searx

commit: 23b9095cbf2d31a1495ee3d63a55bd81548cd367
parent 5bffa9ca33f8ec98baebca13ee9b16262bfe4e8d
Author: Thomas Pointhuber <thomas.pointhuber@gmx.at>
Date:   Mon, 24 Aug 2015 11:28:55 +0200

[fix] improve result handling of startpage engine

Diffstat:
M
searx/engines/startpage.py
6
+++++-

1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/searx/engines/startpage.py b/searx/engines/startpage.py
@@ -66,7 +66,11 @@ def response(resp):
         url = link.attrib.get('href')
 
         # block google-ad url's
-        if re.match("^http(s|)://www.google.[a-z]+/aclk.*$", url):
+        if re.match("^http(s|)://(www\.)?google\.[a-z]+/aclk.*$", url):
+            continue
+
+        # block startpage search url's
+        if re.match("^http(s|)://(www\.)?startpage\.com/do/search\?.*$", url):
             continue
 
         title = escape(extract_text(link))