]> jfr.im git - yt-dlp.git/blob - test/test_subtitles.py
[test] Use `pytest` instead of `nosetests` (#482)
[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 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 )
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
43 def getInfoDict(self):
44 info_dict = self.DL.extract_info(self.url, download=False)
45 return info_dict
46
47 def getSubtitles(self):
48 info_dict = self.getInfoDict()
49 subtitles = info_dict['requested_subtitles']
50 if not subtitles:
51 return subtitles
52 for sub_info in subtitles.values():
53 if sub_info.get('data') is None:
54 uf = self.DL.urlopen(sub_info['url'])
55 sub_info['data'] = uf.read().decode('utf-8')
56 return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
57
58
59 @is_download_test
60 class TestYoutubeSubtitles(BaseTestSubtitles):
61 url = 'QRS8MkLhQmM'
62 IE = YoutubeIE
63
64 def test_youtube_allsubtitles(self):
65 self.DL.params['writesubtitles'] = True
66 self.DL.params['allsubtitles'] = True
67 subtitles = self.getSubtitles()
68 self.assertEqual(len(subtitles.keys()), 13)
69 self.assertEqual(md5(subtitles['en']), '688dd1ce0981683867e7fe6fde2a224b')
70 self.assertEqual(md5(subtitles['it']), '31324d30b8430b309f7f5979a504a769')
71 for lang in ['fr', 'de']:
72 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
73
74 def test_youtube_subtitles_ttml_format(self):
75 self.DL.params['writesubtitles'] = True
76 self.DL.params['subtitlesformat'] = 'ttml'
77 subtitles = self.getSubtitles()
78 self.assertEqual(md5(subtitles['en']), 'c97ddf1217390906fa9fbd34901f3da2')
79
80 def test_youtube_subtitles_vtt_format(self):
81 self.DL.params['writesubtitles'] = True
82 self.DL.params['subtitlesformat'] = 'vtt'
83 subtitles = self.getSubtitles()
84 self.assertEqual(md5(subtitles['en']), 'ae1bd34126571a77aabd4d276b28044d')
85
86 def test_youtube_automatic_captions(self):
87 self.url = '8YoUxe5ncPo'
88 self.DL.params['writeautomaticsub'] = True
89 self.DL.params['subtitleslangs'] = ['it']
90 subtitles = self.getSubtitles()
91 self.assertTrue(subtitles['it'] is not None)
92
93 def test_youtube_no_automatic_captions(self):
94 self.url = 'QRS8MkLhQmM'
95 self.DL.params['writeautomaticsub'] = True
96 subtitles = self.getSubtitles()
97 self.assertTrue(not subtitles)
98
99 def test_youtube_translated_subtitles(self):
100 # This video has a subtitles track, which can be translated
101 self.url = 'i0ZabxXmH4Y'
102 self.DL.params['writeautomaticsub'] = True
103 self.DL.params['subtitleslangs'] = ['it']
104 subtitles = self.getSubtitles()
105 self.assertTrue(subtitles['it'] is not None)
106
107 def test_youtube_nosubtitles(self):
108 self.DL.expect_warning('video doesn\'t have subtitles')
109 self.url = 'n5BB19UTcdA'
110 self.DL.params['writesubtitles'] = True
111 self.DL.params['allsubtitles'] = True
112 subtitles = self.getSubtitles()
113 self.assertFalse(subtitles)
114
115
116 @is_download_test
117 class TestDailymotionSubtitles(BaseTestSubtitles):
118 url = 'http://www.dailymotion.com/video/xczg00'
119 IE = DailymotionIE
120
121 def test_allsubtitles(self):
122 self.DL.params['writesubtitles'] = True
123 self.DL.params['allsubtitles'] = True
124 subtitles = self.getSubtitles()
125 self.assertTrue(len(subtitles.keys()) >= 6)
126 self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
127 self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
128 for lang in ['es', 'fr', 'de']:
129 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
130
131 def test_nosubtitles(self):
132 self.DL.expect_warning('video doesn\'t have subtitles')
133 self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
134 self.DL.params['writesubtitles'] = True
135 self.DL.params['allsubtitles'] = True
136 subtitles = self.getSubtitles()
137 self.assertFalse(subtitles)
138
139
140 @is_download_test
141 class TestTedSubtitles(BaseTestSubtitles):
142 url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
143 IE = TEDIE
144
145 def test_allsubtitles(self):
146 self.DL.params['writesubtitles'] = True
147 self.DL.params['allsubtitles'] = True
148 subtitles = self.getSubtitles()
149 self.assertTrue(len(subtitles.keys()) >= 28)
150 self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
151 self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
152 for lang in ['es', 'fr', 'de']:
153 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
154
155
156 @is_download_test
157 class TestVimeoSubtitles(BaseTestSubtitles):
158 url = 'http://vimeo.com/76979871'
159 IE = VimeoIE
160
161 def test_allsubtitles(self):
162 self.DL.params['writesubtitles'] = True
163 self.DL.params['allsubtitles'] = True
164 subtitles = self.getSubtitles()
165 self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
166 self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
167 self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
168
169 def test_nosubtitles(self):
170 self.DL.expect_warning('video doesn\'t have subtitles')
171 self.url = 'http://vimeo.com/56015672'
172 self.DL.params['writesubtitles'] = True
173 self.DL.params['allsubtitles'] = True
174 subtitles = self.getSubtitles()
175 self.assertFalse(subtitles)
176
177
178 @is_download_test
179 class TestWallaSubtitles(BaseTestSubtitles):
180 url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
181 IE = WallaIE
182
183 def test_allsubtitles(self):
184 self.DL.expect_warning('Automatic Captions not supported by this server')
185 self.DL.params['writesubtitles'] = True
186 self.DL.params['allsubtitles'] = True
187 subtitles = self.getSubtitles()
188 self.assertEqual(set(subtitles.keys()), set(['heb']))
189 self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
190
191 def test_nosubtitles(self):
192 self.DL.expect_warning('video doesn\'t have subtitles')
193 self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
194 self.DL.params['writesubtitles'] = True
195 self.DL.params['allsubtitles'] = True
196 subtitles = self.getSubtitles()
197 self.assertFalse(subtitles)
198
199
200 @is_download_test
201 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
202 url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
203 IE = CeskaTelevizeIE
204
205 def test_allsubtitles(self):
206 self.DL.expect_warning('Automatic Captions not supported by this server')
207 self.DL.params['writesubtitles'] = True
208 self.DL.params['allsubtitles'] = True
209 subtitles = self.getSubtitles()
210 self.assertEqual(set(subtitles.keys()), set(['cs']))
211 self.assertTrue(len(subtitles['cs']) > 20000)
212
213 def test_nosubtitles(self):
214 self.DL.expect_warning('video doesn\'t have subtitles')
215 self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
216 self.DL.params['writesubtitles'] = True
217 self.DL.params['allsubtitles'] = True
218 subtitles = self.getSubtitles()
219 self.assertFalse(subtitles)
220
221
222 @is_download_test
223 class TestLyndaSubtitles(BaseTestSubtitles):
224 url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
225 IE = LyndaIE
226
227 def test_allsubtitles(self):
228 self.DL.params['writesubtitles'] = True
229 self.DL.params['allsubtitles'] = True
230 subtitles = self.getSubtitles()
231 self.assertEqual(set(subtitles.keys()), set(['en']))
232 self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
233
234
235 @is_download_test
236 class TestNPOSubtitles(BaseTestSubtitles):
237 url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
238 IE = NPOIE
239
240 def test_allsubtitles(self):
241 self.DL.params['writesubtitles'] = True
242 self.DL.params['allsubtitles'] = True
243 subtitles = self.getSubtitles()
244 self.assertEqual(set(subtitles.keys()), set(['nl']))
245 self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
246
247
248 @is_download_test
249 class TestMTVSubtitles(BaseTestSubtitles):
250 url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
251 IE = ComedyCentralIE
252
253 def getInfoDict(self):
254 return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
255
256 def test_allsubtitles(self):
257 self.DL.params['writesubtitles'] = True
258 self.DL.params['allsubtitles'] = True
259 subtitles = self.getSubtitles()
260 self.assertEqual(set(subtitles.keys()), set(['en']))
261 self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
262
263
264 @is_download_test
265 class TestNRKSubtitles(BaseTestSubtitles):
266 url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
267 IE = NRKTVIE
268
269 def test_allsubtitles(self):
270 self.DL.params['writesubtitles'] = True
271 self.DL.params['allsubtitles'] = True
272 subtitles = self.getSubtitles()
273 self.assertEqual(set(subtitles.keys()), set(['no']))
274 self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
275
276
277 @is_download_test
278 class TestRaiPlaySubtitles(BaseTestSubtitles):
279 IE = RaiPlayIE
280
281 def test_subtitles_key(self):
282 self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
283 self.DL.params['writesubtitles'] = True
284 self.DL.params['allsubtitles'] = True
285 subtitles = self.getSubtitles()
286 self.assertEqual(set(subtitles.keys()), set(['it']))
287 self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
288
289 def test_subtitles_array_key(self):
290 self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
291 self.DL.params['writesubtitles'] = True
292 self.DL.params['allsubtitles'] = True
293 subtitles = self.getSubtitles()
294 self.assertEqual(set(subtitles.keys()), set(['it']))
295 self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
296
297
298 @is_download_test
299 class TestVikiSubtitles(BaseTestSubtitles):
300 url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
301 IE = VikiIE
302
303 def test_allsubtitles(self):
304 self.DL.params['writesubtitles'] = True
305 self.DL.params['allsubtitles'] = True
306 subtitles = self.getSubtitles()
307 self.assertEqual(set(subtitles.keys()), set(['en']))
308 self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
309
310
311 @is_download_test
312 class TestThePlatformSubtitles(BaseTestSubtitles):
313 # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
314 # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
315 url = 'theplatform:JFUjUE1_ehvq'
316 IE = ThePlatformIE
317
318 def test_allsubtitles(self):
319 self.DL.params['writesubtitles'] = True
320 self.DL.params['allsubtitles'] = True
321 subtitles = self.getSubtitles()
322 self.assertEqual(set(subtitles.keys()), set(['en']))
323 self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
324
325
326 @is_download_test
327 class TestThePlatformFeedSubtitles(BaseTestSubtitles):
328 url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
329 IE = ThePlatformFeedIE
330
331 def test_allsubtitles(self):
332 self.DL.params['writesubtitles'] = True
333 self.DL.params['allsubtitles'] = True
334 subtitles = self.getSubtitles()
335 self.assertEqual(set(subtitles.keys()), set(['en']))
336 self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
337
338
339 @is_download_test
340 class TestRtveSubtitles(BaseTestSubtitles):
341 url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
342 IE = RTVEALaCartaIE
343
344 def test_allsubtitles(self):
345 print('Skipping, only available from Spain')
346 return
347 self.DL.params['writesubtitles'] = True
348 self.DL.params['allsubtitles'] = True
349 subtitles = self.getSubtitles()
350 self.assertEqual(set(subtitles.keys()), set(['es']))
351 self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
352
353
354 @is_download_test
355 class TestDemocracynowSubtitles(BaseTestSubtitles):
356 url = 'http://www.democracynow.org/shows/2015/7/3'
357 IE = DemocracynowIE
358
359 def test_allsubtitles(self):
360 self.DL.params['writesubtitles'] = True
361 self.DL.params['allsubtitles'] = True
362 subtitles = self.getSubtitles()
363 self.assertEqual(set(subtitles.keys()), set(['en']))
364 self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
365
366 def test_subtitles_in_page(self):
367 self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
368 self.DL.params['writesubtitles'] = True
369 self.DL.params['allsubtitles'] = True
370 subtitles = self.getSubtitles()
371 self.assertEqual(set(subtitles.keys()), set(['en']))
372 self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
373
374
375 if __name__ == '__main__':
376 unittest.main()