mirror of
https://github.com/searxng/searxng.git
synced 2024-12-21 00:36:29 +00:00
[enh] startpage engine added
This commit is contained in:
parent
e85972b5e1
commit
57eaeb9b74
1 changed files with 28 additions and 0 deletions
28
searx/engines/startpage.py
Normal file
28
searx/engines/startpage.py
Normal file
|
@ -0,0 +1,28 @@
|
|||
from urllib import quote
|
||||
from lxml import html
|
||||
from urlparse import urljoin
|
||||
from cgi import escape
|
||||
|
||||
base_url = 'http://startpage.com/'
|
||||
search_url = base_url+'do/search'
|
||||
|
||||
def request(query, params):
|
||||
global search_url
|
||||
query = quote(query.replace(' ', '+'), safe='+')
|
||||
params['url'] = search_url
|
||||
params['method'] = 'POST'
|
||||
params['data'] = {'query': query}
|
||||
return params
|
||||
|
||||
|
||||
def response(resp):
|
||||
global base_url
|
||||
results = []
|
||||
dom = html.fromstring(resp.text)
|
||||
for result in dom.xpath('//div[@class="result"]'):
|
||||
link = result.xpath('.//h3/a')[0]
|
||||
url = urljoin(base_url, link.attrib.get('href'))
|
||||
title = ' '.join(link.xpath('.//text()'))
|
||||
content = escape(' '.join(result.xpath('.//p[@class="desc"]//text()')))
|
||||
results.append({'url': url, 'title': title, 'content': content})
|
||||
return results
|
Loading…
Reference in a new issue