[youtube] Improve metadata extraction for age gate content (closes #21943)

This commit is contained in:
Sergey M․ 2019-07-31 00:14:08 +07:00
parent 85c2c4b4ab
commit c2d125d99f
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -1700,6 +1700,15 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
def extract_token(v_info): def extract_token(v_info):
return dict_get(v_info, ('account_playback_token', 'accountPlaybackToken', 'token')) return dict_get(v_info, ('account_playback_token', 'accountPlaybackToken', 'token'))
def extract_player_response(player_response, video_id):
pl_response = str_or_none(player_response)
if not pl_response:
return
pl_response = self._parse_json(pl_response, video_id, fatal=False)
if isinstance(pl_response, dict):
add_dash_mpd_pr(pl_response)
return pl_response
player_response = {} player_response = {}
# Get video info # Get video info
@ -1722,7 +1731,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
note='Refetching age-gated info webpage', note='Refetching age-gated info webpage',
errnote='unable to download video info webpage') errnote='unable to download video info webpage')
video_info = compat_parse_qs(video_info_webpage) video_info = compat_parse_qs(video_info_webpage)
pl_response = video_info.get('player_response', [None])[0]
player_response = extract_player_response(pl_response, video_id)
add_dash_mpd(video_info) add_dash_mpd(video_info)
view_count = extract_view_count(video_info)
else: else:
age_gate = False age_gate = False
video_info = None video_info = None
@ -1745,11 +1757,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
is_live = True is_live = True
sts = ytplayer_config.get('sts') sts = ytplayer_config.get('sts')
if not player_response: if not player_response:
pl_response = str_or_none(args.get('player_response')) player_response = extract_player_response(args.get('player_response'), video_id)
if pl_response:
pl_response = self._parse_json(pl_response, video_id, fatal=False)
if isinstance(pl_response, dict):
player_response = pl_response
if not video_info or self._downloader.params.get('youtube_include_dash_manifest', True): if not video_info or self._downloader.params.get('youtube_include_dash_manifest', True):
add_dash_mpd_pr(player_response) add_dash_mpd_pr(player_response)
# We also try looking in get_video_info since it may contain different dashmpd # We also try looking in get_video_info since it may contain different dashmpd
@ -1781,9 +1789,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
get_video_info = compat_parse_qs(video_info_webpage) get_video_info = compat_parse_qs(video_info_webpage)
if not player_response: if not player_response:
pl_response = get_video_info.get('player_response', [None])[0] pl_response = get_video_info.get('player_response', [None])[0]
if isinstance(pl_response, dict): player_response = extract_player_response(pl_response, video_id)
player_response = pl_response
add_dash_mpd_pr(player_response)
add_dash_mpd(get_video_info) add_dash_mpd(get_video_info)
if view_count is None: if view_count is None:
view_count = extract_view_count(get_video_info) view_count = extract_view_count(get_video_info)