Balance Takehome 💰

Balance Takehome is a prototype GraphQL API to connect with the merge.dev Unified API to access and retrieve Accounting data from common models like Company Info, Accounts and Transactions.

Local Requirements

• Docker/Compose
• Python 3.9 or greater

Local Execution

• Build images and start docker containers using:

cd balance-takehome
docker-compose up --build

• Access pgAdmin through http://localhost:5050 to verify the schema creation and the ingested data:

• Access GraphiQL (In-Browser GraphQL IDE) through http://localhost:8000/graphql and execute the following queries:

1.
query {
  balanceBreakdown {
      ... on DateTimeType {
          now
      }
      ... on BankAccountsList {
        accounts {
          name
          currentBalance
        }
      }
  }
}

2.
query {
  incomeAndExpenses(month:10, year:2023) {
    income
    expenses
  }
}

3.
query {
  transactionsForInterval(startDate:"2023-03-01", endDate:"2023-03-21") {
    ... on TransactionsList {
      transactions {        
        transactionDate
        name
        transactionTo
        transactionFrom
        amount
      }
    }
  }
}

Local Development

• Execute the unit tests session through pytest:

docker exec balance_takehome_app pytest

• Execute the pre-configured Black python formatter:

docker exec balance_takehome_app black main.py app tests

• Execute the isort utility to manage imports:

docker exec balance_takehome_app isort main.py app tests

• Execute a static type checking analysis using mypy:

docker exec balance_takehome_app mypy main.py app tests

• After any database schema/tables change, you'd have to execute the following migration commands:

docker exec balance_takehome_app alembic revision --autogenerate
docker exec balance_takehome_app alembic upgrade head

Technologies / Tradeoffs

FastAPI

• FastAPI provides high performance and good coding time. Also, through its async capabilities and web server (ASGI Standard - Asynchronous Server Gateway Interface), we can schedule a coroutine that executes the polling process with real time confidence just as a Celery task.

PostgreSQL

• PostgreSQL is widely used as a relational database for API's supported by ORM's to abstract the interaction. Using it we gain flexibility and particular capabilities/utilities like the on_conflict_do_update method to execute the UPSERT operation using constraints.

SQLModel - SQLAlchemy & Pydantic

• The SQLModel library mixes the features from SQLAlchemy & Pydantic. It provides a layer of communication (between both libraries) that allows the creation of database tables through type annotations, just as simple as a dataclass or a Pydantic BaseModel. Avoiding code duplication and supporting a clean codebase.

UML

Entity–relationship model

Architecture