searx

commit: b51ba32f619e6b7a927444475b0ee986d4d13a60
parent d4b6ab2be47c7b7ac6ae222905a8848fc428cd69
Author: a01200356 <a01200356@itesm.mx>
Date:   Tue, 29 Dec 2015 20:59:51 -0600

Wolfram Alpha (no API needed now)

Diffstat:
A
searx/engines/wolframalpha_noapi.py
66
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
M
searx/settings.yml
8
++++----

2 files changed, 70 insertions(+), 4 deletions(-)
diff --git a/searx/engines/wolframalpha_noapi.py b/searx/engines/wolframalpha_noapi.py
@@ -0,0 +1,66 @@
+# WolframAlpha (Maths)
+#
+# @website     http://www.wolframalpha.com/
+#
+# @using-api   no
+# @results     HTML, JS
+# @stable      no
+# @parse       answer
+
+import re
+import json
+from urllib import urlencode
+from lxml import html
+from searx.engines.xpath import extract_text
+
+# search-url
+url = 'http://www.wolframalpha.com/'
+search_url = url+'input/?{query}'
+
+
+# do search-request
+def request(query, params):
+    params['url'] = search_url.format(query=urlencode({'i': query}))
+
+    return params
+
+
+# tries to find answer under the pattern given
+def extract_answer(script_list, pattern):
+    answer = None
+
+    # get line that matches the pattern
+    for script in script_list:
+        try:
+            line = re.search(pattern, script.text_content()).group(1)
+        except AttributeError:
+            continue
+
+        # extract answer from json
+        answer = line[line.find('{') : line.rfind('}')+1]
+        answer = json.loads(answer.encode('unicode-escape'))
+        answer = answer['stringified'].decode('unicode-escape')
+
+    return answer
+
+
+# get response from search-request
+def response(resp):
+
+    dom = html.fromstring(resp.text)
+
+    # the answer is inside a js script
+    scripts = dom.xpath('//script')
+
+    results = []
+
+    # answer can be located in different 'pods', although by default it should be in pod_0200
+    answer = extract_answer(scripts, 'pod_0200\.push(.*)\n')
+    if not answer:
+        answer = extract_answer(scripts, 'pod_0100\.push(.*)\n')
+        if answer:
+            results.append({'answer': answer})
+    else:
+        results.append({'answer': answer})
+    
+    return results
diff --git a/searx/settings.yml b/searx/settings.yml
@@ -302,11 +302,11 @@ engines:
 
 # You can use the engine using the official stable API, but you need an API key
 # See : http://products.wolframalpha.com/api/
-#  - name : wolframalpha
-#    shortcut : wa
-#    engine : wolframalpha_api
+  - name : wolframalpha
+    shortcut : wa
+    engine : wolframalpha_noapi
 #    api_key: 'apikey' # required!
-#    timeout: 6.0
+    timeout: 6.0
 
 #The blekko technology and team have joined IBM Watson! -> https://blekko.com/
 #  - name : blekko images