Ponysearch/searx/engines/vimeo.py

68 lines
2.1 KiB
Python
Raw Normal View History

# Vimeo (Videos)
#
# @website https://vimeo.com/
# @provide-api yes (http://developer.vimeo.com/api),
# they have a maximum count of queries/hour
#
# @using-api no (TODO, rewrite to api)
# @results HTML (using search portal)
# @stable no (HTML can change)
# @parse url, title, publishedDate, thumbnail, embedded
#
# @todo rewrite to api
# @todo set content-parameter with correct data
2016-12-11 02:33:04 +01:00
from json import loads
2014-03-18 15:56:22 +01:00
from dateutil import parser
2016-11-30 18:43:03 +01:00
from searx.url_utils import urlencode
# engine dependent config
categories = ['videos']
paging = True
# search-url
2016-12-11 02:33:04 +01:00
base_url = 'https://vimeo.com/'
search_url = base_url + '/search/page:{pageno}?{query}'
2016-12-11 02:33:04 +01:00
embedded_url = '<iframe data-src="//player.vimeo.com/video/{videoid}" ' +\
'width="540" height="304" frameborder="0" ' +\
'webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>'
2014-01-20 02:31:20 +01:00
# do search-request
def request(query, params):
params['url'] = search_url.format(pageno=params['pageno'],
query=urlencode({'q': query}))
return params
2014-01-20 02:31:20 +01:00
# get response from search-request
def response(resp):
results = []
2016-12-11 02:33:04 +01:00
data_start_pos = resp.text.find('{"filtered"')
data_end_pos = resp.text.find(';\n', data_start_pos + 1)
data = loads(resp.text[data_start_pos:data_end_pos])
2014-01-11 11:14:46 +01:00
# parse results
2016-12-11 02:33:04 +01:00
for result in data['filtered']['data']:
result = result[result['type']]
videoid = result['uri'].split('/')[-1]
url = base_url + videoid
2016-12-11 02:33:04 +01:00
title = result['name']
thumbnail = result['pictures']['sizes'][-1]['link']
publishedDate = parser.parse(result['created_time'])
embedded = embedded_url.format(videoid=videoid)
2014-03-18 15:56:22 +01:00
# append result
2014-01-20 02:31:20 +01:00
results.append({'url': url,
'title': title,
'content': '',
2014-01-20 02:31:20 +01:00
'template': 'videos.html',
2014-03-18 15:56:22 +01:00
'publishedDate': publishedDate,
'embedded': embedded,
2014-01-20 02:31:20 +01:00
'thumbnail': thumbnail})
# return results
return results