recode_arabic-win_to_utf8.py

#!/usr/bin/python
# -*- coding: utf-8 -*-
# recode a file from arabic windows(windows-1256) to utf8
import sys, os
def winArabicToUtf8(s, ar2fa=True):
    u = s.decode('windows-1256')
    if ar2fa:
        for item in [ 
            (u'ي',u'ی'),
            (u'ك',u'ک'),
            (u'ۀ',u'هٔ'),
            (u' ِ', u'ِ'),
        ]:
            u = u.replace(item[0], item[1])
    return u.encode('utf8')

s = file(sys.argv[1]).read()
ws = winArabicToUtf8(s)
newName = sys.argv[1]
if newName[-4:]=='.txt':
  newName = newName[:-4]
newName += '.utf8.txt'
file(newName, 'w').write(ws)