Skip to main content

Track account credits in Django

Project description

Accounts credit tracking for Django
===========================

Accounts can be used to implement a variety of interesting components, including:

* Giftcards
* Web accounts
* Loyalty schemes

Basically anything that involves tracking the movement of funds within a closed
system.

This package uses [double-entry bookkeeping](http://en.wikipedia.org/wiki/Double-entry_bookkeeping_system)
where every transaction is recorded twice (once for the source and once for the
destination). This ensures the books always balance and there is full audit
trail of all transactional activity.

If your project manages money, you should be using a library like this. Your
finance people will thank you.

Features
--------

* An account has a credit limit which defaults to zero. Accounts can be set up
with no credit limit so that they are a 'source' of money within the system.
At least one account must be set up without a credit limit in order for money
to move around the system.

* Accounts can have:
- No users assigned
- A single "primary" user - this is the most common case
- A set of users assigned

* A user can have multiple accounts

* An account can have a start and end date to allow its usage in a limited time
window

* Accounts can be categorised

Installation
------------

TODO: make available on pypi for pip installation.

You should set-up a cronjob that calls:

./manage.py close_expired_accounts

to close any expired accounts and transfer their funds to the 'expired'
account.

API
---

Create account instances using the manager:

``` python

from decimal import Decimal
import datetime

from django.contrib.auth.models import User

from accounts import models

anonymous_account = models.Account.objects.create()

barry = User.objects.get(username="barry")
user_account = models.Account.objects.create(primary_user=barry)

no_credit_limit_account = models.Account.objects.create(credit_limit=None)
credit_limit_account = models.Account.objects.create(credit_limit=Decimal('1000.00'))

today = datetime.date.today()
next_week = today + datetime.timedelta(days=7)
date_limited_account = models.Account.objects.create(
start_date=today, end_date=next_week)
```

Transfer funds using the facade:

``` python

from accounts import facade

staff_member = User.objects.get(username="staff")
trans = facade.transfer(source=no_credit_limit_account,
destination=user_account,
amount=Decimal('10.00'),
user=staff_member)
```

Reverse transfers:

``` python
facade.reverse(trans, user=staff_member,
description="Just an example")
```

If the proposed transfer is invalid, an exception will be raised. All
exceptions are subclasses of `accounts.exceptions.AccountException`. Your
client code should look for exceptions of this type and handle them
appropriately.

Client code should only use the `accounts.models.Budget` class and the
two functions from `accounts.facade` - nothing else should be required.

Error handling
--------------

Note that the transfer operation is wrapped in its own database transaction to
ensure that only complete transfers are written out. When using Django's
transaction middleware, you need to be careful. If you have an unhandled
exception, then account transfers will still be committed even though nothing
else will be. To handle this, you need to make sure that, if an exception
occurs during your post-payment code, then you roll-back any transfers.

Here's a toy example:

``` python
from accounts import facade

def submit(self, order_total):
# Take payment first
transfer = facade.transfer(self.get_user_account(),
self.get_merchant_account(),
order_total)
# Create order models
try:
self.place_order()
except Exception, e:
# Something went wrong placing the order. Roll-back the previous
# transfer
facade.reverse(transfer)
```

In this situation, you'll end up with two transfers being created but no order.
While this isn't ideal, it's the best way of handling exceptions that occur
during order placement.

Multi-transfer payments
-----------------------

Projects will often allow users to have multiple accounts and pay for an order
using more than one. This will involve several transfers and needs some careful handling
in your application code.

It normally makes sense to write your own wrapper around the accounts API to encapsulate
your business logic and error handling. Here's an example:

``` python
from decimal import Decimal as D
from accounts import models, exceptions, facade


def redeem(order_number, user, amount):
# Get user's non-empty accounts ordered with the first to expire first
accounts = models.Account.active.filter(
user=user, balance__gt=0).order_by('end_date')

# Build up a list of potential transfers that cover the requested amount
transfers = []
amount_to_allocate = amount
for account in accounts:
to_transfer = min(account.balance, amount_to_allocate)
transfers.append((account, to_transfer))
amount_to_allocate -= to_transfer
if amount_to_allocate == D('0.00'):
break
if amount_to_allocate > D('0.00'):
raise exceptions.InsufficientFunds()

# Execute transfers to some 'Sales' account
destination = models.Account.objects.get(name="Sales")
completed_transfers = []
try:
for account, amount in transfers:
transfer = facade.transfer(
account, destination, amount, user=user,
description="Order %s" % order_number)
completed_transfers.append(transfer)
except exceptions.AccountException, transfer_exc:
# Something went wrong with one of the transfers (possibly a race condition).
# We try and roll back all completed ones to get us back to a clean state.
try:
for transfer in completed_transfers:
facade.reverse(transfer)
except Exception, reverse_exc:
# Uh oh: No man's land. We could be left with a partial redemption. This will
# require an admin to intervene. Make sure your logger mails admins on error.
logger.error("Order %s, transfers failed (%s) and reverse failed (%s)",
order_number, transfer_exc, reverse_exc)
logger.exception(reverse_exc)

# Raise an exception so that your client code can inform the user appropriately.
raise RedemptionFailed()
else:
# All transfers completed ok
return completed_transfers
```

As you can see, there is some careful handling of the scenario where not all transfers can be
executed.

``` python

try:
transfers = api.redeem(order_number, user, total_incl_tax)
except Exception:
# Inform user of failed payment
else:
for transfer in transfers:
source_type, __ = SourceType.objects.get_or_create(name="Accounts")
source = Source(
source_type=source_type,
amount_allocated=transfer.amount,
amount_debited=transfer.amount,
reference=transfer.reference)
self.add_payment_source(source)
```

Core accounts and account types
-------------------------------

A post-syncdb signal will create the common structure for account types and
accounts. Some names can be controlled with settings, as indicated in parentheses.

- **Assets**

- **Sales**
- Redemptions (`ACCOUNTS_REDEMPTIONS_NAME`) - where money is
transferred to when an account is used to pay for something.
- Lapsed (`ACCOUNTS_LAPSED_NAME`) - where money is transferred to
when an account expires. This is done by the
'close_expired_accounts' management command. The name of this
account can be set using the `ACCOUNTS_LAPSED_NAME`.

- **Cash**
- "Bank" (`ACCOUNTS_BANK_NAME`) - the source account for creating new
accounts that are paid for by the customer (eg a giftcard). This
account will not have a credit limit and will normally have a
negative balance as money is only transferred out.

- **Unpaid** - This contains accounts that are used as sources for other
accounts but aren't paid for by the customer. For instance, you might
allow admins to create new accounts in the dashboard. An account of this
type will be the source account for the initial transfer.

- **Liabilities**

- **Deferred income** - This contains customer accounts/giftcards. You may want to create
additional account types within this type to categorise accounts.

Example transactions
--------------------

Consider the following accounts and account types:

- **Assets**
- **Sales**
- Redemptions
- Lapsed
- **Cash**
- Bank
- **Unpaid**
- Merchant funded
- **Liabilities**
- **Deferred income**

Note that all accounts start with a balance of 0 and the sum of all balances will always be zero.

*A customer purchases a £50 giftcard*

- A new account is created of type 'Deferred income' with an end date
- £50 is transferred from the Bank to this new account

*A customer pays for a £30 order using their £50 giftcard*

- £30 is transferred from the giftcard account to the redemptions account

*The customer's giftcard expires with £20 still on it*

- £20 is transferred from the giftcard account to the lapsed account

*The customer phones up to complain and a staff member creates a new giftcard for £20*

- A new account is created of type 'Deferred income'
- £20 is transferred from the "Merchant funded" account to this new account

Settings
--------

There are settings to control the naming and initial unpaid and deferred income account
types:

* `ACCOUNTS_MIN_INITIAL_VALUE` The minimum value that can be used to create an
account (or for a top-up)

* `ACCOUNTS_MAX_INITIAL_VALUE` The maximum value that can be transferred to an
account.

Contributing
------------

Fork repo, set-up virtualenv and run:

make install

Run tests with:

./runtests.py

History
-------

This project used to depend on the e-commerce framework [Oscar](https://github.com/tangentlabs/django-oscar) in a number of places. You may be interested in the upstream repos if you want Oscar integration. This repo was forked off for people who want account balances without Oscar integration.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

django-account-balances-0.3.tar.gz (36.9 kB view details)

Uploaded Source

File details

Details for the file django-account-balances-0.3.tar.gz.

File metadata

File hashes

Hashes for django-account-balances-0.3.tar.gz
Algorithm Hash digest
SHA256 a3279c15054df257996549041f507512f89ed92126fe15d5a3072f8a1011967a
MD5 6bb8c230f44a2d64bd1cc012940730c8
BLAKE2b-256 1ae471126431c2dc063f52e276e415ae276bbb1ff0c6cbd39f088e717124eb15

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page