from PIL import Image
import os
import numpy as np
import tensorflow as tf


data = "DATA"

exit()


def predict(values, dic):
    # diabetes
    if len(values) == 8:
        dic2 = {'NewBMI_Obesity 1': 0, 'NewBMI_Obesity 2': 0, 'NewBMI_Obesity 3': 0, 'NewBMI_Overweight': 0,
                'NewBMI_Underweight': 0, 'NewInsulinScore_Normal': 0, 'NewGlucose_Low': 0,
                'NewGlucose_Normal': 0, 'NewGlucose_Overweight': 0, 'NewGlucose_Secret': 0}

        if dic['BMI'] <= 18.5:
            dic2['NewBMI_Underweight'] = 1
        elif 18.5 < dic['BMI'] <= 24.9:
            pass
        elif 24.9 < dic['BMI'] <= 29.9:
            dic2['NewBMI_Overweight'] = 1
        elif 29.9 < dic['BMI'] <= 34.9:
            dic2['NewBMI_Obesity 1'] = 1
        elif 34.9 < dic['BMI'] <= 39.9:
            dic2['NewBMI_Obesity 2'] = 1
        elif dic['BMI'] > 39.9:
            dic2['NewBMI_Obesity 3'] = 1

        if 16 <= dic['Insulin'] <= 166:
            dic2['NewInsulinScore_Normal'] = 1

        if dic['Glucose'] <= 70:
            dic2['NewGlucose_Low'] = 1
        elif 70 < dic['Glucose'] <= 99:
            dic2['NewGlucose_Normal'] = 1
        elif 99 < dic['Glucose'] <= 126:
            dic2['NewGlucose_Overweight'] = 1
        elif dic['Glucose'] > 126:
            dic2['NewGlucose_Secret'] = 1

        dic.update(dic2)
        values2 = list(map(float, list(dic.values())))

        model = pickle.load(open('models/diabetes.pkl','rb'))
        values = np.asarray(values2)
        return model.predict(values.reshape(1, -1))[0]

    # breast_cancer
    elif len(values) == 22:
        model = pickle.load(open('models/breast_cancer.pkl','rb'))
        values = np.asarray(values)
        return model.predict(values.reshape(1, -1))[0]

    # heart disease
    elif len(values) == 13:
        model = pickle.load(open('models/heart.pkl','rb'))
        values = np.asarray(values)
        return model.predict(values.reshape(1, -1))[0]

    # kidney disease
    elif len(values) == 24:
        model = pickle.load(open('models/kidney.pkl','rb'))
        values = np.asarray(values)
        return model.predict(values.reshape(1, -1))[0]

    # liver disease
    elif len(values) == 10:
        model = pickle.load(open('models/liver.pkl','rb'))
        values = np.asarray(values)
        return model.predict(values.reshape(1, -1))[0]


def predict_api():
    try:
        to_predict_dict = request.form.to_dict()

        for key, value in to_predict_dict.items():
            try:
                to_predict_dict[key] = int(value)
            except ValueError:
                to_predict_dict[key] = float(value)

        to_predict_list = list(map(float, list(to_predict_dict.values())))
        pred = predict(to_predict_list, to_predict_dict)

        return {'prediction': pred}

    except Exception as e:
        return {'error': str(e)}


def malariapredict_api():
    try:
        img = Image.open(request.files['image'])
        img.save("uploads/image.jpg")
        img_path = os.path.join(os.path.dirname(__file__), 'uploads/image.jpg')
        os.path.isfile(img_path)
        img = tf.keras.utils.load_img(img_path, target_size=(128, 128))
        img = tf.keras.utils.img_to_array(img)
        img = np.expand_dims(img, axis=0)

        model = tf.keras.models.load_model("models/malaria.h5")
        pred = np.argmax(model.predict(img))

        return {'prediction': int(pred)}

    except Exception as e:
        return {'error': str(e)}


def pneumoniapredict_api():
    try:
        img = Image.open(request.files['image']).convert('L')
        img.save("uploads/image.jpg")
        img_path = os.path.join(os.path.dirname(__file__), 'uploads/image.jpg')
        os.path.isfile(img_path)
        img = tf.keras.utils.load_img(img_path, target_size=(128, 128))
        img = tf.keras.utils.img_to_array(img)
        img = np.expand_dims(img, axis=0)

        model = tf.keras.models.load_model("models/pneumonia.h5")
        pred = np.argmax(model.predict(img))

        return {'prediction': int(pred)}

    except Exception as e:
        return {'error': str(e)}