Commit | Line | Data |
a687059c |
1 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
2 | * confused with the original package (see point 3 below). Thanks, Henry! |
3 | */ |
4 | |
5 | /* Additional note: this code is very heavily munged from Henry's version |
6 | * in places. In some spots I've traded clarity for efficiency, so don't |
7 | * blame Henry for some of the lack of readability. |
8 | */ |
9 | |
79072805 |
10 | /* $RCSfile: regexec.c,v $$Revision: 4.1 $$Date: 92/08/07 18:26:32 $ |
a687059c |
11 | * |
12 | * $Log: regexec.c,v $ |
79072805 |
13 | * Revision 4.1 92/08/07 18:26:32 lwall |
14 | * |
2b69d0c2 |
15 | * Revision 4.0.1.4 92/06/08 15:25:50 lwall |
16 | * patch20: pattern modifiers i and g didn't interact right |
17 | * patch20: in some cases $` and $' didn't get set by match |
18 | * patch20: /x{0}/ was wrongly interpreted as /x{0,}/ |
19 | * |
f0fcb552 |
20 | * Revision 4.0.1.3 91/11/05 18:23:55 lwall |
21 | * patch11: prepared for ctype implementations that don't define isascii() |
22 | * patch11: initial .* in pattern had dependency on value of $* |
23 | * |
9ef589d8 |
24 | * Revision 4.0.1.2 91/06/07 11:50:33 lwall |
25 | * patch4: new copyright notice |
26 | * patch4: // wouldn't use previous pattern if it started with a null character |
27 | * |
35c8bce7 |
28 | * Revision 4.0.1.1 91/04/12 09:07:39 lwall |
29 | * patch1: regexec only allocated space for 9 subexpresssions |
30 | * |
fe14fcc3 |
31 | * Revision 4.0 91/03/20 01:39:16 lwall |
32 | * 4.0 baseline. |
a687059c |
33 | * |
34 | */ |
f0fcb552 |
35 | /*SUPPRESS 112*/ |
a687059c |
36 | /* |
37 | * regcomp and regexec -- regsub and regerror are not used in perl |
38 | * |
39 | * Copyright (c) 1986 by University of Toronto. |
40 | * Written by Henry Spencer. Not derived from licensed software. |
41 | * |
42 | * Permission is granted to anyone to use this software for any |
43 | * purpose on any computer system, and to redistribute it freely, |
44 | * subject to the following restrictions: |
45 | * |
46 | * 1. The author is not responsible for the consequences of use of |
47 | * this software, no matter how awful, even if they arise |
48 | * from defects in it. |
49 | * |
50 | * 2. The origin of this software must not be misrepresented, either |
51 | * by explicit claim or by omission. |
52 | * |
53 | * 3. Altered versions must be plainly marked as such, and must not |
54 | * be misrepresented as being the original software. |
55 | * |
56 | **** Alterations to Henry's code are... |
57 | **** |
9ef589d8 |
58 | **** Copyright (c) 1991, Larry Wall |
a687059c |
59 | **** |
9ef589d8 |
60 | **** You may distribute under the terms of either the GNU General Public |
61 | **** License or the Artistic License, as specified in the README file. |
a687059c |
62 | * |
63 | * Beware that some of this code is subtly aware of the way operator |
64 | * precedence is structured in regular expressions. Serious changes in |
65 | * regular-expression syntax might require a total rethink. |
66 | */ |
67 | #include "EXTERN.h" |
68 | #include "perl.h" |
69 | #include "regcomp.h" |
70 | |
71 | #ifndef STATIC |
72 | #define STATIC static |
73 | #endif |
74 | |
75 | #ifdef DEBUGGING |
79072805 |
76 | I32 regnarrate = 0; |
a687059c |
77 | #endif |
78 | |
79 | /* |
80 | * regexec and friends |
81 | */ |
82 | |
83 | /* |
a687059c |
84 | * Forwards. |
85 | */ |
79072805 |
86 | STATIC I32 regtry(); |
87 | STATIC I32 regmatch(); |
88 | STATIC I32 regrepeat(); |
a687059c |
89 | |
90 | /* |
91 | - regexec - match a regexp against a string |
92 | */ |
79072805 |
93 | I32 |
a687059c |
94 | regexec(prog, stringarg, strend, strbeg, minend, screamer, safebase) |
95 | register regexp *prog; |
96 | char *stringarg; |
97 | register char *strend; /* pointer to null at end of string */ |
98 | char *strbeg; /* real beginning of string */ |
79072805 |
99 | I32 minend; /* end of match must be at least minend after stringarg */ |
100 | SV *screamer; |
101 | I32 safebase; /* no need to remember string in subbase */ |
a687059c |
102 | { |
103 | register char *s; |
79072805 |
104 | register I32 i; |
a687059c |
105 | register char *c; |
106 | register char *string = stringarg; |
79072805 |
107 | register I32 tmp; |
108 | I32 minlen = 0; /* must match at least this many chars */ |
109 | I32 dontbother = 0; /* how many characters not to try at end */ |
a687059c |
110 | |
111 | /* Be paranoid... */ |
112 | if (prog == NULL || string == NULL) { |
463ee0b2 |
113 | croak("NULL regexp parameter"); |
a687059c |
114 | return(0); |
115 | } |
116 | |
ac58e20f |
117 | if (string == strbeg) /* is ^ valid at stringarg? */ |
118 | regprev = '\n'; |
0a12ae7d |
119 | else { |
ac58e20f |
120 | regprev = stringarg[-1]; |
0a12ae7d |
121 | if (!multiline && regprev == '\n') |
122 | regprev = '\0'; /* force ^ to NOT match */ |
123 | } |
a687059c |
124 | regprecomp = prog->precomp; |
125 | /* Check validity of program. */ |
126 | if (UCHARAT(prog->program) != MAGIC) { |
127 | FAIL("corrupted regexp program"); |
128 | } |
129 | |
130 | if (prog->do_folding) { |
a687059c |
131 | i = strend - string; |
132 | New(1101,c,i+1,char); |
2b69d0c2 |
133 | Copy(string, c, i+1, char); |
a687059c |
134 | string = c; |
135 | strend = string + i; |
136 | for (s = string; s < strend; s++) |
ac58e20f |
137 | if (isUPPER(*s)) |
a687059c |
138 | *s = tolower(*s); |
139 | } |
140 | |
141 | /* If there is a "must appear" string, look for it. */ |
142 | s = string; |
79072805 |
143 | if (prog->regmust != Nullsv && |
9ef589d8 |
144 | (!(prog->reganch & ROPT_ANCH) |
145 | || (multiline && prog->regback >= 0)) ) { |
ac58e20f |
146 | if (stringarg == strbeg && screamer) { |
79072805 |
147 | if (screamfirst[BmRARE(prog->regmust)] >= 0) |
a687059c |
148 | s = screaminstr(screamer,prog->regmust); |
149 | else |
150 | s = Nullch; |
151 | } |
152 | #ifndef lint |
153 | else |
79072805 |
154 | s = fbm_instr((unsigned char*)s, (unsigned char*)strend, |
a687059c |
155 | prog->regmust); |
156 | #endif |
157 | if (!s) { |
79072805 |
158 | ++BmUSEFUL(prog->regmust); /* hooray */ |
a687059c |
159 | goto phooey; /* not present */ |
160 | } |
161 | else if (prog->regback >= 0) { |
162 | s -= prog->regback; |
163 | if (s < string) |
164 | s = string; |
79072805 |
165 | minlen = prog->regback + SvCUR(prog->regmust); |
a687059c |
166 | } |
79072805 |
167 | else if (--BmUSEFUL(prog->regmust) < 0) { /* boo */ |
168 | sv_free(prog->regmust); |
169 | prog->regmust = Nullsv; /* disable regmust */ |
a687059c |
170 | s = string; |
171 | } |
172 | else { |
173 | s = string; |
79072805 |
174 | minlen = SvCUR(prog->regmust); |
a687059c |
175 | } |
176 | } |
177 | |
178 | /* Mark beginning of line for ^ . */ |
ac58e20f |
179 | regbol = string; |
a687059c |
180 | |
181 | /* Mark end of line for $ (and such) */ |
182 | regeol = strend; |
183 | |
184 | /* see how far we have to get to not match where we matched before */ |
185 | regtill = string+minend; |
186 | |
35c8bce7 |
187 | /* Allocate our backreference arrays */ |
188 | if ( regmyp_size < prog->nparens + 1 ) { |
189 | /* Allocate or enlarge the arrays */ |
190 | regmyp_size = prog->nparens + 1; |
191 | if ( regmyp_size < 10 ) regmyp_size = 10; /* minimum */ |
192 | if ( regmystartp ) { |
193 | /* reallocate larger */ |
194 | Renew(regmystartp,regmyp_size,char*); |
195 | Renew(regmyendp, regmyp_size,char*); |
196 | } |
197 | else { |
198 | /* Initial allocation */ |
199 | New(1102,regmystartp,regmyp_size,char*); |
200 | New(1102,regmyendp, regmyp_size,char*); |
201 | } |
202 | |
203 | } |
204 | |
a687059c |
205 | /* Simplest case: anchored match need be tried only once. */ |
206 | /* [unless multiline is set] */ |
9ef589d8 |
207 | if (prog->reganch & ROPT_ANCH) { |
a687059c |
208 | if (regtry(prog, string)) |
209 | goto got_it; |
f0fcb552 |
210 | else if (multiline || (prog->reganch & ROPT_IMPLICIT)) { |
a687059c |
211 | if (minlen) |
212 | dontbother = minlen - 1; |
213 | strend -= dontbother; |
214 | /* for multiline we only have to try after newlines */ |
215 | if (s > string) |
216 | s--; |
00bf170e |
217 | while (s < strend) { |
218 | if (*s++ == '\n') { |
219 | if (s < strend && regtry(prog, s)) |
a687059c |
220 | goto got_it; |
221 | } |
222 | } |
223 | } |
224 | goto phooey; |
225 | } |
226 | |
227 | /* Messy cases: unanchored match. */ |
228 | if (prog->regstart) { |
9ef589d8 |
229 | if (prog->reganch & ROPT_SKIP) { /* we have /x+whatever/ */ |
00bf170e |
230 | /* it must be a one character string */ |
463ee0b2 |
231 | i = SvPVX(prog->regstart)[0]; |
00bf170e |
232 | while (s < strend) { |
233 | if (*s == i) { |
234 | if (regtry(prog, s)) |
235 | goto got_it; |
236 | s++; |
237 | while (s < strend && *s == i) |
238 | s++; |
239 | } |
240 | s++; |
241 | } |
242 | } |
79072805 |
243 | else if (SvPOK(prog->regstart) == 3) { |
00bf170e |
244 | /* We know what string it must start with. */ |
a687059c |
245 | #ifndef lint |
79072805 |
246 | while ((s = fbm_instr((unsigned char*)s, |
a687059c |
247 | (unsigned char*)strend, prog->regstart)) != NULL) |
248 | #else |
249 | while (s = Nullch) |
250 | #endif |
251 | { |
252 | if (regtry(prog, s)) |
253 | goto got_it; |
254 | s++; |
255 | } |
256 | } |
257 | else { |
463ee0b2 |
258 | c = SvPVX(prog->regstart); |
a687059c |
259 | while ((s = ninstr(s, strend, |
79072805 |
260 | c, c + SvCUR(prog->regstart) )) != NULL) { |
a687059c |
261 | if (regtry(prog, s)) |
262 | goto got_it; |
263 | s++; |
264 | } |
265 | } |
266 | goto phooey; |
267 | } |
f0fcb552 |
268 | /*SUPPRESS 560*/ |
a687059c |
269 | if (c = prog->regstclass) { |
79072805 |
270 | I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
00bf170e |
271 | |
a687059c |
272 | if (minlen) |
273 | dontbother = minlen - 1; |
274 | strend -= dontbother; /* don't bother with what can't match */ |
00bf170e |
275 | tmp = 1; |
a687059c |
276 | /* We know what class it must start with. */ |
277 | switch (OP(c)) { |
00bf170e |
278 | case ANYOF: |
a687059c |
279 | c = OPERAND(c); |
280 | while (s < strend) { |
ac58e20f |
281 | i = UCHARAT(s); |
00bf170e |
282 | if (!(c[i >> 3] & (1 << (i&7)))) { |
283 | if (tmp && regtry(prog, s)) |
a687059c |
284 | goto got_it; |
00bf170e |
285 | else |
286 | tmp = doevery; |
287 | } |
288 | else |
289 | tmp = 1; |
a687059c |
290 | s++; |
291 | } |
292 | break; |
293 | case BOUND: |
294 | if (minlen) |
295 | dontbother++,strend--; |
296 | if (s != string) { |
297 | i = s[-1]; |
ac58e20f |
298 | tmp = isALNUM(i); |
a687059c |
299 | } |
300 | else |
ac58e20f |
301 | tmp = isALNUM(regprev); /* assume not alphanumeric */ |
a687059c |
302 | while (s < strend) { |
303 | i = *s; |
ac58e20f |
304 | if (tmp != isALNUM(i)) { |
a687059c |
305 | tmp = !tmp; |
306 | if (regtry(prog, s)) |
307 | goto got_it; |
308 | } |
309 | s++; |
310 | } |
ae986130 |
311 | if ((minlen || tmp) && regtry(prog,s)) |
a687059c |
312 | goto got_it; |
313 | break; |
314 | case NBOUND: |
315 | if (minlen) |
316 | dontbother++,strend--; |
317 | if (s != string) { |
318 | i = s[-1]; |
ac58e20f |
319 | tmp = isALNUM(i); |
a687059c |
320 | } |
321 | else |
ac58e20f |
322 | tmp = isALNUM(regprev); /* assume not alphanumeric */ |
a687059c |
323 | while (s < strend) { |
324 | i = *s; |
ac58e20f |
325 | if (tmp != isALNUM(i)) |
a687059c |
326 | tmp = !tmp; |
327 | else if (regtry(prog, s)) |
328 | goto got_it; |
329 | s++; |
330 | } |
ae986130 |
331 | if ((minlen || !tmp) && regtry(prog,s)) |
a687059c |
332 | goto got_it; |
333 | break; |
334 | case ALNUM: |
335 | while (s < strend) { |
336 | i = *s; |
00bf170e |
337 | if (isALNUM(i)) { |
338 | if (tmp && regtry(prog, s)) |
a687059c |
339 | goto got_it; |
00bf170e |
340 | else |
341 | tmp = doevery; |
342 | } |
343 | else |
344 | tmp = 1; |
a687059c |
345 | s++; |
346 | } |
347 | break; |
348 | case NALNUM: |
349 | while (s < strend) { |
350 | i = *s; |
00bf170e |
351 | if (!isALNUM(i)) { |
352 | if (tmp && regtry(prog, s)) |
a687059c |
353 | goto got_it; |
00bf170e |
354 | else |
355 | tmp = doevery; |
356 | } |
357 | else |
358 | tmp = 1; |
a687059c |
359 | s++; |
360 | } |
361 | break; |
362 | case SPACE: |
363 | while (s < strend) { |
00bf170e |
364 | if (isSPACE(*s)) { |
365 | if (tmp && regtry(prog, s)) |
a687059c |
366 | goto got_it; |
00bf170e |
367 | else |
368 | tmp = doevery; |
369 | } |
370 | else |
371 | tmp = 1; |
a687059c |
372 | s++; |
373 | } |
374 | break; |
375 | case NSPACE: |
376 | while (s < strend) { |
00bf170e |
377 | if (!isSPACE(*s)) { |
378 | if (tmp && regtry(prog, s)) |
a687059c |
379 | goto got_it; |
00bf170e |
380 | else |
381 | tmp = doevery; |
382 | } |
383 | else |
384 | tmp = 1; |
a687059c |
385 | s++; |
386 | } |
387 | break; |
388 | case DIGIT: |
389 | while (s < strend) { |
00bf170e |
390 | if (isDIGIT(*s)) { |
391 | if (tmp && regtry(prog, s)) |
a687059c |
392 | goto got_it; |
00bf170e |
393 | else |
394 | tmp = doevery; |
395 | } |
396 | else |
397 | tmp = 1; |
a687059c |
398 | s++; |
399 | } |
400 | break; |
401 | case NDIGIT: |
402 | while (s < strend) { |
00bf170e |
403 | if (!isDIGIT(*s)) { |
404 | if (tmp && regtry(prog, s)) |
a687059c |
405 | goto got_it; |
00bf170e |
406 | else |
407 | tmp = doevery; |
408 | } |
409 | else |
410 | tmp = 1; |
a687059c |
411 | s++; |
412 | } |
413 | break; |
414 | } |
415 | } |
416 | else { |
663a0e37 |
417 | if (minlen) |
418 | dontbother = minlen - 1; |
a687059c |
419 | strend -= dontbother; |
420 | /* We don't know much -- general case. */ |
421 | do { |
422 | if (regtry(prog, s)) |
423 | goto got_it; |
424 | } while (s++ < strend); |
425 | } |
426 | |
427 | /* Failure. */ |
428 | goto phooey; |
429 | |
430 | got_it: |
2b69d0c2 |
431 | prog->subbeg = strbeg; |
432 | prog->subend = strend; |
a687059c |
433 | if ((!safebase && (prog->nparens || sawampersand)) || prog->do_folding){ |
434 | strend += dontbother; /* uncheat */ |
435 | if (safebase) /* no need for $digit later */ |
436 | s = strbeg; |
437 | else if (strbeg != prog->subbase) { |
438 | i = strend - string + (stringarg - strbeg); |
439 | s = nsavestr(strbeg,i); /* so $digit will work later */ |
440 | if (prog->subbase) |
441 | Safefree(prog->subbase); |
9ef589d8 |
442 | prog->subbeg = prog->subbase = s; |
00bf170e |
443 | prog->subend = s+i; |
a687059c |
444 | } |
2b69d0c2 |
445 | else { |
446 | i = strend - string + (stringarg - strbeg); |
447 | prog->subbeg = s = prog->subbase; |
448 | prog->subend = s+i; |
449 | } |
a687059c |
450 | s += (stringarg - strbeg); |
451 | for (i = 0; i <= prog->nparens; i++) { |
452 | if (prog->endp[i]) { |
453 | prog->startp[i] = s + (prog->startp[i] - string); |
454 | prog->endp[i] = s + (prog->endp[i] - string); |
455 | } |
456 | } |
457 | if (prog->do_folding) |
458 | Safefree(string); |
459 | } |
460 | return(1); |
461 | |
462 | phooey: |
463 | if (prog->do_folding) |
464 | Safefree(string); |
465 | return(0); |
466 | } |
467 | |
468 | /* |
469 | - regtry - try match at specific point |
470 | */ |
79072805 |
471 | static I32 /* 0 failure, 1 success */ |
a687059c |
472 | regtry(prog, string) |
473 | regexp *prog; |
474 | char *string; |
475 | { |
79072805 |
476 | register I32 i; |
a687059c |
477 | register char **sp; |
478 | register char **ep; |
479 | |
480 | reginput = string; |
481 | regstartp = prog->startp; |
482 | regendp = prog->endp; |
483 | reglastparen = &prog->lastparen; |
484 | prog->lastparen = 0; |
485 | |
486 | sp = prog->startp; |
487 | ep = prog->endp; |
488 | if (prog->nparens) { |
fe14fcc3 |
489 | for (i = prog->nparens; i >= 0; i--) { |
a687059c |
490 | *sp++ = NULL; |
491 | *ep++ = NULL; |
492 | } |
493 | } |
494 | if (regmatch(prog->program + 1) && reginput >= regtill) { |
495 | prog->startp[0] = string; |
496 | prog->endp[0] = reginput; |
497 | return(1); |
498 | } else |
499 | return(0); |
500 | } |
501 | |
502 | /* |
503 | - regmatch - main matching routine |
504 | * |
505 | * Conceptually the strategy is simple: check to see whether the current |
506 | * node matches, call self recursively to see whether the rest matches, |
507 | * and then act accordingly. In practice we make some effort to avoid |
508 | * recursion, in particular by going through "ordinary" nodes (that don't |
509 | * need to know whether the rest of the match failed) by a loop instead of |
510 | * by recursion. |
511 | */ |
512 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
513 | * maybe save a little bit of pushing and popping on the stack. It also takes |
514 | * advantage of machines that use a register save mask on subroutine entry. |
515 | */ |
79072805 |
516 | static I32 /* 0 failure, 1 success */ |
a687059c |
517 | regmatch(prog) |
518 | char *prog; |
519 | { |
520 | register char *scan; /* Current node. */ |
521 | char *next; /* Next node. */ |
79072805 |
522 | register I32 nextchar; |
523 | register I32 n; /* no or next */ |
524 | register I32 ln; /* len or last */ |
a687059c |
525 | register char *s; /* operand or save */ |
526 | register char *locinput = reginput; |
527 | |
528 | nextchar = *locinput; |
529 | scan = prog; |
530 | #ifdef DEBUGGING |
531 | if (scan != NULL && regnarrate) |
532 | fprintf(stderr, "%s(\n", regprop(scan)); |
533 | #endif |
534 | while (scan != NULL) { |
535 | #ifdef DEBUGGING |
536 | if (regnarrate) |
537 | fprintf(stderr, "%s...\n", regprop(scan)); |
538 | #endif |
539 | |
540 | #ifdef REGALIGN |
541 | next = scan + NEXT(scan); |
542 | if (next == scan) |
543 | next = NULL; |
544 | #else |
545 | next = regnext(scan); |
546 | #endif |
547 | |
548 | switch (OP(scan)) { |
549 | case BOL: |
ac58e20f |
550 | if (locinput == regbol ? regprev == '\n' : |
a687059c |
551 | ((nextchar || locinput < regeol) && |
552 | locinput[-1] == '\n') ) |
553 | { |
fe14fcc3 |
554 | /* regtill = regbol; */ |
a687059c |
555 | break; |
556 | } |
557 | return(0); |
558 | case EOL: |
559 | if ((nextchar || locinput < regeol) && nextchar != '\n') |
560 | return(0); |
00bf170e |
561 | if (!multiline && regeol - locinput > 1) |
562 | return 0; |
fe14fcc3 |
563 | /* regtill = regbol; */ |
a687059c |
564 | break; |
565 | case ANY: |
566 | if ((nextchar == '\0' && locinput >= regeol) || |
567 | nextchar == '\n') |
568 | return(0); |
569 | nextchar = *++locinput; |
570 | break; |
571 | case EXACTLY: |
572 | s = OPERAND(scan); |
573 | ln = *s++; |
574 | /* Inline the first character, for speed. */ |
575 | if (*s != nextchar) |
576 | return(0); |
00bf170e |
577 | if (regeol - locinput < ln) |
a687059c |
578 | return 0; |
579 | if (ln > 1 && bcmp(s, locinput, ln) != 0) |
580 | return(0); |
581 | locinput += ln; |
582 | nextchar = *locinput; |
583 | break; |
584 | case ANYOF: |
a687059c |
585 | s = OPERAND(scan); |
586 | if (nextchar < 0) |
587 | nextchar = UCHARAT(locinput); |
588 | if (s[nextchar >> 3] & (1 << (nextchar&7))) |
589 | return(0); |
34de22dd |
590 | if (!nextchar && locinput >= regeol) |
a687059c |
591 | return 0; |
34de22dd |
592 | nextchar = *++locinput; |
a687059c |
593 | break; |
594 | case ALNUM: |
595 | if (!nextchar) |
596 | return(0); |
ac58e20f |
597 | if (!isALNUM(nextchar)) |
a687059c |
598 | return(0); |
599 | nextchar = *++locinput; |
600 | break; |
601 | case NALNUM: |
602 | if (!nextchar && locinput >= regeol) |
603 | return(0); |
ac58e20f |
604 | if (isALNUM(nextchar)) |
a687059c |
605 | return(0); |
606 | nextchar = *++locinput; |
607 | break; |
608 | case NBOUND: |
609 | case BOUND: |
610 | if (locinput == regbol) /* was last char in word? */ |
ac58e20f |
611 | ln = isALNUM(regprev); |
a687059c |
612 | else |
ac58e20f |
613 | ln = isALNUM(locinput[-1]); |
614 | n = isALNUM(nextchar); /* is next char in word? */ |
a687059c |
615 | if ((ln == n) == (OP(scan) == BOUND)) |
616 | return(0); |
617 | break; |
618 | case SPACE: |
619 | if (!nextchar && locinput >= regeol) |
620 | return(0); |
ac58e20f |
621 | if (!isSPACE(nextchar)) |
a687059c |
622 | return(0); |
623 | nextchar = *++locinput; |
624 | break; |
625 | case NSPACE: |
626 | if (!nextchar) |
627 | return(0); |
ac58e20f |
628 | if (isSPACE(nextchar)) |
a687059c |
629 | return(0); |
630 | nextchar = *++locinput; |
631 | break; |
632 | case DIGIT: |
ac58e20f |
633 | if (!isDIGIT(nextchar)) |
a687059c |
634 | return(0); |
635 | nextchar = *++locinput; |
636 | break; |
637 | case NDIGIT: |
638 | if (!nextchar && locinput >= regeol) |
639 | return(0); |
ac58e20f |
640 | if (isDIGIT(nextchar)) |
a687059c |
641 | return(0); |
642 | nextchar = *++locinput; |
643 | break; |
644 | case REF: |
fe14fcc3 |
645 | n = ARG1(scan); /* which paren pair */ |
a687059c |
646 | s = regmystartp[n]; |
647 | if (!s) |
648 | return(0); |
649 | if (!regmyendp[n]) |
650 | return(0); |
651 | if (s == regmyendp[n]) |
652 | break; |
653 | /* Inline the first character, for speed. */ |
654 | if (*s != nextchar) |
655 | return(0); |
656 | ln = regmyendp[n] - s; |
657 | if (locinput + ln > regeol) |
658 | return 0; |
659 | if (ln > 1 && bcmp(s, locinput, ln) != 0) |
660 | return(0); |
661 | locinput += ln; |
662 | nextchar = *locinput; |
663 | break; |
664 | |
665 | case NOTHING: |
666 | break; |
667 | case BACK: |
668 | break; |
fe14fcc3 |
669 | case OPEN: |
670 | n = ARG1(scan); /* which paren pair */ |
a687059c |
671 | reginput = locinput; |
672 | |
673 | regmystartp[n] = locinput; /* for REF */ |
674 | if (regmatch(next)) { |
675 | /* |
676 | * Don't set startp if some later |
677 | * invocation of the same parentheses |
678 | * already has. |
679 | */ |
680 | if (regstartp[n] == NULL) |
681 | regstartp[n] = locinput; |
682 | return(1); |
683 | } else |
684 | return(0); |
685 | /* NOTREACHED */ |
fe14fcc3 |
686 | case CLOSE: { |
687 | n = ARG1(scan); /* which paren pair */ |
a687059c |
688 | reginput = locinput; |
689 | |
690 | regmyendp[n] = locinput; /* for REF */ |
691 | if (regmatch(next)) { |
692 | /* |
693 | * Don't set endp if some later |
694 | * invocation of the same parentheses |
695 | * already has. |
696 | */ |
697 | if (regendp[n] == NULL) { |
698 | regendp[n] = locinput; |
699 | if (n > *reglastparen) |
700 | *reglastparen = n; |
701 | } |
702 | return(1); |
703 | } else |
704 | return(0); |
705 | } |
706 | /*NOTREACHED*/ |
707 | case BRANCH: { |
708 | if (OP(next) != BRANCH) /* No choice. */ |
709 | next = NEXTOPER(scan); /* Avoid recursion. */ |
710 | else { |
711 | do { |
712 | reginput = locinput; |
713 | if (regmatch(NEXTOPER(scan))) |
714 | return(1); |
715 | #ifdef REGALIGN |
f0fcb552 |
716 | /*SUPPRESS 560*/ |
a687059c |
717 | if (n = NEXT(scan)) |
718 | scan += n; |
719 | else |
720 | scan = NULL; |
721 | #else |
722 | scan = regnext(scan); |
723 | #endif |
724 | } while (scan != NULL && OP(scan) == BRANCH); |
725 | return(0); |
726 | /* NOTREACHED */ |
727 | } |
728 | } |
729 | break; |
79072805 |
730 | #ifdef NOTYET |
731 | case MINCURLY: |
732 | ln = ARG1(scan); /* min to match */ |
733 | n = -ARG2(scan); /* max to match */ |
734 | scan = NEXTOPER(scan) + 4; |
735 | goto repeat; |
736 | #endif |
00bf170e |
737 | case CURLY: |
738 | ln = ARG1(scan); /* min to match */ |
739 | n = ARG2(scan); /* max to match */ |
740 | scan = NEXTOPER(scan) + 4; |
741 | goto repeat; |
a687059c |
742 | case STAR: |
00bf170e |
743 | ln = 0; |
2b69d0c2 |
744 | n = 32767; |
00bf170e |
745 | scan = NEXTOPER(scan); |
746 | goto repeat; |
a687059c |
747 | case PLUS: |
748 | /* |
749 | * Lookahead to avoid useless match attempts |
750 | * when we know what character comes next. |
751 | */ |
00bf170e |
752 | ln = 1; |
2b69d0c2 |
753 | n = 32767; |
00bf170e |
754 | scan = NEXTOPER(scan); |
755 | repeat: |
a687059c |
756 | if (OP(next) == EXACTLY) |
757 | nextchar = *(OPERAND(next)+1); |
758 | else |
759 | nextchar = -1000; |
a687059c |
760 | reginput = locinput; |
79072805 |
761 | if (n < 0) { |
762 | n = -n; |
763 | while (n >= ln) { |
764 | /* If it could work, try it. */ |
765 | if (nextchar == -1000 || |
766 | *reginput == nextchar) |
767 | if (regmatch(next)) |
768 | return(1); |
769 | /* Couldn't or didn't -- back up. */ |
770 | ln++; |
771 | reginput = locinput + ln; |
772 | } |
773 | } |
774 | else { |
775 | n = regrepeat(scan, n); |
776 | if (!multiline && OP(next) == EOL && ln < n) |
777 | ln = n; /* why back off? */ |
778 | while (n >= ln) { |
779 | /* If it could work, try it. */ |
780 | if (nextchar == -1000 || |
781 | *reginput == nextchar) |
782 | if (regmatch(next)) |
783 | return(1); |
784 | /* Couldn't or didn't -- back up. */ |
785 | n--; |
786 | reginput = locinput + n; |
787 | } |
a687059c |
788 | } |
789 | return(0); |
790 | case END: |
791 | reginput = locinput; /* put where regtry can find it */ |
792 | return(1); /* Success! */ |
793 | default: |
794 | printf("%x %d\n",scan,scan[1]); |
795 | FAIL("regexp memory corruption"); |
796 | } |
797 | |
798 | scan = next; |
799 | } |
800 | |
801 | /* |
802 | * We get here only if there's trouble -- normally "case END" is |
803 | * the terminating point. |
804 | */ |
805 | FAIL("corrupted regexp pointers"); |
806 | /*NOTREACHED*/ |
807 | #ifdef lint |
808 | return 0; |
809 | #endif |
810 | } |
811 | |
812 | /* |
813 | - regrepeat - repeatedly match something simple, report how many |
814 | */ |
815 | /* |
816 | * [This routine now assumes that it will only match on things of length 1. |
817 | * That was true before, but now we assume scan - reginput is the count, |
818 | * rather than incrementing count on every character.] |
819 | */ |
79072805 |
820 | static I32 |
00bf170e |
821 | regrepeat(p, max) |
a687059c |
822 | char *p; |
79072805 |
823 | I32 max; |
a687059c |
824 | { |
825 | register char *scan; |
826 | register char *opnd; |
79072805 |
827 | register I32 c; |
a687059c |
828 | register char *loceol = regeol; |
829 | |
830 | scan = reginput; |
2b69d0c2 |
831 | if (max != 32767 && max < loceol - scan) |
00bf170e |
832 | loceol = scan + max; |
a687059c |
833 | opnd = OPERAND(p); |
834 | switch (OP(p)) { |
835 | case ANY: |
836 | while (scan < loceol && *scan != '\n') |
837 | scan++; |
838 | break; |
839 | case EXACTLY: /* length of string is 1 */ |
840 | opnd++; |
841 | while (scan < loceol && *opnd == *scan) |
842 | scan++; |
843 | break; |
844 | case ANYOF: |
a687059c |
845 | c = UCHARAT(scan); |
846 | while (scan < loceol && !(opnd[c >> 3] & (1 << (c & 7)))) { |
847 | scan++; |
848 | c = UCHARAT(scan); |
849 | } |
850 | break; |
851 | case ALNUM: |
0a12ae7d |
852 | while (scan < loceol && isALNUM(*scan)) |
a687059c |
853 | scan++; |
854 | break; |
855 | case NALNUM: |
ac58e20f |
856 | while (scan < loceol && !isALNUM(*scan)) |
a687059c |
857 | scan++; |
858 | break; |
859 | case SPACE: |
ac58e20f |
860 | while (scan < loceol && isSPACE(*scan)) |
a687059c |
861 | scan++; |
862 | break; |
863 | case NSPACE: |
ac58e20f |
864 | while (scan < loceol && !isSPACE(*scan)) |
a687059c |
865 | scan++; |
866 | break; |
867 | case DIGIT: |
0a12ae7d |
868 | while (scan < loceol && isDIGIT(*scan)) |
a687059c |
869 | scan++; |
870 | break; |
871 | case NDIGIT: |
ac58e20f |
872 | while (scan < loceol && !isDIGIT(*scan)) |
a687059c |
873 | scan++; |
874 | break; |
875 | default: /* Oh dear. Called inappropriately. */ |
876 | FAIL("internal regexp foulup"); |
877 | /* NOTREACHED */ |
878 | } |
879 | |
880 | c = scan - reginput; |
881 | reginput = scan; |
882 | |
883 | return(c); |
884 | } |
885 | |
886 | /* |
887 | - regnext - dig the "next" pointer out of a node |
888 | * |
889 | * [Note, when REGALIGN is defined there are two places in regmatch() |
890 | * that bypass this code for speed.] |
891 | */ |
892 | char * |
893 | regnext(p) |
894 | register char *p; |
895 | { |
79072805 |
896 | register I32 offset; |
a687059c |
897 | |
898 | if (p == ®dummy) |
899 | return(NULL); |
900 | |
901 | offset = NEXT(p); |
902 | if (offset == 0) |
903 | return(NULL); |
904 | |
905 | #ifdef REGALIGN |
906 | return(p+offset); |
907 | #else |
908 | if (OP(p) == BACK) |
909 | return(p-offset); |
910 | else |
911 | return(p+offset); |
912 | #endif |
913 | } |