In CDI, how do I configure my beans?
Let's say I have this code:
class JawaBotApp {
private void init( String configFilePathString ) throws JawaBotException {
ConfigBean cb = new JaxbConfigPersister(configFilePathString).load();
JawaBotApp.jawaBot = JawaBot.create( cb );
}
}
class JawaBot {
public static JawaBot create( ConfigBean cb ) throws JawaBotException{
JawaBot bot = new JawaBot();
bot.applyConfig(cb);
bot.init();
return bot;
}
}
How would I convert it so both could be CDI beans?
I thought about annotating the create() with #Produces, however that would need to have it non-static, and rewrite it so the ConfigBean gets injected, which would need to rewrite JaxbConfigPersister or create a wrapper object... Too much work for nothing.
Is there better approach?
Something like:
class JawaBot {
#Inject public JavaBot(#JawaConfig String configFilePathString) {
...
}
}
Then you just need to produce an #JawaConfig String somewhere which represents your configuration. This could then be mocked out with something like an alternative or specialize to give you a diff config, or just some silly little #Producer that looks at some other external file/setting.
Then you just #Inject a JawaBot when you want it instead of all the other setup, and everything just lives in the injected constructor.
Related
I am trying to write integration test case for one of my rest application which uses mongodb internally to persist the data
#DataMongoTest
#SpringBootTest(webEnvironment = WebEnvironment.RANDOM_PORT)
public class MainControllerTest {
#LocalServerPort
private int port = 8080;
/* some test cases*/
}
but I am getting below error
java.lang.IllegalStateException: Configuration error: found multiple declarations of #BootstrapWith for test class [com.sample.core.controller.MainControllerTest]: [#org.springframework.test.context.BootstrapWith(value=class org.springframework.boot.test.autoconfigure.data.mongo.DataMongoTestContextBootstrapper), #org.springframework.test.context.BootstrapWith(value=class org.springframework.boot.test.context.SpringBootTestContextBootstrapper)]
looks like these two are mutually exclusive, so how to do the integration testing .
Use #AutoConfigureDataMongo with #SpringBootTest and this will resolve this ambiguity issue. #SpringBootTest and #DataMongoTest cannot be used together.
Answering to a very old post hoping it may help others.
#AutoConfigureDataMongo will connect to real database. In order to still use the embedded mongo, one can initiate the embedded mongoDb manually.
#SpringBootTest(classes = SubscriptionEventApplication.class, webEnvironment = SpringBootTest.WebEnvironment.RANDOM_PORT)
class SubscriptionEventApiIntegrationTest {
#BeforeAll
static void setup() throws Exception {
startEmbeddedMongoDbManually();
}
private static void startEmbeddedMongoDbManually() throws IOException {
final String connectionString = "mongodb://%s:%d";
final String ip = "localhost";
final int port = 27017;
ImmutableMongodConfig mongodConfig = MongodConfig
.builder()
.version(Version.V3_5_5)
.net(new Net(ip, port, Network.localhostIsIPv6()))
.build();
MongodStarter starter = MongodStarter.getDefaultInstance();
mongodExecutable = starter.prepare(mongodConfig);
mongodExecutable.start();
mongoTemplate = new MongoTemplate(MongoClients.create(String.format(connectionString, ip, port)), "test");
}
#AfterAll
static void clean() {
mongodExecutable.stop();
}
#Test
public void test() {
.....
}
}
Purushothaman suggested starting embedded MongoDB server manually. I am suggesting to start it automatically using #DataMongoTest, but creating WebTestClient manually instead.
Kotlin code below, translates to Java trivially:
#DataMongoTest
// #ContextConfiguration may not be needed for your case.
#ContextConfiguration(
classes = [
Application::class,
MainController::class,
// Add more needed classes for your tests here.
// ...
]
)
#TestPropertySource(properties = ["spring.mongodb.embedded.version=4.0.12"])
class MainControllerTest(
#Autowired
private val mainController: MainController,
// Add more beans needed for your tests here.
// ...
) {
// Creating a WebTestClient is easy and
// can be done in different ways.
// Here is one of the possible ways.
private val webTestClient: WebTestClient =
WebTestClient.bindToController(mainController).build()
#Test
fun someTest() {
// ...
}
}
Class that i want to mock:
TestClass.java
public class testClass(){
public String getDescription(String input){
String value = this.getDetails(input); // i am not going to change this line, hence want to mock this.
//below this i have some complexity logic, which i would like to fix cyclomatic complexity issue
}
private String getDetails(String input){
return "More details for the "+input;
}
}
My questions is how do i mock "this.getDetails(input)" to return some string for testing purpose?
If you've got a class that is big and complex enough that you need to mock a small piece of it, take that as a hint that you're violating the Single Responsibility Principle and properly split up the classes. If you use dependency injection, you can then supply whatever implementation you'd like.
public class TestClass {
/**
* Computes a detail string based on an input. Supply this in the constructor
* for full DI, relax visibility, or add a setter.
*/
private final Function<String, String> detailFunction;
public String getDescription(String input){
String value = detailFunction.apply(input);
// ...
}
}
As a lightweight alternative, you can test an override or spy of your actual class.
#Test public void testTestClassWithOverride() {
TestClass instanceUnderTest = new TestClass() {
#Override public String getDescription(String input) {
return "Predictable value";
}
};
// test your instanceUnderTest here
}
#Test public void testTestClassWithSpy() {
TestClass spyUnderTest = Mockito.spy(new TestClass());
doReturn("Predictable value").when(spyUnderTest).getDescription(anyString());
// test your spyUnderTest here
}
Bear in mind that, though this is an option for you, it shouldn't be your first option: Rather than testing your actual class, you're testing a one-off variant of it, and you've made it so other consumers can subclass your TestClass as well. If possible, write the flexibility you need into the class itself and treat your test as a consumer that plays by the same rules.
First of all, it is a bad practice to make a so-called "partials mocks". This illustrates that your code doesn't follow single responsibility principle that leads to your code being not (or hardly) testable.
I would suggest you to extract getDescription method from your class and use it indirectly via dependency inversion or more concrete - dependency injection (for instance by employing Spring Framework):
public class TestClass() {
private DetailsServiceProvider detailsServiceProvider;
public TestClass(DetailsServiceProvider detailsServiceProvider) {
this.detailsServiceProvider = detailsServiceProvider;
}
public String getDescription(String input) {
String value = detailsServiceProvider.getDetails(input); // i am not going to change this line, hence want to mock this.
//below this i have some complexity logic, which i would like to fix cyclomatic complexity issue
}
}
public interface DetailsServiceProvider {
String getDetails(String input);
}
public class DetailsServiceProviderImpl implements DetailsServiceProvider{
#Override
public String getDetails(String input) {
return "More details for the "+input;
}
}
Then in your test, you could simply:
#Test
public void test() {
DetailsServiceProvider mockedProvider = Mockito.mock(DetailsServiceProvider.class);
//TODO: add scenarios for the mocked object
TestClass target = new TestClass(mockedProvider);
String description = target.getDescription();
//TODO: add assertions
}
If you do not want to struggle with the preferred approach you could use #Spy in Mockito. This will create exactly what you want - a partial mock for your object where part of the methods will be real and another part - mocks:
#Test
public void test() {
TestClass partialMockedObject = Mockito.spy(new TestClass());
Mockito.doReturn("test details").when(partialMockedObject).getDetails();
String description = partialMockedObject.getDescription();
//TODO: add assertions
}
Again, this method is not desired but can be used if no other options are given. Note that this requires getDetails() to be visible in tests, meaning that the private modifier won't work here.
I am making a framework for making fractals in processing, however, I need to use functions as parameters for a constructor of a class.
Something like:
class Fractal {
String name;
void initialize;
Fractal(String Name, void setup) {
...
}
}
I'm going to guess you're coming from a JavaScript background?
Traditionally, Java didn't really have a way to do this. Instead you'd pass an anonymous instance of an interface, like this:
interface Runner{
public void run();
}
class Fractal {
String name;
Runner initialize;
Fractal(String name, Runner setup) {
...
}
}
Runner r = new Runner(){
public void run(){
// whatever
}
}
Fractal fractal = new Fractal("name here", r);
Note that Java provides a Runnable interface that you can use instead of creating your own, but I wanted to spell it out here to make it more obvious.
As of Java 8, you can pass a reference to a function as a parameter. This is called a lambda function. Googling "Java lambda function" will return a ton of results.
From this answer:
public void pass() {
run(()-> System.out.println("Hello world"));
}
public void run(Runnable function) {
function.run();
}
Depending on how you're using Processing, you might be stuck with the first approach though, since I don't think the Processing editor supports Java 8 yet.
My application has configurations which are loaded using parsing annotations into a file using Jackson's fasterxml annotations. For example:
public class RootConfiguration extends Configuration {
#JsonProperty
#NotEmpty
public String foo;
#JsonProperty
public BarConfiguration bar;
public class BarConfiguration extends Configuration {
#JsonProperty
public String baz;
}
}
The configuration is then injected into providers in my Module that help me bind those properties to places in the code that use them. Like so:
#Provides
#Named("config")
public RootConfiguration provideRootConfiguration(RootConfiguration configuration) {
return configuration;
}
#Provides
#Named("config.foo")
public String provideFooConfiguration(RootConfiguration configuration) {
return configuration.foo;
}
#Provides
#Named("config.bar")
public BarConfiguration provideBarConfiguration(RootConfiguration configuration) {
return configuration.bar;
}
And so on.
I'm looking for a framework to help me avoid this tedious work.
I would imagine something that looks like this:
#Configuration(value = "config", bindSubProperties = true)
public class RootConfiguration extends Configuration {
...
That would use Reflection to bind any sub fields in my class as guice Names.
I've looked into Governator's annotations for configurations but as far as I can see they need to be applied to every configuration that I want to bind, which saves me some coding, but is essentially the same (I still have to manually specify the path for each and every configuration I want to bind).
Before I roll out my own implementation for this, is there something that will give me what I need?
Note: I'm using this for a Dropwizard project so the constraint on using Jackson to map the configuration to POJOs is rather tight (unless I move the application configuration outside of the config yaml).
I don't know of any tool that would do this for you, but you could do it yourself pretty easily with something like this:
void bindConfiguration() {
for (Field field : RootConfiguration.class.getFields() {
bindConfiguration(TypeLiteral.get(field.getGenericType()), field);
}
}
<T> void bindConfiguration(TypeLiteral<T> type, Field field) {
bind(type)
.annotatedWith(Names.named("config." + field.getName()))
.toProvider(new ConfigurationProvider<T>(field))
.in(Singleton.class);
}
class ConfigurationProvider<T> implements Provider<T> {
private final Field field;
#Inject RootConfiguration configuration;
ConfigurationProvider(Field field) {
this.field = field;
}
#Override
public T get() {
return (T) field.get(configuration);
}
}
I am writing integration tests to test existing Routes. The recommended way of getting the response looks something like this (via Camel In Action section 6.4.1):
public class TestGetClaim extends CamelTestSupport {
#Produce(uri = "seda:getClaimListStart")
protected ProducerTemplate producer;
#Test
public void testNormalClient() {
NotifyBuilder notify = new NotifyBuilder(context).whenDone(1).create();
producer.sendBody(new ClientRequestBean("TESTCLIENT", "Y", "A"));
boolean matches = notify.matches(5, TimeUnit.SECONDS);
assertTrue(matches);
BrowsableEndpoint be = context.getEndpoint("seda:getClaimListResponse", BrowsableEndpoint.class);
List<Exchange> list = be.getExchanges();
assertEquals(1, list.size());
System.out.println("***RESPONSE is type "+list.get(0).getIn().getBody().getClass().getName());
}
}
The test runs but I get nothing back. The assertTrue(matches) fails after the 5 second timeout.
If I rewrite the test to look like this I get a response:
#Test
public void testNormalClient() {
producer.sendBody(new ClientRequestBean("TESTCLIENT", "Y", "A"));
Object resp = context.createConsumerTemplate().receiveBody("seda:getClaimListResponse");
System.out.println("***RESPONSE is type "+resp.getClass().getName());
}
The documentation is a little light around this so can anyone tell me what I am doing wrong with the first approach? Is there anything wrong with following the second approach instead?
Thanks.
UPDATE
I have broken this down and it looks like the problem is with the mix of seda as the start endpoint in combination with the use of a recipientList in the Route. I've also changed the construction of the NotifyBuilder (I had the wrong endpoint specified).
If I change the start endpoint to
direct instead of seda then the test will work; or
If I comment out the recipientList
then the test will work.
Here's a stripped down version of my Route that reproduces this issue:
public class TestRouteBuilder extends RouteBuilder {
#Override
public void configure() throws Exception {
// from("direct:start") //works
from("seda:start") //doesn't work
.recipientList(simple("exec:GetClaimList.bat?useStderrOnEmptyStdout=true&args=${body.client}"))
.to("seda:finish");
}
}
Note that if I change the source code of the NotifyTest from the "Camel In Action" source to have a route builder like this then it also fails.
Try to use "seda:getClaimListResponse" in the getEndpoint to be sure the endpoint uri is 100% correct
FWIW: It appears that notifyBuilder in conjunction with seda queues are not quite working: a test class to illustrate:
public class NotifyBuilderTest extends CamelTestSupport {
// Try these out!
// String inputURI = "seda:foo"; // Fails
// String inputURI = "direct:foo"; // Passes
#Test
public void testNotifyBuilder() {
NotifyBuilder b = new NotifyBuilder(context).from(inputURI)
.whenExactlyCompleted(1).create();
assertFalse( b.matches() );
template.sendBody(inputURI, "Test");
assertTrue( b.matches() );
b.reset();
assertFalse( b.matches() );
template.sendBody(inputURI, "Test2");
assertTrue( b.matches() );
}
#Override
protected RouteBuilder createRouteBuilder() throws Exception {
return new RouteBuilder() {
#Override
public void configure() throws Exception {
from(inputURI).to("mock:foo");
}
};
}
}