Solution to Onboarding Project #1

csv_to_json.py

#! /usr/bin/env python
import csv
import fire
import json
import pyjq

"""CSV TO JSON CLI

Usage:
    python csv_to_json.py convert --csv_file="file.csv" --json_file="file.json"
    python csv_to_json.py convert --csv_file="file.csv" --json_file="file.json" --verbose=False --json_indent_spaces=2
    python csv_to_json.py convert --csv_file="file.csv" --json_file="file.json" --jq_fn="first" --jq_str='.data[] | {"param_name": ."First Name"}'

Options:
    <csv_file>  (required) Location of existing input CSV file you want to convert to JSON.
    <json_file> (required) Location of new JSON file (existing files will be overwritten!).
    <verbose> (default: True) True if you want to print to stdout.
    <json_indent_spaces> (default: 4) Number of spaces to indent json output (ignored if json_minified=True).
    <json_minified> (default: False) True if you want to minify whitespace in the json output.

    The following *optional* arguments are for running JQ queries against the new JSON object:
    <jq_fn> JQ query method to run, must be one of: 'all' | 'first' | 'one' | 'compile'.
    <jq_str> JQ query string, to be passed to the JQ query method.
    <jq_default> JQ default value if nothing is found from the JQ query.
"""

JQ_FNS = ('all', 'first', 'one', 'compile')


def convert(csv_file, json_file, verbose=True, json_indent_spaces=4,
            json_minified=False, jq_fn=None, jq_str=None, jq_default=None):

    if verbose:
        print('Reading file {}...'.format(csv_file))

    with open(csv_file, 'r') as thefile:
        json_data = list(csv.DictReader(thefile))

    if jq_fn:
        json_data = apply_jq_query(jq_fn, jq_str, jq_default, json_data)

    if verbose:
        print('Writing to file {}:'.format(json_file))
        print(json_data)

    with open(json_file, 'w') as output_file:
        kwargs_dict = {
            'obj': json_data,
        }

        if not json_minified:
            kwargs_dict['indent'] = json_indent_spaces

        json_output = json.dumps(**kwargs_dict)

        output_file.write(json_output)


def apply_jq_query(jq_fn, jq_str, jq_default, data):
    if not jq_str or jq_fn not in JQ_FNS:
        raise Exception('Invalid JQ arguments')

    pyjq_method = getattr(pyjq, jq_fn)
    data = pyjq_method(jq_str, {'data': data}, jq_default)

    return data


if __name__ == '__main__':
    fire.Fire()