.gitlab-ci.yml

@@ -11,7 +11,7 @@ include:
 
 # For jobs that run backend scripts directly
 .backend-setup:
-  image: registry.gitlab.com/teklia/arkindex/backend/base:django-4.1.4
+  image: registry.gitlab.com/teklia/arkindex/backend/base:django-4.1.5
 
   cache:
     paths:

Dockerfile

-FROM registry.gitlab.com/teklia/arkindex/backend/base:django-4.1.4 as build
+FROM registry.gitlab.com/teklia/arkindex/backend/base:django-4.1.5 as build
 
 RUN mkdir build
 ADD . build
 RUN cd build && python3 setup.py sdist
 
-FROM registry.gitlab.com/teklia/arkindex/backend/base:django-4.1.4
+FROM registry.gitlab.com/teklia/arkindex/backend/base:django-4.1.5
 ARG PONOS_BRANCH=master
 ARG PONOS_ID=10017043
 ARG TRANSKRIBUS_BRANCH=master

Dockerfile.binary

@@ -60,7 +60,7 @@ RUN python -m nuitka \
       arkindex/manage.py
 
 # Start over from a clean setup
-FROM registry.gitlab.com/teklia/arkindex/backend/base:django-4.1.4 as build
+FROM registry.gitlab.com/teklia/arkindex/backend/base:django-4.1.5 as build
 
 # Import files from compilation
 RUN mkdir /usr/share/arkindex

arkindex/documents/api/ml.py

@@ -424,7 +424,10 @@ class ClassificationCreate(CreateAPIView):
 @extend_schema(tags=['classifications'])
 class ClassificationBulk(CreateAPIView):
     """
-    Create multiple classifications at once on the same element with the same classifier.
+    Create multiple classifications at once on the same element.
+
+    Exactly one of `worker_version` or `worker_run_id` fields must be set.
+    If `worker_run_id` is set, the worker version will be deduced from it.
     """
     serializer_class = ClassificationsSerializer
     permission_classes = (IsVerified, )

arkindex/documents/serializers/ml.py

@@ -653,6 +653,14 @@ class ClassificationsSerializer(serializers.Serializer):
         if len(ml_class_ids) != len(set(ml_class_ids)):
             errors['classifications'].append('Duplicated ML classes are not allowed from the same worker version or worker run.')
 
+        worker_run = data.get('worker_run', None)
+        if not worker_run and not data.get('worker_version'):
+            errors['non_field_errors'] = [
+                'Exactly one of `worker_version` or `worker_run_id` must be set.'
+            ]
+        elif worker_run:
+            data['worker_version'] = WorkerVersion(id=worker_run.version_id)
+
         # Check that all ML classes exist in one query
         ml_class_count = (
             MLClass

arkindex/documents/tests/test_bulk_classification.py

@@ -114,7 +114,31 @@ class TestBulkClassification(FixtureAPITestCase):
             ],
         )
 
-    def test_worker_version_xor_worker_run(self):
+    def test_worker_version_or_worker_run(self):
+        """Either a worker run or a worker version is required"""
+        self.client.force_login(self.user)
+        with self.assertNumQueries(6):
+            response = self.client.post(
+                reverse('api:classification-bulk'),
+                format='json',
+                data={
+                    "parent": str(self.page.id),
+                    "classifications": [
+                        {
+                            'ml_class': str(self.cat_class.id),
+                            "confidence": 0.42,
+                        }
+                    ]
+                }
+            )
+            self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+
+        self.assertDictEqual(response.json(), {
+            'non_field_errors': ['Exactly one of `worker_version` or `worker_run_id` must be set.'],
+        })
+
+    def test_worker_version_and_worker_run(self):
+        """Worker run and worker version cannot be set at the same time"""
         self.client.force_login(self.user)
         with self.assertNumQueries(7):
             response = self.client.post(
@@ -162,21 +186,21 @@ class TestBulkClassification(FixtureAPITestCase):
             )
             self.assertEqual(response.status_code, status.HTTP_201_CREATED)
 
-        first_cl, second_cl = self.page.classifications.order_by('-confidence').values_list('id', flat=True)
+        first_cl, second_cl = self.page.classifications.order_by('-confidence').all()
         self.assertEqual(response.json(), {
             'parent': str(self.page.id),
-            'worker_version': None,
+            'worker_version': str(self.worker_version.id),
             'worker_run_id': str(self.worker_run.id),
             'classifications': [
                 {
-                    'id': str(first_cl),
+                    'id': str(first_cl.id),
                     'ml_class': str(self.dog_class.id),
                     'confidence': 0.99,
                     'high_confidence': True,
                     'state': 'pending',
                 },
                 {
-                    'id': str(second_cl),
+                    'id': str(second_cl.id),
                     'ml_class': str(self.cat_class.id),
                     'confidence': 0.42,
                     'high_confidence': False,
@@ -198,6 +222,11 @@ class TestBulkClassification(FixtureAPITestCase):
                 ('cat', 0.42, False, self.worker_version.id, self.worker_run.id),
             ],
         )
+        # Worker run is set, and worker version is deduced from it
+        self.assertEqual(first_cl.worker_version, self.worker_version)
+        self.assertEqual(second_cl.worker_version, self.worker_version)
+        self.assertEqual(first_cl.worker_run, self.worker_run)
+        self.assertEqual(second_cl.worker_run, self.worker_run)
 
     def test_worker_run_not_found(self):
         self.client.force_login(self.user)

arkindex/process/serializers/workers.py

@@ -94,6 +94,10 @@ class UserConfigurationFieldSerializer(serializers.Serializer):
             UserConfigurationFieldType.List: [serializers.ListField(), list]
         }
 
+        if not isinstance(data, dict):
+            errors['__all__'] = [f'User configuration field definitions should be of type dict, not {type(data).__name__}.']
+            raise ValidationError(errors)
+
         for field in data:
             if field not in allowed_fields:
                 errors[field].append(

arkindex/process/tests/test_workers.py

@@ -1013,6 +1013,29 @@ class TestWorkersWorkerVersions(FixtureAPITestCase):
         self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
         self.assertEqual(response.json(), {"configuration": {"user_configuration": [["Expected a dictionary of items but got type \"str\"."]]}})
 
+    def test_invalid_user_configuration_item_not_dict(self):
+        self.client.force_login(self.internal_user)
+        response = self.client.post(
+            reverse("api:worker-versions", kwargs={"pk": str(self.worker_2.id)}),
+            data={
+                "revision_id": str(self.rev2.id),
+                "configuration": {
+                    "user_configuration": {
+                        "secrets": ["aaaaaa"]
+                    }
+                },
+                "gpu_usage": "disabled",
+            },
+            format="json",
+        )
+        self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+        self.assertEqual(response.json(), {"configuration": {
+            "user_configuration": [{
+                "secrets":
+                {'__all__': ['User configuration field definitions should be of type dict, not list.']}
+            }]
+        }})
+
     def test_invalid_user_configuration_wrong_field_type(self):
         self.client.force_login(self.internal_user)
         response = self.client.post(

base/requirements.txt

 boto3==1.18.13
 cryptography==3.4.7
-Django==4.1.4
+Django==4.1.5
 lxml==4.6.3
 psycopg2-binary==2.9.1