test_subtitles.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415
  1. #!/usr/bin/env python3
  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, is_download_test
  9. from hypervideo_dl.extractor import (
  10. YoutubeIE,
  11. DailymotionIE,
  12. TEDIE,
  13. VimeoIE,
  14. WallaIE,
  15. CeskaTelevizeIE,
  16. LyndaIE,
  17. NPOIE,
  18. PBSIE,
  19. ComedyCentralIE,
  20. NRKTVIE,
  21. RaiPlayIE,
  22. VikiIE,
  23. ThePlatformIE,
  24. ThePlatformFeedIE,
  25. RTVEALaCartaIE,
  26. DemocracynowIE,
  27. )
  28. @is_download_test
  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. @is_download_test
  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']), '688dd1ce0981683867e7fe6fde2a224b')
  59. self.assertEqual(md5(subtitles['it']), '31324d30b8430b309f7f5979a504a769')
  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']), 'c97ddf1217390906fa9fbd34901f3da2')
  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']), 'ae1bd34126571a77aabd4d276b28044d')
  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_no_automatic_captions(self):
  79. self.url = 'QRS8MkLhQmM'
  80. self.DL.params['writeautomaticsub'] = True
  81. subtitles = self.getSubtitles()
  82. self.assertTrue(not subtitles)
  83. def test_youtube_translated_subtitles(self):
  84. # This video has a subtitles track, which can be translated
  85. self.url = 'i0ZabxXmH4Y'
  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. @is_download_test
  98. class TestDailymotionSubtitles(BaseTestSubtitles):
  99. url = 'http://www.dailymotion.com/video/xczg00'
  100. IE = DailymotionIE
  101. def test_allsubtitles(self):
  102. self.DL.params['writesubtitles'] = True
  103. self.DL.params['allsubtitles'] = True
  104. subtitles = self.getSubtitles()
  105. self.assertTrue(len(subtitles.keys()) >= 6)
  106. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  107. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  108. for lang in ['es', 'fr', 'de']:
  109. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  110. def test_nosubtitles(self):
  111. self.DL.expect_warning('video doesn\'t have subtitles')
  112. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  113. self.DL.params['writesubtitles'] = True
  114. self.DL.params['allsubtitles'] = True
  115. subtitles = self.getSubtitles()
  116. self.assertFalse(subtitles)
  117. @is_download_test
  118. class TestTedSubtitles(BaseTestSubtitles):
  119. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  120. IE = TEDIE
  121. def test_allsubtitles(self):
  122. self.DL.params['writesubtitles'] = True
  123. self.DL.params['allsubtitles'] = True
  124. subtitles = self.getSubtitles()
  125. self.assertTrue(len(subtitles.keys()) >= 28)
  126. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  127. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  128. for lang in ['es', 'fr', 'de']:
  129. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  130. @is_download_test
  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. @is_download_test
  149. class TestWallaSubtitles(BaseTestSubtitles):
  150. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  151. IE = WallaIE
  152. def test_allsubtitles(self):
  153. self.DL.expect_warning('Automatic Captions not supported by this server')
  154. self.DL.params['writesubtitles'] = True
  155. self.DL.params['allsubtitles'] = True
  156. subtitles = self.getSubtitles()
  157. self.assertEqual(set(subtitles.keys()), set(['heb']))
  158. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  159. def test_nosubtitles(self):
  160. self.DL.expect_warning('video doesn\'t have subtitles')
  161. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  162. self.DL.params['writesubtitles'] = True
  163. self.DL.params['allsubtitles'] = True
  164. subtitles = self.getSubtitles()
  165. self.assertFalse(subtitles)
  166. @is_download_test
  167. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  168. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  169. IE = CeskaTelevizeIE
  170. def test_allsubtitles(self):
  171. self.DL.expect_warning('Automatic Captions not supported by this server')
  172. self.DL.params['writesubtitles'] = True
  173. self.DL.params['allsubtitles'] = True
  174. subtitles = self.getSubtitles()
  175. self.assertEqual(set(subtitles.keys()), set(['cs']))
  176. self.assertTrue(len(subtitles['cs']) > 20000)
  177. def test_nosubtitles(self):
  178. self.DL.expect_warning('video doesn\'t have subtitles')
  179. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  180. self.DL.params['writesubtitles'] = True
  181. self.DL.params['allsubtitles'] = True
  182. subtitles = self.getSubtitles()
  183. self.assertFalse(subtitles)
  184. @is_download_test
  185. class TestLyndaSubtitles(BaseTestSubtitles):
  186. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  187. IE = LyndaIE
  188. def test_allsubtitles(self):
  189. self.DL.params['writesubtitles'] = True
  190. self.DL.params['allsubtitles'] = True
  191. subtitles = self.getSubtitles()
  192. self.assertEqual(set(subtitles.keys()), set(['en']))
  193. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  194. @is_download_test
  195. class TestNPOSubtitles(BaseTestSubtitles):
  196. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  197. IE = NPOIE
  198. def test_allsubtitles(self):
  199. self.DL.params['writesubtitles'] = True
  200. self.DL.params['allsubtitles'] = True
  201. subtitles = self.getSubtitles()
  202. self.assertEqual(set(subtitles.keys()), set(['nl']))
  203. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  204. @is_download_test
  205. class TestMTVSubtitles(BaseTestSubtitles):
  206. url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
  207. IE = ComedyCentralIE
  208. def getInfoDict(self):
  209. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  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(['en']))
  215. self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
  216. @is_download_test
  217. class TestNRKSubtitles(BaseTestSubtitles):
  218. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  219. IE = NRKTVIE
  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(['no']))
  225. self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
  226. @is_download_test
  227. class TestRaiPlaySubtitles(BaseTestSubtitles):
  228. IE = RaiPlayIE
  229. def test_subtitles_key(self):
  230. self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  231. self.DL.params['writesubtitles'] = True
  232. self.DL.params['allsubtitles'] = True
  233. subtitles = self.getSubtitles()
  234. self.assertEqual(set(subtitles.keys()), set(['it']))
  235. self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  236. def test_subtitles_array_key(self):
  237. self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
  238. self.DL.params['writesubtitles'] = True
  239. self.DL.params['allsubtitles'] = True
  240. subtitles = self.getSubtitles()
  241. self.assertEqual(set(subtitles.keys()), set(['it']))
  242. self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
  243. @is_download_test
  244. class TestVikiSubtitles(BaseTestSubtitles):
  245. url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
  246. IE = VikiIE
  247. def test_allsubtitles(self):
  248. self.DL.params['writesubtitles'] = True
  249. self.DL.params['allsubtitles'] = True
  250. subtitles = self.getSubtitles()
  251. self.assertEqual(set(subtitles.keys()), set(['en']))
  252. self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
  253. @is_download_test
  254. class TestThePlatformSubtitles(BaseTestSubtitles):
  255. # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
  256. # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
  257. url = 'theplatform:JFUjUE1_ehvq'
  258. IE = ThePlatformIE
  259. def test_allsubtitles(self):
  260. self.DL.params['writesubtitles'] = True
  261. self.DL.params['allsubtitles'] = True
  262. subtitles = self.getSubtitles()
  263. self.assertEqual(set(subtitles.keys()), set(['en']))
  264. self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
  265. @is_download_test
  266. class TestThePlatformFeedSubtitles(BaseTestSubtitles):
  267. url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
  268. IE = ThePlatformFeedIE
  269. def test_allsubtitles(self):
  270. self.DL.params['writesubtitles'] = True
  271. self.DL.params['allsubtitles'] = True
  272. subtitles = self.getSubtitles()
  273. self.assertEqual(set(subtitles.keys()), set(['en']))
  274. self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
  275. @is_download_test
  276. class TestRtveSubtitles(BaseTestSubtitles):
  277. url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
  278. IE = RTVEALaCartaIE
  279. def test_allsubtitles(self):
  280. print('Skipping, only available from Spain')
  281. return
  282. self.DL.params['writesubtitles'] = True
  283. self.DL.params['allsubtitles'] = True
  284. subtitles = self.getSubtitles()
  285. self.assertEqual(set(subtitles.keys()), set(['es']))
  286. self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
  287. @is_download_test
  288. class TestDemocracynowSubtitles(BaseTestSubtitles):
  289. url = 'http://www.democracynow.org/shows/2015/7/3'
  290. IE = DemocracynowIE
  291. def test_allsubtitles(self):
  292. self.DL.params['writesubtitles'] = True
  293. self.DL.params['allsubtitles'] = True
  294. subtitles = self.getSubtitles()
  295. self.assertEqual(set(subtitles.keys()), set(['en']))
  296. self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
  297. def test_subtitles_in_page(self):
  298. self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
  299. self.DL.params['writesubtitles'] = True
  300. self.DL.params['allsubtitles'] = True
  301. subtitles = self.getSubtitles()
  302. self.assertEqual(set(subtitles.keys()), set(['en']))
  303. self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
  304. @is_download_test
  305. class TestPBSSubtitles(BaseTestSubtitles):
  306. url = 'https://www.pbs.org/video/how-fantasy-reflects-our-world-picecq/'
  307. IE = PBSIE
  308. def test_allsubtitles(self):
  309. self.DL.params['writesubtitles'] = True
  310. self.DL.params['allsubtitles'] = True
  311. subtitles = self.getSubtitles()
  312. self.assertEqual(set(subtitles.keys()), set(['en']))
  313. def test_subtitles_dfxp_format(self):
  314. self.DL.params['writesubtitles'] = True
  315. self.DL.params['subtitlesformat'] = 'dfxp'
  316. subtitles = self.getSubtitles()
  317. self.assertIn(md5(subtitles['en']), ['643b034254cdc3768ff1e750b6b5873b'])
  318. def test_subtitles_vtt_format(self):
  319. self.DL.params['writesubtitles'] = True
  320. self.DL.params['subtitlesformat'] = 'vtt'
  321. subtitles = self.getSubtitles()
  322. self.assertIn(
  323. md5(subtitles['en']), ['937a05711555b165d4c55a9667017045', 'f49ea998d6824d94959c8152a368ff73'])
  324. def test_subtitles_srt_format(self):
  325. self.DL.params['writesubtitles'] = True
  326. self.DL.params['subtitlesformat'] = 'srt'
  327. subtitles = self.getSubtitles()
  328. self.assertIn(md5(subtitles['en']), ['2082c21b43759d9bf172931b2f2ca371'])
  329. def test_subtitles_sami_format(self):
  330. self.DL.params['writesubtitles'] = True
  331. self.DL.params['subtitlesformat'] = 'sami'
  332. subtitles = self.getSubtitles()
  333. self.assertIn(md5(subtitles['en']), ['4256b16ac7da6a6780fafd04294e85cd'])
  334. if __name__ == '__main__':
  335. unittest.main()