Dataset sets management API endpoints

closes #1720 (closed)

requires !2264 (merged)

EDIT: contrary to what it says in the issue, the endpoints require contributor access to the dataset's corpus, excepted for DELETE which requires admin access.

TODO

• Update Dataset endpoints documentation (no more sets editing there)

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(