Tracking Used Invites

Last updated last month

A relatively frequent thing people would love to know, is "what invite someone used to join". Unfortunately, the Discord API does not provide the information about the invite used to join a server. And to add insult to injury, the API doesn't provide any event for invites (neither creation nor change).

To get around this, we need to do two separate steps. The first one is to fetch all of the invites for each guild and store it in a temporary object. The second is to fetch a guild's invite whenever someone joins, and find which one has a new use on it. Thankfully, that's actually pretty simple!

While the below is a fair approximation of invite tracking, it's still not perfect. There are 2 things it doesn't track:

  • Temporary one-use invites (when right-clicking someone, and doing Invite To => Server). Those exist only for a few moments and cannot be tracked at all.

  • Invites created after the bot loaded. That would require fetching on a loop which is dangerous for API spam.

Caching Invites

The first part is to fetch all the invites and keep them cached in an object. This is done in the ready event. First, however, we must ensure that the cache is initalized outside of the ready event.

// Initialize the invite cache
const invites = {};
// A pretty useful method to create a delay without blocking the whole script.
const wait = require('util').promisify(setTimeout);
client.on('ready', () => {
// "ready" isn't really ready. We need to wait a spell.
wait(1000);
// Load all invites for all guilds and save them to the cache.
client.guilds.forEach(g => {
g.fetchInvites().then(guildInvites => {
invites[g.id] = guildInvites;
});
});
});

Catching New Members

So now that we have our invites object, we're ready to listen to the guildMemberAdd event. When a new member joins, we need to fetch all of the guild's invites once again. Then, we look through our cached invites and see which one has been used, by comparing the current invite's use with the cached ones.

client.on('guildMemberAdd', member => {
// To compare, we need to load the current invite list.
member.guild.fetchInvites().then(guildInvites => {
// This is the *existing* invites for the guild.
const ei = invites[member.guild.id];
// Update the cached invites for the guild.
invites[member.guild.id] = guildInvites;
// Look through the invites, find the one for which the uses went up.
const invite = guildInvites.find(i => ei.get(i.code).uses < i.uses);
// This is just to simplify the message being sent below (inviter doesn't have a tag property)
const inviter = client.users.get(invite.inviter.id);
// Get the log channel (change to your liking)
const logChannel = member.guild.channels.find(channel => channel.name === "join-logs");
// A real basic message with the information we need.
logChannel.send(`${member.user.tag} joined using invite code ${invite.code} from ${inviter.tag}. Invite was used ${invite.uses} times since its creation.`);
});
});

And... well, that's pretty much it. But....

There's a Catch

So here's the problem. Each time you fetch invites, you're hitting the Discord API with a request for information. While that's not an issue for small bots, it might as the bot grows. I'm not saying that using this code would get you banned from the API - there is no inherent problem with querying the API if it's not abuse. However, there are a few technical issues with performance especially.

The more guilds you have, the more invites are in each guild, the more data you're receiving from the API. Rember, because of the way the ready event works, you need to wait a bit before running the fetchInvite method, and the more guilds you have, the more time you need to wait. During this time, member joins won't correctly register because the cache doesn't exist.

Furthermore, the cache itself grows in size, so you're using more memory. On top of which, it takes more time to sort through the invites the more invites exist. It might make the bot appear as being slower to respond to members joining.

So to conclude, the above code works perfectly well and it will not get you in trouble with Discord, but I wouldn't recommend implementing this on larger bots, especially if you're worried about memory and performance.