Added regexp to remove all scripts elements

oscarotero · oscarotero · commit d037640387f7 · 2013-12-28T12:02:52.000+01:00
diff --git a/Embed/Url.php b/Embed/Url.php
@@ -154,6 +154,9 @@ public function getHtmlContent () {
 					$response = preg_replace('/<head[^>]*>/','<head><META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=utf-8">', $response);
 				}
 
+				//Remove all script elements (thanks https://github.com/jasny)
+				$response = preg_replace('%<script\b(?:"(?:[^"\\\\\\\\]++|\\\\\\\\.)*+"|\'(?:[^\\\\\\\\]++|\\\\\\\\.)*+\'|[^>"]++)*>(?:"(?:[^"\\\\\\\\]++|\\\\\\\\.)*+"|\'(?:[^\\\\\\\\]++|\\\\\\\\.)*+\'|//.*?\n|/\*(?:[^\*]++|\*)*?\*/|[^<"/]++|/|(?R)|<)*?</\s*script>%si', '', $response);
+
 				$this->htmlContent->loadHTML($response);
 				libxml_use_internal_errors($errors);
 

Original file line number	Diff line number	Diff line change
`@@ -154,6 +154,9 @@ public function getHtmlContent () {`
`154`	`154`	`$response = preg_replace('/<head[^>]*>/','<head><META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=utf-8">', $response);`
`155`	`155`	`}`
`156`	`156`
	`157`	`+ //Remove all script elements (thanks https://github.com/jasny)`
	`158`	`+ $response = preg_replace('%<script\b(?:"(?:[^"\\\\\\\\]++\|\\\\\\\\.)+"\|\'(?:[^\\\\\\\\]++\|\\\\\\\\.)+\'\|[^>"]++)>(?:"(?:[^"\\\\\\\\]++\|\\\\\\\\.)+"\|\'(?:[^\\\\\\\\]++\|\\\\\\\\.)+\'\|//.?\n\|/\(?:[^\]++\|\)?\/\|[^<"/]++\|/\|(?R)\|<)?</\s*script>%si', '', $response);`
	`159`	`+`
`157`	`160`	`$this->htmlContent->loadHTML($response);`
`158`	`161`	`libxml_use_internal_errors($errors);`
`159`	`162`