[ie/tiktok] Fix audio-only format extraction (#7712)

Closes #6608
Authored by: bashonly
This commit is contained in:
bashonly 2023-07-29 11:14:16 -05:00 committed by GitHub
parent 127a224606
commit b09bd0c196
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -205,15 +205,16 @@ def parse_url_key(url_key):
known_resolutions = {} known_resolutions = {}
def mp3_meta(url): def audio_meta(url):
ext = determine_ext(url, default_ext='m4a')
return { return {
'format_note': 'Music track', 'format_note': 'Music track',
'ext': 'mp3', 'ext': ext,
'acodec': 'mp3', 'acodec': 'aac' if ext == 'm4a' else ext,
'vcodec': 'none', 'vcodec': 'none',
'width': None, 'width': None,
'height': None, 'height': None,
} if determine_ext(url) == 'mp3' else {} } if ext == 'mp3' or '-music-' in url else {}
def extract_addr(addr, add_meta={}): def extract_addr(addr, add_meta={}):
parsed_meta, res = parse_url_key(addr.get('url_key', '')) parsed_meta, res = parse_url_key(addr.get('url_key', ''))
@ -231,7 +232,7 @@ def extract_addr(addr, add_meta={}):
**add_meta, **parsed_meta, **add_meta, **parsed_meta,
'format_note': join_nonempty( 'format_note': join_nonempty(
add_meta.get('format_note'), '(API)' if 'aweme/v1' in url else None, delim=' '), add_meta.get('format_note'), '(API)' if 'aweme/v1' in url else None, delim=' '),
**mp3_meta(url), **audio_meta(url),
} for url in addr.get('url_list') or []] } for url in addr.get('url_list') or []]
# Hack: Add direct video links first to prioritize them when removing duplicate formats # Hack: Add direct video links first to prioritize them when removing duplicate formats
@ -527,6 +528,7 @@ class TikTokIE(TikTokBaseIE):
'repost_count': int, 'repost_count': int,
'comment_count': int, 'comment_count': int,
}, },
'params': {'skip_download': True}, # XXX: unable to download video data: HTTP Error 403: Forbidden
}, { }, {
# Video without title and description # Video without title and description
'url': 'https://www.tiktok.com/@pokemonlife22/video/7059698374567611694', 'url': 'https://www.tiktok.com/@pokemonlife22/video/7059698374567611694',
@ -600,7 +602,7 @@ class TikTokIE(TikTokBaseIE):
}, { }, {
# only available via web # only available via web
'url': 'https://www.tiktok.com/@moxypatch/video/7206382937372134662', 'url': 'https://www.tiktok.com/@moxypatch/video/7206382937372134662',
'md5': '8d8c0be14127020cd9f5def4a2e6b411', 'md5': '6aba7fad816e8709ff2c149679ace165',
'info_dict': { 'info_dict': {
'id': '7206382937372134662', 'id': '7206382937372134662',
'ext': 'mp4', 'ext': 'mp4',
@ -637,8 +639,8 @@ class TikTokIE(TikTokBaseIE):
'uploader_id': '86328792343818240', 'uploader_id': '86328792343818240',
'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAA-0bQT0CqebTRr6I4IkYvMDMKSRSJHLNPBo5HrSklJwyA2psXLSZG5FP-LMNpHnJd', 'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAA-0bQT0CqebTRr6I4IkYvMDMKSRSJHLNPBo5HrSklJwyA2psXLSZG5FP-LMNpHnJd',
'channel_id': 'MS4wLjABAAAA-0bQT0CqebTRr6I4IkYvMDMKSRSJHLNPBo5HrSklJwyA2psXLSZG5FP-LMNpHnJd', 'channel_id': 'MS4wLjABAAAA-0bQT0CqebTRr6I4IkYvMDMKSRSJHLNPBo5HrSklJwyA2psXLSZG5FP-LMNpHnJd',
'creator': 't8', 'creator': 'tate mcrae',
'artist': 't8', 'artist': 'tate mcrae',
'track': 'original sound', 'track': 'original sound',
'upload_date': '20220609', 'upload_date': '20220609',
'timestamp': 1654805899, 'timestamp': 1654805899,
@ -650,6 +652,31 @@ class TikTokIE(TikTokBaseIE):
'thumbnail': r're:^https://.+\.webp', 'thumbnail': r're:^https://.+\.webp',
}, },
'params': {'format': 'bytevc1_1080p_808907-0'}, 'params': {'format': 'bytevc1_1080p_808907-0'},
}, {
# Slideshow, audio-only m4a format
'url': 'https://www.tiktok.com/@hara_yoimiya/video/7253412088251534594',
'md5': '2ff8fe0174db2dbf49c597a7bef4e47d',
'info_dict': {
'id': '7253412088251534594',
'ext': 'm4a',
'title': 'я ред флаг простите #переписка #щитпост #тревожныйтиппривязанности #рекомендации ',
'description': 'я ред флаг простите #переписка #щитпост #тревожныйтиппривязанности #рекомендации ',
'uploader': 'hara_yoimiya',
'uploader_id': '6582536342634676230',
'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAAIAlDxriiPWLE-p8p1R_0Bx8qWKfi-7zwmGhzU8Mv25W8sNxjfIKrol31qTczzuLB',
'channel_id': 'MS4wLjABAAAAIAlDxriiPWLE-p8p1R_0Bx8qWKfi-7zwmGhzU8Mv25W8sNxjfIKrol31qTczzuLB',
'creator': 'лампочка',
'artist': 'Øneheart',
'album': 'watching the stars',
'track': 'watching the stars',
'upload_date': '20230708',
'timestamp': 1688816612,
'view_count': int,
'like_count': int,
'comment_count': int,
'repost_count': int,
'thumbnail': r're:^https://.+\.webp',
},
}, { }, {
# Auto-captions available # Auto-captions available
'url': 'https://www.tiktok.com/@hankgreen1/video/7047596209028074758', 'url': 'https://www.tiktok.com/@hankgreen1/video/7047596209028074758',