test_subtitles.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332
  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_no_writesubtitles(self):
  106. subtitles = self.getSubtitles()
  107. self.assertEqual(subtitles, None)
  108. def test_subtitles(self):
  109. self.DL.params['writesubtitles'] = True
  110. subtitles = self.getSubtitles()
  111. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  112. def test_subtitles_lang(self):
  113. self.DL.params['writesubtitles'] = True
  114. self.DL.params['subtitleslangs'] = ['fr']
  115. subtitles = self.getSubtitles()
  116. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  117. def test_allsubtitles(self):
  118. self.DL.params['writesubtitles'] = True
  119. self.DL.params['allsubtitles'] = True
  120. subtitles = self.getSubtitles()
  121. self.assertTrue(len(subtitles.keys()) >= 6)
  122. def test_list_subtitles(self):
  123. self.DL.expect_warning('Automatic Captions not supported by this server')
  124. self.DL.params['listsubtitles'] = True
  125. info_dict = self.getInfoDict()
  126. self.assertEqual(info_dict, None)
  127. def test_automatic_captions(self):
  128. self.DL.expect_warning('Automatic Captions not supported by this server')
  129. self.DL.params['writeautomaticsub'] = True
  130. self.DL.params['subtitleslang'] = ['en']
  131. subtitles = self.getSubtitles()
  132. self.assertTrue(len(subtitles.keys()) == 0)
  133. def test_nosubtitles(self):
  134. self.DL.expect_warning('video doesn\'t have subtitles')
  135. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  136. self.DL.params['writesubtitles'] = True
  137. self.DL.params['allsubtitles'] = True
  138. subtitles = self.getSubtitles()
  139. self.assertEqual(len(subtitles), 0)
  140. def test_multiple_langs(self):
  141. self.DL.params['writesubtitles'] = True
  142. langs = ['es', 'fr', 'de']
  143. self.DL.params['subtitleslangs'] = langs
  144. subtitles = self.getSubtitles()
  145. for lang in langs:
  146. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  147. class TestTedSubtitles(BaseTestSubtitles):
  148. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  149. IE = TEDIE
  150. def test_no_writesubtitles(self):
  151. subtitles = self.getSubtitles()
  152. self.assertFalse(subtitles)
  153. def test_subtitles(self):
  154. self.DL.params['writesubtitles'] = True
  155. subtitles = self.getSubtitles()
  156. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  157. def test_subtitles_lang(self):
  158. self.DL.params['writesubtitles'] = True
  159. self.DL.params['subtitleslangs'] = ['fr']
  160. subtitles = self.getSubtitles()
  161. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  162. def test_allsubtitles(self):
  163. self.DL.params['writesubtitles'] = True
  164. self.DL.params['allsubtitles'] = True
  165. subtitles = self.getSubtitles()
  166. self.assertTrue(len(subtitles.keys()) >= 28)
  167. def test_list_subtitles(self):
  168. self.DL.params['listsubtitles'] = True
  169. info_dict = self.getInfoDict()
  170. self.assertEqual(info_dict, None)
  171. def test_multiple_langs(self):
  172. self.DL.params['writesubtitles'] = True
  173. langs = ['es', 'fr', 'de']
  174. self.DL.params['subtitleslangs'] = langs
  175. subtitles = self.getSubtitles()
  176. for lang in langs:
  177. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  178. class TestBlipTVSubtitles(BaseTestSubtitles):
  179. url = 'http://blip.tv/a/a-6603250'
  180. IE = BlipTVIE
  181. def test_list_subtitles(self):
  182. self.DL.expect_warning('Automatic Captions not supported by this server')
  183. self.DL.params['listsubtitles'] = True
  184. info_dict = self.getInfoDict()
  185. self.assertEqual(info_dict, None)
  186. def test_allsubtitles(self):
  187. self.DL.expect_warning('Automatic Captions not supported by this server')
  188. self.DL.params['writesubtitles'] = True
  189. self.DL.params['allsubtitles'] = True
  190. subtitles = self.getSubtitles()
  191. self.assertEqual(set(subtitles.keys()), set(['en']))
  192. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  193. class TestVimeoSubtitles(BaseTestSubtitles):
  194. url = 'http://vimeo.com/76979871'
  195. IE = VimeoIE
  196. def test_subtitles(self):
  197. self.DL.params['writesubtitles'] = True
  198. subtitles = self.getSubtitles()
  199. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  200. def test_subtitles_lang(self):
  201. self.DL.params['writesubtitles'] = True
  202. self.DL.params['subtitleslangs'] = ['fr']
  203. subtitles = self.getSubtitles()
  204. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  205. def test_allsubtitles(self):
  206. self.DL.params['writesubtitles'] = True
  207. self.DL.params['allsubtitles'] = True
  208. subtitles = self.getSubtitles()
  209. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  210. def test_nosubtitles(self):
  211. self.DL.expect_warning('video doesn\'t have subtitles')
  212. self.url = 'http://vimeo.com/56015672'
  213. self.DL.params['writesubtitles'] = True
  214. self.DL.params['allsubtitles'] = True
  215. subtitles = self.getSubtitles()
  216. self.assertEqual(len(subtitles), 0)
  217. def test_multiple_langs(self):
  218. self.DL.params['writesubtitles'] = True
  219. langs = ['es', 'fr', 'de']
  220. self.DL.params['subtitleslangs'] = langs
  221. subtitles = self.getSubtitles()
  222. for lang in langs:
  223. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  224. class TestWallaSubtitles(BaseTestSubtitles):
  225. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  226. IE = WallaIE
  227. def test_list_subtitles(self):
  228. self.DL.expect_warning('Automatic Captions not supported by this server')
  229. self.DL.params['listsubtitles'] = True
  230. info_dict = self.getInfoDict()
  231. self.assertEqual(info_dict, None)
  232. def test_allsubtitles(self):
  233. self.DL.expect_warning('Automatic Captions not supported by this server')
  234. self.DL.params['writesubtitles'] = True
  235. self.DL.params['allsubtitles'] = True
  236. subtitles = self.getSubtitles()
  237. self.assertEqual(set(subtitles.keys()), set(['heb']))
  238. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  239. def test_nosubtitles(self):
  240. self.DL.expect_warning('video doesn\'t have subtitles')
  241. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  242. self.DL.params['writesubtitles'] = True
  243. self.DL.params['allsubtitles'] = True
  244. subtitles = self.getSubtitles()
  245. self.assertEqual(len(subtitles), 0)
  246. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  247. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  248. IE = CeskaTelevizeIE
  249. def test_list_subtitles(self):
  250. self.DL.expect_warning('Automatic Captions not supported by this server')
  251. self.DL.params['listsubtitles'] = True
  252. info_dict = self.getInfoDict()
  253. self.assertEqual(info_dict, None)
  254. def test_allsubtitles(self):
  255. self.DL.expect_warning('Automatic Captions not supported by this server')
  256. self.DL.params['writesubtitles'] = True
  257. self.DL.params['allsubtitles'] = True
  258. subtitles = self.getSubtitles()
  259. self.assertEqual(set(subtitles.keys()), set(['cs']))
  260. self.assertTrue(len(subtitles['cs']) > 20000)
  261. def test_nosubtitles(self):
  262. self.DL.expect_warning('video doesn\'t have subtitles')
  263. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  264. self.DL.params['writesubtitles'] = True
  265. self.DL.params['allsubtitles'] = True
  266. subtitles = self.getSubtitles()
  267. self.assertEqual(len(subtitles), 0)
  268. if __name__ == '__main__':
  269. unittest.main()