Deleted assets are now no longer included in responses of other assets' relationships (#646)

* Do not serialize direct relationships to soft-deleted assets

* Do not materialiaze soft deleted assets in server responses

* Add a migration script

Author: PGijsbers

alembic/alembic/versions/8f9ac801a283_cascade_delete_ai_resource.py

@@ -0,0 +1,57 @@
+"""cascade delete ai resource
+
+Revision ID: 8f9ac801a283
+Revises: 95fa6a3c7eee
+Create Date: 2025-10-30 14:10:21.536071
+
+"""
+
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = "8f9ac801a283"
+down_revision: Union[str, None] = "95fa6a3c7eee"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def recreate_constraints(with_on_delete_cascade: bool):
+    for table in ["part", "relevant"]:
+        op.drop_constraint(
+            f"ai_resource_{table}_link_ibfk_1", f"ai_resource_{table}_link", type_="foreignkey"
+        )
+        op.drop_constraint(
+            f"ai_resource_{table}_link_ibfk_2", f"ai_resource_{table}_link", type_="foreignkey"
+        )
+
+        op.create_foreign_key(
+            f"ai_resource_{table}_link_ibfk_1",
+            f"ai_resource_{table}_link",
+            "ai_resource",
+            ["parent_identifier"],
+            ["identifier"],
+            onupdate="CASCADE",
+            ondelete="CASCADE" if with_on_delete_cascade else None,
+        )
+        local_col = "child_identifier" if table == "part" else "relevant_identifier"
+        op.create_foreign_key(
+            f"ai_resource_{table}_link_ibfk_2",
+            f"ai_resource_{table}_link",
+            "ai_resource",
+            [local_col],
+            ["identifier"],
+            onupdate="CASCADE",
+            ondelete="CASCADE" if with_on_delete_cascade else None,
+        )
+
+
+def upgrade():
+    recreate_constraints(with_on_delete_cascade=True)
+
+
+def downgrade():
+    recreate_constraints(with_on_delete_cascade=False)

alembic/alembic/versions/95fa6a3c7eee_extend_registration_link.py

@@ -1,7 +1,8 @@
 """extend registration link
 
 Revision ID: 95fa6a3c7eee
-Revises: c7b4844dee44
+
+Revises: 79b2dda7e3be
 Create Date: 2025-10-30 08:30:38.564333
 
 """
@@ -14,7 +15,7 @@
 
 # revision identifiers, used by Alembic.
 revision: str = "95fa6a3c7eee"
-down_revision: Union[str, None] = "c7b4844dee44"
+down_revision: Union[str, None] = "79b2dda7e3be"
 branch_labels: Union[str, Sequence[str], None] = None
 depends_on: Union[str, Sequence[str], None] = None
 

src/database/model/ai_resource/resource_table.py

@@ -13,13 +13,17 @@ class AIResourcePartLink(SQLModel, table=True):  # type: ignore [call-arg]
     __tablename__ = "ai_resource_part_link"
     parent_identifier: str = Field(
         max_length=IDENTIFIER_LENGTH,
-        sa_column_args=[ForeignKey("ai_resource.identifier", onupdate="CASCADE")],
+        sa_column_args=[
+            ForeignKey("ai_resource.identifier", onupdate="CASCADE", ondelete="CASCADE")
+        ],
         sa_column_kwargs=dict(nullable=True, index=True),
         primary_key=True,
     )
     child_identifier: str = Field(
         max_length=IDENTIFIER_LENGTH,
-        sa_column_args=[ForeignKey("ai_resource.identifier", onupdate="CASCADE")],
+        sa_column_args=[
+            ForeignKey("ai_resource.identifier", onupdate="CASCADE", ondelete="CASCADE")
+        ],
         sa_column_kwargs=dict(nullable=True, index=True),
         primary_key=True,
     )
@@ -29,13 +33,17 @@ class AIResourceRelevantLink(SQLModel, table=True):  # type: ignore [call-arg]
     __tablename__ = "ai_resource_relevant_link"
     parent_identifier: str = Field(
         max_length=IDENTIFIER_LENGTH,
-        sa_column_args=[ForeignKey("ai_resource.identifier", onupdate="CASCADE")],
+        sa_column_args=[
+            ForeignKey("ai_resource.identifier", onupdate="CASCADE", ondelete="CASCADE")
+        ],
         sa_column_kwargs=dict(nullable=True, index=True),
         primary_key=True,
     )
     relevant_identifier: str = Field(
         max_length=IDENTIFIER_LENGTH,
-        sa_column_args=[ForeignKey("ai_resource.identifier", onupdate="CASCADE")],
+        sa_column_args=[
+            ForeignKey("ai_resource.identifier", onupdate="CASCADE", ondelete="CASCADE")
+        ],
         sa_column_kwargs=dict(nullable=True, index=True),
         primary_key=True,
     )

src/database/model/helper_functions.py

@@ -1,12 +1,16 @@
+from functools import cache
+from http import HTTPStatus
 from typing import Type, TYPE_CHECKING
 
+from fastapi import HTTPException
 from pydantic import create_model
 from sqlalchemy import Column, Integer, ForeignKey, String
 from sqlmodel import SQLModel, Field
 
 from database.model.field_length import IDENTIFIER_LENGTH
 
 if TYPE_CHECKING:
+    from database.model.concept.concept import AIoDConcept
     from database.model.relationships import _ResourceRelationship
 
 
@@ -84,3 +88,27 @@ def non_abstract_subclasses(cls):
             yield grand_child
         if not has_grandchild:
             yield child
+
+
+@cache
+def get_asset_type_by_abbreviation() -> dict[str, type["AIoDConcept"]]:
+    from database.model.concept.concept import AIoDConcept
+
+    return {
+        cls.__abbreviation__: cls
+        for cls in non_abstract_subclasses(AIoDConcept)
+        if hasattr(cls, "__abbreviation__")
+    }
+
+
+def get_asset_by_identifier(identifier, session):
+    asset_type_map = get_asset_type_by_abbreviation()
+    prefix = identifier.split("_")[0]
+    model_class = asset_type_map.get(prefix)
+    if not model_class:
+        raise HTTPException(
+            status_code=HTTPStatus.NOT_FOUND,
+            detail=f"Unknown asset type with identifier '{identifier}'",
+        )
+    resource = session.get(model_class, identifier)
+    return model_class, resource

src/database/model/relationships.py

@@ -194,7 +194,7 @@ def create_triggers(self, parent_class: Type[SQLModel], field_name: str):
 @dataclasses.dataclass
 class ManyToMany(_ResourceRelationshipList):
     """
-    Configuration for handling many-to-one relationships to another table.
+    Configuration for handling many-to-many relationships to another table.
 
     Args:
         on_delete_trigger_orphan_deletion(Callable): automatically delete orphans of the

src/database/model/serializers.py

@@ -8,8 +8,11 @@
 from starlette.status import HTTP_404_NOT_FOUND
 
 from authentication import KeycloakUser
-from database.model.helper_functions import get_relationships
+from database.model.helper_functions import get_relationships, get_asset_by_identifier
 from database.model.named_relation import NamedRelation, Taxonomy
+from database.model.ai_resource.resource_table import AIResourceORM
+from database.session import DbSession
+
 
 MODEL = TypeVar("MODEL", bound=SQLModel)
 
@@ -99,7 +102,7 @@ def deserialize(
 
     @staticmethod
     def deserialize_ids(clazz: type[SQLModel], session: Session, ids: list[int]):
-        query = select(clazz).where(clazz.identifier.in_(ids))  # noqa
+        query = select(clazz).where(clazz.identifier.in_(ids))
         existing = session.scalars(query).all()
         ids_not_found = set(ids) - {e.identifier for e in existing}
         if any(ids_not_found):
@@ -253,16 +256,30 @@ def create_getter_dict(attribute_serializers: Dict[str, Serializer]):
     object."""
     attribute_names = set(attribute_serializers.keys())
 
+    def is_soft_deleted(item) -> bool:
+        if hasattr(item, "date_deleted") and item.date_deleted:
+            return True
+        if isinstance(item, AIResourceORM):
+            with DbSession() as session:
+                clazz_, resource = get_asset_by_identifier(item.identifier, session)
+                return resource.date_deleted is not None
+
+        return False  # Not sure what cases are not covered here.
+
     class GetterDictSerializer(GetterDict):
         def get(self, key: Any, default: Any = None) -> Any:
-            # if key == "has_part" and hasattr(self._obj, "ai_resource"):
-            #     return [p.identifier for p in self._obj.ai_resource.has_part]
             if key in attribute_names:
                 serializer = attribute_serializers[key]
                 attribute_value = serializer.value(model=self._obj, attribute_name=key)
                 if attribute_value is not None:
                     if isinstance(attribute_value, list):
-                        return [serializer.serialize(v) for v in attribute_value]
+                        return [
+                            serializer.serialize(v)
+                            for v in attribute_value
+                            if not is_soft_deleted(v)
+                        ]
+                    if is_soft_deleted(attribute_value):
+                        return None
                     return serializer.serialize(attribute_value)
             return super().get(key, default)
 

src/database/review.py

@@ -9,7 +9,8 @@
 from database.model.field_length import NORMAL, LONG
 from database.model.concept.concept import AIoDConcept
 from database.model.helper_functions import non_abstract_subclasses
-from routers.helper_functions import get_all_asset_schemas, get_asset_type_by_abbreviation
+from routers.helper_functions import get_all_asset_schemas
+from database.model.helper_functions import get_asset_type_by_abbreviation
 
 REQUIRED_NUMBER_OF_REVIEWS = 1
 

src/routers/asset_router.py

@@ -6,7 +6,7 @@
 from authentication import KeycloakUser, get_user_or_none, get_user_or_raise, get_user_by_username
 from database.authorization import user_can_administer, set_permission, Permission, register_user
 from database.session import get_session
-from routers.helper_functions import get_asset_type_by_abbreviation
+from database.model.helper_functions import get_asset_by_identifier
 from routers.resource_routers import versioned_routers
 from database.model.concept.aiod_entry import EntryStatus
 from database.authorization import user_can_read, PermissionType
@@ -118,16 +118,4 @@ def asset(
             detail=f"No router found to deserialize asset of type '{model_class.__name__}'",
         )
 
-    def get_asset_by_identifier(identifier, session):
-        asset_type_map = get_asset_type_by_abbreviation()
-        prefix = identifier.split("_")[0]
-        model_class = asset_type_map.get(prefix)
-        if not model_class:
-            raise HTTPException(
-                status_code=HTTPStatus.NOT_FOUND,
-                detail=f"Unknown asset type with identifier '{identifier}'",
-            )
-        resource = session.get(model_class, identifier)
-        return model_class, resource
-
     return router

src/routers/bookmark_router.py

@@ -13,7 +13,7 @@
 from datetime import datetime
 
 from dependencies.pagination import PaginationParams
-from routers.helper_functions import get_asset_type_by_abbreviation
+from database.model.helper_functions import get_asset_type_by_abbreviation
 from versioning import Version
 
 

src/routers/helper_functions.py

@@ -29,15 +29,6 @@ def get_all_asset_schemas():
     ]
 
 
-@cache
-def get_asset_type_by_abbreviation() -> dict[str, type[AIoDConcept]]:
-    return {
-        cls.__abbreviation__: cls
-        for cls in non_abstract_subclasses(AIoDConcept)
-        if hasattr(cls, "__abbreviation__")
-    }
-
-
 @cache
 def get_router_by_type() -> dict[type[AIoDConcept], type["ResourceRouter"]]:
     from routers.resource_routers import versioned_routers  # avoid cyclical import