From ae38fc35040cf84ebb7f371422cd43090a7d2d9d Mon Sep 17 00:00:00 2001
From: ml bonhomme <bonhomme@teklia.com>
Date: Thu, 28 Mar 2024 19:07:54 +0000
Subject: [PATCH] Dataset sets management API endpoints
---
arkindex/project/api_v1.py | 4 +
arkindex/training/api.py | 117 +++-
arkindex/training/serializers.py | 13 +-
.../training/tests/test_dataset_sets_api.py | 547 ++++++++++++++++++
4 files changed, 670 insertions(+), 11 deletions(-)
create mode 100644 arkindex/training/tests/test_dataset_sets_api.py
diff --git a/arkindex/project/api_v1.py b/arkindex/project/api_v1.py
index 87433cb65a..70684fd8de 100644
--- a/arkindex/project/api_v1.py
+++ b/arkindex/project/api_v1.py
@@ -108,6 +108,8 @@ from arkindex.training.api import (
DatasetClone,
DatasetElementDestroy,
DatasetElements,
+ DatasetSetCreate,
+ DatasetSets,
DatasetUpdate,
ElementDatasetSets,
MetricValueBulkCreate,
@@ -189,6 +191,8 @@ api = [
path("datasets/<uuid:pk>/clone/", DatasetClone.as_view(), name="dataset-clone"),
path("datasets/<uuid:pk>/elements/", DatasetElements.as_view(), name="dataset-elements"),
path("datasets/<uuid:dataset>/elements/<uuid:element>/", DatasetElementDestroy.as_view(), name="dataset-element"),
+ path("datasets/<uuid:pk>/sets/", DatasetSetCreate.as_view(), name="dataset-sets"),
+ path("datasets/<uuid:dataset>/sets/<uuid:set>/", DatasetSets.as_view(), name="dataset-set"),
# Moderation
path("classifications/", ClassificationCreate.as_view(), name="classification-create"),
diff --git a/arkindex/training/api.py b/arkindex/training/api.py
index d8e952f157..66b6d00a37 100644
--- a/arkindex/training/api.py
+++ b/arkindex/training/api.py
@@ -42,6 +42,7 @@ from arkindex.training.serializers import (
DatasetElementInfoSerializer,
DatasetElementSerializer,
DatasetSerializer,
+ DatasetSetSerializer,
ElementDatasetSetSerializer,
MetricValueBulkSerializer,
MetricValueCreateSerializer,
@@ -656,11 +657,6 @@ class CorpusDataset(CorpusACLMixin, ListCreateAPIView):
"""
Update a dataset. The dataset must not be in `complete` state.
- The sets array can only be updated following those cases:
- * Adding or removing sets, nothing specific is done.
- * Updating a single set within the array,
- all elements linked to the previous set will be moved to the new one.
-
Requires a **contributor** access to the dataset's corpus.
"""
),
@@ -670,11 +666,6 @@ class CorpusDataset(CorpusACLMixin, ListCreateAPIView):
"""
Partially update a dataset. The dataset must not be in `complete` state.
- The sets array can only be updated following those cases:
- * Adding or removing sets, nothing specific is done.
- * Updating a single set within the array,
- all elements linked to the previous set will be moved to the new one.
-
Requires a **contributor** access to the dataset's corpus.
"""
)
@@ -747,6 +738,112 @@ class DatasetElementCursorPagination(CountCursorPagination):
ordering = ("element_id", "id")
+class DatasetSetBase():
+ permission_classes = (IsVerified, )
+ serializer_class = DatasetSetSerializer
+
+ @cached_property
+ def dataset(self):
+ dataset_id = self.kwargs["pk"] if self.request.method == "POST" else self.kwargs["dataset"]
+ dataset = get_object_or_404(
+ Dataset.objects.filter(corpus__in=Corpus.objects.readable(self.request.user))
+ .prefetch_related("sets")
+ .select_related("corpus"),
+ pk=dataset_id
+ )
+ if self.request.method == "DELETE" and not Corpus.objects.admin(self.request.user).filter(pk=dataset.corpus_id).exists():
+ raise PermissionDenied(detail="You do not have admin access to this dataset.")
+ elif self.request.method != "DELETE" and not Corpus.objects.writable(self.request.user).filter(pk=dataset.corpus_id).exists():
+ raise PermissionDenied(detail="You do not have contributor access to this dataset.")
+ if dataset.state != DatasetState.Open:
+ raise ValidationError(detail="You can only add or update sets from a dataset in an open state.")
+ return dataset
+
+ def get_serializer_context(self):
+ context = super().get_serializer_context()
+ # Ignore this step when generating the schema with OpenAPI
+ if not self.kwargs:
+ return context
+ context["dataset"] = self.dataset
+ return context
+
+
+@extend_schema(tags=["datasets"])
+@extend_schema_view(
+ post=extend_schema(
+ operation_id="CreateDatasetSet",
+ description=dedent(
+ """
+ Sets can only be created in **open** datasets.
+
+ Requires **contributor** access to the dataset's corpus.
+ """
+ )
+ ),
+)
+class DatasetSetCreate(DatasetSetBase, CreateAPIView):
+ permission_classes = (IsVerified, )
+
+
+@extend_schema(tags=["datasets"])
+@extend_schema_view(
+ delete=extend_schema(
+ operation_id="DestroyDatasetSet",
+ description=dedent(
+ """
+ Delete a set from a dataset. Sets can only be deleted from **open** datasets.
+
+ Requires **admin** access to the dataset's corpus.
+
+ It is not possible to delete a dataset set if that set is selected in a process, or
+ if it is the only set in the dataset.
+ """
+ )
+ ),
+ put=extend_schema(
+ operation_id="UpdateDatasetSet",
+ description=dedent(
+ """
+ Sets can only be updated if the dataset is in the **open** state.
+
+ Requires **contributor** access to the dataset's corpus.
+ """
+ )
+ ),
+ patch=extend_schema(
+ operation_id="PartialUpdateDatasetSet",
+ description=dedent(
+ """
+ Sets can only be updated if the dataset is in the **open** state.
+
+ Requires **contributor** access to the dataset's corpus.
+ """
+ )
+ )
+)
+class DatasetSets(DatasetSetBase, RetrieveUpdateDestroyAPIView):
+ lookup_url_kwarg = "set"
+ queryset = DatasetSet.objects.none()
+
+ def get_queryset(self):
+ return self.dataset.sets.all()
+
+ def check_object_permissions(self, request, obj):
+ super().check_object_permissions(request, obj)
+ if request.method == "DELETE":
+ if ProcessDatasetSet.objects.filter(set_id=obj.id).exists():
+ raise ValidationError("This dataset set is selected in a process and cannot be deleted.")
+ if self.dataset.sets.count() == 1:
+ raise ValidationError("This dataset set is the only one in its dataset and cannot be deleted.")
+
+ def destroy(self, request, *args, **kwargs):
+ instance = self.get_object()
+ # Delete the dataset set's DatasetElement objects
+ DatasetElement.objects.filter(set_id=instance.id).delete()
+ instance.delete()
+ return Response(status=status.HTTP_204_NO_CONTENT)
+
+
@extend_schema(tags=["datasets"])
@extend_schema_view(
get=extend_schema(
diff --git a/arkindex/training/serializers.py b/arkindex/training/serializers.py
index f6b3831216..ce877bff5f 100644
--- a/arkindex/training/serializers.py
+++ b/arkindex/training/serializers.py
@@ -482,9 +482,20 @@ class MetricValueBulkSerializer(serializers.Serializer):
class DatasetSetSerializer(serializers.ModelSerializer):
+ dataset = serializers.HiddenField(
+ default=_dataset_from_context,
+ write_only=True,
+ )
+
class Meta:
model = DatasetSet
- fields = ("id", "name",)
+ fields = ("id", "name", "dataset", )
+ read_only_fields = ("id", )
+
+ def validate_name(self, name):
+ if any(ds.name == name for ds in self.context["dataset"].sets.all()):
+ raise ValidationError("A set with this name already exists in this dataset.")
+ return name
class DatasetSerializer(serializers.ModelSerializer):
diff --git a/arkindex/training/tests/test_dataset_sets_api.py b/arkindex/training/tests/test_dataset_sets_api.py
new file mode 100644
index 0000000000..17f43037e4
--- /dev/null
+++ b/arkindex/training/tests/test_dataset_sets_api.py
@@ -0,0 +1,547 @@
+from unittest.mock import call, patch
+
+from django.urls import reverse
+from rest_framework import status
+
+from arkindex.documents.models import Corpus
+from arkindex.process.models import Process, ProcessMode
+from arkindex.project.tests import FixtureAPITestCase
+from arkindex.training.models import Dataset, DatasetSet, DatasetState
+from arkindex.users.models import Role, User
+
+
+class TestDatasetSetsAPI(FixtureAPITestCase):
+ """
+ Test dataset sets API endpoints
+ """
+ @classmethod
+ def setUpTestData(cls):
+ super().setUpTestData()
+ cls.page = cls.corpus.elements.get(name="Volume 1, page 1r")
+ cls.line = cls.corpus.elements.filter(type__slug="text_line").first()
+ cls.dataset = Dataset.objects.first()
+ private_user = User.objects.create(email="ikarigendou@mail.com", display_name="Commander Ikari", verified_email=True)
+ cls.private_corpus = Corpus.objects.create(name="Private Project")
+ cls.private_dataset = Dataset.objects.create(corpus=cls.private_corpus, name="Private Dataset", creator=private_user)
+ cls.train_set = cls.dataset.sets.get(name="training")
+ cls.train_set.set_elements.create(element=cls.page)
+ cls.train_set.set_elements.create(element=cls.line)
+ cls.dataset_process = Process.objects.create(
+ creator=cls.user,
+ corpus=cls.corpus,
+ mode=ProcessMode.Files,
+ )
+ cls.dataset_process.process_sets.create(set=cls.train_set)
+
+ # CreateDatasetSet
+
+ def test_create_requires_login(self):
+ with self.assertNumQueries(0):
+ response = self.client.post(
+ reverse("api:dataset-sets", kwargs={"pk": self.dataset.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+ self.assertDictEqual(response.json(), {"detail": "Authentication credentials were not provided."})
+
+ def test_create_requires_verified(self):
+ user = User.objects.create(email="not_verified@mail.com", display_name="Not Verified", verified_email=False)
+ self.client.force_login(user)
+ with self.assertNumQueries(2):
+ response = self.client.post(
+ reverse("api:dataset-sets", kwargs={"pk": self.dataset.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ 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_create_dataset_doesnt_exist(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.post(
+ reverse("api:dataset-sets", kwargs={"pk": "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ @patch("arkindex.users.managers.BaseACLManager.filter_rights")
+ def test_create_no_read_access(self, filter_rights_mock):
+ filter_rights_mock.return_value = Corpus.objects.filter(pk=self.corpus.pk)
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.post(
+ reverse("api:dataset-sets", kwargs={"pk": str(self.private_dataset.pk)}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ self.assertEqual(filter_rights_mock.call_count, 1)
+ self.assertEqual(filter_rights_mock.call_args, call(self.user, Corpus, Role.Guest.value))
+
+ @patch("arkindex.users.managers.BaseACLManager.filter_rights")
+ def test_create_requires_contributor_access(self, filter_rights_mock):
+ filter_rights_mock.side_effect = [Corpus.objects.all(), Corpus.objects.filter(pk=self.corpus.pk)]
+ self.private_corpus.memberships.update_or_create(user=self.user, defaults={"level": Role.Guest.value})
+ self.client.force_login(self.user)
+ with self.assertNumQueries(5):
+ response = self.client.post(
+ reverse("api:dataset-sets", kwargs={"pk": self.private_dataset.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+
+ self.assertDictEqual(response.json(), {"detail": "You do not have contributor access to this dataset."})
+
+ self.assertEqual(filter_rights_mock.call_count, 2)
+ self.assertEqual(filter_rights_mock.call_args_list, [call(self.user, Corpus, Role.Guest.value), call(self.user, Corpus, Role.Contributor.value)])
+
+ def test_create_requires_open_dataset(self):
+ self.client.force_login(self.user)
+
+ for state in set(DatasetState) - {DatasetState.Open}:
+ with self.subTest(state=state):
+ self.dataset.state = state
+ self.dataset.save()
+
+ with self.assertNumQueries(5):
+ response = self.client.post(
+ reverse("api:dataset-sets", kwargs={"pk": self.dataset.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+
+ self.assertEqual(response.json(), ["You can only add or update sets from a dataset in an open state."])
+
+ def test_create_duplicate_set_name(self):
+ self.client.force_login(self.user)
+
+ with self.assertNumQueries(5):
+ response = self.client.post(
+ reverse("api:dataset-sets", kwargs={"pk": self.dataset.pk}),
+ data={"name": "training"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+
+ self.assertDictEqual(response.json(), {"name": ["A set with this name already exists in this dataset."]})
+
+ def test_create(self):
+ self.client.force_login(self.user)
+
+ with self.assertNumQueries(6):
+ response = self.client.post(
+ reverse("api:dataset-sets", kwargs={"pk": self.dataset.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_201_CREATED)
+
+ created_set = DatasetSet.objects.get(dataset=self.dataset, name="Unit-00")
+ self.assertDictEqual(response.json(), {
+ "id": str(created_set.id),
+ "name": "Unit-00"
+ })
+
+ # UpdateDatasetSet
+
+ def test_update_requires_login(self):
+ with self.assertNumQueries(0):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+ self.assertDictEqual(response.json(), {"detail": "Authentication credentials were not provided."})
+
+ def test_update_requires_verified(self):
+ user = User.objects.create(email="not_verified@mail.com", display_name="Not Verified", verified_email=False)
+ self.client.force_login(user)
+ with self.assertNumQueries(2):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ 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_update_dataset_doesnt_exist(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa", "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ def test_update_set_doesnt_exist(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(6):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.id, "set": "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ @patch("arkindex.users.managers.BaseACLManager.filter_rights")
+ def test_update_no_read_access(self, filter_rights_mock):
+ filter_rights_mock.return_value = Corpus.objects.filter(pk=self.private_corpus.pk)
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ self.assertEqual(filter_rights_mock.call_count, 1)
+ self.assertEqual(filter_rights_mock.call_args, call(self.user, Corpus, Role.Guest.value))
+
+ @patch("arkindex.users.managers.BaseACLManager.filter_rights")
+ def test_update_requires_contributor_access(self, filter_rights_mock):
+ filter_rights_mock.side_effect = [Corpus.objects.all(), Corpus.objects.filter(pk=self.private_corpus.pk)]
+ self.private_corpus.memberships.update_or_create(user=self.user, defaults={"level": Role.Guest.value})
+ self.client.force_login(self.user)
+ with self.assertNumQueries(5):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+
+ self.assertDictEqual(response.json(), {"detail": "You do not have contributor access to this dataset."})
+
+ self.assertEqual(filter_rights_mock.call_count, 2)
+ self.assertEqual(filter_rights_mock.call_args_list, [call(self.user, Corpus, Role.Guest.value), call(self.user, Corpus, Role.Contributor.value)])
+
+ def test_update_requires_open_dataset(self):
+ self.client.force_login(self.user)
+
+ for state in set(DatasetState) - {DatasetState.Open}:
+ with self.subTest(state=state):
+ self.dataset.state = state
+ self.dataset.save()
+
+ with self.assertNumQueries(5):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+
+ self.assertEqual(response.json(), ["You can only add or update sets from a dataset in an open state."])
+
+ def test_update_duplicate_set_name(self):
+ self.client.force_login(self.user)
+
+ with self.assertNumQueries(6):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "validation"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+
+ self.assertDictEqual(response.json(), {"name": ["A set with this name already exists in this dataset."]})
+
+ def test_update(self):
+ self.client.force_login(self.user)
+
+ with self.assertNumQueries(7):
+ response = self.client.put(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_200_OK)
+
+ self.assertDictEqual(response.json(), {
+ "id": str(self.train_set.id),
+ "name": "Unit-00"
+ })
+ self.train_set.refresh_from_db()
+ self.assertEqual(self.train_set.name, "Unit-00")
+
+ # PartialUpdateDatasetSet
+
+ def test_partial_update_requires_login(self):
+ with self.assertNumQueries(0):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+ self.assertDictEqual(response.json(), {"detail": "Authentication credentials were not provided."})
+
+ def test_partial_update_requires_verified(self):
+ user = User.objects.create(email="not_verified@mail.com", display_name="Not Verified", verified_email=False)
+ self.client.force_login(user)
+ with self.assertNumQueries(2):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ 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_partial_update_dataset_doesnt_exist(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa", "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ def test_partial_update_set_doesnt_exist(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(6):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.id, "set": "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ @patch("arkindex.users.managers.BaseACLManager.filter_rights")
+ def test_partial_update_no_read_access(self, filter_rights_mock):
+ filter_rights_mock.return_value = Corpus.objects.filter(pk=self.private_corpus.pk)
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ self.assertEqual(filter_rights_mock.call_count, 1)
+ self.assertEqual(filter_rights_mock.call_args, call(self.user, Corpus, Role.Guest.value))
+
+ @patch("arkindex.users.managers.BaseACLManager.filter_rights")
+ def test_partial_update_requires_contributor_access(self, filter_rights_mock):
+ filter_rights_mock.side_effect = [Corpus.objects.all(), Corpus.objects.filter(pk=self.private_corpus.pk)]
+ self.private_corpus.memberships.update_or_create(user=self.user, defaults={"level": Role.Guest.value})
+ self.client.force_login(self.user)
+ with self.assertNumQueries(5):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+
+ self.assertDictEqual(response.json(), {"detail": "You do not have contributor access to this dataset."})
+
+ self.assertEqual(filter_rights_mock.call_count, 2)
+ self.assertEqual(filter_rights_mock.call_args_list, [call(self.user, Corpus, Role.Guest.value), call(self.user, Corpus, Role.Contributor.value)])
+
+ def test_partial_update_requires_open_dataset(self):
+ self.client.force_login(self.user)
+
+ for state in set(DatasetState) - {DatasetState.Open}:
+ with self.subTest(state=state):
+ self.dataset.state = state
+ self.dataset.save()
+
+ with self.assertNumQueries(5):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+
+ self.assertEqual(response.json(), ["You can only add or update sets from a dataset in an open state."])
+
+ def test_partial_update_duplicate_set_name(self):
+ self.client.force_login(self.user)
+
+ with self.assertNumQueries(6):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "validation"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+
+ self.assertDictEqual(response.json(), {"name": ["A set with this name already exists in this dataset."]})
+
+ def test_partial_update(self):
+ self.client.force_login(self.user)
+
+ with self.assertNumQueries(7):
+ response = self.client.patch(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ data={"name": "Unit-00"},
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_200_OK)
+
+ self.assertDictEqual(response.json(), {
+ "id": str(self.train_set.id),
+ "name": "Unit-00"
+ })
+ self.train_set.refresh_from_db()
+ self.assertEqual(self.train_set.name, "Unit-00")
+
+ # DestroyDatasetSet
+
+ def test_destroy_requires_login(self):
+ with self.assertNumQueries(0):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+ self.assertDictEqual(response.json(), {"detail": "Authentication credentials were not provided."})
+
+ def test_destroy_requires_verified(self):
+ user = User.objects.create(email="not_verified@mail.com", display_name="Not Verified", verified_email=False)
+ self.client.force_login(user)
+ with self.assertNumQueries(2):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ format="json"
+ )
+ 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_destroy_dataset_doesnt_exist(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa", "set": self.train_set.pk}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ def test_destroy_set_doesnt_exist(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(6):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.id, "set": "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ @patch("arkindex.users.managers.BaseACLManager.filter_rights")
+ def test_destroy_no_read_access(self, filter_rights_mock):
+ filter_rights_mock.return_value = Corpus.objects.filter(pk=self.private_corpus.pk)
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+ self.assertDictEqual(response.json(), {"detail": "Not found."})
+
+ self.assertEqual(filter_rights_mock.call_count, 1)
+ self.assertEqual(filter_rights_mock.call_args, call(self.user, Corpus, Role.Guest.value))
+
+ @patch("arkindex.users.managers.BaseACLManager.filter_rights")
+ def test_destroy_requires_admin_access(self, filter_rights_mock):
+ for role in [Role.Guest, Role.Contributor]:
+ filter_rights_mock.side_effect = [Corpus.objects.all(), Corpus.objects.filter(pk=self.private_corpus.pk)]
+ with self.subTest(role=role):
+ filter_rights_mock.reset_mock()
+ self.private_corpus.memberships.update_or_create(user=self.user, defaults={"level": role.value})
+ self.client.force_login(self.user)
+ with self.assertNumQueries(5):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+
+ self.assertDictEqual(response.json(), {"detail": "You do not have admin access to this dataset."})
+
+ self.assertEqual(filter_rights_mock.call_count, 2)
+ self.assertEqual(filter_rights_mock.call_args_list, [call(self.user, Corpus, Role.Guest.value), call(self.user, Corpus, Role.Admin.value)])
+
+ def test_destroy_requires_open_dataset(self):
+ self.client.force_login(self.user)
+
+ for state in set(DatasetState) - {DatasetState.Open}:
+ with self.subTest(state=state):
+ self.dataset.state = state
+ self.dataset.save()
+
+ with self.assertNumQueries(5):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+
+ self.assertEqual(response.json(), ["You can only add or update sets from a dataset in an open state."])
+
+ def test_destroy_set_in_process_forbidden(self):
+ self.client.force_login(self.user)
+
+ with self.assertNumQueries(7):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+ self.assertEqual(response.json(), ["This dataset set is selected in a process and cannot be deleted."])
+
+ # Check that self.train_set still exists
+ self.train_set.refresh_from_db()
+
+ def test_destroy_only_set_forbidden(self):
+ self.client.force_login(self.user)
+ test_dataset = self.corpus.datasets.create(name="Tokyo 3", description="第３新æ±äº¬å¸‚", creator=self.user)
+ test_set = test_dataset.sets.create(name="NERV HQ")
+
+ with self.assertNumQueries(7):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": test_dataset.pk, "set": test_set.pk}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+ self.assertEqual(response.json(), ["This dataset set is the only one in its dataset and cannot be deleted."])
+
+ # Check that self.train_set still exists
+ test_set.refresh_from_db()
+
+ def test_destroy(self):
+ self.client.force_login(self.user)
+ # Remove train_set from dataset_process
+ self.dataset_process.process_sets.all().delete()
+ with self.assertNumQueries(9):
+ response = self.client.delete(
+ reverse("api:dataset-set", kwargs={"dataset": self.dataset.pk, "set": self.train_set.pk}),
+ format="json"
+ )
+ self.assertEqual(response.status_code, status.HTTP_204_NO_CONTENT)
+
+ with self.assertRaises(DatasetSet.DoesNotExist):
+ self.train_set.refresh_from_db()
--
GitLab