|
@@ -0,0 +1,143 @@
|
|
1
|
+# -*- coding: utf-8 -*-
|
|
2
|
+from collections import defaultdict
|
|
3
|
+from datetime import datetime
|
|
4
|
+import mock
|
|
5
|
+from searx.engines import yahoo_news
|
|
6
|
+from searx.testing import SearxTestCase
|
|
7
|
+
|
|
8
|
+
|
|
9
|
+class TestYahooNewsEngine(SearxTestCase):
|
|
10
|
+
|
|
11
|
+ def test_request(self):
|
|
12
|
+ query = 'test_query'
|
|
13
|
+ dicto = defaultdict(dict)
|
|
14
|
+ dicto['pageno'] = 1
|
|
15
|
+ dicto['language'] = 'fr_FR'
|
|
16
|
+ params = yahoo_news.request(query, dicto)
|
|
17
|
+ self.assertIn('url', params)
|
|
18
|
+ self.assertIn(query, params['url'])
|
|
19
|
+ self.assertIn('news.search.yahoo.com', params['url'])
|
|
20
|
+ self.assertIn('fr', params['url'])
|
|
21
|
+ self.assertIn('cookies', params)
|
|
22
|
+ self.assertIn('sB', params['cookies'])
|
|
23
|
+ self.assertIn('fr', params['cookies']['sB'])
|
|
24
|
+
|
|
25
|
+ dicto['language'] = 'all'
|
|
26
|
+ params = yahoo_news.request(query, dicto)
|
|
27
|
+ self.assertIn('cookies', params)
|
|
28
|
+ self.assertIn('sB', params['cookies'])
|
|
29
|
+ self.assertIn('en', params['cookies']['sB'])
|
|
30
|
+ self.assertIn('en', params['url'])
|
|
31
|
+
|
|
32
|
+ def test_response(self):
|
|
33
|
+ self.assertRaises(AttributeError, yahoo_news.response, None)
|
|
34
|
+ self.assertRaises(AttributeError, yahoo_news.response, [])
|
|
35
|
+ self.assertRaises(AttributeError, yahoo_news.response, '')
|
|
36
|
+ self.assertRaises(AttributeError, yahoo_news.response, '[]')
|
|
37
|
+
|
|
38
|
+ response = mock.Mock(text='<html></html>')
|
|
39
|
+ self.assertEqual(yahoo_news.response(response), [])
|
|
40
|
+
|
|
41
|
+ html = """
|
|
42
|
+ <div class="res">
|
|
43
|
+ <div>
|
|
44
|
+ <h3>
|
|
45
|
+ <a class="yschttl spt" href="http://this.is.the.url" target="_blank">
|
|
46
|
+ This is
|
|
47
|
+ the <b>title</b>...
|
|
48
|
+ </a>
|
|
49
|
+ </h3>
|
|
50
|
+ </div>
|
|
51
|
+ <span class="url">Business via Yahoo! Finance</span> <span class="timestamp">Feb 03 09:45am</span>
|
|
52
|
+ <div class="abstr">
|
|
53
|
+ This is the content
|
|
54
|
+ </div>
|
|
55
|
+ </div>
|
|
56
|
+ """
|
|
57
|
+ response = mock.Mock(text=html)
|
|
58
|
+ results = yahoo_news.response(response)
|
|
59
|
+ self.assertEqual(type(results), list)
|
|
60
|
+ self.assertEqual(len(results), 1)
|
|
61
|
+ self.assertEqual(results[0]['title'], 'This is the title...')
|
|
62
|
+ self.assertEqual(results[0]['url'], 'http://this.is.the.url/')
|
|
63
|
+ self.assertEqual(results[0]['content'], 'This is the content')
|
|
64
|
+
|
|
65
|
+ html = """
|
|
66
|
+ <div class="res">
|
|
67
|
+ <div>
|
|
68
|
+ <h3>
|
|
69
|
+ <a class="yschttl spt" href="http://this.is.the.url" target="_blank">
|
|
70
|
+ This is
|
|
71
|
+ the <b>title</b>...
|
|
72
|
+ </a>
|
|
73
|
+ </h3>
|
|
74
|
+ </div>
|
|
75
|
+ <span class="url">Business via Yahoo!</span> <span class="timestamp">2 hours, 22 minutes ago</span>
|
|
76
|
+ <div class="abstr">
|
|
77
|
+ This is the content
|
|
78
|
+ </div>
|
|
79
|
+ </div>
|
|
80
|
+ <div class="res">
|
|
81
|
+ <div>
|
|
82
|
+ <h3>
|
|
83
|
+ <a class="yschttl spt" href="http://this.is.the.url" target="_blank">
|
|
84
|
+ This is
|
|
85
|
+ the <b>title</b>...
|
|
86
|
+ </a>
|
|
87
|
+ </h3>
|
|
88
|
+ </div>
|
|
89
|
+ <span class="url">Business via Yahoo!</span> <span class="timestamp">22 minutes ago</span>
|
|
90
|
+ <div class="abstr">
|
|
91
|
+ This is the content
|
|
92
|
+ </div>
|
|
93
|
+ </div>
|
|
94
|
+ <div class="res">
|
|
95
|
+ <div>
|
|
96
|
+ <h3>
|
|
97
|
+ <a class="yschttl spt" href="http://this.is.the.url" target="_blank">
|
|
98
|
+ This is
|
|
99
|
+ the <b>title</b>...
|
|
100
|
+ </a>
|
|
101
|
+ </h3>
|
|
102
|
+ </div>
|
|
103
|
+ <span class="url">Business via Yahoo!</span> <span class="timestamp">Feb 03 09:45am 1900</span>
|
|
104
|
+ <div class="abstr">
|
|
105
|
+ This is the content
|
|
106
|
+ </div>
|
|
107
|
+ </div>
|
|
108
|
+ """
|
|
109
|
+ response = mock.Mock(text=html)
|
|
110
|
+ results = yahoo_news.response(response)
|
|
111
|
+ self.assertEqual(type(results), list)
|
|
112
|
+ self.assertEqual(len(results), 3)
|
|
113
|
+ self.assertEqual(results[0]['title'], 'This is the title...')
|
|
114
|
+ self.assertEqual(results[0]['url'], 'http://this.is.the.url/')
|
|
115
|
+ self.assertEqual(results[0]['content'], 'This is the content')
|
|
116
|
+ self.assertEqual(results[2]['publishedDate'].year, datetime.now().year)
|
|
117
|
+
|
|
118
|
+ html = """
|
|
119
|
+ <li class="b_algo" u="0|5109|4755453613245655|UAGjXgIrPH5yh-o5oNHRx_3Zta87f_QO">
|
|
120
|
+ <div Class="sa_mc">
|
|
121
|
+ <div class="sb_tlst">
|
|
122
|
+ <h2>
|
|
123
|
+ <a href="http://this.should.be.the.link/" h="ID=SERP,5124.1">
|
|
124
|
+ <strong>This</strong> should be the title</a>
|
|
125
|
+ </h2>
|
|
126
|
+ </div>
|
|
127
|
+ <div class="sb_meta">
|
|
128
|
+ <cite>
|
|
129
|
+ <strong>this</strong>.meta.com</cite>
|
|
130
|
+ <span class="c_tlbxTrg">
|
|
131
|
+ <span class="c_tlbxH" H="BASE:CACHEDPAGEDEFAULT" K="SERP,5125.1">
|
|
132
|
+ </span>
|
|
133
|
+ </span>
|
|
134
|
+ </div>
|
|
135
|
+ <p>
|
|
136
|
+ <strong>This</strong> should be the content.</p>
|
|
137
|
+ </div>
|
|
138
|
+ </li>
|
|
139
|
+ """
|
|
140
|
+ response = mock.Mock(text=html)
|
|
141
|
+ results = yahoo_news.response(response)
|
|
142
|
+ self.assertEqual(type(results), list)
|
|
143
|
+ self.assertEqual(len(results), 0)
|