@@ -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
@@ -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
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'\
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',
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
50+ search_path = search_string.format(
51+ query=urlencode({'q': query}),
52+ count=number_of_results,
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'
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
57− number_of_results=number_of_results)
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)
72 for result in dom.xpath('//div[@class="dg_u"]'):
79 for result in dom.xpath('//div[@class="dg_u"]'):
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]
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'})
85− if len(results) >= number_of_results:
91+ except:
88 return results
94 return results