static void _single_char(struct parse_sp *ps, unsigned int c, const char *ptr)
{
ps->type = 0;
ps->cursor = ptr + 1;
dm_bit_clear_all(ps->charset);
dm_bit_set(ps->charset, c);
}
dm_bitset_t dm_bitset_create(struct dm_pool *mem, unsigned num_bits)
{
unsigned n = (num_bits / DM_BITS_PER_INT) + 2;
size_t size = sizeof(int) * n;
dm_bitset_t bs;
if (mem)
bs = dm_pool_zalloc(mem, size);
else
bs = dm_malloc(size);
if (!bs)
return NULL;
*bs = num_bits;
if (!mem)
dm_bit_clear_all(bs);
return bs;
}
/*
* Get the next token from the regular expression.
* Returns: 1 success, 0 end of input, -1 error.
*/
static int _rx_get_token(struct parse_sp *ps)
{
int neg = 0, range = 0;
char c, lc = 0;
const char *ptr = ps->cursor;
if (ptr == ps->rx_end) { /* end of input ? */
ps->type = -1;
return 0;
}
switch (*ptr) {
/* charsets and ncharsets */
case '[':
ptr++;
if (*ptr == '^') {
dm_bit_set_all(ps->charset);
/* never transition on zero */
dm_bit_clear(ps->charset, 0);
neg = 1;
ptr++;
} else
dm_bit_clear_all(ps->charset);
while ((ptr < ps->rx_end) && (*ptr != ']')) {
if (*ptr == '\\') {
/* an escaped character */
ptr++;
switch (*ptr) {
case 'n':
c = '\n';
break;
case 'r':
c = '\r';
break;
case 't':
c = '\t';
break;
default:
c = *ptr;
}
} else if (*ptr == '-' && lc) {
/* we've got a range on our hands */
range = 1;
ptr++;
if (ptr == ps->rx_end) {
log_error("Incomplete range"
"specification");
return -1;
}
c = *ptr;
} else
c = *ptr;
if (range) {
/* add lc - c into the bitset */
if (lc > c) {
char tmp = c;
c = lc;
lc = tmp;
}
for (; lc <= c; lc++) {
if (neg)
dm_bit_clear(ps->charset, lc);
else
dm_bit_set(ps->charset, lc);
}
range = 0;
} else {
/* add c into the bitset */
if (neg)
dm_bit_clear(ps->charset, c);
else
dm_bit_set(ps->charset, c);
}
ptr++;
lc = c;
}
if (ptr >= ps->rx_end) {
ps->type = -1;
return -1;
}
ps->type = 0;
ps->cursor = ptr + 1;
break;
/* These characters are special, we just return their ASCII
codes as the type. Sorted into ascending order to help the
compiler */
case '(':
case ')':
case '*':
case '+':
case '?':
case '|':
ps->type = (int) *ptr;
ps->cursor = ptr + 1;
break;
case '^':
_single_char(ps, HAT_CHAR, ptr);
break;
case '$':
_single_char(ps, DOLLAR_CHAR, ptr);
break;
case '.':
/* The 'all but newline' character set */
ps->type = 0;
ps->cursor = ptr + 1;
dm_bit_set_all(ps->charset);
dm_bit_clear(ps->charset, (int) '\n');
dm_bit_clear(ps->charset, (int) '\r');
dm_bit_clear(ps->charset, 0);
break;
case '\\':
/* escaped character */
ptr++;
if (ptr >= ps->rx_end) {
log_error("Badly quoted character at end "
"of expression");
ps->type = -1;
return -1;
}
ps->type = 0;
ps->cursor = ptr + 1;
dm_bit_clear_all(ps->charset);
switch (*ptr) {
case 'n':
dm_bit_set(ps->charset, (int) '\n');
break;
case 'r':
dm_bit_set(ps->charset, (int) '\r');
break;
case 't':
dm_bit_set(ps->charset, (int) '\t');
break;
default:
dm_bit_set(ps->charset, (int) *ptr);
}
break;
default:
/* add a single character to the bitset */
ps->type = 0;
ps->cursor = ptr + 1;
dm_bit_clear_all(ps->charset);
dm_bit_set(ps->charset, (int) *ptr);
break;
}
return 1;
}
