576d37f · searx · git.iko.soy

[fix] unescape htmlentities in wolframalpha_noapi's answer

commit576d37f256649b570a9c8591a795acd85ac499bc authora01200356 <a01200356@itesm.mx> dateSun 3 Jan 2016, 15:58 (11 years ago) parent16d6e75

2 files changed+8−2

unifiedsplit

searx/engines/wolframalpha_noapi.py+6−1

@@ -12,6 +12,7 @@ from re import search, sub

1212from json import loads

1313from urllib import urlencode

1414from lxml import html

15+import HTMLParser

1516

1617# search-url

1718url = 'http://www.wolframalpha.com/'

@@ -62,7 +63,11 @@ def response(resp):

6263 # extract answer from json

6364 answer = line[line.find('{'):line.rfind('}')+1]

6465 answer = loads(answer.encode('unicode-escape'))

65− answer = answer['stringified'].decode('unicode-escape')

66+ answer = answer['stringified']

67+

68+ # clean plaintext answer

69+ h = HTMLParser.HTMLParser()

70+ answer = h.unescape(answer.decode('unicode-escape'))

6671 answer = sub(r'\\', '', answer)

6772

6873 results.append({'answer': answer})

searx/settings.yml+2−1

@@ -305,9 +305,10 @@ engines:

305305 # You can use the engine using the official stable API, but you need an API key

306306 # See : http://products.wolframalpha.com/api/

307307 # engine : wolframalpha_api

308− # api_key: '5952JX-X52L3VKWT8' # required!

308+ # api_key: '' # required!

309309 engine : wolframalpha_noapi

310310 timeout: 6.0

311+ disabled : True

311312

312313#The blekko technology and team have joined IBM Watson! -> https://blekko.com/

313314# - name : blekko images