Spring demo microservice

Table of contents

1. Introduction
2. About the API
3. Language Version
   1. JDK 17
   2. Managing JDKs with SDKMAN!
   3. Available Toolchains
4. How to Build
5. Testing
   1. Why JUnit 5?
   2. Why AssertJ?
   3. Why Testcontainers?
   4. What is JaCoCo?
      1. JaCoCo logging
   5. Mutation Testing
   6. Future-proofing
6. Required services
   1. Docker Compose
   2. Keycloak
7. Running the microservice
   1. Running the application with Gradle
   2. Consuming the API
8. Contributing

Introduction

This project is about showcasing some of the best practices and technologies that can be used to ensure code quality while developing a microservice in Java with Spring Boot. Therefore, the focus isn't on creating a complex API but rather on how to achieve a high code quality standard, using a variety of mature and free libraries and technologies.

The project has been implemented with:

• Language: Java 17
• Framework: Spring Boot
• Database: MongoDB
• Build tool: Gradle
• Security: Spring Security + Keycloak

Relevant testing information and libraries will be described in the appropriate section of this document.

If you are not familiar with Gradle, the most relevant commands and tasks will be explained in the following sections. If you need further help, check its documentation or execute the help task:

./gradlew help

About the API

The API is rather simple: this microservice allows users to keep track of their "notes", and offers endpoints for the basic CRUD operations.

Endpoints are secured by OAuth2, so each request must include a valid access token to be processed. These tokens are issued by Keycloak, and the instructions on how to obtain them will be detailed in a following section of this document.

No frontend application was developed, but a Swagger UI is available in order to explore and consume the API.

Language Version

Gradle's toolchain concept is used in order to define which java version will be used to build the project. This is declared in the build.gradle.kts file:

java {
    toolchain {
        languageVersion.set(JavaLanguageVersion.of(17))
    }
}

JDK 17

Since the project is implemented with Java 17, a JDK for Java 17 must be available on your machine in order to build this project.

Don't worry! If a matching JDK is not found by Gradle on your machine, it will automatically download it before building the project. If you want Gradle to download a distribution JDK from a specific vendor, you can do so by setting the vendor property in the build.gradle.kts file:

java {
    toolchain {
        languageVersion.set(JavaLanguageVersion.of(17))
        vendor.set(JvmVendorSpec.GRAAL_VM)
    }
}

Managing JDKs with SDKMAN!

Most developers want to control which JDKs are installed on their machines. If you would like to manage JDKs by yourself, you can use tools such as SDKMAN! to simplify that task. This is not meant to compare SDKMAN! with any other tool, but rather explain how you can easily manage SDKs with a mature tool.

The installation is very straightforward, and you can easily install any JDK with a single command. For example, in order to install GraalVM 21.3.0 with Java 17 you can execute the following command:

sdk install java 21.3.0.r17-grl 

After installing it, SDKMAN! will ask if you want to set it as the default JDK on your machine. If you chose to not set it as the default JDK, you can later set it as the active JDK in your terminal window by running the use command:

sdk use java 21.3.0.r17-grl

More info on SDKMAN usage can be found here

Available Toolchains

In order to check which toolchains are available to Gradle, you can execute the javaToolchains task:

./gradlew javaToolchains

Here's an example of the output you should expect from the task when multiple toolchains are available:

> Task :javaToolchains

 + GraalVM Community JDK 11.0.13+7-jvmci-21.3-b05
     | Location:           /Users/username/.sdkman/candidates/java/21.3.0.r11-grl
     | Language Version:   11
     | Vendor:             GraalVM Community
     | Is JDK:             true
     | Detected by:        Current JVM

 + GraalVM Community JDK 17.0.1+12-jvmci-21.3-b05
     | Location:           /Users/username/.sdkman/candidates/java/21.3.0.r17-grl
     | Language Version:   17
     | Vendor:             GraalVM Community
     | Is JDK:             true
     | Detected by:        SDKMAN!

 + OpenJDK 18-ea+24-1608
     | Location:           /Users/username/.sdkman/candidates/java/18.ea.24-open
     | Language Version:   18
     | Vendor:             Oracle
     | Is JDK:             true
     | Detected by:        SDKMAN!

 + Oracle JDK 1.8.0_162-b12
     | Location:           /Library/Java/JavaVirtualMachines/jdk1.8.0_162.jdk/Contents/Home
     | Language Version:   8
     | Vendor:             Oracle
     | Is JDK:             true
     | Detected by:        macOS java_home

How to Build

Gradle was chosen as the build tool for this project and its wrapper has been included in the repository. There are at least two very good reasons to advocate for wrapper inclusion and usage:

• Project can be immediately built after cloning is finished, there's no need for the developer to install Gradle beforehand.
• The wrapper will always match the expected Gradle version that is meant to be used when building the project.

In order to build the project, execute the following command:

./gradlew build

Please note that the build task will also execute the tests. If you simply want to compile the source code and resources, you can do so by executing the assemble task:

./gradlew assemble

Testing

Testing is crucial to ensure code is reliable, maintainable, reusable, extensible. All of these factors influence the code's quality standard.

Many libraries can be used to improve the testing experience. In this project some great libraries and tools are showcased, such as:

• JUnit 5 to create and run our tests
• AssertJ to write amazing assertions in our test cases
• Testcontainers to reproduce a real environment for our tests
• JaCoCo to generate the coverage report of our project
• Pitest to execute mutation testing
• Codecov to keep track of our repository's coverage

Integration and Unit tests can be found in the src/test directory. Usually, they could've been split into different source sets, such as src/test for unit tests and src/it for integration tests, but that was not deemed necessary for this project's purpose, given its size, complexity and scope. However, this might change in the future.

In order to run the tests in this project, execute the test task:

./gradlew test

When execution ends, reports will be available at:

• build/reports/tests/test for JUnit reports
• build/reports/jacoco/test for JaCoCo reports

Why JUnit 5?

JUnit is the most used testing platform for Java, with version 5 being its most recent major version. It has many features and advantages over JUnit 4, some of which will be used in our tests, such as:

• @ParameterizedTest: this allows the execution of the same test case with any number of varying inputs (example: testing the same endpoint with different payloads)
• Assertions.assertAll(...): executes all assertions in a block without failing the test as soon as the first assertion fails, allowing for all failed assertions to be grouped and displayed together
• Assertions.assertThrows(...): which helps the assertion of an exception and its properties

Why AssertJ?

The purpose of this project is not to compare assertion libraries, such as AssertJ, Truth, Hamcrest and others.

AssertJ is an amazing library, and is already included when importing the Spring Boot's testing dependency, so that's why it was used in this project.

Writing meaningful and helpful assertions with ease is the goal, and AssertJ does a great job at it.

Why Testcontainers?

You want your API to be reliable, extensible, maintainable, reusable, and that assumption has many consequences.

One way to ensure that you don't break anything when developing new features or fixing bugs is to have a very thorough test suite, and having a continuous integration pipeline that automatically runs those tests and lets you know when things go wrong.

However, what if your tests do not reproduce the real environment? The execution of the test suite can be successful and once you deploy the new version to your real world production environment, things that you were lead to assume were fine still run the risk of breaking because the environment they were deployed to was nothing like what you tested against.

One of the ways to reduce this kind of problem is to use real instances of the services that the application relies on, with the same version as the ones in your production environment. For example, instead of testing an application against an embedded database, have it connect to a real database server.

This is where Testcontainers comes in handy, allowing you to spin up disposable Docker containers that will live only during the execution of your test suite.

In this case, the existing org.testcontainers:mongodb dependency was used to import its MongoDBContainer class which was extended in the ExtendedMongoDBContainer class in order to ensure that the appropriate URL connection value was made available to Spring Data once the container was ready to accept connections.

For the Keycloak authorization server, the com.github.dasniko:testcontainers-keycloak dependency was used. Similarly to the MongoDB case, the KeycloakContainer was extended by the ExtendedKeycloakContainer class, which is used to configure Keycloak's test realm as well as making its auth URL available to Spring Security.

IMPORTANT!

Using testcontainers means that most of the project's tests REQUIRE docker to be available in the host machine in order to run at all.

The Docker images being used in the tests are:

• mongo:5.0.4
• quay.io/keycloak/keycloak:15.0.2

If they are not found in the host machine, testcontainers will automatically download them. This can give the impression that the tasks are taking too long to start or finish, so if you wish to manually download the images before executing the tests, you can by executing the pull command:

docker pull mongo:5.0.4
docker pull quay.io/keycloak/keycloak:15.0.2

What is JaCoCo?

JaCoCo is an amazing tool that generates code coverage reports for java. In order to generate these reports, the JaCoCo Gradle plugin is used, and is declared in the build.gradle.kts plugins block.

The plugin provides two useful tasks:

• jacocoTestReport: Generates the report and depends on the output of tests. In this project, the gradlew test task is finalized by this task, so that the coverage report is always generated after test execution
• jacocoTestCoverageVerification: verifies if the coverage rules are met. If a project declares min 99% coverage, the task will fail if the coverage is below that threshold

In this project, the plugin is configured to generate both HTML and XML reports.

• HTML provides the best visual experience, as it allows a developer to navigate through the packages and classes, checking all the coverage details.
• The XML report is created so that is exported to Codecov when the GitHub Actions workflow is executed

These reports can be found in build/reports/jacoco, and they include information about:

• Class coverage
• Method coverage
• Branch coverage
• Line coverage
• Instruction coverage
• Complexity coverage

More information on what those counters mean can be found here.

JaCoCo logging

Another neat plugin used is jacocolog, which provides thejacocoLogTestCoverage task which prints all those metrics in the console in case you are only looking for the consolidated coverage numbers

Having near 100% code coverage does not mean that a software is bug-free or flawless. But it usually means that a team is invested into making it reliable, and that type of confidence in the code base provides a lot of comfort when planning and building new features or simply fixing bugs.

Mutation Testing

This is not a very widespread concept, but is a rather powerful one. Below is the explanation of why you should care about it, from the folks at Pitest:

Traditional test coverage (i.e line, statement, branch, etc.) measures only which code is executed by your tests. It does not check that your tests are actually able to detect faults in the executed code. It is therefore only able to identify code that is definitely not tested.

The most extreme examples of the problem are tests with no assertions. Fortunately these are uncommon in most code bases. Much more common is code that is only partially tested by its suite. A suite that only partially tests code can still execute all its branches (examples).

As it is actually able to detect whether each statement is meaningfully tested, mutation testing is the gold standard against which all other types of coverage are measured.

In order to run mutation testing, the Pitest Gradle plugin is used. It provides the pitest gradle task, which runs the mutated tests, and the generated report can be found inbuild/reports/pitest/index.html

IMPORTANT!

Given many mutations are generated for each test case, the execution time of the pitest task is much longer than the time from the test task. Since this may take several minutes, there's no GitHub action configured for it.

The current version of the Pitest Gradle plugin is not compatible with the Gradle Toolchain concept. Therefore, if the JVM used by Gradle to run does not match the language version declared by the toolchain, the execution of pitest task will fail. In order to avoid this, make sure you set the JDK used by Gradle to be compatible with the toolchain version. If you are using SDKMAN! this can be easily done by executing the sdk use java command before you invoke the pitest task.

Future-proofing

One way to future-proof your application is to test it against early access builds of future JDKs. This can save you a lot of time when migrating to newer versions of the JDK, as it will make it clear if your current code and dependencies are compatible with the next release.

Since this project is written in Java 17, the only early access builds it can be tested against are from JDK 18 (at the time of writing this). Therefore, a Gradle task called testsOn18 has been created and declared in the build.gradle.kts file. This task sets the Java Toolchain to use Java Language Version 18 and executes the project's tests.

To run the tests on JDK 18, execute the following command:

./gradlew testsOn18

Once the tests finish, you can access the reports at build/reports/tests/testsOn18

Please note that since an early access build of the JDK is required to test against a future version of the JDK, you must install it manually, as Gradle's toolchain isn't capable of downloading one. This might change in the future, as Gradle has an open issue for this feature.

If you are using SDKMAN! as suggested in the previous sections, you can easily download an early access build of the JDK. At the time of writing, the latest build was 18.ea.25 which can be installed by executing:

sdk install java 18.ea.25-open

Once installed, Gradle's Toolchain is capable of automatically locating the early access build of the JDK and use it when required.

IMPORTANT

The current version of JaCoCo is 0.8.7 and it isn't compatible with Java 18, therefore an exception will be thrown when running the tests on 18. However, this doesn't mean that the tests have failed or won't be executed, just that JaCoCo won't be able to generate its reports from the test output. This should be fixed when JaCoCo version 0.8.8 is released.

Required services

This microservice requires both MongoDB and Keycloak servers to be available in order to function. If you have both servers installed or available on a remote URL, you can use such instances, just don't forget to update corresponding properties on src/main/resources/application.yml (or another profile's file) with the appropriate connection URLs.

If you don't want to alter the properties file, you can also provide the URIs as JVM properties

-Dkeycloak.auth-server-url=http://some.host:8080/auth
-Dspring.data.mongodb.uri=mongodb://user:[email protected]:27017/spring-demo

or even as application arguments.

--keycloak.auth-server-url=http://some.host:8080/auth
--spring.data.mongodb.uri=mongodb://user:[email protected]:27017/spring-demo

Docker Compose

However, you can also easily spin up both MongoDB and Keycloak as docker containers. A file called docker-compose.yml has been added to the project, and you can use Docker Compose in order to take advantage of it.

To start the containers with Docker Compose, execute:

docker-compose up -d

To stop them, execute:

docker-compose down

Please note that the services described in the docker-compose.yml file do not have persistent storage configured, which will cause modifications to the databases to be lost once the containers are stopped. This is intentional behaviour.

If you'd like data to be persisted when the containers are stopped, you can use the docker-compose-persistent.yml file instead. To do so, execute the commands above with the -f flag:

docker-compose -f docker-compose-persistent.yml up -d
docker-compose -f docker-compose-persistent.yml down

The main difference in this approach is that PostgreSQL is used as the backing database for Keycloak, running on its own container, and that both MongoDB and PostgreSQL containers are mounting data volumes, which are managed by the Docker Engine, in the appropriate container directories

Keycloak

Keycloak uses the concept of realm to manage Users, Clients, Roles and other security details. There are many ways to configure a realm:

• You can do it manually
• You can use Keycloak's REST API (or one if its existing clients)
• You can import an existing realm configuration.

In the realm.json file, you can inspect the basic Keycloak realm configuration needed in order for the microservice to work as expected. This file is automatically imported by Keycloak when it starts if you use the project's provided docker compose files.

During tests, instead of importing the realm, it is created programmatically. This is done in the com.github.jorgebsa.spring.demo.util.ExtendedKeycloakContainer class. To achieve this, the Keycloak REST client from the org.keycloak:keycloak-admin-client dependency is used.

Running the microservice

Once your MongoDB and Keycloak instances are reachable, there are a few ways to launch the application:

• You can run it directly from your IDE, by invoking the main method of com.github.jorgebsa.spring.demo.Application
• You can run it using the Spring Boot Gradle plugin

Running the application with Gradle

Is as easy as invoking a simple task from the Spring Boot Gradle Plugin:

./gradlew bootRun

Consuming the API

After the application starts, you can send HTTP requests directly to it or access the Swagger UI available at http://localhost:8080/swagger-ui.html.

Remember that the endpoints are secured, therefore each request must include an access token in its Authorization header. To obtain an access token, you must request one from Keycloak with the appropriate payload, for example:

POST http://localhost:8081/auth/realms/tests/protocol/openid-connect/token
Accept: application/json
Cache-Control: no-cache
Content-Type: application/x-www-form-urlencoded

client_id=spring-demo&client_secret=the-secret&username=the-admin&password=123Admin&grant_type=password

You are looking for the access_token value from the response body, which will be in the following format:

{
  "access_token": "eyJhbGciOiJSUzI1NiIsInR5cCIgOiAiSldUIiwia2lkIiA6ICJ2WENxeHpWT25aNW92a2Y5N0x0UEdZX0JXUkNuSWVRbUYwNU5nUUVzb2hRIn0.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.CII62FsOdCfe9qIyf9BfzBJUbfWjODgP6-BuhqFjQAeqt12yIXGcGXcyy4ZD55uYaBywMovOofMlmhIT_IGx6bRAHLwjYq6sH-TW_5wC_42rrZ4---UWkQn1zn8atdYJnDNHagEPRqMnzvE0R38Nk2otFiVpiYGodP_K-LuoORPKtqpb4LdbiR9kM-uhIuYnqxv_4cJuhH_3-wBFBuhI2uuAi12aHwrgfgWKDkAgc0I4uwNaQ8jIKG7MkeJhPOu81p1kYM99CK2mDYFhjggroR_wdwlvIG33Z_V2tVvHxuNdUCeg7JXTwGMT6t2xRGApW0I0IqeEQ_AdBitxBEeTaA",
  "expires_in": 35999,
  "refresh_expires_in": 1800,
  "refresh_token": "eyJhbGciOiJIUzI1NiIsInR5cCIgOiAiSldUIiwia2lkIiA6ICJmMGVjMzU3OC1kZTA1LTRkZmItOWFjMC1kZWEwMTYzY2EzYjgifQ.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.RCiVcFlkK3hkHtFhVJHKze1oBf_upmB6vxVuZyCWYxk",
  "token_type": "Bearer",
  "not-before-policy": 0,
  "session_state": "9fec5839-23be-4cdb-8b9d-1489896336bb",
  "scope": "profile email"
}

Once you have the access token, you can use it to Authorize Swagger UI's requests by clicking on the Authorize button and pasting the value into the field. If you opt to work with traditional HTTP requests, make sure to set the value of the Authorization header to be of type Bearer, here's an example with the access token from above:

GET http://localhost:8080/notes
Accept: application/json
Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCIgOiAiSldUIiwia2lkIiA6ICJmMGVjMzU3OC1kZTA1LTRkZmItOWFjMC1kZWEwMTYzY2EzYjgifQ.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.RCiVcFlkK3hkHtFhVJHKze1oBf_upmB6vxVuZyCWYxk

Contributing

This project is meant to be a way of showcasing some of the best practices and technologies I've come across in the past few years. I'm always interested in learning more, discussing different points of views and approaches. If you can contribute to this repository in any meaningful way, be it by showcasing something new or improving what has been implemented already, you are more than welcome to open a pull request and start a discussion!