convert text file to specific encoding

convertto.py

#!/usr/bin/env python
#-*-encoding:utf-8-*-

import chardet
import argparse
import os
parser = argparse.ArgumentParser()
parser.add_argument("dir", help="directory where all files will be converted")
parser.add_argument("-r", "--recursive", action="store_true", help="do convertion recursively")
parser.add_argument("-e", "--extension", help="only convert files with specific extension")
parser.add_argument("-E", "--encoding", help="convert files to specific encoding, supporting utf-8, gbk and unicode, default is utf-8")
args = parser.parse_args()
if args.encoding is None or args.encoding.lower() not in ['utf-8','gbk','unicode','gb2312']:
    args.encoding = 'utf-8'
## must be absolute path
def getFiles(dir, ext="txt"):
    if not (os.path.exists(dir) or os.path.isdir(dir)):
        return
    files = []
    dirs = []
    for file in os.listdir(dir):
        abspath = '/'.join([dir,file])
        if os.path.isdir(abspath):
            dirs.append(abspath)
        elif ext=="" or abspath[-len(ext):]==ext:
            files.append(abspath)
    return (dirs,files)
    
def convert(filepath, encoding):
    if os.path.isfile(filepath):
        f = open(filepath,"r")
        s = f.read()
        f.close()
        f = open(filepath+"."+encoding,"w")
        try:
            f.write(s.decode(chardet.detect(s).get('encoding','GBK')).encode(encoding))
        except:
            print "convert error ",filepath
        f.close()

def convertAll(files, encoding):
    for file in files:
        convert(file, encoding)
        
def handle(arg):
    (dirs,files) = getFiles(arg.dir, arg.extension)
    convertAll(files, arg.encoding)
    if arg.recursive:
        while dirs:
            newdirs = []
            for dir in dirs:
                (ddir,files) = getFiles(dir, ext=arg.extension)
                convertAll(files, arg.encoding)
                newdirs.extend(ddir)
            dirs = newdirs

if __name__=="__main__":
    handle(args)