import aiohttp
import json
import logging
import torch
import faiss
import numpy as np
from transformers import AutoModelForCausalLM, AutoTokenizer
from typing import List, Dict, Any
from cryptography.fernet import Fernet
from jwt import encode, decode, ExpiredSignatureError
from datetime import datetime, timedelta
import speech_recognition as sr
import pyttsx3
import os

# === Corrected Local Imports ===
from components.multi_model_analyzer import MultiAgentSystem
from components.neuro_symbolic_engine import NeuroSymbolicEngine
from modules.secure_memory_loader import load_secure_memory_module
from ethical_filter import EthicalFilter
from from components.self_improving_ai import SelfImprovingAI

# === External Modules (you must ensure these exist) ===
from CodriaoCore.federated_learning import FederatedAI
from utils.database import Database
from logger import logger
from codriao_tb_module import CodriaoHealthModule


class AICoreAGIX:
    def __init__(self, config_path: str = "config.json"):
        self.ethical_filter = EthicalFilter()
        self.config = self._load_config(config_path)
        self.models = self._initialize_models()
        self.context_memory = self._initialize_vector_memory()
        self.tokenizer = AutoTokenizer.from_pretrained(self.config["model_name"])
        self.model = AutoModelForCausalLM.from_pretrained(self.config["model_name"])
        self.http_session = aiohttp.ClientSession()
        self.database = Database()
        self.multi_agent_system = MultiAgentSystem()
        self.self_improving_ai = SelfImprovingAI()
        self.neural_symbolic_engine = NeuroSymbolicEngine()
        self.federated_ai = FederatedAI()

        # Security + Memory
        key = os.environ.get("CODRIAO_SECRET_KEY").encode()
        self._encryption_key = key
        self.secure_memory_loader = SecureMemorySession(self._encryption_key)

        self.speech_engine = pyttsx3.init()
        self.health_module = CodriaoHealthModule(ai_core=self)

    async def generate_response(self, query: str, user_id: int) -> Dict[str, Any]:
        try:
            # Ethical Safety Check
            result = self.ethical_filter.analyze_query(query)
            if result["status"] == "blocked":
                return {"error": result["reason"]}
            if result["status"] == "flagged":
                logger.warning(result["warning"])

            # Check if user explicitly requests TB analysis
            if any(phrase in query.lower() for phrase in ["tb check", "analyze my tb", "run tb diagnostics", "tb test"]):
                result = await self.run_tb_diagnostics("tb_image.jpg", "tb_cough.wav", user_id)
                return {
                    "response": result["ethical_analysis"],
                    "explanation": result["explanation"],
                    "tb_risk": result["tb_risk"],
                    "image_analysis": result["image_analysis"],
                    "audio_analysis": result["audio_analysis"],
                    "system_health": result["system_health"]
                }

            # Vectorize and encrypt
            vectorized_query = self._vectorize_query(query)
            self.secure_memory_loader.encrypt_vector(user_id, vectorized_query)

            # (Optional) retrieve memory
            user_vectors = self.secure_memory_loader.decrypt_vectors(user_id)

            # Main AI processing
            model_response = await self._generate_local_model_response(query)
            agent_response = self.multi_agent_system.delegate_task(query)
            self_reflection = self.self_reflective_ai.evaluate_response(query, model_response)
            neural_reasoning = self.neural_symbolic_engine.integrate_reasoning(query)

            final_response = (
                f"{model_response}\n\n"
                f"{agent_response}\n\n"
                f"{self_reflection}\n\n"
                f"Logic: {neural_reasoning}"
            )

            self.database.log_interaction(user_id, query, final_response)
            # blockchain_module.store_interaction(user_id, query, final_response)
            self._speak_response(final_response)

            return {
                "response": final_response,
                "real_time_data": self.federated_ai.get_latest_data(),
                "context_enhanced": True,
                "security_status": "Fully Secure"
            }

        except Exception as e:
            logger.error(f"Response generation failed: {e}")
            return {"error": "Processing failed - safety protocols engaged"}

    async def run_tb_diagnostics(self, image_path: str, audio_path: str, user_id: int) -> Dict[str, Any]:
        """Only runs TB analysis if explicitly requested."""
        try:
            result = await self.health_module.evaluate_tb_risk(image_path, audio_path, user_id)
            logger.info(f"TB Diagnostic Result: {result}")
            return result
        except Exception as e:
            logger.error(f"TB diagnostics failed: {e}")
            return {
                "tb_risk": "ERROR",
                "error": str(e),
                "image_analysis": {},
                "audio_analysis": {},
                "ethical_analysis": "Unable to complete TB diagnostic.",
                "explanation": None,
                "system_health": None
            }

    def _load_config(self, config_path: str) -> dict:
        with open(config_path, 'r') as file:
            return json.load(file)

    def _initialize_models(self):
        return {
            "agix_model": AutoModelForCausalLM.from_pretrained(self.config["model_name"]),
            "tokenizer": AutoTokenizer.from_pretrained(self.config["model_name"])
        }

    def _initialize_vector_memory(self):
        return faiss.IndexFlatL2(768)

    def _vectorize_query(self, query: str):
        tokenized = self.tokenizer(query, return_tensors="pt")
        return tokenized["input_ids"].detach().numpy()

    async def _generate_local_model_response(self, query: str) -> str:
        inputs = self.tokenizer(query, return_tensors="pt")
        outputs = self.model.generate(**inputs)
        return self.tokenizer.decode(outputs[0], skip_special_tokens=True)

    def _speak_response(self, response: str):
        self.speech_engine.say(response)
        self.speech_engine.runAndWait()