From 7b284b949ffc9e4adb48b4232017757675ff086a Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Mon, 9 Jun 2014 15:03:49 +0530
Subject: [PATCH] Handle invalid bytes when in index_to_soup() for
 JavascriptRecipe

---
 src/calibre/web/feeds/jsnews.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/calibre/web/feeds/jsnews.py b/src/calibre/web/feeds/jsnews.py
index bb8d891d06..4993598fa8 100644
--- a/src/calibre/web/feeds/jsnews.py
+++ b/src/calibre/web/feeds/jsnews.py
@@ -16,6 +16,7 @@ from calibre.web.feeds import feeds_from_index
 from calibre.web.feeds.news import BasicNewsRecipe
 from calibre.web.fetch.javascript import fetch_page,  AbortFetch, links_from_selectors
 from calibre.ebooks.chardet import xml_to_unicode, strip_encoding_declarations
+from calibre.utils.cleantext import clean_xml_chars
 
 def image_data_to_url(data, base='cover'):
     from calibre.utils.imghdr import what
@@ -221,7 +222,7 @@ class JavascriptRecipe(BasicNewsRecipe):
         if raw:
             return html
         import html5lib
-        root = html5lib.parse(html, treebuilder='lxml', namespaceHTMLElements=False).getroot()
+        root = html5lib.parse(clean_xml_chars(html), treebuilder='lxml', namespaceHTMLElements=False).getroot()
         return root
 
 # ***************************** Internal API *****************************