[yt-dlp.git] / test / test_subtitles.py

#!/usr/bin/env python3

# Allow direct execution
import os
import sys
import unittest

sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))


from test.helper import FakeYDL, is_download_test, md5
from yt_dlp.extractor import (
    NPOIE,
    NRKTVIE,
    PBSIE,
    CeskaTelevizeIE,
    ComedyCentralIE,
    DailymotionIE,
    DemocracynowIE,
    LyndaIE,
    RaiPlayIE,
    RTVEALaCartaIE,
    TedTalkIE,
    ThePlatformFeedIE,
    ThePlatformIE,
    VikiIE,
    VimeoIE,
    WallaIE,
    YoutubeIE,
)


@is_download_test
class BaseTestSubtitles(unittest.TestCase):
    url = None
    IE = None

    def setUp(self):
        self.DL = FakeYDL()
        self.ie = self.IE()
        self.DL.add_info_extractor(self.ie)
        if not self.IE.working():
            print('Skipping: %s marked as not _WORKING' % self.IE.ie_key())
            self.skipTest('IE marked as not _WORKING')

    def getInfoDict(self):
        info_dict = self.DL.extract_info(self.url, download=False)
        return info_dict

    def getSubtitles(self):
        info_dict = self.getInfoDict()
        subtitles = info_dict['requested_subtitles']
        if not subtitles:
            return subtitles
        for sub_info in subtitles.values():
            if sub_info.get('data') is None:
                uf = self.DL.urlopen(sub_info['url'])
                sub_info['data'] = uf.read().decode()
        return {l: sub_info['data'] for l, sub_info in subtitles.items()}


@is_download_test
class TestYoutubeSubtitles(BaseTestSubtitles):
    # Available subtitles for QRS8MkLhQmM:
    # Language formats
    # ru       vtt, ttml, srv3, srv2, srv1, json3
    # fr       vtt, ttml, srv3, srv2, srv1, json3
    # en       vtt, ttml, srv3, srv2, srv1, json3
    # nl       vtt, ttml, srv3, srv2, srv1, json3
    # de       vtt, ttml, srv3, srv2, srv1, json3
    # ko       vtt, ttml, srv3, srv2, srv1, json3
    # it       vtt, ttml, srv3, srv2, srv1, json3
    # zh-Hant  vtt, ttml, srv3, srv2, srv1, json3
    # hi       vtt, ttml, srv3, srv2, srv1, json3
    # pt-BR    vtt, ttml, srv3, srv2, srv1, json3
    # es-MX    vtt, ttml, srv3, srv2, srv1, json3
    # ja       vtt, ttml, srv3, srv2, srv1, json3
    # pl       vtt, ttml, srv3, srv2, srv1, json3
    url = 'QRS8MkLhQmM'
    IE = YoutubeIE

    def test_youtube_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(len(subtitles.keys()), 13)
        self.assertEqual(md5(subtitles['en']), 'ae1bd34126571a77aabd4d276b28044d')
        self.assertEqual(md5(subtitles['it']), '0e0b667ba68411d88fd1c5f4f4eab2f9')
        for lang in ['fr', 'de']:
            self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)

    def _test_subtitles_format(self, fmt, md5_hash, lang='en'):
        self.DL.params['writesubtitles'] = True
        self.DL.params['subtitlesformat'] = fmt
        subtitles = self.getSubtitles()
        self.assertEqual(md5(subtitles[lang]), md5_hash)

    def test_youtube_subtitles_ttml_format(self):
        self._test_subtitles_format('ttml', 'c97ddf1217390906fa9fbd34901f3da2')

    def test_youtube_subtitles_vtt_format(self):
        self._test_subtitles_format('vtt', 'ae1bd34126571a77aabd4d276b28044d')

    def test_youtube_subtitles_json3_format(self):
        self._test_subtitles_format('json3', '688dd1ce0981683867e7fe6fde2a224b')

    def _test_automatic_captions(self, url, lang):
        self.url = url
        self.DL.params['writeautomaticsub'] = True
        self.DL.params['subtitleslangs'] = [lang]
        subtitles = self.getSubtitles()
        self.assertTrue(subtitles[lang] is not None)

    def test_youtube_automatic_captions(self):
        # Available automatic captions for 8YoUxe5ncPo:
        # Language formats (all in vtt, ttml, srv3, srv2, srv1, json3)
        # gu, zh-Hans, zh-Hant, gd, ga, gl, lb, la, lo, tt, tr,
        # lv, lt, tk, th, tg, te, fil, haw, yi, ceb, yo, de, da,
        # el, eo, en, eu, et, es, ru, rw, ro, bn, be, bg, uk, jv,
        # bs, ja, or, xh, co, ca, cy, cs, ps, pt, pa, vi, pl, hy,
        # hr, ht, hu, hmn, hi, ha, mg, uz, ml, mn, mi, mk, ur,
        # mt, ms, mr, ug, ta, my, af, sw, is, am,
        #                                         *it*, iw, sv, ar,
        # su, zu, az, id, ig, nl, no, ne, ny, fr, ku, fy, fa, fi,
        # ka, kk, sr, sq, ko, kn, km, st, sk, si, so, sn, sm, sl,
        # ky, sd
        # ...
        self._test_automatic_captions('8YoUxe5ncPo', 'it')

    @unittest.skip('Video unavailable')
    def test_youtube_translated_subtitles(self):
        # This video has a subtitles track, which can be translated (#4555)
        self._test_automatic_captions('Ky9eprVWzlI', 'it')

    def test_youtube_nosubtitles(self):
        self.DL.expect_warning('video doesn\'t have subtitles')
        # Available automatic captions for 8YoUxe5ncPo:
        # ...
        # 8YoUxe5ncPo has no subtitles
        self.url = '8YoUxe5ncPo'
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertFalse(subtitles)


@is_download_test
class TestDailymotionSubtitles(BaseTestSubtitles):
    url = 'http://www.dailymotion.com/video/xczg00'
    IE = DailymotionIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertTrue(len(subtitles.keys()) >= 6)
        self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
        self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
        for lang in ['es', 'fr', 'de']:
            self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)

    def test_nosubtitles(self):
        self.DL.expect_warning('video doesn\'t have subtitles')
        self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertFalse(subtitles)


@is_download_test
@unittest.skip('IE broken')
class TestTedSubtitles(BaseTestSubtitles):
    url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
    IE = TedTalkIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertTrue(len(subtitles.keys()) >= 28)
        self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
        self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
        for lang in ['es', 'fr', 'de']:
            self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)


@is_download_test
class TestVimeoSubtitles(BaseTestSubtitles):
    url = 'http://vimeo.com/76979871'
    IE = VimeoIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'de', 'en', 'es', 'fr'})
        self.assertEqual(md5(subtitles['en']), '386cbc9320b94e25cb364b97935e5dd1')
        self.assertEqual(md5(subtitles['fr']), 'c9b69eef35bc6641c0d4da8a04f9dfac')

    def test_nosubtitles(self):
        self.DL.expect_warning('video doesn\'t have subtitles')
        self.url = 'http://vimeo.com/68093876'
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertFalse(subtitles)


@is_download_test
@unittest.skip('IE broken')
class TestWallaSubtitles(BaseTestSubtitles):
    url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
    IE = WallaIE

    def test_allsubtitles(self):
        self.DL.expect_warning('Automatic Captions not supported by this server')
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'heb'})
        self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')

    def test_nosubtitles(self):
        self.DL.expect_warning('video doesn\'t have subtitles')
        self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertFalse(subtitles)


@is_download_test
@unittest.skip('IE broken')
class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
    url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
    IE = CeskaTelevizeIE

    def test_allsubtitles(self):
        self.DL.expect_warning('Automatic Captions not supported by this server')
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'cs'})
        self.assertTrue(len(subtitles['cs']) > 20000)

    def test_nosubtitles(self):
        self.DL.expect_warning('video doesn\'t have subtitles')
        self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertFalse(subtitles)


@is_download_test
@unittest.skip('IE broken')
class TestLyndaSubtitles(BaseTestSubtitles):
    url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
    IE = LyndaIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'en'})
        self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')


@is_download_test
@unittest.skip('IE broken')
class TestNPOSubtitles(BaseTestSubtitles):
    url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
    IE = NPOIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'nl'})
        self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')


@is_download_test
@unittest.skip('IE broken')
class TestMTVSubtitles(BaseTestSubtitles):
    url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
    IE = ComedyCentralIE

    def getInfoDict(self):
        return super().getInfoDict()['entries'][0]

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'en'})
        self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')


@is_download_test
class TestNRKSubtitles(BaseTestSubtitles):
    url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
    IE = NRKTVIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'nb-ttv'})
        self.assertEqual(md5(subtitles['nb-ttv']), '67e06ff02d0deaf975e68f6cb8f6a149')


@is_download_test
class TestRaiPlaySubtitles(BaseTestSubtitles):
    IE = RaiPlayIE

    def test_subtitles_key(self):
        self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'it'})
        self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')

    def test_subtitles_array_key(self):
        self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'it'})
        self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')


@is_download_test
@unittest.skip('IE broken - DRM only')
class TestVikiSubtitles(BaseTestSubtitles):
    url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
    IE = VikiIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'en'})
        self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')


@is_download_test
class TestThePlatformSubtitles(BaseTestSubtitles):
    # from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
    # (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
    url = 'theplatform:JFUjUE1_ehvq'
    IE = ThePlatformIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'en'})
        self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')


@is_download_test
@unittest.skip('IE broken')
class TestThePlatformFeedSubtitles(BaseTestSubtitles):
    url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
    IE = ThePlatformFeedIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'en'})
        self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')


@is_download_test
class TestRtveSubtitles(BaseTestSubtitles):
    url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
    IE = RTVEALaCartaIE

    def test_allsubtitles(self):
        print('Skipping, only available from Spain')
        return
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'es'})
        self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')


@is_download_test
class TestDemocracynowSubtitles(BaseTestSubtitles):
    url = 'http://www.democracynow.org/shows/2015/7/3'
    IE = DemocracynowIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'en'})
        self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')

    def test_subtitles_in_page(self):
        self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'en'})
        self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')


@is_download_test
class TestPBSSubtitles(BaseTestSubtitles):
    url = 'https://www.pbs.org/video/how-fantasy-reflects-our-world-picecq/'
    IE = PBSIE

    def test_allsubtitles(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['allsubtitles'] = True
        subtitles = self.getSubtitles()
        self.assertEqual(set(subtitles.keys()), {'en'})

    def test_subtitles_dfxp_format(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['subtitlesformat'] = 'dfxp'
        subtitles = self.getSubtitles()
        self.assertIn(md5(subtitles['en']), ['643b034254cdc3768ff1e750b6b5873b'])

    def test_subtitles_vtt_format(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['subtitlesformat'] = 'vtt'
        subtitles = self.getSubtitles()
        self.assertIn(
            md5(subtitles['en']), ['937a05711555b165d4c55a9667017045', 'f49ea998d6824d94959c8152a368ff73'])

    def test_subtitles_srt_format(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['subtitlesformat'] = 'srt'
        subtitles = self.getSubtitles()
        self.assertIn(md5(subtitles['en']), ['2082c21b43759d9bf172931b2f2ca371'])

    def test_subtitles_sami_format(self):
        self.DL.params['writesubtitles'] = True
        self.DL.params['subtitlesformat'] = 'sami'
        subtitles = self.getSubtitles()
        self.assertIn(md5(subtitles['en']), ['4256b16ac7da6a6780fafd04294e85cd'])


if __name__ == '__main__':
    unittest.main()
Commit	Line	Data
cc52de43	1	#!/usr/bin/env python3
54007a45	2
ae84f879 JMF	3	# Allow direct execution
	4	import os
	5	import sys
	6	import unittest
ae84f879	7
f8271158	8	sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
ae84f879 JMF	9
ae84f879 JMF	10
ac668111	11	from test.helper import FakeYDL, is_download_test, md5
7a5c1cfe	12	from yt_dlp.extractor import (
b9b42f2e	13	NPOIE,
f8271158	14	NRKTVIE,
a7e999be	15	PBSIE,
f8271158	16	CeskaTelevizeIE,
0af25f78	17	ComedyCentralIE,
f8271158	18	DailymotionIE,
	19	DemocracynowIE,
	20	LyndaIE,
c4d6fc6d	21	RaiPlayIE,
25ac63ed	22	RTVEALaCartaIE,
f8271158	23	TedTalkIE,
	24	ThePlatformFeedIE,
	25	ThePlatformIE,
	26	VikiIE,
	27	VimeoIE,
	28	WallaIE,
	29	YoutubeIE,
ae84f879 JMF	30	)
	31
	32
060ac762	33	@is_download_test
ae84f879 JMF	34	class BaseTestSubtitles(unittest.TestCase):
	35	url = None
	36	IE = None
5f6a1245	37
ae84f879 JMF	38	def setUp(self):
ae84f879 JMF	39	self.DL = FakeYDL()
a504ced0 JMF	40	self.ie = self.IE()
a504ced0 JMF	41	self.DL.add_info_extractor(self.ie)
6d1b3489	42	if not self.IE.working():
	43	print('Skipping: %s marked as not _WORKING' % self.IE.ie_key())
	44	self.skipTest('IE marked as not _WORKING')
ae84f879 JMF	45
ae84f879 JMF	46	def getInfoDict(self):
a504ced0	47	info_dict = self.DL.extract_info(self.url, download=False)
ae84f879 JMF	48	return info_dict
	49
	50	def getSubtitles(self):
	51	info_dict = self.getInfoDict()
c84dd8a9	52	subtitles = info_dict['requested_subtitles']
a504ced0 JMF	53	if not subtitles:
	54	return subtitles
	55	for sub_info in subtitles.values():
	56	if sub_info.get('data') is None:
	57	uf = self.DL.urlopen(sub_info['url'])
0f06bcd7	58	sub_info['data'] = uf.read().decode()
86e5f3ed	59	return {l: sub_info['data'] for l, sub_info in subtitles.items()}
ae84f879 JMF	60
ae84f879 JMF	61
060ac762	62	@is_download_test
ae84f879	63	class TestYoutubeSubtitles(BaseTestSubtitles):
6d1b3489	64	# Available subtitles for QRS8MkLhQmM:
	65	# Language formats
	66	# ru vtt, ttml, srv3, srv2, srv1, json3
	67	# fr vtt, ttml, srv3, srv2, srv1, json3
	68	# en vtt, ttml, srv3, srv2, srv1, json3
	69	# nl vtt, ttml, srv3, srv2, srv1, json3
	70	# de vtt, ttml, srv3, srv2, srv1, json3
	71	# ko vtt, ttml, srv3, srv2, srv1, json3
	72	# it vtt, ttml, srv3, srv2, srv1, json3
	73	# zh-Hant vtt, ttml, srv3, srv2, srv1, json3
	74	# hi vtt, ttml, srv3, srv2, srv1, json3
	75	# pt-BR vtt, ttml, srv3, srv2, srv1, json3
	76	# es-MX vtt, ttml, srv3, srv2, srv1, json3
	77	# ja vtt, ttml, srv3, srv2, srv1, json3
	78	# pl vtt, ttml, srv3, srv2, srv1, json3
ae84f879 JMF	79	url = 'QRS8MkLhQmM'
	80	IE = YoutubeIE
	81
ae84f879 JMF	82	def test_youtube_allsubtitles(self):
	83	self.DL.params['writesubtitles'] = True
	84	self.DL.params['allsubtitles'] = True
	85	subtitles = self.getSubtitles()
	86	self.assertEqual(len(subtitles.keys()), 13)
6d1b3489	87	self.assertEqual(md5(subtitles['en']), 'ae1bd34126571a77aabd4d276b28044d')
6d1b3489	88	self.assertEqual(md5(subtitles['it']), '0e0b667ba68411d88fd1c5f4f4eab2f9')
920d21b9	89	for lang in ['fr', 'de']:
b7bb76df	90	self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
ae84f879	91
6d1b3489	92	def _test_subtitles_format(self, fmt, md5_hash, lang='en'):
ae84f879	93	self.DL.params['writesubtitles'] = True
6d1b3489	94	self.DL.params['subtitlesformat'] = fmt
ae84f879	95	subtitles = self.getSubtitles()
6d1b3489	96	self.assertEqual(md5(subtitles[lang]), md5_hash)
	97
	98	def test_youtube_subtitles_ttml_format(self):
	99	self._test_subtitles_format('ttml', 'c97ddf1217390906fa9fbd34901f3da2')
ae84f879 JMF	100
ae84f879 JMF	101	def test_youtube_subtitles_vtt_format(self):
6d1b3489	102	self._test_subtitles_format('vtt', 'ae1bd34126571a77aabd4d276b28044d')
ae84f879	103
6d1b3489	104	def test_youtube_subtitles_json3_format(self):
6d1b3489	105	self._test_subtitles_format('json3', '688dd1ce0981683867e7fe6fde2a224b')
ae84f879	106
6d1b3489	107	def _test_automatic_captions(self, url, lang):
6d1b3489	108	self.url = url
8848f808	109	self.DL.params['writeautomaticsub'] = True
6d1b3489	110	self.DL.params['subtitleslangs'] = [lang]
8848f808	111	subtitles = self.getSubtitles()
6d1b3489	112	self.assertTrue(subtitles[lang] is not None)
8848f808	113
6d1b3489	114	def test_youtube_automatic_captions(self):
	115	# Available automatic captions for 8YoUxe5ncPo:
	116	# Language formats (all in vtt, ttml, srv3, srv2, srv1, json3)
	117	# gu, zh-Hans, zh-Hant, gd, ga, gl, lb, la, lo, tt, tr,
	118	# lv, lt, tk, th, tg, te, fil, haw, yi, ceb, yo, de, da,
	119	# el, eo, en, eu, et, es, ru, rw, ro, bn, be, bg, uk, jv,
	120	# bs, ja, or, xh, co, ca, cy, cs, ps, pt, pa, vi, pl, hy,
	121	# hr, ht, hu, hmn, hi, ha, mg, uz, ml, mn, mi, mk, ur,
	122	# mt, ms, mr, ug, ta, my, af, sw, is, am,
	123	# it, iw, sv, ar,
	124	# su, zu, az, id, ig, nl, no, ne, ny, fr, ku, fy, fa, fi,
	125	# ka, kk, sr, sq, ko, kn, km, st, sk, si, so, sn, sm, sl,
	126	# ky, sd
	127	# ...
	128	self._test_automatic_captions('8YoUxe5ncPo', 'it')
	129
	130	@unittest.skip('Video unavailable')
7d900ef1	131	def test_youtube_translated_subtitles(self):
6d1b3489	132	# This video has a subtitles track, which can be translated (#4555)
6d1b3489	133	self._test_automatic_captions('Ky9eprVWzlI', 'it')
7d900ef1	134
ae84f879	135	def test_youtube_nosubtitles(self):
a0f59cdc	136	self.DL.expect_warning('video doesn\'t have subtitles')
6d1b3489	137	# Available automatic captions for 8YoUxe5ncPo:
	138	# ...
	139	# 8YoUxe5ncPo has no subtitles
	140	self.url = '8YoUxe5ncPo'
ae84f879 JMF	141	self.DL.params['writesubtitles'] = True
	142	self.DL.params['allsubtitles'] = True
	143	subtitles = self.getSubtitles()
4d171848	144	self.assertFalse(subtitles)
ae84f879	145
ae84f879	146
060ac762	147	@is_download_test
ae84f879 JMF	148	class TestDailymotionSubtitles(BaseTestSubtitles):
	149	url = 'http://www.dailymotion.com/video/xczg00'
	150	IE = DailymotionIE
	151
ae84f879 JMF	152	def test_allsubtitles(self):
	153	self.DL.params['writesubtitles'] = True
	154	self.DL.params['allsubtitles'] = True
	155	subtitles = self.getSubtitles()
8fb474fb	156	self.assertTrue(len(subtitles.keys()) >= 6)
b7bb76df JMF	157	self.assertEqual(md5(subtitles['en']), '976553874490cba125086bbfea3ff76f')
	158	self.assertEqual(md5(subtitles['fr']), '594564ec7d588942e384e920e5341792')
	159	for lang in ['es', 'fr', 'de']:
	160	self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
ae84f879	161
ae84f879	162	def test_nosubtitles(self):
a0f59cdc	163	self.DL.expect_warning('video doesn\'t have subtitles')
ae84f879 JMF	164	self.url = 'http://www.dailymotion.com/video/x12u166_le-zapping-tele-star-du-08-aout-2013_tv'
	165	self.DL.params['writesubtitles'] = True
	166	self.DL.params['allsubtitles'] = True
	167	subtitles = self.getSubtitles()
4d171848	168	self.assertFalse(subtitles)
ae84f879	169
ae84f879	170
060ac762	171	@is_download_test
6d1b3489	172	@unittest.skip('IE broken')
ae84f879 JMF	173	class TestTedSubtitles(BaseTestSubtitles):
ae84f879 JMF	174	url = 'http://www.ted.com/talks/dan_dennett_on_our_consciousness.html'
4259402c	175	IE = TedTalkIE
ae84f879	176
ae84f879 JMF	177	def test_allsubtitles(self):
	178	self.DL.params['writesubtitles'] = True
	179	self.DL.params['allsubtitles'] = True
	180	subtitles = self.getSubtitles()
0321213c	181	self.assertTrue(len(subtitles.keys()) >= 28)
b7bb76df JMF	182	self.assertEqual(md5(subtitles['en']), '4262c1665ff928a2dada178f62cb8d14')
	183	self.assertEqual(md5(subtitles['fr']), '66a63f7f42c97a50f8c0e90bc7797bb5')
	184	for lang in ['es', 'fr', 'de']:
a0f59cdc	185	self.assertTrue(subtitles.get(lang) is not None, 'Subtitles for \'%s\' not extracted' % lang)
ae84f879	186
b4bcffef	187
060ac762	188	@is_download_test
4edff4cf S	189	class TestVimeoSubtitles(BaseTestSubtitles):
	190	url = 'http://vimeo.com/76979871'
	191	IE = VimeoIE
b0268cb6	192
4edff4cf S	193	def test_allsubtitles(self):
	194	self.DL.params['writesubtitles'] = True
	195	self.DL.params['allsubtitles'] = True
	196	subtitles = self.getSubtitles()
86e5f3ed	197	self.assertEqual(set(subtitles.keys()), {'de', 'en', 'es', 'fr'})
6d1b3489	198	self.assertEqual(md5(subtitles['en']), '386cbc9320b94e25cb364b97935e5dd1')
6d1b3489	199	self.assertEqual(md5(subtitles['fr']), 'c9b69eef35bc6641c0d4da8a04f9dfac')
4edff4cf	200
4edff4cf	201	def test_nosubtitles(self):
a0f59cdc	202	self.DL.expect_warning('video doesn\'t have subtitles')
6d1b3489	203	self.url = 'http://vimeo.com/68093876'
4edff4cf S	204	self.DL.params['writesubtitles'] = True
	205	self.DL.params['allsubtitles'] = True
	206	subtitles = self.getSubtitles()
4d171848	207	self.assertFalse(subtitles)
4edff4cf	208
4edff4cf	209
060ac762	210	@is_download_test
6d1b3489	211	@unittest.skip('IE broken')
0ca41c3d	212	class TestWallaSubtitles(BaseTestSubtitles):
7bc8780c S	213	url = 'http://vod.walla.co.il/movie/2705958/the-yes-men'
	214	IE = WallaIE
	215
7bc8780c	216	def test_allsubtitles(self):
a0f59cdc	217	self.DL.expect_warning('Automatic Captions not supported by this server')
7bc8780c S	218	self.DL.params['writesubtitles'] = True
	219	self.DL.params['allsubtitles'] = True
	220	subtitles = self.getSubtitles()
86e5f3ed	221	self.assertEqual(set(subtitles.keys()), {'heb'})
7bc8780c S	222	self.assertEqual(md5(subtitles['heb']), 'e758c5d7cb982f6bef14f377ec7a3920')
	223
	224	def test_nosubtitles(self):
a0f59cdc	225	self.DL.expect_warning('video doesn\'t have subtitles')
7bc8780c S	226	self.url = 'http://vod.walla.co.il/movie/2642630/one-direction-all-for-one'
	227	self.DL.params['writesubtitles'] = True
	228	self.DL.params['allsubtitles'] = True
	229	subtitles = self.getSubtitles()
4d171848	230	self.assertFalse(subtitles)
7bc8780c S	231
7bc8780c S	232
060ac762	233	@is_download_test
6d1b3489	234	@unittest.skip('IE broken')
0b54a5b1 S	235	class TestCeskaTelevizeSubtitles(BaseTestSubtitles):
	236	url = 'http://www.ceskatelevize.cz/ivysilani/10600540290-u6-uzasny-svet-techniky'
	237	IE = CeskaTelevizeIE
	238
0b54a5b1 S	239	def test_allsubtitles(self):
	240	self.DL.expect_warning('Automatic Captions not supported by this server')
	241	self.DL.params['writesubtitles'] = True
	242	self.DL.params['allsubtitles'] = True
	243	subtitles = self.getSubtitles()
86e5f3ed	244	self.assertEqual(set(subtitles.keys()), {'cs'})
8fb474fb	245	self.assertTrue(len(subtitles['cs']) > 20000)
0b54a5b1 S	246
	247	def test_nosubtitles(self):
	248	self.DL.expect_warning('video doesn\'t have subtitles')
	249	self.url = 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220'
	250	self.DL.params['writesubtitles'] = True
	251	self.DL.params['allsubtitles'] = True
	252	subtitles = self.getSubtitles()
4d171848	253	self.assertFalse(subtitles)
0b54a5b1 S	254
0b54a5b1 S	255
060ac762	256	@is_download_test
6d1b3489	257	@unittest.skip('IE broken')
311c3938 JMF	258	class TestLyndaSubtitles(BaseTestSubtitles):
	259	url = 'http://www.lynda.com/Bootstrap-tutorials/Using-exercise-files/110885/114408-4.html'
	260	IE = LyndaIE
	261
	262	def test_allsubtitles(self):
	263	self.DL.params['writesubtitles'] = True
	264	self.DL.params['allsubtitles'] = True
	265	subtitles = self.getSubtitles()
86e5f3ed	266	self.assertEqual(set(subtitles.keys()), {'en'})
311c3938 JMF	267	self.assertEqual(md5(subtitles['en']), '09bbe67222259bed60deaa26997d73a7')
	268
	269
060ac762	270	@is_download_test
6d1b3489	271	@unittest.skip('IE broken')
b9b42f2e JMF	272	class TestNPOSubtitles(BaseTestSubtitles):
	273	url = 'http://www.npo.nl/nos-journaal/28-08-2014/POW_00722860'
	274	IE = NPOIE
	275
	276	def test_allsubtitles(self):
	277	self.DL.params['writesubtitles'] = True
	278	self.DL.params['allsubtitles'] = True
	279	subtitles = self.getSubtitles()
86e5f3ed	280	self.assertEqual(set(subtitles.keys()), {'nl'})
b9b42f2e JMF	281	self.assertEqual(md5(subtitles['nl']), 'fc6435027572b63fb4ab143abd5ad3f4')
	282
	283
060ac762	284	@is_download_test
6d1b3489	285	@unittest.skip('IE broken')
0af25f78	286	class TestMTVSubtitles(BaseTestSubtitles):
518d5ba5	287	url = 'http://www.cc.com/video-clips/p63lk0/adam-devine-s-house-party-chasing-white-swans'
0af25f78 JMF	288	IE = ComedyCentralIE
	289
	290	def getInfoDict(self):
86e5f3ed	291	return super().getInfoDict()['entries'][0]
0af25f78 JMF	292
	293	def test_allsubtitles(self):
	294	self.DL.params['writesubtitles'] = True
	295	self.DL.params['allsubtitles'] = True
	296	subtitles = self.getSubtitles()
86e5f3ed	297	self.assertEqual(set(subtitles.keys()), {'en'})
518d5ba5	298	self.assertEqual(md5(subtitles['en']), '78206b8d8a0cfa9da64dc026eea48961')
0af25f78 JMF	299
0af25f78 JMF	300
060ac762	301	@is_download_test
01561da1 JMF	302	class TestNRKSubtitles(BaseTestSubtitles):
	303	url = 'http://tv.nrk.no/serie/ikke-gjoer-dette-hjemme/DMPV73000411/sesong-2/episode-1'
	304	IE = NRKTVIE
	305
	306	def test_allsubtitles(self):
	307	self.DL.params['writesubtitles'] = True
	308	self.DL.params['allsubtitles'] = True
	309	subtitles = self.getSubtitles()
6d1b3489	310	self.assertEqual(set(subtitles.keys()), {'nb-ttv'})
6d1b3489	311	self.assertEqual(md5(subtitles['nb-ttv']), '67e06ff02d0deaf975e68f6cb8f6a149')
01561da1 JMF	312
01561da1 JMF	313
060ac762	314	@is_download_test
c4d6fc6d	315	class TestRaiPlaySubtitles(BaseTestSubtitles):
c4d6fc6d	316	IE = RaiPlayIE
afbdd3ac	317
00dd0cd5	318	def test_subtitles_key(self):
00dd0cd5	319	self.url = 'http://www.raiplay.it/video/2014/04/Report-del-07042014-cb27157f-9dd0-4aee-b788-b1f67643a391.html'
afbdd3ac JMF	320	self.DL.params['writesubtitles'] = True
	321	self.DL.params['allsubtitles'] = True
	322	subtitles = self.getSubtitles()
86e5f3ed	323	self.assertEqual(set(subtitles.keys()), {'it'})
afbdd3ac JMF	324	self.assertEqual(md5(subtitles['it']), 'b1d90a98755126b61e667567a1f6680a')
afbdd3ac JMF	325
00dd0cd5	326	def test_subtitles_array_key(self):
	327	self.url = 'https://www.raiplay.it/video/2020/12/Report---04-01-2021-2e90f1de-8eee-4de4-ac0e-78d21db5b600.html'
	328	self.DL.params['writesubtitles'] = True
	329	self.DL.params['allsubtitles'] = True
	330	subtitles = self.getSubtitles()
86e5f3ed	331	self.assertEqual(set(subtitles.keys()), {'it'})
00dd0cd5	332	self.assertEqual(md5(subtitles['it']), '4b3264186fbb103508abe5311cfcb9cd')
00dd0cd5	333
afbdd3ac	334
060ac762	335	@is_download_test
6d1b3489	336	@unittest.skip('IE broken - DRM only')
4f7cea6c JMF	337	class TestVikiSubtitles(BaseTestSubtitles):
	338	url = 'http://www.viki.com/videos/1060846v-punch-episode-18'
	339	IE = VikiIE
	340
	341	def test_allsubtitles(self):
	342	self.DL.params['writesubtitles'] = True
	343	self.DL.params['allsubtitles'] = True
	344	subtitles = self.getSubtitles()
86e5f3ed	345	self.assertEqual(set(subtitles.keys()), {'en'})
80970e53	346	self.assertEqual(md5(subtitles['en']), '53cb083a5914b2d84ef1ab67b880d18a')
4f7cea6c JMF	347
4f7cea6c JMF	348
060ac762	349	@is_download_test
8807f127 JMF	350	class TestThePlatformSubtitles(BaseTestSubtitles):
	351	# from http://www.3playmedia.com/services-features/tools/integrations/theplatform/
	352	# (see http://theplatform.com/about/partners/type/subtitles-closed-captioning/)
	353	url = 'theplatform:JFUjUE1_ehvq'
	354	IE = ThePlatformIE
	355
	356	def test_allsubtitles(self):
	357	self.DL.params['writesubtitles'] = True
	358	self.DL.params['allsubtitles'] = True
	359	subtitles = self.getSubtitles()
86e5f3ed	360	self.assertEqual(set(subtitles.keys()), {'en'})
8807f127 JMF	361	self.assertEqual(md5(subtitles['en']), '97e7670cbae3c4d26ae8bcc7fdd78d4b')
	362
	363
060ac762	364	@is_download_test
6d1b3489	365	@unittest.skip('IE broken')
f908b74f YCH	366	class TestThePlatformFeedSubtitles(BaseTestSubtitles):
	367	url = 'http://feed.theplatform.com/f/7wvmTC/msnbc_video-p-test?form=json&pretty=true&range=-40&byGuid=n_hardball_5biden_140207'
	368	IE = ThePlatformFeedIE
	369
	370	def test_allsubtitles(self):
	371	self.DL.params['writesubtitles'] = True
	372	self.DL.params['allsubtitles'] = True
	373	subtitles = self.getSubtitles()
86e5f3ed	374	self.assertEqual(set(subtitles.keys()), {'en'})
f908b74f YCH	375	self.assertEqual(md5(subtitles['en']), '48649a22e82b2da21c9a67a395eedade')
	376
	377
060ac762	378	@is_download_test
25ac63ed JMF	379	class TestRtveSubtitles(BaseTestSubtitles):
	380	url = 'http://www.rtve.es/alacarta/videos/los-misterios-de-laura/misterios-laura-capitulo-32-misterio-del-numero-17-2-parte/2428621/'
	381	IE = RTVEALaCartaIE
	382
	383	def test_allsubtitles(self):
	384	print('Skipping, only available from Spain')
	385	return
	386	self.DL.params['writesubtitles'] = True
	387	self.DL.params['allsubtitles'] = True
	388	subtitles = self.getSubtitles()
86e5f3ed	389	self.assertEqual(set(subtitles.keys()), {'es'})
25ac63ed JMF	390	self.assertEqual(md5(subtitles['es']), '69e70cae2d40574fb7316f31d6eb7fca')
	391
	392
060ac762	393	@is_download_test
66d041f2 YCH	394	class TestDemocracynowSubtitles(BaseTestSubtitles):
	395	url = 'http://www.democracynow.org/shows/2015/7/3'
	396	IE = DemocracynowIE
	397
	398	def test_allsubtitles(self):
	399	self.DL.params['writesubtitles'] = True
	400	self.DL.params['allsubtitles'] = True
	401	subtitles = self.getSubtitles()
86e5f3ed	402	self.assertEqual(set(subtitles.keys()), {'en'})
6d1b3489	403	self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')
66d041f2 YCH	404
	405	def test_subtitles_in_page(self):
	406	self.url = 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree'
	407	self.DL.params['writesubtitles'] = True
	408	self.DL.params['allsubtitles'] = True
	409	subtitles = self.getSubtitles()
86e5f3ed	410	self.assertEqual(set(subtitles.keys()), {'en'})
6d1b3489	411	self.assertEqual(md5(subtitles['en']), 'a3cc4c0b5eadd74d9974f1c1f5101045')
66d041f2 YCH	412
66d041f2 YCH	413
a7e999be	414	@is_download_test
	415	class TestPBSSubtitles(BaseTestSubtitles):
	416	url = 'https://www.pbs.org/video/how-fantasy-reflects-our-world-picecq/'
	417	IE = PBSIE
	418
	419	def test_allsubtitles(self):
	420	self.DL.params['writesubtitles'] = True
	421	self.DL.params['allsubtitles'] = True
	422	subtitles = self.getSubtitles()
86e5f3ed	423	self.assertEqual(set(subtitles.keys()), {'en'})
a7e999be	424
	425	def test_subtitles_dfxp_format(self):
	426	self.DL.params['writesubtitles'] = True
	427	self.DL.params['subtitlesformat'] = 'dfxp'
	428	subtitles = self.getSubtitles()
	429	self.assertIn(md5(subtitles['en']), ['643b034254cdc3768ff1e750b6b5873b'])
	430
	431	def test_subtitles_vtt_format(self):
	432	self.DL.params['writesubtitles'] = True
	433	self.DL.params['subtitlesformat'] = 'vtt'
	434	subtitles = self.getSubtitles()
	435	self.assertIn(
	436	md5(subtitles['en']), ['937a05711555b165d4c55a9667017045', 'f49ea998d6824d94959c8152a368ff73'])
	437
	438	def test_subtitles_srt_format(self):
	439	self.DL.params['writesubtitles'] = True
	440	self.DL.params['subtitlesformat'] = 'srt'
	441	subtitles = self.getSubtitles()
	442	self.assertIn(md5(subtitles['en']), ['2082c21b43759d9bf172931b2f2ca371'])
	443
	444	def test_subtitles_sami_format(self):
	445	self.DL.params['writesubtitles'] = True
	446	self.DL.params['subtitlesformat'] = 'sami'
	447	subtitles = self.getSubtitles()
	448	self.assertIn(md5(subtitles['en']), ['4256b16ac7da6a6780fafd04294e85cd'])
	449
	450
ae84f879 JMF	451	if __name__ == '__main__':
ae84f879 JMF	452	unittest.main()