Timed Text Captions to SRT Subtitles converter script

tt2srt.py

#!/usr/bin/env python
# Usage: python tt2srt.py source.xml output.srt
# FROM: https://gist.github.com/adammw/915259

from xml.dom.minidom import parse
import sys

def fixTime(time):
    parts = time.split(':')
    frames = int(parts[3])    # 00..24
    millisec = frames*40
    time = ":".join(parts[0:3]) + ',' + '%03d' % millisec
    return time

i=1
dom = parse(sys.argv[1])
out = open(sys.argv[2], 'w')
styling = dom.getElementsByTagName("styling")[0]
styles = {}
for style in styling.childNodes:
    if style.nodeName == 'style':
        if style.attributes['tts:color']:
            styles[style.attributes['id'].value] = style.attributes['tts:color'].value
body = dom.getElementsByTagName("body")[0]
paras = body.getElementsByTagName("p")
for para in paras:
    out.write(str(i) + "\n")
    out.write(fixTime(para.attributes['begin'].value) + ' --> ' + fixTime(para.attributes['end'].value) + "\n")
    for child in para.childNodes:
        if child.nodeName == 'span':
            fontOpen = False
            if child.attributes['style']:
                styleid = child.attributes['style'].value
                if styleid in styles:
                    out.write('<font color="' + styles[styleid] + '">')
                    fontOpen = True
            for child2 in child.childNodes:
                if child2.nodeName == 'br':
                    out.write("\n")
                elif child2.nodeName == '#text':
                    rawText = unicode(child2.data).encode('utf-8').strip()
                    if len(rawText) > 0:
                        out.write(rawText)
            if fontOpen:
                out.write('</font>')
        elif child.nodeName == 'br':
            out.write("\n")
    out.write("\n\n")
    i += 1