Hibernate.orgCommunity Documentation

Chapter 7. Collection mapping

Table of Contents

7.1. Persistent collections
7.2. How to map collections
7.2.1. Collection foreign keys
7.2.2. Indexed collections
7.2.3. Collections of basic types and embeddable objects
7.3. Advanced collection mappings
7.3.1. Sorted collections
7.3.2. Bidirectional associations
7.3.3. Bidirectional associations with indexed collections
7.3.4. Ternary associations
7.3.5. Using an <idbag>
7.4. Collection examples

Naturally Hibernate also allows to persist collections. These persistent collections can contain almost any other Hibernate type, including: basic types, custom types, components and references to other entities. The distinction between value and reference semantics is in this context very important. An object in a collection might be handled with "value" semantics (its life cycle fully depends on the collection owner), or it might be a reference to another entity with its own life cycle. In the latter case, only the "link" between the two objects is considered to be a state held by the collection.

As a requirement persistent collection-valued fields must be declared as an interface type (see Example 7.2, “Collection mapping using @OneToMany and @JoinColumn”). The actual interface might be java.util.Set, java.util.Collection, java.util.List, java.util.Map, java.util.SortedSet, java.util.SortedMap or anything you like ("anything you like" means you will have to write an implementation of org.hibernate.usertype.UserCollectionType).

Notice how in Example 7.2, “Collection mapping using @OneToMany and @JoinColumn” the instance variable parts was initialized with an instance of HashSet. This is the best way to initialize collection valued properties of newly instantiated (non-persistent) instances. When you make the instance persistent, by calling persist(), Hibernate will actually replace the HashSet with an instance of Hibernate's own implementation of Set. Be aware of the following error:

The persistent collections injected by Hibernate behave like HashMap, HashSet, TreeMap, TreeSet or ArrayList, depending on the interface type.

Collections instances have the usual behavior of value types. They are automatically persisted when referenced by a persistent object and are automatically deleted when unreferenced. If a collection is passed from one persistent object to another, its elements might be moved from one table to another. Two entities cannot share a reference to the same collection instance. Due to the underlying relational model, collection-valued properties do not support null value semantics. Hibernate does not distinguish between a null collection reference and an empty collection.

Using annotations you can map Collections, Lists, Maps and Sets of associated entities using @OneToMany and @ManyToMany. For collections of a basic or embeddable type use @ElementCollection. In the simplest case a collection mapping looks like this:

Product describes a unidirectional relationship with Part using the join column PART_ID. In this unidirectional one to many scenario you can also use a join table as seen in Example 7.3, “Collection mapping using @OneToMany and @JoinTable”.

Without describing any physical mapping (no @JoinColumn or @JoinTable), a unidirectional one to many with join table is used. The table name is the concatenation of the owner table name, _, and the other side table name. The foreign key name(s) referencing the owner table is the concatenation of the owner table, _, and the owner primary key column(s) name. The foreign key name(s) referencing the other side is the concatenation of the owner property name, _, and the other side primary key column(s) name. A unique constraint is added to the foreign key referencing the other side table to reflect the one to many.

Lets have a look now how collections are mapped using Hibernate mapping files. In this case the first step is to chose the right mapping element. It depends on the type of interface. For example, a <set> element is used for mapping properties of type Set.

In Example 7.4, “Mapping a Set using <set>” a one-to-many association links the Product and Part entities. This association requires the existence of a foreign key column and possibly an index column to the Part table. This mapping loses certain semantics of normal Java collections:

  • An instance of the contained entity class cannot belong to more than one instance of the collection.

  • An instance of the contained entity class cannot appear at more than one value of the collection index.

Looking closer at the used <one-to-many> tag we see that it has the following options.

The <one-to-many> element does not need to declare any columns. Nor is it necessary to specify the table name anywhere.

Apart from the <set> tag as shown in Example 7.4, “Mapping a Set using <set>”, there is also <list>, <map>, <bag>, <array> and <primitive-array> mapping elements. The <map> element is representative:

Example 7.6. Elements of the <map> mapping

    name="propertyName"                                    (1)
    table="table_name"                                     (2)
    schema="schema_name"                                   (3)
    lazy="true|extra|false"                                (4)
    inverse="true|false"                                   (5)
    sort="unsorted|natural|comparatorClass"                (7)
    order-by="column_name asc|desc"                        (8)
    where="arbitrary sql where condition"                  (9)
    fetch="join|select|subselect"                          (10)
    batch-size="N"                                         (11)
    access="field|property|ClassName"                      (12)
    optimistic-lock="true|false"                           (13)
    mutable="true|false"                                   (14)

    <key .... />
    <map-key .... />
    <element .... />


name: the collection property name


table (optional - defaults to property name): the name of the collection table. It is not used for one-to-many associations.


schema (optional): the name of a table schema to override the schema declared on the root element


lazy (optional - defaults to true): disables lazy fetching and specifies that the association is always eagerly fetched. It can also be used to enable "extra-lazy" fetching where most operations do not initialize the collection. This is suitable for large collections.


inverse (optional - defaults to false): marks this collection as the "inverse" end of a bidirectional association.


cascade (optional - defaults to none): enables operations to cascade to child entities.


sort (optional): specifies a sorted collection with natural sort order or a given comparator class.


order-by (optional): specifies a table column or columns that define the iteration order of the Map, Set or bag, together with an optional asc or desc.


where (optional): specifies an arbitrary SQL WHERE condition that is used when retrieving or removing the collection. This is useful if the collection needs to contain only a subset of the available data.


fetch (optional, defaults to select): chooses between outer-join fetching, fetching by sequential select, and fetching by sequential subselect.


batch-size (optional, defaults to 1): specifies a "batch size" for lazily fetching instances of this collection.


access (optional - defaults to property): the strategy Hibernate uses for accessing the collection property value.


optimistic-lock (optional - defaults to true): specifies that changes to the state of the collection results in increments of the owning entity's version. For one-to-many associations you may want to disable this setting.


mutable (optional - defaults to true): a value of false specifies that the elements of the collection never change. This allows for minor performance optimization in some cases.

After exploring the basic mapping of collections in the preceding paragraphs we will now focus details like physical mapping considerations, indexed collections and collections of value types.

In the following paragraphs we have a closer look at the indexed collections List and Map how the their index can be mapped in Hibernate.

Lists can be mapped in two different ways:

  • as ordered lists, where the order is not materialized in the database

  • as indexed lists, where the order is materialized in the database

To order lists in memory, add @javax.persistence.OrderBy to your property. This annotation takes as parameter a list of comma separated properties (of the target entity) and orders the collection accordingly (eg firstname asc, age desc, weight asc nulls last), if the string is empty, the collection will be ordered by the primary key of the target entity.

To store the index value in a dedicated column, use the @javax.persistence.OrderColumn annotation on your property. This annotations describes the column name and attributes of the column keeping the index value. This column is hosted on the table containing the association foreign key. If the column name is not specified, the default is the name of the referencing property, followed by underscore, followed by ORDER (in the following example, it would be orders_ORDER).


We recommend you to convert the legacy @org.hibernate.annotations.IndexColumn usages to the JPA standard @javax.persistence.OrderColumn.

If you are leveraging a custom list index base (maybe currently using the org.hibernate.annotations.IndexColumn.literal attribute), you can specify this using the @org.hibernate.annotations.ListIndexBase in conjunction with @javax.persistence.OrderColumn. The default base is 0 like in Java.

Looking again at the Hibernate mapping file equivalent, the index of an array or list is always of type integer and is mapped using the <list-index> element. The mapped column contains sequential integers that are numbered from zero by default.

If your table does not have an index column, and you still wish to use List as the property type, you can map the property as a Hibernate <bag>. A bag does not retain its order when it is retrieved from the database, but it can be optionally sorted or ordered.

The question with Maps is where the key value is stored. There are several options. Maps can borrow their keys from one of the associated entity properties or have dedicated columns to store an explicit key.

To use one of the target entity property as a key of the map, use @MapKey(name="myProperty"), where myProperty is a property name in the target entity. When using @MapKey without the name attribute, the target entity primary key is used. The map key uses the same column as the property pointed out. There is no additional column defined to hold the map key, because the map key represent a target property. Be aware that once loaded, the key is no longer kept in sync with the property. In other words, if you change the property value, the key will not change automatically in your Java model.

Alternatively the map key is mapped to a dedicated column or columns. In order to customize the mapping use one of the following annotations:

  • @MapKeyColumn if the map key is a basic type. If you don't specify the column name, the name of the property followed by underscore followed by KEY is used (for example orders_KEY).

  • @MapKeyEnumerated / @MapKeyTemporal if the map key type is respectively an enum or a Date.

  • @MapKeyJoinColumn/@MapKeyJoinColumns if the map key type is another entity.

  • @AttributeOverride/@AttributeOverrides when the map key is a embeddable object. Use key. as a prefix for your embeddable object property names.

You can also use @MapKeyClass to define the type of the key if you don't use generics.

Example 7.11. Map key as basic type using @MapKeyColumn


public class Customer {
   @Id @GeneratedValue public Integer getId() { return id; }
   public void setId(Integer id) { this.id = id; }
   private Integer id;
   @OneToMany @JoinTable(name="Cust_Order")
   public Map<String,Order> getOrders() { return orders; }
   public void setOrders(Map<String,Order> orders) { this.orders = orders; }
   private Map<String,Order> orders;
public class Order {
   @Id @GeneratedValue public Integer getId() { return id; }
   public void setId(Integer id) { this.id = id; }
   private Integer id;
   public String getNumber() { return number; }
   public void setNumber(String number) { this.number = number; }
   private String number;
   public Customer getCustomer() { return customer; }
   public void setCustomer(Customer customer) { this.customer = customer; }
   private Customer number;
-- Table schema
|-------------| |----------| |---------------|
| Order       | | Customer | | Cust_Order    |
|-------------| |----------| |---------------|
| id          | | id       | | customer_id   |
| number      | |----------| | order_id      |
| customer_id |              | orders_number |
|-------------|              |---------------|


We recommend you to migrate from @org.hibernate.annotations.MapKey / @org.hibernate.annotation.MapKeyManyToMany to the new standard approach described above

Using Hibernate mapping files there exists equivalent concepts to the descibed annotations. You have to use <map-key>, <map-key-many-to-many> and <composite-map-key>. <map-key> is used for any basic type, <map-key-many-to-many> for an entity reference and <composite-map-key> for a composite type.

In some situations you don't need to associate two entities but simply create a collection of basic types or embeddable objects. Use the @ElementCollection for this case.

The collection table holding the collection data is set using the @CollectionTable annotation. If omitted the collection table name defaults to the concatenation of the name of the containing entity and the name of the collection attribute, separated by an underscore. In our example, it would be User_nicknames.

The column holding the basic type is set using the @Column annotation. If omitted, the column name defaults to the property name: in our example, it would be nicknames.

But you are not limited to basic types, the collection type can be any embeddable object. To override the columns of the embeddable object in the collection table, use the @AttributeOverride annotation.

Such an embeddable object cannot contains a collection itself.


in @AttributeOverride, you must use the value. prefix to override properties of the embeddable object used in the map value and the key. prefix to override properties of the embeddable object used in the map key.


public class User {
      @AttributeOverride(name="key.street1", column=@Column(name="fld_street")),
      @AttributeOverride(name="value.stars", column=@Column(name="fld_note"))
   public Map<Address,Rating> getFavHomes() { ... }


We recommend you to migrate from @org.hibernate.annotations.CollectionOfElements to the new @ElementCollection annotation.

Using the mapping file approach a collection of values is mapped using the <element> tag. For example:

Hibernate supports collections implementing java.util.SortedMap and java.util.SortedSet. With annotations you declare a sort comparator using @Sort. You chose between the comparator types unsorted, natural or custom. If you want to use your own comparator implementation, you'll also have to specify the implementation class using the comparator attribute. Note that you need to use either a SortedSet or a SortedMap interface.

Using Hibernate mapping files you specify a comparator in the mapping file with <sort>:

Allowed values of the sort attribute are unsorted, natural and the name of a class implementing java.util.Comparator.


Sorted collections actually behave like java.util.TreeSet or java.util.TreeMap.

If you want the database itself to order the collection elements, use the order-by attribute of set, bag or map mappings. This solution is implemented using LinkedHashSet or LinkedHashMap and performs the ordering in the SQL query and not in the memory.


The value of the order-by attribute is an SQL ordering, not an HQL ordering.

Associations can even be sorted by arbitrary criteria at runtime using a collection filter():

A bidirectional association allows navigation from both "ends" of the association. Two kinds of bidirectional association are supported:


set or bag valued at one end and single-valued at the other


set or bag valued at both ends

Often there exists a many to one association which is the owner side of a bidirectional relationship. The corresponding one to many association is in this case annotated by @OneToMany(mappedBy=...)

Troop has a bidirectional one to many relationship with Soldier through the troop property. You don't have to (must not) define any physical mapping in the mappedBy side.

To map a bidirectional one to many, with the one-to-many side as the owning side, you have to remove the mappedBy element and set the many to one @JoinColumn as insertable and updatable to false. This solution is not optimized and will produce additional UPDATE statements.

How does the mappping of a bidirectional mapping look like in Hibernate mapping xml? There you define a bidirectional one-to-many association by mapping a one-to-many association to the same table column(s) as a many-to-one association and declaring the many-valued end inverse="true".

Mapping one end of an association with inverse="true" does not affect the operation of cascades as these are orthogonal concepts.

A many-to-many association is defined logically using the @ManyToMany annotation. You also have to describe the association table and the join conditions using the @JoinTable annotation. If the association is bidirectional, one side has to be the owner and one side has to be the inverse end (ie. it will be ignored when updating the relationship values in the association table):

In this example @JoinTable defines a name, an array of join columns, and an array of inverse join columns. The latter ones are the columns of the association table which refer to the Employee primary key (the "other side"). As seen previously, the other side don't have to (must not) describe the physical mapping: a simple mappedBy argument containing the owner side property name bind the two.

As any other annotations, most values are guessed in a many to many relationship. Without describing any physical mapping in a unidirectional many to many the following rules applied. The table name is the concatenation of the owner table name, _ and the other side table name. The foreign key name(s) referencing the owner table is the concatenation of the owner table name, _ and the owner primary key column(s). The foreign key name(s) referencing the other side is the concatenation of the owner property name, _, and the other side primary key column(s). These are the same rules used for a unidirectional one to many relationship.

A Store_City is used as the join table. The Store_id column is a foreign key to the Store table. The implantedIn_id column is a foreign key to the City table.

Without describing any physical mapping in a bidirectional many to many the following rules applied. The table name is the concatenation of the owner table name, _ and the other side table name. The foreign key name(s) referencing the owner table is the concatenation of the other side property name, _, and the owner primary key column(s). The foreign key name(s) referencing the other side is the concatenation of the owner property name, _, and the other side primary key column(s). These are the same rules used for a unidirectional one to many relationship.

A Store_Customer is used as the join table. The stores_id column is a foreign key to the Store table. The customers_id column is a foreign key to the Customer table.

Using Hibernate mapping files you can map a bidirectional many-to-many association by mapping two many-to-many associations to the same database table and declaring one end as inverse.


You cannot select an indexed collection.

Example 7.27, “Many to many association using Hibernate mapping files” shows a bidirectional many-to-many association that illustrates how each category can have many items and each item can be in many categories:

Changes made only to the inverse end of the association are not persisted. This means that Hibernate has two representations in memory for every bidirectional association: one link from A to B and another link from B to A. This is easier to understand if you think about the Java object model and how a many-to-many relationship in Javais created:

The non-inverse side is used to save the in-memory representation to the database.

There are some additional considerations for bidirectional mappings with indexed collections (where one end is represented as a <list> or <map>) when using Hibernate mapping files. If there is a property of the child class that maps to the index column you can use inverse="true" on the collection mapping:

If there is no such property on the child class, the association cannot be considered truly bidirectional. That is, there is information available at one end of the association that is not available at the other end. In this case, you cannot map the collection inverse="true". Instead, you could use the following mapping:

Note that in this mapping, the collection-valued end of the association is responsible for updates to the foreign key.

This section covers collection examples.

The following class has a collection of Child instances:

If each child has, at most, one parent, the most natural mapping is a one-to-many association:

This maps to the following table definitions:

If the parent is required, use a bidirectional one-to-many association:

Notice the NOT NULL constraint:

Alternatively, if this association must be unidirectional you can enforce the NOT NULL constraint.

On the other hand, if a child has multiple parents, a many-to-many association is appropriate.

Table definitions:

For more examples and a complete explanation of a parent/child relationship mapping, see Chapter 23, Example: Parent/Child for more information. Even more complex association mappings are covered in the next chapter.