Pretty-print a facebook message.json, fixing up broken encoding

message_json_parser.py

import sys
import json
from datetime import datetime


def fixup_str(text):
    return text.encode('latin1').decode('utf8')


def fixup_list(l):
    return [fixup(e) for e in l]


def fixup_dict(dct):
    return {fixup_str(k): fixup(v) for k, v in dct.items()}


def fixup(e):
    if isinstance(e, dict):
        return fixup_dict(e)
    if isinstance(e, list):
        return fixup_list(e)
    if isinstance(e, str):
        return fixup_str(e)
    return e


def filter_messages(messages):
    for message in messages:
        if message['type'] == 'Generic':
            yield message


def format_message(message):
    sender = message['sender_name']
    timestamp = datetime.fromtimestamp(message['timestamp_ms']/1000)
    content = message['content']
    return f'{timestamp}\t{sender}\t{content}'


def main(args):
    fname = 'message.json'
    if args:
        fname = args[0]

    with open(fname, 'r') as f:
        messages = fixup(json.loads(f.read()))['messages']

    for message in filter_messages(messages):
        print(format_message(message))


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

Hey. Fixed some issues in my fork:

• https://gist.github.com/glensc/3ef3590f8e3c4bed0550b99e40220c32

See details here:

• https://gist.github.com/glensc/3ef3590f8e3c4bed0550b99e40220c32/revisions