test_subtitles.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356
  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. YoutubeIE,
  11. DailymotionIE,
  12. TEDIE,
  13. VimeoIE,
  14. WallaIE,
  15. CeskaTelevizeIE,
  16. LyndaIE,
  17. NPOIE,
  18. ComedyCentralIE,
  19. NRKTVIE,
  20. RaiPlayIE,
  21. VikiIE,
  22. ThePlatformIE,
  23. ThePlatformFeedIE,
  24. RTVEALaCartaIE,
  25. DemocracynowIE,
  26. )
  27. class BaseTestSubtitles(unittest.TestCase):
  28. url = None
  29. IE = None
  30. def setUp(self):
  31. self.DL = FakeYDL()
  32. self.ie = self.IE()
  33. self.DL.add_info_extractor(self.ie)
  34. if not self.IE.working():
  35. print('Skipping: %s marked as not _WORKING' % self.IE.ie_key())
  36. self.skipTest('IE marked as not _WORKING')
  37. def getInfoDict(self):
  38. info_dict = self.DL.extract_info(self.url, download=False)
  39. return info_dict
  40. def getSubtitles(self):
  41. info_dict = self.getInfoDict()
  42. subtitles = info_dict['requested_subtitles']
  43. if not subtitles:
  44. return subtitles
  45. for sub_info in subtitles.values():
  46. if sub_info.get('data') is None:
  47. uf = self.DL.urlopen(sub_info['url'])
  48. sub_info['data'] = uf.read().decode('utf-8')
  49. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  50. class TestYoutubeSubtitles(BaseTestSubtitles):
  51. url = 'QRS8MkLhQmM'
  52. IE = YoutubeIE
  53. def test_youtube_allsubtitles(self):
  54. self.DL.params['writesubtitles'] = True
  55. self.DL.params['allsubtitles'] = True
  56. subtitles = self.getSubtitles()
  57. self.assertEqual(len(subtitles.keys()), 13)
  58. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  59. self.assertEqual(md5(subtitles['it']), '6d752b98c31f1cf8d597050c7a2cb4b5')
  60. for lang in ['fr', 'de']:
  61. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  62. def test_youtube_subtitles_ttml_format(self):
  63. self.DL.params['writesubtitles'] = True
  64. self.DL.params['subtitlesformat'] = 'ttml'
  65. subtitles = self.getSubtitles()
  66. self.assertEqual(md5(subtitles['en']), 'e306f8c42842f723447d9f63ad65df54')
  67. def test_youtube_subtitles_vtt_format(self):
  68. self.DL.params['writesubtitles'] = True
  69. self.DL.params['subtitlesformat'] = 'vtt'
  70. subtitles = self.getSubtitles()
  71. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  72. def test_youtube_automatic_captions(self):
  73. self.url = '8YoUxe5ncPo'
  74. self.DL.params['writeautomaticsub'] = True
  75. self.DL.params['subtitleslangs'] = ['it']
  76. subtitles = self.getSubtitles()
  77. self.assertTrue(subtitles['it'] is not None)
  78. def test_youtube_translated_subtitles(self):
  79. # This video has a subtitles track, which can be translated
  80. self.url = 'Ky9eprVWzlI'
  81. self.DL.params['writeautomaticsub'] = True
  82. self.DL.params['subtitleslangs'] = ['it']
  83. subtitles = self.getSubtitles()
  84. self.assertTrue(subtitles['it'] is not None)
  85. def test_youtube_nosubtitles(self):
  86. self.DL.expect_warning('video doesn\'t have subtitles')
  87. self.url = 'n5BB19UTcdA'
  88. self.DL.params['writesubtitles'] = True
  89. self.DL.params['allsubtitles'] = True
  90. subtitles = self.getSubtitles()
  91. self.assertFalse(subtitles)
  92. class TestDailymotionSubtitles(BaseTestSubtitles):
  93. url = 'http://www.dailymotion.com/video/xczg00'
  94. IE = DailymotionIE
  95. def test_allsubtitles(self):
  96. self.DL.params['writesubtitles'] = True
  97. self.DL.params['allsubtitles'] = True
  98. subtitles = self.getSubtitles()
  99. self.assertTrue(len(subtitles.keys()) >= 6)
  100. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  101. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  102. for lang in ['es', 'fr', 'de']:
  103. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  104. def test_nosubtitles(self):
  105. self.DL.expect_warning('video doesn\'t have subtitles')
  106. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  107. self.DL.params['writesubtitles'] = True
  108. self.DL.params['allsubtitles'] = True
  109. subtitles = self.getSubtitles()
  110. self.assertFalse(subtitles)
  111. class TestTedSubtitles(BaseTestSubtitles):
  112. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  113. IE = TEDIE
  114. def test_allsubtitles(self):
  115. self.DL.params['writesubtitles'] = True
  116. self.DL.params['allsubtitles'] = True
  117. subtitles = self.getSubtitles()
  118. self.assertTrue(len(subtitles.keys()) >= 28)
  119. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  120. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  121. for lang in ['es', 'fr', 'de']:
  122. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  123. class TestVimeoSubtitles(BaseTestSubtitles):
  124. url = 'http://vimeo.com/76979871'
  125. IE = VimeoIE
  126. def test_allsubtitles(self):
  127. self.DL.params['writesubtitles'] = True
  128. self.DL.params['allsubtitles'] = True
  129. subtitles = self.getSubtitles()
  130. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  131. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  132. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  133. def test_nosubtitles(self):
  134. self.DL.expect_warning('video doesn\'t have subtitles')
  135. self.url = 'http://vimeo.com/56015672'
  136. self.DL.params['writesubtitles'] = True
  137. self.DL.params['allsubtitles'] = True
  138. subtitles = self.getSubtitles()
  139. self.assertFalse(subtitles)
  140. class TestWallaSubtitles(BaseTestSubtitles):
  141. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  142. IE = WallaIE
  143. def test_allsubtitles(self):
  144. self.DL.expect_warning('Automatic Captions not supported by this server')
  145. self.DL.params['writesubtitles'] = True
  146. self.DL.params['allsubtitles'] = True
  147. subtitles = self.getSubtitles()
  148. self.assertEqual(set(subtitles.keys()), set(['heb']))
  149. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  150. def test_nosubtitles(self):
  151. self.DL.expect_warning('video doesn\'t have subtitles')
  152. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  153. self.DL.params['writesubtitles'] = True
  154. self.DL.params['allsubtitles'] = True
  155. subtitles = self.getSubtitles()
  156. self.assertFalse(subtitles)
  157. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  158. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  159. IE = CeskaTelevizeIE
  160. def test_allsubtitles(self):
  161. self.DL.expect_warning('Automatic Captions not supported by this server')
  162. self.DL.params['writesubtitles'] = True
  163. self.DL.params['allsubtitles'] = True
  164. subtitles = self.getSubtitles()
  165. self.assertEqual(set(subtitles.keys()), set(['cs']))
  166. self.assertTrue(len(subtitles['cs']) > 20000)
  167. def test_nosubtitles(self):
  168. self.DL.expect_warning('video doesn\'t have subtitles')
  169. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  170. self.DL.params['writesubtitles'] = True
  171. self.DL.params['allsubtitles'] = True
  172. subtitles = self.getSubtitles()
  173. self.assertFalse(subtitles)
  174. class TestLyndaSubtitles(BaseTestSubtitles):
  175. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  176. IE = LyndaIE
  177. def test_allsubtitles(self):
  178. self.DL.params['writesubtitles'] = True
  179. self.DL.params['allsubtitles'] = True
  180. subtitles = self.getSubtitles()
  181. self.assertEqual(set(subtitles.keys()), set(['en']))
  182. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  183. class TestNPOSubtitles(BaseTestSubtitles):
  184. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  185. IE = NPOIE
  186. def test_allsubtitles(self):
  187. self.DL.params['writesubtitles'] = True
  188. self.DL.params['allsubtitles'] = True
  189. subtitles = self.getSubtitles()
  190. self.assertEqual(set(subtitles.keys()), set(['nl']))
  191. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  192. class TestMTVSubtitles(BaseTestSubtitles):
  193. url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
  194. IE = ComedyCentralIE
  195. def getInfoDict(self):
  196. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  197. def test_allsubtitles(self):
  198. self.DL.params['writesubtitles'] = True
  199. self.DL.params['allsubtitles'] = True
  200. subtitles = self.getSubtitles()
  201. self.assertEqual(set(subtitles.keys()), set(['en']))
  202. self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
  203. class TestNRKSubtitles(BaseTestSubtitles):
  204. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  205. IE = NRKTVIE
  206. def test_allsubtitles(self):
  207. self.DL.params['writesubtitles'] = True
  208. self.DL.params['allsubtitles'] = True
  209. subtitles = self.getSubtitles()
  210. self.assertEqual(set(subtitles.keys()), set(['no']))
  211. self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
  212. class TestRaiPlaySubtitles(BaseTestSubtitles):
  213. IE = RaiPlayIE
  214. def test_subtitles_key(self):
  215. self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  216. self.DL.params['writesubtitles'] = True
  217. self.DL.params['allsubtitles'] = True
  218. subtitles = self.getSubtitles()
  219. self.assertEqual(set(subtitles.keys()), set(['it']))
  220. self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  221. def test_subtitles_array_key(self):
  222. self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
  223. self.DL.params['writesubtitles'] = True
  224. self.DL.params['allsubtitles'] = True
  225. subtitles = self.getSubtitles()
  226. self.assertEqual(set(subtitles.keys()), set(['it']))
  227. self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
  228. class TestVikiSubtitles(BaseTestSubtitles):
  229. url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
  230. IE = VikiIE
  231. def test_allsubtitles(self):
  232. self.DL.params['writesubtitles'] = True
  233. self.DL.params['allsubtitles'] = True
  234. subtitles = self.getSubtitles()
  235. self.assertEqual(set(subtitles.keys()), set(['en']))
  236. self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
  237. class TestThePlatformSubtitles(BaseTestSubtitles):
  238. # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
  239. # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
  240. url = 'theplatform:JFUjUE1_ehvq'
  241. IE = ThePlatformIE
  242. def test_allsubtitles(self):
  243. self.DL.params['writesubtitles'] = True
  244. self.DL.params['allsubtitles'] = True
  245. subtitles = self.getSubtitles()
  246. self.assertEqual(set(subtitles.keys()), set(['en']))
  247. self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
  248. class TestThePlatformFeedSubtitles(BaseTestSubtitles):
  249. url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
  250. IE = ThePlatformFeedIE
  251. def test_allsubtitles(self):
  252. self.DL.params['writesubtitles'] = True
  253. self.DL.params['allsubtitles'] = True
  254. subtitles = self.getSubtitles()
  255. self.assertEqual(set(subtitles.keys()), set(['en']))
  256. self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
  257. class TestRtveSubtitles(BaseTestSubtitles):
  258. url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
  259. IE = RTVEALaCartaIE
  260. def test_allsubtitles(self):
  261. print('Skipping, only available from Spain')
  262. return
  263. self.DL.params['writesubtitles'] = True
  264. self.DL.params['allsubtitles'] = True
  265. subtitles = self.getSubtitles()
  266. self.assertEqual(set(subtitles.keys()), set(['es']))
  267. self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
  268. class TestDemocracynowSubtitles(BaseTestSubtitles):
  269. url = 'http://www.democracynow.org/shows/2015/7/3'
  270. IE = DemocracynowIE
  271. def test_allsubtitles(self):
  272. self.DL.params['writesubtitles'] = True
  273. self.DL.params['allsubtitles'] = True
  274. subtitles = self.getSubtitles()
  275. self.assertEqual(set(subtitles.keys()), set(['en']))
  276. self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
  277. def test_subtitles_in_page(self):
  278. self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
  279. self.DL.params['writesubtitles'] = True
  280. self.DL.params['allsubtitles'] = True
  281. subtitles = self.getSubtitles()
  282. self.assertEqual(set(subtitles.keys()), set(['en']))
  283. self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
  284. if __name__ == '__main__':
  285. unittest.main()