EDIT: As of 2023-02-09, Twitter's API is no longer Free. This article is no longer an effective way to learn coding or develop new skills. I am so sad because Twitter bots are one of the first ways I had fun writing code.
This article will stay up only because it's my first, and it marks an important part of my career change.
RIP Shania Bot.
- Pj
“TWITTER BOTS ARE RUINING TWITTER” is a very accurate, very true statement. I’ve had countless people instinctively flinch when I tell them I make Twitter bots. They’re often responsible for spreading disinformation, hurting artists’ merch sales, and basically ruining the human experience of the bird site.
So why are we making one? To make a nice bot, of course!
YOUR bot isn’t going to sow discord, or do anything malicious, right?
Answer me.
Ok, you’re good. So, let’s talk about how to do this and why. Twitter bots are a great way to practice using an API and developing some coding skills along the way. You can usually get the bot going for under 100 lines of code, so it’s especially good for beginners. In this tutorial, I’ll show you how to use Node.js and a few npms to have your bot tweeting positivity and fun to counteract all those other nasty bots.
We’ll be using:
Visual Studio Code
Node.js
Twit
node-schedule
DotEnv
Twitter
Twitter Developer portal
GitHub & GitHub desktop
Linode
Setting up a Twitter Account
This is where you decide what you want your account to be named and what it will do automatically. This bot will tweet a link to a video game soundtrack once a week (pulled randomly from an array of links) and will respond to a user who asks for a link with a random link from the same array. We’ll call our bot @SNESSoundtracks. Go to Twitter and follow the instructions for making a brand new account for your bot.
Twitter Developer Portal
The developer portal is where you register your new account to be able to use the Twitter API. This is an important step because it unlocks the accounts ability to tweet by using the API with code instead of manually tweeting. You know, the whole point of a bot?
So here’s the process, you’ll head to the developer page. Once there you’ll see a link near the profile picture on the top right that says “apply.”
Click that, and it’ll take you to a page where you apply to have your Twitter profile become an active developer profile. Follow the instructions on screen until you have an authenticated developer account. Once that’s done, create the app and fill out the descriptions for your app. Really, all you have to say is that you’re doing it as a hobby and it’s just for learning to code or practicing a bot. You’re not giving info to the government or really consuming streaming data for analytical purposes. When you’ve finished, the developer portal should look like this.
The Keys and the Code and the Node Package Manager
Ok, so you’ve created the app, verified your email, and now you’re reaching for the keys. I’m gonna stop you right there.
It’s easier to get the keys after you have somewhere to put them. We’re gonna be keeping them in a dotenv file for safekeeping from people who would try to get access to your keys. So let’s get started on the code and come back to the keys.
The easiest way I’ve found to start a new project is to create a new repo on GitHub.com and then pull it locally using GitHub desktop. To do this, start a new repository on github.com and make sure to create a readme, a license for use, and to include a .gitignore file for node. Then click the green dropdown menu button labeled "Code" to the right side of your screen. If you have Github desktop installed, you can click the option to Open with Github Desktop. Then you can choose where to save your repo locally. Now your source control is on your machine and on Github.
We’re gonna create some variables for these keys you’ll be copying from the Twitter developer page later. You’ll have four keys you need in order to connect your app to the code you’ve written. Create a file in your project called “.env”. Inside, add four variables: access_token, access_token_secret, consumer_secret, consumer_key. The consumer keys will pair with the API keys.
Something important about a dotenv file, you won’t need quotes around the values of your variables, as long as there aren’t spaces. Make sure to write it like this: “consumer_key=whateveryourkeyis”. Your dotenv file is going to be your one source of truth for the keys. The config file will point to the .env values, and because you created a gitignore, you won’t upload your keys to GitHub.
Up next, we’re going to install all the npm packages we need, so enter the following into your terminal in VSCode.
"Npm init -y"
"Npm install node"
"Npm install node-schedule"
"Npm install twit"
"Npm install dotenv"
This installs all the dependencies we’ll need to be able to have a functional Twitter bot that can read the API, post on a schedule, use the .env file, and use node.js. We have one more thing to adjust, and that’s to create some scripts inside the package.json file code block. Open it and add the following to the scripts section.
"test": "index.js",
"start": "node ./index.js",
"develop": "NODE_ENV=develop node index.js"
That index.js file doesn’t exist yet, so go ahead and create it and name it whatever you like. I’ll be calling mine snes.js. Go ahead and make a config.js at the same time and input the following into it.
module.exports = {
consumer_key: process.env.consumer_key,
consumer_secret: process.env.consumer_secret,
access_token: process.env.access_token,
access_token_secret: process.env.access_token_secret,
};
Let’s write some code.
Code It Up.
console.log("SNES Soundtracks booting up");
//making sure npm run develop works
if (process.env.NODE_ENV === "develop") {
require("dotenv").config();
};
//rules for node-schedule
var schedule = require("node-schedule");
var rule = new schedule.RecurrenceRule();
rule.dayOfWeek = 1,
rule.hour = 10;
rule.tz = "Etc/GMT+4";
//array to pull soundtracks from
var soundtrackArray = [ "an array of youtube video URLs"];
var soundtrackArrayLength = soundtrackArray.length;
var soundtrackArrayElement = Math.floor(Math.random() * soundtrackArrayLength);
At the beginning, I log a start up message to the console just so I know it’s running. Next is an if statement to use the dotenv when the node environment is ‘develop’, which is handled in the scripts of the json file from earlier. We set up a few variables for the node-schedule so the bot can tweet on a set day and time. I pulled this directly from the docs for node-schedule. Basically, it will tweet every Monday at 10 am Eastern Daylight Savings Time. Finally, I set up an array for the bot to pull from at random with the last three lines of this section. I removed the URLs in the interest of length.
EDIT: I DISCOVERED A BUG HERE. By running the random number here in the code, I create a problem whereby the random number is selected at runtime rather than every time the Bot fires up. This meant the bot was tweeting the same soundtrack every time it tweeted. To fix this, I moved var soundtrackArrayElement = Math.floor(Math.random() * soundtrackArrayLength);
inside the two functions, pressStart and pressSelect. This allows the number to be chosen randomly upon the function running, rather than when the bot is first run. Now, back to the code!
// Create a Twitter object to connect to Twitter API
var Twit = require('twit');
// Pulling keys from another file
var config = require('./config.js');
// Making a Twit object for connection to the API
var T = new Twit(config);
// Setting up a user stream
var stream = T.stream('statuses/filter', { track: '@SnesSoundtracks' });
// Now looking for tweet events
// See: https://dev.Twitter.com/streaming/userstreams
stream.on('tweet', pressStart);
This is where we start using Twit. We create an object called Twit that requires the npm, and then pass a configure that requires the configuration file into the object. We then use new to create “T”, an instance of the object from before. From now on, when we want to use something from Twit, we simply use T.whatever in order to call up the property, field, or method we need from their library. We set up a stream so that we are monitoring specifically @SnesSoundtracks while the code is running. Finally, we create an event listener with stream.on, and use a string parameter to name it, and input a function we’ve called, “pressStart”. pressStart is defined in the next set of code.
function pressStart(tweet) {
var id = tweet.id_str;
var text = tweet.text;
var name = tweet.user.screen_name;
let regex = /(please)/gi;
let playerOne = text.match(regex) || [];
let playerTwo = playerOne.length > 0;
//this helps with errors, so you can see if the regex matched and if playerTwo is true or false
console.log(playerOne);
console.log(playerTwo);
// checks text of tweet for mention of SNESSoundtracks
if (text.includes('@SnesSoundtracks') && playerTwo === true) {
// Start a reply back to the sender
var replyText = ("@" + name + " Here's your soundtrack!" + soundtrackArray[soundtrackArrayElement]);
// Post that tweet
T.post('statuses/update', { status: replyText, in_reply_to_status_id: id }, gameOver);
} else {
console.log("uh-uh-uh, they didn't say the magic word.");
};
function gameOver(err, reply) {
if (err) {
console.log(err.message);
console.log("Game Over");
} else {
console.log('Tweeted: ' + reply.text);
}
};
}
pressStart contains a few local variables, a bit of logic, and a final function that must be included in the T.post method. You can use an unnamed function there and it will do the same thing, but I went ahead and wrote on separately for readability. Essentially, the function gameOver gives us a chance to log an error if it occurs or to log the tweet that was sent out.
pressStart takes “tweet” as a parameter. This is the tweet that another user writes that tags SnesSoundtracks. That tweet has tons of data attached to it, data that Twit helps us parse through. The first three variables are the id of the tweet, the text of the tweet, and the username of the person who wrote the tweet. We will need those three in order to respond accurately as a comment to the original tweet by the other user.
Up next is a regex for whatever word you want to activate the bot to reply. I chose “please,” so that as long as the user is polite, they’ll get a random soundtrack.
The regex has “g” and “i” at the end so it ignores capitalization and checks globally for the word please. playerOne is a variable that can either be an empty array or will use .match to create an array with one element, the word “please”. playerTwo is a boolean that verifies whether the array playerOne has an element or not.
The logic dictates that the tweet text contains the bot’s name and an array of at least one element was passed into playerTwo. If both of these come back as true, then we proceed to an area where the variable replyText is created, which includes a random element of the array, as well as the username of the person being replied to and a short message. replyText is passed into an object that contains two properties: status and in_reply_to_status_id. Status is the actual text to be posted to the tweet, in our case the variable replyText is our status. In_reply_to_status_id is defined as id, which is a variable from the beginning of the pressStart function. Id is a unique identifier of a tweet from Twitter’s API. This allows Twit to identify which tweet the bot will reply to as a comment. Finally, the else statement at the end will log a quote from Jurassic Park to the console if the user doesn’t say please. I thought about having the bot tweet this to the user but decided against it. Instead, it’s a little fun just for me.
function pressSelect() {
var weeklyReplyText = soundtrackArray[soundtrackArrayElement] + " Here's your soundtrack for the week!";
T.post('statuses/update', { status: weeklyReplyText }, gameOver2);
function gameOver2(err, reply) {
if (err) {
console.log(err.message);
console.log("Game Over");
} else {
console.log('Tweeted: ' + reply.text);
}
}
}
const job1 = schedule.scheduleJob(rule, pressSelect);
job1.on("Every Day Tweet", pressSelect);
Here is the function used to tweet on a schedule, which I’ve named pressSelect. pressSelect has the replyText variable, slightly changed to be a tweet rather than a comment, but uses the same array to pull from. The gameOver function is also present, though renamed just to be safe. Since gameOver and gameOver2 are local variables within their respective functions, there shouldn’t be any issues. However, they both do the same thing.
The final part of this code is creating a variable called job1. job1 is the scheduleJob method from the node-schedule object at the top of the code. I pass in the rule created and pressSelect as parameters. We then use an event listener with pressSelect passed in again.
Running the Code
To test your code and ensure it works, type “npm run develop” into the terminal. If you get a Twit error about consumer keys, ensure there are no spaces between the variable, equals sign, and key itself in your .env file. If the error persists, you may have copied your keys wrong. You can always generate them again and copy them directly into the .env file. If you’d like to test pressSelect on its own and make sure it works, you can just comment out the last two lines of the code and call pressSelect directly. This way, you don’t have to wait for whatever day and hour you scheduled node-schedule for.
Once it’s running, to test the way it responds to other users, log in to another Twitter account and tweet at your bot. You should be able to see some action in the terminal that tells you it’s working, followed by the response on Twitter.
Hosting Your Code
You could certainly let this code run for a long while from your local machine, but it’d be better if you had it hosted somewhere else. One great option is Linode.
After signing up for a Linode account, the home page will look like this.
Create a Linode, and pick Ubuntu 20.14 from the first dropdown menu. Select the region closest to you and then choose Nanode, the smallest option. It’s only $5 a month and has more than enough space and RAM for your Twitter bot and any other projects you’d want on it. When it asks for an Ubuntu label, name it after your project. Create a root password and click create. Make sure to remember the root password as you’ll have to use it every time to get into the Linode.
Once it boots up it’s time to get inside and clean up a bit.
Installing, Updating, and Preparing for the bot.
Linode provides a LISH console to use to access your Linode. On the top right of the Linode homepage, you’ll see a link for “Launch LISH Console.”
Click that and you have access to the Ubuntu instance you just created. You can also SSH into it from any terminal using the SSH access text just below the LISH.
Once you’re in, you should run “apt update” to check for updates available, and then “apt dist-upgrade.” It's recommended to make a new username for Linux so you're not running everything through the root, but in this case, all you're doing is a Twitter bot, so it's not hugely important. If you'd like to do it the right way, you can run make a new user and give it admin rights by following this video.
If your code is on Github, you can use git to clone your file, or you can use the GUI interface of Filezilla to transfer your code to the Linux instance. Since this is a beginner tutorial, let’s use Filezilla to transfer our files.
Using Filezilla
Here's how Filezilla looks once you open it.
Inside Filezilla, you can see input fields for Host, Username, Password, and Port. Host is the IP Address for your nanode, username should be "root" or whatever username you set up earlier, and password is what you set at the beginning. Use “22” as the port number. Upon connecting, you’ll get a message saying the key is unknown. This is normal, but you should double-check that the IP Address is correct. If it is, check the box to remember your Linode on Filezilla.
Go to the directory where you’re keeping the files for your Twitter bot. You’ll need the .env, config.js, the package.json, package-lock.json, the node_modules folder, and the index file for your bot. My index file is called snes.js. The files I’m moving over are highlighted in the following picture.
Once you highlight the files you want to transfer, you can click and drag your files to the right where you can see your Linux machine's files. Once you’ve transferred the files, you can use the LISH console or any terminal with an SSH connection to run your bot the same way you would from a terminal in Visual Studio Code. Type "npm run develop" and your code will start up using the .env variables you set up earlier. This makes the bot run and it will continue running on the server as long as the server remains uninterrupted.
Finale
Your bot should be good to go! What does yours do? Let me know @metzinaround on Twitter about the bot you built!
Top comments (1)
This is awesome! Thanks for writing this!