Issue #737 leverage pystac for deeper spatial dimension detection

Author: soxofaan

openeo/metadata.py

@@ -655,12 +655,17 @@ def is_band_asset(asset: pystac.Asset) -> bool:
     else:
         raise ValueError(stac_object)
 
-    # At least assume there are spatial dimensions
-    # TODO: are there conditions in which we even should not assume the presence of spatial dimensions?
-    dimensions = [
-        SpatialDimension(name="x", extent=[None, None]),
-        SpatialDimension(name="y", extent=[None, None]),
-    ]
+    dimensions = []
+
+    spatial_dimensions = _StacMetadataParser().get_spatial_dimensions(stac_object)
+    # Unless spatial dimensions are explicitly absent from the STAC metadata:
+    # assume we at least have spatial dimensions ("x" and "y", per openEO API recommendation).
+    if spatial_dimensions is None:
+        spatial_dimensions = [
+            SpatialDimension(name="x", extent=[None, None]),
+            SpatialDimension(name="y", extent=[None, None]),
+        ]
+    dimensions.extend(spatial_dimensions)
 
     # TODO: conditionally include band dimension when there was actual indication of band metadata?
     band_dimension = BandDimension(name="bands", bands=bands)
@@ -780,3 +785,17 @@ def get_temporal_dimension(self, stac_obj: pystac.STACObject) -> Union[TemporalD
             if len(temporal_dims) == 1:
                 name, extent = temporal_dims[0]
                 return TemporalDimension(name=name, extent=extent)
+
+    def get_spatial_dimensions(self, stac_obj: pystac.STACObject) -> Union[List[SpatialDimension], None]:
+        if _PYSTAC_1_9_EXTENSION_INTERFACE:
+            if stac_obj.ext.has("cube") and hasattr(stac_obj.ext, "cube"):
+                return [
+                    SpatialDimension(
+                        name=n,
+                        extent=d.extent or [None, None],
+                        crs=d.reference_system or SpatialDimension.DEFAULT_CRS,
+                        step=d.step,
+                    )
+                    for (n, d) in stac_obj.ext.cube.dimensions.items()
+                    if d.dim_type == pystac.extensions.datacube.DimensionType.SPATIAL
+                ]

tests/test_metadata.py

@@ -937,6 +937,93 @@ def test_metadata_from_stac_temporal_dimension(tmp_path, stac_dict, expected):
         assert not metadata.has_temporal_dimension()
 
 
+@pytest.mark.skipif(
+    not _PYSTAC_1_9_EXTENSION_INTERFACE,
+    reason="No backport of implementation/test below PySTAC 1.9 extension interface",
+)
+@pytest.mark.parametrize(
+    ["stac_dict", "expected"],
+    [
+        (
+            # Item without cube:dimensions metadata -> assume spatial dimensions x and y
+            StacDummyBuilder.item(),
+            {"x": [None, None], "y": [None, None]},
+        ),
+        (
+            StacDummyBuilder.item(
+                cube_dimensions={
+                    "t": {"type": "temporal", "extent": ["2024-04-04", "2024-06-06"]},
+                    "x": {"type": "spatial", "axis": "x", "extent": [-10, 20]},
+                    "y": {"type": "spatial", "axis": "y", "extent": [30, 50]},
+                }
+            ),
+            {"x": [-10, 20], "y": [30, 50]},
+        ),
+        (
+            # Custom dimension names
+            StacDummyBuilder.item(
+                cube_dimensions={
+                    "TTT": {"type": "temporal", "extent": ["2024-04-04", "2024-06-06"]},
+                    "XXX": {"type": "spatial", "axis": "x", "extent": [-10, 20]},
+                    "YYY": {"type": "spatial", "axis": "y", "extent": [30, 50]},
+                }
+            ),
+            {"XXX": [-10, 20], "YYY": [30, 50]},
+        ),
+        (
+            # Explicitly absent spatial dimensions
+            StacDummyBuilder.item(
+                cube_dimensions={
+                    "t": {"type": "temporal", "extent": ["2024-04-04", "2024-06-06"]},
+                }
+            ),
+            {},
+        ),
+        (
+            # No cube:dimensions metadata -> assume spatial dimensions x and y
+            StacDummyBuilder.collection(),
+            {"x": [None, None], "y": [None, None]},
+        ),
+        (
+            StacDummyBuilder.collection(
+                cube_dimensions={
+                    "t": {"type": "temporal", "extent": ["2024-04-04", "2024-06-06"]},
+                    "x": {"type": "spatial", "axis": "x", "extent": [-10, 20]},
+                    "y": {"type": "spatial", "axis": "y", "extent": [30, 50]},
+                }
+            ),
+            {"x": [-10, 20], "y": [30, 50]},
+        ),
+        (
+            # Explicitly absent spatial dimensions
+            StacDummyBuilder.collection(
+                cube_dimensions={
+                    "t": {"type": "temporal", "extent": ["2024-04-04", "2024-06-06"]},
+                }
+            ),
+            {},
+        ),
+        (
+            StacDummyBuilder.catalog(),
+            {"x": [None, None], "y": [None, None]},
+        ),
+        (
+            # Note: a catalog is not supposed to have datacube extension enabled, but we should not choke on that
+            StacDummyBuilder.catalog(stac_extensions=[StacDummyBuilder._EXT_DATACUBE]),
+            {"x": [None, None], "y": [None, None]},
+        ),
+    ],
+)
+def test_metadata_from_stac_spatial_dimensions(tmp_path, stac_dict, expected):
+    path = tmp_path / "stac.json"
+    # TODO #738 real request mocking of STAC resources compatible with pystac?
+    path.write_text(json.dumps(stac_dict))
+    metadata = metadata_from_stac(str(path))
+    dims = metadata.spatial_dimensions
+    assert all(isinstance(d, SpatialDimension) for d in dims)
+    assert {d.name: d.extent for d in dims} == expected
+
+
 @pytest.mark.parametrize(
     ["kwargs", "expected_x", "expected_y"],
     [