Add init time to parquet files (#349)

I forgot to test for the presence of the initialization time in the
parquet output.

src/unified_graphics/etl/diag.py

@@ -239,6 +239,7 @@ def prep_dataframe(ds: xr.Dataset) -> pd.DataFrame:
     df = ds.to_dataframe()
 
     df["loop"] = ds.loop
+    df["initialization_time"] = datetime.fromisoformat(ds.initialization_time)
 
     # FIXME: Clean the string columns Station_ID, Provider_Name, Subprovider_Name
 

tests/etl/test_save.py

@@ -40,6 +40,7 @@ def parquet_file(model, data_path):
 
 def dataset_to_table(dataset: xr.Dataset) -> pd.DataFrame:
     df = dataset.to_dataframe()
+    df["initialization_time"] = datetime.fromisoformat(dataset.initialization_time)
     df["loop"] = dataset.loop
 
     return df.astype({"loop": "category"})
@@ -258,7 +259,9 @@ def dataset(self, model, diag_dataset, session, zarr_file):
 
     @pytest.fixture(scope="class")
     def dataframe(self, dataset):
-        return pd.concat(map(dataset_to_table, dataset))
+        return pd.concat(map(dataset_to_table, dataset)).sort_values(
+            "initialization_time"
+        )
 
     @pytest.mark.parametrize(
         "init_time,expected", (("2022-05-05T14:00", 0), ("2022-05-05T15:00", 1))
@@ -272,7 +275,7 @@ def test_parquet(self, dataframe, parquet_file):
         result = pd.read_parquet(
             parquet_file / "ps",
             filters=(("loop", "=", "ges"),),
-        )
+        ).sort_values("initialization_time")
 
         pd.testing.assert_frame_equal(result, dataframe)