2014-03-29 16:30:49 +01:00
|
|
|
from lxml import etree
|
|
|
|
from requests import get
|
|
|
|
from json import loads
|
2014-03-29 17:04:33 +01:00
|
|
|
from urllib import urlencode
|
2014-03-29 16:30:49 +01:00
|
|
|
|
|
|
|
|
|
|
|
def dbpedia(query):
|
|
|
|
# dbpedia autocompleter
|
2014-03-29 17:04:33 +01:00
|
|
|
autocomplete_url = 'http://lookup.dbpedia.org/api/search.asmx/KeywordSearch?' # noqa
|
2014-03-29 16:30:49 +01:00
|
|
|
|
2014-03-29 17:04:33 +01:00
|
|
|
response = get(autocomplete_url
|
|
|
|
+ urlencode(dict(QueryString=query)))
|
2014-03-29 16:30:49 +01:00
|
|
|
|
|
|
|
results = []
|
|
|
|
|
|
|
|
if response.ok:
|
|
|
|
dom = etree.fromstring(response.content)
|
|
|
|
results = dom.xpath('//a:Result/a:Label//text()',
|
|
|
|
namespaces={'a': 'http://lookup.dbpedia.org/'})
|
|
|
|
|
|
|
|
return results
|
|
|
|
|
|
|
|
|
2014-09-07 23:56:06 +02:00
|
|
|
def duckduckgo(query):
|
|
|
|
# wikipedia autocompleter
|
|
|
|
url = 'https://ac.duckduckgo.com/ac/?{0}&type=list'
|
|
|
|
|
|
|
|
resp = loads(get(url.format(urlencode(dict(q=query)))).text)
|
|
|
|
if len(resp) > 1:
|
|
|
|
return resp[1]
|
|
|
|
return []
|
|
|
|
|
|
|
|
|
2014-03-29 16:30:49 +01:00
|
|
|
def google(query):
|
|
|
|
# google autocompleter
|
2014-03-29 17:04:33 +01:00
|
|
|
autocomplete_url = 'http://suggestqueries.google.com/complete/search?client=toolbar&' # noqa
|
2014-03-29 16:30:49 +01:00
|
|
|
|
2014-03-29 17:04:33 +01:00
|
|
|
response = get(autocomplete_url
|
|
|
|
+ urlencode(dict(q=query)))
|
2014-03-29 16:30:49 +01:00
|
|
|
|
|
|
|
results = []
|
|
|
|
|
|
|
|
if response.ok:
|
2014-03-29 17:04:33 +01:00
|
|
|
dom = etree.fromstring(response.text)
|
2014-03-29 16:30:49 +01:00
|
|
|
results = dom.xpath('//suggestion/@data')
|
|
|
|
|
|
|
|
return results
|
|
|
|
|
|
|
|
|
|
|
|
def wikipedia(query):
|
|
|
|
# wikipedia autocompleter
|
2014-03-29 17:04:33 +01:00
|
|
|
url = 'https://en.wikipedia.org/w/api.php?action=opensearch&{0}&limit=10&namespace=0&format=json' # noqa
|
2014-03-29 16:30:49 +01:00
|
|
|
|
2014-03-29 17:04:33 +01:00
|
|
|
resp = loads(get(url.format(urlencode(dict(q=query)))).text)
|
2014-03-29 17:15:59 +01:00
|
|
|
if len(resp) > 1:
|
|
|
|
return resp[1]
|
|
|
|
return []
|
2014-03-29 16:30:49 +01:00
|
|
|
|
|
|
|
|
|
|
|
backends = {'dbpedia': dbpedia,
|
2014-09-07 23:56:06 +02:00
|
|
|
'duckduckgo': duckduckgo,
|
2014-03-29 16:30:49 +01:00
|
|
|
'google': google,
|
|
|
|
'wikipedia': wikipedia
|
|
|
|
}
|