/*
- p_b_term - parse one term of a bracketed character list
*/
static void
p_b_term(struct parse *p, cset *cs)
{
char c;
char start, finish;
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;
default:
c = '\0';
break;
}
switch (c) {
case ':': /* character class */
NEXT2();
REQUIRE(MORE(), REG_EBRACK);
c = PEEK();
REQUIRE(c != '-' && c != ']', REG_ECTYPE);
p_b_cclass(p, cs);
REQUIRE(MORE(), REG_EBRACK);
REQUIRE(EATTWO(':', ']'), REG_ECTYPE);
break;
case '=': /* equivalence class */
NEXT2();
REQUIRE(MORE(), REG_EBRACK);
c = PEEK();
REQUIRE(c != '-' && c != ']', REG_ECOLLATE);
p_b_eclass(p, cs);
REQUIRE(MORE(), REG_EBRACK);
REQUIRE(EATTWO('=', ']'), REG_ECOLLATE);
break;
default: /* symbol, ordinary character, or range */
/* xxx revision needed for multichar stuff */
start = p_b_symbol(p);
if (SEE('-') && MORE2() && PEEK2() != ']') {
/* range */
NEXT();
if (EAT('-'))
finish = '-';
else
finish = p_b_symbol(p);
} else
finish = start;
/* xxx what about signed chars here... */
REQUIRE(start <= finish, REG_ERANGE);
for (i = start; i <= finish; i++)
CHadd(cs, i);
break;
}
}
/*
- p_ere - ERE parser top level, concatenation and alternation
== static void p_ere(struct parse *p, int stop, size_t reclimit);
*/
static void
p_ere(
struct parse *p,
int stop, /* character this ERE should end at */
size_t reclimit)
{
char c;
sopno prevback = 0; /* pacify gcc */
sopno prevfwd = 0; /* pacify gcc */
sopno conc;
int first = 1; /* is this the first alternative? */
_DIAGASSERT(p != NULL);
if (reclimit++ > RECLIMIT || p->error == REG_ESPACE) {
p->error = REG_ESPACE;
return;
}
for (;;) {
/* do a bunch of concatenated expressions */
conc = HERE();
while (MORE() && (c = PEEK()) != '|' && c != stop)
p_ere_exp(p, reclimit);
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 - ERE parser top level, concatenation and alternation
*/
static void
p_ere(struct parse *p, int stop) /* character this ERE should end at */
{
char c;
sopno prevback;
sopno prevfwd;
sopno conc;
int first = 1; /* is this the first alternative? */
for (;;) {
/* 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(struct parse *p)
{
char c;
sopno pos;
int count;
int count2;
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®_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((uch)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((uch)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((uch)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((uch)PEEK2())) ) )
return;
SETERROR(REG_BADRPT);
}
