]> jfr.im git - yt-dlp.git/blob - test/test_subtitles.py
[cleanup] Consistent style for file heads
[yt-dlp.git] / 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('Skipping: %s marked as not _WORKING' % self.IE.ie_key())
44 self.skipTest('IE marked as not _WORKING')
45
46 def getInfoDict(self):
47 info_dict = self.DL.extract_info(self.url, download=False)
48 return info_dict
49
50 def getSubtitles(self):
51 info_dict = self.getInfoDict()
52 subtitles = info_dict['requested_subtitles']
53 if not subtitles:
54 return subtitles
55 for sub_info in subtitles.values():
56 if sub_info.get('data') is None:
57 uf = self.DL.urlopen(sub_info['url'])
58 sub_info['data'] = uf.read().decode()
59 return {l: sub_info['data'] for l, sub_info in subtitles.items()}
60
61
62 @is_download_test
63 class TestYoutubeSubtitles(BaseTestSubtitles):
64 # Available subtitles for QRS8MkLhQmM:
65 # Language formats
66 # ru vtt, ttml, srv3, srv2, srv1, json3
67 # fr vtt, ttml, srv3, srv2, srv1, json3
68 # en vtt, ttml, srv3, srv2, srv1, json3
69 # nl vtt, ttml, srv3, srv2, srv1, json3
70 # de vtt, ttml, srv3, srv2, srv1, json3
71 # ko vtt, ttml, srv3, srv2, srv1, json3
72 # it vtt, ttml, srv3, srv2, srv1, json3
73 # zh-Hant vtt, ttml, srv3, srv2, srv1, json3
74 # hi vtt, ttml, srv3, srv2, srv1, json3
75 # pt-BR vtt, ttml, srv3, srv2, srv1, json3
76 # es-MX vtt, ttml, srv3, srv2, srv1, json3
77 # ja vtt, ttml, srv3, srv2, srv1, json3
78 # pl vtt, ttml, srv3, srv2, srv1, json3
79 url = 'QRS8MkLhQmM'
80 IE = YoutubeIE
81
82 def test_youtube_allsubtitles(self):
83 self.DL.params['writesubtitles'] = True
84 self.DL.params['allsubtitles'] = True
85 subtitles = self.getSubtitles()
86 self.assertEqual(len(subtitles.keys()), 13)
87 self.assertEqual(md5(subtitles['en']), 'ae1bd34126571a77aabd4d276b28044d')
88 self.assertEqual(md5(subtitles['it']), '0e0b667ba68411d88fd1c5f4f4eab2f9')
89 for lang in ['fr', 'de']:
90 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
91
92 def _test_subtitles_format(self, fmt, md5_hash, lang='en'):
93 self.DL.params['writesubtitles'] = True
94 self.DL.params['subtitlesformat'] = fmt
95 subtitles = self.getSubtitles()
96 self.assertEqual(md5(subtitles[lang]), md5_hash)
97
98 def test_youtube_subtitles_ttml_format(self):
99 self._test_subtitles_format('ttml', 'c97ddf1217390906fa9fbd34901f3da2')
100
101 def test_youtube_subtitles_vtt_format(self):
102 self._test_subtitles_format('vtt', 'ae1bd34126571a77aabd4d276b28044d')
103
104 def test_youtube_subtitles_json3_format(self):
105 self._test_subtitles_format('json3', '688dd1ce0981683867e7fe6fde2a224b')
106
107 def _test_automatic_captions(self, url, lang):
108 self.url = url
109 self.DL.params['writeautomaticsub'] = True
110 self.DL.params['subtitleslangs'] = [lang]
111 subtitles = self.getSubtitles()
112 self.assertTrue(subtitles[lang] is not None)
113
114 def test_youtube_automatic_captions(self):
115 # Available automatic captions for 8YoUxe5ncPo:
116 # Language formats (all in vtt, ttml, srv3, srv2, srv1, json3)
117 # gu, zh-Hans, zh-Hant, gd, ga, gl, lb, la, lo, tt, tr,
118 # lv, lt, tk, th, tg, te, fil, haw, yi, ceb, yo, de, da,
119 # el, eo, en, eu, et, es, ru, rw, ro, bn, be, bg, uk, jv,
120 # bs, ja, or, xh, co, ca, cy, cs, ps, pt, pa, vi, pl, hy,
121 # hr, ht, hu, hmn, hi, ha, mg, uz, ml, mn, mi, mk, ur,
122 # mt, ms, mr, ug, ta, my, af, sw, is, am,
123 # *it*, iw, sv, ar,
124 # su, zu, az, id, ig, nl, no, ne, ny, fr, ku, fy, fa, fi,
125 # ka, kk, sr, sq, ko, kn, km, st, sk, si, so, sn, sm, sl,
126 # ky, sd
127 # ...
128 self._test_automatic_captions('8YoUxe5ncPo', 'it')
129
130 @unittest.skip('Video unavailable')
131 def test_youtube_translated_subtitles(self):
132 # This video has a subtitles track, which can be translated (#4555)
133 self._test_automatic_captions('Ky9eprVWzlI', 'it')
134
135 def test_youtube_nosubtitles(self):
136 self.DL.expect_warning('video doesn\'t have subtitles')
137 # Available automatic captions for 8YoUxe5ncPo:
138 # ...
139 # 8YoUxe5ncPo has no subtitles
140 self.url = '8YoUxe5ncPo'
141 self.DL.params['writesubtitles'] = True
142 self.DL.params['allsubtitles'] = True
143 subtitles = self.getSubtitles()
144 self.assertFalse(subtitles)
145
146
147 @is_download_test
148 class TestDailymotionSubtitles(BaseTestSubtitles):
149 url = 'http://www.dailymotion.com/video/xczg00'
150 IE = DailymotionIE
151
152 def test_allsubtitles(self):
153 self.DL.params['writesubtitles'] = True
154 self.DL.params['allsubtitles'] = True
155 subtitles = self.getSubtitles()
156 self.assertTrue(len(subtitles.keys()) >= 6)
157 self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
158 self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
159 for lang in ['es', 'fr', 'de']:
160 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
161
162 def test_nosubtitles(self):
163 self.DL.expect_warning('video doesn\'t have subtitles')
164 self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
165 self.DL.params['writesubtitles'] = True
166 self.DL.params['allsubtitles'] = True
167 subtitles = self.getSubtitles()
168 self.assertFalse(subtitles)
169
170
171 @is_download_test
172 @unittest.skip('IE broken')
173 class TestTedSubtitles(BaseTestSubtitles):
174 url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
175 IE = TedTalkIE
176
177 def test_allsubtitles(self):
178 self.DL.params['writesubtitles'] = True
179 self.DL.params['allsubtitles'] = True
180 subtitles = self.getSubtitles()
181 self.assertTrue(len(subtitles.keys()) >= 28)
182 self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
183 self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
184 for lang in ['es', 'fr', 'de']:
185 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
186
187
188 @is_download_test
189 class TestVimeoSubtitles(BaseTestSubtitles):
190 url = 'http://vimeo.com/76979871'
191 IE = VimeoIE
192
193 def test_allsubtitles(self):
194 self.DL.params['writesubtitles'] = True
195 self.DL.params['allsubtitles'] = True
196 subtitles = self.getSubtitles()
197 self.assertEqual(set(subtitles.keys()), {'de', 'en', 'es', 'fr'})
198 self.assertEqual(md5(subtitles['en']), '386cbc9320b94e25cb364b97935e5dd1')
199 self.assertEqual(md5(subtitles['fr']), 'c9b69eef35bc6641c0d4da8a04f9dfac')
200
201 def test_nosubtitles(self):
202 self.DL.expect_warning('video doesn\'t have subtitles')
203 self.url = 'http://vimeo.com/68093876'
204 self.DL.params['writesubtitles'] = True
205 self.DL.params['allsubtitles'] = True
206 subtitles = self.getSubtitles()
207 self.assertFalse(subtitles)
208
209
210 @is_download_test
211 @unittest.skip('IE broken')
212 class TestWallaSubtitles(BaseTestSubtitles):
213 url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
214 IE = WallaIE
215
216 def test_allsubtitles(self):
217 self.DL.expect_warning('Automatic Captions not supported by this server')
218 self.DL.params['writesubtitles'] = True
219 self.DL.params['allsubtitles'] = True
220 subtitles = self.getSubtitles()
221 self.assertEqual(set(subtitles.keys()), {'heb'})
222 self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
223
224 def test_nosubtitles(self):
225 self.DL.expect_warning('video doesn\'t have subtitles')
226 self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
227 self.DL.params['writesubtitles'] = True
228 self.DL.params['allsubtitles'] = True
229 subtitles = self.getSubtitles()
230 self.assertFalse(subtitles)
231
232
233 @is_download_test
234 @unittest.skip('IE broken')
235 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
236 url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
237 IE = CeskaTelevizeIE
238
239 def test_allsubtitles(self):
240 self.DL.expect_warning('Automatic Captions not supported by this server')
241 self.DL.params['writesubtitles'] = True
242 self.DL.params['allsubtitles'] = True
243 subtitles = self.getSubtitles()
244 self.assertEqual(set(subtitles.keys()), {'cs'})
245 self.assertTrue(len(subtitles['cs']) > 20000)
246
247 def test_nosubtitles(self):
248 self.DL.expect_warning('video doesn\'t have subtitles')
249 self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
250 self.DL.params['writesubtitles'] = True
251 self.DL.params['allsubtitles'] = True
252 subtitles = self.getSubtitles()
253 self.assertFalse(subtitles)
254
255
256 @is_download_test
257 @unittest.skip('IE broken')
258 class TestLyndaSubtitles(BaseTestSubtitles):
259 url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
260 IE = LyndaIE
261
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()), {'en'})
267 self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
268
269
270 @is_download_test
271 @unittest.skip('IE broken')
272 class TestNPOSubtitles(BaseTestSubtitles):
273 url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
274 IE = NPOIE
275
276 def test_allsubtitles(self):
277 self.DL.params['writesubtitles'] = True
278 self.DL.params['allsubtitles'] = True
279 subtitles = self.getSubtitles()
280 self.assertEqual(set(subtitles.keys()), {'nl'})
281 self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
282
283
284 @is_download_test
285 @unittest.skip('IE broken')
286 class TestMTVSubtitles(BaseTestSubtitles):
287 url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
288 IE = ComedyCentralIE
289
290 def getInfoDict(self):
291 return super().getInfoDict()['entries'][0]
292
293 def test_allsubtitles(self):
294 self.DL.params['writesubtitles'] = True
295 self.DL.params['allsubtitles'] = True
296 subtitles = self.getSubtitles()
297 self.assertEqual(set(subtitles.keys()), {'en'})
298 self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
299
300
301 @is_download_test
302 class TestNRKSubtitles(BaseTestSubtitles):
303 url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
304 IE = NRKTVIE
305
306 def test_allsubtitles(self):
307 self.DL.params['writesubtitles'] = True
308 self.DL.params['allsubtitles'] = True
309 subtitles = self.getSubtitles()
310 self.assertEqual(set(subtitles.keys()), {'nb-ttv'})
311 self.assertEqual(md5(subtitles['nb-ttv']), '67e06ff02d0deaf975e68f6cb8f6a149')
312
313
314 @is_download_test
315 class TestRaiPlaySubtitles(BaseTestSubtitles):
316 IE = RaiPlayIE
317
318 def test_subtitles_key(self):
319 self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
320 self.DL.params['writesubtitles'] = True
321 self.DL.params['allsubtitles'] = True
322 subtitles = self.getSubtitles()
323 self.assertEqual(set(subtitles.keys()), {'it'})
324 self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
325
326 def test_subtitles_array_key(self):
327 self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
328 self.DL.params['writesubtitles'] = True
329 self.DL.params['allsubtitles'] = True
330 subtitles = self.getSubtitles()
331 self.assertEqual(set(subtitles.keys()), {'it'})
332 self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
333
334
335 @is_download_test
336 @unittest.skip('IE broken - DRM only')
337 class TestVikiSubtitles(BaseTestSubtitles):
338 url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
339 IE = VikiIE
340
341 def test_allsubtitles(self):
342 self.DL.params['writesubtitles'] = True
343 self.DL.params['allsubtitles'] = True
344 subtitles = self.getSubtitles()
345 self.assertEqual(set(subtitles.keys()), {'en'})
346 self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
347
348
349 @is_download_test
350 class TestThePlatformSubtitles(BaseTestSubtitles):
351 # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
352 # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
353 url = 'theplatform:JFUjUE1_ehvq'
354 IE = ThePlatformIE
355
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()), {'en'})
361 self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
362
363
364 @is_download_test
365 @unittest.skip('IE broken')
366 class TestThePlatformFeedSubtitles(BaseTestSubtitles):
367 url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
368 IE = ThePlatformFeedIE
369
370 def test_allsubtitles(self):
371 self.DL.params['writesubtitles'] = True
372 self.DL.params['allsubtitles'] = True
373 subtitles = self.getSubtitles()
374 self.assertEqual(set(subtitles.keys()), {'en'})
375 self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
376
377
378 @is_download_test
379 class TestRtveSubtitles(BaseTestSubtitles):
380 url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
381 IE = RTVEALaCartaIE
382
383 def test_allsubtitles(self):
384 print('Skipping, only available from Spain')
385 return
386 self.DL.params['writesubtitles'] = True
387 self.DL.params['allsubtitles'] = True
388 subtitles = self.getSubtitles()
389 self.assertEqual(set(subtitles.keys()), {'es'})
390 self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
391
392
393 @is_download_test
394 class TestDemocracynowSubtitles(BaseTestSubtitles):
395 url = 'http://www.democracynow.org/shows/2015/7/3'
396 IE = DemocracynowIE
397
398 def test_allsubtitles(self):
399 self.DL.params['writesubtitles'] = True
400 self.DL.params['allsubtitles'] = True
401 subtitles = self.getSubtitles()
402 self.assertEqual(set(subtitles.keys()), {'en'})
403 self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')
404
405 def test_subtitles_in_page(self):
406 self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
407 self.DL.params['writesubtitles'] = True
408 self.DL.params['allsubtitles'] = True
409 subtitles = self.getSubtitles()
410 self.assertEqual(set(subtitles.keys()), {'en'})
411 self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')
412
413
414 @is_download_test
415 class TestPBSSubtitles(BaseTestSubtitles):
416 url = 'https://www.pbs.org/video/how-fantasy-reflects-our-world-picecq/'
417 IE = PBSIE
418
419 def test_allsubtitles(self):
420 self.DL.params['writesubtitles'] = True
421 self.DL.params['allsubtitles'] = True
422 subtitles = self.getSubtitles()
423 self.assertEqual(set(subtitles.keys()), {'en'})
424
425 def test_subtitles_dfxp_format(self):
426 self.DL.params['writesubtitles'] = True
427 self.DL.params['subtitlesformat'] = 'dfxp'
428 subtitles = self.getSubtitles()
429 self.assertIn(md5(subtitles['en']), ['643b034254cdc3768ff1e750b6b5873b'])
430
431 def test_subtitles_vtt_format(self):
432 self.DL.params['writesubtitles'] = True
433 self.DL.params['subtitlesformat'] = 'vtt'
434 subtitles = self.getSubtitles()
435 self.assertIn(
436 md5(subtitles['en']), ['937a05711555b165d4c55a9667017045', 'f49ea998d6824d94959c8152a368ff73'])
437
438 def test_subtitles_srt_format(self):
439 self.DL.params['writesubtitles'] = True
440 self.DL.params['subtitlesformat'] = 'srt'
441 subtitles = self.getSubtitles()
442 self.assertIn(md5(subtitles['en']), ['2082c21b43759d9bf172931b2f2ca371'])
443
444 def test_subtitles_sami_format(self):
445 self.DL.params['writesubtitles'] = True
446 self.DL.params['subtitlesformat'] = 'sami'
447 subtitles = self.getSubtitles()
448 self.assertIn(md5(subtitles['en']), ['4256b16ac7da6a6780fafd04294e85cd'])
449
450
451 if __name__ == '__main__':
452 unittest.main()