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 | |
fe14fcc3 |
10 | /* $Header: regexec.c,v 4.0 91/03/20 01:39:16 lwall Locked $ |
a687059c |
11 | * |
12 | * $Log: regexec.c,v $ |
fe14fcc3 |
13 | * Revision 4.0 91/03/20 01:39:16 lwall |
14 | * 4.0 baseline. |
a687059c |
15 | * |
16 | */ |
17 | |
18 | /* |
19 | * regcomp and regexec -- regsub and regerror are not used in perl |
20 | * |
21 | * Copyright (c) 1986 by University of Toronto. |
22 | * Written by Henry Spencer. Not derived from licensed software. |
23 | * |
24 | * Permission is granted to anyone to use this software for any |
25 | * purpose on any computer system, and to redistribute it freely, |
26 | * subject to the following restrictions: |
27 | * |
28 | * 1. The author is not responsible for the consequences of use of |
29 | * this software, no matter how awful, even if they arise |
30 | * from defects in it. |
31 | * |
32 | * 2. The origin of this software must not be misrepresented, either |
33 | * by explicit claim or by omission. |
34 | * |
35 | * 3. Altered versions must be plainly marked as such, and must not |
36 | * be misrepresented as being the original software. |
37 | * |
38 | **** Alterations to Henry's code are... |
39 | **** |
40 | **** Copyright (c) 1989, Larry Wall |
41 | **** |
42 | **** You may distribute under the terms of the GNU General Public License |
43 | **** as specified in the README file that comes with the perl 3.0 kit. |
44 | * |
45 | * Beware that some of this code is subtly aware of the way operator |
46 | * precedence is structured in regular expressions. Serious changes in |
47 | * regular-expression syntax might require a total rethink. |
48 | */ |
49 | #include "EXTERN.h" |
50 | #include "perl.h" |
51 | #include "regcomp.h" |
52 | |
53 | #ifndef STATIC |
54 | #define STATIC static |
55 | #endif |
56 | |
57 | #ifdef DEBUGGING |
58 | int regnarrate = 0; |
59 | #endif |
60 | |
ac58e20f |
61 | #define isALNUM(c) (isascii(c) && (isalpha(c) || isdigit(c) || c == '_')) |
62 | #define isSPACE(c) (isascii(c) && isspace(c)) |
63 | #define isDIGIT(c) (isascii(c) && isdigit(c)) |
64 | #define isUPPER(c) (isascii(c) && isupper(c)) |
65 | |
a687059c |
66 | /* |
67 | * regexec and friends |
68 | */ |
69 | |
70 | /* |
71 | * Global work variables for regexec(). |
72 | */ |
73 | static char *regprecomp; |
74 | static char *reginput; /* String-input pointer. */ |
ac58e20f |
75 | static char regprev; /* char before regbol, \n if none */ |
a687059c |
76 | static char *regbol; /* Beginning of input, for ^ check. */ |
77 | static char *regeol; /* End of input, for $ check. */ |
78 | static char **regstartp; /* Pointer to startp array. */ |
79 | static char **regendp; /* Ditto for endp. */ |
80 | static char *reglastparen; /* Similarly for lastparen. */ |
81 | static char *regtill; |
82 | |
83 | static char *regmystartp[10]; /* For remembering backreferences. */ |
84 | static char *regmyendp[10]; |
85 | |
86 | /* |
87 | * Forwards. |
88 | */ |
89 | STATIC int regtry(); |
90 | STATIC int regmatch(); |
91 | STATIC int regrepeat(); |
92 | |
93 | extern int multiline; |
94 | |
95 | /* |
96 | - regexec - match a regexp against a string |
97 | */ |
98 | int |
99 | regexec(prog, stringarg, strend, strbeg, minend, screamer, safebase) |
100 | register regexp *prog; |
101 | char *stringarg; |
102 | register char *strend; /* pointer to null at end of string */ |
103 | char *strbeg; /* real beginning of string */ |
104 | int minend; /* end of match must be at least minend after stringarg */ |
105 | STR *screamer; |
106 | int safebase; /* no need to remember string in subbase */ |
107 | { |
108 | register char *s; |
109 | register int i; |
110 | register char *c; |
111 | register char *string = stringarg; |
112 | register int tmp; |
113 | int minlen = 0; /* must match at least this many chars */ |
114 | int dontbother = 0; /* how many characters not to try at end */ |
a687059c |
115 | |
116 | /* Be paranoid... */ |
117 | if (prog == NULL || string == NULL) { |
118 | fatal("NULL regexp parameter"); |
119 | return(0); |
120 | } |
121 | |
ac58e20f |
122 | if (string == strbeg) /* is ^ valid at stringarg? */ |
123 | regprev = '\n'; |
0a12ae7d |
124 | else { |
ac58e20f |
125 | regprev = stringarg[-1]; |
0a12ae7d |
126 | if (!multiline && regprev == '\n') |
127 | regprev = '\0'; /* force ^ to NOT match */ |
128 | } |
a687059c |
129 | regprecomp = prog->precomp; |
130 | /* Check validity of program. */ |
131 | if (UCHARAT(prog->program) != MAGIC) { |
132 | FAIL("corrupted regexp program"); |
133 | } |
134 | |
135 | if (prog->do_folding) { |
136 | safebase = FALSE; |
137 | i = strend - string; |
138 | New(1101,c,i+1,char); |
139 | (void)bcopy(string, c, i+1); |
140 | string = c; |
141 | strend = string + i; |
142 | for (s = string; s < strend; s++) |
ac58e20f |
143 | if (isUPPER(*s)) |
a687059c |
144 | *s = tolower(*s); |
145 | } |
146 | |
147 | /* If there is a "must appear" string, look for it. */ |
148 | s = string; |
34de22dd |
149 | if (prog->regmust != Nullstr && |
150 | (!(prog->reganch & 1) || (multiline && prog->regback >= 0)) ) { |
ac58e20f |
151 | if (stringarg == strbeg && screamer) { |
a687059c |
152 | if (screamfirst[prog->regmust->str_rare] >= 0) |
153 | s = screaminstr(screamer,prog->regmust); |
154 | else |
155 | s = Nullch; |
156 | } |
157 | #ifndef lint |
158 | else |
159 | s = fbminstr((unsigned char*)s, (unsigned char*)strend, |
160 | prog->regmust); |
161 | #endif |
162 | if (!s) { |
163 | ++prog->regmust->str_u.str_useful; /* hooray */ |
164 | goto phooey; /* not present */ |
165 | } |
166 | else if (prog->regback >= 0) { |
167 | s -= prog->regback; |
168 | if (s < string) |
169 | s = string; |
170 | minlen = prog->regback + prog->regmust->str_cur; |
171 | } |
172 | else if (--prog->regmust->str_u.str_useful < 0) { /* boo */ |
173 | str_free(prog->regmust); |
174 | prog->regmust = Nullstr; /* disable regmust */ |
175 | s = string; |
176 | } |
177 | else { |
178 | s = string; |
179 | minlen = prog->regmust->str_cur; |
180 | } |
181 | } |
182 | |
183 | /* Mark beginning of line for ^ . */ |
ac58e20f |
184 | regbol = string; |
a687059c |
185 | |
186 | /* Mark end of line for $ (and such) */ |
187 | regeol = strend; |
188 | |
189 | /* see how far we have to get to not match where we matched before */ |
190 | regtill = string+minend; |
191 | |
192 | /* Simplest case: anchored match need be tried only once. */ |
193 | /* [unless multiline is set] */ |
00bf170e |
194 | if (prog->reganch & 1) { |
a687059c |
195 | if (regtry(prog, string)) |
196 | goto got_it; |
197 | else if (multiline) { |
198 | if (minlen) |
199 | dontbother = minlen - 1; |
200 | strend -= dontbother; |
201 | /* for multiline we only have to try after newlines */ |
202 | if (s > string) |
203 | s--; |
00bf170e |
204 | while (s < strend) { |
205 | if (*s++ == '\n') { |
206 | if (s < strend && regtry(prog, s)) |
a687059c |
207 | goto got_it; |
208 | } |
209 | } |
210 | } |
211 | goto phooey; |
212 | } |
213 | |
214 | /* Messy cases: unanchored match. */ |
215 | if (prog->regstart) { |
00bf170e |
216 | if (prog->reganch & 2) { /* we have /x+whatever/ */ |
217 | /* it must be a one character string */ |
218 | i = prog->regstart->str_ptr[0]; |
219 | while (s < strend) { |
220 | if (*s == i) { |
221 | if (regtry(prog, s)) |
222 | goto got_it; |
223 | s++; |
224 | while (s < strend && *s == i) |
225 | s++; |
226 | } |
227 | s++; |
228 | } |
229 | } |
230 | else if (prog->regstart->str_pok == 3) { |
231 | /* We know what string it must start with. */ |
a687059c |
232 | #ifndef lint |
233 | while ((s = fbminstr((unsigned char*)s, |
234 | (unsigned char*)strend, prog->regstart)) != NULL) |
235 | #else |
236 | while (s = Nullch) |
237 | #endif |
238 | { |
239 | if (regtry(prog, s)) |
240 | goto got_it; |
241 | s++; |
242 | } |
243 | } |
244 | else { |
245 | c = prog->regstart->str_ptr; |
246 | while ((s = ninstr(s, strend, |
247 | c, c + prog->regstart->str_cur )) != NULL) { |
248 | if (regtry(prog, s)) |
249 | goto got_it; |
250 | s++; |
251 | } |
252 | } |
253 | goto phooey; |
254 | } |
255 | if (c = prog->regstclass) { |
00bf170e |
256 | int doevery = (prog->reganch & 2) == 0; |
257 | |
a687059c |
258 | if (minlen) |
259 | dontbother = minlen - 1; |
260 | strend -= dontbother; /* don't bother with what can't match */ |
00bf170e |
261 | tmp = 1; |
a687059c |
262 | /* We know what class it must start with. */ |
263 | switch (OP(c)) { |
00bf170e |
264 | case ANYOF: |
a687059c |
265 | c = OPERAND(c); |
266 | while (s < strend) { |
ac58e20f |
267 | i = UCHARAT(s); |
00bf170e |
268 | if (!(c[i >> 3] & (1 << (i&7)))) { |
269 | if (tmp && regtry(prog, s)) |
a687059c |
270 | goto got_it; |
00bf170e |
271 | else |
272 | tmp = doevery; |
273 | } |
274 | else |
275 | tmp = 1; |
a687059c |
276 | s++; |
277 | } |
278 | break; |
279 | case BOUND: |
280 | if (minlen) |
281 | dontbother++,strend--; |
282 | if (s != string) { |
283 | i = s[-1]; |
ac58e20f |
284 | tmp = isALNUM(i); |
a687059c |
285 | } |
286 | else |
ac58e20f |
287 | tmp = isALNUM(regprev); /* assume not alphanumeric */ |
a687059c |
288 | while (s < strend) { |
289 | i = *s; |
ac58e20f |
290 | if (tmp != isALNUM(i)) { |
a687059c |
291 | tmp = !tmp; |
292 | if (regtry(prog, s)) |
293 | goto got_it; |
294 | } |
295 | s++; |
296 | } |
ae986130 |
297 | if ((minlen || tmp) && regtry(prog,s)) |
a687059c |
298 | goto got_it; |
299 | break; |
300 | case NBOUND: |
301 | if (minlen) |
302 | dontbother++,strend--; |
303 | if (s != string) { |
304 | i = s[-1]; |
ac58e20f |
305 | tmp = isALNUM(i); |
a687059c |
306 | } |
307 | else |
ac58e20f |
308 | tmp = isALNUM(regprev); /* assume not alphanumeric */ |
a687059c |
309 | while (s < strend) { |
310 | i = *s; |
ac58e20f |
311 | if (tmp != isALNUM(i)) |
a687059c |
312 | tmp = !tmp; |
313 | else if (regtry(prog, s)) |
314 | goto got_it; |
315 | s++; |
316 | } |
ae986130 |
317 | if ((minlen || !tmp) && regtry(prog,s)) |
a687059c |
318 | goto got_it; |
319 | break; |
320 | case ALNUM: |
321 | while (s < strend) { |
322 | i = *s; |
00bf170e |
323 | if (isALNUM(i)) { |
324 | if (tmp && regtry(prog, s)) |
a687059c |
325 | goto got_it; |
00bf170e |
326 | else |
327 | tmp = doevery; |
328 | } |
329 | else |
330 | tmp = 1; |
a687059c |
331 | s++; |
332 | } |
333 | break; |
334 | case NALNUM: |
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 SPACE: |
349 | while (s < strend) { |
00bf170e |
350 | if (isSPACE(*s)) { |
351 | if (tmp && regtry(prog, s)) |
a687059c |
352 | goto got_it; |
00bf170e |
353 | else |
354 | tmp = doevery; |
355 | } |
356 | else |
357 | tmp = 1; |
a687059c |
358 | s++; |
359 | } |
360 | break; |
361 | case NSPACE: |
362 | while (s < strend) { |
00bf170e |
363 | if (!isSPACE(*s)) { |
364 | if (tmp && regtry(prog, s)) |
a687059c |
365 | goto got_it; |
00bf170e |
366 | else |
367 | tmp = doevery; |
368 | } |
369 | else |
370 | tmp = 1; |
a687059c |
371 | s++; |
372 | } |
373 | break; |
374 | case DIGIT: |
375 | while (s < strend) { |
00bf170e |
376 | if (isDIGIT(*s)) { |
377 | if (tmp && regtry(prog, s)) |
a687059c |
378 | goto got_it; |
00bf170e |
379 | else |
380 | tmp = doevery; |
381 | } |
382 | else |
383 | tmp = 1; |
a687059c |
384 | s++; |
385 | } |
386 | break; |
387 | case NDIGIT: |
388 | while (s < strend) { |
00bf170e |
389 | if (!isDIGIT(*s)) { |
390 | if (tmp && regtry(prog, s)) |
a687059c |
391 | goto got_it; |
00bf170e |
392 | else |
393 | tmp = doevery; |
394 | } |
395 | else |
396 | tmp = 1; |
a687059c |
397 | s++; |
398 | } |
399 | break; |
400 | } |
401 | } |
402 | else { |
663a0e37 |
403 | if (minlen) |
404 | dontbother = minlen - 1; |
a687059c |
405 | strend -= dontbother; |
406 | /* We don't know much -- general case. */ |
407 | do { |
408 | if (regtry(prog, s)) |
409 | goto got_it; |
410 | } while (s++ < strend); |
411 | } |
412 | |
413 | /* Failure. */ |
414 | goto phooey; |
415 | |
416 | got_it: |
417 | if ((!safebase && (prog->nparens || sawampersand)) || prog->do_folding){ |
418 | strend += dontbother; /* uncheat */ |
419 | if (safebase) /* no need for $digit later */ |
420 | s = strbeg; |
421 | else if (strbeg != prog->subbase) { |
422 | i = strend - string + (stringarg - strbeg); |
423 | s = nsavestr(strbeg,i); /* so $digit will work later */ |
424 | if (prog->subbase) |
425 | Safefree(prog->subbase); |
426 | prog->subbase = s; |
00bf170e |
427 | prog->subend = s+i; |
a687059c |
428 | } |
429 | else |
430 | s = prog->subbase; |
431 | s += (stringarg - strbeg); |
432 | for (i = 0; i <= prog->nparens; i++) { |
433 | if (prog->endp[i]) { |
434 | prog->startp[i] = s + (prog->startp[i] - string); |
435 | prog->endp[i] = s + (prog->endp[i] - string); |
436 | } |
437 | } |
438 | if (prog->do_folding) |
439 | Safefree(string); |
440 | } |
441 | return(1); |
442 | |
443 | phooey: |
444 | if (prog->do_folding) |
445 | Safefree(string); |
446 | return(0); |
447 | } |
448 | |
449 | /* |
450 | - regtry - try match at specific point |
451 | */ |
452 | static int /* 0 failure, 1 success */ |
453 | regtry(prog, string) |
454 | regexp *prog; |
455 | char *string; |
456 | { |
457 | register int i; |
458 | register char **sp; |
459 | register char **ep; |
460 | |
461 | reginput = string; |
462 | regstartp = prog->startp; |
463 | regendp = prog->endp; |
464 | reglastparen = &prog->lastparen; |
465 | prog->lastparen = 0; |
466 | |
467 | sp = prog->startp; |
468 | ep = prog->endp; |
469 | if (prog->nparens) { |
fe14fcc3 |
470 | for (i = prog->nparens; i >= 0; i--) { |
a687059c |
471 | *sp++ = NULL; |
472 | *ep++ = NULL; |
473 | } |
474 | } |
475 | if (regmatch(prog->program + 1) && reginput >= regtill) { |
476 | prog->startp[0] = string; |
477 | prog->endp[0] = reginput; |
478 | return(1); |
479 | } else |
480 | return(0); |
481 | } |
482 | |
483 | /* |
484 | - regmatch - main matching routine |
485 | * |
486 | * Conceptually the strategy is simple: check to see whether the current |
487 | * node matches, call self recursively to see whether the rest matches, |
488 | * and then act accordingly. In practice we make some effort to avoid |
489 | * recursion, in particular by going through "ordinary" nodes (that don't |
490 | * need to know whether the rest of the match failed) by a loop instead of |
491 | * by recursion. |
492 | */ |
493 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
494 | * maybe save a little bit of pushing and popping on the stack. It also takes |
495 | * advantage of machines that use a register save mask on subroutine entry. |
496 | */ |
497 | static int /* 0 failure, 1 success */ |
498 | regmatch(prog) |
499 | char *prog; |
500 | { |
501 | register char *scan; /* Current node. */ |
502 | char *next; /* Next node. */ |
503 | register int nextchar; |
504 | register int n; /* no or next */ |
505 | register int ln; /* len or last */ |
506 | register char *s; /* operand or save */ |
507 | register char *locinput = reginput; |
508 | |
509 | nextchar = *locinput; |
510 | scan = prog; |
511 | #ifdef DEBUGGING |
512 | if (scan != NULL && regnarrate) |
513 | fprintf(stderr, "%s(\n", regprop(scan)); |
514 | #endif |
515 | while (scan != NULL) { |
516 | #ifdef DEBUGGING |
517 | if (regnarrate) |
518 | fprintf(stderr, "%s...\n", regprop(scan)); |
519 | #endif |
520 | |
521 | #ifdef REGALIGN |
522 | next = scan + NEXT(scan); |
523 | if (next == scan) |
524 | next = NULL; |
525 | #else |
526 | next = regnext(scan); |
527 | #endif |
528 | |
529 | switch (OP(scan)) { |
530 | case BOL: |
ac58e20f |
531 | if (locinput == regbol ? regprev == '\n' : |
a687059c |
532 | ((nextchar || locinput < regeol) && |
533 | locinput[-1] == '\n') ) |
534 | { |
fe14fcc3 |
535 | /* regtill = regbol; */ |
a687059c |
536 | break; |
537 | } |
538 | return(0); |
539 | case EOL: |
540 | if ((nextchar || locinput < regeol) && nextchar != '\n') |
541 | return(0); |
00bf170e |
542 | if (!multiline && regeol - locinput > 1) |
543 | return 0; |
fe14fcc3 |
544 | /* regtill = regbol; */ |
a687059c |
545 | break; |
546 | case ANY: |
547 | if ((nextchar == '\0' && locinput >= regeol) || |
548 | nextchar == '\n') |
549 | return(0); |
550 | nextchar = *++locinput; |
551 | break; |
552 | case EXACTLY: |
553 | s = OPERAND(scan); |
554 | ln = *s++; |
555 | /* Inline the first character, for speed. */ |
556 | if (*s != nextchar) |
557 | return(0); |
00bf170e |
558 | if (regeol - locinput < ln) |
a687059c |
559 | return 0; |
560 | if (ln > 1 && bcmp(s, locinput, ln) != 0) |
561 | return(0); |
562 | locinput += ln; |
563 | nextchar = *locinput; |
564 | break; |
565 | case ANYOF: |
a687059c |
566 | s = OPERAND(scan); |
567 | if (nextchar < 0) |
568 | nextchar = UCHARAT(locinput); |
569 | if (s[nextchar >> 3] & (1 << (nextchar&7))) |
570 | return(0); |
34de22dd |
571 | if (!nextchar && locinput >= regeol) |
a687059c |
572 | return 0; |
34de22dd |
573 | nextchar = *++locinput; |
a687059c |
574 | break; |
575 | case ALNUM: |
576 | if (!nextchar) |
577 | return(0); |
ac58e20f |
578 | if (!isALNUM(nextchar)) |
a687059c |
579 | return(0); |
580 | nextchar = *++locinput; |
581 | break; |
582 | case NALNUM: |
583 | if (!nextchar && locinput >= regeol) |
584 | return(0); |
ac58e20f |
585 | if (isALNUM(nextchar)) |
a687059c |
586 | return(0); |
587 | nextchar = *++locinput; |
588 | break; |
589 | case NBOUND: |
590 | case BOUND: |
591 | if (locinput == regbol) /* was last char in word? */ |
ac58e20f |
592 | ln = isALNUM(regprev); |
a687059c |
593 | else |
ac58e20f |
594 | ln = isALNUM(locinput[-1]); |
595 | n = isALNUM(nextchar); /* is next char in word? */ |
a687059c |
596 | if ((ln == n) == (OP(scan) == BOUND)) |
597 | return(0); |
598 | break; |
599 | case SPACE: |
600 | if (!nextchar && locinput >= regeol) |
601 | return(0); |
ac58e20f |
602 | if (!isSPACE(nextchar)) |
a687059c |
603 | return(0); |
604 | nextchar = *++locinput; |
605 | break; |
606 | case NSPACE: |
607 | if (!nextchar) |
608 | return(0); |
ac58e20f |
609 | if (isSPACE(nextchar)) |
a687059c |
610 | return(0); |
611 | nextchar = *++locinput; |
612 | break; |
613 | case DIGIT: |
ac58e20f |
614 | if (!isDIGIT(nextchar)) |
a687059c |
615 | return(0); |
616 | nextchar = *++locinput; |
617 | break; |
618 | case NDIGIT: |
619 | if (!nextchar && locinput >= regeol) |
620 | return(0); |
ac58e20f |
621 | if (isDIGIT(nextchar)) |
a687059c |
622 | return(0); |
623 | nextchar = *++locinput; |
624 | break; |
625 | case REF: |
fe14fcc3 |
626 | n = ARG1(scan); /* which paren pair */ |
a687059c |
627 | s = regmystartp[n]; |
628 | if (!s) |
629 | return(0); |
630 | if (!regmyendp[n]) |
631 | return(0); |
632 | if (s == regmyendp[n]) |
633 | break; |
634 | /* Inline the first character, for speed. */ |
635 | if (*s != nextchar) |
636 | return(0); |
637 | ln = regmyendp[n] - s; |
638 | if (locinput + ln > regeol) |
639 | return 0; |
640 | if (ln > 1 && bcmp(s, locinput, ln) != 0) |
641 | return(0); |
642 | locinput += ln; |
643 | nextchar = *locinput; |
644 | break; |
645 | |
646 | case NOTHING: |
647 | break; |
648 | case BACK: |
649 | break; |
fe14fcc3 |
650 | case OPEN: |
651 | n = ARG1(scan); /* which paren pair */ |
a687059c |
652 | reginput = locinput; |
653 | |
654 | regmystartp[n] = locinput; /* for REF */ |
655 | if (regmatch(next)) { |
656 | /* |
657 | * Don't set startp if some later |
658 | * invocation of the same parentheses |
659 | * already has. |
660 | */ |
661 | if (regstartp[n] == NULL) |
662 | regstartp[n] = locinput; |
663 | return(1); |
664 | } else |
665 | return(0); |
666 | /* NOTREACHED */ |
fe14fcc3 |
667 | case CLOSE: { |
668 | n = ARG1(scan); /* which paren pair */ |
a687059c |
669 | reginput = locinput; |
670 | |
671 | regmyendp[n] = locinput; /* for REF */ |
672 | if (regmatch(next)) { |
673 | /* |
674 | * Don't set endp if some later |
675 | * invocation of the same parentheses |
676 | * already has. |
677 | */ |
678 | if (regendp[n] == NULL) { |
679 | regendp[n] = locinput; |
680 | if (n > *reglastparen) |
681 | *reglastparen = n; |
682 | } |
683 | return(1); |
684 | } else |
685 | return(0); |
686 | } |
687 | /*NOTREACHED*/ |
688 | case BRANCH: { |
689 | if (OP(next) != BRANCH) /* No choice. */ |
690 | next = NEXTOPER(scan); /* Avoid recursion. */ |
691 | else { |
692 | do { |
693 | reginput = locinput; |
694 | if (regmatch(NEXTOPER(scan))) |
695 | return(1); |
696 | #ifdef REGALIGN |
697 | if (n = NEXT(scan)) |
698 | scan += n; |
699 | else |
700 | scan = NULL; |
701 | #else |
702 | scan = regnext(scan); |
703 | #endif |
704 | } while (scan != NULL && OP(scan) == BRANCH); |
705 | return(0); |
706 | /* NOTREACHED */ |
707 | } |
708 | } |
709 | break; |
00bf170e |
710 | case CURLY: |
711 | ln = ARG1(scan); /* min to match */ |
712 | n = ARG2(scan); /* max to match */ |
713 | scan = NEXTOPER(scan) + 4; |
714 | goto repeat; |
a687059c |
715 | case STAR: |
00bf170e |
716 | ln = 0; |
717 | n = 0; |
718 | scan = NEXTOPER(scan); |
719 | goto repeat; |
a687059c |
720 | case PLUS: |
721 | /* |
722 | * Lookahead to avoid useless match attempts |
723 | * when we know what character comes next. |
724 | */ |
00bf170e |
725 | ln = 1; |
726 | n = 0; |
727 | scan = NEXTOPER(scan); |
728 | repeat: |
a687059c |
729 | if (OP(next) == EXACTLY) |
730 | nextchar = *(OPERAND(next)+1); |
731 | else |
732 | nextchar = -1000; |
a687059c |
733 | reginput = locinput; |
00bf170e |
734 | n = regrepeat(scan, n); |
0a12ae7d |
735 | if (!multiline && OP(next) == EOL && ln < n) |
00bf170e |
736 | ln = n; /* why back off? */ |
a687059c |
737 | while (n >= ln) { |
738 | /* If it could work, try it. */ |
739 | if (nextchar == -1000 || *reginput == nextchar) |
740 | if (regmatch(next)) |
741 | return(1); |
742 | /* Couldn't or didn't -- back up. */ |
743 | n--; |
744 | reginput = locinput + n; |
745 | } |
746 | return(0); |
747 | case END: |
748 | reginput = locinput; /* put where regtry can find it */ |
749 | return(1); /* Success! */ |
750 | default: |
751 | printf("%x %d\n",scan,scan[1]); |
752 | FAIL("regexp memory corruption"); |
753 | } |
754 | |
755 | scan = next; |
756 | } |
757 | |
758 | /* |
759 | * We get here only if there's trouble -- normally "case END" is |
760 | * the terminating point. |
761 | */ |
762 | FAIL("corrupted regexp pointers"); |
763 | /*NOTREACHED*/ |
764 | #ifdef lint |
765 | return 0; |
766 | #endif |
767 | } |
768 | |
769 | /* |
770 | - regrepeat - repeatedly match something simple, report how many |
771 | */ |
772 | /* |
773 | * [This routine now assumes that it will only match on things of length 1. |
774 | * That was true before, but now we assume scan - reginput is the count, |
775 | * rather than incrementing count on every character.] |
776 | */ |
777 | static int |
00bf170e |
778 | regrepeat(p, max) |
a687059c |
779 | char *p; |
00bf170e |
780 | int max; |
a687059c |
781 | { |
782 | register char *scan; |
783 | register char *opnd; |
784 | register int c; |
785 | register char *loceol = regeol; |
786 | |
787 | scan = reginput; |
00bf170e |
788 | if (max && max < loceol - scan) |
789 | loceol = scan + max; |
a687059c |
790 | opnd = OPERAND(p); |
791 | switch (OP(p)) { |
792 | case ANY: |
793 | while (scan < loceol && *scan != '\n') |
794 | scan++; |
795 | break; |
796 | case EXACTLY: /* length of string is 1 */ |
797 | opnd++; |
798 | while (scan < loceol && *opnd == *scan) |
799 | scan++; |
800 | break; |
801 | case ANYOF: |
a687059c |
802 | c = UCHARAT(scan); |
803 | while (scan < loceol && !(opnd[c >> 3] & (1 << (c & 7)))) { |
804 | scan++; |
805 | c = UCHARAT(scan); |
806 | } |
807 | break; |
808 | case ALNUM: |
0a12ae7d |
809 | while (scan < loceol && isALNUM(*scan)) |
a687059c |
810 | scan++; |
811 | break; |
812 | case NALNUM: |
ac58e20f |
813 | while (scan < loceol && !isALNUM(*scan)) |
a687059c |
814 | scan++; |
815 | break; |
816 | case SPACE: |
ac58e20f |
817 | while (scan < loceol && isSPACE(*scan)) |
a687059c |
818 | scan++; |
819 | break; |
820 | case NSPACE: |
ac58e20f |
821 | while (scan < loceol && !isSPACE(*scan)) |
a687059c |
822 | scan++; |
823 | break; |
824 | case DIGIT: |
0a12ae7d |
825 | while (scan < loceol && isDIGIT(*scan)) |
a687059c |
826 | scan++; |
827 | break; |
828 | case NDIGIT: |
ac58e20f |
829 | while (scan < loceol && !isDIGIT(*scan)) |
a687059c |
830 | scan++; |
831 | break; |
832 | default: /* Oh dear. Called inappropriately. */ |
833 | FAIL("internal regexp foulup"); |
834 | /* NOTREACHED */ |
835 | } |
836 | |
837 | c = scan - reginput; |
838 | reginput = scan; |
839 | |
840 | return(c); |
841 | } |
842 | |
843 | /* |
844 | - regnext - dig the "next" pointer out of a node |
845 | * |
846 | * [Note, when REGALIGN is defined there are two places in regmatch() |
847 | * that bypass this code for speed.] |
848 | */ |
849 | char * |
850 | regnext(p) |
851 | register char *p; |
852 | { |
853 | register int offset; |
854 | |
855 | if (p == ®dummy) |
856 | return(NULL); |
857 | |
858 | offset = NEXT(p); |
859 | if (offset == 0) |
860 | return(NULL); |
861 | |
862 | #ifdef REGALIGN |
863 | return(p+offset); |
864 | #else |
865 | if (OP(p) == BACK) |
866 | return(p-offset); |
867 | else |
868 | return(p+offset); |
869 | #endif |
870 | } |