csv2ofx is a Python library and command line interface program that converts CSV files to OFX and QIF files for importing into GnuCash or similar financial accounting programs. csv2ofx has built in support for importing csv files from mint, yoodlee, and xero.
csv2ofx has been tested and is known to work on Python 3.7, 3.8, and 3.9; and PyPy3.7.
(You are using a virtualenv, right?)
sudo pip install csv2ofx
csv2ofx is intended to be used either directly from Python or from the command line.
normal OFX usage
import itertools as it
from meza.io import read_csv, IterStringIO
from csv2ofx import utils
from csv2ofx.ofx import OFX
from csv2ofx.mappings.default import mapping
ofx = OFX(mapping)
records = read_csv('path/to/file.csv', has_header=True)
groups = ofx.gen_groups(records)
trxns = ofx.gen_trxns(groups)
cleaned_trxns = ofx.clean_trxns(trxns)
data = utils.gen_data(cleaned_trxns)
content = it.chain([ofx.header(), ofx.gen_body(data), ofx.footer()])
for line in IterStringIO(content):
print(line)normal QIF usage
import itertools as it
from tabutils.io import read_csv, IterStringIO
from csv2ofx import utils
from csv2ofx.qif import QIF
from csv2ofx.mappings.default import mapping
qif = QIF(mapping)
records = read_csv('path/to/file.csv', has_header=True)
groups = qif.gen_groups(records)
trxns = qif.gen_trxns(groups)
cleaned_trxns = qif.clean_trxns(trxns)
data = utils.gen_data(cleaned_trxns)
content = it.chain([qif.gen_body(data), qif.footer()])
for line in IterStringIO(content):
print(line)show help
csv2ofx -h
usage: csv2ofx [options] <source> <dest>
description: csv2ofx converts a csv file to ofx and qif
positional arguments:
source the source csv file (defaults to stdin)
dest the output file (defaults to stdout)
optional arguments:
-h, --help show this help message and exit
-a TYPE, --account TYPE
default account type 'CHECKING' for OFX and 'Bank' for QIF.
-e DATE, --end DATE end date
-l LANGUAGE, --language LANGUAGE
the language
-s DATE, --start DATE
the start date
-y, --dayfirst interpret the first value in ambiguous dates (e.g. 01/05/09) as the day
-m MAPPING, --mapping MAPPING
the account mapping
-x FILE_PATH, --custom FILE_PATH
path to a custom mapping file
-c FIELD_NAME, --collapse FIELD_NAME
field used to combine transactions within a split for double entry statements
-C ROWS, --chunksize ROWS
number of rows to process at a time (default: 2 ** 14)
-r ROWS, --first-row ROWS
number of initial rows to skip (default: 0)
-r ROWS, --last-row ROWS
the final rows to process, negative values count from the end (default: inf)
-O COLS, --first-col COLS
number of initial cols to skip (default: 0)
-L, --list-mappings list the available mappings
-V, --version show version and exit
-q, --qif enables 'QIF' output instead of 'OFX'
-o, --overwrite overwrite destination file if it exists
-D SERVER_DATE, --server-date SERVER_DATE
OFX server date (default: source file mtime)
-E ENCODING, --encoding ENCODING
File encoding (default: utf-8)
-d, --debug display the options and arguments passed to the parser
-v, --verbose verbose outputnormal usage
csv2ofx file.csv file.ofx
print output to stdout
csv2ofx ~/Downloads/transactions.csv
read input from stdin
cat file.csv | csv2ofx
qif output
csv2ofx -q file.csv
specify date range from one year ago to yesterday with qif output
csv2ofx -s '-1 year' -e yesterday -q file.csv
use yoodlee settings
csv2ofx -m yoodlee file.csv
Some banks, like UBS Switzerland, may provide CSV exports that are not
readily tractable by csv2ofx because of extra header or trailing lines,
redundant or unwanted columns. These input files can be preprocessed with the
shipped utilz/csvtrim shell script. F.i., with mapping ubs-ch-fr:
csvtrim untrimmed.csv | csv2ofx -m ubs-ch-fr
If you would like to import csv files with field names different from the default, you can modify the mapping file or create your own. New mappings must be placed in the csv2ofx/mappings folder (otherwise you must use the ). The mapping object consists of a dictionary whose keys are OFX/QIF attributes and whose values are functions which should return the corresponding value from a record (csv row). The mapping function will take in a record, e.g.,
{'Account': 'savings 2', 'Date': '1/3/15', 'Amount': 5000}The most basic mapping function just returns a specific field or value, e.g.,
from operator import itemgetter
mapping = {
'bank': 'BetterBank',
'account': itemgetter('Account'),
'date': itemgetter('Date'),
'amount': itemgetter('Amount')}But more complex parsing is also possible, e.g.,
mapping = {
'account': lambda r: r['Details'].split(':')[0],
'date': lambda r: '%s/%s/%s' % (r['Month'], r['Day'], r['Year']),
'amount': lambda r: r['Amount'] * 2,
'first_row': 1,
'last_row': 10,
'filter': lambda r: r['Amount'] > 10,
}| attribute | description | default field | example |
|---|---|---|---|
account |
transaction account | Account | BetterBank Checking |
date |
transaction date | Date | itemgetter('Transaction Date') |
amount |
transaction amount | Amount | itemgetter('Transaction Amount') |
| attribute | description | default field | default value | example |
|---|---|---|---|---|
desc |
transaction description | Reference | n/a | shell station |
payee |
transaction payee | Description | n/a | Shell |
notes |
transaction notes | Notes | n/a | for gas |
check_num |
the check or transaction number | Row | n/a | 2 |
id |
transaction id | check_num |
Num | n/a |
bank |
the bank name | n/a | account |
Bank |
account |
transaction account type | n/a | checking | savings |
account_id |
transaction account id | n/a | hash of account |
bb_checking |
type |
transaction type (either debit or credit) | n/a | CREDIT if amount > 0 else DEBIT | debit |
balance |
account balance | n/a | n/a | $23.00 |
class |
transaction class | n/a | n/a | travel |
| attribute | description | default value | example |
|---|---|---|---|
has_header |
does the csv file have a header row | True | |
is_split |
does the csv file contain split (double entry) transactions | False | |
currency |
the currency ISO code | USD | GBP |
delimiter |
the csv field delimiter | , | ; |
date_fmt |
custom QIF date output format | %m/%d/%y | %m/%d/%Y |
dayfirst |
interpret the first value in ambiguous dates (e.g. 01/05/09) as the day (ignored if parse_fmt is present) |
False | True |
parse_fmt |
transaction date parsing format | %m/%d/%Y | |
first_row |
the first row to process (zero based) | 0 | 2 |
last_row |
the last row to process (zero based, negative values count from the end) | inf | -2 |
first_col |
the first column to process (zero based) | 0 | 2 |
filter |
keep transactions for which function returns true | lambda tr: tr['amount'] > 10 |
csv2ofx comes with a built in task manager manage.py.
pip install -r dev-requirements.txt
Run python linter and nose tests
manage lint
manage testPlease mimic the coding style/conventions used in this repo. If you add new classes or functions, please add the appropriate doc blocks with examples. Also, make sure the python linter and nose tests pass.
Ready to contribute? Here's how:
- Fork and clone.
git clone git@github.com:<your_username>/csv2ofx.git
cd csv2ofx- Setup a new virtualenv
mkvirtualenv -i pkutils csv2ofx
activate csv2ofx
python setup.py develop
pip install -r dev-requirements.txt- Create a branch for local development
git checkout -b name-of-your-bugfix-or-feature- Make your changes, run linter and tests (see above), and submit a pull request through the GitHub website.
How to contribute a mapping:
- Add the mapping in
csv2ofx/mappings/ - Add a simple example CSV file in
data/test/. - Add the OFX or QIF file that results from the mapping and example CSV file in
data/converted/. - Add a
csv2ofxcall for your mapping to the tests intests/test.py, inPRE_TESTS. If you added an OFX (not QIF) converted file, pay attention to the-e(end date) and-D(server date) arguments in the test- otherwise tests may pass on your workstation and fail on the build server. - Ensure your test succeeds (see above).
csv2ofx is distributed under the MIT License, the same as meza.