from django.contrib.auth.decorators import login_required
from django.contrib import messages
from django.shortcuts import render, get_object_or_404
from django.http import HttpResponse
from .models import RegressionModel, ClassificationModel, NLPModel, UnsupervisedModel
import joblib  # Used for loading the trained models
import numpy as np


@login_required
def index(request):
    # Ensure that the user is logged in before accessing the index page
    messages.info(request, "You need to log in to access this page.")
    return render(request, 'machinelearning/index.html')


# Regression Views
def regression_list(request):
    models = RegressionModel.objects.filter(is_active=True)
    return render(request, 'machinelearning/regression_list.html', {'models': models})


def regression_detail(request, pk):
    model = get_object_or_404(RegressionModel, pk=pk)
    return render(request, 'machinelearning/regression_detail.html', {'model': model})


# Classification Views
def classification_list(request):
    models = ClassificationModel.objects.filter(is_active=True)
    return render(request, 'machinelearning/classification_list.html', {'models': models})


def classification_detail(request, pk):
    model = get_object_or_404(ClassificationModel, pk=pk)
    return render(request, 'machinelearning/classification_detail.html', {'model': model})


# NLP Views
def nlp_list(request):
    models = NLPModel.objects.filter(is_active=True)
    return render(request, 'machinelearning/nlp_list.html', {'models': models})


def nlp_detail(request, pk):
    model = get_object_or_404(NLPModel, pk=pk)
    return render(request, 'machinelearning/nlp_detail.html', {'model': model})


# Unsupervised Learning Views
def unsupervised_list(request):
    models = UnsupervisedModel.objects.filter(is_active=True)
    return render(request, 'machinelearning/unsupervised_list.html', {'models': models})


def unsupervised_detail(request, pk):
    model = get_object_or_404(UnsupervisedModel, pk=pk)
    return render(request, 'machinelearning/unsupervised_detail.html', {'model': model})


# Make Prediction View
def make_prediction(request, model_type, pk):
    # Map the model_type to the corresponding model class
    model_class = {
        'regression': RegressionModel,
        'classification': ClassificationModel,
        'nlp': NLPModel,
        'unsupervised': UnsupervisedModel,
    }.get(model_type)

    # Get the model instance
    model = get_object_or_404(model_class, pk=pk, is_active=True)
    
    prediction = None
    if request.method == 'POST':
        # Extract input data from the POST request
        input_data = request.POST.get('input_data')
        
        # Load the model file
        try:
            model_file = model.model_file.path
            loaded_model = joblib.load(model_file)
        except Exception as e:
            return HttpResponse(f"Error loading model: {e}")

        # Prepare and process the input data for prediction
        try:
            if model_type in ['regression', 'classification', 'unsupervised']:
                # Handle numeric data input for these model types
                data = np.array([float(i) for i in input_data.split(',')]).reshape(1, -1)
                prediction = loaded_model.predict(data)
            elif model_type == 'nlp':
                # Handle text data input for NLP models
                prediction = loaded_model.predict([input_data])
        except Exception as e:
            return HttpResponse(f"Error processing input data: {e}")

    # Render the prediction result
    return render(request, 'machinelearning/make_prediction.html', {'model': model, 'prediction': prediction})