]> jfr.im git - yt-dlp.git/blob - test/test_subtitles.py
Merge branch 'alura' of https://github.com/hugohaa/youtube-dl into hugohaa-alura
[yt-dlp.git] / 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_dlc.extractor import (
14 YoutubeIE,
15 DailymotionIE,
16 TEDIE,
17 VimeoIE,
18 WallaIE,
19 CeskaTelevizeIE,
20 LyndaIE,
21 NPOIE,
22 ComedyCentralIE,
23 NRKTVIE,
24 RaiPlayIE,
25 VikiIE,
26 ThePlatformIE,
27 ThePlatformFeedIE,
28 RTVEALaCartaIE,
29 DemocracynowIE,
30 )
31
32
33 class BaseTestSubtitles(unittest.TestCase):
34 url = None
35 IE = None
36
37 def setUp(self):
38 self.DL = FakeYDL()
39 self.ie = self.IE()
40 self.DL.add_info_extractor(self.ie)
41
42 def getInfoDict(self):
43 info_dict = self.DL.extract_info(self.url, download=False)
44 return info_dict
45
46 def getSubtitles(self):
47 info_dict = self.getInfoDict()
48 subtitles = info_dict['requested_subtitles']
49 if not subtitles:
50 return subtitles
51 for sub_info in subtitles.values():
52 if sub_info.get('data') is None:
53 uf = self.DL.urlopen(sub_info['url'])
54 sub_info['data'] = uf.read().decode('utf-8')
55 return dict((l, sub_info['data']) for l, sub_info in subtitles.items())
56
57
58 class TestYoutubeSubtitles(BaseTestSubtitles):
59 url = 'QRS8MkLhQmM'
60 IE = YoutubeIE
61
62 def test_youtube_allsubtitles(self):
63 self.DL.params['writesubtitles'] = True
64 self.DL.params['allsubtitles'] = True
65 subtitles = self.getSubtitles()
66 self.assertEqual(len(subtitles.keys()), 13)
67 self.assertEqual(md5(subtitles['en']), '688dd1ce0981683867e7fe6fde2a224b')
68 self.assertEqual(md5(subtitles['it']), '31324d30b8430b309f7f5979a504a769')
69 for lang in ['fr', 'de']:
70 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
71
72 def test_youtube_subtitles_ttml_format(self):
73 self.DL.params['writesubtitles'] = True
74 self.DL.params['subtitlesformat'] = 'ttml'
75 subtitles = self.getSubtitles()
76 self.assertEqual(md5(subtitles['en']), 'c97ddf1217390906fa9fbd34901f3da2')
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']), 'ae1bd34126571a77aabd4d276b28044d')
83
84 def test_youtube_automatic_captions(self):
85 self.url = '8YoUxe5ncPo'
86 self.DL.params['writeautomaticsub'] = True
87 self.DL.params['subtitleslangs'] = ['it']
88 subtitles = self.getSubtitles()
89 self.assertTrue(subtitles['it'] is not None)
90
91 def test_youtube_no_automatic_captions(self):
92 self.url = 'QRS8MkLhQmM'
93 self.DL.params['writeautomaticsub'] = True
94 subtitles = self.getSubtitles()
95 self.assertTrue(not subtitles)
96
97 def test_youtube_translated_subtitles(self):
98 # This video has a subtitles track, which can be translated
99 self.url = 'i0ZabxXmH4Y'
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.assertFalse(subtitles)
112
113
114 class TestDailymotionSubtitles(BaseTestSubtitles):
115 url = 'http://www.dailymotion.com/video/xczg00'
116 IE = DailymotionIE
117
118 def test_allsubtitles(self):
119 self.DL.params['writesubtitles'] = True
120 self.DL.params['allsubtitles'] = True
121 subtitles = self.getSubtitles()
122 self.assertTrue(len(subtitles.keys()) >= 6)
123 self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
124 self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
125 for lang in ['es', 'fr', 'de']:
126 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
127
128 def test_nosubtitles(self):
129 self.DL.expect_warning('video doesn\'t have subtitles')
130 self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
131 self.DL.params['writesubtitles'] = True
132 self.DL.params['allsubtitles'] = True
133 subtitles = self.getSubtitles()
134 self.assertFalse(subtitles)
135
136
137 class TestTedSubtitles(BaseTestSubtitles):
138 url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
139 IE = TEDIE
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()) >= 28)
146 self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
147 self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
148 for lang in ['es', 'fr', 'de']:
149 self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
150
151
152 class TestVimeoSubtitles(BaseTestSubtitles):
153 url = 'http://vimeo.com/76979871'
154 IE = VimeoIE
155
156 def test_allsubtitles(self):
157 self.DL.params['writesubtitles'] = True
158 self.DL.params['allsubtitles'] = True
159 subtitles = self.getSubtitles()
160 self.assertEqual(set(subtitles.keys()), set(['de', 'en', 'es', 'fr']))
161 self.assertEqual(md5(subtitles['en']), '8062383cf4dec168fc40a088aa6d5888')
162 self.assertEqual(md5(subtitles['fr']), 'b6191146a6c5d3a452244d853fde6dc8')
163
164 def test_nosubtitles(self):
165 self.DL.expect_warning('video doesn\'t have subtitles')
166 self.url = 'http://vimeo.com/56015672'
167 self.DL.params['writesubtitles'] = True
168 self.DL.params['allsubtitles'] = True
169 subtitles = self.getSubtitles()
170 self.assertFalse(subtitles)
171
172
173 class TestWallaSubtitles(BaseTestSubtitles):
174 url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
175 IE = WallaIE
176
177 def test_allsubtitles(self):
178 self.DL.expect_warning('Automatic Captions not supported by this server')
179 self.DL.params['writesubtitles'] = True
180 self.DL.params['allsubtitles'] = True
181 subtitles = self.getSubtitles()
182 self.assertEqual(set(subtitles.keys()), set(['heb']))
183 self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
184
185 def test_nosubtitles(self):
186 self.DL.expect_warning('video doesn\'t have subtitles')
187 self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
188 self.DL.params['writesubtitles'] = True
189 self.DL.params['allsubtitles'] = True
190 subtitles = self.getSubtitles()
191 self.assertFalse(subtitles)
192
193
194 class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
195 url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
196 IE = CeskaTelevizeIE
197
198 def test_allsubtitles(self):
199 self.DL.expect_warning('Automatic Captions not supported by this server')
200 self.DL.params['writesubtitles'] = True
201 self.DL.params['allsubtitles'] = True
202 subtitles = self.getSubtitles()
203 self.assertEqual(set(subtitles.keys()), set(['cs']))
204 self.assertTrue(len(subtitles['cs']) > 20000)
205
206 def test_nosubtitles(self):
207 self.DL.expect_warning('video doesn\'t have subtitles')
208 self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
209 self.DL.params['writesubtitles'] = True
210 self.DL.params['allsubtitles'] = True
211 subtitles = self.getSubtitles()
212 self.assertFalse(subtitles)
213
214
215 class TestLyndaSubtitles(BaseTestSubtitles):
216 url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
217 IE = LyndaIE
218
219 def test_allsubtitles(self):
220 self.DL.params['writesubtitles'] = True
221 self.DL.params['allsubtitles'] = True
222 subtitles = self.getSubtitles()
223 self.assertEqual(set(subtitles.keys()), set(['en']))
224 self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
225
226
227 class TestNPOSubtitles(BaseTestSubtitles):
228 url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
229 IE = NPOIE
230
231 def test_allsubtitles(self):
232 self.DL.params['writesubtitles'] = True
233 self.DL.params['allsubtitles'] = True
234 subtitles = self.getSubtitles()
235 self.assertEqual(set(subtitles.keys()), set(['nl']))
236 self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
237
238
239 class TestMTVSubtitles(BaseTestSubtitles):
240 url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
241 IE = ComedyCentralIE
242
243 def getInfoDict(self):
244 return super(TestMTVSubtitles, self).getInfoDict()['entries'][0]
245
246 def test_allsubtitles(self):
247 self.DL.params['writesubtitles'] = True
248 self.DL.params['allsubtitles'] = True
249 subtitles = self.getSubtitles()
250 self.assertEqual(set(subtitles.keys()), set(['en']))
251 self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
252
253
254 class TestNRKSubtitles(BaseTestSubtitles):
255 url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
256 IE = NRKTVIE
257
258 def test_allsubtitles(self):
259 self.DL.params['writesubtitles'] = True
260 self.DL.params['allsubtitles'] = True
261 subtitles = self.getSubtitles()
262 self.assertEqual(set(subtitles.keys()), set(['no']))
263 self.assertEqual(md5(subtitles['no']), '544fa917d3197fcbee64634559221cc2')
264
265
266 class TestRaiPlaySubtitles(BaseTestSubtitles):
267 url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
268 IE = RaiPlayIE
269
270 def test_allsubtitles(self):
271 self.DL.params['writesubtitles'] = True
272 self.DL.params['allsubtitles'] = True
273 subtitles = self.getSubtitles()
274 self.assertEqual(set(subtitles.keys()), set(['it']))
275 self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
276
277
278 class TestVikiSubtitles(BaseTestSubtitles):
279 url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
280 IE = VikiIE
281
282 def test_allsubtitles(self):
283 self.DL.params['writesubtitles'] = True
284 self.DL.params['allsubtitles'] = True
285 subtitles = self.getSubtitles()
286 self.assertEqual(set(subtitles.keys()), set(['en']))
287 self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
288
289
290 class TestThePlatformSubtitles(BaseTestSubtitles):
291 # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
292 # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
293 url = 'theplatform:JFUjUE1_ehvq'
294 IE = ThePlatformIE
295
296 def test_allsubtitles(self):
297 self.DL.params['writesubtitles'] = True
298 self.DL.params['allsubtitles'] = True
299 subtitles = self.getSubtitles()
300 self.assertEqual(set(subtitles.keys()), set(['en']))
301 self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
302
303
304 class TestThePlatformFeedSubtitles(BaseTestSubtitles):
305 url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
306 IE = ThePlatformFeedIE
307
308 def test_allsubtitles(self):
309 self.DL.params['writesubtitles'] = True
310 self.DL.params['allsubtitles'] = True
311 subtitles = self.getSubtitles()
312 self.assertEqual(set(subtitles.keys()), set(['en']))
313 self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
314
315
316 class TestRtveSubtitles(BaseTestSubtitles):
317 url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
318 IE = RTVEALaCartaIE
319
320 def test_allsubtitles(self):
321 print('Skipping, only available from Spain')
322 return
323 self.DL.params['writesubtitles'] = True
324 self.DL.params['allsubtitles'] = True
325 subtitles = self.getSubtitles()
326 self.assertEqual(set(subtitles.keys()), set(['es']))
327 self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
328
329
330 class TestDemocracynowSubtitles(BaseTestSubtitles):
331 url = 'http://www.democracynow.org/shows/2015/7/3'
332 IE = DemocracynowIE
333
334 def test_allsubtitles(self):
335 self.DL.params['writesubtitles'] = True
336 self.DL.params['allsubtitles'] = True
337 subtitles = self.getSubtitles()
338 self.assertEqual(set(subtitles.keys()), set(['en']))
339 self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
340
341 def test_subtitles_in_page(self):
342 self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
343 self.DL.params['writesubtitles'] = True
344 self.DL.params['allsubtitles'] = True
345 subtitles = self.getSubtitles()
346 self.assertEqual(set(subtitles.keys()), set(['en']))
347 self.assertEqual(md5(subtitles['en']), 'acaca989e24a9e45a6719c9b3d60815c')
348
349
350 if __name__ == '__main__':
351 unittest.main()