test_subtitles.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349
  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. )
  18. class BaseTestSubtitles(unittest.TestCase):
  19. url = None
  20. IE = None
  21. def setUp(self):
  22. self.DL = FakeYDL()
  23. self.ie = self.IE(self.DL)
  24. def getInfoDict(self):
  25. info_dict = self.ie.extract(self.url)
  26. return info_dict
  27. def getSubtitles(self):
  28. info_dict = self.getInfoDict()
  29. return info_dict['subtitles']
  30. class TestYoutubeSubtitles(BaseTestSubtitles):
  31. url = 'QRS8MkLhQmM'
  32. IE = YoutubeIE
  33. def test_youtube_no_writesubtitles(self):
  34. self.DL.params['writesubtitles'] = False
  35. subtitles = self.getSubtitles()
  36. self.assertEqual(subtitles, None)
  37. def test_youtube_subtitles(self):
  38. self.DL.params['writesubtitles'] = True
  39. subtitles = self.getSubtitles()
  40. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  41. def test_youtube_subtitles_lang(self):
  42. self.DL.params['writesubtitles'] = True
  43. self.DL.params['subtitleslangs'] = ['it']
  44. subtitles = self.getSubtitles()
  45. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  46. def test_youtube_allsubtitles(self):
  47. self.DL.params['writesubtitles'] = True
  48. self.DL.params['allsubtitles'] = True
  49. subtitles = self.getSubtitles()
  50. self.assertEqual(len(subtitles.keys()), 13)
  51. def test_youtube_subtitles_sbv_format(self):
  52. self.DL.params['writesubtitles'] = True
  53. self.DL.params['subtitlesformat'] = 'sbv'
  54. subtitles = self.getSubtitles()
  55. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  56. def test_youtube_subtitles_vtt_format(self):
  57. self.DL.params['writesubtitles'] = True
  58. self.DL.params['subtitlesformat'] = 'vtt'
  59. subtitles = self.getSubtitles()
  60. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  61. def test_youtube_list_subtitles(self):
  62. self.DL.expect_warning('Video doesn\'t have automatic captions')
  63. self.DL.params['listsubtitles'] = True
  64. info_dict = self.getInfoDict()
  65. self.assertEqual(info_dict, None)
  66. def test_youtube_automatic_captions(self):
  67. self.url = '8YoUxe5ncPo'
  68. self.DL.params['writeautomaticsub'] = True
  69. self.DL.params['subtitleslangs'] = ['it']
  70. subtitles = self.getSubtitles()
  71. self.assertTrue(subtitles['it'] is not None)
  72. def test_youtube_translated_subtitles(self):
  73. # This video has a subtitles track, which can be translated
  74. self.url = 'Ky9eprVWzlI'
  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_nosubtitles(self):
  80. self.DL.expect_warning('video doesn\'t have subtitles')
  81. self.url = 'n5BB19UTcdA'
  82. self.DL.params['writesubtitles'] = True
  83. self.DL.params['allsubtitles'] = True
  84. subtitles = self.getSubtitles()
  85. self.assertEqual(len(subtitles), 0)
  86. def test_youtube_multiple_langs(self):
  87. self.url = 'QRS8MkLhQmM'
  88. self.DL.params['writesubtitles'] = True
  89. langs = ['it', 'fr', 'de']
  90. self.DL.params['subtitleslangs'] = langs
  91. subtitles = self.getSubtitles()
  92. for lang in langs:
  93. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  94. class TestDailymotionSubtitles(BaseTestSubtitles):
  95. url = 'http://www.dailymotion.com/video/xczg00'
  96. IE = DailymotionIE
  97. def test_no_writesubtitles(self):
  98. subtitles = self.getSubtitles()
  99. self.assertEqual(subtitles, None)
  100. def test_subtitles(self):
  101. self.DL.params['writesubtitles'] = True
  102. subtitles = self.getSubtitles()
  103. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  104. def test_subtitles_lang(self):
  105. self.DL.params['writesubtitles'] = True
  106. self.DL.params['subtitleslangs'] = ['fr']
  107. subtitles = self.getSubtitles()
  108. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  109. def test_allsubtitles(self):
  110. self.DL.params['writesubtitles'] = True
  111. self.DL.params['allsubtitles'] = True
  112. subtitles = self.getSubtitles()
  113. self.assertEqual(len(subtitles.keys()), 5)
  114. def test_list_subtitles(self):
  115. self.DL.expect_warning('Automatic Captions not supported by this server')
  116. self.DL.params['listsubtitles'] = True
  117. info_dict = self.getInfoDict()
  118. self.assertEqual(info_dict, None)
  119. def test_automatic_captions(self):
  120. self.DL.expect_warning('Automatic Captions not supported by this server')
  121. self.DL.params['writeautomaticsub'] = True
  122. self.DL.params['subtitleslang'] = ['en']
  123. subtitles = self.getSubtitles()
  124. self.assertTrue(len(subtitles.keys()) == 0)
  125. def test_nosubtitles(self):
  126. self.DL.expect_warning('video doesn\'t have subtitles')
  127. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  128. self.DL.params['writesubtitles'] = True
  129. self.DL.params['allsubtitles'] = True
  130. subtitles = self.getSubtitles()
  131. self.assertEqual(len(subtitles), 0)
  132. def test_multiple_langs(self):
  133. self.DL.params['writesubtitles'] = True
  134. langs = ['es', 'fr', 'de']
  135. self.DL.params['subtitleslangs'] = langs
  136. subtitles = self.getSubtitles()
  137. for lang in langs:
  138. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  139. class TestTedSubtitles(BaseTestSubtitles):
  140. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  141. IE = TEDIE
  142. def test_no_writesubtitles(self):
  143. subtitles = self.getSubtitles()
  144. self.assertEqual(subtitles, None)
  145. def test_subtitles(self):
  146. self.DL.params['writesubtitles'] = True
  147. subtitles = self.getSubtitles()
  148. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  149. def test_subtitles_lang(self):
  150. self.DL.params['writesubtitles'] = True
  151. self.DL.params['subtitleslangs'] = ['fr']
  152. subtitles = self.getSubtitles()
  153. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  154. def test_allsubtitles(self):
  155. self.DL.params['writesubtitles'] = True
  156. self.DL.params['allsubtitles'] = True
  157. subtitles = self.getSubtitles()
  158. self.assertTrue(len(subtitles.keys()) >= 28)
  159. def test_list_subtitles(self):
  160. self.DL.expect_warning('Automatic Captions not supported by this server')
  161. self.DL.params['listsubtitles'] = True
  162. info_dict = self.getInfoDict()
  163. self.assertEqual(info_dict, None)
  164. def test_automatic_captions(self):
  165. self.DL.expect_warning('Automatic Captions not supported by this server')
  166. self.DL.params['writeautomaticsub'] = True
  167. self.DL.params['subtitleslang'] = ['en']
  168. subtitles = self.getSubtitles()
  169. self.assertTrue(len(subtitles.keys()) == 0)
  170. def test_multiple_langs(self):
  171. self.DL.params['writesubtitles'] = True
  172. langs = ['es', 'fr', 'de']
  173. self.DL.params['subtitleslangs'] = langs
  174. subtitles = self.getSubtitles()
  175. for lang in langs:
  176. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  177. class TestBlipTVSubtitles(BaseTestSubtitles):
  178. url = 'http://blip.tv/a/a-6603250'
  179. IE = BlipTVIE
  180. def test_list_subtitles(self):
  181. self.DL.expect_warning('Automatic Captions not supported by this server')
  182. self.DL.params['listsubtitles'] = True
  183. info_dict = self.getInfoDict()
  184. self.assertEqual(info_dict, None)
  185. def test_allsubtitles(self):
  186. self.DL.expect_warning('Automatic Captions not supported by this server')
  187. self.DL.params['writesubtitles'] = True
  188. self.DL.params['allsubtitles'] = True
  189. subtitles = self.getSubtitles()
  190. self.assertEqual(set(subtitles.keys()), set(['en']))
  191. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  192. class TestVimeoSubtitles(BaseTestSubtitles):
  193. url = 'http://vimeo.com/76979871'
  194. IE = VimeoIE
  195. def test_no_writesubtitles(self):
  196. subtitles = self.getSubtitles()
  197. self.assertEqual(subtitles, None)
  198. def test_subtitles(self):
  199. self.DL.params['writesubtitles'] = True
  200. subtitles = self.getSubtitles()
  201. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  202. def test_subtitles_lang(self):
  203. self.DL.params['writesubtitles'] = True
  204. self.DL.params['subtitleslangs'] = ['fr']
  205. subtitles = self.getSubtitles()
  206. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  207. def test_allsubtitles(self):
  208. self.DL.params['writesubtitles'] = True
  209. self.DL.params['allsubtitles'] = True
  210. subtitles = self.getSubtitles()
  211. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  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_automatic_captions(self):
  218. self.DL.expect_warning('Automatic Captions not supported by this server')
  219. self.DL.params['writeautomaticsub'] = True
  220. self.DL.params['subtitleslang'] = ['en']
  221. subtitles = self.getSubtitles()
  222. self.assertTrue(len(subtitles.keys()) == 0)
  223. def test_nosubtitles(self):
  224. self.DL.expect_warning('video doesn\'t have subtitles')
  225. self.url = 'http://vimeo.com/56015672'
  226. self.DL.params['writesubtitles'] = True
  227. self.DL.params['allsubtitles'] = True
  228. subtitles = self.getSubtitles()
  229. self.assertEqual(len(subtitles), 0)
  230. def test_multiple_langs(self):
  231. self.DL.params['writesubtitles'] = True
  232. langs = ['es', 'fr', 'de']
  233. self.DL.params['subtitleslangs'] = langs
  234. subtitles = self.getSubtitles()
  235. for lang in langs:
  236. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  237. class TestWallaSubtitles(BaseTestSubtitles):
  238. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  239. IE = WallaIE
  240. def test_list_subtitles(self):
  241. self.DL.expect_warning('Automatic Captions not supported by this server')
  242. self.DL.params['listsubtitles'] = True
  243. info_dict = self.getInfoDict()
  244. self.assertEqual(info_dict, None)
  245. def test_allsubtitles(self):
  246. self.DL.expect_warning('Automatic Captions not supported by this server')
  247. self.DL.params['writesubtitles'] = True
  248. self.DL.params['allsubtitles'] = True
  249. subtitles = self.getSubtitles()
  250. self.assertEqual(set(subtitles.keys()), set(['heb']))
  251. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  252. def test_nosubtitles(self):
  253. self.DL.expect_warning('video doesn\'t have subtitles')
  254. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  255. self.DL.params['writesubtitles'] = True
  256. self.DL.params['allsubtitles'] = True
  257. subtitles = self.getSubtitles()
  258. self.assertEqual(len(subtitles), 0)
  259. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  260. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  261. IE = CeskaTelevizeIE
  262. def test_list_subtitles(self):
  263. self.DL.expect_warning('Automatic Captions not supported by this server')
  264. self.DL.params['listsubtitles'] = True
  265. info_dict = self.getInfoDict()
  266. self.assertEqual(info_dict, None)
  267. def test_allsubtitles(self):
  268. self.DL.expect_warning('Automatic Captions not supported by this server')
  269. self.DL.params['writesubtitles'] = True
  270. self.DL.params['allsubtitles'] = True
  271. subtitles = self.getSubtitles()
  272. self.assertEqual(set(subtitles.keys()), set(['cs']))
  273. self.assertEqual(md5(subtitles['cs']), 'cc3957b2c6dff1db71e5f2e83d467480')
  274. def test_nosubtitles(self):
  275. self.DL.expect_warning('video doesn\'t have subtitles')
  276. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  277. self.DL.params['writesubtitles'] = True
  278. self.DL.params['allsubtitles'] = True
  279. subtitles = self.getSubtitles()
  280. self.assertEqual(len(subtitles), 0)
  281. if __name__ == '__main__':
  282. unittest.main()