2021-01-13 10:31:25 +00:00
|
|
|
# SPDX-License-Identifier: AGPL-3.0-or-later
|
2017-08-05 19:48:07 +00:00
|
|
|
"""
|
|
|
|
Bing (Videos)
|
|
|
|
"""
|
|
|
|
|
|
|
|
from json import loads
|
|
|
|
from lxml import html
|
2020-08-06 15:42:46 +00:00
|
|
|
from urllib.parse import urlencode
|
2018-03-01 04:30:48 +00:00
|
|
|
from searx.utils import match_language
|
2017-08-05 19:48:07 +00:00
|
|
|
|
2020-11-02 10:19:53 +00:00
|
|
|
from searx.engines.bing import language_aliases
|
2021-12-27 08:26:22 +00:00
|
|
|
from searx.engines.bing import (
|
|
|
|
_fetch_supported_languages,
|
|
|
|
supported_languages_url,
|
|
|
|
) # NOQA # pylint: disable=unused-import
|
2017-08-05 19:48:07 +00:00
|
|
|
|
2021-01-13 10:31:25 +00:00
|
|
|
# about
|
|
|
|
about = {
|
|
|
|
"website": 'https://www.bing.com/videos',
|
|
|
|
"wikidata_id": 'Q4914152',
|
|
|
|
"official_api_documentation": 'https://www.microsoft.com/en-us/bing/apis/bing-video-search-api',
|
|
|
|
"use_official_api": False,
|
|
|
|
"require_api_key": False,
|
|
|
|
"results": 'HTML',
|
|
|
|
}
|
|
|
|
|
2017-08-05 19:48:07 +00:00
|
|
|
categories = ['videos']
|
|
|
|
paging = True
|
|
|
|
safesearch = True
|
|
|
|
time_range_support = True
|
2019-07-27 15:49:30 +00:00
|
|
|
number_of_results = 28
|
2017-08-05 19:48:07 +00:00
|
|
|
|
2019-07-27 15:49:30 +00:00
|
|
|
base_url = 'https://www.bing.com/'
|
2021-12-27 08:26:22 +00:00
|
|
|
search_string = 'videos/search' '?{query}' '&count={count}' '&first={first}' '&scope=video' '&FORM=QBLH'
|
2017-08-05 19:48:07 +00:00
|
|
|
time_range_string = '&qft=+filterui:videoage-lt{interval}'
|
2021-12-27 08:26:22 +00:00
|
|
|
time_range_dict = {'day': '1440', 'week': '10080', 'month': '43200', 'year': '525600'}
|
2017-08-05 19:48:07 +00:00
|
|
|
|
|
|
|
# safesearch definitions
|
2021-12-27 08:26:22 +00:00
|
|
|
safesearch_types = {2: 'STRICT', 1: 'DEMOTE', 0: 'OFF'}
|
2017-08-05 19:48:07 +00:00
|
|
|
|
|
|
|
|
|
|
|
# do search-request
|
|
|
|
def request(query, params):
|
2019-07-27 15:49:30 +00:00
|
|
|
offset = ((params['pageno'] - 1) * number_of_results) + 1
|
|
|
|
|
2021-12-27 08:26:22 +00:00
|
|
|
search_path = search_string.format(query=urlencode({'q': query}), count=number_of_results, first=offset)
|
2017-08-05 19:48:07 +00:00
|
|
|
|
|
|
|
# safesearch cookie
|
2021-12-27 08:26:22 +00:00
|
|
|
params['cookies']['SRCHHPGUSR'] = 'ADLT=' + safesearch_types.get(params['safesearch'], 'DEMOTE')
|
2017-08-05 19:48:07 +00:00
|
|
|
|
|
|
|
# language cookie
|
2018-11-26 05:32:48 +00:00
|
|
|
language = match_language(params['language'], supported_languages, language_aliases).lower()
|
2018-03-01 04:30:48 +00:00
|
|
|
params['cookies']['_EDGE_S'] = 'mkt=' + language + '&F=1'
|
2017-08-05 19:48:07 +00:00
|
|
|
|
|
|
|
# query and paging
|
2019-07-27 15:49:30 +00:00
|
|
|
params['url'] = base_url + search_path
|
2017-08-05 19:48:07 +00:00
|
|
|
|
|
|
|
# time range
|
|
|
|
if params['time_range'] in time_range_dict:
|
|
|
|
params['url'] += time_range_string.format(interval=time_range_dict[params['time_range']])
|
|
|
|
|
2020-03-01 16:28:59 +00:00
|
|
|
# bing videos did not like "older" versions < 70.0.1 when selectin other
|
|
|
|
# languages then 'en' .. very strange ?!?!
|
|
|
|
params['headers']['User-Agent'] = 'Mozilla/5.0 (X11; Linux x86_64; rv:73.0.1) Gecko/20100101 Firefox/73.0.1'
|
|
|
|
|
2017-08-05 19:48:07 +00:00
|
|
|
return params
|
|
|
|
|
|
|
|
|
|
|
|
# get response from search-request
|
|
|
|
def response(resp):
|
|
|
|
results = []
|
|
|
|
|
|
|
|
dom = html.fromstring(resp.text)
|
|
|
|
|
|
|
|
for result in dom.xpath('//div[@class="dg_u"]'):
|
2019-07-27 15:49:30 +00:00
|
|
|
try:
|
|
|
|
metadata = loads(result.xpath('.//div[@class="vrhdata"]/@vrhm')[0])
|
|
|
|
info = ' - '.join(result.xpath('.//div[@class="mc_vtvc_meta_block"]//span/text()')).strip()
|
|
|
|
content = '{0} - {1}'.format(metadata['du'], info)
|
|
|
|
thumbnail = '{0}th?id={1}'.format(base_url, metadata['thid'])
|
2021-12-27 08:26:22 +00:00
|
|
|
results.append(
|
|
|
|
{
|
|
|
|
'url': metadata['murl'],
|
|
|
|
'thumbnail': thumbnail,
|
|
|
|
'title': metadata.get('vt', ''),
|
|
|
|
'content': content,
|
|
|
|
'template': 'videos.html',
|
|
|
|
}
|
|
|
|
)
|
2019-07-27 15:49:30 +00:00
|
|
|
|
|
|
|
except:
|
|
|
|
continue
|
2017-08-05 19:48:07 +00:00
|
|
|
|
|
|
|
return results
|