In this blog post, we’ll learn how to make SUSI AI Twitch bot. SUSI.AI is an intelligent Open Source personal assistant. SUSI AI Bots are built to enable users to chat with SUSI on different clients. Twitch is a live streaming video platform. We can integrate SUSI AI to the live chat on any of its channels.

Pre-requisites:

1. A Main Twitch account (This is the account on whose channel users would be able to talk to SUSI)
2. A separate Twitch account for SUSI.
3. GitHub Account

Create a GitHub from here – Sign up on GitHub

4. Heroku Account

To create Heroku account, go to Heroku and click on Sign Up.

5. Node.js

Install Node.js from https://nodejs.org/en/ if it isn’t installed already on your computer.

To check if node is already installed or not, open terminal and type the command:

node -v

If you see something like this – (version can be different)

v9.4.0

Then Node.js is installed on your computer and you can follow along.

Procedure:

You can either fork susi_twitchbot repository and then simply deploy it on Heroku to create SUSI AI Twitch bot or you can create a new repository on your account then deploy it. The following section will describe how to create a new repository for SUSI AI Twitch bot. If you want to deploy susi_twitchbot directly then skip this section and directly move on to the deployment section.

Creating the SUSI AI Twitch codebase:

1. Create a folder on your computer with any name. Open terminal and change your current directory to the new folder that you just created.

2. Type npm init command and enter details like name, version etc. (preferably just keep pressing enter key and let the values stay default)

3. Create a file with the same name that you wrote in entry point (index.js by default). NOTE – It should be in the same folder that you created earlier.

4. In the same folder, type the following commands in command line –

npm i -s tmi.js

We need tmi.js module to easily interact with Twitch messaging interface. For more details, visit https://www.tmijs.org/.

npm i -s request

We need request module to make a GET request to the SUSI API and retrieve JSON data for a particular query.

npm i -s express

We need express module to create successful connection with Heroku through the port provided by it.

5. Open package.json file. It should look like this:

(Adding “start” script is necessary for deploying the app on Heroku)

{
  "name": "susi_twitchbot",
  "version": "1.0.0",
  "description": "SUSI.AI Twitchbot",
  "main": "index.js",
  "scripts": {
    "start": "node index.js"
  },
  "author": "",
  "license": "ISC",
  "dependencies": {
    "express": "^4.16.3",
    "request": "^2.86.0",
    "tmi.js": "^1.2.1"
  }
}

6. Copy the following code into the file that you created i.e. index.js

var tmi = require('tmi.js');
var request = require('request');
const express = require('express');
const app = express();
const userChannel = process.env.CHANNEL;

var ans;
var options = {
        options: {
                debug: true
        },
        connection: {
                reconnect: true
        },
        identity: {
                username: process.env.USERNAME,
                password: process.env.OAUTH_TOKEN
        },
        channels: [userChannel]
};

var client = new tmi.client(options);
// Connect the client to the server
client.connect();
client.on('chat', function(channel, userstate, message, self){
        if(message.includes("@"+process.env.USERNAME)){ // checking if SUSI is tagged
                var u = message.split("@" + process.env.USERNAME + " ");
                // Setting options to make a successful call to SUSI API
                var options1 = {
                        method: 'GET',
                        url: 'http://api.susi.ai/susi/chat.json',
                        qs:
                        {
                                timezoneOffset: '-300',
                                q: u[1]
                        }
                };
                request(options1, function(error, response, body) {
                        if (error) throw new Error(error);
                        if((JSON.parse(body)).answers[0])
                                ans = userstate['display-name'] + " " + (JSON.parse(body)).answers[0].actions[0].expression;
                        else
                                ans = userstate['display-name'] + " Sorry, I could not understand what you just said."
                
                        client.action(userChannel, ans);
                });
        }
});

client.on('connected', function(address, port){
        client.action(userChannel, `Hi, I'm SUSI. Mention me using @${process.env.USERNAME} to chat with me.`);
});
const port = process.env.PORT || 3000;
app.listen(port, () => {
   console.log(`Listening on ${port}`);
});

7. Before we can deploy this bot to Heroku, we have to make a GitHub repository for it. For making a github repository for the chatbot, follow these steps:

In the command line, change current directory to the folder we created above for the bot and type in the following commands.

git init
git add .
git commit -m "initial"

Now, you have to create a Github repository, follow these steps to do that –

1. Go to https://github.com/ and login.
2. Create a new repository. Choose any name.
3. Get the URL for remote repository and copy it.

Again go to the command line, change current directory to the folder we created above for the bot and type in the following commands.

git remote add origin <URL for remote repository that you just copied>
git remote -v
git push -u origin master

Setting up SUSI bot account on Twitch:

In order to set up SUSI AI bot on Twitch, go to http://twitchapps.com/tmi, click on “Connect with Twitch” and login to the account that will be used as SUSI AI bot. This will generate an OAuth token. Save it for later use.

Deploying the Bot on Heroku:

1. Got to Heroku and login.
2. Go to dashboard and create a new app.
3. After creating an app, go to Deploy and choose “GitHub” for Deployment method.

4. Search for the repository that you created on GitHub or select susi_twitchbot after forking it. Connect to it in “App connected to GitHub”.
5. Enable Automatic deployment.

6. Now go to Settings and add the following config vars “CHANNEL”, “OAUTH_TOKEN”, “USERNAME” and “HEROKU_URL”.

• The key of “CHANNEL” is the channel name on which you’d like to talk to SUSI.
• The key of “OAUTH_TOKEN” is the OAuth key that you generated earlier. (Remember to include “oauth:” in the key)
• The key of “USERNAME” is the username of the Twitch account you created for SUSI AI.
• The key of “HEROKU_URL” is “http://<your_app_name>.herokuapp.com” (write ‘http’ and not ‘https’).

7. Go to Deploy, Press “Deploy Branch” under “Manual Deployment”.

Now your SUSI AI Twitch bot is successfully deployed on Heroku. Go to your Twitch channel and mention your SUSI account using ‘@’ and start talking to SUSI!

Here, ‘dns4044′ is the owner of channel, ‘testdns’ is some user on the channel and ‘susiaibot’ is SUSI AI Twitch bot.

References:

• Tmi.js documentation – https://docs.tmijs.org/
• Heroku documentation – https://devcenter.heroku.com/articles/getting-started-with-nodejs
• SUSI AI Twitch repository – https://github.com/fossasia/susi_twitchbot