]> jfr.im git - yt-dlp.git/blob - test/test_subtitles.py
[cleanup] Sort imports
[yt-dlp.git] / test / test_subtitles.py
1 #!/usr/bin/env python3
2 # Allow direct execution
3 import os
4 import sys
5 import unittest
6
7 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
8
9 from test.helper import FakeYDL, is_download_test, md5
10
11 from yt_dlp.extractor import (
12 NPOIE,
13 NRKTVIE,
14 PBSIE,
15 CeskaTelevizeIE,
16 ComedyCentralIE,
17 DailymotionIE,
18 DemocracynowIE,
19 LyndaIE,
20 RaiPlayIE,
21 RTVEALaCartaIE,
22 TedTalkIE,
23 ThePlatformFeedIE,
24 ThePlatformIE,
25 VikiIE,
26 VimeoIE,
27 WallaIE,
28 YoutubeIE,
29 )
30
31
32 @is_download_test
33 class BaseTestSubtitles(unittest.TestCase):
34 url = None
35 IE = None
36
37 def setUp(self):
38 self.DL = FakeYDL()
39 self.ie = self.IE()
40 self.DL.add_info_extractor(self.ie)
41
42 def getInfoDict(self):
43 info_dict = self.DL.extract_info(self.url, download=False)
44 return info_dict
45
46 def getSubtitles(self):
47 info_dict = self.getInfoDict()
48 subtitles = info_dict['requested_subtitles']
49 if not subtitles:
50 return subtitles
51 for sub_info in subtitles.values():
52 if sub_info.get('data') is None:
53 uf = self.DL.urlopen(sub_info['url'])
54 sub_info['data'] = uf.read().decode('utf-8')
55 return {l: sub_info['data'] for l, sub_info in subtitles.items()}
56
57
58 @is_download_test
59 class TestYoutubeSubtitles(BaseTestSubtitles):
60 url = 'QRS8MkLhQmM'
61 IE = YoutubeIE
62
63 def test_youtube_allsubtitles(self):
64 self.DL.params['writesubtitles'] = True
65 self.DL.params['allsubtitles'] = True
66 subtitles = self.getSubtitles()
67 self.assertEqual(len(subtitles.keys()), 13)
68 self.assertEqual(md5(subtitles['en']), '688dd1ce0981683867e7fe6fde2a224b')
69 self.assertEqual(md5(subtitles['it']), '31324d30b8430b309f7f5979a504a769')
70 for lang in ['fr', 'de']:
71 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
72
73 def test_youtube_subtitles_ttml_format(self):
74 self.DL.params['writesubtitles'] = True
75 self.DL.params['subtitlesformat'] = 'ttml'
76 subtitles = self.getSubtitles()
77 self.assertEqual(md5(subtitles['en']), 'c97ddf1217390906fa9fbd34901f3da2')
78
79 def test_youtube_subtitles_vtt_format(self):
80 self.DL.params['writesubtitles'] = True
81 self.DL.params['subtitlesformat'] = 'vtt'
82 subtitles = self.getSubtitles()
83 self.assertEqual(md5(subtitles['en']), 'ae1bd34126571a77aabd4d276b28044d')
84
85 def test_youtube_automatic_captions(self):
86 self.url = '8YoUxe5ncPo'
87 self.DL.params['writeautomaticsub'] = True
88 self.DL.params['subtitleslangs'] = ['it']
89 subtitles = self.getSubtitles()
90 self.assertTrue(subtitles['it'] is not None)
91
92 def test_youtube_no_automatic_captions(self):
93 self.url = 'QRS8MkLhQmM'
94 self.DL.params['writeautomaticsub'] = True
95 subtitles = self.getSubtitles()
96 self.assertTrue(not subtitles)
97
98 def test_youtube_translated_subtitles(self):
99 # This video has a subtitles track, which can be translated
100 self.url = 'i0ZabxXmH4Y'
101 self.DL.params['writeautomaticsub'] = True
102 self.DL.params['subtitleslangs'] = ['it']
103 subtitles = self.getSubtitles()
104 self.assertTrue(subtitles['it'] is not None)
105
106 def test_youtube_nosubtitles(self):
107 self.DL.expect_warning('video doesn\'t have subtitles')
108 self.url = 'n5BB19UTcdA'
109 self.DL.params['writesubtitles'] = True
110 self.DL.params['allsubtitles'] = True
111 subtitles = self.getSubtitles()
112 self.assertFalse(subtitles)
113
114
115 @is_download_test
116 class TestDailymotionSubtitles(BaseTestSubtitles):
117 url = 'http://www.dailymotion.com/video/xczg00'
118 IE = DailymotionIE
119
120 def test_allsubtitles(self):
121 self.DL.params['writesubtitles'] = True
122 self.DL.params['allsubtitles'] = True
123 subtitles = self.getSubtitles()
124 self.assertTrue(len(subtitles.keys()) >= 6)
125 self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
126 self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
127 for lang in ['es', 'fr', 'de']:
128 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
129
130 def test_nosubtitles(self):
131 self.DL.expect_warning('video doesn\'t have subtitles')
132 self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
133 self.DL.params['writesubtitles'] = True
134 self.DL.params['allsubtitles'] = True
135 subtitles = self.getSubtitles()
136 self.assertFalse(subtitles)
137
138
139 @is_download_test
140 class TestTedSubtitles(BaseTestSubtitles):
141 url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
142 IE = TedTalkIE
143
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 self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
150 self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
151 for lang in ['es', 'fr', 'de']:
152 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
153
154
155 @is_download_test
156 class TestVimeoSubtitles(BaseTestSubtitles):
157 url = 'http://vimeo.com/76979871'
158 IE = VimeoIE
159
160 def test_allsubtitles(self):
161 self.DL.params['writesubtitles'] = True
162 self.DL.params['allsubtitles'] = True
163 subtitles = self.getSubtitles()
164 self.assertEqual(set(subtitles.keys()), {'de', 'en', 'es', 'fr'})
165 self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
166 self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
167
168 def test_nosubtitles(self):
169 self.DL.expect_warning('video doesn\'t have subtitles')
170 self.url = 'http://vimeo.com/56015672'
171 self.DL.params['writesubtitles'] = True
172 self.DL.params['allsubtitles'] = True
173 subtitles = self.getSubtitles()
174 self.assertFalse(subtitles)
175
176
177 @is_download_test
178 class TestWallaSubtitles(BaseTestSubtitles):
179 url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
180 IE = WallaIE
181
182 def test_allsubtitles(self):
183 self.DL.expect_warning('Automatic Captions not supported by this server')
184 self.DL.params['writesubtitles'] = True
185 self.DL.params['allsubtitles'] = True
186 subtitles = self.getSubtitles()
187 self.assertEqual(set(subtitles.keys()), {'heb'})
188 self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
189
190 def test_nosubtitles(self):
191 self.DL.expect_warning('video doesn\'t have subtitles')
192 self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
193 self.DL.params['writesubtitles'] = True
194 self.DL.params['allsubtitles'] = True
195 subtitles = self.getSubtitles()
196 self.assertFalse(subtitles)
197
198
199 @is_download_test
200 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
201 url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
202 IE = CeskaTelevizeIE
203
204 def test_allsubtitles(self):
205 self.DL.expect_warning('Automatic Captions not supported by this server')
206 self.DL.params['writesubtitles'] = True
207 self.DL.params['allsubtitles'] = True
208 subtitles = self.getSubtitles()
209 self.assertEqual(set(subtitles.keys()), {'cs'})
210 self.assertTrue(len(subtitles['cs']) > 20000)
211
212 def test_nosubtitles(self):
213 self.DL.expect_warning('video doesn\'t have subtitles')
214 self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
215 self.DL.params['writesubtitles'] = True
216 self.DL.params['allsubtitles'] = True
217 subtitles = self.getSubtitles()
218 self.assertFalse(subtitles)
219
220
221 @is_download_test
222 class TestLyndaSubtitles(BaseTestSubtitles):
223 url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
224 IE = LyndaIE
225
226 def test_allsubtitles(self):
227 self.DL.params['writesubtitles'] = True
228 self.DL.params['allsubtitles'] = True
229 subtitles = self.getSubtitles()
230 self.assertEqual(set(subtitles.keys()), {'en'})
231 self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
232
233
234 @is_download_test
235 class TestNPOSubtitles(BaseTestSubtitles):
236 url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
237 IE = NPOIE
238
239 def test_allsubtitles(self):
240 self.DL.params['writesubtitles'] = True
241 self.DL.params['allsubtitles'] = True
242 subtitles = self.getSubtitles()
243 self.assertEqual(set(subtitles.keys()), {'nl'})
244 self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
245
246
247 @is_download_test
248 class TestMTVSubtitles(BaseTestSubtitles):
249 url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
250 IE = ComedyCentralIE
251
252 def getInfoDict(self):
253 return super().getInfoDict()['entries'][0]
254
255 def test_allsubtitles(self):
256 self.DL.params['writesubtitles'] = True
257 self.DL.params['allsubtitles'] = True
258 subtitles = self.getSubtitles()
259 self.assertEqual(set(subtitles.keys()), {'en'})
260 self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
261
262
263 @is_download_test
264 class TestNRKSubtitles(BaseTestSubtitles):
265 url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
266 IE = NRKTVIE
267
268 def test_allsubtitles(self):
269 self.DL.params['writesubtitles'] = True
270 self.DL.params['allsubtitles'] = True
271 subtitles = self.getSubtitles()
272 self.assertEqual(set(subtitles.keys()), {'no'})
273 self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
274
275
276 @is_download_test
277 class TestRaiPlaySubtitles(BaseTestSubtitles):
278 IE = RaiPlayIE
279
280 def test_subtitles_key(self):
281 self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
282 self.DL.params['writesubtitles'] = True
283 self.DL.params['allsubtitles'] = True
284 subtitles = self.getSubtitles()
285 self.assertEqual(set(subtitles.keys()), {'it'})
286 self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
287
288 def test_subtitles_array_key(self):
289 self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
290 self.DL.params['writesubtitles'] = True
291 self.DL.params['allsubtitles'] = True
292 subtitles = self.getSubtitles()
293 self.assertEqual(set(subtitles.keys()), {'it'})
294 self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
295
296
297 @is_download_test
298 class TestVikiSubtitles(BaseTestSubtitles):
299 url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
300 IE = VikiIE
301
302 def test_allsubtitles(self):
303 self.DL.params['writesubtitles'] = True
304 self.DL.params['allsubtitles'] = True
305 subtitles = self.getSubtitles()
306 self.assertEqual(set(subtitles.keys()), {'en'})
307 self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
308
309
310 @is_download_test
311 class TestThePlatformSubtitles(BaseTestSubtitles):
312 # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
313 # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
314 url = 'theplatform:JFUjUE1_ehvq'
315 IE = ThePlatformIE
316
317 def test_allsubtitles(self):
318 self.DL.params['writesubtitles'] = True
319 self.DL.params['allsubtitles'] = True
320 subtitles = self.getSubtitles()
321 self.assertEqual(set(subtitles.keys()), {'en'})
322 self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
323
324
325 @is_download_test
326 class TestThePlatformFeedSubtitles(BaseTestSubtitles):
327 url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
328 IE = ThePlatformFeedIE
329
330 def test_allsubtitles(self):
331 self.DL.params['writesubtitles'] = True
332 self.DL.params['allsubtitles'] = True
333 subtitles = self.getSubtitles()
334 self.assertEqual(set(subtitles.keys()), {'en'})
335 self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
336
337
338 @is_download_test
339 class TestRtveSubtitles(BaseTestSubtitles):
340 url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
341 IE = RTVEALaCartaIE
342
343 def test_allsubtitles(self):
344 print('Skipping, only available from Spain')
345 return
346 self.DL.params['writesubtitles'] = True
347 self.DL.params['allsubtitles'] = True
348 subtitles = self.getSubtitles()
349 self.assertEqual(set(subtitles.keys()), {'es'})
350 self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
351
352
353 @is_download_test
354 class TestDemocracynowSubtitles(BaseTestSubtitles):
355 url = 'http://www.democracynow.org/shows/2015/7/3'
356 IE = DemocracynowIE
357
358 def test_allsubtitles(self):
359 self.DL.params['writesubtitles'] = True
360 self.DL.params['allsubtitles'] = True
361 subtitles = self.getSubtitles()
362 self.assertEqual(set(subtitles.keys()), {'en'})
363 self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
364
365 def test_subtitles_in_page(self):
366 self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
367 self.DL.params['writesubtitles'] = True
368 self.DL.params['allsubtitles'] = True
369 subtitles = self.getSubtitles()
370 self.assertEqual(set(subtitles.keys()), {'en'})
371 self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
372
373
374 @is_download_test
375 class TestPBSSubtitles(BaseTestSubtitles):
376 url = 'https://www.pbs.org/video/how-fantasy-reflects-our-world-picecq/'
377 IE = PBSIE
378
379 def test_allsubtitles(self):
380 self.DL.params['writesubtitles'] = True
381 self.DL.params['allsubtitles'] = True
382 subtitles = self.getSubtitles()
383 self.assertEqual(set(subtitles.keys()), {'en'})
384
385 def test_subtitles_dfxp_format(self):
386 self.DL.params['writesubtitles'] = True
387 self.DL.params['subtitlesformat'] = 'dfxp'
388 subtitles = self.getSubtitles()
389 self.assertIn(md5(subtitles['en']), ['643b034254cdc3768ff1e750b6b5873b'])
390
391 def test_subtitles_vtt_format(self):
392 self.DL.params['writesubtitles'] = True
393 self.DL.params['subtitlesformat'] = 'vtt'
394 subtitles = self.getSubtitles()
395 self.assertIn(
396 md5(subtitles['en']), ['937a05711555b165d4c55a9667017045', 'f49ea998d6824d94959c8152a368ff73'])
397
398 def test_subtitles_srt_format(self):
399 self.DL.params['writesubtitles'] = True
400 self.DL.params['subtitlesformat'] = 'srt'
401 subtitles = self.getSubtitles()
402 self.assertIn(md5(subtitles['en']), ['2082c21b43759d9bf172931b2f2ca371'])
403
404 def test_subtitles_sami_format(self):
405 self.DL.params['writesubtitles'] = True
406 self.DL.params['subtitlesformat'] = 'sami'
407 subtitles = self.getSubtitles()
408 self.assertIn(md5(subtitles['en']), ['4256b16ac7da6a6780fafd04294e85cd'])
409
410
411 if __name__ == '__main__':
412 unittest.main()