Our data model is separated into schemas on two databases. The schemas are used in isolation except for a few single-key relationships that bridge between the two. There are no write transactions that will span both databases.
Similar to this question Doing a join over 2 tables in different databases using Hibernate, we want to use Hibernate to handle joining the entities. We cannot use the database solution (Federated views on DB2).
We have set up Hibernate with two separate database configurations (Doctor and Patient), which works perfectly when using DAOs to explicitly access a particular session.
We want to use Hibernate to automatically retrieve the entity when we call DoctorBO.getExam().getPatient()
Where examination contains an id pointing to the Patient table on the other database.
One way I've tried doing this is using a custom UserType:
public class DistributedUserType implements UserType, ParameterizedType
{
public static final String CLASS = "CLASS";
public static final String SESSION = "SESSION";
private Class<? extends DistributedEntity> returnedClass;
private String session;
/** {@inheritDoc} */
@Override
public int[] sqlTypes()
{
// The column will only be the id
return new int[] { java.sql.Types.BIGINT };
}
/** {@inheritDoc} */
@Override
public Class<? extends DistributedEntity> returnedClass()
{
// Set by typedef parameter
return returnedClass;
}
/** {@inheritDoc} */
@Override
public boolean equals(Object x, Object y) throws HibernateException
{
if (x == y)
{
return true;
}
if ((x == null) || (y == null))
{
return false;
}
Long xId = ((DistributedEntity) x).getId();
Long yId = ((DistributedEntity) y).getId();
if (xId.equals(yId))
{
return true;
}
else
{
return false;
}
}
/** {@inheritDoc} */
@Override
public int hashCode(Object x) throws HibernateException
{
assert (x != null);
return x.hashCode();
}
/** {@inheritDoc} */
@Override
public Object nullSafeGet(ResultSet rs, String[] names, Object owner) throws HibernateException, SQLException
{
Long id = rs.getLong(names[0]);
return HibernateUtils.getSession(session).get(returnedClass, id);
}
/** {@inheritDoc} */
@Override
public void nullSafeSet(PreparedStatement st, Object value, int index) throws HibernateException, SQLException
{
DistributedEntity de = (DistributedEntity) value;
st.setLong(index, de.getId());
}
/** {@inheritDoc} */
@Override
public Object deepCopy(Object value) throws HibernateException
{
return value;
}
/** {@inheritDoc} */
@Override
public boolean isMutable()
{
return false;
}
/** {@inheritDoc} */
@Override
public Serializable disassemble(Object value) throws HibernateException
{
return (Serializable) value;
}
/** {@inheritDoc} */
@Override
public Object assemble(Serializable cached, Object owner) throws HibernateException
{
return cached;
}
/** {@inheritDoc} */
@Override
public Object replace(Object original, Object target, Object owner) throws HibernateException
{
return original;
}
/** {@inheritDoc} */
@Override
public void setParameterValues(Properties parameters)
{
String clazz = (String) parameters.get(CLASS);
try
{
returnedClass = ReflectHelper.classForName(clazz);
}
catch (ClassNotFoundException e)
{
throw new IllegalArgumentException("Class: " + clazz + " is not a known class type.");
}
session = (String) parameters.get(SESSION);
}
}
Which would then be used:
@TypeDef(name = "testUserType", typeClass = DistributedUserType.class, parameters = {
@Parameter(name = DistributedUserType.CLASS, value = PatientBO.CLASSNAME),
@Parameter(name = DistributedUserType.SESSION, value = HibernateUtils.PATIENT_SESS) })
@Type(type = "testUserType")
@Column(name = "PATIENT_ID")
private PatientBO patient;
The UserType works - the data is loaded correctly with only the Id of the field persisted to the database. I have tested very simple examples of doctor.getExam().getPatient()
and doctor.getExam().setPatient()
and both seem to work great, however I think this is a terrible hack and I do not have adequate knowledge of Hibernate to know if this is safe to use.
Is there a better way to achieve what we want? Is the way I've described here adequate, or will it cause difficulties in the future?
I don't think it's a good idea. You're trying to make "as if" everything was in a single database, whereas it's not the case. And you make "as if" there was a real toOne
association between an exam and a patient, although it's not a real association.
Although you are conscious of this fact, other or future developers won't necessarily be, and will wonder why it's not possible to make a query such as
select e from Exam e left join fetch e.patient
or
select e from Exam e where e.patient.name like 'Smith%'
In short, your pseudo-association only fulfills a very small part of the contract a regular association offers, and this will, IMO, cause more confusion than comfort.
Nothing stops you from having a utility method like
Patient getExamPatient(Exam e)
that does the same thing, but makes it clear that there is no real asociation between both entities.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With