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 | |
56953603 |
22 | #ifdef IN_XSUB_RE |
23 | # define Perl_pregcomp my_regcomp |
24 | # define Perl_regdump my_regdump |
25 | # define Perl_regprop my_regprop |
26 | #endif |
27 | |
f0fcb552 |
28 | /*SUPPRESS 112*/ |
a687059c |
29 | /* |
e50aee73 |
30 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c |
31 | * |
32 | * Copyright (c) 1986 by University of Toronto. |
33 | * Written by Henry Spencer. Not derived from licensed software. |
34 | * |
35 | * Permission is granted to anyone to use this software for any |
36 | * purpose on any computer system, and to redistribute it freely, |
37 | * subject to the following restrictions: |
38 | * |
39 | * 1. The author is not responsible for the consequences of use of |
40 | * this software, no matter how awful, even if they arise |
41 | * from defects in it. |
42 | * |
43 | * 2. The origin of this software must not be misrepresented, either |
44 | * by explicit claim or by omission. |
45 | * |
46 | * 3. Altered versions must be plainly marked as such, and must not |
47 | * be misrepresented as being the original software. |
48 | * |
49 | * |
50 | **** Alterations to Henry's code are... |
51 | **** |
9607fc9c |
52 | **** Copyright (c) 1991-1997, Larry Wall |
a687059c |
53 | **** |
9ef589d8 |
54 | **** You may distribute under the terms of either the GNU General Public |
55 | **** License or the Artistic License, as specified in the README file. |
56 | |
a687059c |
57 | * |
58 | * Beware that some of this code is subtly aware of the way operator |
59 | * precedence is structured in regular expressions. Serious changes in |
60 | * regular-expression syntax might require a total rethink. |
61 | */ |
62 | #include "EXTERN.h" |
63 | #include "perl.h" |
64 | #include "INTERN.h" |
c277df42 |
65 | |
66 | #define REG_COMP_C |
a687059c |
67 | #include "regcomp.h" |
68 | |
d4cce5f1 |
69 | #ifdef op |
11343788 |
70 | #undef op |
d4cce5f1 |
71 | #endif /* op */ |
11343788 |
72 | |
fe14fcc3 |
73 | #ifdef MSDOS |
74 | # if defined(BUGGY_MSC6) |
75 | /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */ |
76 | # pragma optimize("a",off) |
77 | /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/ |
78 | # pragma optimize("w",on ) |
79 | # endif /* BUGGY_MSC6 */ |
80 | #endif /* MSDOS */ |
81 | |
a687059c |
82 | #ifndef STATIC |
83 | #define STATIC static |
84 | #endif |
85 | |
86 | #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?') |
87 | #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \ |
88 | ((*s) == '{' && regcurly(s))) |
2b69d0c2 |
89 | #ifdef atarist |
90 | #define PERL_META "^$.[()|?+*\\" |
91 | #else |
a687059c |
92 | #define META "^$.[()|?+*\\" |
2b69d0c2 |
93 | #endif |
a687059c |
94 | |
35c8bce7 |
95 | #ifdef SPSTART |
96 | #undef SPSTART /* dratted cpp namespace... */ |
97 | #endif |
a687059c |
98 | /* |
99 | * Flags to be passed up and down. |
100 | */ |
a687059c |
101 | #define WORST 0 /* Worst case. */ |
821b33a5 |
102 | #define HASWIDTH 0x1 /* Known to match non-null strings. */ |
a0d0e21e |
103 | #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ |
104 | #define SPSTART 0x4 /* Starts with * or +. */ |
105 | #define TRYAGAIN 0x8 /* Weeded out a declaration. */ |
a687059c |
106 | |
107 | /* |
e50aee73 |
108 | * Forward declarations for pregcomp()'s friends. |
a687059c |
109 | */ |
a0d0e21e |
110 | |
76e3520e |
111 | #ifndef PERL_OBJECT |
c277df42 |
112 | static regnode *reg _((I32, I32 *)); |
113 | static regnode *reganode _((U8, U32)); |
114 | static regnode *regatom _((I32 *)); |
115 | static regnode *regbranch _((I32 *, I32)); |
116 | static void regc _((U8, char *)); |
117 | static regnode *regclass _((void)); |
a0d0e21e |
118 | STATIC I32 regcurly _((char *)); |
c277df42 |
119 | static regnode *reg_node _((U8)); |
120 | static regnode *regpiece _((I32 *)); |
121 | static void reginsert _((U8, regnode *)); |
122 | static void regoptail _((regnode *, regnode *)); |
c277df42 |
123 | static void regtail _((regnode *, regnode *)); |
873ef191 |
124 | static char* regwhite _((char *, char *)); |
a0d0e21e |
125 | static char* nextchar _((void)); |
3bd495df |
126 | static void re_croak2 _((const char* pat1,const char* pat2,...)) __attribute__((noreturn)); |
76e3520e |
127 | #endif |
a687059c |
128 | |
c277df42 |
129 | /* Length of a variant. */ |
130 | |
76e3520e |
131 | #ifndef PERL_OBJECT |
c277df42 |
132 | typedef struct { |
133 | I32 len_min; |
134 | I32 len_delta; |
135 | I32 pos_min; |
136 | I32 pos_delta; |
137 | SV *last_found; |
138 | I32 last_end; /* min value, <0 unless valid. */ |
139 | I32 last_start_min; |
140 | I32 last_start_max; |
141 | SV **longest; /* Either &l_fixed, or &l_float. */ |
142 | SV *longest_fixed; |
143 | I32 offset_fixed; |
144 | SV *longest_float; |
145 | I32 offset_float_min; |
146 | I32 offset_float_max; |
147 | I32 flags; |
148 | } scan_data_t; |
76e3520e |
149 | #endif |
c277df42 |
150 | |
151 | static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; |
152 | |
153 | #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL) |
154 | #define SF_BEFORE_SEOL 0x1 |
155 | #define SF_BEFORE_MEOL 0x2 |
156 | #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL) |
157 | #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL) |
158 | |
09b7f37c |
159 | #ifdef NO_UNARY_PLUS |
160 | # define SF_FIX_SHIFT_EOL (0+2) |
161 | # define SF_FL_SHIFT_EOL (0+4) |
162 | #else |
163 | # define SF_FIX_SHIFT_EOL (+2) |
164 | # define SF_FL_SHIFT_EOL (+4) |
165 | #endif |
c277df42 |
166 | |
167 | #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL) |
168 | #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL) |
169 | |
170 | #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL) |
171 | #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */ |
172 | #define SF_IS_INF 0x40 |
173 | #define SF_HAS_PAR 0x80 |
174 | #define SF_IN_PAR 0x100 |
175 | #define SF_HAS_EVAL 0x200 |
4bfe0158 |
176 | #define SCF_DO_SUBSTR 0x400 |
c277df42 |
177 | |
76e3520e |
178 | STATIC void |
c277df42 |
179 | scan_commit(scan_data_t *data) |
180 | { |
181 | STRLEN l = SvCUR(data->last_found); |
182 | STRLEN old_l = SvCUR(*data->longest); |
183 | |
184 | if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) { |
185 | sv_setsv(*data->longest, data->last_found); |
186 | if (*data->longest == data->longest_fixed) { |
187 | data->offset_fixed = l ? data->last_start_min : data->pos_min; |
188 | if (data->flags & SF_BEFORE_EOL) |
189 | data->flags |
190 | |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL); |
191 | else |
192 | data->flags &= ~SF_FIX_BEFORE_EOL; |
193 | } else { |
194 | data->offset_float_min = l ? data->last_start_min : data->pos_min; |
195 | data->offset_float_max = (l |
196 | ? data->last_start_max |
197 | : data->pos_min + data->pos_delta); |
198 | if (data->flags & SF_BEFORE_EOL) |
199 | data->flags |
200 | |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL); |
201 | else |
202 | data->flags &= ~SF_FL_BEFORE_EOL; |
203 | } |
204 | } |
205 | SvCUR_set(data->last_found, 0); |
206 | data->last_end = -1; |
207 | data->flags &= ~SF_BEFORE_EOL; |
208 | } |
209 | |
c277df42 |
210 | /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set |
211 | to the position after last scanned or to NULL. */ |
212 | |
76e3520e |
213 | STATIC I32 |
c277df42 |
214 | study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags) |
215 | /* scanp: Start here (read-write). */ |
216 | /* deltap: Write maxlen-minlen here. */ |
217 | /* last: Stop before this one. */ |
218 | { |
219 | I32 min = 0, pars = 0, code; |
220 | regnode *scan = *scanp, *next; |
221 | I32 delta = 0; |
222 | int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF); |
223 | I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0; |
224 | scan_data_t data_fake; |
225 | |
226 | while (scan && OP(scan) != END && scan < last) { |
227 | /* Peephole optimizer: */ |
228 | |
229 | if (regkind[(U8)OP(scan)] == EXACT) { |
230 | regnode *n = regnext(scan); |
231 | U32 stringok = 1; |
232 | #ifdef DEBUGGING |
233 | regnode *stop = scan; |
234 | #endif |
235 | |
236 | next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; |
237 | /* Skip NOTHING, merge EXACT*. */ |
238 | while (n && |
239 | ( regkind[(U8)OP(n)] == NOTHING || |
240 | (stringok && (OP(n) == OP(scan)))) |
241 | && NEXT_OFF(n) |
242 | && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) { |
243 | if (OP(n) == TAIL || n > next) |
244 | stringok = 0; |
245 | if (regkind[(U8)OP(n)] == NOTHING) { |
246 | NEXT_OFF(scan) += NEXT_OFF(n); |
247 | next = n + NODE_STEP_REGNODE; |
248 | #ifdef DEBUGGING |
249 | if (stringok) |
250 | stop = n; |
251 | #endif |
252 | n = regnext(n); |
253 | } else { |
254 | int oldl = *OPERAND(scan); |
255 | regnode *nnext = regnext(n); |
256 | |
257 | if (oldl + *OPERAND(n) > U8_MAX) |
258 | break; |
259 | NEXT_OFF(scan) += NEXT_OFF(n); |
260 | *OPERAND(scan) += *OPERAND(n); |
261 | next = n + (*OPERAND(n) + 2 - 1)/sizeof(regnode) + 2; |
262 | /* Now we can overwrite *n : */ |
263 | Move(OPERAND(n) + 1, OPERAND(scan) + oldl + 1, |
264 | *OPERAND(n) + 1, char); |
265 | #ifdef DEBUGGING |
266 | if (stringok) |
267 | stop = next - 1; |
268 | #endif |
269 | n = nnext; |
270 | } |
271 | } |
272 | #ifdef DEBUGGING |
273 | /* Allow dumping */ |
274 | n = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2; |
275 | while (n <= stop) { |
ca04da08 |
276 | /* Purify reports a benign UMR here sometimes, because we |
277 | * don't initialize the OP() slot of a node when that node |
278 | * is occupied by just the trailing null of the string in |
279 | * an EXACT node */ |
c277df42 |
280 | if (regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) { |
281 | OP(n) = OPTIMIZED; |
282 | NEXT_OFF(n) = 0; |
283 | } |
284 | n++; |
285 | } |
286 | #endif |
287 | |
288 | } |
289 | if (OP(scan) != CURLYX) { |
290 | int max = (reg_off_by_arg[OP(scan)] ? I32_MAX : U16_MAX); |
291 | int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan)); |
292 | int noff; |
293 | regnode *n = scan; |
294 | |
295 | /* Skip NOTHING and LONGJMP. */ |
296 | while ((n = regnext(n)) |
297 | && ((regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n))) |
298 | || ((OP(n) == LONGJMP) && (noff = ARG(n)))) |
299 | && off + noff < max) |
300 | off += noff; |
301 | if (reg_off_by_arg[OP(scan)]) |
302 | ARG(scan) = off; |
303 | else |
304 | NEXT_OFF(scan) = off; |
305 | } |
306 | if (OP(scan) == BRANCH || OP(scan) == BRANCHJ |
307 | || OP(scan) == IFTHEN || OP(scan) == SUSPEND) { |
308 | next = regnext(scan); |
309 | code = OP(scan); |
310 | |
311 | if (OP(next) == code || code == IFTHEN || code == SUSPEND) { |
312 | I32 max1 = 0, min1 = I32_MAX, num = 0; |
313 | |
314 | if (flags & SCF_DO_SUBSTR) |
315 | scan_commit(data); |
316 | while (OP(scan) == code) { |
317 | I32 deltanext, minnext; |
318 | |
319 | num++; |
320 | data_fake.flags = 0; |
321 | next = regnext(scan); |
322 | scan = NEXTOPER(scan); |
323 | if (code != BRANCH) |
324 | scan = NEXTOPER(scan); |
325 | /* We suppose the run is continuous, last=next...*/ |
326 | minnext = study_chunk(&scan, &deltanext, next, |
327 | &data_fake, 0); |
328 | if (min1 > minnext) |
329 | min1 = minnext; |
330 | if (max1 < minnext + deltanext) |
331 | max1 = minnext + deltanext; |
332 | if (deltanext == I32_MAX) |
333 | is_inf = 1; |
334 | scan = next; |
335 | if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) |
336 | pars++; |
405ff068 |
337 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 |
338 | data->flags |= SF_HAS_EVAL; |
339 | if (code == SUSPEND) |
340 | break; |
341 | } |
342 | if (code == IFTHEN && num < 2) /* Empty ELSE branch */ |
343 | min1 = 0; |
344 | if (flags & SCF_DO_SUBSTR) { |
345 | data->pos_min += min1; |
346 | data->pos_delta += max1 - min1; |
347 | if (max1 != min1 || is_inf) |
348 | data->longest = &(data->longest_float); |
349 | } |
350 | min += min1; |
351 | delta += max1 - min1; |
352 | } else if (code == BRANCHJ) /* single branch is optimized. */ |
353 | scan = NEXTOPER(NEXTOPER(scan)); |
354 | else /* single branch is optimized. */ |
355 | scan = NEXTOPER(scan); |
356 | continue; |
357 | } else if (OP(scan) == EXACT) { |
358 | min += *OPERAND(scan); |
359 | if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */ |
360 | I32 l = *OPERAND(scan); |
361 | |
362 | /* The code below prefers earlier match for fixed |
363 | offset, later match for variable offset. */ |
364 | if (data->last_end == -1) { /* Update the start info. */ |
365 | data->last_start_min = data->pos_min; |
366 | data->last_start_max = is_inf |
367 | ? I32_MAX : data->pos_min + data->pos_delta; |
368 | } |
161b471a |
369 | sv_catpvn(data->last_found, (char *)(OPERAND(scan)+1), l); |
c277df42 |
370 | data->last_end = data->pos_min + l; |
371 | data->pos_min += l; /* As in the first entry. */ |
372 | data->flags &= ~SF_BEFORE_EOL; |
373 | } |
374 | } else if (regkind[(U8)OP(scan)] == EXACT) { |
375 | if (flags & SCF_DO_SUBSTR) |
376 | scan_commit(data); |
377 | min += *OPERAND(scan); |
378 | if (data && (flags & SCF_DO_SUBSTR)) |
379 | data->pos_min += *OPERAND(scan); |
380 | } else if (strchr(varies,OP(scan))) { |
381 | I32 mincount, maxcount, minnext, deltanext, pos_before, fl; |
382 | regnode *oscan = scan; |
383 | |
384 | switch (regkind[(U8)OP(scan)]) { |
385 | case WHILEM: |
386 | scan = NEXTOPER(scan); |
387 | goto finish; |
388 | case PLUS: |
389 | if (flags & SCF_DO_SUBSTR) { |
390 | next = NEXTOPER(scan); |
391 | if (OP(next) == EXACT) { |
392 | mincount = 1; |
393 | maxcount = REG_INFTY; |
394 | next = regnext(scan); |
395 | scan = NEXTOPER(scan); |
396 | goto do_curly; |
397 | } |
398 | } |
399 | if (flags & SCF_DO_SUBSTR) |
400 | data->pos_min++; |
401 | min++; |
402 | /* Fall through. */ |
403 | case STAR: |
404 | is_inf = 1; |
405 | scan = regnext(scan); |
406 | if (flags & SCF_DO_SUBSTR) { |
407 | scan_commit(data); |
408 | data->longest = &(data->longest_float); |
409 | } |
410 | goto optimize_curly_tail; |
411 | case CURLY: |
412 | mincount = ARG1(scan); |
413 | maxcount = ARG2(scan); |
414 | next = regnext(scan); |
415 | scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; |
416 | do_curly: |
417 | if (flags & SCF_DO_SUBSTR) { |
418 | if (mincount == 0) scan_commit(data); |
419 | pos_before = data->pos_min; |
420 | } |
421 | if (data) { |
422 | fl = data->flags; |
423 | data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL); |
424 | if (is_inf) |
425 | data->flags |= SF_IS_INF; |
426 | } |
427 | /* This will finish on WHILEM, setting scan, or on NULL: */ |
428 | minnext = study_chunk(&scan, &deltanext, last, data, |
429 | mincount == 0 |
430 | ? (flags & ~SCF_DO_SUBSTR) : flags); |
431 | if (!scan) /* It was not CURLYX, but CURLY. */ |
432 | scan = next; |
433 | if (dowarn && (minnext + deltanext == 0) |
821b33a5 |
434 | && !(data->flags & (SF_HAS_PAR|SF_IN_PAR)) |
435 | && maxcount <= 10000) /* Complement check for big count */ |
c277df42 |
436 | warn("Strange *+?{} on zero-length expression"); |
437 | min += minnext * mincount; |
438 | is_inf |= (maxcount == REG_INFTY && (minnext + deltanext) > 0 |
439 | || deltanext == I32_MAX); |
440 | delta += (minnext + deltanext) * maxcount - minnext * mincount; |
441 | |
442 | /* Try powerful optimization CURLYX => CURLYN. */ |
c277df42 |
443 | if ( OP(oscan) == CURLYX && data |
444 | && data->flags & SF_IN_PAR |
445 | && !(data->flags & SF_HAS_EVAL) |
446 | && !deltanext && minnext == 1 ) { |
447 | /* Try to optimize to CURLYN. */ |
448 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; |
449 | regnode *nxt1 = nxt, *nxt2; |
450 | |
451 | /* Skip open. */ |
452 | nxt = regnext(nxt); |
453 | if (!strchr(simple,OP(nxt)) |
454 | && !(regkind[(U8)OP(nxt)] == EXACT |
455 | && *OPERAND(nxt) == 1)) |
456 | goto nogo; |
457 | nxt2 = nxt; |
458 | nxt = regnext(nxt); |
459 | if (OP(nxt) != CLOSE) |
460 | goto nogo; |
461 | /* Now we know that nxt2 is the only contents: */ |
462 | oscan->flags = ARG(nxt); |
463 | OP(oscan) = CURLYN; |
464 | OP(nxt1) = NOTHING; /* was OPEN. */ |
465 | #ifdef DEBUGGING |
466 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ |
467 | NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */ |
468 | NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */ |
469 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ |
470 | OP(nxt + 1) = OPTIMIZED; /* was count. */ |
471 | NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */ |
472 | #endif |
473 | } |
c277df42 |
474 | nogo: |
475 | |
476 | /* Try optimization CURLYX => CURLYM. */ |
477 | if ( OP(oscan) == CURLYX && data |
c277df42 |
478 | && !(data->flags & SF_HAS_PAR) |
c277df42 |
479 | && !(data->flags & SF_HAS_EVAL) |
480 | && !deltanext ) { |
481 | /* XXXX How to optimize if data == 0? */ |
482 | /* Optimize to a simpler form. */ |
483 | regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */ |
484 | regnode *nxt2; |
485 | |
486 | OP(oscan) = CURLYM; |
487 | while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/ |
488 | && (OP(nxt2) != WHILEM)) |
489 | nxt = nxt2; |
490 | OP(nxt2) = SUCCEED; /* Whas WHILEM */ |
c277df42 |
491 | /* Need to optimize away parenths. */ |
492 | if (data->flags & SF_IN_PAR) { |
493 | /* Set the parenth number. */ |
494 | regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/ |
495 | |
496 | if (OP(nxt) != CLOSE) |
497 | FAIL("panic opt close"); |
498 | oscan->flags = ARG(nxt); |
499 | OP(nxt1) = OPTIMIZED; /* was OPEN. */ |
500 | OP(nxt) = OPTIMIZED; /* was CLOSE. */ |
501 | #ifdef DEBUGGING |
502 | OP(nxt1 + 1) = OPTIMIZED; /* was count. */ |
503 | OP(nxt + 1) = OPTIMIZED; /* was count. */ |
504 | NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */ |
505 | NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */ |
506 | #endif |
507 | #if 0 |
508 | while ( nxt1 && (OP(nxt1) != WHILEM)) { |
509 | regnode *nnxt = regnext(nxt1); |
510 | |
511 | if (nnxt == nxt) { |
512 | if (reg_off_by_arg[OP(nxt1)]) |
513 | ARG_SET(nxt1, nxt2 - nxt1); |
514 | else if (nxt2 - nxt1 < U16_MAX) |
515 | NEXT_OFF(nxt1) = nxt2 - nxt1; |
516 | else |
517 | OP(nxt) = NOTHING; /* Cannot beautify */ |
518 | } |
519 | nxt1 = nnxt; |
520 | } |
521 | #endif |
522 | /* Optimize again: */ |
523 | study_chunk(&nxt1, &deltanext, nxt, NULL, 0); |
524 | } else |
525 | oscan->flags = 0; |
c277df42 |
526 | } |
527 | if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) |
528 | pars++; |
529 | if (flags & SCF_DO_SUBSTR) { |
530 | SV *last_str = Nullsv; |
531 | int counted = mincount != 0; |
532 | |
533 | if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */ |
534 | I32 b = pos_before >= data->last_start_min |
535 | ? pos_before : data->last_start_min; |
536 | STRLEN l; |
537 | char *s = SvPV(data->last_found, l); |
538 | |
539 | l -= b - data->last_start_min; |
540 | /* Get the added string: */ |
541 | last_str = newSVpv(s + b - data->last_start_min, l); |
542 | if (deltanext == 0 && pos_before == b) { |
543 | /* What was added is a constant string */ |
544 | if (mincount > 1) { |
545 | SvGROW(last_str, (mincount * l) + 1); |
546 | repeatcpy(SvPVX(last_str) + l, |
547 | SvPVX(last_str), l, mincount - 1); |
548 | SvCUR(last_str) *= mincount; |
549 | /* Add additional parts. */ |
550 | SvCUR_set(data->last_found, |
551 | SvCUR(data->last_found) - l); |
552 | sv_catsv(data->last_found, last_str); |
553 | data->last_end += l * (mincount - 1); |
554 | } |
555 | } |
556 | } |
557 | /* It is counted once already... */ |
558 | data->pos_min += minnext * (mincount - counted); |
559 | data->pos_delta += - counted * deltanext + |
560 | (minnext + deltanext) * maxcount - minnext * mincount; |
561 | if (mincount != maxcount) { |
562 | scan_commit(data); |
563 | if (mincount && last_str) { |
564 | sv_setsv(data->last_found, last_str); |
565 | data->last_end = data->pos_min; |
566 | data->last_start_min = |
567 | data->pos_min - SvCUR(last_str); |
568 | data->last_start_max = is_inf |
569 | ? I32_MAX |
570 | : data->pos_min + data->pos_delta |
571 | - SvCUR(last_str); |
572 | } |
573 | data->longest = &(data->longest_float); |
574 | } |
575 | } |
405ff068 |
576 | if (data && (fl & SF_HAS_EVAL)) |
c277df42 |
577 | data->flags |= SF_HAS_EVAL; |
578 | optimize_curly_tail: |
c277df42 |
579 | if (OP(oscan) != CURLYX) { |
580 | while (regkind[(U8)OP(next = regnext(oscan))] == NOTHING |
581 | && NEXT_OFF(next)) |
582 | NEXT_OFF(oscan) += NEXT_OFF(next); |
583 | } |
c277df42 |
584 | continue; |
585 | default: /* REF only? */ |
586 | if (flags & SCF_DO_SUBSTR) { |
587 | scan_commit(data); |
588 | data->longest = &(data->longest_float); |
589 | } |
590 | is_inf = 1; |
591 | break; |
592 | } |
593 | } else if (strchr(simple,OP(scan))) { |
594 | if (flags & SCF_DO_SUBSTR) { |
595 | scan_commit(data); |
596 | data->pos_min++; |
597 | } |
598 | min++; |
599 | } else if (regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) { |
600 | data->flags |= (OP(scan) == MEOL |
601 | ? SF_BEFORE_MEOL |
602 | : SF_BEFORE_SEOL); |
603 | } else if (regkind[(U8)OP(scan)] == BRANCHJ |
604 | && (scan->flags || data) |
605 | && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) { |
606 | I32 deltanext, minnext; |
607 | regnode *nscan; |
608 | |
609 | data_fake.flags = 0; |
610 | next = regnext(scan); |
611 | nscan = NEXTOPER(NEXTOPER(scan)); |
612 | minnext = study_chunk(&nscan, &deltanext, last, &data_fake, 0); |
613 | if (scan->flags) { |
614 | if (deltanext) { |
615 | FAIL("variable length lookbehind not implemented"); |
616 | } else if (minnext > U8_MAX) { |
617 | FAIL2("lookbehind longer than %d not implemented", U8_MAX); |
618 | } |
619 | scan->flags = minnext; |
620 | } |
621 | if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR)) |
622 | pars++; |
405ff068 |
623 | if (data && (data_fake.flags & SF_HAS_EVAL)) |
c277df42 |
624 | data->flags |= SF_HAS_EVAL; |
625 | } else if (OP(scan) == OPEN) { |
626 | pars++; |
627 | } else if (OP(scan) == CLOSE && ARG(scan) == is_par) { |
c277df42 |
628 | next = regnext(scan); |
629 | |
630 | if ( next && (OP(next) != WHILEM) && next < last) |
c277df42 |
631 | is_par = 0; /* Disable optimization */ |
632 | } else if (OP(scan) == EVAL) { |
633 | if (data) |
634 | data->flags |= SF_HAS_EVAL; |
635 | } |
636 | /* Else: zero-length, ignore. */ |
637 | scan = regnext(scan); |
638 | } |
639 | |
640 | finish: |
641 | *scanp = scan; |
642 | *deltap = is_inf ? I32_MAX : delta; |
643 | if (flags & SCF_DO_SUBSTR && is_inf) |
644 | data->pos_delta = I32_MAX - data->pos_min; |
645 | if (is_par > U8_MAX) |
646 | is_par = 0; |
647 | if (is_par && pars==1 && data) { |
648 | data->flags |= SF_IN_PAR; |
649 | data->flags &= ~SF_HAS_PAR; |
650 | } else if (pars && data) { |
651 | data->flags |= SF_HAS_PAR; |
652 | data->flags &= ~SF_IN_PAR; |
653 | } |
654 | return min; |
655 | } |
656 | |
76e3520e |
657 | STATIC I32 |
c277df42 |
658 | add_data(I32 n, char *s) |
659 | { |
77d41b28 |
660 | if (regcomp_rx->data) { |
661 | Renewc(regcomp_rx->data, |
662 | sizeof(*regcomp_rx->data) + sizeof(void*) * (regcomp_rx->data->count + n - 1), |
c277df42 |
663 | char, struct reg_data); |
77d41b28 |
664 | Renew(regcomp_rx->data->what, regcomp_rx->data->count + n, U8); |
665 | regcomp_rx->data->count += n; |
c277df42 |
666 | } else { |
77d41b28 |
667 | Newc(1207, regcomp_rx->data, sizeof(*regcomp_rx->data) + sizeof(void*) * (n - 1), |
c277df42 |
668 | char, struct reg_data); |
77d41b28 |
669 | New(1208, regcomp_rx->data->what, n, U8); |
670 | regcomp_rx->data->count = n; |
c277df42 |
671 | } |
77d41b28 |
672 | Copy(s, regcomp_rx->data->what + regcomp_rx->data->count - n, n, U8); |
673 | return regcomp_rx->data->count - n; |
c277df42 |
674 | } |
675 | |
a687059c |
676 | /* |
e50aee73 |
677 | - pregcomp - compile a regular expression into internal code |
a687059c |
678 | * |
679 | * We can't allocate space until we know how big the compiled form will be, |
680 | * but we can't compile it (and thus know how big it is) until we've got a |
681 | * place to put the code. So we cheat: we compile it twice, once with code |
682 | * generation turned off and size counting turned on, and once "for real". |
683 | * This also means that we don't allocate space until we are sure that the |
684 | * thing really will compile successfully, and we never have to move the |
685 | * code and thus invalidate pointers into it. (Note that it has to be in |
686 | * one piece because free() must be able to free it all.) [NB: not true in perl] |
687 | * |
688 | * Beware that the optimization-preparation code in here knows about some |
689 | * of the structure of the compiled regexp. [I'll say.] |
690 | */ |
691 | regexp * |
8ac85365 |
692 | pregcomp(char *exp, char *xend, PMOP *pm) |
a687059c |
693 | { |
a0d0e21e |
694 | register regexp *r; |
c277df42 |
695 | regnode *scan; |
696 | SV **longest; |
697 | SV *longest_fixed; |
698 | SV *longest_float; |
699 | regnode *first; |
a0d0e21e |
700 | I32 flags; |
a0d0e21e |
701 | I32 minlen = 0; |
702 | I32 sawplus = 0; |
703 | I32 sawopen = 0; |
704 | |
705 | if (exp == NULL) |
c277df42 |
706 | FAIL("NULL regexp argument"); |
a0d0e21e |
707 | |
bbce6d69 |
708 | regprecomp = savepvn(exp, xend - exp); |
c277df42 |
709 | DEBUG_r(PerlIO_printf(Perl_debug_log, "compiling RE `%*s'\n", |
710 | xend - exp, regprecomp)); |
a0d0e21e |
711 | regflags = pm->op_pmflags; |
bbce6d69 |
712 | regsawback = 0; |
713 | |
c277df42 |
714 | regseen = 0; |
715 | seen_zerolen = *exp == '^' ? -1 : 0; |
716 | extralen = 0; |
717 | |
bbce6d69 |
718 | /* First pass: determine size, legality. */ |
77d41b28 |
719 | regcomp_parse = exp; |
a0d0e21e |
720 | regxend = xend; |
a0d0e21e |
721 | regnaughty = 0; |
a0d0e21e |
722 | regnpar = 1; |
723 | regsize = 0L; |
724 | regcode = ®dummy; |
c277df42 |
725 | regc((U8)MAGIC, (char*)regcode); |
a0d0e21e |
726 | if (reg(0, &flags) == NULL) { |
727 | Safefree(regprecomp); |
728 | regprecomp = Nullch; |
729 | return(NULL); |
730 | } |
c277df42 |
731 | DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", regsize)); |
732 | |
733 | DEBUG_r( |
734 | if (!colorset) { |
735 | int i = 0; |
76e3520e |
736 | char *s = PerlEnv_getenv("TERMCAP_COLORS"); |
c277df42 |
737 | |
738 | colorset = 1; |
739 | if (s) { |
740 | colors[0] = s = savepv(s); |
741 | while (++i < 4) { |
742 | s = strchr(s, '\t'); |
743 | if (!s) |
744 | FAIL("Not enough TABs in TERMCAP_COLORS"); |
745 | *s = '\0'; |
746 | colors[i] = ++s; |
747 | } |
748 | } else { |
749 | while (i < 4) |
750 | colors[i++] = ""; |
751 | } |
752 | /* Reset colors: */ |
753 | PerlIO_printf(Perl_debug_log, "%s%s%s%s", |
754 | colors[0],colors[1],colors[2],colors[3]); |
755 | } |
756 | ); |
a0d0e21e |
757 | |
c277df42 |
758 | /* Small enough for pointer-storage convention? |
759 | If extralen==0, this means that we will not need long jumps. */ |
c277df42 |
760 | if (regsize >= 0x10000L && extralen) |
761 | regsize += extralen; |
762 | else |
763 | extralen = 0; |
a0d0e21e |
764 | |
bbce6d69 |
765 | /* Allocate space and initialize. */ |
c277df42 |
766 | Newc(1001, r, sizeof(regexp) + (unsigned)regsize * sizeof(regnode), |
767 | char, regexp); |
a0d0e21e |
768 | if (r == NULL) |
769 | FAIL("regexp out of space"); |
c277df42 |
770 | r->refcnt = 1; |
bbce6d69 |
771 | r->prelen = xend - exp; |
a0d0e21e |
772 | r->precomp = regprecomp; |
773 | r->subbeg = r->subbase = NULL; |
ef64f398 |
774 | r->nparens = regnpar - 1; /* set early to validate backrefs */ |
77d41b28 |
775 | regcomp_rx = r; |
bbce6d69 |
776 | |
777 | /* Second pass: emit code. */ |
77d41b28 |
778 | regcomp_parse = exp; |
bbce6d69 |
779 | regxend = xend; |
780 | regnaughty = 0; |
a0d0e21e |
781 | regnpar = 1; |
782 | regcode = r->program; |
c277df42 |
783 | regc((U8)MAGIC, (char*) regcode++); |
784 | r->data = 0; |
a0d0e21e |
785 | if (reg(0, &flags) == NULL) |
786 | return(NULL); |
787 | |
788 | /* Dig out information for optimizations. */ |
789 | pm->op_pmflags = regflags; |
a0d0e21e |
790 | r->reganch = 0; |
c277df42 |
791 | r->regstclass = NULL; |
a0d0e21e |
792 | r->naughty = regnaughty >= 10; /* Probably an expensive pattern. */ |
c277df42 |
793 | scan = r->program + 1; /* First BRANCH. */ |
2779dcf1 |
794 | |
795 | /* XXXX To minimize changes to RE engine we always allocate |
796 | 3-units-long substrs field. */ |
797 | Newz(1004, r->substrs, 1, struct reg_substr_data); |
798 | |
c277df42 |
799 | if (OP(scan) != BRANCH) { /* Only one top-level choice. */ |
800 | scan_data_t data; |
801 | I32 fake; |
c5254dd6 |
802 | STRLEN longest_float_length, longest_fixed_length; |
a0d0e21e |
803 | |
c277df42 |
804 | StructCopy(&zero_scan_data, &data, scan_data_t); |
a0d0e21e |
805 | first = scan; |
c277df42 |
806 | /* Skip introductions and multiplicators >= 1. */ |
a0d0e21e |
807 | while ((OP(first) == OPEN && (sawopen = 1)) || |
808 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || |
809 | (OP(first) == PLUS) || |
810 | (OP(first) == MINMOD) || |
811 | (regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
812 | if (OP(first) == PLUS) |
813 | sawplus = 1; |
814 | else |
815 | first += regarglen[(U8)OP(first)]; |
816 | first = NEXTOPER(first); |
a687059c |
817 | } |
818 | |
a0d0e21e |
819 | /* Starting-point info. */ |
820 | again: |
c277df42 |
821 | if (OP(first) == EXACT); /* Empty, get anchored substr later. */ |
a0d0e21e |
822 | else if (strchr(simple+2,OP(first))) |
823 | r->regstclass = first; |
bbce6d69 |
824 | else if (regkind[(U8)OP(first)] == BOUND || |
825 | regkind[(U8)OP(first)] == NBOUND) |
a0d0e21e |
826 | r->regstclass = first; |
827 | else if (regkind[(U8)OP(first)] == BOL) { |
c277df42 |
828 | r->reganch |= (OP(first) == MBOL ? ROPT_ANCH_MBOL: ROPT_ANCH_BOL); |
a0d0e21e |
829 | first = NEXTOPER(first); |
774d564b |
830 | goto again; |
831 | } |
832 | else if (OP(first) == GPOS) { |
833 | r->reganch |= ROPT_ANCH_GPOS; |
834 | first = NEXTOPER(first); |
835 | goto again; |
a0d0e21e |
836 | } |
837 | else if ((OP(first) == STAR && |
838 | regkind[(U8)OP(NEXTOPER(first))] == ANY) && |
839 | !(r->reganch & ROPT_ANCH) ) |
840 | { |
841 | /* turn .* into ^.* with an implied $*=1 */ |
774d564b |
842 | r->reganch |= ROPT_ANCH_BOL | ROPT_IMPLICIT; |
a0d0e21e |
843 | first = NEXTOPER(first); |
774d564b |
844 | goto again; |
a0d0e21e |
845 | } |
846 | if (sawplus && (!sawopen || !regsawback)) |
847 | r->reganch |= ROPT_SKIP; /* x+ must match 1st of run */ |
848 | |
c277df42 |
849 | /* Scan is after the zeroth branch, first is atomic matcher. */ |
850 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %d\n", |
851 | first - scan + 1)); |
a0d0e21e |
852 | /* |
853 | * If there's something expensive in the r.e., find the |
854 | * longest literal string that must appear and make it the |
855 | * regmust. Resolve ties in favor of later strings, since |
856 | * the regstart check works with the beginning of the r.e. |
857 | * and avoiding duplication strengthens checking. Not a |
858 | * strong reason, but sufficient in the absence of others. |
859 | * [Now we resolve ties in favor of the earlier string if |
c277df42 |
860 | * it happens that c_offset_min has been invalidated, since the |
a0d0e21e |
861 | * earlier string may buy us something the later one won't.] |
862 | */ |
a0d0e21e |
863 | minlen = 0; |
a687059c |
864 | |
c277df42 |
865 | data.longest_fixed = newSVpv("",0); |
866 | data.longest_float = newSVpv("",0); |
867 | data.last_found = newSVpv("",0); |
868 | data.longest = &(data.longest_fixed); |
869 | first = scan; |
870 | |
871 | minlen = study_chunk(&first, &fake, scan + regsize, /* Up to end */ |
872 | &data, SCF_DO_SUBSTR); |
873 | if ( regnpar == 1 && data.longest == &(data.longest_fixed) |
874 | && data.last_start_min == 0 && data.last_end > 0 |
875 | && !seen_zerolen |
876 | && (!(regseen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS))) |
877 | r->reganch |= ROPT_CHECK_ALL; |
878 | scan_commit(&data); |
879 | SvREFCNT_dec(data.last_found); |
880 | |
c5254dd6 |
881 | longest_float_length = SvCUR(data.longest_float); |
882 | if (longest_float_length |
c277df42 |
883 | || (data.flags & SF_FL_BEFORE_EOL |
884 | && (!(data.flags & SF_FL_BEFORE_MEOL) |
885 | || (regflags & PMf_MULTILINE)))) { |
886 | if (SvCUR(data.longest_fixed) |
887 | && data.offset_fixed == data.offset_float_min) |
888 | goto remove; /* Like in (a)+. */ |
889 | |
890 | r->float_substr = data.longest_float; |
891 | r->float_min_offset = data.offset_float_min; |
892 | r->float_max_offset = data.offset_float_max; |
2779dcf1 |
893 | fbm_compile(r->float_substr, 0); |
c277df42 |
894 | BmUSEFUL(r->float_substr) = 100; |
895 | if (data.flags & SF_FL_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
896 | && (!(data.flags & SF_FL_BEFORE_MEOL) |
897 | || (regflags & PMf_MULTILINE))) |
898 | SvTAIL_on(r->float_substr); |
899 | } else { |
900 | remove: |
901 | r->float_substr = Nullsv; |
902 | SvREFCNT_dec(data.longest_float); |
c5254dd6 |
903 | longest_float_length = 0; |
a0d0e21e |
904 | } |
c277df42 |
905 | |
c5254dd6 |
906 | longest_fixed_length = SvCUR(data.longest_fixed); |
907 | if (longest_fixed_length |
c277df42 |
908 | || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
909 | && (!(data.flags & SF_FIX_BEFORE_MEOL) |
910 | || (regflags & PMf_MULTILINE)))) { |
911 | r->anchored_substr = data.longest_fixed; |
912 | r->anchored_offset = data.offset_fixed; |
2779dcf1 |
913 | fbm_compile(r->anchored_substr, 0); |
c277df42 |
914 | BmUSEFUL(r->anchored_substr) = 100; |
915 | if (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */ |
916 | && (!(data.flags & SF_FIX_BEFORE_MEOL) |
917 | || (regflags & PMf_MULTILINE))) |
918 | SvTAIL_on(r->anchored_substr); |
919 | } else { |
920 | r->anchored_substr = Nullsv; |
921 | SvREFCNT_dec(data.longest_fixed); |
c5254dd6 |
922 | longest_fixed_length = 0; |
a0d0e21e |
923 | } |
c277df42 |
924 | |
925 | /* A temporary algorithm prefers floated substr to fixed one to dig more info. */ |
c5254dd6 |
926 | if (longest_fixed_length > longest_float_length) { |
c277df42 |
927 | r->check_substr = r->anchored_substr; |
928 | r->check_offset_min = r->check_offset_max = r->anchored_offset; |
929 | if (r->reganch & ROPT_ANCH_SINGLE) |
930 | r->reganch |= ROPT_NOSCAN; |
931 | } else { |
932 | r->check_substr = r->float_substr; |
933 | r->check_offset_min = data.offset_float_min; |
934 | r->check_offset_max = data.offset_float_max; |
a0d0e21e |
935 | } |
c277df42 |
936 | } else { |
937 | /* Several toplevels. Best we can is to set minlen. */ |
938 | I32 fake; |
939 | |
940 | DEBUG_r(PerlIO_printf(Perl_debug_log, "\n")); |
941 | scan = r->program + 1; |
942 | minlen = study_chunk(&scan, &fake, scan + regsize, NULL, 0); |
943 | r->check_substr = r->anchored_substr = r->float_substr = Nullsv; |
a0d0e21e |
944 | } |
945 | |
a0d0e21e |
946 | r->minlen = minlen; |
c277df42 |
947 | if (regseen & REG_SEEN_GPOS) |
948 | r->reganch |= ROPT_GPOS_SEEN; |
949 | if (regseen & REG_SEEN_LOOKBEHIND) |
950 | r->reganch |= ROPT_LOOKBEHIND_SEEN; |
ce862d02 |
951 | if (regseen & REG_SEEN_EVAL) |
952 | r->reganch |= ROPT_EVAL_SEEN; |
a0d0e21e |
953 | Newz(1002, r->startp, regnpar, char*); |
954 | Newz(1002, r->endp, regnpar, char*); |
955 | DEBUG_r(regdump(r)); |
956 | return(r); |
a687059c |
957 | } |
958 | |
959 | /* |
960 | - reg - regular expression, i.e. main body or parenthesized thing |
961 | * |
962 | * Caller must absorb opening parenthesis. |
963 | * |
964 | * Combining parenthesis handling with the base level of regular expression |
965 | * is a trifle forced, but the need to tie the tails of the branches to what |
966 | * follows makes it hard to avoid. |
967 | */ |
76e3520e |
968 | STATIC regnode * |
8ac85365 |
969 | reg(I32 paren, I32 *flagp) |
c277df42 |
970 | /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */ |
a687059c |
971 | { |
c277df42 |
972 | register regnode *ret; /* Will be the head of the group. */ |
973 | register regnode *br; |
974 | register regnode *lastbr; |
975 | register regnode *ender = 0; |
a0d0e21e |
976 | register I32 parno = 0; |
c277df42 |
977 | I32 flags, oregflags = regflags, have_branch = 0, open = 0; |
978 | char c; |
a0d0e21e |
979 | |
821b33a5 |
980 | *flagp = 0; /* Tentatively. */ |
a0d0e21e |
981 | |
982 | /* Make an OPEN node, if parenthesized. */ |
983 | if (paren) { |
77d41b28 |
984 | if (*regcomp_parse == '?') { |
985 | regcomp_parse++; |
986 | paren = *regcomp_parse++; |
c277df42 |
987 | ret = NULL; /* For look-ahead/behind. */ |
a0d0e21e |
988 | switch (paren) { |
c277df42 |
989 | case '<': |
c277df42 |
990 | regseen |= REG_SEEN_LOOKBEHIND; |
77d41b28 |
991 | if (*regcomp_parse == '!') |
c277df42 |
992 | paren = ','; |
77d41b28 |
993 | if (*regcomp_parse != '=' && *regcomp_parse != '!') |
c277df42 |
994 | goto unknown; |
77d41b28 |
995 | regcomp_parse++; |
a0d0e21e |
996 | case '=': |
997 | case '!': |
c277df42 |
998 | seen_zerolen++; |
999 | case ':': |
1000 | case '>': |
a0d0e21e |
1001 | break; |
1002 | case '$': |
1003 | case '@': |
c277df42 |
1004 | FAIL2("Sequence (?%c...) not implemented", (int)paren); |
a0d0e21e |
1005 | break; |
1006 | case '#': |
77d41b28 |
1007 | while (*regcomp_parse && *regcomp_parse != ')') |
1008 | regcomp_parse++; |
1009 | if (*regcomp_parse != ')') |
c277df42 |
1010 | FAIL("Sequence (?#... not terminated"); |
a0d0e21e |
1011 | nextchar(); |
1012 | *flagp = TRYAGAIN; |
1013 | return NULL; |
c277df42 |
1014 | case '{': |
1015 | { |
1016 | dTHR; |
1017 | I32 count = 1, n = 0; |
1018 | char c; |
77d41b28 |
1019 | char *s = regcomp_parse; |
c277df42 |
1020 | SV *sv; |
1021 | OP_4tree *sop, *rop; |
1022 | |
1023 | seen_zerolen++; |
ce862d02 |
1024 | regseen |= REG_SEEN_EVAL; |
77d41b28 |
1025 | while (count && (c = *regcomp_parse)) { |
1026 | if (c == '\\' && regcomp_parse[1]) |
1027 | regcomp_parse++; |
c277df42 |
1028 | else if (c == '{') |
1029 | count++; |
1030 | else if (c == '}') |
1031 | count--; |
77d41b28 |
1032 | regcomp_parse++; |
c277df42 |
1033 | } |
77d41b28 |
1034 | if (*regcomp_parse != ')') |
c277df42 |
1035 | FAIL("Sequence (?{...}) not terminated or not {}-balanced"); |
1036 | if (!SIZE_ONLY) { |
1037 | AV *av; |
1038 | |
77d41b28 |
1039 | if (regcomp_parse - 1 - s) |
1040 | sv = newSVpv(s, regcomp_parse - 1 - s); |
c277df42 |
1041 | else |
1042 | sv = newSVpv("", 0); |
1043 | |
1044 | rop = sv_compile_2op(sv, &sop, "re", &av); |
1045 | |
1046 | n = add_data(3, "nso"); |
77d41b28 |
1047 | regcomp_rx->data->data[n] = (void*)rop; |
1048 | regcomp_rx->data->data[n+1] = (void*)av; |
1049 | regcomp_rx->data->data[n+2] = (void*)sop; |
c277df42 |
1050 | SvREFCNT_dec(sv); |
cc6b7395 |
1051 | } else { /* First pass */ |
e4d48cc9 |
1052 | if (curcop == &compiling) { |
1053 | if (!(hints & HINT_RE_EVAL)) |
1054 | FAIL("Eval-group not allowed, use re 'eval'"); |
1055 | } |
1056 | else { |
1057 | FAIL("Eval-group not allowed at run time"); |
1058 | } |
cc6b7395 |
1059 | if (tainted) |
1060 | FAIL("Eval-group in insecure regular expression"); |
c277df42 |
1061 | } |
1062 | |
1063 | nextchar(); |
c277df42 |
1064 | return reganode(EVAL, n); |
1065 | } |
1066 | case '(': |
1067 | { |
77d41b28 |
1068 | if (regcomp_parse[0] == '?') { |
1069 | if (regcomp_parse[1] == '=' || regcomp_parse[1] == '!' |
1070 | || regcomp_parse[1] == '<' |
1071 | || regcomp_parse[1] == '{') { /* Lookahead or eval. */ |
c277df42 |
1072 | I32 flag; |
1073 | |
1074 | ret = reg_node(LOGICAL); |
1075 | regtail(ret, reg(1, &flag)); |
1076 | goto insert_if; |
1077 | } |
77d41b28 |
1078 | } else if (regcomp_parse[0] >= '1' && regcomp_parse[0] <= '9' ) { |
1079 | parno = atoi(regcomp_parse++); |
c277df42 |
1080 | |
77d41b28 |
1081 | while (isDIGIT(*regcomp_parse)) |
1082 | regcomp_parse++; |
c277df42 |
1083 | ret = reganode(GROUPP, parno); |
1084 | if ((c = *nextchar()) != ')') |
1085 | FAIL2("Switch (?(number%c not recognized", c); |
1086 | insert_if: |
1087 | regtail(ret, reganode(IFTHEN, 0)); |
1088 | br = regbranch(&flags, 1); |
1089 | if (br == NULL) |
1090 | br = reganode(LONGJMP, 0); |
1091 | else |
1092 | regtail(br, reganode(LONGJMP, 0)); |
1093 | c = *nextchar(); |
1094 | if (c == '|') { |
1095 | lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */ |
1096 | regbranch(&flags, 1); |
1097 | regtail(ret, lastbr); |
1098 | c = *nextchar(); |
1099 | } else |
1100 | lastbr = NULL; |
1101 | if (c != ')') |
1102 | FAIL("Switch (?(condition)... contains too many branches"); |
1103 | ender = reg_node(TAIL); |
1104 | regtail(br, ender); |
1105 | if (lastbr) { |
1106 | regtail(lastbr, ender); |
1107 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); |
1108 | } else |
1109 | regtail(ret, ender); |
1110 | return ret; |
1111 | } else { |
77d41b28 |
1112 | FAIL2("Unknown condition for (?(%.2s", regcomp_parse); |
c277df42 |
1113 | } |
1114 | } |
1b1626e4 |
1115 | case 0: |
c277df42 |
1116 | FAIL("Sequence (? incomplete"); |
1b1626e4 |
1117 | break; |
a0d0e21e |
1118 | default: |
77d41b28 |
1119 | --regcomp_parse; |
1120 | while (*regcomp_parse && strchr("iogcmsx", *regcomp_parse)) { |
1121 | if (*regcomp_parse != 'o') |
1122 | pmflag(®flags, *regcomp_parse); |
1123 | ++regcomp_parse; |
48c036b1 |
1124 | } |
c277df42 |
1125 | unknown: |
77d41b28 |
1126 | if (*regcomp_parse != ')') |
1127 | FAIL2("Sequence (?%c...) not recognized", *regcomp_parse); |
a0d0e21e |
1128 | nextchar(); |
1129 | *flagp = TRYAGAIN; |
1130 | return NULL; |
1131 | } |
1132 | } |
1133 | else { |
1134 | parno = regnpar; |
1135 | regnpar++; |
1136 | ret = reganode(OPEN, parno); |
c277df42 |
1137 | open = 1; |
a0d0e21e |
1138 | } |
1139 | } else |
1140 | ret = NULL; |
1141 | |
1142 | /* Pick up the branches, linking them together. */ |
c277df42 |
1143 | br = regbranch(&flags, 1); |
a0d0e21e |
1144 | if (br == NULL) |
1145 | return(NULL); |
77d41b28 |
1146 | if (*regcomp_parse == '|') { |
c277df42 |
1147 | if (!SIZE_ONLY && extralen) { |
1148 | reginsert(BRANCHJ, br); |
1149 | } else |
1150 | reginsert(BRANCH, br); |
1151 | have_branch = 1; |
1152 | if (SIZE_ONLY) |
1153 | extralen += 1; /* For BRANCHJ-BRANCH. */ |
1154 | } else if (paren == ':') { |
1155 | *flagp |= flags&SIMPLE; |
1156 | } |
1157 | if (open) { /* Starts with OPEN. */ |
1158 | regtail(ret, br); /* OPEN -> first. */ |
1159 | } else if (paren != '?') /* Not Conditional */ |
a0d0e21e |
1160 | ret = br; |
821b33a5 |
1161 | if (flags&HASWIDTH) |
1162 | *flagp |= HASWIDTH; |
a0d0e21e |
1163 | *flagp |= flags&SPSTART; |
c277df42 |
1164 | lastbr = br; |
77d41b28 |
1165 | while (*regcomp_parse == '|') { |
c277df42 |
1166 | if (!SIZE_ONLY && extralen) { |
1167 | ender = reganode(LONGJMP,0); |
1168 | regtail(NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */ |
1169 | } |
1170 | if (SIZE_ONLY) |
1171 | extralen += 2; /* Account for LONGJMP. */ |
a0d0e21e |
1172 | nextchar(); |
c277df42 |
1173 | br = regbranch(&flags, 0); |
a687059c |
1174 | if (br == NULL) |
a0d0e21e |
1175 | return(NULL); |
c277df42 |
1176 | regtail(lastbr, br); /* BRANCH -> BRANCH. */ |
1177 | lastbr = br; |
821b33a5 |
1178 | if (flags&HASWIDTH) |
1179 | *flagp |= HASWIDTH; |
a687059c |
1180 | *flagp |= flags&SPSTART; |
a0d0e21e |
1181 | } |
1182 | |
c277df42 |
1183 | if (have_branch || paren != ':') { |
1184 | /* Make a closing node, and hook it on the end. */ |
1185 | switch (paren) { |
1186 | case ':': |
1187 | ender = reg_node(TAIL); |
1188 | break; |
1189 | case 1: |
1190 | ender = reganode(CLOSE, parno); |
1191 | break; |
1192 | case '<': |
c277df42 |
1193 | case ',': |
1194 | case '=': |
1195 | case '!': |
c277df42 |
1196 | *flagp &= ~HASWIDTH; |
821b33a5 |
1197 | /* FALL THROUGH */ |
1198 | case '>': |
1199 | ender = reg_node(SUCCEED); |
c277df42 |
1200 | break; |
1201 | case 0: |
1202 | ender = reg_node(END); |
1203 | break; |
1204 | } |
1205 | regtail(lastbr, ender); |
a0d0e21e |
1206 | |
c277df42 |
1207 | if (have_branch) { |
1208 | /* Hook the tails of the branches to the closing node. */ |
1209 | for (br = ret; br != NULL; br = regnext(br)) { |
1210 | regoptail(br, ender); |
1211 | } |
1212 | } |
a0d0e21e |
1213 | } |
c277df42 |
1214 | |
1215 | { |
1216 | char *p; |
1217 | static char parens[] = "=!<,>"; |
1218 | |
1219 | if (paren && (p = strchr(parens, paren))) { |
1220 | int node = ((p - parens) % 2) ? UNLESSM : IFMATCH; |
1221 | int flag = (p - parens) > 1; |
1222 | |
1223 | if (paren == '>') |
1224 | node = SUSPEND, flag = 0; |
1225 | reginsert(node,ret); |
c277df42 |
1226 | ret->flags = flag; |
c277df42 |
1227 | regtail(ret, reg_node(TAIL)); |
1228 | } |
a0d0e21e |
1229 | } |
1230 | |
1231 | /* Check for proper termination. */ |
77d41b28 |
1232 | if (paren && (regcomp_parse >= regxend || *nextchar() != ')')) { |
a0d0e21e |
1233 | FAIL("unmatched () in regexp"); |
77d41b28 |
1234 | } else if (!paren && regcomp_parse < regxend) { |
1235 | if (*regcomp_parse == ')') { |
a0d0e21e |
1236 | FAIL("unmatched () in regexp"); |
1237 | } else |
1238 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
1239 | /* NOTREACHED */ |
1240 | } |
c277df42 |
1241 | if (paren != 0) { |
1242 | regflags = oregflags; |
1243 | } |
a687059c |
1244 | |
a0d0e21e |
1245 | return(ret); |
a687059c |
1246 | } |
1247 | |
1248 | /* |
1249 | - regbranch - one alternative of an | operator |
1250 | * |
1251 | * Implements the concatenation operator. |
1252 | */ |
76e3520e |
1253 | STATIC regnode * |
c277df42 |
1254 | regbranch(I32 *flagp, I32 first) |
a687059c |
1255 | { |
c277df42 |
1256 | register regnode *ret; |
1257 | register regnode *chain = NULL; |
1258 | register regnode *latest; |
1259 | I32 flags = 0, c = 0; |
a0d0e21e |
1260 | |
c277df42 |
1261 | if (first) |
1262 | ret = NULL; |
1263 | else { |
1264 | if (!SIZE_ONLY && extralen) |
1265 | ret = reganode(BRANCHJ,0); |
1266 | else |
1267 | ret = reg_node(BRANCH); |
1268 | } |
1269 | |
1270 | if (!first && SIZE_ONLY) |
1271 | extralen += 1; /* BRANCHJ */ |
1272 | |
1273 | *flagp = WORST; /* Tentatively. */ |
a0d0e21e |
1274 | |
77d41b28 |
1275 | regcomp_parse--; |
a0d0e21e |
1276 | nextchar(); |
77d41b28 |
1277 | while (regcomp_parse < regxend && *regcomp_parse != '|' && *regcomp_parse != ')') { |
a0d0e21e |
1278 | flags &= ~TRYAGAIN; |
1279 | latest = regpiece(&flags); |
1280 | if (latest == NULL) { |
1281 | if (flags & TRYAGAIN) |
1282 | continue; |
1283 | return(NULL); |
c277df42 |
1284 | } else if (ret == NULL) |
1285 | ret = latest; |
a0d0e21e |
1286 | *flagp |= flags&HASWIDTH; |
c277df42 |
1287 | if (chain == NULL) /* First piece. */ |
a0d0e21e |
1288 | *flagp |= flags&SPSTART; |
1289 | else { |
1290 | regnaughty++; |
1291 | regtail(chain, latest); |
a687059c |
1292 | } |
a0d0e21e |
1293 | chain = latest; |
c277df42 |
1294 | c++; |
1295 | } |
1296 | if (chain == NULL) { /* Loop ran zero times. */ |
1297 | chain = reg_node(NOTHING); |
1298 | if (ret == NULL) |
1299 | ret = chain; |
1300 | } |
1301 | if (c == 1) { |
1302 | *flagp |= flags&SIMPLE; |
a0d0e21e |
1303 | } |
a687059c |
1304 | |
a0d0e21e |
1305 | return(ret); |
a687059c |
1306 | } |
1307 | |
1308 | /* |
1309 | - regpiece - something followed by possible [*+?] |
1310 | * |
1311 | * Note that the branching code sequences used for ? and the general cases |
1312 | * of * and + are somewhat optimized: they use the same NOTHING node as |
1313 | * both the endmarker for their branch list and the body of the last branch. |
1314 | * It might seem that this node could be dispensed with entirely, but the |
1315 | * endmarker role is not redundant. |
1316 | */ |
76e3520e |
1317 | STATIC regnode * |
8ac85365 |
1318 | regpiece(I32 *flagp) |
a687059c |
1319 | { |
c277df42 |
1320 | register regnode *ret; |
a0d0e21e |
1321 | register char op; |
1322 | register char *next; |
1323 | I32 flags; |
77d41b28 |
1324 | char *origparse = regcomp_parse; |
a0d0e21e |
1325 | char *maxpos; |
1326 | I32 min; |
c277df42 |
1327 | I32 max = REG_INFTY; |
a0d0e21e |
1328 | |
1329 | ret = regatom(&flags); |
1330 | if (ret == NULL) { |
1331 | if (flags & TRYAGAIN) |
1332 | *flagp |= TRYAGAIN; |
1333 | return(NULL); |
1334 | } |
1335 | |
77d41b28 |
1336 | op = *regcomp_parse; |
a0d0e21e |
1337 | |
77d41b28 |
1338 | if (op == '{' && regcurly(regcomp_parse)) { |
1339 | next = regcomp_parse + 1; |
a0d0e21e |
1340 | maxpos = Nullch; |
1341 | while (isDIGIT(*next) || *next == ',') { |
1342 | if (*next == ',') { |
1343 | if (maxpos) |
1344 | break; |
1345 | else |
1346 | maxpos = next; |
a687059c |
1347 | } |
a0d0e21e |
1348 | next++; |
1349 | } |
1350 | if (*next == '}') { /* got one */ |
1351 | if (!maxpos) |
1352 | maxpos = next; |
77d41b28 |
1353 | regcomp_parse++; |
1354 | min = atoi(regcomp_parse); |
a0d0e21e |
1355 | if (*maxpos == ',') |
1356 | maxpos++; |
1357 | else |
77d41b28 |
1358 | maxpos = regcomp_parse; |
a0d0e21e |
1359 | max = atoi(maxpos); |
1360 | if (!max && *maxpos != '0') |
c277df42 |
1361 | max = REG_INFTY; /* meaning "infinity" */ |
1362 | else if (max >= REG_INFTY) |
1363 | FAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1); |
77d41b28 |
1364 | regcomp_parse = next; |
a0d0e21e |
1365 | nextchar(); |
1366 | |
1367 | do_curly: |
1368 | if ((flags&SIMPLE)) { |
1369 | regnaughty += 2 + regnaughty / 2; |
1370 | reginsert(CURLY, ret); |
1371 | } |
1372 | else { |
1373 | regnaughty += 4 + regnaughty; /* compound interest */ |
c277df42 |
1374 | regtail(ret, reg_node(WHILEM)); |
1375 | if (!SIZE_ONLY && extralen) { |
1376 | reginsert(LONGJMP,ret); |
1377 | reginsert(NOTHING,ret); |
1378 | NEXT_OFF(ret) = 3; /* Go over LONGJMP. */ |
1379 | } |
a0d0e21e |
1380 | reginsert(CURLYX,ret); |
c277df42 |
1381 | if (!SIZE_ONLY && extralen) |
1382 | NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */ |
1383 | regtail(ret, reg_node(NOTHING)); |
1384 | if (SIZE_ONLY) |
1385 | extralen += 3; |
a0d0e21e |
1386 | } |
c277df42 |
1387 | ret->flags = 0; |
a0d0e21e |
1388 | |
1389 | if (min > 0) |
821b33a5 |
1390 | *flagp = WORST; |
1391 | if (max > 0) |
1392 | *flagp |= HASWIDTH; |
a0d0e21e |
1393 | if (max && max < min) |
c277df42 |
1394 | FAIL("Can't do {n,m} with n > m"); |
1395 | if (!SIZE_ONLY) { |
1396 | ARG1_SET(ret, min); |
1397 | ARG2_SET(ret, max); |
a687059c |
1398 | } |
a687059c |
1399 | |
a0d0e21e |
1400 | goto nest_check; |
a687059c |
1401 | } |
a0d0e21e |
1402 | } |
a687059c |
1403 | |
a0d0e21e |
1404 | if (!ISMULT1(op)) { |
1405 | *flagp = flags; |
a687059c |
1406 | return(ret); |
a0d0e21e |
1407 | } |
bb20fd44 |
1408 | |
c277df42 |
1409 | #if 0 /* Now runtime fix should be reliable. */ |
bb20fd44 |
1410 | if (!(flags&HASWIDTH) && op != '?') |
c277df42 |
1411 | FAIL("regexp *+ operand could be empty"); |
1412 | #endif |
bb20fd44 |
1413 | |
a0d0e21e |
1414 | nextchar(); |
1415 | |
821b33a5 |
1416 | *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH); |
a0d0e21e |
1417 | |
1418 | if (op == '*' && (flags&SIMPLE)) { |
1419 | reginsert(STAR, ret); |
c277df42 |
1420 | ret->flags = 0; |
a0d0e21e |
1421 | regnaughty += 4; |
1422 | } |
1423 | else if (op == '*') { |
1424 | min = 0; |
1425 | goto do_curly; |
1426 | } else if (op == '+' && (flags&SIMPLE)) { |
1427 | reginsert(PLUS, ret); |
c277df42 |
1428 | ret->flags = 0; |
a0d0e21e |
1429 | regnaughty += 3; |
1430 | } |
1431 | else if (op == '+') { |
1432 | min = 1; |
1433 | goto do_curly; |
1434 | } else if (op == '?') { |
1435 | min = 0; max = 1; |
1436 | goto do_curly; |
1437 | } |
1438 | nest_check: |
c277df42 |
1439 | if (dowarn && !SIZE_ONLY && !(flags&HASWIDTH) && max > 10000) { |
a0d0e21e |
1440 | warn("%.*s matches null string many times", |
77d41b28 |
1441 | regcomp_parse - origparse, origparse); |
a0d0e21e |
1442 | } |
1443 | |
77d41b28 |
1444 | if (*regcomp_parse == '?') { |
a0d0e21e |
1445 | nextchar(); |
1446 | reginsert(MINMOD, ret); |
c277df42 |
1447 | regtail(ret, ret + NODE_STEP_REGNODE); |
a0d0e21e |
1448 | } |
77d41b28 |
1449 | if (ISMULT2(regcomp_parse)) |
a0d0e21e |
1450 | FAIL("nested *?+ in regexp"); |
1451 | |
1452 | return(ret); |
a687059c |
1453 | } |
1454 | |
1455 | /* |
1456 | - regatom - the lowest level |
1457 | * |
1458 | * Optimization: gobbles an entire sequence of ordinary characters so that |
1459 | * it can turn them into a single node, which is smaller to store and |
1460 | * faster to run. Backslashed characters are exceptions, each becoming a |
1461 | * separate node; the code is simpler that way and it's not worth fixing. |
1462 | * |
1463 | * [Yes, it is worth fixing, some scripts can run twice the speed.] |
1464 | */ |
76e3520e |
1465 | STATIC regnode * |
8ac85365 |
1466 | regatom(I32 *flagp) |
a687059c |
1467 | { |
c277df42 |
1468 | register regnode *ret = 0; |
a0d0e21e |
1469 | I32 flags; |
1470 | |
1471 | *flagp = WORST; /* Tentatively. */ |
1472 | |
1473 | tryagain: |
77d41b28 |
1474 | switch (*regcomp_parse) { |
a0d0e21e |
1475 | case '^': |
c277df42 |
1476 | seen_zerolen++; |
a0d0e21e |
1477 | nextchar(); |
1478 | if (regflags & PMf_MULTILINE) |
c277df42 |
1479 | ret = reg_node(MBOL); |
a0d0e21e |
1480 | else if (regflags & PMf_SINGLELINE) |
c277df42 |
1481 | ret = reg_node(SBOL); |
a0d0e21e |
1482 | else |
c277df42 |
1483 | ret = reg_node(BOL); |
a0d0e21e |
1484 | break; |
1485 | case '$': |
77d41b28 |
1486 | if (regcomp_parse[1]) |
c277df42 |
1487 | seen_zerolen++; |
a0d0e21e |
1488 | nextchar(); |
1489 | if (regflags & PMf_MULTILINE) |
c277df42 |
1490 | ret = reg_node(MEOL); |
a0d0e21e |
1491 | else if (regflags & PMf_SINGLELINE) |
c277df42 |
1492 | ret = reg_node(SEOL); |
a0d0e21e |
1493 | else |
c277df42 |
1494 | ret = reg_node(EOL); |
a0d0e21e |
1495 | break; |
1496 | case '.': |
1497 | nextchar(); |
1498 | if (regflags & PMf_SINGLELINE) |
c277df42 |
1499 | ret = reg_node(SANY); |
a0d0e21e |
1500 | else |
c277df42 |
1501 | ret = reg_node(ANY); |
a0d0e21e |
1502 | regnaughty++; |
1503 | *flagp |= HASWIDTH|SIMPLE; |
1504 | break; |
1505 | case '[': |
77d41b28 |
1506 | regcomp_parse++; |
a0d0e21e |
1507 | ret = regclass(); |
1508 | *flagp |= HASWIDTH|SIMPLE; |
1509 | break; |
1510 | case '(': |
1511 | nextchar(); |
1512 | ret = reg(1, &flags); |
1513 | if (ret == NULL) { |
1514 | if (flags & TRYAGAIN) |
1515 | goto tryagain; |
1516 | return(NULL); |
1517 | } |
c277df42 |
1518 | *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE); |
a0d0e21e |
1519 | break; |
1520 | case '|': |
1521 | case ')': |
1522 | if (flags & TRYAGAIN) { |
1523 | *flagp |= TRYAGAIN; |
1524 | return NULL; |
1525 | } |
77d41b28 |
1526 | FAIL2("internal urp in regexp at /%s/", regcomp_parse); |
a0d0e21e |
1527 | /* Supposed to be caught earlier. */ |
1528 | break; |
85afd4ae |
1529 | case '{': |
77d41b28 |
1530 | if (!regcurly(regcomp_parse)) { |
1531 | regcomp_parse++; |
85afd4ae |
1532 | goto defchar; |
1533 | } |
1534 | /* FALL THROUGH */ |
a0d0e21e |
1535 | case '?': |
1536 | case '+': |
1537 | case '*': |
3115e423 |
1538 | FAIL("?+*{} follows nothing in regexp"); |
a0d0e21e |
1539 | break; |
1540 | case '\\': |
77d41b28 |
1541 | switch (*++regcomp_parse) { |
a0d0e21e |
1542 | case 'A': |
c277df42 |
1543 | seen_zerolen++; |
1544 | ret = reg_node(SBOL); |
a0d0e21e |
1545 | *flagp |= SIMPLE; |
1546 | nextchar(); |
1547 | break; |
1548 | case 'G': |
c277df42 |
1549 | ret = reg_node(GPOS); |
1550 | regseen |= REG_SEEN_GPOS; |
a0d0e21e |
1551 | *flagp |= SIMPLE; |
1552 | nextchar(); |
1553 | break; |
1554 | case 'Z': |
c277df42 |
1555 | ret = reg_node(SEOL); |
a0d0e21e |
1556 | *flagp |= SIMPLE; |
1557 | nextchar(); |
1558 | break; |
b85d18e9 |
1559 | case 'z': |
1560 | ret = reg_node(EOS); |
1561 | *flagp |= SIMPLE; |
1562 | seen_zerolen++; /* Do not optimize RE away */ |
1563 | nextchar(); |
1564 | break; |
a0d0e21e |
1565 | case 'w': |
c277df42 |
1566 | ret = reg_node((regflags & PMf_LOCALE) ? ALNUML : ALNUM); |
a0d0e21e |
1567 | *flagp |= HASWIDTH|SIMPLE; |
1568 | nextchar(); |
1569 | break; |
1570 | case 'W': |
c277df42 |
1571 | ret = reg_node((regflags & PMf_LOCALE) ? NALNUML : NALNUM); |
a0d0e21e |
1572 | *flagp |= HASWIDTH|SIMPLE; |
1573 | nextchar(); |
1574 | break; |
1575 | case 'b': |
c277df42 |
1576 | seen_zerolen++; |
1577 | ret = reg_node((regflags & PMf_LOCALE) ? BOUNDL : BOUND); |
a0d0e21e |
1578 | *flagp |= SIMPLE; |
1579 | nextchar(); |
1580 | break; |
1581 | case 'B': |
c277df42 |
1582 | seen_zerolen++; |
1583 | ret = reg_node((regflags & PMf_LOCALE) ? NBOUNDL : NBOUND); |
a0d0e21e |
1584 | *flagp |= SIMPLE; |
1585 | nextchar(); |
1586 | break; |
1587 | case 's': |
c277df42 |
1588 | ret = reg_node((regflags & PMf_LOCALE) ? SPACEL : SPACE); |
a0d0e21e |
1589 | *flagp |= HASWIDTH|SIMPLE; |
1590 | nextchar(); |
1591 | break; |
1592 | case 'S': |
c277df42 |
1593 | ret = reg_node((regflags & PMf_LOCALE) ? NSPACEL : NSPACE); |
a0d0e21e |
1594 | *flagp |= HASWIDTH|SIMPLE; |
1595 | nextchar(); |
1596 | break; |
1597 | case 'd': |
c277df42 |
1598 | ret = reg_node(DIGIT); |
a0d0e21e |
1599 | *flagp |= HASWIDTH|SIMPLE; |
1600 | nextchar(); |
1601 | break; |
1602 | case 'D': |
c277df42 |
1603 | ret = reg_node(NDIGIT); |
a0d0e21e |
1604 | *flagp |= HASWIDTH|SIMPLE; |
1605 | nextchar(); |
1606 | break; |
1607 | case 'n': |
1608 | case 'r': |
1609 | case 't': |
1610 | case 'f': |
1611 | case 'e': |
1612 | case 'a': |
1613 | case 'x': |
1614 | case 'c': |
1615 | case '0': |
1616 | goto defchar; |
1617 | case '1': case '2': case '3': case '4': |
1618 | case '5': case '6': case '7': case '8': case '9': |
1619 | { |
77d41b28 |
1620 | I32 num = atoi(regcomp_parse); |
a0d0e21e |
1621 | |
1622 | if (num > 9 && num >= regnpar) |
1623 | goto defchar; |
1624 | else { |
ef64f398 |
1625 | if (!SIZE_ONLY && num > regcomp_rx->nparens) |
1626 | FAIL("reference to nonexistent group"); |
a0d0e21e |
1627 | regsawback = 1; |
c8756f30 |
1628 | ret = reganode((regflags & PMf_FOLD) |
1629 | ? ((regflags & PMf_LOCALE) ? REFFL : REFF) |
1630 | : REF, num); |
a0d0e21e |
1631 | *flagp |= HASWIDTH; |
77d41b28 |
1632 | while (isDIGIT(*regcomp_parse)) |
1633 | regcomp_parse++; |
1634 | regcomp_parse--; |
a0d0e21e |
1635 | nextchar(); |
1636 | } |
1637 | } |
1638 | break; |
1639 | case '\0': |
77d41b28 |
1640 | if (regcomp_parse >= regxend) |
a0d0e21e |
1641 | FAIL("trailing \\ in regexp"); |
1642 | /* FALL THROUGH */ |
1643 | default: |
1644 | goto defchar; |
1645 | } |
1646 | break; |
4633a7c4 |
1647 | |
1648 | case '#': |
1649 | if (regflags & PMf_EXTENDED) { |
77d41b28 |
1650 | while (regcomp_parse < regxend && *regcomp_parse != '\n') regcomp_parse++; |
1651 | if (regcomp_parse < regxend) |
4633a7c4 |
1652 | goto tryagain; |
1653 | } |
1654 | /* FALL THROUGH */ |
1655 | |
a0d0e21e |
1656 | default: { |
1657 | register I32 len; |
c277df42 |
1658 | register U8 ender; |
a0d0e21e |
1659 | register char *p; |
c277df42 |
1660 | char *oldp, *s; |
a0d0e21e |
1661 | I32 numlen; |
1662 | |
77d41b28 |
1663 | regcomp_parse++; |
a0d0e21e |
1664 | |
1665 | defchar: |
c277df42 |
1666 | ret = reg_node((regflags & PMf_FOLD) |
bbce6d69 |
1667 | ? ((regflags & PMf_LOCALE) ? EXACTFL : EXACTF) |
1668 | : EXACT); |
161b471a |
1669 | s = (char *) OPERAND(ret); |
c277df42 |
1670 | regc(0, s++); /* save spot for len */ |
77d41b28 |
1671 | for (len = 0, p = regcomp_parse - 1; |
a0d0e21e |
1672 | len < 127 && p < regxend; |
1673 | len++) |
1674 | { |
1675 | oldp = p; |
5b5a24f7 |
1676 | |
1677 | if (regflags & PMf_EXTENDED) |
1678 | p = regwhite(p, regxend); |
a0d0e21e |
1679 | switch (*p) { |
1680 | case '^': |
1681 | case '$': |
1682 | case '.': |
1683 | case '[': |
1684 | case '(': |
1685 | case ')': |
1686 | case '|': |
1687 | goto loopdone; |
1688 | case '\\': |
1689 | switch (*++p) { |
1690 | case 'A': |
1691 | case 'G': |
1692 | case 'Z': |
b85d18e9 |
1693 | case 'z': |
a0d0e21e |
1694 | case 'w': |
1695 | case 'W': |
1696 | case 'b': |
1697 | case 'B': |
1698 | case 's': |
1699 | case 'S': |
1700 | case 'd': |
1701 | case 'D': |
1702 | --p; |
1703 | goto loopdone; |
1704 | case 'n': |
1705 | ender = '\n'; |
1706 | p++; |
a687059c |
1707 | break; |
a0d0e21e |
1708 | case 'r': |
1709 | ender = '\r'; |
1710 | p++; |
a687059c |
1711 | break; |
a0d0e21e |
1712 | case 't': |
1713 | ender = '\t'; |
1714 | p++; |
a687059c |
1715 | break; |
a0d0e21e |
1716 | case 'f': |
1717 | ender = '\f'; |
1718 | p++; |
a687059c |
1719 | break; |
a0d0e21e |
1720 | case 'e': |
1721 | ender = '\033'; |
1722 | p++; |
a687059c |
1723 | break; |
a0d0e21e |
1724 | case 'a': |
1725 | ender = '\007'; |
1726 | p++; |
a687059c |
1727 | break; |
a0d0e21e |
1728 | case 'x': |
1729 | ender = scan_hex(++p, 2, &numlen); |
1730 | p += numlen; |
a687059c |
1731 | break; |
a0d0e21e |
1732 | case 'c': |
1733 | p++; |
bbce6d69 |
1734 | ender = UCHARAT(p++); |
1735 | ender = toCTRL(ender); |
a687059c |
1736 | break; |
a0d0e21e |
1737 | case '0': case '1': case '2': case '3':case '4': |
1738 | case '5': case '6': case '7': case '8':case '9': |
1739 | if (*p == '0' || |
1740 | (isDIGIT(p[1]) && atoi(p) >= regnpar) ) { |
1741 | ender = scan_oct(p, 3, &numlen); |
1742 | p += numlen; |
1743 | } |
1744 | else { |
1745 | --p; |
1746 | goto loopdone; |
a687059c |
1747 | } |
1748 | break; |
a0d0e21e |
1749 | case '\0': |
1750 | if (p >= regxend) |
a687059c |
1751 | FAIL("trailing \\ in regexp"); |
1752 | /* FALL THROUGH */ |
a0d0e21e |
1753 | default: |
1754 | ender = *p++; |
1755 | break; |
1756 | } |
1757 | break; |
a687059c |
1758 | default: |
a0d0e21e |
1759 | ender = *p++; |
1760 | break; |
a687059c |
1761 | } |
5b5a24f7 |
1762 | if (regflags & PMf_EXTENDED) |
1763 | p = regwhite(p, regxend); |
a0d0e21e |
1764 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
1765 | if (len) |
1766 | p = oldp; |
1767 | else { |
1768 | len++; |
c277df42 |
1769 | regc(ender, s++); |
a0d0e21e |
1770 | } |
1771 | break; |
a687059c |
1772 | } |
c277df42 |
1773 | regc(ender, s++); |
a0d0e21e |
1774 | } |
1775 | loopdone: |
77d41b28 |
1776 | regcomp_parse = p - 1; |
a0d0e21e |
1777 | nextchar(); |
1778 | if (len < 0) |
1779 | FAIL("internal disaster in regexp"); |
1780 | if (len > 0) |
1781 | *flagp |= HASWIDTH; |
1782 | if (len == 1) |
1783 | *flagp |= SIMPLE; |
c277df42 |
1784 | if (!SIZE_ONLY) |
a0d0e21e |
1785 | *OPERAND(ret) = len; |
c277df42 |
1786 | regc('\0', s++); |
1787 | if (SIZE_ONLY) { |
c277df42 |
1788 | regsize += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
c277df42 |
1789 | } else { |
1790 | regcode += (len + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
1791 | } |
a687059c |
1792 | } |
a0d0e21e |
1793 | break; |
1794 | } |
a687059c |
1795 | |
a0d0e21e |
1796 | return(ret); |
a687059c |
1797 | } |
1798 | |
873ef191 |
1799 | STATIC char * |
8ac85365 |
1800 | regwhite(char *p, char *e) |
5b5a24f7 |
1801 | { |
1802 | while (p < e) { |
1803 | if (isSPACE(*p)) |
1804 | ++p; |
1805 | else if (*p == '#') { |
1806 | do { |
1807 | p++; |
1808 | } while (p < e && *p != '\n'); |
1809 | } |
1810 | else |
1811 | break; |
1812 | } |
1813 | return p; |
1814 | } |
1815 | |
76e3520e |
1816 | STATIC regnode * |
8ac85365 |
1817 | regclass(void) |
a687059c |
1818 | { |
c277df42 |
1819 | register char *opnd, *s; |
8ac85365 |
1820 | register I32 Class; |
a0d0e21e |
1821 | register I32 lastclass = 1234; |
1822 | register I32 range = 0; |
c277df42 |
1823 | register regnode *ret; |
a0d0e21e |
1824 | register I32 def; |
1825 | I32 numlen; |
1826 | |
161b471a |
1827 | s = opnd = (char *) OPERAND(regcode); |
c277df42 |
1828 | ret = reg_node(ANYOF); |
8ac85365 |
1829 | for (Class = 0; Class < 33; Class++) |
c277df42 |
1830 | regc(0, s++); |
77d41b28 |
1831 | if (*regcomp_parse == '^') { /* Complement of range. */ |
a0d0e21e |
1832 | regnaughty++; |
77d41b28 |
1833 | regcomp_parse++; |
c277df42 |
1834 | if (!SIZE_ONLY) |
bbce6d69 |
1835 | *opnd |= ANYOF_INVERT; |
1836 | } |
c277df42 |
1837 | if (!SIZE_ONLY) { |
1838 | regcode += ANY_SKIP; |
bbce6d69 |
1839 | if (regflags & PMf_FOLD) |
1840 | *opnd |= ANYOF_FOLD; |
1841 | if (regflags & PMf_LOCALE) |
1842 | *opnd |= ANYOF_LOCALE; |
c277df42 |
1843 | } else { |
1844 | regsize += ANY_SKIP; |
a0d0e21e |
1845 | } |
77d41b28 |
1846 | if (*regcomp_parse == ']' || *regcomp_parse == '-') |
a0d0e21e |
1847 | goto skipcond; /* allow 1st char to be ] or - */ |
77d41b28 |
1848 | while (regcomp_parse < regxend && *regcomp_parse != ']') { |
a0d0e21e |
1849 | skipcond: |
77d41b28 |
1850 | Class = UCHARAT(regcomp_parse++); |
1851 | if (Class == '[' && regcomp_parse + 1 < regxend && |
4599a1de |
1852 | /* I smell either [: or [= or [. -- POSIX has been here, right? */ |
77d41b28 |
1853 | (*regcomp_parse == ':' || *regcomp_parse == '=' || *regcomp_parse == '.')) { |
1854 | char posixccc = *regcomp_parse; |
1855 | char* posixccs = regcomp_parse++; |
4599a1de |
1856 | |
77d41b28 |
1857 | while (regcomp_parse < regxend && *regcomp_parse != posixccc) |
1858 | regcomp_parse++; |
1859 | if (regcomp_parse == regxend) |
4599a1de |
1860 | /* Grandfather lone [:, [=, [. */ |
77d41b28 |
1861 | regcomp_parse = posixccs; |
4599a1de |
1862 | else { |
77d41b28 |
1863 | regcomp_parse++; /* skip over the posixccc */ |
1864 | if (*regcomp_parse == ']') { |
4599a1de |
1865 | /* Not Implemented Yet. |
1866 | * (POSIX Extended Character Classes, that is) |
1867 | * The text between e.g. [: and :] would start |
77d41b28 |
1868 | * at posixccs + 1 and stop at regcomp_parse - 2. */ |
4599a1de |
1869 | if (dowarn && !SIZE_ONLY) |
1870 | warn("Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc); |
77d41b28 |
1871 | regcomp_parse++; /* skip over the ending ] */ |
4599a1de |
1872 | } |
1873 | } |
1874 | } |
8ac85365 |
1875 | if (Class == '\\') { |
77d41b28 |
1876 | Class = UCHARAT(regcomp_parse++); |
8ac85365 |
1877 | switch (Class) { |
a0d0e21e |
1878 | case 'w': |
ae5c130c |
1879 | if (!SIZE_ONLY) { |
1880 | if (regflags & PMf_LOCALE) |
bbce6d69 |
1881 | *opnd |= ANYOF_ALNUML; |
ae5c130c |
1882 | else { |
1883 | for (Class = 0; Class < 256; Class++) |
1884 | if (isALNUM(Class)) |
1885 | ANYOF_SET(opnd, Class); |
1886 | } |
bbce6d69 |
1887 | } |
a0d0e21e |
1888 | lastclass = 1234; |
1889 | continue; |
1890 | case 'W': |
ae5c130c |
1891 | if (!SIZE_ONLY) { |
1892 | if (regflags & PMf_LOCALE) |
bbce6d69 |
1893 | *opnd |= ANYOF_NALNUML; |
ae5c130c |
1894 | else { |
1895 | for (Class = 0; Class < 256; Class++) |
1896 | if (!isALNUM(Class)) |
1897 | ANYOF_SET(opnd, Class); |
1898 | } |
bbce6d69 |
1899 | } |
a0d0e21e |
1900 | lastclass = 1234; |
1901 | continue; |
1902 | case 's': |
ae5c130c |
1903 | if (!SIZE_ONLY) { |
1904 | if (regflags & PMf_LOCALE) |
bbce6d69 |
1905 | *opnd |= ANYOF_SPACEL; |
ae5c130c |
1906 | else { |
1907 | for (Class = 0; Class < 256; Class++) |
1908 | if (isSPACE(Class)) |
1909 | ANYOF_SET(opnd, Class); |
1910 | } |
bbce6d69 |
1911 | } |
a0d0e21e |
1912 | lastclass = 1234; |
1913 | continue; |
1914 | case 'S': |
ae5c130c |
1915 | if (!SIZE_ONLY) { |
1916 | if (regflags & PMf_LOCALE) |
bbce6d69 |
1917 | *opnd |= ANYOF_NSPACEL; |
ae5c130c |
1918 | else { |
1919 | for (Class = 0; Class < 256; Class++) |
1920 | if (!isSPACE(Class)) |
1921 | ANYOF_SET(opnd, Class); |
1922 | } |
bbce6d69 |
1923 | } |
a0d0e21e |
1924 | lastclass = 1234; |
1925 | continue; |
1926 | case 'd': |
ae5c130c |
1927 | if (!SIZE_ONLY) { |
1928 | for (Class = '0'; Class <= '9'; Class++) |
1929 | ANYOF_SET(opnd, Class); |
1930 | } |
a0d0e21e |
1931 | lastclass = 1234; |
1932 | continue; |
1933 | case 'D': |
ae5c130c |
1934 | if (!SIZE_ONLY) { |
1935 | for (Class = 0; Class < '0'; Class++) |
1936 | ANYOF_SET(opnd, Class); |
1937 | for (Class = '9' + 1; Class < 256; Class++) |
1938 | ANYOF_SET(opnd, Class); |
1939 | } |
a0d0e21e |
1940 | lastclass = 1234; |
1941 | continue; |
1942 | case 'n': |
8ac85365 |
1943 | Class = '\n'; |
a0d0e21e |
1944 | break; |
1945 | case 'r': |
8ac85365 |
1946 | Class = '\r'; |
a0d0e21e |
1947 | break; |
1948 | case 't': |
8ac85365 |
1949 | Class = '\t'; |
a0d0e21e |
1950 | break; |
1951 | case 'f': |
8ac85365 |
1952 | Class = '\f'; |
a0d0e21e |
1953 | break; |
1954 | case 'b': |
8ac85365 |
1955 | Class = '\b'; |
a0d0e21e |
1956 | break; |
1957 | case 'e': |
8ac85365 |
1958 | Class = '\033'; |
a0d0e21e |
1959 | break; |
1960 | case 'a': |
8ac85365 |
1961 | Class = '\007'; |
a0d0e21e |
1962 | break; |
1963 | case 'x': |
77d41b28 |
1964 | Class = scan_hex(regcomp_parse, 2, &numlen); |
1965 | regcomp_parse += numlen; |
a0d0e21e |
1966 | break; |
1967 | case 'c': |
77d41b28 |
1968 | Class = UCHARAT(regcomp_parse++); |
8ac85365 |
1969 | Class = toCTRL(Class); |
a0d0e21e |
1970 | break; |
1971 | case '0': case '1': case '2': case '3': case '4': |
1972 | case '5': case '6': case '7': case '8': case '9': |
77d41b28 |
1973 | Class = scan_oct(--regcomp_parse, 3, &numlen); |
1974 | regcomp_parse += numlen; |
a0d0e21e |
1975 | break; |
1976 | } |
1977 | } |
1978 | if (range) { |
8ac85365 |
1979 | if (lastclass > Class) |
a0d0e21e |
1980 | FAIL("invalid [] range in regexp"); |
1981 | range = 0; |
1982 | } |
1983 | else { |
8ac85365 |
1984 | lastclass = Class; |
77d41b28 |
1985 | if (*regcomp_parse == '-' && regcomp_parse+1 < regxend && |
1986 | regcomp_parse[1] != ']') { |
1987 | regcomp_parse++; |
a0d0e21e |
1988 | range = 1; |
1989 | continue; /* do it next time */ |
1990 | } |
a687059c |
1991 | } |
ae5c130c |
1992 | if (!SIZE_ONLY) { |
1993 | for ( ; lastclass <= Class; lastclass++) |
1994 | ANYOF_SET(opnd, lastclass); |
1995 | } |
8ac85365 |
1996 | lastclass = Class; |
a0d0e21e |
1997 | } |
77d41b28 |
1998 | if (*regcomp_parse != ']') |
a0d0e21e |
1999 | FAIL("unmatched [] in regexp"); |
2000 | nextchar(); |
ae5c130c |
2001 | /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */ |
2002 | if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) { |
2003 | for (Class = 0; Class < 256; ++Class) { |
2004 | if (ANYOF_TEST(opnd, Class)) { |
2005 | I32 cf = fold[Class]; |
2006 | ANYOF_SET(opnd, cf); |
2007 | } |
2008 | } |
2009 | *opnd &= ~ANYOF_FOLD; |
2010 | } |
2011 | /* optimize inverted simple patterns (e.g. [^a-z]) */ |
2012 | if (!SIZE_ONLY && (*opnd & 0xFF) == ANYOF_INVERT) { |
2013 | for (Class = 0; Class < 32; ++Class) |
2014 | opnd[1 + Class] ^= 0xFF; |
2015 | *opnd = 0; |
2016 | } |
a0d0e21e |
2017 | return ret; |
2018 | } |
2019 | |
76e3520e |
2020 | STATIC char* |
8ac85365 |
2021 | nextchar(void) |
a0d0e21e |
2022 | { |
77d41b28 |
2023 | char* retval = regcomp_parse++; |
a0d0e21e |
2024 | |
4633a7c4 |
2025 | for (;;) { |
77d41b28 |
2026 | if (*regcomp_parse == '(' && regcomp_parse[1] == '?' && |
2027 | regcomp_parse[2] == '#') { |
2028 | while (*regcomp_parse && *regcomp_parse != ')') |
2029 | regcomp_parse++; |
2030 | regcomp_parse++; |
4633a7c4 |
2031 | continue; |
2032 | } |
2033 | if (regflags & PMf_EXTENDED) { |
77d41b28 |
2034 | if (isSPACE(*regcomp_parse)) { |
2035 | regcomp_parse++; |
748a9306 |
2036 | continue; |
2037 | } |
77d41b28 |
2038 | else if (*regcomp_parse == '#') { |
2039 | while (*regcomp_parse && *regcomp_parse != '\n') |
2040 | regcomp_parse++; |
2041 | regcomp_parse++; |
748a9306 |
2042 | continue; |
2043 | } |
748a9306 |
2044 | } |
4633a7c4 |
2045 | return retval; |
a0d0e21e |
2046 | } |
a687059c |
2047 | } |
2048 | |
2049 | /* |
c277df42 |
2050 | - reg_node - emit a node |
a0d0e21e |
2051 | */ |
76e3520e |
2052 | STATIC regnode * /* Location. */ |
c277df42 |
2053 | reg_node(U8 op) |
a687059c |
2054 | { |
c277df42 |
2055 | register regnode *ret; |
2056 | register regnode *ptr; |
a687059c |
2057 | |
a0d0e21e |
2058 | ret = regcode; |
c277df42 |
2059 | if (SIZE_ONLY) { |
2060 | SIZE_ALIGN(regsize); |
c277df42 |
2061 | regsize += 1; |
a0d0e21e |
2062 | return(ret); |
2063 | } |
a687059c |
2064 | |
c277df42 |
2065 | NODE_ALIGN_FILL(ret); |
a0d0e21e |
2066 | ptr = ret; |
c277df42 |
2067 | FILL_ADVANCE_NODE(ptr, op); |
a0d0e21e |
2068 | regcode = ptr; |
a687059c |
2069 | |
a0d0e21e |
2070 | return(ret); |
a687059c |
2071 | } |
2072 | |
2073 | /* |
a0d0e21e |
2074 | - reganode - emit a node with an argument |
2075 | */ |
76e3520e |
2076 | STATIC regnode * /* Location. */ |
c277df42 |
2077 | reganode(U8 op, U32 arg) |
fe14fcc3 |
2078 | { |
c277df42 |
2079 | register regnode *ret; |
2080 | register regnode *ptr; |
fe14fcc3 |
2081 | |
a0d0e21e |
2082 | ret = regcode; |
c277df42 |
2083 | if (SIZE_ONLY) { |
2084 | SIZE_ALIGN(regsize); |
c277df42 |
2085 | regsize += 2; |
a0d0e21e |
2086 | return(ret); |
2087 | } |
fe14fcc3 |
2088 | |
c277df42 |
2089 | NODE_ALIGN_FILL(ret); |
a0d0e21e |
2090 | ptr = ret; |
c277df42 |
2091 | FILL_ADVANCE_NODE_ARG(ptr, op, arg); |
a0d0e21e |
2092 | regcode = ptr; |
fe14fcc3 |
2093 | |
a0d0e21e |
2094 | return(ret); |
fe14fcc3 |
2095 | } |
2096 | |
2097 | /* |
a0d0e21e |
2098 | - regc - emit (if appropriate) a byte of code |
2099 | */ |
76e3520e |
2100 | STATIC void |
c277df42 |
2101 | regc(U8 b, char* s) |
a687059c |
2102 | { |
c277df42 |
2103 | if (!SIZE_ONLY) |
2104 | *s = b; |
a687059c |
2105 | } |
2106 | |
2107 | /* |
a0d0e21e |
2108 | - reginsert - insert an operator in front of already-emitted operand |
2109 | * |
2110 | * Means relocating the operand. |
2111 | */ |
76e3520e |
2112 | STATIC void |
c277df42 |
2113 | reginsert(U8 op, regnode *opnd) |
a687059c |
2114 | { |
c277df42 |
2115 | register regnode *src; |
2116 | register regnode *dst; |
2117 | register regnode *place; |
2118 | register int offset = regarglen[(U8)op]; |
2119 | |
2120 | /* (regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */ |
2121 | |
2122 | if (SIZE_ONLY) { |
2123 | regsize += NODE_STEP_REGNODE + offset; |
a0d0e21e |
2124 | return; |
2125 | } |
a687059c |
2126 | |
a0d0e21e |
2127 | src = regcode; |
c277df42 |
2128 | regcode += NODE_STEP_REGNODE + offset; |
a0d0e21e |
2129 | dst = regcode; |
2130 | while (src > opnd) |
c277df42 |
2131 | StructCopy(--src, --dst, regnode); |
a0d0e21e |
2132 | |
2133 | place = opnd; /* Op node, where operand used to be. */ |
c277df42 |
2134 | src = NEXTOPER(place); |
2135 | FILL_ADVANCE_NODE(place, op); |
2136 | Zero(src, offset, regnode); |
a687059c |
2137 | } |
2138 | |
2139 | /* |
c277df42 |
2140 | - regtail - set the next-pointer at the end of a node chain of p to val. |
a0d0e21e |
2141 | */ |
76e3520e |
2142 | STATIC void |
c277df42 |
2143 | regtail(regnode *p, regnode *val) |
a687059c |
2144 | { |
c277df42 |
2145 | register regnode *scan; |
2146 | register regnode *temp; |
a0d0e21e |
2147 | register I32 offset; |
2148 | |
c277df42 |
2149 | if (SIZE_ONLY) |
a0d0e21e |
2150 | return; |
2151 | |
2152 | /* Find last node. */ |
2153 | scan = p; |
2154 | for (;;) { |
2155 | temp = regnext(scan); |
2156 | if (temp == NULL) |
2157 | break; |
2158 | scan = temp; |
2159 | } |
a687059c |
2160 | |
c277df42 |
2161 | if (reg_off_by_arg[OP(scan)]) { |
2162 | ARG_SET(scan, val - scan); |
2163 | } else { |
2164 | NEXT_OFF(scan) = val - scan; |
2165 | } |
a687059c |
2166 | } |
2167 | |
2168 | /* |
a0d0e21e |
2169 | - regoptail - regtail on operand of first argument; nop if operandless |
2170 | */ |
76e3520e |
2171 | STATIC void |
c277df42 |
2172 | regoptail(regnode *p, regnode *val) |
a687059c |
2173 | { |
a0d0e21e |
2174 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
c277df42 |
2175 | if (p == NULL || SIZE_ONLY) |
2176 | return; |
2177 | if (regkind[(U8)OP(p)] == BRANCH) { |
2178 | regtail(NEXTOPER(p), val); |
2179 | } else if ( regkind[(U8)OP(p)] == BRANCHJ) { |
2180 | regtail(NEXTOPER(NEXTOPER(p)), val); |
2181 | } else |
a0d0e21e |
2182 | return; |
a687059c |
2183 | } |
2184 | |
2185 | /* |
2186 | - regcurly - a little FSA that accepts {\d+,?\d*} |
2187 | */ |
79072805 |
2188 | STATIC I32 |
8ac85365 |
2189 | regcurly(register char *s) |
a687059c |
2190 | { |
2191 | if (*s++ != '{') |
2192 | return FALSE; |
f0fcb552 |
2193 | if (!isDIGIT(*s)) |
a687059c |
2194 | return FALSE; |
f0fcb552 |
2195 | while (isDIGIT(*s)) |
a687059c |
2196 | s++; |
2197 | if (*s == ',') |
2198 | s++; |
f0fcb552 |
2199 | while (isDIGIT(*s)) |
a687059c |
2200 | s++; |
2201 | if (*s != '}') |
2202 | return FALSE; |
2203 | return TRUE; |
2204 | } |
2205 | |
a687059c |
2206 | |
76e3520e |
2207 | STATIC regnode * |
c277df42 |
2208 | dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l) |
2209 | { |
35ff7856 |
2210 | #ifdef DEBUGGING |
c277df42 |
2211 | register char op = EXACT; /* Arbitrary non-END op. */ |
2212 | register regnode *next, *onode; |
2213 | |
2214 | while (op != END && (!last || node < last)) { |
2215 | /* While that wasn't END last time... */ |
2216 | |
2217 | NODE_ALIGN(node); |
2218 | op = OP(node); |
2219 | if (op == CLOSE) |
2220 | l--; |
2221 | next = regnext(node); |
2222 | /* Where, what. */ |
2223 | if (OP(node) == OPTIMIZED) |
2224 | goto after_print; |
2225 | regprop(sv, node); |
2226 | PerlIO_printf(Perl_debug_log, "%4d%*s%s", node - start, |
2227 | 2*l + 1, "", SvPVX(sv)); |
2228 | if (next == NULL) /* Next ptr. */ |
2229 | PerlIO_printf(Perl_debug_log, "(0)"); |
2230 | else |
2231 | PerlIO_printf(Perl_debug_log, "(%d)", next - start); |
2232 | (void)PerlIO_putc(Perl_debug_log, '\n'); |
2233 | after_print: |
2234 | if (regkind[(U8)op] == BRANCHJ) { |
2235 | register regnode *nnode = (OP(next) == LONGJMP |
2236 | ? regnext(next) |
2237 | : next); |
2238 | if (last && nnode > last) |
2239 | nnode = last; |
2240 | node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1); |
2241 | } else if (regkind[(U8)op] == BRANCH) { |
2242 | node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1); |
2243 | } else if ( op == CURLY) { /* `next' might be very big: optimizer */ |
2244 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
2245 | NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1); |
2246 | } else if (regkind[(U8)op] == CURLY && op != CURLYX) { |
2247 | node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS, |
2248 | next, sv, l + 1); |
2249 | } else if ( op == PLUS || op == STAR) { |
2250 | node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1); |
2251 | } else if (op == ANYOF) { |
2252 | node = NEXTOPER(node); |
2253 | node += ANY_SKIP; |
2254 | } else if (regkind[(U8)op] == EXACT) { |
2255 | /* Literal string, where present. */ |
2256 | node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode); |
2257 | node = NEXTOPER(node); |
2258 | } else { |
2259 | node = NEXTOPER(node); |
2260 | node += regarglen[(U8)op]; |
2261 | } |
2262 | if (op == CURLYX || op == OPEN) |
2263 | l++; |
2264 | else if (op == WHILEM) |
2265 | l--; |
2266 | } |
17c3b450 |
2267 | #endif /* DEBUGGING */ |
c277df42 |
2268 | return node; |
2269 | } |
2270 | |
a687059c |
2271 | /* |
fd181c75 |
2272 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c |
2273 | */ |
2274 | void |
8ac85365 |
2275 | regdump(regexp *r) |
a687059c |
2276 | { |
35ff7856 |
2277 | #ifdef DEBUGGING |
46fc3d4c |
2278 | SV *sv = sv_newmortal(); |
a687059c |
2279 | |
c277df42 |
2280 | (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0); |
a0d0e21e |
2281 | |
2282 | /* Header fields of interest. */ |
c277df42 |
2283 | if (r->anchored_substr) |
2284 | PerlIO_printf(Perl_debug_log, "anchored `%s%s%s'%s at %d ", |
2285 | colors[0], |
2286 | SvPVX(r->anchored_substr), |
2287 | colors[1], |
2288 | SvTAIL(r->anchored_substr) ? "$" : "", |
2289 | r->anchored_offset); |
2290 | if (r->float_substr) |
2291 | PerlIO_printf(Perl_debug_log, "floating `%s%s%s'%s at %d..%u ", |
2292 | colors[0], |
2293 | SvPVX(r->float_substr), |
2294 | colors[1], |
2295 | SvTAIL(r->float_substr) ? "$" : "", |
2296 | r->float_min_offset, r->float_max_offset); |
2297 | if (r->check_substr) |
2298 | PerlIO_printf(Perl_debug_log, |
2299 | r->check_substr == r->float_substr |
2300 | ? "(checking floating" : "(checking anchored"); |
2301 | if (r->reganch & ROPT_NOSCAN) |
2302 | PerlIO_printf(Perl_debug_log, " noscan"); |
2303 | if (r->reganch & ROPT_CHECK_ALL) |
2304 | PerlIO_printf(Perl_debug_log, " isall"); |
2305 | if (r->check_substr) |
2306 | PerlIO_printf(Perl_debug_log, ") "); |
2307 | |
46fc3d4c |
2308 | if (r->regstclass) { |
2309 | regprop(sv, r->regstclass); |
2310 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv)); |
2311 | } |
774d564b |
2312 | if (r->reganch & ROPT_ANCH) { |
2313 | PerlIO_printf(Perl_debug_log, "anchored"); |
2314 | if (r->reganch & ROPT_ANCH_BOL) |
2315 | PerlIO_printf(Perl_debug_log, "(BOL)"); |
c277df42 |
2316 | if (r->reganch & ROPT_ANCH_MBOL) |
2317 | PerlIO_printf(Perl_debug_log, "(MBOL)"); |
774d564b |
2318 | if (r->reganch & ROPT_ANCH_GPOS) |
2319 | PerlIO_printf(Perl_debug_log, "(GPOS)"); |
2320 | PerlIO_putc(Perl_debug_log, ' '); |
2321 | } |
c277df42 |
2322 | if (r->reganch & ROPT_GPOS_SEEN) |
2323 | PerlIO_printf(Perl_debug_log, "GPOS "); |
a0d0e21e |
2324 | if (r->reganch & ROPT_SKIP) |
760ac839 |
2325 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e |
2326 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 |
2327 | PerlIO_printf(Perl_debug_log, "implicit "); |
760ac839 |
2328 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
ce862d02 |
2329 | if (r->reganch & ROPT_EVAL_SEEN) |
2330 | PerlIO_printf(Perl_debug_log, "with eval "); |
760ac839 |
2331 | PerlIO_printf(Perl_debug_log, "\n"); |
17c3b450 |
2332 | #endif /* DEBUGGING */ |
a687059c |
2333 | } |
2334 | |
2335 | /* |
a0d0e21e |
2336 | - regprop - printable representation of opcode |
2337 | */ |
46fc3d4c |
2338 | void |
c277df42 |
2339 | regprop(SV *sv, regnode *o) |
a687059c |
2340 | { |
35ff7856 |
2341 | #ifdef DEBUGGING |
a0d0e21e |
2342 | register char *p = 0; |
2343 | |
46fc3d4c |
2344 | sv_setpv(sv, ":"); |
11343788 |
2345 | switch (OP(o)) { |
a0d0e21e |
2346 | case BOL: |
2347 | p = "BOL"; |
2348 | break; |
2349 | case MBOL: |
2350 | p = "MBOL"; |
2351 | break; |
2352 | case SBOL: |
2353 | p = "SBOL"; |
2354 | break; |
2355 | case EOL: |
2356 | p = "EOL"; |
2357 | break; |
b85d18e9 |
2358 | case EOS: |
2359 | p = "EOS"; |
2360 | break; |
a0d0e21e |
2361 | case MEOL: |
2362 | p = "MEOL"; |
2363 | break; |
2364 | case SEOL: |
2365 | p = "SEOL"; |
2366 | break; |
2367 | case ANY: |
2368 | p = "ANY"; |
2369 | break; |
2370 | case SANY: |
2371 | p = "SANY"; |
2372 | break; |
2373 | case ANYOF: |
2374 | p = "ANYOF"; |
2375 | break; |
2376 | case BRANCH: |
2377 | p = "BRANCH"; |
2378 | break; |
bbce6d69 |
2379 | case EXACT: |
c277df42 |
2380 | sv_catpvf(sv, "EXACT <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]); |
bbce6d69 |
2381 | break; |
2382 | case EXACTF: |
c277df42 |
2383 | sv_catpvf(sv, "EXACTF <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]); |
bbce6d69 |
2384 | break; |
2385 | case EXACTFL: |
c277df42 |
2386 | sv_catpvf(sv, "EXACTFL <%s%s%s>", colors[0], OPERAND(o) + 1, colors[1]); |
a0d0e21e |
2387 | break; |
2388 | case NOTHING: |
2389 | p = "NOTHING"; |
2390 | break; |
c277df42 |
2391 | case TAIL: |
2392 | p = "TAIL"; |
2393 | break; |
a0d0e21e |
2394 | case BACK: |
2395 | p = "BACK"; |
2396 | break; |
2397 | case END: |
2398 | p = "END"; |
2399 | break; |
a0d0e21e |
2400 | case BOUND: |
2401 | p = "BOUND"; |
2402 | break; |
bbce6d69 |
2403 | case BOUNDL: |
2404 | p = "BOUNDL"; |
2405 | break; |
a0d0e21e |
2406 | case NBOUND: |
2407 | p = "NBOUND"; |
2408 | break; |
bbce6d69 |
2409 | case NBOUNDL: |
2410 | p = "NBOUNDL"; |
a0d0e21e |
2411 | break; |
2412 | case CURLY: |
5dc0d613 |
2413 | sv_catpvf(sv, "CURLY {%d,%d}", ARG1(o), ARG2(o)); |
a0d0e21e |
2414 | break; |
c277df42 |
2415 | case CURLYM: |
c277df42 |
2416 | sv_catpvf(sv, "CURLYM[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o)); |
c277df42 |
2417 | break; |
2418 | case CURLYN: |
c277df42 |
2419 | sv_catpvf(sv, "CURLYN[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o)); |
c277df42 |
2420 | break; |
a0d0e21e |
2421 | case CURLYX: |
5dc0d613 |
2422 | sv_catpvf(sv, "CURLYX {%d,%d}", ARG1(o), ARG2(o)); |
a0d0e21e |
2423 | break; |
2424 | case REF: |
c277df42 |
2425 | sv_catpvf(sv, "REF%d", ARG(o)); |
a0d0e21e |
2426 | break; |
c8756f30 |
2427 | case REFF: |
c277df42 |
2428 | sv_catpvf(sv, "REFF%d", ARG(o)); |
c8756f30 |
2429 | break; |
2430 | case REFFL: |
c277df42 |
2431 | sv_catpvf(sv, "REFFL%d", ARG(o)); |
c8756f30 |
2432 | break; |
a0d0e21e |
2433 | case OPEN: |
c277df42 |
2434 | sv_catpvf(sv, "OPEN%d", ARG(o)); |
a0d0e21e |
2435 | break; |
2436 | case CLOSE: |
c277df42 |
2437 | sv_catpvf(sv, "CLOSE%d", ARG(o)); |
a0d0e21e |
2438 | p = NULL; |
2439 | break; |
2440 | case STAR: |
2441 | p = "STAR"; |
2442 | break; |
2443 | case PLUS: |
2444 | p = "PLUS"; |
2445 | break; |
2446 | case MINMOD: |
2447 | p = "MINMOD"; |
2448 | break; |
774d564b |
2449 | case GPOS: |
2450 | p = "GPOS"; |
a0d0e21e |
2451 | break; |
2452 | case UNLESSM: |
c277df42 |
2453 | sv_catpvf(sv, "UNLESSM[-%d]", o->flags); |
a0d0e21e |
2454 | break; |
2455 | case IFMATCH: |
c277df42 |
2456 | sv_catpvf(sv, "IFMATCH[-%d]", o->flags); |
a0d0e21e |
2457 | break; |
2458 | case SUCCEED: |
2459 | p = "SUCCEED"; |
2460 | break; |
2461 | case WHILEM: |
2462 | p = "WHILEM"; |
2463 | break; |
bbce6d69 |
2464 | case DIGIT: |
2465 | p = "DIGIT"; |
2466 | break; |
2467 | case NDIGIT: |
2468 | p = "NDIGIT"; |
2469 | break; |
2470 | case ALNUM: |
2471 | p = "ALNUM"; |
2472 | break; |
2473 | case NALNUM: |
2474 | p = "NALNUM"; |
2475 | break; |
2476 | case SPACE: |
2477 | p = "SPACE"; |
2478 | break; |
2479 | case NSPACE: |
2480 | p = "NSPACE"; |
2481 | break; |
2482 | case ALNUML: |
2483 | p = "ALNUML"; |
2484 | break; |
2485 | case NALNUML: |
2486 | p = "NALNUML"; |
2487 | break; |
2488 | case SPACEL: |
2489 | p = "SPACEL"; |
2490 | break; |
2491 | case NSPACEL: |
2492 | p = "NSPACEL"; |
2493 | break; |
c277df42 |
2494 | case EVAL: |
2495 | p = "EVAL"; |
2496 | break; |
2497 | case LONGJMP: |
2498 | p = "LONGJMP"; |
2499 | break; |
2500 | case BRANCHJ: |
2501 | p = "BRANCHJ"; |
2502 | break; |
2503 | case IFTHEN: |
2504 | p = "IFTHEN"; |
2505 | break; |
2506 | case GROUPP: |
2507 | sv_catpvf(sv, "GROUPP%d", ARG(o)); |
2508 | break; |
2509 | case LOGICAL: |
2510 | p = "LOGICAL"; |
2511 | break; |
2512 | case SUSPEND: |
2513 | p = "SUSPEND"; |
2514 | break; |
2515 | case RENUM: |
2516 | p = "RENUM"; |
2517 | break; |
2518 | case OPTIMIZED: |
2519 | p = "OPTIMIZED"; |
2520 | break; |
a0d0e21e |
2521 | default: |
2522 | FAIL("corrupted regexp opcode"); |
2523 | } |
46fc3d4c |
2524 | if (p) |
2525 | sv_catpv(sv, p); |
17c3b450 |
2526 | #endif /* DEBUGGING */ |
35ff7856 |
2527 | } |
a687059c |
2528 | |
2b69d0c2 |
2529 | void |
8ac85365 |
2530 | pregfree(struct regexp *r) |
a687059c |
2531 | { |
c277df42 |
2532 | if (!r || (--r->refcnt > 0)) |
a0d0e21e |
2533 | return; |
c277df42 |
2534 | if (r->precomp) |
a0d0e21e |
2535 | Safefree(r->precomp); |
c277df42 |
2536 | if (r->subbase) |
a0d0e21e |
2537 | Safefree(r->subbase); |
a193d654 |
2538 | if (r->substrs) { |
2539 | if (r->anchored_substr) |
2540 | SvREFCNT_dec(r->anchored_substr); |
2541 | if (r->float_substr) |
2542 | SvREFCNT_dec(r->float_substr); |
2779dcf1 |
2543 | Safefree(r->substrs); |
a193d654 |
2544 | } |
c277df42 |
2545 | if (r->data) { |
2546 | int n = r->data->count; |
2547 | while (--n >= 0) { |
2548 | switch (r->data->what[n]) { |
2549 | case 's': |
2550 | SvREFCNT_dec((SV*)r->data->data[n]); |
2551 | break; |
2552 | case 'o': |
2553 | op_free((OP_4tree*)r->data->data[n]); |
2554 | break; |
2555 | case 'n': |
2556 | break; |
2557 | default: |
2558 | FAIL2("panic: regfree data code '%c'", r->data->what[n]); |
2559 | } |
2560 | } |
2561 | Safefree(r->data->what); |
2562 | Safefree(r->data); |
a0d0e21e |
2563 | } |
2564 | Safefree(r->startp); |
2565 | Safefree(r->endp); |
2566 | Safefree(r); |
a687059c |
2567 | } |
c277df42 |
2568 | |
2569 | /* |
2570 | - regnext - dig the "next" pointer out of a node |
2571 | * |
2572 | * [Note, when REGALIGN is defined there are two places in regmatch() |
2573 | * that bypass this code for speed.] |
2574 | */ |
2575 | regnode * |
2576 | regnext(register regnode *p) |
2577 | { |
2578 | register I32 offset; |
2579 | |
2580 | if (p == ®dummy) |
2581 | return(NULL); |
2582 | |
2583 | offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p)); |
2584 | if (offset == 0) |
2585 | return(NULL); |
2586 | |
c277df42 |
2587 | return(p+offset); |
c277df42 |
2588 | } |
2589 | |
01f988be |
2590 | STATIC void |
c277df42 |
2591 | re_croak2(const char* pat1,const char* pat2,...) |
c277df42 |
2592 | { |
2593 | va_list args; |
2594 | STRLEN l1 = strlen(pat1); |
2595 | STRLEN l2 = strlen(pat2); |
2596 | char buf[512]; |
2597 | char *message; |
2598 | |
2599 | if (l1 > 510) |
2600 | l1 = 510; |
2601 | if (l1 + l2 > 510) |
2602 | l2 = 510 - l1; |
2603 | Copy(pat1, buf, l1 , char); |
2604 | Copy(pat2, buf + l1, l2 , char); |
3b818b81 |
2605 | buf[l1 + l2] = '\n'; |
2606 | buf[l1 + l2 + 1] = '\0'; |
c277df42 |
2607 | va_start(args, pat2); |
c277df42 |
2608 | message = mess(buf, &args); |
2609 | va_end(args); |
2610 | l1 = strlen(message); |
2611 | if (l1 > 512) |
2612 | l1 = 512; |
2613 | Copy(message, buf, l1 , char); |
2614 | buf[l1] = '\0'; /* Overwrite \n */ |
2615 | croak("%s", buf); |
2616 | } |