How to generate some JPA entities with a multi join?

Question

i would like to create an application in this context : Zk 6, Spring v3.1.1, JPA 2.0, Hibernate 4.1.4, all with annotations but i have some pb with JPA concept.

Here are a type of case study :

3 tables, all linked via a join table ; we are dealing with cardinality 0, n.

So we have T_E_USER, T_E_TYPE and T_E_AIR. Each table has a numeric ID, and a simple VARCHAR field.
A join table is created with T_J_USR_TPE_AIR with the 3 ID referenced by foreign keys forming a composed primary key.

I'm using Hibernate Tools for generate my entities (version JPA). And that's where the problems start ....

I have, in each entity class, an attribute of type set with annotation @ OneToMany.

I have a class representing the join that has an id attribute of complex type (another class) with an annotation EmbeddedId for a composite key. And attributes representing the three entities with annotations @ ManyToOne.

Here are my questions, because that's where I'm confused:

which should i set into the "mappedBy" attribute in the annotation @ OneToMany of my entities?
Am I forced to do a class entity representing the join?
How does the CASCADE? Is it possible to use it in this context to enrich the join table "automatically"? Or should I manually instantiate the class representative of the join in order to persist the information myself?

A big thank you in advance for any kind soul who could give me a helping hand.

Thank you for your answers but one said "yes" when the other says "no" lol

Here's what I did during the day but I have not yet been tested.

In each entity table, i added a @OneToMany relation with mappedBy setted to the attribute defined in "join" entity :

@OneToMany(fetch = FetchType.LAZY, 
       mappedBy = "aircraft",
       cascade = { CascadeType.REMOVE })
private Set<UserConfig> userConfigs = new HashSet<UserConfig>(0);

...

@OneToMany(fetch = FetchType.LAZY, 
       mappedBy = "userAccount",
       cascade = { CascadeType.REMOVE })
private Set<UserConfig> userConfigs = new HashSet<UserConfig>(0);

...

@OneToMany(fetch = FetchType.LAZY, 
       mappedBy = "referenceType",
       cascade = { CascadeType.REMOVE })
private Set<UserConfig> userConfigs = new HashSet<UserConfig>(0);

And i created a new Entity for the join table.

@Entity
@Table(name = "T_J_USR_RFT_AIR_URA")
public class UserConfig implements java.io.Serializable {

@EmbeddedId
@AttributeOverrides({
        @AttributeOverride(name = "airId", 
                           column = @Column(name = "URA_AIR_ID", nullable = false)),
        @AttributeOverride(name = "usrId", 
                           column = @Column(name = "URA_USR_ID", nullable = false)),
        @AttributeOverride(name = "rftId", 
                           column = @Column(name = "URA_RFT_ID", nullable = false))
})  
private UserConfigId id;

@ManyToOne(fetch = FetchType.LAZY)
@JoinColumn(name = "URA_RFT_ID", nullable = false, insertable = false, updatable = false)   
private ReferenceType referenceType;

@ManyToOne(fetch = FetchType.LAZY)
@JoinColumn(name = "URA_USR_ID", nullable = false, insertable = false, updatable = false)   
private UserAccount userAccount;

@ManyToOne(fetch = FetchType.LAZY)
@JoinColumn(name = "URA_AIR_ID", nullable = false, insertable = false, updatable = false)   
private Aircraft aircraft;

...
getter & setter
}

Where UserConfigId is :

@Embeddable
public class UserConfigId implements java.io.Serializable {

@Column(name = "URA_AIR_ID", nullable = false)  
private Integer airId;

@Column(name = "URA_USR_ID", nullable = false)
private Integer usrId;

@Column(name = "URA_RFT_ID", nullable = false)
private Integer rftId;

...
getter & setter
}

What do you think about this practice ?

I just used "cascade" if an object of the join table is deleted in order to delete all element associated in the join.

Is it all right ?

Anyway thank you Tom, i will analyzed your link. Thank you JMelnyk too.

You are welcome if you want to demonstrate what are the best practices for this case.

There was another question about three-way join tables recently - see stackoverflow.com/q/11638412/116639 . I believe that the only way to handle this in JPA is to make the join table an entity, sadly. — Tom Anderson
T_E_USER, T_E_TYPE and T_E_AIR were for an easiest explanation. In reality, it is the table's name :) The real entity/class name are UserAccount, ReferenceType and Aircraft :) The purpose is to configure some reference type on an aircraft for a user. — MychaL
Your solution looks good. Some points, though: (1) plural attributes like sets are lazy by default, so you don't need to explicitly make them lazy, (2) if you are using JPA2, an easier way to handle the use of foreign keys in a primary key is using @MapsId, and (3) you probably want equals and hashCode implementations on UserConfig, although i'm not sure about that. — Tom Anderson
yeah, there are some implementations for equals & hashcode in each class...and i'm trying to work on business-id, not on table-id. Thank you Tom ! — MychaL

Tom Anderson Tom Anderson · Accepted Answer · 2012-08-07T09:16:30

Three-way joins are tricky. I think what you've done, using an entity for the join table, is probably the right thing to do. To answer your questions:

Your @OneToMany attributes refer to the entity mapping the join table; they should be mappedBy the appropriate @ManyToOne attribute in that entity.
Yes, unfortunately, an entity for the join table is the best way to do this.
Cascades can be used to automatically add objects to the database, but not to create objects. You will need to create instances of the join entity in code.

How to generate some JPA entities with a multi join?

2 Answers