outbox-streaming

Reliably send messages to message/task brokers, like Kafka or Celery

Roadmap

Done

• ✅ Kafka + SQLAlchemy
• ✅ Kafka + SQLAlchemy + asyncio

In progress

• ⏹ Celery + SQLAlchemy
• ⏹ Celery + SQLAlchemy + asyncio

Planned

• 🆕 Kafka + Django ORM
• 🆕 Celery + Django ORM
• 🆕 Dramatiq + SQLAlchemy
• 🆕 Dramatiq + SQLAlchemy + asyncio
• 🆕 Dramatiq + Django
• 🆕 RabbitMQ + SQLAlchemy
• 🆕 RabbitMQ + SQLAlchemy + asyncio
• 🆕 RabbitMQ + Djagno

Example FastAPI + Kafka + SQLAlchemy

from fastapi import FastAPI
from outbox_streaming.kafka.sqlalchemy import SQLAlchemyKafkaOutbox
from sqlalchemy import orm

from app import models, db
from app.config import config
from app.schemas import TodoCreate

app = FastAPI()

# create instance of SQLAlchemyKafkaOutbox
outbox = SQLAlchemyKafkaOutbox(
    engine=db.engine,
    kafka_servers=config.KAFKA_SERVERS,
)

# Run separate tread that monitor outbox table and publish messages to Kafka. It's not recommended for production,
# but it's handy for development
outbox.publisher.run_daemon()

Session = orm.sessionmaker(bind=db.engine)

# create outbox tables 
outbox.storage.create_tables(engine=db.engine)


@app.post('/todos')
def create_todo(create: TodoCreate) -> str:

    with Session() as session:

        # create new object
        todo = models.Todo(text=create.text)

        session.add(todo)

        # create kafka event in outbox table
        outbox.save(
            session=session,
            topic='todo_created',
            value=todo.to_dict(),
        )

        # commit changes in database
        session.commit()
        
    # publisher will pick up kafka message from outbox table and will send it kafka topic

    return "OK"