 | else |
2447 | sv_catpvf(listsv, |
2448 | "!utf8::%.*s\n", n, PL_regcomp_parse); |
2449 | } |
2450 | PL_regcomp_parse = e + 1; |
2451 | lastvalue = 123456; |
2452 | continue; |
2453 | case 'n': |
2454 | value = '\n'; |
2455 | break; |
2456 | case 'r': |
2457 | value = '\r'; |
2458 | break; |
2459 | case 't': |
2460 | value = '\t'; |
2461 | break; |
2462 | case 'f': |
2463 | value = '\f'; |
2464 | break; |
2465 | case 'b': |
2466 | value = '\b'; |
2467 | break; |
2468 | case 'e': |
2469 | value = '\033'; |
2470 | break; |
2471 | case 'a': |
2472 | value = '\007'; |
2473 | break; |
2474 | case 'x': |
2475 | if (*PL_regcomp_parse == '{') { |
2476 | e = strchr(PL_regcomp_parse++, '}'); |
2477 | if (!e) |
2478 | FAIL("Missing right brace on \\x{}"); |
2479 | value = scan_hex(PL_regcomp_parse + 1, e - PL_regcomp_parse, &numlen); |
2480 | PL_regcomp_parse = e + 1; |
2481 | } |
2482 | else { |
2483 | value = scan_hex(PL_regcomp_parse, 2, &numlen); |
2484 | PL_regcomp_parse += numlen; |
2485 | } |
2486 | break; |
2487 | case 'c': |
2488 | value = UCHARAT(PL_regcomp_parse++); |
2489 | value = toCTRL(value); |
2490 | break; |
2491 | case '0': case '1': case '2': case '3': case '4': |
2492 | case '5': case '6': case '7': case '8': case '9': |
2493 | value = scan_oct(--PL_regcomp_parse, 3, &numlen); |
2494 | PL_regcomp_parse += numlen; |
2495 | break; |
2496 | } |
2497 | } |
2498 | if (range) { |
2499 | if (lastvalue > value) |
2500 | FAIL("invalid [] range in regexp"); |
2501 | if (!SIZE_ONLY) |
2502 | sv_catpvf(listsv, "%04x\t%04x\n", lastvalue, value); |
2503 | lastvalue = value; |
2504 | range = 0; |
2505 | } |
2506 | else { |
2507 | lastvalue = value; |
2508 | if (*PL_regcomp_parse == '-' && PL_regcomp_parse+1 < PL_regxend && |
2509 | PL_regcomp_parse[1] != ']') { |
2510 | PL_regcomp_parse++; |
2511 | range = 1; |
2512 | continue; /* do it next time */ |
2513 | } |
2514 | if (!SIZE_ONLY) |
2515 | sv_catpvf(listsv, "%04x\n", value); |
2516 | } |
2517 | } |
a0ed51b3 |
2518 | |
2519 | ret = reganode(ANYOFUTF8, 0); |
2520 | |
2521 | if (!SIZE_ONLY) { |
2522 | SV *rv = swash_init("utf8", "", listsv, 1, 0); |
2523 | SvREFCNT_dec(listsv); |
2524 | n = add_data(1,"s"); |
2525 | PL_regcomp_rx->data->data[n] = (void*)rv; |
2526 | ARG1_SET(ret, flags); |
2527 | ARG2_SET(ret, n); |
2528 | } |
2529 | |
2530 | return ret; |
2531 | } |
2532 | |
76e3520e |
2533 | STATIC char* |
8ac85365 |
2534 | nextchar(void) |
a0d0e21e |
2535 | { |
5c0ca799 |
2536 | dTHR; |
3280af22 |
2537 | char* retval = PL_regcomp_parse++; |
a0d0e21e |
2538 | |
4633a7c4 |
2539 | for (;;) { |
3280af22 |
2540 | if (*PL_regcomp_parse == '(' && PL_regcomp_parse[1] == '?' && |
2541 | PL_regcomp_parse[2] == '#') { |
2542 | while (*PL_regcomp_parse && *PL_regcomp_parse != ')') |
2543 | PL_regcomp_parse++; |
2544 | PL_regcomp_parse++; |
4633a7c4 |
2545 | continue; |
2546 | } |
3280af22 |
2547 | if (PL_regflags & PMf_EXTENDED) { |
2548 | if (isSPACE(*PL_regcomp_parse)) { |
2549 | PL_regcomp_parse++; |
748a9306 |
2550 | continue; |
2551 | } |
3280af22 |
2552 | else if (*PL_regcomp_parse == '#') { |
2553 | while (*PL_regcomp_parse && *PL_regcomp_parse != '\n') |
2554 | PL_regcomp_parse++; |
2555 | PL_regcomp_parse++; |
748a9306 |
2556 | continue; |
2557 | } |
748a9306 |
2558 | } |
4633a7c4 |
2559 | return retval; |
a0d0e21e |
2560 | } |
a687059c |
2561 | } |
2562 | |
2563 | /* |
c277df42 |
2564 | - reg_node - emit a node |
a0d0e21e |
2565 | */ |
76e3520e |
2566 | STATIC regnode * /* Location. */ |
c277df42 |
2567 | reg_node(U8 op) |
a687059c |
2568 | { |
5c0ca799 |
2569 | dTHR; |
c277df42 |
2570 | register regnode *ret; |
2571 | register regnode *ptr; |
a687059c |
2572 | |
3280af22 |
2573 | ret = PL_regcode; |
c277df42 |
2574 | if (SIZE_ONLY) { |
6b88bc9c |
2575 | SIZE_ALIGN(PL_regsize); |
3280af22 |
2576 | PL_regsize += 1; |
a0d0e21e |
2577 | return(ret); |
2578 | } |
a687059c |
2579 | |
c277df42 |
2580 | NODE_ALIGN_FILL(ret); |
a0d0e21e |
2581 | ptr = ret; |
c277df42 |
2582 | FILL_ADVANCE_NODE(ptr, op); |
3280af22 |
2583 | PL_regcode = ptr; |
a687059c |
2584 | |
a0d0e21e |
2585 | return(ret); |
a687059c |
2586 | } |
2587 | |
2588 | /* |
a0d0e21e |
2589 | - reganode - emit a node with an argument |
2590 | */ |
76e3520e |
2591 | STATIC regnode * /* Location. */ |
c277df42 |
2592 | reganode(U8 op, U32 arg) |
fe14fcc3 |
2593 | { |
5c0ca799 |
2594 | dTHR; |
c277df42 |
2595 | register regnode *ret; |
2596 | register regnode *ptr; |
fe14fcc3 |
2597 | |
3280af22 |
2598 | ret = PL_regcode; |
c277df42 |
2599 | if (SIZE_ONLY) { |
6b88bc9c |
2600 | SIZE_ALIGN(PL_regsize); |
3280af22 |
2601 | PL_regsize += 2; |
a0d0e21e |
2602 | return(ret); |
2603 | } |
fe14fcc3 |
2604 | |
c277df42 |
2605 | NODE_ALIGN_FILL(ret); |
a0d0e21e |
2606 | ptr = ret; |
c277df42 |
2607 | FILL_ADVANCE_NODE_ARG(ptr, op, arg); |
3280af22 |
2608 | PL_regcode = ptr; |
fe14fcc3 |
2609 | |
a0d0e21e |
2610 | return(ret); |
fe14fcc3 |
2611 | } |
2612 | |
2613 | /* |
a0ed51b3 |
2614 | - regc - emit (if appropriate) a Unicode character |
2615 | */ |
2616 | STATIC void |
2617 | reguni(UV uv, char* s, I32* lenp) |
2618 | { |
c485e607 |
2619 | dTHR; |
a0ed51b3 |
2620 | if (SIZE_ONLY) { |
dfe13c55 |
2621 | U8 tmpbuf[10]; |
a0ed51b3 |
2622 | *lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf; |
2623 | } |
2624 | else |
dfe13c55 |
2625 | *lenp = uv_to_utf8((U8*)s, uv) - (U8*)s; |
a0ed51b3 |
2626 | |
2627 | } |
2628 | |
2629 | /* |
a0d0e21e |
2630 | - regc - emit (if appropriate) a byte of code |
2631 | */ |
76e3520e |
2632 | STATIC void |
c277df42 |
2633 | regc(U8 b, char* s) |
a687059c |
2634 | { |
5c0ca799 |
2635 | dTHR; |
c277df42 |
2636 | if (!SIZE_ONLY) |
2637 | *s = b; |
a687059c |
2638 | } |
2639 | |
2640 | /* |
a0d0e21e |
2641 | - reginsert - insert an operator in front of already-emitted operand |
2642 | * |
2643 | * Means relocating the operand. |
2644 | */ |
76e3520e |
2645 | STATIC void |
c277df42 |
2646 | reginsert(U8 op, regnode *opnd) |
a687059c |
2647 | { |
5c0ca799 |
2648 | dTHR; |
c277df42 |
2649 | register regnode *src; |
2650 | register regnode *dst; |
2651 | register regnode *place; |
2652 | register int offset = regarglen[(U8)op]; |
2653 | |
2654 | /* (regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */ |
2655 | |
2656 | if (SIZE_ONLY) { |
3280af22 |
2657 | PL_regsize += NODE_STEP_REGNODE + offset; |
a0d0e21e |
2658 | return; |
2659 | } |
a687059c |
2660 | |
3280af22 |
2661 | src = PL_regcode; |
2662 | PL_regcode += NODE_STEP_REGNODE + offset; |
2663 | dst = PL_regcode; |
a0d0e21e |
2664 | while (src > opnd) |
c277df42 |
2665 | StructCopy(--src, --dst, regnode); |
a0d0e21e |
2666 | |
2667 | place = opnd; /* Op node, where operand used to be. */ |
c277df42 |
2668 | src = NEXTOPER(place); |
2669 | FILL_ADVANCE_NODE(place, op); |
2670 | Zero(src, offset, regnode); |
a687059c |
2671 | } |
2672 | |
2673 | /* |
c277df42 |
2674 | - regtail - set the next-pointer at the end of a node chain of p to val. |
a0d0e21e |
2675 | */ |
76e3520e |
2676 | STATIC void |
c277df42 |
2677 | regtail(regnode *p, regnode *val) |
a687059c |
2678 | { |
5c0ca799 |
2679 | dTHR; |
c277df42 |
2680 | register regnode *scan; |
2681 | register regnode *temp; |
a0d0e21e |
2682 | register I32 offset; |
2683 | |
c277df42 |
2684 | if (SIZE_ONLY) |
a0d0e21e |
2685 | return; |
2686 | |
2687 | /* Find last node. */ |
2688 | scan = p; |
2689 | for (;;) { |
2690 | temp = regnext(scan); |
2691 | if (temp == NULL) |
2692 | break; |
2693 | scan = temp; |
2694 | } |
a687059c |
2695 | |
c277df42 |
2696 | if (reg_off_by_arg[OP(scan)]) { |
2697 | ARG_SET(scan, val - scan); |
a0ed51b3 |
2698 | } |
2699 | else { |
c277df42 |
2700 | NEXT_OFF(scan) = val - scan; |
2701 | } |
a687059c |
2702 | } |
2703 | |
2704 | /* |
a0d0e21e |
2705 | - regoptail - regtail on operand of first argument; nop if operandless |
2706 | */ |
76e3520e |
2707 | STATIC void |
c277df42 |
2708 | regoptail(regnode *p, regnode *val) |
a687059c |
2709 | { |
5c0ca799 |
2710 | dTHR; |
a0d0e21e |
2711 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
c277df42 |
2712 | if (p == NULL || SIZE_ONLY) |
2713 | return; |
2714 | if (regkind[(U8)OP(p)] == BRANCH) { |
2715 | regtail(NEXTOPER(p), val); |
a0ed51b3 |
2716 | } |
2717 | else if ( regkind[(U8)OP(p)] == BRANCHJ) { |
c277df42 |
2718 | regtail(NEXTOPER(NEXTOPER(p)), val); |
a0ed51b3 |
2719 | } |
2720 | else |
a0d0e21e |
2721 | return; |
a687059c |
2722 | } |
2723 | |
2724 | /* |
2725 | - regcurly - a little FSA that accepts {\d+,?\d*} |
2726 | */ |
79072805 |
2727 | STATIC I32 |
8ac85365 |
2728 | regcurly(register char *s) |
a687059c |
2729 | { |
2730 | if (*s++ != '{') |
2731 | return FALSE; |
f0fcb552 |
2732 | if (!isDIGIT(*s)) |
a687059c |
2733 | return FALSE; |
f0fcb552 |
2734 | while (isDIGIT(*s)) |
a687059c |
2735 | s++; |
2736 | if (*s == ',') |
2737 | s++; |
f0fcb552 |
2738 | while (isDIGIT(*s)) |
a687059c |
2739 | s++; |
2740 | if (*s != '}') |
2741 | return FALSE; |
2742 | return TRUE; |
2743 | } |
2744 | |
a687059c |
2745 | |
76e3520e |
2746 | STATIC regnode * |
c277df42 |
2747 | dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l) |
2748 | { |
35ff7856 |
2749 | #ifdef DEBUGGING |
c277df42 |
2750 | register char op = EXACT; /* Arbitrary non-END op. */ |
2751 | register regnode *next, *onode; |
2752 | |
2753 | while (op != END && (!last || node < last)) { |
2754 | /* While that wasn't END last time... */ |
2755 | |
2756 | NODE_ALIGN(node); |
2757 | op = OP(node); |
2758 | if (op == CLOSE) |
2759 | l--; |
2760 | next = regnext(node); |
2761 | /* Where, what. */ |
2762 | if (OP(node) == OPTIMIZED) |
2763 | goto after_print; |
2764 | regprop(sv, node); |
54dc92de |
2765 | PerlIO_printf(Perl_debug_log, "%4d:%*s%s", node - start, |
c277df42 |
2766 | 2*l + 1, "", SvPVX(sv)); |
2767 | if (next == NULL) /* Next ptr. */ |
2768 | PerlIO_printf(Perl_debug_log, "(0)"); |
2769 | else |
2770 | PerlIO_printf(Perl_debug_log, "(%d)", next - start); |
2771 | (void)PerlIO_putc(Perl_debug_log, '\n'); |
2772 | after_print: |
2773 | if (regkind[(U8)op] == BRANCHJ) { |
2774 | register regnode *nnode = (OP(next) == LONGJMP |
2775 | ? regnext(next) |
2776 | : next); |
2777 | if (last && nnode > last) |
2778 | nnode = last; |
2779 | node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1); |
a0ed51b3 |
2780 | } |
2781 | else if (regkind[(U8)op] == BRANCH) { |
c277df42 |
2782 | node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1); |
a0ed51b3 |
2783 | } |
2784 | else if ( op == CURLY) { /* `next' might be very big: optimizer */ |
c277df42 |
2785 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
2786 | NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1); |
a0ed51b3 |
2787 | } |
2788 | else if (regkind[(U8)op] == CURLY && op != CURLYX) { |
c277df42 |
2789 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
2790 | next, sv, l + 1); |
a0ed51b3 |
2791 | } |
2792 | else if ( op == PLUS || op == STAR) { |
c277df42 |
2793 | node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1); |
a0ed51b3 |
2794 | } |
2795 | else if (op == ANYOF) { |
c277df42 |
2796 | node = NEXTOPER(node); |
2797 | node += ANY_SKIP; |
a0ed51b3 |
2798 | } |
2799 | else if (regkind[(U8)op] == EXACT) { |
c277df42 |
2800 | /* Literal string, where present. */ |
2801 | node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
2802 | node = NEXTOPER(node); |
a0ed51b3 |
2803 | } |
2804 | else { |
c277df42 |
2805 | node = NEXTOPER(node); |
2806 | node += regarglen[(U8)op]; |
2807 | } |
2808 | if (op == CURLYX || op == OPEN) |
2809 | l++; |
2810 | else if (op == WHILEM) |
2811 | l--; |
2812 | } |
17c3b450 |
2813 | #endif /* DEBUGGING */ |
c277df42 |
2814 | return node; |
2815 | } |
2816 | |
a687059c |
2817 | /* |
fd181c75 |
2818 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c |
2819 | */ |
2820 | void |
8ac85365 |
2821 | regdump(regexp *r) |
a687059c |
2822 | { |
35ff7856 |
2823 | #ifdef DEBUGGING |
5c0ca799 |
2824 | dTHR; |
46fc3d4c |
2825 | SV *sv = sv_newmortal(); |
a687059c |
2826 | |
c277df42 |
2827 | (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0); |
a0d0e21e |
2828 | |
2829 | /* Header fields of interest. */ |
c277df42 |
2830 | if (r->anchored_substr) |
2831 | PerlIO_printf(Perl_debug_log, "anchored `%s%s%s'%s at %d ", |
3280af22 |
2832 | PL_colors[0], |
c277df42 |
2833 | SvPVX(r->anchored_substr), |
3280af22 |
2834 | PL_colors[1], |
c277df42 |
2835 | SvTAIL(r->anchored_substr) ? "$" : "", |
2836 | r->anchored_offset); |
2837 | if (r->float_substr) |
2838 | PerlIO_printf(Perl_debug_log, "floating `%s%s%s'%s at %d..%u ", |
3280af22 |
2839 | PL_colors[0], |
c277df42 |
2840 | SvPVX(r->float_substr), |
3280af22 |
2841 | PL_colors[1], |
c277df42 |
2842 | SvTAIL(r->float_substr) ? "$" : "", |
2843 | r->float_min_offset, r->float_max_offset); |
2844 | if (r->check_substr) |
2845 | PerlIO_printf(Perl_debug_log, |
2846 | r->check_substr == r->float_substr |
2847 | ? "(checking floating" : "(checking anchored"); |
2848 | if (r->reganch & ROPT_NOSCAN) |
2849 | PerlIO_printf(Perl_debug_log, " noscan"); |
2850 | if (r->reganch & ROPT_CHECK_ALL) |
2851 | PerlIO_printf(Perl_debug_log, " isall"); |
2852 | if (r->check_substr) |
2853 | PerlIO_printf(Perl_debug_log, ") "); |
2854 | |
46fc3d4c |
2855 | if (r->regstclass) { |
2856 | regprop(sv, r->regstclass); |
2857 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv)); |
2858 | } |
774d564b |
2859 | if (r->reganch & ROPT_ANCH) { |
2860 | PerlIO_printf(Perl_debug_log, "anchored"); |
2861 | if (r->reganch & ROPT_ANCH_BOL) |
2862 | PerlIO_printf(Perl_debug_log, "(BOL)"); |
c277df42 |
2863 | if (r->reganch & ROPT_ANCH_MBOL) |
2864 | PerlIO_printf(Perl_debug_log, "(MBOL)"); |
774d564b |
2865 | if (r->reganch & ROPT_ANCH_GPOS) |
2866 | PerlIO_printf(Perl_debug_log, "(GPOS)"); |
2867 | PerlIO_putc(Perl_debug_log, ' '); |
2868 | } |
c277df42 |
2869 | if (r->reganch & ROPT_GPOS_SEEN) |
2870 | PerlIO_printf(Perl_debug_log, "GPOS "); |
a0d0e21e |
2871 | if (r->reganch & ROPT_SKIP) |
760ac839 |
2872 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e |
2873 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 |
2874 | PerlIO_printf(Perl_debug_log, "implicit "); |
760ac839 |
2875 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
ce862d02 |
2876 | if (r->reganch & ROPT_EVAL_SEEN) |
2877 | PerlIO_printf(Perl_debug_log, "with eval "); |
760ac839 |
2878 | PerlIO_printf(Perl_debug_log, "\n"); |
17c3b450 |
2879 | #endif /* DEBUGGING */ |
a687059c |
2880 | } |
2881 | |
2882 | /* |
a0d0e21e |
2883 | - regprop - printable representation of opcode |
2884 | */ |
46fc3d4c |
2885 | void |
c277df42 |
2886 | regprop(SV *sv, regnode *o) |
a687059c |
2887 | { |
35ff7856 |
2888 | #ifdef DEBUGGING |
5c0ca799 |
2889 | dTHR; |
a0d0e21e |
2890 | register char *p = 0; |
2891 | |
54dc92de |
2892 | sv_setpvn(sv, "", 0); |
11343788 |
2893 | switch (OP(o)) { |
a0d0e21e |
2894 | case BOL: |
2895 | p = "BOL"; |
2896 | break; |
2897 | case MBOL: |
2898 | p = "MBOL"; |
2899 | break; |
2900 | case SBOL: |
2901 | p = "SBOL"; |
2902 | break; |
2903 | case EOL: |
2904 | p = "EOL"; |
2905 | break; |
b85d18e9 |
2906 | case EOS: |
2907 | p = "EOS"; |
2908 | break; |
a0d0e21e |
2909 | case MEOL: |
2910 | p = "MEOL"; |
2911 | break; |
2912 | case SEOL: |
2913 | p = "SEOL"; |
2914 | break; |
2915 | case ANY: |
2916 | p = "ANY"; |
2917 | break; |
2918 | case SANY: |
2919 | p = "SANY"; |
2920 | break; |
a0ed51b3 |
2921 | case ANYUTF8: |
2922 | p = "ANYUTF8"; |
2923 | break; |
2924 | case SANYUTF8: |
2925 | p = "SANYUTF8"; |
2926 | break; |
2927 | case ANYOFUTF8: |
2928 | p = "ANYOFUTF8"; |
2929 | break; |
a0d0e21e |
2930 | case ANYOF: |
2931 | p = "ANYOF"; |
2932 | break; |
2933 | case BRANCH: |
2934 | p = "BRANCH"; |
2935 | break; |
bbce6d69 |
2936 | case EXACT: |
3280af22 |
2937 | sv_catpvf(sv, "EXACT <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]); |
bbce6d69 |
2938 | break; |
2939 | case EXACTF: |
3280af22 |
2940 | sv_catpvf(sv, "EXACTF <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]); |
bbce6d69 |
2941 | break; |
2942 | case EXACTFL: |
3280af22 |
2943 | sv_catpvf(sv, "EXACTFL <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]); |
a0d0e21e |
2944 | break; |
2945 | case NOTHING: |
2946 | p = "NOTHING"; |
2947 | break; |
c277df42 |
2948 | case TAIL: |
2949 | p = "TAIL"; |
2950 | break; |
a0d0e21e |
2951 | case BACK: |
2952 | p = "BACK"; |
2953 | break; |
2954 | case END: |
2955 | p = "END"; |
2956 | break; |
a0d0e21e |
2957 | case BOUND: |
2958 | p = "BOUND"; |
2959 | break; |
bbce6d69 |
2960 | case BOUNDL: |
2961 | p = "BOUNDL"; |
2962 | break; |
a0d0e21e |
2963 | case NBOUND: |
2964 | p = "NBOUND"; |
2965 | break; |
bbce6d69 |
2966 | case NBOUNDL: |
2967 | p = "NBOUNDL"; |
a0d0e21e |
2968 | break; |
2969 | case CURLY: |
5dc0d613 |
2970 | sv_catpvf(sv, "CURLY {%d,%d}", ARG1(o), ARG2(o)); |
a0d0e21e |
2971 | break; |
c277df42 |
2972 | case CURLYM: |
c277df42 |
2973 | sv_catpvf(sv, "CURLYM[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o)); |
c277df42 |
2974 | break; |
2975 | case CURLYN: |
c277df42 |
2976 | sv_catpvf(sv, "CURLYN[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o)); |
c277df42 |
2977 | break; |
a0d0e21e |
2978 | case CURLYX: |
5dc0d613 |
2979 | sv_catpvf(sv, "CURLYX {%d,%d}", ARG1(o), ARG2(o)); |
a0d0e21e |
2980 | break; |
2981 | case REF: |
c277df42 |
2982 | sv_catpvf(sv, "REF%d", ARG(o)); |
a0d0e21e |
2983 | break; |
c8756f30 |
2984 | case REFF: |
c277df42 |
2985 | sv_catpvf(sv, "REFF%d", ARG(o)); |
c8756f30 |
2986 | break; |
2987 | case REFFL: |
c277df42 |
2988 | sv_catpvf(sv, "REFFL%d", ARG(o)); |
c8756f30 |
2989 | break; |
a0d0e21e |
2990 | case OPEN: |
c277df42 |
2991 | sv_catpvf(sv, "OPEN%d", ARG(o)); |
a0d0e21e |
2992 | break; |
2993 | case CLOSE: |
c277df42 |
2994 | sv_catpvf(sv, "CLOSE%d", ARG(o)); |
a0d0e21e |
2995 | p = NULL; |
2996 | break; |
2997 | case STAR: |
2998 | p = "STAR"; |
2999 | break; |
3000 | case PLUS: |
3001 | p = "PLUS"; |
3002 | break; |
3003 | case MINMOD: |
3004 | p = "MINMOD"; |
3005 | break; |
774d564b |
3006 | case GPOS: |
3007 | p = "GPOS"; |
a0d0e21e |
3008 | break; |
3009 | case UNLESSM: |
c277df42 |
3010 | sv_catpvf(sv, "UNLESSM[-%d]", o->flags); |
a0d0e21e |
3011 | break; |
3012 | case IFMATCH: |
c277df42 |
3013 | sv_catpvf(sv, "IFMATCH[-%d]", o->flags); |
a0d0e21e |
3014 | break; |
3015 | case SUCCEED: |
3016 | p = "SUCCEED"; |
3017 | break; |
3018 | case WHILEM: |
3019 | p = "WHILEM"; |
3020 | break; |
bbce6d69 |
3021 | case DIGIT: |
3022 | p = "DIGIT"; |
3023 | break; |
3024 | case NDIGIT: |
3025 | p = "NDIGIT"; |
3026 | break; |
3027 | case ALNUM: |
3028 | p = "ALNUM"; |
3029 | break; |
3030 | case NALNUM: |
3031 | p = "NALNUM"; |
3032 | break; |
3033 | case SPACE: |
3034 | p = "SPACE"; |
3035 | break; |
3036 | case NSPACE: |
3037 | p = "NSPACE"; |
3038 | break; |
3039 | case ALNUML: |
3040 | p = "ALNUML"; |
3041 | break; |
3042 | case NALNUML: |
3043 | p = "NALNUML"; |
3044 | break; |
3045 | case SPACEL: |
3046 | p = "SPACEL"; |
3047 | break; |
3048 | case NSPACEL: |
3049 | p = "NSPACEL"; |
3050 | break; |
c277df42 |
3051 | case EVAL: |
3052 | p = "EVAL"; |
3053 | break; |
3054 | case LONGJMP: |
3055 | p = "LONGJMP"; |
3056 | break; |
3057 | case BRANCHJ: |
3058 | p = "BRANCHJ"; |
3059 | break; |
3060 | case IFTHEN: |
3061 | p = "IFTHEN"; |
3062 | break; |
3063 | case GROUPP: |
3064 | sv_catpvf(sv, "GROUPP%d", ARG(o)); |
3065 | break; |
3066 | case LOGICAL: |
0f5d15d6 |
3067 | sv_catpvf(sv, "LOGICAL[%d]", o->flags); |
c277df42 |
3068 | break; |
3069 | case SUSPEND: |
3070 | p = "SUSPEND"; |
3071 | break; |
3072 | case RENUM: |
3073 | p = "RENUM"; |
3074 | break; |
3075 | case OPTIMIZED: |
3076 | p = "OPTIMIZED"; |
3077 | break; |
a0d0e21e |
3078 | default: |
3079 | FAIL("corrupted regexp opcode"); |
3080 | } |
46fc3d4c |
3081 | if (p) |
3082 | sv_catpv(sv, p); |
17c3b450 |
3083 | #endif /* DEBUGGING */ |
35ff7856 |
3084 | } |
a687059c |
3085 | |
2b69d0c2 |
3086 | void |
8ac85365 |
3087 | pregfree(struct regexp *r) |
a687059c |
3088 | { |
5c0ca799 |
3089 | dTHR; |
c277df42 |
3090 | if (!r || (--r->refcnt > 0)) |
a0d0e21e |
3091 | return; |
c277df42 |
3092 | if (r->precomp) |
a0d0e21e |
3093 | Safefree(r->precomp); |
c277df42 |
3094 | if (r->subbase) |
a0d0e21e |
3095 | Safefree(r->subbase); |
a193d654 |
3096 | if (r->substrs) { |
3097 | if (r->anchored_substr) |
3098 | SvREFCNT_dec(r->anchored_substr); |
3099 | if (r->float_substr) |
3100 | SvREFCNT_dec(r->float_substr); |
2779dcf1 |
3101 | Safefree(r->substrs); |
a193d654 |
3102 | } |
c277df42 |
3103 | if (r->data) { |
3104 | int n = r->data->count; |
3105 | while (--n >= 0) { |
3106 | switch (r->data->what[n]) { |
3107 | case 's': |
3108 | SvREFCNT_dec((SV*)r->data->data[n]); |
3109 | break; |
3110 | case 'o': |
3111 | op_free((OP_4tree*)r->data->data[n]); |
3112 | break; |
3113 | case 'n': |
3114 | break; |
3115 | default: |
3116 | FAIL2("panic: regfree data code '%c'", r->data->what[n]); |
3117 | } |
3118 | } |
3119 | Safefree(r->data->what); |
3120 | Safefree(r->data); |
a0d0e21e |
3121 | } |
3122 | Safefree(r->startp); |
3123 | Safefree(r->endp); |
3124 | Safefree(r); |
a687059c |
3125 | } |
c277df42 |
3126 | |
3127 | /* |
3128 | - regnext - dig the "next" pointer out of a node |
3129 | * |
3130 | * [Note, when REGALIGN is defined there are two places in regmatch() |
3131 | * that bypass this code for speed.] |
3132 | */ |
3133 | regnode * |
3134 | regnext(register regnode *p) |
3135 | { |
5c0ca799 |
3136 | dTHR; |
c277df42 |
3137 | register I32 offset; |
3138 | |
3280af22 |
3139 | if (p == &PL_regdummy) |
c277df42 |
3140 | return(NULL); |
3141 | |
3142 | offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p)); |
3143 | if (offset == 0) |
3144 | return(NULL); |
3145 | |
c277df42 |
3146 | return(p+offset); |
c277df42 |
3147 | } |
3148 | |
01f988be |
3149 | STATIC void |
c277df42 |
3150 | re_croak2(const char* pat1,const char* pat2,...) |
c277df42 |
3151 | { |
3152 | va_list args; |
3153 | STRLEN l1 = strlen(pat1); |
3154 | STRLEN l2 = strlen(pat2); |
3155 | char buf[512]; |
3156 | char *message; |
3157 | |
3158 | if (l1 > 510) |
3159 | l1 = 510; |
3160 | if (l1 + l2 > 510) |
3161 | l2 = 510 - l1; |
3162 | Copy(pat1, buf, l1 , char); |
3163 | Copy(pat2, buf + l1, l2 , char); |
3b818b81 |
3164 | buf[l1 + l2] = '\n'; |
3165 | buf[l1 + l2 + 1] = '\0'; |
c277df42 |
3166 | va_start(args, pat2); |
c277df42 |
3167 | message = mess(buf, &args); |
3168 | va_end(args); |
3169 | l1 = strlen(message); |
3170 | if (l1 > 512) |
3171 | l1 = 512; |
3172 | Copy(message, buf, l1 , char); |
3173 | buf[l1] = '\0'; /* Overwrite \n */ |
3174 | croak("%s", buf); |
3175 | } |
a0ed51b3 |
3176 | |
3177 | /* XXX Here's a total kludge. But we need to re-enter for swash routines. */ |
3178 | |
3179 | void |
3180 | save_re_context(void) |
c485e607 |
3181 | { |
3182 | dTHR; |
a0ed51b3 |
3183 | SAVEPPTR(PL_bostr); |
3184 | SAVEPPTR(PL_regprecomp); /* uncompiled string. */ |
3185 | SAVEI32(PL_regnpar); /* () count. */ |
3186 | SAVEI32(PL_regsize); /* Code size. */ |
3187 | SAVEI16(PL_regflags); /* are we folding, multilining? */ |
3188 | SAVEPPTR(PL_reginput); /* String-input pointer. */ |
3189 | SAVEPPTR(PL_regbol); /* Beginning of input, for ^ check. */ |
3190 | SAVEPPTR(PL_regeol); /* End of input, for $ check. */ |
3191 | SAVESPTR(PL_regstartp); /* Pointer to startp array. */ |
3192 | SAVESPTR(PL_regendp); /* Ditto for endp. */ |
3193 | SAVESPTR(PL_reglastparen); /* Similarly for lastparen. */ |
3194 | SAVEPPTR(PL_regtill); /* How far we are required to go. */ |
3195 | SAVEI32(PL_regprev); /* char before regbol, \n if none */ |
3196 | SAVESPTR(PL_reg_start_tmp); /* from regexec.c */ |
3197 | PL_reg_start_tmp = 0; |
3198 | SAVEFREEPV(PL_reg_start_tmp); |
3199 | SAVEI32(PL_reg_start_tmpl); /* from regexec.c */ |
3200 | PL_reg_start_tmpl = 0; |
3201 | SAVESPTR(PL_regdata); |
3202 | SAVEI32(PL_reg_flags); /* from regexec.c */ |
3203 | SAVEI32(PL_reg_eval_set); /* from regexec.c */ |
3204 | SAVEI32(PL_regnarrate); /* from regexec.c */ |
3205 | SAVESPTR(PL_regprogram); /* from regexec.c */ |
3206 | SAVEINT(PL_regindent); /* from regexec.c */ |
3207 | SAVESPTR(PL_regcc); /* from regexec.c */ |
3208 | SAVESPTR(PL_curcop); |
3209 | SAVESPTR(PL_regcomp_rx); /* from regcomp.c */ |
3210 | SAVEI32(PL_regseen); /* from regcomp.c */ |
3211 | SAVEI32(PL_regsawback); /* Did we see \1, ...? */ |
3212 | SAVEI32(PL_regnaughty); /* How bad is this pattern? */ |
3213 | SAVESPTR(PL_regcode); /* Code-emit pointer; ®dummy = don't */ |
3214 | SAVEPPTR(PL_regxend); /* End of input for compile */ |
3215 | SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */ |
3216 | } |