Initial finetuning of apply_metadata docs #769

Author: soxofaan

docs/examples/udf/udf_modify_spatial.py

@@ -7,7 +7,6 @@
 
 
 def apply_metadata(input_metadata: CollectionMetadata, context: dict) -> CollectionMetadata:
-
     xstep = input_metadata.get("x", "step")
     ystep = input_metadata.get("y", "step")
     new_metadata = {
@@ -24,8 +23,6 @@ def fancy_upsample_function(array: np.array, factor: int = 2) -> np.array:
 
 
 def apply_datacube(cube: XarrayDataCube, context: dict) -> XarrayDataCube:
-    array: xarray.DataArray = cube.get_array()
-
     cubearray: xarray.DataArray = cube.get_array().copy() + 60
 
     # We make prediction and transform numpy array back to datacube
@@ -37,7 +34,7 @@ def apply_datacube(cube: XarrayDataCube, context: dict) -> XarrayDataCube:
     if cubearray.data.ndim == 4 and cubearray.data.shape[0] == 1:
         cubearray = cubearray[0]
     predicted_array = fancy_upsample_function(cubearray.data, 2)
-    inspect(predicted_array, "test message")
+    inspect(data=predicted_array, message="predicted array")
     coord_x = np.linspace(
         start=cube.get_array().coords["x"].min(),
         stop=cube.get_array().coords["x"].max() + init_pixel_size_x,
@@ -50,6 +47,10 @@ def apply_datacube(cube: XarrayDataCube, context: dict) -> XarrayDataCube:
         num=predicted_array.shape[-1],
         endpoint=False,
     )
-    predicted_cube = xarray.DataArray(predicted_array, dims=["bands", "x", "y"], coords=dict(x=coord_x, y=coord_y))
+    predicted_cube = xarray.DataArray(
+        predicted_array,
+        dims=["bands", "x", "y"],
+        coords=dict(x=coord_x, y=coord_y),
+    )
 
     return XarrayDataCube(predicted_cube)

docs/udf.rst

@@ -294,7 +294,9 @@ instead of the original digital number range (thousands):
 
 UDF's that transform cube metadata
 ==================================
-This is a new/experimental feature so may still be subject to change.
+
+.. warning::
+    This is a new/experimental feature so may still be subject to change.
 
 In some cases, a UDF can have impact on the metadata of a cube, but this can not always
 be easily inferred by process graph evaluation logic without running the actual
@@ -314,9 +316,9 @@ To invoke a UDF like this, the apply_neighborhood method is most suitable:
 
 .. code-block:: python
 
-    udf_code = Path("udf_modify_spatial.py").read_text()
+    udf = openeo.UDF.from_file("udf_modify_spatial.py", runtime="Python-Jep")
     cube_updated = cube.apply_neighborhood(
-        lambda data: data.run_udf(udf=udf_code, runtime="Python-Jep", context=dict()),
+        udf,
         size=[
             {"dimension": "x", "value": 128, "unit": "px"},
             {"dimension": "y", "value": 128, "unit": "px"},