[ie] Resolve language to ISO639-2 for ISM formats (#11359)

Closes #11356
Authored by: bashonly
This commit is contained in:
bashonly 2024-10-26 18:02:21 +00:00 committed by GitHub
parent 6abef74232
commit 21cdcf03a2
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -47,6 +47,7 @@
FormatSorter, FormatSorter,
GeoRestrictedError, GeoRestrictedError,
GeoUtils, GeoUtils,
ISO639Utils,
LenientJSONDecoder, LenientJSONDecoder,
Popen, Popen,
RegexNotFoundError, RegexNotFoundError,
@ -3071,7 +3072,11 @@ def _parse_ism_formats_and_subtitles(self, ism_doc, ism_url, ism_id=None):
url_pattern = stream.attrib['Url'] url_pattern = stream.attrib['Url']
stream_timescale = int_or_none(stream.get('TimeScale')) or timescale stream_timescale = int_or_none(stream.get('TimeScale')) or timescale
stream_name = stream.get('Name') stream_name = stream.get('Name')
stream_language = stream.get('Language', 'und') # IsmFD expects ISO 639 Set 2 language codes (3-character length)
# See: https://github.com/yt-dlp/yt-dlp/issues/11356
stream_language = stream.get('Language') or 'und'
if len(stream_language) != 3:
stream_language = ISO639Utils.short2long(stream_language) or 'und'
for track in stream.findall('QualityLevel'): for track in stream.findall('QualityLevel'):
KNOWN_TAGS = {'255': 'AACL', '65534': 'EC-3'} KNOWN_TAGS = {'255': 'AACL', '65534': 'EC-3'}
fourcc = track.get('FourCC') or KNOWN_TAGS.get(track.get('AudioTag')) fourcc = track.get('FourCC') or KNOWN_TAGS.get(track.get('AudioTag'))