Adding AverageShape to daops (#114)

* Added average_shape to daops/ops

* Added tests in daops/test_average.py for average_shape

* Updated AUTHORS.rst and HISTORY.rst

* Update Docstrings

---------

Co-authored-by: charlesgauthier-udm <charlles.gauthier.1@umontreal.ca>

AUTHORS.rst

@@ -17,3 +17,4 @@ Contributors
 ------------
 
 * Trevor James Smith smith.trevorj@ouranos.ca
+* Charles Gauthier gauthier.charles@ouranos.ca

HISTORY.rst

@@ -1,5 +1,14 @@
 Version History
 ===============
+v0.10.1 (unreleased)
+-------------------
+
+New Features
+^^^^^^^^^^^^
+
+* Add clisops.ops.average_shape to daops.ops.average
+
+
 
 v0.10.0 (2023-11-27)
 -------------------

daops/ops/average.py

@@ -1,4 +1,5 @@
 from clisops.ops.average import average_over_dims as clisops_average_over_dims
+from clisops.ops.average import average_shape as clisops_average_shape
 from clisops.ops.average import average_time as clisops_average_time
 from roocs_utils.parameter import collection_parameter
 from roocs_utils.parameter import dimension_parameter
@@ -7,6 +8,8 @@
 
 __all__ = [
     "average_over_dims",
+    "average_time",
+    "average_shape"
 ]
 
 
@@ -74,7 +77,71 @@ def average_over_dims(
     """
 
     result_set = Average(**locals()).calculate()
+    return result_set
+
+
+class AverageShape(Operation):
+    def _resolve_params(self, collection, **params):
+        """
+        Resolve the input parameters to `self.params` and parameterise
+        collection parameter and set to `self.collection`.
+        """
+        shape = params.get("shape")
+        collection = collection_parameter.CollectionParameter(collection)
+
+        self.collection = collection
+        self.params = {
+            "shape": shape,
+            "variable": params.get("variable"),
+        }
+
+    def get_operation_callable(self):
+        return clisops_average_shape
+
+
+def average_shape(
+    collection,
+    shape,
+    variable=None,
+    output_dir=None,
+    output_type="netcdf",
+    split_method="time:auto",
+    file_namer="standard",
+    apply_fixes=True,
+):
+    """
+    Average input dataset over indicated shape.
 
+    Parameters
+    ----------
+    collection: Collection of datasets to process, sequence or string of comma separated dataset identifiers.
+    shape: Path to shape file, or directly a geodataframe to perform average within.
+    variable: Variables to average. If None, average over all data variables.
+    output_dir: str or path like object describing output directory for output files.
+    output_type: {"netcdf", "nc", "zarr", "xarray"}
+    split_method: {"time:auto"}
+    file_namer: {"standard", "simple"}
+    apply_fixes: Boolean. If True fixes will be applied to datasets if needed. Default is True.
+
+    Returns
+    -------
+    List of outputs in the selected type: a list of xarray Datasets or file paths.
+
+
+    Examples
+    --------
+    | collection: ("cmip6.cmip..cas.fgoals-g3.historical.r1i1p1fi.Amon.tas.gn.v20190818",)
+    | shape: "path_to_shape"
+    | ignore_undetected_dims: (-5.,49.,10.,65)
+    | output_type: "netcdf"
+    | output_dir: "/cache/wps/procs/req0111"
+    | split_method: "time:auto"
+    | file_namer: "standard"
+    | apply_fixes: True
+
+    """
+    a = AverageShape(**locals())
+    result_set = AverageShape(**locals()).calculate()
     return result_set
 
 

tests/test_operations/test_average.py

@@ -2,11 +2,14 @@
 
 import pytest
 import xarray as xr
+import geopandas as gpd
+from shapely import Polygon, MultiPolygon
 from roocs_utils.exceptions import InvalidParameterValue
 
 from daops import CONFIG
 from daops.ops.average import average_over_dims
 from daops.ops.average import average_time
+from daops.ops.average import average_shape
 from tests._common import CMIP5_DAY
 from tests._common import CMIP6_MONTH
 
@@ -15,6 +18,17 @@
     "cmip5.output1.MOHC.HadGEM2-ES.rcp85.mon.atmos.Amon.r1i1p1.latest.tas",
     "cmip5.output1.MOHC.HadGEM2-ES.historical.mon.land.Lmon.r1i1p1.latest.rh",
 ]
+CMIP6_IDS = [
+    "CMIP6.CMIP.CAS.FGOALS-g3.historical.r1i1p1f1.Amon.tas.gn.v20190818"
+]
+
+POLY = Polygon([[5.8671874999999996, 57.326521225217064],
+                [-15.468749999999998, 48.45835188280866],
+                [-16.171875, 24.84656534821976],
+                [-3.8671874999999996, 13.581920900545844],
+                [21.796875, 25.799891182088334],
+                [22.8515625, 52.482780222078226],
+                [5.8671874999999996, 57.326521225217064]])
 
 
 def _check_output_nc(result, fname="output_001.nc"):
@@ -94,6 +108,38 @@ def test_average_level(tmpdir):
     )
 
 
+@pytest.mark.online
+def test_average_shape(tmpdir):
+    # Save POLY to tmpdir
+    tmp_poly_path = os.path.join(tmpdir,"tmppoly.json")
+    gpd.GeoDataFrame([{'geometry': POLY}]).to_file(tmp_poly_path)
+
+    result = average_shape(
+        CMIP6_IDS[0],
+        shape=tmp_poly_path,
+        variable=None,
+        output_dir=tmpdir,
+        file_namer='simple',
+        apply_fixes=False
+    )
+    _check_output_nc(result)
+    ds = xr.open_dataset(result.file_uris[0], use_cftime=True)
+    assert "geom" in ds.dims
+
+
+@pytest.mark.online
+def test_average_shape_none(tmpdir):
+    with pytest.raises(InvalidParameterValue) as exc:
+        result = average_shape(
+            CMIP6_IDS[0],
+            shape=None,
+            variable=None,
+            output_dir=tmpdir,
+            file_namer='simple',
+            apply_fixes=False)
+    assert str(exc.value) == "At least one area for averaging must be provided"
+
+
 @pytest.mark.online
 def test_average_time_month(tmpdir):
     ds = xr.open_mfdataset(CMIP5_DAY, use_cftime=True, combine="by_coords")