A project which solved an issue of mine which was tracking my finance. This Finance Tracking application gives overviews of expenses and income to give its users an easy way to explore their data.
I have always been interested in finance and personal economics. Therefore, for the past five years, I have been tracking every single expense or income that I have made. And like most people, I then visualized that data in excel. But creating advanced views with dynamic categories became extremely hard and required a lot of work.
An easy solution for this (kind of time-consuming 😊) was to in a more programmatic way explore and manage my personal finance. The application allows me to, in a much faster and easy way track month-to-month expenses and income.
- MYSQL database to store transactions and loans
- The website is made using React and website authentication is made using Firebase
- Eight different API endpoints were developed to manage the data between the application and the database
- Python Pandas to clean the old transaction data
The overview page is what the user is first presented with when entering the website. The overview view gives the user an insight into last month's income and expenses and a bar chart for this year's income and expenses. The page also includes two sections where the user can either add a new transaction or upload an excel document. The second part of the page includes a list of transactions where the user can easily navigate and see the different transactions made. The table gives the user the option to edit each transaction.
The two most important pages are the yearly and total overview. Both these pages give deep insight into exactly which income/expense categories exist and exactly how much each source is. With the yearly overview, the user can explore how much the user has made an impact on their NET growth. The total view gives more insight into comparing the years towards each other.
A big part of keeping track of personal finance is knowing exactly where all the existing money is located. As you grow older you usually will end up with more and more different bank accounts, investment portfolios, and saving accounts. The accounts overview gives the exact amount that exists in each account. One thing that I had a really hard time with before when tracking in excel is that even after tracking all the monthly expenses the accounts wouldn’t always line up with the real-world amount. Therefore, the transaction page also gives it user a way to edit the balance and then track it as an adjustment.
Another big thing in managing finance is that it must be easy to create new, edit or delete transactions. The application with help of API endpoints can easily perform all of these actions against the database. A biig functionality added to the application is uploading an excel document. The user can upload an excel document with a list of expenses and income which the application can post to the database. This makes it easy to manage the finance from month to month.
The Terraform configuration under infra/terraform/ can provision an Aurora PostgreSQL cluster along with an on-demand bastion host for direct database access. The repo ships with a couple of make targets and helper scripts to streamline the workflow.
# create the bastion (public subnet, security groups, IAM profile, etc.)
make tf-enable-bastion
# remove the bastion and associated networking resources when you're done
make tf-disable-bastionmake bastion-shellThis wraps aws ssm start-session and automatically resolves the bastion instance ID from Terraform outputs. You need the AWS CLI and Session Manager plugin installed locally (see AWS docs).
Use the SSM-based uplink helper to push any local file onto the host:
# upload temp.py to /home/ec2-user/ on the bastion
make bastion-copy FILE=temp.py
# optionally pick a different destination directory
make bastion-copy FILE=path/to/data.sql REMOTE=/tmpBehind the scenes this calls scripts/bastion_copy.py, which base64-encodes the file and posts it via aws ssm send-command, so no SSH keys are required.
The Amazon Linux 2023 image used for the bastion ships with Python 3 but not pip. After connecting via make bastion-shell, run:
sudo dnf install -y python3-pip postgresql-devel gcc
python3 -m pip install --user boto3 psycopg2-binaryThis installs the packages into the bastion user’s home directory so you can run helper scripts such as temp.py.
# assumes the Terraform stack has created the SSM parameters under /finance-tracker/<env>/db/*
python3 temp.py --timeout 30
# fetch credentials without connecting
python3 temp.py --skip-connectionThe script pulls credentials from SSM, prints the resolved endpoint/user/database, and then attempts a simple SELECT version(); against the Aurora cluster. Errors from PostgreSQL are shown verbatim to simplify troubleshooting (e.g., invalid password, timeout).
Always disable the bastion after you are done testing to avoid leaving unnecessary public infrastructure running:
make tf-disable-bastion