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 | |
35c8bce7 |
10 | /* $RCSfile: regexec.c,v $$Revision: 4.0.1.1 $$Date: 91/04/12 09:07:39 $ |
a687059c |
11 | * |
12 | * $Log: regexec.c,v $ |
35c8bce7 |
13 | * Revision 4.0.1.1 91/04/12 09:07:39 lwall |
14 | * patch1: regexec only allocated space for 9 subexpresssions |
15 | * |
fe14fcc3 |
16 | * Revision 4.0 91/03/20 01:39:16 lwall |
17 | * 4.0 baseline. |
a687059c |
18 | * |
19 | */ |
20 | |
21 | /* |
22 | * regcomp and regexec -- regsub and regerror are not used in perl |
23 | * |
24 | * Copyright (c) 1986 by University of Toronto. |
25 | * Written by Henry Spencer. Not derived from licensed software. |
26 | * |
27 | * Permission is granted to anyone to use this software for any |
28 | * purpose on any computer system, and to redistribute it freely, |
29 | * subject to the following restrictions: |
30 | * |
31 | * 1. The author is not responsible for the consequences of use of |
32 | * this software, no matter how awful, even if they arise |
33 | * from defects in it. |
34 | * |
35 | * 2. The origin of this software must not be misrepresented, either |
36 | * by explicit claim or by omission. |
37 | * |
38 | * 3. Altered versions must be plainly marked as such, and must not |
39 | * be misrepresented as being the original software. |
40 | * |
41 | **** Alterations to Henry's code are... |
42 | **** |
43 | **** Copyright (c) 1989, Larry Wall |
44 | **** |
45 | **** You may distribute under the terms of the GNU General Public License |
46 | **** as specified in the README file that comes with the perl 3.0 kit. |
47 | * |
48 | * Beware that some of this code is subtly aware of the way operator |
49 | * precedence is structured in regular expressions. Serious changes in |
50 | * regular-expression syntax might require a total rethink. |
51 | */ |
52 | #include "EXTERN.h" |
53 | #include "perl.h" |
54 | #include "regcomp.h" |
55 | |
56 | #ifndef STATIC |
57 | #define STATIC static |
58 | #endif |
59 | |
60 | #ifdef DEBUGGING |
61 | int regnarrate = 0; |
62 | #endif |
63 | |
ac58e20f |
64 | #define isALNUM(c) (isascii(c) && (isalpha(c) || isdigit(c) || c == '_')) |
65 | #define isSPACE(c) (isascii(c) && isspace(c)) |
66 | #define isDIGIT(c) (isascii(c) && isdigit(c)) |
67 | #define isUPPER(c) (isascii(c) && isupper(c)) |
68 | |
a687059c |
69 | /* |
70 | * regexec and friends |
71 | */ |
72 | |
73 | /* |
74 | * Global work variables for regexec(). |
75 | */ |
76 | static char *regprecomp; |
77 | static char *reginput; /* String-input pointer. */ |
ac58e20f |
78 | static char regprev; /* char before regbol, \n if none */ |
a687059c |
79 | static char *regbol; /* Beginning of input, for ^ check. */ |
80 | static char *regeol; /* End of input, for $ check. */ |
81 | static char **regstartp; /* Pointer to startp array. */ |
82 | static char **regendp; /* Ditto for endp. */ |
83 | static char *reglastparen; /* Similarly for lastparen. */ |
84 | static char *regtill; |
85 | |
35c8bce7 |
86 | static int regmyp_size = 0; |
87 | static char **regmystartp = Null(char**); |
88 | static char **regmyendp = Null(char**); |
a687059c |
89 | |
90 | /* |
91 | * Forwards. |
92 | */ |
93 | STATIC int regtry(); |
94 | STATIC int regmatch(); |
95 | STATIC int regrepeat(); |
96 | |
97 | extern int multiline; |
98 | |
99 | /* |
100 | - regexec - match a regexp against a string |
101 | */ |
102 | int |
103 | regexec(prog, stringarg, strend, strbeg, minend, screamer, safebase) |
104 | register regexp *prog; |
105 | char *stringarg; |
106 | register char *strend; /* pointer to null at end of string */ |
107 | char *strbeg; /* real beginning of string */ |
108 | int minend; /* end of match must be at least minend after stringarg */ |
109 | STR *screamer; |
110 | int safebase; /* no need to remember string in subbase */ |
111 | { |
112 | register char *s; |
113 | register int i; |
114 | register char *c; |
115 | register char *string = stringarg; |
116 | register int tmp; |
117 | int minlen = 0; /* must match at least this many chars */ |
118 | int dontbother = 0; /* how many characters not to try at end */ |
a687059c |
119 | |
120 | /* Be paranoid... */ |
121 | if (prog == NULL || string == NULL) { |
122 | fatal("NULL regexp parameter"); |
123 | return(0); |
124 | } |
125 | |
ac58e20f |
126 | if (string == strbeg) /* is ^ valid at stringarg? */ |
127 | regprev = '\n'; |
0a12ae7d |
128 | else { |
ac58e20f |
129 | regprev = stringarg[-1]; |
0a12ae7d |
130 | if (!multiline && regprev == '\n') |
131 | regprev = '\0'; /* force ^ to NOT match */ |
132 | } |
a687059c |
133 | regprecomp = prog->precomp; |
134 | /* Check validity of program. */ |
135 | if (UCHARAT(prog->program) != MAGIC) { |
136 | FAIL("corrupted regexp program"); |
137 | } |
138 | |
139 | if (prog->do_folding) { |
140 | safebase = FALSE; |
141 | i = strend - string; |
142 | New(1101,c,i+1,char); |
143 | (void)bcopy(string, c, i+1); |
144 | string = c; |
145 | strend = string + i; |
146 | for (s = string; s < strend; s++) |
ac58e20f |
147 | if (isUPPER(*s)) |
a687059c |
148 | *s = tolower(*s); |
149 | } |
150 | |
151 | /* If there is a "must appear" string, look for it. */ |
152 | s = string; |
34de22dd |
153 | if (prog->regmust != Nullstr && |
154 | (!(prog->reganch & 1) || (multiline && prog->regback >= 0)) ) { |
ac58e20f |
155 | if (stringarg == strbeg && screamer) { |
a687059c |
156 | if (screamfirst[prog->regmust->str_rare] >= 0) |
157 | s = screaminstr(screamer,prog->regmust); |
158 | else |
159 | s = Nullch; |
160 | } |
161 | #ifndef lint |
162 | else |
163 | s = fbminstr((unsigned char*)s, (unsigned char*)strend, |
164 | prog->regmust); |
165 | #endif |
166 | if (!s) { |
167 | ++prog->regmust->str_u.str_useful; /* hooray */ |
168 | goto phooey; /* not present */ |
169 | } |
170 | else if (prog->regback >= 0) { |
171 | s -= prog->regback; |
172 | if (s < string) |
173 | s = string; |
174 | minlen = prog->regback + prog->regmust->str_cur; |
175 | } |
176 | else if (--prog->regmust->str_u.str_useful < 0) { /* boo */ |
177 | str_free(prog->regmust); |
178 | prog->regmust = Nullstr; /* disable regmust */ |
179 | s = string; |
180 | } |
181 | else { |
182 | s = string; |
183 | minlen = prog->regmust->str_cur; |
184 | } |
185 | } |
186 | |
187 | /* Mark beginning of line for ^ . */ |
ac58e20f |
188 | regbol = string; |
a687059c |
189 | |
190 | /* Mark end of line for $ (and such) */ |
191 | regeol = strend; |
192 | |
193 | /* see how far we have to get to not match where we matched before */ |
194 | regtill = string+minend; |
195 | |
35c8bce7 |
196 | /* Allocate our backreference arrays */ |
197 | if ( regmyp_size < prog->nparens + 1 ) { |
198 | /* Allocate or enlarge the arrays */ |
199 | regmyp_size = prog->nparens + 1; |
200 | if ( regmyp_size < 10 ) regmyp_size = 10; /* minimum */ |
201 | if ( regmystartp ) { |
202 | /* reallocate larger */ |
203 | Renew(regmystartp,regmyp_size,char*); |
204 | Renew(regmyendp, regmyp_size,char*); |
205 | } |
206 | else { |
207 | /* Initial allocation */ |
208 | New(1102,regmystartp,regmyp_size,char*); |
209 | New(1102,regmyendp, regmyp_size,char*); |
210 | } |
211 | |
212 | } |
213 | |
a687059c |
214 | /* Simplest case: anchored match need be tried only once. */ |
215 | /* [unless multiline is set] */ |
00bf170e |
216 | if (prog->reganch & 1) { |
a687059c |
217 | if (regtry(prog, string)) |
218 | goto got_it; |
219 | else if (multiline) { |
220 | if (minlen) |
221 | dontbother = minlen - 1; |
222 | strend -= dontbother; |
223 | /* for multiline we only have to try after newlines */ |
224 | if (s > string) |
225 | s--; |
00bf170e |
226 | while (s < strend) { |
227 | if (*s++ == '\n') { |
228 | if (s < strend && regtry(prog, s)) |
a687059c |
229 | goto got_it; |
230 | } |
231 | } |
232 | } |
233 | goto phooey; |
234 | } |
235 | |
236 | /* Messy cases: unanchored match. */ |
237 | if (prog->regstart) { |
00bf170e |
238 | if (prog->reganch & 2) { /* we have /x+whatever/ */ |
239 | /* it must be a one character string */ |
240 | i = prog->regstart->str_ptr[0]; |
241 | while (s < strend) { |
242 | if (*s == i) { |
243 | if (regtry(prog, s)) |
244 | goto got_it; |
245 | s++; |
246 | while (s < strend && *s == i) |
247 | s++; |
248 | } |
249 | s++; |
250 | } |
251 | } |
252 | else if (prog->regstart->str_pok == 3) { |
253 | /* We know what string it must start with. */ |
a687059c |
254 | #ifndef lint |
255 | while ((s = fbminstr((unsigned char*)s, |
256 | (unsigned char*)strend, prog->regstart)) != NULL) |
257 | #else |
258 | while (s = Nullch) |
259 | #endif |
260 | { |
261 | if (regtry(prog, s)) |
262 | goto got_it; |
263 | s++; |
264 | } |
265 | } |
266 | else { |
267 | c = prog->regstart->str_ptr; |
268 | while ((s = ninstr(s, strend, |
269 | c, c + prog->regstart->str_cur )) != NULL) { |
270 | if (regtry(prog, s)) |
271 | goto got_it; |
272 | s++; |
273 | } |
274 | } |
275 | goto phooey; |
276 | } |
277 | if (c = prog->regstclass) { |
00bf170e |
278 | int doevery = (prog->reganch & 2) == 0; |
279 | |
a687059c |
280 | if (minlen) |
281 | dontbother = minlen - 1; |
282 | strend -= dontbother; /* don't bother with what can't match */ |
00bf170e |
283 | tmp = 1; |
a687059c |
284 | /* We know what class it must start with. */ |
285 | switch (OP(c)) { |
00bf170e |
286 | case ANYOF: |
a687059c |
287 | c = OPERAND(c); |
288 | while (s < strend) { |
ac58e20f |
289 | i = UCHARAT(s); |
00bf170e |
290 | if (!(c[i >> 3] & (1 << (i&7)))) { |
291 | if (tmp && regtry(prog, s)) |
a687059c |
292 | goto got_it; |
00bf170e |
293 | else |
294 | tmp = doevery; |
295 | } |
296 | else |
297 | tmp = 1; |
a687059c |
298 | s++; |
299 | } |
300 | break; |
301 | case BOUND: |
302 | if (minlen) |
303 | dontbother++,strend--; |
304 | if (s != string) { |
305 | i = s[-1]; |
ac58e20f |
306 | tmp = isALNUM(i); |
a687059c |
307 | } |
308 | else |
ac58e20f |
309 | tmp = isALNUM(regprev); /* assume not alphanumeric */ |
a687059c |
310 | while (s < strend) { |
311 | i = *s; |
ac58e20f |
312 | if (tmp != isALNUM(i)) { |
a687059c |
313 | tmp = !tmp; |
314 | if (regtry(prog, s)) |
315 | goto got_it; |
316 | } |
317 | s++; |
318 | } |
ae986130 |
319 | if ((minlen || tmp) && regtry(prog,s)) |
a687059c |
320 | goto got_it; |
321 | break; |
322 | case NBOUND: |
323 | if (minlen) |
324 | dontbother++,strend--; |
325 | if (s != string) { |
326 | i = s[-1]; |
ac58e20f |
327 | tmp = isALNUM(i); |
a687059c |
328 | } |
329 | else |
ac58e20f |
330 | tmp = isALNUM(regprev); /* assume not alphanumeric */ |
a687059c |
331 | while (s < strend) { |
332 | i = *s; |
ac58e20f |
333 | if (tmp != isALNUM(i)) |
a687059c |
334 | tmp = !tmp; |
335 | else if (regtry(prog, s)) |
336 | goto got_it; |
337 | s++; |
338 | } |
ae986130 |
339 | if ((minlen || !tmp) && regtry(prog,s)) |
a687059c |
340 | goto got_it; |
341 | break; |
342 | case ALNUM: |
343 | while (s < strend) { |
344 | i = *s; |
00bf170e |
345 | if (isALNUM(i)) { |
346 | if (tmp && regtry(prog, s)) |
a687059c |
347 | goto got_it; |
00bf170e |
348 | else |
349 | tmp = doevery; |
350 | } |
351 | else |
352 | tmp = 1; |
a687059c |
353 | s++; |
354 | } |
355 | break; |
356 | case NALNUM: |
357 | while (s < strend) { |
358 | i = *s; |
00bf170e |
359 | if (!isALNUM(i)) { |
360 | if (tmp && regtry(prog, s)) |
a687059c |
361 | goto got_it; |
00bf170e |
362 | else |
363 | tmp = doevery; |
364 | } |
365 | else |
366 | tmp = 1; |
a687059c |
367 | s++; |
368 | } |
369 | break; |
370 | case SPACE: |
371 | while (s < strend) { |
00bf170e |
372 | if (isSPACE(*s)) { |
373 | if (tmp && regtry(prog, s)) |
a687059c |
374 | goto got_it; |
00bf170e |
375 | else |
376 | tmp = doevery; |
377 | } |
378 | else |
379 | tmp = 1; |
a687059c |
380 | s++; |
381 | } |
382 | break; |
383 | case NSPACE: |
384 | while (s < strend) { |
00bf170e |
385 | if (!isSPACE(*s)) { |
386 | if (tmp && regtry(prog, s)) |
a687059c |
387 | goto got_it; |
00bf170e |
388 | else |
389 | tmp = doevery; |
390 | } |
391 | else |
392 | tmp = 1; |
a687059c |
393 | s++; |
394 | } |
395 | break; |
396 | case DIGIT: |
397 | while (s < strend) { |
00bf170e |
398 | if (isDIGIT(*s)) { |
399 | if (tmp && regtry(prog, s)) |
a687059c |
400 | goto got_it; |
00bf170e |
401 | else |
402 | tmp = doevery; |
403 | } |
404 | else |
405 | tmp = 1; |
a687059c |
406 | s++; |
407 | } |
408 | break; |
409 | case NDIGIT: |
410 | while (s < strend) { |
00bf170e |
411 | if (!isDIGIT(*s)) { |
412 | if (tmp && regtry(prog, s)) |
a687059c |
413 | goto got_it; |
00bf170e |
414 | else |
415 | tmp = doevery; |
416 | } |
417 | else |
418 | tmp = 1; |
a687059c |
419 | s++; |
420 | } |
421 | break; |
422 | } |
423 | } |
424 | else { |
663a0e37 |
425 | if (minlen) |
426 | dontbother = minlen - 1; |
a687059c |
427 | strend -= dontbother; |
428 | /* We don't know much -- general case. */ |
429 | do { |
430 | if (regtry(prog, s)) |
431 | goto got_it; |
432 | } while (s++ < strend); |
433 | } |
434 | |
435 | /* Failure. */ |
436 | goto phooey; |
437 | |
438 | got_it: |
439 | if ((!safebase && (prog->nparens || sawampersand)) || prog->do_folding){ |
440 | strend += dontbother; /* uncheat */ |
441 | if (safebase) /* no need for $digit later */ |
442 | s = strbeg; |
443 | else if (strbeg != prog->subbase) { |
444 | i = strend - string + (stringarg - strbeg); |
445 | s = nsavestr(strbeg,i); /* so $digit will work later */ |
446 | if (prog->subbase) |
447 | Safefree(prog->subbase); |
448 | prog->subbase = s; |
00bf170e |
449 | prog->subend = s+i; |
a687059c |
450 | } |
451 | else |
452 | s = prog->subbase; |
453 | s += (stringarg - strbeg); |
454 | for (i = 0; i <= prog->nparens; i++) { |
455 | if (prog->endp[i]) { |
456 | prog->startp[i] = s + (prog->startp[i] - string); |
457 | prog->endp[i] = s + (prog->endp[i] - string); |
458 | } |
459 | } |
460 | if (prog->do_folding) |
461 | Safefree(string); |
462 | } |
463 | return(1); |
464 | |
465 | phooey: |
466 | if (prog->do_folding) |
467 | Safefree(string); |
468 | return(0); |
469 | } |
470 | |
471 | /* |
472 | - regtry - try match at specific point |
473 | */ |
474 | static int /* 0 failure, 1 success */ |
475 | regtry(prog, string) |
476 | regexp *prog; |
477 | char *string; |
478 | { |
479 | register int i; |
480 | register char **sp; |
481 | register char **ep; |
482 | |
483 | reginput = string; |
484 | regstartp = prog->startp; |
485 | regendp = prog->endp; |
486 | reglastparen = &prog->lastparen; |
487 | prog->lastparen = 0; |
488 | |
489 | sp = prog->startp; |
490 | ep = prog->endp; |
491 | if (prog->nparens) { |
fe14fcc3 |
492 | for (i = prog->nparens; i >= 0; i--) { |
a687059c |
493 | *sp++ = NULL; |
494 | *ep++ = NULL; |
495 | } |
496 | } |
497 | if (regmatch(prog->program + 1) && reginput >= regtill) { |
498 | prog->startp[0] = string; |
499 | prog->endp[0] = reginput; |
500 | return(1); |
501 | } else |
502 | return(0); |
503 | } |
504 | |
505 | /* |
506 | - regmatch - main matching routine |
507 | * |
508 | * Conceptually the strategy is simple: check to see whether the current |
509 | * node matches, call self recursively to see whether the rest matches, |
510 | * and then act accordingly. In practice we make some effort to avoid |
511 | * recursion, in particular by going through "ordinary" nodes (that don't |
512 | * need to know whether the rest of the match failed) by a loop instead of |
513 | * by recursion. |
514 | */ |
515 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
516 | * maybe save a little bit of pushing and popping on the stack. It also takes |
517 | * advantage of machines that use a register save mask on subroutine entry. |
518 | */ |
519 | static int /* 0 failure, 1 success */ |
520 | regmatch(prog) |
521 | char *prog; |
522 | { |
523 | register char *scan; /* Current node. */ |
524 | char *next; /* Next node. */ |
525 | register int nextchar; |
526 | register int n; /* no or next */ |
527 | register int ln; /* len or last */ |
528 | register char *s; /* operand or save */ |
529 | register char *locinput = reginput; |
530 | |
531 | nextchar = *locinput; |
532 | scan = prog; |
533 | #ifdef DEBUGGING |
534 | if (scan != NULL && regnarrate) |
535 | fprintf(stderr, "%s(\n", regprop(scan)); |
536 | #endif |
537 | while (scan != NULL) { |
538 | #ifdef DEBUGGING |
539 | if (regnarrate) |
540 | fprintf(stderr, "%s...\n", regprop(scan)); |
541 | #endif |
542 | |
543 | #ifdef REGALIGN |
544 | next = scan + NEXT(scan); |
545 | if (next == scan) |
546 | next = NULL; |
547 | #else |
548 | next = regnext(scan); |
549 | #endif |
550 | |
551 | switch (OP(scan)) { |
552 | case BOL: |
ac58e20f |
553 | if (locinput == regbol ? regprev == '\n' : |
a687059c |
554 | ((nextchar || locinput < regeol) && |
555 | locinput[-1] == '\n') ) |
556 | { |
fe14fcc3 |
557 | /* regtill = regbol; */ |
a687059c |
558 | break; |
559 | } |
560 | return(0); |
561 | case EOL: |
562 | if ((nextchar || locinput < regeol) && nextchar != '\n') |
563 | return(0); |
00bf170e |
564 | if (!multiline && regeol - locinput > 1) |
565 | return 0; |
fe14fcc3 |
566 | /* regtill = regbol; */ |
a687059c |
567 | break; |
568 | case ANY: |
569 | if ((nextchar == '\0' && locinput >= regeol) || |
570 | nextchar == '\n') |
571 | return(0); |
572 | nextchar = *++locinput; |
573 | break; |
574 | case EXACTLY: |
575 | s = OPERAND(scan); |
576 | ln = *s++; |
577 | /* Inline the first character, for speed. */ |
578 | if (*s != nextchar) |
579 | return(0); |
00bf170e |
580 | if (regeol - locinput < ln) |
a687059c |
581 | return 0; |
582 | if (ln > 1 && bcmp(s, locinput, ln) != 0) |
583 | return(0); |
584 | locinput += ln; |
585 | nextchar = *locinput; |
586 | break; |
587 | case ANYOF: |
a687059c |
588 | s = OPERAND(scan); |
589 | if (nextchar < 0) |
590 | nextchar = UCHARAT(locinput); |
591 | if (s[nextchar >> 3] & (1 << (nextchar&7))) |
592 | return(0); |
34de22dd |
593 | if (!nextchar && locinput >= regeol) |
a687059c |
594 | return 0; |
34de22dd |
595 | nextchar = *++locinput; |
a687059c |
596 | break; |
597 | case ALNUM: |
598 | if (!nextchar) |
599 | return(0); |
ac58e20f |
600 | if (!isALNUM(nextchar)) |
a687059c |
601 | return(0); |
602 | nextchar = *++locinput; |
603 | break; |
604 | case NALNUM: |
605 | if (!nextchar && locinput >= regeol) |
606 | return(0); |
ac58e20f |
607 | if (isALNUM(nextchar)) |
a687059c |
608 | return(0); |
609 | nextchar = *++locinput; |
610 | break; |
611 | case NBOUND: |
612 | case BOUND: |
613 | if (locinput == regbol) /* was last char in word? */ |
ac58e20f |
614 | ln = isALNUM(regprev); |
a687059c |
615 | else |
ac58e20f |
616 | ln = isALNUM(locinput[-1]); |
617 | n = isALNUM(nextchar); /* is next char in word? */ |
a687059c |
618 | if ((ln == n) == (OP(scan) == BOUND)) |
619 | return(0); |
620 | break; |
621 | case SPACE: |
622 | if (!nextchar && locinput >= regeol) |
623 | return(0); |
ac58e20f |
624 | if (!isSPACE(nextchar)) |
a687059c |
625 | return(0); |
626 | nextchar = *++locinput; |
627 | break; |
628 | case NSPACE: |
629 | if (!nextchar) |
630 | return(0); |
ac58e20f |
631 | if (isSPACE(nextchar)) |
a687059c |
632 | return(0); |
633 | nextchar = *++locinput; |
634 | break; |
635 | case DIGIT: |
ac58e20f |
636 | if (!isDIGIT(nextchar)) |
a687059c |
637 | return(0); |
638 | nextchar = *++locinput; |
639 | break; |
640 | case NDIGIT: |
641 | if (!nextchar && locinput >= regeol) |
642 | return(0); |
ac58e20f |
643 | if (isDIGIT(nextchar)) |
a687059c |
644 | return(0); |
645 | nextchar = *++locinput; |
646 | break; |
647 | case REF: |
fe14fcc3 |
648 | n = ARG1(scan); /* which paren pair */ |
a687059c |
649 | s = regmystartp[n]; |
650 | if (!s) |
651 | return(0); |
652 | if (!regmyendp[n]) |
653 | return(0); |
654 | if (s == regmyendp[n]) |
655 | break; |
656 | /* Inline the first character, for speed. */ |
657 | if (*s != nextchar) |
658 | return(0); |
659 | ln = regmyendp[n] - s; |
660 | if (locinput + ln > regeol) |
661 | return 0; |
662 | if (ln > 1 && bcmp(s, locinput, ln) != 0) |
663 | return(0); |
664 | locinput += ln; |
665 | nextchar = *locinput; |
666 | break; |
667 | |
668 | case NOTHING: |
669 | break; |
670 | case BACK: |
671 | break; |
fe14fcc3 |
672 | case OPEN: |
673 | n = ARG1(scan); /* which paren pair */ |
a687059c |
674 | reginput = locinput; |
675 | |
676 | regmystartp[n] = locinput; /* for REF */ |
677 | if (regmatch(next)) { |
678 | /* |
679 | * Don't set startp if some later |
680 | * invocation of the same parentheses |
681 | * already has. |
682 | */ |
683 | if (regstartp[n] == NULL) |
684 | regstartp[n] = locinput; |
685 | return(1); |
686 | } else |
687 | return(0); |
688 | /* NOTREACHED */ |
fe14fcc3 |
689 | case CLOSE: { |
690 | n = ARG1(scan); /* which paren pair */ |
a687059c |
691 | reginput = locinput; |
692 | |
693 | regmyendp[n] = locinput; /* for REF */ |
694 | if (regmatch(next)) { |
695 | /* |
696 | * Don't set endp if some later |
697 | * invocation of the same parentheses |
698 | * already has. |
699 | */ |
700 | if (regendp[n] == NULL) { |
701 | regendp[n] = locinput; |
702 | if (n > *reglastparen) |
703 | *reglastparen = n; |
704 | } |
705 | return(1); |
706 | } else |
707 | return(0); |
708 | } |
709 | /*NOTREACHED*/ |
710 | case BRANCH: { |
711 | if (OP(next) != BRANCH) /* No choice. */ |
712 | next = NEXTOPER(scan); /* Avoid recursion. */ |
713 | else { |
714 | do { |
715 | reginput = locinput; |
716 | if (regmatch(NEXTOPER(scan))) |
717 | return(1); |
718 | #ifdef REGALIGN |
719 | if (n = NEXT(scan)) |
720 | scan += n; |
721 | else |
722 | scan = NULL; |
723 | #else |
724 | scan = regnext(scan); |
725 | #endif |
726 | } while (scan != NULL && OP(scan) == BRANCH); |
727 | return(0); |
728 | /* NOTREACHED */ |
729 | } |
730 | } |
731 | break; |
00bf170e |
732 | case CURLY: |
733 | ln = ARG1(scan); /* min to match */ |
734 | n = ARG2(scan); /* max to match */ |
735 | scan = NEXTOPER(scan) + 4; |
736 | goto repeat; |
a687059c |
737 | case STAR: |
00bf170e |
738 | ln = 0; |
739 | n = 0; |
740 | scan = NEXTOPER(scan); |
741 | goto repeat; |
a687059c |
742 | case PLUS: |
743 | /* |
744 | * Lookahead to avoid useless match attempts |
745 | * when we know what character comes next. |
746 | */ |
00bf170e |
747 | ln = 1; |
748 | n = 0; |
749 | scan = NEXTOPER(scan); |
750 | repeat: |
a687059c |
751 | if (OP(next) == EXACTLY) |
752 | nextchar = *(OPERAND(next)+1); |
753 | else |
754 | nextchar = -1000; |
a687059c |
755 | reginput = locinput; |
00bf170e |
756 | n = regrepeat(scan, n); |
0a12ae7d |
757 | if (!multiline && OP(next) == EOL && ln < n) |
00bf170e |
758 | ln = n; /* why back off? */ |
a687059c |
759 | while (n >= ln) { |
760 | /* If it could work, try it. */ |
761 | if (nextchar == -1000 || *reginput == nextchar) |
762 | if (regmatch(next)) |
763 | return(1); |
764 | /* Couldn't or didn't -- back up. */ |
765 | n--; |
766 | reginput = locinput + n; |
767 | } |
768 | return(0); |
769 | case END: |
770 | reginput = locinput; /* put where regtry can find it */ |
771 | return(1); /* Success! */ |
772 | default: |
773 | printf("%x %d\n",scan,scan[1]); |
774 | FAIL("regexp memory corruption"); |
775 | } |
776 | |
777 | scan = next; |
778 | } |
779 | |
780 | /* |
781 | * We get here only if there's trouble -- normally "case END" is |
782 | * the terminating point. |
783 | */ |
784 | FAIL("corrupted regexp pointers"); |
785 | /*NOTREACHED*/ |
786 | #ifdef lint |
787 | return 0; |
788 | #endif |
789 | } |
790 | |
791 | /* |
792 | - regrepeat - repeatedly match something simple, report how many |
793 | */ |
794 | /* |
795 | * [This routine now assumes that it will only match on things of length 1. |
796 | * That was true before, but now we assume scan - reginput is the count, |
797 | * rather than incrementing count on every character.] |
798 | */ |
799 | static int |
00bf170e |
800 | regrepeat(p, max) |
a687059c |
801 | char *p; |
00bf170e |
802 | int max; |
a687059c |
803 | { |
804 | register char *scan; |
805 | register char *opnd; |
806 | register int c; |
807 | register char *loceol = regeol; |
808 | |
809 | scan = reginput; |
00bf170e |
810 | if (max && max < loceol - scan) |
811 | loceol = scan + max; |
a687059c |
812 | opnd = OPERAND(p); |
813 | switch (OP(p)) { |
814 | case ANY: |
815 | while (scan < loceol && *scan != '\n') |
816 | scan++; |
817 | break; |
818 | case EXACTLY: /* length of string is 1 */ |
819 | opnd++; |
820 | while (scan < loceol && *opnd == *scan) |
821 | scan++; |
822 | break; |
823 | case ANYOF: |
a687059c |
824 | c = UCHARAT(scan); |
825 | while (scan < loceol && !(opnd[c >> 3] & (1 << (c & 7)))) { |
826 | scan++; |
827 | c = UCHARAT(scan); |
828 | } |
829 | break; |
830 | case ALNUM: |
0a12ae7d |
831 | while (scan < loceol && isALNUM(*scan)) |
a687059c |
832 | scan++; |
833 | break; |
834 | case NALNUM: |
ac58e20f |
835 | while (scan < loceol && !isALNUM(*scan)) |
a687059c |
836 | scan++; |
837 | break; |
838 | case SPACE: |
ac58e20f |
839 | while (scan < loceol && isSPACE(*scan)) |
a687059c |
840 | scan++; |
841 | break; |
842 | case NSPACE: |
ac58e20f |
843 | while (scan < loceol && !isSPACE(*scan)) |
a687059c |
844 | scan++; |
845 | break; |
846 | case DIGIT: |
0a12ae7d |
847 | while (scan < loceol && isDIGIT(*scan)) |
a687059c |
848 | scan++; |
849 | break; |
850 | case NDIGIT: |
ac58e20f |
851 | while (scan < loceol && !isDIGIT(*scan)) |
a687059c |
852 | scan++; |
853 | break; |
854 | default: /* Oh dear. Called inappropriately. */ |
855 | FAIL("internal regexp foulup"); |
856 | /* NOTREACHED */ |
857 | } |
858 | |
859 | c = scan - reginput; |
860 | reginput = scan; |
861 | |
862 | return(c); |
863 | } |
864 | |
865 | /* |
866 | - regnext - dig the "next" pointer out of a node |
867 | * |
868 | * [Note, when REGALIGN is defined there are two places in regmatch() |
869 | * that bypass this code for speed.] |
870 | */ |
871 | char * |
872 | regnext(p) |
873 | register char *p; |
874 | { |
875 | register int offset; |
876 | |
877 | if (p == ®dummy) |
878 | return(NULL); |
879 | |
880 | offset = NEXT(p); |
881 | if (offset == 0) |
882 | return(NULL); |
883 | |
884 | #ifdef REGALIGN |
885 | return(p+offset); |
886 | #else |
887 | if (OP(p) == BACK) |
888 | return(p-offset); |
889 | else |
890 | return(p+offset); |
891 | #endif |
892 | } |