Merge branch 'clear-process' into 'master'

New ClearProcess endpoint

Closes #1045

See merge request !1717

arkindex/dataimport/api.py

@@ -23,6 +23,7 @@ from rest_framework import permissions, status
 from rest_framework.exceptions import NotFound, PermissionDenied, ValidationError
 from rest_framework.generics import (
     CreateAPIView,
+    DestroyAPIView,
     GenericAPIView,
     ListAPIView,
     ListCreateAPIView,
@@ -1741,3 +1742,38 @@ class ApplyProcessTemplate(ProcessACLMixin, WorkerACLMixin, CreateAPIView):
             status=status.HTTP_200_OK,
             data=DataImportSerializer(dataimport, context={'request': self.request}).data
         )
+
+
+@extend_schema_view(
+    delete=extend_schema(
+        operation_id='ClearProcess',
+        tags=['imports'],
+        responses={
+            204: None,
+        },
+    )
+)
+class ClearProcess(ProcessACLMixin, DestroyAPIView):
+    """
+    Remove all templates and/or workers from a (not started) process
+    """
+    permission_classes = (IsVerified, )
+    queryset = DataImport.objects.all()
+
+    def check_object_permissions(self, request, process):
+        super().check_object_permissions(request, process)
+
+        access_level = self.process_access_level(process)
+        required_access = Role.Admin.value
+
+        if not access_level:
+            raise NotFound
+        if access_level < required_access:
+            raise PermissionDenied(detail='You do not have a sufficient access level to this process.')
+        if process.workflow_id is not None:
+            raise ValidationError({'__all__': ['A process can only be cleared before getting started.']})
+
+    def destroy(self, request, *args, **kwargs):
+        process = self.get_object()
+        process.clear()
+        return Response(status=status.HTTP_204_NO_CONTENT)

arkindex/dataimport/models.py

@@ -395,6 +395,11 @@ class DataImport(IndexableModel):
         else:
             self.start()
 
+    def clear(self):
+        self.worker_runs.all().delete()
+        self.template_id = None
+        self.save()
+
 
 class DataImportElement(models.Model):
     """

arkindex/dataimport/tests/test_imports.py

@@ -65,6 +65,7 @@ class TestImports(FixtureAPITestCase):
         )
         cls.page_type = ElementType.objects.get(corpus=cls.corpus, slug='page')
         cls.recognizer = WorkerVersion.objects.get(worker__slug='reco')
+        cls.dla = WorkerVersion.objects.get(worker__slug='dla')
         cls.version_gpu = WorkerVersion.objects.get(worker__slug='worker-gpu')
         cls.workers_process = cls.corpus.imports.get(mode=DataImportMode.Workers)
         cls.version_with_model = WorkerVersion.objects.get(worker__slug='generic')
@@ -1469,3 +1470,133 @@ class TestImports(FixtureAPITestCase):
                 }
             ]
         )
+
+    def test_clear_process(self):
+        process = self.corpus.imports.create(
+            creator=self.user,
+            mode=DataImportMode.Workers,
+            element_type=self.corpus.types.get(slug='page')
+        )
+        process.worker_runs.create(version=self.dla, parents=[], configuration=None)
+        process.worker_runs.create(version=self.recognizer, parents=[], configuration=None)
+        self.assertEqual(process.worker_runs.count(), 2)
+
+        self.client.force_login(self.user)
+        with self.assertNumQueries(9):
+            response = self.client.delete(reverse('api:clear-process', kwargs={'pk': str(process.id)}))
+        self.assertEqual(response.status_code, status.HTTP_204_NO_CONTENT)
+        process.refresh_from_db()
+        self.assertEqual(process.worker_runs.count(), 0)
+
+    def test_clear_process_requires_login(self):
+        process = self.corpus.imports.create(
+            creator=self.user,
+            mode=DataImportMode.Workers,
+            element_type=self.corpus.types.get(slug='page')
+        )
+        process.worker_runs.create(version=self.dla, parents=[], configuration=None)
+        process.worker_runs.create(version=self.recognizer, parents=[], configuration=None)
+        self.assertEqual(process.worker_runs.count(), 2)
+
+        with self.assertNumQueries(0):
+            response = self.client.delete(reverse('api:clear-process', kwargs={'pk': str(process.id)}))
+        self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+
+    def test_clear_process_requires_verified(self):
+        process = self.corpus.imports.create(
+            creator=self.user,
+            mode=DataImportMode.Workers,
+            element_type=self.corpus.types.get(slug='page')
+        )
+        process.worker_runs.create(version=self.dla, parents=[], configuration=None)
+        process.worker_runs.create(version=self.recognizer, parents=[], configuration=None)
+        self.assertEqual(process.worker_runs.count(), 2)
+
+        self.user.verified_email = False
+        self.user.save()
+        self.client.force_login(self.user)
+        with self.assertNumQueries(2):
+            response = self.client.delete(reverse('api:clear-process', kwargs={'pk': str(process.id)}))
+        self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+        self.assertDictEqual(response.json(), {'detail': 'You do not have permission to perform this action.'})
+
+    def test_clear_process_does_not_exist(self):
+        process = self.corpus.imports.create(
+            creator=self.user,
+            mode=DataImportMode.Workers,
+            element_type=self.corpus.types.get(slug='page')
+        )
+        process.worker_runs.create(version=self.dla, parents=[], configuration=None)
+        process.worker_runs.create(version=self.recognizer, parents=[], configuration=None)
+        self.assertEqual(process.worker_runs.count(), 2)
+
+        fake = uuid.uuid4()
+
+        self.client.force_login(self.user)
+        with self.assertNumQueries(3):
+            response = self.client.delete(reverse('api:clear-process', kwargs={'pk': str(fake)}))
+        self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+
+    def test_clear_process_not_unscheduled(self):
+        """
+        Cannot clear a process that has already started
+        """
+        process = self.corpus.imports.create(
+            creator=self.user,
+            mode=DataImportMode.Workers,
+            element_type=self.corpus.types.get(slug='page')
+        )
+        process.worker_runs.create(version=self.dla, parents=[], configuration=None)
+        process.worker_runs.create(version=self.recognizer, parents=[], configuration=None)
+        self.assertEqual(process.worker_runs.count(), 2)
+        process.start()
+        process.workflow.tasks.update(state=State.Running)
+        self.assertEqual(process.state, State.Running)
+
+        self.client.force_login(self.user)
+        with self.assertNumQueries(7):
+            response = self.client.delete(reverse('api:clear-process', kwargs={'pk': str(process.id)}))
+        self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+        self.assertDictEqual(response.json(), {'__all__': ['A process can only be cleared before getting started.']})
+
+    def test_clear_process_unscheduled_workflow(self):
+        """
+        Cannot clear a process that has a workflow, even unscheduled
+        """
+        process = self.corpus.imports.create(
+            creator=self.user,
+            mode=DataImportMode.Workers,
+            element_type=self.corpus.types.get(slug='page')
+        )
+        process.worker_runs.create(version=self.dla, parents=[], configuration=None)
+        process.worker_runs.create(version=self.recognizer, parents=[], configuration=None)
+        self.assertEqual(process.worker_runs.count(), 2)
+        process.start()
+        self.assertEqual(process.state, State.Unscheduled)
+
+        self.client.force_login(self.user)
+        with self.assertNumQueries(7):
+            response = self.client.delete(reverse('api:clear-process', kwargs={'pk': str(process.id)}))
+        self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+        self.assertDictEqual(response.json(), {'__all__': ['A process can only be cleared before getting started.']})
+
+    def test_clear_process_requires_permissions(self):
+        process = self.corpus.imports.create(
+            creator=self.user,
+            mode=DataImportMode.Workers,
+            element_type=self.corpus.types.get(slug='page')
+        )
+        process.worker_runs.create(version=self.dla, parents=[], configuration=None)
+        process.worker_runs.create(version=self.recognizer, parents=[], configuration=None)
+        self.assertEqual(process.worker_runs.count(), 2)
+
+        user2 = User.objects.create_user('email@mail.com', 'bob')
+        user2.verified_email = True
+        user2.save()
+
+        self.corpus.memberships.create(user=user2, level=Role.Contributor.value)
+        self.client.force_login(user2)
+        with self.assertNumQueries(6):
+            response = self.client.delete(reverse('api:clear-process', kwargs={'pk': str(process.id)}))
+        self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+        self.assertDictEqual(response.json(), {'detail': 'You do not have a sufficient access level to this process.'})

arkindex/project/api_v1.py

@@ -4,6 +4,7 @@ from django.views.generic.base import RedirectView
 from arkindex.dataimport.api import (
     ApplyProcessTemplate,
     AvailableRepositoriesList,
+    ClearProcess,
     CorpusWorkersActivity,
     CorpusWorkerVersionList,
     CorpusWorkflow,
@@ -239,6 +240,7 @@ api = [
     path('process/<uuid:pk>/activity-stats/', ProcessWorkersActivity.as_view(), name='process-activity-stats'),
     path('process/<uuid:pk>/template/', CreateProcessTemplate.as_view(), name='create-process-template'),
     path('process/<uuid:pk>/apply/', ApplyProcessTemplate.as_view(), name='apply-process-template'),
+    path('imports/<uuid:pk>/clear/', ClearProcess.as_view(), name='clear-process'),
 
     # ML models training
     path('modelversion/<uuid:pk>/', ModelVersionsRetrieve.as_view(), name='model-version-retrieve'),