test_subtitles.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350
  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. )
  21. class BaseTestSubtitles(unittest.TestCase):
  22. url = None
  23. IE = None
  24. def setUp(self):
  25. self.DL = FakeYDL()
  26. self.ie = self.IE()
  27. self.DL.add_info_extractor(self.ie)
  28. def getInfoDict(self):
  29. info_dict = self.DL.extract_info(self.url, download=False)
  30. return info_dict
  31. def getSubtitles(self):
  32. info_dict = self.getInfoDict()
  33. subtitles = info_dict['requested_subtitles']
  34. if not subtitles:
  35. return subtitles
  36. for sub_info in subtitles.values():
  37. if sub_info.get('data') is None:
  38. uf = self.DL.urlopen(sub_info['url'])
  39. sub_info['data'] = uf.read().decode('utf-8')
  40. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  41. class TestYoutubeSubtitles(BaseTestSubtitles):
  42. url = 'QRS8MkLhQmM'
  43. IE = YoutubeIE
  44. def test_youtube_subtitles(self):
  45. self.DL.params['writesubtitles'] = True
  46. subtitles = self.getSubtitles()
  47. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  48. def test_youtube_subtitles_lang(self):
  49. self.DL.params['writesubtitles'] = True
  50. self.DL.params['subtitleslangs'] = ['it']
  51. subtitles = self.getSubtitles()
  52. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  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. def test_youtube_subtitles_sbv_format(self):
  59. self.DL.params['writesubtitles'] = True
  60. self.DL.params['subtitlesformat'] = 'sbv'
  61. subtitles = self.getSubtitles()
  62. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  63. def test_youtube_subtitles_vtt_format(self):
  64. self.DL.params['writesubtitles'] = True
  65. self.DL.params['subtitlesformat'] = 'vtt'
  66. subtitles = self.getSubtitles()
  67. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  68. def test_youtube_list_subtitles(self):
  69. self.DL.expect_warning('Video doesn\'t have automatic captions')
  70. self.DL.params['listsubtitles'] = True
  71. info_dict = self.getInfoDict()
  72. self.assertEqual(info_dict, None)
  73. def test_youtube_automatic_captions(self):
  74. self.url = '8YoUxe5ncPo'
  75. self.DL.params['writeautomaticsub'] = True
  76. self.DL.params['subtitleslangs'] = ['it']
  77. subtitles = self.getSubtitles()
  78. self.assertTrue(subtitles['it'] is not None)
  79. def test_youtube_translated_subtitles(self):
  80. # This video has a subtitles track, which can be translated
  81. self.url = 'Ky9eprVWzlI'
  82. self.DL.params['writeautomaticsub'] = True
  83. self.DL.params['subtitleslangs'] = ['it']
  84. subtitles = self.getSubtitles()
  85. self.assertTrue(subtitles['it'] is not None)
  86. def test_youtube_nosubtitles(self):
  87. self.DL.expect_warning('video doesn\'t have subtitles')
  88. self.url = 'n5BB19UTcdA'
  89. self.DL.params['writesubtitles'] = True
  90. self.DL.params['allsubtitles'] = True
  91. subtitles = self.getSubtitles()
  92. self.assertEqual(len(subtitles), 0)
  93. def test_youtube_multiple_langs(self):
  94. self.url = 'QRS8MkLhQmM'
  95. self.DL.params['writesubtitles'] = True
  96. langs = ['it', 'fr', 'de']
  97. self.DL.params['subtitleslangs'] = langs
  98. subtitles = self.getSubtitles()
  99. for lang in langs:
  100. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  101. class TestDailymotionSubtitles(BaseTestSubtitles):
  102. url = 'http://www.dailymotion.com/video/xczg00'
  103. IE = DailymotionIE
  104. def test_subtitles(self):
  105. self.DL.params['writesubtitles'] = True
  106. subtitles = self.getSubtitles()
  107. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  108. def test_subtitles_lang(self):
  109. self.DL.params['writesubtitles'] = True
  110. self.DL.params['subtitleslangs'] = ['fr']
  111. subtitles = self.getSubtitles()
  112. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  113. def test_allsubtitles(self):
  114. self.DL.params['writesubtitles'] = True
  115. self.DL.params['allsubtitles'] = True
  116. subtitles = self.getSubtitles()
  117. self.assertTrue(len(subtitles.keys()) >= 6)
  118. def test_nosubtitles(self):
  119. self.DL.expect_warning('video doesn\'t have subtitles')
  120. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  121. self.DL.params['writesubtitles'] = True
  122. self.DL.params['allsubtitles'] = True
  123. subtitles = self.getSubtitles()
  124. self.assertEqual(len(subtitles), 0)
  125. def test_multiple_langs(self):
  126. self.DL.params['writesubtitles'] = True
  127. langs = ['es', 'fr', 'de']
  128. self.DL.params['subtitleslangs'] = langs
  129. subtitles = self.getSubtitles()
  130. for lang in langs:
  131. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  132. class TestTedSubtitles(BaseTestSubtitles):
  133. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  134. IE = TEDIE
  135. def test_no_writesubtitles(self):
  136. subtitles = self.getSubtitles()
  137. self.assertFalse(subtitles)
  138. def test_subtitles(self):
  139. self.DL.params['writesubtitles'] = True
  140. subtitles = self.getSubtitles()
  141. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  142. def test_subtitles_lang(self):
  143. self.DL.params['writesubtitles'] = True
  144. self.DL.params['subtitleslangs'] = ['fr']
  145. subtitles = self.getSubtitles()
  146. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  147. def test_allsubtitles(self):
  148. self.DL.params['writesubtitles'] = True
  149. self.DL.params['allsubtitles'] = True
  150. subtitles = self.getSubtitles()
  151. self.assertTrue(len(subtitles.keys()) >= 28)
  152. def test_list_subtitles(self):
  153. self.DL.params['listsubtitles'] = True
  154. info_dict = self.getInfoDict()
  155. self.assertEqual(info_dict, None)
  156. def test_multiple_langs(self):
  157. self.DL.params['writesubtitles'] = True
  158. langs = ['es', 'fr', 'de']
  159. self.DL.params['subtitleslangs'] = langs
  160. subtitles = self.getSubtitles()
  161. for lang in langs:
  162. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  163. class TestBlipTVSubtitles(BaseTestSubtitles):
  164. url = 'http://blip.tv/a/a-6603250'
  165. IE = BlipTVIE
  166. def test_list_subtitles(self):
  167. self.DL.params['listsubtitles'] = True
  168. info_dict = self.getInfoDict()
  169. self.assertEqual(info_dict, None)
  170. def test_allsubtitles(self):
  171. self.DL.params['writesubtitles'] = True
  172. self.DL.params['allsubtitles'] = True
  173. subtitles = self.getSubtitles()
  174. self.assertEqual(set(subtitles.keys()), set(['en']))
  175. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  176. class TestVimeoSubtitles(BaseTestSubtitles):
  177. url = 'http://vimeo.com/76979871'
  178. IE = VimeoIE
  179. def test_subtitles(self):
  180. self.DL.params['writesubtitles'] = True
  181. subtitles = self.getSubtitles()
  182. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  183. def test_subtitles_lang(self):
  184. self.DL.params['writesubtitles'] = True
  185. self.DL.params['subtitleslangs'] = ['fr']
  186. subtitles = self.getSubtitles()
  187. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  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(['de', 'en', 'es', 'fr']))
  193. def test_nosubtitles(self):
  194. self.DL.expect_warning('video doesn\'t have subtitles')
  195. self.url = 'http://vimeo.com/56015672'
  196. self.DL.params['writesubtitles'] = True
  197. self.DL.params['allsubtitles'] = True
  198. subtitles = self.getSubtitles()
  199. self.assertEqual(len(subtitles), 0)
  200. def test_multiple_langs(self):
  201. self.DL.params['writesubtitles'] = True
  202. langs = ['es', 'fr', 'de']
  203. self.DL.params['subtitleslangs'] = langs
  204. subtitles = self.getSubtitles()
  205. for lang in langs:
  206. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  207. class TestWallaSubtitles(BaseTestSubtitles):
  208. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  209. IE = WallaIE
  210. def test_list_subtitles(self):
  211. self.DL.expect_warning('Automatic Captions not supported by this server')
  212. self.DL.params['listsubtitles'] = True
  213. info_dict = self.getInfoDict()
  214. self.assertEqual(info_dict, None)
  215. def test_allsubtitles(self):
  216. self.DL.expect_warning('Automatic Captions not supported by this server')
  217. self.DL.params['writesubtitles'] = True
  218. self.DL.params['allsubtitles'] = True
  219. subtitles = self.getSubtitles()
  220. self.assertEqual(set(subtitles.keys()), set(['heb']))
  221. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  222. def test_nosubtitles(self):
  223. self.DL.expect_warning('video doesn\'t have subtitles')
  224. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  225. self.DL.params['writesubtitles'] = True
  226. self.DL.params['allsubtitles'] = True
  227. subtitles = self.getSubtitles()
  228. self.assertEqual(len(subtitles), 0)
  229. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  230. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  231. IE = CeskaTelevizeIE
  232. def test_list_subtitles(self):
  233. self.DL.expect_warning('Automatic Captions not supported by this server')
  234. self.DL.params['listsubtitles'] = True
  235. info_dict = self.getInfoDict()
  236. self.assertEqual(info_dict, None)
  237. def test_allsubtitles(self):
  238. self.DL.expect_warning('Automatic Captions not supported by this server')
  239. self.DL.params['writesubtitles'] = True
  240. self.DL.params['allsubtitles'] = True
  241. subtitles = self.getSubtitles()
  242. self.assertEqual(set(subtitles.keys()), set(['cs']))
  243. self.assertTrue(len(subtitles['cs']) > 20000)
  244. def test_nosubtitles(self):
  245. self.DL.expect_warning('video doesn\'t have subtitles')
  246. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  247. self.DL.params['writesubtitles'] = True
  248. self.DL.params['allsubtitles'] = True
  249. subtitles = self.getSubtitles()
  250. self.assertEqual(len(subtitles), 0)
  251. class TestLyndaSubtitles(BaseTestSubtitles):
  252. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  253. IE = LyndaIE
  254. def test_allsubtitles(self):
  255. self.DL.params['writesubtitles'] = True
  256. self.DL.params['allsubtitles'] = True
  257. subtitles = self.getSubtitles()
  258. self.assertEqual(set(subtitles.keys()), set(['en']))
  259. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  260. class TestNPOSubtitles(BaseTestSubtitles):
  261. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  262. IE = NPOIE
  263. def test_allsubtitles(self):
  264. self.DL.params['writesubtitles'] = True
  265. self.DL.params['allsubtitles'] = True
  266. subtitles = self.getSubtitles()
  267. self.assertEqual(set(subtitles.keys()), set(['nl']))
  268. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  269. class TestMTVSubtitles(BaseTestSubtitles):
  270. url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
  271. IE = ComedyCentralIE
  272. def getInfoDict(self):
  273. return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  274. def test_allsubtitles(self):
  275. self.DL.params['writesubtitles'] = True
  276. self.DL.params['allsubtitles'] = True
  277. subtitles = self.getSubtitles()
  278. self.assertEqual(set(subtitles.keys()), set(['en']))
  279. self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
  280. if __name__ == '__main__':
  281. unittest.main()