import gradio as gr
import os
from huggingface_hub import InferenceClient
import cohere

HF_API_KEY = os.getenv("HF_API_KEY")
COHERE_API_KEY = os.getenv("COHERE_API_KEY")  # Get Cohere API key

models = ["HuggingFaceH4/zephyr-7b-beta", "meta-llama/Llama-3.2-3B-Instruct", "mistralai/Mistral-7B-Instruct-v0.3"]
client_hf = InferenceClient(model=models[2], token=HF_API_KEY)  # HF Client
client_cohere = cohere.Client(COHERE_API_KEY)  # Cohere Client

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
    use_cohere,  # Checkbox value
):
    messages = [{"role": "system", "content": system_message}]
    
    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

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

    if use_cohere:  # If Cohere is selected
        cohere_response = client_cohere.chat(
            message=message,
            model="command-r",  # Or "command" depending on your plan
            temperature=temperature,
            max_tokens=max_tokens
        )
        response = cohere_response.text
        yield response  # Yield full response (Cohere doesn't stream)
    
    else:  # If HF is selected
        for message in client_hf.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            token = message.choices[0].delta.content
            response += token
            yield response

# Gradio UI
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        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"),
        gr.Checkbox(label="Use Cohere API"),  # Checkbox to switch API
    ],
)

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