test_subtitles.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308
  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_subtitles(self):
  42. self.DL.params['writesubtitles'] = True
  43. subtitles = self.getSubtitles()
  44. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  45. def test_youtube_subtitles_lang(self):
  46. self.DL.params['writesubtitles'] = True
  47. self.DL.params['subtitleslangs'] = ['it']
  48. subtitles = self.getSubtitles()
  49. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  50. def test_youtube_allsubtitles(self):
  51. self.DL.params['writesubtitles'] = True
  52. self.DL.params['allsubtitles'] = True
  53. subtitles = self.getSubtitles()
  54. self.assertEqual(len(subtitles.keys()), 13)
  55. def test_youtube_subtitles_sbv_format(self):
  56. self.DL.params['writesubtitles'] = True
  57. self.DL.params['subtitlesformat'] = 'sbv'
  58. subtitles = self.getSubtitles()
  59. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  60. def test_youtube_subtitles_vtt_format(self):
  61. self.DL.params['writesubtitles'] = True
  62. self.DL.params['subtitlesformat'] = 'vtt'
  63. subtitles = self.getSubtitles()
  64. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  65. def test_youtube_list_subtitles(self):
  66. self.DL.expect_warning('Video doesn\'t have automatic captions')
  67. self.DL.params['listsubtitles'] = True
  68. info_dict = self.getInfoDict()
  69. self.assertEqual(info_dict, None)
  70. def test_youtube_automatic_captions(self):
  71. self.url = '8YoUxe5ncPo'
  72. self.DL.params['writeautomaticsub'] = True
  73. self.DL.params['subtitleslangs'] = ['it']
  74. subtitles = self.getSubtitles()
  75. self.assertTrue(subtitles['it'] is not None)
  76. def test_youtube_translated_subtitles(self):
  77. # This video has a subtitles track, which can be translated
  78. self.url = 'Ky9eprVWzlI'
  79. self.DL.params['writeautomaticsub'] = True
  80. self.DL.params['subtitleslangs'] = ['it']
  81. subtitles = self.getSubtitles()
  82. self.assertTrue(subtitles['it'] is not None)
  83. def test_youtube_nosubtitles(self):
  84. self.DL.expect_warning('video doesn\'t have subtitles')
  85. self.url = 'n5BB19UTcdA'
  86. self.DL.params['writesubtitles'] = True
  87. self.DL.params['allsubtitles'] = True
  88. subtitles = self.getSubtitles()
  89. self.assertEqual(len(subtitles), 0)
  90. def test_youtube_multiple_langs(self):
  91. self.url = 'QRS8MkLhQmM'
  92. self.DL.params['writesubtitles'] = True
  93. langs = ['it', 'fr', 'de']
  94. self.DL.params['subtitleslangs'] = langs
  95. subtitles = self.getSubtitles()
  96. for lang in langs:
  97. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  98. class TestDailymotionSubtitles(BaseTestSubtitles):
  99. url = 'http://www.dailymotion.com/video/xczg00'
  100. IE = DailymotionIE
  101. def test_subtitles(self):
  102. self.DL.params['writesubtitles'] = True
  103. subtitles = self.getSubtitles()
  104. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  105. def test_subtitles_lang(self):
  106. self.DL.params['writesubtitles'] = True
  107. self.DL.params['subtitleslangs'] = ['fr']
  108. subtitles = self.getSubtitles()
  109. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  110. def test_allsubtitles(self):
  111. self.DL.params['writesubtitles'] = True
  112. self.DL.params['allsubtitles'] = True
  113. subtitles = self.getSubtitles()
  114. self.assertTrue(len(subtitles.keys()) >= 6)
  115. def test_nosubtitles(self):
  116. self.DL.expect_warning('video doesn\'t have subtitles')
  117. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  118. self.DL.params['writesubtitles'] = True
  119. self.DL.params['allsubtitles'] = True
  120. subtitles = self.getSubtitles()
  121. self.assertEqual(len(subtitles), 0)
  122. def test_multiple_langs(self):
  123. self.DL.params['writesubtitles'] = True
  124. langs = ['es', 'fr', 'de']
  125. self.DL.params['subtitleslangs'] = langs
  126. subtitles = self.getSubtitles()
  127. for lang in langs:
  128. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  129. class TestTedSubtitles(BaseTestSubtitles):
  130. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  131. IE = TEDIE
  132. def test_no_writesubtitles(self):
  133. subtitles = self.getSubtitles()
  134. self.assertFalse(subtitles)
  135. def test_subtitles(self):
  136. self.DL.params['writesubtitles'] = True
  137. subtitles = self.getSubtitles()
  138. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  139. def test_subtitles_lang(self):
  140. self.DL.params['writesubtitles'] = True
  141. self.DL.params['subtitleslangs'] = ['fr']
  142. subtitles = self.getSubtitles()
  143. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  144. def test_allsubtitles(self):
  145. self.DL.params['writesubtitles'] = True
  146. self.DL.params['allsubtitles'] = True
  147. subtitles = self.getSubtitles()
  148. self.assertTrue(len(subtitles.keys()) >= 28)
  149. def test_list_subtitles(self):
  150. self.DL.params['listsubtitles'] = True
  151. info_dict = self.getInfoDict()
  152. self.assertEqual(info_dict, None)
  153. def test_multiple_langs(self):
  154. self.DL.params['writesubtitles'] = True
  155. langs = ['es', 'fr', 'de']
  156. self.DL.params['subtitleslangs'] = langs
  157. subtitles = self.getSubtitles()
  158. for lang in langs:
  159. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  160. class TestBlipTVSubtitles(BaseTestSubtitles):
  161. url = 'http://blip.tv/a/a-6603250'
  162. IE = BlipTVIE
  163. def test_list_subtitles(self):
  164. self.DL.params['listsubtitles'] = True
  165. info_dict = self.getInfoDict()
  166. self.assertEqual(info_dict, None)
  167. def test_allsubtitles(self):
  168. self.DL.params['writesubtitles'] = True
  169. self.DL.params['allsubtitles'] = True
  170. subtitles = self.getSubtitles()
  171. self.assertEqual(set(subtitles.keys()), set(['en']))
  172. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  173. class TestVimeoSubtitles(BaseTestSubtitles):
  174. url = 'http://vimeo.com/76979871'
  175. IE = VimeoIE
  176. def test_subtitles(self):
  177. self.DL.params['writesubtitles'] = True
  178. subtitles = self.getSubtitles()
  179. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  180. def test_subtitles_lang(self):
  181. self.DL.params['writesubtitles'] = True
  182. self.DL.params['subtitleslangs'] = ['fr']
  183. subtitles = self.getSubtitles()
  184. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  185. def test_allsubtitles(self):
  186. self.DL.params['writesubtitles'] = True
  187. self.DL.params['allsubtitles'] = True
  188. subtitles = self.getSubtitles()
  189. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  190. def test_nosubtitles(self):
  191. self.DL.expect_warning('video doesn\'t have subtitles')
  192. self.url = 'http://vimeo.com/56015672'
  193. self.DL.params['writesubtitles'] = True
  194. self.DL.params['allsubtitles'] = True
  195. subtitles = self.getSubtitles()
  196. self.assertEqual(len(subtitles), 0)
  197. def test_multiple_langs(self):
  198. self.DL.params['writesubtitles'] = True
  199. langs = ['es', 'fr', 'de']
  200. self.DL.params['subtitleslangs'] = langs
  201. subtitles = self.getSubtitles()
  202. for lang in langs:
  203. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  204. class TestWallaSubtitles(BaseTestSubtitles):
  205. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  206. IE = WallaIE
  207. def test_list_subtitles(self):
  208. self.DL.expect_warning('Automatic Captions not supported by this server')
  209. self.DL.params['listsubtitles'] = True
  210. info_dict = self.getInfoDict()
  211. self.assertEqual(info_dict, None)
  212. def test_allsubtitles(self):
  213. self.DL.expect_warning('Automatic Captions not supported by this server')
  214. self.DL.params['writesubtitles'] = True
  215. self.DL.params['allsubtitles'] = True
  216. subtitles = self.getSubtitles()
  217. self.assertEqual(set(subtitles.keys()), set(['heb']))
  218. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  219. def test_nosubtitles(self):
  220. self.DL.expect_warning('video doesn\'t have subtitles')
  221. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  222. self.DL.params['writesubtitles'] = True
  223. self.DL.params['allsubtitles'] = True
  224. subtitles = self.getSubtitles()
  225. self.assertEqual(len(subtitles), 0)
  226. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  227. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  228. IE = CeskaTelevizeIE
  229. def test_list_subtitles(self):
  230. self.DL.expect_warning('Automatic Captions not supported by this server')
  231. self.DL.params['listsubtitles'] = True
  232. info_dict = self.getInfoDict()
  233. self.assertEqual(info_dict, None)
  234. def test_allsubtitles(self):
  235. self.DL.expect_warning('Automatic Captions not supported by this server')
  236. self.DL.params['writesubtitles'] = True
  237. self.DL.params['allsubtitles'] = True
  238. subtitles = self.getSubtitles()
  239. self.assertEqual(set(subtitles.keys()), set(['cs']))
  240. self.assertTrue(len(subtitles['cs']) > 20000)
  241. def test_nosubtitles(self):
  242. self.DL.expect_warning('video doesn\'t have subtitles')
  243. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  244. self.DL.params['writesubtitles'] = True
  245. self.DL.params['allsubtitles'] = True
  246. subtitles = self.getSubtitles()
  247. self.assertEqual(len(subtitles), 0)
  248. if __name__ == '__main__':
  249. unittest.main()