test/test_InfoExtractor.py

   1 #!/usr/bin/env python
   2
   3 from __future__ import unicode_literals
   4
   5 # Allow direct execution
   6 import os
   7 import sys
   8 import unittest
   9 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  10
  11 from test.helper import FakeYDL
  12 from youtube_dl.extractor.common import InfoExtractor
  13 from youtube_dl.extractor import YoutubeIE, get_info_extractor
  14
  15
  16 class TestIE(InfoExtractor):
  17     pass
  18
  19
  20 class TestInfoExtractor(unittest.TestCase):
  21     def setUp(self):
  22         self.ie = TestIE(FakeYDL())
  23
  24     def test_ie_key(self):
  25         self.assertEqual(get_info_extractor(YoutubeIE.ie_key()), YoutubeIE)
  26
  27     def test_html_search_regex(self):
  28         html = '<p id="foo">Watch this <a href="http://www.youtube.com/watch?v=BaW_jenozKc">video</a></p>'
  29         search = lambda re, *args: self.ie._html_search_regex(re, html, *args)
  30         self.assertEqual(search(r'<p id="foo">(.+?)</p>', 'foo'), 'Watch this video')
  31
  32     def test_opengraph(self):
  33         ie = self.ie
  34         html = '''
  35             <meta name="og:title" content='Foo'/>
  36             <meta content="Some video's description " name="og:description"/>
  37             <meta property='og:image' content='http://domain.com/pic.jpg?key1=val1&amp;key2=val2'/>
  38             <meta content='application/x-shockwave-flash' property='og:video:type'>
  39             <meta content='Foo' property=og:foobar>
  40             <meta name="og:test1" content='foo > < bar'/>
  41             <meta name="og:test2" content="foo >//< bar"/>
  42             '''
  43         self.assertEqual(ie._og_search_title(html), 'Foo')
  44         self.assertEqual(ie._og_search_description(html), 'Some video\'s description ')
  45         self.assertEqual(ie._og_search_thumbnail(html), 'http://domain.com/pic.jpg?key1=val1&key2=val2')
  46         self.assertEqual(ie._og_search_video_url(html, default=None), None)
  47         self.assertEqual(ie._og_search_property('foobar', html), 'Foo')
  48         self.assertEqual(ie._og_search_property('test1', html), 'foo > < bar')
  49         self.assertEqual(ie._og_search_property('test2', html), 'foo >//< bar')
  50
  51     def test_html_search_meta(self):
  52         ie = self.ie
  53         html = '''
  54             <meta name="a" content="1" />
  55             <meta name='b' content='2'>
  56             <meta name="c" content='3'>
  57             <meta name=d content='4'>
  58             <meta property="e" content='5' >
  59             <meta content="6" name="f">
  60         '''
  61
  62         self.assertEqual(ie._html_search_meta('a', html), '1')
  63         self.assertEqual(ie._html_search_meta('b', html), '2')
  64         self.assertEqual(ie._html_search_meta('c', html), '3')
  65         self.assertEqual(ie._html_search_meta('d', html), '4')
  66         self.assertEqual(ie._html_search_meta('e', html), '5')
  67         self.assertEqual(ie._html_search_meta('f', html), '6')
  68
  69 if __name__ == '__main__':
  70     unittest.main()