Hosting: Vercel Serverless Functions

This tutorial will guide you on how to deploy your bot to Vercel by using Vercel Serverless Functions, assuming that you already have a Vercel account.

Project Structure

The only prerequisite to get started with Vercel Serverless Functions is to move your code to the api/ directory as shown below. You can also see Vercel’s documentation for more on this.

.
├── node_modules/
├── build/
├── api/
│   └── bot.ts
├── package.json
├── package-lock.json
└── tsconfig.json

If you are using TypeScript, you might as well want to install @vercel/node as a dev dependency, but it is not mandatory for following this guide.

Configuring Vercel

The next step is to create a vercel.json file at the top level of your project. For our example structure, its content would be:

{
  "functions": {
    "api/bot.ts": {
      "memory": 1024,
      "maxDuration": 10
    }
  }
}

If you want to use Vercel’s free subscription, your memory and maxDuration configurations might look like above to not bypass its limits.

If you want to learn more about the vercel.json configuration file, see its documentation.

Configuring TypeScript

In our tsconfig.json, we have to specify our output directory as build/, and our root directory as api/. This is important since we will specify them in Vercel’s deploy options.

{
  "compilerOptions": {
    "target": "ES2019",
    "module": "commonjs",
    "rootDir": "./api",
    "moduleResolution": "node",
    "resolveJsonModule": true,
    "outDir": "./build",
    "allowSyntheticDefaultImports": true,
    "esModuleInterop": true,
    "strict": true,
    "skipLibCheck": true
  }
}

The Main File

Regardless of using TypeScript or JavaScript, we should have a source file through which our bot runs. It should look roughly like this:

import { Bot, webhookCallback } from "grammy";

const token = process.env.BOT_TOKEN;
if (!token) throw new Error("BOT_TOKEN is unset");

const bot = new Bot(token);

export default webhookCallback(bot, "std/http");

Vercel Edge Functions provides limited support for grammY

You can still use the core grammY package and a number of plugins, but others may be incompatible due to Node.js-only dependencies that might not be supported by Vercel’s Edge Runtime.

Currently, we don’t have a comprehensive list of compatible plugins, so you need to test it by yourself.

Add this line to snippet above if you want to switch to Edge Functions:

export const config = {
  runtime: "edge",
};

In Vercel’s Dashboard

Assuming that you already have a Vercel account your GitHub is connected to, add a new project and select your bot’s repository. In Build & Development Settings:

• Output directory: build
• Install command: npm install

Don’t forget to add the secrets such as your bot token as environment variables in the settings. Once you have done that, you can deploy it!

Setting the Webhook

The last step is to connect your Vercel app with Telegram. Modify the below URL to your credentials and visit it from your browser:

https://api.telegram.org/bot<BOT_TOKEN>/setWebhook?url=<HOST_URL>

The HOST_URL is a little tricky, because you need to use your Vercel app domain following with the route to the bot code, for example https://appname.vercel.app/api/bot. Where bot is referring to your bot.ts or bot.js file.

You should then see a response like this:

{
  "ok": true,
  "result": true,
  "description": "Webhook was set"
}

Congratulations! Your bot should now be up and running.