test_subtitles.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371
  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. ThePlatformFeedIE,
  25. RTVEALaCartaIE,
  26. FunnyOrDieIE,
  27. DemocracynowIE,
  28. )
  29. class BaseTestSubtitles(unittest.TestCase):
  30. url = None
  31. IE = None
  32. def setUp(self):
  33. self.DL = FakeYDL()
  34. self.ie = self.IE()
  35. self.DL.add_info_extractor(self.ie)
  36. def getInfoDict(self):
  37. info_dict = self.DL.extract_info(self.url, download=False)
  38. return info_dict
  39. def getSubtitles(self):
  40. info_dict = self.getInfoDict()
  41. subtitles = info_dict['requested_subtitles']
  42. if not subtitles:
  43. return subtitles
  44. for sub_info in subtitles.values():
  45. if sub_info.get('data') is None:
  46. uf = self.DL.urlopen(sub_info['url'])
  47. sub_info['data'] = uf.read().decode('utf-8')
  48. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  49. class TestYoutubeSubtitles(BaseTestSubtitles):
  50. url = 'QRS8MkLhQmM'
  51. IE = YoutubeIE
  52. def test_youtube_allsubtitles(self):
  53. self.DL.params['writesubtitles'] = True
  54. self.DL.params['allsubtitles'] = True
  55. subtitles = self.getSubtitles()
  56. self.assertEqual(len(subtitles.keys()), 13)
  57. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  58. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  59. for lang in ['it', 'fr', 'de']:
  60. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  61. def test_youtube_subtitles_sbv_format(self):
  62. self.DL.params['writesubtitles'] = True
  63. self.DL.params['subtitlesformat'] = 'sbv'
  64. subtitles = self.getSubtitles()
  65. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  66. def test_youtube_subtitles_vtt_format(self):
  67. self.DL.params['writesubtitles'] = True
  68. self.DL.params['subtitlesformat'] = 'vtt'
  69. subtitles = self.getSubtitles()
  70. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  71. def test_youtube_automatic_captions(self):
  72. self.url = '8YoUxe5ncPo'
  73. self.DL.params['writeautomaticsub'] = True
  74. self.DL.params['subtitleslangs'] = ['it']
  75. subtitles = self.getSubtitles()
  76. self.assertTrue(subtitles['it'] is not None)
  77. def test_youtube_translated_subtitles(self):
  78. # This video has a subtitles track, which can be translated
  79. self.url = 'Ky9eprVWzlI'
  80. self.DL.params['writeautomaticsub'] = True
  81. self.DL.params['subtitleslangs'] = ['it']
  82. subtitles = self.getSubtitles()
  83. self.assertTrue(subtitles['it'] is not None)
  84. def test_youtube_nosubtitles(self):
  85. self.DL.expect_warning('video doesn\'t have subtitles')
  86. self.url = 'n5BB19UTcdA'
  87. self.DL.params['writesubtitles'] = True
  88. self.DL.params['allsubtitles'] = True
  89. subtitles = self.getSubtitles()
  90. self.assertFalse(subtitles)
  91. class TestDailymotionSubtitles(BaseTestSubtitles):
  92. url = 'http://www.dailymotion.com/video/xczg00'
  93. IE = DailymotionIE
  94. def test_allsubtitles(self):
  95. self.DL.params['writesubtitles'] = True
  96. self.DL.params['allsubtitles'] = True
  97. subtitles = self.getSubtitles()
  98. self.assertTrue(len(subtitles.keys()) >= 6)
  99. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  100. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  101. for lang in ['es', 'fr', 'de']:
  102. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  103. def test_nosubtitles(self):
  104. self.DL.expect_warning('video doesn\'t have subtitles')
  105. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  106. self.DL.params['writesubtitles'] = True
  107. self.DL.params['allsubtitles'] = True
  108. subtitles = self.getSubtitles()
  109. self.assertFalse(subtitles)
  110. class TestTedSubtitles(BaseTestSubtitles):
  111. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  112. IE = TEDIE
  113. def test_allsubtitles(self):
  114. self.DL.params['writesubtitles'] = True
  115. self.DL.params['allsubtitles'] = True
  116. subtitles = self.getSubtitles()
  117. self.assertTrue(len(subtitles.keys()) >= 28)
  118. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  119. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  120. for lang in ['es', 'fr', 'de']:
  121. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  122. class TestBlipTVSubtitles(BaseTestSubtitles):
  123. url = 'http://blip.tv/a/a-6603250'
  124. IE = BlipTVIE
  125. def test_allsubtitles(self):
  126. self.DL.params['writesubtitles'] = True
  127. self.DL.params['allsubtitles'] = True
  128. subtitles = self.getSubtitles()
  129. self.assertEqual(set(subtitles.keys()), set(['en']))
  130. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  131. class TestVimeoSubtitles(BaseTestSubtitles):
  132. url = 'http://vimeo.com/76979871'
  133. IE = VimeoIE
  134. def test_allsubtitles(self):
  135. self.DL.params['writesubtitles'] = True
  136. self.DL.params['allsubtitles'] = True
  137. subtitles = self.getSubtitles()
  138. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  139. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  140. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  141. def test_nosubtitles(self):
  142. self.DL.expect_warning('video doesn\'t have subtitles')
  143. self.url = 'http://vimeo.com/56015672'
  144. self.DL.params['writesubtitles'] = True
  145. self.DL.params['allsubtitles'] = True
  146. subtitles = self.getSubtitles()
  147. self.assertFalse(subtitles)
  148. class TestWallaSubtitles(BaseTestSubtitles):
  149. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  150. IE = WallaIE
  151. def test_allsubtitles(self):
  152. self.DL.expect_warning('Automatic Captions not supported by this server')
  153. self.DL.params['writesubtitles'] = True
  154. self.DL.params['allsubtitles'] = True
  155. subtitles = self.getSubtitles()
  156. self.assertEqual(set(subtitles.keys()), set(['heb']))
  157. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  158. def test_nosubtitles(self):
  159. self.DL.expect_warning('video doesn\'t have subtitles')
  160. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  161. self.DL.params['writesubtitles'] = True
  162. self.DL.params['allsubtitles'] = True
  163. subtitles = self.getSubtitles()
  164. self.assertFalse(subtitles)
  165. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  166. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  167. IE = CeskaTelevizeIE
  168. def test_allsubtitles(self):
  169. self.DL.expect_warning('Automatic Captions not supported by this server')
  170. self.DL.params['writesubtitles'] = True
  171. self.DL.params['allsubtitles'] = True
  172. subtitles = self.getSubtitles()
  173. self.assertEqual(set(subtitles.keys()), set(['cs']))
  174. self.assertTrue(len(subtitles['cs']) > 20000)
  175. def test_nosubtitles(self):
  176. self.DL.expect_warning('video doesn\'t have subtitles')
  177. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  178. self.DL.params['writesubtitles'] = True
  179. self.DL.params['allsubtitles'] = True
  180. subtitles = self.getSubtitles()
  181. self.assertFalse(subtitles)
  182. class TestLyndaSubtitles(BaseTestSubtitles):
  183. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  184. IE = LyndaIE
  185. def test_allsubtitles(self):
  186. self.DL.params['writesubtitles'] = True
  187. self.DL.params['allsubtitles'] = True
  188. subtitles = self.getSubtitles()
  189. self.assertEqual(set(subtitles.keys()), set(['en']))
  190. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  191. class TestNPOSubtitles(BaseTestSubtitles):
  192. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  193. IE = NPOIE
  194. def test_allsubtitles(self):
  195. self.DL.params['writesubtitles'] = True
  196. self.DL.params['allsubtitles'] = True
  197. subtitles = self.getSubtitles()
  198. self.assertEqual(set(subtitles.keys()), set(['nl']))
  199. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  200. class TestMTVSubtitles(BaseTestSubtitles):
  201. url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
  202. IE = ComedyCentralIE
  203. def getInfoDict(self):
  204. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  205. def test_allsubtitles(self):
  206. self.DL.params['writesubtitles'] = True
  207. self.DL.params['allsubtitles'] = True
  208. subtitles = self.getSubtitles()
  209. self.assertEqual(set(subtitles.keys()), set(['en']))
  210. self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
  211. class TestNRKSubtitles(BaseTestSubtitles):
  212. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  213. IE = NRKTVIE
  214. def test_allsubtitles(self):
  215. self.DL.params['writesubtitles'] = True
  216. self.DL.params['allsubtitles'] = True
  217. subtitles = self.getSubtitles()
  218. self.assertEqual(set(subtitles.keys()), set(['no']))
  219. self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
  220. class TestRaiSubtitles(BaseTestSubtitles):
  221. url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  222. IE = RaiIE
  223. def test_allsubtitles(self):
  224. self.DL.params['writesubtitles'] = True
  225. self.DL.params['allsubtitles'] = True
  226. subtitles = self.getSubtitles()
  227. self.assertEqual(set(subtitles.keys()), set(['it']))
  228. self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  229. class TestVikiSubtitles(BaseTestSubtitles):
  230. url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
  231. IE = VikiIE
  232. def test_allsubtitles(self):
  233. self.DL.params['writesubtitles'] = True
  234. self.DL.params['allsubtitles'] = True
  235. subtitles = self.getSubtitles()
  236. self.assertEqual(set(subtitles.keys()), set(['en']))
  237. self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
  238. class TestThePlatformSubtitles(BaseTestSubtitles):
  239. # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
  240. # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
  241. url = 'theplatform:JFUjUE1_ehvq'
  242. IE = ThePlatformIE
  243. def test_allsubtitles(self):
  244. self.DL.params['writesubtitles'] = True
  245. self.DL.params['allsubtitles'] = True
  246. subtitles = self.getSubtitles()
  247. self.assertEqual(set(subtitles.keys()), set(['en']))
  248. self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
  249. class TestThePlatformFeedSubtitles(BaseTestSubtitles):
  250. url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
  251. IE = ThePlatformFeedIE
  252. def test_allsubtitles(self):
  253. self.DL.params['writesubtitles'] = True
  254. self.DL.params['allsubtitles'] = True
  255. subtitles = self.getSubtitles()
  256. self.assertEqual(set(subtitles.keys()), set(['en']))
  257. self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
  258. class TestRtveSubtitles(BaseTestSubtitles):
  259. url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
  260. IE = RTVEALaCartaIE
  261. def test_allsubtitles(self):
  262. print('Skipping, only available from Spain')
  263. return
  264. self.DL.params['writesubtitles'] = True
  265. self.DL.params['allsubtitles'] = True
  266. subtitles = self.getSubtitles()
  267. self.assertEqual(set(subtitles.keys()), set(['es']))
  268. self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
  269. class TestFunnyOrDieSubtitles(BaseTestSubtitles):
  270. url = 'http://www.funnyordie.com/videos/224829ff6d/judd-apatow-will-direct-your-vine'
  271. IE = FunnyOrDieIE
  272. def test_allsubtitles(self):
  273. self.DL.params['writesubtitles'] = True
  274. self.DL.params['allsubtitles'] = True
  275. subtitles = self.getSubtitles()
  276. self.assertEqual(set(subtitles.keys()), set(['en']))
  277. self.assertEqual(md5(subtitles['en']), 'c5593c193eacd353596c11c2d4f9ecc4')
  278. class TestDemocracynowSubtitles(BaseTestSubtitles):
  279. url = 'http://www.democracynow.org/shows/2015/7/3'
  280. IE = DemocracynowIE
  281. def test_allsubtitles(self):
  282. self.DL.params['writesubtitles'] = True
  283. self.DL.params['allsubtitles'] = True
  284. subtitles = self.getSubtitles()
  285. self.assertEqual(set(subtitles.keys()), set(['en']))
  286. self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
  287. def test_subtitles_in_page(self):
  288. self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
  289. self.DL.params['writesubtitles'] = True
  290. self.DL.params['allsubtitles'] = True
  291. subtitles = self.getSubtitles()
  292. self.assertEqual(set(subtitles.keys()), set(['en']))
  293. self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
  294. if __name__ == '__main__':
  295. unittest.main()