/**
* 12 2.4.8
*/
static int h5_state_tag_open(h5_state_t* hs)
{
char ch;
TRACE();
ch = hs->s[hs->pos];
if (ch == CHAR_BANG) {
hs->pos += 1;
return h5_state_markup_declaration_open(hs);
} else if (ch == CHAR_SLASH) {
hs->pos += 1;
hs->is_close = 1;
return h5_state_end_tag_open(hs);
} else if (ch == CHAR_QUESTION) {
hs->pos += 1;
return h5_state_bogus_comment(hs);
} else if (ch == CHAR_PERCENT) {
/* this is not in spec.. alternative comment format used
by IE <= 9 and Safari < 4.0.3 */
hs->pos += 1;
return h5_state_bogus_comment2(hs);
} else if ((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z')) {
return h5_state_tag_name(hs);
} else {
/* user input mistake in configuring state */
if (hs->pos == 0) {
return h5_state_data(hs);
}
hs->token_start = hs->s + hs->pos - 1;
hs->token_len = 1;
hs->token_type = DATA_TEXT;
hs->state = h5_state_data;
return 1;
}
}
/**
* 12.2.4.9
*/
static int h5_state_end_tag_open(h5_state_t* hs)
{
char ch;
TRACE();
if (hs->pos >= hs->len) {
return 0;
}
ch = hs->s[hs->pos];
if (ch == CHAR_GT) {
return h5_state_data(hs);
} else if ((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z')) {
return h5_state_tag_name(hs);
}
hs->is_close = 0;
return h5_state_bogus_comment(hs);
}
/**
* 8.2.4.45
*/
static int h5_state_markup_declaration_open(h5_state_t* hs)
{
size_t remaining;
TRACE();
remaining = hs->len - hs->pos;
if (remaining >= 7 &&
/* case insensitive */
(hs->s[hs->pos + 0] == 'D' || hs->s[hs->pos + 0] == 'd') &&
(hs->s[hs->pos + 1] == 'O' || hs->s[hs->pos + 1] == 'o') &&
(hs->s[hs->pos + 2] == 'C' || hs->s[hs->pos + 2] == 'c') &&
(hs->s[hs->pos + 3] == 'T' || hs->s[hs->pos + 3] == 't') &&
(hs->s[hs->pos + 4] == 'Y' || hs->s[hs->pos + 4] == 'y') &&
(hs->s[hs->pos + 5] == 'P' || hs->s[hs->pos + 5] == 'p') &&
(hs->s[hs->pos + 6] == 'E' || hs->s[hs->pos + 6] == 'e')
) {
return h5_state_doctype(hs);
} else if (remaining >= 7 &&
/* upper case required */
hs->s[hs->pos + 0] == '[' &&
hs->s[hs->pos + 1] == 'C' &&
hs->s[hs->pos + 2] == 'D' &&
hs->s[hs->pos + 3] == 'A' &&
hs->s[hs->pos + 4] == 'T' &&
hs->s[hs->pos + 5] == 'A' &&
hs->s[hs->pos + 6] == '['
) {
hs->pos += 7;
return h5_state_cdata(hs);
} else if (remaining >= 2 &&
hs->s[hs->pos + 0] == '-' &&
hs->s[hs->pos + 1] == '-') {
hs->pos += 2;
return h5_state_comment(hs);
}
return h5_state_bogus_comment(hs);
}