on master help sign in
[fix] bing_videos (#1579)
204a2cbFrank de Lange
searx/engines/bing_videos.py+28−22
@@ -13,7 +13,6 @@
13from json import loads
13from json import loads
14from lxml import html
14from lxml import html
15from searx.engines.bing_images import _fetch_supported_languages, supported_languages_url
15from searx.engines.bing_images import _fetch_supported_languages, supported_languages_url
16−from searx.engines.xpath import extract_text
17from searx.url_utils import urlencode
16from searx.url_utils import urlencode
18from searx.utils import match_language
17from searx.utils import match_language
19
18
@@ -22,11 +21,16 @@ categories = ['videos']
22paging = True
21paging = True
23safesearch = True
22safesearch = True
24time_range_support = True
23time_range_support = True
25−number_of_results = 10
24+number_of_results = 28
26language_support = True
25language_support = True
27
26
28−search_url = 'https://www.bing.com/videos/asyncv2?{query}&async=content&'\
27+base_url = 'https://www.bing.com/'
29− 'first={offset}&count={number_of_results}&CW=1366&CH=25&FORM=R5VR5'
28+search_string = 'videos/search'\
29+ '?{query}'\
30+ '&count={count}'\
31+ '&first={first}'\
32+ '&scope=video'\
33+ '&FORM=QBLH'
30time_range_string = '&qft=+filterui:videoage-lt{interval}'
34time_range_string = '&qft=+filterui:videoage-lt{interval}'
31time_range_dict = {'day': '1440',
35time_range_dict = {'day': '1440',
32 'week': '10080',
36 'week': '10080',
@@ -41,7 +45,12 @@ safesearch_types = {2: 'STRICT',
41
45
42# do search-request
46# do search-request
43def request(query, params):
47def request(query, params):
44− offset = (params['pageno'] - 1) * 10 + 1
48+ offset = ((params['pageno'] - 1) * number_of_results) + 1
49+
50+ search_path = search_string.format(
51+ query=urlencode({'q': query}),
52+ count=number_of_results,
53+ first=offset)
45
54
46 # safesearch cookie
55 # safesearch cookie
47 params['cookies']['SRCHHPGUSR'] = \
56 params['cookies']['SRCHHPGUSR'] = \
@@ -52,9 +61,7 @@ def request(query, params):
52 params['cookies']['_EDGE_S'] = 'mkt=' + language + '&F=1'
61 params['cookies']['_EDGE_S'] = 'mkt=' + language + '&F=1'
53
62
54 # query and paging
63 # query and paging
55− params['url'] = search_url.format(query=urlencode({'q': query}),
64+ params['url'] = base_url + search_path
56− offset=offset,
57− number_of_results=number_of_results)
58
65
59 # time range
66 # time range
60 if params['time_range'] in time_range_dict:
67 if params['time_range'] in time_range_dict:
@@ -70,19 +77,18 @@ def response(resp):
70 dom = html.fromstring(resp.text)
77 dom = html.fromstring(resp.text)
71
78
72 for result in dom.xpath('//div[@class="dg_u"]'):
79 for result in dom.xpath('//div[@class="dg_u"]'):
80+ try:
73− url = result.xpath('./div[@class="mc_vtvc"]/a/@href')[0]
81+ metadata = loads(result.xpath('.//div[@class="vrhdata"]/@vrhm')[0])
74− url = 'https://bing.com' + url
82+ info = ' - '.join(result.xpath('.//div[@class="mc_vtvc_meta_block"]//span/text()')).strip()
75− title = extract_text(result.xpath('./div/a/div/div[@class="mc_vtvc_title"]/@title'))
83+ content = '{0} - {1}'.format(metadata['du'], info)
76− content = extract_text(result.xpath('./div/a/div/div/div/div/text()'))
84+ thumbnail = '{0}th?id={1}'.format(base_url, metadata['thid'])
77− thumbnail = result.xpath('./div/a/div/div/img/@src')[0]
78−
79− results.append({'url': url,
85+ results.append({'url': metadata['murl'],
80− 'title': title,
86+ 'thumbnail': thumbnail,
81− 'content': content,
87+ 'title': metadata.get('vt', ''),
82− 'thumbnail': thumbnail,
88+ 'content': content,
83− 'template': 'videos.html'})
89+ 'template': 'videos.html'})
84−
90+
85− if len(results) >= number_of_results:
91+ except:
86− break
92+ continue
87
93
88 return results
94 return results
tests/unit/engines/test_bing_videos.py+13−41
@@ -29,7 +29,7 @@ class TestBingVideosEngine(SearxTestCase):
29 dicto['time_range'] = 'day'
29 dicto['time_range'] = 'day'
30 dicto['safesearch'] = 2
30 dicto['safesearch'] = 2
31 params = bing_videos.request(query, dicto)
31 params = bing_videos.request(query, dicto)
32− self.assertTrue('first=11' in params['url'])
32+ self.assertTrue('first=29' in params['url'])
33 self.assertTrue('1440' in params['url'])
33 self.assertTrue('1440' in params['url'])
34 self.assertIn('SRCHHPGUSR', params['cookies'])
34 self.assertIn('SRCHHPGUSR', params['cookies'])
35 self.assertTrue('STRICT' in params['cookies']['SRCHHPGUSR'])
35 self.assertTrue('STRICT' in params['cookies']['SRCHHPGUSR'])
@@ -48,53 +48,25 @@ class TestBingVideosEngine(SearxTestCase):
48
48
49 html = """
49 html = """
50 <div class="dg_u">
50 <div class="dg_u">
51− <div id="mc_vtvc_1" class="mc_vtvc">
51+ <div>
52− <a class="mc_vtvc_link" href="/video">
52+ <a>
53− <div class="mc_vtvc_th">
53+ <div>
54− <div class="cico">
54+ <div>
55− <img src="thumb_1.jpg" />
55+ <div class="mc_vtvc_meta_block">
56− </div>
56+ <div><span>100 views</span><span>1 year ago</span></div><div><span>ExampleTube</span><span>Channel 1<span></div> #noqa
57− <div class="mc_vtvc_ban_lo">
58− <div class="vtbc">
59− <div class="mc_bc_w b_smText">
60− <div class="mc_bc pivot bpi_2">
61− <span title="">
62− <span class="mv_vtvc_play cipg "></span>
63− </span>
64− </div>
65− <div class="mc_bc items">10:06</div>
66− </div>
67 </div>
57 </div>
68 </div>
58 </div>
69− </div>
59+ <div class="vrhdata" vrhm='{"du":"01:11","murl":"https://www.example.com/watch?v=DEADBEEF","thid":"OVP.BINGTHUMB1","vt":"Title 1"}'></div> # noqa
70− <div class="mc_vtvc_meta">
71− <div class="mc_vtvc_title" title="Title 1"></div>
72− <div class="mc_vtvc_meta_block_area">
73− <div class="mc_vtvc_meta_block">
74− <div class="mc_vtvc_meta_row">
75− <span>65,696,000+ views</span>
76− <span>1 year ago</span>
77− </div>
78− <div class="mc_vtvc_meta_row mc_vtvc_meta_row_channel">Content 1</div>
79− <div class="mc_vtvc_meta_row"><span>
80− <div class="cico mc_vtvc_src_ico">
81− <div></div>
82− </div>
83− <span>YouTube</span>
84− </span></div>
85− </div>
86− </div>
87 </div>
60 </div>
88− <div class="vrhdata"></div>
89− </a>
61+ </a>
90− </div>
91 </div>
62 </div>
63+ </div>
92 """
64 """
93 response = mock.Mock(text=html)
65 response = mock.Mock(text=html)
94 results = bing_videos.response(response)
66 results = bing_videos.response(response)
95 self.assertEqual(type(results), list)
67 self.assertEqual(type(results), list)
96 self.assertEqual(len(results), 1)
68 self.assertEqual(len(results), 1)
97 self.assertEqual(results[0]['title'], 'Title 1')
69 self.assertEqual(results[0]['title'], 'Title 1')
98− self.assertEqual(results[0]['url'], 'https://bing.com/video')
70+ self.assertEqual(results[0]['url'], 'https://www.example.com/watch?v=DEADBEEF')
99− self.assertEqual(results[0]['content'], 'Content 1')
71+ self.assertEqual(results[0]['content'], '01:11 - 100 views - 1 year ago - ExampleTube - Channel 1')
100− self.assertEqual(results[0]['thumbnail'], 'thumb_1.jpg')
72+ self.assertEqual(results[0]['thumbnail'], 'https://www.bing.com/th?id=OVP.BINGTHUMB1')