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 | |
f0fcb552 |
22 | /*SUPPRESS 112*/ |
a687059c |
23 | /* |
e50aee73 |
24 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c |
25 | * |
26 | * Copyright (c) 1986 by University of Toronto. |
27 | * Written by Henry Spencer. Not derived from licensed software. |
28 | * |
29 | * Permission is granted to anyone to use this software for any |
30 | * purpose on any computer system, and to redistribute it freely, |
31 | * subject to the following restrictions: |
32 | * |
33 | * 1. The author is not responsible for the consequences of use of |
34 | * this software, no matter how awful, even if they arise |
35 | * from defects in it. |
36 | * |
37 | * 2. The origin of this software must not be misrepresented, either |
38 | * by explicit claim or by omission. |
39 | * |
40 | * 3. Altered versions must be plainly marked as such, and must not |
41 | * be misrepresented as being the original software. |
42 | * |
43 | * |
44 | **** Alterations to Henry's code are... |
45 | **** |
a0d0e21e |
46 | **** Copyright (c) 1991-1994, Larry Wall |
a687059c |
47 | **** |
9ef589d8 |
48 | **** You may distribute under the terms of either the GNU General Public |
49 | **** License or the Artistic License, as specified in the README file. |
50 | |
a687059c |
51 | * |
52 | * Beware that some of this code is subtly aware of the way operator |
53 | * precedence is structured in regular expressions. Serious changes in |
54 | * regular-expression syntax might require a total rethink. |
55 | */ |
56 | #include "EXTERN.h" |
57 | #include "perl.h" |
58 | #include "INTERN.h" |
59 | #include "regcomp.h" |
60 | |
fe14fcc3 |
61 | #ifdef MSDOS |
62 | # if defined(BUGGY_MSC6) |
63 | /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */ |
64 | # pragma optimize("a",off) |
65 | /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/ |
66 | # pragma optimize("w",on ) |
67 | # endif /* BUGGY_MSC6 */ |
68 | #endif /* MSDOS */ |
69 | |
a687059c |
70 | #ifndef STATIC |
71 | #define STATIC static |
72 | #endif |
73 | |
74 | #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?') |
75 | #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \ |
76 | ((*s) == '{' && regcurly(s))) |
2b69d0c2 |
77 | #ifdef atarist |
78 | #define PERL_META "^$.[()|?+*\\" |
79 | #else |
a687059c |
80 | #define META "^$.[()|?+*\\" |
2b69d0c2 |
81 | #endif |
a687059c |
82 | |
35c8bce7 |
83 | #ifdef SPSTART |
84 | #undef SPSTART /* dratted cpp namespace... */ |
85 | #endif |
a687059c |
86 | /* |
87 | * Flags to be passed up and down. |
88 | */ |
a687059c |
89 | #define WORST 0 /* Worst case. */ |
a0d0e21e |
90 | #define HASWIDTH 0x1 /* Known never to match null string. */ |
91 | #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ |
92 | #define SPSTART 0x4 /* Starts with * or +. */ |
93 | #define TRYAGAIN 0x8 /* Weeded out a declaration. */ |
a687059c |
94 | |
95 | /* |
e50aee73 |
96 | * Forward declarations for pregcomp()'s friends. |
a687059c |
97 | */ |
a0d0e21e |
98 | |
99 | static char *reg _((I32, I32 *)); |
100 | static char *reganode _((char, unsigned short)); |
101 | static char *regatom _((I32 *)); |
102 | static char *regbranch _((I32 *)); |
103 | static void regc _((char)); |
104 | static char *regclass _((void)); |
105 | STATIC I32 regcurly _((char *)); |
106 | static char *regnode _((char)); |
107 | static char *regpiece _((I32 *)); |
108 | static void reginsert _((char, char *)); |
109 | static void regoptail _((char *, char *)); |
bbce6d69 |
110 | static void regset _((char *, I32)); |
a0d0e21e |
111 | static void regtail _((char *, char *)); |
112 | static char* nextchar _((void)); |
a687059c |
113 | |
114 | /* |
e50aee73 |
115 | - pregcomp - compile a regular expression into internal code |
a687059c |
116 | * |
117 | * We can't allocate space until we know how big the compiled form will be, |
118 | * but we can't compile it (and thus know how big it is) until we've got a |
119 | * place to put the code. So we cheat: we compile it twice, once with code |
120 | * generation turned off and size counting turned on, and once "for real". |
121 | * This also means that we don't allocate space until we are sure that the |
122 | * thing really will compile successfully, and we never have to move the |
123 | * code and thus invalidate pointers into it. (Note that it has to be in |
124 | * one piece because free() must be able to free it all.) [NB: not true in perl] |
125 | * |
126 | * Beware that the optimization-preparation code in here knows about some |
127 | * of the structure of the compiled regexp. [I'll say.] |
128 | */ |
129 | regexp * |
e50aee73 |
130 | pregcomp(exp,xend,pm) |
a0d0e21e |
131 | char* exp; |
132 | char* xend; |
133 | PMOP* pm; |
a687059c |
134 | { |
a0d0e21e |
135 | register regexp *r; |
136 | register char *scan; |
137 | register SV *longish; |
138 | SV *longest; |
139 | register I32 len; |
140 | register char *first; |
141 | I32 flags; |
142 | I32 backish; |
143 | I32 backest; |
144 | I32 curback; |
145 | I32 minlen = 0; |
146 | I32 sawplus = 0; |
147 | I32 sawopen = 0; |
148 | |
149 | if (exp == NULL) |
150 | croak("NULL regexp argument"); |
151 | |
bbce6d69 |
152 | regprecomp = savepvn(exp, xend - exp); |
a0d0e21e |
153 | regflags = pm->op_pmflags; |
bbce6d69 |
154 | regsawback = 0; |
155 | |
156 | /* First pass: determine size, legality. */ |
a0d0e21e |
157 | regparse = exp; |
158 | regxend = xend; |
a0d0e21e |
159 | regnaughty = 0; |
a0d0e21e |
160 | regnpar = 1; |
161 | regsize = 0L; |
162 | regcode = ®dummy; |
163 | regc((char)MAGIC); |
164 | if (reg(0, &flags) == NULL) { |
165 | Safefree(regprecomp); |
166 | regprecomp = Nullch; |
167 | return(NULL); |
168 | } |
169 | |
170 | /* Small enough for pointer-storage convention? */ |
171 | if (regsize >= 32767L) /* Probably could be 65535L. */ |
172 | FAIL("regexp too big"); |
173 | |
bbce6d69 |
174 | /* Allocate space and initialize. */ |
a0d0e21e |
175 | Newc(1001, r, sizeof(regexp) + (unsigned)regsize, char, regexp); |
176 | if (r == NULL) |
177 | FAIL("regexp out of space"); |
bbce6d69 |
178 | r->prelen = xend - exp; |
a0d0e21e |
179 | r->precomp = regprecomp; |
180 | r->subbeg = r->subbase = NULL; |
bbce6d69 |
181 | |
182 | /* Second pass: emit code. */ |
a0d0e21e |
183 | regparse = exp; |
bbce6d69 |
184 | regxend = xend; |
185 | regnaughty = 0; |
a0d0e21e |
186 | regnpar = 1; |
187 | regcode = r->program; |
188 | regc((char)MAGIC); |
189 | if (reg(0, &flags) == NULL) |
190 | return(NULL); |
191 | |
192 | /* Dig out information for optimizations. */ |
193 | pm->op_pmflags = regflags; |
a0d0e21e |
194 | r->regstart = Nullsv; /* Worst-case defaults. */ |
195 | r->reganch = 0; |
196 | r->regmust = Nullsv; |
197 | r->regback = -1; |
198 | r->regstclass = Nullch; |
199 | r->naughty = regnaughty >= 10; /* Probably an expensive pattern. */ |
200 | scan = r->program+1; /* First BRANCH. */ |
201 | if (OP(regnext(scan)) == END) {/* Only one top-level choice. */ |
202 | scan = NEXTOPER(scan); |
203 | |
204 | first = scan; |
205 | while ((OP(first) == OPEN && (sawopen = 1)) || |
206 | (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) || |
207 | (OP(first) == PLUS) || |
208 | (OP(first) == MINMOD) || |
209 | (regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) { |
210 | if (OP(first) == PLUS) |
211 | sawplus = 1; |
212 | else |
213 | first += regarglen[(U8)OP(first)]; |
214 | first = NEXTOPER(first); |
a687059c |
215 | } |
216 | |
a0d0e21e |
217 | /* Starting-point info. */ |
218 | again: |
bbce6d69 |
219 | if (OP(first) == EXACT) { |
a0d0e21e |
220 | r->regstart = newSVpv(OPERAND(first)+1,*OPERAND(first)); |
bbce6d69 |
221 | if (SvCUR(r->regstart) > !sawstudy) |
222 | fbm_compile(r->regstart); |
223 | (void)SvUPGRADE(r->regstart, SVt_PVBM); |
a0d0e21e |
224 | } |
225 | else if (strchr(simple+2,OP(first))) |
226 | r->regstclass = first; |
bbce6d69 |
227 | else if (regkind[(U8)OP(first)] == BOUND || |
228 | regkind[(U8)OP(first)] == NBOUND) |
a0d0e21e |
229 | r->regstclass = first; |
230 | else if (regkind[(U8)OP(first)] == BOL) { |
231 | r->reganch = ROPT_ANCH; |
232 | first = NEXTOPER(first); |
233 | goto again; |
234 | } |
235 | else if ((OP(first) == STAR && |
236 | regkind[(U8)OP(NEXTOPER(first))] == ANY) && |
237 | !(r->reganch & ROPT_ANCH) ) |
238 | { |
239 | /* turn .* into ^.* with an implied $*=1 */ |
240 | r->reganch = ROPT_ANCH | ROPT_IMPLICIT; |
241 | first = NEXTOPER(first); |
242 | goto again; |
243 | } |
244 | if (sawplus && (!sawopen || !regsawback)) |
245 | r->reganch |= ROPT_SKIP; /* x+ must match 1st of run */ |
246 | |
760ac839 |
247 | DEBUG_r(PerlIO_printf(Perl_debug_log, "first %d next %d offset %d\n", |
a0d0e21e |
248 | OP(first), OP(NEXTOPER(first)), first - scan)); |
249 | /* |
250 | * If there's something expensive in the r.e., find the |
251 | * longest literal string that must appear and make it the |
252 | * regmust. Resolve ties in favor of later strings, since |
253 | * the regstart check works with the beginning of the r.e. |
254 | * and avoiding duplication strengthens checking. Not a |
255 | * strong reason, but sufficient in the absence of others. |
256 | * [Now we resolve ties in favor of the earlier string if |
257 | * it happens that curback has been invalidated, since the |
258 | * earlier string may buy us something the later one won't.] |
259 | */ |
260 | longish = newSVpv("",0); |
261 | longest = newSVpv("",0); |
262 | len = 0; |
263 | minlen = 0; |
264 | curback = 0; |
265 | backish = 0; |
266 | backest = 0; |
267 | while (OP(scan) != END) { |
268 | if (OP(scan) == BRANCH) { |
269 | if (OP(regnext(scan)) == BRANCH) { |
270 | curback = -30000; |
271 | while (OP(scan) == BRANCH) |
272 | scan = regnext(scan); |
00bf170e |
273 | } |
a0d0e21e |
274 | else /* single branch is ok */ |
275 | scan = NEXTOPER(scan); |
748a9306 |
276 | continue; |
a0d0e21e |
277 | } |
278 | if (OP(scan) == UNLESSM) { |
279 | curback = -30000; |
280 | scan = regnext(scan); |
748a9306 |
281 | continue; |
a0d0e21e |
282 | } |
bbce6d69 |
283 | if (OP(scan) == EXACT) { |
a0d0e21e |
284 | char *t; |
a687059c |
285 | |
a0d0e21e |
286 | first = scan; |
287 | while (OP(t = regnext(scan)) == CLOSE) |
288 | scan = t; |
289 | minlen += *OPERAND(first); |
290 | if (curback - backish == len) { |
291 | sv_catpvn(longish, OPERAND(first)+1, |
292 | *OPERAND(first)); |
293 | len += *OPERAND(first); |
294 | curback += *OPERAND(first); |
295 | first = regnext(scan); |
2b69d0c2 |
296 | } |
a0d0e21e |
297 | else if (*OPERAND(first) >= len + (curback >= 0)) { |
298 | len = *OPERAND(first); |
299 | sv_setpvn(longish, OPERAND(first)+1,len); |
300 | backish = curback; |
301 | curback += len; |
302 | first = regnext(scan); |
9ef589d8 |
303 | } |
a0d0e21e |
304 | else |
305 | curback += *OPERAND(first); |
306 | } |
307 | else if (strchr(varies,OP(scan))) { |
308 | curback = -30000; |
a687059c |
309 | len = 0; |
a0d0e21e |
310 | if (SvCUR(longish) > SvCUR(longest)) { |
79072805 |
311 | sv_setsv(longest,longish); |
34de22dd |
312 | backest = backish; |
313 | } |
a0d0e21e |
314 | sv_setpvn(longish,"",0); |
315 | if (OP(scan) == PLUS && strchr(simple,OP(NEXTOPER(scan)))) |
316 | minlen++; |
317 | else if (regkind[(U8)OP(scan)] == CURLY && |
318 | strchr(simple,OP(NEXTOPER(scan)+4))) |
319 | minlen += ARG1(scan); |
320 | } |
321 | else if (strchr(simple,OP(scan))) { |
322 | curback++; |
323 | minlen++; |
324 | len = 0; |
325 | if (SvCUR(longish) > SvCUR(longest)) { |
326 | sv_setsv(longest,longish); |
327 | backest = backish; |
fe14fcc3 |
328 | } |
a0d0e21e |
329 | sv_setpvn(longish,"",0); |
330 | } |
331 | scan = regnext(scan); |
a687059c |
332 | } |
333 | |
a0d0e21e |
334 | /* Prefer earlier on tie, unless we can tail match latter */ |
335 | |
bbce6d69 |
336 | if (SvCUR(longish) + (regkind[(U8)OP(first)] == EOL) |
337 | > SvCUR(longest)) |
a0d0e21e |
338 | { |
339 | sv_setsv(longest,longish); |
340 | backest = backish; |
341 | } |
342 | else |
343 | sv_setpvn(longish,"",0); |
344 | if (SvCUR(longest) |
bbce6d69 |
345 | && (!r->regstart |
346 | || !fbm_instr((unsigned char*) SvPVX(r->regstart), |
347 | (unsigned char *) (SvPVX(r->regstart) |
348 | + SvCUR(r->regstart)), |
349 | longest))) |
a0d0e21e |
350 | { |
351 | r->regmust = longest; |
352 | if (backest < 0) |
353 | backest = -1; |
354 | r->regback = backest; |
bbce6d69 |
355 | if (SvCUR(longest) > !(sawstudy || regkind[(U8)OP(first)] == EOL)) |
356 | fbm_compile(r->regmust); |
a0d0e21e |
357 | (void)SvUPGRADE(r->regmust, SVt_PVBM); |
358 | BmUSEFUL(r->regmust) = 100; |
359 | if (regkind[(U8)OP(first)] == EOL && SvCUR(longish)) |
360 | SvTAIL_on(r->regmust); |
361 | } |
362 | else { |
363 | SvREFCNT_dec(longest); |
364 | longest = Nullsv; |
365 | } |
366 | SvREFCNT_dec(longish); |
367 | } |
368 | |
a0d0e21e |
369 | r->nparens = regnpar - 1; |
370 | r->minlen = minlen; |
371 | Newz(1002, r->startp, regnpar, char*); |
372 | Newz(1002, r->endp, regnpar, char*); |
373 | DEBUG_r(regdump(r)); |
374 | return(r); |
a687059c |
375 | } |
376 | |
377 | /* |
378 | - reg - regular expression, i.e. main body or parenthesized thing |
379 | * |
380 | * Caller must absorb opening parenthesis. |
381 | * |
382 | * Combining parenthesis handling with the base level of regular expression |
383 | * is a trifle forced, but the need to tie the tails of the branches to what |
384 | * follows makes it hard to avoid. |
385 | */ |
386 | static char * |
387 | reg(paren, flagp) |
79072805 |
388 | I32 paren; /* Parenthesized? */ |
389 | I32 *flagp; |
a687059c |
390 | { |
a0d0e21e |
391 | register char *ret; |
392 | register char *br; |
393 | register char *ender = 0; |
394 | register I32 parno = 0; |
395 | I32 flags; |
396 | |
397 | *flagp = HASWIDTH; /* Tentatively. */ |
398 | |
399 | /* Make an OPEN node, if parenthesized. */ |
400 | if (paren) { |
401 | if (*regparse == '?') { |
402 | regparse++; |
748a9306 |
403 | paren = *regparse++; |
a0d0e21e |
404 | ret = NULL; |
405 | switch (paren) { |
406 | case ':': |
407 | case '=': |
408 | case '!': |
409 | break; |
410 | case '$': |
411 | case '@': |
412 | croak("Sequence (?%c...) not implemented", paren); |
413 | break; |
414 | case '#': |
415 | while (*regparse && *regparse != ')') |
416 | regparse++; |
417 | if (*regparse != ')') |
748a9306 |
418 | croak("Sequence (?#... not terminated"); |
a0d0e21e |
419 | nextchar(); |
420 | *flagp = TRYAGAIN; |
421 | return NULL; |
422 | default: |
423 | --regparse; |
424 | while (*regparse && strchr("iogmsx", *regparse)) |
425 | pmflag(®flags, *regparse++); |
426 | if (*regparse != ')') |
427 | croak("Sequence (?%c...) not recognized", *regparse); |
428 | nextchar(); |
429 | *flagp = TRYAGAIN; |
430 | return NULL; |
431 | } |
432 | } |
433 | else { |
434 | parno = regnpar; |
435 | regnpar++; |
436 | ret = reganode(OPEN, parno); |
437 | } |
438 | } else |
439 | ret = NULL; |
440 | |
441 | /* Pick up the branches, linking them together. */ |
442 | br = regbranch(&flags); |
443 | if (br == NULL) |
444 | return(NULL); |
445 | if (ret != NULL) |
446 | regtail(ret, br); /* OPEN -> first. */ |
447 | else |
448 | ret = br; |
449 | if (!(flags&HASWIDTH)) |
450 | *flagp &= ~HASWIDTH; |
451 | *flagp |= flags&SPSTART; |
452 | while (*regparse == '|') { |
453 | nextchar(); |
a687059c |
454 | br = regbranch(&flags); |
455 | if (br == NULL) |
a0d0e21e |
456 | return(NULL); |
457 | regtail(ret, br); /* BRANCH -> BRANCH. */ |
a687059c |
458 | if (!(flags&HASWIDTH)) |
a0d0e21e |
459 | *flagp &= ~HASWIDTH; |
a687059c |
460 | *flagp |= flags&SPSTART; |
a0d0e21e |
461 | } |
462 | |
463 | /* Make a closing node, and hook it on the end. */ |
464 | switch (paren) { |
465 | case ':': |
466 | ender = regnode(NOTHING); |
467 | break; |
468 | case 1: |
469 | ender = reganode(CLOSE, parno); |
470 | break; |
471 | case '=': |
472 | case '!': |
473 | ender = regnode(SUCCEED); |
474 | *flagp &= ~HASWIDTH; |
475 | break; |
476 | case 0: |
477 | ender = regnode(END); |
478 | break; |
479 | } |
480 | regtail(ret, ender); |
481 | |
482 | /* Hook the tails of the branches to the closing node. */ |
483 | for (br = ret; br != NULL; br = regnext(br)) |
484 | regoptail(br, ender); |
485 | |
486 | if (paren == '=') { |
487 | reginsert(IFMATCH,ret); |
488 | regtail(ret, regnode(NOTHING)); |
489 | } |
490 | else if (paren == '!') { |
491 | reginsert(UNLESSM,ret); |
492 | regtail(ret, regnode(NOTHING)); |
493 | } |
494 | |
495 | /* Check for proper termination. */ |
4633a7c4 |
496 | if (paren && (regparse >= regxend || *nextchar() != ')')) { |
a0d0e21e |
497 | FAIL("unmatched () in regexp"); |
498 | } else if (!paren && regparse < regxend) { |
499 | if (*regparse == ')') { |
500 | FAIL("unmatched () in regexp"); |
501 | } else |
502 | FAIL("junk on end of regexp"); /* "Can't happen". */ |
503 | /* NOTREACHED */ |
504 | } |
a687059c |
505 | |
a0d0e21e |
506 | return(ret); |
a687059c |
507 | } |
508 | |
509 | /* |
510 | - regbranch - one alternative of an | operator |
511 | * |
512 | * Implements the concatenation operator. |
513 | */ |
514 | static char * |
515 | regbranch(flagp) |
79072805 |
516 | I32 *flagp; |
a687059c |
517 | { |
a0d0e21e |
518 | register char *ret; |
519 | register char *chain; |
520 | register char *latest; |
521 | I32 flags = 0; |
522 | |
523 | *flagp = WORST; /* Tentatively. */ |
524 | |
525 | ret = regnode(BRANCH); |
526 | chain = NULL; |
527 | regparse--; |
528 | nextchar(); |
529 | while (regparse < regxend && *regparse != '|' && *regparse != ')') { |
530 | flags &= ~TRYAGAIN; |
531 | latest = regpiece(&flags); |
532 | if (latest == NULL) { |
533 | if (flags & TRYAGAIN) |
534 | continue; |
535 | return(NULL); |
536 | } |
537 | *flagp |= flags&HASWIDTH; |
538 | if (chain == NULL) /* First piece. */ |
539 | *flagp |= flags&SPSTART; |
540 | else { |
541 | regnaughty++; |
542 | regtail(chain, latest); |
a687059c |
543 | } |
a0d0e21e |
544 | chain = latest; |
545 | } |
546 | if (chain == NULL) /* Loop ran zero times. */ |
547 | (void) regnode(NOTHING); |
a687059c |
548 | |
a0d0e21e |
549 | return(ret); |
a687059c |
550 | } |
551 | |
552 | /* |
553 | - regpiece - something followed by possible [*+?] |
554 | * |
555 | * Note that the branching code sequences used for ? and the general cases |
556 | * of * and + are somewhat optimized: they use the same NOTHING node as |
557 | * both the endmarker for their branch list and the body of the last branch. |
558 | * It might seem that this node could be dispensed with entirely, but the |
559 | * endmarker role is not redundant. |
560 | */ |
561 | static char * |
562 | regpiece(flagp) |
79072805 |
563 | I32 *flagp; |
a687059c |
564 | { |
a0d0e21e |
565 | register char *ret; |
566 | register char op; |
567 | register char *next; |
568 | I32 flags; |
569 | char *origparse = regparse; |
570 | char *maxpos; |
571 | I32 min; |
572 | I32 max = 32767; |
573 | |
574 | ret = regatom(&flags); |
575 | if (ret == NULL) { |
576 | if (flags & TRYAGAIN) |
577 | *flagp |= TRYAGAIN; |
578 | return(NULL); |
579 | } |
580 | |
581 | op = *regparse; |
582 | if (op == '(' && regparse[1] == '?' && regparse[2] == '#') { |
583 | while (op && op != ')') |
584 | op = *++regparse; |
585 | if (op) { |
586 | nextchar(); |
587 | op = *regparse; |
588 | } |
589 | } |
590 | |
591 | if (op == '{' && regcurly(regparse)) { |
592 | next = regparse + 1; |
593 | maxpos = Nullch; |
594 | while (isDIGIT(*next) || *next == ',') { |
595 | if (*next == ',') { |
596 | if (maxpos) |
597 | break; |
598 | else |
599 | maxpos = next; |
a687059c |
600 | } |
a0d0e21e |
601 | next++; |
602 | } |
603 | if (*next == '}') { /* got one */ |
604 | if (!maxpos) |
605 | maxpos = next; |
606 | regparse++; |
607 | min = atoi(regparse); |
608 | if (*maxpos == ',') |
609 | maxpos++; |
610 | else |
611 | maxpos = regparse; |
612 | max = atoi(maxpos); |
613 | if (!max && *maxpos != '0') |
614 | max = 32767; /* meaning "infinity" */ |
615 | regparse = next; |
616 | nextchar(); |
617 | |
618 | do_curly: |
619 | if ((flags&SIMPLE)) { |
620 | regnaughty += 2 + regnaughty / 2; |
621 | reginsert(CURLY, ret); |
622 | } |
623 | else { |
624 | regnaughty += 4 + regnaughty; /* compound interest */ |
625 | regtail(ret, regnode(WHILEM)); |
626 | reginsert(CURLYX,ret); |
627 | regtail(ret, regnode(NOTHING)); |
628 | } |
629 | |
630 | if (min > 0) |
631 | *flagp = (WORST|HASWIDTH); |
632 | if (max && max < min) |
633 | croak("Can't do {n,m} with n > m"); |
634 | if (regcode != ®dummy) { |
00bf170e |
635 | #ifdef REGALIGN |
a0d0e21e |
636 | *(unsigned short *)(ret+3) = min; |
637 | *(unsigned short *)(ret+5) = max; |
00bf170e |
638 | #else |
a0d0e21e |
639 | ret[3] = min >> 8; ret[4] = min & 0377; |
640 | ret[5] = max >> 8; ret[6] = max & 0377; |
00bf170e |
641 | #endif |
a687059c |
642 | } |
a687059c |
643 | |
a0d0e21e |
644 | goto nest_check; |
a687059c |
645 | } |
a0d0e21e |
646 | } |
a687059c |
647 | |
a0d0e21e |
648 | if (!ISMULT1(op)) { |
649 | *flagp = flags; |
a687059c |
650 | return(ret); |
a0d0e21e |
651 | } |
652 | nextchar(); |
653 | |
654 | *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH); |
655 | |
656 | if (op == '*' && (flags&SIMPLE)) { |
657 | reginsert(STAR, ret); |
658 | regnaughty += 4; |
659 | } |
660 | else if (op == '*') { |
661 | min = 0; |
662 | goto do_curly; |
663 | } else if (op == '+' && (flags&SIMPLE)) { |
664 | reginsert(PLUS, ret); |
665 | regnaughty += 3; |
666 | } |
667 | else if (op == '+') { |
668 | min = 1; |
669 | goto do_curly; |
670 | } else if (op == '?') { |
671 | min = 0; max = 1; |
672 | goto do_curly; |
673 | } |
674 | nest_check: |
675 | if (dowarn && regcode != ®dummy && !(flags&HASWIDTH) && max > 10000) { |
676 | warn("%.*s matches null string many times", |
677 | regparse - origparse, origparse); |
678 | } |
679 | |
680 | if (*regparse == '?') { |
681 | nextchar(); |
682 | reginsert(MINMOD, ret); |
683 | #ifdef REGALIGN |
684 | regtail(ret, ret + 4); |
685 | #else |
686 | regtail(ret, ret + 3); |
687 | #endif |
688 | } |
689 | if (ISMULT2(regparse)) |
690 | FAIL("nested *?+ in regexp"); |
691 | |
692 | return(ret); |
a687059c |
693 | } |
694 | |
695 | /* |
696 | - regatom - the lowest level |
697 | * |
698 | * Optimization: gobbles an entire sequence of ordinary characters so that |
699 | * it can turn them into a single node, which is smaller to store and |
700 | * faster to run. Backslashed characters are exceptions, each becoming a |
701 | * separate node; the code is simpler that way and it's not worth fixing. |
702 | * |
703 | * [Yes, it is worth fixing, some scripts can run twice the speed.] |
704 | */ |
705 | static char * |
706 | regatom(flagp) |
79072805 |
707 | I32 *flagp; |
a687059c |
708 | { |
a0d0e21e |
709 | register char *ret = 0; |
710 | I32 flags; |
711 | |
712 | *flagp = WORST; /* Tentatively. */ |
713 | |
714 | tryagain: |
715 | switch (*regparse) { |
716 | case '^': |
717 | nextchar(); |
718 | if (regflags & PMf_MULTILINE) |
719 | ret = regnode(MBOL); |
720 | else if (regflags & PMf_SINGLELINE) |
721 | ret = regnode(SBOL); |
722 | else |
723 | ret = regnode(BOL); |
724 | break; |
725 | case '$': |
726 | nextchar(); |
727 | if (regflags & PMf_MULTILINE) |
728 | ret = regnode(MEOL); |
729 | else if (regflags & PMf_SINGLELINE) |
730 | ret = regnode(SEOL); |
731 | else |
732 | ret = regnode(EOL); |
733 | break; |
734 | case '.': |
735 | nextchar(); |
736 | if (regflags & PMf_SINGLELINE) |
737 | ret = regnode(SANY); |
738 | else |
739 | ret = regnode(ANY); |
740 | regnaughty++; |
741 | *flagp |= HASWIDTH|SIMPLE; |
742 | break; |
743 | case '[': |
744 | regparse++; |
745 | ret = regclass(); |
746 | *flagp |= HASWIDTH|SIMPLE; |
747 | break; |
748 | case '(': |
749 | nextchar(); |
750 | ret = reg(1, &flags); |
751 | if (ret == NULL) { |
752 | if (flags & TRYAGAIN) |
753 | goto tryagain; |
754 | return(NULL); |
755 | } |
756 | *flagp |= flags&(HASWIDTH|SPSTART); |
757 | break; |
758 | case '|': |
759 | case ')': |
760 | if (flags & TRYAGAIN) { |
761 | *flagp |= TRYAGAIN; |
762 | return NULL; |
763 | } |
764 | croak("internal urp in regexp at /%s/", regparse); |
765 | /* Supposed to be caught earlier. */ |
766 | break; |
767 | case '?': |
768 | case '+': |
769 | case '*': |
770 | FAIL("?+* follows nothing in regexp"); |
771 | break; |
772 | case '\\': |
773 | switch (*++regparse) { |
774 | case 'A': |
775 | ret = regnode(SBOL); |
776 | *flagp |= SIMPLE; |
777 | nextchar(); |
778 | break; |
779 | case 'G': |
780 | ret = regnode(GBOL); |
781 | *flagp |= SIMPLE; |
782 | nextchar(); |
783 | break; |
784 | case 'Z': |
785 | ret = regnode(SEOL); |
786 | *flagp |= SIMPLE; |
787 | nextchar(); |
788 | break; |
789 | case 'w': |
bbce6d69 |
790 | ret = regnode((regflags & PMf_LOCALE) ? ALNUML : ALNUM); |
a0d0e21e |
791 | *flagp |= HASWIDTH|SIMPLE; |
792 | nextchar(); |
793 | break; |
794 | case 'W': |
bbce6d69 |
795 | ret = regnode((regflags & PMf_LOCALE) ? NALNUML : NALNUM); |
a0d0e21e |
796 | *flagp |= HASWIDTH|SIMPLE; |
797 | nextchar(); |
798 | break; |
799 | case 'b': |
bbce6d69 |
800 | ret = regnode((regflags & PMf_LOCALE) ? BOUNDL : BOUND); |
a0d0e21e |
801 | *flagp |= SIMPLE; |
802 | nextchar(); |
803 | break; |
804 | case 'B': |
bbce6d69 |
805 | ret = regnode((regflags & PMf_LOCALE) ? NBOUNDL : NBOUND); |
a0d0e21e |
806 | *flagp |= SIMPLE; |
807 | nextchar(); |
808 | break; |
809 | case 's': |
bbce6d69 |
810 | ret = regnode((regflags & PMf_LOCALE) ? SPACEL : SPACE); |
a0d0e21e |
811 | *flagp |= HASWIDTH|SIMPLE; |
812 | nextchar(); |
813 | break; |
814 | case 'S': |
bbce6d69 |
815 | ret = regnode((regflags & PMf_LOCALE) ? NSPACEL : NSPACE); |
a0d0e21e |
816 | *flagp |= HASWIDTH|SIMPLE; |
817 | nextchar(); |
818 | break; |
819 | case 'd': |
820 | ret = regnode(DIGIT); |
821 | *flagp |= HASWIDTH|SIMPLE; |
822 | nextchar(); |
823 | break; |
824 | case 'D': |
825 | ret = regnode(NDIGIT); |
826 | *flagp |= HASWIDTH|SIMPLE; |
827 | nextchar(); |
828 | break; |
829 | case 'n': |
830 | case 'r': |
831 | case 't': |
832 | case 'f': |
833 | case 'e': |
834 | case 'a': |
835 | case 'x': |
836 | case 'c': |
837 | case '0': |
838 | goto defchar; |
839 | case '1': case '2': case '3': case '4': |
840 | case '5': case '6': case '7': case '8': case '9': |
841 | { |
842 | I32 num = atoi(regparse); |
843 | |
844 | if (num > 9 && num >= regnpar) |
845 | goto defchar; |
846 | else { |
847 | regsawback = 1; |
848 | ret = reganode(REF, num); |
849 | *flagp |= HASWIDTH; |
850 | while (isDIGIT(*regparse)) |
a687059c |
851 | regparse++; |
a0d0e21e |
852 | regparse--; |
853 | nextchar(); |
854 | } |
855 | } |
856 | break; |
857 | case '\0': |
858 | if (regparse >= regxend) |
859 | FAIL("trailing \\ in regexp"); |
860 | /* FALL THROUGH */ |
861 | default: |
862 | goto defchar; |
863 | } |
864 | break; |
4633a7c4 |
865 | |
866 | case '#': |
867 | if (regflags & PMf_EXTENDED) { |
868 | while (regparse < regxend && *regparse != '\n') regparse++; |
869 | if (regparse < regxend) |
870 | goto tryagain; |
871 | } |
872 | /* FALL THROUGH */ |
873 | |
a0d0e21e |
874 | default: { |
875 | register I32 len; |
876 | register char ender; |
877 | register char *p; |
878 | char *oldp; |
879 | I32 numlen; |
880 | |
881 | regparse++; |
882 | |
883 | defchar: |
bbce6d69 |
884 | ret = regnode((regflags & PMf_FOLD) |
885 | ? ((regflags & PMf_LOCALE) ? EXACTFL : EXACTF) |
886 | : EXACT); |
a0d0e21e |
887 | regc(0); /* save spot for len */ |
888 | for (len = 0, p = regparse - 1; |
889 | len < 127 && p < regxend; |
890 | len++) |
891 | { |
892 | oldp = p; |
893 | switch (*p) { |
894 | case '^': |
895 | case '$': |
896 | case '.': |
897 | case '[': |
898 | case '(': |
899 | case ')': |
900 | case '|': |
901 | goto loopdone; |
902 | case '\\': |
903 | switch (*++p) { |
904 | case 'A': |
905 | case 'G': |
906 | case 'Z': |
907 | case 'w': |
908 | case 'W': |
909 | case 'b': |
910 | case 'B': |
911 | case 's': |
912 | case 'S': |
913 | case 'd': |
914 | case 'D': |
915 | --p; |
916 | goto loopdone; |
917 | case 'n': |
918 | ender = '\n'; |
919 | p++; |
a687059c |
920 | break; |
a0d0e21e |
921 | case 'r': |
922 | ender = '\r'; |
923 | p++; |
a687059c |
924 | break; |
a0d0e21e |
925 | case 't': |
926 | ender = '\t'; |
927 | p++; |
a687059c |
928 | break; |
a0d0e21e |
929 | case 'f': |
930 | ender = '\f'; |
931 | p++; |
a687059c |
932 | break; |
a0d0e21e |
933 | case 'e': |
934 | ender = '\033'; |
935 | p++; |
a687059c |
936 | break; |
a0d0e21e |
937 | case 'a': |
938 | ender = '\007'; |
939 | p++; |
a687059c |
940 | break; |
a0d0e21e |
941 | case 'x': |
942 | ender = scan_hex(++p, 2, &numlen); |
943 | p += numlen; |
a687059c |
944 | break; |
a0d0e21e |
945 | case 'c': |
946 | p++; |
bbce6d69 |
947 | ender = UCHARAT(p++); |
948 | ender = toCTRL(ender); |
a687059c |
949 | break; |
a0d0e21e |
950 | case '0': case '1': case '2': case '3':case '4': |
951 | case '5': case '6': case '7': case '8':case '9': |
952 | if (*p == '0' || |
953 | (isDIGIT(p[1]) && atoi(p) >= regnpar) ) { |
954 | ender = scan_oct(p, 3, &numlen); |
955 | p += numlen; |
956 | } |
957 | else { |
958 | --p; |
959 | goto loopdone; |
a687059c |
960 | } |
961 | break; |
a0d0e21e |
962 | case '\0': |
963 | if (p >= regxend) |
a687059c |
964 | FAIL("trailing \\ in regexp"); |
965 | /* FALL THROUGH */ |
a0d0e21e |
966 | default: |
967 | ender = *p++; |
968 | break; |
969 | } |
970 | break; |
4633a7c4 |
971 | case '#': |
972 | if (regflags & PMf_EXTENDED) { |
973 | while (p < regxend && *p != '\n') p++; |
974 | } |
975 | /* FALL THROUGH */ |
a0d0e21e |
976 | case ' ': case '\t': case '\n': case '\r': case '\f': case '\v': |
977 | if (regflags & PMf_EXTENDED) { |
978 | p++; |
979 | len--; |
980 | continue; |
981 | } |
982 | /* FALL THROUGH */ |
a687059c |
983 | default: |
a0d0e21e |
984 | ender = *p++; |
985 | break; |
a687059c |
986 | } |
a0d0e21e |
987 | if (ISMULT2(p)) { /* Back off on ?+*. */ |
988 | if (len) |
989 | p = oldp; |
990 | else { |
991 | len++; |
992 | regc(ender); |
993 | } |
994 | break; |
a687059c |
995 | } |
a0d0e21e |
996 | regc(ender); |
997 | } |
998 | loopdone: |
999 | regparse = p - 1; |
1000 | nextchar(); |
1001 | if (len < 0) |
1002 | FAIL("internal disaster in regexp"); |
1003 | if (len > 0) |
1004 | *flagp |= HASWIDTH; |
1005 | if (len == 1) |
1006 | *flagp |= SIMPLE; |
1007 | if (regcode != ®dummy) |
1008 | *OPERAND(ret) = len; |
1009 | regc('\0'); |
a687059c |
1010 | } |
a0d0e21e |
1011 | break; |
1012 | } |
a687059c |
1013 | |
a0d0e21e |
1014 | return(ret); |
a687059c |
1015 | } |
1016 | |
1017 | static void |
bbce6d69 |
1018 | regset(opnd, c) |
1019 | char *opnd; |
79072805 |
1020 | register I32 c; |
a687059c |
1021 | { |
bbce6d69 |
1022 | if (opnd == ®dummy) |
1023 | return; |
1024 | c &= 0xFF; |
1025 | opnd[1 + (c >> 3)] |= (1 << (c & 7)); |
a687059c |
1026 | } |
1027 | |
1028 | static char * |
1029 | regclass() |
1030 | { |
bbce6d69 |
1031 | register char *opnd; |
a0d0e21e |
1032 | register I32 class; |
1033 | register I32 lastclass = 1234; |
1034 | register I32 range = 0; |
1035 | register char *ret; |
1036 | register I32 def; |
1037 | I32 numlen; |
1038 | |
1039 | ret = regnode(ANYOF); |
bbce6d69 |
1040 | opnd = regcode; |
1041 | for (class = 0; class < 33; class++) |
1042 | regc(0); |
a0d0e21e |
1043 | if (*regparse == '^') { /* Complement of range. */ |
1044 | regnaughty++; |
1045 | regparse++; |
bbce6d69 |
1046 | if (opnd != ®dummy) |
1047 | *opnd |= ANYOF_INVERT; |
1048 | } |
1049 | if (opnd != ®dummy) { |
1050 | if (regflags & PMf_FOLD) |
1051 | *opnd |= ANYOF_FOLD; |
1052 | if (regflags & PMf_LOCALE) |
1053 | *opnd |= ANYOF_LOCALE; |
a0d0e21e |
1054 | } |
a0d0e21e |
1055 | if (*regparse == ']' || *regparse == '-') |
1056 | goto skipcond; /* allow 1st char to be ] or - */ |
1057 | while (regparse < regxend && *regparse != ']') { |
1058 | skipcond: |
1059 | class = UCHARAT(regparse++); |
1060 | if (class == '\\') { |
1061 | class = UCHARAT(regparse++); |
1062 | switch (class) { |
1063 | case 'w': |
bbce6d69 |
1064 | if (regflags & PMf_LOCALE) { |
1065 | if (opnd != ®dummy) |
1066 | *opnd |= ANYOF_ALNUML; |
1067 | } |
1068 | else { |
1069 | for (class = 0; class < 256; class++) |
1070 | if (isALNUM(class)) |
1071 | regset(opnd, class); |
1072 | } |
a0d0e21e |
1073 | lastclass = 1234; |
1074 | continue; |
1075 | case 'W': |
bbce6d69 |
1076 | if (regflags & PMf_LOCALE) { |
1077 | if (opnd != ®dummy) |
1078 | *opnd |= ANYOF_NALNUML; |
1079 | } |
1080 | else { |
1081 | for (class = 0; class < 256; class++) |
1082 | if (!isALNUM(class)) |
1083 | regset(opnd, class); |
1084 | } |
a0d0e21e |
1085 | lastclass = 1234; |
1086 | continue; |
1087 | case 's': |
bbce6d69 |
1088 | if (regflags & PMf_LOCALE) { |
1089 | if (opnd != ®dummy) |
1090 | *opnd |= ANYOF_SPACEL; |
1091 | } |
1092 | else { |
1093 | for (class = 0; class < 256; class++) |
1094 | if (isSPACE(class)) |
1095 | regset(opnd, class); |
1096 | } |
a0d0e21e |
1097 | lastclass = 1234; |
1098 | continue; |
1099 | case 'S': |
bbce6d69 |
1100 | if (regflags & PMf_LOCALE) { |
1101 | if (opnd != ®dummy) |
1102 | *opnd |= ANYOF_NSPACEL; |
1103 | } |
1104 | else { |
1105 | for (class = 0; class < 256; class++) |
1106 | if (!isSPACE(class)) |
1107 | regset(opnd, class); |
1108 | } |
a0d0e21e |
1109 | lastclass = 1234; |
1110 | continue; |
1111 | case 'd': |
1112 | for (class = '0'; class <= '9'; class++) |
bbce6d69 |
1113 | regset(opnd, class); |
a0d0e21e |
1114 | lastclass = 1234; |
1115 | continue; |
1116 | case 'D': |
1117 | for (class = 0; class < '0'; class++) |
bbce6d69 |
1118 | regset(opnd, class); |
a0d0e21e |
1119 | for (class = '9' + 1; class < 256; class++) |
bbce6d69 |
1120 | regset(opnd, class); |
a0d0e21e |
1121 | lastclass = 1234; |
1122 | continue; |
1123 | case 'n': |
1124 | class = '\n'; |
1125 | break; |
1126 | case 'r': |
1127 | class = '\r'; |
1128 | break; |
1129 | case 't': |
1130 | class = '\t'; |
1131 | break; |
1132 | case 'f': |
1133 | class = '\f'; |
1134 | break; |
1135 | case 'b': |
1136 | class = '\b'; |
1137 | break; |
1138 | case 'e': |
1139 | class = '\033'; |
1140 | break; |
1141 | case 'a': |
1142 | class = '\007'; |
1143 | break; |
1144 | case 'x': |
1145 | class = scan_hex(regparse, 2, &numlen); |
1146 | regparse += numlen; |
1147 | break; |
1148 | case 'c': |
bbce6d69 |
1149 | class = UCHARAT(regparse++); |
1150 | class = toCTRL(class); |
a0d0e21e |
1151 | break; |
1152 | case '0': case '1': case '2': case '3': case '4': |
1153 | case '5': case '6': case '7': case '8': case '9': |
1154 | class = scan_oct(--regparse, 3, &numlen); |
1155 | regparse += numlen; |
1156 | break; |
1157 | } |
1158 | } |
1159 | if (range) { |
1160 | if (lastclass > class) |
1161 | FAIL("invalid [] range in regexp"); |
1162 | range = 0; |
1163 | } |
1164 | else { |
1165 | lastclass = class; |
1166 | if (*regparse == '-' && regparse+1 < regxend && |
1167 | regparse[1] != ']') { |
a687059c |
1168 | regparse++; |
a0d0e21e |
1169 | range = 1; |
1170 | continue; /* do it next time */ |
1171 | } |
a687059c |
1172 | } |
bbce6d69 |
1173 | for ( ; lastclass <= class; lastclass++) |
1174 | regset(opnd, lastclass); |
a0d0e21e |
1175 | lastclass = class; |
1176 | } |
1177 | if (*regparse != ']') |
1178 | FAIL("unmatched [] in regexp"); |
1179 | nextchar(); |
1180 | return ret; |
1181 | } |
1182 | |
1183 | static char* |
1184 | nextchar() |
1185 | { |
1186 | char* retval = regparse++; |
1187 | |
4633a7c4 |
1188 | for (;;) { |
1189 | if (*regparse == '(' && regparse[1] == '?' && |
1190 | regparse[2] == '#') { |
1191 | while (*regparse && *regparse != ')') |
748a9306 |
1192 | regparse++; |
4633a7c4 |
1193 | regparse++; |
1194 | continue; |
1195 | } |
1196 | if (regflags & PMf_EXTENDED) { |
1197 | if (isSPACE(*regparse)) { |
748a9306 |
1198 | regparse++; |
1199 | continue; |
1200 | } |
1201 | else if (*regparse == '#') { |
1202 | while (*regparse && *regparse != '\n') |
1203 | regparse++; |
1204 | regparse++; |
1205 | continue; |
1206 | } |
748a9306 |
1207 | } |
4633a7c4 |
1208 | return retval; |
a0d0e21e |
1209 | } |
a687059c |
1210 | } |
1211 | |
1212 | /* |
a0d0e21e |
1213 | - regnode - emit a node |
1214 | */ |
1215 | #ifdef CAN_PROTOTYPE |
1216 | static char * /* Location. */ |
1217 | regnode(char op) |
1218 | #else |
a687059c |
1219 | static char * /* Location. */ |
1220 | regnode(op) |
1221 | char op; |
a0d0e21e |
1222 | #endif |
a687059c |
1223 | { |
a0d0e21e |
1224 | register char *ret; |
1225 | register char *ptr; |
a687059c |
1226 | |
a0d0e21e |
1227 | ret = regcode; |
1228 | if (ret == ®dummy) { |
a687059c |
1229 | #ifdef REGALIGN |
a0d0e21e |
1230 | if (!(regsize & 1)) |
1231 | regsize++; |
a687059c |
1232 | #endif |
a0d0e21e |
1233 | regsize += 3; |
1234 | return(ret); |
1235 | } |
a687059c |
1236 | |
1237 | #ifdef REGALIGN |
1238 | #ifndef lint |
a0d0e21e |
1239 | if (!((long)ret & 1)) |
1240 | *ret++ = 127; |
a687059c |
1241 | #endif |
1242 | #endif |
a0d0e21e |
1243 | ptr = ret; |
1244 | *ptr++ = op; |
1245 | *ptr++ = '\0'; /* Null "next" pointer. */ |
1246 | *ptr++ = '\0'; |
1247 | regcode = ptr; |
a687059c |
1248 | |
a0d0e21e |
1249 | return(ret); |
a687059c |
1250 | } |
1251 | |
1252 | /* |
a0d0e21e |
1253 | - reganode - emit a node with an argument |
1254 | */ |
1255 | #ifdef CAN_PROTOTYPE |
1256 | static char * /* Location. */ |
1257 | reganode(char op, unsigned short arg) |
1258 | #else |
fe14fcc3 |
1259 | static char * /* Location. */ |
1260 | reganode(op, arg) |
1261 | char op; |
1262 | unsigned short arg; |
a0d0e21e |
1263 | #endif |
fe14fcc3 |
1264 | { |
a0d0e21e |
1265 | register char *ret; |
1266 | register char *ptr; |
fe14fcc3 |
1267 | |
a0d0e21e |
1268 | ret = regcode; |
1269 | if (ret == ®dummy) { |
fe14fcc3 |
1270 | #ifdef REGALIGN |
a0d0e21e |
1271 | if (!(regsize & 1)) |
1272 | regsize++; |
fe14fcc3 |
1273 | #endif |
a0d0e21e |
1274 | regsize += 5; |
1275 | return(ret); |
1276 | } |
fe14fcc3 |
1277 | |
1278 | #ifdef REGALIGN |
1279 | #ifndef lint |
a0d0e21e |
1280 | if (!((long)ret & 1)) |
1281 | *ret++ = 127; |
fe14fcc3 |
1282 | #endif |
1283 | #endif |
a0d0e21e |
1284 | ptr = ret; |
1285 | *ptr++ = op; |
1286 | *ptr++ = '\0'; /* Null "next" pointer. */ |
1287 | *ptr++ = '\0'; |
fe14fcc3 |
1288 | #ifdef REGALIGN |
a0d0e21e |
1289 | *(unsigned short *)(ret+3) = arg; |
fe14fcc3 |
1290 | #else |
a0d0e21e |
1291 | ret[3] = arg >> 8; ret[4] = arg & 0377; |
fe14fcc3 |
1292 | #endif |
a0d0e21e |
1293 | ptr += 2; |
1294 | regcode = ptr; |
fe14fcc3 |
1295 | |
a0d0e21e |
1296 | return(ret); |
fe14fcc3 |
1297 | } |
1298 | |
1299 | /* |
a0d0e21e |
1300 | - regc - emit (if appropriate) a byte of code |
1301 | */ |
1302 | #ifdef CAN_PROTOTYPE |
1303 | static void |
1304 | regc(char b) |
1305 | #else |
a687059c |
1306 | static void |
1307 | regc(b) |
1308 | char b; |
a0d0e21e |
1309 | #endif |
a687059c |
1310 | { |
a0d0e21e |
1311 | if (regcode != ®dummy) |
1312 | *regcode++ = b; |
1313 | else |
1314 | regsize++; |
a687059c |
1315 | } |
1316 | |
1317 | /* |
a0d0e21e |
1318 | - reginsert - insert an operator in front of already-emitted operand |
1319 | * |
1320 | * Means relocating the operand. |
1321 | */ |
1322 | #ifdef CAN_PROTOTYPE |
1323 | static void |
1324 | reginsert(char op, char *opnd) |
1325 | #else |
a687059c |
1326 | static void |
1327 | reginsert(op, opnd) |
1328 | char op; |
1329 | char *opnd; |
a0d0e21e |
1330 | #endif |
a687059c |
1331 | { |
a0d0e21e |
1332 | register char *src; |
1333 | register char *dst; |
1334 | register char *place; |
1335 | register int offset = (regkind[(U8)op] == CURLY ? 4 : 0); |
a687059c |
1336 | |
a0d0e21e |
1337 | if (regcode == ®dummy) { |
a687059c |
1338 | #ifdef REGALIGN |
a0d0e21e |
1339 | regsize += 4 + offset; |
a687059c |
1340 | #else |
a0d0e21e |
1341 | regsize += 3 + offset; |
a687059c |
1342 | #endif |
a0d0e21e |
1343 | return; |
1344 | } |
a687059c |
1345 | |
a0d0e21e |
1346 | src = regcode; |
a687059c |
1347 | #ifdef REGALIGN |
a0d0e21e |
1348 | regcode += 4 + offset; |
a687059c |
1349 | #else |
a0d0e21e |
1350 | regcode += 3 + offset; |
a687059c |
1351 | #endif |
a0d0e21e |
1352 | dst = regcode; |
1353 | while (src > opnd) |
1354 | *--dst = *--src; |
1355 | |
1356 | place = opnd; /* Op node, where operand used to be. */ |
1357 | *place++ = op; |
1358 | *place++ = '\0'; |
1359 | *place++ = '\0'; |
1360 | while (offset-- > 0) |
a687059c |
1361 | *place++ = '\0'; |
2b69d0c2 |
1362 | #ifdef REGALIGN |
a0d0e21e |
1363 | *place++ = '\177'; |
2b69d0c2 |
1364 | #endif |
a687059c |
1365 | } |
1366 | |
1367 | /* |
a0d0e21e |
1368 | - regtail - set the next-pointer at the end of a node chain |
1369 | */ |
a687059c |
1370 | static void |
1371 | regtail(p, val) |
1372 | char *p; |
1373 | char *val; |
1374 | { |
a0d0e21e |
1375 | register char *scan; |
1376 | register char *temp; |
1377 | register I32 offset; |
1378 | |
1379 | if (p == ®dummy) |
1380 | return; |
1381 | |
1382 | /* Find last node. */ |
1383 | scan = p; |
1384 | for (;;) { |
1385 | temp = regnext(scan); |
1386 | if (temp == NULL) |
1387 | break; |
1388 | scan = temp; |
1389 | } |
a687059c |
1390 | |
1391 | #ifdef REGALIGN |
a0d0e21e |
1392 | offset = val - scan; |
a687059c |
1393 | #ifndef lint |
a0d0e21e |
1394 | *(short*)(scan+1) = offset; |
a687059c |
1395 | #else |
a0d0e21e |
1396 | offset = offset; |
a687059c |
1397 | #endif |
1398 | #else |
a0d0e21e |
1399 | if (OP(scan) == BACK) |
1400 | offset = scan - val; |
1401 | else |
1402 | offset = val - scan; |
1403 | *(scan+1) = (offset>>8)&0377; |
1404 | *(scan+2) = offset&0377; |
a687059c |
1405 | #endif |
1406 | } |
1407 | |
1408 | /* |
a0d0e21e |
1409 | - regoptail - regtail on operand of first argument; nop if operandless |
1410 | */ |
a687059c |
1411 | static void |
1412 | regoptail(p, val) |
1413 | char *p; |
1414 | char *val; |
1415 | { |
a0d0e21e |
1416 | /* "Operandless" and "op != BRANCH" are synonymous in practice. */ |
1417 | if (p == NULL || p == ®dummy || regkind[(U8)OP(p)] != BRANCH) |
1418 | return; |
1419 | regtail(NEXTOPER(p), val); |
a687059c |
1420 | } |
1421 | |
1422 | /* |
1423 | - regcurly - a little FSA that accepts {\d+,?\d*} |
1424 | */ |
79072805 |
1425 | STATIC I32 |
a687059c |
1426 | regcurly(s) |
1427 | register char *s; |
1428 | { |
1429 | if (*s++ != '{') |
1430 | return FALSE; |
f0fcb552 |
1431 | if (!isDIGIT(*s)) |
a687059c |
1432 | return FALSE; |
f0fcb552 |
1433 | while (isDIGIT(*s)) |
a687059c |
1434 | s++; |
1435 | if (*s == ',') |
1436 | s++; |
f0fcb552 |
1437 | while (isDIGIT(*s)) |
a687059c |
1438 | s++; |
1439 | if (*s != '}') |
1440 | return FALSE; |
1441 | return TRUE; |
1442 | } |
1443 | |
1444 | #ifdef DEBUGGING |
1445 | |
1446 | /* |
fd181c75 |
1447 | - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form |
a687059c |
1448 | */ |
1449 | void |
1450 | regdump(r) |
1451 | regexp *r; |
1452 | { |
a0d0e21e |
1453 | register char *s; |
bbce6d69 |
1454 | register char op = EXACT; /* Arbitrary non-END op. */ |
a0d0e21e |
1455 | register char *next; |
a687059c |
1456 | |
1457 | |
a0d0e21e |
1458 | s = r->program + 1; |
1459 | while (op != END) { /* While that wasn't END last time... */ |
a687059c |
1460 | #ifdef REGALIGN |
a0d0e21e |
1461 | if (!((long)s & 1)) |
1462 | s++; |
a687059c |
1463 | #endif |
a0d0e21e |
1464 | op = OP(s); |
760ac839 |
1465 | PerlIO_printf(Perl_debug_log, "%2d%s", s-r->program, regprop(s)); /* Where, what. */ |
a0d0e21e |
1466 | next = regnext(s); |
1467 | s += regarglen[(U8)op]; |
1468 | if (next == NULL) /* Next ptr. */ |
760ac839 |
1469 | PerlIO_printf(Perl_debug_log, "(0)"); |
a0d0e21e |
1470 | else |
760ac839 |
1471 | PerlIO_printf(Perl_debug_log, "(%d)", (s-r->program)+(next-s)); |
a0d0e21e |
1472 | s += 3; |
1473 | if (op == ANYOF) { |
bbce6d69 |
1474 | s += 33; |
a687059c |
1475 | } |
bbce6d69 |
1476 | if (regkind[(U8)op] == EXACT) { |
a0d0e21e |
1477 | /* Literal string, where present. */ |
1478 | s++; |
760ac839 |
1479 | (void)PerlIO_putc(Perl_debug_log, ' '); |
1480 | (void)PerlIO_putc(Perl_debug_log, '<'); |
a0d0e21e |
1481 | while (*s != '\0') { |
760ac839 |
1482 | (void)PerlIO_putc(Perl_debug_log,*s); |
a0d0e21e |
1483 | s++; |
1484 | } |
760ac839 |
1485 | (void)PerlIO_putc(Perl_debug_log, '>'); |
a0d0e21e |
1486 | s++; |
1487 | } |
760ac839 |
1488 | (void)PerlIO_putc(Perl_debug_log, '\n'); |
a0d0e21e |
1489 | } |
1490 | |
1491 | /* Header fields of interest. */ |
1492 | if (r->regstart) |
760ac839 |
1493 | PerlIO_printf(Perl_debug_log, "start `%s' ", SvPVX(r->regstart)); |
a0d0e21e |
1494 | if (r->regstclass) |
760ac839 |
1495 | PerlIO_printf(Perl_debug_log, "stclass `%s' ", regprop(r->regstclass)); |
a0d0e21e |
1496 | if (r->reganch & ROPT_ANCH) |
760ac839 |
1497 | PerlIO_printf(Perl_debug_log, "anchored "); |
a0d0e21e |
1498 | if (r->reganch & ROPT_SKIP) |
760ac839 |
1499 | PerlIO_printf(Perl_debug_log, "plus "); |
a0d0e21e |
1500 | if (r->reganch & ROPT_IMPLICIT) |
760ac839 |
1501 | PerlIO_printf(Perl_debug_log, "implicit "); |
a0d0e21e |
1502 | if (r->regmust != NULL) |
760ac839 |
1503 | PerlIO_printf(Perl_debug_log, "must have \"%s\" back %ld ", SvPVX(r->regmust), |
a0d0e21e |
1504 | (long) r->regback); |
760ac839 |
1505 | PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen); |
1506 | PerlIO_printf(Perl_debug_log, "\n"); |
a687059c |
1507 | } |
1508 | |
1509 | /* |
a0d0e21e |
1510 | - regprop - printable representation of opcode |
1511 | */ |
a687059c |
1512 | char * |
1513 | regprop(op) |
1514 | char *op; |
1515 | { |
a0d0e21e |
1516 | register char *p = 0; |
1517 | |
1518 | (void) strcpy(buf, ":"); |
1519 | |
1520 | switch (OP(op)) { |
1521 | case BOL: |
1522 | p = "BOL"; |
1523 | break; |
1524 | case MBOL: |
1525 | p = "MBOL"; |
1526 | break; |
1527 | case SBOL: |
1528 | p = "SBOL"; |
1529 | break; |
1530 | case EOL: |
1531 | p = "EOL"; |
1532 | break; |
1533 | case MEOL: |
1534 | p = "MEOL"; |
1535 | break; |
1536 | case SEOL: |
1537 | p = "SEOL"; |
1538 | break; |
1539 | case ANY: |
1540 | p = "ANY"; |
1541 | break; |
1542 | case SANY: |
1543 | p = "SANY"; |
1544 | break; |
1545 | case ANYOF: |
1546 | p = "ANYOF"; |
1547 | break; |
1548 | case BRANCH: |
1549 | p = "BRANCH"; |
1550 | break; |
bbce6d69 |
1551 | case EXACT: |
1552 | p = "EXACT"; |
1553 | break; |
1554 | case EXACTF: |
1555 | p = "EXACTF"; |
1556 | break; |
1557 | case EXACTFL: |
1558 | p = "EXACTFL"; |
a0d0e21e |
1559 | break; |
1560 | case NOTHING: |
1561 | p = "NOTHING"; |
1562 | break; |
1563 | case BACK: |
1564 | p = "BACK"; |
1565 | break; |
1566 | case END: |
1567 | p = "END"; |
1568 | break; |
a0d0e21e |
1569 | case BOUND: |
1570 | p = "BOUND"; |
1571 | break; |
bbce6d69 |
1572 | case BOUNDL: |
1573 | p = "BOUNDL"; |
1574 | break; |
a0d0e21e |
1575 | case NBOUND: |
1576 | p = "NBOUND"; |
1577 | break; |
bbce6d69 |
1578 | case NBOUNDL: |
1579 | p = "NBOUNDL"; |
a0d0e21e |
1580 | break; |
1581 | case CURLY: |
1582 | (void)sprintf(buf+strlen(buf), "CURLY {%d,%d}", ARG1(op),ARG2(op)); |
1583 | p = NULL; |
1584 | break; |
1585 | case CURLYX: |
1586 | (void)sprintf(buf+strlen(buf), "CURLYX {%d,%d}", ARG1(op),ARG2(op)); |
1587 | p = NULL; |
1588 | break; |
1589 | case REF: |
1590 | (void)sprintf(buf+strlen(buf), "REF%d", ARG1(op)); |
1591 | p = NULL; |
1592 | break; |
1593 | case OPEN: |
1594 | (void)sprintf(buf+strlen(buf), "OPEN%d", ARG1(op)); |
1595 | p = NULL; |
1596 | break; |
1597 | case CLOSE: |
1598 | (void)sprintf(buf+strlen(buf), "CLOSE%d", ARG1(op)); |
1599 | p = NULL; |
1600 | break; |
1601 | case STAR: |
1602 | p = "STAR"; |
1603 | break; |
1604 | case PLUS: |
1605 | p = "PLUS"; |
1606 | break; |
1607 | case MINMOD: |
1608 | p = "MINMOD"; |
1609 | break; |
1610 | case GBOL: |
1611 | p = "GBOL"; |
1612 | break; |
1613 | case UNLESSM: |
1614 | p = "UNLESSM"; |
1615 | break; |
1616 | case IFMATCH: |
1617 | p = "IFMATCH"; |
1618 | break; |
1619 | case SUCCEED: |
1620 | p = "SUCCEED"; |
1621 | break; |
1622 | case WHILEM: |
1623 | p = "WHILEM"; |
1624 | break; |
bbce6d69 |
1625 | case DIGIT: |
1626 | p = "DIGIT"; |
1627 | break; |
1628 | case NDIGIT: |
1629 | p = "NDIGIT"; |
1630 | break; |
1631 | case ALNUM: |
1632 | p = "ALNUM"; |
1633 | break; |
1634 | case NALNUM: |
1635 | p = "NALNUM"; |
1636 | break; |
1637 | case SPACE: |
1638 | p = "SPACE"; |
1639 | break; |
1640 | case NSPACE: |
1641 | p = "NSPACE"; |
1642 | break; |
1643 | case ALNUML: |
1644 | p = "ALNUML"; |
1645 | break; |
1646 | case NALNUML: |
1647 | p = "NALNUML"; |
1648 | break; |
1649 | case SPACEL: |
1650 | p = "SPACEL"; |
1651 | break; |
1652 | case NSPACEL: |
1653 | p = "NSPACEL"; |
1654 | break; |
a0d0e21e |
1655 | default: |
1656 | FAIL("corrupted regexp opcode"); |
1657 | } |
1658 | if (p != NULL) |
1659 | (void) strcat(buf, p); |
1660 | return(buf); |
a687059c |
1661 | } |
1662 | #endif /* DEBUGGING */ |
1663 | |
2b69d0c2 |
1664 | void |
e50aee73 |
1665 | pregfree(r) |
a687059c |
1666 | struct regexp *r; |
1667 | { |
a0d0e21e |
1668 | if (!r) |
1669 | return; |
1670 | if (r->precomp) { |
1671 | Safefree(r->precomp); |
1672 | r->precomp = Nullch; |
1673 | } |
1674 | if (r->subbase) { |
1675 | Safefree(r->subbase); |
1676 | r->subbase = Nullch; |
1677 | } |
1678 | if (r->regmust) { |
1679 | SvREFCNT_dec(r->regmust); |
1680 | r->regmust = Nullsv; |
1681 | } |
1682 | if (r->regstart) { |
1683 | SvREFCNT_dec(r->regstart); |
1684 | r->regstart = Nullsv; |
1685 | } |
1686 | Safefree(r->startp); |
1687 | Safefree(r->endp); |
1688 | Safefree(r); |
a687059c |
1689 | } |