jvillegasd / transaction-logs

Simple bank transaction logs system using Python core and SQLAlchemy

Transaction logs

This project consist in create a simple bank transaction system, where users can interact with their transactions and filter them. Python 3.10 was used for this project.

Requirements

• As a user I want to be able to login and see the transactions, listed from most recent to least recent.

• As a user I can return my balance as a mathematical result of the executed transactions.

• As a user I want to be able to list the transactions in a range of dates.

• As a user I want to be able to list the transactions by type.

• As a user I want to be able to filter the expenses by merchant.

System requirements

• You must ensure that a user cannot see the transactions of other users

• There are three types of transactions in this simple system which are deposits, withdrawals and expenses.

• he results of the transactions must be paged being the page size passed as a query parameter, if no parameter present then assume 10 as page size

• You cannot withdraw/waste more money than the account has, therefore negative balances are not allowed.

• If a transaction results in a negative balance, it must be rejected.

• Validate that the date ranges make sense for the transaction filter.

Data importer

CRUD for transactions is not required, instead, I designed and developed a data importer class that loads seeds in JSON files for the models created for the project. This DataImporter class was designed as a generic way to loads models seeds, so it is reusable.

Persist data

Loaded seeds are stored in a local PostgreSQL database mounted by Docker (you can see this service in details in docker compose file), so backend can interact with the data through queries.

Sessions

In order to maintain sessions for logged users, a simple session provided for Flask was used.

Libraries

A core rule was to do not use external libraries for the project beyond the framework. So, I chose Flask as Backend micro-framework and SQLAlchemy as ORM library (Unlike Django or FastAPI, Flask lacks of ORM). Schemas, filters, serializers are made from vanilla, my own implementations.

Documentation

Core functions, method and classes have their own docstring for code documentation. API Postman collection is saved in folder docs inside repo.

Unit testing

Python in-build library unittest was used for implement unit tests for this backend. I implemented different test cases for API endpoints and DataImporter in order to cover most cases. Tests inherit from a BaseCase class that implement common functions that all tests can use. This class allows unit tests to load models seeds on demand. You can execute tests with the following command:

python3 -m unittest

If you use Docker for run this project, you can run tests with this command:

docker exec transaction-api python3 -m unittest

Run the project

In order to make easier to run this project in multiple environment, a Docker image and a Docker compose file were created. You can build and run this project with this command:

docker-compose up -d

Environment variables

On this section, you can see how an Environment vars file looks like:

APP_ENV=dev
PORT=3001
SECRET_KEY=u3gsKSs8QSSdOXkw6nxB9Gq2xuCF8UQ6

POSTGRES_USER=admin
POSTGRES_PASSWORD=admin123
POSTGRES_HOST=transaction-db
POSTGRES_PORT=5432
POSTGRES_DB=transaction-logs-db

Filtering

A vanilla filtering algorithm was created for handle dynamic filters for ORM queries. For complex filtering such as range comparison, I implemented a bracket notation to accomplish this. On the following table, you will see the available filtering notation. Filters are received as Query parameters!

• For direct filter, use normal query parameter notation, example:
  • GET - /api/transactions?transaction_type=expense
• For range filtering, you have these operations:
  • '[lt]' = 'Lower than'
  • '[le]' = 'Lower or equal than'
  • '[gt]' = 'Greater than'
  • '[ge]' = 'Greater or equal than'
  • '[eq]' = 'Equals to'
  • Example: GET - /api/transactions?created_at[ge]=12-15-2022&created_at[le]=12-25-2022
• PD: Dates have format: month-day-year

Pagination

Pagination is handle via Query parameters using Cursor technique for quick database queries. The cursor is the field created_at in timestamp. Example:

GET - /api/transactions?per_page=5&next_cursor=1671112380.617523

You can combine pagination with filters queries!