on master help sign in
Merge pull request #609 from LuccoJ/betterwolfram

Improving Wolfram Alpha search hit content

commit8f48c518aa049a37e6e7721cb17ceef92c519ca8 authorAdam Tauber <asciimoo@gmail.com> dateSun 11 Sep 2016, 00:29 (10 years ago) parent390ad59f67db1e
4 files changed+32−12
searx/engines/wolframalpha_api.py+12−4 searx/engines/wolframalpha_noapi.py+12−4 tests/unit/engines/test_wolframalpha_api.py+4−2 tests/unit/engines/test_wolframalpha_noapi.py+4−2
searx/engines/wolframalpha_api.py+12−4
@@ -18,10 +18,10 @@ api_key = '' # defined in settings.yml
1818
1919# xpath variables
2020failure_xpath = '/queryresult[attribute::success="false"]'
21−answer_xpath = '//pod[attribute::primary="true"]/subpod/plaintext'
2221input_xpath = '//pod[starts-with(attribute::id, "Input")]/subpod/plaintext'
2322pods_xpath = '//pod'
2423subpods_xpath = './subpod'
24+pod_primary_xpath = './@primary'
2525pod_id_xpath = './@id'
2626pod_title_xpath = './@title'
2727plaintext_xpath = './plaintext'
@@ -75,13 +75,15 @@ def response(resp):
7575 try:
7676 infobox_title = search_results.xpath(input_xpath)[0].text
7777 except:
78− infobox_title = None
78+ infobox_title = ""
7979
8080 pods = search_results.xpath(pods_xpath)
8181 result_chunks = []
82+ result_content = ""
8283 for pod in pods:
8384 pod_id = pod.xpath(pod_id_xpath)[0]
8485 pod_title = pod.xpath(pod_title_xpath)[0]
86+ pod_is_result = pod.xpath(pod_primary_xpath)
8587
8688 subpods = pod.xpath(subpods_xpath)
8789 if not subpods:
@@ -94,6 +96,10 @@ def response(resp):
9496
9597 if content and pod_id not in image_pods:
9698
99+ if pod_is_result or not result_content:
100+ if pod_id != "Input":
101+ result_content = "%s: %s" % (pod_title, content)
102+
97103 # if no input pod was found, title is first plaintext pod
98104 if not infobox_title:
99105 infobox_title = content
@@ -109,6 +115,8 @@ def response(resp):
109115 if not result_chunks:
110116 return []
111117
118+ title = "Wolfram|Alpha (%s)" % infobox_title
119+
112120 # append infobox
113121 results.append({'infobox': infobox_title,
114122 'attributes': result_chunks,
@@ -116,7 +124,7 @@ def response(resp):
116124
117125 # append link to site
118126 results.append({'url': resp.request.headers['Referer'].decode('utf8'),
119− 'title': 'Wolfram|Alpha',
120− 'content': infobox_title})
127+ 'title': title,
128+ 'content': result_content})
121129
122130 return results
searx/engines/wolframalpha_noapi.py+12−4
@@ -8,9 +8,11 @@
88# @stable no
99# @parse url, infobox
1010
11+from cgi import escape
1112from json import loads
1213from time import time
1314from urllib import urlencode
15+from lxml.etree import XML
1416
1517from searx.poolrequests import get as http_get
1618
@@ -34,7 +36,7 @@ search_url = url + 'input/json.jsp'\
3436referer_url = url + 'input/?{query}'
3537
3638token = {'value': '',
37− 'last_updated': 0}
39+ 'last_updated': None}
3840
3941# pods to display as image in infobox
4042# this pods do return a plaintext, but they look better and are more useful as images
@@ -80,10 +82,12 @@ def response(resp):
8082
8183 # TODO handle resp_json['queryresult']['assumptions']
8284 result_chunks = []
83− infobox_title = None
85+ infobox_title = ""
86+ result_content = ""
8487 for pod in resp_json['queryresult']['pods']:
8588 pod_id = pod.get('id', '')
8689 pod_title = pod.get('title', '')
90+ pod_is_result = pod.get('primary', None)
8791
8892 if 'subpods' not in pod:
8993 continue
@@ -97,6 +101,10 @@ def response(resp):
97101 if subpod['plaintext'] != '(requires interactivity)':
98102 result_chunks.append({'label': pod_title, 'value': subpod['plaintext']})
99103
104+ if pod_is_result or not result_content:
105+ if pod_id != "Input":
106+ result_content = pod_title + ': ' + subpod['plaintext']
107+
100108 elif 'img' in subpod:
101109 result_chunks.append({'label': pod_title, 'image': subpod['img']})
102110
@@ -108,7 +116,7 @@ def response(resp):
108116 'urls': [{'title': 'Wolfram|Alpha', 'url': resp.request.headers['Referer'].decode('utf8')}]})
109117
110118 results.append({'url': resp.request.headers['Referer'].decode('utf8'),
111− 'title': 'Wolfram|Alpha',
112− 'content': infobox_title})
119+ 'title': 'Wolfram|Alpha (' + infobox_title + ')',
120+ 'content': result_content})
113121
114122 return results
tests/unit/engines/test_wolframalpha_api.py+4−2
@@ -103,7 +103,8 @@ class TestWolframAlphaAPIEngine(SearxTestCase):
103103 self.assertEqual(referer_url, results[0]['urls'][0]['url'])
104104 self.assertEqual('Wolfram|Alpha', results[0]['urls'][0]['title'])
105105 self.assertEqual(referer_url, results[1]['url'])
106− self.assertEqual('Wolfram|Alpha', results[1]['title'])
106+ self.assertEqual('Wolfram|Alpha (input_plaintext)', results[1]['title'])
107+ self.assertIn('result_plaintext', results[1]['content'])
107108
108109 # test calc
109110 xml = """<?xml version='1.0' encoding='UTF-8'?>
@@ -161,4 +162,5 @@ class TestWolframAlphaAPIEngine(SearxTestCase):
161162 self.assertEqual(referer_url, results[0]['urls'][0]['url'])
162163 self.assertEqual('Wolfram|Alpha', results[0]['urls'][0]['title'])
163164 self.assertEqual(referer_url, results[1]['url'])
164− self.assertEqual('Wolfram|Alpha', results[1]['title'])
165+ self.assertEqual('Wolfram|Alpha (integral_plaintext)', results[1]['title'])
166+ self.assertIn('integral_plaintext', results[1]['content'])
tests/unit/engines/test_wolframalpha_noapi.py+4−2
@@ -140,7 +140,8 @@ class TestWolframAlphaNoAPIEngine(SearxTestCase):
140140 self.assertEqual(referer_url, results[0]['urls'][0]['url'])
141141 self.assertEqual('Wolfram|Alpha', results[0]['urls'][0]['title'])
142142 self.assertEqual(referer_url, results[1]['url'])
143− self.assertEqual('Wolfram|Alpha', results[1]['title'])
143+ self.assertEqual('Wolfram|Alpha (input_plaintext)', results[1]['title'])
144+ self.assertIn('result_plaintext', results[1]['content'])
144145
145146 # test calc
146147 json = r"""
@@ -219,4 +220,5 @@ class TestWolframAlphaNoAPIEngine(SearxTestCase):
219220 self.assertEqual(referer_url, results[0]['urls'][0]['url'])
220221 self.assertEqual('Wolfram|Alpha', results[0]['urls'][0]['title'])
221222 self.assertEqual(referer_url, results[1]['url'])
222− self.assertEqual('Wolfram|Alpha', results[1]['title'])
223+ self.assertEqual('Wolfram|Alpha (integral_plaintext)', results[1]['title'])
224+ self.assertIn('integral_plaintext', results[1]['content'])