import gradio as gr
from main import (
    graph_app,
    HumanMessage,
    AIMessage,
    SystemMessage,
    DEFAULT_SYSTEM_PROMPT,
    app as fastapi_app
)

def chat_fn(message, history):
    messages = [SystemMessage(content=DEFAULT_SYSTEM_PROMPT)]
    if history:
        for turn in history:
            if isinstance(turn, dict):
                role = turn.get("role", "user")
                content = turn.get("content", "")
                if role == "assistant":
                    messages.append(AIMessage(content=content))
                else:
                    messages.append(HumanMessage(content=content))
            elif isinstance(turn, (list, tuple)) and len(turn) == 2:
                if turn[0]:
                    messages.append(HumanMessage(content=str(turn[0])))
                if turn[1]:
                    messages.append(AIMessage(content=str(turn[1])))

    messages.append(HumanMessage(content=message))
    output = graph_app.invoke({"messages": messages})
    return output["messages"][-1].content

# Interfaz interactiva de Chat de Gradio (Modo Ilimitado 24/7 en CPU Basic)
demo = gr.ChatInterface(
    fn=chat_fn,
    title="⚡ Qwen 2.5 (0.5B) • Intelecto Conciso",
    description="Chatbot superinteligente, hiperpreciso y conciso en español. Ilimitado 24/7 con LangGraph y llama-cpp-python en Hugging Face Spaces (16 GB RAM).",
    examples=["¿Qué es la entropía?", "¿Por qué el cielo es azul?", "¿Qué es la vida?"],
)

# Integrar las rutas de FastAPI (/chat, /healthz, /docs) en la misma app
demo.app.include_router(fastapi_app.router)

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