[mtv] 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 )
25
26
27 class BaseTestSubtitles(unittest.TestCase):
28     url = None
29     IE = None
30
31     def setUp(self):
32         self.DL = FakeYDL()
33         self.ie = self.IE()
34         self.DL.add_info_extractor(self.ie)
35
36     def getInfoDict(self):
37         info_dict = self.DL.extract_info(self.url, download=False)
38         return info_dict
39
40     def getSubtitles(self):
41         info_dict = self.getInfoDict()
42         subtitles = info_dict['requested_subtitles']
43         if not subtitles:
44             return subtitles
45         for sub_info in subtitles.values():
46             if sub_info.get('data') is None:
47                 uf = self.DL.urlopen(sub_info['url'])
48                 sub_info['data'] = uf.read().decode('utf-8')
49         return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
50
51
52 class TestYoutubeSubtitles(BaseTestSubtitles):
53     url = 'QRS8MkLhQmM'
54     IE = YoutubeIE
55
56     def test_youtube_subtitles(self):
57         self.DL.params['writesubtitles'] = True
58         subtitles = self.getSubtitles()
59         self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
60
61     def test_youtube_subtitles_lang(self):
62         self.DL.params['writesubtitles'] = True
63         self.DL.params['subtitleslangs'] = ['it']
64         subtitles = self.getSubtitles()
65         self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
66
67     def test_youtube_allsubtitles(self):
68         self.DL.params['writesubtitles'] = True
69         self.DL.params['allsubtitles'] = True
70         subtitles = self.getSubtitles()
71         self.assertEqual(len(subtitles.keys()), 13)
72
73     def test_youtube_subtitles_sbv_format(self):
74         self.DL.params['writesubtitles'] = True
75         self.DL.params['subtitlesformat'] = 'sbv'
76         subtitles = self.getSubtitles()
77         self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
78
79     def test_youtube_subtitles_vtt_format(self):
80         self.DL.params['writesubtitles'] = True
81         self.DL.params['subtitlesformat'] = 'vtt'
82         subtitles = self.getSubtitles()
83         self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
84
85     def test_youtube_list_subtitles(self):
86         self.DL.expect_warning('Video doesn\'t have automatic captions')
87         self.DL.params['listsubtitles'] = True
88         info_dict = self.getInfoDict()
89         self.assertEqual(info_dict, None)
90
91     def test_youtube_automatic_captions(self):
92         self.url = '8YoUxe5ncPo'
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_translated_subtitles(self):
99         # This video has a subtitles track, which can be translated
100         self.url = 'Ky9eprVWzlI'
101         self.DL.params['writeautomaticsub'] = True
102         self.DL.params['subtitleslangs'] = ['it']
103         subtitles = self.getSubtitles()
104         self.assertTrue(subtitles['it'] is not None)
105
106     def test_youtube_nosubtitles(self):
107         self.DL.expect_warning('video doesn\'t have subtitles')
108         self.url = 'n5BB19UTcdA'
109         self.DL.params['writesubtitles'] = True
110         self.DL.params['allsubtitles'] = True
111         subtitles = self.getSubtitles()
112         self.assertEqual(len(subtitles), 0)
113
114     def test_youtube_multiple_langs(self):
115         self.url = 'QRS8MkLhQmM'
116         self.DL.params['writesubtitles'] = True
117         langs = ['it', 'fr', 'de']
118         self.DL.params['subtitleslangs'] = langs
119         subtitles = self.getSubtitles()
120         for lang in langs:
121             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
122
123
124 class TestDailymotionSubtitles(BaseTestSubtitles):
125     url = 'http://www.dailymotion.com/video/xczg00'
126     IE = DailymotionIE
127
128     def test_subtitles(self):
129         self.DL.params['writesubtitles'] = True
130         subtitles = self.getSubtitles()
131         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
132
133     def test_subtitles_lang(self):
134         self.DL.params['writesubtitles'] = True
135         self.DL.params['subtitleslangs'] = ['fr']
136         subtitles = self.getSubtitles()
137         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
138
139     def test_allsubtitles(self):
140         self.DL.params['writesubtitles'] = True
141         self.DL.params['allsubtitles'] = True
142         subtitles = self.getSubtitles()
143         self.assertTrue(len(subtitles.keys()) >= 6)
144
145     def test_nosubtitles(self):
146         self.DL.expect_warning('video doesn\'t have subtitles')
147         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
148         self.DL.params['writesubtitles'] = True
149         self.DL.params['allsubtitles'] = True
150         subtitles = self.getSubtitles()
151         self.assertEqual(len(subtitles), 0)
152
153     def test_multiple_langs(self):
154         self.DL.params['writesubtitles'] = True
155         langs = ['es', 'fr', 'de']
156         self.DL.params['subtitleslangs'] = langs
157         subtitles = self.getSubtitles()
158         for lang in langs:
159             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
160
161
162 class TestTedSubtitles(BaseTestSubtitles):
163     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
164     IE = TEDIE
165
166     def test_no_writesubtitles(self):
167         subtitles = self.getSubtitles()
168         self.assertFalse(subtitles)
169
170     def test_subtitles(self):
171         self.DL.params['writesubtitles'] = True
172         subtitles = self.getSubtitles()
173         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
174
175     def test_subtitles_lang(self):
176         self.DL.params['writesubtitles'] = True
177         self.DL.params['subtitleslangs'] = ['fr']
178         subtitles = self.getSubtitles()
179         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
180
181     def test_allsubtitles(self):
182         self.DL.params['writesubtitles'] = True
183         self.DL.params['allsubtitles'] = True
184         subtitles = self.getSubtitles()
185         self.assertTrue(len(subtitles.keys()) >= 28)
186
187     def test_list_subtitles(self):
188         self.DL.params['listsubtitles'] = True
189         info_dict = self.getInfoDict()
190         self.assertEqual(info_dict, None)
191
192     def test_multiple_langs(self):
193         self.DL.params['writesubtitles'] = True
194         langs = ['es', 'fr', 'de']
195         self.DL.params['subtitleslangs'] = langs
196         subtitles = self.getSubtitles()
197         for lang in langs:
198             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
199
200
201 class TestBlipTVSubtitles(BaseTestSubtitles):
202     url = 'http://blip.tv/a/a-6603250'
203     IE = BlipTVIE
204
205     def test_list_subtitles(self):
206         self.DL.params['listsubtitles'] = True
207         info_dict = self.getInfoDict()
208         self.assertEqual(info_dict, None)
209
210     def test_allsubtitles(self):
211         self.DL.params['writesubtitles'] = True
212         self.DL.params['allsubtitles'] = True
213         subtitles = self.getSubtitles()
214         self.assertEqual(set(subtitles.keys()), set(['en']))
215         self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
216
217
218 class TestVimeoSubtitles(BaseTestSubtitles):
219     url = 'http://vimeo.com/76979871'
220     IE = VimeoIE
221
222     def test_subtitles(self):
223         self.DL.params['writesubtitles'] = True
224         subtitles = self.getSubtitles()
225         self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
226
227     def test_subtitles_lang(self):
228         self.DL.params['writesubtitles'] = True
229         self.DL.params['subtitleslangs'] = ['fr']
230         subtitles = self.getSubtitles()
231         self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
232
233     def test_allsubtitles(self):
234         self.DL.params['writesubtitles'] = True
235         self.DL.params['allsubtitles'] = True
236         subtitles = self.getSubtitles()
237         self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
238
239     def test_nosubtitles(self):
240         self.DL.expect_warning('video doesn\'t have subtitles')
241         self.url = 'http://vimeo.com/56015672'
242         self.DL.params['writesubtitles'] = True
243         self.DL.params['allsubtitles'] = True
244         subtitles = self.getSubtitles()
245         self.assertEqual(len(subtitles), 0)
246
247     def test_multiple_langs(self):
248         self.DL.params['writesubtitles'] = True
249         langs = ['es', 'fr', 'de']
250         self.DL.params['subtitleslangs'] = langs
251         subtitles = self.getSubtitles()
252         for lang in langs:
253             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
254
255
256 class TestWallaSubtitles(BaseTestSubtitles):
257     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
258     IE = WallaIE
259
260     def test_list_subtitles(self):
261         self.DL.expect_warning('Automatic Captions not supported by this server')
262         self.DL.params['listsubtitles'] = True
263         info_dict = self.getInfoDict()
264         self.assertEqual(info_dict, None)
265
266     def test_allsubtitles(self):
267         self.DL.expect_warning('Automatic Captions not supported by this server')
268         self.DL.params['writesubtitles'] = True
269         self.DL.params['allsubtitles'] = True
270         subtitles = self.getSubtitles()
271         self.assertEqual(set(subtitles.keys()), set(['heb']))
272         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
273
274     def test_nosubtitles(self):
275         self.DL.expect_warning('video doesn\'t have subtitles')
276         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
277         self.DL.params['writesubtitles'] = True
278         self.DL.params['allsubtitles'] = True
279         subtitles = self.getSubtitles()
280         self.assertEqual(len(subtitles), 0)
281
282
283 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
284     url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
285     IE = CeskaTelevizeIE
286
287     def test_list_subtitles(self):
288         self.DL.expect_warning('Automatic Captions not supported by this server')
289         self.DL.params['listsubtitles'] = True
290         info_dict = self.getInfoDict()
291         self.assertEqual(info_dict, None)
292
293     def test_allsubtitles(self):
294         self.DL.expect_warning('Automatic Captions not supported by this server')
295         self.DL.params['writesubtitles'] = True
296         self.DL.params['allsubtitles'] = True
297         subtitles = self.getSubtitles()
298         self.assertEqual(set(subtitles.keys()), set(['cs']))
299         self.assertTrue(len(subtitles['cs']) > 20000)
300
301     def test_nosubtitles(self):
302         self.DL.expect_warning('video doesn\'t have subtitles')
303         self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
304         self.DL.params['writesubtitles'] = True
305         self.DL.params['allsubtitles'] = True
306         subtitles = self.getSubtitles()
307         self.assertEqual(len(subtitles), 0)
308
309
310 class TestLyndaSubtitles(BaseTestSubtitles):
311     url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
312     IE = LyndaIE
313
314     def test_allsubtitles(self):
315         self.DL.params['writesubtitles'] = True
316         self.DL.params['allsubtitles'] = True
317         subtitles = self.getSubtitles()
318         self.assertEqual(set(subtitles.keys()), set(['en']))
319         self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
320
321
322 class TestNPOSubtitles(BaseTestSubtitles):
323     url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
324     IE = NPOIE
325
326     def test_allsubtitles(self):
327         self.DL.params['writesubtitles'] = True
328         self.DL.params['allsubtitles'] = True
329         subtitles = self.getSubtitles()
330         self.assertEqual(set(subtitles.keys()), set(['nl']))
331         self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
332
333
334 class TestMTVSubtitles(BaseTestSubtitles):
335     url = 'http://www.cc.com/video-clips/kllhuv/stand-up-greg-fitzsimmons--uncensored---too-good-of-a-mother'
336     IE = ComedyCentralIE
337
338     def getInfoDict(self):
339         return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
340
341     def test_allsubtitles(self):
342         self.DL.params['writesubtitles'] = True
343         self.DL.params['allsubtitles'] = True
344         subtitles = self.getSubtitles()
345         self.assertEqual(set(subtitles.keys()), set(['en']))
346         self.assertEqual(md5(subtitles['en']), 'b9f6ca22a6acf597ec76f61749765e65')
347
348
349 if __name__ == '__main__':
350     unittest.main()