.pre-commit-config.yaml

 repos:
-  - repo: https://github.com/pre-commit/mirrors-isort
-    rev: v5.10.1
-    hooks:
-      - id: isort
-  - repo: https://github.com/pycqa/flake8
-    rev: 3.9.2
-    hooks:
-      - id: flake8
-        additional_dependencies:
-          - 'flake8-copyright==0.2.2'
-          - 'flake8-debugger==3.1.0'
-          - 'flake8-quotes==3.3.2'
   - repo: https://github.com/astral-sh/ruff-pre-commit
     # Ruff version.
     rev: v0.3.7

arkindex/documents/api/ml.py

@@ -440,7 +440,7 @@ class ManageClassificationsSelection(SelectionMixin, CorpusACLMixin, CreateAPIVi
         mode = serializer.validated_data["mode"]
         if mode == ClassificationMode.Create:
             return self.create(corpus, request, *args, **kwargs)
-        elif mode == ClassificationMode.Validate:
+        if mode == ClassificationMode.Validate:
             elements = self.get_selection(corpus.id)
             Classification.objects.filter(
                 element__in=elements,

arkindex/documents/dates.py

@@ -45,7 +45,7 @@ class InterpretedDate(object):
         """
         if self.month and self.day:
             return DatePrecision.Day
-        elif self.month:
+        if self.month:
             return DatePrecision.Month
         return DatePrecision.Year
 

arkindex/documents/serializers/elements.py

@@ -667,8 +667,7 @@ class ElementSerializer(ElementTinySerializer):
 
             validated_data.update(image=image, polygon=polygon)
 
-        instance = super().update(instance, validated_data)
-        return instance
+        return super().update(instance, validated_data)
 
 
 class ElementNeighborsSerializer(serializers.ModelSerializer):

arkindex/documents/serializers/entities.py

@@ -32,8 +32,7 @@ class EntityTypeSerializer(serializers.ModelSerializer):
         corpus = data.get("corpus")
         if self.instance and corpus:
             raise ValidationError({"corpus": ["It is not possible to update an Entity Type's corpus."]})
-        data = super().to_internal_value(data)
-        return data
+        return super().to_internal_value(data)
 
     def validate(self, data):
         name = data.get("name") if "name" in data else self.instance.name

arkindex/documents/tasks.py

@@ -21,7 +21,7 @@ from arkindex.documents.models import (
     Transcription,
     TranscriptionEntity,
 )
-from arkindex.ponos.models import Task
+from arkindex.ponos.models import Artifact, Task
 from arkindex.process.models import Process, ProcessDatasetSet, ProcessElement, WorkerActivity, WorkerRun
 from arkindex.training.models import DatasetElement, DatasetSet
 from arkindex.users.models import User
@@ -41,6 +41,10 @@ def corpus_delete(corpus_id: str) -> None:
         corpus.top_level_type_id = None
         corpus.save(update_fields=["top_level_type_id"])
 
+    # Set process.element to NULL in all processes, to avoid IntegrityErrors when deleting elements, as
+    # elements are deleted before processes.
+    Process.objects.filter(corpus_id=corpus_id).exclude(element_id=None).update(element_id=None)
+
     # Delete all related objects, bypassing RESTRICTs deletion rules
     # and Django's way of loading everything into memory before deleting.
     querysets = [
@@ -67,6 +71,8 @@ def corpus_delete(corpus_id: str) -> None:
         Selection.objects.filter(element__corpus_id=corpus_id),
         corpus.memberships.all(),
         corpus.exports.all(),
+        # Delete task artifacts
+        Artifact.objects.filter(task__process__corpus_id=corpus_id),
         # ProcessDatasetSet M2M
         ProcessDatasetSet.objects.filter(set__dataset__corpus_id=corpus_id),
         ProcessDatasetSet.objects.filter(process__corpus_id=corpus_id),

arkindex/documents/tests/test_edit_elementpath.py

@@ -363,7 +363,7 @@ class TestEditElementPath(FixtureTestCase):
                 path_id = elements["B"].paths.get().id
                 if path1.id == path_id:
                     return str(path1.path[0])
-                elif path2.id == path_id:
+                if path2.id == path_id:
                     return str(path2.path[0])
                 raise AssertionError("Unexpected top-level path ID")
 

arkindex/images/managers.py

@@ -47,7 +47,7 @@ class ImageServerManager(models.Manager):
                 f'The URL "{url}" does not match any existing image server. '
                 "Please ask an instance administrator to register the IIIF server for this image."
             )
-        elif servers_count > 1:
+        if servers_count > 1:
             raise self.model.MultipleObjectsReturned(
                 f'The URL "{url}" matched multiple existing image servers'
             )

arkindex/images/models.py

@@ -108,7 +108,7 @@ class ImageServer(models.Model):
         src_url, dest_url = map(urllib.parse.urlsplit, (self.url, other.url))
         if src_url.scheme != dest_url.scheme:
             raise ValueError("Cannot merge into an image server of a different protocol")
-        elif src_url.netloc != dest_url.netloc:
+        if src_url.netloc != dest_url.netloc:
             raise ValueError("Cannot merge into an image server on a different domain")
 
         # Check paths

arkindex/images/serializers.py

@@ -68,7 +68,7 @@ class ImageSerializer(serializers.ModelSerializer):
         if not self.instance:
             # Force the Unchecked status when creating a new image
             return S3FileStatus.Unchecked
-        elif value == S3FileStatus.Checked:
+        if value == S3FileStatus.Checked:
             # Perform image validation if we are updating an existing image to Checked
             try:
                 if self.instance.server.s3_bucket and not self.instance.server.s3_read_only_bucket:

arkindex/process/api.py

@@ -402,8 +402,9 @@ class ProcessDetails(ProcessACLMixin, ProcessQuerysetMixin, RetrieveUpdateDestro
 
 
 class ProcessRetry(ProcessACLMixin, ProcessQuerysetMixin, GenericAPIView):
-    """
-    Retry a process. Can only be used on processes with Error, Failed, Stopped or Completed states.\n\n
+    r"""
+    Retry a process. Can only be used on processes with Error, Failed, Stopped or Completed states.
+
     Requires an **admin** access to the process and **guest** access to the process' farm.
     """
     permission_classes = (IsVerified, )
@@ -427,9 +428,9 @@ class ProcessRetry(ProcessACLMixin, ProcessQuerysetMixin, GenericAPIView):
         # Allow 'retrying' a process that has no Ponos tasks (that has never been started)
         if len(process.tasks.all()) and state in (State.Unscheduled, State.Pending):
             raise ValidationError({"__all__": ["This process is already pending"]})
-        elif state == State.Running:
+        if state == State.Running:
             raise ValidationError({"__all__": ["This process is already running"]})
-        elif state == State.Stopping:
+        if state == State.Stopping:
             raise ValidationError({"__all__": ["This process is stopping"]})
 
     @extend_schema(
@@ -476,8 +477,9 @@ class FilesProcess(CreateAPIView):
     )
 )
 class CorpusProcess(SelectionMixin, CorpusACLMixin, CreateAPIView):
-    """
-    Create a distributed process from elements of an Arkindex corpus.\n\n
+    r"""
+    Create a distributed process from elements of an Arkindex corpus.
+
     Requires an **admin** access to the corpus.
     """
     permission_classes = (IsVerified, )
@@ -542,8 +544,9 @@ class CorpusProcess(SelectionMixin, CorpusACLMixin, CreateAPIView):
     ),
 )
 class StartProcess(CorpusACLMixin, CreateAPIView):
-    """
-    Start a process, used to build a Workflow with Workers.\n\n
+    r"""
+    Start a process, used to build a Workflow with Workers.
+
     Requires an **admin** access to the corpus of this process.
     """
     permission_classes = (IsVerified, )
@@ -1555,8 +1558,9 @@ class WorkerRunDetails(ProcessACLMixin, RetrieveUpdateDestroyAPIView):
     tags=["process"]
 ))
 class ListProcessElements(CorpusACLMixin, ListAPIView):
-    """
-    List all elements for a process with workers.\n\n
+    r"""
+    List all elements for a process with workers.
+
     Requires an **admin** access to the process corpus.
     """
     pagination_class = CountCursorPagination
@@ -1579,8 +1583,7 @@ class ListProcessElements(CorpusACLMixin, ListAPIView):
     def get_serializer_class(self):
         if self.with_image:
             return ProcessElementSerializer
-        else:
-            return ProcessElementLightSerializer
+        return ProcessElementLightSerializer
 
     def get_queryset(self):
         if not self.has_admin_access(self.process.corpus):
@@ -1829,8 +1832,9 @@ class WorkerActivityBase(ListAPIView):
     )
 )
 class CorpusWorkersActivity(CorpusACLMixin, WorkerActivityBase):
-    """
-    Retrieve corpus wise statistics about the activity of all its worker processes.\n
+    r"""
+    Retrieve corpus wise statistics about the activity of all its worker processes.
+
     Requires a **guest** access.
     """
 
@@ -1846,8 +1850,9 @@ class CorpusWorkersActivity(CorpusACLMixin, WorkerActivityBase):
     )
 )
 class ProcessWorkersActivity(ProcessACLMixin, WorkerActivityBase):
-    """
-    Retrieve process statistics about the activity of its workers.\n
+    r"""
+    Retrieve process statistics about the activity of its workers.
+
     Requires a **guest** access.
     """
 

arkindex/process/models.py

@@ -232,8 +232,7 @@ class Process(IndexableModel):
         """
         if self.has_prefetched_tasks:
             return max((t.expiry for t in self.tasks.all()), default=None)
-        else:
-            return self.tasks.aggregate(models.Max("expiry"))["expiry__max"]
+        return self.tasks.aggregate(models.Max("expiry"))["expiry__max"]
 
     def get_last_run(self) -> int:
         """

arkindex/process/serializers/imports.py

@@ -545,7 +545,7 @@ class ApplyProcessTemplateSerializer(ProcessACLMixin, serializers.Serializer):
         access_level = self.process_access_level(process)
         if not access_level:
             raise ValidationError(detail="Process with this ID does not exist.")
-        elif access_level < Role.Contributor.value:
+        if access_level < Role.Contributor.value:
             raise PermissionDenied(detail="You do not have a contributor access to this process.")
 
         errors = []
@@ -627,7 +627,7 @@ class CorpusProcessSerializer(serializers.Serializer):
         access_level = get_max_level(self.context["request"].user, corpus)
         if not access_level:
             raise ValidationError(["Corpus with this ID does not exist."])
-        elif access_level < Role.Admin.value:
+        if access_level < Role.Admin.value:
             raise ValidationError(["You do not have an admin access to this corpus."])
         return corpus
 

arkindex/project/argparse.py

@@ -37,8 +37,7 @@ class ModelArgument(object):
         text_filter = {f"{self.text_search_field}__{self.text_search_lookup}": arg}
         if self.many:
             return qs.filter(**text_filter)
-        else:
-            return qs.get(**text_filter)
+        return qs.get(**text_filter)
 
 
 class CorpusArgument(ModelArgument):

arkindex/project/serializer_fields.py

@@ -304,4 +304,4 @@ class NullField(serializers.CharField):
     def to_internal_value(self, data):
         if data is not None:
             self.fail("invalid")
-        return None
+        return

arkindex/project/tests/test_config.py

@@ -26,7 +26,7 @@ class TestConfig(TestCase):
                 data = data.value
             if data is None:
                 return self.represent_none(data)
-            elif isinstance(data, (bool, int, float, bytes, str)):
+            if isinstance(data, (bool, int, float, bytes, str)):
                 return self.represent_data(data)
             return self.represent_str(str(data))
 

arkindex/sql_validation/corpus_delete.sql

@@ -10,6 +10,11 @@ FROM "documents_corpus"
 WHERE "documents_corpus"."id" = '{corpus_id}'::uuid
 LIMIT 21;
 
+UPDATE "process_process"
+SET "element_id" = NULL
+WHERE ("process_process"."corpus_id" = '{corpus_id}'::uuid
+       AND NOT ("process_process"."element_id" IS NULL));
+
 SELECT "django_content_type"."id",
        "django_content_type"."app_label",
        "django_content_type"."model"
@@ -152,6 +157,15 @@ DELETE
 FROM "documents_corpusexport"
 WHERE "documents_corpusexport"."corpus_id" = '{corpus_id}'::uuid;
 
+DELETE
+FROM "ponos_artifact"
+WHERE "ponos_artifact"."id" IN
+        (SELECT U0."id"
+         FROM "ponos_artifact" U0
+         INNER JOIN "ponos_task" U1 ON (U0."task_id" = U1."id")
+         INNER JOIN "process_process" U2 ON (U1."process_id" = U2."id")
+         WHERE U2."corpus_id" = '{corpus_id}'::uuid);
+
 DELETE
 FROM "process_processdatasetset"
 WHERE "process_processdatasetset"."id" IN

arkindex/sql_validation/corpus_delete_top_level_type.sql

@@ -14,6 +14,11 @@ UPDATE "documents_corpus"
 SET "top_level_type_id" = NULL
 WHERE "documents_corpus"."id" = '{corpus_id}'::uuid;
 
+UPDATE "process_process"
+SET "element_id" = NULL
+WHERE ("process_process"."corpus_id" = '{corpus_id}'::uuid
+       AND NOT ("process_process"."element_id" IS NULL));
+
 SELECT "django_content_type"."id",
        "django_content_type"."app_label",
        "django_content_type"."model"
@@ -156,6 +161,15 @@ DELETE
 FROM "documents_corpusexport"
 WHERE "documents_corpusexport"."corpus_id" = '{corpus_id}'::uuid;
 
+DELETE
+FROM "ponos_artifact"
+WHERE "ponos_artifact"."id" IN
+        (SELECT U0."id"
+         FROM "ponos_artifact" U0
+         INNER JOIN "ponos_task" U1 ON (U0."task_id" = U1."id")
+         INNER JOIN "process_process" U2 ON (U1."process_id" = U2."id")
+         WHERE U2."corpus_id" = '{corpus_id}'::uuid);
+
 DELETE
 FROM "process_processdatasetset"
 WHERE "process_processdatasetset"."id" IN

arkindex/training/admin.py

@@ -48,8 +48,7 @@ class DatasetAdmin(admin.ModelAdmin):
         self.readonly_fields = self.__class__.readonly_fields
         if obj is not None:
             self.readonly_fields += ("unique_elements",)
-        form = super().get_form(request, obj=None, **kwargs)
-        return form
+        return super().get_form(request, obj=None, **kwargs)
 
 
 admin.site.register(Model, ModelAdmin)

arkindex/training/api.py

@@ -417,8 +417,7 @@ class ModelsList(TrainingModelMixin, ListCreateAPIView):
                     "id": str(existing_model.id),
                     "name": "A model with this name already exists",
                 })
-            else:
-                raise PermissionDenied()
+            raise PermissionDenied()
         return serializer.save()
 
 
@@ -753,7 +752,7 @@ class DatasetSetBase():
         )
         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():
+        if 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.")