From 52a57ee045e02844a8f650a9d3ae30e0092d86cd Mon Sep 17 00:00:00 2001
From: Cqoicebordel <Cqoicebordel@users.noreply.github.com>
Date: Fri, 30 Jan 2015 21:00:49 +0100
Subject: [PATCH] Replace every bunch of whitespaces with only one space in
 HTML text

---
 searx/utils.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/searx/utils.py b/searx/utils.py
index 59d4b85b..ef221ef8 100644
--- a/searx/utils.py
+++ b/searx/utils.py
@@ -119,6 +119,8 @@ class HTMLTextExtractor(HTMLParser):
 
 
 def html_to_text(html):
+    html = html.replace('\n', ' ')
+    html = ' '.join(html.split())
     s = HTMLTextExtractor()
     s.feed(html)
     return s.get_text()