test_subtitles.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363
  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. )
  22. class BaseTestSubtitles(unittest.TestCase):
  23. url = None
  24. IE = None
  25. def setUp(self):
  26. self.DL = FakeYDL()
  27. self.ie = self.IE()
  28. self.DL.add_info_extractor(self.ie)
  29. def getInfoDict(self):
  30. info_dict = self.DL.extract_info(self.url, download=False)
  31. return info_dict
  32. def getSubtitles(self):
  33. info_dict = self.getInfoDict()
  34. subtitles = info_dict['requested_subtitles']
  35. if not subtitles:
  36. return subtitles
  37. for sub_info in subtitles.values():
  38. if sub_info.get('data') is None:
  39. uf = self.DL.urlopen(sub_info['url'])
  40. sub_info['data'] = uf.read().decode('utf-8')
  41. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  42. class TestYoutubeSubtitles(BaseTestSubtitles):
  43. url = 'QRS8MkLhQmM'
  44. IE = YoutubeIE
  45. def test_youtube_subtitles(self):
  46. self.DL.params['writesubtitles'] = True
  47. subtitles = self.getSubtitles()
  48. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  49. def test_youtube_subtitles_lang(self):
  50. self.DL.params['writesubtitles'] = True
  51. self.DL.params['subtitleslangs'] = ['it']
  52. subtitles = self.getSubtitles()
  53. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  54. def test_youtube_allsubtitles(self):
  55. self.DL.params['writesubtitles'] = True
  56. self.DL.params['allsubtitles'] = True
  57. subtitles = self.getSubtitles()
  58. self.assertEqual(len(subtitles.keys()), 13)
  59. def test_youtube_subtitles_sbv_format(self):
  60. self.DL.params['writesubtitles'] = True
  61. self.DL.params['subtitlesformat'] = 'sbv'
  62. subtitles = self.getSubtitles()
  63. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  64. def test_youtube_subtitles_vtt_format(self):
  65. self.DL.params['writesubtitles'] = True
  66. self.DL.params['subtitlesformat'] = 'vtt'
  67. subtitles = self.getSubtitles()
  68. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  69. def test_youtube_list_subtitles(self):
  70. self.DL.expect_warning('Video doesn\'t have automatic captions')
  71. self.DL.params['listsubtitles'] = True
  72. info_dict = self.getInfoDict()
  73. self.assertEqual(info_dict, None)
  74. def test_youtube_automatic_captions(self):
  75. self.url = '8YoUxe5ncPo'
  76. self.DL.params['writeautomaticsub'] = True
  77. self.DL.params['subtitleslangs'] = ['it']
  78. subtitles = self.getSubtitles()
  79. self.assertTrue(subtitles['it'] is not None)
  80. def test_youtube_translated_subtitles(self):
  81. # This video has a subtitles track, which can be translated
  82. self.url = 'Ky9eprVWzlI'
  83. self.DL.params['writeautomaticsub'] = True
  84. self.DL.params['subtitleslangs'] = ['it']
  85. subtitles = self.getSubtitles()
  86. self.assertTrue(subtitles['it'] is not None)
  87. def test_youtube_nosubtitles(self):
  88. self.DL.expect_warning('video doesn\'t have subtitles')
  89. self.url = 'n5BB19UTcdA'
  90. self.DL.params['writesubtitles'] = True
  91. self.DL.params['allsubtitles'] = True
  92. subtitles = self.getSubtitles()
  93. self.assertEqual(len(subtitles), 0)
  94. def test_youtube_multiple_langs(self):
  95. self.url = 'QRS8MkLhQmM'
  96. self.DL.params['writesubtitles'] = True
  97. langs = ['it', 'fr', 'de']
  98. self.DL.params['subtitleslangs'] = langs
  99. subtitles = self.getSubtitles()
  100. for lang in langs:
  101. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  102. class TestDailymotionSubtitles(BaseTestSubtitles):
  103. url = 'http://www.dailymotion.com/video/xczg00'
  104. IE = DailymotionIE
  105. def test_subtitles(self):
  106. self.DL.params['writesubtitles'] = True
  107. subtitles = self.getSubtitles()
  108. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  109. def test_subtitles_lang(self):
  110. self.DL.params['writesubtitles'] = True
  111. self.DL.params['subtitleslangs'] = ['fr']
  112. subtitles = self.getSubtitles()
  113. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  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()) >= 6)
  119. def test_nosubtitles(self):
  120. self.DL.expect_warning('video doesn\'t have subtitles')
  121. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  122. self.DL.params['writesubtitles'] = True
  123. self.DL.params['allsubtitles'] = True
  124. subtitles = self.getSubtitles()
  125. self.assertEqual(len(subtitles), 0)
  126. def test_multiple_langs(self):
  127. self.DL.params['writesubtitles'] = True
  128. langs = ['es', 'fr', 'de']
  129. self.DL.params['subtitleslangs'] = langs
  130. subtitles = self.getSubtitles()
  131. for lang in langs:
  132. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  133. class TestTedSubtitles(BaseTestSubtitles):
  134. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  135. IE = TEDIE
  136. def test_no_writesubtitles(self):
  137. subtitles = self.getSubtitles()
  138. self.assertFalse(subtitles)
  139. def test_subtitles(self):
  140. self.DL.params['writesubtitles'] = True
  141. subtitles = self.getSubtitles()
  142. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  143. def test_subtitles_lang(self):
  144. self.DL.params['writesubtitles'] = True
  145. self.DL.params['subtitleslangs'] = ['fr']
  146. subtitles = self.getSubtitles()
  147. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  148. def test_allsubtitles(self):
  149. self.DL.params['writesubtitles'] = True
  150. self.DL.params['allsubtitles'] = True
  151. subtitles = self.getSubtitles()
  152. self.assertTrue(len(subtitles.keys()) >= 28)
  153. def test_list_subtitles(self):
  154. self.DL.params['listsubtitles'] = True
  155. info_dict = self.getInfoDict()
  156. self.assertEqual(info_dict, None)
  157. def test_multiple_langs(self):
  158. self.DL.params['writesubtitles'] = True
  159. langs = ['es', 'fr', 'de']
  160. self.DL.params['subtitleslangs'] = langs
  161. subtitles = self.getSubtitles()
  162. for lang in langs:
  163. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  164. class TestBlipTVSubtitles(BaseTestSubtitles):
  165. url = 'http://blip.tv/a/a-6603250'
  166. IE = BlipTVIE
  167. def test_list_subtitles(self):
  168. self.DL.params['listsubtitles'] = True
  169. info_dict = self.getInfoDict()
  170. self.assertEqual(info_dict, None)
  171. def test_allsubtitles(self):
  172. self.DL.params['writesubtitles'] = True
  173. self.DL.params['allsubtitles'] = True
  174. subtitles = self.getSubtitles()
  175. self.assertEqual(set(subtitles.keys()), set(['en']))
  176. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  177. class TestVimeoSubtitles(BaseTestSubtitles):
  178. url = 'http://vimeo.com/76979871'
  179. IE = VimeoIE
  180. def test_subtitles(self):
  181. self.DL.params['writesubtitles'] = True
  182. subtitles = self.getSubtitles()
  183. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  184. def test_subtitles_lang(self):
  185. self.DL.params['writesubtitles'] = True
  186. self.DL.params['subtitleslangs'] = ['fr']
  187. subtitles = self.getSubtitles()
  188. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  189. def test_allsubtitles(self):
  190. self.DL.params['writesubtitles'] = True
  191. self.DL.params['allsubtitles'] = True
  192. subtitles = self.getSubtitles()
  193. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  194. def test_nosubtitles(self):
  195. self.DL.expect_warning('video doesn\'t have subtitles')
  196. self.url = 'http://vimeo.com/56015672'
  197. self.DL.params['writesubtitles'] = True
  198. self.DL.params['allsubtitles'] = True
  199. subtitles = self.getSubtitles()
  200. self.assertEqual(len(subtitles), 0)
  201. def test_multiple_langs(self):
  202. self.DL.params['writesubtitles'] = True
  203. langs = ['es', 'fr', 'de']
  204. self.DL.params['subtitleslangs'] = langs
  205. subtitles = self.getSubtitles()
  206. for lang in langs:
  207. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  208. class TestWallaSubtitles(BaseTestSubtitles):
  209. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  210. IE = WallaIE
  211. def test_list_subtitles(self):
  212. self.DL.expect_warning('Automatic Captions not supported by this server')
  213. self.DL.params['listsubtitles'] = True
  214. info_dict = self.getInfoDict()
  215. self.assertEqual(info_dict, None)
  216. def test_allsubtitles(self):
  217. self.DL.expect_warning('Automatic Captions not supported by this server')
  218. self.DL.params['writesubtitles'] = True
  219. self.DL.params['allsubtitles'] = True
  220. subtitles = self.getSubtitles()
  221. self.assertEqual(set(subtitles.keys()), set(['heb']))
  222. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  223. def test_nosubtitles(self):
  224. self.DL.expect_warning('video doesn\'t have subtitles')
  225. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  226. self.DL.params['writesubtitles'] = True
  227. self.DL.params['allsubtitles'] = True
  228. subtitles = self.getSubtitles()
  229. self.assertEqual(len(subtitles), 0)
  230. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  231. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  232. IE = CeskaTelevizeIE
  233. def test_list_subtitles(self):
  234. self.DL.expect_warning('Automatic Captions not supported by this server')
  235. self.DL.params['listsubtitles'] = True
  236. info_dict = self.getInfoDict()
  237. self.assertEqual(info_dict, None)
  238. def test_allsubtitles(self):
  239. self.DL.expect_warning('Automatic Captions not supported by this server')
  240. self.DL.params['writesubtitles'] = True
  241. self.DL.params['allsubtitles'] = True
  242. subtitles = self.getSubtitles()
  243. self.assertEqual(set(subtitles.keys()), set(['cs']))
  244. self.assertTrue(len(subtitles['cs']) > 20000)
  245. def test_nosubtitles(self):
  246. self.DL.expect_warning('video doesn\'t have subtitles')
  247. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  248. self.DL.params['writesubtitles'] = True
  249. self.DL.params['allsubtitles'] = True
  250. subtitles = self.getSubtitles()
  251. self.assertEqual(len(subtitles), 0)
  252. class TestLyndaSubtitles(BaseTestSubtitles):
  253. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  254. IE = LyndaIE
  255. def test_allsubtitles(self):
  256. self.DL.params['writesubtitles'] = True
  257. self.DL.params['allsubtitles'] = True
  258. subtitles = self.getSubtitles()
  259. self.assertEqual(set(subtitles.keys()), set(['en']))
  260. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  261. class TestNPOSubtitles(BaseTestSubtitles):
  262. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  263. IE = NPOIE
  264. def test_allsubtitles(self):
  265. self.DL.params['writesubtitles'] = True
  266. self.DL.params['allsubtitles'] = True
  267. subtitles = self.getSubtitles()
  268. self.assertEqual(set(subtitles.keys()), set(['nl']))
  269. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  270. class TestMTVSubtitles(BaseTestSubtitles):
  271. url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
  272. IE = ComedyCentralIE
  273. def getInfoDict(self):
  274. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  275. def test_allsubtitles(self):
  276. self.DL.params['writesubtitles'] = True
  277. self.DL.params['allsubtitles'] = True
  278. subtitles = self.getSubtitles()
  279. self.assertEqual(set(subtitles.keys()), set(['en']))
  280. self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
  281. class TestNRKSubtitles(BaseTestSubtitles):
  282. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  283. IE = NRKTVIE
  284. def test_allsubtitles(self):
  285. self.DL.params['writesubtitles'] = True
  286. self.DL.params['allsubtitles'] = True
  287. subtitles = self.getSubtitles()
  288. self.assertEqual(set(subtitles.keys()), set(['no']))
  289. self.assertEqual(md5(subtitles['no']), '1d221e6458c95c5494dcd38e6a1f129a')
  290. if __name__ == '__main__':
  291. unittest.main()