dkbparse

PDF parser for DKB bank and VISA statements

Can be handy for further processing your bank data (e.g. with beancount).

Features

• Can be used as shell script
• CSV export

Requirements

You will need to have Python 3 and pdftotext installed.

Usage

Shell Script

You can run dkbparse.py as shell script and pass a list of directories that contain DKB PDF statements. The script will write all parsed transactions as CSV to stdout:

$ ./dkbparse.py ~/dkb/account/1234567890/ ~/dkb/visa/4998xxxxxxxx1234/

account,year,statement,transaction,booked,valued,value,type,payee,comment
0000001234567890,2019,01,001,2019-01-02,2018-12-31,-250.00,Überweisung,Tante Helga,ÜBERWEISUNG DATUM 29.12.2018, 05.33 UHR 1.TAN 123456
...

Library

Alternatively, you can pass a list of directories that contain DKB PDF files (VISA and/or bank statements) to dkbparse.scan_dir(). All directories are scanned recursively. The function returns a list of transactions and a list of bank statements that were parsed:

import os
from dkbparse import scan_dir

# folder with DKB PDFs
dirpath = os.path.expanduser('~/dkb/visa/4998xxxxxxxx1234') 
transactions, statements = dkbparse.scan_dirs([dirpath])

A transaction dict looks like this

{
    'account': '0000001234567890',
    'year': '2019',
    'statement': '01',
    'transaction': '001',
    'booked': datetime.date(2019, 1, 2), 
    'valued': datetime.date(2018, 12, 31), 
    'type': 'Überweisung', 
    'value': Decimal('-250.00'), 
    'payee': 'Tante Helga',
    'comment': 'ÜBERWEISUNG DATUM 29.12.2018, 05.33 UHR 1.TAN 123456'
}

A statement dict looks like this

{   
    'account': '9451359782',
    'year': '2019'
    'no': '1',
    'from': datetime.datetime(2018, 12, 29, 0, 0),
    'to': datetime.datetime(2019, 1, 3, 0, 0),
    'balance_old': Decimal('1443.81'),
    'balance_new': Decimal('487.33'),
    'file': 'Kontoauszug_9451359782_Nr_2019_001_per_2019_03_01.pdf'
}

The script will output an error if the sum of all transactions of a statement does not correspond to the stated balance difference. This will help you to identify cases where the parser fails to parse a statement.

Performance

The script scans around 100 PDF files (or 2500 transactions) per second.

Limitations

The main limitation of this approach is that the output depends on the version of the installed pdftotext (Poppler). Different results on different platforms can not be excluded. For example, using pdftotext version 0.86.1, I occasionally get the following error, which appears to be a known issue.

Syntax Warning: FoFiType1::parse a line has more than 255 characters, we don't support this