Merge pull request #15 from oscarotero/remove_scripts

oscarotero · oscarotero · commit 152b87d52b75 · 2013-12-29T04:43:56.000-08:00
Remove scripts, comments and cdata sections to avoid problems in the DOMDocument
diff --git a/Embed/Url.php b/Embed/Url.php
@@ -153,7 +153,10 @@ public function getHtmlContent () {
 					$response = mb_convert_encoding($response, 'HTML-ENTITIES', 'UTF-8');
 					$response = preg_replace('/<head[^>]*>/','<head><META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=utf-8">', $response);
 				}
-
+    
+				//Remove all script elements, CDATA sections and comments (thanks https://github.com/jasny)
+				$response = preg_replace(array('%<!--(?:[^-]++|-)*?-->|<!\[CDATA\[(?:[^\]]++|\])*?\]\]>%si', '%<script\b(?:"(?:[^"\\\\]++|\\\\.)*+"|\'(?:[^\'\\\\]++|\\\\.)*+\'|[^>"\']++)*>(?:[^<]++|<)*?</\s*script\s*>%si'), '', $response);
+                
 				$this->htmlContent->loadHTML($response);
 				libxml_use_internal_errors($errors);
 
