strip_cc.py

from __future__ import annotations

import argparse
import os
import subprocess
from pathlib import Path


def strip_cc(file: Path | os.PathLike[str]) -> None:
    """
    Strip EIA-608 closed captions from a H.264 MP4 file
    while preserving x264 encoding settings.
    """
    file = Path(file)

    with open(file, "rb") as fd:
        data = fd.read(60000)

    needs_strip = False
    encoding_settings = None
    if b"1GA94" in data:
        needs_strip = True
        if b"x264" in data:
            i = data.index(b"x264")
            encoding_settings = data[i : i + data[i:].index(b"\x00")]

    if not needs_strip:
        return

    bitstream_filters = ["filter_units=remove_types=6"]
    if encoding_settings:
        bitstream_filters.append(
            "h264_metadata=sei_user_data={magic}+{settings}".format(
                magic="dc45e9bde6d948b7962cd820d923eeef",
                settings=encoding_settings.replace(b":", rb"\\:").replace(b",", rb"\,").decode(),
            )
        )

    fixed_file = file.with_suffix(".fixed.mp4")

    subprocess.run(
        [
            "ffmpeg",
            "-y",
            "-nostdin",
            "-hide_banner",
            "-loglevel",
            "error",
            "-stats",
            "-i",
            file,
            "-c",
            "copy",
            "-bsf:v",
            ",".join(bitstream_filters),
            fixed_file,
        ],
        check=True,
    )

    file.unlink()
    fixed_file.rename(file)


def main() -> None:
    parser = argparse.ArgumentParser()
    parser.add_argument("input", type=Path, help="input file")
    args = parser.parse_args()

    strip_cc(args.input)


if __name__ == "__main__":
    main()