app.py

import os
import zipfile
import tempfile
import uuid
import rasterio
import numpy as np
import matplotlib.pyplot as plt
import gradio as gr
from rasterio.enums import Resampling

def generate_temp_path(suffix=".tif"):
    return os.path.join(tempfile.gettempdir(), f"{uuid.uuid4().hex}{suffix}")

def read_band(path):
    with rasterio.open(path) as src:
        return src.read(1).astype(np.float32), src.profile

def resample_to_10m(band_path, ref_shape, ref_transform):
    with rasterio.open(band_path) as src:
        data = src.read(
            out_shape=(1, ref_shape[0], ref_shape[1]),
            resampling=Resampling.bilinear
        )
        return data[0].astype(np.float32)

def normalize(array):
    array /= 10000.0
    return np.clip(array, 0, 1)

def array_to_plot(img, title, cmap=None):
    fig = plt.figure(figsize=(6, 6))
    if cmap:
        plt.imshow(img, cmap=cmap)
        plt.colorbar()
    else:
        plt.imshow(img)
    plt.title(title)
    plt.axis('off')
    return fig

def save_tif(path, array, profile, count=3):
    profile = profile.copy()
    profile.update({
        'driver': 'GTiff',
        'count': count,
        'dtype': rasterio.float32,
        'compress': 'deflate',
        'predictor': 2,
        'tiled': True,
        'blockxsize': 512,
        'blockysize': 512
    })
    with rasterio.open(path, 'w', **profile) as dst:
        if count == 1:
            dst.write(array, 1)
        else:
            for i in range(count):
                dst.write(array[:, :, i], i + 1)

def process_visualization(zip_file_path, vis_type):
    with tempfile.TemporaryDirectory() as temp_dir:
        with zipfile.ZipFile(zip_file_path, 'r') as zip_ref:
            zip_ref.extractall(temp_dir)

        dirs = [d for d in os.listdir(temp_dir) if d.endswith(".SAFE")]
        if not dirs:
            raise Exception(".SAFE folder not found inside the zip file.")
        extract_dir = os.path.join(temp_dir, dirs[0])

        granule_dir = os.path.join(extract_dir, "GRANULE")
        granule_path = [os.path.join(granule_dir, d) for d in os.listdir(granule_dir) if d.startswith("L2A_")][0]
        img_data_dir = os.path.join(granule_path, "IMG_DATA")

        res_paths = {
            "R10m": os.path.join(img_data_dir, "R10m"),
            "R20m": os.path.join(img_data_dir, "R20m")
        }

        # Geometry reference
        b4, profile = read_band(os.path.join(res_paths["R10m"], [f for f in os.listdir(res_paths["R10m"]) if "_B04" in f][0]))

        if vis_type == "Natural Color (B4, B3, B2)":
            b2, _ = read_band(os.path.join(res_paths["R10m"], [f for f in os.listdir(res_paths["R10m"]) if "_B02" in f][0]))
            b3, _ = read_band(os.path.join(res_paths["R10m"], [f for f in os.listdir(res_paths["R10m"]) if "_B03" in f][0]))
            rgb = np.stack([b4, b3, b2], axis=-1)
            rgb_plot = array_to_plot(normalize(rgb), vis_type)
            tif_path = generate_temp_path(".tif")
            save_tif(tif_path, rgb, profile, count=3)
            return rgb_plot, tif_path

        elif vis_type == "False Color Vegetation (B8, B4, B3)":
            b3, _ = read_band(os.path.join(res_paths["R10m"], [f for f in os.listdir(res_paths["R10m"]) if "_B03" in f][0]))
            b8, _ = read_band(os.path.join(res_paths["R10m"], [f for f in os.listdir(res_paths["R10m"]) if "_B08" in f][0]))
            fcv = np.stack([b8, b4, b3], axis=-1)
            fcv_plot = array_to_plot(normalize(fcv), vis_type)
            tif_path = generate_temp_path(".tif")
            save_tif(tif_path, fcv, profile, count=3)
            return fcv_plot, tif_path

        elif vis_type == "False Color SWIR (B12, B8, B4)":
            b8, _ = read_band(os.path.join(res_paths["R10m"], [f for f in os.listdir(res_paths["R10m"]) if "_B08" in f][0]))
            b12_path = os.path.join(res_paths["R20m"], [f for f in os.listdir(res_paths["R20m"]) if "_B12" in f][0])
            b12 = resample_to_10m(b12_path, b4.shape, profile["transform"])
            fcswir = np.stack([b12, b8, b4], axis=-1)
            swir_plot = array_to_plot(normalize(fcswir), vis_type)
            tif_path = generate_temp_path(".tif")
            save_tif(tif_path, fcswir, profile, count=3)
            return swir_plot, tif_path

        elif vis_type == "NDVI":
            b8, _ = read_band(os.path.join(res_paths["R10m"], [f for f in os.listdir(res_paths["R10m"]) if "_B08" in f][0]))
            ndvi = (b8 - b4) / (b8 + b4 + 1e-6)
            ndvi_plot = array_to_plot(ndvi, "NDVI", cmap='RdYlGn')
            tif_path = generate_temp_path(".tif")
            save_tif(tif_path, ndvi, profile, count=1)
            return ndvi_plot, tif_path

        else:
            raise ValueError("Invalid visualization type.")

# === Gradio Interface ===
demo = gr.Interface(
    fn=process_visualization,
    inputs=[
        gr.File(label="Sentinel-2 Archive (.zip)", type="filepath"),
        gr.Dropdown(
            choices=[
                "Natural Color (B4, B3, B2)",
                "False Color Vegetation (B8, B4, B3)",
                "False Color SWIR (B12, B8, B4)",
                "NDVI"
            ],
            label="Visualization Type"
        )
    ],
    outputs=[
        gr.Plot(label="Preview"),
        gr.File(label="Download GeoTIFF")
    ],
    title="Sentinel-2 Viewer + GeoTIFF Export",
    description="Upload a .SAFE.zip file, choose a visualization type, and download the corresponding GeoTIFF file."
)

if __name__ == "__main__":
    demo.launch()