log anonymizer

#!/usr/bin/env python3

"""
Log Anonymizer reads a logfile, where the first field must be a valid IPv4
or IPv6 address. It anonymizes 16 bits with IPv4 addresses and
32 bits with IPv6 addresses.

If required the ip address could be hashed.
It is more anonymized if you use the random generated token.

By default the file is not overwritten.
If you use the command without -o, then the anonymized logfile
is printed to console.
"""


import os
import ipaddress
import hashlib
import secrets
import argparse
import sys
from pathlib import Path


def log_anonymizer(log: Path, hash_ip: bool=True, salt: bool=True):
    def anonymize(ip, ipv4_mask=16, ipv6_mask=32):
        ip = ipaddress.ip_address(ip)
        mask = 2 ** ip.max_prefixlen - 1
        if ip.version == 4:
            mask -= 2 ** ipv4_mask - 1
        elif ip.version == 6:
            mask -= 2 ** ipv6_mask - 1
        return ipaddress.ip_address(int(ip) & mask)
    if salt:
        salt_val = secrets.token_hex()
    with open(log) as fd:
        for idx, line in enumerate(fd, start=1):
            try:
                ip, rest = line.strip().split(maxsplit=1)
            except ValueError:
                print("Error with line no {}".format(idx), file=sys.stderr)
                yield line.strip()
                continue
            try:
                ip = str(anonymize(ip))
            except ValueError:
                print("Could not parse ip in line no {}".format(idx), file=sys.stderr)
                yield line.strip()
                continue
            if salt:
                ip += salt_val
            if salt or hash_ip:
                ip = hashlib.sha256(ip.encode()).hexdigest()
            yield " ".join((ip, rest))



if __name__ == "__main__":
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("logfile", type=Path, help="Logfile to anonymize")
    parser.add_argument("-H", action="store_true", help="hash ip address with sha256 after anonymize")
    parser.add_argument("-s", action="store_true", help="Add random salt before hashing")
    parser.add_argument("-o", action="store_true", help="Overwrite file.")
    args = parser.parse_args()
    log = log_anonymizer(args.logfile, args.H, args.s)
    if args.o:
        new_name = args.logfile.name + ".anon"
        new_file = args.logfile.with_name(new_name)
        with new_file.open("w") as fd:
            for line in log:
                fd.write(line + os.linesep)
        new_file.rename(args.logfile)
    else:
        for line in log:
            print(line)