Check if all values in a groupby column are NaN or not NaN (#526)

intake_esm/cat.py

@@ -1,5 +1,6 @@
 import datetime
 import enum
+import functools
 import json
 import os
 import typing
@@ -12,6 +13,28 @@
 from ._search import search, search_apply_require_all_on
 
 
+def _allnan_or_nonan(df, column: str) -> bool:
+    """Check if all values in a column are NaN or not NaN
+
+    Returns
+    -------
+    bool
+        Whether the dataframe column has all NaNs or no NaN valles
+
+    Raises
+    ------
+    ValueError
+        When the column has a mix of NaNs non NaN values
+    """
+    if df[column].isnull().all():
+        return False
+    if df[column].isnull().any():
+        raise ValueError(
+            f'The data in the {column} column should either be all NaN or there should be no NaNs'
+        )
+    return True
+
+
 class AggregationType(str, enum.Enum):
     join_new = 'join_new'
     join_existing = 'join_existing'
@@ -286,6 +309,15 @@ def _cast_agg_columns_with_iterables(self) -> None:
 
     @property
     def grouped(self) -> typing.Union[pd.core.groupby.DataFrameGroupBy, pd.DataFrame]:
+
+        if self.aggregation_control.groupby_attrs:
+            self.aggregation_control.groupby_attrs = list(
+                filter(
+                    functools.partial(_allnan_or_nonan, self.df),
+                    self.aggregation_control.groupby_attrs,
+                )
+            )
+
         if self.aggregation_control.groupby_attrs and set(
             self.aggregation_control.groupby_attrs
         ) != set(self.df.columns):

intake_esm/utils.py

@@ -3,28 +3,6 @@
 import sys
 
 
-def _allnan_or_nonan(df, column: str) -> bool:
-    """Check if all values in a column are NaN or not NaN
-
-    Returns
-    -------
-    bool
-        Whether the dataframe column has all NaNs or no NaN valles
-
-    Raises
-    ------
-    ValueError
-        When the column has a mix of NaNs non NaN values
-    """
-    if df[column].isnull().all():
-        return False
-    if df[column].isnull().any():
-        raise ValueError(
-            f'The data in the {column} column should either be all NaN or there should be no NaNs'
-        )
-    return True
-
-
 def show_versions(file=sys.stdout):  # pragma: no cover
     """print the versions of intake-esm and its dependencies.
        Adapted from xarray/util/print_versions.py