Copyright © 2016-2018

Copies of this document may be made for your own use and for distribution to others, provided that you do not charge any fee for such copies and further provided that each copy contains this Copyright Notice, whether distributed in print or electronically.

1. Introduction

The Holon JPA Datastore is the Java Persistence API reference implementation of the Datastore API.

See the Datastore documentation for further information about the Datastore API.

The JPA Datastore uses the EntityManagerFactory API as a factory to obtain EntityManager API instances to perform data access and management operations through the standard JPA JPQL language and the EntityManager API operations.

The Holon Platform JPA modules and components require JPA version 2.0 or higher, so a JPA 2.0+ compliant persistence provider (ORM) is needed at runtime.

1.1. Sources and contributions

The Holon Platform JPA Datastore module source code is available from the GitHub repository https://github.com/holon-platform/holon-datastore-jpa.

See the repository README file for information about:

  • The source code structure.

  • How to build the module artifacts from sources.

  • Where to find the code examples.

  • How to contribute to the module development.

2. Obtaining the artifacts

The Holon Platform uses Maven for projects build and configuration. All the platform artifacts are published in the Maven Central Repository, so there is no need to explicitly declare additional repositories in your project pom file.

At the top of each section of this documentation you will find the Maven coordinates (group id, artifact id and version) to obtain the artifact(s) as a dependency for your project.

A BOM (Bill Of Materials) pom is provided to import the available dependencies for a specific version in your projects. The Maven coordinates for the core BOM are the following:

Maven coordinates:

<groupId>com.holon-platform.jpa</groupId>
<artifactId>holon-datastore-jpa-bom</artifactId>
<version>5.2.1</version>

The BOM can be imported in a Maven project in the following way:

<dependencyManagement>
  <dependencies>
    <dependency>
      <groupId>com.holon-platform.jpa</groupId>
      <artifactId>holon-datastore-jpa-bom</artifactId>
      <version>5.2.1</version>
      <type>pom</type>
      <scope>import</scope>
    </dependency>
  </dependencies>
</dependencyManagement>

2.1. Using the Platform BOM

The Holon Platform provides an overall Maven BOM (Bill of Materials) to easily obtain all the available platform artifacts.

3. What’s new in version 5.2.x

  • A basic support for database locks is now available. See Lock support.

  • The new JpaTransactionFactory interface is now available to customize the JPA Datastore transactions lifecycle. It can be configured using the JPA Datastore builder.

  • Support for JDK 9+ module system using Automatic-Module-Name.

4. What’s new in version 5.1.x

5. JPA Datastore

The holon-datastore-jpa artifact is the main entry point to use the JPA Datastore API implementation.

Maven coordinates:

<groupId>com.holon-platform.jpa</groupId>
<artifactId>holon-datastore-jpa</artifactId>
<version>5.2.1</version>

The JpaDatastore interface represents the JPA Datastore API implementation, extending the core Datastore API.

The JpaDatastore API, besides the standard Datastore API operations, provides methods to:

  • Create and configure a JpaDatastore API instance, using the provided builder.

  • Directly working with JpaDatastore managed EntityManagers, with a consistent and integrated EntityManager instance and related persistence context lifecycle management. See JpaDatastore API.

If you want to reach the goal of a complete abstraction from the persistence store technology and the persistence model, the core Datastore API interface should be used instead of the specific JpaDatastore API by your application code. This way, the concrete Datastore API implementation may be replaced by a different one at any time, without any change to the codebase.

5.1. Setup and configuration

To create a JPA Datastore instance, the builder() static method of the JpaDatastore API can be used to obtain the JPA Datastore builder API.

The JPA Datastore builder provides a JpaDatastore instance:

JpaDatastore datastore = JpaDatastore.builder() (1)
    // Datastore configuration omitted
    .build();
1 Obtain the JPA Datastore builder to configure and create a new JPA Datastore instance

But, as stated in the previous section, to reach the goal of a complete abstraction from the persistence store technology and the persistence model, the core Datastore API interface should be used by your application code, instead of the specific JpaDatastore API.

So you can simply obtain the JPA Datastore implementation as a core Datastore API implementation:

Datastore datastore = JpaDatastore.builder() (1)
    // Datastore configuration omitted
    .build();
1 Obtain the JPA Datastore builder to configure and create a new JPA Datastore instance and expose it as core Datastore API type

5.1.1. Common Datastore configuration options

The JPA Datastore builder API extends the core Datastore builder API, which provides common Datastore configuration settings as listed below.

Builder method Arguments Description

dataContextId

The data context id String value

Set the data context id to which the Datastore is bound. Can be used, for example, to declare configuration properties for multiple Datastore instances.

traceEnabled

true or false

Whether to enable Datastore operations tracing. When enabled, the JPA Datastore will log any JPQL operation performed using the Datastore API.

configuration

A DatastoreConfigProperties instance

Set the DatastoreConfigProperties type configuration property set instance to use in order to read the Datastore configuration properties. See the Datastore configuration documentation section for details. This configuration properties can be used as an alternative for the programmatic configuration performed with the previous builder methods.

Example of base JPA Datastore configuration:

Datastore datastore = JpaDatastore.builder()
    // DataSource configuration omitted
    .dataContextId("mydataContextId") (1)
    .traceEnabled(true) (2)
    .build();
1 Set a data context id for the Datastore
2 Activate operations tracing in log

The configuration properties can also be provided through an external configuration property source, using the properties provided by the DatastoreConfigProperties property set.

For example, supposing to have a properties file named datastore.properties like this:

holon.datastore.trace=true

We can use it as configuration property source to enable the Datastore tracing mode:

Datastore datastore = JpaDatastore.builder()
    // DataSource configuration omitted
    .configuration(DatastoreConfigProperties.builder().withPropertySource("datastore.properties").build()) (1)
    .build();
1 Use the datastore.properties file as configuration property source

5.1.2. EntityManagerFactory configuration

The JPA Datastore implementation relies on the standard JPA EntityManagerFactory API to obtain and manage the EntityManager instances to be used to perform the Datastore operations.

The EntityManagerFactory reference is the only required JPA Datastore configuration attribute and can be provided using the JPA Datastore builder API.

EntityManagerFactory entityManagerFactory = createOrObtainEntityManagerFactory();

Datastore datastore = JpaDatastore.builder() //
    .entityManagerFactory(entityManagerFactory) (1)
    .build();
1 Set the EntityManagerFactory instance to use

5.1.3. ORM platform configuration

For some internal operations, the JPA Datastore needs to know the ORM platform which manages the concrete EntityManagerFactory (and accorrdingly the EntityManager) instances (for example Hibernate, Eclipselink and so on). The ORM platform is used, for example, to auto-detect the ORM dialect to use if not directly specified at JPA Datastore configuration time.

The ORM platform can be explicitly specified using the JPA Datastore platform builder method, using the ORMPlatform enumeration.

EntityManagerFactory entityManagerFactory = createOrObtainEntityManagerFactory();

Datastore datastore = JpaDatastore.builder() //
    .entityManagerFactory(entityManagerFactory) //
    .platform(ORMPlatform.HIBERNATE) (1)
    .build();
1 Set Hibernate as ORM platform

ORM platform auto detection:

When the ORM platform is not explicitly specified, the JPA Datastore will auto-detect it using the configured EntityManagerFactory.

5.1.4. ORM Dialect configuration

To ensure operations consistency and efficiency, the JPA Datastore uses the ORMDialect API abstraction in order to resolve each ORM platform specificity and JPQL language difference.

Normally, the ORMDialect implementation to use is auto-detected by the JPA Datastore, relying on the ORM platform to which the EntityManagerFactory instance is bound, as described in the previous section.

The ORMDialect implementation to use can be also explicitly configured using the JPA Datastore builder. This can be done in two ways:

1. ORMDialect configuration using the JPA Datastore builder:

The ORM dialect to use can be directly configured using the JPA Datastore builder, either using the dialect(ORMDialect dialect) or the dialect(String dialectClassName) method.

The ORMDialect API provides a static methods to obtain the ORM dialect for a specific ORM platform or to directly obtain a specific dialect implementation within the available ones.

EntityManagerFactory entityManagerFactory = createOrObtainEntityManagerFactory();

Datastore datastore = JpaDatastore.builder() //
    .entityManagerFactory(entityManagerFactory) //
    .dialect(ORMDialect.hibernate()) (1)
    .dialect("com.holonplatform.datastore.jpa.dialect.HibernateDialect") (2)
    .build();
1 Configure the ORM dialect using a specific dialect implementation
2 Configure the ORM dialect using the dialect class name

1. ORMDialect configuration using a configuration property:

The ORM dialect to use can be also configured using the default holon.datastore.dialect Datastore configuration property, available from the DatastoreConfigProperties property set.

The fully qualified dialect class name must be provided as property value.

For example, supposing to have a properties file named datastore.properties like this:

holon.datastore.dialect=com.holonplatform.datastore.jpa.dialect.HibernateDialect
holon.datastore.trace=true

We can use it as configuration property source to configure the JPA Datastore dialect:

EntityManagerFactory entityManagerFactory = createOrObtainEntityManagerFactory();

Datastore datastore = JpaDatastore.builder() //
    .entityManagerFactory(entityManagerFactory) //
    .configuration(DatastoreConfigProperties.builder().withPropertySource("datastore.properties").build()) (1)
    .build();
1 Use the datastore.properties file as configuration property source

5.1.5. Builtin ORM dialects

The Holon JPA Datastore module provides a set of builtin ORM dialects for the most common ORM platforms. The currently available ORM dialect implementations are:

ORM platform Dialect class

Hibernate

HibernateDialect

Eclipselink

EclipselinkDialect

Apache OpenJPA

OpenJPADialect

Datanucleus

DatanucleusDialect

5.1.6. Auto-flush mode

The JPA Datastore API can be configured to automatically flush the persistence context into the concrete persistence source when a data management operation of the Datastore API (such as save or delete) is performed.

When the auto-flush mode is enabled, the EntityManager.flush() operation is invoked just after the Datastore API operation execution, forcing the persistence context synchronization to the underlying database.

EntityManagerFactory entityManagerFactory = createOrObtainEntityManagerFactory();

Datastore datastore = JpaDatastore.builder() //
    .entityManagerFactory(entityManagerFactory) //
    .autoFlush(true) (1)
    .build();
1 Enable the auto-flush mode for the JPA Datastore API instance

5.1.7. EntityManager lifecyle

The default JPA Datastore strategy to handle the EntityManager lifecycle is defined as follows:

  • For each Datastore API operation, a new EntityManager instance is obtained from the configured EntityManagerFactory, using the createEntityManager() method, before the operation execution. The EntityManager instance is then closed, using the EntityManager.close() method, right after the operation execution.

  • If the Datastore API operation is executed within a Datastore managed transaction (see Transactions management), the EntityManager instance is created at transaction start, shared between any transaction operation and closed only at transaction end.

To customize the EntityManager lifecycle, the EntityManagerInitializer and EntityManagerFinalizer interfaces can be used to control the EntityManager instance creation and/or the EntityManager instance finalization for any JPA Datastore API operation.

The EntityManagerInitializer and EntityManagerFinalizer to use can be configured using the JPA Datastore builder API.

EntityManagerFactory entityManagerFactory = createOrObtainEntityManagerFactory();

Datastore datastore = JpaDatastore.builder() //
    .entityManagerFactory(entityManagerFactory) //
    .entityManagerInitializer(emf -> emf.createEntityManager()) (1)
    .entityManagerFinalizer(em -> em.close()) (2)
    .build();
1 Configure a custom EntityManagerInitializer
2 Configure a custom EntityManagerFinalizer

5.2. Data model attributes naming conventions

Since the JPA Datastore implementation relies on the JPA architecture and APIs to manage the concrete data model and persistence source, the JPA data model definition, i.e. the relational schema mapping through Java objects using entity beans representations, is used to resolve the Datastore API operations meta-language into valid JPA operation definitions, for example using the JPQL language.

This means that the JPA data model (the JPA entity beans definitions) has to be defined and made available to the EntityManagerFactory used by the JPA Datastore to allow the Datastore API operations execution, for example using a persistence.xml file. This is the convention used by JPA, and it does not directly concern the JPA Datastore implementation. See the JPA reference documentation for details.

For this reason, the JPA Datastore implementation relies on the following conventions regarding *DataTarget*s and *Path*s definitions:

  • The DataTarget name is interpreted as the JPA entity name (i.e. the simple entity class name or the name specified using the name attribute of the javax.persistence.Entity annotation).

  • The Path name is interpreted as a JPA entity attribute name, supporting nested classes through the conventional dot notation (e.g. address.city).

For example, given a table definition as follows:

create table test (
  code numeric(20) primary key,
  text varchar(100)
)

And the corresponding JPA entity mapping definition:

@Entity
public class Test {

  @Id
  @Column(name = "code")
  private Long id;

  @Column(name = "text")
  private String value;

  // getters and setters omitted

}

According to the naming conventions described above, a property model and a DataTarget for the example JPA entity can be defined as follows:

public static final NumericProperty<Long> ID = NumericProperty.longType("id"); (1)
public static final StringProperty VALUE = StringProperty.create("value"); (2)

public static final PropertySet<?> TEST = PropertySet.of(ID, VALUE);

public static final DataTarget<?> TARGET = DataTarget.named("Test"); (3)
1 Path property definition for the id JPA entity attribute name
2 Path property definition for the value JPA entity attribute name
3 DataTarget definition for the Test JPA entity name

If the JPA entity name is specified using the name attribute of the javax.persistence.Entity annotation, that name has to be used as DataTarget name.

For example, if myName is specified as JPA entity name:

@Entity(name = "myName")

The DataTarget definition will be:

DataTarget<?> TARGET = DataTarget.named("myName");
See JpaTarget to learn how to use the JPA entity class definition rather than the JPA entity name to declare a DataTarget.

Now the property model definition can be used with the Datastore API operations in an abstract and implementation-independent way. For example:

Datastore datastore = getJpaDatastore();

Stream<Long> ids = datastore.query().target(TARGET).filter(ID.gt(0L)).sort(VALUE.asc()).stream(ID); (1)
Stream<PropertyBox> results = datastore.query().target(TARGET).filter(ID.gt(0L)).sort(VALUE.asc()).stream(TEST); (2)

PropertyBox valueToSave = PropertyBox.builder(TEST).set(ID, 1L).set(VALUE, "test").build();
datastore.save(TARGET, valueToSave); (3)
1 Perform a query and obtain a stream of ID property values
2 Perform a query and obtain a stream of PropertyBox values using the TEST property set, which corresponds to the Test JPA entity attributes definitions
3 Insert or update a value (which corresponds to a JPA Test instance) using the save operation
See Using the Bean introspection APIs to obtain a JPA entity Property model to learn how to create a property model definition for a JPA entity class using the Holon Platform Bean introspection APIs.

5.3. JpaTarget

The JpaTarget interface can be used to define a DataTarget using a JPA entity class definition rather than the entity name.

The entity name used in persistence operations will be the entity class simple name or the name specified using the name attribute of the javax.persistence.Entity annotation, if available.

Fo example, for the Test JPA entity class defined in the example of the previous section, a JpaTarget can be created in the following way:

JpaTarget<Test> TARGET = JpaTarget.of(Test.class);

The JpaTarget path type will be the JPA entity class. A JpaTarget definition can be used with Datastore API operations just like any other DataTarget definition.

5.3.1. Using a JPA entity class as Query projection

By default, the Datastore API relies on the PropertyBox type to represent and handle a set of data model attribute values. For this reason, the PropertyBox type is the default type used in query projections to obtain a set of property values.

The JpaTarget definition can be also used as query projection, to obtain results as JPA entity class instances, instead of PropertyBox type instances.

If you want to reach the goal of a complete abstraction from the persistence store technology and the persistence model, the PropertyBox type should be used to represent a set of data model attributes values. This way, the concrete Datastore API implementation may be replaced by a different one at any time, without any change to the codebase.
public static final JpaTarget<Test> JPA_TARGET = JpaTarget.of(Test.class); (1)

void query() {
  Datastore datastore = getJpaDatastore();

  List<Test> resultEntities = datastore.query(JPA_TARGET).filter(ID.gt(0L)).list(JPA_TARGET); (2)
}
1 JpaTarget definition using the Test JPA entity class
2 Perform a query and obtain the results as Test JPA entity instances, using the JpaTarget as query projection

5.4. Using the Bean introspection APIs to obtain a JPA entity Property model

The Holon Platform Bean introspection API can be used to obtain a property model definition from any Java Bean class, including a JPA entity class.

The BeanPropertySet obtained using the bean introspection API can be used to refer to the JPA entity attributes as properties for Datastore API operations definition and execution.

static //
@Entity public class Test {

  public static final BeanPropertySet<Test> PROPERTIES = BeanPropertySet.create(Test.class); (1)

  public static final DataTarget<Test> TARGET = JpaTarget.of(Test.class); (2)

  @Id
  private Long id;

  private String value;

  // getters and setters omitted

}

void operations() {
  Datastore datastore = getJpaDatastore();

  PropertyBox value = PropertyBox.builder(Test.PROPERTIES).set(Test.PROPERTIES.property("id"), 1L)
      .set(Test.PROPERTIES.property("value"), "test").build(); (3)
  datastore.save(Test.TARGET, value);

  List<PropertyBox> values = datastore.query(Test.TARGET).filter(Test.PROPERTIES.property("id").gt(0L))
      .sort(Test.PROPERTIES.property("value").desc()).list(Test.PROPERTIES); (4)
}
1 Create a BeanPropertySet for the Test JPA entity class
2 Create a DataTarget for the Test JPA entity class
3 The BeanPropertySet can be used to obtain the property definitions which corresponds to the JPA entity class attributes
4 Perform a query using the property model provided by the BeanPropertySet

The JPA Datastore API implementation automatically registers the JPA bean post processors provided by the Holon Platform JPA Module, used to inspect standard JPA entity annotations (for example @Id, @EmbeddedId and @Transient) and configure the bean properties accordingly.

See the Holon Platform JPA Module documentation for details.

5.5. JPA Write options

The data manipulation Datastore API operations supports a set of write options to provide additional operation configuration attributes or behavior specifications.

Besides the standard write options, the JPA Datastore API implementation supports an additional FLUSH, available from the JpaWriteOption enumeration.

This write option can be provided a Datastore API data manipulation operation to synchronize the persistence context to the underlying database right after the operation execution, automatically invoking the EntityManager.flush() method on the current EntityManager instance.

PropertyBox valueToSave = PropertyBox.builder(TEST).set(ID, 1L).set(VALUE, "test").build();

getJpaDatastore().save(TARGET, valueToSave, JpaWriteOption.FLUSH); (1)
1 The FLUSH write option is provided to synchronize the persistence context to the underlying database after the operation execution

5.6. Relational expressions

As relational Datastore, the JPA Datastore API supports core relational expressions for data access and manipulation:

1. Sub-query:

The SubQuery interface can be used to represent a sub-query, which can be used in a query definition to express query restrictions (filters) that involve a sub-query as filter operand.

See the core sub query expression documentation for further information on sub query expressions.

2. Alias and Joins:

The RelationalTarget interface can be used to declare alias and joins for a DataTarget expression.

See the core alias and joins documentation for further information on alias and join expressions.

The supported join types are dependent from the JPA API specification version in use. For example, the left join type is only supported by the JPA specification version 2.1 or above.

5.7. Auto-generated ids

The JPA Datastore API supports the retrieving of auto-generated id column values, if supported by the underlying JDBC driver and ORM platform.

The auto-generated id values can be obtained from the OperationResult object, returned by Datastore data manipulation operations, through the getInsertedKeys() and related methods.

Datastore datastore = getJpaDatastore();

PropertyBox value = PropertyBox.builder(TEST).set(VALUE, "test").build();

OperationResult result = datastore.insert(TARGET, value, JpaWriteOption.FLUSH); (1)

Optional<Long> idValue = result.getInsertedKey(ID); (2)
1 Perform an insert operation, using the FLUSH write option to flush the persistence context to the underlyng database after operation execution to ensure the key auto-generation is triggered
2 Get the auto-generated key value which corresponds to the ID property definition, if available

The default BRING_BACK_GENERATED_IDS WriteOption can be provided to the Datastore API operation to bring back any auto-generated key value into the PropertyBox instance which was the subject of the operation, if a corresponding PathProperty (using the path name) is available in the PropertyBox property set.

Datastore datastore = getJpaDatastore();

PropertyBox value = PropertyBox.builder(TEST).set(VALUE, "test").build();

OperationResult result = datastore.insert(TARGET, value, JpaWriteOption.FLUSH,
    DefaultWriteOption.BRING_BACK_GENERATED_IDS); (1)

Long idValue = value.getValue(ID); (2)
1 Perform an insert operation using the BRING_BACK_GENERATED_IDS write option and the FLUSH write option to ensure the id auto-generation is triggered
2 The ID property value of the inserted PropertyBox value is updated with the auto-generated value, if available

5.8. Transactions management

The JPA Datastore API implementation is transactional, so it supports transactions management through the Transactional API, which can be used to manage transactions at a higher level, in an abstract and implementation-independent way.

See the Transactional Datastore documentation section for information on transactions management with the Transactional API.

final Datastore datastore = getJpaDatastore(); // build or obtain a JPA Datastore

datastore.requireTransactional().withTransaction(tx -> { (1)
  PropertyBox value = buildPropertyBoxValue();
  datastore.save(TARGET, value);

  tx.commit(); (2)
});

OperationResult result = datastore.requireTransactional().withTransaction(tx -> { (3)

  PropertyBox value = buildPropertyBoxValue();
  return datastore.save(TARGET, value);

}, TransactionConfiguration.withAutoCommit()); (4)
1 Obtain the Transactional API to execute one or more Datastore operation within a transaction
2 Commit the transaction
3 Obtain the Transactional API to execute the Datastore operation within a transaction and return a value
4 The transaction is configured with the auto commit mode, this way the transaction is automatically committed at the transactional operation end if no error occurred

5.9. Lock support

The JPA Datastore implementation supports base query-level locks through the core LockQuery extension.

The LockQuery API extends the standard Query API and provides the following additional methods to configure the query results row locks:

  • lock(LockMode lockMode, long timeout): Configures the lock mode to use for query execution and allows to specify an optional lock timeout.

  • tryLock(LockMode lockMode, long timeout): Try to perform the query operation using given lock mode and optional timeout, returning true if the lock was successfully acquired or false otherwise.

The LockQuery API is a Datastore commodity, automatically registered in the JPA Datastore implementation. So a LockQuery type implementation can be otained as follows:

Datastore datastore = getJpaDatastore();

LockQuery lockQuery = datastore.create(LockQuery.class); (1)
1 Obtain a new LockQuery implementation as a Datastore commodity
See the Datastore commodities definition and registration documentation section to learn how the Datastore commodity architecture can be used to provide extensions to the default Datastore API.

The LockQuery implementation, even when used as a standard Query API, provides builtin lock exceptions translation support, using the LockAcquisitionException type to notify any lock acquisition error propagated through the ORM.

final NumericProperty<Long> ID = NumericProperty.longType("id");
final StringProperty VALUE = StringProperty.create("value");

Datastore datastore = getJpaDatastore();

Optional<PropertyBox> result = datastore.create(LockQuery.class) (1)
    .target(DataTarget.named("test")).filter(ID.eq(1L)) //
    .lock() (2)
    .findOne(ID, VALUE); (3)

result = datastore.create(LockQuery.class).target(DataTarget.named("test")) //
    .filter(ID.eq(1L)).lock(3000) (4)
    .findOne(ID, VALUE);

boolean lockAcquired = datastore.create(LockQuery.class).target(DataTarget.named("test")).filter(ID.eq(1L))
    .tryLock(0); (5)
1 Obtain a new LockQuery
2 Configure the query lock, using default mode and timeout
3 Execute query: if the lock cannot be acquired, a LockAcquisitionException is thrown
4 Configure the query lock setting 3 seconds (3000 milliseconds) as lock timeout
5 Try to acquire a lock on the rows returned by the query, setting 0 as lock timeout (no wait)

5.10. JpaDatastore API

The specialized JpaDatastore API, which extends the standard Datastore API, makes available an additional methods through the EntityManagerHandler interface to execute operations using a Datastore managed EntityManager.

The JPA Datastore will take care of the EntityManager instance provision and finalization. See the EntityManager lifecyle section for information about the EntityManager lifecycle management in the JPA Datastore API implementation.

JpaDatastore datastore = getJpaDatastore(); // build or obtain a JpaDatastore

datastore.withEntityManager(em -> { (1)
  em.persist(new Test());
});

Test result = datastore.withEntityManager(em -> { (2)
  return em.find(Test.class, 1);
});
1 Execute an operation using a JPA Datastore managed EntityManager instance
2 Execute an operation using a JPA Datastore managed EntityManager instance and return a result
If you want to reach the goal of a complete abstraction from the persistence store technology and the persistence model, the core Datastore API interface should be used instead of the specific JpaDatastore API by your application code. This way, the concrete Datastore API implementation may be replaced by a different one at any time, without any change to the codebase.

5.11. Extending the JPA Datastore API

5.12. Expression resolvers

The Datastore API can be extended using the ExpressionResolver API, to add new expression resolution strategies, modify existing ones and to handle new Expression types.

See the Datastore API extension documentation section for details.

5.12.1. JPA Expression resolvers registration

A new ExpressionResolver can be registered in the JPA Datastore API in two ways:

1. Using the JPA Datastore API instance:

An ExpressionResolver can be registered either using the Datastore builder API at Datastore configuration time:

Datastore datastore = JpaDatastore.builder() //
    .withExpressionResolver(new MyExpressionResolver()) (1)
    .build();
1 Register and new ExpressionResolver

Or using the Datastore API itself, which extends the ExpressionResolverSupport API:

datastore.addExpressionResolver(new MyExpressionResolver()); (1)
1 Register and new ExpressionResolver

2. Automatic registration using Java service extensions:

The JPA Datastore supports ExpressionResolver automatic registration using the JpaDatastoreExpressionResolver base type and the default Java service extensions modality.

To automatically register an ExpressionResolver this way, a class implementing JpaDatastoreExpressionResolver has to be created and its fully qualified name must be specified in a file named com.holonplatform.datastore.jpa.config.JpaDatastoreExpressionResolver, placed in the META-INF/services folder in classpath.

When this registration method is used, the expression resolvers defined this way will be registered for any JPA Datastore API instance.

5.12.2. Specific expression resolvers registration

All the default Datastore API operation representations supports operation specific expression resolvers registration, through the ExpressionResolverSupport API.

An ExpressionResolver registered for a specific Datastore API operation execution will be available only for the execution of that operation, and will be ignored by any other Datastore API operation.

For example, to register an expression resolver only for a single Query execution, the Query builder API can be used:

long result = datastore.query().target(DataTarget.named("Test")) //
    .withExpressionResolver(new MyExpressionResolver()) (1)
    .count();
1 Register an expression resolver only for the specific Query operation definition

5.12.3. Expression resolvers priority

According to the standard convention, the javax.annotation.Priority annotation can be used on ExpressionResolver classes to indicate in what order the expression resolvers bound to the same type resolution pair (i.e. the expression type handled by a resolver and the target expression type into which it will be resolved) must be applied.

The less is the javax.annotation.Priority number assigned to a resolver, the higher will be it’s priority order.

All the default JPA Datastore expression resolvers have the minimum priority order, allowing to override their behavior and resolution strategies with custom expression resolvers with a higher assigned priority order (i.e. a priority number less then Integer.MAX_VALUE).

5.12.4. Expression validation

The internal JPA Datastore JPQL composer engine will perform validation on any Expression instance to resolve and each corresponding resolved Expression instance, using the default expression validate() method.

So the validate() method can be used to implement custom expression validation logic and throw an InvalidExpressionException when validation fails.

5.12.5. JPA Datastore expressions

Besides the standard Datastore API expressions, such as DataTarget, QueryFilter and QuerySort, which can be used to extend the Datastore API with new expression implementations and new resolution strategies, the JPA Datastore API can be extended using a set of specific JPQL resolution expressions, used by the internal JPQL composer engine to implement the resolution and composition strategy to obtain JPQL statements from the Datastore API meta-language expressions.

These JPQL expression are available from the com.holonplatform.datastore.jpa.jpql.expression package of the holon-datastore-jpa artifact.

The JPQLExpression is the expression which represents an JPQL statement part, used to compose the actual JPQL statement which will be executed using the EntityManager API. So this is the final target expression used by the JPQL composer engine to obtain a JPQL statement part from other, more abstract, expression types.

The JPQLExpression type can be used to directly resolve an abstract Datastore API expression into a JPQL statement part.

For example, supposing to have a KeyIs class which represents a QueryFilter expression type to represent the expression "the key JPA attribute value is equal to a given Long type value":

class KeyIs implements QueryFilter {

  private final Long value;

  public KeyIs(Long value) {
    this.value = value;
  }

  public Long getValue() {
    return value;
  }

  @Override
  public void validate() throws InvalidExpressionException {
    if (value == null) {
      throw new InvalidExpressionException("Kay value must be not null");
    }
  }

}

We want to create an ExpressionResolver class to resolve the KeyIs expression directly into a JPQL WHERE statemet part, using the JPQLExpression type. Using the convenience create method of the ExpressionResolver API, we can do it in the following way:

final NumericProperty<Long> KEY = NumericProperty.create("key", long.class);

final ExpressionResolver<KeyIs, JPQLExpression> keyIsResolver = ExpressionResolver.create( //
    KeyIs.class, (1)
    JPQLExpression.class, (2)
    (keyIs, context) -> {
      String path = JPQLResolutionContext.isJPQLResolutionContext(context)
          .flatMap(ctx -> ctx.isStatementCompositionContext()).flatMap(ctx -> ctx.getAliasOrRoot(KEY))
          .map(alias -> alias + ".key").orElse("key");
      return Optional.of(JPQLExpression.create(path + " = " + keyIs.getValue()));
    }); (3)
1 Expression type to resolve
2 Target expression type
3 Expression resolution logic: since we resolve the KeyIs expression directly into JPQLExpression type, the JPQL WHERE clause part is provided
In this example, the JPQLResolutionContext resolution context extension is used to obtain th alias of property path: see JPA Expression resolution context for details.

After the ExpressionResolver is registered in the Datastore API, the new KeyIs expression can be used in the Datastore API operations which support the QueryFilter expression type just like any other filter expression. For example, in a Query expression:

Datastore datastore = JpaDatastore.builder().withExpressionResolver(keyIsResolver) (1)
    .build();

Query query = datastore.query().filter(new KeyIs(1L)); (2)
1 Register the new expression resolver
2 Use the KeyIs expression in a query definition

The JPQL composer engine will translate the given KeyIs expression in the JPQL WHERE statement part (alias).key = 1, using the previously defined expression resolver.

Other expression types are used to represent elements of a query or a Datastore operation, to be resolved into a final JPQLExpression type. These expression types often represent an intermediate expression type, between the highest abstract layer (i.e. an expression of the Datastore API meta-language) and the final JPQL statement part representation (i.e. the JPQLExpression type).

Some examples are:

  • JPQLLiteral to represent a literal value.

  • JPQLParameter to represent a parameter value.

  • JPQLFunction to represent a JPQL function.

  • JPQLProjection to represent a JPQL SELECT projection.

  • JPQLStatement to represent a full JPQL statement with parameters support.

  • JPQLQueryDefinition and JPQLQuery to represent the definition of a JPQL query and the query statement representation with the JPQLResultConverter to use to covert the query results in the expected type.

For example, let’s see how the JPQLFunction expression can be used as an intermediate JPQL expression type to define a new JPQL function.

We want to define a Trim function to map the JPQL TRIM(both from x) function.

We will use the QueryFunction expression type to represent the function, since it is the default function expression representation of the Datastore API.

public class Trim implements QueryFunction<String, String> {

  private final TypedExpression<String> expression; (1)

  public Trim(TypedExpression<String> expression) {
    super();
    this.expression = expression;
  }

  @Override
  public Class<? extends String> getType() {
    return String.class;
  }

  @Override
  public void validate() throws InvalidExpressionException { (2)
    if (expression == null) {
      throw new InvalidExpressionException("Null function expression");
    }

  }

  @Override
  public List<TypedExpression<? extends String>> getExpressionArguments() { (3)
    return Collections.singletonList(expression);
  }

}
1 The expression to be trimmed
2 Validate the argument
3 The unique function argument will be the expression to trim

Now we create an expression resolver to resolve the QueryFunction expression into a JPQLFunction intermediate JPQL expression type:

public class TrimResolver implements ExpressionResolver<Trim, JPQLFunction> {

  @Override
  public Optional<JPQLFunction> resolve(Trim expression, ResolutionContext context)
      throws InvalidExpressionException {
    return Optional.of(JPQLFunction.create(args -> { (1)
      StringBuilder sb = new StringBuilder();
      sb.append("TRIM(both from ");
      sb.append(args.get(0));
      sb.append(")");
      return sb.toString();
    }));
  }

  @Override
  public Class<? extends Trim> getExpressionType() {
    return Trim.class;
  }

  @Override
  public Class<? extends JPQLFunction> getResolvedType() {
    return JPQLFunction.class;
  }

}
1 Use the JPQLFunction create method to provide the JPQL function representation using the function arguments. The function arguments are already provided by the JPQL composition engine as serialized JPQL expression tokens, according to the actual arguments expression types

Now the function can be registered in the Datastore API and used the same way as any other QueryFunction expression implementation.

final StringProperty STR = StringProperty.create("stringAttribute");
final DataTarget<?> TARGET = DataTarget.named("Test");

Datastore datastore = JpaDatastore.builder() //
    .withExpressionResolver(new TrimResolver()) (1)
    .build();

Stream<String> trimmedValues = datastore.query(TARGET).stream(new Trim(STR)); (2)
1 Register the Trim function expression resolver in Datastore
2 The Trim function is used to provide the trimmed STR property value as query projection

5.12.6. JPA Expression resolution context

The JPA Datastore API makes available an extension of the standard expression ResolutionContext, to provide a set of configuration attributes and JPQL resolution context specific operations. This resolution context extension is represented by the JPQLResolutionContext API.

When used within the JPA Datastore API, the ResolutionContext provided to the registered expression resolvers is a JPQLResolutionContext for any JPA Datastore API standard operation. To obtain a JPQLResolutionContext, the isJPQLResolutionContext method can be used.

The JPQLStatementResolutionContext API is a further context extension which provides methods related to JPQL statements composition. It can be obtained from a JPQLResolutionContext through the isStatementCompositionContext() method.

The statement resolution context provides methods, for example, to obtain the alias for a property path used in a query.

@Override
public Optional<JPQLExpression> resolve(JPQLExpression expression, ResolutionContext context) (1)
    throws InvalidExpressionException {

  JPQLResolutionContext.isJPQLResolutionContext(context).ifPresent(ctx -> { (2)
    ORMPlatform paltform = ctx.getORMPlatform().orElse(null); (3)

    ctx.isStatementCompositionContext().ifPresent(sctx -> { (4)
      Optional<String> alias = sctx.getAliasOrRoot(A_PROPERTY); (5)
    });
  });

  return Optional.empty();
}
1 Let’s suppose we are implementing an expression resolver resolution method
2 Check and obtain the current ResolutionContext as a JPQLResolutionContext
3 As an example, get the current ORMPlatform is requested
4 If the context is a JPQL statement resolution context, it can be obtained as a JPQLStatementResolutionContext
5 Get the alias assigned to the A_PROPERTY property in the current JPQL statement, if available

5.13. Commodity factories

The JPA Datastore API supports Datastore commodities registration using the JpaDatastoreCommodityFactory type.

See the Datastore commodities definition and registration documentation section to learn how the Datastore commodity architecture can be used to provide extensions to the default Datastore API.

The JpaDatastoreCommodityFactory type provides a specialized JpaDatastoreCommodityContext API as Datastore commodity context, to make available a set of JPA Datastore specific configuration attributes and references, for example:

  • The EntityManagerFactory bound the JPA Datastore.

  • The configured (or auto-detected) ORM platform.

  • The available expression resolvers

  • A set of APIs and configuration methods available from the JPQL composer engine.

Furthermore, it makes available some API methods to invoke some JPA Datastore operations, such as the withEntityManager(EntityManagerOperation<R> operation) method to execute an operation using a Datastore managed EntityManager instance or methods to inspect and create other Datastore commodities.

interface MyCommodity extends DatastoreCommodity { (1)

  CriteriaBuilder getCriteriaBuilder();

}

class MyCommodityImpl implements MyCommodity { (2)

  private final EntityManagerFactory entityManagerFactory;

  public MyCommodityImpl(EntityManagerFactory entityManagerFactory) {
    super();
    this.entityManagerFactory = entityManagerFactory;
  }

  @Override
  public CriteriaBuilder getCriteriaBuilder() {
    return entityManagerFactory.getCriteriaBuilder();
  }

}

class MyCommodityFactory implements JpaDatastoreCommodityFactory<MyCommodity> { (3)

  @Override
  public Class<? extends MyCommodity> getCommodityType() {
    return MyCommodity.class;
  }

  @Override
  public MyCommodity createCommodity(JpaDatastoreCommodityContext context)
      throws CommodityConfigurationException {
    EntityManagerFactory entityManagerFactory = context.getEntityManagerFactory();
    return new MyCommodityImpl(entityManagerFactory);
  }

}
1 Datastore commodity API
2 Commodity implementation
3 Commodity factory implementation

A Datastore commodity factory class which extends the JpaDatastoreCommodityFactory interface can be registered in a JPA Datastore in two ways:

1. Direct registration using the JPA Datastore API builder:

The JPA Datastore API supports the commodity factory registration using the withCommodity builder method.

Datastore datastore = JpaDatastore.builder() //
    .withCommodity(new MyCommodityFactory()) (1)
    .build();
1 Register the MyCommodityFactory commodity factory in given JPA Datastore implementation

2. Automatic registration using the Java service extensions:

To automatically register an commodity factory using the standard Java service extensions based method, a class implementing JpaDatastoreCommodityFactory has to be created and its qualified full name must be specified in a file named com.holonplatform.datastore.jpa.config.JpaDatastoreCommodityFactory, placed in the META-INF/services folder of the classpath.

When this registration method is used, the commodity factories defined this way will be registered for any JPA Datastore API instance.

6. Spring ecosystem integration

The holon-datastore-jpa-spring artifact provides integration with the Spring framework for the JPA Datastore API.

Maven coordinates:

<groupId>com.holon-platform.jpa</groupId>
<artifactId>holon-datastore-jpa-spring</artifactId>
<version>5.2.1</version>

6.1. Integration with the Spring JPA infrastructure

When a JPA Datastore API is configured as a Spring bean using the facilities described in this documentation section, a consistent integration with the Spring JPA infrastructure is automatically provided.

In particular, the EntityManager instance used for the JPA Datastore API operations is obtained as a Spring transactional EntityManager proxy, using the EntityManagerFactory configured for the JPA Datastore implementation.

This allows to seamlessy integrate with the Spring transaction synchronization architecture, since the shared EntityManager proxy provides the following behavior: it will delegate all calls to the current transactional EntityManager, if any; otherwise it will fall back to a newly created EntityManager per operation.

This way, the JPA Datastore API can be seamlessly used along with the Spring transactions management conventions, for example when using a Spring PlatformTransactionManager or the @Transactional annotation.

6.2. JPA Datastore configuration

The EnableJpaDatastore annotation can be used on Spring configuration classes to enable automatic JPA Datastore beans configuration.

By default, the JPA Datastore bean name configured with the @EnableJpaDatastore annotation will be jpaDatastore.

6.2.1. EntityManagerFactory

The EntityManagerFactory bean to be used to configure the JPA Datastore API is obtained as follows:

  • If the entityManagerFactoryReference attribute of the @EnableJpaDatastore annotation is setted, the provided bean definition name will be used as the EntityManagerFactory bean definition name to use.

@EnableJpaDatastore(entityManagerFactoryReference = "myEntityManagerFactory") (1)
@Configuration
class Config {

  @Bean(name = "myEntityManagerFactory")
  public FactoryBean<EntityManagerFactory> entityManagerFactory(DataSource dataSource) {
    LocalContainerEntityManagerFactoryBean emf = new LocalContainerEntityManagerFactoryBean();
    emf.setDataSource(dataSource);
    emf.setJpaVendorAdapter(new HibernateJpaVendorAdapter());
    emf.setPackagesToScan("com.example.entities");
    return emf;
  }

}

@Autowired
Datastore datastore; (2)
1 Provide the EntityManagerFactory bean definition name to use with the JPA Datastore
2 The JPA Datastore is configured and made available, for example, using dependency injection
  • Otherwise, the default entityManagerFactory bean definition name will be used to lookup for the EntityManagerFactory bean to use.

@EnableJpaDatastore
@Configuration
class Config {

  @Bean
  public FactoryBean<EntityManagerFactory> entityManagerFactory(DataSource dataSource) { (1)
    LocalContainerEntityManagerFactoryBean emf = new LocalContainerEntityManagerFactoryBean();
    emf.setDataSource(dataSource);
    emf.setJpaVendorAdapter(new HibernateJpaVendorAdapter());
    emf.setPackagesToScan("com.example.entities");
    return emf;
  }

}

@Autowired
Datastore datastore; (2)
1 The default entityManagerFactory bean definition name is used for the EntityManagerFactory bean definition
2 The JPA Datastore is configured and made available, for example, using dependency injection

If a EntityManagerFactory bean definition whith the required name is not present in Spring context, an initialization error is thrown.

See the Full JPA bean stack auto-configuration section to learn how to auto-configure a complete JPA bean stack in the Spring context.

6.2.2. Multiple JPA Datastores configuration

When more than one JPA Datastore bean has to be configured using the @EnableJpaDatastore annotation, the dataContextId attribute can be used to assign a different data context id to each JPA Datastore bean definition, in order to:

  • Provide different sets of configuration properties using the same Spring environment.

  • Provide a default name pattern matching strategy with the EntityManagerFactory bean definition to use for each JPA Datastore to configure: if not directly specified with the entityManagerFactoryReference attribute, the EntityManagerFactory bean definition to use for each JPA Datastore will be detected in Spring context using the bean name pattern: entityManagerFactory_{datacontextid} where {datacontextid} is equal to the dataContextId attribute of the @EnableJpaDatastore annotation.

When a data context id is defined, a Spring qualifier named the same as the data context id will be associated to the generated JPA Datastore bean definitions, and such qualifier can be later used to obtain the right Datastore instance through dependency injection.

Furthermore, the JPA Datastore bean definitions will be named using the data context id as suffix, according to the name pattern: jpaDatastore_{datacontextid}.

@Configuration class Config {

  @Configuration
  @EnableJpaDatastore(dataContextId = "one") (1)
  static class Config1 {

    @Bean(name = "entityManagerFactory_one")
    public EntityManagerFactory entityManagerFactory() {
      return buildEntityManagerFactory();
    }

  }

  @Configuration
  @EnableJpaDatastore(dataContextId = "two") (2)
  static class Config2 {

    @Bean(name = "entityManagerFactory_two")
    public EntityManagerFactory entityManagerFactory() {
      return buildEntityManagerFactory();
    }

  }

}

@Autowired
@Qualifier("one")
Datastore datastore1; (3)

@Autowired
@Qualifier("two")
Datastore datastore2;
1 Configure the first JPA Datastore using one as data context id: by default the bean named entityManagerFactory_one will be used as EntityManagerFactory
2 Configure the first JPA Datastore using two as data context id: by default the bean named entityManagerFactory_two will be used as EntityManagerFactory
3 A specific Datastore type bean reference can be obtained using the data context id as qualifier

6.2.3. Auto-flush mode

The JPA Datastore Auto-flush mode can be configured using the autoFlush attribute of the @EnableJpaDatastore annotation.

@EnableJpa(autoFlush = true) (1)
@Configuration
class Config {

  // ...

}
1 Enable the JPA Datastore auto-flush mode

6.2.4. Transactional JPA Datastore operations

The transactional attribute of the @EnableJpaDatastore annotation can be used to control the transactional configuration of a set of the Datastore API operations.

When setted to true, the Spring @Transactional annotation behavior is automatically added to the following Datastore API operation methods:

  • refresh

  • insert

  • update

  • save

  • delete

The default REQUIRED propagation behavior is used, thus allowing the method calls to participate in an existing transaction or to be executed in a new one when the Spring transactions infrastructure is used.

The transactional attribute is true by default
@EnableJpa
@PropertySource("jdbc.properties")
@EnableTransactionManagement (1)
@Configuration
class Config {

}

@Autowired
Datastore datastore;

void doTransactionally() {
  datastore.insert(DataTarget.named("Test"), buildPropertyBoxValue()); (2)
}
1 Enables Spring’s annotation-driven transaction management capability.
2 The insert method is transactional by default, so the @Transactional annotation is not explicitly required here

The transactional attribute can be used to disable this default behavior:

@EnableJpaDatastore(transactional = false)
@Configuration
class Config2 {

}

6.2.5. Primary mode

The @EnableJpaDatastore annotation provides a primary() attribute which can be used to control the primary mode of the JPA Datastore bean registration.

If the primary mode is set to PrimaryMode.TRUE, the Datastore bean created with the corresponding annotation will be marked as primary in the Spring application context, meaning that will be the one provided by Spring in case of multiple available candidates, when no specific bean name or qualifier is specified in the dependency injection declaration.

This behaviour is similar to the one obtained with the Spring @Primary annotation at bean definition time.

By default, the primary mode is set to PrimaryMode.AUTO, meaning that the registred JPA Datastore bean will be marked as primary only when the EntityManagerFactory bean to which is bound is registered as primary candidate bean.

6.2.6. JPA Datastore configuration properties

When a JPA Datastore bean is configured using the @EnableJpaDatastore annotation, the Spring environment is automatically used as configuration properties source.

This way, many Datastore configuration settings can be provided using a configuration property with the proper name and value.

The supported configuration properties are:

1. The standard Datastore configuration properties, avaible from the DatastoreConfigProperties property set (See Datastore configuration).

The configuration property prefix is holon.datastore and the following properties are available:

Table 1. Datastore configuration properties
Name Type Meaning

holon.datastore. trace

Boolean (true / false)

Enable/disable Datastore operations tracing.

holon.datastore. dialect

String

The fully qualified class name of the ORM Dialect to use. See ORM Dialect configuration.

2. An additional set of properties, provided by the JpaDatastoreConfigProperties property set, which can be used as an alternative for the @EnableJpaDatastore annotation attributes described in the previous sections.

Table 2. JPA Datastore configuration properties
Name Type Meaning

holon.datastore.jpa. primary

Boolean (true / false)

Mark the JPA Datastore bean as primary candidate for dependency injection when more than one definition is available. IF not setted to true, the AUTO strategy will be used: the JPA Datastore bean will be marked as primary only when the EntityManagerFactory bean to which is bound is registered as primary candidate bean.

holon.datastore.jpa. auto-flush

Boolean (true / false)

Enable or disable the JPA Datastore auto-flush mode. See Auto-flush mode.

holon.datastore.jpa. transactional

Boolean (true / false)

Whether to add the Spring @Transactional behavior to the suitable Datastore API methods. See Transactional JPA Datastore operations.

Example of Datastore configuration properties:

holon.datastore.trace=true (1)
holon.datastore.dialect=my.dialect.class.Name (2)

holon.datastore.jpa.auto-flush=true (3)
holon.datastore.jpa.transactional=false (4)
1 Enable tracing
2 Set the ORM dialect class name
3 Enable the auto-flush mode
4 Disable the automatic transactional behavior of the Datastore operations

6.2.7. Datastore extension and configuration using the Spring context

The JPA Datastore implementation supports the standard Holon Platform Datastore Spring integration features for Datastore beans configuration and extension, which includes:

@DatastoreResolver (1)
class MyFilterExpressionResolver implements QueryFilterResolver<MyFilter> {

  @Override
  public Class<? extends MyFilter> getExpressionType() {
    return MyFilter.class;
  }

  @Override
  public Optional<QueryFilter> resolve(MyFilter expression, ResolutionContext context)
      throws InvalidExpressionException {
    // implement actual MyFilter expression resolution
    return Optional.empty();
  }

}

@Component
class MyDatastorePostProcessor implements DatastorePostProcessor { (2)

  @Override
  public void postProcessDatastore(ConfigurableDatastore datastore, String datastoreBeanName) {
    // configure Datastore
  }

}

@DatastoreCommodityFactory (3)
class MyCommodityFactory implements JpaDatastoreCommodityFactory<MyCommodity> {

  @Override
  public Class<? extends MyCommodity> getCommodityType() {
    return MyCommodity.class;
  }

  @Override
  public MyCommodity createCommodity(JpaDatastoreCommodityContext context)
      throws CommodityConfigurationException {
    // create commodity instance
    return new MyCommodity();
  }

}
1 Automatically register a Datastore expression resolver using the @DatastoreResolver annotation
2 Post process Datastore configuration using a DatastorePostProcessor type Spring bean
3 Automatically register a Datastore commodity factory using the @DatastoreCommodityFactory annotation

6.3. Full JPA bean stack auto-configuration

The EnableJpa annotation can be used on Spring configuration classes to setup a full JPA enviroment bean stack:

  • DataSource: If a javax.sql.DataSource type bean is not already registered in Spring context, a DataSource instance is created and configured using the Holon Platform DataSource configuration properties. See the Holon Platform JDBC Module documentation for details.

  • EntityManagerFactory: A JPA javax.persistence.EntityManagerFactory bean is registered and configured, along with a suitable Spring JpaVendorAdapter instance for ORM configuration. A default Spring LocalContainerEntityManagerFactoryBean is used as EntityManagerFactory factory bean.

  • PlatformTransactionManager: A org.springframework.transaction.PlatformTransactionManager bean is registered to enable the Spring transactions management APIs and conventions;

  • Datastore: A JPA Datastore is configured and bound to the EntityManagerFactory bean of the JPA stack.

In a typical scenario, to enable a full JPA stack in Spring context, the @EnableJpa annotation can be used as foolows:

@EnableJpa(entityPackageClasses = TestEntity.class) (1)
@PropertySource("jdbc.properties")
@Configuration
class Config {

}

@Autowired
DataSource dataSource;

@PersistenceUnit
EntityManagerFactory entityManagerFactory;

@Autowired
PlatformTransactionManager transactionManager;

@Autowired
Datastore datastore;
1 Specify the base package to use to auto-detect the JPA entity definition classes to mapped into the EntityManagerFactory instance

The DataSource configuration properties in the example above are expected to be provided through the jdbc.properties file. For example:

holon.datasource.url=jdbc:h2:mem:test
holon.datasource.username=sa

See the next sections to learn about all the available configuration attributes to fine tune the JPA stack beans configuration.

6.3.1. DataSource configuration

A DataSource type bean is automatically configured if:

  • A javax.sql.DataSource type bean is not already registered in the Spring context.

  • The dataSourceReference attribute of the @EnableJpa annotation is not specified. If specified, it indicates the DataSource bean name to use, which must be available in the Spring context.

In order to auto-configure the DataSource bean, a suitable set of configuration properties must be available in the Spring Environment.

See the Holon Platform DataSource configuration properties section of the Holon Platform JDBC Module documentation for details.

6.3.2. EntityManagerFactory configuration

A Spring LocalContainerEntityManagerFactoryBean instance is used as javax.persistence.EntityManagerFactory implementation to ensure a full integration with Spring JPA architecture and provide functionalities such as automatic JPA persistence unit configuration without the need of a persistence.xml configuration file.

If a persistence.xml configuration file is not available, automatic persistence unit entity classes configuration can be performed using one of the following @EnableJpa annotation attributes:

  • entityPackages : providing a list of package names to scan in order to map JPA entity classes into the EntityManagerFactory,

  • entityPackageClasses : providing a list of classes from which to obtain the package names to scan in order to map JPA entity classes into the EntityManagerFactory. Represents a type-safe alternative to entityPackages attribute.

Two more annotation attributes are available for persistence unit configuration:

  • validationMode : Specify the JPA 2.0 validation mode for the persistence unit;

  • sharedCacheMode : Specify the JPA 2.0 shared cache mode for the persistence unit.

@EnableJpa(entityPackageClasses = TestEntity.class, validationMode = ValidationMode.NONE)
@PropertySource("jdbc.properties")
@Configuration
class Config {

}

6.3.3. Persistence provider (ORM) configuration

By default, the JPA Persistence provider (ORM) to use is auto-detected from classpath and a suitable Spring JpaVendorAdapter instance is configured and bound to the EntityManagerFactory bean.

If more than one persistence provider implementation is present in classpath (or to explicitly specify a persistence provider implementation to use, anyway) the orm configuration property can be used. See JPA configuration properties for details.

6.3.4. Transaction manager configuration

A Spring JPA PlatformTransactionManager is auto-configured and bound to the EntityManagerFactory bean to enable the Spring’s transactions support.

A set of attributes are made available by the @EnableJpa annotation to fine tune the transaction manager configuration:

Attribute name Type Meaning Default

transactionSynchronization

int

Set when the transaction manager should activate the thread-bound transaction synchronization support.

-1 (always)

defaultTimeout

int

Specify the default timeout that the transaction manager should apply if there is no timeout specified at the transaction level, in seconds.

-1 (TransactionDefinition.TIMEOUT_DEFAULT)

validateExistingTransaction

boolean

Set whether existing transactions should be validated before participating in them.

false

failEarlyOnGlobalRollbackOnly

boolean

Set whether to fail early in case of the transaction being globally marked as rollback-only.

false

rollbackOnCommitFailure

boolean

Set whether a rollback should be performed on failure of the transaction manager commit call.

false

6.3.5. JPA Datastore configuration

By default, a JPA Datastore implementation is automatically configured using the EntityManagerFactory bean and made available as Spring bean. The JPA Datastore implementation is configured for a seamless integration with the Spring JPA infrastructure, as described in Integration with the Spring JPA infrastructure.

To disable automatic JPA Datastore configuration, the enableDatastore attribute of the @EnableJpa annotation can be used.

@EnableJpa(entityPackageClasses = TestEntity.class, enableDatastore = false) (1)
@PropertySource("jdbc.properties")
@Configuration
class Config {

}
1 Disable the JPA Datastore bean auto configuration

The @EnableJpa annotation makes available two additional configuration attributes related to the JPA Datastore implementation setup:

  • autoFlush : whether to enable the JPA Datastore auto flush mode. See Auto-flush mode for details.

  • transactionalDatastore : whether to add transactional behaviour to data manipulation Datastore API operations. See Transactional JPA Datastore operations for details.

6.3.6. JPA configuration properties

The JpaConfigProperties interface provides a set of configuration properties to be used with JPA stack beans auto-configuration. It extends a default ConfigPropertySet bound to the property name prefix holon.jpa.

The available configuration properties are listed below:

Table 3. JPA configuration properties
Name Type Meaning

holon.jpa. orm

ORMPlatform enumeration

ORM platform to use as persistence provider.

holon.jpa. dialect

String

ORM dialect class name to use.

holon.jpa. database

com.holonplatform.jdbc.DatabasePlatform enumeration

Database platform to which the DataSource is connected (auto-detected by default).

holon.jpa. generate-ddl

Boolean (true/false)

Whether to initialize the database schema on startup.

holon.jpa. show-sql

Boolean (true/false)

Whether to instruct JPA ORM engine to show executed SQL statements, if supported by the ORM platform.

The JpaConfigProperties can be loaded from a number a sources using the default ConfigPropertySet builder interface.

Using the Spring integration, all Environment registered PropertySources will be enabled as a JpaConfigProperties source.

6.3.7. Multiple JPA bean stacks configuration

When more than one JPA stack has to be configured using the @EnableJpa annotation, the dataContextId attribute can be used to assign a different data context id to each JPA bean definitions stack, in order to:

  • Provide different sets of configuration properties using the same Spring environment.

  • Provide a default name pattern strategy for the JPA beans stack using the bean name pattern: beanDefinitionName_{datacontextid} where {datacontextid} is equal to the dataContextId attribute of the @EnableJpa annotation.

When a data context id is defined, a Spring qualifier named the same as the data context id will be associated to each JPA bean definition, and such qualifier can be later used to obtain the right bean instance through dependency injection.

For example, given the following jdbc.properties configuration property source file:

holon.datasource.one.url=jdbc:h2:mem:testdbm1
holon.datasource.one.username=sa

holon.datasource.two.url=jdbc:h2:mem:testdbm2
holon.datasource.two.username=sa

To configure two JPA bean stacks, one bound to the one data context id and the other to the two data context id, the @EnableJpa annotation can be used as follows:

@Configuration
@PropertySource("jdbc.properties")
static class Config {

  @Configuration
  @EnableJpa(dataContextId = "one", entityPackageClasses = TestEntity1.class)
  static class Config1 {
  }

  @Configuration
  @EnableJpa(dataContextId = "two", entityPackageClasses = TestEntity2.class)
  static class Config2 {
  }

}

@Autowired
@Qualifier("one")
Datastore datastore1;

@Autowired
@Qualifier("two")
Datastore datastore2;

6.3.8. Programmatic JPA Datastore bean configuration

When a JPA Datastore is not istantiated and configured in a Spring context using the @EnableJpaDatastore annotation, but rather providing the bean implementation programmatically (for example using the JpaDatastore builder API), the Spring integration features described in the previous sections must be explicitly enabled:

@Configuration
@EnableJpaDatastore
@EnableDatastoreConfiguration (1)
class Config {

  @Bean
  public Datastore jpaDatastore(EntityManagerFactory entityManagerFactory) {
    return JpaDatastore.builder().entityManagerFactory(entityManagerFactory)
        .entityManagerHandler(SpringEntityManagerLifecycleHandler.create()) (2)
        .build();
  }

}
1 Use the @EnableDatastoreConfiguration to automatically configure the Datastore with auto-detected Datastore configuration context beans
2 Set a SpringEntityManagerLifecycleHandler instance as Datastore EntityManager lifecycle handler to use Spring managed EntityManager proxies

7. Spring Boot integration

The holon-datastore-jpa-spring-boot artifact provides integration with Spring Boot for JPA stack and JPA Datastore beans auto-configuration.

To enable Spring Boot JPA auto-configuration features, the following artifact must be included in your project dependencies:

Maven coordinates:

<groupId>com.holon-platform.jpa</groupId>
<artifactId>holon-datastore-jpa-spring-boot</artifactId>
<version>5.2.1</version>

7.1. JPA Datastore auto-configuration

A JPA Datastore bean is auto-configured only when:

  • A JpaDatastore type bean is not already available from the Spring application context.

  • A valid EntityManagerFactory type bean is available from the Spring application context.

The JPA Datastore auto-configuration strategy is the same adopted by the @EnableJpaDatastore annotation, including the Spring environment configuration properties support. See JPA Datastore configuration for details.

For example, given an application.yml configuration file as follows:

holon:
  datastore:
    trace: true

The Holon platform Spring Boot auto-configuration classes will auto configure a JPA Datastore bean bound to the EntityManagerFactory bean (which must be available in the Spring application context), enabling operations tracing according to the holon.datastore.trace property value.

@Configuration
@EnableAutoConfiguration
class Config {

  @Bean
  public DataSource dataSource() {
    return buildDataSource();
  }

  @Bean
  public FactoryBean<EntityManagerFactory> entityManagerFactory(DataSource dataSource) {
    LocalContainerEntityManagerFactoryBean emf = new LocalContainerEntityManagerFactoryBean();
    emf.setDataSource(dataSource);
    emf.setJpaVendorAdapter(new HibernateJpaVendorAdapter());
    emf.setPackagesToScan("com.example.entities");
    return emf;
  }

}

@Autowired
Datastore datastore; (1)
1 A JPA Datastore type bean is auto-configured ad available from the Spring context

To disable this auto-configuration feature the JpaDatastoreAutoConfiguration class can be excluded:

@EnableAutoConfiguration(exclude={JpaDatastoreAutoConfiguration.class})

7.2. Full JPA beans stack auto-configuration

When the holon-datastore-jpa-spring-boot artifact is available in classpath, a full JPA beans stack is auto-configured when an EntityManagerFactory type bean is not already registered in the Spring application context.

The JPA beans stack configuration strategy is the same of the one adopted by the @EnableJpa annotation, including the Spring environment configuration properties support. See Full JPA bean stack auto-configuration for details.

For example, given an application.yml configuration file as follows:

spring:
  datasource:
    url: "jdbc:h2:mem:test"
    username: "sa"

holon:
  datastore:
    trace: true

The Holon platform Spring Boot auto-configuration classes will auto configure a JPA beans stack using the DataSource auto-configured by the standard Spring Boot classes spring.datasource.* configuration properties. The Datastore operations tracing is enabled according to the holon.datastore.trace property value.

@Configuration
@EnableAutoConfiguration
class Config {

}

@Autowired
DataSource dataSource;

@PersistenceUnit
EntityManagerFactory entityManagerFactory;

@Autowired
PlatformTransactionManager transactionManager;

@Autowired
Datastore datastore;

The Holon Platform DataSource auto-configuration capabilities can also be used, replacing the spring.datasource. configuration properties with the holon.datasource. ones. For example:

holon:
  datasource:
    url: "jdbc:h2:mem:test"
    username: "sa"

  datastore:
    trace: true

To disable this auto-configuration feature the JpaAutoConfiguration class can be excluded:

@EnableAutoConfiguration(exclude={JpaAutoConfiguration.class})

7.3. Multiple JPA beans stack auto configuration

When the DataSource type beans are auto-configured using the Holon Platform JDBC Module features (see the Holon Platform JDBC Module Spring Boot integration documentation section), the auto-configuration of multiple JPA bean stacks is available out-of-the-box.

The data context id convention is used to provide multiple DataSource and JPA beans stack auto-configuration capabilities: when multiple DataSource type beans are registered, each of them bound to a data context id, the Spring Boot auto-configuration classes will provide to configure a JPA beans stack for each detected DataSource bean, binding the same data context id to the JPA stack bean definitions.

According to the data context id convention, each DataSource and JPA stack bean will be qualified with the corresponding data context id, so that the specific bean instance can be later obtained using the data context id as Spring bean qualifier name.

For example, given an application.yml configuration file as follows:

holon:
  datasource:
    one:
      url: "jdbc:h2:mem:test1"
      username: "sa"
    two:
      url: "jdbc:h2:mem:test2"
      username: "sa"

The auto-configuration feature will configure two DataSource beans:

  • One DataSource bean instance using the one data context id configuration properties, qualified with the one qualifier.

  • Another DataSource bean instance using the two data context id configuration properties, qualified with the two qualifier.

And two corresponding JPA beans stacks:

  • One backed by the DataSource bound to the one data context id, qualified with the one qualifier.

  • Another backed by the DataSource bound to the two data context id, qualified with the two qualifier.

So, for example, the DataSource and JPA Datastore beans can be obtained as follows:

// 'one' data context id:
@Autowired @Qualifier("one")
DataSource dataSource1;

@Autowired @Qualifier("one")
Datastore datastore1;

// 'two' data context id:
@Autowired @Qualifier("two")
DataSource dataSource2;

@Autowired @Qualifier("two")
Datastore datastore2;

7.3.1. JPA entities scan

When more than one JPA beans stack is auto-configured, the JpaEntityScan repeatable annotation can be used to provide the JPA entity set to map to each auto-configured EntityManagerFactory, using the data context id value to identify each EntityManagerFactory bean.

So, for example, given an application.yml configuration file as follows:

holon:
  datasource:
    one:
      url: "jdbc:h2:mem:test1"
      username: "sa"
    two:
      url: "jdbc:h2:mem:test2"
      username: "sa"

Two DataSource beans and two JPA beans stacks will be auto-configured using the one and two data context ids.

The @JpaEntityScan annotation can be used to specify the packages to scan for JPA entity classes mapping, one for the one data context id and another for the two data context id:

@Configuration
@EnableAutoConfiguration
@JpaEntityScan(value = "one", basePackageClasses = TestEntity1.class) (1)
@JpaEntityScan(value = "two", basePackageClasses = TestEntity2.class) (2)
class Config {

}
1 Use the base package of the TestEntity1 class to scan for the JPA entity classes to map to the EntityManagerFactory bean associated to the one data context id
2 Use the base package of the TestEntity2 class to scan for the JPA entity classes to map to the EntityManagerFactory bean associated to the two data context id

7.4. Spring Boot starters

The following starter artifacts are available to provide a quick project configuration setup using Maven dependency system:

1. JPA starter using Hibernate provides the dependencies to the Holon JPA Spring Boot integration artifact holon-datastore-jpa-spring-boot and the dependencies to use Hibernate ORM as persistence provider.

Furthermore, it provides dependencies for:

Maven coordinates:

<groupId>com.holon-platform.jpa</groupId>
<artifactId>holon-starter-jpa-hibernate</artifactId>
<version>5.2.1</version>

2. JPA starter using Eclipselink provides the same dependencies as the previous JPA starter, but using Eclipselink as persistence provider, instead of Hibernate.

Maven coordinates:

<groupId>com.holon-platform.jpa</groupId>
<artifactId>holon-starter-jpa-eclipselink</artifactId>
<version>5.2.1</version>

8. Loggers

By default, the Holon platform uses the SLF4J API for logging. The use of SLF4J is optional: it is enabled when the presence of SLF4J is detected in the classpath. Otherwise, logging will fall back to JUL (java.util.logging).

The logger name for the JPA Datastore module is com.holonplatform.datastore.jpa.

9. System requirements

9.1. Java

The Holon Platform JPA Datastore module requires Java 8 or higher.

9.2. JPA

The Java Persistence API version 2.0 or higher is required for the JPA Datastore module proper use. To use most recent JPA features, such as left joins and the ON clause, the Java Persistence API version 2.1 is required.

9.3. Persistence providers

Altough any JPA 2.0 or higher compliant persistence provider (ORM) is supported, the Holon Platform JPA Datastore module is tested and certified with the followings: