The repository pattern used to be one of the core J2EE patterns and could be found in most enterprise applications reading and writing data to persistent stores. While the Java Persistence API (JPA) as part of Java EE 5+ has replaced many aspects of the repository pattern, it is still a good approach to centralize complex query logic related to specific entities.
The DeltaSpike Data module is intended to help you simplifying your repository layer. While you will have complex queries in a repository requiring your full attention, there will also be many simple ones often requiring boilerplate code and clutter. This is where the DeltaSpike Data module will help you keeping your repository lean so you can focus on the though things.
The code sample below will give you a quick overview on the common usage scenarios of the data module:
@Repository public interface PersonRepository extends EntityRepository<Person, Long> { List<Person> findByAgeBetweenAndGender(int minAge, int maxAge, Gender gender); @Query("select p from Person p where p.ssn = ?1") Person findBySSN(String ssn); @Query(named=Person.BY_FULL_NAME) Person findByFullName(String firstName, String lastName); }
As you see in the sample, there are several usage scenarios outlined here:
The implementation of the method is done automatically by the CDI extension. A client can declare a dependency to the interface only. The details on how to use those features are outlines in the following chapters.
The simplest way using the DeltaSpike Data module is to run your application in a Java EE container supporting at least the Java EE 6 Web Profile. Other configurations like running it inside Tomcat or even a Java SE application should be possible - you need to include a JPA provider as well as a CDI container to your application manually.
Also note that in order to use abstract classes as repositories, this currently requires the presence of the javassist library in your classpath.
CAUTION:
Using DeltaSpike Data in an EAR deployment is currently restricted to annotation-based entities.
If you are using Maven as your build tool, you can add the following dependencies to your pom.xml
file to include the DeltaSpike data module:
<dependency> <groupId>org.apache.deltaspike.modules</groupId> <artifactId>deltaspike-data-module-api</artifactId> <version>${deltaspike.version}</version> <scope>compile</scope> </dependency> <dependency> <groupId>org.apache.deltaspike.modules</groupId> <artifactId>deltaspike-data-module-impl</artifactId> <version>${deltaspike.version}</version> <scope>runtime</scope> </dependency>
TIP:
Substitute the expression
${deltaspike.version}
with the most recent or appropriate version of DeltaSpike. Alternatively, you can create a Maven user-defined property to satisfy this substitution so you can centrally manage the version.
Including the API at compile time and only include the implementation at runtime protects you from inadvertantly depending on an implementation class.
DeltaSpike Data requires an EntityManager
exposed via a CDI producer - which is common practice
in Java EE 6 applications.
public class EntityManagerProducer { @PersistenceUnit private EntityManagerFactory emf; @Produces // you can also make this @RequestScoped public EntityManager create() { return emf.createEntityManager(); } public void close(@Disposes EntityManager em) { if (em.isOpen()) { em.close(); } } }
This allows the EntityManager
to be injected over CDI instead of only being used with a
@PersistenceContext
annotation. Using multiple EntityManager
is explored in more detail
in a following section.
If you use a JTA DataSource with your EntityManager
, you also have to configure the
TransactionStrategy
your repositories use. Adapt your beans.xml
for this:
<beans> <alternatives> <class>org.apache.deltaspike.jpa.impl.transaction.BeanManagedUserTransactionStrategy</class> </alternatives> </beans>
You're now ready to use repositories in your application!
With the DeltaSpike Data module, it is possible to make a repository out of basically any abstract class or interface (using a concrete class will work too, but you won't be able to use most of the CDI extension features). All that is required is to mark the type as such with a simple annotation:
@Repository(forEntity = Person.class) public abstract class PersonRepository { ... } @Repository(forEntity = Person.class) public interface PersonRepository { ... }
The @Repository
annotation tells the extension that this is a repository for the Person
entity.
Any method defined on the repository will be processed by the framework. The annotation does not
require to set the entity class (we'll see later why) but if there are just plain classes or
interfaces this is the only way to tell the framework what entity the repository relates to.
In order to simplify this, DeltaSpike Data provides several base types.
EntityRepository
interface¶Although mainly intended to hold complex query logic, working with both a repository and an EntityManager
in the service layer might unnecessarily clutter code. In order to avoid this for the most common cases,
DeltaSpike Data provides base types which can be used to replace the entity manager.
The top base type is the EntityRepository
interface, providing common methods used with an EntityManager
.
The following code shows the most important methods of the interface:
public interface EntityRepository<E, PK extends Serializable> { E save(E entity); void remove(E entity); void refresh(E entity); void flush(); E findBy(PK primaryKey); List<E> findAll(); List<E> findBy(E example, SingularAttribute<E, ?>... attributes); List<E> findByLike(E example, SingularAttribute<E, ?>... attributes); Long count(); Long count(E example, SingularAttribute<E, ?>... attributes); Long countLike(E example, SingularAttribute<E, ?>... attributes); }
The concrete repository can then extend this basic interface. For our Person repository, this might look like the following:
@Repository public interface PersonRepository extends EntityRepository<Person, Long> { Person findBySsn(String ssn); }
TIP:
Annotations on interfaces do not inherit. If the
EntityRepository
interface is extended by another interface adding some more common methods, it is not possible to simply add the annotation there. It needs to go on each concrete repository. The same is not true if a base class is introduced, as we see in the next chapter.
AbstractEntityRepository
class¶This class is an implementation of the EntityRepository
interface and provides additional functionality
when custom query logic needs also to be implemented in the repository.
public abstract class PersonRepository extends AbstractEntityRepository<Person, Long> { public Person findBySSN(String ssn) { return typedQuery("select p from Person p where p.ssn = ?1") .setParameter(1, ssn) .getResultList(); } }
EntityManager
¶While most applications will run just fine with a single EntityManager
, there might be setups
where multiple data sources are used. This can be configured with the EntityManagerConfig
annotation:
@Repository @EntityManagerConfig(entityManagerResolver = CrmEntityManagerResolver.class, flushMode = FlushModeType.COMMIT) public interface PersonRepository extends EntityRepository<Person, Long> { ... } public class CrmEntityManagerResolver implements EntityManagerResolver { @Inject @CustomerData // Qualifier - assumes a producer is around... private EntityManager em; @Override public EntityManager resolveEntityManager() { return em; } }
Again, note that annotations on interfaces do not inherit, so it's not possible to create something like a base
CrmRepository
interface with the @EntityManagerConfig
and then extending / implementing this interface.
EntityManager
methods¶While the EntityRepository
methods should cover most interactions normally done with an EntityManager
,
for some specific cases it might still be useful to have one or the other method available. For this case,
it's possible to extend / implement the EntityManagerDelegate
interface for repositories, which offers
most other methods available in a JPA 2.0 EntityManager
:
@Repository public interface PersonRepository extends EntityRepository<Person, Long>, EntityManagerDelegate<Person> { ... }
Good naming is a difficult aspects in software engineering. A good method name usually makes comments unnecessary and states exactly what the method does. And with method expressions, the method name is actually the implementation!
Let's start by looking at a (simplified for readability) example:
@Entity public class Person { @Id @GeneratedValue private Long id; private String name; private Integer age; private Gender gender; } @Repository public interface PersonRepository extends EntityRepository<Person, Long> { List<Person> findByNameLikeAndAgeBetweenAndGender(String name, int minAge, int maxAge, Gender gender); }
Looking at the method name, this can easily be read as query all Persons which have a name like the given name parameter, their age is between a min and a max age and having a specific gender. The DeltaSpike Data module can translate method names following a given format and directly generate the query implementation out of it (in EBNF-like form):
(Entity|List<Entity>) findBy(Property[Comparator]){Operator Property [Comparator]}
Or in more concrete words:
findBy
keyword (or related findOptionalBy
, findAnyBy
).And
or Or
.Other assumptions taken by the expression evaluator:
Following comparators are currently supported to be used in method expressions:
Name | # of Arguments | Description |
Equal | 1 | Property must be equal to argument value. If the operator is omitted in the expression, this is assumed as default. |
NotEqual | 1 | Property must be not equal to argument value. |
Like | 1 | Property must be like the argument value. Use the %-wildcard in the argument. |
GreaterThan | 1 | Property must be greater than argument value. |
GreaterThanEquals | 1 | Property must be greater than or equal to argument value. |
LessThan | 1 | Property must be less than argument value. |
LessThanEquals | 1 | Property must be less than or equal to argument value. |
Between | 2 | Property must be between the two argument values. |
IsNull | 0 | Property must be null. |
IsNotNull | 0 | Property must be non-null. |
Note that DeltaSpike will validate those expressions during startup, so you will notice early in case you have a typo in those expressions.
Beside comparators it's also possible to sort queries by using the OrderBy
keyword, followed
by the attribute name and the direction (Asc
or Desc
).
@Repository public interface PersonRepository extends EntityRepository<Person, Long> { List<Person> findByLastNameLikeOrderByAgeAscLastNameDesc(String lastName); }
To create a comparison on a nested property, the traversal parts can be separated by a _
:
@Repository public interface PersonRepository extends EntityRepository<Person, Long> { List<Person> findByCompany_companyName(String companyName); }
DeltaSpike supports query options on method expressions. If you want to page a query, you can change the first result as well as the maximum number of results returned:
@Repository public interface PersonRepository extends EntityRepository<Person, Long> { List<Person> findByNameLike(String name, @FirstResult int start, @MaxResults int pageSize); }
In case the findBy
prefix does not comply with your team conventions, this can be adapted:
@Repository(methodPrefix = "fetchWith") public interface PersonRepository extends EntityRepository<Person, Long> { List<Person> fetchWithNameLike(String name, @FirstResult int start, @MaxResults int pageSize); }
While method expressions are fine for simple queries, they will often reach their limit once things get slightly more complex. Another aspect is the way you want to use JPA: The recommended approach using JPA for best performance is over named queries. To help incorporate those use cases, the DeltaSpike Data module supports also annotating methods for more control on the generated query.
The simples way to define a specific query is by annotating a method and providing the JPQL query string which has to be executed. In code, this looks like the following sample:
public interface PersonRepository extends EntityRepository<Person, Long> { @Query("select count(p) from Person p where p.age > ?1") Long countAllOlderThan(int minAge); }
The parameter binding in the query corresponds to the argument index in the method.
You can also refer to a named query which is constructed and executed automatically. The @Query
annotation has a named attribute which corresponds to the query name:
@Entity @NamedQueries({ @NamedQuery(name = Person.BY_MIN_AGE, query = "select count(p) from Person p where p.age > ?1 order by p.age asc") }) public class Person { public static final String BY_MIN_AGE = "person.byMinAge"; ... } @Repository public interface PersonRepository extends EntityRepository<Person, Long> { @Query(named = Person.BY_MIN_AGE) Long countAllOlderThan(int minAge); }
Same as before, the parameter binding corresponds to the argument index in the method. If the named
query requires named parameters to be used, this can be done by annotating the arguments with the
@QueryParam
annotation.
TIP:
Java does not preserve method parameter names (yet), that's why the annotation is needed.
@NamedQuery(name = Person.BY_MIN_AGE, query = "select count(p) from Person p where p.age > :minAge order by p.age asc") ... @Repository public interface PersonRepository extends EntityRepository<Person, Long> { @Query(named = Person.BY_MIN_AGE) Long countAllOlderThan(@QueryParam("minAge") int minAge); }
It is also possible to set a native SQL query in the annotation. The @Query
annotation has a native attribute
which flags that the query is not JPQL but plain SQL:
@Entity @Table(name = "PERSON_TABLE") public class Person { ... } @Repository public interface PersonRepository extends EntityRepository<Person, Long> { @Query(value = "SELECT * FROM PERSON_TABLE p WHERE p.AGE > ?1", isNative = true) List<Person> findAllOlderThan(int minAge); }
Beside providing a query string or reference, the @Query
annotation provides also two more attributes:
@Repository public interface PersonRepository extends EntityRepository<Person, Long> { @Query(named = Person.BY_MIN_AGE, max = 10, lock = LockModeType.PESSIMISTIC_WRITE) List<Person> findAllForUpdate(int minAge); }
Name | Description |
max | Limits the number of results. |
lock | Use a specific LockModeType to execute the query. |
Note that these options can also be applied to method expressions.
All the query options you have seen so far are more or less static. But sometimes you might want
to apply certain query options dynamically. For example, sorting criteria could come from a user
selection so they cannot be known beforehand. DeltaSpike allows you to apply query options at runtime by
using the QueryResult
result type:
@Repository public interface PersonRepository extends EntityRepository<Person, Long> { @Query("select p from Person p where p.age between ?1 and ?2") QueryResult<Person> findAllByAge(int minAge, int maxAge); }
Once you have obtained a QueryResult
, you can apply further options to the query:
List<Person> result = personRepository.findAllByAge(18, 65) .sortAsc(Person_.lastName) .sortDesc(Person_.age) .lockMode(LockModeType.WRITE) .hint("org.hibernate.timeout", Integer.valueOf(10)) .getResultList();
CAUTION:
Note that sorting is only applicable to method expressions or non-named queries. For named queries it might be possible, but is currently only supported for Hibernate, EclipseLink and OpenJPA.
Note that the QueryResult
return type can also be used with method expressions.
We introduced the QueryResult
type in the last chapter, which can also be used for pagination:
// Query API style QueryResult<Person> paged = personRepository.findByAge(age) .maxResults(10) .firstResult(50); // or paging style QueryResult<Person> paged = personRepository.findByAge(age) .withPageSize(10) // equivalent to maxResults .toPage(5); int totalPages = paged.countPages();
While reading entities and updating them one by one might be fine for many use cases, applying bulk
updates or deletes is also a common usage scenario for repositories. DeltaSpike supports this with a special
marking annotation @Modifying
:
@Repository public interface PersonRepository extends EntityRepository<Person, Long> { @Modifying @Query("update Person as p set p.classifier = ?1 where p.classifier = ?2") int updateClassifier(Classifier current, Classifier next); }
Bulk operation query methods can either return void or int, which counts the number of entities affected by the bulk operation.
The JPA spec requires to throw exceptions in case the getSingleResult()
method does either return
no or more than one result. This can result in tedious handling with try-catch blocks or have potential
impact on your transaction (as the RuntimeException
might roll it back).
DeltaSpike Data gives the option to change this to the way it makes most sense for the current usecase. While the default behavior is still fully aligned with JPA, it's also possible to request optional query results.
With this option, the query returns null
instead of throwing a NoResultException
when there is no
result returned. It's usable with method expressions, Query
annotations and QueryResult<E>
calls.
@Repository(forEntity = Person.class) public interface PersonRepository { Person findOptionalBySsn(String ssn); @Query(named = Person.BY_NAME, singleResult = SingleResultType.OPTIONAL) Person findByName(String firstName, String lastName); }
For method expressions, the findOptionalBy
prefix can be used. For @Query
annotations, the singleResult
attribute can be overridden with the SingleResultType.OPTIONAL
enum.
In case the query returns more than one result, a NonUniqueResultException
is still thrown.
If the caller does not really mind what kind if result is returned, it's also possible to request any
result from the query. If there is no result, same as for optional queries null
is returned. In case
there is more than one result, any result is returned, or more concretely the first result out of the
result list.
@Repository(forEntity = Person.class) public interface PersonRepository { Person findAnyByLastName(String lastName); @Query(named = Person.BY_NAME, singleResult = SingleResultType.ANY) Person findByName(String firstName, String lastName); }
For method expressions, the findAnyBy
prefix can be used. For @Query
annotations, the singleResult
attribute can be overridden with the SingleResultType.ANY
enum.
This option will not throw an exception.
If you call any method expression, @Query
-annotated method or a method from the EntityRepository
, the
repository will figure out if a transaction is needed or not, and if so, if there is already one ongoing.
The Data module uses the TransactionStrategy
provided by the JPA Module for this. See the JPA module
documentation for more details.
CAUTION:
Some containers do not support
BeanManagedUserTransactionStrategy
! As JTA has still some portability issues even in Java EE 7, it might be required that you implement your ownTransactionStrategy
. We will think about providing an acceptable solution for this.
CAUTION:
Annotating Repository methods with
@Transactional
is not yet supported, but will follow.
If you need to open a transaction on a concrete repository method, we currently recommend creating an extension
(see next chapter) which uses @Transactional
and might look like the following sample.
public class TxExtension<E> implements TxRepository // this is your extension interface { @Inject private EntityManager em; @Override @Transactional public List<E> transactional(ListResultCallback callback) { return callback.execute(); } }
Repositories can then implement the TxRepository
interface and call their queries in the
transactional
method (where the callback implementation can be e.g. in an anonymous class).
While repositories defines several base interfaces, there might still be the odd convenience method that is missing. This is actually intentional - things should not get overloaded for each and every use case. That's why in DeltaSpike you can define your own reusable methods.
For example, you might want to use the QueryDsl library in your repositories:
import com.mysema.query.jpa.impl.JPAQuery; public interface QueryDslSupport { JPAQuery jpaQuery(); } @Repository(forEntity = Person.class) public interface PersonRepository extends QueryDslSupport { ... }
The first step is to define an interface which contains the extra methods for your repositories (as shown above):
public interface QueryDslSupport { JPAQuery jpaQuery(); }
As a next step, you need to provide an implementation for this interface once. It's also important
that this implementation implements the DelegateQueryHandler
interface (don't worry, this is just
an empty marker interface):
public class QueryDslRepositoryExtension<E> implements QueryDslSupport, DelegateQueryHandler { @Inject private QueryInvocationContext context; @Override public JPAQuery jpaQuery() { return new JPAQuery(context.getEntityManager()); } }
As you see in the sample, you can inject a QueryInvocationContext
which contains utility methods
like accessing the current EntityManager
and entity class.
Note that, if you define multiple extensions with equivalent method signatures, there is no specific order in which the implementation is selected.
While repositories are primarily intended to work with Entities, it might be preferable in some cases to have an additional mapping layer on top of them, e.g. because the Entities are quite complex but the service layer needs only a limited view on it, or because the Entities are exposed over a remote interface and there should not be a 1:1 view on the domain model.
DeltaSpike Data allows to directly plugin in such a mapping mechanism without the need to specify additional mapping methods:
@Repository(forEntity = Person.class) @MappingConfig(PersonDtoMapper.class) public interface PersonRepository { PersonDto findBySsn(String ssn); List<PersonDto> findByLastName(String lastName); }
The PersonDtoMapper
class has to implement the QueryInOutMapper
interface:
public class PersonDtoMapper implements QueryInOutMapper<Person> { @Override public Object mapResult(Person result) { ... // converts Person into a PersonDto } ... @Override public Object mapResultList(List<Simple> result) { ... // result lists can also be mapped into something different // than a collection. } @Override public boolean mapsParameter(Object parameter) { return parameter != null && ( parameter instanceof PersonDto || parameter instanceof PersonId); } @Override public Object mapParameter(Object parameter) { ... // converts query parameters if required } }
The mapper can also be used to transform query parameters. Parameters are converted before executing queries and calling repository extensions.
Note that those mapper classes are treated as CDI Beans, so it is possible to use injection
in those beans (you might e.g. inject an EntityManager
or other mappers). As the @MappingConfig
refers to the mapper class directly, the mapper must be uniquely identifiable by its class.
It's also possible to combine mappings with the base Repository classes:
@Repository(forEntity = Person.class) @MappingConfig(PersonDtoMapper.class) public interface PersonRepository extends EntityRepository<PersonDto, PersonId> { ... }
In this case, the forEntity
attribute in the @Repository
annotation is mandatory. Also it is up
to the mapper to convert parameters correctly (in this example, a conversion from a PersonDto
parameter to Person
entity and from PersonId
to Long
is necessary).
In many cases it's just required to map a DTO object back and forth. For this case, the SimpleQueryInOutMapperBase
class
can be subclassed, which only requires to override two methods:
public class PersonMapper extends SimpleQueryInOutMapperBase<Person, PersonDto> { @Override protected Object getPrimaryKey(PersonDto dto) { return dto.getId(); } @Override protected PersonDto toDto(Person entity) { ... } @Override protected Person toEntity(Person entity, PersonDto dto) { ... return entity; } }
The first method, getPrimaryKey
, identifies the primary key of an incoming DTO (this might need mapping too).
If there is a primary key in the DTO, Data tries to retrieve the Entity and feed it to the toEntity
method,
so the entity to be mapped is attached to the persistence context. If there is no primary key, a new
instance of the Entity is created. In any case, there is no need to map the primary key to the entity (it either
does not exist or is already populated for an existing entity).
Beside automatic query generation, the DeltaSpike Data module also provides a DSL-like API to create JPA 2 Criteria queries. It takes advantage of the JPA 2 meta model, which helps creating type safe queries.
TIP:
The JPA meta model can easily be generated with an annotation processor. Hibernate or EclipseLink provide such a processor, which can be integrated into your compile and build cycle.
Note that this criteria API is not intended to replace the standard criteria API - it's rather a utility API that should make life easier on the most common cases for a custom query. The JPA criteria API's strongest point is certainly its type safety - which comes at the cost of readability. We're trying to provide a middle way here. A less powerful API, but still type safe and readable.
The API is centered around the Criteria class and is targeted to provide a fluent interface to write criteria queries:
@Repository(forEntity = Person.class) public abstract class PersonRepository implements CriteriaSupport<Person> { public List<Person> findAdultFamilyMembers(String name, Integer minAge) { return criteria() .like(Person_.name, "%" + name + "%") .gtOrEq(Person_.age, minAge) .eq(Person_.validated, Boolean.TRUE) .orderDesc(Person_.age) .getResultList(); } }
Following comparators are supported by the API:
Name | Description |
.eq(..., ...) | Property value must be equal to the given value |
.in(..., ..., ..., ...) | Property value must be in one of the given values. |
.notEq(..., ...) | Negates equality |
.like(..., ...) | A SQL `like` equivalent comparator. Use % on the value. |
.notLike(..., ...) | Negates the like value |
.lt(..., ...) | Property value must be less than the given value. |
.ltOrEq(..., ...) | Property value must be less than or equal to the given value. |
.gt(..., ...) | Property value must be greater than the given value. |
.ltOrEq(..., ...) | Property value must be greater than or equal to the given value. |
.between(..., ..., ...) | Property value must be between the two given values. |
.isNull(...) | Property must be `null` |
.isNotNull(...) | Property must be non-`null` |
.isEmpty(...) | Collection property must be empty |
.isNotEmpty(...) | Collection property must be non-empty |
The query result can be modified with the following settings:
Name | Description |
.orderAsc(...) | Sorts the result ascending by the given property. Note that this can be applied to several properties |
.orderDesc(...) | Sorts the result descending by the given property. Note that this can be applied to several properties |
.distinct() | Sets distinct to true on the query. |
Once all comparators and query options are applied, the createQuery()
method is called.
This creates a JPA TypedQuery object for the repository entity. If required, further processing can be applied here.
For simple cases, restricting on the repository entity only works out fine, but once the Data model gets more complicated, the query will have to consider relations to other entities. The module's criteria API therefore supports joins as shown in the sample below:
@Repository public abstract class PersonRepository extends AbstractEntityRepository<Person, Long> { public List<Person> findByCompanyName(String companyName) { return criteria() .join(Person_.company, where(Company.class) .eq(Company_.name, companyName) ) .eq(Person_.validated, Boolean.TRUE) .getResultList(); } }
Beside the inner and outer joins, also fetch joins are supported. Those are slighly simpler as seen in the next sample:
public abstract class PersonRepository extends AbstractEntityRepository<Person, Long> { public Person findBySSN(String ssn) { return criteria() .fetch(Person_.familyMembers) .eq(Person_.ssn, ssn) .distinct() .getSingleResult(); } }
By default, all query operators are concatenated as an and conjunction to the query. The DeltaSpike criteria API also allows to add groups of disjunctions.
public abstract class PersonRepository extends AbstractEntityRepository<Person, Long> { public List<Person> findAdults() { return criteria() .or( criteria(). .gtOrEq(Person_.age, 18) .eq(Person_.origin, Country.SWITZERLAND), criteria(). .gtOrEq(Person_.age, 21) .eq(Person_.origin, Country.USA) ) .getResultList(); } }
It might not always be appropriate to retrieve full entities - you might also be interested in scalar values or by modified entity attributes. The Criteria interface allows this with the selection method:
public abstract class PersonRepository extends AbstractEntityRepository<Person, Long> { public Statistics ageStatsFor(Segment segment) { return criteria() .select(Statistics.class, avg(Person_.age), min(Person_.age), max(Person_.age)) .eq(Person_.segment, segment) .getSingleResult(); } public List<Object[]> personViewForFamily(String name) { return criteria() .select(upper(Person_.name), attribute(Person_.age), substring(Person_.firstname, 1)) .like(Person_.name, name) .getResultList(); } }
There are also several functions supported which can be used in the selection clause:
Name | Description |
abs(...) | Absolute value. Applicable to Number attributes. |
avg(...) | Average value. Applicable to Number attributes. |
count(...) | Count function. Applicable to Number attributes. |
max(...) | Max value. Applicable to Number attributes. |
min(...) | Min value. Applicable to Number attributes. |
modulo(...) | Modulo function. Applicable to Integer attributes. |
neg(...) | Negative value. Applicable to Number attributes. |
sum(...) | Sum function. Applicable to Number attributes. |
lower(...) | String to lowercase. Applicable to String attributes. |
substring(int from, ...) | Substring starting from. Applicable to String attributes. |
substring(int from, int to, ...) | Substring starting from ending to. Applicable to String attributes. |
upper(...) | String to uppercase. Applicable to String attributes. |
currDate() | The DB sysdate. Returns a Date object. |
currTime() | The DB sysdate. Returns a Time object. |
currTStamp() | The DB sysdate. Returns a Timestamp object. |
A common requirement for entities is tracking what is being done with them. DeltaSpike provides a convenient way to support this requirement.
TIP:
DeltaSpike does not support creating revisions of entities. If this is a requirement for your audits, have a look at Hibernate Envers.
DeltaSpike uses an entity listener to update auditing data before entities get created or update. The entity listener must be activated before it can be used. This can either be done globally for all entities of a persistent unit or per entity.
Activation per persistence unit in orm.xml
:
<entity-mappings xmlns="http://java.sun.com/xml/ns/persistence/orm" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://java.sun.com/xml/ns/persistence/orm http://java.sun.com/xml/ns/persistence/orm_2_0.xsd" version="2.0"> <persistence-unit-metadata> <persistence-unit-defaults> <entity-listeners> <entity-listener class="org.apache.deltaspike.data.impl.audit.AuditEntityListener" /> </entity-listeners> </persistence-unit-defaults> </persistence-unit-metadata> </entity-mappings>
Activation per entity:
@Entity @EntityListeners(AuditEntityListener.class) public class AuditedEntity { ... }
Note that for this variant, you need a compile dependency on the impl module. Alternatively, also the per entity listener can be configured by XML.
All that has to be done now is annotating the entity properties which are used to audit the entity.
To keep track on creation and modification times, following annotations can be used:
@Entity public class AuditedEntity { ... @Temporal(TemporalType.TIMESTAMP) @CreatedOn private Date created; @Temporal(TemporalType.TIMESTAMP) @ModifiedOn private Date updated; ... }
In case the modification date should also be set during entity creation, the annotation can be customized:
@ModifiedOn(setOnCreate=true)
Beside keeping track of when a change has happened, it's also often critical to track who's responsible for the change. Annotate a user tracking field with the following annotation:
@Entity public class AuditedEntity { ... @ModifiedBy private String auditUser; ... }
Now a little help is needed. The entity listener needs to be able to resolve the current user - there must be a bean available of the matching type for the annotation property, exposed over a special CDI qualifier:
public class UserProvider { @Inject private User user; @Produces @CurrentUser public String currentUser() { return user.getUsername(); } ... }
TIP:
The JPA Spec does not recommend to modify entity relations from within a lifecycle callback. If you expose another entity here, make sure that your persistence provider supports this. Also you should ensure that the entity is attached to a persistent context. Also, be aware that the CDI container will proxy a scoped bean, which might confuse the persistence provider when persisting / updating the target entity.