I have this table in a legacy database:



It's very similar to the Category that appears in the book "Java persistence with hibernate", however, it has only two levels:

It has a composite key using the columns ID and TYPE. The ones named _PARENT have a relationship with the previous ones in this way:

- If ID and TYPE are equals to ID_PARENT and TYPE_PARENT, it means that we have a "parent category" (because its father is itself)
- If they're different, we have a "child category" (because its father is the one whose ID and TYPE are stored in ID_PARNT and TYPE_PARENT)

Is it possible to create some kind of formula to modelize something like this?



Thanks in advance

Solved. At least the technical problem exposed.

You can find the solution in the very book "Java persistence with hibernate", maybe the actual problem is that I have a low level in reading comprehension (or the book is too dense, even though fantastic)

At the end of the 202 page you can read: "Sometimes, especially in legacy schemas, you don’t have the freedom to include
an extra discriminator column in your entity tables. In this case, you can apply a formula to calculate a discriminator value for each row:"

That is the solution: a DiscriminatorFormula using a SQL CASE/WHEN that results in literals that will be used as DiscriminatorValues in the child classes.



NOTE: I made an additional mistake in the previous code, the org.hibernate.annotations.DiscriminatorFormula "must be placed at the root entity" (as the API said).

Something similar applies to DiscriminatorValue: "this annotation can only be specified on a concrete entity class" (as the API said).

Another problem remains, but it is a data problem.

As it had been done in the book, there is a relationship between categories:


However, there is a chance to manipulate the database manually, and according to my first example, someone can insert a Child category whose related category isn't a Parent Category (according to the discriminator formula)

At this moment Hibernate cannot instantiate an appropriate object and the program simply crashes.

The actual problem is that in the error log is difficult to see that the there is a data problem.

Is it possible to catch the creation of the object before it's created in order to ensure that there won't be any problem?

Actually, I don't want to create a different kind of object, I only want to create a good warning containing the row that is creating the problem.

I tried with an Interceptor (extends EmptyInterceptor) executing a Query (entityManager.createNativeQuery) in the public boolean onLoad(Object entity, Serializable id, Object[] state, String[] propertyNames, Type[] types) method before returning super.onLoad(entity, id, state, propertyNames, types);.

But it failed (and the transaction is rollbacked)

Can anyone help me with this?