It is a minimal async/sync event processing framework. Has no dependencies and uses nothing except pure Python 3.8.

TL;DR Do not have much time? See recap on examples and recap on conditions. Now go and use aioevproc! :)

Simplest example for a single async handler, just echo the message text:

from aioevproc import EventsProcessor, handler, Event

class EchoTelegramBot(EventsProcessor):
    @handler(lambda event: 'message' in event and 'text' in event['message'])
    async def echo_message(self, event: Event) -> None:
        await self.reply_to_message(text=event['message']['text'])

A little bit more complex Telegram bot example, see the explanation below:

from aioevproc import EventsProcessor, handler, Event
from contextlib import asynccontextmanager, contextmanager

class TelegramBot(EventsProcessor):
    # synchronous middleware for any exception: log exception
    @handler()
    @contextmanager
    def log_exception(self, event: Event) -> Generator[None, None, None]:
        try:
            yield
        except:
            logging.exception('Error!')

    # async middleware for any exception: send excuse message to the user
    @handler()
    @asynccontextmanager
    def send_excuse_message(self, event: Event) -> AsyncGenerator[None, None]:
        try:
            yield
        except:
            await self.send_message('Sorry!')

    # synchronous handler for all updates: log message
    @handler()
    def log_update_id(self, event: Event) -> Literal[True]:
        logging.info(event['update_id'])
        return True  # call following handlers

    # async handler to check if user is admin for update with messages and cb
    @handler(lambda event: 'message' in event or 'callback_query' in event)
    async def check_admin(self, event: Event) -> bool:
        # the next handler will be called only if this returns True
        return event['message']['from_user']['id'] in await self.get_admins()
 
    # async handler to echo updates containing a message
    @handler(lambda event: 'message' in event and 'text' in event['message'])
    async def echo_message(self, event: Event) -> None:
        # if the update contains a message then echo it
        await self.reply_to_message(text=event['message']['text'])

    # async handler to answer a callback query
    @handler(lambda event: 'callback_query' in event)
    async def echo_message(self, event: Event) -> None:
        # if the update does not contain a message but a callback query, answer
        await self.answer_callback_query(event['callback_query']['id'])

handler decorates methods of EventsProcessor subclasses. The method can be one of: async function (like check_admin, handle_message and echo_message in the example above), sync function (log_update_id), async context manager (send_excuse_message) or sync context manager (log_exception).

All of the handlers are called in the same order as they are declared in the class body. Middlewares follow the same rule: they are entered in the order of declaration and exited in the reversed order (in a recursive manner).

Sync and async handlers may return a value: if it is not a truthy value then none of the following handlers will be called and event processing will be stopped at the handler which did not return truthy value.

Please notice: if you return nothing from the sync/async handler method (means you implicitly return None) then none of the following handlers will be called. This is an intended default behavior since usually an event requires a single handler. None is a falsy (not truthy) value.

Returning True from the handler is useful for logging purposes: the logging method should not block further processing of the event. This is shown in the example below (log_update_id) as well as the filtering use case for admins: if the user is not an admin then check_admin will return False and no further processing will be done.

Middlewares are based on context managers and are intended to be used for exceptions handling. Also use them when some related actions are required before and after the event is processed by other handlers: for example, for measuring the execution time.

Let's sum up on the examples:

1. aioevproc supports both sync and async handlers and middlewares.
2. Every handler or middleware has to be a method of EventsProcessor subclass.
3. If the handler does not return a truthy value then the following handlers are not called.
4. Middlewares are sync/async context managers.
5. Handlers and middlewares are called in the same order as they are declared.

Handler usually has to be applied to certain types of events, not all. The following handler will be applied only to updates containing a message:

@handler(lambda event: 'message' in event)
async def handle_event(self, event: Event) -> None:
    pass

If the condition check fails then the next handler condition will be checked:

@handler(lambda event: False)
def always_skipped(self, event: Event) -> Literal[False]:
    # this handler is never called since its predicate always evaluates to False
    return False  # has no effect since this handler is not called

# since previous handler condition check failed this one will be checked next
@handler(lambda event: 'edited_message' in event)
def log_message_edit(self, event: Event) -> None:
    pass

Please notice: if the handler condition check failed then the handler's return value does not affect the next handlers. The return value of the handler affects the next handlers only if the handler itself is called (meaning that its condition check is passed).

You can specify multiple predicates in a handler call: this will make handler to be called only if all of the predicates evaluate to a truthy value for the event. Example below shows the handers which will be applied only to updates with text messages:

@handler(
        lambda event: 'message' in event,
        lambda event: 'text' in event['message'],
)
async def handle_event(self, event: Event) -> None:
    pass

The predicates are evaluated in the same order as they are declared. So the above pair of conditions is equivalent to 'message' in event and 'text' in event['message']. This means that specifying multiple predicates for a single handler call implements AND semantics (conditions conjunction).

If you need to apply single handler if any of the conditions is true, use multiple handler calls:

@handler(lambda event: 'message' in event)
@handler(lambda event: 'callback_query' in event)
async def handle_event(self, event: Event) -> None:
    pass

This will apply the handler for either update with a message or update with a callback query. This form implements OR semantics (conditions disjunction).

Please notice: the implementation of aioevproc checks handlers predicates in the same order as they are declared. First 'message' in event will be checked and after that the 'callback_query' in event predicate will be evaluated. This is a reversed order to how Python applies decorators: Python applies the most inner decorator first. But aioevproc applies the most outer handler call first since it is more intuitive.

If you need a handler to be applied unconditionally then use just handler() without arguments.

Please notice: you cannot use handler() without arguments on a handler with any other handler call with arguments since this has no sense:

@handler()  # will raise an AssertionError
@handler(lambda event: 'message' in event)
async def handle_event(self, event: Event) -> None:
    pass

Don't forget to return True from unconditionally applied handler to not ignore all of the following handlers!

Let's sum up on conditions:

1. Single handler call accepts multiple predicates as arguments. The handler then will be called only if all of the predicates are true (AND semantics).
2. If a handler method (or middleware) is decorated with multiple handler calls then the handler will be called if any of the handlers' conditions is true (OR semantics).
3. OR and AND semantics can be combined.
4. If the handler's conditions check failed then the handler is skipped and the next handlers' conditions are checked until the matching handler is found.
5. All the conditions are checked in the same order as they are declared. The most outer handler decorator is applied first.
6. Handler decorated with handler() w/o arguments is applied unconditionally.

pip install aioevproc

From project root directory: python -m unittest discover -s tests/unit