Merge pull request #535 from CatalystCode/enhancement/csv-import-with-out-of-order-columns

Enhancement/Enable import of CSVs with arbitrary column order

app/api/tests/data/example_out_of_order_columns.csv

@@ -0,0 +1,4 @@
+label,foo,text,bar,baz
+Positive,foo1,AAA,barA,baz
+Positive,foo2,BBB,barB,bazz
+Negative,foo3,CCC,barC,bazzz

app/api/tests/test_api.py

@@ -1067,6 +1067,21 @@ class TestUploader(APITestCase):
                                 file_format='csv',
                                 expected_status=status.HTTP_201_CREATED)
 
+    def test_can_upload_classification_csv_with_out_of_order_columns(self):
+        self.upload_test_helper(project_id=self.classification_project.id,
+                                filename='example_out_of_order_columns.csv',
+                                file_format='csv',
+                                expected_status=status.HTTP_201_CREATED)
+
+        self.label_test_helper(
+            project_id=self.classification_project.id,
+            expected_labels=[
+                {'text': 'Positive'},
+                {'text': 'Negative'},
+            ],
+            expected_label_keys=[],
+        )
+
     def test_can_upload_csv_with_non_utf8_encoding(self):
         self.upload_test_helper(project_id=self.classification_project.id,
                                 filename='example.utf16.csv',

app/api/utils.py

@@ -356,8 +356,9 @@ class ExcelParser(FileParser):
                 data.append({'text': row[0]})
             # Text, labels and metadata columns
             elif len(row) == len(columns) and len(row) >= 2:
-                text, label = row[:2]
-                meta = json.dumps(dict(zip(columns[2:], row[2:])))
+                datum = dict(zip(columns, row))
+                text, label = datum.pop('text'), datum.pop('label')
+                meta = json.dumps(datum)
                 j = {'text': text, 'labels': [label], 'meta': meta}
                 data.append(j)
             else: