#!/usr/bin/env python
"""
charguess.py - Guess Character Encoding
Author: Sean B. Palmer, inamidst.com
Guesses between: ascii, iso-8859-1, cp1252, and utf-8
"""

import sys, os.path, re

r_hibyte = re.compile(r'[\x80-\xff]')
r_winbyte = re.compile(r'[\x80-\xbf]')

def charguess(fn): 
   ascii = True
   utf8 = True
   cp1252 = False
   iso88591 = False

   if not os.path.exists(fn): 
      print >> sys.stderr, "Error! Not found: %s" % fn
      return 

   if not os.path.isfile(fn): 
      if os.path.isdir(fn): 
         print >> sys.stderr, "Error! Directory: %s" % fn
      else: print >> sys.stderr, "Error! Not a regular file: %s" % fn
      return

   f = open(fn, 'rb')
   for line in f: 
      # If there's an r_hibyte, it's not ascii
      # If it doesn't decode as utf-8, it's not utf-8
      # If there's an r_winbyte, it's cp1252
      # Otherwise, iso-8859-1

      if r_hibyte.search(line): 
         ascii = False

         try: line.decode('utf-8')
         except UnicodeDecodeError: 
            utf8 = False
            if r_winbyte.search(line): 
               cp1252 = True
               break
            else: iso88591 = True
   f.close()

   if ascii: return 'ascii'
   elif utf8: return 'utf-8'
   elif cp1252: return 'cp1252'
   else: return 'iso-8859-1'

def main(): 
   for fn in sys.argv[1:]: 
      result = charguess(fn)
      if result: print result + ':', fn

if __name__ == '__main__': 
   main()