Key-frame based FFmpeg Cut

cut.py

# WARNING: NOT QUITE WORKING...

import sys
import subprocess
import bisect
from pathlib import Path

def run_command(command_list):
    success = True
    stdout = b""
    print("executing: ", " ".join(command_list))
    try:
        stdout = subprocess.check_output(command_list)
    except subprocess.CalledProcessError:
        success = False
    return stdout.decode(), success

# https://stackoverflow.com/questions/63548027/cut-a-video-in-between-key-frames-without-re-encoding-the-full-video-using-ffpme
def get_keyframe_list(input_file):
    probe_command = ["ffprobe", "-v", "error", "-select_streams", "v:0", "-skip_frame", "nokey", "-show_entries", "frame=pkt_pts_time", "-of", "csv=p=0", input_file]
    output, _ = run_command(probe_command)
    keyframes = [float(line) for line in output.split("\n") if line]
    return keyframes

def find_closest_range(keyframe_list, start_sec, end_sec):
    if len(keyframe_list) < 2:
        raise Exception("too few keyframe...")
    # some binary search magic that seems to work?
    start_index = bisect.bisect_left(keyframe_list, start_sec)
    end_index = bisect.bisect(keyframe_list, end_sec)-1
    return keyframe_list[start_index], keyframe_list[end_index]

def time_to_sec(time_str):
    ms = 0
    if ":" not in time_str:
        raise Exception("invalid time format")
    if "." in time_str:
        time_str, ms = time_str.split(".")
        ms = float("0." + ms)
    
    times = time_str.split(":")
    hour, min, sec = 0, int(times[-2]), int(times[-1])
    if len(times) == 3:
        hour = int(times[0])
    precise_second = hour*60*60 + min*60 + sec + ms
    return precise_second

def ffmpeg_encode_cut(in_file, start, end, out_file):
    # EDIT ENCODE PARAMETER HERE!
    cut_command = ["ffmpeg", "-i", in_file,"-ss", str(start), "-to", str(end), "-c:v", "libx264", "-crf", "18", "-y", out_file]
    # print(" ".join(cut_command))
    output, success = run_command(cut_command)
    return success

def ffmpeg_stream_copy_cut(in_file, start, end, out_file):
    cut_command = ["ffmpeg", "-ss", str(start), "-i", in_file, "-to", str(end), "-c:v", "copy", "-c:a", "copy", "-y", out_file]
    output, success = run_command(cut_command)
    return success

def ffmpeg_concat(segments, out_file):
    segments = [segment for segment in segments if Path(segment).exists()]
    with open("concat.txt", "w") as f:
        for segment in segments:
            f.write("file '{}'\n".format(segment))
    concat_command = ["ffmpeg", "-f", "concat", "-safe", "0", "-i", "concat.txt", "-c", "copy", "-y", out_file]
    output, success = run_command(concat_command)
    return success


def do_cut(file, start, end):
    keyframes = get_keyframe_list(file)
    start_sec, end_sec = time_to_sec(start), time_to_sec(end)
    start_frame, end_frame = find_closest_range(keyframes, start_sec, end_sec)
    print(f"{start_sec=} {start_frame=} {end_frame=} {end_sec=}")
    ffmpeg_encode_cut(file, start_sec, start_frame, "before.mp4")
    ffmpeg_stream_copy_cut(file, start_frame, end_frame, "inside.mp4")
    ffmpeg_encode_cut(file, end_frame, end_sec, "after.mp4")
    ffmpeg_concat(["before.mp4", "inside.mp4", "after.mp4"], "output.mp4")

def main():
    print("hello world")
    if len(sys.argv) < 4:
        print("usage: cut.py [file] [start] [end]")
        print("example: cut.py input.mp4 00:01:02.02 00:01:04.88")
        print("time format: (hh:)mm:ss(.ms)")
        sys.exit(1)
    do_cut(sys.argv[1], sys.argv[2], sys.argv[3])

if __name__ == "__main__":
    main()