Added thumbnail maker

s-boomi · s-boomi · commit 2794c3707e2e · 2025-11-13T17:57:07.000+01:00
diff --git a/notebooks/catalog_reading.py b/notebooks/catalog_reading.py
@@ -8,13 +8,14 @@
 def _():
     import marimo as mo
     import pystac
+    from pathlib import Path
 
-    return mo, pystac
+    return Path, mo, pystac
 
 
 @app.cell
-def _(pystac):
-    catalog = pystac.Catalog.from_file("./data/processed/catalog.json")
+def _(Path, pystac):
+    catalog = pystac.Catalog.from_file(Path.home() / "data-folder/catalog/catalog.json")
     catalog.describe()
     return (catalog,)
 
diff --git a/notebooks/omega_cubes.py b/notebooks/omega_cubes.py
@@ -886,6 +886,73 @@ def _(dt, ex_nc_ds_l2, re):
     return
 
 
+@app.cell
+def _(ex_nc_ds_l2, plt):
+    ex_nc_ds_l2.Reflectance.mean("wavelength").plot(cmap=plt.cm.get_cmap("viridis"))
+    plt.title("Reflectance map based on wavelengh mean")
+    return
+
+
+@app.cell
+def _(Path, ex_nc_ds_l2, mo, np, plt):
+    # import numpy as np
+    from PIL import Image
+    from typing import Literal
+    from tempfile import TemporaryDirectory
+
+    def convert_to_thumbnail(
+        data: np.ndarray,
+        resize_dims: tuple[int, int],
+        mode: Literal["L", "RGB", "RGBA"] = "L",
+        cmap: str | None = None,
+    ) -> Image.Image:
+        """
+        Converts a 2D or 3D NumPy array into a resized PNG-style image.
+        Applies a matplotlib colormap if provided.
+        """
+        # --- Ensure numpy array ---
+        data = np.asarray(data, dtype=float)
+
+        # --- Normalize input data to 0-1 ---
+        data = (data - np.nanmin(data)) / (np.nanmax(data) - np.nanmin(data) + 1e-8)
+
+        # --- Apply colormap if requested ---
+        if cmap is not None:
+            cm = plt.get_cmap(cmap)
+            result = cm(data)[..., :4]  # includes alpha
+            result = (result * 255).astype(np.uint8)
+            if mode == "RGB":
+                result = result[..., :3]
+        else:
+            result = (data * 255).astype(np.uint8)
+            if mode in ["RGB", "RGBA"]:
+                result = np.stack([result] * (3 if mode == "RGB" else 4), axis=-1)
+
+        # --- Resize with high-quality interpolation ---
+        img = Image.fromarray(result, mode=mode)
+        img = img.resize(resize_dims, Image.Resampling.LANCZOS)
+
+        return img
+
+    tempdir = TemporaryDirectory()
+    resized_img = convert_to_thumbnail(
+        ex_nc_ds_l2.Reflectance.mean("wavelength").values,
+        (256, 256),
+        mode="RGB",
+        cmap="rainbow",
+    )
+
+    resized_img.save(Path(tempdir.name) / "test.png")
+    mo.image(src=Path(tempdir.name) / "test.png")
+    return
+
+
+@app.cell
+def _(ex_nc_ds_l2):
+    getattr(ex_nc_ds_l2.Reflectance, "mean")("wavelength").values
+    return
+
+
 @app.cell
 def _():
     return
diff --git a/notebooks/projections.py b/notebooks/projections.py
@@ -23,6 +23,16 @@ def _(WktIoHandler):
     return
 
 
+@app.cell
+def _():
+    from stac_validator import stac_validator
+
+    stac = stac_validator.StacValidate("./data/processed/catalog.json", extensions=True)
+    stac.run()
+    stac.message
+    return
+
+
 @app.cell
 def _():
     return
diff --git a/src/psup_stac_converter/omega/_base.py b/src/psup_stac_converter/omega/_base.py
@@ -25,6 +25,7 @@
 from psup_stac_converter.extensions import apply_sci, apply_ssys
 from psup_stac_converter.informations.data_providers import providers as data_providers
 from psup_stac_converter.settings import create_logger
+from psup_stac_converter.utils.file_utils import convert_arr_to_thumbnail
 from psup_stac_converter.utils.io import PsupIoHandler
 from psup_stac_converter.utils.models import (
     CubedataVariable,
@@ -123,6 +124,10 @@ def __init__(
         self.nc_metadata_folder = (
             psup_io_handler.output_folder / f"{self.metadata_folder_prefix}nc"
         )
+        self.thumbnail_folder = (
+            psup_io_handler.output_folder / f"{self.metadata_folder_prefix}_thumbnail"
+        )
+        self.thumbnail_dims = (256, 256)
         self.log.debug(f".sav metadata folder: {self.sav_metadata_folder}")
         self.log.debug(f".nc metadata folder: {self.nc_metadata_folder}")
         if not self.sav_metadata_folder.exists():
@@ -454,6 +459,38 @@ def create_stac_item(self, orbit_cube_idx: str, **kwargs) -> pystac.Item:
         except OmegaCubeDataMissingError:
             self.log.warning(f"IDL.sav not found for {orbit_cube_idx}. Skipping.")
 
+        thumbnail_location = (
+            self.thumbnail_folder
+            / f"{orbit_cube_idx}_{self.thumbnail_dims[0]}x{self.thumbnail_dims[1]}.png"
+        )
+
+        # Normally the thumbnail should be generated
+        # but if not, the file is open
+        if not thumbnail_location.exists():
+            nc_data = self.open_file(orbit_cube_idx, "nc")
+            thumbnail_strategy = "mean"
+
+            # define thumbnail strategy
+            # By default, takes the reflectance cube
+            self.make_thumbnail(
+                orbit_cube_idx=orbit_cube_idx,
+                data=getattr(nc_data.Reflectance, thumbnail_strategy)(
+                    "wavelength"
+                ).values,
+                dims=self.thumbnail_dims,
+            )
+
+            nc_data.close()
+
+        # Thumbnail
+        thumbn_asset = pystac.Asset(
+            href=thumbnail_location.as_posix(),
+            media_type=pystac.MediaType.PNG,
+            roles=["thumbnail"],
+            description="PNG thumbnail preview for visualizations",
+        )
+        pystac_item.add_asset("thumbnail", thumbn_asset)
+
         # extensions
         pystac_item = cast(pystac.Item, apply_ssys(pystac_item))
 
@@ -476,13 +513,17 @@ def create_stac_item(self, orbit_cube_idx: str, **kwargs) -> pystac.Item:
         return pystac_item
 
     def find_cubedata_from_ncfile(
-        self, orbit_cube_idx: str
+        self, orbit_cube_idx: str, thumbnail_strategy: str = "mean"
     ) -> dict[str, dict[str, Dimension | Variable]]:
         """From the NetCDF file, extracts the cubedata information needed for the
-        generated STAC item
+        generated STAC item.
 
         Args:
-            orbit_cube_idx (str): _description_
+            orbit_cube_idx (str): the ID of the orbit cube
+
+        Returns:
+            dict[str, dict[str, Dimension | Variable]]: A dict containing
+            "dimensions", "variables" and "extras" as main keys
         """
         dimensions = {}
         variables = {}
@@ -574,6 +615,16 @@ def find_cubedata_from_ncfile(
             # Add some extras if you want
             extras = self.find_extra_nc_data(nc_data)
 
+            # define thumbnail strategy
+            # By default, takes the reflectance cube
+            self.make_thumbnail(
+                orbit_cube_idx=orbit_cube_idx,
+                data=getattr(nc_data.Reflectance, thumbnail_strategy)(
+                    "wavelength"
+                ).values,
+                dims=self.thumbnail_dims,
+            )
+
             nc_data.close()
         except OSError as ose:
             self.log.error(f"[{ose.__class__.__name__}] {ose}")
@@ -620,3 +671,29 @@ def retrieve_nc_info_from_saved_state(self, orbit_cube_idx: str) -> dict[str, An
                 nc_info = {}
 
         return nc_info
+
+    def make_thumbnail(
+        self,
+        orbit_cube_idx: str,
+        data: np.ndarray,
+        dims: tuple[int, int],
+        mode: Literal["L", "RGB", "RGBA"] = "RGB",
+        cmap: str = "viridis",
+        fmt: str = "png",
+    ):
+        """Creates a thumbnail for a datacube out of one of its 2D data arrays.
+
+        Args:
+            orbit_cube_idx (str): _description_
+            data (np.ndarray): _description_
+            dims (tuple[int, int]): _description_
+            mode (str, optional): _description_. Defaults to "RGB".
+            cmap (str, optional): _description_. Defaults to "viridis".
+            fmt (str, optional): _description_. Defaults to "png".
+        """
+        thumbnail = convert_arr_to_thumbnail(
+            data=data, resize_dims=dims, mode=mode, cmap=cmap
+        )
+        thumbnail.save(
+            self.thumbnail_folder / f"{orbit_cube_idx}_{dims[0]}x{dims[1]}.{fmt}"
+        )
diff --git a/src/psup_stac_converter/omega/mineral_maps.py b/src/psup_stac_converter/omega/mineral_maps.py
@@ -34,7 +34,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""This data product is a global NIR 1-micrometer albedo map of Mars based on reflectance data acquired by the Mars Express OMEGA hyperspectral camera from January 2004 to August 2010""",
         raster_keywords=["albedo", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/albedo_r1080_equ_map_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/albedo_r1080_equ_map_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -50,7 +50,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""This data product is a global ferric oxide spectral parameter map of Mars based on reflectance data acquired by the Mars Express OMEGA hyperspectral camera from January 2004 to August 2010. This ferric oxide spectral parameter (DB530) is based on the strength of the 0.53 micrometer ferric absorption edge.""",
         raster_keywords=["ferric oxides", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/ferric_bd530_equ_map_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/ferric_bd530_equ_map_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -66,7 +66,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""This data product is a global nanophase ferric oxide (dust) spectral parameter map of Mars based on reflectance data acquired by the Mars Express OMEGA hyperspectral camera from January 2004 to August 2010. This nanophase ferric oxide spectral parameter (NNPHS) is based on the absorption feature centered at 0.86 micrometer.""",
         raster_keywords=["nanophase ferric oxides", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/ferric_nnphs_equ_map_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/ferric_nnphs_equ_map_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -82,7 +82,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""This data product is a global olivine spectral parameter map of Mars based on reflectance data acquired by the Mars Express OMEGA hyperspectral camera from January 2004 to August 2010. This olivine spectral parameter (OSP1) detects Mg-rich and/or small grain size and/or low abundance olivine.""",
         raster_keywords=["olivine", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/olivine_osp1_equ_map_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/olivine_osp1_equ_map_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -98,7 +98,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""This data product is a global olivine spectral parameter map of Mars based on reflectance data acquired by the Mars Express OMEGA hyperspectral camera from January 2004 to August 2010. This olivine spectral parameter (OSP2) is sensitive to olivine with high iron content and/or large grain size and/or high abundance.""",
         raster_keywords=["olivine", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/olivine_osp2_equ_map_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/olivine_osp2_equ_map_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -114,7 +114,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""This data product is a global olivine spectral parameter map of Mars based on reflectance data acquired by the Mars Express OMEGA hyperspectral camera from January 2004 to August 2010. This olivine spectral parameter (OSP3) determines the full band depth at 1.36 micrometer relative to a continuum. It preferentially detects olivine with a large Fe content and/or with large grain size and/or with high abundance.""",
         raster_keywords=["olivine", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/olivine_osp3_equ_map_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/olivine_osp3_equ_map_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -130,7 +130,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""This data product is a global pyroxene spectral parameter map of Mars based on reflectance data acquired by the Mars Express OMEGA hyperspectral camera from January 2004 to August 2010. This pyroxene spectral parameter (BD2000) is based on its 2 micrometer absorption band due to both high-calcium and low-calcium pyroxene.""",
         raster_keywords=["pyroxene", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/pyroxene_bd2000_equ_map_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/pyroxene_bd2000_equ_map_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -146,7 +146,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""60 ppd global map of Solar Albedo from OMEGA data fileld with TES 20 ppd solar albedo global maps (Putzig and Mellon, 2007b) (21600x10800 pixels). This map is 100% filled. Variable name is "albedo". "latitude" and "longitude" indicate the coordinates of the centers of the pixels. Reference : Vincendon et al., Icarus, 2015""",
         raster_keywords=["albedo", "filled", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/albedo_filled_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/albedo_filled_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -162,7 +162,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""60 ppd global map of Solar Albedo from OMEGA data only (21600 x 10800 pixels). This map is filled at 94.79% (rest are NaN). Variable name is "albedo". "latitude" and "longitude" indicate the coordinates of the centers of the pixels. Reference : Vincendon et al., Icarus, 2015.""",
         raster_keywords=["albedo", "unfilled", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/albedo_unfilled_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/albedo_unfilled_reduce.png"
         ),
         publication=Publication(
             citation="""
@@ -178,7 +178,7 @@ class OmegaMineralMapDesc(BaseModel):
         raster_lng_description="""40 ppd global maps of surface emissivity at 5.03 mic. Holes are set to NaN.""",
         raster_keywords=["emissivity", "5mic", "global"],
         thumbnail=HttpUrl(
-            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/emissivite_5.03mic_OMEGA0_reduce.png/sitools/upload/download-thumb.png"
+            "http://psup.ias.u-psud.fr/sitools/datastorage/user/storage/marsdata/omega/png/emissivite_5.03mic_OMEGA0_reduce.png"
         ),
         publication=Publication(
             citation="""
diff --git a/src/psup_stac_converter/utils/file_utils.py b/src/psup_stac_converter/utils/file_utils.py
@@ -2,9 +2,12 @@
 from pathlib import Path
 from typing import Literal
 
+import matplotlib.pyplot as plt
+import numpy as np
 import rasterio
 from astropy.io import fits
 from attr import asdict
+from PIL import Image
 from rasterio.transform import from_gcps
 from rich.console import Console
 
@@ -118,3 +121,34 @@ def fits_header_to_dict(
             fits_obj[header_key] = header_val
 
     return fits_obj
+
+
+def convert_arr_to_thumbnail(
+    data: np.ndarray,
+    resize_dims: tuple[int, int],
+    mode: Literal["L", "RGB", "RGBA"] = "L",
+    cmap: str | None = None,
+) -> Image.Image:
+    """
+    Converts a 2D or 3D NumPy array into a resized PNG-style image.
+    Applies a matplotlib colormap if provided.
+    """
+    data = np.asarray(data, dtype=float)
+
+    data = (data - np.nanmin(data)) / (np.nanmax(data) - np.nanmin(data) + 1e-8)
+
+    if cmap is not None:
+        cm = plt.get_cmap(cmap)
+        result = cm(data)[..., :4]  # includes alpha
+        result = (result * 255).astype(np.uint8)
+        if mode == "RGB":
+            result = result[..., :3]
+    else:
+        result = (data * 255).astype(np.uint8)
+        if mode in ["RGB", "RGBA"]:
+            result = np.stack([result] * (3 if mode == "RGB" else 4), axis=-1)
+
+    img = Image.fromarray(result, mode=mode)
+    img = img.resize(resize_dims, Image.Resampling.LANCZOS)
+
+    return img
