import gradio as gr
from huggingface_hub import InferenceClient

"""
For more information on `huggingface_hub` Inference API support, please check the docs:
https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
"""

def respond(message, history, token, model, system_message, max_tokens, temperature, top_p):
    """
    Handle chat responses using the Hugging Face Inference API.
    
    Parameters:
    - message: The user's current message.
    - history: List of previous user-assistant message pairs.
    - token: HF API token for authentication.
    - model: Model name to use for inference.
    - system_message: System prompt to initialize the chat.
    - max_tokens: Maximum number of tokens to generate.
    - temperature: Sampling temperature.
    - top_p: Top-p (nucleus) sampling parameter.
    
    Yields:
    - Incremental responses for streaming in the chat interface.
    """
    # Check for missing token
    if not token:
        yield "Please provide an HF API Token."
        return

    # Use default model if none provided
    if not model:
        model = "meta-llama/Llama-3.1-8B-Instruct"

    # Initialize the InferenceClient
    try:
        client = InferenceClient(model=model, token=token)
    except Exception as e:
        yield f"Error initializing client: {str(e)}"
        return

    # Build the message history, starting with the system message
    messages = [{"role": "system", "content": system_message}]

    for val in history:
        if val[0]:  # User message
            messages.append({"role": "user", "content": val[0]})
        if val[1]:  # Assistant message
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message})

    response = ""

    # Make the API call with streaming
    try:
        for message in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            # Check for non-empty content in the delta
            if message.choices and message.choices[0].delta.content is not None:
                token = message.choices[0].delta.content
                response += token
                yield response
    except Exception as e:
        yield f"Error during API call: {str(e)}"

# Define input components
token_input = gr.Textbox(type="password", label="HF API Token")
model_input = gr.Textbox(label="Model Name", value="HuggingFaceH4/zephyr-7b-beta")

"""
For information on how to customize the ChatInterface, peruse the gradio docs:
https://www.gradio.app/docs/chatinterface
"""
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        token_input,
        model_input,
        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
    ],
)

if __name__ == "__main__":
    demo.launch()