Middlewares

aiogram provides powerful mechanism for customizing event handlers via middlewares.

Middlewares in bot framework seems like Middlewares mechanism in web-frameworks (like aiohttp, fastapi, Django or etc.) with small difference - here is implemented two layers of middlewares (before and after filters).

Info

Middleware is function that triggered on every event received from Telegram Bot API in many points on processing pipeline.

Base theory

As many books and other literature in internet says:

Middleware is reusable software that leverages patterns and frameworks to bridge the gap between the functional requirements of applications and the underlying operating systems, network protocol stacks, and databases.

Middleware can modify, extend or reject processing event in many places of pipeline.

Basics

Middleware instance can be applied for every type of Telegram Event (Update, Message, etc.) in two places

  1. Outer scope - before processing filters (<router>.<event>.outer_middleware(...))
  2. Inner scope - after processing filters but before handler (<router>.<event>.middleware(...))

middlewares

(Click on image to zoom it)

Warning

Middleware should be subclass of BaseMiddleware (from aiogram import BaseMiddleware) or any async callable

Arguments specification

Argument Type Description
handler Callable[[T, Dict[str, Any]], Awaitable[Any]] Wrapped handler in middlewares chain
event T Incoming event (Subclass of TelegramObject)
data Dict[str, Any] Contextual data. Will be mapped to handler arguments

Examples

Danger

Middleware should always call await handler(event, data) to propagate event for next middleware/handler

Class-based

from aiogram import BaseMiddleware
from aiogram.api.types import Message


class CounterMiddleware(BaseMiddleware[Message]):
    def __init__(self) -> None:
        self.counter = 0

    async def __call__(
        self, 
        handler: Callable[[Message, Dict[str, Any]], Awaitable[Any]],
        event: Message,
        data: Dict[str, Any]
    ) -> Any:
        self.counter += 1
        data['counter'] = self.counter
        return await handler(event, data)

and then

router = Router()
router.message.middleware(CounterMiddleware())

Function-based

@dispatcher.update.outer_middleware()
async def database_transaction_middleware(
    handler: Callable[[Update, Dict[str, Any]], Awaitable[Any]],
    event: Update,
    data: Dict[str, Any]
) -> Any:
    async with database.transaction():
        return await handler(event, data)

Facts

  1. Middlewares from outer scope will be called on every incoming event
  2. Middlewares from inner scope will be called only when filters pass
  3. Inner middlewares is always calls for Update event type in due to all incoming updates going to specific event type handler through built in update handler