From 04106ff6f52b03d0d5b009f07a7b6ab11107f0c0 Mon Sep 17 00:00:00 2001
From: Michael Vogel <icarus@dabo.de>
Date: Mon, 8 Dec 2014 22:37:49 +0100
Subject: [PATCH] Moving the clean up code to a central place.

---
 include/Scrape.php       | 5 -----
 library/HTML5/Parser.php | 6 ++++++
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/include/Scrape.php b/include/Scrape.php
index ad9e030a39..99784af336 100644
--- a/include/Scrape.php
+++ b/include/Scrape.php
@@ -263,11 +263,6 @@ 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: ' . $e);
diff --git a/library/HTML5/Parser.php b/library/HTML5/Parser.php
index 5f9ca560e5..c7faf875ad 100644
--- a/library/HTML5/Parser.php
+++ b/library/HTML5/Parser.php
@@ -17,6 +17,12 @@ class HTML5_Parser
      * @return Parsed HTML as DOMDocument
      */
     static public function parse($text, $builder = null) {
+
+	// Cleanup invalid HTML
+	$doc = new DOMDocument();
+	@$doc->loadHTML($text);
+	$text = $doc->saveHTML();
+
         $tokenizer = new HTML5_Tokenizer($text, $builder);
         $tokenizer->parse();
         return $tokenizer->save();