]> asedeno.scripts.mit.edu Git - youtube-dl.git/blob - test/test_subtitles.py
c250473be1ff8c81fa1bf17231065f2a2420dec0
[youtube-dl.git] / test / test_subtitles.py
1 #!/usr/bin/env python
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
11
12
13 from youtube_dl.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 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         if not self.IE.working():
42             print('Skipping: %s marked as not _WORKING' % self.IE.ie_key())
43             self.skipTest('IE marked as not _WORKING')
44
45     def getInfoDict(self):
46         info_dict = self.DL.extract_info(self.url, download=False)
47         return info_dict
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('utf-8')
58         return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
59
60
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']), '3cb210999d3e021bd6c7f0ea751eab06')
71         self.assertEqual(md5(subtitles['it']), '6d752b98c31f1cf8d597050c7a2cb4b5')
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']), 'e306f8c42842f723447d9f63ad65df54')
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']), '3cb210999d3e021bd6c7f0ea751eab06')
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_translated_subtitles(self):
95         # This video has a subtitles track, which can be translated
96         self.url = 'Ky9eprVWzlI'
97         self.DL.params['writeautomaticsub'] = True
98         self.DL.params['subtitleslangs'] = ['it']
99         subtitles = self.getSubtitles()
100         self.assertTrue(subtitles['it'] is not None)
101
102     def test_youtube_nosubtitles(self):
103         self.DL.expect_warning('video doesn\'t have subtitles')
104         self.url = 'n5BB19UTcdA'
105         self.DL.params['writesubtitles'] = True
106         self.DL.params['allsubtitles'] = True
107         subtitles = self.getSubtitles()
108         self.assertFalse(subtitles)
109
110
111 class TestDailymotionSubtitles(BaseTestSubtitles):
112     url = 'http://www.dailymotion.com/video/xczg00'
113     IE = DailymotionIE
114
115     def test_allsubtitles(self):
116         self.DL.params['writesubtitles'] = True
117         self.DL.params['allsubtitles'] = True
118         subtitles = self.getSubtitles()
119         self.assertTrue(len(subtitles.keys()) >= 6)
120         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
121         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
122         for lang in ['es', 'fr', 'de']:
123             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
124
125     def test_nosubtitles(self):
126         self.DL.expect_warning('video doesn\'t have subtitles')
127         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
128         self.DL.params['writesubtitles'] = True
129         self.DL.params['allsubtitles'] = True
130         subtitles = self.getSubtitles()
131         self.assertFalse(subtitles)
132
133
134 class TestTedSubtitles(BaseTestSubtitles):
135     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
136     IE = TEDIE
137
138     def test_allsubtitles(self):
139         self.DL.params['writesubtitles'] = True
140         self.DL.params['allsubtitles'] = True
141         subtitles = self.getSubtitles()
142         self.assertTrue(len(subtitles.keys()) >= 28)
143         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
144         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
145         for lang in ['es', 'fr', 'de']:
146             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
147
148
149 class TestVimeoSubtitles(BaseTestSubtitles):
150     url = 'http://vimeo.com/76979871'
151     IE = VimeoIE
152
153     def test_allsubtitles(self):
154         self.DL.params['writesubtitles'] = True
155         self.DL.params['allsubtitles'] = True
156         subtitles = self.getSubtitles()
157         self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
158         self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
159         self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
160
161     def test_nosubtitles(self):
162         self.DL.expect_warning('video doesn\'t have subtitles')
163         self.url = 'http://vimeo.com/56015672'
164         self.DL.params['writesubtitles'] = True
165         self.DL.params['allsubtitles'] = True
166         subtitles = self.getSubtitles()
167         self.assertFalse(subtitles)
168
169
170 class TestWallaSubtitles(BaseTestSubtitles):
171     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
172     IE = WallaIE
173
174     def test_allsubtitles(self):
175         self.DL.expect_warning('Automatic Captions not supported by this server')
176         self.DL.params['writesubtitles'] = True
177         self.DL.params['allsubtitles'] = True
178         subtitles = self.getSubtitles()
179         self.assertEqual(set(subtitles.keys()), set(['heb']))
180         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
181
182     def test_nosubtitles(self):
183         self.DL.expect_warning('video doesn\'t have subtitles')
184         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
185         self.DL.params['writesubtitles'] = True
186         self.DL.params['allsubtitles'] = True
187         subtitles = self.getSubtitles()
188         self.assertFalse(subtitles)
189
190
191 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
192     url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
193     IE = CeskaTelevizeIE
194
195     def test_allsubtitles(self):
196         self.DL.expect_warning('Automatic Captions not supported by this server')
197         self.DL.params['writesubtitles'] = True
198         self.DL.params['allsubtitles'] = True
199         subtitles = self.getSubtitles()
200         self.assertEqual(set(subtitles.keys()), set(['cs']))
201         self.assertTrue(len(subtitles['cs']) > 20000)
202
203     def test_nosubtitles(self):
204         self.DL.expect_warning('video doesn\'t have subtitles')
205         self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
206         self.DL.params['writesubtitles'] = True
207         self.DL.params['allsubtitles'] = True
208         subtitles = self.getSubtitles()
209         self.assertFalse(subtitles)
210
211
212 class TestLyndaSubtitles(BaseTestSubtitles):
213     url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
214     IE = LyndaIE
215
216     def test_allsubtitles(self):
217         self.DL.params['writesubtitles'] = True
218         self.DL.params['allsubtitles'] = True
219         subtitles = self.getSubtitles()
220         self.assertEqual(set(subtitles.keys()), set(['en']))
221         self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
222
223
224 class TestNPOSubtitles(BaseTestSubtitles):
225     url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
226     IE = NPOIE
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(['nl']))
233         self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
234
235
236 class TestMTVSubtitles(BaseTestSubtitles):
237     url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
238     IE = ComedyCentralIE
239
240     def getInfoDict(self):
241         return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
242
243     def test_allsubtitles(self):
244         self.DL.params['writesubtitles'] = True
245         self.DL.params['allsubtitles'] = True
246         subtitles = self.getSubtitles()
247         self.assertEqual(set(subtitles.keys()), set(['en']))
248         self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
249
250
251 class TestNRKSubtitles(BaseTestSubtitles):
252     url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
253     IE = NRKTVIE
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()), set(['no']))
260         self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
261
262
263 class TestRaiPlaySubtitles(BaseTestSubtitles):
264     IE = RaiPlayIE
265
266     def test_subtitles_key(self):
267         self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
268         self.DL.params['writesubtitles'] = True
269         self.DL.params['allsubtitles'] = True
270         subtitles = self.getSubtitles()
271         self.assertEqual(set(subtitles.keys()), set(['it']))
272         self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
273
274     def test_subtitles_array_key(self):
275         self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
276         self.DL.params['writesubtitles'] = True
277         self.DL.params['allsubtitles'] = True
278         subtitles = self.getSubtitles()
279         self.assertEqual(set(subtitles.keys()), set(['it']))
280         self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
281
282
283 class TestVikiSubtitles(BaseTestSubtitles):
284     url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
285     IE = VikiIE
286
287     def test_allsubtitles(self):
288         self.DL.params['writesubtitles'] = True
289         self.DL.params['allsubtitles'] = True
290         subtitles = self.getSubtitles()
291         self.assertEqual(set(subtitles.keys()), set(['en']))
292         self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
293
294
295 class TestThePlatformSubtitles(BaseTestSubtitles):
296     # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
297     # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
298     url = 'theplatform:JFUjUE1_ehvq'
299     IE = ThePlatformIE
300
301     def test_allsubtitles(self):
302         self.DL.params['writesubtitles'] = True
303         self.DL.params['allsubtitles'] = True
304         subtitles = self.getSubtitles()
305         self.assertEqual(set(subtitles.keys()), set(['en']))
306         self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
307
308
309 class TestThePlatformFeedSubtitles(BaseTestSubtitles):
310     url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
311     IE = ThePlatformFeedIE
312
313     def test_allsubtitles(self):
314         self.DL.params['writesubtitles'] = True
315         self.DL.params['allsubtitles'] = True
316         subtitles = self.getSubtitles()
317         self.assertEqual(set(subtitles.keys()), set(['en']))
318         self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
319
320
321 class TestRtveSubtitles(BaseTestSubtitles):
322     url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
323     IE = RTVEALaCartaIE
324
325     def test_allsubtitles(self):
326         print('Skipping, only available from Spain')
327         return
328         self.DL.params['writesubtitles'] = True
329         self.DL.params['allsubtitles'] = True
330         subtitles = self.getSubtitles()
331         self.assertEqual(set(subtitles.keys()), set(['es']))
332         self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
333
334
335 class TestDemocracynowSubtitles(BaseTestSubtitles):
336     url = 'http://www.democracynow.org/shows/2015/7/3'
337     IE = DemocracynowIE
338
339     def test_allsubtitles(self):
340         self.DL.params['writesubtitles'] = True
341         self.DL.params['allsubtitles'] = True
342         subtitles = self.getSubtitles()
343         self.assertEqual(set(subtitles.keys()), set(['en']))
344         self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
345
346     def test_subtitles_in_page(self):
347         self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
348         self.DL.params['writesubtitles'] = True
349         self.DL.params['allsubtitles'] = True
350         subtitles = self.getSubtitles()
351         self.assertEqual(set(subtitles.keys()), set(['en']))
352         self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
353
354
355 if __name__ == '__main__':
356     unittest.main()