From ac6292d614633c55f9c6e4a64c2a106dae479b15 Mon Sep 17 00:00:00 2001
From: Karl Hobley <karl@torchbox.com>
Date: Fri, 11 Jul 2014 16:40:36 +0100
Subject: [PATCH] Replaced lxml with html5lib

---
 setup.py                         | 2 +-
 wagtail/wagtailcore/whitelist.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/setup.py b/setup.py
index 129a48468a..b37d4f256b 100644
--- a/setup.py
+++ b/setup.py
@@ -32,7 +32,7 @@ install_requires = [
     "django-treebeard==2.0",
     "Pillow>=2.3.0",
     "beautifulsoup4>=4.3.2",
-    "lxml>=3.3.0",
+    "html5lib==0.999",
     "Unidecode>=0.04.14",
     "six==1.7.3",
     'requests==2.3.0',
diff --git a/wagtail/wagtailcore/whitelist.py b/wagtail/wagtailcore/whitelist.py
index 4aaff780d8..dba8982fdc 100644
--- a/wagtail/wagtailcore/whitelist.py
+++ b/wagtail/wagtailcore/whitelist.py
@@ -81,7 +81,7 @@ class Whitelister(object):
     def clean(cls, html):
         """Clean up an HTML string to contain just the allowed elements /
         attributes"""
-        doc = BeautifulSoup(html, 'lxml')
+        doc = BeautifulSoup(html, 'html5lib')
         cls.clean_node(doc, doc)
         return doc.decode()