static gchar *html_find_str_case(HTMLParser *parser, const gchar *str)
{
gchar *p;
while ((p = strcasestr(parser->bufp, str)) == NULL) {
if (html_read_line(parser) == HTML_EOF)
return NULL;
}
return p;
}
static gchar *html_find_char(HTMLParser *parser, gchar ch)
{
gchar *p;
while ((p = strchr(parser->bufp, ch)) == NULL) {
if (html_read_line(parser) == HTML_EOF)
return NULL;
}
return p;
}
static void html_get_parenthesis(HTMLParser *parser, gchar *buf, gint len)
{
gchar *p;
buf[0] = '\0';
g_return_if_fail(*parser->bufp == '<');
/* ignore comment / CSS / script stuff */
if (!strncmp(parser->bufp, "<!--", 4)) {
parser->bufp += 4;
while ((p = strstr(parser->bufp, "-->")) == NULL)
if (html_read_line(parser) == HTML_EOF) return;
parser->bufp = p + 3;
return;
}
if (!g_ascii_strncasecmp(parser->bufp, "<style", 6)) {
parser->bufp += 6;
while ((p = strcasestr(parser->bufp, "</style>")) == NULL)
if (html_read_line(parser) == HTML_EOF) return;
parser->bufp = p + 8;
return;
}
if (!g_ascii_strncasecmp(parser->bufp, "<script", 7)) {
parser->bufp += 7;
while ((p = strcasestr(parser->bufp, "</script>")) == NULL)
if (html_read_line(parser) == HTML_EOF) return;
parser->bufp = p + 9;
return;
}
parser->bufp++;
while ((p = strchr(parser->bufp, '>')) == NULL)
if (html_read_line(parser) == HTML_EOF) return;
strncpy2(buf, parser->bufp, MIN(p - parser->bufp + 1, len));
g_strstrip(buf);
parser->bufp = p + 1;
}
const gchar *html_parse(HTMLParser *parser)
{
parser->state = HTML_NORMAL;
g_string_truncate(parser->str, 0);
if (*parser->bufp == '\0') {
g_string_truncate(parser->buf, 0);
parser->bufp = parser->buf->str;
if (html_read_line(parser) == HTML_EOF)
return NULL;
}
while (*parser->bufp != '\0') {
switch (*parser->bufp) {
case '<':
if (parser->str->len == 0)
html_parse_tag(parser);
else
return parser->str->str;
break;
case '&':
html_parse_special(parser);
break;
case ' ':
case '\t':
case '\r':
case '\n':
if (parser->bufp[0] == '\r' && parser->bufp[1] == '\n')
parser->bufp++;
if (!parser->pre) {
if (!parser->newline)
parser->space = TRUE;
parser->bufp++;
break;
}
/* fallthrough */
default:
html_append_char(parser, *parser->bufp++);
}
}
return parser->str->str;
}
