Allow user to have non-str in pandas passed to Labeler (capitalone#343)

* fix: allow user to have non-str in pandas passed to Labeler

* fix: naming to be sample as not necessarily string

dataprofiler/labelers/data_processing.py

@@ -1434,11 +1434,11 @@ def convert_to_structured_analysis(self, sentences, results, label_mapping,
         if 'conf' in results:
             confs_out = np.zeros((len(results['pred']), num_labels))
 
-        for i, label_sentences in enumerate(zip(results['pred'], sentences)):
-            column_labels, sentence = label_sentences
+        for i, label_samples in enumerate(zip(results['pred'], sentences)):
+            column_labels, sample = label_samples
 
             # get count of all labels in prediction
-            column_label_counter = Counter(column_labels[:len(sentence)])
+            column_label_counter = Counter(column_labels[:len(str(sample))])
             column_label_counter.pop(ignore_value, None)
             modes = [entity_id for entity_id, count in
                      column_label_counter.most_common() if

dataprofiler/tests/labelers/test_integration_struct_data_labeler.py

@@ -63,6 +63,14 @@ def test_fit_with_default_model(self):
         # validate epoch id
         self.assertEqual(2, default.model._epoch_id)
 
+    def test_labeling_mixed_df(self):
+        labeler = dp.labelers.StructuredDataLabeler()
+
+        # test int, float, str, dict
+        data = pd.DataFrame([1, 1.1, 'string', {'random': 3}])
+        predictions = labeler.predict(data)
+        self.assertEqual(4, len(predictions['pred']))
+
     def test_data_labeler_change_labels(self):
         """test changing labels of data labeler with fitting data"""
         # constructing default StructuredDataLabeler()
@@ -391,5 +399,6 @@ def test_warning_tf_multiple_dp_with_update(self):
 
         profile1.update_profile(data)
 
+
 if __name__ == '__main__':
     unittest.main()