searx

commit: 52a57ee045e02844a8f650a9d3ae30e0092d86cd
parent a3d444ab85dbb85dc3200c686ec3323dbb7008cb
Author: Cqoicebordel <Cqoicebordel@users.noreply.github.com>
Date:   Fri, 30 Jan 2015 21:00:49 +0100

Replace every bunch of whitespaces with only one space in HTML text

Diffstat:
M
searx/utils.py
2
++

1 file changed, 2 insertions(+), 0 deletions(-)
diff --git 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()