A serverless Telegram bot that provides Muslim prayer times and sends notifications when prayers are approaching.
Problem
- As a Muslim, I strive to pray on time, but my busy schedule sometimes causes me to miss prayer times.
- Although many apps are available for prayer reminders, as a developer, I spend most of my time on a laptop.
- I wanted a way to receive prayer notifications directly on my laptop.
- Since I was already using Telegram extensively, I decided to build a Telegram bot that would send me prayer time notifications.
Now that you have the context, let’s dive into how I built this bot.
Serverless
The bot is designed to be serverless, meaning it runs entirely in a cloud environment without the need for dedicated servers or manual infrastructure management.
This approach provides:
- Easy deployment
- Automatic scaling
- Cost efficiency (which is why I migrated from a VPS to a serverless setup)
Here’s an overview of the bot’s architecture:
| Component | Purpose |
|---|---|
dispatcher | Handles incoming messages and user interactions. |
reminder | Sends prayer time reminders to subscribed users. |
loader | Loads prayer times from S3 and stores them in YDB. |
YDB
I chose YDB for several reasons:
- Serverless: No need to manage virtual machines, clusters, or scaling — you pay only for what you use.
- SQL Support: YDB supports familiar SQL queries, making integration straightforward without complex query builders.
- Cost: It’s free up to a generous quota — perfect for prototyping and small-to-medium scale applications.
- Exploration: I wanted to explore new technology and expand my database experience beyond PostgreSQL and MongoDB.
The bot uses two main tables:
chats table
| Column | Type | Purpose |
|---|---|---|
chat_id | Int64 | Unique identifier for the Telegram chat. |
bot_id | Int64 | Unique identifier for the bot instance. |
language_code | Utf8 | Language selected by the user for localized messages. |
state | Utf8 | Current user interaction state (e.g., awaiting input, viewing results). |
reminder_offset | Int32 | Time offset for prayer reminders (in minutes). |
reminder_message_id | Int32 | ID of the reminder message sent (used to delete old reminders). |
subscribed | Bool | Whether the user is subscribed to notifications. |
subscribed_at | Datetime | Timestamp of when the user subscribed. |
created_at | Datetime | Timestamp of when the chat record was created. |
prayers table
| Column | Type | Purpose |
|---|---|---|
bot_id | Int64 | Unique identifier for the bot instance. |
prayer_date | Date | Date for which prayer times are calculated. |
fajr | Datetime | Fajr prayer time. |
shuruq | Datetime | Shuruq (sunrise) time. |
duhr | Datetime | Duhr prayer time. |
asr | Datetime | Asr prayer time. |
maghrib | Datetime | Maghrib prayer time. |
isha | Datetime | Isha prayer time. |
Languages
The bot supports multiple languages for user interaction.
Translation templates are located in:
serverless/dispatcher/internal/handler/languages/
Currently supported languages:
- Arabic (
ar) - English (
en) - Spanish (
es) - French (
fr) - Russian (
ru) - Turkish (
tr) - Tatar (
tt) - Uzbek (
uz)
Each language is represented as a YAML file containing the necessary translations.
Deployment
As mentioned earlier, the bot runs serverlessly on Yandex Cloud.
For infrastructure management, I used Terraform to provision all necessary resources.
You can view the Terraform configuration file here.