Metadata-Version: 2.4
Name: gradiochat
Version: 0.3.1
Summary: App to easily create different Gradiochat apps with different context, system messages and gradio themes.
Project-URL: Homepage, https://github.com/Hopsakee/gradiochat
Project-URL: Documentation, https://hopsakee.github.io/gradiochat/
Author-email: Hopsakee <jdejong@posteo.nl>
License-Expression: Apache-2.0
License-File: LICENSE
Classifier: Programming Language :: Python :: 3
Requires-Python: >=3.10
Requires-Dist: fastcore>=1.7.29
Requires-Dist: gradio>=5.20.1
Requires-Dist: ollama>=0.4.7
Requires-Dist: openai>=1.65.4
Requires-Dist: pydantic>=2.10.6
Requires-Dist: python-dotenv>=1.0.1
Requires-Dist: requests>=2.32.3
Requires-Dist: together>=1.4.6
Description-Content-Type: text/markdown

# gradiochat


<!-- WARNING: THIS FILE WAS AUTOGENERATED! DO NOT EDIT! -->

## Features

- Easy-to-use interface
- Supports integration with various AI models
- Real-time chat capabilities
- Open-source and customizable

## Documentation

Documentation can be found hosted on this GitHub
[repository](https://github.com/Hopsakee/gradiochat)’s
[pages](https://Hopsakee.github.io/gradiochat/). Additionally you can
find package manager specific guidelines on
[pypi](https://pypi.org/project/gradiochat/) respectively.

## How to use

This comprehensive guide explains how to use the GradioChat package to
create customizable LLM-powered chat applications with Gradio.
GradioChat provides a simple yet powerful framework for building chat
interfaces that can connect to various language models.

#### Installation

Install the package using `pip` or `uv` using the explanation below.

Install latest from the GitHub
[repository](https://github.com/Hopsakee/gradiochat):

``` sh
$ pip install git+https://github.com/Hopsakee/gradiochat.git
```

or from [pypi](https://pypi.org/project/gradiochat/)

``` sh
$ pip install gradiochat
```

#### Quick start

Here’s a minimal example to get you started.

``` python
# Eval is false to prevent testing when nbdev_test or nbdev_prepare is run. The api_key is stored in a .env file and that is not accessible at test time.
from gradiochat.config import ModelConfig, ChatAppConfig
from gradiochat.ui import create_chat_app
from pathlib import Path

# Create model configuration
model_config = ModelConfig(
    model_name="mistralai/Mistral-7B-Instruct-v0.2",
    provider="huggingface",
    api_key_env_var="HF_API_KEY"  # Optional: Set in .env file or environment
)

# Create chat application configuration
config = ChatAppConfig(
    app_name="My Chat App",
    description="A simple chat application powered by Mistral",
    system_prompt="You are a helpful assistant.",
    model=model_config
)

# Create and launch the chat application
app = create_chat_app(config)
app.build_interface().launch()
```

    /home/jelle/code/gradiochat/src/gradiochat/ui.py:89: UserWarning: You have not specified a value for the `type` parameter. Defaulting to the 'tuples' format for chatbot messages, but this is deprecated and will be removed in a future version of Gradio. Please set type='messages' instead, which uses openai-style dictionaries with 'role' and 'content' keys.
      chatbot = gr.Chatbot(

    * Running on local URL:  http://127.0.0.1:7860

    To create a public link, set `share=True` in `launch()`.

<div><iframe src="http://127.0.0.1:7860/" width="100%" height="500" allow="autoplay; camera; microphone; clipboard-read; clipboard-write;" frameborder="0" allowfullscreen></iframe></div>

### Configuration

The core of GradioChat is its configuration system which uses Pydantic
for validation.

#### ModelConfig

The `ModelConfig` class defines how to connect to a language model:

``` python
from gradiochat.config import ModelConfig

# HuggingFace model
hf_model = ModelConfig(
    model_name="mistralai/Mistral-7B-Instruct-v0.2",
    provider="huggingface",
    api_key_env_var="HF_API_KEY",  # Will read from environment variable
    api_base_url=None,  # Optional: Custom API endpoint
    max_completion_tokens=1024,
    temperature=0.7
)
```

### Message

The Message class represents a single message in a conversation:

``` python
from gradiochat.config import Message

# Create a system message
system_msg = Message(
    role="system",
    content="You are a helpful assistant."
)

# Create a user message
user_msg = Message(
    role="user",
    content="Hello, can you help me with Python?"
)

# Create an assistant message
assistant_msg = Message(
    role="assistant",
    content="Of course! I'd be happy to help with Python. What would you like to know?"
)
```

### ChatAppConfig

The `ChatAppConfig` class is the main configuration for your chat
application:

``` python
from gradiochat.config import ChatAppConfig, ModelConfig
from pathlib import Path

# Create model configuration
model_config = ModelConfig(
    model_name="mistralai/Mistral-7B-Instruct-v0.2",
    provider="huggingface",
    api_key_env_var="HF_API_KEY"
)

# Create chat application configuration
config = ChatAppConfig(
    app_name="Python Helper",
    description="Get help with Python programming",
    system_prompt="You are a Python expert who helps users with programming questions.",
    starter_prompt="Hello! I'm your Python assistant. Ask me any Python-related question.",
    context_files=[Path("docs/python_tips.md")],  # Optional: Add context from files
    model=model_config,
    theme=None,  # Optional: Custom Gradio theme
    logo_path=Path("assets/logo.png"),  # Optional: Path to logo image
    show_system_prompt=True,  # Whether to show system prompt in UI
    show_context=True  # Whether to show context in UI
)
```

## Creating a Chat Application

### Using Environment Variables

For API keys, it’s recommended to use environment variables. You can
create a ’ `.env` file in your project root:

`HF_API_KEY=your_huggingface_api_key_here`

Then load it in your application:

``` python
from dotenv import load_dotenv

load_dotenv()  # Load environment variables from .env file

# Now create your ModelConfig with api_key_env_var
```

### Adding Context Files

You can provide additional context to your LLM by adding markdown files:

``` python
from pathlib import Path

config = ChatAppConfig(
    # ... other parameters
    context_files=[
        Path("docs/product_info.md"),
        Path("docs/faq.md")
    ],
    # ... other parameters
)
```

## Customization

### Custom Themes

You can customize the appearance of your chat application using [Gradio
themes](https://www.gradio.app/guides/theming-guide). You can build
those yourself with help from the `gradio_themebuilder` or you can use
one of the predefined themes in `gradio_themes`. The predifined themes
are listed below.

    - themeWDODelta

``` python
import gradio as gr

my_theme = gr.themes.Base(
    primary_hue="fuchsia",
)

# Use the theme in your config
config = ChatAppConfig(
    # ... other parameters
    theme=my_theme,
    # ... other parameters
)
```

## API Reference

### BaseChatApp

The BaseChatApp class provides the core functionality for chat
applications:

``` python
from gradiochat.app import BaseChatApp
from gradiochat.config import ChatAppConfig

# Create configuration
config = ChatAppConfig(...)

# Create base app
base_app = BaseChatApp(config)

# Generate a response
response = base_app.generate_response("What is Python?")

# Generate a streaming response
# IMPORTANT: I don't actually think this already works. To be continued.
for chunk in base_app.generate_stream("Tell me about Python"):
    print(chunk, end="", flush=True)
```

### GradioChat

The `GradioChat` class provides the Gradio UI for the chat application:

``` python
from gradiochat.ui import GradioChat
from gradiochat.app import BaseChatApp

# Create base app
base_app = BaseChatApp(config)

# Create Gradio interface
gradio_app = GradioChat(base_app)

# Build and launch the interface
interface = gradio_app.build_interface()
interface.launch()
```

### LLM Clients

The package currently supports HuggingFace models through the
`HuggingFaceClient` class:

``` python
from gradiochat.app import HuggingFaceClient
from gradiochat.config import ModelConfig, Message

# Create model config
model_config = ModelConfig(
    model_name="mistralai/Mistral-7B-Instruct-v0.2",
    provider="huggingface",
    api_key_env_var="HF_API_KEY"
)

# Create client
client = HuggingFaceClient(model_config)

# Generate a completion
messages = [
    Message(role="system", content="You are a helpful assistant."),
    Message(role="user", content="What is Python?")
]
response = client.chat_completion(messages)
```

## Complete Example

Here’s a complete example that demonstrates most features:

``` python
import gradio as gr
from gradiochat.config import ModelConfig, ChatAppConfig
from gradiochat.gradio_themes import themeWDODelta
from pathlib import Path
from dotenv import load_dotenv

# Load environment variables
load_dotenv()

# Create a custom theme
theme = themeWDODelta

# Create model configuration
model_config = ModelConfig(
    model_name="mistralai/Mistral-7B-Instruct-v0.2",
    provider="huggingface",
    api_key_env_var="HF_API_KEY",
    max_completion_tokens=2048,
    temperature=0.8
)

# Create chat application configuration
config = ChatAppConfig(
    app_name="Python Expert",
    description="Get expert help with Python programming",
    system_prompt="You are a Python expert who helps users with programming questions. Provide clear, concise, and accurate information.",
    starter_prompt="Hello! I'm your Python assistant. How can I help you today?",
    context_files=[Path("docs/python_reference.md")],
    model=model_config,
    theme=theme,
    logo_path=Path("assets/python_logo.png"),
    show_system_prompt=True,
    show_context=True
)

# Create and launch the chat application
from gradiochat.ui import create_chat_app
app = create_chat_app(config)
app.build_interface().launch(share=True)
```

To close all existing clients.

``` python
gr.close_all()
```

## Developer Guide

I built this using `nbdev` [notebook development
tutorial](https://nbdev.fast.ai/getting_started.html) from the company
[fast.ai](https://www.fast.ai).

I also used the `uv` [python package and environment
manager](https://docs.astral.sh/uv/).

Officially `nbdev` doesn’t support `uv`. It works with `conda` or `pip`,
preferably with a single environment for all your Python projects if I
understand correctly. The `nbdev` package then handles the dependencies
and such via the `setings.ini` and `setup.py` and `requirements.txt`.
But don’t take my word for that, dive into the actual documentation.

I stumbled into some quircks trying to combine `nbdev` and `uv`. Most of
those are probably a result from `nbdev` needing `settings.ini` and/or
`settings.py`, while `uv` uses `pyproject.toml`. All of this, is to say.
You might possibly run into some issues running this package because I
wanted to do something that’s not officially supported or possible.

While developing I learned a few things that seemed to work after some
struggles.

1.  Make sure to use Python 3.10 if you want to use Gradio. This is
    strange. With `uv` I can use 3.11 and Gradio just fine. But the
    moment I push to Github the `nbdev` CI action fails with the
    statement that my app can at most support 3.10.16. I think that is
    because in Github `pip` is used and not `uv`.
2.  I also think you need Python 3.10 if you want to deploy to
    HuggingSpaces. Lower is possible, higher is not yet supported by
    HuggingSpaces.
3.  I thought that we needed Python 3.11 to be able to combine `nbdev`
    with `uv`. But I haven’t noticed any issues yet.

#### Install gradiochat in Development mode

``` sh
# make sure gradiochat package is installed in development mode
$ pip install -e .

# make changes under nbs/ directory
# ...

# compile to have changes apply to gradiochat
$ nbdev_prepare
```
