Пример #1
0
/*
 - freezeset - final processing on a set of characters
 *
 * The main task here is merging identical sets.  This is usually a waste
 * of time (although the hash code minimizes the overhead), but can win
 * big if REG_ICASE is being used.  REG_ICASE, by the way, is why the hash
 * is done using addition rather than xor -- all ASCII [aA] sets xor to
 * the same value!
 */
static int			/* set number */
freezeset(struct parse *p, cset *cs)
{
	uch h = cs->hash;
	size_t i;
	cset *top = &p->g->sets[p->g->ncsets];
	cset *cs2;
	size_t css = (size_t)p->g->csetsize;

	/* look for an earlier one which is the same */
	for (cs2 = &p->g->sets[0]; cs2 < top; cs2++)
		if (cs2->hash == h && cs2 != cs) {
			/* maybe */
			for (i = 0; i < css; i++)
				if (!!CHIN(cs2, i) != !!CHIN(cs, i))
					break;		/* no */
			if (i == css)
				break;			/* yes */
		}

	if (cs2 < top) {	/* found one */
		freeset(p, cs);
		cs = cs2;
	}

	return((int)(cs - p->g->sets));
}
Пример #2
0
/*
 - nch - number of characters in a set
 */
static int nch(struct parse *p, cset *cs) {
	int i;
	size_t css = (size_t)p->g->csetsize;
	int n = 0;

	for (i = 0; i < css; i++)
		if (CHIN(cs, i))
			n++;
	return(n);
}
Пример #3
0
/*
 - firstch - return first character in a set (which must have at least one)
 */
static int			/* character; there is no "none" value */
firstch(struct parse *p, cset *cs)
{
	size_t i;
	size_t css = (size_t)p->g->csetsize;

	for (i = 0; i < css; i++)
		if (CHIN(cs, i))
			return((char)i);
	assert(never);
	return(0);		/* arbitrary */
}
Пример #4
0
/*
 - nch - number of characters in a set
 == static int nch(struct parse *p, cset *cs);
 */
static int
nch(
    struct parse *p,
    cset *cs)
{
	size_t i;
	size_t css;
	int n = 0;

	_DIAGASSERT(p != NULL);
	_DIAGASSERT(cs != NULL);

	css = (size_t)p->g->csetsize;

	for (i = 0; i < css; i++)
		if (CHIN(cs, i))
			n++;
	return(n);
}
Пример #5
0
/*
 - p_bracket - parse a bracketed character list
 *
 * Note a significant property of this code:  if the allocset() did SETERROR,
 * no set operations are done.
 */
static void
p_bracket(struct parse *p)
{
	cset *cs;
	int invert = 0;

	/* Dept of Truly Sickening Special-Case Kludges */
	if (p->next + 5 < p->end && strncmp(p->next, "[:<:]]", 6) == 0) {
		EMIT(OBOW, 0);
		NEXTn(6);
		return;
	}
	if (p->next + 5 < p->end && strncmp(p->next, "[:>:]]", 6) == 0) {
		EMIT(OEOW, 0);
		NEXTn(6);
		return;
	}

	if ((cs = allocset(p)) == NULL) {
		/* allocset did set error status in p */
		return;
	}

	if (EAT('^'))
		invert++;	/* make note to invert set at end */
	if (EAT(']'))
		CHadd(cs, ']');
	else if (EAT('-'))
		CHadd(cs, '-');
	while (MORE() && PEEK() != ']' && !SEETWO('-', ']'))
		p_b_term(p, cs);
	if (EAT('-'))
		CHadd(cs, '-');
	MUSTEAT(']', REG_EBRACK);

	if (p->error != 0) {	/* don't mess things up further */
		freeset(p, cs);
		return;
	}

	if (p->g->cflags&REG_ICASE) {
		int i;
		int ci;

		for (i = p->g->csetsize - 1; i >= 0; i--)
			if (CHIN(cs, i) && isalpha(i)) {
				ci = othercase(i);
				if (ci != i)
					CHadd(cs, ci);
			}
		if (cs->multis != NULL)
			mccase(p, cs);
	}
	if (invert) {
		int i;

		for (i = p->g->csetsize - 1; i >= 0; i--)
			if (CHIN(cs, i))
				CHsub(cs, i);
			else
				CHadd(cs, i);
		if (p->g->cflags&REG_NEWLINE)
			CHsub(cs, '\n');
		if (cs->multis != NULL)
			mcinvert(p, cs);
	}

	assert(cs->multis == NULL);		/* xxx */

	if (nch(p, cs) == 1) {		/* optimize singleton sets */
		ordinary(p, firstch(p, cs));
		freeset(p, cs);
	} else
		EMIT(OANYOF, freezeset(p, cs));
}
Пример #6
0
/*
 - step - map set of states reachable before char to set reachable after
 == static states step(struct re_guts *g, sopno start, sopno stop, \
 ==	states bef, int ch, states aft);
 == #define	BOL	(OUT-1)
 == #define	EOL	(BOL-1)
 == #define	BOLEOL	(BOL-2)
 == #define	NOTHING	(BOL-3)
 == #define	BOW	(BOL-4)
 == #define	EOW	(BOL-5)
 == #define	BADCHAR	(BOL-6)
 == #define	NONCHAR(c)	((c) <= OUT)
 */
static states
step(struct re_guts *g,
	sopno start,		/* start state within strip */
	sopno stop,		/* state after stop state within strip */
	states bef,		/* states reachable before */
	wint_t ch,		/* character or NONCHAR code */
	states aft)		/* states already known reachable after */
{
	cset *cs;
	sop s;
	sopno pc;
	onestate here;		/* note, macros know this name */
	sopno look;
	int i;

	for (pc = start, INIT(here, pc); pc != stop; pc++, INC(here)) {
		s = g->strip[pc];
		switch (OP(s)) {
		case OEND:
			assert(pc == stop-1);
			break;
		case OCHAR:
			/* only characters can match */
			assert(!NONCHAR(ch) || ch != OPND(s));
			if (ch == OPND(s))
				FWD(aft, bef, 1);
			break;
		case OBOL:
			if (ch == BOL || ch == BOLEOL)
				FWD(aft, bef, 1);
			break;
		case OEOL:
			if (ch == EOL || ch == BOLEOL)
				FWD(aft, bef, 1);
			break;
		case OBOW:
			if (ch == BOW)
				FWD(aft, bef, 1);
			break;
		case OEOW:
			if (ch == EOW)
				FWD(aft, bef, 1);
			break;
		case OANY:
			if (!NONCHAR(ch))
				FWD(aft, bef, 1);
			break;
		case OANYOF:
			cs = &g->sets[OPND(s)];
			if (!NONCHAR(ch) && CHIN(cs, ch))
				FWD(aft, bef, 1);
			break;
		case OBACK_:		/* ignored here */
		case O_BACK:
			FWD(aft, aft, 1);
			break;
		case OPLUS_:		/* forward, this is just an empty */
			FWD(aft, aft, 1);
			break;
		case O_PLUS:		/* both forward and back */
			FWD(aft, aft, 1);
			i = ISSETBACK(aft, OPND(s));
			BACK(aft, aft, OPND(s));
			if (!i && ISSETBACK(aft, OPND(s))) {
				/* oho, must reconsider loop body */
				pc -= OPND(s) + 1;
				INIT(here, pc);
			}
			break;
		case OQUEST_:		/* two branches, both forward */
			FWD(aft, aft, 1);
			FWD(aft, aft, OPND(s));
			break;
		case O_QUEST:		/* just an empty */
			FWD(aft, aft, 1);
			break;
		case OLPAREN:		/* not significant here */
		case ORPAREN:
			FWD(aft, aft, 1);
			break;
		case OCH_:		/* mark the first two branches */
			FWD(aft, aft, 1);
			assert(OP(g->strip[pc+OPND(s)]) == OOR2);
			FWD(aft, aft, OPND(s));
			break;
		case OOR1:		/* done a branch, find the O_CH */
			if (ISSTATEIN(aft, here)) {
				for (look = 1;
						OP(s = g->strip[pc+look]) != O_CH;
						look += OPND(s))
					assert(OP(s) == OOR2);
				FWD(aft, aft, look);
			}
			break;
		case OOR2:		/* propagate OCH_'s marking */
			FWD(aft, aft, 1);
			if (OP(g->strip[pc+OPND(s)]) != O_CH) {
				assert(OP(g->strip[pc+OPND(s)]) == OOR2);
				FWD(aft, aft, OPND(s));
			}
			break;
		case O_CH:		/* just empty */
			FWD(aft, aft, 1);
			break;
		default:		/* ooooops... */
			assert(nope);
			break;
		}
	}

	return(aft);
}
Пример #7
0
/*
 - backref - figure out what matched what, figuring in back references
 == static const char *backref(struct match *m, const char *start, \
 ==	const char *stop, sopno startst, sopno stopst, sopno lev);
 */
static const char *		/* == stop (success) or NULL (failure) */
backref(struct match *m,
	const char *start,
	const char *stop,
	sopno startst,
	sopno stopst,
	sopno lev,		/* PLUS nesting level */
	int rec)
{
	int i;
	sopno ss;		/* start sop of current subRE */
	const char *sp;		/* start of string matched by it */
	sopno ssub;		/* start sop of subsubRE */
	sopno esub;		/* end sop of subsubRE */
	const char *ssp;	/* start of string matched by subsubRE */
	const char *dp;
	size_t len;
	int hard;
	sop s;
	regoff_t offsave;
	cset *cs;
	wint_t wc;

	AT("back", start, stop, startst, stopst);
	sp = start;

	/* get as far as we can with easy stuff */
	hard = 0;
	for (ss = startst; !hard && ss < stopst; ss++)
		switch (OP(s = m->g->strip[ss])) {
		case OCHAR:
			if (sp == stop)
				return(NULL);
			sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
			if (wc != OPND(s))
				return(NULL);
			break;
		case OANY:
			if (sp == stop)
				return(NULL);
			sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
			if (wc == BADCHAR)
				return (NULL);
			break;
		case OANYOF:
			if (sp == stop)
				return (NULL);
			cs = &m->g->sets[OPND(s)];
			sp += XMBRTOWC(&wc, sp, stop - sp, &m->mbs, BADCHAR);
			if (wc == BADCHAR || !CHIN(cs, wc))
				return(NULL);
			break;
		case OBOL:
			if ( (sp == m->beginp && !(m->eflags&REG_NOTBOL)) ||
					(sp < m->endp && *(sp-1) == '\n' &&
						(m->g->cflags&REG_NEWLINE)) )
				{ /* yes */ }
			else
				return(NULL);
			break;
		case OEOL:
			if ( (sp == m->endp && !(m->eflags&REG_NOTEOL)) ||
					(sp < m->endp && *sp == '\n' &&
						(m->g->cflags&REG_NEWLINE)) )
				{ /* yes */ }
			else
				return(NULL);
			break;
		case OBOW:
			if (( (sp == m->beginp && !(m->eflags&REG_NOTBOL)) ||
					(sp < m->endp && *(sp-1) == '\n' &&
						(m->g->cflags&REG_NEWLINE)) ||
					(sp > m->beginp &&
							!ISWORD(*(sp-1))) ) &&
					(sp < m->endp && ISWORD(*sp)) )
				{ /* yes */ }
			else
				return(NULL);
			break;
		case OEOW:
			if (( (sp == m->endp && !(m->eflags&REG_NOTEOL)) ||
					(sp < m->endp && *sp == '\n' &&
						(m->g->cflags&REG_NEWLINE)) ||
					(sp < m->endp && !ISWORD(*sp)) ) &&
					(sp > m->beginp && ISWORD(*(sp-1))) )
				{ /* yes */ }
			else
				return(NULL);
			break;
		case O_QUEST:
			break;
		case OOR1:	/* matches null but needs to skip */
			ss++;
			s = m->g->strip[ss];
			do {
				assert(OP(s) == OOR2);
				ss += OPND(s);
			} while (OP(s = m->g->strip[ss]) != O_CH);
			/* note that the ss++ gets us past the O_CH */
			break;
		default:	/* have to make a choice */
			hard = 1;
			break;
		}
	if (!hard) {		/* that was it! */
		if (sp != stop)
			return(NULL);
		return(sp);
	}
	ss--;			/* adjust for the for's final increment */

	/* the hard stuff */
	AT("hard", sp, stop, ss, stopst);
	s = m->g->strip[ss];
	switch (OP(s)) {
	case OBACK_:		/* the vilest depths */
		i = OPND(s);
		assert(0 < i && i <= m->g->nsub);
		if (m->pmatch[i].rm_eo == -1)
			return(NULL);
		assert(m->pmatch[i].rm_so != -1);
		len = m->pmatch[i].rm_eo - m->pmatch[i].rm_so;
		if (len == 0 && rec++ > MAX_RECURSION)
			return(NULL);
		assert(stop - m->beginp >= len);
		if (sp > stop - len)
			return(NULL);	/* not enough left to match */
		ssp = m->offp + m->pmatch[i].rm_so;
		if (memcmp(sp, ssp, len) != 0)
			return(NULL);
		while (m->g->strip[ss] != SOP(O_BACK, i))
			ss++;
		return(backref(m, sp+len, stop, ss+1, stopst, lev, rec));
		break;
	case OQUEST_:		/* to null or not */
		dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
		if (dp != NULL)
			return(dp);	/* not */
		return(backref(m, sp, stop, ss+OPND(s)+1, stopst, lev, rec));
		break;
	case OPLUS_:
		assert(m->lastpos != NULL);
		assert(lev+1 <= m->g->nplus);
		m->lastpos[lev+1] = sp;
		return(backref(m, sp, stop, ss+1, stopst, lev+1, rec));
		break;
	case O_PLUS:
		if (sp == m->lastpos[lev])	/* last pass matched null */
			return(backref(m, sp, stop, ss+1, stopst, lev-1, rec));
		/* try another pass */
		m->lastpos[lev] = sp;
		dp = backref(m, sp, stop, ss-OPND(s)+1, stopst, lev, rec);
		if (dp == NULL)
			return(backref(m, sp, stop, ss+1, stopst, lev-1, rec));
		else
			return(dp);
		break;
	case OCH_:		/* find the right one, if any */
		ssub = ss + 1;
		esub = ss + OPND(s) - 1;
		assert(OP(m->g->strip[esub]) == OOR1);
		for (;;) {	/* find first matching branch */
			dp = backref(m, sp, stop, ssub, esub, lev, rec);
			if (dp != NULL)
				return(dp);
			/* that one missed, try next one */
			if (OP(m->g->strip[esub]) == O_CH)
				return(NULL);	/* there is none */
			esub++;
			assert(OP(m->g->strip[esub]) == OOR2);
			ssub = esub + 1;
			esub += OPND(m->g->strip[esub]);
			if (OP(m->g->strip[esub]) == OOR2)
				esub--;
			else
				assert(OP(m->g->strip[esub]) == O_CH);
		}
		break;
	case OLPAREN:		/* must undo assignment if rest fails */
		i = OPND(s);
		assert(0 < i && i <= m->g->nsub);
		offsave = m->pmatch[i].rm_so;
		m->pmatch[i].rm_so = sp - m->offp;
		dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
		if (dp != NULL)
			return(dp);
		m->pmatch[i].rm_so = offsave;
		return(NULL);
		break;
	case ORPAREN:		/* must undo assignment if rest fails */
		i = OPND(s);
		assert(0 < i && i <= m->g->nsub);
		offsave = m->pmatch[i].rm_eo;
		m->pmatch[i].rm_eo = sp - m->offp;
		dp = backref(m, sp, stop, ss+1, stopst, lev, rec);
		if (dp != NULL)
			return(dp);
		m->pmatch[i].rm_eo = offsave;
		return(NULL);
		break;
	default:		/* uh oh */
		assert(nope);
		break;
	}

	/* "can't happen" */
	assert(nope);
	/* NOTREACHED */
	return "shut up gcc";
}
Пример #8
0
/*
 * s_print - print the strip for debugging
 */
static void
s_print(struct re_guts *g, FILE *d)
{
	sop *s;
	cset *cs;
	int done = 0;
	sop opnd;
	int col = 0;
	ssize_t last;
	sopno offset = 2;
#	define	GAP()	{	if (offset % 5 == 0) { \
					if (col > 40) { \
						fprintf(d, "\n\t"); \
						col = 0; \
					} else { \
						fprintf(d, " "); \
						col++; \
					} \
				} else \
					col++; \
				offset++; \
			}

	if (OP(g->strip[0]) != OEND)
		fprintf(d, "missing initial OEND!\n");
	for (s = &g->strip[1]; !done; s++) {
		opnd = OPND(*s);
		switch (OP(*s)) {
		case OEND:
			fprintf(d, "\n");
			done = 1;
			break;
		case OCHAR:
			if (strchr("\\|()^$.[+*?{}!<> ", (char)opnd) != NULL)
				fprintf(d, "\\%c", (char)opnd);
			else
				fprintf(d, "%s", regchar((char)opnd));
			break;
		case OBOL:
			fprintf(d, "^");
			break;
		case OEOL:
			fprintf(d, "$");
			break;
		case OBOW:
			fprintf(d, "\\{");
			break;
		case OEOW:
			fprintf(d, "\\}");
			break;
		case OANY:
			fprintf(d, ".");
			break;
		case OANYOF:
			fprintf(d, "[(%ld)", (long)opnd);
#ifdef __NetBSD__
			cs = &g->sets[opnd];
			last = -1;
			for (size_t i = 0; i < g->csetsize+1; i++)	/* +1 flushes */
				if (CHIN(cs, i) && i < g->csetsize) {
					if (last < 0) {
						fprintf(d, "%s", regchar(i));
						last = i;
					}
				} else {
					if (last >= 0) {
						if (last != (ssize_t)i - 1)
							fprintf(d, "-%s",
							    regchar(i - 1));
						last = -1;
					}
				}
#endif
			fprintf(d, "]");
			break;
		case OBACK_:
			fprintf(d, "(\\<%ld>", (long)opnd);
			break;
		case O_BACK:
			fprintf(d, "<%ld>\\)", (long)opnd);
			break;
		case OPLUS_:
			fprintf(d, "(+");
			if (OP(*(s+opnd)) != O_PLUS)
				fprintf(d, "<%ld>", (long)opnd);
			break;
		case O_PLUS:
			if (OP(*(s-opnd)) != OPLUS_)
				fprintf(d, "<%ld>", (long)opnd);
			fprintf(d, "+)");
			break;
		case OQUEST_:
			fprintf(d, "(?");
			if (OP(*(s+opnd)) != O_QUEST)
				fprintf(d, "<%ld>", (long)opnd);
			break;
		case O_QUEST:
			if (OP(*(s-opnd)) != OQUEST_)
				fprintf(d, "<%ld>", (long)opnd);
			fprintf(d, "?)");
			break;
		case OLPAREN:
			fprintf(d, "((<%ld>", (long)opnd);
			break;
		case ORPAREN:
			fprintf(d, "<%ld>))", (long)opnd);
			break;
		case OCH_:
			fprintf(d, "<");
			if (OP(*(s+opnd)) != OOR2)
				fprintf(d, "<%ld>", (long)opnd);
			break;
		case OOR1:
			if (OP(*(s-opnd)) != OOR1 && OP(*(s-opnd)) != OCH_)
				fprintf(d, "<%ld>", (long)opnd);
			fprintf(d, "|");
			break;
		case OOR2:
			fprintf(d, "|");
			if (OP(*(s+opnd)) != OOR2 && OP(*(s+opnd)) != O_CH)
				fprintf(d, "<%ld>", (long)opnd);
			break;
		case O_CH:
			if (OP(*(s-opnd)) != OOR1)
				fprintf(d, "<%ld>", (long)opnd);
			fprintf(d, ">");
			break;
		default:
#ifdef __FreeBSD__
			fprintf(d, "!%ld(%ld)!", OP(*s), opnd);
#else
			fprintf(d, "!%d(%d)!", OP(*s), opnd);
#endif
			break;
		}
		if (!done)
			GAP();
	}
}