Update for xarray backend (#165)

* Update for xarray backend This commit adds more strict checks to coord/dim attrs to make sure the Dataset and/or DataArray dimension sizes matches the what is in the input files. * Adding test The test is for a ValueError being raised when opening a test grib2 file, 2024101012_Milton_Adv22_e70_cum_dat.grb, using the backend. The error should be raised when no filter for duration is used. * Update xarray_backend.py This commit offers a different method suggested by Adam. --------- Co-authored-by: Eric Engle <[email protected]>
NOAA-MDL · Jan 31, 2025 · 97a12f3 · 97a12f3
1 parent 45f0c80
commit 97a12f3
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 0 deletions.
diff --git a/src/grib2io/xarray_backend.py b/src/grib2io/xarray_backend.py
@@ -525,6 +525,21 @@ def build_da_without_coords(index, cube, filename) -> xr.DataArray:
     constant_meta_names = [k for k in cube.__dataclass_fields__.keys() if cube[k] is None]
     dims = {k: len(cube[k]) for k in dim_names}
 
+    dims_total = 1
+    dims_to_filter = []
+    for dim_name, dim_len, in dims.items():
+        if dim_name not in {'x','y','station'}:
+            dims_total *= dim_len
+            dims_to_filter.append(dim_name)
+
+    # Check number of GRIB2 message indexed compared to non-X/Y
+    # dimensions.
+    if dims_total != len(index):
+        raise ValueError(
+            f"DataArray dimensions are not compatible with number of GRIB2 messages; DataArray has {dims_total} "
+            f"and GRIB2 index has {len(index)}. Consider applying a filter for dimensions: {dims_to_filter}"
+            )
+
     data = OnDiskArray(filename, index, cube)
     lock = LOCK
     data = GribBackendArray(data, lock)

diff --git a/tests/data/2024101012_Milton_Adv22_e70_cum_dat.grb b/tests/data/2024101012_Milton_Adv22_e70_cum_dat.grb
diff --git a/tests/test_xarray_backend.py b/tests/test_xarray_backend.py
@@ -46,3 +46,8 @@ def test_interp_with_openmp_threads(request):
         assert da.shape == (1, 1597, 2345)
     except(ModuleNotFoundError):
         pytest.skip()
+
+def test_valueerror_multiple_durations_to_filter(request):
+    data = request.config.rootdir / 'tests' / 'data'
+    with pytest.raises(ValueError, match=r"DataArray dimensions are not compatible with number of GRIB2 messages; DataArray has 4 and GRIB2 index has 2. Consider applying a filter for dimensions: \['leadTime', 'duration'\]"):
+        ds = xr.open_dataset(data / "2024101012_Milton_Adv22_e70_cum_dat.grb", engine="grib2io")