2015-02-04 19:41:40 +00:00
|
|
|
# -*- coding: utf-8 -*-
|
|
|
|
|
from collections import defaultdict
|
|
|
|
|
import mock
|
|
|
|
|
from searx.engines import yahoo
|
|
|
|
|
from searx.testing import SearxTestCase
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TestYahooEngine(SearxTestCase):
|
|
|
|
|
|
|
|
|
|
def test_parse_url(self):
|
|
|
|
|
test_url = 'http://r.search.yahoo.com/_ylt=A0LEb9JUSKcAEGRXNyoA;_ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb' +\
|
|
|
|
|
'2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10/RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=' +\
|
|
|
|
|
'dtcJsfP4mEeBOjnVfUQ-'
|
|
|
|
|
url = yahoo.parse_url(test_url)
|
|
|
|
|
self.assertEqual('https://this.is.the.url/', url)
|
|
|
|
|
|
|
|
|
|
test_url = 'http://r.search.yahoo.com/_ylt=A0LElb9JUSKcAEGRXNyoA;_ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb' +\
|
|
|
|
|
'2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10/RU=https%3a%2f%2fthis.is.the.url%2f/RS=' +\
|
|
|
|
|
'dtcJsfP4mEeBOjnVfUQ-'
|
|
|
|
|
url = yahoo.parse_url(test_url)
|
|
|
|
|
self.assertEqual('https://this.is.the.url/', url)
|
|
|
|
|
|
|
|
|
|
test_url = 'https://this.is.the.url/'
|
|
|
|
|
url = yahoo.parse_url(test_url)
|
|
|
|
|
self.assertEqual('https://this.is.the.url/', url)
|
|
|
|
|
|
|
|
|
|
def test_request(self):
|
|
|
|
|
query = 'test_query'
|
|
|
|
|
dicto = defaultdict(dict)
|
|
|
|
|
dicto['pageno'] = 1
|
|
|
|
|
dicto['language'] = 'fr_FR'
|
|
|
|
|
params = yahoo.request(query, dicto)
|
|
|
|
|
self.assertIn('url', params)
|
|
|
|
|
self.assertIn(query, params['url'])
|
|
|
|
|
self.assertIn('search.yahoo.com', params['url'])
|
|
|
|
|
self.assertIn('fr', params['url'])
|
|
|
|
|
self.assertIn('cookies', params)
|
|
|
|
|
self.assertIn('sB', params['cookies'])
|
|
|
|
|
self.assertIn('fr', params['cookies']['sB'])
|
|
|
|
|
|
|
|
|
|
dicto['language'] = 'all'
|
|
|
|
|
params = yahoo.request(query, dicto)
|
|
|
|
|
self.assertIn('cookies', params)
|
|
|
|
|
self.assertIn('sB', params['cookies'])
|
|
|
|
|
self.assertIn('en', params['cookies']['sB'])
|
|
|
|
|
self.assertIn('en', params['url'])
|
|
|
|
|
|
|
|
|
|
def test_response(self):
|
|
|
|
|
self.assertRaises(AttributeError, yahoo.response, None)
|
|
|
|
|
self.assertRaises(AttributeError, yahoo.response, [])
|
|
|
|
|
self.assertRaises(AttributeError, yahoo.response, '')
|
|
|
|
|
self.assertRaises(AttributeError, yahoo.response, '[]')
|
|
|
|
|
|
|
|
|
|
response = mock.Mock(text='<html></html>')
|
|
|
|
|
self.assertEqual(yahoo.response(response), [])
|
|
|
|
|
|
|
|
|
|
html = """
|
2015-02-20 11:34:13 +00:00
|
|
|
<ol class="reg mb-15 searchCenterMiddle">
|
|
|
|
|
<li class="first">
|
|
|
|
|
<div class="dd algo fst Sr">
|
|
|
|
|
<div class="compTitle">
|
|
|
|
|
<h3 class="title"><a class=" td-u" href="http://r.search.yahoo.com/_ylt=A0LEb9JUSKcAEGRXNyoA;
|
|
|
|
|
_ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10
|
|
|
|
|
/RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=dtcJsfP4mEeBOjnVfUQ-"
|
|
|
|
|
target="_blank" data-bid="54e712e13671c">
|
|
|
|
|
<b><b>This is the title</b></b></a>
|
2015-02-04 19:41:40 +00:00
|
|
|
</h3>
|
|
|
|
|
</div>
|
2015-02-20 11:34:13 +00:00
|
|
|
<div class="compText aAbs">
|
|
|
|
|
<p class="lh-18"><b><b>This is the </b>content</b>
|
|
|
|
|
</p>
|
2015-02-04 19:41:40 +00:00
|
|
|
</div>
|
|
|
|
|
</div>
|
2015-02-20 11:34:13 +00:00
|
|
|
</li>
|
|
|
|
|
<li>
|
|
|
|
|
<div class="dd algo lst Sr">
|
|
|
|
|
<div class="compTitle">
|
|
|
|
|
<h3 class="title"><a class=" td-u" href="http://r.search.yahoo.com/_ylt=AwrBT7zgEudUW.wAe2ZXNyoA;
|
|
|
|
|
_ylu=X3oDMTBybGY3bmpvBGNvbG8DYmYxBHBvcwMyBHZ0aWQDBHNlYwNzcg--/RV=2\/RE=1424458593/RO=10
|
|
|
|
|
/RU=https%3a%2f%2fthis.is.the.second.url%2f/RK=0/RS=jIctjj_cBH1Efj88GCgHKp3__Qk-"
|
|
|
|
|
target="_blank" data-bid="54e712e136926">
|
|
|
|
|
This is the second <b><b>title</b></b></a>
|
|
|
|
|
</h3>
|
|
|
|
|
</div>
|
|
|
|
|
<div class="compText aAbs">
|
|
|
|
|
<p class="lh-18">This is the second content</p>
|
|
|
|
|
</div>
|
2015-02-04 19:41:40 +00:00
|
|
|
</div>
|
2015-02-20 11:34:13 +00:00
|
|
|
</li>
|
|
|
|
|
</ol>
|
|
|
|
|
<div class="dd assist fst lst AlsoTry" data-bid="54e712e138d04">
|
|
|
|
|
<div class="compTitle mb-4 h-17">
|
|
|
|
|
<h3 class="title">Also Try</h3> </div>
|
|
|
|
|
<table class="compTable m-0 ac-1st td-u fz-ms">
|
|
|
|
|
<tbody>
|
|
|
|
|
<tr>
|
|
|
|
|
<td class="w-50p pr-28"><a href="https://search.yahoo.com/"><B>This is the </B>suggestion<B></B></a>
|
|
|
|
|
</td>
|
|
|
|
|
</tr>
|
|
|
|
|
</table>
|
|
|
|
|
</div>
|
2015-02-04 19:41:40 +00:00
|
|
|
"""
|
|
|
|
|
response = mock.Mock(text=html)
|
|
|
|
|
results = yahoo.response(response)
|
2015-02-20 11:34:13 +00:00
|
|
|
print results
|
2015-02-04 19:41:40 +00:00
|
|
|
self.assertEqual(type(results), list)
|
2015-02-20 11:34:13 +00:00
|
|
|
self.assertEqual(len(results), 3)
|
2015-02-04 19:41:40 +00:00
|
|
|
self.assertEqual(results[0]['title'], 'This is the title')
|
|
|
|
|
self.assertEqual(results[0]['url'], 'https://this.is.the.url/')
|
|
|
|
|
self.assertEqual(results[0]['content'], 'This is the content')
|
2015-02-20 11:34:13 +00:00
|
|
|
self.assertEqual(results[1]['title'], 'This is the second title')
|
|
|
|
|
self.assertEqual(results[1]['url'], 'https://this.is.the.second.url/')
|
|
|
|
|
self.assertEqual(results[1]['content'], 'This is the second content')
|
|
|
|
|
self.assertEqual(results[2]['suggestion'], 'This is the suggestion')
|
2015-02-04 19:41:40 +00:00
|
|
|
|
|
|
|
|
html = """
|
2015-02-20 11:34:13 +00:00
|
|
|
<ol class="reg mb-15 searchCenterMiddle">
|
|
|
|
|
<li class="first">
|
|
|
|
|
<div class="dd algo fst Sr">
|
|
|
|
|
<div class="compTitle">
|
|
|
|
|
<h3 class="title"><a class=" td-u" href="http://r.search.yahoo.com/_ylt=A0LEb9JUSKcAEGRXNyoA;
|
|
|
|
|
_ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10
|
|
|
|
|
/RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=dtcJsfP4mEeBOjnVfUQ-"
|
|
|
|
|
target="_blank" data-bid="54e712e13671c">
|
|
|
|
|
<b><b>This is the title</b></b></a>
|
2015-02-04 19:41:40 +00:00
|
|
|
</h3>
|
|
|
|
|
</div>
|
2015-02-20 11:34:13 +00:00
|
|
|
<div class="compText aAbs">
|
|
|
|
|
<p class="lh-18"><b><b>This is the </b>content</b>
|
|
|
|
|
</p>
|
2015-02-04 19:41:40 +00:00
|
|
|
</div>
|
|
|
|
|
</div>
|
2015-02-20 11:34:13 +00:00
|
|
|
</li>
|
|
|
|
|
</ol>
|
2015-02-04 19:41:40 +00:00
|
|
|
"""
|
|
|
|
|
response = mock.Mock(text=html)
|
|
|
|
|
results = yahoo.response(response)
|
|
|
|
|
self.assertEqual(type(results), list)
|
|
|
|
|
self.assertEqual(len(results), 1)
|
|
|
|
|
self.assertEqual(results[0]['title'], 'This is the title')
|
|
|
|
|
self.assertEqual(results[0]['url'], 'https://this.is.the.url/')
|
|
|
|
|
self.assertEqual(results[0]['content'], 'This is the content')
|
|
|
|
|
|
|
|
|
|
html = """
|
|
|
|
|
<li class="b_algo" u="0|5109|4755453613245655|UAGjXgIrPH5yh-o5oNHRx_3Zta87f_QO">
|
|
|
|
|
</li>
|
|
|
|
|
"""
|
|
|
|
|
response = mock.Mock(text=html)
|
|
|
|
|
results = yahoo.response(response)
|
|
|
|
|
self.assertEqual(type(results), list)
|
|
|
|
|
self.assertEqual(len(results), 0)
|
