django-telegram-bot

Sexy Django + python-telegram-bot + Celery + Redis + Postgres + Dokku + GitHub Actions template. Production-ready Telegram bot with database, admin panel and a bunch of useful built-in methods.

⭐ graph:

Check the example bot that uses the code from Main branch: t.me/djangotelegrambot

Features

Database: Postgres, Sqlite3, MySQL - you decide!
Admin panel (thanks to Django)
Background jobs using Celery
Production-ready deployment using Dokku
Telegram API usage in polling or webhook mode
Export all users in .csv
Native telegram commands in menu
- In order to edit or delete these commands you'll need to use set_my_commands bot's method just like in tgbot.dispatcher.setup_my_commands

Built-in Telegram bot methods:

/broadcast — send message to all users (admin command)
/export_users — bot sends you info about your users in .csv file (admin command)
/stats — show basic bot stats
/ask_for_location — log user location when received and reverse geocode it to get country, city, etc.

Content

How to run

Quickstart: Polling & SQLite

The fastest way to run the bot is to run it in polling mode using SQLite database without all Celery workers for background jobs. This should be enough for quickstart:

git clone https://github.com/ohld/django-telegram-bot
cd django-telegram-bot

Create virtual environment (optional)

python3 -m venv dtb_venv
source dtb_venv/bin/activate

Install all requirements:

pip install -r requirements.txt

Create .env file in root directory and copy-paste this or just run cp .env_example .env, don't forget to change telegram token:

DJANGO_DEBUG=True
DATABASE_URL=sqlite:///db.sqlite3
TELEGRAM_TOKEN=<PASTE YOUR TELEGRAM TOKEN HERE>

Run migrations to setup SQLite database:

python manage.py migrate

Create superuser to get access to admin panel:

python manage.py createsuperuser

Run bot in polling mode:

python run_polling.py

If you want to open Django admin panel which will be located on http://localhost:8000/tgadmin/:

python manage.py runserver

Run locally using docker-compose

If you want just to run all the things locally, you can use Docker-compose which will start all containers for you.

Create .env file.

You can switch to PostgreSQL just by uncommenting it's DATABASE_URL and commenting SQLite variable.

cp .env_example .env

Docker-compose

To run all services (Django, Postgres, Redis, Celery) at once:

docker-compose up -d --build

Check status of the containers.

docker ps -a

It should look similar to this:

Try visit Django-admin panel.

Enter django shell:

docker exec -it dtb_django bash

Create superuser for Django admin panel

python manage.py createsuperuser

To see logs of the container:

docker logs -f dtb_django

Deploy to Production

Production stack will include these technologies:

Postgres as main database for Django
Celery + Redis + easy scalable workers
Dokku as PaaS (will build app from sources and deploy it with zero downtime)

All app's services that are going to be launched in production can be found in Procfile file. It includes Django webserver (Telegram event processing + admin panel) and Celery workers (background and periodic jobs).

What is Dokku and how it works

Dokku is an open-source version of Heroku.

I really like Heroku deployment approach:

you push commit to Main branch of your Repo
in couple minutes your new app is running
if something breaks during deployment - old app will not be shut down

You can achieve the same approach with Dokku + Github Actions (just to trigger deployment).

Dokku uses buildpacks technology to create a Docker image from the code. No Dockerfile needed. Speaking about Python, it requires requirements.txt, Procfile files to run the things up. Also files DOKKU_SCALE and runtime.txt are useful to tweak configs to make the deployed app even better. E.g. in DOKKU_SCALE you can specify how many app instances should be run behind built-in load balancer.

One disadvantage of Dokku that you should be warned about is that it can work with one server only. You can't just scale your app up to 2 machines using only small config change. You still can use several servers by providing correct .env URLs to deployed apps (e.g. DATABASE_URL) but it will require more time to setup.

Deploy using Dokku: step-by-step

I assume that you already have Dokku installed on your server. Let's also assume that the address of your server is <YOURDOMAIN.COM> (you will need a domain to setup HTTPs for Telegram webhook support). I'd recommend to have at least 2GB RAM and 2 CPU cores.

Create Dokku app

dokku apps:create dtb

You might need to added .env variables to app, e.g. to specify Telegram token:

dokku config:set dtb TELEGRAM_TOKEN=.....

Postgres and Redis

Postgres and Redis are configured as Dokku plugins on a server. They will automatically add REDIS_URL & DATABASE_URL .env vars to the app after being linked. You might need to install these Dokku plugins before. Install Postgres, install Redis.

dokku postgres:create dtb
dokku postgres:link dtb dtb

dokku redis:create dtb
dokku redis:link dtb dtb

Deploy on commit with Github Actions

Go to file .github/workflows/dokku.yml:

Enter your host name (address of your server),
Deployed dokku app name (in our case this is dtb),
Set SSH_PRIVATE_KEY secret variable via GitHub repo settings. This private key should have the root ssh access to your server.

This will trigger Dokku's zero-downtime deployment. You would probably need to fork this repo to change file.

After that you should see a green arrow ✅ at Github Actions tab that would mean your app is deployed successfully. If you see a red cross ❌ you can find the deployed logs in Github Actions tab and find out what went wrong.

HTTPS & Telegram bot webhook

Why you need to setup webhook

Basic polling approach is really handy and can speed up development of Telegram bots. But it doesn't scale. Better approach is to allow Telegram servers push events (webhook messages) to your server when something happens with your Telegram bot. You can use built-in Dokku load-balancer to parallel event processing.

HTTPS using Letsencrypt plugin

For Telegram bot API webhook usage you'll need a https which can be setup using Letsencrypt Dokku plugin. You will need to attach a domain to your Django app before and specify a email (required by Letsencrypt) - you will receive notifications when certificates would become old. Make sure you achieved a successful deployment first (your app runs at <YOURDOMAIN.COM>, check in browser).

dokku domains:add dtb <YOURDOMAIN.COM>
dokku config:set --global DOKKU_LETSENCRYPT_EMAIL=<YOUR@EMAIL.COM>
dokku letsencrypt:enable dtb

Setup Telegram Bot API webhook URL

You need to tell Telegram servers where to send events of your Telegram bot. Just open in the browser:

https://api.telegram.org/bot<TELEGRAM_TOKEN>/setWebhook?url=https://<YOURDOMAIN.COM>/super_secter_webhook/

After deployment

You can be sure that your app is deployed successfully if you see a green arrow at the latest workflow at Github Actions tab.

You would need to create a superuser to access an admin panel at https://<YOURDOMAIN.COM>/tgadmin. This can be done using a standard way using django shell:

Open shell in deployed app

dokku enter dtb web

Create Django super user

Being inside a container:

python manage.py createsuperuser

After that you can open admin panel of your deployed app which is located at https://<YOURDOMAIN.COM>/tgadmin.

Read app logs

dokku logs dtb -t

AlexeyGoto/django-telegram-bot