[#76] Emit warning if field in rollUp doesn't exist in the schema

Author: Bjwebb

flattentool/schema.py

@@ -99,6 +99,9 @@ def parse_schema_dict(self, parent_name, schema_dict, parent_id_fields=None):
                         fields = self.parse_schema_dict(parent_name+'/'+property_name+'[]',
                                 property_schema_dict['items'],
                                 parent_id_fields=id_fields)
+
+                        rolledUp = set()
+
                         for field, child_title in fields:
                             if self.use_titles:
                                 if not child_title:
@@ -110,7 +113,14 @@ def parse_schema_dict(self, parent_name, schema_dict, parent_id_fields=None):
                             if child_title:
                                 self.sub_sheets[sub_sheet_name].titles[child_title] = field
                             if self.rollup and 'rollUp' in property_schema_dict and field in property_schema_dict['rollUp']:
+                                rolledUp.add(field)
                                 yield property_name+'[]/'+field, (title+':'+child_title if title and child_title else None)
+
+                        # Check that all items in rollUp are in the schema
+                        if self.rollup and 'rollUp' in property_schema_dict:
+                            missedRollUp = set(property_schema_dict['rollUp']) - rolledUp
+                            if missedRollUp:
+                                warn('{} in rollUp but not in schema'.format(', '.join(missedRollUp)))
                     else:
                         raise ValueError
                 elif 'string' in property_type_set:

flattentool/tests/test_schema_parser.py

@@ -341,6 +341,35 @@ def test_rollup():
     assert set(parser.sub_sheets) == set(['testA'])
     assert set(parser.sub_sheets['testA']) == set(['ocid', 'testB', 'testC'])
 
+def test_bad_rollup(recwarn):
+    '''
+    When rollUp is specified, but the field is missing in the schema, we expect
+    a warning.
+
+    '''
+    parser = SchemaParser(root_schema_dict={
+        'properties': {
+            'testA': {
+                'type': 'array',
+                'rollUp': [ 'testB' ],
+                'items': {
+                    'type': 'object',
+                    'properties': {
+                        'testC': type_string
+                    }
+                }
+            },
+        }
+    }, rollup=True)
+    parser.parse()
+
+    w = recwarn.pop(UserWarning)
+    assert 'testB in rollUp but not in schema' in text_type(w.message)
+
+    assert set(parser.main_sheet) == set()
+    assert set(parser.sub_sheets) == set(['testA'])
+    assert set(parser.sub_sheets['testA']) == set(['ocid', 'testC'])
+
 
 def test_sub_sheet_custom_id():
     parser = SchemaParser(root_schema_dict={