[youtube] 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 )
22
23
24 class BaseTestSubtitles(unittest.TestCase):
25     url = None
26     IE = None
27
28     def setUp(self):
29         self.DL = FakeYDL()
30         self.ie = self.IE()
31         self.DL.add_info_extractor(self.ie)
32
33     def getInfoDict(self):
34         info_dict = self.DL.extract_info(self.url, download=False)
35         return info_dict
36
37     def getSubtitles(self):
38         info_dict = self.getInfoDict()
39         subtitles = info_dict['requested_subtitles']
40         if not subtitles:
41             return subtitles
42         for sub_info in subtitles.values():
43             if sub_info.get('data') is None:
44                 uf = self.DL.urlopen(sub_info['url'])
45                 sub_info['data'] = uf.read().decode('utf-8')
46         return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
47
48
49 class TestYoutubeSubtitles(BaseTestSubtitles):
50     url = 'QRS8MkLhQmM'
51     IE = YoutubeIE
52
53     def test_youtube_subtitles(self):
54         self.DL.params['writesubtitles'] = True
55         subtitles = self.getSubtitles()
56         self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
57
58     def test_youtube_subtitles_lang(self):
59         self.DL.params['writesubtitles'] = True
60         self.DL.params['subtitleslangs'] = ['it']
61         subtitles = self.getSubtitles()
62         self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
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
70     def test_youtube_subtitles_sbv_format(self):
71         self.DL.params['writesubtitles'] = True
72         self.DL.params['subtitlesformat'] = 'sbv'
73         subtitles = self.getSubtitles()
74         self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
75
76     def test_youtube_subtitles_vtt_format(self):
77         self.DL.params['writesubtitles'] = True
78         self.DL.params['subtitlesformat'] = 'vtt'
79         subtitles = self.getSubtitles()
80         self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
81
82     def test_youtube_list_subtitles(self):
83         self.DL.expect_warning('Video doesn\'t have automatic captions')
84         self.DL.params['listsubtitles'] = True
85         info_dict = self.getInfoDict()
86         self.assertEqual(info_dict, None)
87
88     def test_youtube_automatic_captions(self):
89         self.url = '8YoUxe5ncPo'
90         self.DL.params['writeautomaticsub'] = True
91         self.DL.params['subtitleslangs'] = ['it']
92         subtitles = self.getSubtitles()
93         self.assertTrue(subtitles['it'] is not None)
94
95     def test_youtube_translated_subtitles(self):
96         # This video has a subtitles track, which can be translated
97         self.url = 'Ky9eprVWzlI'
98         self.DL.params['writeautomaticsub'] = True
99         self.DL.params['subtitleslangs'] = ['it']
100         subtitles = self.getSubtitles()
101         self.assertTrue(subtitles['it'] is not None)
102
103     def test_youtube_nosubtitles(self):
104         self.DL.expect_warning('video doesn\'t have subtitles')
105         self.url = 'n5BB19UTcdA'
106         self.DL.params['writesubtitles'] = True
107         self.DL.params['allsubtitles'] = True
108         subtitles = self.getSubtitles()
109         self.assertEqual(len(subtitles), 0)
110
111     def test_youtube_multiple_langs(self):
112         self.url = 'QRS8MkLhQmM'
113         self.DL.params['writesubtitles'] = True
114         langs = ['it', 'fr', 'de']
115         self.DL.params['subtitleslangs'] = langs
116         subtitles = self.getSubtitles()
117         for lang in langs:
118             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
119
120
121 class TestDailymotionSubtitles(BaseTestSubtitles):
122     url = 'http://www.dailymotion.com/video/xczg00'
123     IE = DailymotionIE
124
125     def test_subtitles(self):
126         self.DL.params['writesubtitles'] = True
127         subtitles = self.getSubtitles()
128         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
129
130     def test_subtitles_lang(self):
131         self.DL.params['writesubtitles'] = True
132         self.DL.params['subtitleslangs'] = ['fr']
133         subtitles = self.getSubtitles()
134         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
135
136     def test_allsubtitles(self):
137         self.DL.params['writesubtitles'] = True
138         self.DL.params['allsubtitles'] = True
139         subtitles = self.getSubtitles()
140         self.assertTrue(len(subtitles.keys()) >= 6)
141
142     def test_nosubtitles(self):
143         self.DL.expect_warning('video doesn\'t have subtitles')
144         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
145         self.DL.params['writesubtitles'] = True
146         self.DL.params['allsubtitles'] = True
147         subtitles = self.getSubtitles()
148         self.assertEqual(len(subtitles), 0)
149
150     def test_multiple_langs(self):
151         self.DL.params['writesubtitles'] = True
152         langs = ['es', 'fr', 'de']
153         self.DL.params['subtitleslangs'] = langs
154         subtitles = self.getSubtitles()
155         for lang in langs:
156             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
157
158
159 class TestTedSubtitles(BaseTestSubtitles):
160     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
161     IE = TEDIE
162
163     def test_no_writesubtitles(self):
164         subtitles = self.getSubtitles()
165         self.assertFalse(subtitles)
166
167     def test_subtitles(self):
168         self.DL.params['writesubtitles'] = True
169         subtitles = self.getSubtitles()
170         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
171
172     def test_subtitles_lang(self):
173         self.DL.params['writesubtitles'] = True
174         self.DL.params['subtitleslangs'] = ['fr']
175         subtitles = self.getSubtitles()
176         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
177
178     def test_allsubtitles(self):
179         self.DL.params['writesubtitles'] = True
180         self.DL.params['allsubtitles'] = True
181         subtitles = self.getSubtitles()
182         self.assertTrue(len(subtitles.keys()) >= 28)
183
184     def test_list_subtitles(self):
185         self.DL.params['listsubtitles'] = True
186         info_dict = self.getInfoDict()
187         self.assertEqual(info_dict, None)
188
189     def test_multiple_langs(self):
190         self.DL.params['writesubtitles'] = True
191         langs = ['es', 'fr', 'de']
192         self.DL.params['subtitleslangs'] = langs
193         subtitles = self.getSubtitles()
194         for lang in langs:
195             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
196
197
198 class TestBlipTVSubtitles(BaseTestSubtitles):
199     url = 'http://blip.tv/a/a-6603250'
200     IE = BlipTVIE
201
202     def test_list_subtitles(self):
203         self.DL.expect_warning('Automatic Captions not supported by this server')
204         self.DL.params['listsubtitles'] = True
205         info_dict = self.getInfoDict()
206         self.assertEqual(info_dict, None)
207
208     def test_allsubtitles(self):
209         self.DL.expect_warning('Automatic Captions not supported by this server')
210         self.DL.params['writesubtitles'] = True
211         self.DL.params['allsubtitles'] = True
212         subtitles = self.getSubtitles()
213         self.assertEqual(set(subtitles.keys()), set(['en']))
214         self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
215
216
217 class TestVimeoSubtitles(BaseTestSubtitles):
218     url = 'http://vimeo.com/76979871'
219     IE = VimeoIE
220
221     def test_subtitles(self):
222         self.DL.params['writesubtitles'] = True
223         subtitles = self.getSubtitles()
224         self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
225
226     def test_subtitles_lang(self):
227         self.DL.params['writesubtitles'] = True
228         self.DL.params['subtitleslangs'] = ['fr']
229         subtitles = self.getSubtitles()
230         self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
231
232     def test_allsubtitles(self):
233         self.DL.params['writesubtitles'] = True
234         self.DL.params['allsubtitles'] = True
235         subtitles = self.getSubtitles()
236         self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
237
238     def test_nosubtitles(self):
239         self.DL.expect_warning('video doesn\'t have subtitles')
240         self.url = 'http://vimeo.com/56015672'
241         self.DL.params['writesubtitles'] = True
242         self.DL.params['allsubtitles'] = True
243         subtitles = self.getSubtitles()
244         self.assertEqual(len(subtitles), 0)
245
246     def test_multiple_langs(self):
247         self.DL.params['writesubtitles'] = True
248         langs = ['es', 'fr', 'de']
249         self.DL.params['subtitleslangs'] = langs
250         subtitles = self.getSubtitles()
251         for lang in langs:
252             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
253
254
255 class TestWallaSubtitles(BaseTestSubtitles):
256     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
257     IE = WallaIE
258
259     def test_list_subtitles(self):
260         self.DL.expect_warning('Automatic Captions not supported by this server')
261         self.DL.params['listsubtitles'] = True
262         info_dict = self.getInfoDict()
263         self.assertEqual(info_dict, None)
264
265     def test_allsubtitles(self):
266         self.DL.expect_warning('Automatic Captions not supported by this server')
267         self.DL.params['writesubtitles'] = True
268         self.DL.params['allsubtitles'] = True
269         subtitles = self.getSubtitles()
270         self.assertEqual(set(subtitles.keys()), set(['heb']))
271         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
272
273     def test_nosubtitles(self):
274         self.DL.expect_warning('video doesn\'t have subtitles')
275         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
276         self.DL.params['writesubtitles'] = True
277         self.DL.params['allsubtitles'] = True
278         subtitles = self.getSubtitles()
279         self.assertEqual(len(subtitles), 0)
280
281
282 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
283     url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
284     IE = CeskaTelevizeIE
285
286     def test_list_subtitles(self):
287         self.DL.expect_warning('Automatic Captions not supported by this server')
288         self.DL.params['listsubtitles'] = True
289         info_dict = self.getInfoDict()
290         self.assertEqual(info_dict, None)
291
292     def test_allsubtitles(self):
293         self.DL.expect_warning('Automatic Captions not supported by this server')
294         self.DL.params['writesubtitles'] = True
295         self.DL.params['allsubtitles'] = True
296         subtitles = self.getSubtitles()
297         self.assertEqual(set(subtitles.keys()), set(['cs']))
298         self.assertTrue(len(subtitles['cs']) > 20000)
299
300     def test_nosubtitles(self):
301         self.DL.expect_warning('video doesn\'t have subtitles')
302         self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
303         self.DL.params['writesubtitles'] = True
304         self.DL.params['allsubtitles'] = True
305         subtitles = self.getSubtitles()
306         self.assertEqual(len(subtitles), 0)
307
308
309 if __name__ == '__main__':
310     unittest.main()