fast-channels 1.0.0

Fast Channels brings Django Channels–style consumers and channel layers to FastAPI and Starlette for real-time apps.

Fast Channels brings Django Channels–style consumers and channel layers to FastAPI, Starlette, and any ASGI-compatible framework for real-time apps.

What are Channel Layers?

Channel layers enhance WebSocket functionality by enabling advanced messaging patterns beyond simple request-response:

• Group Messaging: Send messages to multiple WebSocket connections simultaneously

• Cross-Instance Communication: Send messages from HTTP endpoints, background workers, or other processes

• Task Distribution: Use as a basic task queue for offloading work to worker processes

• Distributed Applications: Build scalable real-time apps without routing everything through a database

As the Django Channels documentation explains: “Channel layers allow you to talk between different instances of an application. They’re a useful part of making a distributed realtime application if you don’t want to have to shuttle all of your messages or events through a database.”

Fast Channels ports this proven architecture from Django Channels, which has been battle-tested in production environments for years. By leveraging Redis for high-performance, reliable message delivery while maintaining the familiar consumer-based programming model, Fast Channels brings Django’s mature WebSocket patterns to the entire ASGI ecosystem.

Features

• Django Channels-Style Consumers: Familiar WebSocket consumer patterns with connect, receive, and disconnect methods

• Channel Layers: Support for Redis-backed channel layers with group messaging

• Redis Integration: Full Redis support with Redis Sentinel for high availability

• Async/Await Support: Built from the ground up for modern Python async programming

• ASGI Compatible: Seamless integration with FastAPI, Starlette, and any ASGI-compatible framework

• Testing Framework: Comprehensive testing utilities for WebSocket consumers

• Full Type Safety: Complete type hints with mypy and pyright support

• Multiple Channel Layer Backends: In-memory for development, Redis for production

Installation

# Basic installation
pip install fast-channels

# Recommended: Install with Redis support for production use
pip install fast-channels[redis]

Optional Dependencies:

# For Redis channel layer support (recommended for production)
pip install fast-channels[redis]

Note: Additional channel layer backends (such as Kafka) may be supported in future releases.

Core Dependencies: Python 3.11+, compatible with any ASGI framework (FastAPI, Starlette, Quart, etc.)

Quick Start

1. Register Channel Layers

# layers.py
import os
from fast_channels.layers import (
    InMemoryChannelLayer,
    RedisChannelLayer,
    RedisPubSubChannelLayer,
    register_channel_layer,
)

def setup_layers():
    # Redis URL from environment
    redis_url = os.getenv("REDIS_URL", "redis://localhost:6379")

    # Register different layer types
    register_channel_layer("memory", InMemoryChannelLayer())
    register_channel_layer("chat", RedisPubSubChannelLayer(hosts=[redis_url], prefix="chat"))
    register_channel_layer("queue", RedisChannelLayer(
        hosts=[redis_url],
        prefix="queue",
        expiry=900,  # 15 minutes
        capacity=1000,
    ))

2. Create WebSocket Consumer

# consumer.py
from fast_channels.consumer.websocket import AsyncWebsocketConsumer

class ChatConsumer(AsyncWebsocketConsumer):
    groups = ["chat_room"]
    channel_layer_alias = "chat"  # Use registered layer

    async def connect(self):
        await self.accept()
        await self.channel_layer.group_send(
            "chat_room",
            {"type": "chat_message", "message": "Someone joined the chat"}
        )

    async def disconnect(self, close_code):
        await self.channel_layer.group_send(
            "chat_room",
            {"type": "chat_message", "message": "Someone left the chat"}
        )

    async def receive(self, text_data=None, bytes_data=None, **kwargs):
        await self.channel_layer.group_send(
            "chat_room",
            {"type": "chat_message", "message": f"Message: {text_data}"}
        )

    async def chat_message(self, event):
        # Send message to WebSocket
        await self.send(event["message"])

3. Integrate with FastAPI

# main.py
from fastapi import FastAPI
from .layers import setup_layers
from .consumer import ChatConsumer

# Setup layers before creating the app
setup_layers()

app = FastAPI()

# Create WebSocket sub-app for better organization
ws_app = FastAPI()
ws_app.add_websocket_route("/chat", ChatConsumer.as_asgi())

# Mount WebSocket routes
app.mount("/ws", ws_app)

Channel Layer Backends

In-Memory (Testing Only)

• Fast and simple for unit tests

• Single-process only - cannot send messages from workers or HTTP endpoints

• No persistence

• Use only for testing group chat functionality or running test suites

Redis Queue Layer (Production - Reliable)

• Message persistence and guaranteed delivery

• Scalable across multiple processes

• Configurable expiry and capacity

• Best for critical messaging

Redis Pub/Sub Layer (Production - Real-time)

• Ultra-low latency messaging

• Real-time broadcasting

• No message persistence

• Best for live chat and notifications

Registration Examples:

from fast_channels.layers import (
    InMemoryChannelLayer,
    RedisChannelLayer,
    RedisPubSubChannelLayer,
    register_channel_layer,
)

# In-memory for development
register_channel_layer("memory", InMemoryChannelLayer())

# Redis Queue Layer for reliable messaging
register_channel_layer("reliable", RedisChannelLayer(
    hosts=["redis://localhost:6379"],
    prefix="app_queue",
    capacity=1500,
    expiry=3600,  # 1 hour
))

# Redis Pub/Sub for real-time chat
register_channel_layer("chat", RedisPubSubChannelLayer(
    hosts=["redis://localhost:6379"],
    prefix="app_chat",
))

# Redis Sentinel for high availability
register_channel_layer("ha_queue", RedisChannelLayer(
    sentinels=[("localhost", 26379)],
    service_name="mymaster",
    sentinel_kwargs={"password": "sentinel_password"},
    connection_kwargs={"password": "redis_password"},
))

Testing

Fast Channels includes comprehensive testing utilities out of the box:

from fast_channels.testing import WebsocketCommunicator
import pytest

@pytest.mark.asyncio
async def test_chat_consumer():
    communicator = WebsocketCommunicator(ChatConsumer, "/ws/chat/")
    connected, subprotocol = await communicator.connect()
    assert connected

    # Test sending a message
    await communicator.send_json_to({
        "message": "hello world"
    })
    response = await communicator.receive_json_from()
    assert response == {"message": "hello world"}

    await communicator.disconnect()

Documentation

Please visit Fast Channels docs for complete documentation, including:

• Detailed consumer patterns

• Advanced channel layer configuration

• Production deployment guides

• Testing best practices

• Migration guides from Django Channels

Comparison with Alternatives

Fast Channels vs. Native FastAPI/Starlette WebSockets

Native FastAPI WebSocket support provides basic connection handling but lacks advanced messaging capabilities:

# Native FastAPI - Limited to direct connections
@app.websocket("/ws")
async def websocket_endpoint(websocket: WebSocket):
    await websocket.accept()
    # Can only send/receive to this specific connection
    # No group messaging or cross-process communication

Fast Channels vs. Broadcaster

Broadcaster is a lightweight pub/sub library, but Fast Channels provides more comprehensive functionality:

Feature	Native WS	Broadcaster	Fast Channels
Basic WebSocket	✅	✅	✅
Simple Pub/Sub	❌	✅	✅
Group Messaging	❌	✅	✅
Consumer Pattern	❌	❌	✅
Message Persistence	❌	❌	✅ (Redis Queue)
Testing Framework	❌	❌	✅
Connection Management	Manual	Manual	Automatic
Type Safety	Manual	Basic	Full
Background Worker Msgs	❌	❌	✅
Structured Event System	❌	❌	✅

Contributing

Contributions are welcome! Please see our CONTRIBUTING.md for detailed development setup and guidelines.

License

This project is licensed under the MIT License - see the LICENSE file for details.