I'm using DynamoDB local for unit testing. It's not bad, but has some drawbacks. Specifically:
You have to somehow start the server before your tests run
The server isn't started and stopped before each test so tests become inter-dependent unless you add code to delete all tables, etc. after each test
All developers need to have it installed
What I want to do is something like put the DynamoDB local jar, and the other jars upon which it depends, in my test/resources directory (I'm writing in Java). Then before each test I'd start it up, running with -inMemory, and after the test I'd stop it. That way anyone pulling down the git repo gets a copy of everything they need to run the tests and each test is independent of the others.
I have found a way to make this work, but it's ugly, so I'm looking for alternatives. The solution I have is to put a .zip file of the DynamoDB local stuff in test/resources, then in the #Before method, I'd extract it to some temporary directory and start a new java process to execute it. That works, but it's ugly and has some drawbacks:
Everyone needs the java executable on their $PATH
I have to unpack a zip to the local disk. Using local disk is often dicey for testing, especially with continuous builds and such.
I have to spawn a process and wait for it to start for each unit test, and then kill that process after each test. Besides being slow, the potential for left-over processes seems ugly.
It seems like there should be an easier way. DynamoDB Local is, after all, just Java code. Can't I somehow ask the JVM to fork itself and look inside the resources to build a classpath? Or, even better, can't I just call the main method of DynamoDB Local from some other thread so this all happens in a single process? Any ideas?
PS: I am aware of Alternator, but it appears to have other drawbacks so I'm inclined to stick with Amazon's supported solution if I can make it work.
In order to use DynamoDBLocal you need to follow these steps.
Get Direct DynamoDBLocal Dependency
Get Native SQLite4Java dependencies
Set sqlite4java.library.path to show native libraries
1. Get Direct DynamoDBLocal Dependency
This one is the easy one. You need this repository as explained here.
<!--Dependency:-->
<dependencies>
<dependency>
<groupId>com.amazonaws</groupId>
<artifactId>DynamoDBLocal</artifactId>
<version>1.11.0.1</version>
<scope></scope>
</dependency>
</dependencies>
<!--Custom repository:-->
<repositories>
<repository>
<id>dynamodb-local</id>
<name>DynamoDB Local Release Repository</name>
<url>https://s3-us-west-2.amazonaws.com/dynamodb-local/release</url>
</repository>
</repositories>
2. Get Native SQLite4Java dependencies
If you do not add these dependencies, your tests will fail with 500 internal error.
First, add these dependencies:
<dependency>
<groupId>com.almworks.sqlite4java</groupId>
<artifactId>sqlite4java</artifactId>
<version>1.0.392</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>com.almworks.sqlite4java</groupId>
<artifactId>sqlite4java-win32-x86</artifactId>
<version>1.0.392</version>
<type>dll</type>
<scope>test</scope>
</dependency>
<dependency>
<groupId>com.almworks.sqlite4java</groupId>
<artifactId>sqlite4java-win32-x64</artifactId>
<version>1.0.392</version>
<type>dll</type>
<scope>test</scope>
</dependency>
<dependency>
<groupId>com.almworks.sqlite4java</groupId>
<artifactId>libsqlite4java-osx</artifactId>
<version>1.0.392</version>
<type>dylib</type>
<scope>test</scope>
</dependency>
<dependency>
<groupId>com.almworks.sqlite4java</groupId>
<artifactId>libsqlite4java-linux-i386</artifactId>
<version>1.0.392</version>
<type>so</type>
<scope>test</scope>
</dependency>
<dependency>
<groupId>com.almworks.sqlite4java</groupId>
<artifactId>libsqlite4java-linux-amd64</artifactId>
<version>1.0.392</version>
<type>so</type>
<scope>test</scope>
</dependency>
Then, add this plugin to get native dependencies to specific folder:
<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-dependency-plugin</artifactId>
<version>2.10</version>
<executions>
<execution>
<id>copy</id>
<phase>test-compile</phase>
<goals>
<goal>copy-dependencies</goal>
</goals>
<configuration>
<includeScope>test</includeScope>
<includeTypes>so,dll,dylib</includeTypes>
<outputDirectory>${project.basedir}/native-libs</outputDirectory>
</configuration>
</execution>
</executions>
</plugin>
</plugins>
</build>
3. Set sqlite4java.library.path to show native libraries
As last step, you need to set sqlite4java.library.path system property to native-libs directory. It is OK to do that just before creating your local server.
System.setProperty("sqlite4java.library.path", "native-libs");
After these steps you can use DynamoDBLocal as you want. Here is a Junit rule that creates local server for that.
import com.amazonaws.auth.BasicAWSCredentials;
import com.amazonaws.services.dynamodbv2.AmazonDynamoDB;
import com.amazonaws.services.dynamodbv2.AmazonDynamoDBClient;
import com.amazonaws.services.dynamodbv2.local.main.ServerRunner;
import com.amazonaws.services.dynamodbv2.local.server.DynamoDBProxyServer;
import org.junit.rules.ExternalResource;
import java.io.IOException;
import java.net.ServerSocket;
/**
* Creates a local DynamoDB instance for testing.
*/
public class LocalDynamoDBCreationRule extends ExternalResource {
private DynamoDBProxyServer server;
private AmazonDynamoDB amazonDynamoDB;
public LocalDynamoDBCreationRule() {
// This one should be copied during test-compile time. If project's basedir does not contains a folder
// named 'native-libs' please try '$ mvn clean install' from command line first
System.setProperty("sqlite4java.library.path", "native-libs");
}
#Override
protected void before() throws Throwable {
try {
final String port = getAvailablePort();
this.server = ServerRunner.createServerFromCommandLineArgs(new String[]{"-inMemory", "-port", port});
server.start();
amazonDynamoDB = new AmazonDynamoDBClient(new BasicAWSCredentials("access", "secret"));
amazonDynamoDB.setEndpoint("http://localhost:" + port);
} catch (Exception e) {
throw new RuntimeException(e);
}
}
#Override
protected void after() {
if (server == null) {
return;
}
try {
server.stop();
} catch (Exception e) {
throw new RuntimeException(e);
}
}
public AmazonDynamoDB getAmazonDynamoDB() {
return amazonDynamoDB;
}
private String getAvailablePort() {
try (final ServerSocket serverSocket = new ServerSocket(0)) {
return String.valueOf(serverSocket.getLocalPort());
} catch (IOException e) {
throw new RuntimeException("Available port was not found", e);
}
}
}
You can use this rule like this
#RunWith(JUnit4.class)
public class UserDAOImplTest {
#ClassRule
public static final LocalDynamoDBCreationRule dynamoDB = new LocalDynamoDBCreationRule();
}
In August 2018 Amazon announced new Docker image with Amazon DynamoDB Local onboard. It does not require downloading and running any JARs as well as adding using third-party OS-specific binaries (I'm talking about sqlite4java).
It is as simple as starting a Docker container before the tests:
docker run -p 8000:8000 amazon/dynamodb-local
You can do that manually for local development, as described above, or use it in your CI pipeline. Many CI services provide an ability to start additional containers during the pipeline that can provide dependencies for your tests. Here is an example for Gitlab CI/CD:
test:
stage: test
image: openjdk:8-alpine
services:
- name: amazon/dynamodb-local
alias: dynamodb-local
script:
- DYNAMODB_LOCAL_URL=http://dynamodb-local:8000 ./gradlew clean test
Or Bitbucket Pipelines:
definitions:
services:
dynamodb-local:
image: amazon/dynamodb-local
…
step:
name: test
image:
name: openjdk:8-alpine
services:
- dynamodb-local
script:
- DYNAMODB_LOCAL_URL=http://localhost:8000 ./gradlew clean test
And so on. The idea is to move all the configuration you can see in other answers out of your build tool and provide the dependency externally. Think of it as of dependency injection / IoC but for the whole service, not just a single bean.
After you've started the container you can create a client pointing to it:
private AmazonDynamoDB createAmazonDynamoDB(final DynamoDBLocal configuration) {
return AmazonDynamoDBClientBuilder
.standard()
.withEndpointConfiguration(
new AwsClientBuilder.EndpointConfiguration(
"http://localhost:8000",
Regions.US_EAST_1.getName()
)
)
.withCredentials(
new AWSStaticCredentialsProvider(
// DynamoDB Local works with any non-null credentials
new BasicAWSCredentials("", "")
)
)
.build();
}
Now to the original questions:
You have to somehow start the server before your tests run
You can just start it manually, or prepare a developsers' script for it. IDEs usually provide a way to run arbitrary commands before executing a task, so you can make IDE to start the container for you. I think that running something locally should not be a top priority in this case, but instead you should focus on configuring CI and let the developers start the container as it's comfortable to them.
The server isn't started and stopped before each test so tests become inter-dependent unless you add code to delete all tables, etc. after each test
That's trueee, but… You should not start and stop such heavyweight things
and recreate tables before / after each test. DB tests are almost always inter-dependent and that's ok for them. Just use unique values for each test case (e.g. set item's hash key to ticket id / specific test case id you're working on). As for the seed data, I'd recommend moving it from the build tool and test code as well. Either make your own image with all the data you need or use AWS CLI to create tables and insert data. Follow the single responsibility principle and dependency injection principles: your test code must not do anything but tests. All the environment (tables and data in this case should be provided for them). Creating a table in a test is wrong, because in a real life that table already exist (unless you're testing a method that actually creates a table, of course).
All developers need to have it installed
Docker should be a must for every developer in 2018, so that's not a problem.
And if you're using JUnit 5, it can be a good idea to use a DynamoDB Local extension that will inject the client in your tests (yes, I'm doing a self-promotion):
Add a dependency on me.madhead.aws-junit5:dynamo-v1
pom.xml:
<dependency>
<groupId>me.madhead.aws-junit5</groupId>
<artifactId>dynamo-v1</artifactId>
<version>6.0.1</version>
<scope>test</scope>
</dependency>
build.gradle
dependencies {
testImplementation("me.madhead.aws-junit5:dynamo-v1:6.0.1")
}
Use the extension in your tests:
#ExtendWith(DynamoDBLocalExtension.class)
class MultipleInjectionsTest {
#DynamoDBLocal(
url = "http://dynamodb-local-1:8000"
)
private AmazonDynamoDB first;
#DynamoDBLocal(
urlEnvironmentVariable = "DYNAMODB_LOCAL_URL"
)
private AmazonDynamoDB second;
#Test
void test() {
first.listTables();
second.listTables();
}
}
This is a restating of bhdrkn's answer for Gradle users (his is based on Maven). It's still the same three steps:
Get Direct DynamoDBLocal Dependency
Get Native SQLite4Java dependencies
Set sqlite4java.library.path to show native libraries
1. Get Direct DynamoDBLocal Dependency
Add to the dependencies section of your build.gradle file...
dependencies {
testCompile "com.amazonaws:DynamoDBLocal:1.+"
}
2. Get Native SQLite4Java dependencies
The sqlite4java libraries will already be downloaded as a dependency of DynamoDBLocal, but the library files need to be copied to the right place. Add to your build.gradle file...
task copyNativeDeps(type: Copy) {
from(configurations.compile + configurations.testCompile) {
include '*.dll'
include '*.dylib'
include '*.so'
}
into 'build/libs'
}
3. Set sqlite4java.library.path to show native libraries
We need to tell Gradle to run copyNativeDeps for testing and tell sqlite4java where to find the files. Add to your build.gradle file...
test {
dependsOn copyNativeDeps
systemProperty "java.library.path", 'build/libs'
}
You can use DynamoDB Local as a Maven test dependency in your test code, as is shown in this announcement. You can run over HTTP:
import com.amazonaws.services.dynamodbv2.local.main.ServerRunner;
import com.amazonaws.services.dynamodbv2.local.server.DynamoDBProxyServer;
final String[] localArgs = { "-inMemory" };
DynamoDBProxyServer server = ServerRunner.createServerFromCommandLineArgs(localArgs);
server.start();
AmazonDynamoDB dynamodb = new AmazonDynamoDBClient();
dynamodb.setEndpoint("http://localhost:8000");
dynamodb.listTables();
server.stop();
You can also run in embedded mode:
import com.amazonaws.services.dynamodbv2.local.embedded.DynamoDBEmbedded;
AmazonDynamoDB dynamodb = DynamoDBEmbedded.create();
dynamodb.listTables();
I have wrapped the answers above into two JUnit rules that does not require changes to the build script as the rules handles the native library stuff. I did this as I found that Idea did not like the Gradle/Maven solutions as it just went off and did its own thing anyhoos.
This means the steps are:
Get the AssortmentOfJUnitRules version 1.5.32 or above dependency
Get the Direct DynamoDBLocal dependency
Add the LocalDynamoDbRule or HttpDynamoDbRule to your JUnit test.
Maven:
<!--Dependency:-->
<dependencies>
<dependency>
<groupId>com.amazonaws</groupId>
<artifactId>DynamoDBLocal</artifactId>
<version>1.11.0.1</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>com.github.mlk</groupId>
<artifactId>assortmentofjunitrules</artifactId>
<version>1.5.36</version>
<scope>test</scope>
</dependency>
</dependencies>
<!--Custom repository:-->
<repositories>
<repository>
<id>dynamodb-local</id>
<name>DynamoDB Local Release Repository</name>
<url>https://s3-us-west-2.amazonaws.com/dynamodb-local/release</url>
</repository>
</repositories>
Gradle:
repositories {
mavenCentral()
maven {
url = "https://s3-us-west-2.amazonaws.com/dynamodb-local/release"
}
}
dependencies {
testCompile "com.github.mlk:assortmentofjunitrules:1.5.36"
testCompile "com.amazonaws:DynamoDBLocal:1.+"
}
Code:
public class LocalDynamoDbRuleTest {
#Rule
public LocalDynamoDbRule ddb = new LocalDynamoDbRule();
#Test
public void test() {
doDynamoStuff(ddb.getClient());
}
}
Try out tempest-testing! It ships a JUnit4 Rule and a JUnit5 Extension. It also supports both AWS SDK v1 and SDK v2.
Tempest provides a library for testing DynamoDB clients
using DynamoDBLocal
. It comes with two implementations:
JVM: This is the preferred option, running a DynamoDBProxyServer backed by sqlite4java,
which is available on most platforms.
Docker: This runs dynamodb-local in a Docker
container.
Feature matrix:
Feature
tempest-testing-jvm
tempest-testing-docker
Start up time
~1s
~10s
Memory usage
Less
More
Dependency
sqlite4java native library
Docker
To use tempest-testing, first add this library as a test dependency:
For AWS SDK 1.x:
dependencies {
testImplementation "app.cash.tempest:tempest-testing-jvm:1.5.2"
testImplementation "app.cash.tempest:tempest-testing-junit5:1.5.2"
}
// Or
dependencies {
testImplementation "app.cash.tempest:tempest-testing-docker:1.5.2"
testImplementation "app.cash.tempest:tempest-testing-junit5:1.5.2"
}
For AWS SDK 2.x:
dependencies {
testImplementation "app.cash.tempest:tempest2-testing-jvm:1.5.2"
testImplementation "app.cash.tempest:tempest2-testing-junit5:1.5.2"
}
// Or
dependencies {
testImplementation "app.cash.tempest:tempest2-testing-docker:1.5.2"
testImplementation "app.cash.tempest:tempest2-testing-junit5:1.5.2"
}
Then in tests annotated with #org.junit.jupiter.api.Test, you may add TestDynamoDb as a test
extension. This extension spins up a
DynamoDB server. It shares the server across tests and keeps it running until the process exits. It
also manages test tables for you, recreating them before each test.
class MyTest {
#RegisterExtension
TestDynamoDb db = new TestDynamoDb.Builder(JvmDynamoDbServer.Factory.INSTANCE) // or DockerDynamoDbServer
// `MusicItem` is annotated with `#DynamoDBTable`. Tempest recreates this table before each test.
.addTable(TestTable.create(MusicItem.TABLE_NAME, MusicItem.class))
.build();
#Test
public void test() {
PutItemRequest request = // ...;
// Talk to the local DynamoDB.
db.dynamoDb().putItem(request);
}
}
It seems like there should be an easier way. DynamoDB Local is, after all, just Java code. Can't I somehow ask the JVM to fork itself and look inside the resources to build a classpath?
You can do something along these lines, but much simpler: programmatically search the classpath for the location of the native libraries, then set the sqlite4java.library.path property before starting DynamoDB. This is the approach implemented in tempest-testing, as well as in this answer (code here) which is why they just work as pure library/classpath dependency and nothing more.
In my case needed access to DynamoDB outside of a JUnit extension, but I still wanted something self-contained in library code, so I extracted the approach it takes:
import com.amazonaws.services.dynamodbv2.local.embedded.DynamoDBEmbedded;
import com.amazonaws.services.dynamodbv2.local.shared.access.AmazonDynamoDBLocal;
import com.google.common.collect.MoreCollectors;
import java.io.File;
import java.util.Arrays;
import java.util.stream.Stream;
import org.junit.jupiter.api.condition.OS;
...
public AmazonDynamoDBLocal embeddedDynamoDb() {
final OS os = Stream.of(OS.values()).filter(OS::isCurrentOs)
.collect(MoreCollectors.onlyElement());
final String prefix;
switch (os) {
case LINUX:
prefix = "libsqlite4java-linux-amd64-";
break;
case MAC:
prefix = "libsqlite4java-osx-";
break;
case WINDOWS:
prefix = "sqlite4java-win32-x64-";
break;
default:
throw new UnsupportedOperationException(os.toString());
}
System.setProperty("sqlite4java.library.path",
Arrays.asList(System.getProperty("java.class.path").split(File.pathSeparator))
.stream()
.map(File::new)
.filter(file -> file.getName().startsWith(prefix))
.collect(MoreCollectors.onlyElement())
.getParent());
return DynamoDBEmbedded.create();
}
Not had a chance to test on a lot of platforms, and the error handling could likely be improved.
It's a pity AWS haven't taken the time to make the library more friendly, as this could easily be done in the library code itself.
For unit testing at work I use Mockito, then just mock the AmazonDynamoDBClient. then mock out the returns using when. like the following:
when(mockAmazonDynamoDBClient.getItem(isA(GetItemRequest.class))).thenAnswer(new Answer<GetItemResult>() {
#Override
public GetItemResult answer(InvocationOnMock invocation) throws Throwable {
GetItemResult result = new GetItemResult();
result.setItem( testResultItem );
return result;
}
});
not sure if that is what your looking for but that's how we do it.
Shortest solution with fix for sqlite4java.SQLiteException UnsatisfiedLinkError if it is a java/kotlin project built with gradle (a changed $PATH is not needed).
repositories {
// ... other dependencies
maven { url 'https://s3-us-west-2.amazonaws.com/dynamodb-local/release' }
}
dependencies {
testImplementation("com.amazonaws:DynamoDBLocal:1.13.6")
}
import org.gradle.internal.os.OperatingSystem
test {
doFirst {
// Fix for: UnsatisfiedLinkError -> provide a valid native lib path
String nativePrefix = OperatingSystem.current().nativePrefix
File nativeLib = sourceSets.test.runtimeClasspath.files.find {it.name.startsWith("libsqlite4java") && it.name.contains(nativePrefix) } as File
systemProperty "sqlite4java.library.path", nativeLib.parent
}
}
Straightforward usage in test classes (src/test):
private lateinit var db: AmazonDynamoDBLocal
#BeforeAll
fun runDb() { db = DynamoDBEmbedded.create() }
#AfterAll
fun shutdownDb() { db.shutdown() }
There are couple of node.js wrappers for DynamoDB Local. These allows to easily execute unit tests combining with task runners like gulp or grunt. Try dynamodb-localhost,
dynamodb-local
I have found that the amazon repo as no index file, so does not seem to function in a way that allows you to bring it in like this:
maven {
url = "https://s3-us-west-2.amazonaws.com/dynamodb-local/release"
}
The only way I could get the dependencies to load is by downloading DynamoDbLocal as a jar and bringing it into my build script like this:
dependencies {
...
runtime files('libs/DynamoDBLocal.jar')
...
}
Of course this means that all the SQLite and Jetty dependencies need to be brought in by hand - I'm still trying to get this right. If anyone knows of a reliable repo for DynamoDbLocal, I would really love to know.
You could also use this lightweight test container 'Dynalite'
https://www.testcontainers.org/modules/databases/dynalite/
From testcontainers:
Dynalite is a clone of DynamoDB, enabling local testing. It's light
and quick to run.
The DynamoDB Gradle dependency already includes the SQLite libraries. You can pretty easily instruct the Java runtime to use it in your Gradle build script. Here's my build.gradle.kts as an example:
import org.apache.tools.ant.taskdefs.condition.Os
plugins {
application
}
repositories {
mavenCentral()
maven {
url = uri("https://s3-us-west-2.amazonaws.com/dynamodb-local/release")
}
}
dependencies {
implementation("com.amazonaws:DynamoDBLocal:[1.12,2.0)")
}
fun getSqlitePath(): String? {
val dirName = when {
Os.isFamily(Os.FAMILY_MAC) -> "libsqlite4java-osx"
Os.isFamily(Os.FAMILY_UNIX) -> "libsqlite4java-linux-amd64"
Os.isFamily(Os.FAMILY_WINDOWS) -> "sqlite4java-win32-x64"
else -> throw kotlin.Exception("DynamoDB emulator cannot run on this platform")
}
return project.configurations.runtimeClasspath.get().find { it.name.contains(dirName) }?.parent
}
application {
mainClass.set("com.amazonaws.services.dynamodbv2.local.main.ServerRunner")
applicationDefaultJvmArgs = listOf("-Djava.library.path=${getSqlitePath()}")
}
tasks.named<JavaExec>("run") {
args("-inMemory")
}
Related
When creating the basic Application using gradle init the following block is defined in the resulting build.gradle file
testing {
suites {
// Configure the built-in test suite
test {
// Use JUnit Jupiter test framework
useJUnitJupiter('5.8.2')
}
}
}
I have the following gradle/libs.versions.toml file
[versions]
guava = "31.0.1-jre"
junit = "5.8.2"
[libraries]
guava = {module = "com.google.guava:guava", version.ref= "guava"}
Now I want to use the junit version defined in the toml file in the build.gradle file, but can't seem to succeed. The Gradle documentation is not really helpful either.
What I tried sofar is:
useJUnitJupiter(versions.junit)
useJUnitJupiter(libs.versions.junit)
useJUnitJupiter(libs.junit)
Thanks in advance for any help
I'm writing a custom gradle plugin in which I want to have a bunch of common for several of my projects tasks and a sort of a 'main' task to control which of these tasks to turn on.
Regular tasks in the plugin are e.g.:
CopyDockerResourcesTask
CopyContainerFilesTask
PerformAnalysisTask
and the 'main' task is:
BaseProjectTask
so then in the project in build.gradle I'd like to be able to do this:
BaseProjectTask {
copyDockerResources = true
copyContainerFiles = true
performAnalysis = true
}
I want the default behaviour of the plugin to be to not to do anything, only add certain tasks if they are turned on in BaseProjectTask.
I wanted to achieve this with adding task dependency in #TaskAction method of BaseProjectTask:
class BaseProjectTask extends DefaultTask {
private final BaseProjectExtension extension
private final Project project
#Optional
#Input
Boolean copyContainerFiles = false
...
#Inject
BaseProjectTask(Project project, BaseProjectExtension extension) {
this.project = project
this.extension = extension
}
#TaskAction
def execute() {
if (copyContainerFiles) {
project.tasks.assemble.dependsOn(project.tasks.copyContainerFiles)
}
...
}
}
Creating task dependency, this line:
project.tasks.assemble.dependsOn(project.tasks.copyContainerFiles)
doesn't work.
Edit:
My current findings are that defining task dependency in #TaskAction is too late as this is execution phase. I could do it in the constructor (this way it works) but its too early as property copyContainerFiles isn't set yet.
Does anyone know a way of adding code in the task class that would be fired in the configuration phase? I think this is what I'm missing.
You need to configure task dependencies during the build configuration phase, as you surmised.
It's not possible to do it in the #TaskAction method. It's fundamental to the way Gradle works that it needs to know how tasks depend on each other before it starts executing the build. That allows Gradle to do some useful things, such as only executing the tasks that are not up to date, or working out what tasks will execute without actually executing them.
In general, tasks should not be aware of one another1.
When you are trying to do this in a plugin using values in a project extension, you must wait until after the project has evaluated so that the build script code executes first. You can do this with project.afterEvaluate()2.
So you can do the following (using Kotlin DSL3):
project.afterEvaluate {
tasks.register("baseTask") {
if (baseProjectExtension.copyDockerResources)
dependsOn(tasks.getByName("copyDockerResources"))
if (baseProjectExtension.copyContainerFiles)
dependsOn(tasks.getByName("copyContainerFiles"))
if (baseProjectExtension.performAnalysis)
dependsOn(tasks.getByName("performAnalysis"))
}
}
1See How to declare dependencies of a Gradle custom task?
2See https://docs.gradle.org/current/userguide/build_lifecycle.html#sec:project_evaluation
3What I am familiar with. Hopefully not too much trouble to convert to Groovy.
I use Elasticsearch 7.11.2 with the following dependencies to write unit tests
implementation 'org.codelibs.elasticsearch.module:lang-painless:7.10.2'
implementation 'org.codelibs.elasticsearch.module:analysis-common:7.10.2'
Spinning up a node looks like
public EmbeddedElastic() throws NodeValidationException, IOException {
tempDir = Files.createTempDirectory("elastic_search_temp").toAbsolutePath().toString();
int port = getAvailableBasePort();
Map<String, String> settings = new HashMap<>();
settings.put("path.home", tempDir);
settings.put("http.port", Integer.toString(port));
settings.put("transport.tcp.port", Integer.toString(port + 1));
settings.put("transport.type", "netty4");
settings.put("http.cors.enabled", Boolean.FALSE.toString());
new PluginNode(settings).start();
client = new RestHighLevelClient(RestClient.builder(new HttpHost("localhost", port, "http")));
System.out.println("Client: " + client.getClass().getName());
Runtime.getRuntime().addShutdownHook(new Thread() {
#Override
public void run() {
// file deletion
}
});
}
private static class PluginNode extends Node {
public PluginNode(Map<String, String> preparedSettings) {
super(InternalSettingsPreparer.prepareEnvironment(Settings.EMPTY, preparedSettings,
null, () -> "node-test"),
Lists.newArrayList(Netty4Plugin.class, ParentJoinPlugin.class, CommonAnalysisPlugin.class,
PainlessPlugin.class),
false);
System.out.println("Started local elastic with PainlessPlugin loaded.");
}
}
Now I am upgrading to ES 7.16.2 and the same code doesn't work well. I presume its because of the lang-painless and analysis-common libraries. Their latest version is only 7.10.
The exception I receive after upgrading the elasticsearch client to 7.16 and spring-data-elasticsearch to 4.3.0
Failed to instantiate [org.springframework.data.elasticsearch.core.ElasticsearchOperations]: Factory method 'elasticsearchTemplate' threw exception; nested exception is java.lang.BootstrapMethodError: java.lang.NoClassDefFoundError: org/elasticsearch/common/xcontent/ToXContentObject
Before I dive deep into fixing it, I am wondering if anyone has actually made unit tests work with 7.16 ES version? I have seen that Elasticsearch Test containers is the recommended way to go
EDIT:
Thanks to Augusto's tip on checking classes in Maven Central, I found out that lang-painless depends on certain libraries from elasticsearch client which are not available in the higher versions of ES client. So its not possible to write unit tests with the same libs.
Elasticsearch moved this class in 7.16 from the org.elasticsearch.common.xcontent package to the org.elasticsearch.xcontent package, a breaking change between 7.15 and 7.16.
Spring Data Elasticsearch 4.3 is built against 7.15.x and so won't work with Elasticsearch 7.16 libraries.
You either need to use Elasticsearch 7.15 or wait for Spring Data Elasticsearch 4.4 (the first milestone for that should be published next week) to be released which is built against 7.16.
And, yes, you should better use testcontainers.
You are probably missing the dependency with this class. It could be that before it came transitevily with the ES client.
Try adding org.elasticsearch:elasticsearch-x-content:7.15.2 to your build tool. A word of caution, you might be missing other dependencies too.
Just as a tip, if you ever get a NoClassDefFoundError, one of the easiest ways to find where that class is defined is to search in maven central. Search for fc:<fully_qualified_class_name>, and it will show all the jars where the class is present.
My plugin mojo test class leverages maven-plugin-test-harness to build the complete maven environment with all pom config, plexus container context and repo access.
The following should all actually work:
the test will refer to a test pom.xml in the plugin project's test resources directory
the mojo will be populated with defaults from the mojo annotations
all specified configuration in the test pom is accepted
the maven project object is initialised
all dependencies from the repo are available
the tests should pass in IntelliJ IDE as well as Maven on CI server
Because of the lack of concrete working examples I've been trying many different fixes using what I've collected from SO, and other blogs or articles online.
I am currently struggling to get the maven to resolve the artifacts. While I've got the dependency list from the maven project object, the artifact list is empty.
This is what I've built up by dissecting AbstractMojoTestCase.
I can't use MojoRule because JUnit5 doesn't use #Rules anymore.
Plus, some of the maven API calls are deprecated, but I couldn't find a new implementation. I think it won't come until mvn4. See the quote below.
#BeforeEach
protected void setUp() throws Exception {
super.setUp();
cleanUp();
ClassLoader classLoader = getClass().getClassLoader();
URL url = classLoader.getResource(TEST_POM);
if (url == null) {
throw new MojoExecutionException(String.format(
"Cannot locate %s", TEST_POM));
}
File pom = new File(url.getFile());
//noinspection deprecation - wait on maven-plugin-testing-harness update
MavenSettingsBuilder mavenSettingsBuilder = (MavenSettingsBuilder)
getContainer().lookup(MavenSettingsBuilder.ROLE);
Settings settings = mavenSettingsBuilder.buildSettings();
MavenExecutionRequest request = new DefaultMavenExecutionRequest();
request.setPom(pom);
request.setLocalRepositoryPath(settings.getLocalRepository());
MavenExecutionRequestPopulator populator =
getContainer().lookup(MavenExecutionRequestPopulator.class);
populator.populateDefaults(request);
DefaultMaven maven = (DefaultMaven) getContainer().lookup(Maven.class);
DefaultRepositorySystemSession repoSession =
(DefaultRepositorySystemSession)
maven.newRepositorySession(request);
LocalRepository localRepository = new LocalRepository(
request.getLocalRepository().getBasedir());
SimpleLocalRepositoryManagerFactory factory =
new SimpleLocalRepositoryManagerFactory();
LocalRepositoryManager localRepositoryManager =
factory.newInstance(repoSession, localRepository);
repoSession.setLocalRepositoryManager(localRepositoryManager);
ProjectBuildingRequest buildingRequest =
request.getProjectBuildingRequest()
.setRepositorySession(repoSession)
.setResolveDependencies(true);
ProjectBuilder projectBuilder = lookup(ProjectBuilder.class);
MavenProject project =
projectBuilder.build(pom, buildingRequest).getProject();
//noinspection deprecation - wait on maven-plugin-testing-harness update
MavenSession session = new MavenSession(getContainer(), repoSession,
request, new DefaultMavenExecutionResult());
session.setCurrentProject(project);
session.setProjects(Collections.singletonList(project));
request.setSystemProperties(System.getProperties());
testMojo = (GenerateConfig) lookupConfiguredMojo(session,
newMojoExecution("configure"));
copyTestProjectResourcesToTarget(getContainer(), project, session);
}
[UPDATE 2017-07-27]: actually this now solves most of my problems.
Only a couple of minor issues now:
the code to grab the settings.xml is marked as #Deprecated so I assume there is a better way of doing it (using the MavenSettingsBuilder.buildSettings())
probably quite a lot of the setup code is duplicating process that occurs anyway when running in native maven, but is required to run with JUnit in IntelliJ.
[UPDATE 2017-08-01]: test now needs to access some property files which would be on the classpath in a live environment in the target/classes dir.
Logically they are test resources in my maven-plugin project, so I have included them under the same directory as the test pom.xml in src/test/resources/my-test-project dir.
That didn't work, so I also tried src/test/resources/my-test-project/src/main/resources but that's also not good.
I am having a hard time establishing what exactly is on the plugin's classpath during the test, or working out how to cause it to be constructed correctly.
[UPDATE 2017-08-02]: although I've answered my own question (as opposed to extending this question), this whole thing isn't finished yet so I'm not marking this as answered quite yet.
And just for the record, these are the dependencies required:
<dependency>
<groupId>org.junit.jupiter</groupId>
<artifactId>junit-jupiter-api</artifactId>
<version>5.0.0-M4</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>org.junit.vintage</groupId>
<artifactId>junit-vintage-engine</artifactId>
<version>4.12.0-M4</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>org.apache.maven</groupId>
<artifactId>maven-plugin-api</artifactId>
<version>3.5.0</version>
</dependency>
<dependency>
<groupId>org.apache.maven.plugin-tools</groupId>
<artifactId>maven-plugin-annotations</artifactId>
<version>3.5</version>
<scope>provided</scope>
</dependency>
<dependency>
<groupId>org.apache.maven.plugin-testing</groupId>
<artifactId>maven-plugin-testing-harness</artifactId>
<version>3.3.0</version>
<scope>test</scope>
<exclusions>
<exclusion>
<groupId>org.codehaus.plexus</groupId>
<artifactId>plexus-container-default</artifactId>
</exclusion>
</exclusions>
</dependency>
<dependency>
<groupId>org.apache.maven</groupId>
<artifactId>maven-core</artifactId>
<version>3.5.0</version>
</dependency>
<dependency>
<groupId>org.apache.maven</groupId>
<artifactId>maven-compat</artifactId>
<version>3.5.0</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>org.twdata.maven</groupId>
<artifactId>mojo-executor</artifactId>
<version>2.3.0</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-resources-plugin</artifactId>
<version>3.0.2</version>
<scope>test</scope>
</dependency>
[UPDATE 2017-08-09]:
I have to add some more functionality and discovered that the test was fine if the dependency it wanted to unpack was in the local repo already, but if not, it won't fetch it.
I now need to determine how to instruct maven to fetch the dependency from the remote repo.
I tried launching the dependency plugin and invoking resolve in the test setup, but it dies badly, I think there must be a simpler way.
No need for so much complexity.
I can't use MojoRule because JUnit5 doesn't use #Rules anymore.
Since version 3.2 you don't need to use #MojoRule, 7 years ago. Just follow the three steps below:
Your test class should extend AbstractMojoTestCase
Before your tests, call super.setUp()
Perform a lookup for your mojo:
MyMojo myMojo = (MyMojo) super.lookupMojo("myGoal", "src/test/resources/its/my-test-mojo.pom.xml");
With that, you can work with Junit 5, Mockito, etc, with no overhead.
Some comments in the maven source code for MavenProject said
With changes during 3.2.2 release MavenProject is closer to being immutable after construction with the removal of all components from this class, and the upfront construction taken care of entirely by the #{ProjectBuilder}. There is still the issue of having to run the lifecycle in order to find all the compile source roots and resource directories but I hope to take care of this during the Maven 4.0 release (jvz).
I figure this whole maven plugin integration test thing is not going to work until then... and so looking around, I found a great blog entry on invoking plugins. So I invoked the maven-resources-plugin directly to get it to copy across what it was meant to. That's what the copyTestProjectResourcesToTarget() call does.
private void copyTestProjectResourcesToTarget(PlexusContainer container,
MavenProject project,
MavenSession session)
throws ComponentLookupException, MojoExecutionException {
logger.fine("generateConfig dependencies: ");
project.getDependencies().forEach(d -> logger.fine(d.getArtifactId()));
Optional<Dependency> resourcesPluginDepOpt =
project.getDependencies().stream()
.filter(d -> Objects.equals(d.getArtifactId(),
MAVEN_RESOURCES_ARTIFACT_ID))
.findFirst();
// don't want to define the version here so we read it from what we have
if (!resourcesPluginDepOpt.isPresent()) {
throw new MojoExecutionException("Require " +
MAVEN_RESOURCES_ARTIFACT_ID);
}
Plugin resourcePlugin = MojoExecutor.plugin(
MojoExecutor.groupId(MAVEN_RESOURCES_GROUP_ID),
MojoExecutor.artifactId(MAVEN_RESOURCES_ARTIFACT_ID),
MojoExecutor.version(resourcesPluginDepOpt.get().getVersion()));
MojoExecutor.executeMojo(resourcePlugin,
MojoExecutor.goal("resources"),
MojoExecutor.configuration(),
MojoExecutor.executionEnvironment(
project, session,
container.lookup(BuildPluginManager.class)));
}
I found the solution to fetching dependencies from the remote repository.
Working with the maven internals like this and judging from the deprecated classes and the amount of duplicated functionality, it gives me the strong impression that maven v4 will make this redundant.
One glitch with this setup routine is that it creates a local repository directory tree in the maven project directory. This is obviously not desirable but will need some more tweaking to solve.
#BeforeEach
public void setUp() throws Exception {
super.setUp();
ClassLoader classLoader = getClass().getClassLoader();
URL url = classLoader.getResource(TEST_POM);
if (url == null) {
throw new MojoExecutionException(String.format(
"Cannot locate %s", TEST_POM));
}
File pom = new File(url.getFile());
Settings settings = getMavenSettings();
if (settings.getLocalRepository() == null) {
settings.setLocalRepository(
org.apache.maven.repository.RepositorySystem
.defaultUserLocalRepository.getAbsolutePath());
}
MavenExecutionRequest request = new DefaultMavenExecutionRequest();
request.setPom(pom);
ArtifactRepository artifactRepository =
new org.apache.maven.artifact.repository.
DefaultArtifactRepository(
"id", settings.getLocalRepository(),
new DefaultRepositoryLayout());
request.setLocalRepository(artifactRepository);
MavenExecutionRequestPopulator populator =
getContainer().lookup(MavenExecutionRequestPopulator.class);
populator.populateFromSettings(request, settings);
DefaultMaven maven = (DefaultMaven)
getContainer().lookup(Maven.class);
DefaultRepositorySystemSession repositorySystemSession =
(DefaultRepositorySystemSession)
maven.newRepositorySession(request);
SimpleLocalRepositoryManagerFactory factory =
new SimpleLocalRepositoryManagerFactory();
LocalRepositoryManager localRepositoryManager =
factory.newInstance(repositorySystemSession,
new LocalRepository(settings.getLocalRepository()));
repositorySystemSession.setLocalRepositoryManager(
localRepositoryManager);
ProjectBuildingRequest buildingRequest =
request.getProjectBuildingRequest()
.setRepositorySession(repositorySystemSession)
.setResolveDependencies(true);
ProjectBuilder projectBuilder = lookup(ProjectBuilder.class);
ProjectBuildingResult projectBuildingResult =
projectBuilder.build(pom, buildingRequest);
MavenProject project = projectBuildingResult.getProject();
MavenSession session = new MavenSession(getContainer(),
repositorySystemSession, request,
new DefaultMavenExecutionResult());
session.setCurrentProject(project);
session.setProjects(Collections.singletonList(project));
request.setSystemProperties(System.getProperties());
testMojo = (GenerateConfig) lookupConfiguredMojo(session,
newMojoExecution("configure"));
testMojo.getLog().debug(String.format("localRepo = %s",
request.getLocalRepository()));
copyTestProjectResourcesToTarget(getContainer(), project, session);
resolveConfigurationFromRepo(repositorySystemSession, project);
}
private Settings getMavenSettings()
throws ComponentLookupException,
IOException,
XmlPullParserException {
org.apache.maven.settings.MavenSettingsBuilder mavenSettingsBuilder
= (org.apache.maven.settings.MavenSettingsBuilder)
getContainer().lookup(
org.apache.maven.settings.MavenSettingsBuilder.ROLE);
return mavenSettingsBuilder.buildSettings();
}
/**
* This is ugly but there seems to be no other way to accomplish it. The
* artifact that the mojo finds on its own will not resolve to a jar file
* on its own in the test harness. So we use aether to resolve it, by
* cloning the maven default artifact into an aether artifact and feeding
* an artifact request to the repo system obtained by the aether service
* locator.
*/
private void resolveConfigurationFromRepo(
DefaultRepositorySystemSession repositorySystemSession,
MavenProject project)
throws ArtifactResolutionException, MojoExecutionException {
org.apache.maven.artifact.Artifact defaultArtifact =
testMojo.getConfigArtifact();
Artifact artifact = new DefaultArtifact(
defaultArtifact.getGroupId(),
defaultArtifact.getArtifactId(),
null,
defaultArtifact.getType(),
defaultArtifact.getVersion());
List<RemoteRepository> remoteArtifactRepositories =
project.getRemoteProjectRepositories();
DefaultServiceLocator locator =
MavenRepositorySystemUtils.newServiceLocator();
locator.addService(RepositoryConnectorFactory.class,
BasicRepositoryConnectorFactory.class);
locator.addService(TransporterFactory.class,
FileTransporterFactory.class);
locator.addService(TransporterFactory.class,
HttpTransporterFactory.class);
RepositorySystem repositorySystem = locator.getService(
RepositorySystem.class);
ArtifactRequest artifactRequest = new ArtifactRequest();
artifactRequest.setArtifact(artifact);
artifactRequest.setRepositories(remoteArtifactRepositories);
ArtifactResult result = repositorySystem.resolveArtifact(
repositorySystemSession, artifactRequest);
defaultArtifact.setFile(result.getArtifact().getFile());
testMojo.getLog().debug( "Resolved artifact " + artifact + " to " +
result.getArtifact().getFile() + " from "
+ result.getRepository() );
}
/**
* Need manual copy of resources because only parts of the maven lifecycle
* happen automatically with this test harness.
*/
private void copyTestProjectResourcesToTarget(PlexusContainer container,
MavenProject project,
MavenSession session)
throws ComponentLookupException, MojoExecutionException {
Optional<Dependency> resourcesPluginDepOpt =
project.getDependencies().stream()
.filter(d -> Objects.equals(d.getArtifactId(),
MAVEN_RESOURCES_ARTIFACT_ID))
.findFirst();
// don't want to define the version here so we read it from what we have
if (!resourcesPluginDepOpt.isPresent()) {
throw new MojoExecutionException("Require " +
MAVEN_RESOURCES_ARTIFACT_ID);
}
Plugin resourcePlugin = MojoExecutor.plugin(
MojoExecutor.groupId(MAVEN_PLUGINS_GROUP_ID),
MojoExecutor.artifactId(MAVEN_RESOURCES_ARTIFACT_ID),
MojoExecutor.version(resourcesPluginDepOpt.get().getVersion()));
MojoExecutor.executeMojo(resourcePlugin,
MojoExecutor.goal("resources"),
MojoExecutor.configuration(),
MojoExecutor.executionEnvironment(
project, session,
container.lookup(BuildPluginManager.class)));
}
and here are the packages used, quite important to use the classes from the right package but easily confused:
import org.apache.maven.DefaultMaven;
import org.apache.maven.Maven;
import org.apache.maven.artifact.repository.ArtifactRepository;
import org.apache.maven.artifact.repository.layout.DefaultRepositoryLayout;
import org.apache.maven.execution.DefaultMavenExecutionRequest;
import org.apache.maven.execution.DefaultMavenExecutionResult;
import org.apache.maven.execution.MavenExecutionRequest;
import org.apache.maven.execution.MavenExecutionRequestPopulator;
import org.apache.maven.execution.MavenSession;
import org.apache.maven.model.Dependency;
import org.apache.maven.model.Plugin;
import org.apache.maven.plugin.BuildPluginManager;
import org.apache.maven.plugin.MojoExecutionException;
import org.apache.maven.plugin.MojoFailureException;
import org.apache.maven.plugin.testing.AbstractMojoTestCase;
import org.apache.maven.project.MavenProject;
import org.apache.maven.project.ProjectBuilder;
import org.apache.maven.project.ProjectBuildingRequest;
import org.apache.maven.project.ProjectBuildingResult;
import org.apache.maven.repository.internal.MavenRepositorySystemUtils;
import org.apache.maven.settings.Settings;
import org.codehaus.plexus.PlexusContainer;
import org.codehaus.plexus.component.repository.exception.ComponentLookupException;
import org.codehaus.plexus.util.xml.pull.XmlPullParserException;
import org.eclipse.aether.DefaultRepositorySystemSession;
import org.eclipse.aether.RepositorySystem;
import org.eclipse.aether.artifact.Artifact;
import org.eclipse.aether.artifact.DefaultArtifact;
import org.eclipse.aether.connector.basic.BasicRepositoryConnectorFactory;
import org.eclipse.aether.impl.DefaultServiceLocator;
import org.eclipse.aether.internal.impl.SimpleLocalRepositoryManagerFactory;
import org.eclipse.aether.repository.LocalRepository;
import org.eclipse.aether.repository.LocalRepositoryManager;
import org.eclipse.aether.repository.RemoteRepository;
import org.eclipse.aether.resolution.ArtifactRequest;
import org.eclipse.aether.resolution.ArtifactResolutionException;
import org.eclipse.aether.resolution.ArtifactResult;
import org.eclipse.aether.spi.connector.RepositoryConnectorFactory;
import org.eclipse.aether.spi.connector.transport.TransporterFactory;
import org.eclipse.aether.transport.file.FileTransporterFactory;
import org.eclipse.aether.transport.http.HttpTransporterFactory;
I want to write unit tests (junit4) for my maven-plugin. All examples i found use "AbstractMojoTestCase" (junit3 :-(). To get rid of this i got answer here. But the problem is how Mojos get instantiated:
MyMojo myMojo = (MyMojo) lookupMojo( "touch", pom );
That means i need a pom for every test case - the pom is the tests input data. But is there a way to mock (i would use Mockito) the project model some how?
Could lookupMojo(String groupId, String artifactId, String version, String goal, PlexusConfiguration pluginConfiguration) be a good starting point? In this case i would mock "PlexusConfiguration", but what methods?
Some maven-plugin testing doku uses classes like "MavenProjectStub". But i can't get a consistent picture of how a mojo is created and to what intefaces it talks on creation.
A perfect solution would be if i could just
#inject
MyMojo testObject;
and just mock all the stuff it need to get it working (primary i need #Parameters)
Based on my experience writing Maven plugin, there are two levels of testing a plugin: via unit test (using mocks) and via integration tests (using the maven-invoker-plugin).
For the integration tests, the maven archetype for new maven plugins already provide a good example out of the box, just execute the following and have a look at it:
mvn archetype:generate \
-DgroupId=sample.plugin \
-DartifactId=hello-maven-plugin \
-DarchetypeGroupId=org.apache.maven.archetypes \
-DarchetypeArtifactId=maven-archetype-plugin
By default you will get integration tests in a profile to start with. An example maven project will also be available (under src\it\simple-it\pom.xml) which can execute your plugin goals. What I suggest is also to enforce the result of your integration test via additional constraints in that pom.xml. For instance: you can add the Maven Enforcer Plugin rule to check against created files, if that makes sense for your plugin.
To answer more specifically to your question on how to write unit tests for custom maven plugins, this is the approach I'm using:
JUnit + Mockito.
Test case running using #RunWith(MockitoJUnitRunner.class)
Mock Maven specific classes (MavenProject, Log, Build, DependencyNode, etc.) using #Mock annotations
Initiate and link your mock objects in a #Before method (typically setUp() method)
Test your plugin :)
As an example, you might have the following mocked objects as class variable of your unit test:
#Mock
private MavenProject project;
#Mock
private Log log;
#Mock
Build build;
Then, in your #Before method you need to add a big of glue code as following:
Mockito.when(this.project.getBuild()).thenReturn(this.build);
For instance, I use to write some custom Enforcer Plugin rules, hence I need
#Mock
private EnforcerRuleHelper helper;
And in the #Before method:
Mockito.when(this.helper.evaluate("${project}")).thenReturn(this.project);
Mockito.when(this.helper.getLog()).thenReturn(this.log);
Mockito.when(this.project.getBuild()).thenReturn(this.build);
Mockito.when(this.helper.getComponent(DependencyGraphBuilder.class)).thenReturn(this.graphBuilder);
Mockito.when(this.graphBuilder.buildDependencyGraph(this.project, null)).thenReturn(this.node);
As such, it will be easy to use these mock objects into your tests. For instance, a must have first dummy test is to test it against an empty build as following (below testing a custom Enforcer rule):
#Test
public void testEmptyBuild() throws Exception {
try {
this.rule.execute(this.helper);
} catch (EnforcerRuleException e) {
Assert.fail("Rule should not fail");
}
}
If you need to test against dependencies of your build, for instance, you might end up writing utility methods as following:
private static DependencyNode generateNode(String groupId, String artifactId, String version) {
DependencyNode node = Mockito.mock(DependencyNode.class);
Artifact artifact = Mockito.mock(Artifact.class);
Mockito.when(node.getArtifact()).thenReturn(artifact);
// mock artifact
Mockito.when(artifact.getGroupId()).thenReturn(groupId);
Mockito.when(artifact.getArtifactId()).thenReturn(artifactId);
Mockito.when(artifact.getVersion()).thenReturn(version);
return node;
}
In order to easily create dependencies into the dependency graph of your build, as following:
List<DependencyNode> nodes = new ArrayList<DependencyNode>();
nodes.add(generateNode("junit", "junit", "4.12"));
Mockito.when(node.getChildren()).thenReturn(nodes);
NOTE: you can improve the utility method if you need further details (like scope or classifier for a dependency).
If you also need to mock configuration of a plugin, because you need to scan existing plugins and their configuration, for instance, you can do it as following:
List<Plugin> plugins = new ArrayList<Plugin>();
Plugin p = new Plugin(); // no need to mock it
p.setArtifactId("maven-surefire-plugin");
Xpp3Dom conf = new Xpp3Dom("configuration");
Xpp3Dom skip = new Xpp3Dom("skip");
skip.setValue("true");
conf.addChild(skip);
p.setConfiguration(conf);
plugins.add(p);
Mockito.when(this.build.getPlugins()).thenReturn(plugins);
I will obviously not cover all the possible cases, but I am sure you got an understanding about approach and usage. Hope it helps.