Core interface for components that manage events in the library.

Module Contents#

class hikari.api.event_manager.EventManager[source]#

Bases: abc.ABC

Base interface for event manager implementations.

This is a listener of a hikari.events.base_events.Event object and consumer of raw event payloads, and is expected to invoke one or more corresponding event listeners where appropriate.

abstract consume_raw_event(event_name, shard, payload)[source]#

Consume a raw event.


The case-insensitive name of the event being triggered.


Object of the shard that received this event.


Payload of the event being triggered.


If there is no consumer for the event.

abstract dispatch(event)[source]#

Dispatch an event.


The event to dispatch.


A future that can be optionally awaited. If awaited, the future will complete once all corresponding event listeners have been invoked. If not awaited, this will schedule the dispatch of the events in the background for later.


We can dispatch custom events by first defining a class that derives from hikari.events.base_events.Event.

import attrs

from hikari.traits import RESTAware
from hikari.events.base_events import Event
from hikari.users import User
from hikari.snowflakes import Snowflake

class EveryoneMentionedEvent(Event):
    app: RESTAware = attrs.field()

    author: User = attrs.field()
    '''The user who mentioned everyone.'''

    content: str = attrs.field()
    '''The message that was sent.'''

    message_id: Snowflake = attrs.field()
    '''The message ID.'''

    channel_id: Snowflake = attrs.field()
    '''The channel ID.'''

We can then dispatch our event as we see fit.

from hikari.events.messages import MessageCreateEvent

async def on_message(event):
    if "@everyone" in event.content or "@here" in event.content:
        event = EveryoneMentionedEvent(


This event can be listened to elsewhere by subscribing to it with EventManager.subscribe.

async def on_everyone_mentioned(event):
    print(event.user, "just pinged everyone in", event.channel_id)
abstract get_listeners(event_type, /, *, polymorphic=True)[source]#

Get the listeners for a given event type, if there are any.


The event type to look for. T must be a subclass of hikari.events.base_events.Event.


If True, this will also return the listeners for all the event types event_type will dispatch. If False, then only listeners for this class specifically are returned. The default is True.

typing.Collection[typing.Callable[[T], typing.Coroutine[typing.Any, typing.Any, None]]]

A copy of the collection of listeners for the event. Will return an empty collection if nothing is registered.

abstract listen(*event_types)[source]#

Generate a decorator to subscribe a callback to an event type.

This is a second-order decorator.


The event types to subscribe to. The implementation may allow this to be undefined. If this is the case, the event type will be inferred instead from the type hints on the function signature. T must be a subclass of hikari.events.base_events.Event.

typing.Callable[[T], T]

A decorator for a coroutine function that passes it to EventManager.subscribe before returning the function reference.

abstract stream(event_type, /, timeout, limit=None)[source]#

Return a stream iterator for the given event and sub-events.


If you use await stream.open() to start the stream then you must also close it with await stream.close() otherwise it may queue events in memory indefinitely.


The event type to listen for. This will listen for subclasses of this type additionally.

timeouttyping.Optional[int, float]

How long this streamer should wait for the next event before ending the iteration. If None then this will continue until explicitly broken from.


The limit for how many events this should queue at one time before dropping extra incoming events, leave this as None for the cache size to be unlimited.


The async iterator to handle streamed events. This must be started with with stream: or stream.open() before asynchronously iterating over it.


with bot.stream(events.ReactionAddEvent, timeout=30).filter(("message_id", message.id)) as stream:
    async for user_id in stream.map("user_id").limit(50):

or using open() and close()

stream = bot.stream(events.ReactionAddEvent, timeout=30).filter(("message_id", message.id))

async for user_id in stream.map("user_id").limit(50)

abstract subscribe(event_type, callback)[source]#

Subscribe a given callback to a given event type.


The event type to listen for. This will also listen for any subclasses of the given type. T must be a subclass of hikari.events.base_events.Event.


Must be a coroutine function to invoke. This should consume an instance of the given event, or an instance of a valid subclass if one exists. Any result is discarded.


The following demonstrates subscribing a callback to message creation events.

from hikari.events.messages import MessageCreateEvent

async def on_message(event):

bot.subscribe(MessageCreateEvent, on_message)
abstract unsubscribe(event_type, callback)[source]#

Unsubscribe a given callback from a given event type, if present.


The event type to unsubscribe from. This must be the same exact type as was originally subscribed with to be removed correctly. T must derive from hikari.events.base_events.Event.


The callback to unsubscribe.


The following demonstrates unsubscribing a callback from a message creation event.

from hikari.events.messages import MessageCreateEvent

async def on_message(event):

bot.unsubscribe(MessageCreateEvent, on_message)
abstract async wait_for(event_type, /, timeout, predicate=None)[source]#

Wait for a given event to occur once, then return the event.


Async predicates are not supported.


The event type to listen for. This will listen for subclasses of this type additionally.


A function taking the event as the single parameter. This should return True if the event is one you want to return, or False if the event should not be returned. If left as None (the default), then the first matching event type that the bot receives (or any subtype) will be the one returned.

timeouttyping.Union[float, int, None]

The amount of time to wait before raising an asyncio.TimeoutError and giving up instead. This is measured in seconds. If None, then no timeout will be waited for (no timeout can result in “leaking” of coroutines that never complete if called in an uncontrolled way, so is not recommended).


The event that was provided.


If the timeout is not None and is reached before an event is received that the predicate returns True for.

class hikari.api.event_manager.EventStream[source]#

Bases: hikari.iterators.LazyIterator[hikari.events.base_events.EventT], abc.ABC

A base abstract class for all event streamers.

Unlike hikari.iterators.LazyIterator (which this extends), an event streamer must be started and closed.

See also




A streamer may either be started and closed using with syntax where EventStream.open and EventStream.close are implicitly called based on context.

with EventStream(app, EventType, timeout=50) as stream:
    async for entry in stream:

A streamer may also be directly started and closed using the EventStream.close and EventStream.open. Note that if you don’t call EventStream.close after opening a streamer when you’re finished with it then it may queue events events in memory indefinitely.

stream = EventStream(app, EventType, timeout=50)
await stream.open()
async for event in stream:

await stream.close()
abstract close()[source]#

Mark this streamer as closed to stop it from queueing and receiving events.

If called on an already closed streamer then this will do nothing.


with streamer may be used as a short-cut for opening and closing a streamer.

abstract filter(*predicates, **attrs)[source]#

Filter the items by one or more conditions.

Each condition is treated as a predicate, being called with each item that this iterator would return when it is requested.

All conditions must evaluate to True for the item to be returned. If this is not met, then the item is discarded and ignored, the next matching item will be returned instead, if there is one.

*predicatestyping.Union[typing.Callable[[ValueT], bool], typing.Tuple[str, typing.Any]]

Predicates to invoke. These are functions that take a value and return True if it is of interest, or False otherwise. These may instead include 2-tuple objects consisting of a str attribute name (nested attributes are referred to using the . operator), and values to compare for equality. This allows you to specify conditions such as members.filter(("user.bot", True)).


Alternative to passing 2-tuples. Cannot specify nested attributes using this method.


The current stream with the new filter applied.

abstract open()[source]#

Mark this streamer as opened to let it start receiving and queueing events.

If called on an already started streamer then this will do nothing.


with streamer may be used as a short-cut for opening and closing a stream.