convert text file encodings

gistfile1.txt

# convert text file encoding
import codecs
import os

from shutil import copyfile

encoding_china = 'gb18030'
encoding_gbk = 'gbk'
encoding_source = encoding_china
encoding_target = 'utf-8'


def con(source, target):
    blocksize = 1048576  # or some other, desired size in bytes
    try:
        with codecs.open(source, "r", encoding_source) as sourceFile:
            with codecs.open(target, "w", encoding_target) as targetFile:
                while True:
                    contents = sourceFile.read(blocksize)
                    if not contents:
                        break
                    targetFile.write(contents)
    except UnicodeDecodeError:
        copyfile(source, target)
        pass


def test():
    source = '/Users/tan/tan/dev/python/dian/source'
    conn(source)
    check(source)
    check(source.replace('source', 'target'))


def check(path):
    files = []
    count_dir(files, path)
    print(len(files))


def conn(path):
    if not os.path.isfile(path):
        target = path.replace('source', 'target')
        if not os.path.exists(target):
            os.mkdir(target)
        for line in os.listdir(path):
            conn(path + '/' + line)
    else:
        con(path, path.replace('source', 'target'))
        print(path)


def count_dir(files, file):
    if not os.path.isfile(file):
        for line in os.listdir(file):
            count_dir(files, file + '/' + line)
    else:
        files.append(file)


test()