from sentinelhub import SHConfig, MimeType, CRS, BBox, DataCollection, SentinelHubRequest, bbox_to_dimensions
import numpy as np
from rasterio import features
import geojson
from shapely.geometry import shape

# Sentinel Hub configuration
config = SHConfig()
config.sh_client_id = '2a2bea5f-7441-4631-87b6-92fb461cdc85'
config.sh_client_secret = 'o889KWLdDrlEgJhxqkUnpx2hvO6QnWHS'

def detect_sugarcane(bbox_coordinates, time_interval=('2024-07-01', '2025-02-11')):
    """
    Enhanced sugarcane detection using multi-index analysis and green band
    
    Parameters:
        bbox_coordinates: tuple (west, south, east, north) in WGS84
        time_interval: tuple (start_date, end_date)
        
    Returns:
        GeoJSON FeatureCollection
    """
    # Initialize request
    bbox = BBox(bbox=bbox_coordinates, crs=CRS.WGS84)
    size = bbox_to_dimensions(bbox, resolution=10)
    
    evalscript = """
    //VERSION=3
    function setup() {
        return {
            input: ["B03", "B04", "B05", "B06", "B08", "B11", "B12", "SCL"],
            output: { bands: 8 }
        };
    }
    function evaluatePixel(sample) {
        return [
            sample.B03, sample.B04, sample.B05,
            sample.B06, sample.B08, sample.B11,
            sample.B12, sample.SCL
        ];
    }
    """

    request = SentinelHubRequest(
        evalscript=evalscript,
        input_data=[
            SentinelHubRequest.input_data(
                data_collection=DataCollection.SENTINEL2_L2A,
                time_interval=time_interval,
                mosaicking_order='leastCC'
            )
        ],
        responses=[
            SentinelHubRequest.output_response('default', MimeType.TIFF)
        ],
        bbox=bbox,
        size=size,
        config=config
    )

    # Process data
    data = request.get_data()[0]
    green = data[:, :, 0]
    red = data[:, :, 1]
    red_edge1 = data[:, :, 2]
    red_edge2 = data[:, :, 3]
    nir = data[:, :, 4]
    swir1 = data[:, :, 5]
    swir2 = data[:, :, 6]
    scl = data[:, :, 7]

    # Cloud masking
    cloud_mask = (scl != 8) & (scl != 9) & (scl != 4) & (scl != 10)
    
    # Calculate indices
    ndvi = (nir - red) / (nir + red + 1e-6)
    ndre = (nir - red_edge1) / (nir + red_edge1 + 1e-6)
    ndwi = (green - nir) / (green + nir + 1e-6)
    ndmi = (nir - swir1) / (nir + swir1 + 1e-6)
    
    # Sugarcane spectral thresholds
    sugarcane_mask = (
        (ndvi > 0.6) &
        (ndre > 0.4) &
        (ndwi < 0.2) &
        (ndmi < 0.15) &
        cloud_mask
    )

    # Calculate affine transform
    west, south, east, north = bbox_coordinates
    width, height = size
    pixel_width = (east - west) / width
    pixel_height = (south - north) / height
    transform = (west, pixel_width, 0, north, 0, -pixel_height)

    # Vectorization
    shapes = features.shapes(sugarcane_mask.astype(np.int16), transform=transform)
    features_list = []

    for shape, value in shapes:
        if value == 1:
            geom = shape(shape.__geo_interface__)
            if geom.area > 10000:  # Minimum field size (1 hectare)
                features_list.append(
                    geojson.Feature(geometry=geom.__geo_interface__, properties={
                        'ndvi': float(ndvi[int(geom.centroid.y), int(geom.centroid.x)]),
                        'ndre': float(ndre[int(geom.centroid.y), int(geom.centroid.x)])
                    })
                )

    return geojson.FeatureCollection(features_list)

# Example usage
if __name__ == "__main__":
    # Define bounding box (WGS84 coordinates)
    bbox = (16.599503, 75.207556, 16.660316, 75.293479)  # Example area
    
    # Detect sugarcane
    sugarcane_geojson = detect_sugarcane(bbox)
    
    # Save to file
    with open('sugarcane.geojson', 'w') as f:
        geojson.dump(sugarcane_geojson, f)