app.py

import os
import joblib
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from flask import Flask, render_template, request, jsonify
from werkzeug.utils import secure_filename
from spectral.io import envi

# Make a directory for uploads if it doesn't exist
script_dir = os.path.dirname(os.path.abspath(__file__))

# Build full path to the pkl files
model_path = os.path.join(script_dir, 'optimized_soil_carbon_model.pkl')
selector_path = os.path.join(script_dir, 'feature_selector.pkl')

# Load the files using full path
model = joblib.load(model_path)
selector = joblib.load(selector_path)

# Flask app setup
app = Flask(__name__)
app.config['UPLOAD_FOLDER'] = 'uploads'

# Ensure required folders exist
os.makedirs('uploads', exist_ok=True)
os.makedirs('static', exist_ok=True)

# Global variables to store map and image shape
org_c_map = None
rows, cols = 0, 0

# Allowed extensions
ALLOWED_EXTENSIONS_HDR = {'hdr'}
ALLOWED_EXTENSIONS_BIN = {'bin'}

# Function to check allowed file extensions
def allowed_file(filename, allowed_ext):
    return '.' in filename and filename.rsplit('.', 1)[1].lower() in allowed_ext

# Function to add spectral indices
def add_spectral_indices(X):
    df = X.copy()
    if 'W700' in df.columns and 'W1700' in df.columns:
        df['NDI_700_1700'] = (df['W700'] - df['W1700']) / (df['W700'] + df['W1700'])
    if 'W600' in df.columns and 'W2100' in df.columns:
        df['SR_600_2100'] = df['W600'] / df['W2100']
    if 'W2200' in df.columns:
        df['CR_2200'] = df['W2200'] / (0.5 * (df.get('W2100', 1) + df.get('W2300', 1)))
    return df

@app.route('/', methods=['GET', 'POST'])
def index():
    global org_c_map, rows, cols
    value = None
    error = None

    if request.method == 'POST':
        # Get uploaded files
        hdr_file = request.files['hdr_file']
        bin_file = request.files['bin_file']

        # Check extensions
        if not (allowed_file(hdr_file.filename, ALLOWED_EXTENSIONS_HDR) and
                allowed_file(bin_file.filename, ALLOWED_EXTENSIONS_BIN)):
            error = "Invalid file type! Please upload a valid .hdr file and .bin file."
            return render_template('index.html', rows=rows, cols=cols, value=value, error=error)

        # Secure filenames
        hdr_filename = secure_filename(hdr_file.filename)
        bin_filename = secure_filename(bin_file.filename)

        # Save files
        hdr_path = os.path.join(app.config['UPLOAD_FOLDER'], hdr_filename)
        bin_path = os.path.join(app.config['UPLOAD_FOLDER'], bin_filename)

        hdr_file.save(hdr_path)
        bin_file.save(bin_path)

        try:
            # Load image
            img = envi.open(hdr_path, bin_path)
            cube = img.load()  # shape: (rows, cols, bands)

            # Get wavelengths
            wavelengths = img.metadata.get("wavelength")
            wavelengths = [float(w) for w in wavelengths]

            # Prepare data
            rows, cols, bands = cube.shape
            pixel_data = cube.reshape(-1, bands)

            band_names = [f"W{int(w)}" for w in wavelengths]
            pixel_df = pd.DataFrame(pixel_data, columns=band_names)

            # Add spectral indices
            pixel_df = add_spectral_indices(pixel_df)

            # Select features
            selected_columns = selector.get_support(indices=True)
            selected_features = pixel_df.columns[selected_columns]
            X_selected = pixel_df[selected_features]

            # Predict
            y_pred_flat = model.predict(X_selected)
            org_c_map = y_pred_flat.reshape(rows, cols)

            # Save map image
            plt.figure(figsize=(10, 6))
            plt.imshow(org_c_map, cmap='YlGn', interpolation='none')
            plt.title("Predicted Soil Organic Carbon Map")
            plt.colorbar(label='Org C (%)')
            plt.axis('off')
            plt.savefig('static/output_map.png')
            plt.close()

        except Exception as e:
            # Clear previous map and dimensions
            org_c_map = None
            rows, cols = 0, 0
            error = f"An error occurred while processing the image: {e}"
            return render_template('index.html', rows=rows, cols=cols, value=value, error=error)


    return render_template('index.html', rows=rows, cols=cols, value=value, error=error)

@app.route('/get_value', methods=['POST'])
def get_value():
    global org_c_map, rows, cols

    # Get requested row/col
    row = int(request.form['row'])
    col = int(request.form['col'])

    # Clip values to valid range
    row = max(0, min(row, rows - 1))
    col = max(0, min(col, cols - 1))

    # Get value
    value = org_c_map[row, col] if org_c_map is not None else None

    # Re-render index with value shown
    return jsonify({'value': f"{value:.3f} %" if value is not None else "N/A"})

if __name__ == '__main__':
    app.run(debug=True)