codau_2_khongdau.py

#!/usr/bin/env python
import re
import sys

patterns = {
    '[àáảãạăắằẵặẳâầấậẫẩ]': 'a',
    '[đ]': 'd',
    '[èéẻẽẹêềếểễệ]': 'e',
    '[ìíỉĩị]': 'i',
    '[òóỏõọôồốổỗộơờớởỡợ]': 'o',
    '[ùúủũụưừứửữự]': 'u',
    '[ỳýỷỹỵ]': 'y'
}

def convert(text):
    """
    Convert from 'Tieng Viet co dau' thanh 'Tieng Viet khong dau'

    text: input string to be converted

    Return: string converted
    """
    output = text
    for regex, replace in patterns.items():
        output = re.sub(regex, replace, output)
        # deal with upper case
        output = re.sub(regex.upper(), replace.upper(), output)
    return output

if __name__ == '__main__':
    print(convert(sys.argv[1]))

mình có làm 1 phiên bản khác hoàn thiện hơn:

https://gist.github.com/phineas-pta/05cad38a29fea000ab6d9e13a6f7e623