Commit | Line | Data |
---|---|---|
a687059c LW |
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 | ||
10 | /* $Header: regexec.c,v 3.0 89/10/18 15:22:53 lwall Locked $ | |
11 | * | |
12 | * $Log: regexec.c,v $ | |
13 | * Revision 3.0 89/10/18 15:22:53 lwall | |
14 | * 3.0 baseline | |
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 | ||
61 | /* | |
62 | * regexec and friends | |
63 | */ | |
64 | ||
65 | /* | |
66 | * Global work variables for regexec(). | |
67 | */ | |
68 | static char *regprecomp; | |
69 | static char *reginput; /* String-input pointer. */ | |
70 | static char *regbol; /* Beginning of input, for ^ check. */ | |
71 | static char *regeol; /* End of input, for $ check. */ | |
72 | static char **regstartp; /* Pointer to startp array. */ | |
73 | static char **regendp; /* Ditto for endp. */ | |
74 | static char *reglastparen; /* Similarly for lastparen. */ | |
75 | static char *regtill; | |
76 | ||
77 | static char *regmystartp[10]; /* For remembering backreferences. */ | |
78 | static char *regmyendp[10]; | |
79 | ||
80 | /* | |
81 | * Forwards. | |
82 | */ | |
83 | STATIC int regtry(); | |
84 | STATIC int regmatch(); | |
85 | STATIC int regrepeat(); | |
86 | ||
87 | extern int multiline; | |
88 | ||
89 | /* | |
90 | - regexec - match a regexp against a string | |
91 | */ | |
92 | int | |
93 | regexec(prog, stringarg, strend, strbeg, minend, screamer, safebase) | |
94 | register regexp *prog; | |
95 | char *stringarg; | |
96 | register char *strend; /* pointer to null at end of string */ | |
97 | char *strbeg; /* real beginning of string */ | |
98 | int minend; /* end of match must be at least minend after stringarg */ | |
99 | STR *screamer; | |
100 | int safebase; /* no need to remember string in subbase */ | |
101 | { | |
102 | register char *s; | |
103 | register int i; | |
104 | register char *c; | |
105 | register char *string = stringarg; | |
106 | register int tmp; | |
107 | int minlen = 0; /* must match at least this many chars */ | |
108 | int dontbother = 0; /* how many characters not to try at end */ | |
109 | int beginning = (string == strbeg); /* is ^ valid at stringarg? */ | |
110 | ||
111 | /* Be paranoid... */ | |
112 | if (prog == NULL || string == NULL) { | |
113 | fatal("NULL regexp parameter"); | |
114 | return(0); | |
115 | } | |
116 | ||
117 | regprecomp = prog->precomp; | |
118 | /* Check validity of program. */ | |
119 | if (UCHARAT(prog->program) != MAGIC) { | |
120 | FAIL("corrupted regexp program"); | |
121 | } | |
122 | ||
123 | if (prog->do_folding) { | |
124 | safebase = FALSE; | |
125 | i = strend - string; | |
126 | New(1101,c,i+1,char); | |
127 | (void)bcopy(string, c, i+1); | |
128 | string = c; | |
129 | strend = string + i; | |
130 | for (s = string; s < strend; s++) | |
131 | if (isupper(*s)) | |
132 | *s = tolower(*s); | |
133 | } | |
134 | ||
135 | /* If there is a "must appear" string, look for it. */ | |
136 | s = string; | |
137 | if (prog->regmust != Nullstr) { | |
138 | if (beginning && screamer) { | |
139 | if (screamfirst[prog->regmust->str_rare] >= 0) | |
140 | s = screaminstr(screamer,prog->regmust); | |
141 | else | |
142 | s = Nullch; | |
143 | } | |
144 | #ifndef lint | |
145 | else | |
146 | s = fbminstr((unsigned char*)s, (unsigned char*)strend, | |
147 | prog->regmust); | |
148 | #endif | |
149 | if (!s) { | |
150 | ++prog->regmust->str_u.str_useful; /* hooray */ | |
151 | goto phooey; /* not present */ | |
152 | } | |
153 | else if (prog->regback >= 0) { | |
154 | s -= prog->regback; | |
155 | if (s < string) | |
156 | s = string; | |
157 | minlen = prog->regback + prog->regmust->str_cur; | |
158 | } | |
159 | else if (--prog->regmust->str_u.str_useful < 0) { /* boo */ | |
160 | str_free(prog->regmust); | |
161 | prog->regmust = Nullstr; /* disable regmust */ | |
162 | s = string; | |
163 | } | |
164 | else { | |
165 | s = string; | |
166 | minlen = prog->regmust->str_cur; | |
167 | } | |
168 | } | |
169 | ||
170 | /* Mark beginning of line for ^ . */ | |
171 | if (beginning) | |
172 | regbol = string; | |
173 | else | |
174 | regbol = NULL; | |
175 | ||
176 | /* Mark end of line for $ (and such) */ | |
177 | regeol = strend; | |
178 | ||
179 | /* see how far we have to get to not match where we matched before */ | |
180 | regtill = string+minend; | |
181 | ||
182 | /* Simplest case: anchored match need be tried only once. */ | |
183 | /* [unless multiline is set] */ | |
184 | if (prog->reganch) { | |
185 | if (regtry(prog, string)) | |
186 | goto got_it; | |
187 | else if (multiline) { | |
188 | if (minlen) | |
189 | dontbother = minlen - 1; | |
190 | strend -= dontbother; | |
191 | /* for multiline we only have to try after newlines */ | |
192 | if (s > string) | |
193 | s--; | |
194 | for (; s < strend; s++) { | |
195 | if (*s == '\n') { | |
196 | if (++s < strend && regtry(prog, s)) | |
197 | goto got_it; | |
198 | } | |
199 | } | |
200 | } | |
201 | goto phooey; | |
202 | } | |
203 | ||
204 | /* Messy cases: unanchored match. */ | |
205 | if (prog->regstart) { | |
206 | /* We know what string it must start with. */ | |
207 | if (prog->regstart->str_pok == 3) { | |
208 | #ifndef lint | |
209 | while ((s = fbminstr((unsigned char*)s, | |
210 | (unsigned char*)strend, prog->regstart)) != NULL) | |
211 | #else | |
212 | while (s = Nullch) | |
213 | #endif | |
214 | { | |
215 | if (regtry(prog, s)) | |
216 | goto got_it; | |
217 | s++; | |
218 | } | |
219 | } | |
220 | else { | |
221 | c = prog->regstart->str_ptr; | |
222 | while ((s = ninstr(s, strend, | |
223 | c, c + prog->regstart->str_cur )) != NULL) { | |
224 | if (regtry(prog, s)) | |
225 | goto got_it; | |
226 | s++; | |
227 | } | |
228 | } | |
229 | goto phooey; | |
230 | } | |
231 | if (c = prog->regstclass) { | |
232 | if (minlen) | |
233 | dontbother = minlen - 1; | |
234 | strend -= dontbother; /* don't bother with what can't match */ | |
235 | /* We know what class it must start with. */ | |
236 | switch (OP(c)) { | |
237 | case ANYOF: case ANYBUT: | |
238 | c = OPERAND(c); | |
239 | while (s < strend) { | |
240 | i = *s; | |
241 | if (!(c[i >> 3] & (1 << (i&7)))) | |
242 | if (regtry(prog, s)) | |
243 | goto got_it; | |
244 | s++; | |
245 | } | |
246 | break; | |
247 | case BOUND: | |
248 | if (minlen) | |
249 | dontbother++,strend--; | |
250 | if (s != string) { | |
251 | i = s[-1]; | |
252 | tmp = (isalpha(i) || isdigit(i) || i == '_'); | |
253 | } | |
254 | else | |
255 | tmp = 0; /* assume not alphanumeric */ | |
256 | while (s < strend) { | |
257 | i = *s; | |
258 | if (tmp != (isalpha(i) || isdigit(i) || i == '_')) { | |
259 | tmp = !tmp; | |
260 | if (regtry(prog, s)) | |
261 | goto got_it; | |
262 | } | |
263 | s++; | |
264 | } | |
265 | if (tmp && regtry(prog,s)) | |
266 | goto got_it; | |
267 | break; | |
268 | case NBOUND: | |
269 | if (minlen) | |
270 | dontbother++,strend--; | |
271 | if (s != string) { | |
272 | i = s[-1]; | |
273 | tmp = (isalpha(i) || isdigit(i) || i == '_'); | |
274 | } | |
275 | else | |
276 | tmp = 0; /* assume not alphanumeric */ | |
277 | while (s < strend) { | |
278 | i = *s; | |
279 | if (tmp != (isalpha(i) || isdigit(i) || i == '_')) | |
280 | tmp = !tmp; | |
281 | else if (regtry(prog, s)) | |
282 | goto got_it; | |
283 | s++; | |
284 | } | |
285 | if (!tmp && regtry(prog,s)) | |
286 | goto got_it; | |
287 | break; | |
288 | case ALNUM: | |
289 | while (s < strend) { | |
290 | i = *s; | |
291 | if (isalpha(i) || isdigit(i) || i == '_') | |
292 | if (regtry(prog, s)) | |
293 | goto got_it; | |
294 | s++; | |
295 | } | |
296 | break; | |
297 | case NALNUM: | |
298 | while (s < strend) { | |
299 | i = *s; | |
300 | if (!isalpha(i) && !isdigit(i) && i != '_') | |
301 | if (regtry(prog, s)) | |
302 | goto got_it; | |
303 | s++; | |
304 | } | |
305 | break; | |
306 | case SPACE: | |
307 | while (s < strend) { | |
308 | if (isspace(*s)) | |
309 | if (regtry(prog, s)) | |
310 | goto got_it; | |
311 | s++; | |
312 | } | |
313 | break; | |
314 | case NSPACE: | |
315 | while (s < strend) { | |
316 | if (!isspace(*s)) | |
317 | if (regtry(prog, s)) | |
318 | goto got_it; | |
319 | s++; | |
320 | } | |
321 | break; | |
322 | case DIGIT: | |
323 | while (s < strend) { | |
324 | if (isdigit(*s)) | |
325 | if (regtry(prog, s)) | |
326 | goto got_it; | |
327 | s++; | |
328 | } | |
329 | break; | |
330 | case NDIGIT: | |
331 | while (s < strend) { | |
332 | if (!isdigit(*s)) | |
333 | if (regtry(prog, s)) | |
334 | goto got_it; | |
335 | s++; | |
336 | } | |
337 | break; | |
338 | } | |
339 | } | |
340 | else { | |
341 | dontbother = minend; | |
342 | strend -= dontbother; | |
343 | /* We don't know much -- general case. */ | |
344 | do { | |
345 | if (regtry(prog, s)) | |
346 | goto got_it; | |
347 | } while (s++ < strend); | |
348 | } | |
349 | ||
350 | /* Failure. */ | |
351 | goto phooey; | |
352 | ||
353 | got_it: | |
354 | if ((!safebase && (prog->nparens || sawampersand)) || prog->do_folding){ | |
355 | strend += dontbother; /* uncheat */ | |
356 | if (safebase) /* no need for $digit later */ | |
357 | s = strbeg; | |
358 | else if (strbeg != prog->subbase) { | |
359 | i = strend - string + (stringarg - strbeg); | |
360 | s = nsavestr(strbeg,i); /* so $digit will work later */ | |
361 | if (prog->subbase) | |
362 | Safefree(prog->subbase); | |
363 | prog->subbase = s; | |
364 | } | |
365 | else | |
366 | s = prog->subbase; | |
367 | s += (stringarg - strbeg); | |
368 | for (i = 0; i <= prog->nparens; i++) { | |
369 | if (prog->endp[i]) { | |
370 | prog->startp[i] = s + (prog->startp[i] - string); | |
371 | prog->endp[i] = s + (prog->endp[i] - string); | |
372 | } | |
373 | } | |
374 | if (prog->do_folding) | |
375 | Safefree(string); | |
376 | } | |
377 | return(1); | |
378 | ||
379 | phooey: | |
380 | if (prog->do_folding) | |
381 | Safefree(string); | |
382 | return(0); | |
383 | } | |
384 | ||
385 | /* | |
386 | - regtry - try match at specific point | |
387 | */ | |
388 | static int /* 0 failure, 1 success */ | |
389 | regtry(prog, string) | |
390 | regexp *prog; | |
391 | char *string; | |
392 | { | |
393 | register int i; | |
394 | register char **sp; | |
395 | register char **ep; | |
396 | ||
397 | reginput = string; | |
398 | regstartp = prog->startp; | |
399 | regendp = prog->endp; | |
400 | reglastparen = &prog->lastparen; | |
401 | prog->lastparen = 0; | |
402 | ||
403 | sp = prog->startp; | |
404 | ep = prog->endp; | |
405 | if (prog->nparens) { | |
406 | for (i = NSUBEXP; i > 0; i--) { | |
407 | *sp++ = NULL; | |
408 | *ep++ = NULL; | |
409 | } | |
410 | } | |
411 | if (regmatch(prog->program + 1) && reginput >= regtill) { | |
412 | prog->startp[0] = string; | |
413 | prog->endp[0] = reginput; | |
414 | return(1); | |
415 | } else | |
416 | return(0); | |
417 | } | |
418 | ||
419 | /* | |
420 | - regmatch - main matching routine | |
421 | * | |
422 | * Conceptually the strategy is simple: check to see whether the current | |
423 | * node matches, call self recursively to see whether the rest matches, | |
424 | * and then act accordingly. In practice we make some effort to avoid | |
425 | * recursion, in particular by going through "ordinary" nodes (that don't | |
426 | * need to know whether the rest of the match failed) by a loop instead of | |
427 | * by recursion. | |
428 | */ | |
429 | /* [lwall] I've hoisted the register declarations to the outer block in order to | |
430 | * maybe save a little bit of pushing and popping on the stack. It also takes | |
431 | * advantage of machines that use a register save mask on subroutine entry. | |
432 | */ | |
433 | static int /* 0 failure, 1 success */ | |
434 | regmatch(prog) | |
435 | char *prog; | |
436 | { | |
437 | register char *scan; /* Current node. */ | |
438 | char *next; /* Next node. */ | |
439 | register int nextchar; | |
440 | register int n; /* no or next */ | |
441 | register int ln; /* len or last */ | |
442 | register char *s; /* operand or save */ | |
443 | register char *locinput = reginput; | |
444 | ||
445 | nextchar = *locinput; | |
446 | scan = prog; | |
447 | #ifdef DEBUGGING | |
448 | if (scan != NULL && regnarrate) | |
449 | fprintf(stderr, "%s(\n", regprop(scan)); | |
450 | #endif | |
451 | while (scan != NULL) { | |
452 | #ifdef DEBUGGING | |
453 | if (regnarrate) | |
454 | fprintf(stderr, "%s...\n", regprop(scan)); | |
455 | #endif | |
456 | ||
457 | #ifdef REGALIGN | |
458 | next = scan + NEXT(scan); | |
459 | if (next == scan) | |
460 | next = NULL; | |
461 | #else | |
462 | next = regnext(scan); | |
463 | #endif | |
464 | ||
465 | switch (OP(scan)) { | |
466 | case BOL: | |
467 | if (locinput == regbol || | |
468 | ((nextchar || locinput < regeol) && | |
469 | locinput[-1] == '\n') ) | |
470 | { | |
471 | regtill--; | |
472 | break; | |
473 | } | |
474 | return(0); | |
475 | case EOL: | |
476 | if ((nextchar || locinput < regeol) && nextchar != '\n') | |
477 | return(0); | |
478 | regtill--; | |
479 | break; | |
480 | case ANY: | |
481 | if ((nextchar == '\0' && locinput >= regeol) || | |
482 | nextchar == '\n') | |
483 | return(0); | |
484 | nextchar = *++locinput; | |
485 | break; | |
486 | case EXACTLY: | |
487 | s = OPERAND(scan); | |
488 | ln = *s++; | |
489 | /* Inline the first character, for speed. */ | |
490 | if (*s != nextchar) | |
491 | return(0); | |
492 | if (locinput + ln > regeol) | |
493 | return 0; | |
494 | if (ln > 1 && bcmp(s, locinput, ln) != 0) | |
495 | return(0); | |
496 | locinput += ln; | |
497 | nextchar = *locinput; | |
498 | break; | |
499 | case ANYOF: | |
500 | case ANYBUT: | |
501 | s = OPERAND(scan); | |
502 | if (nextchar < 0) | |
503 | nextchar = UCHARAT(locinput); | |
504 | if (s[nextchar >> 3] & (1 << (nextchar&7))) | |
505 | return(0); | |
506 | nextchar = *++locinput; | |
507 | if (!nextchar && locinput > regeol) | |
508 | return 0; | |
509 | break; | |
510 | case ALNUM: | |
511 | if (!nextchar) | |
512 | return(0); | |
513 | if (!isalpha(nextchar) && !isdigit(nextchar) && | |
514 | nextchar != '_') | |
515 | return(0); | |
516 | nextchar = *++locinput; | |
517 | break; | |
518 | case NALNUM: | |
519 | if (!nextchar && locinput >= regeol) | |
520 | return(0); | |
521 | if (isalpha(nextchar) || isdigit(nextchar) || | |
522 | nextchar == '_') | |
523 | return(0); | |
524 | nextchar = *++locinput; | |
525 | break; | |
526 | case NBOUND: | |
527 | case BOUND: | |
528 | if (locinput == regbol) /* was last char in word? */ | |
529 | ln = 0; | |
530 | else | |
531 | ln = (isalpha(locinput[-1]) || | |
532 | isdigit(locinput[-1]) || | |
533 | locinput[-1] == '_' ); | |
534 | n = (isalpha(nextchar) || isdigit(nextchar) || | |
535 | nextchar == '_' ); /* is next char in word? */ | |
536 | if ((ln == n) == (OP(scan) == BOUND)) | |
537 | return(0); | |
538 | break; | |
539 | case SPACE: | |
540 | if (!nextchar && locinput >= regeol) | |
541 | return(0); | |
542 | if (!isspace(nextchar)) | |
543 | return(0); | |
544 | nextchar = *++locinput; | |
545 | break; | |
546 | case NSPACE: | |
547 | if (!nextchar) | |
548 | return(0); | |
549 | if (isspace(nextchar)) | |
550 | return(0); | |
551 | nextchar = *++locinput; | |
552 | break; | |
553 | case DIGIT: | |
554 | if (!isdigit(nextchar)) | |
555 | return(0); | |
556 | nextchar = *++locinput; | |
557 | break; | |
558 | case NDIGIT: | |
559 | if (!nextchar && locinput >= regeol) | |
560 | return(0); | |
561 | if (isdigit(nextchar)) | |
562 | return(0); | |
563 | nextchar = *++locinput; | |
564 | break; | |
565 | case REF: | |
566 | case REF+1: | |
567 | case REF+2: | |
568 | case REF+3: | |
569 | case REF+4: | |
570 | case REF+5: | |
571 | case REF+6: | |
572 | case REF+7: | |
573 | case REF+8: | |
574 | case REF+9: | |
575 | n = OP(scan) - REF; | |
576 | s = regmystartp[n]; | |
577 | if (!s) | |
578 | return(0); | |
579 | if (!regmyendp[n]) | |
580 | return(0); | |
581 | if (s == regmyendp[n]) | |
582 | break; | |
583 | /* Inline the first character, for speed. */ | |
584 | if (*s != nextchar) | |
585 | return(0); | |
586 | ln = regmyendp[n] - s; | |
587 | if (locinput + ln > regeol) | |
588 | return 0; | |
589 | if (ln > 1 && bcmp(s, locinput, ln) != 0) | |
590 | return(0); | |
591 | locinput += ln; | |
592 | nextchar = *locinput; | |
593 | break; | |
594 | ||
595 | case NOTHING: | |
596 | break; | |
597 | case BACK: | |
598 | break; | |
599 | case OPEN+1: | |
600 | case OPEN+2: | |
601 | case OPEN+3: | |
602 | case OPEN+4: | |
603 | case OPEN+5: | |
604 | case OPEN+6: | |
605 | case OPEN+7: | |
606 | case OPEN+8: | |
607 | case OPEN+9: | |
608 | n = OP(scan) - OPEN; | |
609 | reginput = locinput; | |
610 | ||
611 | regmystartp[n] = locinput; /* for REF */ | |
612 | if (regmatch(next)) { | |
613 | /* | |
614 | * Don't set startp if some later | |
615 | * invocation of the same parentheses | |
616 | * already has. | |
617 | */ | |
618 | if (regstartp[n] == NULL) | |
619 | regstartp[n] = locinput; | |
620 | return(1); | |
621 | } else | |
622 | return(0); | |
623 | /* NOTREACHED */ | |
624 | case CLOSE+1: | |
625 | case CLOSE+2: | |
626 | case CLOSE+3: | |
627 | case CLOSE+4: | |
628 | case CLOSE+5: | |
629 | case CLOSE+6: | |
630 | case CLOSE+7: | |
631 | case CLOSE+8: | |
632 | case CLOSE+9: { | |
633 | n = OP(scan) - CLOSE; | |
634 | reginput = locinput; | |
635 | ||
636 | regmyendp[n] = locinput; /* for REF */ | |
637 | if (regmatch(next)) { | |
638 | /* | |
639 | * Don't set endp if some later | |
640 | * invocation of the same parentheses | |
641 | * already has. | |
642 | */ | |
643 | if (regendp[n] == NULL) { | |
644 | regendp[n] = locinput; | |
645 | if (n > *reglastparen) | |
646 | *reglastparen = n; | |
647 | } | |
648 | return(1); | |
649 | } else | |
650 | return(0); | |
651 | } | |
652 | /*NOTREACHED*/ | |
653 | case BRANCH: { | |
654 | if (OP(next) != BRANCH) /* No choice. */ | |
655 | next = NEXTOPER(scan); /* Avoid recursion. */ | |
656 | else { | |
657 | do { | |
658 | reginput = locinput; | |
659 | if (regmatch(NEXTOPER(scan))) | |
660 | return(1); | |
661 | #ifdef REGALIGN | |
662 | if (n = NEXT(scan)) | |
663 | scan += n; | |
664 | else | |
665 | scan = NULL; | |
666 | #else | |
667 | scan = regnext(scan); | |
668 | #endif | |
669 | } while (scan != NULL && OP(scan) == BRANCH); | |
670 | return(0); | |
671 | /* NOTREACHED */ | |
672 | } | |
673 | } | |
674 | break; | |
675 | case STAR: | |
676 | case PLUS: | |
677 | /* | |
678 | * Lookahead to avoid useless match attempts | |
679 | * when we know what character comes next. | |
680 | */ | |
681 | if (OP(next) == EXACTLY) | |
682 | nextchar = *(OPERAND(next)+1); | |
683 | else | |
684 | nextchar = -1000; | |
685 | ln = (OP(scan) == STAR) ? 0 : 1; | |
686 | reginput = locinput; | |
687 | n = regrepeat(NEXTOPER(scan)); | |
688 | while (n >= ln) { | |
689 | /* If it could work, try it. */ | |
690 | if (nextchar == -1000 || *reginput == nextchar) | |
691 | if (regmatch(next)) | |
692 | return(1); | |
693 | /* Couldn't or didn't -- back up. */ | |
694 | n--; | |
695 | reginput = locinput + n; | |
696 | } | |
697 | return(0); | |
698 | case END: | |
699 | reginput = locinput; /* put where regtry can find it */ | |
700 | return(1); /* Success! */ | |
701 | default: | |
702 | printf("%x %d\n",scan,scan[1]); | |
703 | FAIL("regexp memory corruption"); | |
704 | } | |
705 | ||
706 | scan = next; | |
707 | } | |
708 | ||
709 | /* | |
710 | * We get here only if there's trouble -- normally "case END" is | |
711 | * the terminating point. | |
712 | */ | |
713 | FAIL("corrupted regexp pointers"); | |
714 | /*NOTREACHED*/ | |
715 | #ifdef lint | |
716 | return 0; | |
717 | #endif | |
718 | } | |
719 | ||
720 | /* | |
721 | - regrepeat - repeatedly match something simple, report how many | |
722 | */ | |
723 | /* | |
724 | * [This routine now assumes that it will only match on things of length 1. | |
725 | * That was true before, but now we assume scan - reginput is the count, | |
726 | * rather than incrementing count on every character.] | |
727 | */ | |
728 | static int | |
729 | regrepeat(p) | |
730 | char *p; | |
731 | { | |
732 | register char *scan; | |
733 | register char *opnd; | |
734 | register int c; | |
735 | register char *loceol = regeol; | |
736 | ||
737 | scan = reginput; | |
738 | opnd = OPERAND(p); | |
739 | switch (OP(p)) { | |
740 | case ANY: | |
741 | while (scan < loceol && *scan != '\n') | |
742 | scan++; | |
743 | break; | |
744 | case EXACTLY: /* length of string is 1 */ | |
745 | opnd++; | |
746 | while (scan < loceol && *opnd == *scan) | |
747 | scan++; | |
748 | break; | |
749 | case ANYOF: | |
750 | case ANYBUT: | |
751 | c = UCHARAT(scan); | |
752 | while (scan < loceol && !(opnd[c >> 3] & (1 << (c & 7)))) { | |
753 | scan++; | |
754 | c = UCHARAT(scan); | |
755 | } | |
756 | break; | |
757 | case ALNUM: | |
758 | while (isalpha(*scan) || isdigit(*scan) || *scan == '_') | |
759 | scan++; | |
760 | break; | |
761 | case NALNUM: | |
762 | while (scan < loceol && (!isalpha(*scan) && !isdigit(*scan) && | |
763 | *scan != '_')) | |
764 | scan++; | |
765 | break; | |
766 | case SPACE: | |
767 | while (scan < loceol && isspace(*scan)) | |
768 | scan++; | |
769 | break; | |
770 | case NSPACE: | |
771 | while (scan < loceol && !isspace(*scan)) | |
772 | scan++; | |
773 | break; | |
774 | case DIGIT: | |
775 | while (isdigit(*scan)) | |
776 | scan++; | |
777 | break; | |
778 | case NDIGIT: | |
779 | while (scan < loceol && !isdigit(*scan)) | |
780 | scan++; | |
781 | break; | |
782 | default: /* Oh dear. Called inappropriately. */ | |
783 | FAIL("internal regexp foulup"); | |
784 | /* NOTREACHED */ | |
785 | } | |
786 | ||
787 | c = scan - reginput; | |
788 | reginput = scan; | |
789 | ||
790 | return(c); | |
791 | } | |
792 | ||
793 | /* | |
794 | - regnext - dig the "next" pointer out of a node | |
795 | * | |
796 | * [Note, when REGALIGN is defined there are two places in regmatch() | |
797 | * that bypass this code for speed.] | |
798 | */ | |
799 | char * | |
800 | regnext(p) | |
801 | register char *p; | |
802 | { | |
803 | register int offset; | |
804 | ||
805 | if (p == ®dummy) | |
806 | return(NULL); | |
807 | ||
808 | offset = NEXT(p); | |
809 | if (offset == 0) | |
810 | return(NULL); | |
811 | ||
812 | #ifdef REGALIGN | |
813 | return(p+offset); | |
814 | #else | |
815 | if (OP(p) == BACK) | |
816 | return(p-offset); | |
817 | else | |
818 | return(p+offset); | |
819 | #endif | |
820 | } |