Merge remote-tracking branch 'upstream/master' into bliptv
[youtube-dl] / 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     BlipTVIE,
15     YoutubeIE,
16     DailymotionIE,
17     TEDIE,
18     VimeoIE,
19     WallaIE,
20     CeskaTelevizeIE,
21     LyndaIE,
22     NPOIE,
23     ComedyCentralIE,
24     NRKTVIE,
25     RaiIE,
26     VikiIE,
27     ThePlatformIE,
28     ThePlatformFeedIE,
29     RTVEALaCartaIE,
30     FunnyOrDieIE,
31     DemocracynowIE,
32 )
33
34
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 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']), '4cd9278a35ba2305f47354ee13472260')
70         self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
71         for lang in ['it', 'fr', 'de']:
72             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
73
74     def test_youtube_subtitles_sbv_format(self):
75         self.DL.params['writesubtitles'] = True
76         self.DL.params['subtitlesformat'] = 'sbv'
77         subtitles = self.getSubtitles()
78         self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
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']), '3cb210999d3e021bd6c7f0ea751eab06')
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_translated_subtitles(self):
94         # This video has a subtitles track, which can be translated
95         self.url = 'Ky9eprVWzlI'
96         self.DL.params['writeautomaticsub'] = True
97         self.DL.params['subtitleslangs'] = ['it']
98         subtitles = self.getSubtitles()
99         self.assertTrue(subtitles['it'] is not None)
100
101     def test_youtube_nosubtitles(self):
102         self.DL.expect_warning('video doesn\'t have subtitles')
103         self.url = 'n5BB19UTcdA'
104         self.DL.params['writesubtitles'] = True
105         self.DL.params['allsubtitles'] = True
106         subtitles = self.getSubtitles()
107         self.assertFalse(subtitles)
108
109
110 class TestDailymotionSubtitles(BaseTestSubtitles):
111     url = 'http://www.dailymotion.com/video/xczg00'
112     IE = DailymotionIE
113
114     def test_allsubtitles(self):
115         self.DL.params['writesubtitles'] = True
116         self.DL.params['allsubtitles'] = True
117         subtitles = self.getSubtitles()
118         self.assertTrue(len(subtitles.keys()) >= 6)
119         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
120         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
121         for lang in ['es', 'fr', 'de']:
122             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
123
124     def test_nosubtitles(self):
125         self.DL.expect_warning('video doesn\'t have subtitles')
126         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
127         self.DL.params['writesubtitles'] = True
128         self.DL.params['allsubtitles'] = True
129         subtitles = self.getSubtitles()
130         self.assertFalse(subtitles)
131
132
133 class TestTedSubtitles(BaseTestSubtitles):
134     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
135     IE = TEDIE
136
137     def test_allsubtitles(self):
138         self.DL.params['writesubtitles'] = True
139         self.DL.params['allsubtitles'] = True
140         subtitles = self.getSubtitles()
141         self.assertTrue(len(subtitles.keys()) >= 28)
142         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
143         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
144         for lang in ['es', 'fr', 'de']:
145             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
146
147
148 class TestBlipTVSubtitles(BaseTestSubtitles):
149     url = 'http://blip.tv/a/a-6603250'
150     IE = BlipTVIE
151
152     def test_allsubtitles(self):
153         self.DL.params['writesubtitles'] = True
154         self.DL.params['allsubtitles'] = True
155         subtitles = self.getSubtitles()
156         self.assertEqual(set(subtitles.keys()), set(['en']))
157         self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
158
159
160 class TestVimeoSubtitles(BaseTestSubtitles):
161     url = 'http://vimeo.com/76979871'
162     IE = VimeoIE
163
164     def test_allsubtitles(self):
165         self.DL.params['writesubtitles'] = True
166         self.DL.params['allsubtitles'] = True
167         subtitles = self.getSubtitles()
168         self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
169         self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
170         self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
171
172     def test_nosubtitles(self):
173         self.DL.expect_warning('video doesn\'t have subtitles')
174         self.url = 'http://vimeo.com/56015672'
175         self.DL.params['writesubtitles'] = True
176         self.DL.params['allsubtitles'] = True
177         subtitles = self.getSubtitles()
178         self.assertFalse(subtitles)
179
180
181 class TestWallaSubtitles(BaseTestSubtitles):
182     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
183     IE = WallaIE
184
185     def test_allsubtitles(self):
186         self.DL.expect_warning('Automatic Captions not supported by this server')
187         self.DL.params['writesubtitles'] = True
188         self.DL.params['allsubtitles'] = True
189         subtitles = self.getSubtitles()
190         self.assertEqual(set(subtitles.keys()), set(['heb']))
191         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
192
193     def test_nosubtitles(self):
194         self.DL.expect_warning('video doesn\'t have subtitles')
195         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
196         self.DL.params['writesubtitles'] = True
197         self.DL.params['allsubtitles'] = True
198         subtitles = self.getSubtitles()
199         self.assertFalse(subtitles)
200
201
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 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 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()), set(['nl']))
244         self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
245
246
247 class TestMTVSubtitles(BaseTestSubtitles):
248     url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
249     IE = ComedyCentralIE
250
251     def getInfoDict(self):
252         return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
253
254     def test_allsubtitles(self):
255         self.DL.params['writesubtitles'] = True
256         self.DL.params['allsubtitles'] = True
257         subtitles = self.getSubtitles()
258         self.assertEqual(set(subtitles.keys()), set(['en']))
259         self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
260
261
262 class TestNRKSubtitles(BaseTestSubtitles):
263     url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
264     IE = NRKTVIE
265
266     def test_allsubtitles(self):
267         self.DL.params['writesubtitles'] = True
268         self.DL.params['allsubtitles'] = True
269         subtitles = self.getSubtitles()
270         self.assertEqual(set(subtitles.keys()), set(['no']))
271         self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
272
273
274 class TestRaiSubtitles(BaseTestSubtitles):
275     url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
276     IE = RaiIE
277
278     def test_allsubtitles(self):
279         self.DL.params['writesubtitles'] = True
280         self.DL.params['allsubtitles'] = True
281         subtitles = self.getSubtitles()
282         self.assertEqual(set(subtitles.keys()), set(['it']))
283         self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
284
285
286 class TestVikiSubtitles(BaseTestSubtitles):
287     url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
288     IE = VikiIE
289
290     def test_allsubtitles(self):
291         self.DL.params['writesubtitles'] = True
292         self.DL.params['allsubtitles'] = True
293         subtitles = self.getSubtitles()
294         self.assertEqual(set(subtitles.keys()), set(['en']))
295         self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
296
297
298 class TestThePlatformSubtitles(BaseTestSubtitles):
299     # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
300     # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
301     url = 'theplatform:JFUjUE1_ehvq'
302     IE = ThePlatformIE
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']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
310
311
312 class TestThePlatformFeedSubtitles(BaseTestSubtitles):
313     url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
314     IE = ThePlatformFeedIE
315
316     def test_allsubtitles(self):
317         self.DL.params['writesubtitles'] = True
318         self.DL.params['allsubtitles'] = True
319         subtitles = self.getSubtitles()
320         self.assertEqual(set(subtitles.keys()), set(['en']))
321         self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
322
323
324 class TestRtveSubtitles(BaseTestSubtitles):
325     url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
326     IE = RTVEALaCartaIE
327
328     def test_allsubtitles(self):
329         print('Skipping, only available from Spain')
330         return
331         self.DL.params['writesubtitles'] = True
332         self.DL.params['allsubtitles'] = True
333         subtitles = self.getSubtitles()
334         self.assertEqual(set(subtitles.keys()), set(['es']))
335         self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
336
337
338 class TestFunnyOrDieSubtitles(BaseTestSubtitles):
339     url = 'http://www.funnyordie.com/videos/224829ff6d/judd-apatow-will-direct-your-vine'
340     IE = FunnyOrDieIE
341
342     def test_allsubtitles(self):
343         self.DL.params['writesubtitles'] = True
344         self.DL.params['allsubtitles'] = True
345         subtitles = self.getSubtitles()
346         self.assertEqual(set(subtitles.keys()), set(['en']))
347         self.assertEqual(md5(subtitles['en']), 'c5593c193eacd353596c11c2d4f9ecc4')
348
349
350 class TestDemocracynowSubtitles(BaseTestSubtitles):
351     url = 'http://www.democracynow.org/shows/2015/7/3'
352     IE = DemocracynowIE
353
354     def test_allsubtitles(self):
355         self.DL.params['writesubtitles'] = True
356         self.DL.params['allsubtitles'] = True
357         subtitles = self.getSubtitles()
358         self.assertEqual(set(subtitles.keys()), set(['en']))
359         self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
360
361     def test_subtitles_in_page(self):
362         self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
363         self.DL.params['writesubtitles'] = True
364         self.DL.params['allsubtitles'] = True
365         subtitles = self.getSubtitles()
366         self.assertEqual(set(subtitles.keys()), set(['en']))
367         self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
368
369
370 if __name__ == '__main__':
371     unittest.main()