A set of tools for building a discord bot with a serverless architecture in mind. If you are looking for a more complete discord sdk, check out discord.py.
Install with pip: pip install discord_sls
The library provides a decorator @bot_handler which can be used to decorate a lambda handler to respond to discord api requests.
It will handle the ping-pong handshake, and will parse the request body into a python object for you to use. The decorator expects you to return Ineraction Callback Data.
Discord expects a quick response to the initial request, if your bot needs longer to handle an interaction you can use the send_command_to_queue function to send the command to a queue for processing in another lambda decorated with @deferred_response_handler. The queue is determined by the LONG_RESPONSE_QUEUE environment variable.
import json
import logging
from discord_sls import Interaction, bot_handler, deferred_response_handler
@bot_handler
def discord_bot(command_body, send_command_to_queue):
command_name = command_body.get("data", {}).get("name")
if command_name == "hello":
return {"content": "Hello Moto"}
elif command_name == "helloasync":
send_command_to_queue(command_body)
return {"content": "Hello..."}
else:
logging.warn(f"unhandled command: {command_name}")
return {"content": "Unknown Command"}
@deferred_response_handler
def long_response_handler(interaction: Interaction):
interaction.follow_up({"content": "Hello...async"})With most serverless architectures you will need to keep your lambdas warm to avoid cold start times. A popular mechanism for doing that is using cloudwatch event rules, the @bot_handler decorator will automatically handle these requests for you.T his is what a SAM template could look like for the rule:
BotKeepWarm:
Type: AWS::Events::Rule
Properties:
Description: Keeps the bot lambda warm
Name: !Sub 'keep-warm-${Stage}'
ScheduleExpression: rate(5 minutes)
Targets:
- Id: KeepWarmDiscordBot
Arn: !GetAtt DiscordBotFunction.Arn