How to configure JPA for testing in Maven - java

Is there a way to set up a second persistence.xml file in a Maven project such that it is used for testing instead of the normal one that is used for deployment?
I tried putting a persistence.xml into src/test/resources/META-INF, which gets copied into target/test-classes/META-INF, but it seems target/classes/META-INF (the copy from the src/main/resources) gets preferred, despite mvn -X test listing the classpath entries in the right order:
[DEBUG] Test Classpath :
[DEBUG] /home/uqpbecke/dev/NetBeansProjects/UserManager/target/test-classes
[DEBUG] /home/uqpbecke/dev/NetBeansProjects/UserManager/target/classes
[DEBUG] /home/uqpbecke/.m2/repository/junit/junit/4.5/junit-4.5.jar
...
I would like to be able to run tests against a simple hsqldb configuration without having to change the deployment version of the JPA configuration, ideally straight after project checkout without any need for local tweaking.

The following will work for Maven 2.1+ (prior to that there wasn't a phase between test and package that you could bind an execution to).
You can use the maven-antrun-plugin to replace the persistence.xml with the test version for the duration of the tests, then restore the proper version before the project is packaged.
This example assumes the production version is src/main/resources/META-INF/persistence.xml and the test version is src/test/resources/META-INF/persistence.xml, so they will be copied to target/classes/META-INF and target/test-classes/META-INF respectively.
It would be more elegant to encapsulate this into a mojo, but as you're only copying a file, it seems like overkill.
<plugin>
<artifactId>maven-antrun-plugin</artifactId>
<version>1.3</version>
<executions>
<execution>
<id>copy-test-persistence</id>
<phase>process-test-resources</phase>
<configuration>
<tasks>
<!--backup the "proper" persistence.xml-->
<copy file="${project.build.outputDirectory}/META-INF/persistence.xml" tofile="${project.build.outputDirectory}/META-INF/persistence.xml.proper"/>
<!--replace the "proper" persistence.xml with the "test" version-->
<copy file="${project.build.testOutputDirectory}/META-INF/persistence.xml" tofile="${project.build.outputDirectory}/META-INF/persistence.xml"/>
</tasks>
</configuration>
<goals>
<goal>run</goal>
</goals>
</execution>
<execution>
<id>restore-persistence</id>
<phase>prepare-package</phase>
<configuration>
<tasks>
<!--restore the "proper" persistence.xml-->
<copy file="${project.build.outputDirectory}/META-INF/persistence.xml.proper" tofile="${project.build.outputDirectory}/META-INF/persistence.xml"/>
</tasks>
</configuration>
<goals>
<goal>run</goal>
</goals>
</execution>
</executions>
</plugin>

In an EE6/CDI/JPA project, a test src/test/resources/META-INF/persistence.xml is picked up just fine without any further configuration.
When using JPA in Spring, the following works in the application context used for testing:
<bean id="entityManagerFactory"
class="org.springframework.orm.jpa.LocalContainerEntityManagerFactoryBean">
<property name="dataSource" ref="dataSource" />
<!--
JPA requires META-INF/persistence.xml, but somehow prefers the one
in classes/META-INF over the one in test-classes/META-INF. Spring
to the rescue, as it allows for setting things differently, like by
referring to "classpath:persistence-TEST.xml". Or, simply referring
to "META-INF/persistence.xml" makes JPA use the test version too:
-->
<property name="persistenceXmlLocation" value="META-INF/persistence.xml" />
<!-- As defined in /src/test/resources/META-INF/persistence.xml -->
<property name="persistenceUnitName" value="myTestPersistenceUnit" />
<property name="jpaVendorAdapter">
<bean
class="org.springframework.orm.jpa.vendor.HibernateJpaVendorAdapter">
</bean>
</property>
</bean>
Here, /src/test/resources/META-INF/persistence.xml (copied into target/test-classes) would be preferred over /src/main/resources/META-INF/persistence.xml (copied into target/classes).
Unfortunately, the location of the persistence.xml file also determines the so-called "persistence unit's root", which then determines which classes are scanned for #Entity annotations. So, using /src/test/resources/META-INF/persistence.xml would scan classes in target/test-classes, not classes in target/classes (where the classes that need to be tested would live).
Hence, for testing, one would need to explicitly add <class> entries to persistence.xml, to avoid java.lang.IllegalArgumentException: Not an entity: class .... The need for <class> entries can be avoided by using a different file name, like persistence-TEST.xml, and put that file in the very same folder as the regular persistence.xml file. The Spring context from your test folder can then just refer to <property name="persistenceXmlLocation" value="META-INF/persistence-TEST.xml" />, and Spring will find it for you in src/main.
As an alternative, one might be able to keep persistence.xml the same for the actual application and the tests, and only define one in src/main. Most configuration such as the drivers, dialect and optional credentials can be done in the Spring context instead. Also settings such as hibernate.hbm2ddl.auto can be passed in the context:
<bean id="dataSource"
class="org.springframework.jdbc.datasource.DriverManagerDataSource">
<!-- For example: com.mysql.jdbc.Driver or org.h2.Driver -->
<property name="driverClassName" value="#{myConfig['db.driver']}" />
<!-- For example: jdbc:mysql://localhost:3306/myDbName or
jdbc:h2:mem:test;DB_CLOSE_DELAY=-1 -->
<property name="url" value="#{myConfig['db.url']}" />
<!-- Ignored for H2 -->
<property name="username" value="#{myConfig['db.username']}" />
<property name="password" value="#{myConfig['db.password']}" />
</bean>
<bean id="jpaAdaptor"
class="org.springframework.orm.jpa.vendor.HibernateJpaVendorAdapter">
<!-- For example: org.hibernate.dialect.MySQL5Dialect or
org.hibernate.dialect.H2Dialect -->
<property name="databasePlatform" value="#{myConfig['db.dialect']}" />
</bean>
<bean id="entityManagerFactory"
class="org.springframework.orm.jpa.LocalContainerEntityManagerFactoryBean">
<property name="dataSource" ref="dataSource" />
<property name="jpaVendorAdapter" ref="jpaAdapter" />
<property name="jpaProperties">
<props>
<!-- For example: validate, update, create or create-drop -->
<prop key="hibernate.hbm2ddl.auto">#{myConfig['db.ddl']}</prop>
<prop key="hibernate.show_sql">#{myConfig['db.showSql']}</prop>
<prop key="hibernate.format_sql">true</prop>
</props>
</property>
</bean>

It seems multiple persistence.xml files is a general problem with JPA, solved only by classloading tricks.
A workaround that works for me is to define multiple persistence units in one persistence.xml file and then make sure that your deployment and test code use a different binding (in Spring you can set the "persistenceUnitName" property on the entity manager factory). It pollutes your deployment file with the test configuration, but if you don't mind that it works ok.

Add a persistance.xml for tests: /src/test/resources/META-INF/persistence.xml
As #Arjan said, that would change persistance unit's root and entity classes would be scanned in target/test-classes. To handle that, add jar-file element to this persistance.xml:
/src/test/resources/META-INF/persistence.xml
<persistence xmlns="http://java.sun.com/xml/ns/persistence"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://java.sun.com/xml/ns/persistence http://java.sun.com/xml/ns/persistence/persistence_2_0.xsd"
version="2.0">
<persistence-unit name="com.some.project">
<provider>org.hibernate.jpa.HibernatePersistenceProvider</provider>
<jar-file>${project.basedir}/target/classes</jar-file>
<properties>
<property name="javax.persistence.jdbc.url" value="jdbc:postgresql://localhost:5432/test_database" />
<property name="javax.persistence.jdbc.driver" value="org.postgresql.Driver" />
<property name="javax.persistence.jdbc.user" value="user" />
<property name="javax.persistence.jdbc.password" value="..." />
</properties>
</persistence-unit>
</persistence>
Then, add filtering of test resources to your pom.xml:
<project>
...
<build>
...
<testResources>
<testResource>
<directory>src/test/resources</directory>
<filtering>true</filtering>
</testResource>
</testResources>
...
</build>
...
</project>
This will work because jar-file can target to directories, not only to jar files.

I prefer the solution of using different persistence.xml for testing and production as Rich Seller post (thanks!!).
But need to change:
<copy file="${project.build.outputDirectory}/META-INF/persistence.xml.proper" tofile="${project.build.outputDirectory}/META-INF/persistence.xml"/>
for:
<move file="${project.build.outputDirectory}/META-INF/persistence.xml.proper" tofile="${project.build.outputDirectory}/META-INF/persistence.xml" overwrite="true"/>
In order persistence.xml.proper not embedded in .jar file

I tried the ClassLoaderProxy approach but had the problem that the JPA annotated classes are not handled as persistent classes by hibernate.
So decided to try it without using persistence.xml. The advantage is that the maven build and the Eclipse JUnit test will work without modifications.
I have a persitent support class for JUnit testing.
public class PersistenceTestSupport {
protected EntityManager em;
protected EntityTransaction et;
/**
* Setup the the {#code EntityManager} and {#code EntityTransaction} for
* local junit testing.
*/
public void setup() {
Properties props = new Properties();
props.put("hibernate.hbm2ddl.auto", "create-drop");
props.put("hibernate.dialect", "org.hibernate.dialect.MySQLDialect");
props.put("hibernate.connection.url", "jdbc:mysql://localhost/db_name");
props.put("hibernate.connection.driver_class", "com.mysql.jdbc.Driver");
props.put("hibernate.connection.username", "user");
props.put("hibernate.connection.password", "****");
Ejb3Configuration cfg = new Ejb3Configuration();
em = cfg.addProperties(props)
.addAnnotatedClass(Class1.class)
.addAnnotatedClass(Class2.class)
...
.addAnnotatedClass(Classn.class)
.buildEntityManagerFactory()
.createEntityManager();
et = em.getTransaction();
}
}
My test classes just extend PersistenceTestSupport and call the setup() in TestCase.setup().
The only drawback is to keep the persistent classes up todate, but for JUnit testing this is acceptable for me.

This answer might sounds silly but I was looking for a way which lets me run those tests from eclipse by Run As -> JUnit Test. This is how I made it:
#BeforeClass
public static void setUp() throws IOException {
Files.copy(new File("target/test-classes/META-INF/persistence.xml"), new File("target/classes/META-INF/persistence.xml"));
// ...
}
I'm just copying the test/persistence.xml to classes/persistence.xml. This works.

Keep two copies of persistence.xml file. One for testing and another for normal build.
The default life cycle copy the build persistence.xml to src/test/resources/META-INF
Create a separate profile which when run will copy the testing persistence.xml to src/test/resources/META-INF

Persistence.xml is used as a starting point to search for entity classes unless you list all classes explicitly and add .
So if you want to override this file with another one, say from src/test/resources, you have to specify every single entity class in this second persistence.xml otherwise no entity class would be found.
Another solution would be to overwrite the file using the maven-resources-plugin ('copy-resources' goal). But then you have to overwrite it twice, once for testing (e.g. phase process-test-classes) and once for the real packaging (phase 'prepare-package').

This is an extension of Rich Seller's answer with proper handling of Hibernate finding multiple persistence.xml files on the classpath and pre-testing state restoration.
Setup:
Create one persistence file for deployment/packaging and one for testing:
src/main/resources/persistence.xml
src/test/resources/persistence-testing.xml
in your pom.xml add this to the plugins section:
<plugin>
<artifactId>maven-antrun-plugin</artifactId>
<version>1.3</version>
<executions>
<execution>
<id>copy-test-persistence</id>
<phase>process-test-resources</phase>
<configuration>
<tasks>
<echo>renaming deployment persistence.xml</echo>
<move file="${project.build.outputDirectory}/META-INF/persistence.xml" tofile="${project.build.outputDirectory}/META-INF/persistence.xml.proper"/>
<echo>replacing deployment persistence.xml with test version</echo>
<copy file="${project.build.testOutputDirectory}/META-INF/persistence-testing.xml" tofile="${project.build.outputDirectory}/META-INF/persistence.xml" overwrite="true"/>
</tasks>
</configuration>
<goals>
<goal>run</goal>
</goals>
</execution>
<execution>
<id>restore-persistence</id>
<phase>prepare-package</phase>
<configuration>
<tasks>
<echo>restoring the deployment persistence.xml</echo>
<move file="${project.build.outputDirectory}/META-INF/persistence.xml.proper" tofile="${project.build.outputDirectory}/META-INF/persistence.xml" overwrite="true"/>
</tasks>
</configuration>
<goals>
<goal>run</goal>
</goals>
</execution>
</executions>
</plugin>
Advantages over other solutions
No extra Java code required
Only one persistence.xml on classpath
Both building and testing work as expected
Describing output on console (echo)
For packaging the state is 100% restored. No leftover files

I'm trying to do the same thing. I have a solution that works for me - yours may vary (and you might not love the solution... it's a bit low-level).
I came across an article on the net where they were using a custom class loader to do something similar which served as inspiration. If anyone can see how to improve then suggestions would be welcome btw. For deployment I rely on container injection of the EntityManager but for testing I create it myself using this code:
final Thread currentThread = Thread.currentThread();
final ClassLoader saveClassLoader = currentThread.getContextClassLoader();
currentThread.setContextClassLoader(new ClassLoaderProxy(saveClassLoader));
EntityManagerFactory emFactory = Persistence.createEntityManagerFactory("test");
em = emFactory.createEntityManager();
Then the ClassLoaderProxy is about as minimal as you can get and just redirects requests for META-INF/persistence.xml to META-INF/test-persist.xml:
public class ClassLoaderProxy extends ClassLoader {
public ClassLoaderProxy(final ClassLoader parent) {
super();
}
#Override
public Enumeration<URL> getResources(final String name) throws IOException {
if (!"META-INF/persistence.xml".equals(name)) {
return super.getResources(name);
} else {
System.out.println("Redirecting persistence.xml to test-persist.xml");
return super.getResources("META-INF/test-persist.xml");
}
}
}
Just to explain this a bit more:
There are two persistence.xml files (one named persistence.xml that is used outside testing and one named test-persist.xml that is used for tests).
The custom class loader is only active for unit tests (for deployment everything is normal)
The custom class loader redirects requests for "META-INF/persistence.xml" to the test version ("META-INF/test-persist.xml").
I was originally hitting some problems because Hibernate will revert back (somehow) to the classloader that was used to load Hibernate (at least I think that is what was going on). I've found that putting the ClassLoader switching code (the first block) as a static block in your Test case it will get loaded before Hibernate but that, depending on your unit test structure you may also need to put the same code in other places (yuck).

Another approach is to use a separate persistence.xml for testing (test/../META-INF/persistence.xml but override the Scanner as follows: -
testing persistence.xml needs to contain
<property name="hibernate.ejb.resource_scanner" value = "...TestScanner" />
Code for new class TestScanner is as follows.
import java.lang.annotation.Annotation;
import java.net.MalformedURLException;
import java.net.URL;
import java.util.Set;
import org.hibernate.ejb.packaging.NamedInputStream;
import org.hibernate.ejb.packaging.NativeScanner;
public class TestScanner extends NativeScanner
{
#Override
public Set <Class <?> >
getClassesInJar (URL jar, Set <Class <? extends Annotation> > annotations)
{ return super.getClassesInJar (getUpdatedURL (jar), annotations); }
#Override
public Set <NamedInputStream>
getFilesInJar (URL jar, Set <String> patterns)
{ return super.getFilesInJar (getUpdatedURL (jar), patterns); }
#Override
public Set <Package>
getPackagesInJar (URL jar, Set <Class <? extends Annotation> > annotations)
{ return super.getPackagesInJar (getUpdatedURL (jar), annotations); }
private URL getUpdatedURL (URL url)
{
String oldURL = url.toExternalForm ();
String newURL = oldURL.replaceAll ("test-classes", "classes");
URL result;
try {
result = newURL.equals (oldURL) ? url : new URL (newURL);
} catch (MalformedURLException e)
{ // Whatever }
return result;
}
}

When using OpenEJB, persistence.xml can be overriden with alternate descriptors: http://tomee.apache.org/alternate-descriptors.html

Another option for this use case would be adding multiple persistence units, one for lets say production and another one for testing and inject the EntityManagerFactory accordingly.
Place both persistence-units into the persistence.xml of the actual project and have your test cases inject the correct EntityManager. The example below illustrates how to do that with guice. Please note that I've left in some mockito mocking for completeness, the mockito specific code has been marked accordingly and is not required for injection.
public class HibernateTestDatabaseProvider extends AbstractModule {
private static final ThreadLocal<EntityManager> ENTITYMANAGER_CACHE = new ThreadLocal<>();
#Override
public void configure() {
}
#Provides
#Singleton
public EntityManagerFactory provideEntityManagerFactory() {
return Persistence.createEntityManagerFactory("my.test.persistence.unit");
}
#Provides
public CriteriaBuilder provideCriteriaBuilder(EntityManagerFactory entityManagerFactory) {
return entityManagerFactory.getCriteriaBuilder();
}
#Provides
public EntityManager provideEntityManager(EntityManagerFactory entityManagerFactory) {
EntityManager entityManager = ENTITYMANAGER_CACHE.get();
if (entityManager == null) {
// prevent commits on the database, requires mockito. Not relevant for this answer
entityManager = spy(entityManagerFactory.createEntityManager());
EntityTransaction et = spy(entityManager.getTransaction());
when(entityManager.getTransaction()).thenReturn(et);
doNothing().when(et).commit();
ENTITYMANAGER_CACHE.set(entityManager);
}
return entityManager;
}
}

put tests in own maven project with its persistence.xml

I'd suggest using different maven profiles where you could filter your database.proprerties files and have one database.properties per profile.
This way you don't have to keep duplicates of any other configuration files except for the .properties.
<properties>
<!-- Used to locate the profile specific configuration file. -->
<build.profile.id>default</build.profile.id>
<!-- Only unit tests are run by default. -->
<skip.integration.tests>true</skip.integration.tests>
<skip.unit.tests>false</skip.unit.tests>
<integration.test.files>**/*IT.java</integration.test.files>
</properties>
<profiles>
<profile>
<id>default</id>
<activation>
<activeByDefault>true</activeByDefault>
</activation>
<properties>
<!--
Specifies the build profile id, which is used to find out the correct properties file.
This is not actually necessary for this example, but it can be used for other purposes.
-->
<build.profile.id>default</build.profile.id>
<skip.integration.tests>true</skip.integration.tests>
<skip.unit.tests>false</skip.unit.tests>
</properties>
<build>
<filters>
<!--
Specifies path to the properties file, which contains profile specific
configuration. In this case, the configuration file should be the default spring/database.properties file
-->
<filter>src/main/resources/META-INF/spring/database.properties</filter>
</filters>
<resources>
<!--
Placeholders found from files located in the configured resource directories are replaced
with values found from the profile specific configuration files.
-->
<resource>
<filtering>true</filtering>
<directory>src/main/resources</directory>
<!--
You can also include only specific files found from the configured directory or
exclude files. This can be done by uncommenting following sections and adding
the configuration under includes and excludes tags.
-->
<!--
<includes>
<include></include>
</includes>
<excludes>
<exclude></exclude>
</excludes>
-->
</resource>
</resources>
</build>
</profile>
<profile>
<id>integration</id>
<properties>
<!--
Specifies the build profile id, which is used to find out the correct properties file.
This is not actually necessary for this example, but it can be used for other purposes.
-->
<build.profile.id>integration</build.profile.id>
<skip.integration.tests>false</skip.integration.tests>
<skip.unit.tests>true</skip.unit.tests>
</properties>
<build>
<filters>
<!--
Specifies path to the properties file, which contains profile specific
configuration. In this case, the configuration file is searched
from spring/profiles/it/ directory.
-->
<filter>src/main/resources/META-INF/spring/profiles/${build.profile.id}/database.properties</filter>
</filters>
<resources>
<!--
Placeholders found from files located in the configured resource directories are replaced
with values found from the profile specific configuration files.
-->
<resource>
<filtering>true</filtering>
<directory>src/main/resources</directory>
<!--
You can also include only specific files found from the configured directory or
exclude files. This can be done by uncommenting following sections and adding
the configuration under includes and excludes tags.
-->
<!--
<includes>
<include></include>
</includes>
<excludes>
<exclude></exclude>
</excludes>
-->
</resource>
</resources>
</build>
</profile>
</profiles>
With the help of surefire for unit tests and failsfe for integration tests, you're done.
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-surefire-plugin</artifactId>
<version>2.12</version>
<configuration>
<junitArtifactName>org.junit:com.springsource.org.junit</junitArtifactName>
<!--see: https://issuetracker.springsource.com/browse/EBR-220-->
<printSummary>false</printSummary>
<redirectTestOutputToFile>true</redirectTestOutputToFile>
<!-- Skips unit tests if the value of skip.unit.tests property is true -->
<skipTests>${skip.unit.tests}</skipTests>
<!-- Excludes integration tests when unit tests are run. -->
<excludes>
<exclude>${integration.test.files}</exclude>
</excludes>
</configuration>
</plugin>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-failsafe-plugin</artifactId>
<version>2.12</version>
<configuration>
<!-- Skips integration tests if the value of skip.integration.tests property is true -->
<skipTests>${skip.integration.tests}</skipTests>
<includes>
<include>${integration.test.files}</include>
</includes>
<forkMode>once</forkMode>
<!--
<reuseForks>false</reuseForks>
<forkCount>1</forkCount>
-->
</configuration>
<executions>
<execution>
<id>integration-test</id>
<goals>
<goal>integration-test</goal>
</goals>
</execution>
<execution>
<id>verify</id>
<goals>
<goal>verify</goal>
</goals>
</execution>
</executions>
</plugin>
Now you need just mvn test for your unit tests and mvn verify -Pintegration for your integration tests.
Obviously you should create the database.properties files in the specified (on the profiles) paths (or elsewhere and change the paths)
Based-on reference: http://www.petrikainulainen.net/programming/tips-and-tricks/creating-profile-specific-configuration-files-with-maven/

I found 2 possibilities without changing classloader/using other Maven plugins/profiles/copy-overwrite files.
TL;DR: check provider name.
At first I started to construct the entityManagerFactory programmatically, like here: create entity manager programmatically without persistence file.
So I did sth very similar:
#BeforeClass
public static void prepare() {
Map<String, Object> configOverrides = new HashMap<>();
configOverrides.put("hibernate.connection.driver_class", "org.h2.Driver");
configOverrides.put("hibernate.connection.url", "jdbc:h2:mem:test;DB_CLOSE_DELAY=-1");
configOverrides.put("hibernate.connection.username", "sa");
configOverrides.put("hibernate.connection.password", "sa");
configOverrides.put("hibernate.dialect", "org.hibernate.dialect.H2Dialect");
configOverrides.put("hibernate.show_sql", "true");
configOverrides.put("hibernate.hbm2ddl.auto", "validate");
factory = new HibernatePersistence().createContainerEntityManagerFactory(
new CustomPersistenceUnitInfo(), configOverrides
);
//factory = Persistence.createEntityManagerFactory("test");
assertNotNull(factory);
}
...
private static class CustomPersistenceUnitInfo implements PersistenceUnitInfo {
#Override
public String getPersistenceUnitName() {
return "test";
}
#Override
public String getPersistenceProviderClassName() {
return "org.hibernate.jpa.HibernatePersistenceProvider";
// <------------note here: this is wrong!
}
#Override
public PersistenceUnitTransactionType getTransactionType() {
return PersistenceUnitTransactionType.RESOURCE_LOCAL;
}
#Override
public DataSource getJtaDataSource() {
return null;
}
#Override
public DataSource getNonJtaDataSource() {
return null;
}
#Override
public List<String> getMappingFileNames() {
return Collections.emptyList();
}
#Override
public List<URL> getJarFileUrls() {
try {
return Collections.list(this.getClass()
.getClassLoader()
.getResources(""));
} catch (IOException e) {
throw new UncheckedIOException(e);
}
}
#Override
public URL getPersistenceUnitRootUrl() {
return null;
}
#Override
public List<String> getManagedClassNames() {
return Arrays.asList(
"com.app.Entity1",
"com.app.Entity2"
);
}
#Override
public boolean excludeUnlistedClasses() {
return true;
}
#Override
public SharedCacheMode getSharedCacheMode() {
return null;
}
#Override
public ValidationMode getValidationMode() {
return null;
}
#Override
public Properties getProperties() {
return null;
}
#Override
public String getPersistenceXMLSchemaVersion() {
return null;
}
#Override
public ClassLoader getClassLoader() {
return null;
}
#Override
public void addTransformer(final ClassTransformer classTransformer) {
}
#Override
public ClassLoader getNewTempClassLoader() {
return null;
}
}
But then, I found it still return null. Why?
Then I found that when I use com.hibernate.ejb.HibernatePersistence class, the provider should not be com.hibernate.jpa.HibernatePersistenceProvider, but com.hibernate.ejb.HibernatePersistence. The class HibernatePersistenceProvider is not even found with IDEA "Open Class", even when it is in the main persistence.xml.
In Ejb3Configuration.class I found:
integration = integration != null ? Collections.unmodifiableMap(integration) : CollectionHelper.EMPTY_MAP;
String provider = (String)integration.get("javax.persistence.provider");
if (provider == null) {
provider = info.getPersistenceProviderClassName();
}
if (provider != null && !provider.trim().startsWith(IMPLEMENTATION_NAME)) { // private static final String IMPLEMENTATION_NAME = HibernatePersistence.class.getName(); which, is, "com.hibernate.ejb.HibernatePersistence"
LOG.requiredDifferentProvider(provider);
return null;
} else {
So I went back to the first solution of persistence.xml, and change provider name, and now it works. It seems that even the provider in main is jpa.xxx, in tests it is not.
So, in summary, 3 things to check:
turn on -X in Maven to check if maven-resources-plugin really copied your src/test/resources/META-INF/persistence.xml into target/test-classes(I think this never fails)
check if hibernate-entitymanager is in your classpath(you can check with mvn dependency:tree -Dincludes=org.hibernate:hibernate-entitymanager.
check provider's name, most important one. Should be org.hibernate.ejb.HibernatePersistence.
<persistence version="2.0"
xmlns="http://java.sun.com/xml/ns/persistence" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://java.sun.com/xml/ns/persistence http://java.sun.com/xml/ns/persistence/persistence_2_0.xsd">
<persistence-unit name="test" transaction-type="RESOURCE_LOCAL">
<provider>org.hibernate.ejb.HibernatePersistence</provider>
<class>com.app.model.Company</class>
...

Related

Testing with Arquillian, how to share Arquillian.xml?

How can the Arquillian configuration file Arquillian.xml be shared between projects and team members?
<arquillian xmlns="http://jboss.org/schema/arquillian"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="
http://jboss.org/schema/arquillian
http://jboss.org/schema/arquillian/arquillian_1_0.xsd">
<container qualifier="jbossas-managed-wildfly-8" default="true">
<configuration>
<property name="jbossHome">C:\test\wildfly-8.1.0.Final</property>
<property name="javaVmArguments">-Djboss.socket.binding.port-offset=2 -Xmx512m -XX:MaxPermSize=128m</property>
<property name="managementPort">9992</property>
</configuration>
</container>
The problem is this points to specific locations on the the disk, and different team members use Wildfly in different locations.
In addition we must duplicate Arquillian.xml for each project that uses it.
We use Arquillian for Maven testing (which could inject the values) and JUnit tests within Eclipse (which cannot inject them).
Any ideas how to do this?
Since there is already Maven support and structure then you can make use of Maven properties and replace of place holder values. It is simple
I guess your Arquillian.xml is under src/test/resources/arquillian.xml right? Then you can replace the absolute values with properties.
<configuration>
<property name="jbossHome">${jboss.home}</property>
</configuration>
The above property can be either defined in the properties section of your pom or can be overridden during mvn executuon using -Djboss.home=C:\myPath
In order though this thing to work, you want Maven automatically for each developer when is about to package arquillian.xml to replace this place-holder ${jboss.home} with a value, that we have either defined on top in the properties section or we have passed it from the command line. This is done through the resource filtering functionality
<build>
<testResources>
<testResource>
<directory>src/test/resources</directory>
<filtering>true</filtering>
</testResource>
<testResources>
</build>
See the simple examples here

Generate java classes with JAXB2 in another maven project

Can someone please help me with my problem?
I have two maven projects A and B.
I want project A to contain all my model classes and in project B i'm creating a contract-first web service with Spring WS. So in project B, I use maven-jaxb2-plugin to generate classes from my schema. It happens that the generate classes in my webservice project (project A) are identical to the ones in my model project (project A) (with no XML annotations).
Because i don't want to have duplicate classes in my web service project (project B), i decided to make this project depends on the model project (project A) and what i want next is, not anymore generate classes to the webservice project but to the model project (project A).
Do you think there isn't another way to do this ?
Can someone please help me doing this if it's possible?
Project A
package project.a;
public class Client {
//...
}
Project B
Class
package project.b;
public class Compte {
//This class manipulates a Client object
//...
}
JAXB2 Maven Plugin
<plugin>
<groupId>org.codehaus.mojo</groupId>
<artifactId>jaxb2-maven-plugin</artifactId>
<version>1.6</version>
<executions>
<execution>
<id>xjc</id>
<goals>
<goal>xjc</goal>
</goals>
</execution>
</executions>
<configuration>
<outputDirectory>src/main/java</outputDirectory>
<packageName>project.a</packageName>
<schemaDirectory>src/main/webapp/WEB-INF/schemas</schemaDirectory>
<clearOutputDir>false</clearOutputDir>
</configuration>
<plugin>
Schema
<element name="client">
<complexType>
<sequence>
<!-- -->
</sequence>
</complexType>
</element>
Project B spring bean configuration
<bean class="org.springframework.ws.client.core.WebServiceTemplate">
<property name="marshaller" ref="marshaller"/>
<property name="unmarshaller" ref="marshaller"/>
<property name="defaultUri" value="http://localhost:8080/project/" />
</bean>
<bean id="marshaller" class="org.springframework.oxm.jaxb.Jaxb2Marshaller">
<property name="contextPath" value="project.a"/>
</bean>
You don't have to recreate model classes again. Just place them in Project A and then import those beans into Project B as:
<import resource="classpath:projectABeanDefinitionFile.xml" />
Assuming you are implementing/going to #XmlRootElement on Compte class (else suggest to go through a tutorial like here),
then change project B file as:
<bean id="marshaller" class="org.springframework.oxm.jaxb.Jaxb2Marshaller">
<property name="classesToBeBound" value="project.b.Compte"/>
</bean>

Filtering a Spring XML file from a dependent project

I have a project MyProject which has a dependency on configuration in another project, BaseProject. Inside BaseProject I have dependencies to many different projects like ErrorLogging, PerformanceLogging, etc... I want to be able to build the top level project (MyProject) and have it filter all the spring xml files in those projects that it has as dependencies. I'm not having any luck. I can see the beans but they are not being filtered. Some of the beans are being filtered with default filters defined in their own poms but non are using the filters from MyProject.
MyProject - This contains the filter files and imports the config from the other projects.
BaseProject - Has spring beans defined which require filtering.
ErrorLogging - Has spring beans defined which require filtering.
When I run a package from MyProject all the spring files are correctly extracted into the jar file but they still contain the property placeholder values ${error.logging.host} for example... The beans in MyProject are correctly filtered. The alternative to this is to define the beans in MyProject but there are about 10 of these projects which use BaseProject and it's beans and I do not want to have to redefine them across 10 seperate projects.
If anyone could shed any light on this issue it'd be great. Thanks
Edit :
To make this clearer, I have a spring beans xml definition inside of the project ErrorLogging called errors-config.xml which defines beans for connecting to databases. This just has place holders for the connection details which should be provided by the filter.properties file that is inside of MyProject.
errors-config.xml is imported as a resource into baseproject-config.xml which sits inside of the BaseProject. Base project and it's config file are imported to MyProject.
I then build MyProject using Maven and I would like the property placeholders inside of errors-config.xml to be replaced with the values in the filter.properties in MyProject. MyProject can successfully filter it's own files but not those of ErrorsLogging project. ErrorsLogging seems to pick up filters from it's own src/main/resources folder instead of that of MyProject.
You can achieve that by unpacking all the dependencies, filtering and packing again, the whole process depends on the structure of your project, for a basic configuration this may suffices:
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-dependency-plugin</artifactId>
<executions>
<execution>
<id>unpack-dependencies</id>
<!--unpack all the dependencies to the target of this project-->
<phase>initialize</phase>
<inherited>false</inherited>
<goals>
<goal>unpack-dependencies</goal>
</goals>
<configuration>
<includeGroupIds>${pom.groupId}</includeGroupIds>
<overWrite>true</overWrite>
<outputDirectory>${project.build.directory}/${artifactId}</outputDirectory>
<includes>**/*.properties,**/*.xml</includes>
</configuration>
</execution>
</executions>
</plugin>
<plugin>
<artifactId>maven-assembly-plugin</artifactId>
<configuration>
<descriptors>
<descriptor>${config.maven.plattform.resources}/assembly/zip.xml</descriptor>
</descriptors>
</configuration>
<executions>
<execution>
<id>zip</id>
<phase>package</phase>
<inherited>true</inherited>
<goals>
<goal>assembly</goal>
</goals>
</execution>
</executions>
</plugin>
This should work as long as you have correctly defined the correct filtering of the resources (which takes places later and also uses the maven-resources-plugin).
You could use the PropertyOverrideConfigurer to override the initial properties.
For example, if you have the folowing datasource definition in errors-config.xml :
<bean id="dataSource" class="com.mchange.v2.c3p0.ComboPooledDataSource">
<property name="driverClass" value="${hibernate.driverClassName}" />
<property name="jdbcUrl" value="${hibernate.url}" />
<property name="user" value="${hibernate.username}" />
<property name="password" value="${hibernate.password}" />
</bean>
You can override the database connection properties in the MyProject context like this :
<bean id="propertyOverideConfigurer" class="org.springframework.beans.factory.config.PropertyOverrideConfigurer">
<property name="locations">
<list>
<value>filter.properties</value>
</list>
</property>
</bean>
And in the filter.properties file you need to specify the bean names and properties you wish to override :
datasource.driverClass = oracle.jdbc.driver.OracleDriver
datasource.jdbcUrl = jdbc:oracle:thin:#localhost:1521:xe
datasource.user = username
datasource.password = password
Hope this helps.

How to instruct Maven to ignore my main/resources/persistence.xml in favor of test/...?

I have two persistence.xml files, for the sake of testing:
src/main/resources/META-INF/persistence.xml
src/test/resources/META-INF/persistence.xml
How to instruct Maven to ignore the first file during testing? Now it is not ignored since OpenEJB says:
ERROR - FAIL ... Finder: #PersistenceContext unitName has multiple matches:
unitName "abc" has 2 possible matches.
Check out the alternate descriptors functionality which is aimed at what you're trying to do.
Try this setup:
src/main/resources/META-INF/persistence.xml
src/main/resources/META-INF/test.persistence.xml
Then you can construct OpenEJB to prefer the test.persistence.xml file by setting the openejb.altdd.prefix System or InitialContext property to test
A different possible solution could be to override the persistence unit properties in the test. With that approach you could avoid the need for a second persistence.xml which can be nice as maintaining two can be a pain.
You can use the Maven approach, but be aware that per spec the persistence provider will only look (aka scan) for #Entity beans in the exact jar or directory where the persistence.xml is found. So be keenly aware that in Maven these are two different locations:
target/classes
target/test-classes
EDIT More details on the overriding capabilities
You can override any property in your test setup via either system properties or the initial context properties (this includes jndi.properties files). The format is:
<unit-name>.<property>=<value>
So for example with the following persistence.xml:
<persistence>
<persistence-unit name="movie-unit">
<provider>org.hibernate.ejb.HibernatePersistence</provider>
<jta-data-source>movieDatabase</jta-data-source>
<non-jta-data-source>movieDatabaseUnmanaged</non-jta-data-source>
<properties>
<property name="hibernate.hbm2ddl.auto" value="create-drop"/>
<property name="hibernate.max_fetch_depth" value="3"/>
</properties>
</persistence-unit>
</persistence>
You can override and add persistence unit properties in your test case. There are currently no facilities for removing them (if you have a need for that let us know – it hasn't really come up so far).
Properties p = new Properties();
p.put(Context.INITIAL_CONTEXT_FACTORY,"org.apache.openejb.client.LocalInitialContextFactory");
p.put("movie-unit.hibernate.hbm2ddl.auto", "update");
p.put("movie-unit.hibernate.dialect", "org.hibernate.dialect.HSQLDialect");
context = new InitialContext(p);
Or alternatively via a jndi.properties file
java.naming.factory.initial=org.apache.openejb.client.LocalInitialContextFactory
movie-unit.hibernate.hbm2ddl.auto = update
movie-unit.hibernate.dialect = org.hibernate.dialect.HSQLDialect
I think you can create two profiles in your pom.xml:
<properties>
<environment>dev</environment>
</properties>
<profiles>
<profile>
<id>prod</id>
<properties>
<environment>test</environment>
</properties>
</profile>
</profiles>
After that, in your src folder, create two folders named dev/resoruces and test/resources and copy your different resources there. After that, add something like this:
<resources>
<resource>
<directory>${basedir}/src/main/resources</directory>
<filtering>false</filtering>
</resource>
<resource>
<directory>${basedir}/src/main/${environment}/resources</directory>
<filtering>true</filtering>
</resource>
</resources>
The ${basedir} depends on the command line parameter, it can be test or dev.
You run the maven command like this: mvn clean package -P test.
I have been testing these and other similar solutions without involving the pom.xml... In my opinion, the best way to solve this issue is to have two application-context.xml (one only to be used in test classes) and to add a custom persistence unit manager bean in the test's application-context.xml. Like this example:
<bean id="pum" class="org.springframework.orm.jpa.persistenceunit.DefaultPersistenceUnitManager">
<property name="persistenceXmlLocation">
<value>classpath*:META-INF/test.persistence.xml</value>
</property>
<property name="defaultDataSource" ref="dataSource"/>
</bean>
This solution runs.
:)
Better add both files - in general, making test/production or debug/profile/production distinction in build makes only trouble.
Better try to use different perasistence unit name for production (say abc-production) and for tests (abc-tests).

Do I need <class> elements in persistence.xml?

I have very simple persistance.xml file:
<?xml version="1.0" encoding="UTF-8"?>
<persistence version="1.0"
xmlns="http://java.sun.com/xml/ns/persistence" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://java.sun.com/xml/ns/persistence http://java.sun.com/xml/ns/persistence/persistence_1_0.xsd">
<persistence-unit name="eventractor" transaction-type="RESOURCE_LOCAL">
<class>pl.michalmech.eventractor.domain.User</class>
<class>pl.michalmech.eventractor.domain.Address</class>
<class>pl.michalmech.eventractor.domain.City</class>
<class>pl.michalmech.eventractor.domain.Country</class>
<properties>
<property name="hibernate.hbm2ddl.auto" value="validate" />
<property name="hibernate.show_sql" value="true" />
</properties>
</persistence-unit>
</persistence>
and it works.
But when I remove <class> elements application doesn't see entities (all classes are annotated with #Entity).
Is there any automatic mechanism to scan for #Entity classes?
The persistence.xml has a jar-file that you can use. From the Java EE 5 tutorial:
<persistence>
<persistence-unit name="OrderManagement">
<description>This unit manages orders and customers.
It does not rely on any vendor-specific features and can
therefore be deployed to any persistence provider.
</description>
<jta-data-source>jdbc/MyOrderDB</jta-data-source>
<jar-file>MyOrderApp.jar</jar-file>
<class>com.widgets.Order</class>
<class>com.widgets.Customer</class>
</persistence-unit>
</persistence>
This file defines a persistence unit
named OrderManagement, which uses a
JTA-aware data source jdbc/MyOrderDB. The jar-file and class elements specify managed persistence classes: entity classes, embeddable classes, and mapped superclasses. The jar-file element specifies JAR files that are visible to the packaged persistence unit that contain managed persistence classes, while the class element explicitly names managed persistence classes.
In the case of Hibernate, have a look at the Chapter2. Setup and configuration too for more details.
EDIT: Actually, If you don't mind not being spec compliant, Hibernate supports auto-detection even in Java SE. To do so, add the hibernate.archive.autodetection property:
<persistence-unit name="eventractor" transaction-type="RESOURCE_LOCAL">
<!-- This is required to be spec compliant, Hibernate however supports
auto-detection even in JSE.
<class>pl.michalmech.eventractor.domain.User</class>
<class>pl.michalmech.eventractor.domain.Address</class>
<class>pl.michalmech.eventractor.domain.City</class>
<class>pl.michalmech.eventractor.domain.Country</class>
-->
<properties>
<!-- Scan for annotated classes and Hibernate mapping XML files -->
<property name="hibernate.archive.autodetection" value="class, hbm"/>
<property name="hibernate.hbm2ddl.auto" value="validate" />
<property name="hibernate.show_sql" value="true" />
</properties>
</persistence-unit>
In Java SE environment, by specification you have to specify all classes as you have done:
A list of all named managed persistence classes must be specified in Java SE environments to insure portability
and
If it is not intended that the annotated persistence classes contained in the root of the persistence unit be included in the persistence unit, the exclude-unlisted-classes element should be used. The exclude-unlisted-classes element is not intended for use in Java SE environments.
(JSR-000220 6.2.1.6)
In Java EE environments, you do not have to do this as the provider scans for annotations for you.
Unofficially, you can try to set <exclude-unlisted-classes>false</exclude-unlisted-classes> in your persistence.xml. This parameter defaults to false in EE and truein SE. Both EclipseLink and Toplink supports this as far I can tell. But you should not rely on it working in SE, according to spec, as stated above.
You can TRY the following (may or may not work in SE-environments):
<persistence-unit name="eventractor" transaction-type="RESOURCE_LOCAL">
<exclude-unlisted-classes>false</exclude-unlisted-classes>
<properties>
<property name="hibernate.hbm2ddl.auto" value="validate" />
<property name="hibernate.show_sql" value="true" />
</properties>
</persistence-unit>
Do I need Class elements in persistence.xml?
No, you don't necessarily. Here is how you do it in Eclipse (Kepler tested):
Right click on the project, click Properties, select JPA, in the Persistence class management tick Discover annotated classes automatically.
For those running JPA in Spring, from version 3.1 onwards, you can set packagesToScan property under LocalContainerEntityManagerFactoryBean and get rid of persistence.xml altogether.
Here's the low-down
You can provide for jar-file element path to a folder with compiled classes. For example I added something like that when I prepared persistence.xml to some integration tests:
<jar-file>file:../target/classes</jar-file>
for JPA 2+ this does the trick
<jar-file></jar-file>
scan all jars in war for annotated #Entity classes
Hibernate doesn't support <exclude-unlisted-classes>false</exclude-unlisted-classes> under SE, (another poster mentioned this works with TopLink and EclipseLink).
There are tools that will auto-generate the list of classes to persistence.xml e.g. the Import Database Schema wizard in IntelliJ. Once you've got your project's initial classes in persistence.xml it should be simple to add/remove single classes by hand as your project progresses.
Not sure if you're doing something similar to what I am doing, but Im generating a load of source java from an XSD using JAXB in a seperate component using Maven. Lets say this artifact is called "base-model"
I wanted to import this artifact containing the java source and run hibernate over all classes in my "base-model" artifact jar and not specify each explicitly. Im adding "base-model" as a dependency for my hibernate component but the trouble is the tag in persistence.xml only allows you to specify absolute paths.
The way I got round it is to copy my "base-model" jar dependency explictly to my target dir and also strip the version of it. So whereas if I build my "base-model" artifact it generate "base-model-1.0-SNAPSHOT.jar", the copy-resources step copies it as "base-model.jar".
So in your pom for the hibernate component:
<!-- We want to copy across all our artifacts containing java code
generated from our scheams. We copy them across and strip the version
so that our persistence.xml can reference them directly in the tag
<jar-file>target/dependency/${artifactId}.jar</jar-file> -->
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-dependency-plugin</artifactId>
<version>2.5.1</version>
<executions>
<execution>
<id>copy-dependencies</id>
<phase>process-resources</phase>
<goals>
<goal>copy-dependencies</goal>
</goals>
</execution>
</executions>
<configuration>
<includeArtifactIds>base-model</includeArtifactIds>
<stripVersion>true</stripVersion>
</configuration>
</plugin>
Then I call the hibernate plugin in the next phase "process-classes":
<!-- Generate the schema DDL -->
<plugin>
<groupId>org.codehaus.mojo</groupId>
<artifactId>hibernate3-maven-plugin</artifactId>
<version>2.2</version>
<executions>
<execution>
<id>generate-ddl</id>
<phase>process-classes</phase>
<goals>
<goal>hbm2ddl</goal>
</goals>
</execution>
</executions>
<configuration>
<components>
<component>
<name>hbm2java</name>
<implementation>annotationconfiguration</implementation>
<outputDirectory>/src/main/java</outputDirectory>
</component>
</components>
<componentProperties>
<persistenceunit>mysql</persistenceunit>
<implementation>jpaconfiguration</implementation>
<create>true</create>
<export>false</export>
<drop>true</drop>
<outputfilename>mysql-schema.sql</outputfilename>
</componentProperties>
</configuration>
</plugin>
and finally in my persistence.xml I can explicitly set the location of the jar thus:
<jar-file>target/dependency/base-model.jar</jar-file>
and add the property:
<property name="hibernate.archive.autodetection" value="class, hbm"/>
It's not a solution but a hint for those using Spring:
I tried to use org.springframework.orm.jpa.LocalContainerEntityManagerFactoryBean with setting persistenceXmlLocation but with this I had to provide the <class> elements (even if the persistenceXmlLocation just pointed to META-INF/persistence.xml).
When not using persistenceXmlLocation I could omit these <class> elements.
I'm not sure this solution is under the spec but I think I can share for others.
dependency tree
my-entities.jar
Contains entity classes only. No META-INF/persistence.xml.
my-services.jar
Depends on my-entities. Contains EJBs only.
my-resources.jar
Depends on my-services. Contains resource classes and META-INF/persistence.xml.
problems
How can we specify <jar-file/> element in my-resources as the version-postfixed artifact name of a transient dependency?
How can we sync the <jar-file/> element's value and the actual transient dependency's one?
solution
direct (redundant?) dependency and resource filtering
I put a property and a dependency in my-resources/pom.xml.
<properties>
<my-entities.version>x.y.z-SNAPSHOT</my-entities.version>
</properties>
<dependencies>
<dependency>
<!-- this is actually a transitive dependency -->
<groupId>...</groupId>
<artifactId>my-entities</artifactId>
<version>${my-entities.version}</version>
<scope>compile</scope> <!-- other values won't work -->
</dependency>
<dependency>
<groupId>...</groupId>
<artifactId>my-services</artifactId>
<version>some.very.sepecific</version>
<scope>compile</scope>
</dependency>
<dependencies>
Now get the persistence.xml ready for being filtered
<?xml version="1.0" encoding="UTF-8"?>
<persistence ...>
<persistence-unit name="myPU" transaction-type="JTA">
...
<jar-file>lib/my-entities-${my-entities.version}.jar</jar-file>
...
</persistence-unit>
</persistence>
Maven Enforcer Plugin
With the dependencyConvergence rule, we can assure that the my-entities' version is same in both direct and transitive.
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-enforcer-plugin</artifactId>
<version>1.4.1</version>
<executions>
<execution>
<id>enforce</id>
<configuration>
<rules>
<dependencyConvergence/>
</rules>
</configuration>
<goals>
<goal>enforce</goal>
</goals>
</execution>
</executions>
</plugin>
Not necessarily in all cases.
I m using Jboss 7.0.8 and Eclipselink 2.7.0. In my case to load entities without adding the same in persistence.xml, I added the following system property in Jboss Standalone XML:
<property name="eclipselink.archive.factory" value="org.jipijapa.eclipselink.JBossArchiveFactoryImpl"/>

Categories