names: add internal id to identify the names

invenio_vocabularies/alembic/3ba812d80559_add_internal_name_id.py

@@ -0,0 +1,36 @@
+#
+# This file is part of Invenio.
+# Copyright (C) 2024 CERN.
+#
+# Invenio is free software; you can redistribute it and/or modify it
+# under the terms of the MIT License; see LICENSE file for more details.
+
+"""Add internal name ID."""
+
+import sqlalchemy as sa
+from alembic import op
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision = "3ba812d80559"
+down_revision = "55a700f897b6"
+branch_labels = ()
+depends_on = None
+
+
+def upgrade():
+    """Upgrade database."""
+    op.add_column(
+        "name_metadata", sa.Column("internal_id", sa.String(length=255), nullable=True)
+    )
+    op.create_unique_constraint(
+        op.f("uq_name_metadata_internal_id"), "name_metadata", ["internal_id"]
+    )
+
+
+def downgrade():
+    """Downgrade database."""
+    op.drop_constraint(
+        op.f("uq_name_metadata_internal_id"), "name_metadata", type_="unique"
+    )
+    op.drop_column("name_metadata", "internal_id")

invenio_vocabularies/contrib/names/components.py

@@ -0,0 +1,24 @@
+#
+# This file is part of Invenio.
+# Copyright (C) 2024 CERN.
+#
+# Invenio is free software; you can redistribute it and/or modify it
+# under the terms of the MIT License; see LICENSE file for more details.
+
+"""Names service components."""
+
+from invenio_records_resources.services.records.components import ServiceComponent
+
+
+class InternalIDComponent(ServiceComponent):
+    """Service component for internal id field."""
+
+    field = "internal_id"
+
+    def create(self, identity, data=None, record=None, **kwargs):
+        """Create handler."""
+        setattr(record, self.field, data.pop(self.field, None))
+
+    def update(self, identity, data=None, record=None, **kwargs):
+        """Update handler."""
+        setattr(record, self.field, data.pop(self.field, None))

invenio_vocabularies/contrib/names/config.py

@@ -21,6 +21,7 @@
 from werkzeug.local import LocalProxy
 
 from ...services.components import PIDComponent
+from .components import InternalIDComponent
 
 names_schemes = LocalProxy(lambda: current_app.config["VOCABULARIES_NAMES_SCHEMES"])
 
@@ -67,6 +68,7 @@ class NamesSearchOptions(SearchOptions):
 
 service_components = [
     # Order of components are important!
+    InternalIDComponent,
     DataComponent,
     PIDComponent,
     RelationsComponent,

invenio_vocabularies/contrib/names/jsonschemas/names/name-v1.0.0.json

@@ -65,4 +65,4 @@
       "uniqueItems": true
     }
   }
-}
+}

invenio_vocabularies/contrib/names/names.py

@@ -13,7 +13,7 @@
 from invenio_records.dumpers import SearchDumper
 from invenio_records.dumpers.indexedat import IndexedAtDumperExt
 from invenio_records.dumpers.relations import RelationDumperExt
-from invenio_records.systemfields import RelationsField
+from invenio_records.systemfields import ModelField, RelationsField
 from invenio_records_resources.factories.factory import RecordTypeFactory
 from invenio_records_resources.records.systemfields import (
     ModelPIDField,
@@ -47,10 +47,12 @@
         # cannot set to nullable=False because it would fail at
         # service level when create({}), see records-resources.
         "pid": db.Column(db.String(255), unique=True),
+        "internal_id": db.Column(db.String(255), unique=True, nullable=True),
     },
     schema_version="1.0.0",
     schema_path="local://names/name-v1.0.0.json",
     index_name="names-name-v2.0.0",
+    record_cls_attrs={"internal_id": ModelField("internal_id", dump=False)},
     record_relations=name_relations,
     record_dumper=SearchDumper(
         model_fields={"pid": ("id", str)},

invenio_vocabularies/contrib/names/schema.py

@@ -35,6 +35,7 @@ class NameSchema(BaseVocabularySchema, ModePIDFieldVocabularyMixin):
     so it does not inherit from it.
     """
 
+    internal_id = fields.Str(allow_none=True)
     name = SanitizedUnicode()
     given_name = SanitizedUnicode()
     family_name = SanitizedUnicode()
@@ -66,7 +67,7 @@ def validate_names(self, data, **kwargs):
             raise ValidationError({"family_name": messages})
 
     @validates_schema
-    def validate_affiliatons(self, data, **kwargs):
+    def validate_affiliations(self, data, **kwargs):
         """Validate names."""
         affiliations = data.get("affiliations", [])
         seen_names = set()

invenio_vocabularies/contrib/names/services.py

@@ -42,10 +42,10 @@ def resolve(self, identity, id_, id_type, many=False):
         else:
             results = self._read_many(identity, search_query, max_records=1)
 
-        # cant use the results_item because it returns dicts intead of records
+        # cant use the results_item because it returns dicts instead of records
         total = results.hits.total["value"]
         if total == 0:
-            # Not a PID but trated as such
+            # Not a PID but treated as such
             raise PIDDoesNotExistError(pid_type=id_type, pid_value=id_)
         if many:
             for result in results:

tests/contrib/names/test_names_service.py

@@ -27,6 +27,7 @@ def test_simple_flow(app, service, identity, name_full_data, example_affiliation
 
     # Create it
     item = service.create(identity, name_full_data)
+
     id_ = item.id
 
     # add dereferenced values
@@ -68,7 +69,7 @@ def test_simple_flow(app, service, identity, name_full_data, example_affiliation
     # Fail to retrieve it
     # - db
     # only the metadata is removed from the record, it is still resolvable
-    base_keys = {"created", "updated", "id", "links", "revision_id"}
+    base_keys = {"created", "updated", "id", "links", "revision_id", "internal_id"}
     deleted_rec = service.read(identity, id_).to_dict()
     assert set(deleted_rec.keys()) == base_keys
     # - search