subtitle_Mediathek_SRT.py

#!/usr/bin/env python3
# Converts German Mediathek (ARD, ZDF, etc.) XML Subtitles into SRT Subtitle format.
# 2018-02-09
# Version 0.0.1

import re, sys
from bs4 import BeautifulSoup

if len(sys.argv) < 2:
    print("Usage: subtitle_Mediathek_SRT.py <Subtitlefile>.xml")
    sys.exit(2)

s = open(sys.argv[1]).read()
s = s.replace('<tt:br/>', '\n')
soup = BeautifulSoup(s, "lxml")
srt = ''

for p_tag in soup.find_all('tt:p', limit=100000):
	# get ID
	srt += str(int(p_tag.get('xml:id')[3:])+1) + '\n'
	# find begin/endtime
	srt += "%s --> %s" % (p_tag.get('begin'), p_tag.get('end'))
	# replace double newline with single newline.
	srt += p_tag.text.replace('\n\n','\n')
	srt += '\n\n'

f = open(sys.argv[1][:-4]+".srt", 'w')
f.write(srt)
f.close()