git.iko.soy/searx
on master help sign in
Yahoo's unit test
9f13af8Cqoicebordel
searx/engines/yahoo.py+2−2
compare with master
unifiedsplit
@@ -35,7 +35,7 @@ suggestion_xpath = '//div[@id="satat"]//a'
35def parse_url(url_string):
35def parse_url(url_string):
36 endings = ['/RS', '/RK']
36 endings = ['/RS', '/RK']
37 endpositions = []
37 endpositions = []
38− start = url_string.find('http', url_string.find('/RU=')+1)
38+ start = url_string.find('http', url_string.find('/RU=') + 1)
39
39
40 for ending in endings:
40 for ending in endings:
41 endpos = url_string.rfind(ending)
41 endpos = url_string.rfind(ending)
@@ -91,7 +91,7 @@ def response(resp):
91 'content': content})
91 'content': content})
92
92
93 # if no suggestion found, return results
93 # if no suggestion found, return results
94− if not suggestion_xpath:
94+ if not dom.xpath(suggestion_xpath):
95 return results
95 return results
96
96
97 # parse suggestion
97 # parse suggestion
searx/tests/engines/test_yahoo.pynew file+154
@@ -0,0 +1,154 @@
1+# -*- coding: utf-8 -*-
2+from collections import defaultdict
3+import mock
4+from searx.engines import yahoo
5+from searx.testing import SearxTestCase
6+
7+
8+class TestYahooEngine(SearxTestCase):
9+
10+ def test_parse_url(self):
11+ test_url = 'http://r.search.yahoo.com/_ylt=A0LEb9JUSKcAEGRXNyoA;_ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb' +\
12+ '2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10/RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=' +\
13+ 'dtcJsfP4mEeBOjnVfUQ-'
14+ url = yahoo.parse_url(test_url)
15+ self.assertEqual('https://this.is.the.url/', url)
16+
17+ test_url = 'http://r.search.yahoo.com/_ylt=A0LElb9JUSKcAEGRXNyoA;_ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb' +\
18+ '2xvA2Jm2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10/RU=https%3a%2f%2fthis.is.the.url%2f/RS=' +\
19+ 'dtcJsfP4mEeBOjnVfUQ-'
20+ url = yahoo.parse_url(test_url)
21+ self.assertEqual('https://this.is.the.url/', url)
22+
23+ test_url = 'https://this.is.the.url/'
24+ url = yahoo.parse_url(test_url)
25+ self.assertEqual('https://this.is.the.url/', url)
26+
27+ def test_request(self):
28+ query = 'test_query'
29+ dicto = defaultdict(dict)
30+ dicto['pageno'] = 1
31+ dicto['language'] = 'fr_FR'
32+ params = yahoo.request(query, dicto)
33+ self.assertIn('url', params)
34+ self.assertIn(query, params['url'])
35+ self.assertIn('search.yahoo.com', params['url'])
36+ self.assertIn('fr', params['url'])
37+ self.assertIn('cookies', params)
38+ self.assertIn('sB', params['cookies'])
39+ self.assertIn('fr', params['cookies']['sB'])
40+
41+ dicto['language'] = 'all'
42+ params = yahoo.request(query, dicto)
43+ self.assertIn('cookies', params)
44+ self.assertIn('sB', params['cookies'])
45+ self.assertIn('en', params['cookies']['sB'])
46+ self.assertIn('en', params['url'])
47+
48+ def test_response(self):
49+ self.assertRaises(AttributeError, yahoo.response, None)
50+ self.assertRaises(AttributeError, yahoo.response, [])
51+ self.assertRaises(AttributeError, yahoo.response, '')
52+ self.assertRaises(AttributeError, yahoo.response, '[]')
53+
54+ response = mock.Mock(text='<html></html>')
55+ self.assertEqual(yahoo.response(response), [])
56+
57+ html = """
58+ <div class="res">
59+ <div>
60+ <h3>
61+ <a id="link-1" class="yschttl spt" href="http://r.search.yahoo.com/_ylt=A0LEVzClb9JUSKcAEGRXNyoA;
62+ _ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb2xvA2JmMQR2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10
63+ /RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=dtcJsfP4mEeBOjnVfUQ-"target="_blank" data-bk="5063.1">
64+ <b>This</b> is the title
65+ </a>
66+ </h3>
67+ </div>
68+ <span class="url" dir="ltr">www.<b>test</b>.com</span>
69+ <div class="abstr">
70+ <b>This</b> is the content
71+ </div>
72+ </div>
73+ <div id="satat" data-bns="Yahoo" data-bk="124.1">
74+ <h2>Also Try</h2>
75+ <table>
76+ <tbody>
77+ <tr>
78+ <td>
79+ <a id="srpnat0" class="" href="https://search.yahoo.com/search=rs-bottom" >
80+ <span>
81+ <b></b>This is <b>the suggestion</b>
82+ </span>
83+ </a>
84+ </td>
85+ </tr>
86+ </tbody>
87+ </table>
88+ </div>
89+ """
90+ response = mock.Mock(text=html)
91+ results = yahoo.response(response)
92+ self.assertEqual(type(results), list)
93+ self.assertEqual(len(results), 2)
94+ self.assertEqual(results[0]['title'], 'This is the title')
95+ self.assertEqual(results[0]['url'], 'https://this.is.the.url/')
96+ self.assertEqual(results[0]['content'], 'This is the content')
97+ self.assertEqual(results[1]['suggestion'], 'This is the suggestion')
98+
99+ html = """
100+ <div class="res">
101+ <div>
102+ <h3>
103+ <a id="link-1" class="yschttl spt" href="http://r.search.yahoo.com/_ylt=A0LEVzClb9JUSKcAEGRXNyoA;
104+ _ylu=X3oDMTEzZm1qazYwBHNlYwNzcgRwb3MDMQRjb2xvA2JmMQR2dGlkA1NNRTcwM18x/RV=2/RE=1423106085/RO=10
105+ /RU=https%3a%2f%2fthis.is.the.url%2f/RK=0/RS=dtcJsfP4mEeBOjnVfUQ-"target="_blank" data-bk="5063.1">
106+ <b>This</b> is the title
107+ </a>
108+ </h3>
109+ </div>
110+ <span class="url" dir="ltr">www.<b>test</b>.com</span>
111+ <div class="abstr">
112+ <b>This</b> is the content
113+ </div>
114+ </div>
115+ <div class="res">
116+ <div>
117+ <h3>
118+ <a id="link-1" class="yschttl spt">
119+ <b>This</b> is the title
120+ </a>
121+ </h3>
122+ </div>
123+ <span class="url" dir="ltr">www.<b>test</b>.com</span>
124+ <div class="abstr">
125+ <b>This</b> is the content
126+ </div>
127+ </div>
128+ <div class="res">
129+ <div>
130+ <h3>
131+ </h3>
132+ </div>
133+ <span class="url" dir="ltr">www.<b>test</b>.com</span>
134+ <div class="abstr">
135+ <b>This</b> is the content
136+ </div>
137+ </div>
138+ """
139+ response = mock.Mock(text=html)
140+ results = yahoo.response(response)
141+ self.assertEqual(type(results), list)
142+ self.assertEqual(len(results), 1)
143+ self.assertEqual(results[0]['title'], 'This is the title')
144+ self.assertEqual(results[0]['url'], 'https://this.is.the.url/')
145+ self.assertEqual(results[0]['content'], 'This is the content')
146+
147+ html = """
148+ <li class="b_algo" u="0|5109|4755453613245655|UAGjXgIrPH5yh-o5oNHRx_3Zta87f_QO">
149+ </li>
150+ """
151+ response = mock.Mock(text=html)
152+ results = yahoo.response(response)
153+ self.assertEqual(type(results), list)
154+ self.assertEqual(len(results), 0)
searx/tests/test_engines.py+1
@@ -27,4 +27,5 @@ from searx.tests.engines.test_twitter import * # noqa
27from searx.tests.engines.test_vimeo import * # noqa
27from searx.tests.engines.test_vimeo import * # noqa
28from searx.tests.engines.test_www500px import * # noqa
28from searx.tests.engines.test_www500px import * # noqa
29from searx.tests.engines.test_youtube import * # noqa
29from searx.tests.engines.test_youtube import * # noqa
30+from searx.tests.engines.test_yahoo import * # noqa
30from searx.tests.engines.test_yahoo_news import * # noqa
31from searx.tests.engines.test_yahoo_news import * # noqa