Skip to content

Latest commit

 

History

History
155 lines (124 loc) · 5.42 KB

README.md

File metadata and controls

155 lines (124 loc) · 5.42 KB
Raw

esthetique-telegram-bot

The telegram chatbot that periodically gives statistics on posted via tg @like bot messages, i.e. esthetique posts, and can warn you about image duplicating as well.

plot

Agenda

Definition

  • tg-@like bot message/esthetique post: message with photo that sent with help of tg-@like bot with 5 buttons: 😡 / 😔 / 😐 / ☺️ / 😍 to the chat (10.08.22 😊 added as well)
  • statistics: bot shows esthetique posts with best score in ascending order
Esthetique example

plot

Feature

  • statistics: it is not possible via standard tg bot features to obtain changed historic information of an already collected message, thus, telethon is here in chat_history FastAPI microservice.
  • image duplicating: I use ImageHash lib as it is in bot. In realtime aiogram bot handle esthetique posts and duplication warnings.

Start

Briefly

Before the start, I suppose that both bots: telethon (chat_history) bot and aiogram (bot) added to the chat. For aiogram bot you have to disable privacy mode, i.e. @BotFather bot >> Bot Settings >> Group Privacy >> Turn off

Next to be mentioned, there is docker-compose file presented, thus, to run:

  • go to deploy/ and prepare your .env according to deploy/.env.example, then you are able to gracefully start compose in daemon mode:
docker-compose -f docker-compose.prod.yml up -d
  • for telethon you should login with pin verified, for this purpose I left script, So, complete login with script and telegram session will be saved into redis, e.g. command:
docker exec -ti deploy_chat_history_1 sh -c "PYTHONPATH=. python scripts/init_session.py"
  • or run the command above initially with tmp container from deploy/ dir.
docker-compose run chat_history sh -c "PYTHONPATH=. python scripts/init_session.py"

TODO: make this login as entrypoint with forwarding verification message via bot to admin

With Cache Warming Up

When you want to populate image hashes with already existed chat and its history I left a not good script for 1 run. The script also replaces script mentioned earlier, coz it creates session for telethon in redis as well.

Before continuing, I suppose:

  • Chat already exists. To be more precisely, chat is in supergroup status
  • Chat history for new members is visible
  • telethon bot account added to chat
  • requirements for both installed in you python3.9 venv or even host
  • redis that stack uses has port allocated to host, deploy/.env has values according, i.e. REDIS_HOST=localhost
  • privacy mode for bot is disabled

To start import run from script directory:

python import_previous_images_hashes.py

todo: too complicated instruction, instead prepare one run dockerfile? or even make use for chat_history image api?

Settings

Settings that may be tunes should be located in deploy/.env and prepared as in deploy/.env.example.

Tip to find chat_id fast: https://api.telegram.org/bot<UR_BOT_TOKEN>/getUpdates

Develop

Develop docker-compose

E.g. 
```yaml
version: '3.8'

volumes:
  redis_data_dev:

x-app: &dev-app
  env_file:
    - ./.env
  depends_on:
    - redis

services:
  chat_history:
    <<: *dev-app
    build:
      context: ../chat_history
      dockerfile: .
    command: uvicorn main:app --reload --host 0.0.0.0 --port 8000
    volumes:
      - ../chat_history/src/:/opt/
    ports:
      - 8000:8000

  bot:
    <<: *dev-app
    build:
      context: ../bot
      dockerfile: .
    volumes:
      - ../bot/src/:/opt/

  redis:
    image: redis:6.0.8-alpine
    volumes:
      - redis_data_dev:/data
    ports:
      - 6379:6379
  ```

Contribution

  • pep8 [120 symbols per line], pls
  • feature branch, e.g. feature/only_members_vote

TODO

  • github compose deploy
  • check that we calculate only chat members votes
  • in-code todos about entrypoint
  • in-code todos about chat_history app (telethon ) verification asking
  • auto title image change
  • telegram logger: send error messages to admins
  • it is possible to hack statistic results via providing similar emojis but the ones not from a template and allowed. Thus, by providing other emojis hacker could manipulate what he wants to be counted in final result.

Acknowledgement