Create record adjacency list relationship

This adds DB-level support for parent-child record hierarchies, which will facilitate API development relating to HasPart/IsPartOf related identifiers.

API tests have been updated to create factory records without parents/children, effectively maintaining the existing API test behaviour. Tests will be reviewed and revised as appropriate with forthcoming record API changes.

migrate/versions/2023_07_19_aa0153edf5a9_create_parent_child_record_relationship.py

@@ -0,0 +1,30 @@
+"""Create parent-child record relationship
+
+Revision ID: aa0153edf5a9
+Revises: 48bf65c26f62
+Create Date: 2023-07-19 13:58:52.518432
+
+"""
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision = 'aa0153edf5a9'
+down_revision = '48bf65c26f62'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # ### commands auto generated by Alembic - adjusted ###
+    op.add_column('record', sa.Column('parent_id', sa.String(), nullable=True))
+    op.create_foreign_key('record_parent_id_fkey', 'record', 'record', ['parent_id'], ['id'], ondelete='RESTRICT')
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - adjusted ###
+    op.drop_constraint('record_parent_id_fkey', 'record', type_='foreignkey')
+    op.drop_column('record', 'parent_id')
+    # ### end Alembic commands ###

odp/db/models/record.py

@@ -47,13 +47,18 @@ class Record(Base):
     schema_type = Column(Enum(SchemaType), nullable=False)
     schema = relationship('Schema')
 
+    # parent-child relationship for HasPart/IsPartOf related identifiers
+    parent_id = Column(String, ForeignKey('record.id', ondelete='RESTRICT'))
+    parent = relationship('Record', remote_side=id)
+    children = relationship('Record', viewonly=True)
+
     # view of associated tags (one-to-many)
     tags = relationship('RecordTag', viewonly=True)
 
     # view of associated catalog records (one-to-many)
     catalog_records = relationship('CatalogRecord', viewonly=True)
 
-    _repr_ = 'id', 'doi', 'sid', 'collection_id', 'schema_id'
+    _repr_ = 'id', 'doi', 'sid', 'collection_id', 'schema_id', 'parent_id'
 
 
 class RecordAudit(Base):

test/api/test_collection.py

@@ -429,7 +429,7 @@ def test_delete_collection(api, collection_batch, scopes, collection_auth, has_r
     del modified_collection_batch[2]
 
     if has_record:
-        RecordFactory(collection=collection_batch[2])
+        RecordFactory(collection=collection_batch[2], is_child_record=False)
 
     r = api(scopes, api_client_collections).delete(f'/collection/{collection_batch[2].id}')
 

test/api/test_provider.py

@@ -237,7 +237,7 @@ def test_delete_provider(api, provider_batch, scopes, has_record):
 
     if has_record:
         if collection := next((c for c in provider_batch[2].collections), None):
-            RecordFactory(collection=collection)
+            RecordFactory(collection=collection, is_child_record=False)
         else:
             has_record = False
 

test/api/test_record.py

@@ -19,7 +19,7 @@ def record_batch():
     """Create and commit a batch of Record instances."""
     records = []
     for _ in range(randint(3, 5)):
-        records += [record := RecordFactory()]
+        records += [record := RecordFactory(is_child_record=False)]
         RecordTagFactory.create_batch(randint(0, 3), record=record)
         CollectionTagFactory.create_batch(randint(0, 3), collection=record.collection)
     return records
@@ -29,14 +29,14 @@ def record_batch():
 def record_batch_no_tags():
     """Create and commit a batch of Record instances
     without any tag instances."""
-    return [RecordFactory() for _ in range(randint(3, 5))]
+    return [RecordFactory(is_child_record=False) for _ in range(randint(3, 5))]
 
 
 @pytest.fixture
 def record_batch_with_ids():
     """Create and commit a batch of Record instances
     with both DOIs and SIDs."""
-    return [RecordFactory(identifiers='both') for _ in range(randint(3, 5))]
+    return [RecordFactory(identifiers='both', is_child_record=False) for _ in range(randint(3, 5))]
 
 
 def record_build(collection=None, collection_tags=None, **id):
@@ -46,6 +46,7 @@ def record_build(collection=None, collection_tags=None, **id):
         **id,
         collection=collection or (collection := CollectionFactory()),
         collection_id=collection.id,
+        is_child_record=False,
     )
     if collection_tags:
         for ct in collection_tags:

test/factories.py

@@ -224,7 +224,7 @@ class Meta:
 class RecordFactory(ODPModelFactory):
     class Meta:
         model = Record
-        exclude = ('identifiers',)
+        exclude = ('identifiers', 'is_child_record')
 
     identifiers = factory.LazyFunction(lambda: choice(('doi', 'sid', 'both')))
     doi = factory.LazyAttributeSequence(lambda r, n: f'10.5555/test-{n}' if r.identifiers in ('doi', 'both') else None)
@@ -238,6 +238,13 @@ class Meta:
                                              SchemaFactory(id=r.schema_id, type='metadata'))
     timestamp = factory.LazyFunction(lambda: datetime.now(timezone.utc))
 
+    is_child_record = factory.LazyFunction(lambda: randint(0, 1))
+    parent = factory.Maybe(
+        'is_child_record',
+        yes_declaration=factory.SubFactory('test.factories.RecordFactory', is_child_record=False),
+        no_declaration=None,
+    )
+
 
 class RecordTagFactory(ODPModelFactory):
     class Meta:

test/test_db.py

@@ -91,9 +91,37 @@ def test_create_provider():
 
 def test_create_record():
     record = RecordFactory()
-    result = Session.execute(select(Record)).scalar_one()
-    assert (result.id, result.doi, result.sid, result.metadata_, result.validity, result.collection_id, result.schema_id, result.schema_type) \
-           == (record.id, record.doi, record.sid, record.metadata_, record.validity, record.collection.id, record.schema.id, record.schema.type)
+    result = Session.execute(
+        select(Record).where(Record.id == record.id)
+    ).scalar_one()
+    assert (
+               result.id,
+               result.doi,
+               result.sid,
+               result.metadata_,
+               result.validity,
+               result.collection_id,
+               result.schema_id,
+               result.schema_type,
+               result.parent_id,
+           ) == (
+               record.id,
+               record.doi,
+               record.sid,
+               record.metadata_,
+               record.validity,
+               record.collection.id,
+               record.schema.id,
+               record.schema.type,
+               record.parent_id,
+           )
+    if record.parent_id:
+        parent = Session.execute(
+            select(Record).where(Record.id == record.parent_id)
+        ).scalar_one()
+        assert result.parent == parent
+        assert result.parent_id == parent.id
+        assert parent.children == [result]
 
 
 def test_create_record_tag():