Web-Based ChatGPT Chatbot with Django

• Web-Based ChatGPT Chatbot
  • Remember the conversation in a session
  • Uses NanumSquareNeo font (Neat looking for Korean)

Screenshot

• A modified example for Korean. Source code is only in English

Chatbot with the background tasks processing and communications via WebSockets. For more details please check my article - Heroku Chatbot with Celery, WebSockets, and Redis.

Supported by Choi Woen-Sug (Ocean Eng., KMOU)🧑‍🏫

Installation

• Requirements
  • Supported system : Debian Linux (e.g. Ubuntu)
  • Tested system : Ubuntu 22.04

Install Redis

curl -fsSL https://packages.redis.io/gpg | sudo gpg --dearmor -o /usr/share/keyrings/redis-archive-keyring.gpg
echo "deb [signed-by=/usr/share/keyrings/redis-archive-keyring.gpg] https://packages.redis.io/deb $(lsb_release -cs) main" | sudo tee /etc/apt/sources.list.d/redis.list
sudo apt update && sudo apt install -y redis

Check redis by running with following command,

redis-cli

You may see redis url and port. To exit, type exit.

Environment Variables

• REDIS URL
  • default redis url and port is 127.0.0.1:6379
• DJANGO SECRET KEY
  • Obtain at https://djecrety.ir/
  • Replafe YOUR_DJGANGO_SECRET_KEY with your key
• OPENAI MODEL
  • Name of the model you want to use.
  • ChatGPT version 3 is text-davinci-003
• OPENAI API KEY
  • Obtain at https://platform.openai.com/account/api-keys
  • Replace YOUR_API_KEY with your key
• PROMPT OPTION
  • It's an option you want to give to add exaplanation to the prompt.
  • It will be added to the prompt to help better responses.
  • Example would be Response in 40 words maximum. You may add as much as you want.

Set variables

export PROMPT_OPTION="Response in 40 words maximum."
export REDIS_URL="redis://127.0.0.1:6379"
export DJANGO_SECRET_KEY="YOUR_DJGANGO_SECRET_KEY"
export OPENAI_MODEL='text-davinci-003'
export OPENAI_API_KEY="YOUR_API_KEY"

• If you want to fix it and don't type every time

  # Prompt option
  echo 'export PROMPT_OPTION="Response in 40 words maximum."' >> ~/.bashrc
  
  # Export Redis URL (default redis url and port is 127.0.0.1:6379)
  echo "export REDIS_URL='redis://127.0.0.1:6379'" >> ~/.bashrc
  
  # Export Django Secret Key
  echo 'export DJANGO_SECRET_KEY="YOUR_DJGANGO_SECRET_KEY"' >> ~/.bashrc
  
  # Export the name of the OpenAI Model
  echo 'export OPENAI_MODEL="text-davinci-003"' >> ~/.bashrc
  
  # Export OpenAI ChatGPT API Key
  echo 'export OPENAI_API_KEY=YOUR_API_KEY' >> ~/.bashrc

Load bashrc for environment variables

source ~/.bashrc

Get source code

git clone git@github.com:woensug-choi/web-chatgpt-chatbot-django.git

Install Dependencies

sudo apt install python3-pip
pip install -r requirements.txt
pip install openai

Run server

set-up database (ONLY ONCE)

python manage.py migration

Finally, run the server

python manage.py runserver 0:8080

Open Browser to access

When running the python manage.py runserver 0:8080, you will be given a url to access. Typical access is http://127.0.0.1:8080. Open with any browser you will have access.

Example of system service to make it run at background

Write a service configuration file

• wirte a service configuration file at /etc/systemd/system/webchatgpt.service

• sudo nano /etc/systmd/system/webchatgpt.service

• You may change the name webchatgpt with anything you want

• CHANGE THE PATH OF manage.py in the following example

  # Example of /etc/systemd/system/webchatgpt.service
  [Unit]
  Description=WebChatGPT
  
  [Service]
  User=ubuntu
  Type=simple
  Restart=always
  ExecStart=/usr/bin/python3 /home/ubuntu/WebChatGPT/manage.py runserver 0:8888
  
  [Install]
  WantedBy=multi-user.target
  

Run the service,

sudo systemctl daemon-reload # Reload the config file
sudo systemctl enable webchatgpt.service # Enable (make it auto start)
sudo systemctl start webchatgpt.service # Star the service now

Obsolete

Deployment

You can host it on Heroku for free (account verification required).