bleh

Author: manishvenu

regional_mom6/regional_mom6.py

@@ -27,6 +27,7 @@
     find_files_by_pattern,
     try_pint_convert,
 )
+from regional_mom6.validate import validate_obc_file, validate_general_file
 
 warnings.filterwarnings("ignore")
 
@@ -1326,8 +1327,29 @@ def setup_initial_condition(
         self.ic_tracers = tracers_out
         self.ic_vels = vel_out
 
-        print("done setting up initial condition.")
-
+        validate_general_file(
+            eta_out,
+            ["eta"],
+            {
+                "eta_t": {"_FillValue": None},
+            },
+        )
+        validate_general_file(
+            tracers_out,
+            ["temp", "salt"],
+            encoding={
+                "temp": {"_FillValue": -1e20, "missing_value": -1e20},
+                "salt": {"_FillValue": -1e20, "missing_value": -1e20},
+            },
+        )
+        validate_general_file(
+            vel_out,
+            ["u", "v"],
+            {
+                "u": {"_FillValue": netCDF4.default_fillvals["f4"]},
+                "v": {"_FillValue": netCDF4.default_fillvals["f4"]},
+            },
+        )
         return
 
     def get_glorys(self, raw_boundaries_path):

regional_mom6/validate.py

@@ -75,11 +75,6 @@ def validate_obc_file(
     if encoding_dict is None:
         encoding_dict = {}
     ds = get_file(file)
-
-    print(
-        "This function identifies variables by if they have the word 'segment' in the name and don't start with nz,dz,lon,lat."
-    )
-
     for var in variable_names:
 
         # check variable name format
@@ -114,5 +109,30 @@ def validate_obc_file(
             )
 
 
+def validate_general_file(
+    file, variable_names: list, encoding_dict=None, surface_var="eta"
+):
+    """Validate initial condition file and other domain spanning files"""
+    if encoding_dict is None:
+        encoding_dict = {}
+    ds = get_file(file)
+
+    for var in variable_names:
+
+        # Add encodings
+        if var in encoding_dict:
+            for key, value in encoding_dict[var].items():
+                ds[var].attrs[key] = value
+
+        # Check if there is a non-NaN fill value
+        _check_fill_value(ds[var])
+
+        # check coordinates
+        _check_coordinates(ds, var_name=var)
+
+        # Check the correct number of dimensions
+        _check_required_dimensions(ds[var], surface=(var == surface_var))
+
+
 def ends_with_3_digits(s: str) -> bool:
     return bool(re.search(r"_\d{3}$", s))

tests/test_validate.py

@@ -138,7 +138,9 @@ def test_validate_obc_file_valid(caplog):
         ds[var].attrs["_FillValue"] = -999.0
         ds[var].attrs["coordinates"] = "lon lat"
 
-    validate_obc_file(ds, ["temp_segment_001"], surface_var="eta_segment_001")
+    validate_obc_file(
+        ds, ["temp_segment_001", "eta_segment_001"], surface_var="eta_segment_001"
+    )
 
 
 def test_validate_obc_file_issues(caplog):
@@ -179,3 +181,25 @@ def test_validate_obc_file_encoding_dict():
     validate_obc_file(ds, ["temp_segment_001"], encoding_dict=encoding_dict)
 
     assert ds["temp_segment_001"].attrs["units"] == "celsius"
+
+
+def test_validate_general_file_valid(caplog):
+    """Valid general file with all required attributes passes"""
+    ds = xr.Dataset(
+        {
+            "temp": (["time", "z", "x", "y"], np.random.rand(2, 3, 4, 5)),
+            "dz_temp": (
+                ["time", "z", "x", "y"],
+                np.random.rand(2, 3, 4, 5),
+            ),
+            "eta": (["time", "x", "y"], np.random.rand(2, 4, 5)),
+            "lon": (["x", "y"], np.random.rand(4, 5)),
+            "lat": (["x", "y"], np.random.rand(4, 5)),
+        }
+    )
+
+    for var in ds.data_vars:
+        ds[var].attrs["_FillValue"] = -999.0
+        ds[var].attrs["coordinates"] = "lon lat"
+
+    validate_obc_file(ds, ["temp", "eta"], surface_var="eta")