FIX bug when flow only has a single tag

Author: mfeurer

openml/flows/flow.py

@@ -308,18 +308,7 @@ def _from_dict(cls, xml_dict):
                 flow = OpenMLFlow._from_dict(component)
                 components[component['oml:identifier']] = flow
         arguments['components'] = components
-
-        tags = []
-        if 'oml:tag' in dic and dic['oml:tag'] is not None:
-            # In case of a single tag xmltodict returns a dict, otherwise a list
-            if isinstance(dic['oml:tag'], dict):
-                oml_tags = [dic['oml:tag']]
-            else:
-                oml_tags = dic['oml:tag']
-
-            for tag in oml_tags:
-                tags.append(tag)
-        arguments['tags'] = tags
+        arguments['tags'] = extract_tags(dic)
 
         arguments['model'] = None
         flow = cls(**arguments)
@@ -385,3 +374,16 @@ def _add_if_nonempty(dic, key, value):
     if value is not None:
         dic[key] = value
 
+
+def extract_tags(dic):
+    if 'oml:tag' in dic and dic['oml:tag'] is not None:
+        if isinstance(dic['oml:tag'], six.string_types):
+            oml_tags = [dic['oml:tag']]
+        elif isinstance(dic['oml:tag'], list):
+            oml_tags = dic['oml:tag']
+        else:
+            raise ValueError('Received not string and non list as tag item')
+
+        return oml_tags
+    else:
+        return None

openml/runs/functions.py

@@ -616,7 +616,7 @@ def _create_run_from_xml(xml):
 
     tags = None
     if 'oml:tag' in run:
-        if isinstance(run['oml:tag'], str):
+        if isinstance(run['oml:tag'], six.string_types):
             tags = [run['oml:tag']]
         elif isinstance(run['oml:tag'], list):
             tags = run['oml:tag']

tests/test_flows/test_flow.py

@@ -320,3 +320,15 @@ def test_sklearn_to_upload_to_flow(self):
         self.assertTrue('openml-python' in new_flow.tags)
         self.assertTrue('unittest' in new_flow.tags)
         new_flow.model.fit(X, y)
+
+    def test_extract_tags(self):
+        flow_xml = "<oml:tag>study_14</oml:tag>"
+        flow_dict = xmltodict.parse(flow_xml)
+        tags = openml.flows.flow.extract_tags(flow_dict)
+        self.assertEqual(tags, ['study_14'])
+
+        flow_xml = "<oml:flow><oml:tag>OpenmlWeka</oml:tag>\n" \
+                   "<oml:tag>weka</oml:tag></oml:flow>"
+        flow_dict = xmltodict.parse(flow_xml)
+        tags = openml.flows.flow.extract_tags(flow_dict['oml:flow'])
+        self.assertEqual(tags, ['OpenmlWeka', 'weka'])