JSON obfuscator

json-obfuscator.py

#!/usr/bin/env python3
'''
Uses the IJSON library to parse JSON from stdin, replacing strings with salted
hashes that are internally consistent (i.e. "cat" is always replaced with the
same hash).  Numbers are not modified.  The output is finally writtent to
stdout.
'''

import sys
import logging
from json import dump
from hashlib import sha1
import ijson
from random import randbytes

logging.basicConfig(level=logging.INFO)
salt = randbytes(10)

logging.info(f'Hashing JSON strings with SHA-1 and a salt of {salt.hex()}')


def hashing_parser(basic_parser):
    for event, value in basic_parser:
        if type(value) is str:
            value = sha1(salt + value.encode('utf-8')).hexdigest()

        yield event, value


basic_parser = ijson.basic_parse(sys.stdin)
hp = hashing_parser(basic_parser)
item_stream = ijson.items(ijson.parse(hp), '')

for item in item_stream:
    dump(item, sys.stdout, indent=2)

logging.info('Finished writing obfuscated JSON to stdout.')