[rtve] Extract subtitles
[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     RTVEALaCartaIE,
29 )
30
31
32 class BaseTestSubtitles(unittest.TestCase):
33     url = None
34     IE = None
35
36     def setUp(self):
37         self.DL = FakeYDL()
38         self.ie = self.IE()
39         self.DL.add_info_extractor(self.ie)
40
41     def getInfoDict(self):
42         info_dict = self.DL.extract_info(self.url, download=False)
43         return info_dict
44
45     def getSubtitles(self):
46         info_dict = self.getInfoDict()
47         subtitles = info_dict['requested_subtitles']
48         if not subtitles:
49             return subtitles
50         for sub_info in subtitles.values():
51             if sub_info.get('data') is None:
52                 uf = self.DL.urlopen(sub_info['url'])
53                 sub_info['data'] = uf.read().decode('utf-8')
54         return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
55
56
57 class TestYoutubeSubtitles(BaseTestSubtitles):
58     url = 'QRS8MkLhQmM'
59     IE = YoutubeIE
60
61     def test_youtube_allsubtitles(self):
62         self.DL.params['writesubtitles'] = True
63         self.DL.params['allsubtitles'] = True
64         subtitles = self.getSubtitles()
65         self.assertEqual(len(subtitles.keys()), 13)
66         self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
67         self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
68         for lang in ['it', 'fr', 'de']:
69             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
70
71     def test_youtube_subtitles_sbv_format(self):
72         self.DL.params['writesubtitles'] = True
73         self.DL.params['subtitlesformat'] = 'sbv'
74         subtitles = self.getSubtitles()
75         self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
76
77     def test_youtube_subtitles_vtt_format(self):
78         self.DL.params['writesubtitles'] = True
79         self.DL.params['subtitlesformat'] = 'vtt'
80         subtitles = self.getSubtitles()
81         self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
82
83     def test_youtube_automatic_captions(self):
84         self.url = '8YoUxe5ncPo'
85         self.DL.params['writeautomaticsub'] = True
86         self.DL.params['subtitleslangs'] = ['it']
87         subtitles = self.getSubtitles()
88         self.assertTrue(subtitles['it'] is not None)
89
90     def test_youtube_translated_subtitles(self):
91         # This video has a subtitles track, which can be translated
92         self.url = 'Ky9eprVWzlI'
93         self.DL.params['writeautomaticsub'] = True
94         self.DL.params['subtitleslangs'] = ['it']
95         subtitles = self.getSubtitles()
96         self.assertTrue(subtitles['it'] is not None)
97
98     def test_youtube_nosubtitles(self):
99         self.DL.expect_warning('video doesn\'t have subtitles')
100         self.url = 'n5BB19UTcdA'
101         self.DL.params['writesubtitles'] = True
102         self.DL.params['allsubtitles'] = True
103         subtitles = self.getSubtitles()
104         self.assertFalse(subtitles)
105
106
107 class TestDailymotionSubtitles(BaseTestSubtitles):
108     url = 'http://www.dailymotion.com/video/xczg00'
109     IE = DailymotionIE
110
111     def test_allsubtitles(self):
112         self.DL.params['writesubtitles'] = True
113         self.DL.params['allsubtitles'] = True
114         subtitles = self.getSubtitles()
115         self.assertTrue(len(subtitles.keys()) >= 6)
116         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
117         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
118         for lang in ['es', 'fr', 'de']:
119             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
120
121     def test_nosubtitles(self):
122         self.DL.expect_warning('video doesn\'t have subtitles')
123         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
124         self.DL.params['writesubtitles'] = True
125         self.DL.params['allsubtitles'] = True
126         subtitles = self.getSubtitles()
127         self.assertFalse(subtitles)
128
129
130 class TestTedSubtitles(BaseTestSubtitles):
131     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
132     IE = TEDIE
133
134     def test_allsubtitles(self):
135         self.DL.params['writesubtitles'] = True
136         self.DL.params['allsubtitles'] = True
137         subtitles = self.getSubtitles()
138         self.assertTrue(len(subtitles.keys()) >= 28)
139         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
140         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
141         for lang in ['es', 'fr', 'de']:
142             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
143
144
145 class TestBlipTVSubtitles(BaseTestSubtitles):
146     url = 'http://blip.tv/a/a-6603250'
147     IE = BlipTVIE
148
149     def test_allsubtitles(self):
150         self.DL.params['writesubtitles'] = True
151         self.DL.params['allsubtitles'] = True
152         subtitles = self.getSubtitles()
153         self.assertEqual(set(subtitles.keys()), set(['en']))
154         self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
155
156
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 class TestWallaSubtitles(BaseTestSubtitles):
179     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
180     IE = WallaIE
181
182     def test_allsubtitles(self):
183         self.DL.expect_warning('Automatic Captions not supported by this server')
184         self.DL.params['writesubtitles'] = True
185         self.DL.params['allsubtitles'] = True
186         subtitles = self.getSubtitles()
187         self.assertEqual(set(subtitles.keys()), set(['heb']))
188         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
189
190     def test_nosubtitles(self):
191         self.DL.expect_warning('video doesn\'t have subtitles')
192         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
193         self.DL.params['writesubtitles'] = True
194         self.DL.params['allsubtitles'] = True
195         subtitles = self.getSubtitles()
196         self.assertFalse(subtitles)
197
198
199 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
200     url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
201     IE = CeskaTelevizeIE
202
203     def test_allsubtitles(self):
204         self.DL.expect_warning('Automatic Captions not supported by this server')
205         self.DL.params['writesubtitles'] = True
206         self.DL.params['allsubtitles'] = True
207         subtitles = self.getSubtitles()
208         self.assertEqual(set(subtitles.keys()), set(['cs']))
209         self.assertTrue(len(subtitles['cs']) > 20000)
210
211     def test_nosubtitles(self):
212         self.DL.expect_warning('video doesn\'t have subtitles')
213         self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
214         self.DL.params['writesubtitles'] = True
215         self.DL.params['allsubtitles'] = True
216         subtitles = self.getSubtitles()
217         self.assertFalse(subtitles)
218
219
220 class TestLyndaSubtitles(BaseTestSubtitles):
221     url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
222     IE = LyndaIE
223
224     def test_allsubtitles(self):
225         self.DL.params['writesubtitles'] = True
226         self.DL.params['allsubtitles'] = True
227         subtitles = self.getSubtitles()
228         self.assertEqual(set(subtitles.keys()), set(['en']))
229         self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
230
231
232 class TestNPOSubtitles(BaseTestSubtitles):
233     url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
234     IE = NPOIE
235
236     def test_allsubtitles(self):
237         self.DL.params['writesubtitles'] = True
238         self.DL.params['allsubtitles'] = True
239         subtitles = self.getSubtitles()
240         self.assertEqual(set(subtitles.keys()), set(['nl']))
241         self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
242
243
244 class TestMTVSubtitles(BaseTestSubtitles):
245     url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
246     IE = ComedyCentralIE
247
248     def getInfoDict(self):
249         return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
250
251     def test_allsubtitles(self):
252         self.DL.params['writesubtitles'] = True
253         self.DL.params['allsubtitles'] = True
254         subtitles = self.getSubtitles()
255         self.assertEqual(set(subtitles.keys()), set(['en']))
256         self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
257
258
259 class TestNRKSubtitles(BaseTestSubtitles):
260     url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
261     IE = NRKTVIE
262
263     def test_allsubtitles(self):
264         self.DL.params['writesubtitles'] = True
265         self.DL.params['allsubtitles'] = True
266         subtitles = self.getSubtitles()
267         self.assertEqual(set(subtitles.keys()), set(['no']))
268         self.assertEqual(md5(subtitles['no']), '1d221e6458c95c5494dcd38e6a1f129a')
269
270
271 class TestRaiSubtitles(BaseTestSubtitles):
272     url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
273     IE = RaiIE
274
275     def test_allsubtitles(self):
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']), 'b1d90a98755126b61e667567a1f6680a')
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 TestRtveSubtitles(BaseTestSubtitles):
310     url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
311     IE = RTVEALaCartaIE
312
313     def test_allsubtitles(self):
314         print('Skipping, only available from Spain')
315         return
316         self.DL.params['writesubtitles'] = True
317         self.DL.params['allsubtitles'] = True
318         subtitles = self.getSubtitles()
319         self.assertEqual(set(subtitles.keys()), set(['es']))
320         self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
321
322
323 if __name__ == '__main__':
324     unittest.main()