test_subtitles.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376
  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. )
  23. class BaseTestSubtitles(unittest.TestCase):
  24. url = None
  25. IE = None
  26. def setUp(self):
  27. self.DL = FakeYDL()
  28. self.ie = self.IE()
  29. self.DL.add_info_extractor(self.ie)
  30. def getInfoDict(self):
  31. info_dict = self.DL.extract_info(self.url, download=False)
  32. return info_dict
  33. def getSubtitles(self):
  34. info_dict = self.getInfoDict()
  35. subtitles = info_dict['requested_subtitles']
  36. if not subtitles:
  37. return subtitles
  38. for sub_info in subtitles.values():
  39. if sub_info.get('data') is None:
  40. uf = self.DL.urlopen(sub_info['url'])
  41. sub_info['data'] = uf.read().decode('utf-8')
  42. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  43. class TestYoutubeSubtitles(BaseTestSubtitles):
  44. url = 'QRS8MkLhQmM'
  45. IE = YoutubeIE
  46. def test_youtube_subtitles(self):
  47. self.DL.params['writesubtitles'] = True
  48. subtitles = self.getSubtitles()
  49. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  50. def test_youtube_subtitles_lang(self):
  51. self.DL.params['writesubtitles'] = True
  52. self.DL.params['subtitleslangs'] = ['it']
  53. subtitles = self.getSubtitles()
  54. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  55. def test_youtube_allsubtitles(self):
  56. self.DL.params['writesubtitles'] = True
  57. self.DL.params['allsubtitles'] = True
  58. subtitles = self.getSubtitles()
  59. self.assertEqual(len(subtitles.keys()), 13)
  60. def test_youtube_subtitles_sbv_format(self):
  61. self.DL.params['writesubtitles'] = True
  62. self.DL.params['subtitlesformat'] = 'sbv'
  63. subtitles = self.getSubtitles()
  64. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  65. def test_youtube_subtitles_vtt_format(self):
  66. self.DL.params['writesubtitles'] = True
  67. self.DL.params['subtitlesformat'] = 'vtt'
  68. subtitles = self.getSubtitles()
  69. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  70. def test_youtube_list_subtitles(self):
  71. self.DL.expect_warning('Video doesn\'t have automatic captions')
  72. self.DL.params['listsubtitles'] = True
  73. info_dict = self.getInfoDict()
  74. self.assertEqual(info_dict, None)
  75. def test_youtube_automatic_captions(self):
  76. self.url = '8YoUxe5ncPo'
  77. self.DL.params['writeautomaticsub'] = True
  78. self.DL.params['subtitleslangs'] = ['it']
  79. subtitles = self.getSubtitles()
  80. self.assertTrue(subtitles['it'] is not None)
  81. def test_youtube_translated_subtitles(self):
  82. # This video has a subtitles track, which can be translated
  83. self.url = 'Ky9eprVWzlI'
  84. self.DL.params['writeautomaticsub'] = True
  85. self.DL.params['subtitleslangs'] = ['it']
  86. subtitles = self.getSubtitles()
  87. self.assertTrue(subtitles['it'] is not None)
  88. def test_youtube_nosubtitles(self):
  89. self.DL.expect_warning('video doesn\'t have subtitles')
  90. self.url = 'n5BB19UTcdA'
  91. self.DL.params['writesubtitles'] = True
  92. self.DL.params['allsubtitles'] = True
  93. subtitles = self.getSubtitles()
  94. self.assertEqual(len(subtitles), 0)
  95. def test_youtube_multiple_langs(self):
  96. self.url = 'QRS8MkLhQmM'
  97. self.DL.params['writesubtitles'] = True
  98. langs = ['it', 'fr', 'de']
  99. self.DL.params['subtitleslangs'] = langs
  100. subtitles = self.getSubtitles()
  101. for lang in langs:
  102. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  103. class TestDailymotionSubtitles(BaseTestSubtitles):
  104. url = 'http://www.dailymotion.com/video/xczg00'
  105. IE = DailymotionIE
  106. def test_subtitles(self):
  107. self.DL.params['writesubtitles'] = True
  108. subtitles = self.getSubtitles()
  109. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  110. def test_subtitles_lang(self):
  111. self.DL.params['writesubtitles'] = True
  112. self.DL.params['subtitleslangs'] = ['fr']
  113. subtitles = self.getSubtitles()
  114. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  115. def test_allsubtitles(self):
  116. self.DL.params['writesubtitles'] = True
  117. self.DL.params['allsubtitles'] = True
  118. subtitles = self.getSubtitles()
  119. self.assertTrue(len(subtitles.keys()) >= 6)
  120. def test_nosubtitles(self):
  121. self.DL.expect_warning('video doesn\'t have subtitles')
  122. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  123. self.DL.params['writesubtitles'] = True
  124. self.DL.params['allsubtitles'] = True
  125. subtitles = self.getSubtitles()
  126. self.assertEqual(len(subtitles), 0)
  127. def test_multiple_langs(self):
  128. self.DL.params['writesubtitles'] = True
  129. langs = ['es', 'fr', 'de']
  130. self.DL.params['subtitleslangs'] = langs
  131. subtitles = self.getSubtitles()
  132. for lang in langs:
  133. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  134. class TestTedSubtitles(BaseTestSubtitles):
  135. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  136. IE = TEDIE
  137. def test_no_writesubtitles(self):
  138. subtitles = self.getSubtitles()
  139. self.assertFalse(subtitles)
  140. def test_subtitles(self):
  141. self.DL.params['writesubtitles'] = True
  142. subtitles = self.getSubtitles()
  143. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  144. def test_subtitles_lang(self):
  145. self.DL.params['writesubtitles'] = True
  146. self.DL.params['subtitleslangs'] = ['fr']
  147. subtitles = self.getSubtitles()
  148. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  149. def test_allsubtitles(self):
  150. self.DL.params['writesubtitles'] = True
  151. self.DL.params['allsubtitles'] = True
  152. subtitles = self.getSubtitles()
  153. self.assertTrue(len(subtitles.keys()) >= 28)
  154. def test_list_subtitles(self):
  155. self.DL.params['listsubtitles'] = True
  156. info_dict = self.getInfoDict()
  157. self.assertEqual(info_dict, None)
  158. def test_multiple_langs(self):
  159. self.DL.params['writesubtitles'] = True
  160. langs = ['es', 'fr', 'de']
  161. self.DL.params['subtitleslangs'] = langs
  162. subtitles = self.getSubtitles()
  163. for lang in langs:
  164. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  165. class TestBlipTVSubtitles(BaseTestSubtitles):
  166. url = 'http://blip.tv/a/a-6603250'
  167. IE = BlipTVIE
  168. def test_list_subtitles(self):
  169. self.DL.params['listsubtitles'] = True
  170. info_dict = self.getInfoDict()
  171. self.assertEqual(info_dict, None)
  172. def test_allsubtitles(self):
  173. self.DL.params['writesubtitles'] = True
  174. self.DL.params['allsubtitles'] = True
  175. subtitles = self.getSubtitles()
  176. self.assertEqual(set(subtitles.keys()), set(['en']))
  177. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  178. class TestVimeoSubtitles(BaseTestSubtitles):
  179. url = 'http://vimeo.com/76979871'
  180. IE = VimeoIE
  181. def test_subtitles(self):
  182. self.DL.params['writesubtitles'] = True
  183. subtitles = self.getSubtitles()
  184. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  185. def test_subtitles_lang(self):
  186. self.DL.params['writesubtitles'] = True
  187. self.DL.params['subtitleslangs'] = ['fr']
  188. subtitles = self.getSubtitles()
  189. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  190. def test_allsubtitles(self):
  191. self.DL.params['writesubtitles'] = True
  192. self.DL.params['allsubtitles'] = True
  193. subtitles = self.getSubtitles()
  194. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  195. def test_nosubtitles(self):
  196. self.DL.expect_warning('video doesn\'t have subtitles')
  197. self.url = 'http://vimeo.com/56015672'
  198. self.DL.params['writesubtitles'] = True
  199. self.DL.params['allsubtitles'] = True
  200. subtitles = self.getSubtitles()
  201. self.assertEqual(len(subtitles), 0)
  202. def test_multiple_langs(self):
  203. self.DL.params['writesubtitles'] = True
  204. langs = ['es', 'fr', 'de']
  205. self.DL.params['subtitleslangs'] = langs
  206. subtitles = self.getSubtitles()
  207. for lang in langs:
  208. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  209. class TestWallaSubtitles(BaseTestSubtitles):
  210. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  211. IE = WallaIE
  212. def test_list_subtitles(self):
  213. self.DL.expect_warning('Automatic Captions not supported by this server')
  214. self.DL.params['listsubtitles'] = True
  215. info_dict = self.getInfoDict()
  216. self.assertEqual(info_dict, None)
  217. def test_allsubtitles(self):
  218. self.DL.expect_warning('Automatic Captions not supported by this server')
  219. self.DL.params['writesubtitles'] = True
  220. self.DL.params['allsubtitles'] = True
  221. subtitles = self.getSubtitles()
  222. self.assertEqual(set(subtitles.keys()), set(['heb']))
  223. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  224. def test_nosubtitles(self):
  225. self.DL.expect_warning('video doesn\'t have subtitles')
  226. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  227. self.DL.params['writesubtitles'] = True
  228. self.DL.params['allsubtitles'] = True
  229. subtitles = self.getSubtitles()
  230. self.assertEqual(len(subtitles), 0)
  231. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  232. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  233. IE = CeskaTelevizeIE
  234. def test_list_subtitles(self):
  235. self.DL.expect_warning('Automatic Captions not supported by this server')
  236. self.DL.params['listsubtitles'] = True
  237. info_dict = self.getInfoDict()
  238. self.assertEqual(info_dict, None)
  239. def test_allsubtitles(self):
  240. self.DL.expect_warning('Automatic Captions not supported by this server')
  241. self.DL.params['writesubtitles'] = True
  242. self.DL.params['allsubtitles'] = True
  243. subtitles = self.getSubtitles()
  244. self.assertEqual(set(subtitles.keys()), set(['cs']))
  245. self.assertTrue(len(subtitles['cs']) > 20000)
  246. def test_nosubtitles(self):
  247. self.DL.expect_warning('video doesn\'t have subtitles')
  248. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  249. self.DL.params['writesubtitles'] = True
  250. self.DL.params['allsubtitles'] = True
  251. subtitles = self.getSubtitles()
  252. self.assertEqual(len(subtitles), 0)
  253. class TestLyndaSubtitles(BaseTestSubtitles):
  254. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  255. IE = LyndaIE
  256. def test_allsubtitles(self):
  257. self.DL.params['writesubtitles'] = True
  258. self.DL.params['allsubtitles'] = True
  259. subtitles = self.getSubtitles()
  260. self.assertEqual(set(subtitles.keys()), set(['en']))
  261. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  262. class TestNPOSubtitles(BaseTestSubtitles):
  263. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  264. IE = NPOIE
  265. def test_allsubtitles(self):
  266. self.DL.params['writesubtitles'] = True
  267. self.DL.params['allsubtitles'] = True
  268. subtitles = self.getSubtitles()
  269. self.assertEqual(set(subtitles.keys()), set(['nl']))
  270. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  271. class TestMTVSubtitles(BaseTestSubtitles):
  272. url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
  273. IE = ComedyCentralIE
  274. def getInfoDict(self):
  275. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  276. def test_allsubtitles(self):
  277. self.DL.params['writesubtitles'] = True
  278. self.DL.params['allsubtitles'] = True
  279. subtitles = self.getSubtitles()
  280. self.assertEqual(set(subtitles.keys()), set(['en']))
  281. self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
  282. class TestNRKSubtitles(BaseTestSubtitles):
  283. url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  284. IE = NRKTVIE
  285. def test_allsubtitles(self):
  286. self.DL.params['writesubtitles'] = True
  287. self.DL.params['allsubtitles'] = True
  288. subtitles = self.getSubtitles()
  289. self.assertEqual(set(subtitles.keys()), set(['no']))
  290. self.assertEqual(md5(subtitles['no']), '1d221e6458c95c5494dcd38e6a1f129a')
  291. class TestRaiSubtitles(BaseTestSubtitles):
  292. url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  293. IE = RaiIE
  294. def test_allsubtitles(self):
  295. self.DL.params['writesubtitles'] = True
  296. self.DL.params['allsubtitles'] = True
  297. subtitles = self.getSubtitles()
  298. self.assertEqual(set(subtitles.keys()), set(['it']))
  299. self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  300. if __name__ == '__main__':
  301. unittest.main()