forked from Ponysearch/Ponysearch
Startpage's unit test
This commit is contained in:
Cqoicebordel
parent
3a4d6045c1
commit
f1c10f4fe4
3 changed files with 146 additions and 8 deletions
|
|
@ -13,6 +13,7 @@
|
||||||
from lxml import html
|
from lxml import html
|
||||||
from cgi import escape
|
from cgi import escape
|
||||||
import re
|
import re
|
||||||
|
from searx.engines.xpath import extract_text
|
||||||
|
|
||||||
# engine dependent config
|
# engine dependent config
|
||||||
categories = ['general']
|
categories = ['general']
|
||||||
|
|
@ -45,8 +46,7 @@ def request(query, params):
|
||||||
|
|
||||||
# set language if specified
|
# set language if specified
|
||||||
if params['language'] != 'all':
|
if params['language'] != 'all':
|
||||||
params['data']['with_language'] = ('lang_' +
|
params['data']['with_language'] = ('lang_' + params['language'].split('_')[0])
|
||||||
params['language'].split('_')[0])
|
|
||||||
|
|
||||||
return params
|
return params
|
||||||
|
|
||||||
|
|
@ -64,18 +64,15 @@ def response(resp):
|
||||||
continue
|
continue
|
||||||
link = links[0]
|
link = links[0]
|
||||||
url = link.attrib.get('href')
|
url = link.attrib.get('href')
|
||||||
try:
|
|
||||||
title = escape(link.text_content())
|
|
||||||
except UnicodeDecodeError:
|
|
||||||
continue
|
|
||||||
|
|
||||||
# block google-ad url's
|
# 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
|
continue
|
||||||
|
|
||||||
|
title = escape(extract_text(link))
|
||||||
|
|
||||||
if result.xpath('./p[@class="desc"]'):
|
if result.xpath('./p[@class="desc"]'):
|
||||||
content = escape(result.xpath('./p[@class="desc"]')[0]
|
content = escape(extract_text(result.xpath('./p[@class="desc"]')))
|
||||||
.text_content())
|
|
||||||
else:
|
else:
|
||||||
content = ''
|
content = ''
|
||||||
|
|
||||||
|
|
|
||||||
140
searx/tests/engines/test_startpage.py
Normal file
140
searx/tests/engines/test_startpage.py
Normal file
|
|
@ -0,0 +1,140 @@
|
||||||
|
# -*- coding: utf-8 -*-
|
||||||
|
from collections import defaultdict
|
||||||
|
import mock
|
||||||
|
from searx.engines import startpage
|
||||||
|
from searx.testing import SearxTestCase
|
||||||
|
|
||||||
|
|
||||||
|
class TestStartpageEngine(SearxTestCase):
|
||||||
|
|
||||||
|
def test_request(self):
|
||||||
|
query = 'test_query'
|
||||||
|
dicto = defaultdict(dict)
|
||||||
|
dicto['pageno'] = 1
|
||||||
|
dicto['language'] = 'fr_FR'
|
||||||
|
params = startpage.request(query, dicto)
|
||||||
|
self.assertIn('url', params)
|
||||||
|
self.assertIn('startpage.com', params['url'])
|
||||||
|
self.assertIn('data', params)
|
||||||
|
self.assertIn('query', params['data'])
|
||||||
|
self.assertIn(query, params['data']['query'])
|
||||||
|
self.assertIn('with_language', params['data'])
|
||||||
|
self.assertIn('lang_fr', params['data']['with_language'])
|
||||||
|
|
||||||
|
dicto['language'] = 'all'
|
||||||
|
params = startpage.request(query, dicto)
|
||||||
|
self.assertNotIn('with_language', params['data'])
|
||||||
|
|
||||||
|
def test_response(self):
|
||||||
|
self.assertRaises(AttributeError, startpage.response, None)
|
||||||
|
self.assertRaises(AttributeError, startpage.response, [])
|
||||||
|
self.assertRaises(AttributeError, startpage.response, '')
|
||||||
|
self.assertRaises(AttributeError, startpage.response, '[]')
|
||||||
|
|
||||||
|
response = mock.Mock(content='<html></html>')
|
||||||
|
self.assertEqual(startpage.response(response), [])
|
||||||
|
|
||||||
|
html = """
|
||||||
|
<div class='result' style=' *width : auto; *margin-right : 10%;'>
|
||||||
|
<h3>
|
||||||
|
<a href='http://this.should.be.the.link/' id='title_2' name='title_2' >
|
||||||
|
This should be the title
|
||||||
|
</a>
|
||||||
|
<span id='title_stars_2' name='title_stars_2'> </span>
|
||||||
|
</h3>
|
||||||
|
<p class='desc'>
|
||||||
|
This should be the content.
|
||||||
|
</p>
|
||||||
|
<p>
|
||||||
|
<span class='url'>www.speed<b>test</b>.net/fr/
|
||||||
|
</span>
|
||||||
|
-
|
||||||
|
<A class="proxy" id="proxy_link" HREF="https://ixquick-proxy.com/do/spg/proxy?ep=&edata=&ek=&ekdata="
|
||||||
|
class='proxy'>
|
||||||
|
Navigation avec Ixquick Proxy
|
||||||
|
</A>
|
||||||
|
-
|
||||||
|
<A HREF="https://ixquick-proxy.com/do/spg/highlight.pl?l=francais&c=hf&cat=web&q=test&rl=NONE&rid=
|
||||||
|
&hlq=https://startpage.com/do/search&mtabp=-1&mtcmd=process_search&mtlanguage=francais&mtengine0=
|
||||||
|
&mtcat=web&u=http:%2F%2Fwww.speedtest.net%2Ffr%2F" class='proxy'>
|
||||||
|
Mis en surbrillance
|
||||||
|
</A>
|
||||||
|
</p>
|
||||||
|
</div>
|
||||||
|
"""
|
||||||
|
response = mock.Mock(content=html)
|
||||||
|
results = startpage.response(response)
|
||||||
|
self.assertEqual(type(results), list)
|
||||||
|
self.assertEqual(len(results), 1)
|
||||||
|
self.assertEqual(results[0]['title'], 'This should be the title')
|
||||||
|
self.assertEqual(results[0]['url'], 'http://this.should.be.the.link/')
|
||||||
|
self.assertEqual(results[0]['content'], 'This should be the content.')
|
||||||
|
|
||||||
|
html = """
|
||||||
|
<div class='result' style=' *width : auto; *margin-right : 10%;'>
|
||||||
|
<h3>
|
||||||
|
<a href='http://www.google.com/aclk?sa=l&ai=C' id='title_2' name='title_2' >
|
||||||
|
This should be the title
|
||||||
|
</a>
|
||||||
|
<span id='title_stars_2' name='title_stars_2'> </span>
|
||||||
|
</h3>
|
||||||
|
<p class='desc'>
|
||||||
|
This should be the content.
|
||||||
|
</p>
|
||||||
|
<p>
|
||||||
|
<span class='url'>www.speed<b>test</b>.net/fr/
|
||||||
|
</span>
|
||||||
|
-
|
||||||
|
<A class="proxy" id="proxy_link" HREF="https://ixquick-proxy.com/do/spg/proxy?ep=&edata=&ek=&ekdata="
|
||||||
|
class='proxy'>
|
||||||
|
Navigation avec Ixquick Proxy
|
||||||
|
</A>
|
||||||
|
-
|
||||||
|
<A HREF="https://ixquick-proxy.com/do/spg/highlight.pl?l=francais&c=hf&cat=web&q=test&rl=NONE&rid=
|
||||||
|
&hlq=https://startpage.com/do/search&mtabp=-1&mtcmd=process_search&mtlanguage=francais&mtengine0=
|
||||||
|
&mtcat=web&u=http:%2F%2Fwww.speedtest.net%2Ffr%2F" class='proxy'>
|
||||||
|
Mis en surbrillance
|
||||||
|
</A>
|
||||||
|
</p>
|
||||||
|
</div>
|
||||||
|
<div class='result' style=' *width : auto; *margin-right : 10%;'>
|
||||||
|
<h3>
|
||||||
|
<span id='title_stars_2' name='title_stars_2'> </span>
|
||||||
|
</h3>
|
||||||
|
<p class='desc'>
|
||||||
|
This should be the content.
|
||||||
|
</p>
|
||||||
|
<p>
|
||||||
|
<span class='url'>www.speed<b>test</b>.net/fr/
|
||||||
|
</span>
|
||||||
|
</p>
|
||||||
|
</div>
|
||||||
|
<div class='result' style=' *width : auto; *margin-right : 10%;'>
|
||||||
|
<h3>
|
||||||
|
<a href='http://this.should.be.the.link/' id='title_2' name='title_2' >
|
||||||
|
This should be the title
|
||||||
|
</a>
|
||||||
|
<span id='title_stars_2' name='title_stars_2'> </span>
|
||||||
|
</h3>
|
||||||
|
<p>
|
||||||
|
<span class='url'>www.speed<b>test</b>.net/fr/
|
||||||
|
</span>
|
||||||
|
-
|
||||||
|
<A class="proxy" id="proxy_link" HREF="https://ixquick-proxy.com/do/spg/proxy?ep=&edata=&ek=&ekdata="
|
||||||
|
class='proxy'>
|
||||||
|
Navigation avec Ixquick Proxy
|
||||||
|
</A>
|
||||||
|
-
|
||||||
|
<A HREF="https://ixquick-proxy.com/do/spg/highlight.pl?l=francais&c=hf&cat=web&q=test&rl=NONE&rid=
|
||||||
|
&hlq=https://startpage.com/do/search&mtabp=-1&mtcmd=process_search&mtlanguage=francais&mtengine0=
|
||||||
|
&mtcat=web&u=http:%2F%2Fwww.speedtest.net%2Ffr%2F" class='proxy'>
|
||||||
|
Mis en surbrillance
|
||||||
|
</A>
|
||||||
|
</p>
|
||||||
|
</div>
|
||||||
|
"""
|
||||||
|
response = mock.Mock(content=html)
|
||||||
|
results = startpage.response(response)
|
||||||
|
self.assertEqual(type(results), list)
|
||||||
|
self.assertEqual(len(results), 1)
|
||||||
|
self.assertEqual(results[0]['content'], '')
|
||||||
|
|
@ -23,6 +23,7 @@ from searx.tests.engines.test_searchcode_code import * # noqa
|
||||||
from searx.tests.engines.test_searchcode_doc import * # noqa
|
from searx.tests.engines.test_searchcode_doc import * # noqa
|
||||||
from searx.tests.engines.test_soundcloud import * # noqa
|
from searx.tests.engines.test_soundcloud import * # noqa
|
||||||
from searx.tests.engines.test_stackoverflow import * # noqa
|
from searx.tests.engines.test_stackoverflow import * # noqa
|
||||||
|
from searx.tests.engines.test_startpage import * # noqa
|
||||||
from searx.tests.engines.test_subtitleseeker import * # noqa
|
from searx.tests.engines.test_subtitleseeker import * # noqa
|
||||||
from searx.tests.engines.test_twitter import * # noqa
|
from searx.tests.engines.test_twitter import * # noqa
|
||||||
from searx.tests.engines.test_vimeo import * # noqa
|
from searx.tests.engines.test_vimeo import * # noqa
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue