2 * SPDX-License-Identifier: BSD-3-Clause
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.
8 * This code is derived from software contributed to Berkeley by
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
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 * 3. 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.
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
35 * @(#)engine.c 8.5 (Berkeley) 3/20/94
38 #include <sys/cdefs.h>
39 __FBSDID("$FreeBSD$");
44 * The matching engine and friends. This file is #included by regexec.c
45 * after suitable #defines of a variety of macros used herein, so that
46 * different state representations can be used without duplicating masses
51 #define stepback sstepback
52 #define matcher smatcher
54 #define dissect sdissect
55 #define backref sbackref
62 #define stepback lstepback
63 #define matcher lmatcher
65 #define dissect ldissect
66 #define backref lbackref
73 #define stepback mstepback
74 #define matcher mmatcher
76 #define dissect mdissect
77 #define backref mbackref
84 /* another structure passed up and down to avoid zillions of parameters */
88 regmatch_t *pmatch; /* [nsub+1] (0 element unused) */
89 const char *offp; /* offsets work from here */
90 const char *beginp; /* start of string -- virtual NUL precedes */
91 const char *endp; /* end of string -- virtual NUL here */
92 const char *coldp; /* can be no match starting before here */
93 const char **lastpos; /* [nplus+1] */
95 states st; /* current states */
96 states fresh; /* states for a fresh start */
97 states tmp; /* temporary */
98 states empty; /* empty set of states */
99 mbstate_t mbs; /* multibyte conversion state */
102 /* ========= begin header generated by ./mkh ========= */
107 /* === engine.c === */
108 static int matcher(struct re_guts *g, const char *string, size_t nmatch, regmatch_t pmatch[], int eflags);
109 static const char *dissect(struct match *m, const char *start, const char *stop, sopno startst, sopno stopst);
110 static const char *backref(struct match *m, const char *start, const char *stop, sopno startst, sopno stopst, sopno lev, int);
111 static const char *walk(struct match *m, const char *start, const char *stop, sopno startst, sopno stopst, bool fast);
112 static states step(struct re_guts *g, sopno start, sopno stop, states bef, wint_t ch, states aft, int sflags);
113 #define MAX_RECURSION 100
116 #define BOLEOL (BOL-2)
117 #define NOTHING (BOL-3)
120 #define BADCHAR (BOL-6)
121 #define NWBND (BOL-7)
122 #define NONCHAR(c) ((c) <= OUT)
128 static void print(struct match *m, const char *caption, states st, int ch, FILE *d);
131 static void at(struct match *m, const char *title, const char *start, const char *stop, sopno startst, sopno stopst);
134 static const char *pchar(int ch);
140 /* ========= end header generated by ./mkh ========= */
143 #define SP(t, s, c) print(m, t, s, c, stdout)
144 #define AT(t, p1, p2, s1, s2) at(m, t, p1, p2, s1, s2)
145 #define NOTE(str) { if (m->eflags®_TRACE) printf("=%s\n", (str)); }
147 #define SP(t, s, c) /* nothing */
148 #define AT(t, p1, p2, s1, s2) /* nothing */
149 #define NOTE(s) /* nothing */
153 * Given a multibyte string pointed to by start, step back nchar characters
154 * from current position pointed to by cur.
157 stepback(const char *start, const char *cur, int nchar)
165 return ((cur - nchar) > start ? cur - nchar : NULL);
168 for (wc = nchar; wc > 0; wc--) {
169 for (mbc = 1; mbc <= MB_CUR_MAX; mbc++) {
170 if ((ret - mbc) < start)
172 memset(&mbs, 0, sizeof(mbs));
173 clen = mbrtowc(NULL, ret - mbc, mbc, &mbs);
174 if (clen != (size_t)-1 && clen != (size_t)-2)
177 if (mbc > MB_CUR_MAX)
186 - matcher - the actual matching engine
187 == static int matcher(struct re_guts *g, const char *string, \
188 == size_t nmatch, regmatch_t pmatch[], int eflags);
190 static int /* 0 success, REG_NOMATCH failure */
191 matcher(struct re_guts *g,
200 struct match *m = &mv;
201 const char *dp = NULL;
202 const sopno gf = g->firststate+1; /* +1 for OEND */
203 const sopno gl = g->laststate;
206 /* Boyer-Moore algorithms variables */
209 const char *mustfirst;
210 const char *mustlast;
214 /* simplify the situation where possible */
215 if (g->cflags®_NOSUB)
217 if (eflags®_STARTEND) {
218 start = string + pmatch[0].rm_so;
219 stop = string + pmatch[0].rm_eo;
222 stop = start + strlen(start);
227 /* prescreening; this does wonders for this rather slow code */
228 if (g->must != NULL) {
229 if (g->charjump != NULL && g->matchjump != NULL) {
231 mustlast = g->must + g->mlen - 1;
232 charjump = g->charjump;
233 matchjump = g->matchjump;
235 for (dp = start+g->mlen-1; dp < stop;) {
236 /* Fast skip non-matches */
237 while (dp < stop && charjump[(int)*dp])
238 dp += charjump[(int)*dp];
244 /* We depend on not being used for
245 * for strings of length 1
247 while (*--dp == *--pp && pp != mustfirst);
252 /* Jump to next possible match */
253 mj = matchjump[pp - mustfirst];
254 cj = charjump[(int)*dp];
255 dp += (cj < mj ? mj : cj);
261 for (dp = start; dp < stop; dp++)
262 if (*dp == g->must[0] &&
263 stop - dp >= g->mlen &&
264 memcmp(dp, g->must, (size_t)g->mlen) == 0)
266 if (dp == stop) /* we didn't find g->must */
271 /* match struct setup */
287 /* Adjust start according to moffset, to speed things up */
288 if (dp != NULL && g->moffset > -1) {
291 nstart = stepback(start, dp, g->moffset);
296 SP("mloop", m->st, *start);
298 /* this loop does only one repetition except for backrefs */
300 endp = walk(m, start, stop, gf, gl, true);
301 if (endp == NULL) { /* a miss */
302 if (m->pmatch != NULL)
303 free((char *)m->pmatch);
304 if (m->lastpos != NULL)
305 free((char *)m->lastpos);
309 if (nmatch == 0 && !g->backrefs)
310 break; /* no further info needed */
313 assert(m->coldp != NULL);
315 NOTE("finding start");
316 endp = walk(m, m->coldp, stop, gf, gl, false);
319 assert(m->coldp < m->endp);
320 m->coldp += XMBRTOWC(NULL, m->coldp,
321 m->endp - m->coldp, &m->mbs, 0);
323 if (nmatch == 1 && !g->backrefs)
324 break; /* no further info needed */
326 /* oh my, he wants the subexpressions... */
327 if (m->pmatch == NULL)
328 m->pmatch = (regmatch_t *)malloc((m->g->nsub + 1) *
330 if (m->pmatch == NULL) {
334 for (i = 1; i <= m->g->nsub; i++)
335 m->pmatch[i].rm_so = m->pmatch[i].rm_eo = -1;
336 if (!g->backrefs && !(m->eflags®_BACKR)) {
338 dp = dissect(m, m->coldp, endp, gf, gl);
340 if (g->nplus > 0 && m->lastpos == NULL)
341 m->lastpos = malloc((g->nplus+1) *
342 sizeof(const char *));
343 if (g->nplus > 0 && m->lastpos == NULL) {
348 NOTE("backref dissect");
349 dp = backref(m, m->coldp, endp, gf, gl, (sopno)0, 0);
354 /* uh-oh... we couldn't find a subexpression-level match */
355 assert(g->backrefs); /* must be back references doing it */
356 assert(g->nplus == 0 || m->lastpos != NULL);
358 if (dp != NULL || endp <= m->coldp)
361 endp = walk(m, m->coldp, endp-1, gf, gl, false);
364 /* try it on a shorter possibility */
366 for (i = 1; i <= m->g->nsub; i++) {
367 assert(m->pmatch[i].rm_so == -1);
368 assert(m->pmatch[i].rm_eo == -1);
371 NOTE("backoff dissect");
372 dp = backref(m, m->coldp, endp, gf, gl, (sopno)0, 0);
374 assert(dp == NULL || dp == endp);
375 if (dp != NULL) /* found a shorter one */
378 /* despite initial appearances, there is no match here */
380 /* recycle starting later */
381 start = m->coldp + XMBRTOWC(NULL, m->coldp,
382 stop - m->coldp, &m->mbs, 0);
383 assert(start <= stop);
386 /* fill in the details if requested */
388 pmatch[0].rm_so = m->coldp - m->offp;
389 pmatch[0].rm_eo = endp - m->offp;
392 assert(m->pmatch != NULL);
393 for (i = 1; i < nmatch; i++)
395 pmatch[i] = m->pmatch[i];
397 pmatch[i].rm_so = -1;
398 pmatch[i].rm_eo = -1;
402 if (m->pmatch != NULL)
403 free((char *)m->pmatch);
404 if (m->lastpos != NULL)
405 free((char *)m->lastpos);
411 - dissect - figure out what matched what, no back references
412 == static const char *dissect(struct match *m, const char *start, \
413 == const char *stop, sopno startst, sopno stopst);
415 static const char * /* == stop (success) always */
416 dissect(struct match *m,
423 sopno ss; /* start sop of current subRE */
424 sopno es; /* end sop of current subRE */
425 const char *sp; /* start of string matched by it */
426 const char *stp; /* string matched by it cannot pass here */
427 const char *rest; /* start of rest of string */
428 const char *tail; /* string unmatched by rest of RE */
429 sopno ssub; /* start sop of subsubRE */
430 sopno esub; /* end sop of subsubRE */
431 const char *ssp; /* start of string matched by subsubRE */
432 const char *sep; /* end of string matched by subsubRE */
433 const char *oldssp; /* previous ssp */
434 const char *dp __unused;
436 AT("diss", start, stop, startst, stopst);
438 for (ss = startst; ss < stopst; ss = es) {
439 /* identify end of subRE */
441 switch (OP(m->g->strip[es])) {
444 es += OPND(m->g->strip[es]);
447 while (OP(m->g->strip[es]) != (sop)O_CH)
448 es += OPND(m->g->strip[es]);
453 /* figure out what it matched */
454 switch (OP(m->g->strip[ss])) {
459 sp += XMBRTOWC(NULL, sp, stop - start, &m->mbs, 0);
472 sp += XMBRTOWC(NULL, sp, stop - start, &m->mbs, 0);
478 /* cases where length of match is hard to find */
482 /* how long could this one be? */
483 rest = walk(m, sp, stp, ss, es, false);
484 assert(rest != NULL); /* it did match */
485 /* could the rest match the rest? */
486 tail = walk(m, rest, stop, es, stopst, false);
489 /* no -- try a shorter match for this one */
491 assert(stp >= sp); /* it did work */
495 /* did innards match? */
496 if (walk(m, sp, rest, ssub, esub, false) != NULL) {
497 dp = dissect(m, sp, rest, ssub, esub);
506 /* how long could this one be? */
507 rest = walk(m, sp, stp, ss, es, false);
508 assert(rest != NULL); /* it did match */
509 /* could the rest match the rest? */
510 tail = walk(m, rest, stop, es, stopst, false);
513 /* no -- try a shorter match for this one */
515 assert(stp >= sp); /* it did work */
521 for (;;) { /* find last match of innards */
522 sep = walk(m, ssp, rest, ssub, esub, false);
523 if (sep == NULL || sep == ssp)
524 break; /* failed or matched null */
525 oldssp = ssp; /* on to next try */
529 /* last successful match */
533 assert(sep == rest); /* must exhaust substring */
534 assert(walk(m, ssp, sep, ssub, esub, false) == rest);
535 dp = dissect(m, ssp, sep, ssub, esub);
542 /* how long could this one be? */
543 rest = walk(m, sp, stp, ss, es, false);
544 assert(rest != NULL); /* it did match */
545 /* could the rest match the rest? */
546 tail = walk(m, rest, stop, es, stopst, false);
549 /* no -- try a shorter match for this one */
551 assert(stp >= sp); /* it did work */
554 esub = ss + OPND(m->g->strip[ss]) - 1;
555 assert(OP(m->g->strip[esub]) == OOR1);
556 for (;;) { /* find first matching branch */
557 if (walk(m, sp, rest, ssub, esub, false) == rest)
558 break; /* it matched all of it */
559 /* that one missed, try next one */
560 assert(OP(m->g->strip[esub]) == OOR1);
562 assert(OP(m->g->strip[esub]) == OOR2);
564 esub += OPND(m->g->strip[esub]);
565 if (OP(m->g->strip[esub]) == (sop)OOR2)
568 assert(OP(m->g->strip[esub]) == O_CH);
570 dp = dissect(m, sp, rest, ssub, esub);
582 i = OPND(m->g->strip[ss]);
583 assert(0 < i && i <= m->g->nsub);
584 m->pmatch[i].rm_so = sp - m->offp;
587 i = OPND(m->g->strip[ss]);
588 assert(0 < i && i <= m->g->nsub);
589 m->pmatch[i].rm_eo = sp - m->offp;
601 #define ISBOW(m, sp) \
602 (sp < m->endp && ISWORD(*sp) && \
603 ((sp == m->beginp && !(m->eflags®_NOTBOL)) || \
604 (sp > m->offp && !ISWORD(*(sp-1)))))
605 #define ISEOW(m, sp) \
606 (((sp == m->endp && !(m->eflags®_NOTEOL)) || \
607 (sp < m->endp && *sp == '\n' && \
608 (m->g->cflags®_NEWLINE)) || \
609 (sp < m->endp && !ISWORD(*sp)) ) && \
610 (sp > m->beginp && ISWORD(*(sp-1)))) \
613 - backref - figure out what matched what, figuring in back references
614 == static const char *backref(struct match *m, const char *start, \
615 == const char *stop, sopno startst, sopno stopst, sopno lev);
617 static const char * /* == stop (success) or NULL (failure) */
618 backref(struct match *m,
623 sopno lev, /* PLUS nesting level */
627 sopno ss; /* start sop of current subRE */
628 const char *sp; /* start of string matched by it */
629 sopno ssub; /* start sop of subsubRE */
630 sopno esub; /* end sop of subsubRE */
631 const char *ssp; /* start of string matched by subsubRE */
640 AT("back", start, stop, startst, stopst);
643 /* get as far as we can with easy stuff */
645 for (ss = startst; !hard && ss < stopst; ss++)
646 switch (OP(s = m->g->strip[ss])) {
650 sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
657 sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
664 cs = &m->g->sets[OPND(s)];
665 sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
666 if (wc == BADCHAR || !CHIN(cs, wc))
670 if (sp == m->beginp && (m->eflags & REG_NOTBOL) == 0)
676 if (sp == m->endp && (m->eflags & REG_NOTEOL) == 0)
682 if ((sp == m->beginp && !(m->eflags®_NOTBOL)) ||
683 (sp > m->offp && sp < m->endp &&
684 *(sp-1) == '\n' && (m->g->cflags®_NEWLINE)))
690 if ( (sp == m->endp && !(m->eflags®_NOTEOL)) ||
691 (sp < m->endp && *sp == '\n' &&
692 (m->g->cflags®_NEWLINE)) )
698 if (ISBOW(m, sp) || ISEOW(m, sp))
704 if (((sp == m->beginp) && !ISWORD(*sp)) ||
705 (sp == m->endp && !ISWORD(*(sp - 1))))
706 { /* yes, beginning/end of subject */ }
707 else if (ISWORD(*(sp - 1)) == ISWORD(*sp))
708 { /* yes, beginning/end of subject */ }
726 case OOR1: /* matches null but needs to skip */
730 assert(OP(s) == OOR2);
732 } while (OP(s = m->g->strip[ss]) != (sop)O_CH);
733 /* note that the ss++ gets us past the O_CH */
735 default: /* have to make a choice */
739 if (!hard) { /* that was it! */
744 ss--; /* adjust for the for's final increment */
747 AT("hard", sp, stop, ss, stopst);
750 case OBACK_: /* the vilest depths */
752 assert(0 < i && i <= m->g->nsub);
753 if (m->pmatch[i].rm_eo == -1)
755 assert(m->pmatch[i].rm_so != -1);
756 len = m->pmatch[i].rm_eo - m->pmatch[i].rm_so;
757 if (len == 0 && rec++ > MAX_RECURSION)
759 assert(stop - m->beginp >= len);
761 return(NULL); /* not enough left to match */
762 ssp = m->offp + m->pmatch[i].rm_so;
763 if (memcmp(sp, ssp, len) != 0)
765 while (m->g->strip[ss] != (sop)SOP(O_BACK, i))
767 return(backref(m, sp+len, stop, ss+1, stopst, lev, rec));
768 case OQUEST_: /* to null or not */
769 dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
771 return(dp); /* not */
772 return(backref(m, sp, stop, ss+OPND(s)+1, stopst, lev, rec));
774 assert(m->lastpos != NULL);
775 assert(lev+1 <= m->g->nplus);
776 m->lastpos[lev+1] = sp;
777 return(backref(m, sp, stop, ss+1, stopst, lev+1, rec));
779 if (sp == m->lastpos[lev]) /* last pass matched null */
780 return(backref(m, sp, stop, ss+1, stopst, lev-1, rec));
781 /* try another pass */
782 m->lastpos[lev] = sp;
783 dp = backref(m, sp, stop, ss-OPND(s)+1, stopst, lev, rec);
785 return(backref(m, sp, stop, ss+1, stopst, lev-1, rec));
788 case OCH_: /* find the right one, if any */
790 esub = ss + OPND(s) - 1;
791 assert(OP(m->g->strip[esub]) == OOR1);
792 for (;;) { /* find first matching branch */
793 dp = backref(m, sp, stop, ssub, esub, lev, rec);
796 /* that one missed, try next one */
797 if (OP(m->g->strip[esub]) == (sop)O_CH)
798 return(NULL); /* there is none */
800 assert(OP(m->g->strip[esub]) == (sop)OOR2);
802 esub += OPND(m->g->strip[esub]);
803 if (OP(m->g->strip[esub]) == (sop)OOR2)
806 assert(OP(m->g->strip[esub]) == O_CH);
810 case OLPAREN: /* must undo assignment if rest fails */
812 assert(0 < i && i <= m->g->nsub);
813 offsave = m->pmatch[i].rm_so;
814 m->pmatch[i].rm_so = sp - m->offp;
815 dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
818 m->pmatch[i].rm_so = offsave;
820 case ORPAREN: /* must undo assignment if rest fails */
822 assert(0 < i && i <= m->g->nsub);
823 offsave = m->pmatch[i].rm_eo;
824 m->pmatch[i].rm_eo = sp - m->offp;
825 dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
828 m->pmatch[i].rm_eo = offsave;
838 return "shut up gcc";
842 - walk - step through the string either quickly or slowly
843 == static const char *walk(struct match *m, const char *start, \
844 == const char *stop, sopno startst, sopno stopst, bool fast);
846 static const char * /* where it ended, or NULL */
847 walk(struct match *m, const char *start, const char *stop, sopno startst,
848 sopno stopst, bool fast)
851 states fresh = m->fresh;
852 states empty = m->empty;
854 const char *p = start;
856 wint_t lastc; /* previous c */
859 const char *matchp; /* last p at which a match ended */
863 AT("slow", start, stop, startst, stopst);
866 SP("sstart", st, *p);
867 st = step(m->g, startst, stopst, st, NOTHING, st, sflags);
871 if (start == m->offp || (start == m->beginp && !(m->eflags®_NOTBOL)))
875 * XXX Wrong if the previous character was multi-byte.
876 * Newline never is (in encodings supported by FreeBSD),
877 * so this only breaks the ISWORD tests below.
879 c = (uch)*(start - 1);
889 clen = XMBRTOWC(&c, p, m->endp - p, &m->mbs, BADCHAR);
891 if (fast && EQ(st, fresh))
894 /* is there an EOL and/or BOL between lastc and c? */
897 if ( (lastc == '\n' && m->g->cflags®_NEWLINE) ||
898 (lastc == OUT && !(m->eflags®_NOTBOL)) ) {
902 if ( (c == '\n' && m->g->cflags®_NEWLINE) ||
903 (c == OUT && !(m->eflags®_NOTEOL)) ) {
904 flagch = (flagch == BOL) ? BOLEOL : EOL;
907 if (lastc == OUT && (m->eflags & REG_NOTBOL) == 0) {
909 /* Step one more for BOS. */
912 if (c == OUT && (m->eflags & REG_NOTEOL) == 0) {
914 /* Step one more for EOS. */
919 st = step(m->g, startst, stopst, st, flagch, st,
921 SP("sboleol", st, c);
924 /* how about a word boundary? */
925 if ( (flagch == BOL || (lastc != OUT && !ISWORD(lastc))) &&
926 (c != OUT && ISWORD(c)) ) {
929 if ( (lastc != OUT && ISWORD(lastc)) &&
930 (flagch == EOL || (c != OUT && !ISWORD(c))) ) {
933 if (flagch == BOW || flagch == EOW) {
934 st = step(m->g, startst, stopst, st, flagch, st, sflags);
935 SP("sboweow", st, c);
937 if (lastc != OUT && c != OUT &&
938 ISWORD(lastc) == ISWORD(c)) {
940 } else if ((lastc == OUT && !ISWORD(c)) ||
941 (c == OUT && !ISWORD(lastc))) {
944 if (flagch == NWBND) {
945 st = step(m->g, startst, stopst, st, flagch, st, sflags);
950 if (ISSET(st, stopst)) {
956 if (EQ(st, empty) || p == stop || clen > (size_t)(stop - p))
957 break; /* NOTE BREAK OUT */
959 /* no, we must deal with this character */
966 st = step(m->g, startst, stopst, tmp, c, st, sflags);
968 assert(EQ(step(m->g, startst, stopst, st, NOTHING, st, sflags),
974 assert(matchp != NULL);
976 if (ISSET(st, stopst))
977 return (p + XMBRTOWC(NULL, p, stop - p, &m->mbs, 0));
985 - step - map set of states reachable before char to set reachable after
986 == static states step(struct re_guts *g, sopno start, sopno stop, \
987 == states bef, int ch, states aft);
988 == #define BOL (OUT-1)
989 == #define EOL (BOL-1)
990 == #define BOLEOL (BOL-2)
991 == #define NOTHING (BOL-3)
992 == #define BOW (BOL-4)
993 == #define EOW (BOL-5)
994 == #define BADCHAR (BOL-6)
995 == #define NONCHAR(c) ((c) <= OUT)
998 step(struct re_guts *g,
999 sopno start, /* start state within strip */
1000 sopno stop, /* state after stop state within strip */
1001 states bef, /* states reachable before */
1002 wint_t ch, /* character or NONCHAR code */
1003 states aft, /* states already known reachable after */
1004 int sflags) /* state flags */
1009 onestate here; /* note, macros know this name */
1013 for (pc = start, INIT(here, pc); pc != stop; pc++, INC(here)) {
1017 assert(pc == stop-1);
1020 /* only characters can match */
1021 assert(!NONCHAR(ch) || ch != OPND(s));
1026 if ((ch == BOL || ch == BOLEOL) && (sflags & SBOS) != 0)
1030 if ((ch == EOL || ch == BOLEOL) && (sflags & SEOS) != 0)
1034 if (ch == BOL || ch == BOLEOL)
1038 if (ch == EOL || ch == BOLEOL)
1050 if (ch == BOW || ch == EOW)
1062 cs = &g->sets[OPND(s)];
1063 if (!NONCHAR(ch) && CHIN(cs, ch))
1066 case OBACK_: /* ignored here */
1070 case OPLUS_: /* forward, this is just an empty */
1073 case O_PLUS: /* both forward and back */
1075 i = ISSETBACK(aft, OPND(s));
1076 BACK(aft, aft, OPND(s));
1077 if (!i && ISSETBACK(aft, OPND(s))) {
1078 /* oho, must reconsider loop body */
1083 case OQUEST_: /* two branches, both forward */
1085 FWD(aft, aft, OPND(s));
1087 case O_QUEST: /* just an empty */
1090 case OLPAREN: /* not significant here */
1094 case OCH_: /* mark the first two branches */
1096 assert(OP(g->strip[pc+OPND(s)]) == (sop)OOR2);
1097 FWD(aft, aft, OPND(s));
1099 case OOR1: /* done a branch, find the O_CH */
1100 if (ISSTATEIN(aft, here)) {
1102 OP(s = g->strip[pc+look]) != (sop)O_CH;
1104 assert(OP(s) == (sop)OOR2);
1105 FWD(aft, aft, look + 1);
1108 case OOR2: /* propagate OCH_'s marking */
1110 if (OP(g->strip[pc+OPND(s)]) != (sop)O_CH) {
1111 assert(OP(g->strip[pc+OPND(s)]) == (sop)OOR2);
1112 FWD(aft, aft, OPND(s));
1115 case O_CH: /* just empty */
1118 default: /* ooooops... */
1129 - print - print a set of states
1131 == static void print(struct match *m, const char *caption, states st, \
1132 == int ch, FILE *d);
1136 print(struct match *m,
1137 const char *caption,
1142 struct re_guts *g = m->g;
1146 if (!(m->eflags®_TRACE))
1149 fprintf(d, "%s", caption);
1151 fprintf(d, " %s", pchar(ch));
1152 for (i = 0; i < g->nstates; i++)
1154 fprintf(d, "%s%lu", (first) ? "\t" : ", ", i);
1161 - at - print current situation
1163 == static void at(struct match *m, const char *title, const char *start, \
1164 == const char *stop, sopno startst, sopno stopst);
1168 at( struct match *m,
1175 if (!(m->eflags®_TRACE))
1178 printf("%s %s-", title, pchar(*start));
1179 printf("%s ", pchar(*stop));
1180 printf("%ld-%ld\n", (long)startst, (long)stopst);
1184 #define PCHARDONE /* never again */
1186 - pchar - make a character printable
1188 == static const char *pchar(int ch);
1191 * Is this identical to regchar() over in debug.c? Well, yes. But a
1192 * duplicate here avoids having a debugging-capable regexec.o tied to
1193 * a matching debug.o, and this is convenient. It all disappears in
1194 * the non-debug compilation anyway, so it doesn't matter much.
1196 static const char * /* -> representation */
1199 static char pbuf[10];
1201 if (isprint((uch)ch) || ch == ' ')
1202 sprintf(pbuf, "%c", ch);
1204 sprintf(pbuf, "\\%o", ch);