dromzeh / rust-discord-bot

A pure-Rust serverless discord chatbot hosted on Cloudflare Workers.

About

A pure-Rust serverless discord chatbot hosted on Cloudflare Workers. With a free account you have up to 100k requests per day. For storing state you can use the bundled workers-rs crate to access KV or Durable objects.

This template is designed for compiling Rust to WebAssembly and publishing the resulting worker to Cloudflare's edge infrastructure.

Setup

1. Signup for a Cloudflare account, in the dashboard setup a subdomain (i.e <mydomain>.workers.dev)
2. Setup a worker project named bot (i.e bot.<mydomain>.workers.dev) or pick your own name and update wrangler.toml
3. Install wrangler CLI with cargo install wrangler and authenticate with cloudflare via wrangler config
4. Create a new discord app at https://discord.com/developers/applications and copy your token/application_id/public_key
5. Pass those secrets to your bot with wrangler secret put DISCORD_TOKEN, wrangler secret put DISCORD_PUBLIC_KEY, wrangler secret put DISCORD_APPLICATION_ID
6. Add bot permissions and grab your Oauth url to invite the bot to your server
7. Publish the demo app with wrangler publish. The template bot contains a single hello command with a dummy autocomplete argument.
8. Put your bot domain https://bot.<mydomain>.workers.dev in the INTERACTIONS ENDPOINT URL in your discord app page from step 4
9. After initial deployment and each time you add a new command on your bot you need to register it with the discord api. To do that simply curl -X POST https://bot.<mydomain>.workers.dev/register

You should now be able to run the /hello command on discord

Adding new commands

To add a new command simply implement the Command trait. For example to add a ping command

1. create a file src/commands/ping.rs

use crate::interaction::{
    InteractionApplicationCommandCallbackData, ApplicationCommandOption, ApplicationCommandOptionChoice, ApplicationCommandOptionType
};
use crate::error::InteractionError;
use crate::command::{Command, CommandInput};

use async_trait::async_trait;


pub(crate) struct Ping {}

#[async_trait(?Send)]
impl Command for Ping {
    async fn respond(&self, _input: &CommandInput) -> Result<InteractionApplicationCommandCallbackData, InteractionError> {
        Ok(InteractionApplicationCommandCallbackData {
            content: Some("Pong".to_string()),
            choices: None,
            embeds: None
        })
    }

    fn name(&self) -> String{
        "ping".into()
    }

    fn description(&self) -> String {
        "Send a ping".into()
    }

    fn options(&self) -> Option<Vec<ApplicationCommandOption>> {
        // add any arguments/choices here, more info at https://discord.com/developers/docs/interactions/application-commands#application-command-object-application-command-option-structure
        None
    }

    async fn autocomplete(&self, _input: &CommandInput) -> Result<Option<InteractionApplicationCommandCallbackData>, InteractionError> {
        None
    }
}

2. add your new module in src/commands/mod.rs
3. Register your command in init_commands in src/command.rs

pub(crate) fn init_commands() -> Vec<Box<dyn Command + Sync>> {
    let mut v : Vec<Box<dyn Command + Sync>> = Vec::new();
    v.push(Box::new(commands::hello::Hello {}));
    // Add this line
    v.push(Box::new(commands::ping::Ping {}));
    v
}

4. publish your package with wrangler publish
5. register your new command with discord with curl -X POST http://bot.<mydomain>.workers.dev/register

You can store and access state using the input context object passed to the respond and autocomplete methods, for example:

let my_val = input.kv_get("my_namespace", "my_key").await?; // the namespace must be first registered on cloudflare dashboard
input.kv_put("my_namespace", "foo", "bar").await?;

Local Dev

With wrangler, you can build, test, and deploy your Worker with the following commands:

# compiles your project to WebAssembly and will warn of any issues
wrangler build 

# run your Worker in an ideal development workflow (with a local server, file watcher & more)
wrangler dev

# deploy your Worker globally to the Cloudflare network (update your wrangler.toml file for configuration)
wrangler publish

you can use ngrok to tunnel traffic into your local machine, more info here

Actions Deployments

You can create an a action to automatically deploy your worker & register your commands on each push to main.

Create a Cloudflare API token and use the Edit Cloudflare Workers template.

Then, create a repository secret with your API Token under CF_API_TOKEN and add the following inside .github/workflows/deploy.yml:

name: Deploy to Cloudflare Workers

on:
  push:
    branches:
      - main

jobs:
    deploy:
        runs-on: ubuntu-latest
        steps:
            - name: Checkout Repository
            - uses: actions/checkout@v3

            - name: Deploy to Cloudflare Workers
              env:
                  CF_API_TOKEN: ${{ secrets.CF_API_TOKEN }}
              run: npm i -g wrangler && npx wrangler publish

            - name: Curl the worker
              run: curl -X POST https://<your_worker_domain>/register

WebAssembly

workers-rs (the Rust SDK for Cloudflare Workers used in this template) is meant to be executed as compiled WebAssembly, and as such so must all the code you write and depend upon. All crates and modules used in Rust-based Workers projects have to compile to the wasm32-unknown-unknown triple.

Read more about this on the workers-rs project README.

Credits

based on stateless-discord-bot