github/yt-dlp
1
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-12-17 23:33:57 -05:00
Code Issues Projects Releases Packages Wiki Activity

[youtube] Avoid false positives when detecting damaged formats

Browse source 
Closes #3083
This commit is contained in:
pukkandan 2022-03-16 19:46:19 +05:30
parent 5dbc77df26
commit a1b2d84360
No known key found for this signature in database
GPG key ID: 7EEE9E1E817D0A39
1 changed files with 13 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

24
yt_dlp/extractor/youtube.py
View file

@ -1297,7 +1297,6 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
}, },
'expected_warnings': [ 'expected_warnings': [
'DASH manifest missing', 'DASH manifest missing',
'Some formats are possibly damaged'
] ]
}, },
# Olympics (https://github.com/ytdl-org/youtube-dl/issues/4431) # Olympics (https://github.com/ytdl-org/youtube-dl/issues/4431)
@ -3013,7 +3012,7 @@ def append_client(client_name):
self.report_warning(last_error) self.report_warning(last_error)
return prs, player_url return prs, player_url
def _extract_formats(self, streaming_data, video_id, player_url, is_live): def _extract_formats(self, streaming_data, video_id, player_url, is_live, duration):
itags, stream_ids = {}, [] itags, stream_ids = {}, []
itag_qualities, res_qualities = {}, {} itag_qualities, res_qualities = {}, {}
q = qualities([ q = qualities([
@ -3024,7 +3023,6 @@ def _extract_formats(self, streaming_data, video_id, player_url, is_live):
'small', 'medium', 'large', 'hd720', 'hd1080', 'hd1440', 'hd2160', 'hd2880', 'highres' 'small', 'medium', 'large', 'hd720', 'hd1080', 'hd1440', 'hd2160', 'hd2880', 'highres'
]) ])
streaming_formats = traverse_obj(streaming_data, (..., ('formats', 'adaptiveFormats'), ...), default=[]) streaming_formats = traverse_obj(streaming_data, (..., ('formats', 'adaptiveFormats'), ...), default=[])
approx_duration = max(traverse_obj(streaming_formats, (..., 'approxDurationMs'), expected_type=float_or_none) or [0]) or None
for fmt in streaming_formats: for fmt in streaming_formats:
if fmt.get('targetDurationSec') or fmt.get('drmFamilies'): if fmt.get('targetDurationSec') or fmt.get('drmFamilies'):
@ -3091,7 +3089,9 @@ def _extract_formats(self, streaming_data, video_id, player_url, is_live):
else -1) else -1)
# Some formats may have much smaller duration than others (possibly damaged during encoding) # Some formats may have much smaller duration than others (possibly damaged during encoding)
# Eg: 2-nOtRESiUc Ref: https://github.com/yt-dlp/yt-dlp/issues/2823 # Eg: 2-nOtRESiUc Ref: https://github.com/yt-dlp/yt-dlp/issues/2823
is_damaged = try_get(fmt, lambda x: float(x['approxDurationMs']) < approx_duration - 10000) # Make sure to avoid false positives with small duration differences.
# Eg: __2ABJjxzNo, ySuUZEjARPY
is_damaged = try_get(fmt, lambda x: float(x['approxDurationMs']) / duration < 500)
if is_damaged: if is_damaged:
self.report_warning(f'{video_id}: Some formats are possibly damaged. They will be deprioritized', only_once=True) self.report_warning(f'{video_id}: Some formats are possibly damaged. They will be deprioritized', only_once=True)
dct = { dct = {
@ -3227,14 +3227,14 @@ def _download_player_responses(self, url, smuggled_data, video_id, webpage_url):
return webpage, master_ytcfg, player_responses, player_url return webpage, master_ytcfg, player_responses, player_url
def _list_formats(self, video_id, microformats, video_details, player_responses, player_url): def _list_formats(self, video_id, microformats, video_details, player_responses, player_url, duration=None):
live_broadcast_details = traverse_obj(microformats, (..., 'liveBroadcastDetails')) live_broadcast_details = traverse_obj(microformats, (..., 'liveBroadcastDetails'))
is_live = get_first(video_details, 'isLive') is_live = get_first(video_details, 'isLive')
if is_live is None: if is_live is None:
is_live = get_first(live_broadcast_details, 'isLiveNow') is_live = get_first(live_broadcast_details, 'isLiveNow')
streaming_data = traverse_obj(player_responses, (..., 'streamingData'), default=[]) streaming_data = traverse_obj(player_responses, (..., 'streamingData'), default=[])
formats = list(self._extract_formats(streaming_data, video_id, player_url, is_live)) formats = list(self._extract_formats(streaming_data, video_id, player_url, is_live, duration))
return live_broadcast_details, is_live, streaming_data, formats return live_broadcast_details, is_live, streaming_data, formats
@ -3315,7 +3315,13 @@ def feed_entry(name):
return self.playlist_result( return self.playlist_result(
entries, video_id, video_title, video_description) entries, video_id, video_title, video_description)
live_broadcast_details, is_live, streaming_data, formats = self._list_formats(video_id, microformats, video_details, player_responses, player_url) duration = int_or_none(
get_first(video_details, 'lengthSeconds')
or get_first(microformats, 'lengthSeconds')
or parse_duration(search_meta('duration'))) or None
live_broadcast_details, is_live, streaming_data, formats = self._list_formats(
video_id, microformats, video_details, player_responses, player_url, duration)
if not formats: if not formats:
if not self.get_param('allow_unplayable_formats') and traverse_obj(streaming_data, (..., 'licenseInfos')): if not self.get_param('allow_unplayable_formats') and traverse_obj(streaming_data, (..., 'licenseInfos')):
@ -3387,10 +3393,6 @@ def feed_entry(name):
get_first(video_details, 'channelId') get_first(video_details, 'channelId')
or get_first(microformats, 'externalChannelId') or get_first(microformats, 'externalChannelId')
or search_meta('channelId')) or search_meta('channelId'))
duration = int_or_none(
get_first(video_details, 'lengthSeconds')
or get_first(microformats, 'lengthSeconds')
or parse_duration(search_meta('duration'))) or None
owner_profile_url = get_first(microformats, 'ownerProfileUrl') owner_profile_url = get_first(microformats, 'ownerProfileUrl')
live_content = get_first(video_details, 'isLiveContent') live_content = get_first(video_details, 'isLiveContent')