1
0
mirror of https://github.com/yt-dlp/yt-dlp synced 2025-07-10 07:44:01 -05:00

[extractor/wistia] Improve extension detection (#5415)

Closes #5053
Authored by: bashonly, Grub4k, pukkandan
This commit is contained in:
bashonly
2022-12-29 16:32:54 +00:00
committed by GitHub
parent 53006b35ea
commit 2647c933b8
2 changed files with 104 additions and 59 deletions

View File

@ -3480,67 +3480,93 @@ def error_to_str(err):
return f'{type(err).__name__}: {err}'
def mimetype2ext(mt):
if mt is None:
def mimetype2ext(mt, default=NO_DEFAULT):
if not isinstance(mt, str):
if default is not NO_DEFAULT:
return default
return None
mt, _, params = mt.partition(';')
mt = mt.strip()
FULL_MAP = {
'audio/mp4': 'm4a',
# Per RFC 3003, audio/mpeg can be .mp1, .mp2 or .mp3. Here use .mp3 as
# it's the most popular one
'audio/mpeg': 'mp3',
'audio/x-wav': 'wav',
'audio/wav': 'wav',
'audio/wave': 'wav',
}
ext = FULL_MAP.get(mt)
if ext is not None:
return ext
SUBTYPE_MAP = {
MAP = {
# video
'3gpp': '3gp',
'smptett+xml': 'tt',
'ttaf+xml': 'dfxp',
'ttml+xml': 'ttml',
'x-flv': 'flv',
'x-mp4-fragmented': 'mp4',
'x-ms-sami': 'sami',
'x-ms-wmv': 'wmv',
'mp2t': 'ts',
'mp4': 'mp4',
'mpeg': 'mpeg',
'mpegurl': 'm3u8',
'x-mpegurl': 'm3u8',
'vnd.apple.mpegurl': 'm3u8',
'quicktime': 'mov',
'webm': 'webm',
'vp9': 'vp9',
'x-flv': 'flv',
'x-m4v': 'm4v',
'x-matroska': 'mkv',
'x-mng': 'mng',
'x-mp4-fragmented': 'mp4',
'x-ms-asf': 'asf',
'x-ms-wmv': 'wmv',
'x-msvideo': 'avi',
# application (streaming playlists)
'dash+xml': 'mpd',
'f4m+xml': 'f4m',
'hds+xml': 'f4m',
'vnd.apple.mpegurl': 'm3u8',
'vnd.ms-sstr+xml': 'ism',
'quicktime': 'mov',
'mp2t': 'ts',
'x-mpegurl': 'm3u8',
# audio
'audio/mp4': 'm4a',
# Per RFC 3003, audio/mpeg can be .mp1, .mp2 or .mp3.
# Using .mp3 as it's the most popular one
'audio/mpeg': 'mp3',
'audio/webm': 'weba',
'audio/x-matroska': 'mka',
'audio/x-mpegurl': 'm3u',
'midi': 'mid',
'ogg': 'ogg',
'wav': 'wav',
'wave': 'wav',
'x-aac': 'aac',
'x-flac': 'flac',
'x-m4a': 'm4a',
'x-realaudio': 'ra',
'x-wav': 'wav',
'filmstrip+json': 'fs',
# image
'avif': 'avif',
'bmp': 'bmp',
'gif': 'gif',
'jpeg': 'jpg',
'png': 'png',
'svg+xml': 'svg',
}
'tiff': 'tif',
'vnd.wap.wbmp': 'wbmp',
'webp': 'webp',
'x-icon': 'ico',
'x-jng': 'jng',
'x-ms-bmp': 'bmp',
_, _, subtype = mt.rpartition('/')
ext = SUBTYPE_MAP.get(subtype.lower())
if ext is not None:
return ext
# caption
'filmstrip+json': 'fs',
'smptett+xml': 'tt',
'ttaf+xml': 'dfxp',
'ttml+xml': 'ttml',
'x-ms-sami': 'sami',
SUFFIX_MAP = {
# misc
'gzip': 'gz',
'json': 'json',
'xml': 'xml',
'zip': 'zip',
'gzip': 'gz',
}
_, _, suffix = subtype.partition('+')
ext = SUFFIX_MAP.get(suffix)
if ext is not None:
return ext
mimetype = mt.partition(';')[0].strip().lower()
_, _, subtype = mimetype.rpartition('/')
ext = traverse_obj(MAP, mimetype, subtype, subtype.rsplit('+')[-1])
if ext:
return ext
elif default is not NO_DEFAULT:
return default
return subtype.replace('+', '.')
@ -3634,7 +3660,7 @@ def get_compatible_ext(*, vcodecs, acodecs, vexts, aexts, preferences=None):
return 'mkv' if allow_mkv else preferences[-1]
def urlhandle_detect_ext(url_handle):
def urlhandle_detect_ext(url_handle, default=NO_DEFAULT):
getheader = url_handle.headers.get
cd = getheader('Content-Disposition')
@ -3645,7 +3671,13 @@ def urlhandle_detect_ext(url_handle):
if e:
return e
return mimetype2ext(getheader('Content-Type'))
meta_ext = getheader('x-amz-meta-name')
if meta_ext:
e = meta_ext.rpartition('.')[2]
if e:
return e
return mimetype2ext(getheader('Content-Type'), default=default)
def encode_data_uri(data, mime_type):