Getting Intimate with Spring Boot and Hibernate

In my previous blog , we looked at how to get started with a simple Spring Boot and Hibernate application. We managed to get our application up and running in a few minutes.

In this article, we will look at how and what Spring Boot does behind the scenes.

Basic requirement

To setup a Spring based application with Hibernate/JPA we usually need the following:

1. Datasource  which connects to a database and details like the database url, password, username – this is really independent of whether we use Spring, JPA.
2. JPA EntityManager to perform repository(CRUD) related operations.
3. Vendor specific (Hibernate) properties
4. Transaction support.

Let us take a look at how these things get configured.

In case of an in memory H2 database, all of the above was configured automatically by Spring Boot. We did not write a single configuration either using Java configuration or XML. Well, how does Spring Boot do everything for us?  Let’s take a look :

Vanishing (configuration) act explained :

1. The starting point is the spring.factories file. This file has a Auto Configure section which Spring Boot uses to determine what should be auto configured. This file is in the META-INF folder which is part of the spring-boot-autoconfigure-<version>.RELEASE.jar. The SpringFactoriesLoader class loads the spring.factories file.

2. Since we are talking about JPA/Hibernate, the spring.factories contains a (among other autoconfigurations) key value pair: org.springframework.boot.autoconfigure.EnableAutoConfiguration=\org.springframework.boot.autoconfigure.orm.jpa.HibernateJpaAutoConfiguration

3. The key EnableAutoConfiguration above is triggered using the @SpringBootApplication annotation on the SpringBootJpaApplication.java file. This is our starting point in our application.

4. The value part above, HibernatJpaAutoConfiguration class which is part of the Spring framework looks like

One of the important annotations above is on line 4 :

@AutoConfigureAfter({ DataSourceAutoConfiguration.class }).

This indicates that we should perform HibernateJpaAutoConfiguration only after reading the DataSourceAutoConfiguration.class or only if a Datasource is found on the classpath. It would make no sense to configure JPA and Hibernate without a datasource/database !

5. The DataSourceAutoConfiguration looks as follows:

6. This is a normal Spring Configuration class which sets up the DataSource only if all of the above annotations satisfy our criteria. We used an H2 database and hence the DataSource.class and EmbeddedDatabaseType.class satisfies the condition.

7. Now the configuration moves back to step 4 where we needed the presence of  datasource/database at minimum. Now on to the next part where the following is checked :

@ConditionalOnClass({ LocalContainerEntityManagerFactoryBean.class, EntityManager.class })

This is configuration specific to JPA classes and the @ConditionalOnClass checks if these classes are there on the class path. There are present in our case through the Spring Data JPA library.

8.Next comes the @Conditional(HibernateEntityManagerCondition.class) . This checks if the JPA provider is Hibernate, this is done via the HibernateEntityManagerCondition.java. In our case it is true since Spring Boot brings in Hibernate as the default provider of JPA and HibernateEntityManager.java is on the classpath.

9.Once these conditions are satisfied, the HibernateJpaAutoConfiguration extends the JpaBaseConfiguration class.

This is annotated with the @EnableConfigurationProperties(JpaProperties.class).  This JpaProperties.class is annotated with @ConfigurationProperties(prefix = “spring.jpa”) . This gets triggered if we were to add any spring.jpa.xxx properties in our application.properties file. We would use this if we were to add any jpa specific properties. We did not add any and hence the default is assumed.

10.The JpaBaseConfiguration class contains all the other remaining configurations. It primarily uses the @ConditionalOnMissingBean annotation to configure the EntityManager, JpaVendorAdapter, LocalContainerEntityManagerFactoryBean, JpaTransactionManager if they have already not been configured. The @ConditionalOnMissingBean when applied on a method means – if this bean has not been configured, then execute the method and configure it.

So in this way, datasource, entitymanager, vendor(hibernate) specific properties and transactions get configured. If you have setup a Spring Hibernate application before, we would do the same using Java configuration or an xml and configure the above mentioned properties. You can configure any specific properties and in that case Spring Boot will skip auto configuring that property.

Conclusion 

All the Harry Houdini illusions or Spring Boot magic starts from the spring.factories file and then through the usage of some powerful annotations like @ConditionalOnMissingBean, ConditionalOnClass,@AutoConfigureAfter , Jpa/Datasource/Transactions gets configured depending on what is found in the classpath. Spring Boot will not configure a property or a bean if it was already configured.

Complaining is a Habit ?

I have heard people complaining about the so-called magic in Spring Boot. Well, when Spring folks gave us xml configurations, we complained ! They gave us configurations via annotations and Java Configuration to get rid of the xml, we still complained ! Now, things get configured automatically…..and we still complain. Complaining can be good at times, the folks at Pivotal have been listening to our complaints and have been giving us fantastic tools/ frameworks. But let’s not forget that the details are out there, let’s explore a little more before complaining.