test_subtitles.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309
  1. #!/usr/bin/env python
  2. from __future__ import unicode_literals
  3. # Allow direct execution
  4. import os
  5. import sys
  6. import unittest
  7. sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  8. from test.helper import FakeYDL, md5
  9. from youtube_dl.extractor import (
  10. BlipTVIE,
  11. YoutubeIE,
  12. DailymotionIE,
  13. TEDIE,
  14. VimeoIE,
  15. WallaIE,
  16. CeskaTelevizeIE,
  17. LyndaIE,
  18. NPOIE,
  19. ComedyCentralIE,
  20. NRKTVIE,
  21. RaiIE,
  22. VikiIE,
  23. ThePlatformIE,
  24. )
  25. class BaseTestSubtitles(unittest.TestCase):
  26. url = None
  27. IE = None
  28. def setUp(self):
  29. self.DL = FakeYDL()
  30. self.ie = self.IE()
  31. self.DL.add_info_extractor(self.ie)
  32. def getInfoDict(self):
  33. info_dict = self.DL.extract_info(self.url, download=False)
  34. return info_dict
  35. def getSubtitles(self):
  36. info_dict = self.getInfoDict()
  37. subtitles = info_dict['requested_subtitles']
  38. if not subtitles:
  39. return subtitles
  40. for sub_info in subtitles.values():
  41. if sub_info.get('data') is None:
  42. uf = self.DL.urlopen(sub_info['url'])
  43. sub_info['data'] = uf.read().decode('utf-8')
  44. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  45. class TestYoutubeSubtitles(BaseTestSubtitles):
  46. url = 'QRS8MkLhQmM'
  47. IE = YoutubeIE
  48. def test_youtube_allsubtitles(self):
  49. self.DL.params['writesubtitles'] = True
  50. self.DL.params['allsubtitles'] = True
  51. subtitles = self.getSubtitles()
  52. self.assertEqual(len(subtitles.keys()), 13)
  53. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  54. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  55. for lang in ['it', 'fr', 'de']:
  56. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  57. def test_youtube_subtitles_sbv_format(self):
  58. self.DL.params['writesubtitles'] = True
  59. self.DL.params['subtitlesformat'] = 'sbv'
  60. subtitles = self.getSubtitles()
  61. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  62. def test_youtube_subtitles_vtt_format(self):
  63. self.DL.params['writesubtitles'] = True
  64. self.DL.params['subtitlesformat'] = 'vtt'
  65. subtitles = self.getSubtitles()
  66. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  67. def test_youtube_automatic_captions(self):
  68. self.url = '8YoUxe5ncPo'
  69. self.DL.params['writeautomaticsub'] = True
  70. self.DL.params['subtitleslangs'] = ['it']
  71. subtitles = self.getSubtitles()
  72. self.assertTrue(subtitles['it'] is not None)
  73. def test_youtube_translated_subtitles(self):
  74. # This video has a subtitles track, which can be translated
  75. self.url = 'Ky9eprVWzlI'
  76. self.DL.params['writeautomaticsub'] = True
  77. self.DL.params['subtitleslangs'] = ['it']
  78. subtitles = self.getSubtitles()
  79. self.assertTrue(subtitles['it'] is not None)
  80. def test_youtube_nosubtitles(self):
  81. self.DL.expect_warning('video doesn\'t have subtitles')
  82. self.url = 'n5BB19UTcdA'
  83. self.DL.params['writesubtitles'] = True
  84. self.DL.params['allsubtitles'] = True
  85. subtitles = self.getSubtitles()
  86. self.assertFalse(subtitles)
  87. class TestDailymotionSubtitles(BaseTestSubtitles):
  88. url = 'http://www.dailymotion.com/video/xczg00'
  89. IE = DailymotionIE
  90. def test_allsubtitles(self):
  91. self.DL.params['writesubtitles'] = True
  92. self.DL.params['allsubtitles'] = True
  93. subtitles = self.getSubtitles()
  94. self.assertTrue(len(subtitles.keys()) >= 6)
  95. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  96. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  97. for lang in ['es', 'fr', 'de']:
  98. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  99. def test_nosubtitles(self):
  100. self.DL.expect_warning('video doesn\'t have subtitles')
  101. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  102. self.DL.params['writesubtitles'] = True
  103. self.DL.params['allsubtitles'] = True
  104. subtitles = self.getSubtitles()
  105. self.assertFalse(subtitles)
  106. class TestTedSubtitles(BaseTestSubtitles):
  107. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  108. IE = TEDIE
  109. def test_allsubtitles(self):
  110. self.DL.params['writesubtitles'] = True
  111. self.DL.params['allsubtitles'] = True
  112. subtitles = self.getSubtitles()
  113. self.assertTrue(len(subtitles.keys()) >= 28)
  114. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  115. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  116. for lang in ['es', 'fr', 'de']:
  117. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  118. class TestBlipTVSubtitles(BaseTestSubtitles):
  119. url = 'http://blip.tv/a/a-6603250'
  120. IE = BlipTVIE
  121. def test_allsubtitles(self):
  122. self.DL.params['writesubtitles'] = True
  123. self.DL.params['allsubtitles'] = True
  124. subtitles = self.getSubtitles()
  125. self.assertEqual(set(subtitles.keys()), set(['en']))
  126. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  127. class TestVimeoSubtitles(BaseTestSubtitles):
  128. url = 'http://vimeo.com/76979871'
  129. IE = VimeoIE
  130. def test_allsubtitles(self):
  131. self.DL.params['writesubtitles'] = True
  132. self.DL.params['allsubtitles'] = True
  133. subtitles = self.getSubtitles()
  134. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  135. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  136. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  137. def test_nosubtitles(self):
  138. self.DL.expect_warning('video doesn\'t have subtitles')
  139. self.url = 'http://vimeo.com/56015672'
  140. self.DL.params['writesubtitles'] = True
  141. self.DL.params['allsubtitles'] = True
  142. subtitles = self.getSubtitles()
  143. self.assertFalse(subtitles)
  144. class TestWallaSubtitles(BaseTestSubtitles):
  145. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  146. IE = WallaIE
  147. def test_allsubtitles(self):
  148. self.DL.expect_warning('Automatic Captions not supported by this server')
  149. self.DL.params['writesubtitles'] = True
  150. self.DL.params['allsubtitles'] = True
  151. subtitles = self.getSubtitles()
  152. self.assertEqual(set(subtitles.keys()), set(['heb']))
  153. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  154. def test_nosubtitles(self):
  155. self.DL.expect_warning('video doesn\'t have subtitles')
  156. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  157. self.DL.params['writesubtitles'] = True
  158. self.DL.params['allsubtitles'] = True
  159. subtitles = self.getSubtitles()
  160. self.assertFalse(subtitles)
  161. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  162. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  163. IE = CeskaTelevizeIE
  164. def test_allsubtitles(self):
  165. self.DL.expect_warning('Automatic Captions not supported by this server')
  166. self.DL.params['writesubtitles'] = True
  167. self.DL.params['allsubtitles'] = True
  168. subtitles = self.getSubtitles()
  169. self.assertEqual(set(subtitles.keys()), set(['cs']))
  170. self.assertTrue(len(subtitles['cs']) > 20000)
  171. def test_nosubtitles(self):
  172. self.DL.expect_warning('video doesn\'t have subtitles')
  173. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  174. self.DL.params['writesubtitles'] = True
  175. self.DL.params['allsubtitles'] = True
  176. subtitles = self.getSubtitles()
  177. self.assertFalse(subtitles)
  178. class TestLyndaSubtitles(BaseTestSubtitles):
  179. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  180. IE = LyndaIE
  181. def test_allsubtitles(self):
  182. self.DL.params['writesubtitles'] = True
  183. self.DL.params['allsubtitles'] = True
  184. subtitles = self.getSubtitles()
  185. self.assertEqual(set(subtitles.keys()), set(['en']))
  186. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  187. class TestNPOSubtitles(BaseTestSubtitles):
  188. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  189. IE = NPOIE
  190. def test_allsubtitles(self):
  191. self.DL.params['writesubtitles'] = True
  192. self.DL.params['allsubtitles'] = True
  193. subtitles = self.getSubtitles()
  194. self.assertEqual(set(subtitles.keys()), set(['nl']))
  195. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  196. class TestMTVSubtitles(BaseTestSubtitles):
  197. url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
  198. IE = ComedyCentralIE
  199. def getInfoDict(self):
  200. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  201. def test_allsubtitles(self):
  202. self.DL.params['writesubtitles'] = True
  203. self.DL.params['allsubtitles'] = True
  204. subtitles = self.getSubtitles()
  205. self.assertEqual(set(subtitles.keys()), set(['en']))
  206. self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
  207. class TestNRKSubtitles(BaseTestSubtitles):
  208. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  209. IE = NRKTVIE
  210. def test_allsubtitles(self):
  211. self.DL.params['writesubtitles'] = True
  212. self.DL.params['allsubtitles'] = True
  213. subtitles = self.getSubtitles()
  214. self.assertEqual(set(subtitles.keys()), set(['no']))
  215. self.assertEqual(md5(subtitles['no']), '1d221e6458c95c5494dcd38e6a1f129a')
  216. class TestRaiSubtitles(BaseTestSubtitles):
  217. url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  218. IE = RaiIE
  219. def test_allsubtitles(self):
  220. self.DL.params['writesubtitles'] = True
  221. self.DL.params['allsubtitles'] = True
  222. subtitles = self.getSubtitles()
  223. self.assertEqual(set(subtitles.keys()), set(['it']))
  224. self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  225. class TestVikiSubtitles(BaseTestSubtitles):
  226. url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
  227. IE = VikiIE
  228. def test_allsubtitles(self):
  229. self.DL.params['writesubtitles'] = True
  230. self.DL.params['allsubtitles'] = True
  231. subtitles = self.getSubtitles()
  232. self.assertEqual(set(subtitles.keys()), set(['en']))
  233. self.assertEqual(md5(subtitles['en']), 'b0b781eeb45efd3f6398a925b259150b')
  234. class TestThePlatformSubtitles(BaseTestSubtitles):
  235. # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
  236. # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
  237. url = 'theplatform:JFUjUE1_ehvq'
  238. IE = ThePlatformIE
  239. def test_allsubtitles(self):
  240. self.DL.params['writesubtitles'] = True
  241. self.DL.params['allsubtitles'] = True
  242. subtitles = self.getSubtitles()
  243. self.assertEqual(set(subtitles.keys()), set(['en']))
  244. self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
  245. if __name__ == '__main__':
  246. unittest.main()