aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorRemita Amine <remitamine@gmail.com>2019-04-27 09:16:17 +0100
committerRemita Amine <remitamine@gmail.com>2019-04-27 09:16:17 +0100
commit822b9d9cb09429645582791dba31f4cbed7583cf (patch)
treee2fcf4121007b62fd44f1ee85caecef06d3befc2
parent5caabd3c701a484271d197f7006ecf831e38136b (diff)
downloadyoutube-dl-822b9d9cb09429645582791dba31f4cbed7583cf.tar.xz
[youtube] improve Youtube Music Auto-generated description parsing(closes #20742)
-rw-r--r--youtube_dl/extractor/youtube.py57
1 files changed, 23 insertions, 34 deletions
diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 438eb5aa7..55eafb866 100644
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -1088,7 +1088,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
},
},
{
- # artist and track fields should return non-null, per issue #20599
+ # Youtube Music Auto-generated description
'url': 'https://music.youtube.com/watch?v=MgNrAu2pzNs',
'info_dict': {
'id': 'MgNrAu2pzNs',
@@ -1109,11 +1109,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
},
},
{
+ # Youtube Music Auto-generated description
# 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',
@@ -1134,6 +1132,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
},
},
{
+ # Youtube Music Auto-generated description
# handle multiple artists on youtube music video
'url': 'https://www.youtube.com/watch?v=74qn0eJSjpA',
'info_dict': {
@@ -1155,6 +1154,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
},
},
{
+ # Youtube Music Auto-generated description
# handle youtube music video with release_year and no release_date
'url': 'https://www.youtube.com/watch?v=-hcAI0g-f5M',
'info_dict': {
@@ -2161,36 +2161,25 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
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])
+
+ # Youtube Music Auto-generated description
+ album = release_date = release_year = None
+ if video_description:
+ mobj = re.search(r'(?s)Provided to YouTube by [^\n]+\n+(?P<track>[^·]+)·(?P<artist>[^\n]+)\n+(?P<album>[^\n]+)(?:.+?℗\s*(?P<release_year>\d{4})(?!\d))?(?:.+?Released on\s*:\s*(?P<release_date>\d{4}-\d{2}-\d{2}))?(.+?\nArtist\s*:\s*(?P<clean_artist>[^\n]+))?', video_description)
+ if mobj:
+ if not track:
+ track = mobj.group('track').strip()
+ if not artist:
+ artist = mobj.group('clean_artist') or ', '.join(a.strip() for a in mobj.group('artist').split('·'))
+ album = mobj.group('album'.strip())
+ release_year = mobj.group('release_year')
+ release_date = mobj.group('release_date')
+ if release_date:
+ release_date = release_date.replace('-', '')
+ if not release_year:
+ release_year = int(release_date[:4])
+ if release_year:
+ release_year = int(release_year)
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>',