I am looking into Spring Data JPA. Consider the below example where I will get all the crud and finder functionality working by default and if I want to customize a finder then that can be also done easily in the interface itself.
#Transactional(readOnly = true)
public interface AccountRepository extends JpaRepository<Account, Long> {
#Query("<JPQ statement here>")
List<Account> findByCustomer(Customer customer);
}
I would like to know how can I add a complete custom method with its implementation for the above AccountRepository? Since its an Interface I cannot implement the method there.
You need to create a separate interface for your custom methods:
public interface AccountRepository
extends JpaRepository<Account, Long>, AccountRepositoryCustom { ... }
public interface AccountRepositoryCustom {
public void customMethod();
}
and provide an implementation class for that interface:
public class AccountRepositoryImpl implements AccountRepositoryCustom {
#Autowired
#Lazy
AccountRepository accountRepository; /* Optional - if you need it */
public void customMethod() { ... }
}
See also:
4.6 Custom Implementations for Spring Data Repositories
Note that the naming scheme has changed between versions. See https://stackoverflow.com/a/52624752/66686 for details.
In addition to axtavt's answer, don't forget you can inject Entity Manager in your custom implementation if you need it to build your queries:
public class AccountRepositoryImpl implements AccountRepositoryCustom {
#PersistenceContext
private EntityManager em;
public void customMethod() {
...
em.createQuery(yourCriteria);
...
}
}
There's a slightly modified solution that does not require additional interfaces.
As specificed in the documented functionality, the Impl suffix allows us to have such clean solution:
Define in you regular #Repository interface, say MyEntityRepository the custom methods (in addition to your Spring Data methods)
Create a class MyEntityRepositoryImpl (the Impl suffix is the magic) anywhere (doesn't even need to be in the same package) that implements the custom methods only and annotate such class with #Component** (#Repository will not work).
This class can even inject MyEntityRepository via #Autowired for use in the custom methods.
Example:
Entity class (for completeness):
package myapp.domain.myentity;
#Entity
public class MyEntity {
#Id private Long id;
#Column private String comment;
}
Repository interface:
package myapp.domain.myentity;
#Repository
public interface MyEntityRepository extends JpaRepository<MyEntity, Long> {
// EXAMPLE SPRING DATA METHOD
List<MyEntity> findByCommentEndsWith(String x);
List<MyEntity> doSomeHql(Long id); // custom method, code at *Impl class below
List<MyEntity> useTheRepo(Long id); // custom method, code at *Impl class below
}
Custom methods implementation bean:
package myapp.infrastructure.myentity;
#Component // Must be #Component !!
public class MyEntityRepositoryImpl { // must have the exact repo name + Impl !!
#PersistenceContext
private EntityManager entityManager;
#Autowired
private MyEntityRepository myEntityRepository;
#SuppressWarnings("unused")
public List<MyEntity> doSomeHql(Long id) {
String hql = "SELECT eFROM MyEntity e WHERE e.id = :id";
TypedQuery<MyEntity> query = entityManager.createQuery(hql, MyEntity.class);
query.setParameter("id", id);
return query.getResultList();
}
#SuppressWarnings("unused")
public List<MyEntity> useTheRepo(Long id) {
List<MyEntity> es = doSomeHql(id);
es.addAll(myEntityRepository.findByCommentEndsWith("DO"));
es.add(myEntityRepository.findById(2L).get());
return es;
}
}
Usage:
// You just autowire the the MyEntityRepository as usual
// (the Impl class is just impl detail, the clients don't even know about it)
#Service
public class SomeService {
#Autowired
private MyEntityRepository myEntityRepository;
public void someMethod(String x, long y) {
// call any method as usual
myEntityRepository.findByCommentEndsWith(x);
myEntityRepository.doSomeHql(y);
}
}
And that's all, no need for any interfaces other than the Spring Data repo one you already have.
The only possible drawbacks I identified are:
The custom methods in the Impl class are marked as unused by the compiler, thus the #SuppressWarnings("unused") suggestion.
You have a limit of one Impl class. (Whereas in the regular fragment interfaces implementation the docs suggest you could have many.)
If you place the Impl class at a different package and your test uses only #DataJpaTest, you have to add #ComponentScan("package.of.the.impl.clazz") to your test, so Spring loads it.
The accepted answer works, but has three problems:
It uses an undocumented Spring Data feature when naming the custom implementation as AccountRepositoryImpl. The documentation clearly states that it has to be called AccountRepositoryCustomImpl, the custom interface name plus Impl
You cannot use constructor injection, only #Autowired, that are considered bad practice
You have a circular dependency inside of the custom implementation (that's why you cannot use constructor injection).
I found a way to make it perfect, though not without using another undocumented Spring Data feature:
public interface AccountRepository extends AccountRepositoryBasic,
AccountRepositoryCustom
{
}
public interface AccountRepositoryBasic extends JpaRepository<Account, Long>
{
// standard Spring Data methods, like findByLogin
}
public interface AccountRepositoryCustom
{
public void customMethod();
}
public class AccountRepositoryCustomImpl implements AccountRepositoryCustom
{
private final AccountRepositoryBasic accountRepositoryBasic;
// constructor-based injection
public AccountRepositoryCustomImpl(
AccountRepositoryBasic accountRepositoryBasic)
{
this.accountRepositoryBasic = accountRepositoryBasic;
}
public void customMethod()
{
// we can call all basic Spring Data methods using
// accountRepositoryBasic
}
}
This is limited in usage, but for simple custom methods you can use default interface methods like:
import demo.database.Customer;
import org.springframework.data.repository.CrudRepository;
public interface CustomerService extends CrudRepository<Customer, Long> {
default void addSomeCustomers() {
Customer[] customers = {
new Customer("Józef", "Nowak", "nowakJ#o2.pl", 679856885, "Rzeszów", "Podkarpackie", "35-061", "Zamknięta 12"),
new Customer("Adrian", "Mularczyk", "adii333#wp.pl", 867569344, "Krosno", "Podkarpackie", "32-442", "Hynka 3/16"),
new Customer("Kazimierz", "Dejna", "sobieski22#weebly.com", 996435876, "Jarosław", "Podkarpackie", "25-122", "Korotyńskiego 11"),
new Customer("Celina", "Dykiel", "celina.dykiel39#yahoo.org", 947845734, "Żywiec", "Śląskie", "54-333", "Polna 29")
};
for (Customer customer : customers) {
save(customer);
}
}
}
EDIT:
In this spring tutorial it is written:
Spring Data JPA also allows you to define other query methods by
simply declaring their method signature.
So it is even possible to just declare method like:
Customer findByHobby(Hobby personHobby);
and if object Hobby is a property of Customer then Spring will automatically define method for you.
Im using the following code in order to access generated find methods from my custom implementation. Getting the implementation through the bean factory prevents circular bean creation problems.
public class MyRepositoryImpl implements MyRepositoryExtensions, BeanFactoryAware {
private BrandRepository myRepository;
public MyBean findOne(int first, int second) {
return myRepository.findOne(new Id(first, second));
}
public void setBeanFactory(BeanFactory beanFactory) throws BeansException {
myRepository = beanFactory.getBean(MyRepository.class);
}
}
Considering your code snippet, please note that you can only pass Native objects to the findBy### method, lets say you want to load a list of accounts that belongs certain costumers, one solution is to do this,
#Query("Select a from Account a where a."#nameoffield"=?1")
List<Account> findByCustomer(String "#nameoffield");
Make sue the name of the table to be queried is thesame as the Entity class.
For further implementations please take a look at this
If you want to be able to do more sophisticated operations you might need access to Spring Data's internals, in which case the following works (as my interim solution to DATAJPA-422):
public class AccountRepositoryImpl implements AccountRepositoryCustom {
#PersistenceContext
private EntityManager entityManager;
private JpaEntityInformation<Account, ?> entityInformation;
#PostConstruct
public void postConstruct() {
this.entityInformation = JpaEntityInformationSupport.getMetadata(Account.class, entityManager);
}
#Override
#Transactional
public Account saveWithReferenceToOrganisation(Account entity, long organisationId) {
entity.setOrganisation(entityManager.getReference(Organisation.class, organisationId));
return save(entity);
}
private Account save(Account entity) {
// save in same way as SimpleJpaRepository
if (entityInformation.isNew(entity)) {
entityManager.persist(entity);
return entity;
} else {
return entityManager.merge(entity);
}
}
}
There is another issue to be considered here. Some people expect that adding custom method to your repository will automatically expose them as REST services under '/search' link. This is unfortunately not the case. Spring doesn't support that currently.
This is 'by design' feature, spring data rest explicitly checks if method is a custom method and doesn't expose it as a REST search link:
private boolean isQueryMethodCandidate(Method method) {
return isQueryAnnotationPresentOn(method) || !isCustomMethod(method) && !isBaseClassMethod(method);
}
This is a qoute of Oliver Gierke:
This is by design. Custom repository methods are no query methods as
they can effectively implement any behavior. Thus, it's currently
impossible for us to decide about the HTTP method to expose the method
under. POST would be the safest option but that's not in line with the
generic query methods (which receive GET).
For more details see this issue: https://jira.spring.io/browse/DATAREST-206
I liked Danila's solution and started using it but nobody else on the team liked having to create 4 classes for each repository. Danila's solution is the only one here that let's you use the Spring Data methods in the Impl class. However, I found a way to do it with just a single class:
public interface UserRepository extends MongoAccess, PagingAndSortingRepository<User> {
List<User> getByUsername(String username);
default List<User> getByUsernameCustom(String username) {
// Can call Spring Data methods!
findAll();
// Can write your own!
MongoOperations operations = getMongoOperations();
return operations.find(new Query(Criteria.where("username").is(username)), User.class);
}
}
You just need some way of getting access to your db bean (in this example, MongoOperations). MongoAccess provides that access to all of your repositories by retrieving the bean directly:
public interface MongoAccess {
default MongoOperations getMongoOperations() {
return BeanAccessor.getSingleton(MongoOperations.class);
}
}
Where BeanAccessor is:
#Component
public class BeanAccessor implements ApplicationContextAware {
private static ApplicationContext applicationContext;
public static <T> T getSingleton(Class<T> clazz){
return applicationContext.getBean(clazz);
}
public static <T> T getSingleton(String beanName, Class<T> clazz){
return applicationContext.getBean(beanName, clazz);
}
#Override
public void setApplicationContext(ApplicationContext applicationContext) throws BeansException {
BeanAccessor.applicationContext = applicationContext;
}
}
Unfortunately, you can't #Autowire in an interface. You could autowire the bean into a MongoAccessImpl and provide a method in the interface to access it, but Spring Data blows up. I don't think it expects to see an Impl associated even indirectly with PagingAndSortingRepository.
I faced with this using mongo and spring. So let's assume we use MongoRepository to provided base crud operations, and let's say we need to implement some custom criteria query operation using mongoTemplate. To achieve one interface to inject repository for crud and custom we need to specify:
Custom interface:
public interface UserCustomRepository {
List<User> findAllUsersBySomeCriteria(UserCriteriaRequest criteriaRequest);
}
UserRepository interface 'must' first extends UserCustomRepository and then MongoRepository
#Repository
public interface UserRepository extends UserCustomRepository, MongoRepository<User, ObjectId> {
}
UserRepositoryImpl must have the same name as what crud interface with *Impl suffix.
#Component
#NoArgsConstructor
#AllArgsConstructor(onConstructor = #__(#Autowired))
public class UserRepositoryImpl implements UserCustomRepository {
private MongoTemplate mongoTemplate;
#Override
public List<User> findAllUsersBySomeCriteria(UserCriteriaRequest criteriaRequest){
//some impl
}
}
Let's impl some service - here we inject only UserRepository interface and use methods from crud repository and custom class impl.
#Service
#NoArgsConstructor
#AllArgsConstructor(onConstructor = #__(#Autowired))
public class UserService {
private UserRepository userReposityry;
public List<User> getUserByCriteria(UserCriteriaRequest request) {
userRepository.findById(request.getUserId); // Crud repository method
userRepository.findAllUsersBySomeCriteria(request); // custom method.
}
}
I extends the SimpleJpaRepository:
public class ExtendedRepositoryImpl<T extends EntityBean> extends SimpleJpaRepository<T, Long>
implements ExtendedRepository<T> {
private final JpaEntityInformation<T, ?> entityInformation;
private final EntityManager em;
public ExtendedRepositoryImpl(final JpaEntityInformation<T, ?> entityInformation,
final EntityManager entityManager) {
super(entityInformation, entityManager);
this.entityInformation = entityInformation;
this.em = entityManager;
}
}
and adds this class to #EnableJpaRepositoryries repositoryBaseClass.
I use SimpleJpaRepository as the base class of repository implementation and add custom method in the interface,eg:
public interface UserRepository {
User FindOrInsert(int userId);
}
#Repository
public class UserRepositoryImpl extends SimpleJpaRepository implements UserRepository {
private RedisClient redisClient;
public UserRepositoryImpl(RedisClient redisClient, EntityManager em) {
super(User.class, em);
this.redisClient = redisClient;
}
#Override
public User FindOrInsert(int userId) {
User u = redisClient.getOrSet("test key.. User.class, () -> {
Optional<User> ou = this.findById(Integer.valueOf(userId));
return ou.get();
});
…………
return u;
}
I am learning to use JPA. And I'm a little confused.
According JPA EntityManager manages transactions. But a design pattern is to inject the EntityManager in DAOs. So how is possible that are different EntityManager to the same transaction?
This is the case I want to solve
I have the DAOs defined
#Repository
JPARepository1 {
#PersistenceContext
protected EntityManager em;
....
.
#Repository
JPARepository2 {
#PersistenceContext
protected EntityManager em;
....
I have a Service
#Service
public class ServiceImpl1 {
#Autowired
private JPARepository1 repo1;
#Autowired
private JPARepository2 repo2;
public void mainMethod(){
Object o= transactionalMethod1();
try{
transactionalMethod2(o);
}catch (Exception e){
transactionalMethod3(o);
}
}
private Object transactionalMethod1(){
....
}
private void transactionalMethod2(Object o){
....
}
private void transactionalMethod3(Object o){
....
}
Then from #Controller I will invoke mainMethod().
What would be the right way to do transactional to transactionalMethod1, transactionalMethod2 and transactionalMethod3,within the same Service and using the same Repository's.
I would like it if there is an exeption in transactionalMethod2, this abort the transaction, but keep the transactions of transactionalMethod1 and transactionalMethod3
Thanks, sorry for my English
Usually you configure one EntityManager, so the wired manager is always the same, the one you configured. The instance of this manager though, is different in every wiring.
So, every transaction in your service uses a different instance of the EntityManager and thus every transaction invoked is seperated from each other.
As so, an exception in transactionalMethod2 doesn't necessarily affects the transactionalMethod1 and transactionalMethod3
What would be the right way to do transactional to transactionalMethod1, transactionalMethod2 and transactionalMethod3,within the same Service and using the same Repository's.
Now, you have two options to do service methods transactions
1) You could annotate your whole #Service like that:
#Service
#Transactional
public class ServiceImpl1 {
....
so every method declared here is also a transaction.
2) You could annotate each method as #Transactional:
#Transactional
private Object transactionalMethod1(){
....
}
#Transactional
private void transactionalMethod2(Object o){
....
}
#Transactional
private void transactionalMethod3(Object o){
....
}
If you want to use a single repository just #Autowired a single one and use it in your #Transactional method. E.g:
#Service
#Transactional
public class ServiceImpl1 {
#Autowired
private JPARepository1 repo1;
public void mainMethod(){
Object o= transactionalMethod1();
try{
transactionalMethod2(o);
}catch (Exception e){
transactionalMethod3(o);
}
}
private Object transactionalMethod1(){
return repo1.findOne();
}
private void transactionalMethod2(Object o){
repo1.create(o);
}
private void transactionalMethod3(Object o){
repo1.delete(o)
}
Because I have some generics which need to be passed to the newly produced object, I am creating a producer. But while producer works, the EntityManager is not injected because producer creates an instance with operator new instead of using CDI.
How can I produce an object with CDI support?
The code:
Qualifier:
#Qualifier
#Retention(RUNTIME)
#Target(
{ FIELD, TYPE, METHOD })
public #interface Multiselector
{
Class<? extends Dbo> clazz();
}
Producer:
#SessionScoped
public class MultiselectorProducer implements Serializable
{
#Produces
#Multiselector(clazz = SpecialDbo.class)
public MultiselectorService<SpecialDbo> produce()
{
return new MultiselectorService<SpecialDbo>(SpecialDbo.class);
}
}
Service class:
#Stateful
#LocalBean
public class MultiselectorService<T extends Dbo> implements Serializable
{
#Inject
private EntityManager em;
private List<T> itemList;
public MultiselectorService()
{
}
public MultiselectorService(Class<? extends Dbo> clazz)
{
itemList = em.createQuery("some Sql String", clazz);
}
....
}
NOTE: The EntityManager is a custom crud service which is otherwise injected correctly
Any improvement suggestions over the code are welcome. Thanks!
You have mixed a lot of unrelated things:
Your service MultiselectorService is an EJB, and you cannot produce it with a producer. EJB is registered once application is created and then depending on the scope it creates instances.
You have a method public void MultiselectorService(Class<? extends Dbo> clazz) with the name similar to constructor, it against convention.
Assume that you have fixed that method to be a constructor, but then line 'itemList = em.createQuery("some Sql String", clazz);' will fail with NPE. Because em will be initialized only after bean creation. there are two ways to do it:
Inject entity manager into constructor (this is against EJB spec, if you will still use EJB)
Execute initialization operation in method with annotation '#PostConstruct'
Do you have a producer for EntityManager ? e.g. you cannot just inject entity manager, you need to provide as a resource for EJB, with annotation #PersistenceContext
I understand what you try to achieve. The main problem is that manually created beans are not managed by container, this means that interceptors and decorators won't apply (e.g. PostConstruct and Transactional annotation will not work). Check here. So far the best way to achieve this is:
public interface SpecialDboMultiselectorService extends MultiselectorService {
}
#Stateless
public class SpecialDboMultiselectorServiceImpl extends MultiselectorServiceImpl<SpecialDbo> implements SpecialDboMultiselectorService
{
public SpecialDboMultiselectorServiceImpl() {
super(SpecialDbo.class);
}
}
public class MultiselectorServiceImpl<T extends Dbo> implements MultiselectorService {
#Inject
private EntityManager em;
private Class<? extends Dbo> clazz;
private List<T> itemList;
public MultiselectorService(Class<? extends Dbo> clazz) {
this.clazz = clazz
}
#PostConstruct
public void init() {
itemList = em.createQuery("some Sql String", clazz);
}
}
and inject SpecialDboMultiselectorService.
In a project of mine I'm trying to switch management of my persistence from application to container. I'm following these instructions: http://docs.oracle.com/javaee/6/tutorial/doc/gkhrg.html
I've read about EntityManager not being thread safe and just want to make sure that my setup is correct. My concern: http://weblogs.java.net/blog/2005/12/19/dont-use-persistencecontext-web-app.
I have a class that produces a persistence context.
#Singleton
public class JpaResourceProducer {
//The "pu" unit is defined with transaction-type="JTA"
#Produces
#PersistenceUnit(unitName = "pu")
#Database
EntityManagerFactory databasePersistenceUnit;
#Produces
#PersistenceContext(unitName = "pu")
#Database
EntityManager databaseEntityManager;
/* Alternative
#PersistenceContext(unitName = "pu")
private EntityManager em;
#Produces
#UserDatabase
public EntityManager create() {
return em;
}
public void close(#Disposes #Database EntityManager em) {
em.close();
}
*/
}
Then I have a jax-rs resource that injects a DAO.
#RequestScoped
#Path("/endpoint")
public class MyResource {
#Inject private Dao dao;
#GET
#Produces({MediaType.APPLICATION_JSON})
public Converter get() {
MyEntity entity = dao.find(1);
Converter converter = new Converter(entity);
return converter;
}
}
And finally a DAO where I inject the EntityManager.
#Singleton
public class JpaDao<T, K extends Serializable> implements Dao<T, K> {
protected Class<T> entityClass;
#Inject
#Database
EntityManager em;
public JpaDao() {
ParameterizedType genericSuperclass = (ParameterizedType) getClass().getGenericSuperclass();
this.entityClass = (Class<T>) genericSuperclass.getActualTypeArguments()[0];
}
public T find(K id) {
return em.find(entityClass, id);
}
....
}
1. Is this a good setup in terms of thread safety and overall performance?
Bonus questions:
In the JpaResourceProducer I have an alternative setup for the EntityManager where I manually close the manager on dispose.
2. Is the container handling close of my EntityManager automatically?
The example from Oracle includes an EntityManagerFactory.
3. Do I really need an EntityManagerFactory when I'm using CMP?
What's your container? From annotations used I presume it is at least Java EE 6 Web Profile compatible, and in that case you're overcomplicating this.
The tutorial presents producting of multiple persistence units. Is this the case in your application? If not, plain injection into EJBs would be more relieable and less magical - I am also unsure about transactional behaviour of a singleton-produced persistence context.
Singleton JpaDAO means, there can be only one database operation at a time in your application, so this is serious performance bottleneck. It should be Stateless instead (and having its persistence context injected by traditional means).
In general, implement your DAOs and Business logic, even your JAX-RS endpoints as session EJBs, and let container handle concurrency and transactions properly.
Also use other sources of examples, e. g. Adam Bien's JavaEE Patterns.