test_subtitles.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404
  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_subtitles(self):
  49. self.DL.params['writesubtitles'] = True
  50. subtitles = self.getSubtitles()
  51. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  52. def test_youtube_subtitles_lang(self):
  53. self.DL.params['writesubtitles'] = True
  54. self.DL.params['subtitleslangs'] = ['it']
  55. subtitles = self.getSubtitles()
  56. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  57. def test_youtube_allsubtitles(self):
  58. self.DL.params['writesubtitles'] = True
  59. self.DL.params['allsubtitles'] = True
  60. subtitles = self.getSubtitles()
  61. self.assertEqual(len(subtitles.keys()), 13)
  62. def test_youtube_subtitles_sbv_format(self):
  63. self.DL.params['writesubtitles'] = True
  64. self.DL.params['subtitlesformat'] = 'sbv'
  65. subtitles = self.getSubtitles()
  66. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  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_list_subtitles(self):
  73. self.DL.expect_warning('Video doesn\'t have automatic captions')
  74. self.DL.params['listsubtitles'] = True
  75. info_dict = self.getInfoDict()
  76. self.assertEqual(info_dict, None)
  77. def test_youtube_automatic_captions(self):
  78. self.url = '8YoUxe5ncPo'
  79. self.DL.params['writeautomaticsub'] = True
  80. self.DL.params['subtitleslangs'] = ['it']
  81. subtitles = self.getSubtitles()
  82. self.assertTrue(subtitles['it'] is not None)
  83. def test_youtube_translated_subtitles(self):
  84. # This video has a subtitles track, which can be translated
  85. self.url = 'Ky9eprVWzlI'
  86. self.DL.params['writeautomaticsub'] = True
  87. self.DL.params['subtitleslangs'] = ['it']
  88. subtitles = self.getSubtitles()
  89. self.assertTrue(subtitles['it'] is not None)
  90. def test_youtube_nosubtitles(self):
  91. self.DL.expect_warning('video doesn\'t have subtitles')
  92. self.url = 'n5BB19UTcdA'
  93. self.DL.params['writesubtitles'] = True
  94. self.DL.params['allsubtitles'] = True
  95. subtitles = self.getSubtitles()
  96. self.assertFalse(subtitles)
  97. def test_youtube_multiple_langs(self):
  98. self.url = 'QRS8MkLhQmM'
  99. self.DL.params['writesubtitles'] = True
  100. langs = ['it', 'fr', 'de']
  101. self.DL.params['subtitleslangs'] = langs
  102. subtitles = self.getSubtitles()
  103. for lang in langs:
  104. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  105. class TestDailymotionSubtitles(BaseTestSubtitles):
  106. url = 'http://www.dailymotion.com/video/xczg00'
  107. IE = DailymotionIE
  108. def test_subtitles(self):
  109. self.DL.params['writesubtitles'] = True
  110. subtitles = self.getSubtitles()
  111. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  112. def test_subtitles_lang(self):
  113. self.DL.params['writesubtitles'] = True
  114. self.DL.params['subtitleslangs'] = ['fr']
  115. subtitles = self.getSubtitles()
  116. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  117. def test_allsubtitles(self):
  118. self.DL.params['writesubtitles'] = True
  119. self.DL.params['allsubtitles'] = True
  120. subtitles = self.getSubtitles()
  121. self.assertTrue(len(subtitles.keys()) >= 6)
  122. def test_nosubtitles(self):
  123. self.DL.expect_warning('video doesn\'t have subtitles')
  124. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  125. self.DL.params['writesubtitles'] = True
  126. self.DL.params['allsubtitles'] = True
  127. subtitles = self.getSubtitles()
  128. self.assertFalse(subtitles)
  129. def test_multiple_langs(self):
  130. self.DL.params['writesubtitles'] = True
  131. langs = ['es', 'fr', 'de']
  132. self.DL.params['subtitleslangs'] = langs
  133. subtitles = self.getSubtitles()
  134. for lang in langs:
  135. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  136. class TestTedSubtitles(BaseTestSubtitles):
  137. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  138. IE = TEDIE
  139. def test_no_writesubtitles(self):
  140. subtitles = self.getSubtitles()
  141. self.assertFalse(subtitles)
  142. def test_subtitles(self):
  143. self.DL.params['writesubtitles'] = True
  144. subtitles = self.getSubtitles()
  145. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  146. def test_subtitles_lang(self):
  147. self.DL.params['writesubtitles'] = True
  148. self.DL.params['subtitleslangs'] = ['fr']
  149. subtitles = self.getSubtitles()
  150. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  151. def test_allsubtitles(self):
  152. self.DL.params['writesubtitles'] = True
  153. self.DL.params['allsubtitles'] = True
  154. subtitles = self.getSubtitles()
  155. self.assertTrue(len(subtitles.keys()) >= 28)
  156. def test_list_subtitles(self):
  157. self.DL.params['listsubtitles'] = True
  158. info_dict = self.getInfoDict()
  159. self.assertEqual(info_dict, None)
  160. def test_multiple_langs(self):
  161. self.DL.params['writesubtitles'] = True
  162. langs = ['es', 'fr', 'de']
  163. self.DL.params['subtitleslangs'] = langs
  164. subtitles = self.getSubtitles()
  165. for lang in langs:
  166. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  167. class TestBlipTVSubtitles(BaseTestSubtitles):
  168. url = 'http://blip.tv/a/a-6603250'
  169. IE = BlipTVIE
  170. def test_list_subtitles(self):
  171. self.DL.params['listsubtitles'] = True
  172. info_dict = self.getInfoDict()
  173. self.assertEqual(info_dict, None)
  174. def test_allsubtitles(self):
  175. self.DL.params['writesubtitles'] = True
  176. self.DL.params['allsubtitles'] = True
  177. subtitles = self.getSubtitles()
  178. self.assertEqual(set(subtitles.keys()), set(['en']))
  179. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  180. class TestVimeoSubtitles(BaseTestSubtitles):
  181. url = 'http://vimeo.com/76979871'
  182. IE = VimeoIE
  183. def test_subtitles(self):
  184. self.DL.params['writesubtitles'] = True
  185. subtitles = self.getSubtitles()
  186. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  187. def test_subtitles_lang(self):
  188. self.DL.params['writesubtitles'] = True
  189. self.DL.params['subtitleslangs'] = ['fr']
  190. subtitles = self.getSubtitles()
  191. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  192. def test_allsubtitles(self):
  193. self.DL.params['writesubtitles'] = True
  194. self.DL.params['allsubtitles'] = True
  195. subtitles = self.getSubtitles()
  196. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  197. def test_nosubtitles(self):
  198. self.DL.expect_warning('video doesn\'t have subtitles')
  199. self.url = 'http://vimeo.com/56015672'
  200. self.DL.params['writesubtitles'] = True
  201. self.DL.params['allsubtitles'] = True
  202. subtitles = self.getSubtitles()
  203. self.assertFalse(subtitles)
  204. def test_multiple_langs(self):
  205. self.DL.params['writesubtitles'] = True
  206. langs = ['es', 'fr', 'de']
  207. self.DL.params['subtitleslangs'] = langs
  208. subtitles = self.getSubtitles()
  209. for lang in langs:
  210. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  211. class TestWallaSubtitles(BaseTestSubtitles):
  212. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  213. IE = WallaIE
  214. def test_list_subtitles(self):
  215. self.DL.expect_warning('Automatic Captions not supported by this server')
  216. self.DL.params['listsubtitles'] = True
  217. info_dict = self.getInfoDict()
  218. self.assertEqual(info_dict, None)
  219. def test_allsubtitles(self):
  220. self.DL.expect_warning('Automatic Captions not supported by this server')
  221. self.DL.params['writesubtitles'] = True
  222. self.DL.params['allsubtitles'] = True
  223. subtitles = self.getSubtitles()
  224. self.assertEqual(set(subtitles.keys()), set(['heb']))
  225. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  226. def test_nosubtitles(self):
  227. self.DL.expect_warning('video doesn\'t have subtitles')
  228. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  229. self.DL.params['writesubtitles'] = True
  230. self.DL.params['allsubtitles'] = True
  231. subtitles = self.getSubtitles()
  232. self.assertFalse(subtitles)
  233. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  234. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  235. IE = CeskaTelevizeIE
  236. def test_list_subtitles(self):
  237. self.DL.expect_warning('Automatic Captions not supported by this server')
  238. self.DL.params['listsubtitles'] = True
  239. info_dict = self.getInfoDict()
  240. self.assertEqual(info_dict, None)
  241. def test_allsubtitles(self):
  242. self.DL.expect_warning('Automatic Captions not supported by this server')
  243. self.DL.params['writesubtitles'] = True
  244. self.DL.params['allsubtitles'] = True
  245. subtitles = self.getSubtitles()
  246. self.assertEqual(set(subtitles.keys()), set(['cs']))
  247. self.assertTrue(len(subtitles['cs']) > 20000)
  248. def test_nosubtitles(self):
  249. self.DL.expect_warning('video doesn\'t have subtitles')
  250. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  251. self.DL.params['writesubtitles'] = True
  252. self.DL.params['allsubtitles'] = True
  253. subtitles = self.getSubtitles()
  254. self.assertFalse(subtitles)
  255. class TestLyndaSubtitles(BaseTestSubtitles):
  256. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  257. IE = LyndaIE
  258. def test_allsubtitles(self):
  259. self.DL.params['writesubtitles'] = True
  260. self.DL.params['allsubtitles'] = True
  261. subtitles = self.getSubtitles()
  262. self.assertEqual(set(subtitles.keys()), set(['en']))
  263. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  264. class TestNPOSubtitles(BaseTestSubtitles):
  265. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  266. IE = NPOIE
  267. def test_allsubtitles(self):
  268. self.DL.params['writesubtitles'] = True
  269. self.DL.params['allsubtitles'] = True
  270. subtitles = self.getSubtitles()
  271. self.assertEqual(set(subtitles.keys()), set(['nl']))
  272. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  273. class TestMTVSubtitles(BaseTestSubtitles):
  274. url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
  275. IE = ComedyCentralIE
  276. def getInfoDict(self):
  277. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  278. def test_allsubtitles(self):
  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']), 'b9f6ca22a6acf597ec76f61749765e65')
  284. class TestNRKSubtitles(BaseTestSubtitles):
  285. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  286. IE = NRKTVIE
  287. def test_allsubtitles(self):
  288. self.DL.params['writesubtitles'] = True
  289. self.DL.params['allsubtitles'] = True
  290. subtitles = self.getSubtitles()
  291. self.assertEqual(set(subtitles.keys()), set(['no']))
  292. self.assertEqual(md5(subtitles['no']), '1d221e6458c95c5494dcd38e6a1f129a')
  293. class TestRaiSubtitles(BaseTestSubtitles):
  294. url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  295. IE = RaiIE
  296. def test_allsubtitles(self):
  297. self.DL.params['writesubtitles'] = True
  298. self.DL.params['allsubtitles'] = True
  299. subtitles = self.getSubtitles()
  300. self.assertEqual(set(subtitles.keys()), set(['it']))
  301. self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  302. class TestVikiSubtitles(BaseTestSubtitles):
  303. url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
  304. IE = VikiIE
  305. def test_allsubtitles(self):
  306. self.DL.params['writesubtitles'] = True
  307. self.DL.params['allsubtitles'] = True
  308. subtitles = self.getSubtitles()
  309. self.assertEqual(set(subtitles.keys()), set(['en']))
  310. self.assertEqual(md5(subtitles['en']), 'b0b781eeb45efd3f6398a925b259150b')
  311. class TestThePlatformSubtitles(BaseTestSubtitles):
  312. # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
  313. # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
  314. url = 'theplatform:JFUjUE1_ehvq'
  315. IE = ThePlatformIE
  316. def test_allsubtitles(self):
  317. self.DL.params['writesubtitles'] = True
  318. self.DL.params['allsubtitles'] = True
  319. subtitles = self.getSubtitles()
  320. self.assertEqual(set(subtitles.keys()), set(['en']))
  321. self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
  322. if __name__ == '__main__':
  323. unittest.main()