[youtube] Extract license (Closes #8725)

This commit is contained in:
Sergey M․ 2016-03-02 23:07:25 +06:00
parent 2bc0c46f98
commit 7caf9830b0

View File

@ -384,6 +384,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader': 'Philipp Hagemeister', 'uploader': 'Philipp Hagemeister',
'uploader_id': 'phihag', 'uploader_id': 'phihag',
'upload_date': '20121002', 'upload_date': '20121002',
'license': 'Standard YouTube License',
'description': 'test chars: "\'/\\ä↭𝕐\ntest URL: https://github.com/rg3/youtube-dl/issues/1892\n\nThis is a test video for youtube-dl.\n\nFor more information, contact phihag@phihag.de .', 'description': 'test chars: "\'/\\ä↭𝕐\ntest URL: https://github.com/rg3/youtube-dl/issues/1892\n\nThis is a test video for youtube-dl.\n\nFor more information, contact phihag@phihag.de .',
'categories': ['Science & Technology'], 'categories': ['Science & Technology'],
'tags': ['youtube-dl'], 'tags': ['youtube-dl'],
@ -402,12 +403,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'upload_date': '20120506', 'upload_date': '20120506',
'title': 'Icona Pop - I Love It (feat. Charli XCX) [OFFICIAL VIDEO]', 'title': 'Icona Pop - I Love It (feat. Charli XCX) [OFFICIAL VIDEO]',
'alt_title': 'I Love It (feat. Charli XCX)', 'alt_title': 'I Love It (feat. Charli XCX)',
'description': 'md5:782e8651347686cba06e58f71ab51773', 'description': 'md5:f3ceb5ef83a08d95b9d146f973157cc8',
'tags': ['Icona Pop i love it', 'sweden', 'pop music', 'big beat records', 'big beat', 'charli', 'tags': ['Icona Pop i love it', 'sweden', 'pop music', 'big beat records', 'big beat', 'charli',
'xcx', 'charli xcx', 'girls', 'hbo', 'i love it', "i don't care", 'icona', 'pop', 'xcx', 'charli xcx', 'girls', 'hbo', 'i love it', "i don't care", 'icona', 'pop',
'iconic ep', 'iconic', 'love', 'it'], 'iconic ep', 'iconic', 'love', 'it'],
'uploader': 'Icona Pop', 'uploader': 'Icona Pop',
'uploader_id': 'IconaPop', 'uploader_id': 'IconaPop',
'license': 'Standard YouTube License',
'creator': 'Icona Pop', 'creator': 'Icona Pop',
} }
}, },
@ -423,6 +425,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'description': 'md5:64249768eec3bc4276236606ea996373', 'description': 'md5:64249768eec3bc4276236606ea996373',
'uploader': 'justintimberlakeVEVO', 'uploader': 'justintimberlakeVEVO',
'uploader_id': 'justintimberlakeVEVO', 'uploader_id': 'justintimberlakeVEVO',
'license': 'Standard YouTube License',
'creator': 'Justin Timberlake', 'creator': 'Justin Timberlake',
'age_limit': 18, 'age_limit': 18,
} }
@ -438,6 +441,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'description': 'md5:09b78bd971f1e3e289601dfba15ca4f7', 'description': 'md5:09b78bd971f1e3e289601dfba15ca4f7',
'uploader': 'SET India', 'uploader': 'SET India',
'uploader_id': 'setindia', 'uploader_id': 'setindia',
'license': 'Standard YouTube License',
'age_limit': 18, 'age_limit': 18,
} }
}, },
@ -451,6 +455,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader': 'Philipp Hagemeister', 'uploader': 'Philipp Hagemeister',
'uploader_id': 'phihag', 'uploader_id': 'phihag',
'upload_date': '20121002', 'upload_date': '20121002',
'license': 'Standard YouTube License',
'description': 'test chars: "\'/\\ä↭𝕐\ntest URL: https://github.com/rg3/youtube-dl/issues/1892\n\nThis is a test video for youtube-dl.\n\nFor more information, contact phihag@phihag.de .', 'description': 'test chars: "\'/\\ä↭𝕐\ntest URL: https://github.com/rg3/youtube-dl/issues/1892\n\nThis is a test video for youtube-dl.\n\nFor more information, contact phihag@phihag.de .',
'categories': ['Science & Technology'], 'categories': ['Science & Technology'],
'tags': ['youtube-dl'], 'tags': ['youtube-dl'],
@ -471,6 +476,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader_id': '8KVIDEO', 'uploader_id': '8KVIDEO',
'description': '', 'description': '',
'uploader': '8KVIDEO', 'uploader': '8KVIDEO',
'license': 'Standard YouTube License',
'title': 'UHDTV TEST 8K VIDEO.mp4' 'title': 'UHDTV TEST 8K VIDEO.mp4'
}, },
'params': { 'params': {
@ -489,6 +495,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader': 'AfrojackVEVO', 'uploader': 'AfrojackVEVO',
'uploader_id': 'AfrojackVEVO', 'uploader_id': 'AfrojackVEVO',
'upload_date': '20131011', 'upload_date': '20131011',
'license': 'Standard YouTube License',
}, },
'params': { 'params': {
'youtube_include_dash_manifest': True, 'youtube_include_dash_manifest': True,
@ -507,6 +514,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader': 'TaylorSwiftVEVO', 'uploader': 'TaylorSwiftVEVO',
'uploader_id': 'TaylorSwiftVEVO', 'uploader_id': 'TaylorSwiftVEVO',
'upload_date': '20140818', 'upload_date': '20140818',
'license': 'Standard YouTube License',
'creator': 'Taylor Swift', 'creator': 'Taylor Swift',
}, },
'params': { 'params': {
@ -523,6 +531,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'upload_date': '20100909', 'upload_date': '20100909',
'uploader': 'The Amazing Atheist', 'uploader': 'The Amazing Atheist',
'uploader_id': 'TheAmazingAtheist', 'uploader_id': 'TheAmazingAtheist',
'license': 'Standard YouTube License',
'title': 'Burning Everyone\'s Koran', 'title': 'Burning Everyone\'s Koran',
'description': 'SUBSCRIBE: http://www.youtube.com/saturninefilms\n\nEven Obama has taken a stand against freedom on this issue: http://www.huffingtonpost.com/2010/09/09/obama-gma-interview-quran_n_710282.html', 'description': 'SUBSCRIBE: http://www.youtube.com/saturninefilms\n\nEven Obama has taken a stand against freedom on this issue: http://www.huffingtonpost.com/2010/09/09/obama-gma-interview-quran_n_710282.html',
} }
@ -538,6 +547,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader': 'The Witcher', 'uploader': 'The Witcher',
'uploader_id': 'WitcherGame', 'uploader_id': 'WitcherGame',
'upload_date': '20140605', 'upload_date': '20140605',
'license': 'Standard YouTube License',
'age_limit': 18, 'age_limit': 18,
}, },
}, },
@ -552,6 +562,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader': 'LloydVEVO', 'uploader': 'LloydVEVO',
'uploader_id': 'LloydVEVO', 'uploader_id': 'LloydVEVO',
'upload_date': '20110629', 'upload_date': '20110629',
'license': 'Standard YouTube License',
'age_limit': 18, 'age_limit': 18,
}, },
}, },
@ -566,6 +577,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'creator': 'deadmau5', 'creator': 'deadmau5',
'description': 'md5:12c56784b8032162bb936a5f76d55360', 'description': 'md5:12c56784b8032162bb936a5f76d55360',
'uploader': 'deadmau5', 'uploader': 'deadmau5',
'license': 'Standard YouTube License',
'title': 'Deadmau5 - Some Chords (HD)', 'title': 'Deadmau5 - Some Chords (HD)',
'alt_title': 'Some Chords', 'alt_title': 'Some Chords',
}, },
@ -581,6 +593,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'upload_date': '20150827', 'upload_date': '20150827',
'uploader_id': 'olympic', 'uploader_id': 'olympic',
'license': 'Standard YouTube License',
'description': 'HO09 - Women - GER-AUS - Hockey - 31 July 2012 - London 2012 Olympic Games', 'description': 'HO09 - Women - GER-AUS - Hockey - 31 July 2012 - London 2012 Olympic Games',
'uploader': 'Olympics', 'uploader': 'Olympics',
'title': 'Hockey - Women - GER-AUS - London 2012 Olympic Games', 'title': 'Hockey - Women - GER-AUS - London 2012 Olympic Games',
@ -600,6 +613,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader_id': 'AllenMeow', 'uploader_id': 'AllenMeow',
'description': 'made by Wacom from Korea | 字幕&加油添醋 by TY\'s Allen | 感謝heylisa00cavey1001同學熱情提供梗及翻譯', 'description': 'made by Wacom from Korea | 字幕&加油添醋 by TY\'s Allen | 感謝heylisa00cavey1001同學熱情提供梗及翻譯',
'uploader': '孫艾倫', 'uploader': '孫艾倫',
'license': 'Standard YouTube License',
'title': '[A-made] 變態妍字幕版 太妍 我就是這樣的人', 'title': '[A-made] 變態妍字幕版 太妍 我就是這樣的人',
}, },
}, },
@ -631,6 +645,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'upload_date': '20150625', 'upload_date': '20150625',
'uploader_id': 'dorappi2000', 'uploader_id': 'dorappi2000',
'uploader': 'dorappi2000', 'uploader': 'dorappi2000',
'license': 'Standard YouTube License',
'formats': 'mincount:33', 'formats': 'mincount:33',
}, },
}, },
@ -645,6 +660,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'uploader': 'Airtek', 'uploader': 'Airtek',
'description': 'Retransmisión en directo de la XVIII media maratón de Zaragoza.', 'description': 'Retransmisión en directo de la XVIII media maratón de Zaragoza.',
'uploader_id': 'UCzTzUmjXxxacNnL8I3m4LnQ', 'uploader_id': 'UCzTzUmjXxxacNnL8I3m4LnQ',
'license': 'Standard YouTube License',
'title': 'Retransmisión XVIII Media maratón Zaragoza 2015', 'title': 'Retransmisión XVIII Media maratón Zaragoza 2015',
}, },
'params': { 'params': {
@ -669,6 +685,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'upload_date': '20150721', 'upload_date': '20150721',
'uploader': 'Beer Games Beer', 'uploader': 'Beer Games Beer',
'uploader_id': 'beergamesbeer', 'uploader_id': 'beergamesbeer',
'license': 'Standard YouTube License',
}, },
}, { }, {
'info_dict': { 'info_dict': {
@ -679,6 +696,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'upload_date': '20150721', 'upload_date': '20150721',
'uploader': 'Beer Games Beer', 'uploader': 'Beer Games Beer',
'uploader_id': 'beergamesbeer', 'uploader_id': 'beergamesbeer',
'license': 'Standard YouTube License',
}, },
}, { }, {
'info_dict': { 'info_dict': {
@ -689,6 +707,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'upload_date': '20150721', 'upload_date': '20150721',
'uploader': 'Beer Games Beer', 'uploader': 'Beer Games Beer',
'uploader_id': 'beergamesbeer', 'uploader_id': 'beergamesbeer',
'license': 'Standard YouTube License',
}, },
}, { }, {
'info_dict': { 'info_dict': {
@ -699,6 +718,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'upload_date': '20150721', 'upload_date': '20150721',
'uploader': 'Beer Games Beer', 'uploader': 'Beer Games Beer',
'uploader_id': 'beergamesbeer', 'uploader_id': 'beergamesbeer',
'license': 'Standard YouTube License',
}, },
}], }],
'params': { 'params': {
@ -733,6 +753,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'upload_date': '20151119', 'upload_date': '20151119',
'uploader_id': 'IronSoulElf', 'uploader_id': 'IronSoulElf',
'uploader': 'IronSoulElf', 'uploader': 'IronSoulElf',
'license': 'Standard YouTube License',
'creator': 'Todd Haberman, Daniel Law Heath & Aaron Kaplan', 'creator': 'Todd Haberman, Daniel Law Heath & Aaron Kaplan',
}, },
'params': { 'params': {
@ -760,6 +781,23 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'skip_download': True, 'skip_download': True,
}, },
}, },
{
# Video licensed under Creative Commons
'url': 'https://www.youtube.com/watch?v=M4gD1WSo5mA',
'info_dict': {
'id': 'M4gD1WSo5mA',
'ext': 'mp4',
'title': 'md5:e41008789470fc2533a3252216f1c1d1',
'description': 'md5:a677553cf0840649b731a3024aeff4cc',
'upload_date': '20150127',
'uploader_id': 'BerkmanCenter',
'uploader': 'BerkmanCenter',
'license': 'Creative Commons Attribution license (reuse allowed)',
},
'params': {
'skip_download': True,
},
},
{ {
'url': 'https://www.youtube.com/watch?feature=player_embedded&v=V36LpHqtcDY', 'url': 'https://www.youtube.com/watch?feature=player_embedded&v=V36LpHqtcDY',
'only_matching': True, 'only_matching': True,
@ -1326,6 +1364,10 @@ def add_dash_mpd(video_info):
upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split()) upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split())
upload_date = unified_strdate(upload_date) upload_date = unified_strdate(upload_date)
video_license = self._html_search_regex(
r'<h4[^>]+class="title"[^>]*>\s*License\s*</h4>\s*<ul[^>]*>\s*<li>(.+?)</li',
video_webpage, 'license', default=None)
m_music = re.search( m_music = re.search(
r'<h4[^>]+class="title"[^>]*>\s*Music\s*</h4>\s*<ul[^>]*>\s*<li>(?P<title>.+?) by (?P<creator>.+?)(?:\(.+?\))?</li', r'<h4[^>]+class="title"[^>]*>\s*Music\s*</h4>\s*<ul[^>]*>\s*<li>(?P<title>.+?) by (?P<creator>.+?)(?:\(.+?\))?</li',
video_webpage) video_webpage)
@ -1601,6 +1643,7 @@ def decrypt_sig(mobj):
'uploader': video_uploader, 'uploader': video_uploader,
'uploader_id': video_uploader_id, 'uploader_id': video_uploader_id,
'upload_date': upload_date, 'upload_date': upload_date,
'license': video_license,
'creator': video_creator, 'creator': video_creator,
'title': video_title, 'title': video_title,
'alt_title': video_alt_title, 'alt_title': video_alt_title,