[YoutubeDL] store the subtitles to download in the 'requested_subtitles' field
[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_no_writesubtitles(self):
54         self.DL.params['writesubtitles'] = False
55         subtitles = self.getSubtitles()
56         self.assertEqual(subtitles, None)
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_no_writesubtitles(self):
131         subtitles = self.getSubtitles()
132         self.assertEqual(subtitles, None)
133
134     def test_subtitles(self):
135         self.DL.params['writesubtitles'] = True
136         subtitles = self.getSubtitles()
137         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
138
139     def test_subtitles_lang(self):
140         self.DL.params['writesubtitles'] = True
141         self.DL.params['subtitleslangs'] = ['fr']
142         subtitles = self.getSubtitles()
143         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
144
145     def test_allsubtitles(self):
146         self.DL.params['writesubtitles'] = True
147         self.DL.params['allsubtitles'] = True
148         subtitles = self.getSubtitles()
149         self.assertTrue(len(subtitles.keys()) >= 6)
150
151     def test_list_subtitles(self):
152         self.DL.expect_warning('Automatic Captions not supported by this server')
153         self.DL.params['listsubtitles'] = True
154         info_dict = self.getInfoDict()
155         self.assertEqual(info_dict, None)
156
157     def test_automatic_captions(self):
158         self.DL.expect_warning('Automatic Captions not supported by this server')
159         self.DL.params['writeautomaticsub'] = True
160         self.DL.params['subtitleslang'] = ['en']
161         subtitles = self.getSubtitles()
162         self.assertTrue(len(subtitles.keys()) == 0)
163
164     def test_nosubtitles(self):
165         self.DL.expect_warning('video doesn\'t have subtitles')
166         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
167         self.DL.params['writesubtitles'] = True
168         self.DL.params['allsubtitles'] = True
169         subtitles = self.getSubtitles()
170         self.assertEqual(len(subtitles), 0)
171
172     def test_multiple_langs(self):
173         self.DL.params['writesubtitles'] = True
174         langs = ['es', 'fr', 'de']
175         self.DL.params['subtitleslangs'] = langs
176         subtitles = self.getSubtitles()
177         for lang in langs:
178             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
179
180
181 class TestTedSubtitles(BaseTestSubtitles):
182     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
183     IE = TEDIE
184
185     def test_no_writesubtitles(self):
186         subtitles = self.getSubtitles()
187         self.assertFalse(subtitles)
188
189     def test_subtitles(self):
190         self.DL.params['writesubtitles'] = True
191         subtitles = self.getSubtitles()
192         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
193
194     def test_subtitles_lang(self):
195         self.DL.params['writesubtitles'] = True
196         self.DL.params['subtitleslangs'] = ['fr']
197         subtitles = self.getSubtitles()
198         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
199
200     def test_allsubtitles(self):
201         self.DL.params['writesubtitles'] = True
202         self.DL.params['allsubtitles'] = True
203         subtitles = self.getSubtitles()
204         self.assertTrue(len(subtitles.keys()) >= 28)
205
206     def test_list_subtitles(self):
207         self.DL.params['listsubtitles'] = True
208         info_dict = self.getInfoDict()
209         self.assertEqual(info_dict, None)
210
211     def test_multiple_langs(self):
212         self.DL.params['writesubtitles'] = True
213         langs = ['es', 'fr', 'de']
214         self.DL.params['subtitleslangs'] = langs
215         subtitles = self.getSubtitles()
216         for lang in langs:
217             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
218
219
220 class TestBlipTVSubtitles(BaseTestSubtitles):
221     url = 'http://blip.tv/a/a-6603250'
222     IE = BlipTVIE
223
224     def test_list_subtitles(self):
225         self.DL.expect_warning('Automatic Captions not supported by this server')
226         self.DL.params['listsubtitles'] = True
227         info_dict = self.getInfoDict()
228         self.assertEqual(info_dict, None)
229
230     def test_allsubtitles(self):
231         self.DL.expect_warning('Automatic Captions not supported by this server')
232         self.DL.params['writesubtitles'] = True
233         self.DL.params['allsubtitles'] = True
234         subtitles = self.getSubtitles()
235         self.assertEqual(set(subtitles.keys()), set(['en']))
236         self.assertEqual(md5(subtitles['en']), '5b75c300af65fe4476dff79478bb93e4')
237
238
239 class TestVimeoSubtitles(BaseTestSubtitles):
240     url = 'http://vimeo.com/76979871'
241     IE = VimeoIE
242
243     def test_subtitles(self):
244         self.DL.params['writesubtitles'] = True
245         subtitles = self.getSubtitles()
246         self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
247
248     def test_subtitles_lang(self):
249         self.DL.params['writesubtitles'] = True
250         self.DL.params['subtitleslangs'] = ['fr']
251         subtitles = self.getSubtitles()
252         self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
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(['de', 'en', 'es', 'fr']))
259
260     def test_nosubtitles(self):
261         self.DL.expect_warning('video doesn\'t have subtitles')
262         self.url = 'http://vimeo.com/56015672'
263         self.DL.params['writesubtitles'] = True
264         self.DL.params['allsubtitles'] = True
265         subtitles = self.getSubtitles()
266         self.assertEqual(len(subtitles), 0)
267
268     def test_multiple_langs(self):
269         self.DL.params['writesubtitles'] = True
270         langs = ['es', 'fr', 'de']
271         self.DL.params['subtitleslangs'] = langs
272         subtitles = self.getSubtitles()
273         for lang in langs:
274             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
275
276
277 class TestWallaSubtitles(BaseTestSubtitles):
278     url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
279     IE = WallaIE
280
281     def test_list_subtitles(self):
282         self.DL.expect_warning('Automatic Captions not supported by this server')
283         self.DL.params['listsubtitles'] = True
284         info_dict = self.getInfoDict()
285         self.assertEqual(info_dict, None)
286
287     def test_allsubtitles(self):
288         self.DL.expect_warning('Automatic Captions not supported by this server')
289         self.DL.params['writesubtitles'] = True
290         self.DL.params['allsubtitles'] = True
291         subtitles = self.getSubtitles()
292         self.assertEqual(set(subtitles.keys()), set(['heb']))
293         self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
294
295     def test_nosubtitles(self):
296         self.DL.expect_warning('video doesn\'t have subtitles')
297         self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
298         self.DL.params['writesubtitles'] = True
299         self.DL.params['allsubtitles'] = True
300         subtitles = self.getSubtitles()
301         self.assertEqual(len(subtitles), 0)
302
303
304 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
305     url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
306     IE = CeskaTelevizeIE
307
308     def test_list_subtitles(self):
309         self.DL.expect_warning('Automatic Captions not supported by this server')
310         self.DL.params['listsubtitles'] = True
311         info_dict = self.getInfoDict()
312         self.assertEqual(info_dict, None)
313
314     def test_allsubtitles(self):
315         self.DL.expect_warning('Automatic Captions not supported by this server')
316         self.DL.params['writesubtitles'] = True
317         self.DL.params['allsubtitles'] = True
318         subtitles = self.getSubtitles()
319         self.assertEqual(set(subtitles.keys()), set(['cs']))
320         self.assertTrue(len(subtitles['cs']) > 20000)
321
322     def test_nosubtitles(self):
323         self.DL.expect_warning('video doesn\'t have subtitles')
324         self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
325         self.DL.params['writesubtitles'] = True
326         self.DL.params['allsubtitles'] = True
327         subtitles = self.getSubtitles()
328         self.assertEqual(len(subtitles), 0)
329
330
331 if __name__ == '__main__':
332     unittest.main()