test_subtitles.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389
  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. )
  24. class BaseTestSubtitles(unittest.TestCase):
  25. url = None
  26. IE = None
  27. def setUp(self):
  28. self.DL = FakeYDL()
  29. self.ie = self.IE()
  30. self.DL.add_info_extractor(self.ie)
  31. def getInfoDict(self):
  32. info_dict = self.DL.extract_info(self.url, download=False)
  33. return info_dict
  34. def getSubtitles(self):
  35. info_dict = self.getInfoDict()
  36. subtitles = info_dict['requested_subtitles']
  37. if not subtitles:
  38. return subtitles
  39. for sub_info in subtitles.values():
  40. if sub_info.get('data') is None:
  41. uf = self.DL.urlopen(sub_info['url'])
  42. sub_info['data'] = uf.read().decode('utf-8')
  43. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  44. class TestYoutubeSubtitles(BaseTestSubtitles):
  45. url = 'QRS8MkLhQmM'
  46. IE = YoutubeIE
  47. def test_youtube_subtitles(self):
  48. self.DL.params['writesubtitles'] = True
  49. subtitles = self.getSubtitles()
  50. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  51. def test_youtube_subtitles_lang(self):
  52. self.DL.params['writesubtitles'] = True
  53. self.DL.params['subtitleslangs'] = ['it']
  54. subtitles = self.getSubtitles()
  55. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  56. def test_youtube_allsubtitles(self):
  57. self.DL.params['writesubtitles'] = True
  58. self.DL.params['allsubtitles'] = True
  59. subtitles = self.getSubtitles()
  60. self.assertEqual(len(subtitles.keys()), 13)
  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_list_subtitles(self):
  72. self.DL.expect_warning('Video doesn\'t have automatic captions')
  73. self.DL.params['listsubtitles'] = True
  74. info_dict = self.getInfoDict()
  75. self.assertEqual(info_dict, None)
  76. def test_youtube_automatic_captions(self):
  77. self.url = '8YoUxe5ncPo'
  78. self.DL.params['writeautomaticsub'] = True
  79. self.DL.params['subtitleslangs'] = ['it']
  80. subtitles = self.getSubtitles()
  81. self.assertTrue(subtitles['it'] is not None)
  82. def test_youtube_translated_subtitles(self):
  83. # This video has a subtitles track, which can be translated
  84. self.url = 'Ky9eprVWzlI'
  85. self.DL.params['writeautomaticsub'] = True
  86. self.DL.params['subtitleslangs'] = ['it']
  87. subtitles = self.getSubtitles()
  88. self.assertTrue(subtitles['it'] is not None)
  89. def test_youtube_nosubtitles(self):
  90. self.DL.expect_warning('video doesn\'t have subtitles')
  91. self.url = 'n5BB19UTcdA'
  92. self.DL.params['writesubtitles'] = True
  93. self.DL.params['allsubtitles'] = True
  94. subtitles = self.getSubtitles()
  95. self.assertEqual(len(subtitles), 0)
  96. def test_youtube_multiple_langs(self):
  97. self.url = 'QRS8MkLhQmM'
  98. self.DL.params['writesubtitles'] = True
  99. langs = ['it', 'fr', 'de']
  100. self.DL.params['subtitleslangs'] = langs
  101. subtitles = self.getSubtitles()
  102. for lang in langs:
  103. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  104. class TestDailymotionSubtitles(BaseTestSubtitles):
  105. url = 'http://www.dailymotion.com/video/xczg00'
  106. IE = DailymotionIE
  107. def test_subtitles(self):
  108. self.DL.params['writesubtitles'] = True
  109. subtitles = self.getSubtitles()
  110. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  111. def test_subtitles_lang(self):
  112. self.DL.params['writesubtitles'] = True
  113. self.DL.params['subtitleslangs'] = ['fr']
  114. subtitles = self.getSubtitles()
  115. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  116. def test_allsubtitles(self):
  117. self.DL.params['writesubtitles'] = True
  118. self.DL.params['allsubtitles'] = True
  119. subtitles = self.getSubtitles()
  120. self.assertTrue(len(subtitles.keys()) >= 6)
  121. def test_nosubtitles(self):
  122. self.DL.expect_warning('video doesn\'t have subtitles')
  123. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  124. self.DL.params['writesubtitles'] = True
  125. self.DL.params['allsubtitles'] = True
  126. subtitles = self.getSubtitles()
  127. self.assertEqual(len(subtitles), 0)
  128. def test_multiple_langs(self):
  129. self.DL.params['writesubtitles'] = True
  130. langs = ['es', 'fr', 'de']
  131. self.DL.params['subtitleslangs'] = langs
  132. subtitles = self.getSubtitles()
  133. for lang in langs:
  134. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  135. class TestTedSubtitles(BaseTestSubtitles):
  136. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  137. IE = TEDIE
  138. def test_no_writesubtitles(self):
  139. subtitles = self.getSubtitles()
  140. self.assertFalse(subtitles)
  141. def test_subtitles(self):
  142. self.DL.params['writesubtitles'] = True
  143. subtitles = self.getSubtitles()
  144. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  145. def test_subtitles_lang(self):
  146. self.DL.params['writesubtitles'] = True
  147. self.DL.params['subtitleslangs'] = ['fr']
  148. subtitles = self.getSubtitles()
  149. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  150. def test_allsubtitles(self):
  151. self.DL.params['writesubtitles'] = True
  152. self.DL.params['allsubtitles'] = True
  153. subtitles = self.getSubtitles()
  154. self.assertTrue(len(subtitles.keys()) >= 28)
  155. def test_list_subtitles(self):
  156. self.DL.params['listsubtitles'] = True
  157. info_dict = self.getInfoDict()
  158. self.assertEqual(info_dict, None)
  159. def test_multiple_langs(self):
  160. self.DL.params['writesubtitles'] = True
  161. langs = ['es', 'fr', 'de']
  162. self.DL.params['subtitleslangs'] = langs
  163. subtitles = self.getSubtitles()
  164. for lang in langs:
  165. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  166. class TestBlipTVSubtitles(BaseTestSubtitles):
  167. url = 'http://blip.tv/a/a-6603250'
  168. IE = BlipTVIE
  169. def test_list_subtitles(self):
  170. self.DL.params['listsubtitles'] = True
  171. info_dict = self.getInfoDict()
  172. self.assertEqual(info_dict, None)
  173. def test_allsubtitles(self):
  174. self.DL.params['writesubtitles'] = True
  175. self.DL.params['allsubtitles'] = True
  176. subtitles = self.getSubtitles()
  177. self.assertEqual(set(subtitles.keys()), set(['en']))
  178. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  179. class TestVimeoSubtitles(BaseTestSubtitles):
  180. url = 'http://vimeo.com/76979871'
  181. IE = VimeoIE
  182. def test_subtitles(self):
  183. self.DL.params['writesubtitles'] = True
  184. subtitles = self.getSubtitles()
  185. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  186. def test_subtitles_lang(self):
  187. self.DL.params['writesubtitles'] = True
  188. self.DL.params['subtitleslangs'] = ['fr']
  189. subtitles = self.getSubtitles()
  190. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  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(['de', 'en', 'es', 'fr']))
  196. def test_nosubtitles(self):
  197. self.DL.expect_warning('video doesn\'t have subtitles')
  198. self.url = 'http://vimeo.com/56015672'
  199. self.DL.params['writesubtitles'] = True
  200. self.DL.params['allsubtitles'] = True
  201. subtitles = self.getSubtitles()
  202. self.assertEqual(len(subtitles), 0)
  203. def test_multiple_langs(self):
  204. self.DL.params['writesubtitles'] = True
  205. langs = ['es', 'fr', 'de']
  206. self.DL.params['subtitleslangs'] = langs
  207. subtitles = self.getSubtitles()
  208. for lang in langs:
  209. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  210. class TestWallaSubtitles(BaseTestSubtitles):
  211. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  212. IE = WallaIE
  213. def test_list_subtitles(self):
  214. self.DL.expect_warning('Automatic Captions not supported by this server')
  215. self.DL.params['listsubtitles'] = True
  216. info_dict = self.getInfoDict()
  217. self.assertEqual(info_dict, None)
  218. def test_allsubtitles(self):
  219. self.DL.expect_warning('Automatic Captions not supported by this server')
  220. self.DL.params['writesubtitles'] = True
  221. self.DL.params['allsubtitles'] = True
  222. subtitles = self.getSubtitles()
  223. self.assertEqual(set(subtitles.keys()), set(['heb']))
  224. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  225. def test_nosubtitles(self):
  226. self.DL.expect_warning('video doesn\'t have subtitles')
  227. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  228. self.DL.params['writesubtitles'] = True
  229. self.DL.params['allsubtitles'] = True
  230. subtitles = self.getSubtitles()
  231. self.assertEqual(len(subtitles), 0)
  232. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  233. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  234. IE = CeskaTelevizeIE
  235. def test_list_subtitles(self):
  236. self.DL.expect_warning('Automatic Captions not supported by this server')
  237. self.DL.params['listsubtitles'] = True
  238. info_dict = self.getInfoDict()
  239. self.assertEqual(info_dict, None)
  240. def test_allsubtitles(self):
  241. self.DL.expect_warning('Automatic Captions not supported by this server')
  242. self.DL.params['writesubtitles'] = True
  243. self.DL.params['allsubtitles'] = True
  244. subtitles = self.getSubtitles()
  245. self.assertEqual(set(subtitles.keys()), set(['cs']))
  246. self.assertTrue(len(subtitles['cs']) > 20000)
  247. def test_nosubtitles(self):
  248. self.DL.expect_warning('video doesn\'t have subtitles')
  249. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  250. self.DL.params['writesubtitles'] = True
  251. self.DL.params['allsubtitles'] = True
  252. subtitles = self.getSubtitles()
  253. self.assertEqual(len(subtitles), 0)
  254. class TestLyndaSubtitles(BaseTestSubtitles):
  255. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  256. IE = LyndaIE
  257. def test_allsubtitles(self):
  258. self.DL.params['writesubtitles'] = True
  259. self.DL.params['allsubtitles'] = True
  260. subtitles = self.getSubtitles()
  261. self.assertEqual(set(subtitles.keys()), set(['en']))
  262. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  263. class TestNPOSubtitles(BaseTestSubtitles):
  264. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  265. IE = NPOIE
  266. def test_allsubtitles(self):
  267. self.DL.params['writesubtitles'] = True
  268. self.DL.params['allsubtitles'] = True
  269. subtitles = self.getSubtitles()
  270. self.assertEqual(set(subtitles.keys()), set(['nl']))
  271. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  272. class TestMTVSubtitles(BaseTestSubtitles):
  273. url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
  274. IE = ComedyCentralIE
  275. def getInfoDict(self):
  276. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  277. def test_allsubtitles(self):
  278. self.DL.params['writesubtitles'] = True
  279. self.DL.params['allsubtitles'] = True
  280. subtitles = self.getSubtitles()
  281. self.assertEqual(set(subtitles.keys()), set(['en']))
  282. self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
  283. class TestNRKSubtitles(BaseTestSubtitles):
  284. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  285. IE = NRKTVIE
  286. def test_allsubtitles(self):
  287. self.DL.params['writesubtitles'] = True
  288. self.DL.params['allsubtitles'] = True
  289. subtitles = self.getSubtitles()
  290. self.assertEqual(set(subtitles.keys()), set(['no']))
  291. self.assertEqual(md5(subtitles['no']), '1d221e6458c95c5494dcd38e6a1f129a')
  292. class TestRaiSubtitles(BaseTestSubtitles):
  293. url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  294. IE = RaiIE
  295. def test_allsubtitles(self):
  296. self.DL.params['writesubtitles'] = True
  297. self.DL.params['allsubtitles'] = True
  298. subtitles = self.getSubtitles()
  299. self.assertEqual(set(subtitles.keys()), set(['it']))
  300. self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  301. class TestVikiSubtitles(BaseTestSubtitles):
  302. url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
  303. IE = VikiIE
  304. def test_allsubtitles(self):
  305. self.DL.params['writesubtitles'] = True
  306. self.DL.params['allsubtitles'] = True
  307. subtitles = self.getSubtitles()
  308. self.assertEqual(set(subtitles.keys()), set(['en']))
  309. self.assertEqual(md5(subtitles['en']), 'b0b781eeb45efd3f6398a925b259150b')
  310. if __name__ == '__main__':
  311. unittest.main()