regcomp.c

This is a source code of VxWorks

		/* do a bunch of concatenated expressions */
		conc = HERE();
		while (MORE() && (c = PEEK()) != '|' && c != stop)
			p_ere_exp(p);
		REQUIRE(HERE() != conc, REG_EMPTY);	/* require nonempty */

		if (!EAT('|'))
			break;		/* NOTE BREAK OUT */

		if (first) {
			INSERT(OCH_, conc);	/* offset is wrong */
			prevfwd = conc;
			prevback = conc;
			first = 0;
		}
		ASTERN(OOR1, prevback);
		prevback = THERE();
		AHEAD(prevfwd);			/* fix previous offset */
		prevfwd = HERE();
		EMIT(OOR2, 0);			/* offset is very wrong */
	}

	if (!first) {		/* tail-end fixups */
		AHEAD(prevfwd);
		ASTERN(O_CH, prevback);
	}

	assert(!MORE() || SEE(stop));
}

/*
 - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op
 == static void p_ere_exp(register struct parse *p);
 */
static void
p_ere_exp(p)
register struct parse *p;
{
	register char c;
	register sopno pos;
	register int count;
	register int count2;
	register sopno subno;
	int wascaret = 0;

	assert(MORE());		/* caller should have ensured this */
	c = GETNEXT();

	pos = HERE();
	switch (c) {
	case '(':
		REQUIRE(MORE(), REG_EPAREN);
		p->g->nsub++;
		subno = p->g->nsub;
		if (subno < NPAREN)
			p->pbegin[subno] = HERE();
		EMIT(OLPAREN, subno);
		if (!SEE(')'))
			p_ere(p, ')');
		if (subno < NPAREN) {
			p->pend[subno] = HERE();
			assert(p->pend[subno] != 0);
		}
		EMIT(ORPAREN, subno);
		MUSTEAT(')', REG_EPAREN);
		break;
#ifndef POSIX_MISTAKE
	case ')':		/* happens only if no current unmatched ( */
		/*
		 * You may ask, why the ifndef?  Because I didn't notice
		 * this until slightly too late for 1003.2, and none of the
		 * other 1003.2 regular-expression reviewers noticed it at
		 * all.  So an unmatched ) is legal POSIX, at least until
		 * we can get it fixed.
		 */
		SETERROR(REG_EPAREN);
		break;
#endif
	case '^':
		EMIT(OBOL, 0);
		p->g->iflags |= USEBOL;
		p->g->nbol++;
		wascaret = 1;
		break;
	case '$':
		EMIT(OEOL, 0);
		p->g->iflags |= USEEOL;
		p->g->neol++;
		break;
	case '|':
		SETERROR(REG_EMPTY);
		break;
	case '*':
	case '+':
	case '?':
		SETERROR(REG_BADRPT);
		break;
	case '.':
		if (p->g->cflags&REG_NEWLINE)
			nonnewline(p);
		else
			EMIT(OANY, 0);
		break;
	case '[':
		p_bracket(p);
		break;
	case '\\':
		REQUIRE(MORE(), REG_EESCAPE);
		c = GETNEXT();
		ordinary(p, c);
		break;
	case '{':		/* okay as ordinary except if digit follows */
		REQUIRE(!MORE() || !isdigit(PEEK()), REG_BADRPT);
		/* FALLTHROUGH */
	default:
		ordinary(p, c);
		break;
	}

	if (!MORE())
		return;
	c = PEEK();
	/* we call { a repetition if followed by a digit */
	if (!( c == '*' || c == '+' || c == '?' ||
				(c == '{' && MORE2() && isdigit(PEEK2())) ))
		return;		/* no repetition, we're done */
	NEXT();

	REQUIRE(!wascaret, REG_BADRPT);
	switch (c) {
	case '*':	/* implemented as +? */
		/* this case does not require the (y|) trick, noKLUDGE */
		INSERT(OPLUS_, pos);
		ASTERN(O_PLUS, pos);
		INSERT(OQUEST_, pos);
		ASTERN(O_QUEST, pos);
		break;
	case '+':
		INSERT(OPLUS_, pos);
		ASTERN(O_PLUS, pos);
		break;
	case '?':
		/* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
		INSERT(OCH_, pos);		/* offset slightly wrong */
		ASTERN(OOR1, pos);		/* this one's right */
		AHEAD(pos);			/* fix the OCH_ */
		EMIT(OOR2, 0);			/* offset very wrong... */
		AHEAD(THERE());			/* ...so fix it */
		ASTERN(O_CH, THERETHERE());
		break;
	case '{':
		count = p_count(p);
		if (EAT(',')) {
			if (isdigit(PEEK())) {
				count2 = p_count(p);
				REQUIRE(count <= count2, REG_BADBR);
			} else		/* single number with comma */
				count2 = INFINITY;
		} else		/* just a single number */
			count2 = count;
		repeat(p, pos, count, count2);
		if (!EAT('}')) {	/* error heuristics */
			while (MORE() && PEEK() != '}')
				NEXT();
			REQUIRE(MORE(), REG_EBRACE);
			SETERROR(REG_BADBR);
		}
		break;
	}

	if (!MORE())
		return;
	c = PEEK();
	if (!( c == '*' || c == '+' || c == '?' ||
				(c == '{' && MORE2() && isdigit(PEEK2())) ) )
		return;
	SETERROR(REG_BADRPT);
}

/*
 - p_str - string (no metacharacters) "parser"
 == static void p_str(register struct parse *p);
 */
static void
p_str(p)
register struct parse *p;
{
	REQUIRE(MORE(), REG_EMPTY);
	while (MORE())
		ordinary(p, GETNEXT());
}

/*
 - p_bre - BRE parser top level, anchoring and concatenation
 == static void p_bre(register struct parse *p, register int end1, \
 ==	register int end2);
 * Giving end1 as OUT essentially eliminates the end1/end2 check.
 *
 * This implementation is a bit of a kludge, in that a trailing $ is first
 * taken as an ordinary character and then revised to be an anchor.  The
 * only undesirable side effect is that '$' gets included as a character
 * category in such cases.  This is fairly harmless; not worth fixing.
 * The amount of lookahead needed to avoid this kludge is excessive.
 */
static void
p_bre(p, end1, end2)
register struct parse *p;
register int end1;		/* first terminating character */
register int end2;		/* second terminating character */
{
	register sopno start = HERE();
	register int first = 1;			/* first subexpression? */
	register int wasdollar = 0;

	if (EAT('^')) {
		EMIT(OBOL, 0);
		p->g->iflags |= USEBOL;
		p->g->nbol++;
	}
	while (MORE() && !SEETWO(end1, end2)) {
		wasdollar = p_simp_re(p, first);
		first = 0;
	}
	if (wasdollar) {	/* oops, that was a trailing anchor */
		DROP(1);
		EMIT(OEOL, 0);
		p->g->iflags |= USEEOL;
		p->g->neol++;
	}

	REQUIRE(HERE() != start, REG_EMPTY);	/* require nonempty */
}

/*
 - p_simp_re - parse a simple RE, an atom possibly followed by a repetition
 == static int p_simp_re(register struct parse *p, int starordinary);
 */
static int			/* was the simple RE an unbackslashed $? */
p_simp_re(p, starordinary)
register struct parse *p;
int starordinary;		/* is a leading * an ordinary character? */
{
	register int c;
	register int count;
	register int count2;
	register sopno pos;
	register int i;
	register sopno subno;
#	define	BACKSL	(1<<CHAR_BIT)

	pos = HERE();		/* repetion op, if any, covers from here */

	assert(MORE());		/* caller should have ensured this */
	c = GETNEXT();
	if (c == '\\') {
		REQUIRE(MORE(), REG_EESCAPE);
		c = BACKSL | (unsigned char)GETNEXT();
	}
	switch (c) {
	case '.':
		if (p->g->cflags&REG_NEWLINE)
			nonnewline(p);
		else
			EMIT(OANY, 0);
		break;
	case '[':
		p_bracket(p);
		break;
	case BACKSL|'{':
		SETERROR(REG_BADRPT);
		break;
	case BACKSL|'(':
		p->g->nsub++;
		subno = p->g->nsub;
		if (subno < NPAREN)
			p->pbegin[subno] = HERE();
		EMIT(OLPAREN, subno);
		/* the MORE here is an error heuristic */
		if (MORE() && !SEETWO('\\', ')'))
			p_bre(p, '\\', ')');
		if (subno < NPAREN) {
			p->pend[subno] = HERE();
			assert(p->pend[subno] != 0);
		}
		EMIT(ORPAREN, subno);
		REQUIRE(EATTWO('\\', ')'), REG_EPAREN);
		break;
	case BACKSL|')':	/* should not get here -- must be user */
	case BACKSL|'}':
		SETERROR(REG_EPAREN);
		break;
	case BACKSL|'1':
	case BACKSL|'2':
	case BACKSL|'3':
	case BACKSL|'4':
	case BACKSL|'5':
	case BACKSL|'6':
	case BACKSL|'7':
	case BACKSL|'8':
	case BACKSL|'9':
		i = (c&~BACKSL) - '0';
		assert(i < NPAREN);
		if (p->pend[i] != 0) {
			assert(i <= p->g->nsub);
			EMIT(OBACK_, i);
			assert(p->pbegin[i] != 0);
			assert(OP(p->strip[p->pbegin[i]]) == OLPAREN);
			assert(OP(p->strip[p->pend[i]]) == ORPAREN);
			(void) dupl(p, p->pbegin[i]+1, p->pend[i]);
			EMIT(O_BACK, i);
		} else
			SETERROR(REG_ESUBREG);
		p->g->backrefs = 1;
		break;
	case '*':
		REQUIRE(starordinary, REG_BADRPT);
		/* FALLTHROUGH */
	default:
		ordinary(p, c &~ BACKSL);
		break;
	}

	if (EAT('*')) {		/* implemented as +? */
		/* this case does not require the (y|) trick, noKLUDGE */
		INSERT(OPLUS_, pos);
		ASTERN(O_PLUS, pos);
		INSERT(OQUEST_, pos);
		ASTERN(O_QUEST, pos);
	} else if (EATTWO('\\', '{')) {
		count = p_count(p);
		if (EAT(',')) {
			if (MORE() && isdigit(PEEK())) {
				count2 = p_count(p);
				REQUIRE(count <= count2, REG_BADBR);
			} else		/* single number with comma */
				count2 = INFINITY;
		} else		/* just a single number */
			count2 = count;
		repeat(p, pos, count, count2);
		if (!EATTWO('\\', '}')) {	/* error heuristics */
			while (MORE() && !SEETWO('\\', '}'))
				NEXT();
			REQUIRE(MORE(), REG_EBRACE);
			SETERROR(REG_BADBR);
		}
	} else if (c == (unsigned char)'$')	/* $ (but not \$) ends it */
		return(1);

	return(0);
}

/*
 - p_count - parse a repetition count
 == static int p_count(register struct parse *p);
 */
static int			/* the value */
p_count(p)
register struct parse *p;
{
	register int count = 0;
	register int ndigits = 0;

	while (MORE() && isdigit(PEEK()) && count <= DUPMAX) {
		count = count*10 + (GETNEXT() - '0');
		ndigits++;
	}

	REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR);
	return(count);
}

/*
 - p_bracket - parse a bracketed character list
 == static void p_bracket(register struct parse *p);
 *
 * Note a significant property of this code:  if the allocset() did SETERROR,
 * no set operations are done.
 */
static void
p_bracket(p)
register struct parse *p;
{
	/* register char c; XXX jcf: unused */
	register cset *cs = allocset(p);
	register 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 (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 */
		return;

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

/*
 - p_b_term - parse one term of a bracketed character list
 == static void p_b_term(register struct parse *p, register cset *cs);
 */
static void
p_b_term(p, cs)
register struct parse *p;
register cset *cs;
{
	register char c;
	register char start, finish;
	register int i;

	/* classify what we've got */
	switch ((MORE()) ? PEEK() : '\0') {
	case '[':
		c = (MORE2()) ? PEEK2() : '\0';
		break;
	case '-':
		SETERROR(REG_ERANGE);
		return;			/* NOTE RETURN */
		break;