test_subtitles.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315
  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()
  24. self.DL.add_info_extractor(self.ie)
  25. def getInfoDict(self):
  26. info_dict = self.DL.extract_info(self.url, download=False)
  27. return info_dict
  28. def getSubtitles(self):
  29. info_dict = self.getInfoDict()
  30. subtitles = info_dict['requested_subtitles']
  31. if not subtitles:
  32. return subtitles
  33. for sub_info in subtitles.values():
  34. if sub_info.get('data') is None:
  35. uf = self.DL.urlopen(sub_info['url'])
  36. sub_info['data'] = uf.read().decode('utf-8')
  37. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  38. class TestYoutubeSubtitles(BaseTestSubtitles):
  39. url = 'QRS8MkLhQmM'
  40. IE = YoutubeIE
  41. def test_youtube_no_writesubtitles(self):
  42. self.DL.params['writesubtitles'] = False
  43. subtitles = self.getSubtitles()
  44. self.assertEqual(subtitles, None)
  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.expect_warning('Automatic Captions not supported by this server')
  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.expect_warning('Automatic Captions not supported by this server')
  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. if __name__ == '__main__':
  255. unittest.main()