logo

youtube-dl

[mirror] Download/Watch videos from video hostersgit clone https://hacktivis.me/git/mirror/youtube-dl.git

test_subtitles.py (15801B)

    

  1. #!/usr/bin/env python
  2. from __future__ import unicode_literals
  4. # Allow direct execution
  5. import os
  6. import sys
  7. import unittest
  8. sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  10. from test.helper import FakeYDL, md5
  13. from youtube_dl.extractor import (
  14.     YoutubeIE,
  15.     DailymotionIE,
  16.     TEDIE,
  17.     VimeoIE,
  18.     WallaIE,
  19.     CeskaTelevizeIE,
  20.     LyndaIE,
  21.     NPOIE,
  22.     ComedyCentralIE,
  23.     NRKTVIE,
  24.     RaiPlayIE,
  25.     VikiIE,
  26.     ThePlatformIE,
  27.     ThePlatformFeedIE,
  28.     RTVEALaCartaIE,
  29.     DemocracynowIE,
  30. )
  33. class BaseTestSubtitles(unittest.TestCase):
  34.     url = None
  35.     IE = None
  37.     def setUp(self):
  38.         self.DL = FakeYDL()
  39.         self.ie = self.IE()
  40.         self.DL.add_info_extractor(self.ie)
  41.         if not self.IE.working():
  42.             print('Skipping: %s marked as not _WORKING' % self.IE.ie_key())
  43.             self.skipTest('IE marked as not _WORKING')
  45.     def getInfoDict(self):
  46.         info_dict = self.DL.extract_info(self.url, download=False)
  47.         return info_dict
  49.     def getSubtitles(self):
  50.         info_dict = self.getInfoDict()
  51.         subtitles = info_dict['requested_subtitles']
  52.         if not subtitles:
  53.             return subtitles
  54.         for sub_info in subtitles.values():
  55.             if sub_info.get('data') is None:
  56.                 uf = self.DL.urlopen(sub_info['url'])
  57.                 sub_info['data'] = uf.read().decode('utf-8')
  58.         return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
  61. class TestYoutubeSubtitles(BaseTestSubtitles):
  62.     # Available subtitles for QRS8MkLhQmM:
  63.     # Language formats
  64.     # ru       vtt, ttml, srv3, srv2, srv1, json3
  65.     # fr       vtt, ttml, srv3, srv2, srv1, json3
  66.     # en       vtt, ttml, srv3, srv2, srv1, json3
  67.     # nl       vtt, ttml, srv3, srv2, srv1, json3
  68.     # de       vtt, ttml, srv3, srv2, srv1, json3
  69.     # ko       vtt, ttml, srv3, srv2, srv1, json3
  70.     # it       vtt, ttml, srv3, srv2, srv1, json3
  71.     # zh-Hant  vtt, ttml, srv3, srv2, srv1, json3
  72.     # hi       vtt, ttml, srv3, srv2, srv1, json3
  73.     # pt-BR    vtt, ttml, srv3, srv2, srv1, json3
  74.     # es-MX    vtt, ttml, srv3, srv2, srv1, json3
  75.     # ja       vtt, ttml, srv3, srv2, srv1, json3
  76.     # pl       vtt, ttml, srv3, srv2, srv1, json3
  77.     url = 'QRS8MkLhQmM'
  78.     IE = YoutubeIE
  80.     def test_youtube_allsubtitles(self):
  81.         self.DL.params['writesubtitles'] = True
  82.         self.DL.params['allsubtitles'] = True
  83.         subtitles = self.getSubtitles()
  84.         self.assertEqual(len(subtitles.keys()), 13)
  85.         self.assertEqual(md5(subtitles['en']), 'ae1bd34126571a77aabd4d276b28044d')
  86.         self.assertEqual(md5(subtitles['it']), '0e0b667ba68411d88fd1c5f4f4eab2f9')
  87.         for lang in ['fr', 'de']:
  88.             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  90.     def _test_subtitles_format(self, fmt, md5_hash, lang='en'):
  91.         self.DL.params['writesubtitles'] = True
  92.         self.DL.params['subtitlesformat'] = fmt
  93.         subtitles = self.getSubtitles()
  94.         self.assertEqual(md5(subtitles[lang]), md5_hash)
  96.     def test_youtube_subtitles_ttml_format(self):
  97.         self._test_subtitles_format('ttml', 'c97ddf1217390906fa9fbd34901f3da2')
  99.     def test_youtube_subtitles_vtt_format(self):
  100.         self._test_subtitles_format('vtt', 'ae1bd34126571a77aabd4d276b28044d')
  102.     def test_youtube_subtitles_json3_format(self):
  103.         self._test_subtitles_format('json3', '688dd1ce0981683867e7fe6fde2a224b')
  105.     def _test_automatic_captions(self, url, lang):
  106.         self.url = url
  107.         self.DL.params['writeautomaticsub'] = True
  108.         self.DL.params['subtitleslangs'] = [lang]
  109.         subtitles = self.getSubtitles()
  110.         self.assertTrue(subtitles[lang] is not None)
  112.     def test_youtube_automatic_captions(self):
  113.         # Available automatic captions for 8YoUxe5ncPo:
  114.         # Language formats (all in vtt, ttml, srv3, srv2, srv1, json3)
  115.         # gu, zh-Hans, zh-Hant, gd, ga, gl, lb, la, lo, tt, tr,
  116.         # lv, lt, tk, th, tg, te, fil, haw, yi, ceb, yo, de, da,
  117.         # el, eo, en, eu, et, es, ru, rw, ro, bn, be, bg, uk, jv,
  118.         # bs, ja, or, xh, co, ca, cy, cs, ps, pt, pa, vi, pl, hy,
  119.         # hr, ht, hu, hmn, hi, ha, mg, uz, ml, mn, mi, mk, ur,
  120.         # mt, ms, mr, ug, ta, my, af, sw, is, am,
  121.         #                                         *it*, iw, sv, ar,
  122.         # su, zu, az, id, ig, nl, no, ne, ny, fr, ku, fy, fa, fi,
  123.         # ka, kk, sr, sq, ko, kn, km, st, sk, si, so, sn, sm, sl,
  124.         # ky, sd
  125.         # ...
  126.         self._test_automatic_captions('8YoUxe5ncPo', 'it')
  128.     @unittest.skip('ASR subs all in all supported langs now')
  129.     def test_youtube_translated_subtitles(self):
  130.         # This video has a subtitles track, which can be translated (#4555)
  131.         self._test_automatic_captions('Ky9eprVWzlI', 'it')
  133.     def test_youtube_nosubtitles(self):
  134.         self.DL.expect_warning('video doesn\'t have subtitles')
  135.         # Available automatic captions for 8YoUxe5ncPo:
  136.         # ...
  137.         # 8YoUxe5ncPo has no subtitles
  138.         self.url = '8YoUxe5ncPo'
  139.         self.DL.params['writesubtitles'] = True
  140.         self.DL.params['allsubtitles'] = True
  141.         subtitles = self.getSubtitles()
  142.         self.assertFalse(subtitles)
  145. class TestDailymotionSubtitles(BaseTestSubtitles):
  146.     url = 'http://www.dailymotion.com/video/xczg00'
  147.     IE = DailymotionIE
  149.     def test_allsubtitles(self):
  150.         self.DL.params['writesubtitles'] = True
  151.         self.DL.params['allsubtitles'] = True
  152.         subtitles = self.getSubtitles()
  153.         self.assertTrue(len(subtitles.keys()) >= 6)
  154.         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
  155.         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
  156.         for lang in ['es', 'fr', 'de']:
  157.             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  159.     def test_nosubtitles(self):
  160.         self.DL.expect_warning('video doesn\'t have subtitles')
  161.         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
  162.         self.DL.params['writesubtitles'] = True
  163.         self.DL.params['allsubtitles'] = True
  164.         subtitles = self.getSubtitles()
  165.         self.assertFalse(subtitles)
  168. @unittest.skip('IE broken')
  169. class TestTedSubtitles(BaseTestSubtitles):
  170.     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
  171.     IE = TEDIE
  173.     def test_allsubtitles(self):
  174.         self.DL.params['writesubtitles'] = True
  175.         self.DL.params['allsubtitles'] = True
  176.         subtitles = self.getSubtitles()
  177.         self.assertTrue(len(subtitles.keys()) >= 28)
  178.         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
  179.         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
  180.         for lang in ['es', 'fr', 'de']:
  181.             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
  184. class TestVimeoSubtitles(BaseTestSubtitles):
  185.     url = 'http://vimeo.com/76979871'
  186.     IE = VimeoIE
  188.     def test_allsubtitles(self):
  189.         self.DL.params['writesubtitles'] = True
  190.         self.DL.params['allsubtitles'] = True
  191.         subtitles = self.getSubtitles()
  192.         self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
  193.         self.assertEqual(md5(subtitles['en']), '386cbc9320b94e25cb364b97935e5dd1')
  194.         self.assertEqual(md5(subtitles['fr']), 'c9b69eef35bc6641c0d4da8a04f9dfac')
  196.     def test_nosubtitles(self):
  197.         self.DL.expect_warning('video doesn\'t have subtitles')
  198.         self.url = 'http://vimeo.com/68093876'
  199.         self.DL.params['writesubtitles'] = True
  200.         self.DL.params['allsubtitles'] = True
  201.         subtitles = self.getSubtitles()
  202.         self.assertFalse(subtitles)
  205. @unittest.skip('IE broken')
  206. class TestWallaSubtitles(BaseTestSubtitles):
  207.     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
  208.     IE = WallaIE
  210.     def test_allsubtitles(self):
  211.         self.DL.expect_warning('Automatic Captions not supported by this server')
  212.         self.DL.params['writesubtitles'] = True
  213.         self.DL.params['allsubtitles'] = True
  214.         subtitles = self.getSubtitles()
  215.         self.assertEqual(set(subtitles.keys()), set(['heb']))
  216.         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
  218.     def test_nosubtitles(self):
  219.         self.DL.expect_warning('video doesn\'t have subtitles')
  220.         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
  221.         self.DL.params['writesubtitles'] = True
  222.         self.DL.params['allsubtitles'] = True
  223.         subtitles = self.getSubtitles()
  224.         self.assertFalse(subtitles)
  227. @unittest.skip('IE broken')
  228. class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
  229.     url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
  230.     IE = CeskaTelevizeIE
  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(['cs']))
  238.         self.assertTrue(len(subtitles['cs']) > 20000)
  240.     def test_nosubtitles(self):
  241.         self.DL.expect_warning('video doesn\'t have subtitles')
  242.         self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
  243.         self.DL.params['writesubtitles'] = True
  244.         self.DL.params['allsubtitles'] = True
  245.         subtitles = self.getSubtitles()
  246.         self.assertFalse(subtitles)
  249. @unittest.skip('IE broken')
  250. class TestLyndaSubtitles(BaseTestSubtitles):
  251.     url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
  252.     IE = LyndaIE
  254.     def test_allsubtitles(self):
  255.         self.DL.params['writesubtitles'] = True
  256.         self.DL.params['allsubtitles'] = True
  257.         subtitles = self.getSubtitles()
  258.         self.assertEqual(set(subtitles.keys()), set(['en']))
  259.         self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
  262. @unittest.skip('IE broken')
  263. class TestNPOSubtitles(BaseTestSubtitles):
  264.     url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
  265.     IE = NPOIE
  267.     def test_allsubtitles(self):
  268.         self.DL.params['writesubtitles'] = True
  269.         self.DL.params['allsubtitles'] = True
  270.         subtitles = self.getSubtitles()
  271.         self.assertEqual(set(subtitles.keys()), set(['nl']))
  272.         self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
  275. @unittest.skip('IE broken')
  276. class TestMTVSubtitles(BaseTestSubtitles):
  277.     url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
  278.     IE = ComedyCentralIE
  280.     def getInfoDict(self):
  281.         return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
  283.     def test_allsubtitles(self):
  284.         self.DL.params['writesubtitles'] = True
  285.         self.DL.params['allsubtitles'] = True
  286.         subtitles = self.getSubtitles()
  287.         self.assertEqual(set(subtitles.keys()), set(['en']))
  288.         self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
  291. class TestNRKSubtitles(BaseTestSubtitles):
  292.     url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
  293.     IE = NRKTVIE
  295.     def test_allsubtitles(self):
  296.         self.DL.params['writesubtitles'] = True
  297.         self.DL.params['allsubtitles'] = True
  298.         self.DL.params['format'] = 'best/bestvideo'
  299.         subtitles = self.getSubtitles()
  300.         self.assertEqual(set(subtitles.keys()), set(['nb-ttv']))
  301.         self.assertEqual(md5(subtitles['nb-ttv']), '67e06ff02d0deaf975e68f6cb8f6a149')
  304. class TestRaiPlaySubtitles(BaseTestSubtitles):
  305.     IE = RaiPlayIE
  307.     def test_subtitles_key(self):
  308.         self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
  309.         self.DL.params['writesubtitles'] = True
  310.         self.DL.params['allsubtitles'] = True
  311.         subtitles = self.getSubtitles()
  312.         self.assertEqual(set(subtitles.keys()), set(['it']))
  313.         self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
  315.     def test_subtitles_array_key(self):
  316.         self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
  317.         self.DL.params['writesubtitles'] = True
  318.         self.DL.params['allsubtitles'] = True
  319.         subtitles = self.getSubtitles()
  320.         self.assertEqual(set(subtitles.keys()), set(['it']))
  321.         self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
  324. @unittest.skip('IE broken - DRM only')
  325. class TestVikiSubtitles(BaseTestSubtitles):
  326.     url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
  327.     IE = VikiIE
  329.     def test_allsubtitles(self):
  330.         self.DL.params['writesubtitles'] = True
  331.         self.DL.params['allsubtitles'] = True
  332.         subtitles = self.getSubtitles()
  333.         self.assertEqual(set(subtitles.keys()), set(['en']))
  334.         self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
  337. class TestThePlatformSubtitles(BaseTestSubtitles):
  338.     # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
  339.     # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
  340.     url = 'theplatform:JFUjUE1_ehvq'
  341.     IE = ThePlatformIE
  343.     def test_allsubtitles(self):
  344.         self.DL.params['writesubtitles'] = True
  345.         self.DL.params['allsubtitles'] = True
  346.         subtitles = self.getSubtitles()
  347.         self.assertEqual(set(subtitles.keys()), set(['en']))
  348.         self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
  351. @unittest.skip('IE broken')
  352. class TestThePlatformFeedSubtitles(BaseTestSubtitles):
  353.     url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
  354.     IE = ThePlatformFeedIE
  356.     def test_allsubtitles(self):
  357.         self.DL.params['writesubtitles'] = True
  358.         self.DL.params['allsubtitles'] = True
  359.         subtitles = self.getSubtitles()
  360.         self.assertEqual(set(subtitles.keys()), set(['en']))
  361.         self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
  364. class TestRtveSubtitles(BaseTestSubtitles):
  365.     url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
  366.     IE = RTVEALaCartaIE
  368.     def test_allsubtitles(self):
  369.         print('Skipping, only available from Spain')
  370.         return
  371.         self.DL.params['writesubtitles'] = True
  372.         self.DL.params['allsubtitles'] = True
  373.         subtitles = self.getSubtitles()
  374.         self.assertEqual(set(subtitles.keys()), set(['es']))
  375.         self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
  378. class TestDemocracynowSubtitles(BaseTestSubtitles):
  379.     url = 'http://www.democracynow.org/shows/2015/7/3'
  380.     IE = DemocracynowIE
  382.     def test_allsubtitles(self):
  383.         self.DL.params['writesubtitles'] = True
  384.         self.DL.params['allsubtitles'] = True
  385.         subtitles = self.getSubtitles()
  386.         self.assertEqual(set(subtitles.keys()), set(['en']))
  387.         self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')
  389.     def test_subtitles_in_page(self):
  390.         self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
  391.         self.DL.params['writesubtitles'] = True
  392.         self.DL.params['allsubtitles'] = True
  393.         subtitles = self.getSubtitles()
  394.         self.assertEqual(set(subtitles.keys()), set(['en']))
  395.         self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')
  398. if __name__ == '__main__':
  399.     unittest.main()