srtmod.py

#!/usr/bin/python3
import sys
import re

import argparse

rxnum = re.compile(r'^(\d+)$')
rxtime = re.compile(r'(\d+):(\d+):(\d+),(\d+) --> (\d+):(\d+):(\d+),(\d+)')
rxcap = re.compile(r'-?\([A-Z\s0-9-]+\)')

def main():
    p = argparse.ArgumentParser(description='')
    p.add_argument('source')
    p.add_argument('-d', '--delay', action='store_true', help='Delay each subtitle by its duration')
    p.add_argument('-c', '--removecap', action='store_true', help='Try to remove SDH captions')
    args = p.parse_args()

    with open(args.source, 'r', encoding='utf-8') as fp:
        subs = readsrt(fp)

    if args.removecap:
        subs = removecap(subs)

    if args.delay:
        subs = delaysubs(subs)

    writesrt(sys.stdout, subs)

def delaysubs(src):
    out = []
    for i, (begin, end, text) in enumerate(src):
        dur = min(end - begin, 3)
        begin = end
        end = begin + dur
        try:
            nb = src[i+1][1]
            if end > nb:
                end = nb
        except IndexError:
            pass
        out.append((begin, end, text))
    return out

def removecap(src):
    out = []
    for i, (begin, end, text) in enumerate(src):
        ntxt = '\n'.join(text)
        ntxt = rxcap.sub('', ntxt)
        if ntxt.strip():
            out.append((begin, end, ntxt.split('\n')))
    return out

def srttime(t):
    secs = int(t)
    frac = t - secs
    mins = secs / 60
    secs %= 60
    hrs = mins / 60
    mins %= 60
    return '%02d:%02d:%02d,%03d' % (hrs, mins, secs, int(frac * 1000))

def readsrt(inf):
    out = []
    cursub = None
    for l in inf:
        l = l.strip()
        if rxnum.match(l):
            cursub = None
            continue

        m = rxtime.match(l)
        if m:
            bh, bm, bs, bf, eh, em, es, ef = m.groups()
            begin = (int(bh) * 3600 + int(bm) * 60 + float(bs + '.' + bf))
            end = (int(eh) * 3600 + int(em) * 60 + float(es + '.' + ef))
            cursub = []
            out.append((begin, end, cursub))
            continue

        if cursub is not None:
            cursub.append(l)
    return out

def writesrt(outf, subs):
    for i, (begin, end, text) in enumerate(subs):
        outf.write('%d\n' % (i + 1))
        outf.write('%s --> %s\n' % (srttime(begin), srttime(end)))
        for l in text:
            outf.write('%s\n' % l)

if __name__ == '__main__':
    main()