Offset timestamps in an .srt file

fix-transcripts.py

#!/usr/bin/env python
from pathlib import Path
import re

srt_timestamp_pattern = r"^(\d{2}):(\d{2}):(\d{2}),(\d{3}) --> (\d{2}):(\d{2}):(\d{2}),(\d{3})$"

def seconds_to_minsec(full_seconds:float) -> list[int, int, int, int]:
    hours = full_seconds // 3600
    full_seconds -= hours * 3600
    minutes = full_seconds // 60
    full_seconds -= minutes * 60
    seconds = int(full_seconds)
    ms = int((full_seconds - seconds) * 1000)
    return [int(x) for x in [hours, minutes, seconds, ms]]

def minsec_to_seconds(minsec:list[int, int, int, int]) -> float:
    return minsec[0] * 3600 + minsec[1] * 60 + minsec[2] + minsec[3] / 1000

def format_hms(hour, min, sec, ms) -> str:
    return f"{hour:0>2}:{min:0>2}:{sec:0>2},{ms:0>3}"

framerate = 25
out_points = [
    [1, 36, 22, 1000 * (7 / framerate)],
    [3, 7, 38, 1000 * (20 / framerate)],
    [4, 27, 54, 1000 * (22 / framerate)],
]

videos = list(Path().glob("*.mp4"))
videos.sort()
for i, video in enumerate(videos):
    if i == 0:
        continue

    duration_to_remove = minsec_to_seconds(out_points[i - 1])

    print(f"Removing {duration_to_remove} seconds from {video}")
    
    srt_filename = video.name.replace("mp4", "srt")
    with open(srt_filename, "r") as unfixed, open(srt_filename.replace(".srt", "_fixed.srt"), "w") as fixed:
        for line in unfixed:
            if "-->" not in line:
                fixed.write(line)
                continue
            
            matched_line = re.match(srt_timestamp_pattern, line)
            assert matched_line is not None, f"Matched line failed on this line: {line}"

            start = [int(matched_line.group(x)) for x in [1, 2, 3, 4]]
            end = [int(matched_line.group(x)) for x in [5, 6, 7, 8]]

            start = minsec_to_seconds(start)
            end = minsec_to_seconds(end)

            start -= duration_to_remove
            end -= duration_to_remove

            start = seconds_to_minsec(start)
            end = seconds_to_minsec(end)

            fixed.write(f"{format_hms(*start)} --> {format_hms(*end)}\n")