Issue #197 add vector cube UDF test to filter out geometries

Author: soxofaan

openeo_driver/testing.py

@@ -533,7 +533,15 @@ def to_geojson_feature_collection(self) -> dict:
 
 
 class ApproxGeoJSONByBounds:
-    """pytest assert helper to build a matcher to check if a certain GeoJSON construct is within expected bounds"""
+    """
+    pytest assert helper to build a matcher to check if a certain GeoJSON construct has expected bounds
+
+    Usage example:
+
+        >>> geometry = {"type": "Polygon",  "coordinates": [...]}
+        # Check that this geometry has bounds (1, 2, 6, 5) with some absolute tolerance
+        >>> assert geometry == ApproxGeoJSONByBounds(1, 2, 6, 5, abs=0.1)
+    """
 
     def __init__(
         self,
@@ -543,8 +551,9 @@ def __init__(
         abs: Optional[float] = None,
     ):
         bounds = args[0] if len(args) == 1 else args
-        assert isinstance(bounds, (list, tuple)) and len(bounds) == 4
-        self.expected_bounds = [float(b) for b in bounds]
+        bounds = [float(b) for b in bounds]
+        assert len(bounds) == 4
+        self.expected_bounds = bounds
         self.rel = rel
         self.abs = abs
         self.expected_types = set(types)
@@ -572,7 +581,6 @@ def __repr__(self):
             msg += "\n" + "\n".join(f"    # {i}" for i in self.actual_info)
         return msg
 
-
 def caplog_with_custom_formatter(caplog: pytest.LogCaptureFixture, format: Union[str, logging.Formatter]):
     """
     Context manager to set a custom formatter on the caplog fixture.

tests/data/geojson/FeatureCollection10.json

@@ -0,0 +1,37 @@
+{
+  "type": "FeatureCollection",
+  "features": [
+    {
+      "type": "Feature",
+      "properties": {"id": "first", "pop": 123},
+      "geometry": {
+        "type": "Polygon",
+        "coordinates": [[[1, 1], [3, 1], [2, 3], [1, 1]]]
+      }
+    },
+    {
+      "type": "Feature",
+      "properties": {"id": "second", "pop": 456},
+      "geometry": {
+        "type": "Polygon",
+        "coordinates": [[[4, 2], [5, 4], [3, 4], [4, 2]]]
+      }
+    },
+    {
+      "type": "Feature",
+      "properties": {"id": "third", "pop": 789},
+      "geometry": {
+        "type": "Polygon",
+        "coordinates": [[[6, 2], [10, 2], [12, 6], [8, 6], [6, 2]]]
+      }
+    },
+    {
+      "type": "Feature",
+      "properties": {"id": "fourth", "pop": 101112},
+      "geometry": {
+        "type": "Polygon",
+        "coordinates": [[[-2, 7], [5, 7], [5, 14], [-2, 14], [-2, 7]]]
+      }
+    }
+  ]
+}

tests/test_views_execute.py

@@ -3592,3 +3592,71 @@ def process_geometries(udf_data: UdfData) -> UdfData:
                 ],
             }
         )
+
+    def test_apply_dimension_run_udf_filter_geometries_dimension_properties(self, api100):
+        """
+        Test to use `apply_dimension(dimension="properties", process=UDF)` to filter out certain
+        entries from geometries dimension.
+
+        Note that, strictly speaking, this approach draws outside the lines of the openEO API spec
+        as apply_dimension only allows changing the cardinality of the provided dimension ("properties" in this case),
+        not any other dimension (like "geometries" in this case).
+        """
+        udf_code = """
+            from openeo.udf import UdfData, FeatureCollection
+            import shapely.geometry
+            def process_geometries(udf_data: UdfData) -> UdfData:
+                [feature_collection] = udf_data.get_feature_collection_list()
+                gdf = feature_collection.data
+                to_intersect = shapely.geometry.box(4, 3, 8, 4)
+                gdf = gdf[gdf["geometry"].intersects(to_intersect)]
+                udf_data.set_feature_collection_list([
+                    FeatureCollection(id="_", data=gdf),
+                ])
+            """
+        udf_code = textwrap.dedent(udf_code)
+        process_graph = {
+            "get_vector_data": {
+                "process_id": "load_uploaded_files",
+                "arguments": {"paths": [str(get_path("geojson/FeatureCollection10.json"))], "format": "GeoJSON"},
+            },
+            "apply_dimension": {
+                "process_id": "apply_dimension",
+                "arguments": {
+                    "data": {"from_node": "get_vector_data"},
+                    "dimension": "properties",
+                    "process": {
+                        "process_graph": {
+                            "runudf1": {
+                                "process_id": "run_udf",
+                                "arguments": {
+                                    "data": {"from_node": "get_vector_data"},
+                                    "udf": udf_code,
+                                    "runtime": "Python",
+                                },
+                                "result": True,
+                            }
+                        },
+                    },
+                },
+                "result": True,
+            },
+        }
+        resp = api100.check_result(process_graph)
+        assert resp.json == DictSubSet(
+            {
+                "type": "FeatureCollection",
+                "features": [
+                    {
+                        "type": "Feature",
+                        "geometry": ApproxGeoJSONByBounds(3, 2, 5, 4, types=["Polygon"], abs=0.1),
+                        "properties": {"id": "second", "pop": 456},
+                    },
+                    {
+                        "type": "Feature",
+                        "geometry": ApproxGeoJSONByBounds(6, 2, 12, 6, types=["Polygon"], abs=0.1),
+                        "properties": {"id": "third", "pop": 789},
+                    },
+                ],
+            }
+        )