jupyter-chat-widget

A simple chat UI widget for Jupyter notebooks with streaming response support, designed to integrate easily with LLM APIs.

Note: this widget uses ipywidgets<8.0.0 for compatibility with Colab.

Installation

pip install jupyter-chat-widget

Quick Start

from jupyter_chat_widget import ChatUI

# Create the chat widget
chat = ChatUI()

# Define a handler for user messages
def handle_message(message: str) -> None:
    # Stream a response token by token
    for word in message.split():
        chat.append(word + " ")

# Connect the handler
chat.connect(handle_message)

Documentation

See the live documentation or the example notebook in examples/.

API Reference

ChatUI

The main chat widget class.

Methods

Example: Streaming Responses

from jupyter_chat_widget import ChatUI
from time import sleep

chat = ChatUI()

def slow_echo(message: str) -> None:
    """Echo back the message word by word."""
    for word in message.split():
        sleep(0.5)  # Simulate processing time
        chat.append(word + " ")
    chat.rewrite(f"You said: {message}")

chat.connect(slow_echo)

Example: Integration with LLM

from jupyter_chat_widget import ChatUI

chat = ChatUI()

def llm_handler(message: str) -> None:
    """Stream responses from an LLM API."""
    # Replace with your LLM API call
    for token in your_llm_api.stream(message):
        chat.append(token)

chat.connect(llm_handler)

Development

# Clone the repository
git clone https://github.com/zansara/jupyter-chat-widget.git
cd jupyter-chat-widget

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run linting
ruff check src tests
ruff format src tests

# Build the package
python -m build

Pre-commit Hooks

# Install pre-commit hooks
pre-commit install

# Run hooks on all files
pre-commit run --all-files

License

MIT License - see LICENSE for details.