logo

searx

My custom branche(s) on searx, a meta-search engine git clone https://hacktivis.me/git/searx.git

test_utils.py (5169B)

    

  1. # -*- coding: utf-8 -*-
  2. import mock
  3. import sys
  4. from searx.testing import SearxTestCase
  5. from searx import utils
  7. if sys.version_info[0] == 3:
  8.     unicode = str
  11. class TestUtils(SearxTestCase):
  13.     def test_gen_useragent(self):
  14.         self.assertIsInstance(utils.gen_useragent(), str)
  15.         self.assertIsNotNone(utils.gen_useragent())
  16.         self.assertTrue(utils.gen_useragent().startswith('Mozilla'))
  18.     def test_searx_useragent(self):
  19.         self.assertIsInstance(utils.searx_useragent(), str)
  20.         self.assertIsNotNone(utils.searx_useragent())
  21.         self.assertTrue(utils.searx_useragent().startswith('searx'))
  23.     def test_highlight_content(self):
  24.         self.assertEqual(utils.highlight_content(0, None), None)
  25.         self.assertEqual(utils.highlight_content(None, None), None)
  26.         self.assertEqual(utils.highlight_content('', None), None)
  27.         self.assertEqual(utils.highlight_content(False, None), None)
  29.         contents = [
  30.             '<html></html>'
  31.             'not<'
  32.         ]
  33.         for content in contents:
  34.             self.assertEqual(utils.highlight_content(content, None), content)
  36.         content = 'a'
  37.         query = b'test'
  38.         self.assertEqual(utils.highlight_content(content, query), content)
  39.         query = b'a test'
  40.         self.assertEqual(utils.highlight_content(content, query), content)
  42.     def test_html_to_text(self):
  43.         html = """
  44.         <a href="/testlink" class="link_access_account">
  45.             <span class="toto">
  46.                 <span>
  47.                     <img src="test.jpg" />
  48.                 </span>
  49.             </span>
  50.             <span class="titi">
  51.                             Test text
  52.             </span>
  53.         </a>
  54.         """
  55.         self.assertIsInstance(utils.html_to_text(html), unicode)
  56.         self.assertIsNotNone(utils.html_to_text(html))
  57.         self.assertEqual(utils.html_to_text(html), "Test text")
  59.     def test_prettify_url(self):
  60.         data = (('https://searx.me/', 'https://searx.me/'),
  61.                 (u'https://searx.me/ű', u'https://searx.me/ű'),
  62.                 ('https://searx.me/' + (100 * 'a'), 'https://searx.me/[...]aaaaaaaaaaaaaaaaa'),
  63.                 (u'https://searx.me/' + (100 * u'ű'), u'https://searx.me/[...]űűűűűűűűűűűűűűűűű'))
  65.         for test_url, expected in data:
  66.             self.assertEqual(utils.prettify_url(test_url, max_length=32), expected)
  68.     def test_match_language(self):
  69.         self.assertEqual(utils.match_language('es', ['es']), 'es')
  70.         self.assertEqual(utils.match_language('es', [], fallback='fallback'), 'fallback')
  71.         self.assertEqual(utils.match_language('ja', ['jp'], {'ja': 'jp'}), 'jp')
  73.         aliases = {'en-GB': 'en-UK', 'he': 'iw'}
  75.         # guess country
  76.         self.assertEqual(utils.match_language('de-DE', ['de']), 'de')
  77.         self.assertEqual(utils.match_language('de', ['de-DE']), 'de-DE')
  78.         self.assertEqual(utils.match_language('es-CO', ['es-AR', 'es-ES', 'es-MX']), 'es-ES')
  79.         self.assertEqual(utils.match_language('es-CO', ['es-MX']), 'es-MX')
  80.         self.assertEqual(utils.match_language('en-UK', ['en-AU', 'en-GB', 'en-US']), 'en-GB')
  81.         self.assertEqual(utils.match_language('en-GB', ['en-AU', 'en-UK', 'en-US'], aliases), 'en-UK')
  83.         # language aliases
  84.         self.assertEqual(utils.match_language('iw', ['he']), 'he')
  85.         self.assertEqual(utils.match_language('he', ['iw'], aliases), 'iw')
  86.         self.assertEqual(utils.match_language('iw-IL', ['he']), 'he')
  87.         self.assertEqual(utils.match_language('he-IL', ['iw'], aliases), 'iw')
  88.         self.assertEqual(utils.match_language('iw', ['he-IL']), 'he-IL')
  89.         self.assertEqual(utils.match_language('he', ['iw-IL'], aliases), 'iw-IL')
  90.         self.assertEqual(utils.match_language('iw-IL', ['he-IL']), 'he-IL')
  91.         self.assertEqual(utils.match_language('he-IL', ['iw-IL'], aliases), 'iw-IL')
  94. class TestHTMLTextExtractor(SearxTestCase):
  96.     def setUp(self):
  97.         self.html_text_extractor = utils.HTMLTextExtractor()
  99.     def test__init__(self):
  100.         self.assertEqual(self.html_text_extractor.result, [])
  102.     def test_handle_charref(self):
  103.         self.html_text_extractor.handle_charref('xF')
  104.         self.assertIn(u'\x0f', self.html_text_extractor.result)
  105.         self.html_text_extractor.handle_charref('XF')
  106.         self.assertIn(u'\x0f', self.html_text_extractor.result)
  108.         self.html_text_extractor.handle_charref('97')
  109.         self.assertIn(u'a', self.html_text_extractor.result)
  111.     def test_handle_entityref(self):
  112.         entity = 'test'
  113.         self.html_text_extractor.handle_entityref(entity)
  114.         self.assertIn(entity, self.html_text_extractor.result)
  117. class TestUnicodeWriter(SearxTestCase):
  119.     def setUp(self):
  120.         self.unicode_writer = utils.UnicodeWriter(mock.MagicMock())
  122.     def test_write_row(self):
  123.         row = [1, 2, 3]
  124.         self.assertEqual(self.unicode_writer.writerow(row), None)
  126.     def test_write_rows(self):
  127.         self.unicode_writer.writerow = mock.MagicMock()
  128.         rows = [1, 2, 3]
  129.         self.unicode_writer.writerows(rows)
  130.         self.assertEqual(self.unicode_writer.writerow.call_count, len(rows))