SaaS Pegasus Chat Demo

This is an example application that you can drop into your Django projects to get a real-time, streaming, asynchronous chatbot in just a few steps.

This application is based on this companion article: Learn to use Websockets with Django by building your own ChatGPT

If you'd like a version that also includes user-based sessions, asynchronous database access, chat history, and loads more functionality---check out SaaS Pegasus: the Django boilerplate for AI apps.

Demo

pegasus-chat-demo.webm

Installation

First install this library:

pip install .

Add the app to settings.INSTALLED_APPS:

INSTALLED_APPS = [
   # other apps here
   "pegasus_chat_demo.apps.PegasusChatDemoConfig",
]

Add the URLs to urls.py:

urlpatterns = [
    # other urls here
    path("chat-demo/", include("pegasus_chat_demo.urls")),
]

Then set up your URL routes. In asgi.py:

from pegasus_chat_demo.routing import websocket_urlpatterns

application = ProtocolTypeRouter(
    {
        "http": django_asgi_app,
        "websocket": AllowedHostsOriginValidator(
            AuthMiddlewareStack(
                URLRouter(
                    websocket_url_patterns,
                )
            )
        ),
    }
)

Finally, to call ChatGPT you need to set OPENAI_API_KEY in settings.py

OPENAI_API_KEY = "sk-***"