From f3e68af10d4960df8ca59de3efb0ae2f53522f12 Mon Sep 17 00:00:00 2001
From: cedricbonhomme <devnull@localhost>
Date: Wed, 28 Apr 2010 08:24:41 +0200
Subject: New regular expression to remove special characters (for instance
 &#8217;).

---
 utils.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'utils.py')

diff --git a/utils.py b/utils.py
index 3b8b376a..fc945f34 100755
--- a/utils.py
+++ b/utils.py
@@ -74,7 +74,8 @@ def remove_html_tags(data):
     Remove HTML tags for the search.
     """
     p = re.compile(r'<[^<]*?/?>')
-    return p.sub('', data)
+    q = re.compile(r'&#[0-9]+;')
+    return p.sub('', q.sub('', data))
 
 def top_words(dic_articles, n=10, size=5):
     """
-- 
cgit