spakky-kafka 6.4.0

Apache Kafka plugin for Spakky framework

Spakky Kafka

Apache Kafka plugin for Spakky Framework.

Installation

pip install spakky-kafka

Or install via Spakky extras:

pip install spakky[kafka]

Configuration

Set environment variables with the SPAKKY_KAFKA__ prefix:

export SPAKKY_KAFKA__GROUP_ID="my-consumer-group"
export SPAKKY_KAFKA__CLIENT_ID="my-app"
export SPAKKY_KAFKA__BOOTSTRAP_SERVERS="localhost:9092"
export SPAKKY_KAFKA__AUTO_OFFSET_RESET="earliest"  # earliest, latest, none

SASL Authentication (Optional)

export SPAKKY_KAFKA__SECURITY_PROTOCOL="SASL_SSL"
export SPAKKY_KAFKA__SASL_MECHANISM="PLAIN"
export SPAKKY_KAFKA__SASL_USERNAME="username"
export SPAKKY_KAFKA__SASL_PASSWORD="password"

Topic Configuration (Optional)

export SPAKKY_KAFKA__NUMBER_OF_PARTITIONS="3"
export SPAKKY_KAFKA__REPLICATION_FACTOR="1"

Usage

Event Publishing

from spakky.core.common.mutability import immutable
from spakky.domain.models.event import AbstractIntegrationEvent
from spakky.event.event_publisher import IEventPublisher
from spakky.core.pod.annotations.pod import Pod

@immutable
class UserCreatedEvent(AbstractIntegrationEvent):
    user_id: int
    email: str

@Pod()
class UserService:
    def __init__(self, publisher: IEventPublisher) -> None:
        self.publisher = publisher

    def create_user(self, email: str) -> User:
        user = User(email=email)
        self.publisher.publish(UserCreatedEvent(user_id=user.id, email=email))
        return user

Event Consuming

from spakky.event.stereotype.event_handler import EventHandler, on_event

@EventHandler()
class UserEventHandler:
    def __init__(self, notification_service: NotificationService) -> None:
        self.notification_service = notification_service

    @on_event(UserCreatedEvent)
    async def on_user_created(self, event: UserCreatedEvent) -> None:
        await self.notification_service.send_welcome_email(event.email)

Async Variants

For async applications, use IAsyncEventPublisher:

from spakky.event.event_publisher import IAsyncEventPublisher

@Pod()
class AsyncUserService:
    def __init__(self, publisher: IAsyncEventPublisher) -> None:
        self.publisher = publisher

    async def create_user(self, email: str) -> User:
        user = User(email=email)
        await self.publisher.publish(UserCreatedEvent(user_id=user.id, email=email))
        return user

Distributed Tracing

spakky-tracing은 필수 의존성으로 자동 설치됩니다. ITracePropagator가 컨테이너에 등록되어 있으면 이벤트 발행/소비 시 TraceContext가 자동으로 전파됩니다.

• 발행 측: IEventTransport.send() 시 현재 TraceContext를 Kafka 메시지 헤더에 주입합니다
• 소비 측: 수신 메시지에서 TraceContext를 추출하여 자식 스팬을 생성합니다
• 헤더가 없으면 새로운 루트 트레이스를 시작합니다

Features

• Automatic topic creation: Topics are created based on event type names
• Sync and Async support: Both synchronous and asynchronous publishers/consumers
• Background service pattern: Consumer polling runs as a background service
• Pydantic serialization: Events are serialized/deserialized using Pydantic
• Confluent Kafka client: Built on the robust confluent-kafka library
• Distributed tracing: spakky-tracing integration for cross-service trace propagation

Components

Component	Description
KafkaEventTransport	Synchronous event transport (IEventTransport)
AsyncKafkaEventTransport	Asynchronous event transport (IAsyncEventTransport)
KafkaEventConsumer	Synchronous event consumer (background service)
AsyncKafkaEventConsumer	Asynchronous event consumer (background service)
KafkaConnectionConfig	Configuration via environment variables

License

MIT License