abacus-minimal 0.10.0

Ledger in Python that follows corporate accounting rules.

abacus-minimal

abacus-minimal aims to be as concise as possible in implementation of double entry book-keeping rules as applied in corporate accounting.

Project goals

• Make valid accounting engine in fewer lines of code (Python or other languages).
• Explain book-keeping rules through code examples.
• Make pathways into accounting for programmers and into programming for accountants.
• Curate various charts of accounts as JSON files and make conversions between them.
• Make free web learning tools in accounting similar to abacus-streamlit.
• Ultimately, lower the book-keeping and financial analytics costs for the businesses.

Non-goals:

• replacing SAP or QBO immediately with this Python code.

Changelog

• 0.10.0 (2024-10-24) -- separates core, chart, entry and book code and tests.

For the 0.11.0 version

• book.income_statement that works before and after period close
• book.balance_sheet with current_profit account when not closed + Chart.current_earnings attribute

Future versions

• cleaner BalancesDict
• Book.increase() and Book.decrease() methods
• Entry.explain() and Entry.validate() methods

Install

pip install abacus-minimal

Latest:

pip install git+https://github.com/epogrebnyak/abacus-minimal.git

Workflow

The steps for using abacus-minimal follow the steps of a typical accounting cycle:

• create a chart of accounts,
• post business transactions to ledger,
• make reconciliations and adjustments,
• close accounts at reporting period end,
• show reports for the financial results,
• save data for the next reporting period.

The complete example code for the workflow is in readme.py. For more implementation details see "Data structures, actions and architecture" section below.

1. Create chart of accounts

Steps involved:

• specify name of the retained earnings account that will accumulate company profits less dividend,
• add account names for assets, capital, liabilities, income and expenses,
• add contra accounts (eg refunds is a contra account to sales).

Code example:

from abacus import Chart

chart = Chart(
    retained_earnings="retained_earnings",
    assets=["cash"],
    capital=["equity"],
    liabilities=["vat_payable"],
    income=["sales"],
    expenses=["salaries"],
)
chart.offset("sales", "refunds")

Chart class is a pydantic model, which means it is easily converted to a JSON file:

chart.save("chart.json")

Equally, you can load an existing chart from file:

chart = Chart.load("chart.json")

2. Post entries to ledger

Steps involved:

• create a data structure that represents state of accounts (ledger),
• record account starting balances from the previous period (skip for a new company),
• record accounting entries that represent business transactions,
• show state of ledger (trial balance or account balances) at any time.

Code example:

from abacus import Book, Entry

# Create book with opening balances
opening_balances = {"cash": 10_000, "equity": 10_000}
book = Book(chart, opening_balances)

# Post entries
entries = [
    Entry("Sold services with VAT")
    .debit("cash", 6000)
    .credit("sales", 5000)
    .credit("vat_payable", 1000),
    Entry("Made client refund").double(debit="refunds", credit="cash", amount=500),
    Entry("Paid salaries").debit("salaries", 1500).credit("cash", 1500),
]
book.post_many(entries)
print(book.trial_balance)

Invalid entries will be rejected with AbacusError raised. The invalid entries are the ones that touch non-existent accounts or the entries where debits and credits are not balanced.

3. Closing accounts

Steps before closing (not in current example):

• make reconciliation entries,
• make adjustment entries for accruals and deferrals.

Closing accounts at period end involves:

• closing contra accounts to temporary accounts,
• closing temporary accounts to the retained earnings account,
• make post-close entries if applicable (not in current example).

Closing accounts was probably the hardest part of the project where I had to refactor code several times to make it more explicit and concise.

4. Reporting and saving

Steps involved:

• show balance sheet and income statement,
• save account balances for the next period.

Saving the book will write chart.json, store.json and balances.json files.

Code example:

# Close at period end and show reports
print(book.income_statement)
book.close()
print(book.balance_sheet)

# Check account balances match expected values
assert book.ledger.balances == {
    "cash": 14000,
    "equity": 10000,
    "vat_payable": 1000,
    "retained_earnings": 3000,
}

# Save everything to JSON files in current folder
book.save(directory=".")

Data structures, actions and architecture

Data structures

Underneath Chart, Entry and Book clasees there are more primitive data structures that make up the core of abacus-minimal:

• ChartDict holds chart of accounts information and ensures uniqueness and consistency of account names.
• SingleEntry specifies amount and debit or credit side.
• MultipleEntry is a list of SingleEntry items where sum of debit and credit entries should match.
• TAccount is the base class for DebitAccount and CreditAccount.
• Ledger is a dictionary that maps account names to accounts and accepts entries for posting.
• TrailBalance and BalancesDict show account names and their balances.
• BalanceSheet and IncomeStatement are financial reports based on ledger state.

Actions

The principal chain of actions in abacus-minimal is shown in a table below. The function signatures (or type annotations) indicate want variables participate in the calculation.

Action	Function signature
Create ledger	ChartDict -> Ledger
Post entries to ledger	Ledger -> [MultipleEntry] -> Ledger
Make a list of closing pairs	(ChartDict, AccountName) -> [(AccountName, AccountName)]
Close ledger at period end	(ChartDict, AccountName) -> Ledger -> (IncomeStatement, Ledger)
Report balance sheet	Ledger -> BalanceSheet
Show trial balance	Ledger -> TrialBalance
Show account balances	Ledger -> BalancesDict.

Architecture

abacus-minimal focuses on enforcing the book-keeping rules, but not on not storing entires.

Unlike production accounting projects abacus-minimal is database-free. For abacus-minimal it would not
matter how entries are saved and where they are coming from -- this responsibility should be taken by some other part of software, eg the medici ledger. The Book class does offer saving of entries to a JSON file, but this is done for demonstration only.

Limitations

Several assumptions and simplifications are used to make abacus-minimal more manageable to develop.

The key assumptions are:

• one currency,
• one level of accounts in chart,
• no intermediate accounts,
• no changes in equity and cash flow statements.

See main.py module docstring for more details.

Alternatives

abacus-minimal takes inspiration from the following great projects:

• hledger and plain text accounting tools,
• medici ledger in JavaScript using Mongo database,
• microbooks API and python-accounting.

Plain text accounting tools are usually for personal finance while abacus-minimal targets accounting for a corporate entity. medici is a high performance ledger, but does not enforce accounting rules on data entry. python-accounting is a production-grade project, tightly coupled to a database.

Big players in accounting software for small and middle-sized companies are Intuit Quickbooks (US) and Xero (Australia). Not everyone is happy with how they work or how much they cost, especially when moving from a desktop version to the cloud.

Many other office automation providers do also have accounting APIs (eg Zoho) and there are open source packages that have accounting functionality (eg Frappe).

Several outlets advertise they provide IFRS-compliant charts of accounts, but usually as Excel files. There are account taxonomies for reporting, but the charts more seldom.

Accounting knowledge

If you are totally new to accounting the suggested friendly course is https://www.accountingcoach.com/.

ACCA and CPA are the international and the US professional qualifications and IFRS and GAAP are the standards for accounting recognition, measurement and disclosure.

Part B-G in the ACCA syllabus for the FFA exam talk about what abacus-minimal is designed for.

Project conventions

I use just command runner to automate code maintenance tasks in this project.

just test and just fix scripts will run the following tools:

• pytest
• mypy
• black and isort --float-to-top (probably should replace with ruff format)
• ruff check
• prettier for markdown formatting
• codedown to extract Python code from README.md.

examples/readme.py is overwritten by the just readme command.