Dataset sets management API endpoints

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"),

arkindex/training/api.py

@@ -42,6 +42,7 @@ from arkindex.training.serializers import (
     DatasetElementInfoSerializer,
     DatasetElementSerializer,
     DatasetSerializer,
+    DatasetSetSerializer,
     ElementDatasetSetSerializer,
     MetricValueBulkSerializer,
     MetricValueCreateSerializer,
@@ -747,6 +748,110 @@ 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 dataset.corpus not in Corpus.objects.admin(self.request.user):
+            raise PermissionDenied(detail="You do not have admin 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 **admin** 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 **admin** 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 **admin** 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(

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):