Extend JSON schema with 'uniqueObjects' keyword and validator

Fixes #34

ckanext/metadata/logic/json_validator.py

@@ -110,12 +110,16 @@ def add_error(node, path, message):
                 error.path.append(required_key)
 
             elif error.schema_path[-1] == 'minItems':
-                error.path.append('__length')
+                error.path.append('__minItems')
                 error.message = 'Array has too few items'
 
             elif error.schema_path[-1] == 'not' and error.validator_value == {}:
                 error.message = 'This key may not be present in the dictionary'
 
+            elif error.schema_path[-1] == 'uniqueObjects':
+                error.path.append('__uniqueObjects')
+                error.message = 'Array has non-unique objects'
+
             add_error(errors, error.path, error.message)
 
         return errors

ckanext/metadata/logic/json_validator_functions.py

@@ -62,6 +62,20 @@ def role_validator(validator, role_name, instance, schema):
         yield jsonschema.ValidationError("Role validation has not been implemented yet")
 
 
+def unique_objects_validator(validator, key_properties, instance, schema):
+    """
+    "uniqueObjects" keyword validator: for an array comprising objects, this checks that the objects
+    are unique with respect to the named properties.
+    """
+    if validator.is_type(instance, 'array') and validator.is_type(schema.get('items', {}), 'object'):
+        key_objects = []
+        for obj in instance:
+            key_object = {k: v for k, v in obj.items() if k in key_properties}
+            if key_object in key_objects:
+                yield jsonschema.ValidationError(_("%r has non-unique objects") % (instance,))
+            key_objects += [key_object]
+
+
 @checks_format('doi')
 def is_doi(instance):
     if not isinstance(instance, basestring):

ckanext/metadata/logic/workflow_validator.py

@@ -1,7 +1,7 @@
 # encoding: utf-8
 
 from ckanext.metadata.logic.json_validator import JSONValidator
-from ckanext.metadata.logic.json_validator_functions import objectid_validator, role_validator
+from ckanext.metadata.logic.json_validator_functions import objectid_validator, role_validator, unique_objects_validator
 
 
 class WorkflowValidator(JSONValidator):
@@ -14,6 +14,7 @@ def _validators(cls):
         return {
             'objectid': objectid_validator,
             'role': role_validator,
+            'uniqueObjects': unique_objects_validator,
         }
 
     @classmethod

ckanext/metadata/tests/__init__.py

@@ -6,6 +6,7 @@
 from paste.deploy.converters import asbool
 import pkg_resources
 from collections import deque
+import traceback
 
 from ckan.tests import factories as ckan_factories
 from ckan.tests.helpers import FunctionalTestBase, call_action
@@ -194,6 +195,7 @@ def _test_action(self, action_name, should_error=False, exception_class=tk.Valid
             else:
                 result = e.message
         except Exception, e:
+            traceback.print_exc()
             assert False, "Unexpected exception %s: %s" % (type(e), e)
         else:
             if should_error:

ckanext/metadata/tests/test_metadata_record_actions.py

@@ -786,7 +786,7 @@ def test_workflow_transition_published(self):
                                               *jsonpatch_ids,
                                               **{
                                                   'errors/identifier': 'This key may not be present in the dictionary',
-                                                  'quality_control/__length': 'Array has too few items',
+                                                  'quality_control/__minItems': 'Array has too few items',
                                               })
         assert_package_has_extra(metadata_record['id'], 'workflow_state_id', '')
 
@@ -799,16 +799,15 @@ def test_workflow_transition_published(self):
                                       value=json.dumps({"userid": self.normal_user['id'], "date": "2018-08-15"}),
                                       )['id']]
 
-        # TODO: the following depends on implementation of userid uniqueness checking
-        # self._test_action('metadata_record_workflow_state_transition', id=metadata_record['id'],
-        #                   workflow_state_id=workflow_state_published['id'])
-        # self._assert_workflow_activity_logged(metadata_record['id'],
-        #                                       workflow_state_published['id'],
-        #                                       *jsonpatch_ids,
-        #                                       **{
-        #                                           'quality_control': 'uniqueness of user id...',
-        #                                       })
-        # assert_package_has_extra(metadata_record['id'], 'workflow_state_id', '')
+        self._test_action('metadata_record_workflow_state_transition', id=metadata_record['id'],
+                          workflow_state_id=workflow_state_published['id'])
+        self._assert_workflow_activity_logged(metadata_record['id'],
+                                              workflow_state_published['id'],
+                                              *jsonpatch_ids,
+                                              **{
+                                                  'quality_control/__uniqueObjects': 'Array has non-unique objects',
+                                              })
+        assert_package_has_extra(metadata_record['id'], 'workflow_state_id', '')
 
         jsonpatch_ids += [call_action('metadata_record_workflow_annotation_create', id=metadata_record['id'],
                                       path='/quality_control',

examples/workflow_state_published_rules.json

@@ -34,8 +34,7 @@
                 "required": ["userid", "date"]
             },
             "minItems": 2,
-            "uniqueItems": true,
-            "$comment": "to fix: as it stands this sub-schema will not enforce different users, because of the date property"
+            "uniqueObjects": ["userid"]
         },
         "validated": {
             "type": "boolean",