[npo] 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 )
24
25
26 class BaseTestSubtitles(unittest.TestCase):
27     url = None
28     IE = None
29
30     def setUp(self):
31         self.DL = FakeYDL()
32         self.ie = self.IE()
33         self.DL.add_info_extractor(self.ie)
34
35     def getInfoDict(self):
36         info_dict = self.DL.extract_info(self.url, download=False)
37         return info_dict
38
39     def getSubtitles(self):
40         info_dict = self.getInfoDict()
41         subtitles = info_dict['requested_subtitles']
42         if not subtitles:
43             return subtitles
44         for sub_info in subtitles.values():
45             if sub_info.get('data') is None:
46                 uf = self.DL.urlopen(sub_info['url'])
47                 sub_info['data'] = uf.read().decode('utf-8')
48         return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
49
50
51 class TestYoutubeSubtitles(BaseTestSubtitles):
52     url = 'QRS8MkLhQmM'
53     IE = YoutubeIE
54
55     def test_youtube_subtitles(self):
56         self.DL.params['writesubtitles'] = True
57         subtitles = self.getSubtitles()
58         self.assertEqual(md5(subtitles['en']), '4cd9278a35ba2305f47354ee13472260')
59
60     def test_youtube_subtitles_lang(self):
61         self.DL.params['writesubtitles'] = True
62         self.DL.params['subtitleslangs'] = ['it']
63         subtitles = self.getSubtitles()
64         self.assertEqual(md5(subtitles['it']), '164a51f16f260476a05b50fe4c2f161d')
65
66     def test_youtube_allsubtitles(self):
67         self.DL.params['writesubtitles'] = True
68         self.DL.params['allsubtitles'] = True
69         subtitles = self.getSubtitles()
70         self.assertEqual(len(subtitles.keys()), 13)
71
72     def test_youtube_subtitles_sbv_format(self):
73         self.DL.params['writesubtitles'] = True
74         self.DL.params['subtitlesformat'] = 'sbv'
75         subtitles = self.getSubtitles()
76         self.assertEqual(md5(subtitles['en']), '13aeaa0c245a8bed9a451cb643e3ad8b')
77
78     def test_youtube_subtitles_vtt_format(self):
79         self.DL.params['writesubtitles'] = True
80         self.DL.params['subtitlesformat'] = 'vtt'
81         subtitles = self.getSubtitles()
82         self.assertEqual(md5(subtitles['en']), '3cb210999d3e021bd6c7f0ea751eab06')
83
84     def test_youtube_list_subtitles(self):
85         self.DL.expect_warning('Video doesn\'t have automatic captions')
86         self.DL.params['listsubtitles'] = True
87         info_dict = self.getInfoDict()
88         self.assertEqual(info_dict, None)
89
90     def test_youtube_automatic_captions(self):
91         self.url = '8YoUxe5ncPo'
92         self.DL.params['writeautomaticsub'] = True
93         self.DL.params['subtitleslangs'] = ['it']
94         subtitles = self.getSubtitles()
95         self.assertTrue(subtitles['it'] is not None)
96
97     def test_youtube_translated_subtitles(self):
98         # This video has a subtitles track, which can be translated
99         self.url = 'Ky9eprVWzlI'
100         self.DL.params['writeautomaticsub'] = True
101         self.DL.params['subtitleslangs'] = ['it']
102         subtitles = self.getSubtitles()
103         self.assertTrue(subtitles['it'] is not None)
104
105     def test_youtube_nosubtitles(self):
106         self.DL.expect_warning('video doesn\'t have subtitles')
107         self.url = 'n5BB19UTcdA'
108         self.DL.params['writesubtitles'] = True
109         self.DL.params['allsubtitles'] = True
110         subtitles = self.getSubtitles()
111         self.assertEqual(len(subtitles), 0)
112
113     def test_youtube_multiple_langs(self):
114         self.url = 'QRS8MkLhQmM'
115         self.DL.params['writesubtitles'] = True
116         langs = ['it', 'fr', 'de']
117         self.DL.params['subtitleslangs'] = langs
118         subtitles = self.getSubtitles()
119         for lang in langs:
120             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
121
122
123 class TestDailymotionSubtitles(BaseTestSubtitles):
124     url = 'http://www.dailymotion.com/video/xczg00'
125     IE = DailymotionIE
126
127     def test_subtitles(self):
128         self.DL.params['writesubtitles'] = True
129         subtitles = self.getSubtitles()
130         self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
131
132     def test_subtitles_lang(self):
133         self.DL.params['writesubtitles'] = True
134         self.DL.params['subtitleslangs'] = ['fr']
135         subtitles = self.getSubtitles()
136         self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
137
138     def test_allsubtitles(self):
139         self.DL.params['writesubtitles'] = True
140         self.DL.params['allsubtitles'] = True
141         subtitles = self.getSubtitles()
142         self.assertTrue(len(subtitles.keys()) >= 6)
143
144     def test_nosubtitles(self):
145         self.DL.expect_warning('video doesn\'t have subtitles')
146         self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
147         self.DL.params['writesubtitles'] = True
148         self.DL.params['allsubtitles'] = True
149         subtitles = self.getSubtitles()
150         self.assertEqual(len(subtitles), 0)
151
152     def test_multiple_langs(self):
153         self.DL.params['writesubtitles'] = True
154         langs = ['es', 'fr', 'de']
155         self.DL.params['subtitleslangs'] = langs
156         subtitles = self.getSubtitles()
157         for lang in langs:
158             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
159
160
161 class TestTedSubtitles(BaseTestSubtitles):
162     url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
163     IE = TEDIE
164
165     def test_no_writesubtitles(self):
166         subtitles = self.getSubtitles()
167         self.assertFalse(subtitles)
168
169     def test_subtitles(self):
170         self.DL.params['writesubtitles'] = True
171         subtitles = self.getSubtitles()
172         self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
173
174     def test_subtitles_lang(self):
175         self.DL.params['writesubtitles'] = True
176         self.DL.params['subtitleslangs'] = ['fr']
177         subtitles = self.getSubtitles()
178         self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
179
180     def test_allsubtitles(self):
181         self.DL.params['writesubtitles'] = True
182         self.DL.params['allsubtitles'] = True
183         subtitles = self.getSubtitles()
184         self.assertTrue(len(subtitles.keys()) >= 28)
185
186     def test_list_subtitles(self):
187         self.DL.params['listsubtitles'] = True
188         info_dict = self.getInfoDict()
189         self.assertEqual(info_dict, None)
190
191     def test_multiple_langs(self):
192         self.DL.params['writesubtitles'] = True
193         langs = ['es', 'fr', 'de']
194         self.DL.params['subtitleslangs'] = langs
195         subtitles = self.getSubtitles()
196         for lang in langs:
197             self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
198
199
200 class TestBlipTVSubtitles(BaseTestSubtitles):
201     url = 'http://blip.tv/a/a-6603250'
202     IE = BlipTVIE
203
204     def test_list_subtitles(self):
205         self.DL.params['listsubtitles'] = True
206         info_dict = self.getInfoDict()
207         self.assertEqual(info_dict, None)
208
209     def test_allsubtitles(self):
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 class TestLyndaSubtitles(BaseTestSubtitles):
310     url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
311     IE = LyndaIE
312
313     def test_allsubtitles(self):
314         self.DL.params['writesubtitles'] = True
315         self.DL.params['allsubtitles'] = True
316         subtitles = self.getSubtitles()
317         self.assertEqual(set(subtitles.keys()), set(['en']))
318         self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
319
320
321 class TestNPOSubtitles(BaseTestSubtitles):
322     url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
323     IE = NPOIE
324
325     def test_allsubtitles(self):
326         self.DL.params['writesubtitles'] = True
327         self.DL.params['allsubtitles'] = True
328         subtitles = self.getSubtitles()
329         self.assertEqual(set(subtitles.keys()), set(['nl']))
330         self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
331
332
333 if __name__ == '__main__':
334     unittest.main()