Add integration test for recipe runner

pyproject.toml

@@ -82,6 +82,7 @@ dev = [
   "types-PyYAML",
   "pytest",
   "pytest-cov",
+  "pytest-mock",
   "pre-commit",
 ]
 docs = [

src/zampy/utils/regrid.py

@@ -105,7 +105,7 @@ def _groupby_regrid(
     ds_out = ds_out.swap_dims(
         {"latitude_bins": "latitude", "longitude_bins": "longitude"}
     )
-    ds_out = ds_out.drop(["latitude_bins", "longitude_bins"])
+    ds_out = ds_out.drop_vars(["latitude_bins", "longitude_bins"])
     return ds_out.transpose("time", "latitude", "longitude", ...)
 
 

tests/test_recipes/generate_test_data.py

@@ -0,0 +1,72 @@
+"""Generates test data for running the recipe tests."""
+from pathlib import Path
+import numpy as np
+import pandas as pd
+import xarray as xr
+from zampy.datasets.dataset_protocol import SpatialBounds
+from zampy.datasets.dataset_protocol import TimeBounds
+
+
+def generate_era5_file(
+    varname: str,
+    time_bounds: TimeBounds,
+    spatial_bounds: SpatialBounds,
+    test_value: float,
+    resolution: float,
+    time_res="1H",
+) -> xr.Dataset:
+    time_coords = pd.date_range(
+        start=time_bounds.start, end=time_bounds.end, freq=time_res, inclusive="left"
+    )
+    lat_coords = np.arange(
+        start=np.round(spatial_bounds.south - 1),
+        stop=np.round(spatial_bounds.north + 1),
+        step=resolution,
+    )
+    lon_coords = np.arange(
+        start=np.round(spatial_bounds.west - 1),
+        stop=np.round(spatial_bounds.east + 1),
+        step=resolution,
+    )
+    data = np.zeros((len(lon_coords), len(lat_coords), len(time_coords))) + test_value
+
+    ds = xr.Dataset(
+        data_vars={ERA5_LOOKUP[varname][1]: (("longitude", "latitude", "time"), data)},
+        coords={
+            "longitude": lon_coords,
+            "latitude": lat_coords,
+            "time": time_coords,
+        },
+    )
+    ds[ERA5_LOOKUP[varname][1]].attrs["units"] = ERA5_LOOKUP[varname][0]
+    ds["latitude"].attrs["units"] = "degrees_north"
+    ds["longitude"].attrs["units"] = "degrees_east"
+
+    return ds
+
+
+ERA5_LOOKUP = {  # name: (unit, fname)
+    "10m_u_component_of_wind": ("m s**-1", "u10"),
+    "10m_v_component_of_wind": ("m s**-1", "v10"),
+    "surface_pressure": ("Pa", "sp"),
+}
+
+
+def generate_era5_files(
+    directory: Path,
+    variables: list[str],
+    spatial_bounds: SpatialBounds,
+    time_bounds: TimeBounds,
+) -> None:
+    data_dir_era5 = directory / "era5"
+    data_dir_era5.mkdir()
+
+    for var in variables:
+        ds = generate_era5_file(
+            varname=var,
+            time_bounds=time_bounds,
+            spatial_bounds=spatial_bounds,
+            test_value=1.0,
+            resolution=0.25,
+        )
+        ds.to_netcdf(path=data_dir_era5 / f"era5_{var}.nc")

tests/test_recipes/recipes/era5_recipe.yml

@@ -0,0 +1,16 @@
+name: "era5_recipe"
+
+download:
+  years: [2020, 2020]
+  bbox: [51, 4, 50, 3] # NESW
+
+  datasets:
+    era5:
+      variables:
+        - 10m_v_component_of_wind
+        - surface_pressure
+
+convert:
+  convention: ALMA
+  frequency: 1H  # outputs at 1 hour frequency. Pandas-like freq-keyword.
+  resolution: 0.5  # output resolution in degrees.

tests/test_recipes/test_simple_recipe.py

@@ -0,0 +1,44 @@
+"""Testing a simple recipe."""
+from pathlib import Path
+from unittest.mock import patch
+import generate_test_data
+import numpy as np
+import xarray as xr
+from zampy.datasets import DATASETS
+from zampy.datasets.dataset_protocol import SpatialBounds
+from zampy.datasets.dataset_protocol import TimeBounds
+from zampy.datasets.dataset_protocol import write_properties_file
+from zampy.recipe import RecipeManager
+
+
+RECIPE_FILE = Path(__file__).parent / "recipes" / "era5_recipe.yml"
+
+
+def test_recipe(tmp_path: Path, mocker):
+    with (patch.object(DATASETS["era5"], "download"),):
+        mocker.patch(
+            "zampy.recipe.config_loader",
+            return_value={"working_directory": str(tmp_path.absolute())},
+        )
+        rm = RecipeManager(str(RECIPE_FILE.absolute()))
+
+        spatial_bounds = SpatialBounds(51, 4, 50, 3)
+        time_bounds = TimeBounds(
+            np.datetime64("2020-01-01T00:00"), np.datetime64("2020-12-31T23:59")
+        )
+        variables = ["10m_v_component_of_wind", "surface_pressure"]
+
+        generate_test_data.generate_era5_files(
+            directory=tmp_path / "download",
+            variables=variables,
+            spatial_bounds=spatial_bounds,
+            time_bounds=time_bounds,
+        )
+        write_properties_file(
+            tmp_path / "download" / "era5", spatial_bounds, time_bounds, variables
+        )
+
+        rm.run()
+
+        ds = xr.open_mfdataset(str(tmp_path / "output" / "era5_recipe" / "*.nc"))
+        assert all(var in ds.data_vars for var in ["Psurf", "Wind_N"])