searxng/searx/engines/vimeo.py

76 lines
2.3 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
from urllib import urlencode
from lxml import html
from HTMLParser import HTMLParser
2014-03-24 11:04:07 +00:00
from searx.engines.xpath import extract_text
2014-03-18 14:56:22 +00:00
from dateutil import parser
# engine dependent config
categories = ['videos']
paging = True
# search-url
base_url = 'http://vimeo.com'
search_url = base_url + '/search/page:{pageno}?{query}'
# specific xpath variables
results_xpath = '//div[@id="browse_content"]/ol/li'
url_xpath = './a/@href'
title_xpath = './a/div[@class="data"]/p[@class="title"]'
content_xpath = './a/img/@src'
publishedDate_xpath = './/p[@class="meta"]//attribute::datetime'
embedded_url = '<iframe data-src="//player.vimeo.com/video{videoid}" ' +\
'width="540" height="304" frameborder="0" ' +\
'webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>'
2014-01-20 01:31:20 +00:00
# do search-request
def request(query, params):
params['url'] = search_url.format(pageno=params['pageno'],
query=urlencode({'q': query}))
return params
2014-01-20 01:31:20 +00:00
# get response from search-request
def response(resp):
results = []
dom = html.fromstring(resp.text)
p = HTMLParser()
2014-01-11 10:14:46 +00:00
# parse results
for result in dom.xpath(results_xpath):
videoid = result.xpath(url_xpath)[0]
url = base_url + videoid
title = p.unescape(extract_text(result.xpath(title_xpath)))
thumbnail = extract_text(result.xpath(content_xpath)[0])
2015-01-31 18:49:54 +00:00
publishedDate = parser.parse(extract_text(result.xpath(publishedDate_xpath)[0]))
embedded = embedded_url.format(videoid=videoid)
2014-03-18 14:56:22 +00:00
# append result
2014-01-20 01:31:20 +00:00
results.append({'url': url,
'title': title,
'content': '',
2014-01-20 01:31:20 +00:00
'template': 'videos.html',
2014-03-18 14:56:22 +00:00
'publishedDate': publishedDate,
'embedded': embedded,
2014-01-20 01:31:20 +00:00
'thumbnail': thumbnail})
# return results
return results