Parse Athena-formatted data into JSON

parse_athena_format.py

#!/usr/bin/env python3
# A script which will parse athena-formatted lines to be json. Assumes all
# fields and values are strings.

import json
import sys
import re


def parseathena(line):
    new = line
    if not new.strip(): return "{}"
    new = re.sub(r'(\w), ', r'\1", ', new)
    new = re.sub(r', (\w)', r', "\1', new)
    new = re.sub(r'{(\w)', r'{"\1', new)
    new = re.sub(r'(\w)}', r'\1"}', new)
    new = re.sub(r'}, (\w)', r'}, "\1', new)
    new = re.sub(r'\[(\w)', '["\\1', new)
    new = re.sub(r'(\w)]', r'\1"]', new)

    new = new.replace('=[', '":[')
    new = new.replace('={', '":{')
    new = new.replace('=', '":"')
    return new


def main():
    for line in sys.stdin:
        line = line.strip()
        if not line:
            continue
        print(json.dumps(json.loads(parseathena(line))))


if __name__ == '__main__': main()