summaryrefslogtreecommitdiff
path: root/searx/engines
diff options
context:
space:
mode:
authorCqoicebordel <Cqoicebordel@users.noreply.github.com>2015-02-06 17:31:10 +0100
committerCqoicebordel <Cqoicebordel@users.noreply.github.com>2015-02-06 17:31:10 +0100
commitf1c10f4fe45f34c12994b9bbc4aca133202fd7ca (patch)
tree155243afe2b7ed1d98160d61f664101214ec47eb /searx/engines
parent3a4d6045c1da950d13d1d14192247389c5932631 (diff)
Startpage's unit test
Diffstat (limited to 'searx/engines')
-rw-r--r--searx/engines/startpage.py13
1 files changed, 5 insertions, 8 deletions
diff --git a/searx/engines/startpage.py b/searx/engines/startpage.py
index d60ecd978..9d5b4befe 100644
--- a/searx/engines/startpage.py
+++ b/searx/engines/startpage.py
@@ -13,6 +13,7 @@
from lxml import html
from cgi import escape
import re
+from searx.engines.xpath import extract_text
# engine dependent config
categories = ['general']
@@ -45,8 +46,7 @@ def request(query, params):
# set language if specified
if params['language'] != 'all':
- params['data']['with_language'] = ('lang_' +
- params['language'].split('_')[0])
+ params['data']['with_language'] = ('lang_' + params['language'].split('_')[0])
return params
@@ -64,18 +64,15 @@ def response(resp):
continue
link = links[0]
url = link.attrib.get('href')
- try:
- title = escape(link.text_content())
- except UnicodeDecodeError:
- continue
# block google-ad url's
if re.match("^http(s|)://www.google.[a-z]+/aclk.*$", url):
continue
+ title = escape(extract_text(link))
+
if result.xpath('./p[@class="desc"]'):
- content = escape(result.xpath('./p[@class="desc"]')[0]
- .text_content())
+ content = escape(extract_text(result.xpath('./p[@class="desc"]')))
else:
content = ''