otteroad 0.2.1

Framework with scalable Kafka consumer/producer logic for IDU FastAPI services.

🦦 otteroad

Kafka framework for FastAPI microservices in the IDU (Institute of Design and Urban Studies).

** The name was inspired by this text.

---

✨ Overview

otteroad is a Kafka framework designed for FastAPI-based microservices. It simplifies integration with Apache Kafka and supports:

• ✅ Unified consumer & producer APIs
• ✅ AVRO + Schema Registry support via Pydantic
• ✅ Pluggable settings from .env, .yaml, or custom config
• ✅ Structured event handlers with lifecycle hooks
• ✅ Flexible handler registry and extensible processing pipeline
• ✅ Designed for FastAPI services but works standalone

---

📦 Installation

pip install otteroad

Or via poetry:

poetry add otteroad

---

⚙️ Configuration

Kafka settings are defined via two classes:

• KafkaConsumerSettings
• KafkaProducerSettings

They can be created from any source:

from otteroad import KafkaConsumerSettings, KafkaProducerSettings

consumer_settings = KafkaConsumerSettings.from_env()
producer_settings = KafkaProducerSettings.from_yaml("config/kafka.yaml")

# define pydantic model/dataclass/dict/etc.
config = {"bootstrap.servers": "localhost: 9092"}
settings = KafkaProducerSettings.from_custom_config(config)

---

📡 Event Models (AVRO + Schema Registry)

Use AvroEventModel as the base for your event schemas. These are strict, typed messages validated via Pydantic.

from typing import ClassVar
from pydantic import Field
from otteroad.avro import AvroEventModel


class TerritoryCreated(AvroEventModel):
    """Model for message indicates that a territory has been created."""

    topic: ClassVar[str] = "urban.events"
    namespace: ClassVar[str] = "territories"
    schema_version: ClassVar[int] = 1
    schema_compatibility: ClassVar[str] = "BACKWARD"

    territory_id: int = Field(..., description="new territory identifier")

---

🧠 Handlers

Handlers process typed events. Extend BaseMessageHandler and implement core logic in handle(). Optional hooks: pre_process, post_process, on_startup, on_shutdown, handle_error.

ℹ️ Note for IDU services: It is strongly recommended to use only models from the models/ directory to ensure schema consistency and maintainability across services.

from otteroad.consumer import BaseMessageHandler
from otteroad.models import TerritoryCreated  # please, use only models from the models/ directory

class TerritoryCreatedHandler(BaseMessageHandler[TerritoryCreated]):
    async def handle(self, event, ctx):
        print(f"Territory created: {event.territory_id}")
        
    async def on_startup(self): ...
    
    async def on_shutdown(self): ...

---

🔄 Consumer

KafkaConsumerService manages lifecycle and worker threads; KafkaConsumerWorker pulls messages, resolves handlers and runs processing logic.

from otteroad import KafkaConsumerService

service = KafkaConsumerService(consumer_settings)
service.register_handler(TerritoryCreatedHandler())
service.add_worker(topics=["urban.events"]).start()

Under the hood, the pipeline is:

receive message -> validate -> pre_process -> handle -> post_process

If an error occurs, custom error handling or DQL logic can be added.

---

🚀 Producer

Use KafkaProducerClient to send strongly typed Avro events:

from otteroad import KafkaProducerClient
from otteroad.models import TerritoryCreated

async def send_event():
    async with KafkaProducerClient(producer_settings) as producer:
        event = TerritoryCreated(territory_id=1)
        await producer.send(event)

---

🧩 FastAPI Integration

For a simple integration example with FastAPI, see:

📄 Example