]> jfr.im git - yt-dlp.git/blob - test/test_subtitles.py
[Ted] Rewrite extractor (#2359)
[yt-dlp.git] / test / test_subtitles.py
1 #!/usr/bin/env python3
2 from __future__ import unicode_literals
3
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__))))
9
10 from test.helper import FakeYDL, md5, is_download_test
11
12
13 from yt_dlp.extractor import (
14 YoutubeIE,
15 DailymotionIE,
16 TedTalkIE,
17 VimeoIE,
18 WallaIE,
19 CeskaTelevizeIE,
20 LyndaIE,
21 NPOIE,
22 PBSIE,
23 ComedyCentralIE,
24 NRKTVIE,
25 RaiPlayIE,
26 VikiIE,
27 ThePlatformIE,
28 ThePlatformFeedIE,
29 RTVEALaCartaIE,
30 DemocracynowIE,
31 )
32
33
34 @is_download_test
35 class BaseTestSubtitles(unittest.TestCase):
36 url = None
37 IE = None
38
39 def setUp(self):
40 self.DL = FakeYDL()
41 self.ie = self.IE()
42 self.DL.add_info_extractor(self.ie)
43
44 def getInfoDict(self):
45 info_dict = self.DL.extract_info(self.url, download=False)
46 return info_dict
47
48 def getSubtitles(self):
49 info_dict = self.getInfoDict()
50 subtitles = info_dict['requested_subtitles']
51 if not subtitles:
52 return subtitles
53 for sub_info in subtitles.values():
54 if sub_info.get('data') is None:
55 uf = self.DL.urlopen(sub_info['url'])
56 sub_info['data'] = uf.read().decode('utf-8')
57 return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
58
59
60 @is_download_test
61 class TestYoutubeSubtitles(BaseTestSubtitles):
62 url = 'QRS8MkLhQmM'
63 IE = YoutubeIE
64
65 def test_youtube_allsubtitles(self):
66 self.DL.params['writesubtitles'] = True
67 self.DL.params['allsubtitles'] = True
68 subtitles = self.getSubtitles()
69 self.assertEqual(len(subtitles.keys()), 13)
70 self.assertEqual(md5(subtitles['en']), '688dd1ce0981683867e7fe6fde2a224b')
71 self.assertEqual(md5(subtitles['it']), '31324d30b8430b309f7f5979a504a769')
72 for lang in ['fr', 'de']:
73 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
74
75 def test_youtube_subtitles_ttml_format(self):
76 self.DL.params['writesubtitles'] = True
77 self.DL.params['subtitlesformat'] = 'ttml'
78 subtitles = self.getSubtitles()
79 self.assertEqual(md5(subtitles['en']), 'c97ddf1217390906fa9fbd34901f3da2')
80
81 def test_youtube_subtitles_vtt_format(self):
82 self.DL.params['writesubtitles'] = True
83 self.DL.params['subtitlesformat'] = 'vtt'
84 subtitles = self.getSubtitles()
85 self.assertEqual(md5(subtitles['en']), 'ae1bd34126571a77aabd4d276b28044d')
86
87 def test_youtube_automatic_captions(self):
88 self.url = '8YoUxe5ncPo'
89 self.DL.params['writeautomaticsub'] = True
90 self.DL.params['subtitleslangs'] = ['it']
91 subtitles = self.getSubtitles()
92 self.assertTrue(subtitles['it'] is not None)
93
94 def test_youtube_no_automatic_captions(self):
95 self.url = 'QRS8MkLhQmM'
96 self.DL.params['writeautomaticsub'] = True
97 subtitles = self.getSubtitles()
98 self.assertTrue(not subtitles)
99
100 def test_youtube_translated_subtitles(self):
101 # This video has a subtitles track, which can be translated
102 self.url = 'i0ZabxXmH4Y'
103 self.DL.params['writeautomaticsub'] = True
104 self.DL.params['subtitleslangs'] = ['it']
105 subtitles = self.getSubtitles()
106 self.assertTrue(subtitles['it'] is not None)
107
108 def test_youtube_nosubtitles(self):
109 self.DL.expect_warning('video doesn\'t have subtitles')
110 self.url = 'n5BB19UTcdA'
111 self.DL.params['writesubtitles'] = True
112 self.DL.params['allsubtitles'] = True
113 subtitles = self.getSubtitles()
114 self.assertFalse(subtitles)
115
116
117 @is_download_test
118 class TestDailymotionSubtitles(BaseTestSubtitles):
119 url = 'http://www.dailymotion.com/video/xczg00'
120 IE = DailymotionIE
121
122 def test_allsubtitles(self):
123 self.DL.params['writesubtitles'] = True
124 self.DL.params['allsubtitles'] = True
125 subtitles = self.getSubtitles()
126 self.assertTrue(len(subtitles.keys()) >= 6)
127 self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
128 self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
129 for lang in ['es', 'fr', 'de']:
130 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
131
132 def test_nosubtitles(self):
133 self.DL.expect_warning('video doesn\'t have subtitles')
134 self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
135 self.DL.params['writesubtitles'] = True
136 self.DL.params['allsubtitles'] = True
137 subtitles = self.getSubtitles()
138 self.assertFalse(subtitles)
139
140
141 @is_download_test
142 class TestTedSubtitles(BaseTestSubtitles):
143 url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
144 IE = TedTalkIE
145
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 self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
152 self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
153 for lang in ['es', 'fr', 'de']:
154 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
155
156
157 @is_download_test
158 class TestVimeoSubtitles(BaseTestSubtitles):
159 url = 'http://vimeo.com/76979871'
160 IE = VimeoIE
161
162 def test_allsubtitles(self):
163 self.DL.params['writesubtitles'] = True
164 self.DL.params['allsubtitles'] = True
165 subtitles = self.getSubtitles()
166 self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
167 self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
168 self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
169
170 def test_nosubtitles(self):
171 self.DL.expect_warning('video doesn\'t have subtitles')
172 self.url = 'http://vimeo.com/56015672'
173 self.DL.params['writesubtitles'] = True
174 self.DL.params['allsubtitles'] = True
175 subtitles = self.getSubtitles()
176 self.assertFalse(subtitles)
177
178
179 @is_download_test
180 class TestWallaSubtitles(BaseTestSubtitles):
181 url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
182 IE = WallaIE
183
184 def test_allsubtitles(self):
185 self.DL.expect_warning('Automatic Captions not supported by this server')
186 self.DL.params['writesubtitles'] = True
187 self.DL.params['allsubtitles'] = True
188 subtitles = self.getSubtitles()
189 self.assertEqual(set(subtitles.keys()), set(['heb']))
190 self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
191
192 def test_nosubtitles(self):
193 self.DL.expect_warning('video doesn\'t have subtitles')
194 self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
195 self.DL.params['writesubtitles'] = True
196 self.DL.params['allsubtitles'] = True
197 subtitles = self.getSubtitles()
198 self.assertFalse(subtitles)
199
200
201 @is_download_test
202 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
203 url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
204 IE = CeskaTelevizeIE
205
206 def test_allsubtitles(self):
207 self.DL.expect_warning('Automatic Captions not supported by this server')
208 self.DL.params['writesubtitles'] = True
209 self.DL.params['allsubtitles'] = True
210 subtitles = self.getSubtitles()
211 self.assertEqual(set(subtitles.keys()), set(['cs']))
212 self.assertTrue(len(subtitles['cs']) > 20000)
213
214 def test_nosubtitles(self):
215 self.DL.expect_warning('video doesn\'t have subtitles')
216 self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
217 self.DL.params['writesubtitles'] = True
218 self.DL.params['allsubtitles'] = True
219 subtitles = self.getSubtitles()
220 self.assertFalse(subtitles)
221
222
223 @is_download_test
224 class TestLyndaSubtitles(BaseTestSubtitles):
225 url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
226 IE = LyndaIE
227
228 def test_allsubtitles(self):
229 self.DL.params['writesubtitles'] = True
230 self.DL.params['allsubtitles'] = True
231 subtitles = self.getSubtitles()
232 self.assertEqual(set(subtitles.keys()), set(['en']))
233 self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
234
235
236 @is_download_test
237 class TestNPOSubtitles(BaseTestSubtitles):
238 url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
239 IE = NPOIE
240
241 def test_allsubtitles(self):
242 self.DL.params['writesubtitles'] = True
243 self.DL.params['allsubtitles'] = True
244 subtitles = self.getSubtitles()
245 self.assertEqual(set(subtitles.keys()), set(['nl']))
246 self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
247
248
249 @is_download_test
250 class TestMTVSubtitles(BaseTestSubtitles):
251 url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
252 IE = ComedyCentralIE
253
254 def getInfoDict(self):
255 return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
256
257 def test_allsubtitles(self):
258 self.DL.params['writesubtitles'] = True
259 self.DL.params['allsubtitles'] = True
260 subtitles = self.getSubtitles()
261 self.assertEqual(set(subtitles.keys()), set(['en']))
262 self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
263
264
265 @is_download_test
266 class TestNRKSubtitles(BaseTestSubtitles):
267 url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
268 IE = NRKTVIE
269
270 def test_allsubtitles(self):
271 self.DL.params['writesubtitles'] = True
272 self.DL.params['allsubtitles'] = True
273 subtitles = self.getSubtitles()
274 self.assertEqual(set(subtitles.keys()), set(['no']))
275 self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
276
277
278 @is_download_test
279 class TestRaiPlaySubtitles(BaseTestSubtitles):
280 IE = RaiPlayIE
281
282 def test_subtitles_key(self):
283 self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
284 self.DL.params['writesubtitles'] = True
285 self.DL.params['allsubtitles'] = True
286 subtitles = self.getSubtitles()
287 self.assertEqual(set(subtitles.keys()), set(['it']))
288 self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
289
290 def test_subtitles_array_key(self):
291 self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
292 self.DL.params['writesubtitles'] = True
293 self.DL.params['allsubtitles'] = True
294 subtitles = self.getSubtitles()
295 self.assertEqual(set(subtitles.keys()), set(['it']))
296 self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
297
298
299 @is_download_test
300 class TestVikiSubtitles(BaseTestSubtitles):
301 url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
302 IE = VikiIE
303
304 def test_allsubtitles(self):
305 self.DL.params['writesubtitles'] = True
306 self.DL.params['allsubtitles'] = True
307 subtitles = self.getSubtitles()
308 self.assertEqual(set(subtitles.keys()), set(['en']))
309 self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
310
311
312 @is_download_test
313 class TestThePlatformSubtitles(BaseTestSubtitles):
314 # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
315 # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
316 url = 'theplatform:JFUjUE1_ehvq'
317 IE = ThePlatformIE
318
319 def test_allsubtitles(self):
320 self.DL.params['writesubtitles'] = True
321 self.DL.params['allsubtitles'] = True
322 subtitles = self.getSubtitles()
323 self.assertEqual(set(subtitles.keys()), set(['en']))
324 self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
325
326
327 @is_download_test
328 class TestThePlatformFeedSubtitles(BaseTestSubtitles):
329 url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
330 IE = ThePlatformFeedIE
331
332 def test_allsubtitles(self):
333 self.DL.params['writesubtitles'] = True
334 self.DL.params['allsubtitles'] = True
335 subtitles = self.getSubtitles()
336 self.assertEqual(set(subtitles.keys()), set(['en']))
337 self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
338
339
340 @is_download_test
341 class TestRtveSubtitles(BaseTestSubtitles):
342 url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
343 IE = RTVEALaCartaIE
344
345 def test_allsubtitles(self):
346 print('Skipping, only available from Spain')
347 return
348 self.DL.params['writesubtitles'] = True
349 self.DL.params['allsubtitles'] = True
350 subtitles = self.getSubtitles()
351 self.assertEqual(set(subtitles.keys()), set(['es']))
352 self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
353
354
355 @is_download_test
356 class TestDemocracynowSubtitles(BaseTestSubtitles):
357 url = 'http://www.democracynow.org/shows/2015/7/3'
358 IE = DemocracynowIE
359
360 def test_allsubtitles(self):
361 self.DL.params['writesubtitles'] = True
362 self.DL.params['allsubtitles'] = True
363 subtitles = self.getSubtitles()
364 self.assertEqual(set(subtitles.keys()), set(['en']))
365 self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
366
367 def test_subtitles_in_page(self):
368 self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
369 self.DL.params['writesubtitles'] = True
370 self.DL.params['allsubtitles'] = True
371 subtitles = self.getSubtitles()
372 self.assertEqual(set(subtitles.keys()), set(['en']))
373 self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
374
375
376 @is_download_test
377 class TestPBSSubtitles(BaseTestSubtitles):
378 url = 'https://www.pbs.org/video/how-fantasy-reflects-our-world-picecq/'
379 IE = PBSIE
380
381 def test_allsubtitles(self):
382 self.DL.params['writesubtitles'] = True
383 self.DL.params['allsubtitles'] = True
384 subtitles = self.getSubtitles()
385 self.assertEqual(set(subtitles.keys()), set(['en']))
386
387 def test_subtitles_dfxp_format(self):
388 self.DL.params['writesubtitles'] = True
389 self.DL.params['subtitlesformat'] = 'dfxp'
390 subtitles = self.getSubtitles()
391 self.assertIn(md5(subtitles['en']), ['643b034254cdc3768ff1e750b6b5873b'])
392
393 def test_subtitles_vtt_format(self):
394 self.DL.params['writesubtitles'] = True
395 self.DL.params['subtitlesformat'] = 'vtt'
396 subtitles = self.getSubtitles()
397 self.assertIn(
398 md5(subtitles['en']), ['937a05711555b165d4c55a9667017045', 'f49ea998d6824d94959c8152a368ff73'])
399
400 def test_subtitles_srt_format(self):
401 self.DL.params['writesubtitles'] = True
402 self.DL.params['subtitlesformat'] = 'srt'
403 subtitles = self.getSubtitles()
404 self.assertIn(md5(subtitles['en']), ['2082c21b43759d9bf172931b2f2ca371'])
405
406 def test_subtitles_sami_format(self):
407 self.DL.params['writesubtitles'] = True
408 self.DL.params['subtitlesformat'] = 'sami'
409 subtitles = self.getSubtitles()
410 self.assertIn(md5(subtitles['en']), ['4256b16ac7da6a6780fafd04294e85cd'])
411
412
413 if __name__ == '__main__':
414 unittest.main()