Merge pull request #38 from sedos-project/fix/none-values-in-fk-column

Fix None values in FK column

CHANGELOG.md

@@ -17,6 +17,10 @@ Here is a template for new release sections
 -
 ```
 
+## [Unreleased]
+### Changed
+- "None" values in possible FK column are overwritten by FK mapping
+
 ## [0.21.0] - 2024-06-25
 ### Changed
 - in case of bandwidth values, first value is used for process

data_adapter/preprocessing.py

@@ -261,10 +261,14 @@ def __unpack_bandwidths(self, df: pd.DataFrame) -> pd.DataFrame:
         Currently only supports to return the first argument of bandwidths
         Different bandwidth types are not supported yet
 
-        Example:
-
+        Parameters
+        ----------
+        df : pd.DataFrame
+            Data containing bandwidths
 
-        Returns:
+        Returns
+        -------
+        pd.Dataframe
             Modified Dataframe
 
         """
@@ -442,11 +446,16 @@ def _get_foreign_keys(process: str, df: pd.DataFrame) -> dict[str, ForeignKey]:
         # Check if Fks are unique (cannot have different FKs per process/subprocess)
         fk_candidates = {}
         for fk_column in fk_column_candidates:
-            if len(df[fk_column].unique()) > 1:
+            column_data_without_none = df[fk_column][~df[fk_column].isnull()]
+            if len(column_data_without_none.unique()) > 1:
                 continue  # no candidate
-            fk = df[fk_column].iloc[0]
+            fk = column_data_without_none.iloc[0]
             if "." not in fk:
                 continue  # no candidate
+            if df[fk_column].isnull().sum() > 0:
+                logging.warning(
+                    f"None values in column '{fk_column}' of process '{process}' will be overwritten by FK values."
+                )
             fk_candidates[fk_column] = ForeignKey(*fk.split("."))
         return fk_candidates