Ponysearch/searx/engines/duckduckgo.py

70 lines
1.7 KiB
Python
Raw Normal View History

## DuckDuckGo (Web)
#
# @website https://duckduckgo.com/
# @provide-api yes (https://duckduckgo.com/api), but not all results from search-site
#
# @using-api no
# @results HTML (using search portal)
# @stable no (HTML can change)
# @parse url, title, content
#
# @todo rewrite to api
2014-09-02 20:21:33 +02:00
# @todo language support (the current used site does not support language-change)
2013-10-23 23:55:37 +02:00
from urllib import urlencode
2014-03-21 16:33:17 +01:00
from lxml.html import fromstring
2013-11-09 18:39:20 +01:00
from searx.utils import html_to_text
2013-10-14 23:09:13 +02:00
# engine dependent config
categories = ['general']
paging = True
locale = 'us-en'
2013-10-14 23:09:13 +02:00
# search-url
url = 'https://duckduckgo.com/html?{query}&s={offset}'
# specific xpath variables
result_xpath = '//div[@class="results_links results_links_deep web-result"]' # noqa
url_xpath = './/a[@class="large"]/@href'
title_xpath = './/a[@class="large"]//text()'
content_xpath = './/div[@class="snippet"]//text()'
2014-03-29 16:38:45 +01:00
# do search-request
2013-10-14 23:09:13 +02:00
def request(query, params):
2014-01-30 01:03:19 +01:00
offset = (params['pageno'] - 1) * 30
params['url'] = url.format(
query=urlencode({'q': query, 'l': locale}),
offset=offset)
2013-10-14 23:09:13 +02:00
return params
# get response from search-request
2013-10-14 23:09:13 +02:00
def response(resp):
2013-10-15 19:11:43 +02:00
results = []
2014-03-21 16:33:17 +01:00
doc = fromstring(resp.text)
2014-09-02 18:12:42 +02:00
# parse results
2014-03-21 16:33:17 +01:00
for r in doc.xpath(result_xpath):
2014-03-21 18:17:13 +01:00
try:
res_url = r.xpath(url_xpath)[-1]
except:
continue
2014-03-21 16:33:17 +01:00
if not res_url:
2013-10-15 19:11:43 +02:00
continue
2014-03-21 16:33:17 +01:00
title = html_to_text(''.join(r.xpath(title_xpath)))
content = html_to_text(''.join(r.xpath(content_xpath)))
# append result
2014-03-21 16:33:17 +01:00
results.append({'title': title,
'content': content,
'url': res_url})
# return results
2013-10-15 19:11:43 +02:00
return results