diff --git a/include/Scrape.php b/include/Scrape.php index 99784af3..ad9e030a 100644 --- a/include/Scrape.php +++ b/include/Scrape.php @@ -263,6 +263,11 @@ function scrape_feed($url) { } try { + // Cleanup invalid HTML + $doc = new DOMDocument(); + @$doc->loadHTML($s); + $s = $doc->saveHTML(); + $dom = HTML5_Parser::parse($s); } catch (DOMException $e) { logger('scrape_feed: parse error: ' . $e);