Automatic SMI to SRT convert script

auto_convert.sh

#!/bin/sh
# Find SMI file recursively script by Taylor Starfield

FindRoot=/top/directory/to/convert

find $FindRoot -name '[^.]*.smi' | while read line; do
  if [ -f "${line%.smi}.srt" ]; then
    continue
  fi
  python convert.py "$line"
  chmod +x "${line%.smi}.srt"

done

convert.py

#!/usr/bin/env python
# -*- coding: UTF-8 -*-
'''
@package smi2srt
@brief this module is for convert .smi subtitle file into .srt subtitle 
	(Request by Alfred Chae)

Started : 2011/08/08
license: GPL

@version: 1.0.0
@author: Moonchang Chae <mcchae@gmail.com>


SMI have this format!
===================================================================================================

SRT have this format!
===================================================================================================
1
00:00:12,000 --> 00:00:15,123
This is the first subtitle

2
00:00:16,000 --> 00:00:18,000
Another subtitle demonstrating tags:
<b>bold</b>, <i>italic</i>, <u>underlined</u>
<font color="#ff0000">red text</font>

3
00:00:20,000 --> 00:00:22,000  X1:40 X2:600 Y1:20 Y2:50
Another subtitle demonstrating position.
'''
__author__ = "MoonChang Chae <mcchae@gmail.com>"
__date__ = "2011/08/08"
__version__ = "1.0.0"
__version_info__ = (1, 0, 0)
__license__ = "GCQVista's NDA"

###################################################################################################
import os
import sys
import re
import chardet #@UnresolvedImport

###################################################################################################
def usage(msg=None, exit_code=1):
	print_msg = """
usage %s smifile.smi [...]
	convert smi into srt subtitle file with same filename.
	By MoonChang Chae <mcchae@gmail.com>
""" % os.path.basename(sys.argv[0])
	if msg:
		print_msg += '%s\n' % msg
	print print_msg
	sys.exit(exit_code)

###################################################################################################
class smiItem(object):
	def __init__(self):
		self.start_ms = 0L
		self.start_ts = '00:00:00,000'
		self.end_ms = 0L
		self.end_ts = '00:00:00,000'
		self.contents = None
		self.linecount = 0
	@staticmethod
	def ms2ts(ms):
		hours = ms / 3600000L
		ms -= hours * 3600000L
		minutes = ms / 60000L
		ms -= minutes * 60000L
		seconds = ms / 1000L
		ms -= seconds * 1000L
		s = '%02d:%02d:%02d,%03d' % (hours, minutes, seconds, ms)
		return s
	def convertSrt(self):
		if self.linecount == 4:
			i=1 #@UnusedVariable
		# 1) convert timestamp
		self.start_ts = smiItem.ms2ts(self.start_ms)
		self.end_ts = smiItem.ms2ts(self.end_ms-10)
		# 2) remove new-line
		self.contents = re.sub(r'\s+', ' ', self.contents)
		# 3) remove web string like "&nbsp";
		self.contents = re.sub(r'&[a-z]{2,5};', '', self.contents)
		# 4) replace "<br>" with '\n';
		self.contents = re.sub(r'(<br>)+', '\n', self.contents, flags=re.IGNORECASE)
		# 5) find all tags
		fndx = self.contents.find('<')
		if fndx >= 0:
			contents = self.contents
			sb = self.contents[0:fndx]
			contents = contents[fndx:]
			while True:
				m = re.match(r'</?([a-z]+)[^>]*>([^<>]*)', contents, flags=re.IGNORECASE)
				if m == None: break
				contents = contents[m.end(2):]
				#if m.group(1).lower() in ['font', 'b', 'i', 'u']:
				if m.group(1).lower() in ['b', 'i', 'u']:
					sb += m.string[0:m.start(2)]
				sb += m.group(2)
			self.contents = sb
		self.contents = self.contents.strip()
		self.contents = self.contents.strip('\n')
	def __repr__(self):
		s = '%d:%d:<%s>:%d' % (self.start_ms, self.end_ms, self.contents, self.linecount)
		return s

###################################################################################################
def convertSMI(smi_file):
	if not os.path.exists(smi_file):
		sys.stderr.write('Cannot find smi file <%s>\n' % smi_file)
		return False
	rndx = smi_file.rfind('.')
	srt_file = '%s.srt' % smi_file[0:rndx]

	ifp = open(smi_file)
	smi_sgml = ifp.read()#.upper()
	ifp.close()
	chdt = chardet.detect(smi_sgml)
	if not chdt.get('encoding'):
		pass
	elif chdt['encoding'] != 'UTF-8':
		smi_sgml = unicode(smi_sgml, chdt['encoding'].lower()).encode('utf-8')

	# skip to first starting tag (skip first 0xff 0xfe ...)
	try:
		fndx = smi_sgml.lower().find('<sync')
	except Exception, e:
		print chdt
		raise e
	if fndx < 0:
		return False
	smi_sgml = smi_sgml[fndx:]
	lines = smi_sgml.split('\n')
	
	srt_list = []
	sync_cont = ''
	si = None
	last_si = None
	linecnt = 0
	for line in lines:
		linecnt += 1
		sndx = line.upper().find('<SYNC')
		if sndx >= 0:
			m = re.search(r'<sync\s+start\s*=\s*(\d+)>(.*)$', line, flags=re.IGNORECASE)
			if not m:
				raise Exception('Invalid format tag of <Sync start=nnnn> with "%s"' % line)
			sync_cont += line[0:sndx]
			last_si = si
			if last_si != None:
				last_si.end_ms = long(m.group(1))
				last_si.contents = sync_cont
				srt_list.append(last_si)
				last_si.linecount = linecnt
				#print '[%06d] %s' % (linecnt, last_si)
			sync_cont = m.group(2)
			si = smiItem()
			si.start_ms = long(m.group(1))
		else:
			sync_cont += line
			
	ofp = open(srt_file, 'w')
	ndx = 1
	for si in srt_list:
		si.convertSrt()
		if si.contents == None or len(si.contents) <= 0:
			continue
		#print si
		sistr = '%d\n%s --> %s\n%s\n\n' % (ndx, si.start_ts, si.end_ts, si.contents)
		#sistr = unicode(sistr, 'utf-8').encode('euc-kr')
		ofp.write(sistr)
		ndx += 1
	ofp.close()
	return True

###################################################################################################
def doConvert():
	if len(sys.argv) <= 1:
		usage()
	for smi_file in sys.argv[1:]:
		if convertSMI(smi_file):
			print "Converting <%s> OK!" % smi_file
		else:
			print "Converting <%s> Failture!" % smi_file
			
	
###################################################################################################
if __name__ == '__main__':
	doConvert()

Automatic convert SMI to SRT script

Just set "FindRoot" to top directory of to convert directory and regist auto_convert.sh to crontab
This script will convert all of SMI file to SRT at same directory with original SMI file where under "FindRoot" directory recursively
But original SMI file will still exist even though converted

Special thanks to Moonchang Chae. I changed some code to solve bug in his code.