app.py

import os
import numpy as np
from flask import Flask, request, render_template, Response
from werkzeug.utils import secure_filename
from tensorflow.keras.preprocessing.image import load_img, img_to_array
from tensorflow.keras.models import load_model
from twilio.rest import Client
from twilio.twiml.messaging_response import MessagingResponse
import requests
from requests.auth import HTTPBasicAuth

# Initialize Flask app
app = Flask(__name__)

# --------------------------
# 1️⃣ Load Trained Model
# --------------------------
model = load_model('model_name.h5')
print('✅ Model loaded successfully!')

labels = {0: 'Healthy', 1: 'Powdery', 2: 'Rust'}
precautions = {
    "Rust": "🌿 Precaution for Rust: Remove infected leaves, avoid overhead watering, and apply a fungicide if necessary.",
    "Powdery": "🌿 Precaution for Powdery Mildew: Improve air circulation, avoid wetting leaves, and use sulfur-based fungicide if needed.",
    "Healthy": "✅ Your leaf looks healthy! Keep monitoring regularly and maintain good farm hygiene."
}


# --------------------------
# 2️⃣ Twilio Setup
# --------------------------
ACCOUNT_SID = 'sid token'
AUTH_TOKEN = 'auth token'  # Replace with your Twilio Auth Token
FROM_WHATSAPP = "whatsapp:no"  # Twilio sandbox number

client = Client(ACCOUNT_SID, AUTH_TOKEN)

# --------------------------
# 3️⃣ Helper: Prediction
# --------------------------
def getResult(image_path):
    img = load_img(image_path, target_size=(225, 225))
    x = img_to_array(img)
    x = x.astype('float32') / 255.
    x = np.expand_dims(x, axis=0)
    predictions = model.predict(x)[0]
    predicted_label = labels[np.argmax(predictions)]
    return predicted_label

# --------------------------
# 4️⃣ Web Interface Route
# --------------------------
@app.route('/', methods=['GET'])
def index():
    return render_template('index.html')

@app.route('/predict', methods=['GET', 'POST'])
def upload():
    if request.method == 'POST':
        f = request.files['file']

        basepath = os.path.dirname(__file__)
        upload_dir = os.path.join(basepath, 'uploads')
        os.makedirs(upload_dir, exist_ok=True)

        file_path = os.path.join(upload_dir, secure_filename(f.filename))
        f.save(file_path)

        predicted_label = getResult(file_path)
        return str(predicted_label)
    return None

# --------------------------
# 5️⃣ WhatsApp Integration Route
# --------------------------
@app.route("/whatsapp", methods=["POST"])
def whatsapp_reply():
    """
    Handles incoming WhatsApp messages from Twilio.
    Farmers can send leaf images and get predictions instantly.
    """
    # Log incoming Twilio data for debugging
    print("Incoming Twilio data:", request.form)

    media_url = request.form.get('MediaUrl0')
    resp = MessagingResponse()
    reply = resp.message()

    if media_url:
        try:
            # Ensure uploads folder exists
            os.makedirs("uploads", exist_ok=True)
            img_path = os.path.join("uploads", "whatsapp_leaf.jpg")

            # Download media with Twilio authentication
            img_data = requests.get(media_url, auth=HTTPBasicAuth(ACCOUNT_SID, AUTH_TOKEN)).content
            with open(img_path, "wb") as f:
                f.write(img_data)

            # Predict leaf disease
            predicted_label = getResult(img_path)
            precaution_text = precautions.get(predicted_label, "")
            reply.body(f"🍃 Prediction: {predicted_label}\n{precaution_text}")

        except Exception as e:
            print("Error processing image:", e)
            reply.body("⚠️ Error processing the image. Please try again.")

    else:
        reply.body("Please send a *leaf image* of apple tree to detect disease 🌿")

    # Return proper TwiML response with Content-Type
    return Response(str(resp), mimetype="text/xml")

# --------------------------
# 6️⃣ Optional: Send proactive WhatsApp message from Flask
# --------------------------
def send_whatsapp_message(to_number, message_text):
    client.messages.create(
        from_=FROM_WHATSAPP,
        body=message_text,
        to=f"whatsapp:{to_number}"
    )

# --------------------------
# 7️⃣ Run Flask App
# --------------------------
if __name__ == '__main__':
    app.run(host="0.0.0.0", port=5000, debug=True)