Ponysearch/searx/tests/engines/test_google.py

164 lines
6.3 KiB
Python
Raw Normal View History

2015-02-11 17:16:52 +01:00
# -*- coding: utf-8 -*-
from collections import defaultdict
import mock
import lxml
from searx.engines import google
from searx.testing import SearxTestCase
class TestGoogleEngine(SearxTestCase):
def test_request(self):
query = 'test_query'
dicto = defaultdict(dict)
dicto['pageno'] = 1
dicto['language'] = 'fr_FR'
params = google.request(query, dicto)
self.assertIn('url', params)
self.assertIn(query, params['url'])
self.assertIn('google.com', params['url'])
2015-05-02 13:21:01 +02:00
self.assertNotIn('PREF', params['cookies'])
2015-02-11 17:16:52 +01:00
self.assertIn('fr', params['headers']['Accept-Language'])
dicto['language'] = 'all'
params = google.request(query, dicto)
self.assertIn('en', params['headers']['Accept-Language'])
2015-05-02 13:21:01 +02:00
self.assertIn('PREF', params['cookies'])
2015-02-11 17:16:52 +01:00
def test_response(self):
self.assertRaises(AttributeError, google.response, None)
self.assertRaises(AttributeError, google.response, [])
self.assertRaises(AttributeError, google.response, '')
self.assertRaises(AttributeError, google.response, '[]')
response = mock.Mock(text='<html></html>')
self.assertEqual(google.response(response), [])
html = """
<li class="g">
<h3 class="r">
<a href="http://this.should.be.the.link/">
<b>This</b> is <b>the</b> title
</a>
</h3>
<div class="s">
<div class="kv" style="margin-bottom:2px">
<cite>
<b>test</b>.psychologies.com/
</cite>
<div class="_nBb">
<div style="display:inline" onclick="google.sham(this);" aria-expanded="false"
aria-haspopup="true" tabindex="0" data-ved="0CBUQ7B0wAA">
<span class="_O0">
</span>
</div>
<div style="display:none" class="am-dropdown-menu" role="menu" tabindex="-1">
<ul>
<li class="_Ykb">
<a class="_Zkb" href="http://www.google.fr/url?url=http://webcache.googleusercontent
.com/search%3Fcache:R1Z_4pGXjuIJ:http://test.psychologies.com/">
En cache
</a>
</li>
<li class="_Ykb">
<a class="_Zkb" href="/search?safe=off&amp;q=related:test.psy.com/">
Pages similaires
</a>
</li>
</ul>
</div>
</div>
</div>
<span class="st">
This should be the content.
</span>
<br>
<div class="osl">
<a href="http://www.google.fr/url?url=http://test.psychologies.com/tests/">
Test Personnalité
</a> -
<a href="http://www.google.fr/url?url=http://test.psychologies.com/test/">
Tests - Moi
</a> -
<a href="http://www.google.fr/url?url=http://test.psychologies.com/test/tests-couple">
Test Couple
</a>
-
<a href="http://www.google.fr/url?url=http://test.psychologies.com/tests/tests-amour">
Test Amour
</a>
</div>
</div>
</li>
<li class="g">
<h3 class="r">
<a href="http://www.google.com/images?q=toto">
<b>This</b>
</a>
</h3>
</li>
<li class="g">
<h3 class="r">
<a href="http://www.google.com/search?q=toto">
<b>This</b> is
</a>
</h3>
</li>
<li class="g">
<h3 class="r">
<a href="">
<b>This</b> is <b>the</b>
</a>
</h3>
</li>
<li class="g">
<h3 class="r">
<a href="/url?q=url">
<b>This</b> is <b>the</b>
</a>
</h3>
</li>
<p class="_Bmc" style="margin:3px 8px">
<a href="/search?num=20&amp;safe=off&amp;q=t&amp;revid=1754833769&amp;sa=X&amp;ei=-&amp;ved=">
suggestion <b>title</b>
</a>
</p>
"""
response = mock.Mock(text=html)
results = google.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 2)
self.assertEqual(results[0]['title'], 'This is the title')
self.assertEqual(results[0]['url'], 'http://this.should.be.the.link/')
self.assertEqual(results[0]['content'], 'This should be the content.')
self.assertEqual(results[1]['suggestion'], 'suggestion title')
html = """
<li class="b_algo" u="0|5109|4755453613245655|UAGjXgIrPH5yh-o5oNHRx_3Zta87f_QO">
</li>
"""
response = mock.Mock(text=html)
results = google.response(response)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 0)
def test_parse_images(self):
html = """
<li>
<div>
<a href="http://www.google.com/url?q=http://this.is.the.url/">
<img style="margin:3px 0;margin-right:6px;padding:0" height="90"
src="https://this.is.the.image/image.jpg" width="60" align="middle" alt="" border="0">
</a>
</div>
</li>
"""
dom = lxml.html.fromstring(html)
results = google.parse_images(dom)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)
self.assertEqual(results[0]['url'], 'http://this.is.the.url/')
self.assertEqual(results[0]['title'], '')
self.assertEqual(results[0]['content'], '')
self.assertEqual(results[0]['img_src'], 'https://this.is.the.image/image.jpg')