SAP Tutorials Platform

Appearance

@PersonalData Cascade ​

The user-anonymization pipeline (AdminService.anonymizeUser, AdminService.anonymizeByDsrRequest) runs through a cascade walker that processes every entity annotated with @PersonalData automatically. Adding a new annotated entity makes it part of the cascade with no JS changes, provided the entity uses the default 'null-personal' action.

Cascade actions ​

Each @PersonalData entity declares its action via @PersonalData.cascade. Four valid values; default is 'null-personal' when the annotation is absent.

ValueSemanticsExample
'null-personal' (default)Set FK = null. Set every IsPotentiallyPersonal field = null. Keep row.CodeCheckSubmissions
'delete'DELETE rows where FK = user.ID.UserMetaData
'audit-only'UPDATE createdBy + modifiedBy = 'ANONYMIZED'. Keep row, keep FK.TaskRecords
'identity-replace'UPDATE specific identity fields with placeholders/null per srv/lib/anonymization.js buildAnonymizationOps.Users

Adding a new @PersonalData entity ​

In db/audit-logging.cds:

cds
annotate ims.YourNewEntity with @PersonalData: {
  EntitySemantics: 'Other'  // or 'DataSubjectDetails' if the row describes a subject
  // No cascade override → default 'null-personal' applies.
} {
  user @PersonalData.FieldSemantics: 'DataSubjectID';            // required: which field is the user FK?
  someField @PersonalData.IsPotentiallyPersonal;                  // optional: any field worth nulling
};

That's it. No code change to srv/admin-service.js or srv/lib/anonymization-cascade.js.

If you need a non-default action, add cascade: '<action>' to the entity-level annotation. See db/audit-logging.cds for the existing examples.

When the walker skips an entity ​

If an entity has @PersonalData but no field annotated with FieldSemantics: 'DataSubjectID' (i.e. there's no FK telling the walker which rows belong to the user being anonymized), the walker logs a warning and emits action: 'skip' for that entity. The deploy log will show:

[anonymization-cascade] WARN  Entity ims.YourEntity has @PersonalData but no FieldSemantics: 'DataSubjectID' field — skipping cascade.

Same applies if cascade: has an unrecognised value. The cascade does NOT fail fast — it logs and continues, so a partial misconfiguration cannot block GDPR compliance.

Reference ​