searxng/searx/engines/vimeo.py

55 lines
1.7 KiB
Python
Raw Normal View History

from urllib import urlencode
from HTMLParser import HTMLParser
from lxml import html
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
base_url = 'http://vimeo.com'
search_url = base_url + '/search?{query}'
2014-01-20 01:31:20 +00:00
url_xpath = None
2014-01-11 10:14:46 +00:00
content_xpath = None
2014-01-20 01:31:20 +00:00
title_xpath = None
2014-01-11 10:14:46 +00:00
results_xpath = ''
2014-01-20 01:31:20 +00:00
content_tpl = '<a href="{0}"> <img src="{2}"/> </a>'
2014-03-18 14:56:22 +00:00
publishedDate_xpath = './/p[@class="meta"]//attribute::datetime'
2014-01-20 01:31:20 +00:00
# the cookie set by vimeo contains all the following values,
# but only __utma seems to be requiered
2014-01-11 10:14:46 +00:00
cookie = {
#'vuid':'918282893.1027205400'
# 'ab_bs':'%7B%223%22%3A279%7D'
2014-01-20 01:31:20 +00:00
'__utma': '00000000.000#0000000.0000000000.0000000000.0000000000.0'
# '__utmb':'18302654.1.10.1388942090'
#, '__utmc':'18302654'
2014-01-20 01:31:20 +00:00
#, '__utmz':'18#302654.1388942090.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)' # noqa
#, '__utml':'search'
}
2014-01-20 01:31:20 +00:00
def request(query, params):
2014-01-20 01:31:20 +00:00
params['url'] = search_url.format(query=urlencode({'q': query}))
2014-01-11 10:14:46 +00:00
params['cookies'] = cookie
return params
2014-01-20 01:31:20 +00:00
def response(resp):
results = []
dom = html.fromstring(resp.text)
2014-01-11 10:14:46 +00:00
p = HTMLParser()
for result in dom.xpath(results_xpath):
url = base_url + result.xpath(url_xpath)[0]
title = p.unescape(extract_text(result.xpath(title_xpath)))
thumbnail = extract_text(result.xpath(content_xpath)[0])
2014-03-18 18:24:01 +00:00
publishedDate = parser.parse(extract_text(
result.xpath(publishedDate_xpath)[0]))
2014-03-18 14:56:22 +00:00
2014-01-20 01:31:20 +00:00
results.append({'url': url,
'title': title,
'content': content_tpl.format(url, title, thumbnail),
'template': 'videos.html',
2014-03-18 14:56:22 +00:00
'publishedDate': publishedDate,
2014-01-20 01:31:20 +00:00
'thumbnail': thumbnail})
return results