Digest a JSON file

json_digest.py

import json
import sys

"""JSON Digest

Python script to digest a JSON structure.

Tasks:
    1. Get each key name and corresponding value
        a. Get the current value.
        b. Get the value's data type.
    2. Get the unique keys names.
        a. Ignore one or more keys.
        b. Set tree level digest limit
"""

# Settings
ignore_keys = ['email mappings']
last_level = 0

# Flags
first_round = True
type_instead = True
full_output = True
exit_on_first_round = False

# Data holders
level_keys = {}


def get_tabs(level):
    return "".join(["  " for x in range(0, level - 1)])


def digest_dictionary(src, level):
    for key in src.keys():
        digest_data(src[key], level, key)


def add_key_to_list(key, level):
    global level_keys

    try:
        if key not in level_keys[level]:
            level_keys[level].append(key)
            print(get_tabs(level) + key)
    except Exception:
        # Create a list per level if there isn't one
        level_keys[level] = [key]
        print(get_tabs(level) + key)


def digest_data(src, level=0, key=""):
    # Settings
    global last_level
    global ignore_keys
    # Flags
    global first_round
    global type_instead
    global full_output

    if key not in ignore_keys:
        # Ig level one is reached and
        # the first_round flag is off
        # exit
        if level == 5 and not first_round and exit_on_first_round:
            return
        else:
            level += 1

            # when the deepest level is reached
            # from the first round turn off the
            # first_round flag
            if (last_level + 1) != level:
                first_round = False

            if type(src) is dict:
                # Before going deeper keep save
                # temporaly the current_level
                last_level = level

                # tab = get_tabs(level)
                output = "{}: # [M] Level:{}".format(key, level)
                # print(tab + output)
                add_key_to_list(key, level)

                digest_dictionary(src, level)

            elif type(src) is list:
                last_level = level

                for item in src:
                    digest_data(item, level, key)

            else:
                if full_output:
                    output = "{}: {} # {} [A]"

                    # TODO: Add show_level flag
                    outlevel = "[L:{}]".format(level)

                    outtype = "{}".format(type(src))
                    # On type_instead set as value the data type
                    outvalue = src if not type_instead else outtype

                    # Soround any string with single quotes
                    if type(src) is str:
                        outvalue = "'{}'".format(outvalue)

                    print(get_tabs(level) + output.format(
                        key, outvalue, outlevel))
                else:
                    # print(get_tabs(level) + key)
                    add_key_to_list(key, level)


def main():
    try:
        with open(sys.argv[1]) as file:
            digest_data(json.loads(file.read()))
    except Exception:
        print("Usage: python json_digest.py <filename>.json")


if __name__ == '__main__':
    main()