X-Git-Url: http://git.bitcoin.ninja/index.cgi?p=youtube-dl;a=blobdiff_plain;f=test%2Ftest_InfoExtractor.py;h=71f6608feae4a5bcad37e96d74e795e7535a5dc7;hp=da6cd39b65c4e23a6cd6059631b2a75ed04b437b;hb=6cd452acffe8d79c895a2ebd0346e2ba7f9e112f;hpb=067aa17edf5a46a8cbc4d6b90864eddf051fa2bc diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index da6cd39b6..71f6608fe 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -107,6 +107,184 @@ class TestInfoExtractor(unittest.TestCase): self.assertRaises(ExtractorError, self.ie._download_json, uri, None) self.assertEqual(self.ie._download_json(uri, None, fatal=False), None) + def test_parse_html5_media_entries(self): + # from https://www.r18.com/ + # with kpbs in label + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.r18.com/', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://awscc3001.r18.com/litevideo/freepv/m/mgm/mgmr105/mgmr105_sm_w.mp4', + 'ext': 'mp4', + 'format_id': '300kbps', + 'height': 240, + 'tbr': 300, + }, { + 'url': 'https://awscc3001.r18.com/litevideo/freepv/m/mgm/mgmr105/mgmr105_dm_w.mp4', + 'ext': 'mp4', + 'format_id': '1000kbps', + 'height': 480, + 'tbr': 1000, + }, { + 'url': 'https://awscc3001.r18.com/litevideo/freepv/m/mgm/mgmr105/mgmr105_dmb_w.mp4', + 'ext': 'mp4', + 'format_id': '1500kbps', + 'height': 740, + 'tbr': 1500, + }], + 'thumbnail': '//pics.r18.com/digital/amateur/mgmr105/mgmr105jp.jpg' + }) + + # from https://www.csfd.cz/ + # with width and height + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.csfd.cz/', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327358_eac647.mp4', + 'ext': 'mp4', + 'width': 640, + 'height': 360, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327360_3d2646.mp4', + 'ext': 'mp4', + 'width': 1280, + 'height': 720, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327356_91f258.mp4', + 'ext': 'mp4', + 'width': 1920, + 'height': 1080, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327359_962b4a.webm', + 'ext': 'webm', + 'width': 640, + 'height': 360, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327361_6feee0.webm', + 'ext': 'webm', + 'width': 1280, + 'height': 720, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327357_8ab472.webm', + 'ext': 'webm', + 'width': 1920, + 'height': 1080, + }], + 'subtitles': { + 'cs': [{'url': 'https://video.csfd.cz/files/subtitles/163/344/163344115_4c388b.srt'}] + }, + 'thumbnail': 'https://img.csfd.cz/files/images/film/video/preview/163/344/163344118_748d20.png?h360' + }) + + # from https://tamasha.com/v/Kkdjw + # with height in label + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://tamasha.com/v/Kkdjw', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://s-v2.tamasha.com/statics/videos_file/19/8f/Kkdjw_198feff8577d0057536e905cce1fb61438dd64e0_n_240.mp4', + }, { + 'url': 'https://s-v2.tamasha.com/statics/videos_file/19/8f/Kkdjw_198feff8577d0057536e905cce1fb61438dd64e0_n_240.mp4', + 'ext': 'mp4', + 'format_id': '240p', + 'height': 240, + }, { + 'url': 'https://s-v2.tamasha.com/statics/videos_file/20/00/Kkdjw_200041c66f657fc967db464d156eafbc1ed9fe6f_n_144.mp4', + 'ext': 'mp4', + 'format_id': '144p', + 'height': 144, + }] + }) + + # from https://www.directvnow.com + # with data-src + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.directvnow.com', + r''' + + ''', None)[0], + { + 'formats': [{ + 'ext': 'mp4', + 'url': 'https://cdn.directv.com/content/dam/dtv/prod/website_directvnow-international/videos/DTVN_hdr_HBO_v3.mp4', + }] + }) + + # from https://www.directvnow.com + # with data-src + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.directvnow.com', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://cdn.directv.com/content/dam/dtv/prod/website_directvnow-international/videos/DTVN_hdr_HBO_v3.mp4', + 'ext': 'mp4', + }] + }) + + # from https://www.klarna.com/uk/ + # with data-video-src + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.directvnow.com', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://www.klarna.com/uk/wp-content/uploads/sites/11/2019/01/KL062_Smooth3_0_DogWalking_5s_920x080_.mp4', + 'ext': 'mp4', + }], + }) + def test_extract_jwplayer_data_realworld(self): # from http://www.suffolk.edu/sjc/ expect_dict( @@ -574,7 +752,8 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ # Also tests duplicate representation ids, see # https://github.com/ytdl-org/youtube-dl/issues/15111 'float_duration', - 'http://unknown/manifest.mpd', + 'http://unknown/manifest.mpd', # mpd_url + None, # mpd_base_url [{ 'manifest_url': 'http://unknown/manifest.mpd', 'ext': 'm4a', @@ -654,7 +833,8 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ ), ( # https://github.com/ytdl-org/youtube-dl/pull/14844 'urls_only', - 'http://unknown/manifest.mpd', + 'http://unknown/manifest.mpd', # mpd_url + None, # mpd_base_url [{ 'manifest_url': 'http://unknown/manifest.mpd', 'ext': 'mp4', @@ -733,15 +913,61 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'width': 1920, 'height': 1080, }] + ), ( + # https://github.com/ytdl-org/youtube-dl/issues/20346 + # Media considered unfragmented even though it contains + # Initialization tag + 'unfragmented', + 'https://v.redd.it/hw1x7rcg7zl21/DASHPlaylist.mpd', # mpd_url + 'https://v.redd.it/hw1x7rcg7zl21', # mpd_base_url + [{ + 'url': 'https://v.redd.it/hw1x7rcg7zl21/audio', + 'manifest_url': 'https://v.redd.it/hw1x7rcg7zl21/DASHPlaylist.mpd', + 'ext': 'm4a', + 'format_id': 'AUDIO-1', + 'format_note': 'DASH audio', + 'container': 'm4a_dash', + 'acodec': 'mp4a.40.2', + 'vcodec': 'none', + 'tbr': 129.87, + 'asr': 48000, + + }, { + 'url': 'https://v.redd.it/hw1x7rcg7zl21/DASH_240', + 'manifest_url': 'https://v.redd.it/hw1x7rcg7zl21/DASHPlaylist.mpd', + 'ext': 'mp4', + 'format_id': 'VIDEO-2', + 'format_note': 'DASH video', + 'container': 'mp4_dash', + 'acodec': 'none', + 'vcodec': 'avc1.4d401e', + 'tbr': 608.0, + 'width': 240, + 'height': 240, + 'fps': 30, + }, { + 'url': 'https://v.redd.it/hw1x7rcg7zl21/DASH_360', + 'manifest_url': 'https://v.redd.it/hw1x7rcg7zl21/DASHPlaylist.mpd', + 'ext': 'mp4', + 'format_id': 'VIDEO-1', + 'format_note': 'DASH video', + 'container': 'mp4_dash', + 'acodec': 'none', + 'vcodec': 'avc1.4d401e', + 'tbr': 804.261, + 'width': 360, + 'height': 360, + 'fps': 30, + }] ) ] - for mpd_file, mpd_url, expected_formats in _TEST_CASES: + for mpd_file, mpd_url, mpd_base_url, expected_formats in _TEST_CASES: with io.open('./test/testdata/mpd/%s.mpd' % mpd_file, mode='r', encoding='utf-8') as f: formats = self.ie._parse_mpd_formats( compat_etree_fromstring(f.read().encode('utf-8')), - mpd_url=mpd_url) + mpd_base_url=mpd_base_url, mpd_url=mpd_url) self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None)