Of course, this won’t run–Hibernate will gag on the mapping, because it doesn’t know how to map a JDBC DATE column to a DateNoMs–as one would expect. There are two things we need at this point: first, an object which Hibernate can use to transform JDBC DATE into a DateNoMs, and an annotation pointing to that “Factory”. The factory class is produced by implementing (in the simplest case) org.hibernate.usertype.UserType. Documentation in this interface is pretty thin, but there are good examples available in the Hibernate distribution. Here’s my implementation. I’m greatly helped by the fact that my class (DateNoMs) is very close to java.util.Date, and java.sql.Date extends java.util.Date.
/**
* Map “things” (currently Oracle Date columns) to the DateNoMs.
*/
public class DateNoMsType implements UserType {
/** @see org.hibernate.usertype.UserType#assemble(java.io.Serializable, Object) */
public Object assemble(Serializable cached, @SuppressWarnings(“unused”) Object owner) {
return cached;
}
/** @see org.hibernate.usertype.UserType#deepCopy(Object) */
public Object deepCopy(Object value) {
if (value==null)
return null;
if (! (value instanceof java.util.Date))
throw new UnsupportedOperationException(“can’t convert “+value.getClass());
return new DateNoMs((java.util.Date)value);
}
/** @see org.hibernate.usertype.UserType#disassemble(Object) */
public Serializable disassemble(Object value) throws HibernateException {
if (! (value instanceof java.util.Date))
throw new UnsupportedOperationException(“can’t convert “+value.getClass());
return new DateNoMs((java.util.Date)value);
}
/** @see org.hibernate.usertype.UserType#equals(Object, Object) */
public boolean equals(Object x, Object y) throws HibernateException {
return x.equals(y);
}
/** @see org.hibernate.usertype.UserType#hashCode(Object) */
public int hashCode(Object value) throws HibernateException {
return value.hashCode();
}
/** @see org.hibernate.usertype.UserType#isMutable() */
public boolean isMutable() {
return true;
}
/** @see org.hibernate.usertype.UserType#nullSafeGet(java.sql.ResultSet, String[], Object) */
public Object nullSafeGet(ResultSet rs, String[] names, @SuppressWarnings(“unused”) Object owner)
throws HibernateException, SQLException {
// assume that we only map to one column, so there’s only one column name
java.sql.Date value = rs.getDate( names[0] );
if (value==null)
return null;
return new DateNoMs(value.getTime());
}
/** @see org.hibernate.usertype.UserType#nullSafeSet(java.sql.PreparedStatement, Object, int) */
public void nullSafeSet(PreparedStatement stmt, Object value, int index)
throws HibernateException, SQLException {
if (value==null) {
stmt.setNull(index, Types.DATE);
return;
}
if (! (value instanceof java.util.Date))
throw new UnsupportedOperationException(“can’t convert “+value.getClass());
stmt.setDate( index, new java.sql.Date( ((java.util.Date)value).getTime()) );
}
/** @see org.hibernate.usertype.UserType#replace(Object, Object, Object) */
public Object replace(Object original,
@SuppressWarnings(“unused”) Object target, @SuppressWarnings(“unused”) Object owner) {
return original;
}
/** @see org.hibernate.usertype.UserType#returnedClass() */
@SuppressWarnings(“unchecked”)
public Class returnedClass() {
return DateNoMs.class;
}
/** @see org.hibernate.usertype.UserType#sqlTypes() */
public int[] sqlTypes() {
return new int[] {Types.DATE};
}
}
The core of this class is the two methods which get and set values associated with my new type: nullSafeSet and nullSafeGet. One key thing to note is that nullSafeGet is supplied with a list of all the column names mapped to the custom datatype in the current query. In my case, there’s only one, but in complex cases, you can map multiple columns to one object (there are examples in the Hibernate documentation).
The final piece of the puzzle is the annotation which tells Hibernate to use the new “Type” class to generate objects of your custom type by adding a new @Type annotation to the column:
@Type(type=”com.gorillalogic.type.DateNoMsType”)
@Column(name = “PAYMENT_DATE”)
private DateNoMs m_paymentDate;
The @Type annotation needs a full path to the class that implements the userType interface; this is the factory for producing the target type of the mapped column.
If you’re going to use your new type in a lot of places, you can shorten the @Type annotation by doing a typedef; you can place this in package-info.java in any package you like (I put mine in the same package as the UserType class). Here’s the line for the type defined above:
@TypeDefs(
{
@TypeDef(name = “dateNoMs”, typeClass = com.gorillalogic.type.DateNoMsType.class
}) package com.gorillalogic.type;
Now my column annotation can look like this:
@Type(type=”dateNoMsType”)
@Column(name = “PAYMENT_DATE”)
private DateNoMs m_paymentDate;
That should be enough to get you started.
