chatGPT-Linebot

Feature

• chatGPT on Line Bot
• Runs on docker-compose
• Published on Docker Hub
• Uses Python3 Django with SQLite
• Only the first response is processed each time

Requirements

• Knowledge of setting up a Line Bot
  • Reference: LINE Bot Tutorial
• Obtain an OpenAI API KEY
• Using ngrok to build tunnel.

Configuration

Linebot config path is located at src/config.json

{
    "OPENAI_API": {
        "openai_api_key": "ENTER YOUR OPENAPI KEY",
        "model": "gpt-3.5-turbo",
        "max_token": 1024,
        "temperature": 0.8,
        "timeout": 30
    },
    "LINE_CHANNEL_SECRET": "YOUR LINE SECRET",
    "LINE_CHANNEL_ACCESS_TOKEN": "YOUR LINE ACCESS TOKEN"
}

• "OPENAI_API" detal can watch Open AI Chat
  • openai_api_key: Your OpenAI key
  • model
  • max_token
  • temperature
  • timeout: Timeout for each request
• "LINE_CHANNEL_SECRET"
• "LINE_CHANNEL_ACCESS_TOKEN" All history session storage in src/history.json.

Ngrok config at init/ngrok/ngrok.yml

# Enter you ngrok authtoken
authtoken: INPUT AUTHTOKEN

Build from Scratch

1. Clone the repository locally
2. Install Docker and Docker Compose
3. Set config.json in src/config.json and ngrok.yml in init/ngrok/ngrok.yml
4. In chatGPT-Linebot directory, run docker-compose up -d
5. Linebot will be listening at localhost:8000/chatGPT/callback
6. Using curl -s localhost:4040/api/tunnels get the ngrok tunnel address, or look at docker compose up output.
7. Edit the Line Webhook URL

Client use

1. Can have a continuous conversation.
2. When you want to end a session, enter #clean to clean the history session.