[youtube] Extract additional meta data from video description on youtube music videos

YouTube music videos often have auto-generated video descriptions that can be
utilized to extract additional information about the video. This is desirable
in order to provide the user with as much meta data as possible. This commit
adds extraction methods for the following fields for youtube music videos:
- artist (fallback extraction methods added)
- track (fallback extraction methods added)
- album (new in this commit)
- release_date (new in this commit)
- release_year (new in this commit)

4 tests have been added to test this new functionality:
- YoutubeIE tests 27, 28, 29, and 30

Resolves: #20599
This commit is contained in:
quinlander 2019-04-22 00:26:48 -04:00 committed by Remita Amine
parent aa05a093bb
commit 5caabd3c70

View File

@ -1086,7 +1086,95 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
'skip_download': True,
'youtube_include_dash_manifest': False,
},
}
},
{
# artist and track fields should return non-null, per issue #20599
'url': 'https://music.youtube.com/watch?v=MgNrAu2pzNs',
'info_dict': {
'id': 'MgNrAu2pzNs',
'ext': 'mp4',
'title': 'Voyeur Girl',
'description': 'md5:7ae382a65843d6df2685993e90a8628f',
'upload_date': '20190312',
'uploader': 'Various Artists - Topic',
'uploader_id': 'UCVWKBi1ELZn0QX2CBLSkiyw',
'artist': 'Stephen',
'track': 'Voyeur Girl',
'album': 'it\'s too much love to know my dear',
'release_date': '20190313',
'release_year': 2019,
},
'params': {
'skip_download': True,
},
},
{
# Retrieve 'artist' field from 'Artist:' in video description
# when it is present on youtube music video
# Some videos have release_date and no release_year -
# (release_year should be extracted from release_date)
# https://github.com/ytdl-org/youtube-dl/pull/20742#issuecomment-485740932
'url': 'https://www.youtube.com/watch?v=k0jLE7tTwjY',
'info_dict': {
'id': 'k0jLE7tTwjY',
'ext': 'mp4',
'title': 'Latch Feat. Sam Smith',
'description': 'md5:3cb1e8101a7c85fcba9b4fb41b951335',
'upload_date': '20150110',
'uploader': 'Various Artists - Topic',
'uploader_id': 'UCNkEcmYdjrH4RqtNgh7BZ9w',
'artist': 'Disclosure',
'track': 'Latch Feat. Sam Smith',
'album': 'Latch Featuring Sam Smith',
'release_date': '20121008',
'release_year': 2012,
},
'params': {
'skip_download': True,
},
},
{
# handle multiple artists on youtube music video
'url': 'https://www.youtube.com/watch?v=74qn0eJSjpA',
'info_dict': {
'id': '74qn0eJSjpA',
'ext': 'mp4',
'title': 'Eastside',
'description': 'md5:290516bb73dcbfab0dcc4efe6c3de5f2',
'upload_date': '20180710',
'uploader': 'Benny Blanco - Topic',
'uploader_id': 'UCzqz_ksRu_WkIzmivMdIS7A',
'artist': 'benny blanco, Halsey, Khalid',
'track': 'Eastside',
'album': 'Eastside',
'release_date': '20180713',
'release_year': 2018,
},
'params': {
'skip_download': True,
},
},
{
# handle youtube music video with release_year and no release_date
'url': 'https://www.youtube.com/watch?v=-hcAI0g-f5M',
'info_dict': {
'id': '-hcAI0g-f5M',
'ext': 'mp4',
'title': 'Put It On Me',
'description': 'md5:93c55acc682ae7b0c668f2e34e1c069e',
'upload_date': '20180426',
'uploader': 'Matt Maeson - Topic',
'uploader_id': 'UCnEkIGqtGcQMLk73Kp-Q5LQ',
'artist': 'Matt Maeson',
'track': 'Put It On Me',
'album': 'The Hearse',
'release_date': None,
'release_year': 2018,
},
'params': {
'skip_download': True,
},
},
]
def __init__(self, *args, **kwargs):
@ -2073,6 +2161,36 @@ def extract_meta(field):
track = extract_meta('Song')
artist = extract_meta('Artist')
album = None
release_date = None
release_year = None
description_info = video_description.split('\n\n')
# If the description of the video has the youtube music auto-generated format, extract additional info
if len(description_info) >= 5 and description_info[-1] == 'Auto-generated by YouTube.':
track_artist = description_info[1].split(' · ')
if len(track_artist) >= 2:
if track is None:
track = track_artist[0]
if artist is None:
artist = re.search(r'Artist: ([^\n]+)', description_info[-2])
if artist:
artist = artist.group(1)
if artist is None:
artist = track_artist[1]
# handle multiple artists
if len(track_artist) > 2:
for i in range(2, len(track_artist)):
artist += ', %s' % track_artist[i]
release_year = re.search(r'℗ ([0-9]+)', video_description)
if release_year:
release_year = int_or_none(release_year.group(1))
album = description_info[2]
if description_info[4].startswith('Released on: '):
release_date = description_info[4].split(': ')[1].replace('-', '')
# extract release_year from release_date if necessary
if release_year is None:
release_year = int_or_none(release_date[0:4])
m_episode = re.search(
r'<div[^>]+id="watch7-headline"[^>]*>\s*<span[^>]*>.*?>(?P<series>[^<]+)</a></b>\s*S(?P<season>\d+)\s*•\s*E(?P<episode>\d+)</span>',
@ -2226,6 +2344,9 @@ def decrypt_sig(mobj):
'episode_number': episode_number,
'track': track,
'artist': artist,
'album': album,
'release_date': release_date,
'release_year': release_year,
}