test_subtitles.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324
  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. RTVEALaCartaIE,
  25. )
  26. class BaseTestSubtitles(unittest.TestCase):
  27. url = None
  28. IE = None
  29. def setUp(self):
  30. self.DL = FakeYDL()
  31. self.ie = self.IE()
  32. self.DL.add_info_extractor(self.ie)
  33. def getInfoDict(self):
  34. info_dict = self.DL.extract_info(self.url, download=False)
  35. return info_dict
  36. def getSubtitles(self):
  37. info_dict = self.getInfoDict()
  38. subtitles = info_dict['requested_subtitles']
  39. if not subtitles:
  40. return subtitles
  41. for sub_info in subtitles.values():
  42. if sub_info.get('data') is None:
  43. uf = self.DL.urlopen(sub_info['url'])
  44. sub_info['data'] = uf.read().decode('utf-8')
  45. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  46. class TestYoutubeSubtitles(BaseTestSubtitles):
  47. url = 'QRS8MkLhQmM'
  48. IE = YoutubeIE
  49. def test_youtube_allsubtitles(self):
  50. self.DL.params['writesubtitles'] = True
  51. self.DL.params['allsubtitles'] = True
  52. subtitles = self.getSubtitles()
  53. self.assertEqual(len(subtitles.keys()), 13)
  54. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  55. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  56. for lang in ['it', 'fr', 'de']:
  57. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  58. def test_youtube_subtitles_sbv_format(self):
  59. self.DL.params['writesubtitles'] = True
  60. self.DL.params['subtitlesformat'] = 'sbv'
  61. subtitles = self.getSubtitles()
  62. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  63. def test_youtube_subtitles_vtt_format(self):
  64. self.DL.params['writesubtitles'] = True
  65. self.DL.params['subtitlesformat'] = 'vtt'
  66. subtitles = self.getSubtitles()
  67. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  68. def test_youtube_automatic_captions(self):
  69. self.url = '8YoUxe5ncPo'
  70. self.DL.params['writeautomaticsub'] = True
  71. self.DL.params['subtitleslangs'] = ['it']
  72. subtitles = self.getSubtitles()
  73. self.assertTrue(subtitles['it'] is not None)
  74. def test_youtube_translated_subtitles(self):
  75. # This video has a subtitles track, which can be translated
  76. self.url = 'Ky9eprVWzlI'
  77. self.DL.params['writeautomaticsub'] = True
  78. self.DL.params['subtitleslangs'] = ['it']
  79. subtitles = self.getSubtitles()
  80. self.assertTrue(subtitles['it'] is not None)
  81. def test_youtube_nosubtitles(self):
  82. self.DL.expect_warning('video doesn\'t have subtitles')
  83. self.url = 'n5BB19UTcdA'
  84. self.DL.params['writesubtitles'] = True
  85. self.DL.params['allsubtitles'] = True
  86. subtitles = self.getSubtitles()
  87. self.assertFalse(subtitles)
  88. class TestDailymotionSubtitles(BaseTestSubtitles):
  89. url = 'http://www.dailymotion.com/video/xczg00'
  90. IE = DailymotionIE
  91. def test_allsubtitles(self):
  92. self.DL.params['writesubtitles'] = True
  93. self.DL.params['allsubtitles'] = True
  94. subtitles = self.getSubtitles()
  95. self.assertTrue(len(subtitles.keys()) >= 6)
  96. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  97. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  98. for lang in ['es', 'fr', 'de']:
  99. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  100. def test_nosubtitles(self):
  101. self.DL.expect_warning('video doesn\'t have subtitles')
  102. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  103. self.DL.params['writesubtitles'] = True
  104. self.DL.params['allsubtitles'] = True
  105. subtitles = self.getSubtitles()
  106. self.assertFalse(subtitles)
  107. class TestTedSubtitles(BaseTestSubtitles):
  108. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  109. IE = TEDIE
  110. def test_allsubtitles(self):
  111. self.DL.params['writesubtitles'] = True
  112. self.DL.params['allsubtitles'] = True
  113. subtitles = self.getSubtitles()
  114. self.assertTrue(len(subtitles.keys()) >= 28)
  115. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  116. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  117. for lang in ['es', 'fr', 'de']:
  118. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  119. class TestBlipTVSubtitles(BaseTestSubtitles):
  120. url = 'http://blip.tv/a/a-6603250'
  121. IE = BlipTVIE
  122. def test_allsubtitles(self):
  123. self.DL.params['writesubtitles'] = True
  124. self.DL.params['allsubtitles'] = True
  125. subtitles = self.getSubtitles()
  126. self.assertEqual(set(subtitles.keys()), set(['en']))
  127. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  128. class TestVimeoSubtitles(BaseTestSubtitles):
  129. url = 'http://vimeo.com/76979871'
  130. IE = VimeoIE
  131. def test_allsubtitles(self):
  132. self.DL.params['writesubtitles'] = True
  133. self.DL.params['allsubtitles'] = True
  134. subtitles = self.getSubtitles()
  135. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  136. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  137. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  138. def test_nosubtitles(self):
  139. self.DL.expect_warning('video doesn\'t have subtitles')
  140. self.url = 'http://vimeo.com/56015672'
  141. self.DL.params['writesubtitles'] = True
  142. self.DL.params['allsubtitles'] = True
  143. subtitles = self.getSubtitles()
  144. self.assertFalse(subtitles)
  145. class TestWallaSubtitles(BaseTestSubtitles):
  146. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  147. IE = WallaIE
  148. def test_allsubtitles(self):
  149. self.DL.expect_warning('Automatic Captions not supported by this server')
  150. self.DL.params['writesubtitles'] = True
  151. self.DL.params['allsubtitles'] = True
  152. subtitles = self.getSubtitles()
  153. self.assertEqual(set(subtitles.keys()), set(['heb']))
  154. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  155. def test_nosubtitles(self):
  156. self.DL.expect_warning('video doesn\'t have subtitles')
  157. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  158. self.DL.params['writesubtitles'] = True
  159. self.DL.params['allsubtitles'] = True
  160. subtitles = self.getSubtitles()
  161. self.assertFalse(subtitles)
  162. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  163. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  164. IE = CeskaTelevizeIE
  165. def test_allsubtitles(self):
  166. self.DL.expect_warning('Automatic Captions not supported by this server')
  167. self.DL.params['writesubtitles'] = True
  168. self.DL.params['allsubtitles'] = True
  169. subtitles = self.getSubtitles()
  170. self.assertEqual(set(subtitles.keys()), set(['cs']))
  171. self.assertTrue(len(subtitles['cs']) > 20000)
  172. def test_nosubtitles(self):
  173. self.DL.expect_warning('video doesn\'t have subtitles')
  174. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  175. self.DL.params['writesubtitles'] = True
  176. self.DL.params['allsubtitles'] = True
  177. subtitles = self.getSubtitles()
  178. self.assertFalse(subtitles)
  179. class TestLyndaSubtitles(BaseTestSubtitles):
  180. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  181. IE = LyndaIE
  182. def test_allsubtitles(self):
  183. self.DL.params['writesubtitles'] = True
  184. self.DL.params['allsubtitles'] = True
  185. subtitles = self.getSubtitles()
  186. self.assertEqual(set(subtitles.keys()), set(['en']))
  187. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  188. class TestNPOSubtitles(BaseTestSubtitles):
  189. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  190. IE = NPOIE
  191. def test_allsubtitles(self):
  192. self.DL.params['writesubtitles'] = True
  193. self.DL.params['allsubtitles'] = True
  194. subtitles = self.getSubtitles()
  195. self.assertEqual(set(subtitles.keys()), set(['nl']))
  196. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  197. class TestMTVSubtitles(BaseTestSubtitles):
  198. url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
  199. IE = ComedyCentralIE
  200. def getInfoDict(self):
  201. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  202. def test_allsubtitles(self):
  203. self.DL.params['writesubtitles'] = True
  204. self.DL.params['allsubtitles'] = True
  205. subtitles = self.getSubtitles()
  206. self.assertEqual(set(subtitles.keys()), set(['en']))
  207. self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
  208. class TestNRKSubtitles(BaseTestSubtitles):
  209. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  210. IE = NRKTVIE
  211. def test_allsubtitles(self):
  212. self.DL.params['writesubtitles'] = True
  213. self.DL.params['allsubtitles'] = True
  214. subtitles = self.getSubtitles()
  215. self.assertEqual(set(subtitles.keys()), set(['no']))
  216. self.assertEqual(md5(subtitles['no']), '1d221e6458c95c5494dcd38e6a1f129a')
  217. class TestRaiSubtitles(BaseTestSubtitles):
  218. url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  219. IE = RaiIE
  220. def test_allsubtitles(self):
  221. self.DL.params['writesubtitles'] = True
  222. self.DL.params['allsubtitles'] = True
  223. subtitles = self.getSubtitles()
  224. self.assertEqual(set(subtitles.keys()), set(['it']))
  225. self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  226. class TestVikiSubtitles(BaseTestSubtitles):
  227. url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
  228. IE = VikiIE
  229. def test_allsubtitles(self):
  230. self.DL.params['writesubtitles'] = True
  231. self.DL.params['allsubtitles'] = True
  232. subtitles = self.getSubtitles()
  233. self.assertEqual(set(subtitles.keys()), set(['en']))
  234. self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
  235. class TestThePlatformSubtitles(BaseTestSubtitles):
  236. # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
  237. # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
  238. url = 'theplatform:JFUjUE1_ehvq'
  239. IE = ThePlatformIE
  240. def test_allsubtitles(self):
  241. self.DL.params['writesubtitles'] = True
  242. self.DL.params['allsubtitles'] = True
  243. subtitles = self.getSubtitles()
  244. self.assertEqual(set(subtitles.keys()), set(['en']))
  245. self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
  246. class TestRtveSubtitles(BaseTestSubtitles):
  247. url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
  248. IE = RTVEALaCartaIE
  249. def test_allsubtitles(self):
  250. print('Skipping, only available from Spain')
  251. return
  252. self.DL.params['writesubtitles'] = True
  253. self.DL.params['allsubtitles'] = True
  254. subtitles = self.getSubtitles()
  255. self.assertEqual(set(subtitles.keys()), set(['es']))
  256. self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
  257. if __name__ == '__main__':
  258. unittest.main()