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