sparebanken-sor-ynab/ynab.py

#!/bin/python3

import sys
import csv
import re
from pathlib import Path

whitelist_cards = ["7756"]

in_header = ['Utført dato', 'Bokført dato', 'Rentedato', 'Beskrivelse', 'Type', 'Undertype', 'Fra konto', 'Avsender', 'Til konto', 'Mottakernavn', 'Beløp inn', 'Beløp ut', 'Valuta', 'Status', 'Melding']
out_header = ['Date', 'Payee', 'Memo', 'Outflow', 'Inflow']


def usage():
    print('Usage: ynab.py <filename>')


def is_reserved(row):
    return row['Status'] == "Reservert"

def convert_memo(original):
    original = original.replace(" Kurs: 1.0000", "")
    words = original.split(" ")

    for i in range(20):
        if words[0] == "":
            # It's empty
            del words[0]
        elif m := re.match(r'\*(\d{4})', words[0]):
            # It's the last four digits of a card
            if m.groups()[0] in whitelist_cards:
                # It's an expected card, ignore it
                del words[0]
            else:
                # It's an unexpected card, move it to the end
                words.append(words.pop(0))
        elif m := re.match(r'\d{2}\.\d{2}', words[0]):
            # It's the date. Move it to the end
            words.append(words.pop(0))
        elif (m1 := re.match(r'^[A-Z]{3}$', words[0])) and (m2 := re.match(r'[\d]+\.[\d]+', words[1])):
            # It's the original currency
            if words[0] == "NOK":
                # It's Norwegian kroner, ignoring
                del words[0]
                del words[0]
            else:
                # It's some other currency, move it to the end
                words.append(words.pop(0))
                words.append(words.pop(0))
        else:
            break
    else:
        raise Exception(f"Infinite loop while parsing \"{original}\"")


    return " ".join(words)


def convert_row(row):
    if is_reserved(row):
        return None

    if row['Valuta'] != 'NOK':
        raise ValueError(f"Unknown currency {row['Valuta']}")

    return [
        row['Bokført dato'],
        row['Beskrivelse'],
        convert_memo(row['Melding']),
        -float(row['Beløp ut'] or '0'),
        float(row['Beløp inn'] or '0'),
    ]


def convert(reader, writer):
    writer.writerow(out_header)

    # Ignore header
    reader.__next__()

    for raw_row in reader:
        # Stop when we hit an empty row
        for field in raw_row:
            if field:
                break
        else:
            break

        # Create dictionary with proper field names
        row = {x[0]: x[1] for x in zip(in_header, raw_row)}

        try:
            if (result := convert_row(row)) is not None:
                writer.writerow(result)
        except Exception as e:
            print(f'Error in row {row}')
            raise e


def main():
    if len(sys.argv) != 2:
        usage()
        exit(1)

    filepath = Path(sys.argv[1])

    new_basename = f'ynab-{filepath.name}'
    new_filepath = filepath.parent / new_basename

    with filepath.open(mode='r', encoding='latin1') as old_file:
        reader = csv.reader(old_file, delimiter=';')

        with new_filepath.open(mode='w') as new_file:
            writer = csv.writer(new_file)
            convert(reader, writer)


if __name__ == '__main__':
    main()