test/test_subtitles.py

   1 #!/usr/bin/env python3
   2
   3 # Allow direct execution
   4 import os
   5 import sys
   6 import unittest
   7
   8 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
   9
  10
  11 from test.helper import FakeYDL, is_download_test, md5
  12 from yt_dlp.extractor import (
  13     NPOIE,
  14     NRKTVIE,
  15     PBSIE,
  16     CeskaTelevizeIE,
  17     ComedyCentralIE,
  18     DailymotionIE,
  19     DemocracynowIE,
  20     LyndaIE,
  21     RaiPlayIE,
  22     RTVEALaCartaIE,
  23     TedTalkIE,
  24     ThePlatformFeedIE,
  25     ThePlatformIE,
  26     VikiIE,
  27     VimeoIE,
  28     WallaIE,
  29     YoutubeIE,
  30 )
  31
  32
  33 @is_download_test
  34 class BaseTestSubtitles(unittest.TestCase):
  35     url = None
  36     IE = None
  37
  38     def setUp(self):
  39         self.DL = FakeYDL()
  40         self.ie = self.IE()
  41         self.DL.add_info_extractor(self.ie)
  42         if not self.IE.working():
  43             print(f'Skipping: {self.IE.ie_key()} marked as not _WORKING')
  44             self.skipTest('IE marked as not _WORKING')
  45
  46     def getInfoDict(self):
  47         return self.DL.extract_info(self.url, download=False)
  48
  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()
  58         return {l: sub_info['data'] for l, sub_info in subtitles.items()}
  59
  60
  61 @is_download_test
  62 class TestYoutubeSubtitles(BaseTestSubtitles):
  63     # Available subtitles for QRS8MkLhQmM:
  64     # Language formats
  65     # ru       vtt, ttml, srv3, srv2, srv1, json3
  66     # fr       vtt, ttml, srv3, srv2, srv1, json3
  67     # en       vtt, ttml, srv3, srv2, srv1, json3
  68     # nl       vtt, ttml, srv3, srv2, srv1, json3
  69     # de       vtt, ttml, srv3, srv2, srv1, json3
  70     # ko       vtt, ttml, srv3, srv2, srv1, json3
  71     # it       vtt, ttml, srv3, srv2, srv1, json3
  72     # zh-Hant  vtt, ttml, srv3, srv2, srv1, json3
  73     # hi       vtt, ttml, srv3, srv2, srv1, json3
  74     # pt-BR    vtt, ttml, srv3, srv2, srv1, json3
  75     # es-MX    vtt, ttml, srv3, srv2, srv1, json3
  76     # ja       vtt, ttml, srv3, srv2, srv1, json3
  77     # pl       vtt, ttml, srv3, srv2, srv1, json3
  78     url = 'QRS8MkLhQmM'
  79     IE = YoutubeIE
  80
  81     def test_youtube_allsubtitles(self):
  82         self.DL.params['writesubtitles'] = True
  83         self.DL.params['allsubtitles'] = True
  84         subtitles = self.getSubtitles()
  85         self.assertEqual(len(subtitles.keys()), 13)
  86         self.assertEqual(md5(subtitles['en']), 'ae1bd34126571a77aabd4d276b28044d')
  87         self.assertEqual(md5(subtitles['it']), '0e0b667ba68411d88fd1c5f4f4eab2f9')
  88         for lang in ['fr', 'de']:
  89             self.assertTrue(subtitles.get(lang) is not None, f'Subtitles for \'{lang}\' not extracted')
  90
  91     def _test_subtitles_format(self, fmt, md5_hash, lang='en'):
  92         self.DL.params['writesubtitles'] = True
  93         self.DL.params['subtitlesformat'] = fmt
  94         subtitles = self.getSubtitles()
  95         self.assertEqual(md5(subtitles[lang]), md5_hash)
  96
  97     def test_youtube_subtitles_ttml_format(self):
  98         self._test_subtitles_format('ttml', 'c97ddf1217390906fa9fbd34901f3da2')
  99
 100     def test_youtube_subtitles_vtt_format(self):
 101         self._test_subtitles_format('vtt', 'ae1bd34126571a77aabd4d276b28044d')
 102
 103     def test_youtube_subtitles_json3_format(self):
 104         self._test_subtitles_format('json3', '688dd1ce0981683867e7fe6fde2a224b')
 105
 106     def _test_automatic_captions(self, url, lang):
 107         self.url = url
 108         self.DL.params['writeautomaticsub'] = True
 109         self.DL.params['subtitleslangs'] = [lang]
 110         subtitles = self.getSubtitles()
 111         self.assertTrue(subtitles[lang] is not None)
 112
 113     def test_youtube_automatic_captions(self):
 114         # Available automatic captions for 8YoUxe5ncPo:
 115         # Language formats (all in vtt, ttml, srv3, srv2, srv1, json3)
 116         # gu, zh-Hans, zh-Hant, gd, ga, gl, lb, la, lo, tt, tr,
 117         # lv, lt, tk, th, tg, te, fil, haw, yi, ceb, yo, de, da,
 118         # el, eo, en, eu, et, es, ru, rw, ro, bn, be, bg, uk, jv,
 119         # bs, ja, or, xh, co, ca, cy, cs, ps, pt, pa, vi, pl, hy,
 120         # hr, ht, hu, hmn, hi, ha, mg, uz, ml, mn, mi, mk, ur,
 121         # mt, ms, mr, ug, ta, my, af, sw, is, am,
 122         #                                         *it*, iw, sv, ar,
 123         # su, zu, az, id, ig, nl, no, ne, ny, fr, ku, fy, fa, fi,
 124         # ka, kk, sr, sq, ko, kn, km, st, sk, si, so, sn, sm, sl,
 125         # ky, sd
 126         # ...
 127         self._test_automatic_captions('8YoUxe5ncPo', 'it')
 128
 129     @unittest.skip('Video unavailable')
 130     def test_youtube_translated_subtitles(self):
 131         # This video has a subtitles track, which can be translated (#4555)
 132         self._test_automatic_captions('Ky9eprVWzlI', 'it')
 133
 134     def test_youtube_nosubtitles(self):
 135         self.DL.expect_warning('video doesn\'t have subtitles')
 136         # Available automatic captions for 8YoUxe5ncPo:
 137         # ...
 138         # 8YoUxe5ncPo has no subtitles
 139         self.url = '8YoUxe5ncPo'
 140         self.DL.params['writesubtitles'] = True
 141         self.DL.params['allsubtitles'] = True
 142         subtitles = self.getSubtitles()
 143         self.assertFalse(subtitles)
 144
 145
 146 @is_download_test
 147 class TestDailymotionSubtitles(BaseTestSubtitles):
 148     url = 'http://www.dailymotion.com/video/xczg00'
 149     IE = DailymotionIE
 150
 151     def test_allsubtitles(self):
 152         self.DL.params['writesubtitles'] = True
 153         self.DL.params['allsubtitles'] = True
 154         subtitles = self.getSubtitles()
 155         self.assertTrue(len(subtitles.keys()) >= 6)
 156         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
 157         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
 158         for lang in ['es', 'fr', 'de']:
 159             self.assertTrue(subtitles.get(lang) is not None, f'Subtitles for \'{lang}\' not extracted')
 160
 161     def test_nosubtitles(self):
 162         self.DL.expect_warning('video doesn\'t have subtitles')
 163         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
 164         self.DL.params['writesubtitles'] = True
 165         self.DL.params['allsubtitles'] = True
 166         subtitles = self.getSubtitles()
 167         self.assertFalse(subtitles)
 168
 169
 170 @is_download_test
 171 @unittest.skip('IE broken')
 172 class TestTedSubtitles(BaseTestSubtitles):
 173     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
 174     IE = TedTalkIE
 175
 176     def test_allsubtitles(self):
 177         self.DL.params['writesubtitles'] = True
 178         self.DL.params['allsubtitles'] = True
 179         subtitles = self.getSubtitles()
 180         self.assertTrue(len(subtitles.keys()) >= 28)
 181         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
 182         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
 183         for lang in ['es', 'fr', 'de']:
 184             self.assertTrue(subtitles.get(lang) is not None, f'Subtitles for \'{lang}\' not extracted')
 185
 186
 187 @is_download_test
 188 class TestVimeoSubtitles(BaseTestSubtitles):
 189     url = 'http://vimeo.com/76979871'
 190     IE = VimeoIE
 191
 192     def test_allsubtitles(self):
 193         self.DL.params['writesubtitles'] = True
 194         self.DL.params['allsubtitles'] = True
 195         subtitles = self.getSubtitles()
 196         self.assertEqual(set(subtitles.keys()), {'de', 'en', 'es', 'fr'})
 197         self.assertEqual(md5(subtitles['en']), '386cbc9320b94e25cb364b97935e5dd1')
 198         self.assertEqual(md5(subtitles['fr']), 'c9b69eef35bc6641c0d4da8a04f9dfac')
 199
 200     def test_nosubtitles(self):
 201         self.DL.expect_warning('video doesn\'t have subtitles')
 202         self.url = 'http://vimeo.com/68093876'
 203         self.DL.params['writesubtitles'] = True
 204         self.DL.params['allsubtitles'] = True
 205         subtitles = self.getSubtitles()
 206         self.assertFalse(subtitles)
 207
 208
 209 @is_download_test
 210 @unittest.skip('IE broken')
 211 class TestWallaSubtitles(BaseTestSubtitles):
 212     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
 213     IE = WallaIE
 214
 215     def test_allsubtitles(self):
 216         self.DL.expect_warning('Automatic Captions not supported by this server')
 217         self.DL.params['writesubtitles'] = True
 218         self.DL.params['allsubtitles'] = True
 219         subtitles = self.getSubtitles()
 220         self.assertEqual(set(subtitles.keys()), {'heb'})
 221         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
 222
 223     def test_nosubtitles(self):
 224         self.DL.expect_warning('video doesn\'t have subtitles')
 225         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
 226         self.DL.params['writesubtitles'] = True
 227         self.DL.params['allsubtitles'] = True
 228         subtitles = self.getSubtitles()
 229         self.assertFalse(subtitles)
 230
 231
 232 @is_download_test
 233 @unittest.skip('IE broken')
 234 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
 235     url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
 236     IE = CeskaTelevizeIE
 237
 238     def test_allsubtitles(self):
 239         self.DL.expect_warning('Automatic Captions not supported by this server')
 240         self.DL.params['writesubtitles'] = True
 241         self.DL.params['allsubtitles'] = True
 242         subtitles = self.getSubtitles()
 243         self.assertEqual(set(subtitles.keys()), {'cs'})
 244         self.assertTrue(len(subtitles['cs']) > 20000)
 245
 246     def test_nosubtitles(self):
 247         self.DL.expect_warning('video doesn\'t have subtitles')
 248         self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
 249         self.DL.params['writesubtitles'] = True
 250         self.DL.params['allsubtitles'] = True
 251         subtitles = self.getSubtitles()
 252         self.assertFalse(subtitles)
 253
 254
 255 @is_download_test
 256 @unittest.skip('IE broken')
 257 class TestLyndaSubtitles(BaseTestSubtitles):
 258     url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
 259     IE = LyndaIE
 260
 261     def test_allsubtitles(self):
 262         self.DL.params['writesubtitles'] = True
 263         self.DL.params['allsubtitles'] = True
 264         subtitles = self.getSubtitles()
 265         self.assertEqual(set(subtitles.keys()), {'en'})
 266         self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
 267
 268
 269 @is_download_test
 270 @unittest.skip('IE broken')
 271 class TestNPOSubtitles(BaseTestSubtitles):
 272     url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
 273     IE = NPOIE
 274
 275     def test_allsubtitles(self):
 276         self.DL.params['writesubtitles'] = True
 277         self.DL.params['allsubtitles'] = True
 278         subtitles = self.getSubtitles()
 279         self.assertEqual(set(subtitles.keys()), {'nl'})
 280         self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
 281
 282
 283 @is_download_test
 284 @unittest.skip('IE broken')
 285 class TestMTVSubtitles(BaseTestSubtitles):
 286     url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
 287     IE = ComedyCentralIE
 288
 289     def getInfoDict(self):
 290         return super().getInfoDict()['entries'][0]
 291
 292     def test_allsubtitles(self):
 293         self.DL.params['writesubtitles'] = True
 294         self.DL.params['allsubtitles'] = True
 295         subtitles = self.getSubtitles()
 296         self.assertEqual(set(subtitles.keys()), {'en'})
 297         self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
 298
 299
 300 @is_download_test
 301 class TestNRKSubtitles(BaseTestSubtitles):
 302     url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
 303     IE = NRKTVIE
 304
 305     def test_allsubtitles(self):
 306         self.DL.params['writesubtitles'] = True
 307         self.DL.params['allsubtitles'] = True
 308         subtitles = self.getSubtitles()
 309         self.assertEqual(set(subtitles.keys()), {'nb-ttv'})
 310         self.assertEqual(md5(subtitles['nb-ttv']), '67e06ff02d0deaf975e68f6cb8f6a149')
 311
 312
 313 @is_download_test
 314 class TestRaiPlaySubtitles(BaseTestSubtitles):
 315     IE = RaiPlayIE
 316
 317     def test_subtitles_key(self):
 318         self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
 319         self.DL.params['writesubtitles'] = True
 320         self.DL.params['allsubtitles'] = True
 321         subtitles = self.getSubtitles()
 322         self.assertEqual(set(subtitles.keys()), {'it'})
 323         self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
 324
 325     def test_subtitles_array_key(self):
 326         self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
 327         self.DL.params['writesubtitles'] = True
 328         self.DL.params['allsubtitles'] = True
 329         subtitles = self.getSubtitles()
 330         self.assertEqual(set(subtitles.keys()), {'it'})
 331         self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
 332
 333
 334 @is_download_test
 335 @unittest.skip('IE broken - DRM only')
 336 class TestVikiSubtitles(BaseTestSubtitles):
 337     url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
 338     IE = VikiIE
 339
 340     def test_allsubtitles(self):
 341         self.DL.params['writesubtitles'] = True
 342         self.DL.params['allsubtitles'] = True
 343         subtitles = self.getSubtitles()
 344         self.assertEqual(set(subtitles.keys()), {'en'})
 345         self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
 346
 347
 348 @is_download_test
 349 class TestThePlatformSubtitles(BaseTestSubtitles):
 350     # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
 351     # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
 352     url = 'theplatform:JFUjUE1_ehvq'
 353     IE = ThePlatformIE
 354
 355     def test_allsubtitles(self):
 356         self.DL.params['writesubtitles'] = True
 357         self.DL.params['allsubtitles'] = True
 358         subtitles = self.getSubtitles()
 359         self.assertEqual(set(subtitles.keys()), {'en'})
 360         self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
 361
 362
 363 @is_download_test
 364 @unittest.skip('IE broken')
 365 class TestThePlatformFeedSubtitles(BaseTestSubtitles):
 366     url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
 367     IE = ThePlatformFeedIE
 368
 369     def test_allsubtitles(self):
 370         self.DL.params['writesubtitles'] = True
 371         self.DL.params['allsubtitles'] = True
 372         subtitles = self.getSubtitles()
 373         self.assertEqual(set(subtitles.keys()), {'en'})
 374         self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
 375
 376
 377 @is_download_test
 378 class TestRtveSubtitles(BaseTestSubtitles):
 379     url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
 380     IE = RTVEALaCartaIE
 381
 382     def test_allsubtitles(self):
 383         print('Skipping, only available from Spain')
 384         return
 385         self.DL.params['writesubtitles'] = True
 386         self.DL.params['allsubtitles'] = True
 387         subtitles = self.getSubtitles()
 388         self.assertEqual(set(subtitles.keys()), {'es'})
 389         self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
 390
 391
 392 @is_download_test
 393 class TestDemocracynowSubtitles(BaseTestSubtitles):
 394     url = 'http://www.democracynow.org/shows/2015/7/3'
 395     IE = DemocracynowIE
 396
 397     def test_allsubtitles(self):
 398         self.DL.params['writesubtitles'] = True
 399         self.DL.params['allsubtitles'] = True
 400         subtitles = self.getSubtitles()
 401         self.assertEqual(set(subtitles.keys()), {'en'})
 402         self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')
 403
 404     def test_subtitles_in_page(self):
 405         self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
 406         self.DL.params['writesubtitles'] = True
 407         self.DL.params['allsubtitles'] = True
 408         subtitles = self.getSubtitles()
 409         self.assertEqual(set(subtitles.keys()), {'en'})
 410         self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')
 411
 412
 413 @is_download_test
 414 class TestPBSSubtitles(BaseTestSubtitles):
 415     url = 'https://www.pbs.org/video/how-fantasy-reflects-our-world-picecq/'
 416     IE = PBSIE
 417
 418     def test_allsubtitles(self):
 419         self.DL.params['writesubtitles'] = True
 420         self.DL.params['allsubtitles'] = True
 421         subtitles = self.getSubtitles()
 422         self.assertEqual(set(subtitles.keys()), {'en'})
 423
 424     def test_subtitles_dfxp_format(self):
 425         self.DL.params['writesubtitles'] = True
 426         self.DL.params['subtitlesformat'] = 'dfxp'
 427         subtitles = self.getSubtitles()
 428         self.assertIn(md5(subtitles['en']), ['643b034254cdc3768ff1e750b6b5873b'])
 429
 430     def test_subtitles_vtt_format(self):
 431         self.DL.params['writesubtitles'] = True
 432         self.DL.params['subtitlesformat'] = 'vtt'
 433         subtitles = self.getSubtitles()
 434         self.assertIn(
 435             md5(subtitles['en']), ['937a05711555b165d4c55a9667017045', 'f49ea998d6824d94959c8152a368ff73'])
 436
 437     def test_subtitles_srt_format(self):
 438         self.DL.params['writesubtitles'] = True
 439         self.DL.params['subtitlesformat'] = 'srt'
 440         subtitles = self.getSubtitles()
 441         self.assertIn(md5(subtitles['en']), ['2082c21b43759d9bf172931b2f2ca371'])
 442
 443     def test_subtitles_sami_format(self):
 444         self.DL.params['writesubtitles'] = True
 445         self.DL.params['subtitlesformat'] = 'sami'
 446         subtitles = self.getSubtitles()
 447         self.assertIn(md5(subtitles['en']), ['4256b16ac7da6a6780fafd04294e85cd'])
 448
 449
 450 if __name__ == '__main__':
 451     unittest.main()