import os
import gradio as gr
from src.const.agent_response_constants import *
from src.rag.agent_chain import ExecutiveAgentChain
from src.rag.utilclasses import LeadAgentQueryResponse
from src.utils.logging import get_logger
from src.cache.cache import Cache

logger = get_logger("chatbot_app")
cache_logger = get_logger("cache_chatbot_app")

class ChatbotApplication:
    def __init__(self, language: str = 'de') -> None:
        self._app = gr.Blocks()
        self._language = language
        self._cache = Cache.get_cache()

        with self._app:
            agent_state = gr.State(None)
            lang_state = gr.State(language)

            with gr.Row():
                lang_selector = gr.Radio(
                    choices=["Deutsch", "English"],
                    value="English" if language == 'en' else 'Deutsch',
                    label="Selected Language",
                    interactive=True,
                )
                reset_button = gr.Button("Reset Conversation")

            chatbot = gr.Chatbot(
                height=600,
                type='messages',
                label="Executive Education Adviser"
            )
            chat = gr.ChatInterface(
                fn=lambda msg, history, agent: self._chat(
                    message=msg,
                    history=history,
                    agent=agent,
                ),
                additional_inputs=[agent_state],
                title="Executive Education Adviser",
                type='messages',
                chatbot=chatbot,
                fill_height=True
            )

            def clear_chat_immediate():
                return [], ""

            def on_lang_change(language):
                lang_code = 'en' if language == 'English' else 'de'
                return switch_language(lang_code)

            def initalize_agent(language):
                agent = ExecutiveAgentChain(language=language)
                greeting = agent.generate_greeting()
                return agent, [{"role": "assistant", "content": greeting}]

            def switch_language(new_language):
                new_agent, greeting = initalize_agent(new_language)
                return (
                    new_agent,
                    new_language,
                    greeting,
                    ""
                )

            lang_selector.change(
                fn=clear_chat_immediate,
                outputs=[chat.chatbot_value],
                queue=True,
            )

            lang_selector.change(
                fn=on_lang_change,
                inputs=[lang_selector],
                outputs=[agent_state, lang_state, chat.chatbot_value],
                queue=True,
            )

            reset_button.click(
                fn=clear_chat_immediate,
                outputs=[chat.chatbot_value],
                queue=True,
            )

            reset_button.click(
                fn=switch_language,
                inputs=[lang_state],
                outputs=[agent_state, lang_state, chat.chatbot_value],
                queue=True,
            )

            # Initialize the agent chain on the app startup
            self._app.load(
                fn=lambda: initalize_agent(self._language),
                outputs=[agent_state, chat.chatbot_value],
            )

    @property
    def app(self) -> gr.Blocks:
        """Expose underlying Gradio Blocks for external runners (e.g., HF Spaces)."""
        return self._app

    def _chat(self, message: str, history: list[dict], agent: ExecutiveAgentChain):
        if agent is None:
            logger.error("Agent not initialized")
            return ["I apologize, but the chatbot is not properly initialized."]

        answers = []
        try:
            logger.info(f"Processing user query: {message[:100]}...")

            preprocess_resp = agent.preprocess_query(message)
            final_response: LeadAgentQueryResponse = None

            current_lang = preprocess_resp.language
            processed_q = preprocess_resp.processed_query

            if preprocess_resp.response:
                # Response comes from preprocessing step
                final_response = preprocess_resp

            elif Cache._settings["enabled"]:
                cached_data = self._cache.get(processed_q, language=current_lang)

                if cached_data:
                    # Cache Hit — restore response with metadata
                    if isinstance(cached_data, dict):
                        final_response = LeadAgentQueryResponse(
                            response=cached_data["response"],
                            language=current_lang,
                            appointment_requested=cached_data.get("appointment_requested", False),
                            relevant_programs=cached_data.get("relevant_programs", []),
                        )
                    else:
                        # Legacy: plain string cache entry
                        final_response = LeadAgentQueryResponse(
                            response=cached_data,
                            language=current_lang,
                        )

            if not final_response:
                # Response needs to be generated by the agent
                final_response = agent.agent_query(processed_q)

            answers.append(final_response.response)
            self._language = final_response.language

            if final_response.confidence_fallback or final_response.max_turns_reached or final_response.appointment_requested:
                html_code = get_booking_widget(language=self._language, programs=final_response.relevant_programs)
                answers.append(gr.HTML(value=html_code))

            if final_response.should_cache and Cache._settings["enabled"]:
                # Caching response with metadata
                self._cache.set(
                    key=processed_q,
                    value={
                        "response": final_response.response,
                        "appointment_requested": final_response.appointment_requested,
                        "relevant_programs": final_response.relevant_programs,
                    },
                    language=current_lang
                )

        except Exception as e:
            logger.error(f"Error processing query: {e}", exc_info=True)
            error_message = (
                "I apologize, but I encountered an error processing your request. "
                "Please try rephrasing your question or contact our admissions team for assistance."
            )
            answers.append(error_message)

        return answers

    def run(self):
        self._app.launch(
            share=os.getenv("GRADIO_SHARE", "false").lower() == "true",
            server_name=os.getenv("SERVER_NAME", "0.0.0.0"),
            server_port=int(os.getenv("PORT", 7860)),
        )