diff options
Diffstat (limited to 'it_html.class')
| -rw-r--r-- | it_html.class | 10 |
1 files changed, 7 insertions, 3 deletions
diff --git a/it_html.class b/it_html.class index 9780d5d..effe920 100644 --- a/it_html.class +++ b/it_html.class @@ -408,11 +408,15 @@ static function sanitize($html) $html = it::replace(array('[\0\s]+' => " "), $html); # \s also matches \r and \n $urlpattern = 'https?://[^">]+'; - if ($tag = it::match("(.*?)<(div|p|ol|ul|li|i|b|strong|h[1-6])\b[^>]*>(.*?)</\\2>(.*)", $html)) + if ($tag = it::match('(<(div|p|ol|ul|li|i|b|strong|h[1-6])\b[^>]*>((?:(?!</?\2\b).|(?R))*?)</\2>)', $html, ['offset_capture' => 1])) { # Simple tags with content, no attributes kept - list($head, $tagname, $content, $tail) = $tag; - $tagname = strtolower($tagname); + $offset = $tag[0][1]; + $length = strlen($tag[0][0]); + $head = substr($html, 0, $offset); + $tail = substr($html, $offset + $length); + $content = $tag[2][0]; + $tagname = strtolower($tag[1][0]); $result .= it_html::sanitize($head) . "<$tagname>" . it_html::sanitize($content) . "</$tagname>" . it_html::sanitize($tail); } else if ($tag = it::match('(.*)<a\b[^>]+?\bhref\s*=\s*"(' . $urlpattern . ')"[^>]*?>(.*?)</a>(.*)', $html)) |