?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
5158 sed dumps core in new multibyte code
[illumos-gate.git] / usr / src / lib / libc / port / locale / engine.c
blob9d9f33cd3f9e79ae7d3baad693424f1fc23a5c9c
1 /*
2 * Copyright 2010 Nexenta Systems, Inc. All rights reserved.
3 * Copyright 2012 Milan Jurik. All rights reserved.
4 * Copyright (c) 1992, 1993, 1994 Henry Spencer.
5 * Copyright (c) 1992, 1993, 1994
6 * The Regents of the University of California. All rights reserved.
7 *
8 * This code is derived from software contributed to Berkeley by
9 * Henry Spencer.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 */
35
36 /*
37 * The matching engine and friends. This file is #included by regexec.c
38 * after suitable #defines of a variety of macros used herein, so that
39 * different state representations can be used without duplicating masses
40 * of code.
41 */
42
43 #ifdef SNAMES
44 #define matcher smatcher
45 #define fast sfast
46 #define slow sslow
47 #define dissect sdissect
48 #define backref sbackref
49 #define step sstep
50 #define print sprint
51 #define at sat
52 #define match smat
53 #endif
54 #ifdef LNAMES
55 #define matcher lmatcher
56 #define fast lfast
57 #define slow lslow
58 #define dissect ldissect
59 #define backref lbackref
60 #define step lstep
61 #define print lprint
62 #define at lat
63 #define match lmat
64 #endif
65 #ifdef MNAMES
66 #define matcher mmatcher
67 #define fast mfast
68 #define slow mslow
69 #define dissect mdissect
70 #define backref mbackref
71 #define step mstep
72 #define print mprint
73 #define at mat
74 #define match mmat
75 #endif
76
77 /* another structure passed up and down to avoid zillions of parameters */
78 struct match {
79 struct re_guts *g;
80 int eflags;
81 regmatch_t *pmatch; /* [nsub+1] (0 element unused) */
82 const char *offp; /* offsets work from here */
83 const char *beginp; /* start of string -- virtual NUL precedes */
84 const char *endp; /* end of string -- virtual NUL here */
85 const char *coldp; /* can be no match starting before here */
86 const char **lastpos; /* [nplus+1] */
87 STATEVARS;
88 states st; /* current states */
89 states fresh; /* states for a fresh start */
90 states tmp; /* temporary */
91 states empty; /* empty set of states */
92 mbstate_t mbs; /* multibyte conversion state */
93 };
94
95 /* ========= begin header generated by ./mkh ========= */
96 #ifdef __cplusplus
97 extern "C" {
98 #endif
99
100 /* === engine.c === */
101 static int matcher(struct re_guts *, const char *, size_t, regmatch_t[], int);
102 static const char *dissect(struct match *, const char *, const char *,
103 sopno, sopno);
104 static const char *backref(struct match *, const char *, const char *, sopno,
105 sopno, sopno, int);
106 static const char *fast(struct match *, const char *, const char *, sopno,
107 sopno);
108 static const char *slow(struct match *, const char *, const char *, sopno,
109 sopno);
110 static states step(struct re_guts *, sopno, sopno, states, wint_t, states);
111 #define MAX_RECURSION 100
112 #define BOL (OUT-1)
113 #define EOL (BOL-1)
114 #define BOLEOL (BOL-2)
115 #define NOTHING (BOL-3)
116 #define BOW (BOL-4)
117 #define EOW (BOL-5)
118 #define BADCHAR (BOL-6)
119 #define NONCHAR(c) ((c) <= OUT)
120 #ifdef REDEBUG
121 static void print(struct match *, const char *, states, int, FILE *);
122 #endif
123 #ifdef REDEBUG
124 static void at(struct match *, const char *, const char *, const char *,
125 sopno, sopno);
126 #endif
127 #ifdef REDEBUG
128 static const char *pchar(int ch);
129 #endif
130
131 #ifdef __cplusplus
132 }
133 #endif
134 /* ========= end header generated by ./mkh ========= */
135
136 #ifdef REDEBUG
137 #define SP(t, s, c) print(m, t, s, c, stdout)
138 #define AT(t, p1, p2, s1, s2) at(m, t, p1, p2, s1, s2)
139 #define NOTE(str) { if (m->eflags&REG_TRACE) printf("=%s\n", (str)); }
140 #else
141 #define SP(t, s, c) /* nothing */
142 #define AT(t, p1, p2, s1, s2) /* nothing */
143 #define NOTE(s) /* nothing */
144 #endif
145
146 /*
147 * matcher - the actual matching engine
148 */
149 static int /* 0 success, REG_NOMATCH failure */
150 matcher(struct re_guts *g,
151 const char *string,
152 size_t nmatch,
153 regmatch_t pmatch[],
154 int eflags)
155 {
156 const char *endp;
157 int i;
158 struct match mv;
159 struct match *m = &mv;
160 const char *dp;
161 const sopno gf = g->firststate+1; /* +1 for OEND */
162 const sopno gl = g->laststate;
163 const char *start;
164 const char *stop;
165 /* Boyer-Moore algorithms variables */
166 const char *pp;
167 int cj, mj;
168 const char *mustfirst;
169 const char *mustlast;
170 int *matchjump;
171 int *charjump;
172
173 /* simplify the situation where possible */
174 if (g->cflags&REG_NOSUB)
175 nmatch = 0;
176
177 if (eflags&REG_STARTEND) {
178 start = string + pmatch[0].rm_so;
179 stop = string + pmatch[0].rm_eo;
180 } else {
181 start = string;
182 stop = start + strlen(start);
183 }
184
185 if (stop < start)
186 return (REG_EFATAL);
187
188 /* prescreening; this does wonders for this rather slow code */
189 if (g->must != NULL) {
190 if (g->charjump != NULL && g->matchjump != NULL) {
191 mustfirst = g->must;
192 mustlast = g->must + g->mlen - 1;
193 charjump = g->charjump;
194 matchjump = g->matchjump;
195 pp = mustlast;
196 for (dp = start+g->mlen-1; dp < stop; ) {
197 /* Fast skip non-matches */
198 while (dp < stop && charjump[(int)*dp])
199 dp += charjump[(int)*dp];
200
201 if (dp >= stop)
202 break;
203
204 /* Greedy matcher */
205 /*
206 * We depend on not being used for
207 * for strings of length 1
208 */
209 while (*--dp == *--pp && pp != mustfirst)
210 ;
211
212 if (*dp == *pp)
213 break;
214
215 /* Jump to next possible match */
216 mj = matchjump[pp - mustfirst];
217 cj = charjump[(int)*dp];
218 dp += (cj < mj ? mj : cj);
219 pp = mustlast;
220 }
221 if (pp != mustfirst)
222 return (REG_NOMATCH);
223 } else {
224 for (dp = start; dp < stop; dp++)
225 if (*dp == g->must[0] &&
226 stop - dp >= g->mlen &&
227 memcmp(dp, g->must, (size_t)g->mlen) == 0)
228 break;
229 if (dp == stop) /* we didn't find g->must */
230 return (REG_NOMATCH);
231 }
232 }
233
234 /* match struct setup */
235 m->g = g;
236 m->eflags = eflags;
237 m->pmatch = NULL;
238 m->lastpos = NULL;
239 m->offp = string;
240 m->beginp = start;
241 m->endp = stop;
242 STATESETUP(m, 4);
243 SETUP(m->st);
244 SETUP(m->fresh);
245 SETUP(m->tmp);
246 SETUP(m->empty);
247 CLEAR(m->empty);
248 ZAPSTATE(&m->mbs);
249
250 /* Adjust start according to moffset, to speed things up */
251 if (g->moffset > -1)
252 start = ((dp - g->moffset) < start) ? start : dp - g->moffset;
253
254 SP("mloop", m->st, *start);
255
256 /* this loop does only one repetition except for backrefs */
257 for (;;) {
258 endp = fast(m, start, stop, gf, gl);
259 if (endp == NULL) { /* a miss */
260 if (m->pmatch != NULL)
261 free((char *)m->pmatch);
262 if (m->lastpos != NULL)
263 free((char *)m->lastpos);
264 STATETEARDOWN(m);
265 return (REG_NOMATCH);
266 }
267 if (nmatch == 0 && !g->backrefs)
268 break; /* no further info needed */
269
270 /* where? */
271 assert(m->coldp != NULL);
272 for (;;) {
273 NOTE("finding start");
274 endp = slow(m, m->coldp, stop, gf, gl);
275 if (endp != NULL)
276 break;
277 assert(m->coldp < m->endp);
278 m->coldp += XMBRTOWC(NULL, m->coldp,
279 m->endp - m->coldp, &m->mbs, 0);
280 }
281 if (nmatch == 1 && !g->backrefs)
282 break; /* no further info needed */
283
284 /* oh my, he wants the subexpressions... */
285 if (m->pmatch == NULL)
286 m->pmatch = (regmatch_t *)malloc((m->g->nsub + 1) *
287 sizeof (regmatch_t));
288 if (m->pmatch == NULL) {
289 STATETEARDOWN(m);
290 return (REG_ESPACE);
291 }
292 for (i = 1; i <= m->g->nsub; i++)
293 m->pmatch[i].rm_so = m->pmatch[i].rm_eo = -1;
294 /* NB: FreeBSD has REG_BACKR, we do not */
295 if (!g->backrefs /* && !(m->eflags&REG_BACKR) */) {
296 NOTE("dissecting");
297 dp = dissect(m, m->coldp, endp, gf, gl);
298 } else {
299 if (g->nplus > 0 && m->lastpos == NULL)
300 m->lastpos = malloc((g->nplus+1) *
301 sizeof (const char *));
302 if (g->nplus > 0 && m->lastpos == NULL) {
303 free(m->pmatch);
304 STATETEARDOWN(m);
305 return (REG_ESPACE);
306 }
307 NOTE("backref dissect");
308 dp = backref(m, m->coldp, endp, gf, gl, (sopno)0, 0);
309 }
310 if (dp != NULL)
311 break;
312
313 /* uh-oh... we couldn't find a subexpression-level match */
314 assert(g->backrefs); /* must be back references doing it */
315 assert(g->nplus == 0 || m->lastpos != NULL);
316 for (;;) {
317 if (dp != NULL || endp <= m->coldp)
318 break; /* defeat */
319 NOTE("backoff");
320 endp = slow(m, m->coldp, endp-1, gf, gl);
321 if (endp == NULL)
322 break; /* defeat */
323 /* try it on a shorter possibility */
324 #ifndef NDEBUG
325 for (i = 1; i <= m->g->nsub; i++) {
326 assert(m->pmatch[i].rm_so == -1);
327 assert(m->pmatch[i].rm_eo == -1);
328 }
329 #endif
330 NOTE("backoff dissect");
331 dp = backref(m, m->coldp, endp, gf, gl, (sopno)0, 0);
332 }
333 assert(dp == NULL || dp == endp);
334 if (dp != NULL) /* found a shorter one */
335 break;
336
337 /* despite initial appearances, there is no match here */
338 NOTE("false alarm");
339 /* recycle starting later */
340 start = m->coldp + XMBRTOWC(NULL, m->coldp,
341 stop - m->coldp, &m->mbs, 0);
342 assert(start <= stop);
343 }
344
345 /* fill in the details if requested */
346 if (nmatch > 0) {
347 pmatch[0].rm_so = m->coldp - m->offp;
348 pmatch[0].rm_eo = endp - m->offp;
349 }
350 if (nmatch > 1) {
351 assert(m->pmatch != NULL);
352 for (i = 1; i < nmatch; i++)
353 if (i <= m->g->nsub)
354 pmatch[i] = m->pmatch[i];
355 else {
356 pmatch[i].rm_so = -1;
357 pmatch[i].rm_eo = -1;
358 }
359 }
360
361 if (m->pmatch != NULL)
362 free((char *)m->pmatch);
363 if (m->lastpos != NULL)
364 free((char *)m->lastpos);
365 STATETEARDOWN(m);
366 return (0);
367 }
368
369 /*
370 * dissect - figure out what matched what, no back references
371 */
372 static const char *
373 dissect(struct match *m, const char *start, const char *stop, sopno startst,
374 sopno stopst)
375 {
376 int i;
377 sopno ss; /* start sop of current subRE */
378 sopno es; /* end sop of current subRE */
379 const char *sp; /* start of string matched by it */
380 const char *stp; /* string matched by it cannot pass here */
381 const char *rest; /* start of rest of string */
382 const char *tail; /* string unmatched by rest of RE */
383 sopno ssub; /* start sop of subsubRE */
384 sopno esub; /* end sop of subsubRE */
385 const char *ssp; /* start of string matched by subsubRE */
386 const char *sep; /* end of string matched by subsubRE */
387 const char *oldssp; /* previous ssp */
388 const char *dp;
389
390 AT("diss", start, stop, startst, stopst);
391 sp = start;
392 for (ss = startst; ss < stopst; ss = es) {
393 /* identify end of subRE */
394 es = ss;
395 switch (OP(m->g->strip[es])) {
396 case OPLUS_:
397 case OQUEST_:
398 es += OPND(m->g->strip[es]);
399 break;
400 case OCH_:
401 while (OP(m->g->strip[es]) != O_CH)
402 es += OPND(m->g->strip[es]);
403 break;
404 }
405 es++;
406
407 /* figure out what it matched */
408 switch (OP(m->g->strip[ss])) {
409 case OEND:
410 assert(0);
411 break;
412 case OCHAR:
413 sp += XMBRTOWC(NULL, sp, stop - start, &m->mbs, 0);
414 break;
415 case OBOL:
416 case OEOL:
417 case OBOW:
418 case OEOW:
419 break;
420 case OANY:
421 case OANYOF:
422 sp += XMBRTOWC(NULL, sp, stop - start, &m->mbs, 0);
423 break;
424 case OBACK_:
425 case O_BACK:
426 assert(0);
427 break;
428 /* cases where length of match is hard to find */
429 case OQUEST_:
430 stp = stop;
431 for (;;) {
432 /* how long could this one be? */
433 rest = slow(m, sp, stp, ss, es);
434 assert(rest != NULL); /* it did match */
435 /* could the rest match the rest? */
436 tail = slow(m, rest, stop, es, stopst);
437 if (tail == stop)
438 break; /* yes! */
439 /* no -- try a shorter match for this one */
440 stp = rest - 1;
441 assert(stp >= sp); /* it did work */
442 }
443 ssub = ss + 1;
444 esub = es - 1;
445 /* did innards match? */
446 if (slow(m, sp, rest, ssub, esub) != NULL) {
447 dp = dissect(m, sp, rest, ssub, esub);
448 assert(dp == rest);
449 #if defined(__lint)
450 (void) dp;
451 #endif
452 } else /* no */
453 assert(sp == rest);
454 sp = rest;
455 break;
456 case OPLUS_:
457 stp = stop;
458 for (;;) {
459 /* how long could this one be? */
460 rest = slow(m, sp, stp, ss, es);
461 assert(rest != NULL); /* it did match */
462 /* could the rest match the rest? */
463 tail = slow(m, rest, stop, es, stopst);
464 if (tail == stop)
465 break; /* yes! */
466 /* no -- try a shorter match for this one */
467 stp = rest - 1;
468 assert(stp >= sp); /* it did work */
469 }
470 ssub = ss + 1;
471 esub = es - 1;
472 ssp = sp;
473 oldssp = ssp;
474 for (;;) { /* find last match of innards */
475 sep = slow(m, ssp, rest, ssub, esub);
476 if (sep == NULL || sep == ssp)
477 break; /* failed or matched null */
478 oldssp = ssp; /* on to next try */
479 ssp = sep;
480 }
481 if (sep == NULL) {
482 /* last successful match */
483 sep = ssp;
484 ssp = oldssp;
485 }
486 assert(sep == rest); /* must exhaust substring */
487 assert(slow(m, ssp, sep, ssub, esub) == rest);
488 dp = dissect(m, ssp, sep, ssub, esub);
489 assert(dp == sep);
490 sp = rest;
491 break;
492 case OCH_:
493 stp = stop;
494 for (;;) {
495 /* how long could this one be? */
496 rest = slow(m, sp, stp, ss, es);
497 assert(rest != NULL); /* it did match */
498 /* could the rest match the rest? */
499 tail = slow(m, rest, stop, es, stopst);
500 if (tail == stop)
501 break; /* yes! */
502 /* no -- try a shorter match for this one */
503 stp = rest - 1;
504 assert(stp >= sp); /* it did work */
505 }
506 ssub = ss + 1;
507 esub = ss + OPND(m->g->strip[ss]) - 1;
508 assert(OP(m->g->strip[esub]) == OOR1);
509 for (;;) { /* find first matching branch */
510 if (slow(m, sp, rest, ssub, esub) == rest)
511 break; /* it matched all of it */
512 /* that one missed, try next one */
513 assert(OP(m->g->strip[esub]) == OOR1);
514 esub++;
515 assert(OP(m->g->strip[esub]) == OOR2);
516 ssub = esub + 1;
517 esub += OPND(m->g->strip[esub]);
518 if (OP(m->g->strip[esub]) == OOR2)
519 esub--;
520 else
521 assert(OP(m->g->strip[esub]) == O_CH);
522 }
523 dp = dissect(m, sp, rest, ssub, esub);
524 assert(dp == rest);
525 sp = rest;
526 break;
527 case O_PLUS:
528 case O_QUEST:
529 case OOR1:
530 case OOR2:
531 case O_CH:
532 assert(0);
533 break;
534 case OLPAREN:
535 i = OPND(m->g->strip[ss]);
536 assert(0 < i && i <= m->g->nsub);
537 m->pmatch[i].rm_so = sp - m->offp;
538 break;
539 case ORPAREN:
540 i = OPND(m->g->strip[ss]);
541 assert(0 < i && i <= m->g->nsub);
542 m->pmatch[i].rm_eo = sp - m->offp;
543 break;
544 default: /* uh oh */
545 assert(0);
546 break;
547 }
548 }
549
550 assert(sp == stop);
551 return (sp);
552 }
553
554 /*
555 * backref - figure out what matched what, figuring in back references
556 */
557 static const char *
558 backref(struct match *m, const char *start, const char *stop, sopno startst,
559 sopno stopst, sopno lev, /* PLUS nesting level */
560 int rec)
561 {
562 int i;
563 sopno ss; /* start sop of current subRE */
564 const char *sp; /* start of string matched by it */
565 sopno ssub; /* start sop of subsubRE */
566 sopno esub; /* end sop of subsubRE */
567 const char *ssp; /* start of string matched by subsubRE */
568 const char *dp;
569 size_t len;
570 int hard;
571 sop s;
572 regoff_t offsave;
573 cset *cs;
574 wint_t wc;
575
576 AT("back", start, stop, startst, stopst);
577 sp = start;
578
579 /* get as far as we can with easy stuff */
580 hard = 0;
581 for (ss = startst; !hard && ss < stopst; ss++)
582 switch (OP(s = m->g->strip[ss])) {
583 case OCHAR:
584 if (sp == stop)
585 return (NULL);
586 sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
587 if (wc != OPND(s))
588 return (NULL);
589 break;
590 case OANY:
591 if (sp == stop)
592 return (NULL);
593 sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
594 if (wc == BADCHAR)
595 return (NULL);
596 break;
597 case OANYOF:
598 if (sp == stop)
599 return (NULL);
600 cs = &m->g->sets[OPND(s)];
601 sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
602 if (wc == BADCHAR || !CHIN(cs, wc))
603 return (NULL);
604 break;
605 case OBOL:
606 if ((sp == m->beginp && !(m->eflags&REG_NOTBOL)) ||
607 (sp < m->endp && *(sp-1) == '\n' &&
608 (m->g->cflags&REG_NEWLINE))) {
609 break;
610 }
611 return (NULL);
612 case OEOL:
613 if ((sp == m->endp && !(m->eflags&REG_NOTEOL)) ||
614 (sp < m->endp && *sp == '\n' &&
615 (m->g->cflags&REG_NEWLINE))) {
616 break;
617 }
618 return (NULL);
619 case OBOW:
620 if (((sp == m->beginp && !(m->eflags&REG_NOTBOL)) ||
621 (sp < m->endp && *(sp-1) == '\n' &&
622 (m->g->cflags&REG_NEWLINE)) ||
623 (sp > m->beginp && !ISWORD(*(sp-1)))) &&
624 (sp < m->endp && ISWORD(*sp))) {
625 break;
626 }
627 return (NULL);
628 case OEOW:
629 if (((sp == m->endp && !(m->eflags&REG_NOTEOL)) ||
630 (sp < m->endp && *sp == '\n' &&
631 (m->g->cflags&REG_NEWLINE)) ||
632 (sp < m->endp && !ISWORD(*sp))) &&
633 (sp > m->beginp && ISWORD(*(sp-1)))) {
634 break;
635 }
636 return (NULL);
637 case O_QUEST:
638 break;
639 case OOR1: /* matches null but needs to skip */
640 ss++;
641 s = m->g->strip[ss];
642 do {
643 assert(OP(s) == OOR2);
644 ss += OPND(s);
645 } while (OP(s = m->g->strip[ss]) != O_CH);
646 /* note that the ss++ gets us past the O_CH */
647 break;
648 default: /* have to make a choice */
649 hard = 1;
650 break;
651 }
652 if (!hard) { /* that was it! */
653 if (sp != stop)
654 return (NULL);
655 return (sp);
656 }
657 ss--; /* adjust for the for's final increment */
658
659 /* the hard stuff */
660 AT("hard", sp, stop, ss, stopst);
661 s = m->g->strip[ss];
662 switch (OP(s)) {
663 case OBACK_: /* the vilest depths */
664 i = OPND(s);
665 assert(0 < i && i <= m->g->nsub);
666 if (m->pmatch[i].rm_eo == -1)
667 return (NULL);
668 assert(m->pmatch[i].rm_so != -1);
669 len = m->pmatch[i].rm_eo - m->pmatch[i].rm_so;
670 if (len == 0 && rec++ > MAX_RECURSION)
671 return (NULL);
672 assert(stop - m->beginp >= len);
673 if (sp > stop - len)
674 return (NULL); /* not enough left to match */
675 ssp = m->offp + m->pmatch[i].rm_so;
676 if (memcmp(sp, ssp, len) != 0)
677 return (NULL);
678 while (m->g->strip[ss] != SOP(O_BACK, i))
679 ss++;
680 return (backref(m, sp+len, stop, ss+1, stopst, lev, rec));
681 case OQUEST_: /* to null or not */
682 dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
683 if (dp != NULL)
684 return (dp); /* not */
685 return (backref(m, sp, stop, ss+OPND(s)+1, stopst, lev, rec));
686 case OPLUS_:
687 assert(m->lastpos != NULL);
688 assert(lev+1 <= m->g->nplus);
689 m->lastpos[lev+1] = sp;
690 return (backref(m, sp, stop, ss+1, stopst, lev+1, rec));
691 case O_PLUS:
692 if (sp == m->lastpos[lev]) /* last pass matched null */
693 return (backref(m, sp, stop, ss+1, stopst, lev-1, rec));
694 /* try another pass */
695 m->lastpos[lev] = sp;
696 dp = backref(m, sp, stop, ss-OPND(s)+1, stopst, lev, rec);
697 if (dp == NULL)
698 return (backref(m, sp, stop, ss+1, stopst, lev-1, rec));
699 return (dp);
700 case OCH_: /* find the right one, if any */
701 ssub = ss + 1;
702 esub = ss + OPND(s) - 1;
703 assert(OP(m->g->strip[esub]) == OOR1);
704 for (;;) { /* find first matching branch */
705 dp = backref(m, sp, stop, ssub, esub, lev, rec);
706 if (dp != NULL)
707 return (dp);
708 /* that one missed, try next one */
709 if (OP(m->g->strip[esub]) == O_CH)
710 return (NULL); /* there is none */
711 esub++;
712 assert(OP(m->g->strip[esub]) == OOR2);
713 ssub = esub + 1;
714 esub += OPND(m->g->strip[esub]);
715 if (OP(m->g->strip[esub]) == OOR2)
716 esub--;
717 else
718 assert(OP(m->g->strip[esub]) == O_CH);
719 }
720 /* NOTREACHED */
721 break;
722 case OLPAREN: /* must undo assignment if rest fails */
723 i = OPND(s);
724 assert(0 < i && i <= m->g->nsub);
725 offsave = m->pmatch[i].rm_so;
726 m->pmatch[i].rm_so = sp - m->offp;
727 dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
728 if (dp != NULL)
729 return (dp);
730 m->pmatch[i].rm_so = offsave;
731 return (NULL);
732 case ORPAREN: /* must undo assignment if rest fails */
733 i = OPND(s);
734 assert(0 < i && i <= m->g->nsub);
735 offsave = m->pmatch[i].rm_eo;
736 m->pmatch[i].rm_eo = sp - m->offp;
737 dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
738 if (dp != NULL)
739 return (dp);
740 m->pmatch[i].rm_eo = offsave;
741 return (NULL);
742 default: /* uh oh */
743 assert(0);
744 break;
745 }
746
747 /* "can't happen" */
748 assert(0);
749 return (NULL);
750 }
751
752 /*
753 * fast - step through the string at top speed
754 */
755 static const char *
756 fast(struct match *m, const char *start, const char *stop, sopno startst,
757 sopno stopst)
758 {
759 states st = m->st;
760 states fresh = m->fresh;
761 states tmp = m->tmp;
762 const char *p = start;
763 wint_t c;
764 wint_t lastc; /* previous c */
765 wint_t flagch;
766 int i;
767 const char *coldp; /* last p after which no match was underway */
768 size_t clen;
769
770 CLEAR(st);
771 SET1(st, startst);
772 SP("fast", st, *p);
773 st = step(m->g, startst, stopst, st, NOTHING, st);
774 ASSIGN(fresh, st);
775 SP("start", st, *p);
776 coldp = NULL;
777 if (start == m->beginp)
778 c = OUT;
779 else {
780 /*
781 * XXX Wrong if the previous character was multi-byte.
782 * Newline never is (in encodings supported by FreeBSD),
783 * so this only breaks the ISWORD tests below.
784 */
785 c = (uch)*(start - 1);
786 }
787 for (;;) {
788 /* next character */
789 lastc = c;
790 if (p == m->endp) {
791 clen = 0;
792 c = OUT;
793 } else
794 clen = XMBRTOWC(&c, p, m->endp - p, &m->mbs, BADCHAR);
795 if (EQ(st, fresh))
796 coldp = p;
797
798 /* is there an EOL and/or BOL between lastc and c? */
799 flagch = '\0';
800 i = 0;
801 if ((lastc == '\n' && m->g->cflags&REG_NEWLINE) ||
802 (lastc == OUT && !(m->eflags&REG_NOTBOL))) {
803 flagch = BOL;
804 i = m->g->nbol;
805 }
806 if ((c == '\n' && m->g->cflags&REG_NEWLINE) ||
807 (c == OUT && !(m->eflags&REG_NOTEOL))) {
808 flagch = (flagch == BOL) ? BOLEOL : EOL;
809 i += m->g->neol;
810 }
811 if (i != 0) {
812 for (; i > 0; i--)
813 st = step(m->g, startst, stopst, st,
814 flagch, st);
815 SP("boleol", st, c);
816 }
817
818 /* how about a word boundary? */
819 if ((flagch == BOL || (lastc != OUT && !ISWORD(lastc))) &&
820 (c != OUT && ISWORD(c))) {
821 flagch = BOW;
822 }
823 if ((lastc != OUT && ISWORD(lastc)) &&
824 (flagch == EOL || (c != OUT && !ISWORD(c)))) {
825 flagch = EOW;
826 }
827 if (flagch == BOW || flagch == EOW) {
828 st = step(m->g, startst, stopst, st, flagch, st);
829 SP("boweow", st, c);
830 }
831
832 /* are we done? */
833 if (ISSET(st, stopst) || p == stop || clen > stop - p)
834 break; /* NOTE BREAK OUT */
835
836 /* no, we must deal with this character */
837 ASSIGN(tmp, st);
838 ASSIGN(st, fresh);
839 assert(c != OUT);
840 st = step(m->g, startst, stopst, tmp, c, st);
841 SP("aft", st, c);
842 assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st));
843 p += clen;
844 }
845
846 assert(coldp != NULL);
847 m->coldp = coldp;
848 if (ISSET(st, stopst))
849 return (p+XMBRTOWC(NULL, p, stop - p, &m->mbs, 0));
850 else
851 return (NULL);
852 }
853
854 /*
855 * slow - step through the string more deliberately
856 */
857 static const char *
858 slow(struct match *m, const char *start, const char *stop, sopno startst,
859 sopno stopst)
860 {
861 states st = m->st;
862 states empty = m->empty;
863 states tmp = m->tmp;
864 const char *p = start;
865 wint_t c;
866 wint_t lastc; /* previous c */
867 wint_t flagch;
868 int i;
869 const char *matchp; /* last p at which a match ended */
870 size_t clen;
871
872 AT("slow", start, stop, startst, stopst);
873 CLEAR(st);
874 SET1(st, startst);
875 SP("sstart", st, *p);
876 st = step(m->g, startst, stopst, st, NOTHING, st);
877 matchp = NULL;
878 if (start == m->beginp)
879 c = OUT;
880 else {
881 /*
882 * XXX Wrong if the previous character was multi-byte.
883 * Newline never is (in encodings supported by FreeBSD),
884 * so this only breaks the ISWORD tests below.
885 */
886 c = (uch)*(start - 1);
887 }
888 for (;;) {
889 /* next character */
890 lastc = c;
891 if (p == m->endp) {
892 c = OUT;
893 clen = 0;
894 } else
895 clen = XMBRTOWC(&c, p, m->endp - p, &m->mbs, BADCHAR);
896
897 /* is there an EOL and/or BOL between lastc and c? */
898 flagch = '\0';
899 i = 0;
900 if ((lastc == '\n' && m->g->cflags&REG_NEWLINE) ||
901 (lastc == OUT && !(m->eflags&REG_NOTBOL))) {
902 flagch = BOL;
903 i = m->g->nbol;
904 }
905 if ((c == '\n' && m->g->cflags&REG_NEWLINE) ||
906 (c == OUT && !(m->eflags&REG_NOTEOL))) {
907 flagch = (flagch == BOL) ? BOLEOL : EOL;
908 i += m->g->neol;
909 }
910 if (i != 0) {
911 for (; i > 0; i--)
912 st = step(m->g, startst, stopst, st,
913 flagch, st);
914 SP("sboleol", st, c);
915 }
916
917 /* how about a word boundary? */
918 if ((flagch == BOL || (lastc != OUT && !ISWORD(lastc))) &&
919 (c != OUT && ISWORD(c))) {
920 flagch = BOW;
921 }
922 if ((lastc != OUT && ISWORD(lastc)) &&
923 (flagch == EOL || (c != OUT && !ISWORD(c)))) {
924 flagch = EOW;
925 }
926 if (flagch == BOW || flagch == EOW) {
927 st = step(m->g, startst, stopst, st, flagch, st);
928 SP("sboweow", st, c);
929 }
930
931 /* are we done? */
932 if (ISSET(st, stopst))
933 matchp = p;
934 if (EQ(st, empty) || p == stop || clen > stop - p)
935 break; /* NOTE BREAK OUT */
936
937 /* no, we must deal with this character */
938 ASSIGN(tmp, st);
939 ASSIGN(st, empty);
940 assert(c != OUT);
941 st = step(m->g, startst, stopst, tmp, c, st);
942 SP("saft", st, c);
943 assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st));
944 p += clen;
945 }
946
947 return (matchp);
948 }
949
950
951 /*
952 * step - map set of states reachable before char to set reachable after
953 */
954 static states
955 step(struct re_guts *g,
956 sopno start, /* start state within strip */
957 sopno stop, /* state after stop state within strip */
958 states bef, /* states reachable before */
959 wint_t ch, /* character or NONCHAR code */
960 states aft) /* states already known reachable after */
961 {
962 cset *cs;
963 sop s;
964 sopno pc;
965 onestate here; /* note, macros know this name */
966 sopno look;
967 int i;
968
969 for (pc = start, INIT(here, pc); pc != stop; pc++, INC(here)) {
970 s = g->strip[pc];
971 switch (OP(s)) {
972 case OEND:
973 assert(pc == stop-1);
974 break;
975 case OCHAR:
976 /* only characters can match */
977 assert(!NONCHAR(ch) || ch != OPND(s));
978 if (ch == OPND(s))
979 FWD(aft, bef, 1);
980 break;
981 case OBOL:
982 if (ch == BOL || ch == BOLEOL)
983 FWD(aft, bef, 1);
984 break;
985 case OEOL:
986 if (ch == EOL || ch == BOLEOL)
987 FWD(aft, bef, 1);
988 break;
989 case OBOW:
990 if (ch == BOW)
991 FWD(aft, bef, 1);
992 break;
993 case OEOW:
994 if (ch == EOW)
995 FWD(aft, bef, 1);
996 break;
997 case OANY:
998 if (!NONCHAR(ch))
999 FWD(aft, bef, 1);
1000 break;
1001 case OANYOF:
1002 cs = &g->sets[OPND(s)];
1003 if (!NONCHAR(ch) && CHIN(cs, ch))
1004 FWD(aft, bef, 1);
1005 break;
1006 case OBACK_: /* ignored here */
1007 case O_BACK:
1008 FWD(aft, aft, 1);
1009 break;
1010 case OPLUS_: /* forward, this is just an empty */
1011 FWD(aft, aft, 1);
1012 break;
1013 case O_PLUS: /* both forward and back */
1014 FWD(aft, aft, 1);
1015 i = ISSETBACK(aft, OPND(s));
1016 BACK(aft, aft, OPND(s));
1017 if (!i && ISSETBACK(aft, OPND(s))) {
1018 /* oho, must reconsider loop body */
1019 pc -= OPND(s) + 1;
1020 INIT(here, pc);
1021 }
1022 break;
1023 case OQUEST_: /* two branches, both forward */
1024 FWD(aft, aft, 1);
1025 FWD(aft, aft, OPND(s));
1026 break;
1027 case O_QUEST: /* just an empty */
1028 FWD(aft, aft, 1);
1029 break;
1030 case OLPAREN: /* not significant here */
1031 case ORPAREN:
1032 FWD(aft, aft, 1);
1033 break;
1034 case OCH_: /* mark the first two branches */
1035 FWD(aft, aft, 1);
1036 assert(OP(g->strip[pc+OPND(s)]) == OOR2);
1037 FWD(aft, aft, OPND(s));
1038 break;
1039 case OOR1: /* done a branch, find the O_CH */
1040 if (ISSTATEIN(aft, here)) {
1041 for (look = 1;
1042 OP(s = g->strip[pc+look]) != O_CH;
1043 look += OPND(s))
1044 assert(OP(s) == OOR2);
1045 FWD(aft, aft, look + 1);
1046 }
1047 break;
1048 case OOR2: /* propagate OCH_'s marking */
1049 FWD(aft, aft, 1);
1050 if (OP(g->strip[pc+OPND(s)]) != O_CH) {
1051 assert(OP(g->strip[pc+OPND(s)]) == OOR2);
1052 FWD(aft, aft, OPND(s));
1053 }
1054 break;
1055 case O_CH: /* just empty */
1056 FWD(aft, aft, 1);
1057 break;
1058 default: /* ooooops... */
1059 assert(0);
1060 break;
1061 }
1062 }
1063
1064 return (aft);
1065 }
1066
1067 #ifdef REDEBUG
1068 /*
1069 * print - print a set of states
1070 */
1071 static void
1072 print(struct match *m, const char *caption, states st, int ch, FILE *d)
1073 {
1074 struct re_guts *g = m->g;
1075 int i;
1076 int first = 1;
1077
1078 if (!(m->eflags&REG_TRACE))
1079 return;
1080
1081 (void) fprintf(d, "%s", caption);
1082 if (ch != '\0')
1083 (void) fprintf(d, " %s", pchar(ch));
1084 for (i = 0; i < g->nstates; i++)
1085 if (ISSET(st, i)) {
1086 (void) fprintf(d, "%s%d", (first) ? "\t" : ", ", i);
1087 first = 0;
1088 }
1089 (void) fprintf(d, "\n");
1090 }
1091
1092 /*
1093 * at - print current situation
1094 */
1095 static void
1096 at(struct match *m, const char *title, const char *start, const char *stop,
1097 sopno startst, sopno stopst)
1098 {
1099 if (!(m->eflags&REG_TRACE))
1100 return;
1101
1102 (void) printf("%s %s-", title, pchar(*start));
1103 (void) printf("%s ", pchar(*stop));
1104 (void) printf("%ld-%ld\n", (long)startst, (long)stopst);
1105 }
1106
1107 #ifndef PCHARDONE
1108 #define PCHARDONE /* never again */
1109 /*
1110 * pchar - make a character printable
1111 *
1112 * Is this identical to regchar() over in debug.c? Well, yes. But a
1113 * duplicate here avoids having a debugging-capable regexec.o tied to
1114 * a matching debug.o, and this is convenient. It all disappears in
1115 * the non-debug compilation anyway, so it doesn't matter much.
1116 */
1117 static const char *
1118 pchar(int ch)
1119 {
1120 static char pbuf[10];
1121
1122 if (isprint((uch)ch) || ch == ' ')
1123 (void) sprintf(pbuf, "%c", ch);
1124 else
1125 (void) sprintf(pbuf, "\\%o", ch);
1126 return (pbuf);
1127 }
1128 #endif
1129 #endif
1130
1131 #undef matcher
1132 #undef fast
1133 #undef slow
1134 #undef dissect
1135 #undef backref
1136 #undef step
1137 #undef print
1138 #undef at
1139 #undef match
Community developed and maintained version of the OS/Net consolidation