@@ -13,7 +13,6 @@
1313from json import loads
1414from lxml import html
1515from searx.engines.bing_images import _fetch_supported_languages, supported_languages_url
16−from searx.engines.xpath import extract_text
1716from searx.url_utils import urlencode
1817from searx.utils import match_language
1918
@@ -22,11 +21,16 @@ categories = ['videos']
2221paging = True
2322safesearch = True
2423time_range_support = True
25−number_of_results = 10
24+number_of_results = 28
2625language_support = True
2726
28−search_url = 'https://www.bing.com/videos/asyncv2?{query}&async=content&'\
29− 'first={offset}&count={number_of_results}&CW=1366&CH=25&FORM=R5VR5'
27+base_url = 'https://www.bing.com/'
28+search_string = 'videos/search'\
29+ '?{query}'\
30+ '&count={count}'\
31+ '&first={first}'\
32+ '&scope=video'\
33+ '&FORM=QBLH'
3034time_range_string = '&qft=+filterui:videoage-lt{interval}'
3135time_range_dict = {'day': '1440',
3236 'week': '10080',
@@ -41,7 +45,12 @@ safesearch_types = {2: 'STRICT',
4145
4246# do search-request
4347def 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)
4554
4655 # safesearch cookie
4756 params['cookies']['SRCHHPGUSR'] = \
@@ -52,9 +61,7 @@ def request(query, params):
5261 params['cookies']['_EDGE_S'] = 'mkt=' + language + '&F=1'
5362
5463 # query and paging
55− params['url'] = search_url.format(query=urlencode({'q': query}),
56− offset=offset,
57− number_of_results=number_of_results)
64+ params['url'] = base_url + search_path
5865
5966 # time range
6067 if params['time_range'] in time_range_dict:
@@ -70,19 +77,18 @@ def response(resp):
7077 dom = html.fromstring(resp.text)
7178
7279 for result in dom.xpath('//div[@class="dg_u"]'):
73− url = result.xpath('./div[@class="mc_vtvc"]/a/@href')[0]
74− url = 'https://bing.com' + url
75− title = extract_text(result.xpath('./div/a/div/div[@class="mc_vtvc_title"]/@title'))
76− content = extract_text(result.xpath('./div/a/div/div/div/div/text()'))
77− thumbnail = result.xpath('./div/a/div/div/img/@src')[0]
78−
79− results.append({'url': url,
80− 'title': title,
81− 'content': content,
82− 'thumbnail': thumbnail,
83− 'template': 'videos.html'})
84−
85− if len(results) >= number_of_results:
86− break
80+ try:
81+ metadata = loads(result.xpath('.//div[@class="vrhdata"]/@vrhm')[0])
82+ info = ' - '.join(result.xpath('.//div[@class="mc_vtvc_meta_block"]//span/text()')).strip()
83+ content = '{0} - {1}'.format(metadata['du'], info)
84+ thumbnail = '{0}th?id={1}'.format(base_url, metadata['thid'])
85+ results.append({'url': metadata['murl'],
86+ 'thumbnail': thumbnail,
87+ 'title': metadata.get('vt', ''),
88+ 'content': content,
89+ 'template': 'videos.html'})
90+
91+ except:
92+ continue
8793
8894 return results