ffmeta

Media metadata files read, written and converted through one model: FFmpeg's ffmetadata, Vorbis comments, and cue sheets.

It is pure -- text in, text out: no ffmpeg, no file opened. A file read becomes one model -- its tags, its streams' tags, its chapters, a cue sheet's disc; written in another format, what that format cannot hold is told as a note, never dropped in silence.

What crosses between them, field by field and with its reasons, is ffsuite's mappings skill.

Installing

pip install ffmeta

Python 3.13 or later; the standard library alone.

Using it

>>> import ffmeta
>>> text = ";FFMETADATA1\ntitle=Album\n\n[CHAPTER]\nTIMEBASE=1/1000\nSTART=0\nEND=60000\ntitle=One\n"
>>> meta = ffmeta.read(text, "ffmetadata", "album.ffmeta")
>>> converted = ffmeta.convert(meta, "ffmetadata", "vorbis")
>>> written = ffmeta.write(converted.meta, "vorbis")
>>> print(written.text, end="")
title=Album
CHAPTER000=00:00:00.000
CHAPTER000NAME=One
>>> converted.notes + written.notes
('chapter ends: Vorbis comments hold none (a chapter ends where the next begins): left out',)
>>> ffmeta.read("title=x\n", "ffmetadata", "bad.ffmeta")
Traceback (most recent call last):
  ...
ffmeta.Error: bad.ffmeta:1: not ffmetadata: the first line must begin ;FFMETADATA: title=x

What crosses into the target is converted; what does not is a note, from convert (a field the target lacks) or write (a structure it lacks). A text that cannot be read is refused: ffmeta.Error, its message the reason.

Licence

Part of ffsuite, beside ffman: its own package. Licensed under either of MIT or Apache-2.0, at your option.