Downloader-Bot

A production-oriented Telegram bot for downloading media from major social platforms, with inline delivery, PostgreSQL-backed state, structured logging, and a queue designed for real-world load.

The project is built on aiogram 3, uses SQLAlchemy + Alembic for persistence, and supports both local development and Docker-based deployment.

Features

• Download media from TikTok, Instagram, YouTube, SoundCloud, Pinterest, and X/Twitter
• Support direct bot usage and Telegram inline mode flows
• Per-user settings for captions, info buttons, URL buttons, MP3 buttons, and auto-delete
• Shared download queue with backpressure, retries, and large-download worker handoff
• PostgreSQL persistence for users, settings, analytics, and file cache
• Admin tools for runtime stats, session metrics, user management, and log access
• Structured event and performance logging for easier debugging and operations

Stack

• Python 3.14
• aiogram 3
• PostgreSQL
• SQLAlchemy 2 + Alembic
• yt-dlp
• aiohttp / httpx
• Docker / Docker Compose
• pytest

Supported flows

User-facing

• /start to initialize the bot
• /settings to configure delivery behavior
• /stats to view usage statistics
• Inline queries for supported services where Telegram inline delivery is available

Admin

• /admin
• /perf
• /session

Architecture

The codebase is split into a few clear layers:

• handlers contains Telegram message, callback, and inline-query flows
• services/platforms contains platform-specific media extraction and download logic
• services/download contains the shared queue and worker logic
• services/storage contains the database models, repositories, caching, and schema bootstrap
• services/runtime contains transient runtime state such as dedupe, pending requests, and live stats
• middlewares contains antiflood, chat tracking, ban checks, and private-chat guard behavior
• tests covers handlers, runtime services, startup, storage, queue behavior, and deployment helpers

Quick Start

1. Clone the repository

git clone https://github.com/Mak5er/Downloader-Bot.git
cd Downloader-Bot

2. Install dependencies

pip install -r requirements.txt

For development and tests:

pip install -r requirements-dev.txt

3. Configure environment

Create a .env file in the project root:

BOT_TOKEN=your_telegram_bot_token
DATABASE_URL=postgresql://user:password@host:5432/dbname
ADMIN_ID=123456789
CUSTOM_API_URL=http://your-telegram-bot-api:8081

# Optional but recommended depending on your deployment/features
MEASUREMENT_ID=
API_SECRET=
CHANNEL_ID=
COBALT_API_URL=
COBALT_API_KEY=

# Optional YouTube access helpers for yt-dlp
YTDLP_YOUTUBE_COOKIES_FILE=cookies/youtube.txt
# Alternative to a cookie file, useful only when running on the same machine as the browser:
# YTDLP_YOUTUBE_COOKIES_FROM_BROWSER=firefox:Profile 1
# YTDLP_YOUTUBE_REMOTE_COMPONENTS=ejs:github
# YTDLP_YOUTUBE_PLAYER_CLIENT=web,android
# YTDLP_YOUTUBE_PO_TOKEN=web.gvs+your_token

# Performance tuning
BOT_POLLING_TASKS_CONCURRENCY_LIMIT=256
BOT_SESSION_CONNECTION_LIMIT=400
DB_POOL_SIZE=32
DB_MAX_OVERFLOW=64
DB_POOL_TIMEOUT=30

# Anti-flood and dedupe
ANTIFLOOD_MESSAGE_LIMIT=4
ANTIFLOOD_MESSAGE_WINDOW_SECONDS=2
ANTIFLOOD_CALLBACK_LIMIT=6
ANTIFLOOD_CALLBACK_WINDOW_SECONDS=2
ANTIFLOOD_INLINE_LIMIT=4
ANTIFLOOD_INLINE_WINDOW_SECONDS=3
ANTIFLOOD_GLOBAL_LIMIT=8
ANTIFLOOD_GLOBAL_WINDOW_SECONDS=3
ANTIFLOOD_COOLDOWN_SECONDS=6
ANTIFLOOD_USER_TTL_SECONDS=180
ANTIFLOOD_MAX_TRACKED_USERS=50000
REQUEST_DEDUPE_ACTIVE_TTL_SECONDS=900
REQUEST_DEDUPE_COMPLETED_TTL_SECONDS=12
REQUEST_DEDUPE_MAX_ENTRIES=50000

# Download queue tuning
DOWNLOAD_QUEUE_MIN_WORKERS=3
DOWNLOAD_QUEUE_MAX_WORKERS=7
DOWNLOAD_QUEUE_MAX_SIZE=250
DOWNLOAD_QUEUE_PER_USER_RATE_LIMIT=4
DOWNLOAD_QUEUE_PER_USER_WINDOW_SECONDS=10
DOWNLOAD_QUEUE_PER_USER_MAX_PENDING=3
DOWNLOAD_QUEUE_PER_USER_PENDING_TIMEOUT_SECONDS=0
DOWNLOAD_QUEUE_SCALE_COOLDOWN_SECONDS=8
DOWNLOAD_QUEUE_IDLE_SCALE_DOWN_SECONDS=35
DOWNLOAD_MAX_WORKERS_CAP=6

# Batch link tuning
BATCH_LINKS_MAX_ITEMS=6
BATCH_LINKS_MIN_CONCURRENCY=1
BATCH_LINKS_MAX_CONCURRENCY=2
BATCH_LINKS_PARALLEL_QUEUE_DEPTH_THRESHOLD=2
BATCH_LINKS_PARALLEL_ACTIVE_JOBS_THRESHOLD=3

4. Run locally

python main.py

On startup the bot initializes the database schema via Alembic migrations automatically.

YouTube Cookies

Some YouTube videos need an authenticated browser session for yt-dlp to work correctly. The bot automatically uses cookies/youtube.txt when that file exists. You can also override the path with YTDLP_YOUTUBE_COOKIES_FILE.

Recommended setup:

1. Install a browser extension that exports cookies in Netscape format, for example "Get cookies.txt LOCALLY".
2. Open youtube.com in the browser profile that has a normal signed-in YouTube session.
3. Export only YouTube cookies for .youtube.com / youtube.com in Netscape cookies.txt format.
4. Put the exported file at cookies/youtube.txt.
5. Keep YTDLP_YOUTUBE_COOKIES_FILE=cookies/youtube.txt in .env, or omit it and let the default path be used.
6. If yt-dlp logs n challenge solving failed or only shows storyboard/image formats, set YTDLP_YOUTUBE_REMOTE_COMPONENTS=ejs:github.
7. Restart the bot after replacing cookies or changing YouTube yt-dlp options.

The cookies directory is kept in git with cookies/.gitkeep, but real cookie files are ignored by git and Docker builds. Treat cookies/youtube.txt like a password: do not commit it, paste it in chats, or bake it into images.

For Docker Compose, the repository mounts ./cookies into the container as /app/cookies, so the same cookies/youtube.txt path works inside the container. The bot startup fixes ownership on that mount because yt-dlp may need to update the cookie jar.

Docker

The repository includes a production-friendly multi-stage Dockerfile and a minimal docker-compose.yml.

Run with the published image:

docker compose pull
docker compose up -d

Or build locally:

docker compose up -d --build

Notes:

• Downloads and logs are stored in named Docker volumes
• The container entrypoint prepares runtime paths and drops privileges before launching the app
• ffmpeg is installed in the runtime image

Performance Profile

For a 4-core i5-7500, 16 GB RAM, and gigabit network, start with:

BOT_POLLING_TASKS_CONCURRENCY_LIMIT=192
BOT_SESSION_CONNECTION_LIMIT=300
DB_POOL_SIZE=24
DB_MAX_OVERFLOW=48
DB_POOL_TIMEOUT=30

DOWNLOAD_QUEUE_MIN_WORKERS=3
DOWNLOAD_QUEUE_MAX_WORKERS=7
DOWNLOAD_QUEUE_MAX_SIZE=250
DOWNLOAD_QUEUE_PER_USER_RATE_LIMIT=4
DOWNLOAD_QUEUE_PER_USER_WINDOW_SECONDS=10
DOWNLOAD_QUEUE_PER_USER_MAX_PENDING=3
DOWNLOAD_QUEUE_PER_USER_PENDING_TIMEOUT_SECONDS=0
DOWNLOAD_QUEUE_SCALE_COOLDOWN_SECONDS=8
DOWNLOAD_QUEUE_IDLE_SCALE_DOWN_SECONDS=35
DOWNLOAD_MAX_WORKERS_CAP=6

BATCH_LINKS_MAX_ITEMS=6
BATCH_LINKS_MIN_CONCURRENCY=1
BATCH_LINKS_MAX_CONCURRENCY=2
BATCH_LINKS_PARALLEL_QUEUE_DEPTH_THRESHOLD=2
BATCH_LINKS_PARALLEL_ACTIVE_JOBS_THRESHOLD=3

This keeps CPU-heavy downloader work bounded while still letting the queue scale up under real demand. Batch links run with limited parallelism only when the queue is healthy; under load they fall back to sequential processing automatically.

Logging and Observability

The bot writes both human-readable and structured logs:

• logs/bot_log.log
• logs/error_log.log
• logs/events_log.jsonl
• logs/perf_log.jsonl

The logging layer adds request-scoped context such as service, flow, and request_id, which makes it easier to trace a single user action across fetch, queue, download, and delivery steps.

Development Notes

• The project expects a custom Telegram Bot API endpoint via CUSTOM_API_URL
• PostgreSQL is required in normal operation
• Some platform flows rely on a Cobalt-compatible backend; set COBALT_API_URL and COBALT_API_KEY as needed
• Test coverage is centered around handlers, queue/runtime behavior, database access, migrations, and startup logic

Run tests with:

pytest

Suggested Improvements

• Add a small health/readiness endpoint or admin health command that checks PostgreSQL connectivity, Telegram Bot API reachability, queue depth, and downloader worker status.
• Expand platform regression tests with saved extractor fixtures for TikTok, Instagram, YouTube, SoundCloud, Pinterest, and X/Twitter so platform-specific breakages are easier to catch before deployment.

Contributing

Issues and pull requests are welcome. If you change runtime behavior, settings, platform handling, or deployment flow, please update the README and related tests in the same change.

Repository Layout

.
|-- handlers/
|-- services/
|   |-- download/
|   |-- inline/
|   |-- links/
|   |-- media/
|   |-- platforms/
|   |-- runtime/
|   `-- storage/
|-- middlewares/
|-- keyboards/
|-- messages/
|-- filters/
|-- tests/
|-- main.py
|-- config.py
|-- Dockerfile
`-- docker-compose.yml

Bot

Public bot: @MaxLoadBot

Support the Project

If Downloader-Bot saves you time, you can support ongoing development and infrastructure costs:

• PayPal: Donate
• TRC20: TS4Ktovpwz9n2Ws8q9YXC3npW8gXi4QyYi
• BEP20: 0xE8F613484f84F1B70A777325771d3A3Ca33979Ab
• Solana: 8pfgWjfvDUpmeszVXbRzbifFzUDzeNeGWuJf6HCcjAF7
• ERC20: 0x9b38804F07A4ca4381a6Ef7F0022a3F4caBc5b6F
• TON: UQBm9KPhtMw-XVVjirUoa09wzrlyWsbeZhKfefl1Uw-qNZ-r

License

Released under the MIT License.