git.iko.soy/searx
on master help sign in
Stackoverflow's unit test
d20ddf9Cqoicebordel
searx/engines/stackoverflow.py+4−4
compare with master
unifiedsplit
@@ -12,6 +12,7 @@ from urlparse import urljoin
12from cgi import escape
12from cgi import escape
13from urllib import urlencode
13from urllib import urlencode
14from lxml import html
14from lxml import html
15+from searx.engines.xpath import extract_text
15
16
16# engine dependent config
17# engine dependent config
17categories = ['it']
18categories = ['it']
@@ -24,8 +25,7 @@ search_url = url+'search?{query}&page={pageno}'
24# specific xpath variables
25# specific xpath variables
25results_xpath = '//div[contains(@class,"question-summary")]'
26results_xpath = '//div[contains(@class,"question-summary")]'
26link_xpath = './/div[@class="result-link"]//a|.//div[@class="summary"]//h3//a'
27link_xpath = './/div[@class="result-link"]//a|.//div[@class="summary"]//h3//a'
27−title_xpath = './/text()'
28−content_xpath = './/div[@class="excerpt"]//text()'
28+content_xpath = './/div[@class="excerpt"]'
29
29
30
30
31# do search-request
31# do search-request
@@ -46,8 +46,8 @@ def response(resp):
46 for result in dom.xpath(results_xpath):
46 for result in dom.xpath(results_xpath):
47 link = result.xpath(link_xpath)[0]
47 link = result.xpath(link_xpath)[0]
48 href = urljoin(url, link.attrib.get('href'))
48 href = urljoin(url, link.attrib.get('href'))
49− title = escape(' '.join(link.xpath(title_xpath)))
49+ title = escape(extract_text(link))
50− content = escape(' '.join(result.xpath(content_xpath)))
50+ content = escape(extract_text(result.xpath(content_xpath)))
51
51
52 # append result
52 # append result
53 results.append({'url': href,
53 results.append({'url': href,
searx/tests/engines/test_stackoverflow.pynew file+106
@@ -0,0 +1,106 @@
1+from collections import defaultdict
2+import mock
3+from searx.engines import stackoverflow
4+from searx.testing import SearxTestCase
5+
6+
7+class TestStackoverflowEngine(SearxTestCase):
8+
9+ def test_request(self):
10+ query = 'test_query'
11+ dicto = defaultdict(dict)
12+ dicto['pageno'] = 0
13+ params = stackoverflow.request(query, dicto)
14+ self.assertTrue('url' in params)
15+ self.assertTrue(query in params['url'])
16+ self.assertTrue('stackoverflow.com' in params['url'])
17+
18+ def test_response(self):
19+ self.assertRaises(AttributeError, stackoverflow.response, None)
20+ self.assertRaises(AttributeError, stackoverflow.response, [])
21+ self.assertRaises(AttributeError, stackoverflow.response, '')
22+ self.assertRaises(AttributeError, stackoverflow.response, '[]')
23+
24+ response = mock.Mock(text='<html></html>')
25+ self.assertEqual(stackoverflow.response(response), [])
26+
27+ html = """
28+ <div class="question-summary search-result" id="answer-id-1783426">
29+ <div class="statscontainer">
30+ <div class="statsarrow"></div>
31+ <div class="stats">
32+ <div class="vote">
33+ <div class="votes answered">
34+ <span class="vote-count-post "><strong>2583</strong></span>
35+ <div class="viewcount">votes</div>
36+ </div>
37+ </div>
38+ </div>
39+ </div>
40+ <div class="summary">
41+ <div class="result-link">
42+ <span>
43+ <a href="/questions/this.is.the.url"
44+ data-searchsession="/questions"
45+ title="Checkout remote Git branch">
46+ This is the title
47+ </a>
48+ </span>
49+ </div>
50+ <div class="excerpt">
51+ This is the content
52+ </div>
53+ <div class="tags user-tags t-git t-git-checkout t-remote-branch">
54+ </div>
55+ <div class="started fr">
56+ answered <span title="2009-11-23 14:26:08Z" class="relativetime">nov 23 '09</span> by
57+ <a href="/users/214090/hallski">hallski</a>
58+ </div>
59+ </div>
60+ </div>
61+ """
62+ response = mock.Mock(text=html)
63+ results = stackoverflow.response(response)
64+ self.assertEqual(type(results), list)
65+ self.assertEqual(len(results), 1)
66+ self.assertEqual(results[0]['title'], 'This is the title')
67+ self.assertEqual(results[0]['url'], 'http://stackoverflow.com/questions/this.is.the.url')
68+ self.assertEqual(results[0]['content'], 'This is the content')
69+
70+ html = """
71+ <div class="statscontainer">
72+ <div class="statsarrow"></div>
73+ <div class="stats">
74+ <div class="vote">
75+ <div class="votes answered">
76+ <span class="vote-count-post "><strong>2583</strong></span>
77+ <div class="viewcount">votes</div>
78+ </div>
79+ </div>
80+ </div>
81+ </div>
82+ <div class="summary">
83+ <div class="result-link">
84+ <span>
85+ <a href="/questions/this.is.the.url"
86+ data-searchsession="/questions"
87+ title="Checkout remote Git branch">
88+ This is the title
89+ </a>
90+ </span>
91+ </div>
92+ <div class="excerpt">
93+ This is the content
94+ </div>
95+ <div class="tags user-tags t-git t-git-checkout t-remote-branch">
96+ </div>
97+ <div class="started fr">
98+ answered <span title="2009-11-23 14:26:08Z" class="relativetime">nov 23 '09</span> by
99+ <a href="/users/214090/hallski">hallski</a>
100+ </div>
101+ </div>
102+ """
103+ response = mock.Mock(text=html)
104+ results = stackoverflow.response(response)
105+ self.assertEqual(type(results), list)
106+ self.assertEqual(len(results), 0)
searx/tests/test_engines.py+1
@@ -16,4 +16,5 @@ from searx.tests.engines.test_mixcloud import * # noqa
16from searx.tests.engines.test_searchcode_code import * # noqa
16from searx.tests.engines.test_searchcode_code import * # noqa
17from searx.tests.engines.test_searchcode_doc import * # noqa
17from searx.tests.engines.test_searchcode_doc import * # noqa
18from searx.tests.engines.test_soundcloud import * # noqa
18from searx.tests.engines.test_soundcloud import * # noqa
19+from searx.tests.engines.test_stackoverflow import * # noqa
19from searx.tests.engines.test_youtube import * # noqa
20from searx.tests.engines.test_youtube import * # noqa