Convert TTML to SRT

requirements.txt

bs4
lxml

ttml2srt.py

import re
import sys
from textwrap import dedent
from typing import List

import bs4


def extract_subs(xml_path: str) -> List[str]:
    with open(xml_path) as fin:
        soup = bs4.BeautifulSoup(fin.read(), 'lxml')
    subs = []
    for i, s in enumerate(soup.select('[begin]'), start=1):
        text, begin, end = s.text.strip(), s['begin'], s['end']
        text = re.sub(r'\n{2,}', '\n', text)
        srt_item = '\n'.join([str(i), f'{begin} --> {end}', text, '\n'])
        subs.append(srt_item)
    return subs


def main():
    if len(sys.argv) < 3:
        print(dedent('''\
        usage: python ttml2srt.py src.xml out.srt
        '''))
        return

    src_path, out_path = sys.argv[1:]
    subs = extract_subs(src_path)

    with open(out_path, 'w') as fout:
        for item in subs:
            fout.write(item)


if __name__ == '__main__':
    main()