fix issue when categorical column have a default value (#115)

* fix issue when categorical column have a default value
OasisLMF · May 16, 2024 · 24607ff · 24607ff
1 parent 6ced5ca
commit 24607ff
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 1 deletion.
diff --git a/ods_tools/oed/source.py b/ods_tools/oed/source.py
@@ -305,7 +305,10 @@ def prepare_df(cls, df, column_to_field, ods_fields):
             if (field_info
                     and field_info['Default'] != 'n/a'
                     and (df[col].isna().any() or (field_info['pd_dtype'] == 'category' and df[col].isnull().any()))):
-                fill_empty(df, col, df[col].dtype.type(field_info['Default']))
+                if df[col].dtype.name == 'category':
+                    fill_empty(df, col, field_info['Default'])
+                else:
+                    fill_empty(df, col, df[col].dtype.type(field_info['Default']))
             elif df[col].dtype.name == 'category':
                 fill_empty(df, col, '')
 

diff --git a/tests/test_ods_package.py b/tests/test_ods_package.py
@@ -111,6 +111,24 @@ def test_load_oed_from_config(self):
                 exposure2 = OedExposure(**config)
             self.assertTrue(exposure.location.dataframe.equals(exposure2.location.dataframe))
 
+    def test_categorical_with_default(self):
+        # UseReinsDates is a string column with a non null default, check default setting works
+        with tempfile.TemporaryDirectory() as tmp_run_dir:
+            config = {
+                'ri_info': base_url + '/SourceReinsInfoOEDPiWind.csv',
+                'use_field': True
+            }
+            exposure = OedExposure(**config)
+            exposure.ri_info.dataframe['UseReinsDates'] = None
+            exposure.ri_info.dataframe.to_csv(os.path.join(tmp_run_dir, 'ri_info.csv'), index=False)
+
+            exposure = OedExposure(**{
+                'ri_info': os.path.join(tmp_run_dir, 'ri_info.csv'),
+                'use_field': True
+            })
+            ri_scope = exposure.ri_info.dataframe
+            self.assertTrue(isinstance(ri_scope, pd.DataFrame))
+
     def test_load_oed_from_df(self):
         location_df = pd.DataFrame({
             'PortNumber': [1, 1],