A script to anonymize emails in a SQL dump

anonymize_email.py

#!/usr/bin/env python3
# coding: utf-8
import argparse
import re
from uuid import uuid4
from hashlib import sha256


SALT = str(uuid4()).encode()


def load_file(file_path):
    with open(file_path, encoding='utf-8') as f:
        return f.read()


def anonymize(email, domain):
    m = sha256()
    m.update(SALT)
    m.update(email.lower().encode())
    return '{}@{}'.format(m.hexdigest(), domain)


def replace_emails(text, fake_domain, allow):
    def replace(m):
        d = m.group(2).lower()

        if d in allow:
            out = m.group(0).lower()
        else:
            out = anonymize(m.group(0), fake_domain)

        print('replace {} => {}'.format(m.group(0), out))
        return out

    e = re.compile(r'([a-zA-Z_\.0-9+]{1,100})@([a-zA-Z_\.0-9\-]{1,100})')
    return e.sub(replace, text)


def write_out(text, file_path):
    with open(file_path, mode='w', encoding='utf-8') as f:
        f.write(text)


def main(args):
    text = load_file(args.input)
    allow = set(args.allow + [args.fake_domain])
    text = replace_emails(text, args.fake_domain, allow)
    write_out(text, args.output)


def parse():
    parser = argparse.ArgumentParser(description="Transform all emails in a"
                                     " text file into fake email address")

    parser.add_argument('-i', '--input', help='Input file', required=True)
    parser.add_argument('-o', '--output', help='Output file', required=True)
    parser.add_argument('-d', '--fake-domain', default='example.com',
                        help='All re-writter emails will be in this domain.')
    parser.add_argument('-a', '--allow', action='append',
                        help='Do not anonymize those domains.')

    return parser.parse_args()

if __name__ == '__main__':
    args = parse()
    main(args)