Indonesian Regency code comparison

regency_code_comparison.py

"""
Indonesia regency code comparison.

Provinces differ for two codes, 90s, covering Papua and Papua Barat.

More information provinces at:
https://gitlab.com/ciptamedia/ciptamedia-dev.gitlab.io/issues/104#note_59625585

This script is to compare the regencies to determine how much they overlapped.
"""

import csv
import os


def load_csv(filename, fieldnames=None):
    with open(filename) as f:
        reader = csv.DictReader(f, fieldnames=fieldnames)
        return list(reader)


def get_wdq_regencies():
    # This file is export of http://tinyurl.com/y9nacntd
    filename = os.path.join('..', 'wdq-regencies.csv')
    data = list(load_csv(filename))
    return data


def get_bps_regencies():
    # This file is from
    # https://github.com/edwardsamuel/Wilayah-Administratif-Indonesia/blob/master/csv/regencies.csv
    filename = os.path.join('..', 'bps-regencies.csv')
    data = list(load_csv(filename,
                         fieldnames=['code', 'province_code', 'name']))
    return data


def check_regency_lists():
    wdq_data = get_wdq_regencies()

    bps_data = get_bps_regencies()

    wdq_dict = {}
    for row in wdq_data:
        wdq_dict[row['rawcode']] = row

    ok = not_ok = 0
    for row in bps_data:
        code = row['code']
        if code not in wdq_dict:
            print('Code %s missing from Wikidata' % code)
            continue

        name = row['name'].lower()
        name = name.replace('kabupaten', '')
        name = name.strip()
        name = name.replace('-', '')
        name = name.replace(' ', '')

        wdq_row = wdq_dict[code]
        wd_name = wdq_row['itemLabel'].lower()
        wd_name = wd_name.replace('-', '')
        wd_name = wd_name.replace(' ', '')

        if wd_name == name:
            ok = ok + 1
            print('%s ok' % wdq_row['itemLabel'])
        else:
            not_ok = not_ok + 1
            print('%s != %s' % (wdq_row['itemLabel'], row['name']))

    print('Matches: %d' % ok)
    print('Not ok: %d' % not_ok)


def main(argv=None):
    check_regency_lists()


if __name__ == '__main__':
    main()