2014-09-01 15:10:05 +02:00
|
|
|
## Google (Web)
|
|
|
|
#
|
|
|
|
# @website https://www.google.com
|
|
|
|
# @provide-api yes (https://developers.google.com/web-search/docs/), deprecated!
|
|
|
|
#
|
|
|
|
# @using-api yes
|
|
|
|
# @results JSON
|
|
|
|
# @stable yes (but deprecated)
|
|
|
|
# @parse url, title, content
|
2014-01-29 19:28:38 +01:00
|
|
|
|
|
|
|
from urllib import urlencode
|
|
|
|
from json import loads
|
|
|
|
|
2014-09-01 15:10:05 +02:00
|
|
|
# engine dependent config
|
2014-01-29 19:28:38 +01:00
|
|
|
categories = ['general']
|
2014-09-01 15:10:05 +02:00
|
|
|
paging = True
|
|
|
|
language_support = True
|
2014-01-29 19:28:38 +01:00
|
|
|
|
2014-09-01 15:10:05 +02:00
|
|
|
# search-url
|
2014-01-29 19:28:38 +01:00
|
|
|
url = 'https://ajax.googleapis.com/'
|
2014-01-31 04:35:23 +01:00
|
|
|
search_url = url + 'ajax/services/search/web?v=2.0&start={offset}&rsz=large&safe=off&filter=off&{query}&hl={language}' # noqa
|
|
|
|
|
2014-01-29 19:28:38 +01:00
|
|
|
|
2014-09-01 15:10:05 +02:00
|
|
|
# do search-request
|
2014-01-29 19:28:38 +01:00
|
|
|
def request(query, params):
|
2014-01-29 21:14:38 +01:00
|
|
|
offset = (params['pageno'] - 1) * 8
|
2014-09-01 15:10:05 +02:00
|
|
|
|
2014-01-31 04:35:23 +01:00
|
|
|
language = 'en-US'
|
|
|
|
if params['language'] != 'all':
|
|
|
|
language = params['language'].replace('_', '-')
|
2014-09-01 15:10:05 +02:00
|
|
|
|
2014-01-29 21:14:38 +01:00
|
|
|
params['url'] = search_url.format(offset=offset,
|
2014-01-31 04:35:23 +01:00
|
|
|
query=urlencode({'q': query}),
|
|
|
|
language=language)
|
2014-09-01 15:10:05 +02:00
|
|
|
|
2014-01-29 19:28:38 +01:00
|
|
|
return params
|
|
|
|
|
|
|
|
|
2014-09-01 15:10:05 +02:00
|
|
|
# get response from search-request
|
2014-01-29 19:28:38 +01:00
|
|
|
def response(resp):
|
|
|
|
results = []
|
2014-09-01 15:10:05 +02:00
|
|
|
|
2014-01-29 19:28:38 +01:00
|
|
|
search_res = loads(resp.text)
|
|
|
|
|
2014-09-01 15:10:05 +02:00
|
|
|
# return empty array if there are no results
|
2014-01-29 19:28:38 +01:00
|
|
|
if not search_res.get('responseData', {}).get('results'):
|
|
|
|
return []
|
|
|
|
|
2014-09-01 15:10:05 +02:00
|
|
|
# parse results
|
2014-01-29 19:28:38 +01:00
|
|
|
for result in search_res['responseData']['results']:
|
2014-09-01 15:10:05 +02:00
|
|
|
# append result
|
2014-01-29 19:28:38 +01:00
|
|
|
results.append({'url': result['unescapedUrl'],
|
|
|
|
'title': result['titleNoFormatting'],
|
|
|
|
'content': result['content']})
|
2014-09-01 15:10:05 +02:00
|
|
|
|
|
|
|
# return results
|
2014-01-29 19:28:38 +01:00
|
|
|
return results
|