test_subtitles.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334
  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. )
  20. class BaseTestSubtitles(unittest.TestCase):
  21. url = None
  22. IE = None
  23. def setUp(self):
  24. self.DL = FakeYDL()
  25. self.ie = self.IE()
  26. self.DL.add_info_extractor(self.ie)
  27. def getInfoDict(self):
  28. info_dict = self.DL.extract_info(self.url, download=False)
  29. return info_dict
  30. def getSubtitles(self):
  31. info_dict = self.getInfoDict()
  32. subtitles = info_dict['requested_subtitles']
  33. if not subtitles:
  34. return subtitles
  35. for sub_info in subtitles.values():
  36. if sub_info.get('data') is None:
  37. uf = self.DL.urlopen(sub_info['url'])
  38. sub_info['data'] = uf.read().decode('utf-8')
  39. return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  40. class TestYoutubeSubtitles(BaseTestSubtitles):
  41. url = 'QRS8MkLhQmM'
  42. IE = YoutubeIE
  43. def test_youtube_subtitles(self):
  44. self.DL.params['writesubtitles'] = True
  45. subtitles = self.getSubtitles()
  46. self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
  47. def test_youtube_subtitles_lang(self):
  48. self.DL.params['writesubtitles'] = True
  49. self.DL.params['subtitleslangs'] = ['it']
  50. subtitles = self.getSubtitles()
  51. self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
  52. def test_youtube_allsubtitles(self):
  53. self.DL.params['writesubtitles'] = True
  54. self.DL.params['allsubtitles'] = True
  55. subtitles = self.getSubtitles()
  56. self.assertEqual(len(subtitles.keys()), 13)
  57. def test_youtube_subtitles_sbv_format(self):
  58. self.DL.params['writesubtitles'] = True
  59. self.DL.params['subtitlesformat'] = 'sbv'
  60. subtitles = self.getSubtitles()
  61. self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
  62. def test_youtube_subtitles_vtt_format(self):
  63. self.DL.params['writesubtitles'] = True
  64. self.DL.params['subtitlesformat'] = 'vtt'
  65. subtitles = self.getSubtitles()
  66. self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
  67. def test_youtube_list_subtitles(self):
  68. self.DL.expect_warning('Video doesn\'t have automatic captions')
  69. self.DL.params['listsubtitles'] = True
  70. info_dict = self.getInfoDict()
  71. self.assertEqual(info_dict, None)
  72. def test_youtube_automatic_captions(self):
  73. self.url = '8YoUxe5ncPo'
  74. self.DL.params['writeautomaticsub'] = True
  75. self.DL.params['subtitleslangs'] = ['it']
  76. subtitles = self.getSubtitles()
  77. self.assertTrue(subtitles['it'] is not None)
  78. def test_youtube_translated_subtitles(self):
  79. # This video has a subtitles track, which can be translated
  80. self.url = 'Ky9eprVWzlI'
  81. self.DL.params['writeautomaticsub'] = True
  82. self.DL.params['subtitleslangs'] = ['it']
  83. subtitles = self.getSubtitles()
  84. self.assertTrue(subtitles['it'] is not None)
  85. def test_youtube_nosubtitles(self):
  86. self.DL.expect_warning('video doesn\'t have subtitles')
  87. self.url = 'n5BB19UTcdA'
  88. self.DL.params['writesubtitles'] = True
  89. self.DL.params['allsubtitles'] = True
  90. subtitles = self.getSubtitles()
  91. self.assertEqual(len(subtitles), 0)
  92. def test_youtube_multiple_langs(self):
  93. self.url = 'QRS8MkLhQmM'
  94. self.DL.params['writesubtitles'] = True
  95. langs = ['it', 'fr', 'de']
  96. self.DL.params['subtitleslangs'] = langs
  97. subtitles = self.getSubtitles()
  98. for lang in langs:
  99. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  100. class TestDailymotionSubtitles(BaseTestSubtitles):
  101. url = 'http://www.dailymotion.com/video/xczg00'
  102. IE = DailymotionIE
  103. def test_subtitles(self):
  104. self.DL.params['writesubtitles'] = True
  105. subtitles = self.getSubtitles()
  106. self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  107. def test_subtitles_lang(self):
  108. self.DL.params['writesubtitles'] = True
  109. self.DL.params['subtitleslangs'] = ['fr']
  110. subtitles = self.getSubtitles()
  111. self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  112. def test_allsubtitles(self):
  113. self.DL.params['writesubtitles'] = True
  114. self.DL.params['allsubtitles'] = True
  115. subtitles = self.getSubtitles()
  116. self.assertTrue(len(subtitles.keys()) >= 6)
  117. def test_nosubtitles(self):
  118. self.DL.expect_warning('video doesn\'t have subtitles')
  119. self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  120. self.DL.params['writesubtitles'] = True
  121. self.DL.params['allsubtitles'] = True
  122. subtitles = self.getSubtitles()
  123. self.assertEqual(len(subtitles), 0)
  124. def test_multiple_langs(self):
  125. self.DL.params['writesubtitles'] = True
  126. langs = ['es', 'fr', 'de']
  127. self.DL.params['subtitleslangs'] = langs
  128. subtitles = self.getSubtitles()
  129. for lang in langs:
  130. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  131. class TestTedSubtitles(BaseTestSubtitles):
  132. url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  133. IE = TEDIE
  134. def test_no_writesubtitles(self):
  135. subtitles = self.getSubtitles()
  136. self.assertFalse(subtitles)
  137. def test_subtitles(self):
  138. self.DL.params['writesubtitles'] = True
  139. subtitles = self.getSubtitles()
  140. self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  141. def test_subtitles_lang(self):
  142. self.DL.params['writesubtitles'] = True
  143. self.DL.params['subtitleslangs'] = ['fr']
  144. subtitles = self.getSubtitles()
  145. self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  146. def test_allsubtitles(self):
  147. self.DL.params['writesubtitles'] = True
  148. self.DL.params['allsubtitles'] = True
  149. subtitles = self.getSubtitles()
  150. self.assertTrue(len(subtitles.keys()) >= 28)
  151. def test_list_subtitles(self):
  152. self.DL.params['listsubtitles'] = True
  153. info_dict = self.getInfoDict()
  154. self.assertEqual(info_dict, None)
  155. def test_multiple_langs(self):
  156. self.DL.params['writesubtitles'] = True
  157. langs = ['es', 'fr', 'de']
  158. self.DL.params['subtitleslangs'] = langs
  159. subtitles = self.getSubtitles()
  160. for lang in langs:
  161. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  162. class TestBlipTVSubtitles(BaseTestSubtitles):
  163. url = 'http://blip.tv/a/a-6603250'
  164. IE = BlipTVIE
  165. def test_list_subtitles(self):
  166. self.DL.params['listsubtitles'] = True
  167. info_dict = self.getInfoDict()
  168. self.assertEqual(info_dict, None)
  169. def test_allsubtitles(self):
  170. self.DL.params['writesubtitles'] = True
  171. self.DL.params['allsubtitles'] = True
  172. subtitles = self.getSubtitles()
  173. self.assertEqual(set(subtitles.keys()), set(['en']))
  174. self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
  175. class TestVimeoSubtitles(BaseTestSubtitles):
  176. url = 'http://vimeo.com/76979871'
  177. IE = VimeoIE
  178. def test_subtitles(self):
  179. self.DL.params['writesubtitles'] = True
  180. subtitles = self.getSubtitles()
  181. self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
  182. def test_subtitles_lang(self):
  183. self.DL.params['writesubtitles'] = True
  184. self.DL.params['subtitleslangs'] = ['fr']
  185. subtitles = self.getSubtitles()
  186. self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
  187. def test_allsubtitles(self):
  188. self.DL.params['writesubtitles'] = True
  189. self.DL.params['allsubtitles'] = True
  190. subtitles = self.getSubtitles()
  191. self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  192. def test_nosubtitles(self):
  193. self.DL.expect_warning('video doesn\'t have subtitles')
  194. self.url = 'http://vimeo.com/56015672'
  195. self.DL.params['writesubtitles'] = True
  196. self.DL.params['allsubtitles'] = True
  197. subtitles = self.getSubtitles()
  198. self.assertEqual(len(subtitles), 0)
  199. def test_multiple_langs(self):
  200. self.DL.params['writesubtitles'] = True
  201. langs = ['es', 'fr', 'de']
  202. self.DL.params['subtitleslangs'] = langs
  203. subtitles = self.getSubtitles()
  204. for lang in langs:
  205. self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  206. class TestWallaSubtitles(BaseTestSubtitles):
  207. url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  208. IE = WallaIE
  209. def test_list_subtitles(self):
  210. self.DL.expect_warning('Automatic Captions not supported by this server')
  211. self.DL.params['listsubtitles'] = True
  212. info_dict = self.getInfoDict()
  213. self.assertEqual(info_dict, None)
  214. def test_allsubtitles(self):
  215. self.DL.expect_warning('Automatic Captions not supported by this server')
  216. self.DL.params['writesubtitles'] = True
  217. self.DL.params['allsubtitles'] = True
  218. subtitles = self.getSubtitles()
  219. self.assertEqual(set(subtitles.keys()), set(['heb']))
  220. self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  221. def test_nosubtitles(self):
  222. self.DL.expect_warning('video doesn\'t have subtitles')
  223. self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  224. self.DL.params['writesubtitles'] = True
  225. self.DL.params['allsubtitles'] = True
  226. subtitles = self.getSubtitles()
  227. self.assertEqual(len(subtitles), 0)
  228. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  229. url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  230. IE = CeskaTelevizeIE
  231. def test_list_subtitles(self):
  232. self.DL.expect_warning('Automatic Captions not supported by this server')
  233. self.DL.params['listsubtitles'] = True
  234. info_dict = self.getInfoDict()
  235. self.assertEqual(info_dict, None)
  236. def test_allsubtitles(self):
  237. self.DL.expect_warning('Automatic Captions not supported by this server')
  238. self.DL.params['writesubtitles'] = True
  239. self.DL.params['allsubtitles'] = True
  240. subtitles = self.getSubtitles()
  241. self.assertEqual(set(subtitles.keys()), set(['cs']))
  242. self.assertTrue(len(subtitles['cs']) > 20000)
  243. def test_nosubtitles(self):
  244. self.DL.expect_warning('video doesn\'t have subtitles')
  245. self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  246. self.DL.params['writesubtitles'] = True
  247. self.DL.params['allsubtitles'] = True
  248. subtitles = self.getSubtitles()
  249. self.assertEqual(len(subtitles), 0)
  250. class TestLyndaSubtitles(BaseTestSubtitles):
  251. url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  252. IE = LyndaIE
  253. def test_allsubtitles(self):
  254. self.DL.params['writesubtitles'] = True
  255. self.DL.params['allsubtitles'] = True
  256. subtitles = self.getSubtitles()
  257. self.assertEqual(set(subtitles.keys()), set(['en']))
  258. self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  259. class TestNPOSubtitles(BaseTestSubtitles):
  260. url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  261. IE = NPOIE
  262. def test_allsubtitles(self):
  263. self.DL.params['writesubtitles'] = True
  264. self.DL.params['allsubtitles'] = True
  265. subtitles = self.getSubtitles()
  266. self.assertEqual(set(subtitles.keys()), set(['nl']))
  267. self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  268. if __name__ == '__main__':
  269. unittest.main()