forked from zaclys/searxng
71 lines
1.6 KiB
Python
71 lines
1.6 KiB
Python
"""
|
|
Blekko (Images)
|
|
|
|
@website https://blekko.com
|
|
@provide-api yes (inofficial)
|
|
|
|
@using-api yes
|
|
@results JSON
|
|
@stable yes
|
|
@parse url, title, img_src
|
|
"""
|
|
|
|
from json import loads
|
|
from urllib import urlencode
|
|
|
|
# engine dependent config
|
|
categories = ['images']
|
|
paging = True
|
|
safesearch = True
|
|
|
|
# search-url
|
|
base_url = 'https://blekko.com'
|
|
search_url = '/api/images?{query}&c={c}'
|
|
|
|
# safesearch definitions
|
|
safesearch_types = {2: '1',
|
|
1: '',
|
|
0: '0'}
|
|
|
|
|
|
# do search-request
|
|
def request(query, params):
|
|
c = (params['pageno'] - 1) * 48
|
|
|
|
params['url'] = base_url +\
|
|
search_url.format(query=urlencode({'q': query}),
|
|
c=c)
|
|
|
|
if params['pageno'] != 1:
|
|
params['url'] += '&page={pageno}'.format(pageno=(params['pageno']-1))
|
|
|
|
# let Blekko know we wan't have profiling
|
|
params['cookies']['tag_lesslogging'] = '1'
|
|
|
|
# parse safesearch argument
|
|
params['cookies']['safesearch'] = safesearch_types.get(params['safesearch'], '')
|
|
|
|
return params
|
|
|
|
|
|
# get response from search-request
|
|
def response(resp):
|
|
results = []
|
|
|
|
search_results = loads(resp.text)
|
|
|
|
# return empty array if there are no results
|
|
if not search_results:
|
|
return []
|
|
|
|
for result in search_results:
|
|
# append result
|
|
results.append({'url': result['page_url'],
|
|
'title': result['title'],
|
|
'content': '',
|
|
'img_src': result['url'],
|
|
'template': 'images.html'})
|
|
|
|
# return results
|
|
return results
|