Merge branch 'issue690-resample-spatial-metadata'

Author: soxofaan

CHANGELOG.md

@@ -11,6 +11,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 
+- Improved tracking of metadata changes with `resample_spatial` and `resample_cube_spatial` ([#690](https://github.com/Open-EO/openeo-python-client/issues/690))
+
 ### Removed
 
 ### Fixed

openeo/_version.py

@@ -1 +1 @@
-__version__ = "0.38.0a1"
+__version__ = "0.38.0a2"

openeo/metadata.py

@@ -12,6 +12,7 @@
 
 from openeo.internal.jupyter import render_component
 from openeo.util import Rfc3339, deep_get
+from openeo.utils.normalize import normalize_resample_resolution
 
 _log = logging.getLogger(__name__)
 
@@ -25,6 +26,8 @@ class DimensionAlreadyExistsException(MetadataException):
 
 
 # TODO: make these dimension classes immutable data classes
+# TODO: align better with STAC datacube extension
+# TODO: align/adapt/integrate with pystac's datacube extension implementation?
 class Dimension:
     """Base class for dimensions."""
 
@@ -58,6 +61,8 @@ def rename_labels(self, target, source) -> Dimension:
 
 
 class SpatialDimension(Dimension):
+    # TODO: align better with STAC datacube extension: e.g. support "axis" (x or y)
+
     DEFAULT_CRS = 4326
 
     def __init__(
@@ -257,6 +262,10 @@ def __init__(self, dimensions: Optional[List[Dimension]] = None):
     def __eq__(self, o: Any) -> bool:
         return isinstance(o, type(self)) and self._dimensions == o._dimensions
 
+    def __str__(self) -> str:
+        bands = self.band_names if self.has_band_dimension() else "no bands dimension"
+        return f"CubeMetadata({bands} - {self.dimension_names()})"
+
     def _clone_and_update(self, dimensions: Optional[List[Dimension]] = None, **kwargs) -> CubeMetadata:
         """Create a new instance (of same class) with copied/updated fields."""
         cls = type(self)
@@ -411,10 +420,39 @@ def drop_dimension(self, name: str = None) -> CubeMetadata:
             raise ValueError("No dimension named {n!r} (valid names: {ns!r})".format(n=name, ns=dimension_names))
         return self._clone_and_update(dimensions=[d for d in self._dimensions if not d.name == name])
 
-    def __str__(self) -> str:
-        bands = self.band_names if self.has_band_dimension() else "no bands dimension"
-        return f"CubeMetadata({bands} - {self.dimension_names()})"
+    def resample_spatial(
+        self,
+        resolution: Union[float, Tuple[float, float], List[float]] = 0.0,
+        projection: Union[int, str, None] = None,
+    ) -> CubeMetadata:
+        resolution = normalize_resample_resolution(resolution)
+        if self._dimensions is None:
+            # Best-effort fallback to work with
+            dimensions = [
+                SpatialDimension(name="x", extent=[None, None]),
+                SpatialDimension(name="y", extent=[None, None]),
+            ]
+        else:
+            # Make sure to work with a copy (to edit in-place)
+            dimensions = list(self._dimensions)
+
+        # Find and replace spatial dimensions
+        spatial_indixes = [i for i, d in enumerate(dimensions) if isinstance(d, SpatialDimension)]
+        if len(spatial_indixes) != 2:
+            raise MetadataException(f"Expected two spatial resolutions but found {spatial_indixes=}")
+        for i in spatial_indixes:
+            dim: SpatialDimension = dimensions[i]
+            dimensions[i] = SpatialDimension(
+                name=dim.name,
+                extent=dim.extent,
+                crs=projection or dim.crs,
+                step=resolution[i] if resolution[i] else dim.step,
+            )
+
+        return self._clone_and_update(dimensions=dimensions)
 
+    def resample_cube_spatial(self, target: CubeMetadata) -> CubeMetadata:
+        return self._clone_and_update(dimensions=list(target._dimensions))
 
 class CollectionMetadata(CubeMetadata):
     """

openeo/rest/datacube.py

@@ -40,6 +40,7 @@
     Band,
     BandDimension,
     CollectionMetadata,
+    CubeMetadata,
     SpatialDimension,
     TemporalDimension,
     metadata_from_stac,
@@ -69,6 +70,10 @@
     from openeo.udf import XarrayDataCube
 
 
+# Sentinel value for arguments that are unset (when `None` has a different meaning)
+_UNSET = object()
+
+
 log = logging.getLogger(__name__)
 
 
@@ -97,7 +102,7 @@ def process(
         self,
         process_id: str,
         arguments: Optional[dict] = None,
-        metadata: Optional[CollectionMetadata] = None,
+        metadata: Optional[CubeMetadata] = _UNSET,
         namespace: Optional[str] = None,
         **kwargs,
     ) -> DataCube:
@@ -111,7 +116,11 @@ def process(
         :return: new DataCube instance
         """
         pg = self._build_pgnode(process_id=process_id, arguments=arguments, namespace=namespace, **kwargs)
-        return DataCube(graph=pg, connection=self._connection, metadata=metadata or self.metadata)
+        return DataCube(
+            graph=pg,
+            connection=self._connection,
+            metadata=self.metadata if metadata is _UNSET else metadata,
+        )
 
     graph_add_node = legacy_alias(process, "graph_add_node", since="0.1.1")
 
@@ -749,16 +758,24 @@ def band(self, band: Union[str, int]) -> DataCube:
 
     @openeo_process
     def resample_spatial(
-            self, resolution: Union[float, Tuple[float, float]], projection: Union[int, str] = None,
-            method: str = 'near', align: str = 'upper-left'
+        self,
+        resolution: Union[float, Tuple[float, float], List[float]] = 0.0,
+        projection: Union[int, str, None] = None,
+        method: str = "near",
+        align: str = "upper-left",
     ) -> DataCube:
-        return self.process('resample_spatial', {
-            'data': THIS,
-            'resolution': resolution,
-            'projection': projection,
-            'method': method,
-            'align': align
-        })
+        metadata = (self.metadata or CubeMetadata()).resample_spatial(resolution=resolution, projection=projection)
+        return self.process(
+            process_id="resample_spatial",
+            arguments={
+                "data": THIS,
+                "resolution": resolution,
+                "projection": projection,
+                "method": method,
+                "align": align,
+            },
+            metadata=metadata,
+        )
 
     def resample_cube_spatial(self, target: DataCube, method: str = "near") -> DataCube:
         """
@@ -773,7 +790,15 @@ def resample_cube_spatial(self, target: DataCube, method: str = "near") -> DataC
         :param method: Resampling method to use.
         :return:
         """
-        return self.process("resample_cube_spatial", {"data": self, "target": target, "method": method})
+        if target.metadata:
+            metadata = (self.metadata or CubeMetadata()).resample_cube_spatial(target=target.metadata)
+        else:
+            metadata = None
+        return self.process(
+            process_id="resample_cube_spatial",
+            arguments={"data": self, "target": target, "method": method},
+            metadata=metadata,
+        )
 
     @openeo_process
     def resample_cube_temporal(

openeo/utils/__init__.py

@@ -0,0 +1,16 @@
+from typing import Tuple, Union
+
+
+def normalize_resample_resolution(
+    resolution: Union[int, float, Tuple[float, float], Tuple[int, int]]
+) -> Tuple[Union[int, float], Union[int, float]]:
+    """Normalize a resolution value, as used in the `resample_spatial` process to a two-element tuple."""
+    if isinstance(resolution, (int, float)):
+        return (resolution, resolution)
+    elif (
+        isinstance(resolution, (list, tuple))
+        and len(resolution) == 2
+        and all(isinstance(r, (int, float)) for r in resolution)
+    ):
+        return tuple(resolution)
+    raise ValueError(f"Invalid resolution {resolution!r}")

openeo/utils/normalize.py

@@ -20,6 +20,7 @@
 
 from openeo import collection_property
 from openeo.api.process import Parameter
+from openeo.metadata import SpatialDimension
 from openeo.rest import BandMathException, OpenEoClientException
 from openeo.rest._testing import build_capabilities
 from openeo.rest.connection import Connection
@@ -730,12 +731,144 @@ def test_apply_kernel(s2cube):
 
 
 def test_resample_spatial(s2cube):
-    im = s2cube.resample_spatial(resolution=[2.0, 3.0], projection=4578)
-    graph = _get_leaf_node(im)
-    assert graph["process_id"] == "resample_spatial"
-    assert "data" in graph["arguments"]
-    assert graph["arguments"]["resolution"] == [2.0, 3.0]
-    assert graph["arguments"]["projection"] == 4578
+    cube = s2cube.resample_spatial(resolution=[2.0, 3.0], projection=4578)
+    assert get_download_graph(cube, drop_load_collection=True, drop_save_result=True) == {
+        "resamplespatial1": {
+            "process_id": "resample_spatial",
+            "arguments": {
+                "data": {"from_node": "loadcollection1"},
+                "resolution": [2.0, 3.0],
+                "projection": 4578,
+                "method": "near",
+                "align": "upper-left",
+            },
+        }
+    }
+
+    assert cube.metadata.spatial_dimensions == [
+        SpatialDimension(name="x", extent=None, crs=4578, step=2.0),
+        SpatialDimension(name="y", extent=None, crs=4578, step=3.0),
+    ]
+
+
+def test_resample_spatial_no_metadata(s2cube_without_metadata):
+    cube = s2cube_without_metadata.resample_spatial(resolution=(3, 5), projection=4578)
+    assert get_download_graph(cube, drop_load_collection=True, drop_save_result=True) == {
+        "resamplespatial1": {
+            "process_id": "resample_spatial",
+            "arguments": {
+                "data": {"from_node": "loadcollection1"},
+                "resolution": [3, 5],
+                "projection": 4578,
+                "method": "near",
+                "align": "upper-left",
+            },
+        }
+    }
+    assert cube.metadata.spatial_dimensions == [
+        SpatialDimension(name="x", extent=[None, None], crs=4578, step=3.0),
+        SpatialDimension(name="y", extent=[None, None], crs=4578, step=5.0),
+    ]
+
+
+def test_resample_cube_spatial(s2cube):
+    cube1 = s2cube.resample_spatial(resolution=[2.0, 3.0], projection=4578)
+    cube2 = s2cube.resample_spatial(resolution=10, projection=32631)
+
+    cube12 = cube1.resample_cube_spatial(target=cube2)
+    assert get_download_graph(cube12, drop_load_collection=True, drop_save_result=True) == {
+        "resamplespatial1": {
+            "process_id": "resample_spatial",
+            "arguments": {
+                "align": "upper-left",
+                "data": {"from_node": "loadcollection1"},
+                "method": "near",
+                "projection": 4578,
+                "resolution": [2.0, 3.0],
+            },
+        },
+        "resamplespatial2": {
+            "process_id": "resample_spatial",
+            "arguments": {
+                "align": "upper-left",
+                "data": {"from_node": "loadcollection1"},
+                "method": "near",
+                "projection": 32631,
+                "resolution": 10,
+            },
+        },
+        "resamplecubespatial1": {
+            "arguments": {
+                "data": {"from_node": "resamplespatial1"},
+                "method": "near",
+                "target": {"from_node": "resamplespatial2"},
+            },
+            "process_id": "resample_cube_spatial",
+        },
+    }
+    assert cube12.metadata.spatial_dimensions == [
+        SpatialDimension(name="x", extent=None, crs=32631, step=10),
+        SpatialDimension(name="y", extent=None, crs=32631, step=10),
+    ]
+
+    cube21 = cube2.resample_cube_spatial(target=cube1)
+    assert get_download_graph(cube21, drop_load_collection=True, drop_save_result=True) == {
+        "resamplespatial1": {
+            "process_id": "resample_spatial",
+            "arguments": {
+                "align": "upper-left",
+                "data": {"from_node": "loadcollection1"},
+                "method": "near",
+                "projection": 32631,
+                "resolution": 10,
+            },
+        },
+        "resamplespatial2": {
+            "process_id": "resample_spatial",
+            "arguments": {
+                "align": "upper-left",
+                "data": {"from_node": "loadcollection1"},
+                "method": "near",
+                "projection": 4578,
+                "resolution": [2.0, 3.0],
+            },
+        },
+        "resamplecubespatial1": {
+            "arguments": {
+                "data": {"from_node": "resamplespatial1"},
+                "method": "near",
+                "target": {"from_node": "resamplespatial2"},
+            },
+            "process_id": "resample_cube_spatial",
+        },
+    }
+    assert cube21.metadata.spatial_dimensions == [
+        SpatialDimension(name="x", extent=None, crs=4578, step=2.0),
+        SpatialDimension(name="y", extent=None, crs=4578, step=3.0),
+    ]
+
+
+def test_resample_cube_spatial_no_source_metadata(s2cube, s2cube_without_metadata):
+    cube = s2cube_without_metadata
+    target = s2cube.resample_spatial(resolution=10, projection=32631)
+    assert cube.metadata is None
+    assert target.metadata is not None
+
+    result = cube.resample_cube_spatial(target=target)
+    assert result.metadata.spatial_dimensions == [
+        SpatialDimension(name="x", extent=None, crs=32631, step=10),
+        SpatialDimension(name="y", extent=None, crs=32631, step=10),
+    ]
+
+
+def test_resample_cube_spatial_no_target_metadata(s2cube, s2cube_without_metadata):
+    cube = s2cube.resample_spatial(resolution=10, projection=32631)
+    target = s2cube_without_metadata
+    assert cube.metadata is not None
+    assert target.metadata is None
+
+    result = cube.resample_cube_spatial(target=target)
+    assert result.metadata is None
 
 
 def test_merge(s2cube, api_version, test_data):