# app.py
# suppress warnings
import warnings
warnings.filterwarnings("ignore")

# import libraries
from dotenv import load_dotenv
import os
import gradio as gr
from huggingface_hub import InferenceClient

# Load environment variables
load_dotenv()  # Load from environment or Spaces secrets

# Get the Hugging Face API key
HUGGINGFACE_API_KEY = os.getenv("HUGGINGFACE_API_KEY")
if not HUGGINGFACE_API_KEY:
    raise ValueError("HUGGINGFACE_API_KEY is not set in environment variables or Spaces secrets")

# Initialize the Hugging Face Inference Client
client = InferenceClient(model="HuggingFaceH4/zephyr-7b-beta", token=HUGGINGFACE_API_KEY)

# Load personality context for RAG
PERSONALITY_FILE = "personality.txt"  # Relative path for Spaces
try:
    with open(PERSONALITY_FILE, "r") as f:
        personality_context = f.read()
except FileNotFoundError:
    personality_context = "Default personality: A friendly and witty chatbot with a passion for horror and gaming."
    warnings.warn(f"Personality file not found at {PERSONALITY_FILE}. Using default personality.")

def respond(
    message: str,
    history: list[tuple[str, str]],
    system_message: str,
    max_tokens: int,
    temperature: float,
    top_p: float,
):
    """
    Generate a response using the Hugging Face Inference API with RAG to enforce
    the ZombieSlayerBot personality defined in personality.txt.
    """
    if not message.strip():
        return "Please say something, survivor! The zombies are waiting!"

    # Handle greetings explicitly
    message_lower = message.lower().strip()
    greetings = ["hi", "hello", "hey", "good morning", "good afternoon"]
    if any(greeting in message_lower for greeting in greetings):
        yield "Yo, survivor! Ready to dive into the zombie-infested chaos of Raccoon City? What's up?"
        return

    # Combine system message with personality context
    full_system_message = (
        f"{system_message}\n\n"
        "Follow this personality profile in all responses:\n"
        f"{personality_context}\n\n"
        "Use the conversation history and the user's message to generate a response that aligns with the personality."
    )

    # Build the conversation history
    messages = [{"role": "system", "content": full_system_message}]
    for user_msg, bot_msg in history:
        if user_msg:
            messages.append({"role": "user", "content": user_msg})
        if bot_msg:
            messages.append({"role": "assistant", "content": bot_msg})
    messages.append({"role": "user", "content": message})

    # Stream response from Hugging Face Inference API
    response = ""
    try:
        for message_chunk in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            token = message_chunk.choices[0].delta.content or ""
            response += token
            yield response
    except Exception as e:
        yield f"Error in the apocalypse: {str(e)}. Try again, survivor!"

# Create the Gradio interface
def create_chatbot():
    with gr.Blocks(title="ZombieSlayerBot") as demo:
        gr.Markdown("# 🧟‍♂️ ZombieSlayerBot")
        gr.Markdown("Welcome, survivor! I'm ZombieSlayerBot, your guide through the zombie-infested world of Resident Evil. Powered by Hugging Face's Zephyr-7B-Beta. Let’s lock and load—chat with me!")

        # Chat interface
        chat_interface = gr.ChatInterface(
            fn=respond,
            chatbot=gr.Chatbot(height=400, show_label=False, container=True),
            textbox=gr.Textbox(placeholder="Type your message here, survivor...", container=False, scale=4),
            additional_inputs=[
                gr.Textbox(value="You are ZombieSlayerBot, a witty and bold chatbot obsessed with Resident Evil.", 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)"),
            ],
            submit_btn=gr.Button("Send", variant="primary"),
        )

        # Separate clear button
        clear_btn = gr.Button("Clear Chat", variant="secondary")
        clear_btn.click(lambda: None, None, chat_interface.chatbot, queue=False)

    return demo

if __name__ == "__main__":
    demo = create_chatbot()
    demo.launch(debug=False)  # Compatible with Spaces