1. Introduction

ArchUnit is a free, simple and extensible library for checking the architecture of your Java code. That is, ArchUnit can check dependencies between packages and classes, layers and slices, check for cyclic dependencies and more. It does so by analyzing given Java bytecode, importing all classes into a Java code structure. ArchUnit’s main focus is to automatically test architecture and coding rules, using any plain Java unit testing framework.

1.1. Module Overview

ArchUnit consists of the following production modules: archunit, archunit-junit4 as well as archunit-junit5-api, archunit-junit5-engine and archunit-junit5-engine-api. Also relevant for end users is the archunit-example module.

1.1.1. Module archunit

This module contains the actual ArchUnit core infrastructure required to write architecture tests: The ClassFileImporter, the domain objects, as well as the rule syntax infrastructure.

1.1.2. Module archunit-junit4

This module contains the infrastructure to integrate with JUnit 4, in particular the ArchUnitRunner to cache imported classes.

1.1.3. Modules archunit-junit5-*

These modules contain the infrastructure to integrate with JUnit 5 and contain the respective infrastructure to cache imported classes between test runs. archunit-junit5-api contains the user API to write tests with ArchUnit’s JUnit 5 support, archunit-junit5-engine contains the runtime engine to run those tests. archunit-junit5-engine-api contains API code for tools that want more detailed control over running ArchUnit JUnit 5 tests, in particular a FieldSelector which can be used to instruct the ArchUnitTestEngine to run a specific rule field (compare JUnit 4 & 5 Support).

1.1.4. Module archunit-example

This module contains example architecture rules and sample code that violates these rules. Look here to get inspiration on how to set up rules for your project, or at ArchUnit-Examples for the last released version.

2. Installation

To use ArchUnit, it is sufficient to include the respective JAR files in the classpath. Most commonly, this is done by adding the dependency to your dependency management tool, which is illustrated for Maven and Gradle below. Alternatively you can obtain the necessary JAR files directly from Maven Central.

2.1. JUnit 4

To use ArchUnit in combination with JUnit 4, include the following dependency from Maven Central:

pom.xml
<dependency>
    <groupId>com.tngtech.archunit</groupId>
    <artifactId>archunit-junit4</artifactId>
    <version>1.2.1</version>
    <scope>test</scope>
</dependency>
build.gradle
dependencies {
    testImplementation 'com.tngtech.archunit:archunit-junit4:1.2.1'
}

2.2. JUnit 5

ArchUnit’s JUnit 5 artifacts follow the pattern of JUnit Jupiter. There is one artifact containing the API, i.e. the compile time dependencies to write tests. Then there is another artifact containing the actual TestEngine used at runtime. Just like JUnit Jupiter ArchUnit offers one convenience artifact transitively including both API and engine with the correct scope, which in turn can be added as a test compile dependency. Thus to include ArchUnit’s JUnit 5 support, simply add the following dependency from Maven Central:

pom.xml
<dependency>
    <groupId>com.tngtech.archunit</groupId>
    <artifactId>archunit-junit5</artifactId>
    <version>1.2.1</version>
    <scope>test</scope>
</dependency>
build.gradle
dependencies {
    testImplementation 'com.tngtech.archunit:archunit-junit5:1.2.1'
}

2.3. Other Test Frameworks

ArchUnit works with any test framework that executes Java code. To use ArchUnit in such a context, include the core ArchUnit dependency from Maven Central:

pom.xml
<dependency>
    <groupId>com.tngtech.archunit</groupId>
    <artifactId>archunit</artifactId>
    <version>1.2.1</version>
    <scope>test</scope>
</dependency>
build.gradle
dependencies {
   testImplementation 'com.tngtech.archunit:archunit:1.2.1'
}

2.4. Maven Plugin

There exists a Maven plugin by Société Générale to run ArchUnit rules straight from Maven. For more information visit their GitHub repo: https://github.com/societe-generale/arch-unit-maven-plugin

3. Getting Started

ArchUnit tests are written the same way as any Java unit test and can be written with any Java unit testing framework. To really understand the ideas behind ArchUnit, one should consult Ideas and Concepts. The following will outline a "technical" getting started.

3.1. Importing Classes

At its core ArchUnit provides infrastructure to import Java bytecode into Java code structures. This can be done using the ClassFileImporter

JavaClasses classes = new ClassFileImporter().importPackages("com.mycompany.myapp");

The ClassFileImporter offers many ways to import classes. Some ways depend on the current project’s classpath, like importPackages(..). However there are other ways that do not, for example:

JavaClasses classes = new ClassFileImporter().importPath("/some/path");

The returned object of type JavaClasses represents a collection of elements of type JavaClass, where JavaClass in turn represents a single imported class file. You can in fact access most properties of the imported class via the public API:

JavaClass clazz = classes.get(Object.class);
System.out.print(clazz.getSimpleName()); // returns 'Object'

3.2. Asserting (Architectural) Constraints

To express architectural rules, like 'Services should only be accessed by Controllers', ArchUnit offers an abstract DSL-like fluent API, which can in turn be evaluated against imported classes. To specify a rule, use the class ArchRuleDefinition as entry point:

import static com.tngtech.archunit.lang.syntax.ArchRuleDefinition.classes;

// ...

ArchRule myRule = classes()
    .that().resideInAPackage("..service..")
    .should().onlyBeAccessed().byAnyPackage("..controller..", "..service..");

The two dots represent any number of packages (compare AspectJ Pointcuts). The returned object of type ArchRule can now be evaluated against a set of imported classes:

myRule.check(importedClasses);

Thus the complete example could look like

@Test
public void Services_should_only_be_accessed_by_Controllers() {
    JavaClasses importedClasses = new ClassFileImporter().importPackages("com.mycompany.myapp");

    ArchRule myRule = classes()
        .that().resideInAPackage("..service..")
        .should().onlyBeAccessed().byAnyPackage("..controller..", "..service..");

    myRule.check(importedClasses);
}

3.3. Using JUnit 4 or JUnit 5

While ArchUnit can be used with any unit testing framework, it provides extended support for writing tests with JUnit 4 and JUnit 5. The main advantage is automatic caching of imported classes between tests (of the same imported classes), as well as reduction of boilerplate code.

To use the JUnit support, declare ArchUnit’s ArchUnitRunner (only JUnit 4), declare the classes to import via @AnalyzeClasses and add the respective rules as fields:

@RunWith(ArchUnitRunner.class) // Remove this line for JUnit 5!!
@AnalyzeClasses(packages = "com.mycompany.myapp")
public class MyArchitectureTest {

    @ArchTest
    public static final ArchRule myRule = classes()
        .that().resideInAPackage("..service..")
        .should().onlyBeAccessed().byAnyPackage("..controller..", "..service..");

}

The JUnit test support will automatically import (or reuse) the specified classes and evaluate any rule annotated with @ArchTest against those classes.

For further information on how to use the JUnit support refer to JUnit Support.

3.4. Using JUnit support with Kotlin

Using the JUnit support with Kotlin is quite similar to Java:

@RunWith(ArchUnitRunner::class) // Remove this line for JUnit 5!!
@AnalyzeClasses(packagesOf = [MyArchitectureTest::class])
class MyArchitectureTest {
    @ArchTest
    val rule_as_field = ArchRuleDefinition.noClasses().should()...

    @ArchTest
    fun rule_as_method(importedClasses: JavaClasses) {
        val rule = ArchRuleDefinition.noClasses().should()...
        rule.check(importedClasses)
    }
}

4. What to Check

The following section illustrates some typical checks you could do with ArchUnit.

4.1. Package Dependency Checks

package deps no access
noClasses().that().resideInAPackage("..source..")
    .should().dependOnClassesThat().resideInAPackage("..foo..")
package deps only access
classes().that().resideInAPackage("..foo..")
    .should().onlyHaveDependentClassesThat().resideInAnyPackage("..source.one..", "..foo..")

4.2. Class Dependency Checks

class naming deps
classes().that().haveNameMatching(".*Bar")
    .should().onlyHaveDependentClassesThat().haveSimpleName("Bar")

4.3. Class and Package Containment Checks

class package contain
classes().that().haveSimpleNameStartingWith("Foo")
    .should().resideInAPackage("com.foo")

4.4. Inheritance Checks

inheritance naming check
classes().that().implement(Connection.class)
    .should().haveSimpleNameEndingWith("Connection")
inheritance access check
classes().that().areAssignableTo(EntityManager.class)
    .should().onlyHaveDependentClassesThat().resideInAnyPackage("..persistence..")

4.5. Annotation Checks

inheritance annotation check
classes().that().areAssignableTo(EntityManager.class)
    .should().onlyHaveDependentClassesThat().areAnnotatedWith(Transactional.class)

4.6. Layer Checks

layer check
layeredArchitecture()
    .consideringAllDependencies()
    .layer("Controller").definedBy("..controller..")
    .layer("Service").definedBy("..service..")
    .layer("Persistence").definedBy("..persistence..")

    .whereLayer("Controller").mayNotBeAccessedByAnyLayer()
    .whereLayer("Service").mayOnlyBeAccessedByLayers("Controller")
    .whereLayer("Persistence").mayOnlyBeAccessedByLayers("Service")

4.7. Cycle Checks

cycle check
slices().matching("com.myapp.(*)..").should().beFreeOfCycles()

5. Ideas and Concepts

ArchUnit is divided into different layers, where the most important ones are the "Core" layer, the "Lang" layer and the "Library" layer. In short the Core layer deals with the basic infrastructure, i.e. how to import byte code into Java objects. The Lang layer contains the rule syntax to specify architecture rules in a succinct way. The Library layer contains more complex predefined rules, like a layered architecture with several layers. The following section will explain these layers in more detail.

5.1. Core

Much of ArchUnit’s core API resembles the Java Reflection API. There are classes like JavaMethod, JavaField, and more, and the public API consists of methods like getName(), getMethods(), getRawType() or getRawParameterTypes(). Additionally ArchUnit extends this API for concepts needed to talk about dependencies between code, like JavaMethodCall, JavaConstructorCall or JavaFieldAccess. For example, it is possible to programmatically iterate over javaClass.getAccessesFromSelf() and react to the imported accesses between this Java class and other Java classes.

To import compiled Java class files, ArchUnit provides the ClassFileImporter, which can for example be used to import packages from the classpath:

JavaClasses classes = new ClassFileImporter().importPackages("com.mycompany.myapp");

For more information refer to The Core API.

5.2. Lang

The Core API is quite powerful and offers a lot of information about the static structure of a Java program. However, tests directly using the Core API lack expressiveness, in particular with respect to architectural rules.

For this reason ArchUnit provides the Lang API, which offers a powerful syntax to express rules in an abstract way. Most parts of the Lang API are composed as fluent APIs, i.e. an IDE can provide valuable suggestions on the possibilities the syntax offers.

An example for a specified architecture rule would be:

ArchRule rule =
    classes().that().resideInAPackage("..service..")
        .should().onlyBeAccessed().byAnyPackage("..controller..", "..service..");

Once a rule is composed, imported Java classes can be checked against it:

JavaClasses importedClasses = new ClassFileImporter().importPackage("com.myapp");
ArchRule rule = // define the rule
rule.check(importedClasses);

The syntax ArchUnit provides is fully extensible and can thus be adjusted to almost any specific need. For further information, please refer to The Lang API.

5.3. Library

The Library API offers predefined complex rules for typical architectural goals. For example a succinct definition of a layered architecture via package definitions. Or rules to slice the code base in a certain way, for example in different areas of the domain, and enforce these slices to be acyclic or independent of each other. More detailed information is provided in The Library API.

6. The Core API

The Core API is itself divided into the domain objects and the actual import.

6.1. Import

As mentioned in Ideas and Concepts the backbone of the infrastructure is the ClassFileImporter, which provides various ways to import Java classes. One way is to import packages from the classpath, or the complete classpath via

JavaClasses classes = new ClassFileImporter().importClasspath();

However, the import process is completely independent of the classpath, so it would be well possible to import any path from the file system:

JavaClasses classes = new ClassFileImporter().importPath("/some/path/to/classes");

The ClassFileImporter offers several other methods to import classes, for example locations can be specified as URLs or as JAR files.

Furthermore specific locations can be filtered out, if they are contained in the source of classes, but should not be imported. A typical use case would be to ignore test classes, when the classpath is imported. This can be achieved by specifying ImportOptions:

ImportOption ignoreTests = new ImportOption() {
    @Override
    public boolean includes(Location location) {
        return !location.contains("/test/"); // ignore any URI to sources that contains '/test/'
    }
};

JavaClasses classes = new ClassFileImporter().withImportOption(ignoreTests).importClasspath();

A Location is principally an URI, i.e. ArchUnit considers sources as File or JAR URIs

For the two common cases to skip importing JAR files and to skip importing test files (for typical setups, like a Maven or Gradle build), there already exist predefined ImportOptions:

new ClassFileImporter()
    .withImportOption(ImportOption.Predefined.DO_NOT_INCLUDE_JARS)
    .withImportOption(ImportOption.Predefined.DO_NOT_INCLUDE_TESTS)
    .importClasspath();

6.1.1. Dealing with Missing Classes

While importing the requested classes (e.g. target/classes or target/test-classes) it can happen that a class within the scope of the import has a reference to a class outside of the scope of the import. This will naturally happen, if the classes of the JDK are not imported, since then for example any dependency on Object.class will be unresolved within the import.

At this point ArchUnit needs to decide how to treat these classes that are missing from the import. By default, ArchUnit searches within the classpath for missing classes and if found imports them. This obviously has the advantage that information about those classes (which interfaces they implement, how they are annotated) is present during rule evaluation.

On the downside this additional lookup from the classpath will cost some performance and in some cases might not make sense (e.g. if information about classes not present in the original import is known to be unnecessary for evaluating rules). Thus ArchUnit can be configured to create stubs instead, i.e. a JavaClass that has all the known information, like the fully qualified name or the method called. However, this stub might naturally lack some information, like superclasses, annotations or other details that cannot be determined without importing the bytecode of this class. This behavior will also happen, if ArchUnit fails to determine the location of a missing class from the classpath.

To find out, how to configure the default behavior, refer to Configuring the Resolution Behavior.

6.2. Domain

The domain objects represent Java code, thus the naming should be pretty straight forward. Most commonly, the ClassFileImporter imports instances of type JavaClass. A rough overview looks like this:

domain overview

Most objects resemble the Java Reflection API, including inheritance relations. Thus a JavaClass has JavaMembers, which can in turn be either JavaField, JavaMethod, JavaConstructor (or JavaStaticInitializer). While not present within the reflection API, it makes sense to introduce an expression for anything that can access other code, which ArchUnit calls 'code unit', and is in fact either a method, a constructor (including the class initializer) or a static initializer of a class (e.g. a static { …​ } block, a static field assignment, etc.).

Furthermore one of the most interesting features of ArchUnit that exceeds the Java Reflection API, is the concept of accesses to another class. On the lowest level accesses can only take place from a code unit (as mentioned, any block of executable code) to either a field (JavaFieldAccess), a method (JavaMethodCall) or constructor (JavaConstructorCall).

ArchUnit imports the whole graph of classes and their relationship to each other. While checking the accesses from a class is pretty isolated (the bytecode offers all this information), checking accesses to a class requires the whole graph to be built first. To distinguish which sort of access is referred to, methods will always clearly state fromSelf and toSelf. For example, every JavaField allows to call JavaField#getAccessesToSelf() to retrieve all code units within the graph that access this specific field. The resolution process through inheritance is not completely straight forward. Consider for example

resolution example

The bytecode will record a field access from ClassAccessing.accessField() to ClassBeingAccessed.accessedField. However, there is no such field, since the field is actually declared in the superclass. This is the reason why a JavaFieldAccess has no JavaField as its target, but a FieldAccessTarget. In other words, ArchUnit models the situation, as it is found within the bytecode, and an access target is not an actual member within another class. If a member is queried for accessesToSelf() though, ArchUnit will resolve the necessary targets and determine, which member is represented by which target. The situation looks roughly like

resolution overview

Two things might seem strange at the first look.

First, why can a target resolve to zero matching members? The reason is that the set of classes that was imported does not need to have all classes involved within this resolution process. Consider the above example, if SuperclassBeingAccessed would not be imported, ArchUnit would have no way of knowing where the actual targeted field resides. Thus in this case the resolution would return zero elements.

Second, why can there be more than one resolved methods for method calls? The reason for this is that a call target might indeed match several methods in those cases, for example:

diamond example

While this situation will always be resolved in a specified way for a real program, ArchUnit cannot do the same. Instead, the resolution will report all candidates that match a specific access target, so in the above example, the call target C.targetMethod() would in fact resolve to two JavaMethods, namely A.targetMethod() and B.targetMethod(). Likewise a check of either A.targetMethod.getCallsToSelf() or B.targetMethod.getCallsToSelf() would return the same call from D.callTargetMethod() to C.targetMethod().

6.2.1. Domain Objects, Reflection and the Classpath

ArchUnit tries to offer a lot of information from the bytecode. For example, a JavaClass provides details like if it is an enum or an interface, modifiers like public or abstract, but also the source, where this class was imported from (namely the URI mentioned in the first section). However, if information is missing, and the classpath is correct, ArchUnit offers some convenience to rely on the reflection API for extended details. For this reason, most Java* objects offer a method reflect(), which will in fact try to resolve the respective object from the Reflection API. For example:

JavaClasses classes = new ClassFileImporter().importClasspath();

// ArchUnit's java.lang.String
JavaClass javaClass = classes.get(String.class);
// Reflection API's java.lang.String
Class<?> stringClass = javaClass.reflect();

// ArchUnit's public int java.lang.String.length()
JavaMethod javaMethod = javaClass.getMethod("length");
// Reflection API's public int java.lang.String.length()
Method lengthMethod = javaMethod.reflect();

However, this will throw an Exception, if the respective classes are missing on the classpath (e.g. because they were just imported from some file path).

This restriction also applies to handling annotations in a more convenient way. Consider the following annotation:

@interface CustomAnnotation {
    String value();
}

If you need to access this annotation without it being on the classpath, you must rely on

JavaAnnotation<?> annotation = javaClass.getAnnotationOfType("some.pkg.CustomAnnotation");
// result is untyped, since it might not be on the classpath (e.g. enums)
Object value = annotation.get("value");

So there is neither type safety nor automatic refactoring support. If this annotation is on the classpath, however, this can be written way more naturally:

CustomAnnotation annotation = javaClass.getAnnotationOfType(CustomAnnotation.class);
String value = annotation.value();

ArchUnit’s own rule APIs (compare The Lang API) never rely on the classpath though. Thus the evaluation of default rules and syntax combinations, described in the next section, does not depend on whether the classes were imported from the classpath or some JAR / folder.

7. The Lang API

7.1. Composing Class Rules

The Core API is pretty powerful with regard to all the details from the bytecode that it provides to tests. However, tests written this way lack conciseness and fail to convey the architectural concept that they should assert. Consider:

Set<JavaClass> services = new HashSet<>();
for (JavaClass clazz : classes) {
    // choose those classes with FQN with infix '.service.'
    if (clazz.getName().contains(".service.")) {
        services.add(clazz);
    }
}

for (JavaClass service : services) {
    for (JavaAccess<?> access : service.getAccessesFromSelf()) {
        String targetName = access.getTargetOwner().getName();

        // fail if the target FQN has the infix ".controller."
        if (targetName.contains(".controller.")) {
            String message = String.format(
                    "Service %s accesses Controller %s in line %d",
                    service.getName(), targetName, access.getLineNumber());
            Assert.fail(message);
        }
    }
}

What we want to express, is the rule "no classes that reside in a package 'service' should access classes that reside in a package 'controller'". Nevertheless, it’s hard to read through that code and distill that information. And the same process has to be done every time someone needs to understand the semantics of this rule.

To solve this shortcoming, ArchUnit offers a high level API to express architectural concepts in a concise way. In fact, we can write code that is almost equivalent to the prose rule text mentioned before:

ArchRule rule = ArchRuleDefinition.noClasses()
    .that().resideInAPackage("..service..")
    .should().accessClassesThat().resideInAPackage("..controller..");

rule.check(importedClasses);

The only difference to colloquial language is the ".." in the package notation, which refers to any number of packages. Thus "..service.." just expresses "any package that contains some sub-package 'service'", e.g. com.myapp.service.any. If this test fails, it will report an AssertionError with the following message:

java.lang.AssertionError: Architecture Violation [Priority: MEDIUM] -
Rule 'no classes that reside in a package '..service..'
should access classes that reside in a package '..controller..'' was violated (1 times):
Method <some.pkg.service.SomeService.callController()>
calls method <some.pkg.controller.SomeController.execute()>
in (SomeService.java:14)

So as a benefit, the assertion error contains the full rule text out of the box and reports all violations including the exact class and line number. The rule API also allows to combine predicates and conditions:

noClasses()
    .that().resideInAPackage("..service..")
    .or().resideInAPackage("..persistence..")
    .should().accessClassesThat().resideInAPackage("..controller..")
    .orShould().accessClassesThat().resideInAPackage("..ui..")

rule.check(importedClasses);

7.2. Composing Member Rules

In addition to a predefined API to write rules about Java classes and their relations, there is an extended API to define rules for members of Java classes. This might be relevant, for example, if methods in a certain context need to be annotated with a specific annotation, or return types implementing a certain interface. The entry point is again ArchRuleDefinition, e.g.

ArchRule rule = ArchRuleDefinition.methods()
    .that().arePublic()
    .and().areDeclaredInClassesThat().resideInAPackage("..controller..")
    .should().beAnnotatedWith(Secured.class);

rule.check(importedClasses);

Besides methods(), ArchRuleDefinition offers the methods members(), fields(), codeUnits(), constructors() – and the corresponding negations noMembers(), noFields(), noMethods(), etc.

7.3. Creating Custom Rules

In fact, most architectural rules take the form

classes that ${PREDICATE} should ${CONDITION}

In other words, we always want to limit imported classes to a relevant subset, and then evaluate some condition to see that all those classes satisfy it. ArchUnit’s API allows you to do just that, by exposing the concepts of DescribedPredicate and ArchCondition. So the rule above is just an application of this generic API:

DescribedPredicate<JavaClass> resideInAPackageService = // define the predicate
ArchCondition<JavaClass> accessClassesThatResideInAPackageController = // define the condition

noClasses().that(resideInAPackageService)
    .should(accessClassesThatResideInAPackageController);

Thus, if the predefined API does not allow to express some concept, it is possible to extend it in any custom way. For example:

DescribedPredicate<JavaClass> haveAFieldAnnotatedWithPayload =
    new DescribedPredicate<JavaClass>("have a field annotated with @Payload"){
        @Override
        public boolean test(JavaClass input) {
            boolean someFieldAnnotatedWithPayload = // iterate fields and check for @Payload
            return someFieldAnnotatedWithPayload;
        }
    };

ArchCondition<JavaClass> onlyBeAccessedBySecuredMethods =
    new ArchCondition<JavaClass>("only be accessed by @Secured methods") {
        @Override
        public void check(JavaClass item, ConditionEvents events) {
            for (JavaMethodCall call : item.getMethodCallsToSelf()) {
                if (!call.getOrigin().isAnnotatedWith(Secured.class)) {
                    String message = String.format(
                        "Method %s is not @Secured", call.getOrigin().getFullName());
                    events.add(SimpleConditionEvent.violated(call, message));
                }
            }
        }
    };

classes().that(haveAFieldAnnotatedWithPayload).should(onlyBeAccessedBySecuredMethods);

If the rule fails, the error message will be built from the supplied descriptions. In the example above, it would be

classes that have a field annotated with @Payload should only be accessed by @Secured methods

7.4. Predefined Predicates and Conditions

Custom predicates and conditions like in the last section can often be composed from predefined elements. ArchUnit’s basic convention for predicates is that they are defined in an inner class Predicates within the type they target. For example, one can find the predicate to check for the simple name of a JavaClass as

JavaClass.Predicates.simpleName(String)

Predicates can be joined using the methods predicate.or(other) and predicate.and(other). So for example a predicate testing for a class with simple name "Foo" that is serializable could be created the following way:

import static com.tngtech.archunit.core.domain.JavaClass.Predicates.assignableTo;
import static com.tngtech.archunit.core.domain.JavaClass.Predicates.simpleName;

DescribedPredicate<JavaClass> serializableNamedFoo =
    simpleName("Foo").and(assignableTo(Serializable.class));

Note that for some properties, there exist interfaces with predicates defined for them. For example the property to have a name is represented by the interface HasName; consequently the predicate to check the name of a JavaClass is the same as the predicate to check the name of a JavaMethod, and resides within

HasName.Predicates.name(String)

This can at times lead to problems with the type system, if predicates are supposed to be joined. Since the or(..) method accepts a type of DescribedPredicate<? super T>, where T is the type of the first predicate. For example:

// Does not compile, because type(..) targets a subtype of HasName
HasName.Predicates.name("").and(JavaClass.Predicates.type(Serializable.class))

// Does compile, because name(..) targets a supertype of JavaClass
JavaClass.Predicates.type(Serializable.class).and(HasName.Predicates.name(""))

// Does compile, because the compiler now sees name(..) as a predicate for JavaClass
DescribedPredicate<JavaClass> name = HasName.Predicates.name("").forSubtype();
name.and(JavaClass.Predicates.type(Serializable.class));

This behavior is somewhat tedious, but unfortunately it is a shortcoming of the Java type system that cannot be circumvented in a satisfying way.

Just like predicates, there exist predefined conditions that can be combined in a similar way. Since ArchCondition is a less generic concept, all predefined conditions can be found within ArchConditions. Examples:

ArchCondition<JavaClass> callEquals =
    ArchConditions.callMethod(Object.class, "equals", Object.class);
ArchCondition<JavaClass> callHashCode =
    ArchConditions.callMethod(Object.class, "hashCode");

ArchCondition<JavaClass> callEqualsOrHashCode = callEquals.or(callHashCode);

7.5. Rules with Custom Concepts

Earlier we stated that most architectural rules take the form

classes that ${PREDICATE} should ${CONDITION}

However, we do not always talk about classes, if we express architectural concepts. We might have custom language, we might talk about modules, about slices, or on the other hand more detailed about fields, methods or constructors. A generic API will never be able to support every imaginable concept out of the box. Thus ArchUnit’s rule API has at its foundation a more generic API that controls the types of objects that our concept targets.

import vs lang

To achieve this, any rule definition is based on a ClassesTransformer that defines how JavaClasses are to be transformed to the desired rule input. In many cases, like the ones mentioned in the sections above, this is the identity transformation, passing classes on to the rule as they are. However, one can supply any custom transformation to express a rule about a different type of input object. For example:

ClassesTransformer<JavaPackage> packages = new AbstractClassesTransformer<JavaPackage>("packages") {
    @Override
    public Iterable<JavaPackage> doTransform(JavaClasses classes) {
        Set<JavaPackage> result = new HashSet<>();
        classes.getDefaultPackage().traversePackageTree(alwaysTrue(), new PackageVisitor() {
            @Override
            public void visit(JavaPackage javaPackage) {
                result.add(javaPackage);
            }
        });
        return result;
    }
};

all(packages).that(containACoreClass()).should(...);

Of course these transformers can represent any custom concept desired:

// how we map classes to business modules
ClassesTransformer<BusinessModule> businessModules = ...

// filter business module dealing with orders
DescribedPredicate<BusinessModule> dealWithOrders = ...

// check that the actual business module is independent of payment
ArchCondition<BusinessModule> beIndependentOfPayment = ...

all(businessModules).that(dealWithOrders).should(beIndependentOfPayment);

7.6. Controlling the Rule Text

If the rule is straight forward, the rule text that is created automatically should be sufficient in many cases. However, for rules that are not common knowledge, it is good practice to document the reason for this rule. This can be done in the following way:

classes().that(haveAFieldAnnotatedWithPayload).should(onlyBeAccessedBySecuredMethods)
    .because("@Secured methods will be intercepted, checking for increased privileges " +
        "and obfuscating sensitive auditing information");

Nevertheless, the generated rule text might sometimes not convey the real intention concisely enough, e.g. if multiple predicates or conditions are joined. It is possible to completely overwrite the rule description in those cases:

classes().that(haveAFieldAnnotatedWithPayload).should(onlyBeAccessedBySecuredMethods)
    .as("Payload may only be accessed in a secure way");

7.7. Ignoring Violations

In legacy projects there might be too many violations to fix at once. Nevertheless, that code should be covered completely by architecture tests to ensure that no further violations will be added to the existing code. One approach to ignore existing violations is to tailor the that(..) clause of the rules in question to ignore certain violations. A more generic approach is to ignore violations based on simple regex matches. For this one can put a file named archunit_ignore_patterns.txt in the root of the classpath. Every line will be interpreted as a regular expression and checked against reported violations. Violations with a message matching the pattern will be ignored. If no violations are left, the check will pass.

For example, suppose the class some.pkg.LegacyService violates a lot of different rules. It is possible to add

archunit_ignore_patterns.txt
.*some\.pkg\.LegacyService.*

All violations mentioning some.pkg.LegacyService will consequently be ignored, and rules that are only violated by such violations will report success instead of failure.

It is possible to add comments to ignore patterns by prefixing the line with a '#':

archunit_ignore_patterns.txt
# There are many known violations where LegacyService is involved; we'll ignore them all
.*some\.pkg\.LegacyService.*

8. The Library API

The Library API offers a growing collection of predefined rules, which offer a more concise API for more complex but common patterns, like a layered architecture or checks for cycles between slices (compare What to Check).

8.1. Architectures

The entrance point for checks of common architectural styles is:

com.tngtech.archunit.library.Architectures

At the moment this only provides a convenient check for a layered architecture and onion architecture. But in the future it might be extended for styles like a pipes and filters, separation of business logic and technical infrastructure, etc.

8.1.1. Layered Architecture

In layered architectures, we define different layers and how those interact with each other. An example setup for a simple 3-tier architecture can be found in Layer Checks.

8.1.2. Onion Architecture

In an "Onion Architecture" (also known as "Hexagonal Architecture" or "Ports and Adapters"), we can define domain packages and adapter packages as follows.

onionArchitecture()
        .domainModels("com.myapp.domain.model..")
        .domainServices("com.myapp.domain.service..")
        .applicationServices("com.myapp.application..")
        .adapter("cli", "com.myapp.adapter.cli..")
        .adapter("persistence", "com.myapp.adapter.persistence..")
        .adapter("rest", "com.myapp.adapter.rest..");

The semantic follows the descriptions in https://jeffreypalermo.com/2008/07/the-onion-architecture-part-1/. More precisely, the following holds:

  • The domain package is the core of the application. It consists of two parts.

    1. The domainModels packages contain the domain entities.

    2. The packages in domainServices contains services that use the entities in the domainModel packages.

  • The applicationServices packages contain services and configuration to run the application and use cases. It can use the items of the domain package but there must not be any dependency from the domain to the application packages.

  • The adapter package contains logic to connect to external systems and/or infrastructure. No adapter may depend on another adapter. Adapters can use both the items of the domain as well as the application packages. Vice versa, neither the domain nor the application packages must contain dependencies on any adapter package.

onion architecture check

8.2. Slices

Currently, there are two "slice" rules offered by the Library API. These are basically rules that slice the code by packages, and contain assertions on those slices. The entrance point is:

com.tngtech.archunit.library.dependencies.SlicesRuleDefinition

The API is based on the idea to sort classes into slices according to one or several package infixes, and then write assertions against those slices. At the moment this is for example:

// sort classes by the first package after 'myapp'
// then check those slices for cyclic dependencies
SlicesRuleDefinition.slices().matching("..myapp.(*)..").should().beFreeOfCycles()

// checks all subpackages of 'myapp' for cycles
SlicesRuleDefinition.slices().matching("..myapp.(**)").should().notDependOnEachOther()

// sort classes by packages between 'myapp' and 'service'
// then check those slices for not having any dependencies on each other
SlicesRuleDefinition.slices().matching("..myapp.(**).service..").should().notDependOnEachOther()

If this constraint is too rigid, e.g. in legacy applications where the package structure is rather inconsistent, it is possible to further customize the slice creation. This can be done by specifying a mapping of JavaClass to SliceIdentifier where classes with the same SliceIdentifier will be sorted into the same slice. Consider this example:

SliceAssignment legacyPackageStructure = new SliceAssignment() {
    // this will specify which classes belong together in the same slice
    @Override
    public SliceIdentifier getIdentifierOf(JavaClass javaClass) {
        if (javaClass.getPackageName().startsWith("com.oldapp")) {
            return SliceIdentifier.of("Legacy");
        }
        if (javaClass.getName().contains(".esb.")) {
            return SliceIdentifier.of("ESB");
        }
        // ... further custom mappings

        // if the class does not match anything, we ignore it
        return SliceIdentifier.ignore();
    }

    // this will be part of the rule description if the test fails
    @Override
    public String getDescription() {
        return "legacy package structure";
    }
};

SlicesRuleDefinition.slices().assignedFrom(legacyPackageStructure).should().beFreeOfCycles()

8.2.1. Configurations

There are two configuration parameters to adjust the behavior of the cycle detection. They can be configured via archunit.properties (compare Advanced Configuration).

archunit.properties
# This will limit the maximum number of cycles to detect and thus required CPU and heap.
# default is 100
cycles.maxNumberToDetect=50

# This will limit the maximum number of dependencies to report per cycle edge.
# Note that ArchUnit will regardless always analyze all dependencies to detect cycles,
# so this purely affects how many dependencies will be printed in the report.
# Also note that this number will quickly affect the required heap since it scales with number.
# of edges and number of cycles
# default is 20
cycles.maxNumberOfDependenciesPerEdge=5

8.2.2. The Cycle Detection Core API

The underlying infrastructure for cycle detection that the slices() rule makes use of can also be accessed without any rule syntax around it. This allows to use the pure cycle detection algorithm in custom checks or libraries. The core class of the cycle detection is

com.tngtech.archunit.library.cycle_detection.CycleDetector

It can be used on a set of a generic type NODE in combination with a generic Set<EDGE> (where EDGE implements Edge<NODE>) representing the edges of the graph:

Set<MyNode> nodes = // ...
Set<Edge<MyNode>> edges = // ...
Cycles<Edge<MyNode>> foundCycles = CycleDetector.detectCycles(nodes, edges);

Edges are parameterized by a generic type EDGE to allow custom edge types that can then transport additional meta-information if needed.

8.3. Modularization Rules

Note: ArchUnit doesn’t strive to be a "competition" for module systems like the Java Platform Module System. Such systems have advantages like checks at compile time versus test time as ArchUnit does. So, if another module system works well in your environment, there is no need to switch over. But ArchUnit can bring JPMS-like features to older code bases, e.g. Java 8 projects, or environments where the JPMS is for some reason no option. It also can accompany a module system by adding additional rules e.g. on the API of a module.

To express the concept of modularization ArchUnit offers ArchModules. The entrypoint into the API is ModuleRuleDefinition, e.g.

ModuleRuleDefinition.modules().definedByPackages("..example.(*)..").should().beFreeOfCycles();

As the example shows, it shares some concepts with the Slices API. For example definedByPackages(..) follows the same semantics as slices().matching(..). Also, the configuration options for cycle detection mentioned in the last section are shared by these APIs. But, it also offers several powerful concepts beyond that API to express many different modularization scenarios.

One example would be to express modules via annotation. We can introduce a custom annotation like @AppModule and follow a convention to annotate the top-level package-info file of each package we consider the root of a module. E.g.

com/myapp/example/module_one/package-info.java
@AppModule(
  name = "Module One",
  allowedDependencies = {"Module Two", "Module Three"},
  exposedPackages = {"..module_one.api.."}
)
package com.myapp.example.module_one;

We can then define a rule using this annotation:

modules()
  .definedByAnnotation(AppModule.class)
  .should().respectTheirAllowedDependenciesDeclaredIn("allowedDependencies",
      consideringOnlyDependenciesInAnyPackage("..example.."))
  .andShould().onlyDependOnEachOtherThroughPackagesDeclaredIn("exposedPackages")

As the example shows, the syntax carries on meta-information (like the annotation of the annotated package-info) into the created ArchModule objects where it can be used to define the rule. In this example, the allowed dependencies are taken from the @AppModule annotation on the respective package-info and compared to the actual module dependencies. Any dependency not listed is reported as violation. Likewise, the exposed packages are taken from the @AppModule annotation and any dependency where the target class’s package doesn’t match any declared package identifier is reported as violation.

Note that the modules() API can be adjusted in many ways to model custom requirements. For further details, please take a look at the examples provided here.

8.3.1. Modularization Core API

The infrastructure to create modules and inspect their dependencies can also be used outside the rule syntax, e.g. for custom checks or utility code:

ArchModules<?> modules = ArchModules.defineByPackages("..example.(*)..").modularize(javaClasses);
ArchModule<?> coreModule = modules.getByIdentifier("core");
Set<? extends ModuleDependency<?>> coreDependencies = coreModule.getModuleDependenciesFromSelf();
coreDependencies.forEach(...);

8.4. General Coding Rules

The Library API also offers a small set of coding rules that might be useful in various projects. Those can be found within

com.tngtech.archunit.library

8.4.1. GeneralCodingRules

The class GeneralCodingRules contains a set of very general rules and conditions for coding. For example:

  • To check that classes do not access System.out or System.err, but use logging instead.

  • To check that classes do not throw generic exceptions, but use specific exceptions instead.

  • To check that classes do not use java.util.logging, but use other libraries like Log4j, Logback, or SLF4J instead

  • To check that classes do not use JodaTime, but use java.time instead.

  • To check that classes do not use field injection, but constructor injection instead.

8.4.2. DependencyRules

The class DependencyRules contains a set of rules and conditions for checking dependencies between classes. For example:

  • To check that classes do not depend on classes from upper packages.

8.4.3. ProxyRules

The class ProxyRules contains a set of rules and conditions for checking the usage of proxy objects. For example:

  • To check that methods that matches a predicate are not called directly from within the same class.

8.5. PlantUML Component Diagrams as rules

The Library API offers a feature that supports PlantUML diagrams. This feature is located in

com.tngtech.archunit.library.plantuml

ArchUnit can derive rules straight from PlantUML diagrams and check to make sure that all imported JavaClasses abide by the dependencies of the diagram. The respective rule can be created in the following way:

URL myDiagram = getClass().getResource("my-diagram.puml");

classes().should(adhereToPlantUmlDiagram(myDiagram, consideringAllDependencies()));

Diagrams supported have to be component diagrams and associate classes to components via stereotypes. The way this works is to use the respective package identifiers (compare ArchConditions.onlyHaveDependenciesInAnyPackage(..)) as stereotypes:

simple plantuml archrule example
@startuml
[Some Source] <<..some.source..>>
[Some Target] <<..some.target..>> as target

[Some Source] --> target
@enduml

Consider this diagram applied as a rule via adhereToPlantUmlDiagram(..), then for example a class some.target.Target accessing some.source.Source would be reported as a violation.

8.5.1. Configurations

There are different ways to deal with dependencies of imported classes not covered by the diagram at all. The behavior of the PlantUML API can be configured by supplying a respective Configuration:

// considers all dependencies possible (including java.lang, java.util, ...)
classes().should(adhereToPlantUmlDiagram(
        mydiagram, consideringAllDependencies())

// considers only dependencies specified in the PlantUML diagram
// (so any unknown dependency will be ignored)
classes().should(adhereToPlantUmlDiagram(
        mydiagram, consideringOnlyDependenciesInDiagram())

// considers only dependencies in any specified package
// (control the set of dependencies to consider, e.g. only com.myapp..)
classes().should(adhereToPlantUmlDiagram(
        mydiagram, consideringOnlyDependenciesInAnyPackage("..some.package.."))

It is possible to further customize which dependencies to ignore:

// there are further ignore flavors available
classes().should(adhereToPlantUmlDiagram(mydiagram).ignoreDependencies(predicate))

A PlantUML diagram used with ArchUnit must abide by a certain set of rules:

  1. Components must be declared in the bracket notation (i.e. [Some Component])

  2. Components must have at least one (possible multiple) stereotype(s). Each stereotype in the diagram must be unique and represent a valid package identifier (e.g. <<..example..>> where .. represents an arbitrary number of packages; compare the core API)

  3. Components may have an optional alias (e.g. [Some Component] <<..example..>> as myalias). The alias must be alphanumeric and must not be quoted.

  4. Components may have an optional color (e.g. [Some Component] <<..example..>> #OrangeRed)

  5. Dependencies must use arrows only consisting of dashes (e.g. -->)

  6. Dependencies may go from left to right --> or right to left <--

  7. Dependencies may consist of any number of dashes (e.g -> or ----->)

  8. Dependencies may contain direction hints (e.g. -up->) or color directives (e.g. -[#green]->)

You can compare this diagram of ArchUnit-Examples.

8.6. Freezing Arch Rules

When rules are introduced in grown projects, there are often hundreds or even thousands of violations, way too many to fix immediately. The only way to tackle such extensive violations is to establish an iterative approach, which prevents the code base from further deterioration.

FreezingArchRule can help in these scenarios by recording all existing violations to a ViolationStore. Consecutive runs will then only report new violations and ignore known violations. If violations are fixed, FreezingArchRule will automatically reduce the known stored violations to prevent any regression.

8.6.1. Usage

To freeze an arbitrary ArchRule just wrap it into a FreezingArchRule:

ArchRule rule = FreezingArchRule.freeze(classes().should()./*complete ArchRule*/);

On the first run all violations of that rule will be stored as the current state. On consecutive runs only new violations will be reported. By default FreezingArchRule will ignore line numbers, i.e. if a violation is just shifted to a different line, it will still count as previously recorded and will not be reported.

8.6.2. Configuration

By default FreezingArchRule will use a simple ViolationStore based on plain text files. This is sufficient to add these files to any version control system to continuously track the progress. You can configure the location of the violation store within archunit.properties (compare Advanced Configuration):

archunit.properties
freeze.store.default.path=/some/path/in/a/vcs/repo

Furthermore, it is possible to configure

archunit.properties
# must be set to true to allow the creation of a new violation store
# default is false
freeze.store.default.allowStoreCreation=true

# can be set to false to forbid updates of the violations stored for frozen rules
# default is true
freeze.store.default.allowStoreUpdate=false

This can help in CI environments to prevent misconfiguration: For example, a CI build should probably never create a new the violation store, but operate on an existing one.

As mentioned in Overriding configuration, these properties can be passed as system properties as needed. For example to allow the creation of the violation store in a specific environment, it is possible to pass the system property via

-Darchunit.freeze.store.default.allowStoreCreation=true

It is also possible to allow all violations to be "refrozen", i.e. the store will just be updated with the current state, and the reported result will be success. Thus, it is effectively the same behavior as if all rules would never have been frozen. This can e.g. make sense, because current violations are consciously accepted and should be added to the store, or because the format of some violations has changed. The respective property to allow refreezing all current violations is freeze.refreeze=true, where the default is false.

8.6.3. Extension

FreezingArchRule provides two extension points to adjust the behavior to custom needs. The first one is the ViolationStore, i.e. the store violations will be recorded to. The second one is the ViolationLineMatcher, i.e. how FreezingArchRule will associate lines of stored violations with lines of actual violations. As mentioned, by default this is a line matcher that ignores the line numbers of violations within the same class.

Violation Store

As mentioned in Configuration, the default ViolationStore is a simple text based store. It can be exchanged though, for example to store violations in a database. To provide your own implementation, implement com.tngtech.archunit.library.freeze.ViolationStore and configure FreezingArchRule to use it. This can either be done programmatically:

FreezingArchRule.freeze(rule).persistIn(customViolationStore);

Alternatively it can be configured via archunit.properties (compare Advanced Configuration):

freeze.store=fully.qualified.name.of.MyCustomViolationStore

You can supply properties to initialize the store by using the namespace freeze.store. For properties

freeze.store.propOne=valueOne
freeze.store.propTwo=valueTwo

the method ViolationStore.initialize(props) will be called with the properties

propOne=valueOne
propTwo=valueTwo
Violation Line Matcher

The ViolationLineMatcher compares lines from occurred violations with lines from the store. The default implementation ignores line numbers and numbers of anonymous classes or lambda expressions, and counts lines as equivalent when all other details match. A custom ViolationLineMatcher can again either be defined programmatically:

FreezingArchRule.freeze(rule).associateViolationLinesVia(customLineMatcher);

or via archunit.properties:

freeze.lineMatcher=fully.qualified.name.of.MyCustomLineMatcher

8.7. Software Architecture Metrics

Similar to code quality metrics, like cyclomatic complexity or method length, software architecture metrics strive to measure the structure and design of software. ArchUnit can be used to calculate some well-known software architecture metrics. The foundation of these metrics is generally some form of componentization, i.e. we partition the classes/methods/fields of a Java application into related units and provide measurements for these units. In ArchUnit this concept is expressed by com.tngtech.archunit.library.metrics.MetricsComponent. For some metrics, like the Cumulative Dependency Metrics by John Lakos, we also need to know the dependencies between those components, which are naturally derived from the dependencies between the elements (e.g. classes) within these components.

A very simple concrete example would be to consider some Java packages as components and the classes within these packages as the contained elements. From the dependencies between the classes we can derive which package depends on which other package.

The following will give a quick overview of the metrics that ArchUnit can calculate. However, for further background information it is recommended to rely on some dedicated literature that explains these metrics in full detail.

8.7.1. Cumulative Dependency Metrics by John Lakos

These are software architecture metrics as defined by John Lakos in his book "Large-Scale C++ Software Design". The basic idea is to calculate the DependsOn value for each component, which is the sum of all components that can be transitively reached from some component including the component itself.

From these values we can derive

  • Cumulative Component Dependency (CCD): The sum of all DependsOn values of all components

  • Average Component Dependency (ACD): The CCD divided by the number of all components

  • Relative Average Component Dependency (RACD): The ACD divided by the number of all components

  • Normalized Cumulative Component Dependency (NCCD): The CCD of the system divided by the CCD of a balanced binary tree with the same number of components

Example
lakos example

Thus these metrics provide some insights into the complexity of the dependency graph of a system. Note that in a cycle all elements have the same DependsOn value which will lead to an increased CCD. In fact for any non-trivial (n >= 5) acyclic graph of components the RACD is bound by 0.6.

How to use the API

The values described for these metrics can be calculated in the following way:

import com.tngtech.archunit.library.metrics.ArchitectureMetrics;
// ...

JavaClasses classes = // ...
Set<JavaPackage> packages = classes.getPackage("com.example").getSubpackages();

// These components can also be created in a package agnostic way, compare MetricsComponents.from(..)
MetricsComponents<JavaClass> components = MetricsComponents.fromPackages(packages);

LakosMetrics metrics = ArchitectureMetrics.lakosMetrics(components);

System.out.println("CCD: " + metrics.getCumulativeComponentDependency());
System.out.println("ACD: " + metrics.getAverageComponentDependency());
System.out.println("RACD: " + metrics.getRelativeAverageComponentDependency());
System.out.println("NCCD: " + metrics.getNormalizedCumulativeComponentDependency());

8.7.2. Component Dependency Metrics by Robert C. Martin

These software architecture metrics were defined by Robert C. Martin in various sources, for example in his book "Clean architecture : a craftsman’s guide to software structure and design".

The foundation are again components, that must in this case contain classes as their elements (i.e. these are purely object-oriented metrics that need a concept of abstract classes).

The metrics are based on the following definitions:

  • Efferent Coupling (Ce): The number of outgoing dependencies to any other component

  • Afferent Coupling (Ca): The number of incoming dependencies from any other component

  • Instability (I): Ce / (Ca + Ce), i.e. the relationship of outgoing dependencies to all dependencies

  • Abstractness (A): num(abstract_classes) / num(all_classes) in the component

  • Distance from Main Sequence (D): | A + I - 1 |, i.e. the normalized distance from the ideal line between (A=1, I=0) and (A=0, I=1)

Note that ArchUnit slightly differs from the original definition. In ArchUnit the Abstractness value is only based on public classes, i.e. classes that are visible from the outside. The reason is that Ce, Ca and I all are metrics with respect to coupling of components. But only classes that are visible to the outside can affect coupling between components, so it makes sense to only consider those classes to calculate the A value.

Example

The following provides some example where the A values assume some random factor of abstract classes within the respective component.

martin example
How to use the API

The values described for these metrics can be calculated in the following way:

import com.tngtech.archunit.library.metrics.ArchitectureMetrics;
// ...

JavaClasses classes = // ...
Set<JavaPackage> packages = classes.getPackage("com.example").getSubpackages();

// These components can also be created in a package agnostic way, compare MetricsComponents.from(..)
MetricsComponents<JavaClass> components = MetricsComponents.fromPackages(packages);

ComponentDependencyMetrics metrics = ArchitectureMetrics.componentDependencyMetrics(components);

System.out.println("Ce: " + metrics.getEfferentCoupling("com.example.component"));
System.out.println("Ca: " + metrics.getAfferentCoupling("com.example.component"));
System.out.println("I: " + metrics.getInstability("com.example.component"));
System.out.println("A: " + metrics.getAbstractness("com.example.component"));
System.out.println("D: " + metrics.getNormalizedDistanceFromMainSequence("com.example.component"));

8.7.3. Visibility Metrics by Herbert Dowalil

These software architecture metrics were defined by Herbert Dowalil in his book "Modulare Softwarearchitektur: Nachhaltiger Entwurf durch Microservices, Modulithen und SOA 2.0". They provide a measure for the Information Hiding Principle, i.e. the relation of visible to hidden elements within a component.

The metrics are composed from the following definitions:

  • Relative Visibility (RV): num(visible_elements) / num(all_elements) for each component

  • Average Relative Visibility (ARV): The average of all RV values

  • Global Relative Visibility (GRV): num(visible_elements) / num(all_elements) over all components

Example
dowalil example
How to use the API

The values described for these metrics can be calculated in the following way:

import com.tngtech.archunit.library.metrics.ArchitectureMetrics;
// ...

JavaClasses classes = // ...
Set<JavaPackage> packages = classes.getPackage("com.example").getSubpackages();

// These components can also be created in a package agnostic way, compare MetricsComponents.from(..)
MetricsComponents<JavaClass> components = MetricsComponents.fromPackages(packages);

VisibilityMetrics metrics = ArchitectureMetrics.visibilityMetrics(components);

System.out.println("RV : " + metrics.getRelativeVisibility("com.example.component"));
System.out.println("ARV: " + metrics.getAverageRelativeVisibility());
System.out.println("GRV: " + metrics.getGlobalRelativeVisibility());

9. JUnit Support

At the moment ArchUnit offers extended support for writing tests with JUnit 4 and JUnit 5. This mainly tackles the problem of caching classes between test runs and to remove some boilerplate.

Consider a straight forward approach to write tests:

@Test
public void rule1() {
    JavaClasses importedClasses = new ClassFileImporter().importClasspath();

    ArchRule rule = classes()...

    rule.check(importedClasses);
}

@Test
public void rule2() {
    JavaClasses importedClasses = new ClassFileImporter().importClasspath();

    ArchRule rule = classes()...

    rule.check(importedClasses);
}

For bigger projects, this will have a significant performance impact, since the import can take a noticeable amount of time. Also rules will always be checked against the imported classes, thus the explicit call of check(importedClasses) is bloat and error prone (i.e. it can be forgotten).

9.1. JUnit 4 & 5 Support

Make sure you follow the installation instructions at Installation, in particular to include the correct dependency for the respective JUnit support.

9.1.1. Writing tests

Tests look and behave very similar between JUnit 4 and 5. The only difference is, that with JUnit 4 it is necessary to add a specific Runner to take care of caching and checking rules, while JUnit 5 picks up the respective TestEngine transparently. A test typically looks the following way:

@RunWith(ArchUnitRunner.class) // Remove this line for JUnit 5!!
@AnalyzeClasses(packages = "com.myapp")
public class ArchitectureTest {

    // ArchRules can just be declared as static fields and will be evaluated
    @ArchTest
    public static final ArchRule rule1 = classes().should()...

    @ArchTest
    public static final ArchRule rule2 = classes().should()...

    @ArchTest
    public static void rule3(JavaClasses classes) {
        // The runner also understands static methods with a single JavaClasses argument
        // reusing the cached classes
    }

}

The JavaClass cache will work in two ways. On the one hand it will cache the classes by test, so they can be reused by several rules declared within the same class. On the other hand, it will cache the classes by location, so a second test that wants to import classes from the same URLs will reuse the classes previously imported as well. Note that this second caching uses soft references, so the classes will be dropped from memory, if the heap runs low. For further information see Controlling the Cache.

9.1.2. Controlling the Import

Which classes will be imported can be controlled in a declarative way through @AnalyzeClasses. If no packages or locations are provided, the whole classpath will be imported. You can specify packages to import as strings:

@AnalyzeClasses(packages = {"com.myapp.subone", "com.myapp.subtwo"})

To better support refactorings, packages can also be declared relative to classes, i.e. the packages these classes reside in will be imported:

@AnalyzeClasses(packagesOf = {SubOneConfiguration.class, SubTwoConfiguration.class})

As a third option, locations can be specified freely by implementing a LocationProvider:

public class MyLocationProvider implements LocationProvider {
    @Override
    public Set<Location> get(Class<?> testClass) {
        // Determine Locations (= URLs) to import
        // Can also consider the actual test class, e.g. to read some custom annotation
    }
}

@AnalyzeClasses(locations = MyLocationProvider.class)

Furthermore, to choose specific classes beneath those locations, ImportOptions can be specified (compare The Core API). For example, to import the classpath, but only consider production code, and only consider code that is directly supplied and does not come from JARs:

@AnalyzeClasses(importOptions = {DoNotIncludeTests.class, DoNotIncludeJars.class})

As explained in The Core API, you can write your own custom implementation of ImportOption and then supply the type to @AnalyzeClasses.

9.1.3. Controlling the Cache

By default, all classes will be cached by location. This means that between different test class runs imported Java classes will be reused, if the exact combination of locations has already been imported.

If the heap runs low, and thus the garbage collector has to do a big sweep in one run, this can cause a noticeable delay. On the other hand, if it is known that no other test class will reuse the imported Java classes, it would make sense to deactivate this cache.

This can be achieved by configuring CacheMode.PER_CLASS, e.g.

@AnalyzeClasses(packages = "com.myapp.special", cacheMode = CacheMode.PER_CLASS)

The Java classes imported during this test run will not be cached by location and just be reused within the same test class. After all tests of this class have been run, the imported Java classes will simply be dropped.

9.1.4. Ignoring Tests

It is possible to skip tests by annotating them with @ArchIgnore, for example:

public class ArchitectureTest {

    // will run
    @ArchTest
    public static final ArchRule rule1 = classes().should()...

    // won't run
    @ArchIgnore
    @ArchTest
    public static final ArchRule rule2 = classes().should()...
}

Note for users of JUnit 5: the annotation @Disabled has no effect here. Instead, @ArchIgnore should be used.

9.1.5. Grouping Rules

Often a project might end up with different categories of rules, for example "service rules" and "persistence rules". It is possible to write one class for each set of rules, and then refer to those sets from another test:

public class ServiceRules {
    @ArchTest
    public static final ArchRule ruleOne = ...

    // further rules
}

public class PersistenceRules {
    @ArchTest
    public static final ArchRule ruleOne = ...

    // further rules
}

@RunWith(ArchUnitRunner.class) // Remove this line for JUnit 5!!
@AnalyzeClasses
public class ArchitectureTest {

    @ArchTest
    static final ArchTests serviceRules = ArchTests.in(ServiceRules.class);

    @ArchTest
    static final ArchTests persistenceRules = ArchTests.in(PersistenceRules.class);

}

The runner will include all @ArchTest annotated members within ServiceRules and PersistenceRules and evaluate them against the classes declared within @AnalyzeClasses on ArchitectureTest. This also allows an easy reuse of a rule library in different projects or modules.

9.1.6. Generating Display Names

ArchUnit offers the possibility to generate more readable names in the test report by replacing underscores in the original rule names by spaces. For example, if a method or field is named

some_Field_or_Method_rule

this will appear as

some Field or Method rule

in the test report.

This is similar to JUnit 5’s @DisplayNameGeneration annotation, but because this display name generation does not fit well with ArchUnit’s rule execution and because we’d like to offer this feature for JUnit 4 as well, you can enable display name generation in ArchUnit with a configuration property (see Advanced Configuration):

archunit.properties
junit.displayName.replaceUnderscoresBySpaces=true

If you omit the property (or set it to false) the original rule names are used as display names.

10. Advanced Configuration

Some behavior of ArchUnit can be centrally configured by adding a file archunit.properties to the root of the classpath (e.g. under src/test/resources). This section will outline some global configuration options.

10.1. Overriding configuration

ArchUnit will use exactly the archunit.properties file returned by the context ClassLoader from the classpath root, via the standard Java resource loading mechanism.

It is possible to override any property from archunit.properties, by passing a system property to the respective JVM process executing ArchUnit:

-Darchunit.propertyName=propertyValue

E.g. to override the property resolveMissingDependenciesFromClassPath described in the next section, it would be possible to pass:

-Darchunit.resolveMissingDependenciesFromClassPath=false

10.2. Configuring the Resolution Behavior

As mentioned in Dealing with Missing Classes, it might be preferable to configure a different import behavior if dealing with missing classes wastes too much performance. One way that can be chosen out of the box is to never resolve any missing class from the classpath:

archunit.properties
resolveMissingDependenciesFromClassPath=false

If you want to resolve just some classes from the classpath (e.g. to import missing classes from your own organization but avoid the performance impact of importing classes from 3rd party packages), it is possible to configure only specific packages to be resolved from the classpath:

archunit.properties
classResolver=com.tngtech.archunit.core.importer.resolvers.SelectedClassResolverFromClasspath
classResolver.args=some.pkg.one,some.pkg.two

This configuration would only resolve the packages some.pkg.one and some.pkg.two from the classpath, and stub all other missing classes.

The last example also demonstrates, how the behavior can be customized freely, for example if classes are imported from a different source and are not on the classpath:

First Supply a custom implementation of

com.tngtech.archunit.core.importer.resolvers.ClassResolver

Then configure it

archunit.properties
classResolver=some.pkg.MyCustomClassResolver

If the resolver needs some further arguments, create a public constructor with one List<String> argument, and supply the concrete arguments as

archunit.properties
classResolver.args=myArgOne,myArgTwo

For further details, compare the sources of SelectedClassResolverFromClasspath.

10.2.1. Configuring the Number of Resolution Iterations

It is also possible to apply a more fine-grained configuration to the import dependency resolution behavior.

In particular, the ArchUnit importer distinguishes 6 types of import dependencies:

  • member types (e.g. the type of a field of a class, type of a method parameter, etc.)

  • accesses to types (e.g. a method calls another method of a different class)

  • supertypes (i.e. superclasses that are extended and interfaces that are implemented)

  • enclosing types (i.e. outer classes of nested classes)

  • annotation types, including parameters of annotations

  • generic signature types (i.e. types that comprise a parameterized type, like List and String for List<String>)

For each of these dependency types it is possible to configure the maximum number of iterations to traverse the dependencies. E.g. let us assume we configure the maximum iterations as 2 for member types and we have a class A that declares a field of type B which in turn holds a field of type C. On the first run we would automatically resolve the type B as a dependency of A. On the second run we would then also resolve C as a member type dependency of B.

The configuration parameters with their defaults are the following:

archunit.properties
import.dependencyResolutionProcess.maxIterationsForMemberTypes = 1
import.dependencyResolutionProcess.maxIterationsForAccessesToTypes = 1
import.dependencyResolutionProcess.maxIterationsForSupertypes = -1
import.dependencyResolutionProcess.maxIterationsForEnclosingTypes = -1
import.dependencyResolutionProcess.maxIterationsForAnnotationTypes = -1
import.dependencyResolutionProcess.maxIterationsForGenericSignatureTypes = -1

Note that setting a property to a negative value (e.g. -1) will not stop the resolution until all types for the given sort of dependency have been resolved. E.g. for generic type signatures a value of -1 will by default lead to all types comprising the signature Map<?, List<? extends String>> (that is Map, List and String) to be automatically resolved.

Setting a property to 0 will disable the automatic resolution, which can lead to types being only partially imported or stubbed. For these types the information is likely not complete or even wrong (e.g. an interface might be reported as class if the bytecode was never analyzed, or annotations might be missing).

On the other hand, bigger (or negative) values can have a massive performance impact. The defaults should provide a reasonable behavior. They include the class graph for all types that are used by members or accesses directly and cut the resolution at that point. However, relevant information for these types is fully imported, no matter how many iterations it takes (e.g. supertypes or generic signatures).

10.3. MD5 Sums of Classes

Sometimes it can be valuable to record the MD5 sums of classes being imported to track unexpected behavior. Since this has a performance impact, it is disabled by default, but it can be activated the following way:

archunit.properties
enableMd5InClassSources=true

If this feature is enabled, the MD5 sum can be queried as

javaClass.getSource().get().getMd5sum()

10.4. Fail Rules on Empty Should

By default, ArchUnit will forbid the should-part of rules to be evaluated against an empty set of classes. The reason is that this can lead to rules that by accident do not check any classes at all. Take for example

classes().that().resideInAPackage("com.myapp.old").should()...

Now consider somebody renames the package old to newer. The rule will now always evaluate successfully without any reported error. However, it actually does not check any classes at all anymore. This is likely not what most users want. Thus, by default ArchUnit will fail checking the rule in this case. If you want to allow evaluating such rules, i.e. where the actual input to the should-clause is empty, you can use one of the following ways:

Allow Empty Should on a Per-Rule Basis

On each ArchRule you can use the method ArchRule.allowEmptyShould(..) to override the behavior for a single rule, e.g.

// create a rule that allows that no classes are passed to the should-clause
classes().that()...should()...allowEmptyShould(true)

Allow Empty Should Globally

To allow all rules to be evaluated without checking any classes you can set the following property:

archunit.properties
archRule.failOnEmptyShould=false

10.5. Custom Error Messages

You can configure a custom format to display the failures of a rule.

First Supply a custom implementation of

com.tngtech.archunit.lang.FailureDisplayFormat

Then configure it

archunit.properties
failureDisplayFormat=some.pkg.MyCustomFailureDisplayFormat

One example would be to shorten the fully qualified class names in failure messages:

private static class SimpleClassNameFailureFormat implements FailureDisplayFormat {
    @Override
    public String formatFailure(HasDescription rule, FailureMessages failureMessages, Priority priority) {
        String failureDetails = failureMessages.stream()
                .map(message -> message.replaceAll("<(?:\\w+\\.)+([A-Z][^>]*)>", "<$1>"))
                .collect(joining(lineSeparator()));

        return String.format("Architecture Violation [Priority: %s] - Rule '%s' was violated (%s):%n%s",
                priority.asString(), rule.getDescription(), failureMessages.getInformationAboutNumberOfViolations(), failureDetails);
    }
}

Note that due to the free format how violation texts can be composed, in particular by custom predicates and conditions, there is at the moment no more sophisticated way than plain text parsing. Users can tailor this to their specific environments where they know which sorts of failure formats can appear in practice.