Validate that the metadata is always a dict

sdmetrics/reports/base_report.py

@@ -55,14 +55,25 @@ def validate(self, real_data, synthetic_data, metadata):
             metadata (dict):
                 The metadata of the table.
         """
-        if not isinstance(metadata, dict):
-            metadata = metadata.to_dict()
-
         self._validate_metadata_matches_data(real_data, synthetic_data, metadata)
 
     def _handle_results(self, verbose):
         raise NotImplementedError
 
+    @staticmethod
+    def _convert_metadata(metadata):
+        """If the metadta is not a dict, try to convert it."""
+        if not isinstance(metadata, dict):
+            try:
+                metadata = metadata.to_dict()
+            except Exception:
+                raise TypeError(
+                    'The provided metadata is not a dictionary and does not have a to_dict method.'
+                    'Please convert the metadata to a dictionary.'
+                )
+
+        return metadata
+
     @staticmethod
     def convert_datetimes(real_data, synthetic_data, metadata):
         """Try to convert all datetime columns to datetime dtype.
@@ -101,6 +112,7 @@ def generate(self, real_data, synthetic_data, metadata, verbose=True):
             verbose (bool):
                 Whether or not to print report summary and progress.
         """
+        metadata = self._convert_metadata(metadata)
         self.validate(real_data, synthetic_data, metadata)
         self.convert_datetimes(real_data, synthetic_data, metadata)
 

tests/unit/reports/test_base_report.py

@@ -152,7 +152,56 @@ def test_convert_datetimes(self):
         pd.testing.assert_frame_equal(real_data, expected_real_data)
         pd.testing.assert_frame_equal(synthetic_data, expected_synthetic_data)
 
-    def test_generate(self):
+    def test__convert_metadata_with_to_dict_method(self):
+        """Test ``_convert_metadata`` when the metadata object has a ``to_dict`` method."""
+        # Setup
+        metadata_example = {
+            'column1': {'sdtype': 'numerical'},
+            'column2': {'sdtype': 'categorical'},
+        }
+
+        class Metadata:
+            def __init__(self):
+                self.columns = metadata_example
+
+            def to_dict(self):
+                return self.columns
+
+        metadata = Metadata()
+
+        # Run
+        converted_metadata = BaseReport._convert_metadata(metadata)
+
+        # Assert
+        assert converted_metadata == metadata_example
+
+    def test__convert_metadata_without_to_dict_method(self):
+        """Test ``_convert_metadata`` when the metadata object has no ``to_dict`` method."""
+        # Setup
+        metadata_example = {
+            'column1': {'sdtype': 'numerical'},
+            'column2': {'sdtype': 'categorical'},
+        }
+
+        class Metadata:
+            def __init__(self):
+                self.columns = metadata_example
+
+        metadata = Metadata()
+
+        # Run and Assert
+        expected_message = re.escape(
+            'The provided metadata is not a dictionary and does not have a to_dict method.'
+            'Please convert the metadata to a dictionary.'
+        )
+        with pytest.raises(TypeError, match=expected_message):
+            BaseReport._convert_metadata(metadata)
+
+        result = BaseReport._convert_metadata(metadata_example)
+        assert result == metadata_example
+
+    @patch('sdmetrics.reports.base_report.BaseReport._convert_metadata')
+    def test_generate(self, mock__convert_metadata):
         """Test the ``generate`` method.
 
         This test checks that the method calls the ``validate`` method and the ``get_score``
@@ -183,11 +232,13 @@ def test_generate(self):
                 'column2': {'sdtype': 'categorical'}
             }
         }
+        mock__convert_metadata.return_value = metadata
 
         # Run
         base_report.generate(real_data, synthetic_data, metadata, verbose=False)
 
         # Assert
+        mock__convert_metadata.assert_called_once_with(metadata)
         mock_validate.assert_called_once_with(real_data, synthetic_data, metadata)
         mock_handle_results.assert_called_once_with(False)
         base_report._properties['Property 1'].get_score.assert_called_with(