[rai] Convert to new subtitles system
[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 )
27
28
29 class BaseTestSubtitles(unittest.TestCase):
30     url = None
31     IE = None
32
33     def setUp(self):
34         self.DL = FakeYDL()
35         self.ie = self.IE()
36         self.DL.add_info_extractor(self.ie)
37
38     def getInfoDict(self):
39         info_dict = self.DL.extract_info(self.url, download=False)
40         return info_dict
41
42     def getSubtitles(self):
43         info_dict = self.getInfoDict()
44         subtitles = info_dict['requested_subtitles']
45         if not subtitles:
46             return subtitles
47         for sub_info in subtitles.values():
48             if sub_info.get('data') is None:
49                 uf = self.DL.urlopen(sub_info['url'])
50                 sub_info['data'] = uf.read().decode('utf-8')
51         return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
52
53
54 class TestYoutubeSubtitles(BaseTestSubtitles):
55     url = 'QRS8MkLhQmM'
56     IE = YoutubeIE
57
58     def test_youtube_subtitles(self):
59         self.DL.params['writesubtitles'] = True
60         subtitles = self.getSubtitles()
61         self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
62
63     def test_youtube_subtitles_lang(self):
64         self.DL.params['writesubtitles'] = True
65         self.DL.params['subtitleslangs'] = ['it']
66         subtitles = self.getSubtitles()
67         self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
68
69     def test_youtube_allsubtitles(self):
70         self.DL.params['writesubtitles'] = True
71         self.DL.params['allsubtitles'] = True
72         subtitles = self.getSubtitles()
73         self.assertEqual(len(subtitles.keys()), 13)
74
75     def test_youtube_subtitles_sbv_format(self):
76         self.DL.params['writesubtitles'] = True
77         self.DL.params['subtitlesformat'] = 'sbv'
78         subtitles = self.getSubtitles()
79         self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
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_list_subtitles(self):
88         self.DL.expect_warning('Video doesn\'t have automatic captions')
89         self.DL.params['listsubtitles'] = True
90         info_dict = self.getInfoDict()
91         self.assertEqual(info_dict, None)
92
93     def test_youtube_automatic_captions(self):
94         self.url = '8YoUxe5ncPo'
95         self.DL.params['writeautomaticsub'] = True
96         self.DL.params['subtitleslangs'] = ['it']
97         subtitles = self.getSubtitles()
98         self.assertTrue(subtitles['it'] is not None)
99
100     def test_youtube_translated_subtitles(self):
101         # This video has a subtitles track, which can be translated
102         self.url = 'Ky9eprVWzlI'
103         self.DL.params['writeautomaticsub'] = True
104         self.DL.params['subtitleslangs'] = ['it']
105         subtitles = self.getSubtitles()
106         self.assertTrue(subtitles['it'] is not None)
107
108     def test_youtube_nosubtitles(self):
109         self.DL.expect_warning('video doesn\'t have subtitles')
110         self.url = 'n5BB19UTcdA'
111         self.DL.params['writesubtitles'] = True
112         self.DL.params['allsubtitles'] = True
113         subtitles = self.getSubtitles()
114         self.assertEqual(len(subtitles), 0)
115
116     def test_youtube_multiple_langs(self):
117         self.url = 'QRS8MkLhQmM'
118         self.DL.params['writesubtitles'] = True
119         langs = ['it', 'fr', 'de']
120         self.DL.params['subtitleslangs'] = langs
121         subtitles = self.getSubtitles()
122         for lang in langs:
123             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
124
125
126 class TestDailymotionSubtitles(BaseTestSubtitles):
127     url = 'http://www.dailymotion.com/video/xczg00'
128     IE = DailymotionIE
129
130     def test_subtitles(self):
131         self.DL.params['writesubtitles'] = True
132         subtitles = self.getSubtitles()
133         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
134
135     def test_subtitles_lang(self):
136         self.DL.params['writesubtitles'] = True
137         self.DL.params['subtitleslangs'] = ['fr']
138         subtitles = self.getSubtitles()
139         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
140
141     def test_allsubtitles(self):
142         self.DL.params['writesubtitles'] = True
143         self.DL.params['allsubtitles'] = True
144         subtitles = self.getSubtitles()
145         self.assertTrue(len(subtitles.keys()) >= 6)
146
147     def test_nosubtitles(self):
148         self.DL.expect_warning('video doesn\'t have subtitles')
149         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
150         self.DL.params['writesubtitles'] = True
151         self.DL.params['allsubtitles'] = True
152         subtitles = self.getSubtitles()
153         self.assertEqual(len(subtitles), 0)
154
155     def test_multiple_langs(self):
156         self.DL.params['writesubtitles'] = True
157         langs = ['es', 'fr', 'de']
158         self.DL.params['subtitleslangs'] = langs
159         subtitles = self.getSubtitles()
160         for lang in langs:
161             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
162
163
164 class TestTedSubtitles(BaseTestSubtitles):
165     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
166     IE = TEDIE
167
168     def test_no_writesubtitles(self):
169         subtitles = self.getSubtitles()
170         self.assertFalse(subtitles)
171
172     def test_subtitles(self):
173         self.DL.params['writesubtitles'] = True
174         subtitles = self.getSubtitles()
175         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
176
177     def test_subtitles_lang(self):
178         self.DL.params['writesubtitles'] = True
179         self.DL.params['subtitleslangs'] = ['fr']
180         subtitles = self.getSubtitles()
181         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
182
183     def test_allsubtitles(self):
184         self.DL.params['writesubtitles'] = True
185         self.DL.params['allsubtitles'] = True
186         subtitles = self.getSubtitles()
187         self.assertTrue(len(subtitles.keys()) >= 28)
188
189     def test_list_subtitles(self):
190         self.DL.params['listsubtitles'] = True
191         info_dict = self.getInfoDict()
192         self.assertEqual(info_dict, None)
193
194     def test_multiple_langs(self):
195         self.DL.params['writesubtitles'] = True
196         langs = ['es', 'fr', 'de']
197         self.DL.params['subtitleslangs'] = langs
198         subtitles = self.getSubtitles()
199         for lang in langs:
200             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
201
202
203 class TestBlipTVSubtitles(BaseTestSubtitles):
204     url = 'http://blip.tv/a/a-6603250'
205     IE = BlipTVIE
206
207     def test_list_subtitles(self):
208         self.DL.params['listsubtitles'] = True
209         info_dict = self.getInfoDict()
210         self.assertEqual(info_dict, None)
211
212     def test_allsubtitles(self):
213         self.DL.params['writesubtitles'] = True
214         self.DL.params['allsubtitles'] = True
215         subtitles = self.getSubtitles()
216         self.assertEqual(set(subtitles.keys()), set(['en']))
217         self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
218
219
220 class TestVimeoSubtitles(BaseTestSubtitles):
221     url = 'http://vimeo.com/76979871'
222     IE = VimeoIE
223
224     def test_subtitles(self):
225         self.DL.params['writesubtitles'] = True
226         subtitles = self.getSubtitles()
227         self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
228
229     def test_subtitles_lang(self):
230         self.DL.params['writesubtitles'] = True
231         self.DL.params['subtitleslangs'] = ['fr']
232         subtitles = self.getSubtitles()
233         self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
234
235     def test_allsubtitles(self):
236         self.DL.params['writesubtitles'] = True
237         self.DL.params['allsubtitles'] = True
238         subtitles = self.getSubtitles()
239         self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
240
241     def test_nosubtitles(self):
242         self.DL.expect_warning('video doesn\'t have subtitles')
243         self.url = 'http://vimeo.com/56015672'
244         self.DL.params['writesubtitles'] = True
245         self.DL.params['allsubtitles'] = True
246         subtitles = self.getSubtitles()
247         self.assertEqual(len(subtitles), 0)
248
249     def test_multiple_langs(self):
250         self.DL.params['writesubtitles'] = True
251         langs = ['es', 'fr', 'de']
252         self.DL.params['subtitleslangs'] = langs
253         subtitles = self.getSubtitles()
254         for lang in langs:
255             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
256
257
258 class TestWallaSubtitles(BaseTestSubtitles):
259     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
260     IE = WallaIE
261
262     def test_list_subtitles(self):
263         self.DL.expect_warning('Automatic Captions not supported by this server')
264         self.DL.params['listsubtitles'] = True
265         info_dict = self.getInfoDict()
266         self.assertEqual(info_dict, None)
267
268     def test_allsubtitles(self):
269         self.DL.expect_warning('Automatic Captions not supported by this server')
270         self.DL.params['writesubtitles'] = True
271         self.DL.params['allsubtitles'] = True
272         subtitles = self.getSubtitles()
273         self.assertEqual(set(subtitles.keys()), set(['heb']))
274         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
275
276     def test_nosubtitles(self):
277         self.DL.expect_warning('video doesn\'t have subtitles')
278         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
279         self.DL.params['writesubtitles'] = True
280         self.DL.params['allsubtitles'] = True
281         subtitles = self.getSubtitles()
282         self.assertEqual(len(subtitles), 0)
283
284
285 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
286     url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
287     IE = CeskaTelevizeIE
288
289     def test_list_subtitles(self):
290         self.DL.expect_warning('Automatic Captions not supported by this server')
291         self.DL.params['listsubtitles'] = True
292         info_dict = self.getInfoDict()
293         self.assertEqual(info_dict, None)
294
295     def test_allsubtitles(self):
296         self.DL.expect_warning('Automatic Captions not supported by this server')
297         self.DL.params['writesubtitles'] = True
298         self.DL.params['allsubtitles'] = True
299         subtitles = self.getSubtitles()
300         self.assertEqual(set(subtitles.keys()), set(['cs']))
301         self.assertTrue(len(subtitles['cs']) > 20000)
302
303     def test_nosubtitles(self):
304         self.DL.expect_warning('video doesn\'t have subtitles')
305         self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
306         self.DL.params['writesubtitles'] = True
307         self.DL.params['allsubtitles'] = True
308         subtitles = self.getSubtitles()
309         self.assertEqual(len(subtitles), 0)
310
311
312 class TestLyndaSubtitles(BaseTestSubtitles):
313     url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
314     IE = LyndaIE
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']), '09bbe67222259bed60deaa26997d73a7')
322
323
324 class TestNPOSubtitles(BaseTestSubtitles):
325     url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
326     IE = NPOIE
327
328     def test_allsubtitles(self):
329         self.DL.params['writesubtitles'] = True
330         self.DL.params['allsubtitles'] = True
331         subtitles = self.getSubtitles()
332         self.assertEqual(set(subtitles.keys()), set(['nl']))
333         self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
334
335
336 class TestMTVSubtitles(BaseTestSubtitles):
337     url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
338     IE = ComedyCentralIE
339
340     def getInfoDict(self):
341         return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
342
343     def test_allsubtitles(self):
344         self.DL.params['writesubtitles'] = True
345         self.DL.params['allsubtitles'] = True
346         subtitles = self.getSubtitles()
347         self.assertEqual(set(subtitles.keys()), set(['en']))
348         self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
349
350
351 class TestNRKSubtitles(BaseTestSubtitles):
352     url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
353     IE = NRKTVIE
354
355     def test_allsubtitles(self):
356         self.DL.params['writesubtitles'] = True
357         self.DL.params['allsubtitles'] = True
358         subtitles = self.getSubtitles()
359         self.assertEqual(set(subtitles.keys()), set(['no']))
360         self.assertEqual(md5(subtitles['no']), '1d221e6458c95c5494dcd38e6a1f129a')
361
362
363 class TestRaiSubtitles(BaseTestSubtitles):
364     url = 'http://www.rai.tv/dl/RaiTV/programmi/media/ContentItem-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
365     IE = RaiIE
366
367     def test_allsubtitles(self):
368         self.DL.params['writesubtitles'] = True
369         self.DL.params['allsubtitles'] = True
370         subtitles = self.getSubtitles()
371         self.assertEqual(set(subtitles.keys()), set(['it']))
372         self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
373
374
375 if __name__ == '__main__':
376     unittest.main()