made load_stac nicer to mismatch in band names

Author: VictorVerhaert

openeo/metadata.py

@@ -218,6 +218,16 @@ def rename_labels(self, target, source) -> Dimension:
     def rename(self, name) -> Dimension:
         return BandDimension(name=name, bands=self.bands)
 
+    def contains_band(self, band: Union[int, str]) -> bool:
+        """
+        Check if the given band name or index is present in the dimension.
+        """
+        try:
+            self.band_index(band)
+            return True
+        except ValueError:
+            return False
+
 
 class GeometryDimension(Dimension):
     # TODO: how to model/store labels of geometry dimension?

openeo/rest/datacube.py

@@ -441,10 +441,23 @@ def load_stac(
         graph = PGNode("load_stac", arguments=arguments)
         try:
             metadata = metadata_from_stac(url)
+            # TODO: also apply spatial/temporal filters to metadata?
+
             if isinstance(bands, list):
-                # TODO: also apply spatial/temporal filters to metadata?
-                metadata = metadata.filter_bands(band_names=bands)
-        except Exception:
+                if not metadata.has_band_dimension():
+                    metadata = metadata.add_dimension(
+                        name="bands",
+                        type="bands",
+                        label=None,
+                    )
+                if all(metadata.band_dimension.contains_band(b) for b in bands):
+                    metadata = metadata.filter_bands(band_names=bands)
+                else:
+                    logging.warning(
+                        "Some bands are not available in the collection metadata. Using requested bands as is."
+                    )
+                    metadata = metadata.rename_labels(dimension="bands", target=bands)
+        except Exception as e:
             log.warning(f"Failed to extract cube metadata from STAC URL {url}", exc_info=True)
             metadata = None
         return cls(graph=graph, connection=connection, metadata=metadata)

tests/rest/test_connection.py

@@ -2957,7 +2957,7 @@ def test_load_stac_no_cube_extension_temporal_dimension(self, con120, tmp_path,
         cube = con120.load_stac(str(stac_path))
         assert cube.metadata.temporal_dimension == TemporalDimension(name="t", extent=dim_extent)
 
-    def test_load_stac_band_filtering(self, con120, tmp_path):
+    def test_load_stac_band_filtering(self, con120, tmp_path, caplog):
         stac_path = tmp_path / "stac.json"
         stac_data = StacDummyBuilder.collection(
             summaries={"eo:bands": [{"name": "B01"}, {"name": "B02"}, {"name": "B03"}]}
@@ -2971,6 +2971,18 @@ def test_load_stac_band_filtering(self, con120, tmp_path):
         cube = con120.load_stac(str(stac_path), bands=["B03", "B02"])
         assert cube.metadata.band_names == ["B03", "B02"]
 
+        caplog.set_level(logging.WARNING)
+        # Test with non-existing bands in the collection metadata
+        cube = con120.load_stac(str(stac_path), bands=["B04"])
+        assert cube.metadata.band_names == ["B04"]
+        assert "Some bands are not available in the collection metadata. Using requested bands as is." in caplog.text
+        caplog.clear()
+
+        cube = con120.load_stac(str(stac_path), bands=["B03", "B04"])
+        assert cube.metadata.band_names == ["B03", "B04"]
+        assert "Some bands are not available in the collection metadata. Using requested bands as is." in caplog.text
+        caplog.clear()
+
     @pytest.mark.parametrize(
         "bands",
         [