Ponysearch/searx/engines/startpage.py

32 lines
1 KiB
Python
Raw Normal View History

2013-11-09 18:39:20 +01:00
from urllib import urlencode
2013-10-19 18:29:39 +02:00
from lxml import html
2013-10-24 23:43:39 +02:00
from urlparse import urlparse
2013-10-19 18:29:39 +02:00
from cgi import escape
2013-10-19 18:30:08 +02:00
base_url = 'https://startpage.com/'
2013-10-19 18:29:39 +02:00
search_url = base_url+'do/search'
def request(query, params):
global search_url
2013-11-09 18:39:20 +01:00
query = urlencode({'q': query})[2:]
2013-10-19 18:29:39 +02:00
params['url'] = search_url
params['method'] = 'POST'
params['data'] = {'query': query}
return params
def response(resp):
global base_url
results = []
2013-10-24 21:00:44 +02:00
dom = html.fromstring(resp.content)
# ads xpath //div[@id="results"]/div[@id="sponsored"]//div[@class="result"]
# not ads : div[@class="result"] are the direct childs of div[@id="results"]
for result in dom.xpath('//div[@id="results"]/div[@class="result"]'):
2013-10-19 18:29:39 +02:00
link = result.xpath('.//h3/a')[0]
2013-10-24 23:43:39 +02:00
url = link.attrib.get('href')
2013-10-19 20:12:22 +02:00
parsed_url = urlparse(url)
title = link.text_content()
content = result.xpath('./p[@class="desc"]')[0].text_content()
2013-10-19 18:29:39 +02:00
results.append({'url': url, 'title': title, 'content': content})
return results