from fastapi import HTTPException, UploadFile
from src.services.preprocess_features import preprocess_features
from src.services.file_verification import file_verification
from src.services.report import summarize_predictions

from src.agents.l1_screener import Screener
from src.agents.l2_supervisor import Supervisor
from src.agents.l3_classifier import Classifier

required_columns = [
    "Header_Length", "Protocol Type", "Time_To_Live", "Rate",
    "fin_flag_number", "syn_flag_number", "rst_flag_number",
    "psh_flag_number", "ack_flag_number", "ece_flag_number",
    "cwr_flag_number", "ack_count", "syn_count", "fin_count",
    "rst_count", "TCP", "UDP", "Tot sum", "Min", "Max", "AVG",
    "Std", "Tot size", "IAT", "Number", "Variance"
]

def global_prediction_on_csv(file: UploadFile):
    try:
        # Vérifier l'extension et les colonnes du fichier
        correct_csv = file_verification(file)

        # augmenter le nombre de features
        data = preprocess_features(correct_csv)


        print("start loading model and predicting")

        screener = Screener()
        l1_summary = summarize_predictions(screener.predict, data)
        
        supervisor = Supervisor()
        l2_summary = summarize_predictions(supervisor.predict, data)
        
        # classifier = Classifier()
        # l3_summary = summarize_predictions(classifier.predict, data)
        

        return {
                "l1": l1_summary,
                "l2": l2_summary,
                # "l3": l3_summary
                }

    except HTTPException:
        raise
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

def single_prediction_controller(data: dict):
    try:
        if not isinstance(data, dict):
            raise HTTPException(status_code=400, detail="Invalid JSON payload")

        missing_columns = [col for col in required_columns if col not in data]
        if missing_columns:
            raise HTTPException(
                status_code=422,
                detail=f"Missing required columns: {missing_columns}"
            )

        # Ici tu peux faire la prédiction
        print("Received data for single prediction:", data)
        return {"message": "All required columns present", "to_do": "Not yet implemented"}

    except HTTPException:
        raise
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))