소스 검색

Merge pull request #370 from Cqoicebordel/fix_vimeo

Fix vimeo
Adam Tauber 9 년 전
부모
커밋
c9b8c7f8de

+ 6
- 6
searx/engines/vimeo.py 파일 보기

@@ -27,11 +27,11 @@ base_url = 'https://vimeo.com'
27 27
 search_url = base_url + '/search/page:{pageno}?{query}'
28 28
 
29 29
 # specific xpath variables
30
-results_xpath = '//div[@id="browse_content"]/ol/li'
31
-url_xpath = './a/@href'
32
-title_xpath = './a/div[@class="data"]/p[@class="title"]'
33
-content_xpath = './a/img/@src'
34
-publishedDate_xpath = './/p[@class="meta"]//attribute::datetime'
30
+results_xpath = '//div[contains(@class,"results_grid")]/ul/li'
31
+url_xpath = './/a/@href'
32
+title_xpath = './/span[@class="title"]'
33
+thumbnail_xpath = './/img[@class="js-clip_thumbnail_image"]/@src'
34
+publishedDate_xpath = './/time/attribute::datetime'
35 35
 
36 36
 embedded_url = '<iframe data-src="//player.vimeo.com/video{videoid}" ' +\
37 37
     'width="540" height="304" frameborder="0" ' +\
@@ -58,7 +58,7 @@ def response(resp):
58 58
         videoid = result.xpath(url_xpath)[0]
59 59
         url = base_url + videoid
60 60
         title = p.unescape(extract_text(result.xpath(title_xpath)))
61
-        thumbnail = extract_text(result.xpath(content_xpath)[0])
61
+        thumbnail = extract_text(result.xpath(thumbnail_xpath)[0])
62 62
         publishedDate = parser.parse(extract_text(result.xpath(publishedDate_xpath)[0]))
63 63
         embedded = embedded_url.format(videoid=videoid)
64 64
 

+ 4
- 0
searx/tests/engines/test_swisscows.py 파일 보기

@@ -23,6 +23,10 @@ class TestSwisscowsEngine(SearxTestCase):
23 23
         self.assertTrue('uiLanguage=browser' in params['url'])
24 24
         self.assertTrue('region=browser' in params['url'])
25 25
 
26
+        dicto['category'] = 'images'
27
+        params = swisscows.request(query, dicto)
28
+        self.assertIn('image', params['url'])
29
+
26 30
     def test_response(self):
27 31
         self.assertRaises(AttributeError, swisscows.response, None)
28 32
         self.assertRaises(AttributeError, swisscows.response, [])

+ 35
- 18
searx/tests/engines/test_vimeo.py 파일 보기

@@ -1,3 +1,4 @@
1
+# -*- coding: utf-8 -*-
1 2
 from collections import defaultdict
2 3
 import mock
3 4
 from searx.engines import vimeo
@@ -25,26 +26,42 @@ class TestVimeoEngine(SearxTestCase):
25 26
         self.assertEqual(vimeo.response(response), [])
26 27
 
27 28
         html = """
28
-        <div id="browse_content" class="" data-search-id="696d5f8366914ec4ffec33cf7652de384976d4f4">
29
-            <ol class="js-browse_list clearfix browse browse_videos browse_videos_thumbnails kane"
29
+        <div id="browse_content" class="results_grid" data-search-id="696d5f8366914ec4ffec33cf7652de384976d4f4">
30
+            <ul class="js-browse_list clearfix browse browse_videos browse_videos_thumbnails kane"
30 31
                 data-stream="c2VhcmNoOjo6ZGVzYzp7InF1ZXJ5IjoidGVzdCJ9">
31
-                <li id="clip_100785455" data-start-page="/search/page:1/sort:relevant/" data-position="1">
32
-                    <a href="/videoid" title="Futurama 3d (test shot)">
33
-                        <img src="http://image.url.webp"
34
-                            srcset="http://i.vimeocdn.com/video/482375085_590x332.webp 2x" alt=""
35
-                            class="thumbnail thumbnail_lg_wide">
36
-                        <div class="data">
37
-                            <p class="title">
38
-                                This is the title
39
-                            </p>
40
-                            <p class="meta">
41
-                                <time datetime="2014-07-15T04:16:27-04:00"
42
-                                    title="mardi 15 juillet 2014 04:16">Il y a 6 mois</time>
43
-                            </p>
44
-                        </div>
45
-                    </a>
32
+                <li data-position="7" data-result-id="clip_79600943">
33
+                    <div class="clip_thumbnail">
34
+                        <a href="/videoid" class="js-result_url">
35
+                            <div class="thumbnail_wrapper">
36
+                                <img src="http://image.url.webp" class="js-clip_thumbnail_image">
37
+                                <div class="overlay overlay_clip_meta">
38
+                                    <div class="meta_data_footer">
39
+                                        <span class="clip_upload_date">
40
+                                            <time datetime="2013-11-17T08:49:09-05:00"
41
+                                                title="dimanche 17 novembre 2013 08:49">Il y a 1 an</time>
42
+                                        </span>
43
+                                        <span class="clip_likes">
44
+                                            <img src="https://f.vimeocdn.com/images_v6/svg/heart-icon.svg">2 215
45
+                                        </span>
46
+                                        <span class="clip_comments">
47
+                                            <img src="https://f.vimeocdn.com/images_v6/svg/comment-icon.svg">75
48
+                                        </span>
49
+                                        <span class="overlay meta_data_footer clip_duration">01:12</span>
50
+                                    </div>
51
+                                </div>
52
+                            </div>
53
+                            <span class="title">This is the title</span>
54
+                        </a>
55
+                    </div>
56
+                    <div class="clip_thumbnail_attribution">
57
+                        <a href="/fedorshmidt">
58
+                            <img src="https://i.vimeocdn.com/portrait/6628061_100x100.jpg" class="avatar">
59
+                            <span class="display_name">Fedor Shmidt</span>
60
+                        </a>
61
+                        <span class="plays">2,1M lectures</span>
62
+                    </div>
46 63
                 </li>
47
-            </ol>
64
+            </ul>
48 65
         </div>
49 66
         """
50 67
         response = mock.Mock(text=html)

+ 18
- 1
searx/tests/engines/test_yahoo_news.py 파일 보기

@@ -29,6 +29,13 @@ class TestYahooNewsEngine(SearxTestCase):
29 29
         self.assertIn('en', params['cookies']['sB'])
30 30
         self.assertIn('en', params['url'])
31 31
 
32
+    def test_sanitize_url(self):
33
+        url = "test.url"
34
+        self.assertEqual(url, yahoo_news.sanitize_url(url))
35
+
36
+        url = "www.yahoo.com/;_ylt=test"
37
+        self.assertEqual("www.yahoo.com/", yahoo_news.sanitize_url(url))
38
+
32 39
     def test_response(self):
33 40
         self.assertRaises(AttributeError, yahoo_news.response, None)
34 41
         self.assertRaises(AttributeError, yahoo_news.response, [])
@@ -57,7 +64,17 @@ class TestYahooNewsEngine(SearxTestCase):
57 64
                    This is the content
58 65
                </div>
59 66
             </li>
60
-        </div>
67
+            <li class="first">
68
+                <div class="compTitle">
69
+                    <h3>
70
+                        <a class="yschttl spt" target="_blank">
71
+                        </a>
72
+                    </h3>
73
+                </div>
74
+                <div class="compText">
75
+               </div>
76
+            </li>
77
+        </ol>
61 78
         """
62 79
         response = mock.Mock(text=html)
63 80
         results = yahoo_news.response(response)

+ 51
- 0
searx/tests/engines/test_youtube_noapi.py 파일 보기

@@ -94,6 +94,57 @@ class TestYoutubeNoAPIEngine(SearxTestCase):
94 94
         html = """
95 95
         <ol id="item-section-063864" class="item-section">
96 96
             <li>
97
+                <div class="yt-lockup yt-lockup-tile yt-lockup-video vve-check clearfix yt-uix-tile"
98
+                data-context-item-id="DIVZCPfAOeM"
99
+                data-visibility-tracking="CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JECx_-GK5uqMpcIB">
100
+                <div class="yt-lockup-dismissable"><div class="yt-lockup-thumbnail contains-addto">
101
+                <a aria-hidden="true" href="/watch?v=DIVZCPfAOeM" class=" yt-uix-sessionlink pf-link"
102
+                data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JFIEdGVzdA">
103
+                <div class="yt-thumb video-thumb"><img src="//i.ytimg.com/vi/DIVZCPfAOeM/mqdefault.jpg"
104
+                width="196" height="110"/></div><span class="video-time" aria-hidden="true">11:35</span></a>
105
+                <span class="thumb-menu dark-overflow-action-menu video-actions">
106
+                </span>
107
+                </div>
108
+                <div class="yt-lockup-content">
109
+                <h3 class="yt-lockup-title">
110
+                <span class="accessible-description" id="description-id-259079"> - Durée : 11:35.</span>
111
+                </h3>
112
+                <div class="yt-lockup-byline">de
113
+                <a href="/user/mheejapan" class=" yt-uix-sessionlink spf-link g-hovercard"
114
+                data-sessionlink="itct=CBgQ3DAYACITCPGXnYau6sUCFZEIHAod-VQASCj0JA" data-ytid="UCzEesu54Hjs0uRKmpy66qeA"
115
+                data-name="">MEHAY SUPERBIKE</a></div><div class="yt-lockup-meta">
116
+                <ul class="yt-lockup-meta-info">
117
+                    <li>il y a 20 heures</li>
118
+                    <li>8 424 vues</li>
119
+                </ul>
120
+                </div>
121
+                <div class="yt-lockup-badges">
122
+                <ul class="yt-badge-list ">
123
+                    <li class="yt-badge-item" >
124
+                        <span class="yt-badge">Nouveauté</span>
125
+                    </li>
126
+                    <li class="yt-badge-item" ><span class="yt-badge " >HD</span></li>
127
+                </ul>
128
+                </div>
129
+                <div class="yt-lockup-action-menu yt-uix-menu-container">
130
+                <div class="yt-uix-menu yt-uix-videoactionmenu hide-until-delayloaded"
131
+                data-video-id="DIVZCPfAOeM" data-menu-content-id="yt-uix-videoactionmenu-menu">
132
+                </div>
133
+                </div>
134
+                </div>
135
+                </div>
136
+                </div>
137
+            </li>
138
+        </ol>
139
+        """
140
+        response = mock.Mock(text=html)
141
+        results = youtube_noapi.response(response)
142
+        self.assertEqual(type(results), list)
143
+        self.assertEqual(len(results), 1)
144
+
145
+        html = """
146
+        <ol id="item-section-063864" class="item-section">
147
+            <li>
97 148
             </li>
98 149
         </ol>
99 150
         """