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

import io


import google.generativeai as genai


def import_and_predict(image_data, model, class_labels):
    size = (256, 256)

    if image_data is not None:
        image = Image.open(io.BytesIO(image_data.read()))
        image = image.resize(size)
        image = np.array(image)
        img_reshape = image / 255.0
        img_reshape = np.expand_dims(img_reshape, axis=0)

        prediction = model.predict(img_reshape)
        st.image(image, width=300)
        predictions_label = class_labels[np.argmax(prediction[0])]
        return predictions_label
    else:
        st.warning("Please upload an image.")
        return None


def get_info_from_gemini(prompt):
  genai.configure(api_key=os.environ.get('gemini_api'))
  model = genai.GenerativeModel('gemini-pro')
  response = model.generate_content(f"{prompt}")
  return response




st.title("Plant Disease Detection")

uploaded_image = st.file_uploader(f"Upload an image", type=["jpg", "jpeg", "png"])
models_path = ['./best_model_100_subset.h5',]



CLASS_LABELS = ['Tomato Early blight', 'Tomato Leaf Mold', 'Tomato YellowLeaf Curl Virus',
                                 'Tomato mosaic virus', 'Tomato healthy']
                
model = tf.keras.models.load_model(models_path[0])




prediction = import_and_predict(uploaded_image, model, CLASS_LABELS)
st.write("disease name: ", prediction)


if prediction != None:
            new_title = '<p style="font-size: 38px">Measures you can take to control: </p>'
            st.markdown(new_title, unsafe_allow_html=True)
            if prediction == CLASS_LABELS[4]:
                st.write("Plant is healthy take good care of it")
            response = get_info_from_gemini(f"cure for the disease {prediction} tell in bulletpoints and estimated cost in inr at last give summary of each measures estimated cost")
            st.write(response.text)