on master help sign in
A bit of utils unit tests
cfe81d7Cqoicebordel
searx/tests/test_utils.py+22
@@ -10,6 +10,11 @@ class TestUtils(SearxTestCase):
10 self.assertIsNotNone(utils.gen_useragent())
10 self.assertIsNotNone(utils.gen_useragent())
11 self.assertTrue(utils.gen_useragent().startswith('Mozilla'))
11 self.assertTrue(utils.gen_useragent().startswith('Mozilla'))
12
12
13+ def test_searx_useragent(self):
14+ self.assertIsInstance(utils.searx_useragent(), str)
15+ self.assertIsNotNone(utils.searx_useragent())
16+ self.assertTrue(utils.searx_useragent().startswith('searx'))
17+
13 def test_highlight_content(self):
18 def test_highlight_content(self):
14 self.assertEqual(utils.highlight_content(0, None), None)
19 self.assertEqual(utils.highlight_content(0, None), None)
15 self.assertEqual(utils.highlight_content(None, None), None)
20 self.assertEqual(utils.highlight_content(None, None), None)
@@ -29,6 +34,23 @@ class TestUtils(SearxTestCase):
29 query = 'a test'
34 query = 'a test'
30 self.assertEqual(utils.highlight_content(content, query), content)
35 self.assertEqual(utils.highlight_content(content, query), content)
31
36
37+ def test_html_to_text(self):
38+ html = """
39+ <a href="/testlink" class="link_access_account">
40+ <span class="toto">
41+ <span>
42+ <img src="test.jpg" />
43+ </span>
44+ </span>
45+ <span class="titi">
46+ Test text
47+ </span>
48+ </a>
49+ """
50+ self.assertIsInstance(utils.html_to_text(html), unicode)
51+ self.assertIsNotNone(utils.html_to_text(html))
52+ self.assertEqual(utils.html_to_text(html), "Test text")
53+
32
54
33class TestHTMLTextExtractor(SearxTestCase):
55class TestHTMLTextExtractor(SearxTestCase):
34
56
searx/utils.py+1−1
@@ -115,7 +115,7 @@ class HTMLTextExtractor(HTMLParser):
115 self.result.append(name)
115 self.result.append(name)
116
116
117 def get_text(self):
117 def get_text(self):
118− return u''.join(self.result)
118+ return u''.join(self.result).strip()
119
119
120
120
121def html_to_text(html):
121def html_to_text(html):