I have a bunch of tests in my project that are all annotated with #SpringBootTest and therefore load up a SpringBoot context.
Now recently I refactored a Test in which I wanted a smaller scope (it´s about process coverage with camunda) to #RunWith(SpringJUnit4ClassRunner.class).
Since this means that no context is loaded automatically I create some beans "manually" with a static inner class configuration. The entire test looks something like this:
#RunWith(SpringJUnit4ClassRunner.class)
#ContextConfiguration(classes = {
ExternalConfiguration.class, MyTest.InternalConfiguration.class
})
public class MyTest{
#Autowired
private SomeBean someInternalBean;
#Configuration
public static class InternalConfiguration{
#Bean
SomeBean someInternalBean() {
return mock(SomeBean .class);
}
}
//Tests
Now, this test runs fine when I run it. BUT when I run any other test ( those still annotated with #SpringBootTest), I get issues with when the ApplicationContext is loaded:
The bean 'someInternalBean', defined in class path resource [.../MyTest$InternalConfiguration.class], could not be registered. A bean with that name has already been defined in file [.../SomeBean.class] and overriding is disabled.
Apparently a bean is created when loading the ApplicationContext because the class is annotated with #Component AND the context loader tries to create another bean from my internal configuration.
I cant allow bean-overriding because my mock beans might overwrite the automatically created beans (which they do, I tried).
How do I circumvent this? I want my SpringJUnit4ClassRunner-tests with their internal configurations to not affect my other #SpringBootTest-tests. I already tried making the configuration beans conditional with #ConditionalOnMissingBean but that did not work.
Turns out those inner configuration classes should not be annotated with #Configuration. Removing the annotation makes it so that the manual bean generation still works and the configuration is no longer picked up by the componentScan.
Related
Is ApplicationContext automatically instantiated in Spring?
If I have my bean defined like this
#Component
public class Car{
...
}
and then I have my config class which tells Spring container where to look for beans through the annotation #ComponentScan
#Configuration
#ComponentScan
public class AppConfig {
...
}
Is Spring automatically creating a context loading all my beans? Or do I have to create it programmatically? If so how do I do it, with something like this?
#Configuration
#ComponentScan
public class AppConfig {
ApplicationContext context = new AnnotationConfigApplicationContext(AppConfig.class);
context.getBean(Car.class);
...
}
Even doing this, there may be a problem, because every time I need the context I have to call new AnnotationConfigApplicationContext...
what is the recommended way to instantiate the context and making him available inside the whole project, maybe as a bean like inside Spring boot app where i can just autowire it.
How Spring Boot can initialize it, load all the beans and let the context available as a bean, ready to be autowired?
No, Application Context isn't automatically instantiated, if you're having a simple and basic Spring Core application. Moreover, your #Configuration class won't scan anything and won't create any beans, if you don't create your Spring Container/Context explicitly with that #Configuration class.
There are several ways of creating Application Context, but the most popular and traditional ones are:
ApplicationContext context = new ClassPathXmlApplicationContext(applicationContext.xml) - implying, that you have your container configuration in the applicationContext.xml file;
ApplicationContext context = new AnnotationConfigApplicationContext(ConfigClass.class); - implying, that your ConfigClass is the #Configuration class.
However, if you have the Spring Boot application annotated with #SpringBootApplication, then the Application Context will be automatically instantiated for you, because:
#SpringBootApplication annotation consists of:
#EnableAutoConfiguration - which enables Spring Boot’s auto-configuration mechanism;
#ComponentScan - which enable #Component scan on the package where the application is located;
#Configuration - allows to register extra beans in the context or import additional configuration classes.
and this will spin up the context for you.
You can obtain the reference to the Spring Context created by Spring Boot, by the factory method you have in your main method: SpringApplication.run(MainClass.class, args);
This returns the reference to the Application Context and you can assign it to variable like this:
ApplicationContext context = SpringApplication.run(MainClass.class, args)
I am trying to create a common library that includes several stuff needed by my microservices. One of those things is the ACL functionality provided with spring-security. My initial thought was to initialize all ACL-related beans from a #Configuration file in the common library and each time a microservice needs this functionality i could use the #Import annotation(to my microservice project) to "enable" it.
Some of these beans require the famous javax.sql.DataSource to work, so in my common library i autowired it as follows:
#Configuration
public class AclConfiguration {
#Autowired
DataSource dataSource
When i decide that i want this configuration to take place i go to my microservice project (let's say RulesApplication) and on the main class (annotated with #SpringBootApplication) i do the following
#SpringBootApplication
#EnableJpaRepositories
#EnableJpaAuditing
#EnableCaching
#Import(AclConfiguration.class)
public class RulesApplication {
.
.
.
The problem is that the DataSource bean cannot be seen from the common library, although it is being created as expected (validated just by removing the #Import).
Everytime i import the configuration from the common library i get a :
Caused by: java.lang.IllegalArgumentException: DataSource required
at org.springframework.util.Assert.notNull(Assert.java:198) ~[spring-core-5.2.5.RELEASE.jar:5.2.5.RELEASE]
Indicating that the DataSource bean is null.
What am i missing here?
Coming up from some digging.. There was never a real problem with the DataSource bean. All the frustration was created by double-defining another Bean in the same class, which led to all the #Autowired beans failure to initialize.
As a result from this research(as other posts mentioned) the bean initialization is working smoothly between shared projects, so most of the times this error will occur from double-defining/badly-defining other beans.
What does this annotation do?
When would I want to use it?
When would I not want to use it?
#RunWith(SpringJUnit4ClassRunner.class)
I can find more usages of this when I Google and do not find a 101 explanation as to what this annotation is supposed to communicate to me or when/why I would use it?
The annotation is used to configure a unit test that required Spring's dependency injection.
From Spring Reference - 10. Unit Testing:
10.1 Creating a Unit Test Class
In order for the unit test to run a batch job, the framework must load the job's ApplicationContext. Two annotations are used to trigger this:
#RunWith(SpringJUnit4ClassRunner.class): Indicates that the class should use Spring's JUnit facilities.
#ContextConfiguration(locations = {...}): Indicates which XML files contain the ApplicationContext.
If you are using annotations rather than XML files, then any class that you are unit testing that requires Spring dependency injection needs to be put into the #ContextConfiguration annotation. For example:
#RunWith(SpringJUnit4ClassRunner.class)
#ContextConfiguration(classes = FooManager.class)
class FooManagerTest {
#Autowired
FooManager fooManager;
Now when you use fooManager in a unit test it will have have a Spring context setup for it.
If fooManager autowires in any beans then those bean's classes also need to be in the #ContextConfiguration annotation. So if fooManager autowires in a FooReporter bean:
#ContextConfiguration(classes = {FooManager.class, FooReporter.class})
If the beans that fooManager autowires in contain state, then you will likely want to reset the state of those beans for each test. In that case you can add the #DirtiesContext annotation to your test class:
#DirtiesContext(classMode = DirtiesContext.ClassMode.AFTER_EACH_TEST_METHOD)
If fooManager or any of its autowired beans reads Spring config then you need to add an initializers list to the #ContextConfiguration annotation, that contains the ConfigFileApplicationContextInitializer class:
#ContextConfiguration(classes = {FooManager.class, FooReporter.class}, initializers = ConfigFileApplicationContextInitializer.class)
To answer the when you would and wouldn't want to use it part of the question.
When to use SpringJUnit4ClassRunner
IMO SpringJUnit4ClassRunner should be used very sparingly. There is a significant overhead involved with starting up a Spring container to run a unit test.
I typically use SpringJUnit4ClassRunner to test:
that components are injected (auto-wired) as expected
that configuration data is injected as expected
When you are injecting components issues can arise if the #Qualifier annotation is not used or used incorrectly, for example.
When loading configuration from multiple yaml files you may want to test that maps are being merged as expected, with the appropriate overrides occurring.
At the very least I always have a simple SpringJUnit4ClassRunner test as a sanity check that the Spring container starts up OK.
When not to use SpringJUnit4ClassRunner
I would not use SpringJUnit4ClassRunner to test the non-Spring related functionality in my code under test. Which in my experience means most of the functionality.
So this means that any autowired components and injected config data needs to be mocked. This can mean quite a bit of setup code for your unit tests. However this setup code only needs to be written once for all the tests in your class under test. It is also much quicker to run unit tests with mocked components.
I keep the mocking simple and use Spock to mock the components. Example groovy code:
import spock.lang.Specification
class FooManagerTest extends Specification {
FooManager cut
void createMockFooReporter() {
FooReporter mockFooReporter = Mock(FooReporter)
mockFooReporter.fooFormatter = Mock(FooFormatter)
}
void setup() {
cut = new FooManager()
cut.fooReporter = createMockFooReporter()
}
void "Basic test"() {
// Do a basic test using 'cut'
}
}
In this example the class under test FooManager has an autowired FooReporter which itself contains an autowired FooFormatter.
I think #RunWith annotation is in order to initialize the context of spring. Because the junit5 is released, you just can replace it with #SpringJUnitConfig.By the way, #RunWith annotation is already replaced by #ExtendWith, but you still can use it.
I have a TestNG listener that implements IInvokedMethodListener. I would like to wire in a Spring bean inside this listener and use it. Unfortunately, this class is instantiated by TestNG and so Spring cannot wire anything in that is annotated using #Autowired. I tried implementing ApplicationContextAware, but that doesn't work either.
Is there a way to wire Spring beans into classes that implement IInvokedMethodListener?
ApplicationContextAware only works for Spring Beans. You can use #Configurable, but that requires AspectJ.
Here's a simple hack that should work: Add a static member to your listener class and inject it via a non-static setter.
public class MyMethodListener implements IInvokedMethodListener {
private static MyBean myBean;
#Autowired
public void setMyBean(MyBean myBean) {
MyMethodListener.myBean = myBean;
}
}
Include a bean of the required type in your application context.
The listener instantiated by TestNG will not be the same instance as the one from the Spring context, but it will have the static member set, provided that context creation has finished before TestNG instantiates the listener.
I had the same problem recently, it is basically Listeners are not maintained by spring. So I did some googling around this concept like "Injecting beans into classes not managed by spring", I got https://dzone.com/articles/autowiring-spring-beans-into-classes-not-managed-by-spring link which explains exactly the same problem and the solution. It worked for me.
In Guice, I have a ProductionModule with my bindings. In my tests, I can load that PLUS a module that overrides a few of the production bindings with mock objects. How do I do such a thing in spring....
For example, load production-spring.xml in test files and then have the test load in test-spring.xml which would ONLY override some of the bindings in production-spring.xml
This tests the integration and make sure changes in production-spring.xml don't break things. These are more automated integration tests then testing a unit and work extremely well.
You can override beans by listing multiple xml files. Beans in the later files will override those loaded before.
#RunWith(SpringJUnit4ClassRunner.class)
#ContextConfiguration(
locations = {"classpath:prodDB.xml",
"classpath:applicationContext.xml",
"classpath:testDb.xml"})
public class SpringTest {
#Autowired
protected DataSource dataSource; //uses the datasource from testDb.xml
}
So in this case testDB.xml overrides the DataSource configured in prodDb.xml. This applies even if you don't use the SpringJUnit4ClassRunner as well:
new ClassPathXmlApplicationContext(new String[]
{"classpath:prodDb.xml",
"classpath:testDb.xml"});
Using a tool like Constretto you can do the same with annotated beans:
#Service
public class FooService...
#Service
#Environment("test")
public class FakeFooService ...
Now, if you run a test with the #Environment("test") annotation on the class, the FakeFooService will be used.
In test environment, you can add your Spring context xml with with overwritten beans before other declarations. And there is also external properties you can change for each environment.