Cleanup the HTML before parsing it.

2014-12-08 14:26:44 +01:00
parent c398974918
commit 620ee6be0b
1 changed files with 5 additions and 0 deletions
--- 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);