Java Tooling for Working with Aspect Models

General Considerations

In this section, the Java APIs for working with Aspect Models are described. All of the components described in the following subsections can either be included in your project using a dedicated dependency (as described in the respective subsection), or you can use the aspect-model-starter artifact that aggregates all necessary dependencies:

Maven

<dependency>
    <groupId>org.eclipse.esmf</groupId>
    <artifactId>esmf-aspect-model-starter</artifactId>
    <version>2.7.0</version>
</dependency>

Gradle Groovy DSL

implementation 'org.eclipse.esmf:esmf-aspect-model-starter:2.7.0'

Gradle Kotlin DSL

implementation("org.eclipse.esmf:esmf-aspect-model-starter:2.7.0")

The error handling in many APIs is done using the Try type provided by the Java library Vavr. This is similar to a java.lang.Optional in that a successful result can be processed using .map() or .flatMap(), but if an error occurs, the object can also provide the original exception using the getCause() method. Compared with throwing exceptions, using a Try<T> as a return type instead has the advantage of enabling composable error handling. Please see the Vavr User Guide for more information.

You generally have multiple options for handling Try<T> results, similar to dealing with Optional:

Show used imports

import java.util.ArrayList;
import java.util.List;

import io.vavr.collection.Stream;
import io.vavr.control.Try;

final Try<List<String>> tryList = someMethodReturningTryOfListOfString();

// Option 1: forEach to execute code on the value
tryList.forEach( valueList -> {
   final List<String> result1 = new ArrayList<>();
   for ( final String element : valueList ) {
      result1.add( element.toUpperCase() );
   }
} );

// Option 2: map/flatMap values
final List<String> result2 = tryList.toStream().flatMap( Stream::ofAll )
      .map( String::toUpperCase ).toJavaList();

// Option 3: "extract" value and throw on failure
final List<String> valueList = tryList.getOrElseThrow( () ->
      new RuntimeException( tryList.getCause() ) );
final List<String> result3 = new ArrayList<>();
for ( final String element : valueList ) {
   result3.add( element.toUpperCase() );
}

Getting the right version

The tooling can be made available in two different versions: release and milestone. The release version represents a stable version of the product and can be referenced in your projects in the usual way from the Maven Central repository. There is also the possibility to have the intermediate builds made available, these are called milestone builds. Instead of Maven Central, these are released via GitHub Packages mechanism.
To be able to use the artifacts released in this way in your projects, first the right repository has to be added to your pom.xml file:

<repositories>
  <repository>
    <id>github</id>
    <name>ESMF SDK</name>
    <url>https://maven.pkg.github.com/eclipse-esmf/esmf-sdk</url>
    <releases><enabled>true</enabled></releases>
    <snapshots><enabled>true</enabled></snapshots>
  </repository>
</repositories>

Then the desired dependencies can be referenced in the usual way. For an example, please refer to Github - Installing a package.

JDK requirements

The esmf-sdk components are built with Java 17 and require a JDK >= 17 such as Adoptium Temurin.

The esmf-sdk can also be used with a Java 17-compatible GraalVM JDK. With GraalVM you need to make the Graal JavaScript component available, as parts of the SDK such as the Aspect Model validation component require embedded JavaScript execution:

For using the esmf-sdk in the GraalVM itself, install the JS component using the GraalVM Updater: gu install js
For using the esmf-sdk with GraalVM native-image, be sure to use the --language:js option when building the native image; for more information see polyglot programming.

Parsing Aspect Model URNs

The aspect-model-urn artifact provides a way to parse and validate Aspect model element URNs as described in the specification.

Show used imports

import org.eclipse.esmf.aspectmodel.urn.AspectModelUrn;

final AspectModelUrn urn = AspectModelUrn.fromUrn( "urn:samm:com.example:1.0.0#Example" );
final String namespace = urn.getNamespace(); // com.example
final String name = urn.getName();           // Example
final String version = urn.getVersion();     // 1.0.0
final String urnPrefix = urn.getUrnPrefix(); // urn:samm:com.example:1.0.0#

To include the artifact, use the following dependencies:

Maven

<dependency>
    <groupId>org.eclipse.esmf</groupId>
    <artifactId>esmf-aspect-model-urn</artifactId>
    <version>2.7.0</version>
</dependency>

Gradle Groovy DSL

implementation 'org.eclipse.esmf:esmf-aspect-model-urn:2.7.0'

Gradle Kotlin DSL

implementation("org.eclipse.esmf:esmf-aspect-model-urn:2.7.0")

Loading and Saving Aspect Models

Aspect models are, like their Meta Model, described using the Resource Description Format (RDF, [rdf11]) and the Terse RDF Triple Language syntax (TTL, [turtle]). There are two ways of working with Aspect models: Either the model is loaded as an RDF model where the abstraction level directly corresponds to the RDF/Turtle serialization, or the RDF is parsed into a native Java Aspect model representation where the abstraction level corresponds to the SAMM concepts. Both approaches have different use cases and advantages and both are supported by the Aspect model Java tooling:

Working on the RDF level Using the Java Aspect model

Working on the RDF level	Using the Java Aspect model
Low level, focus on power and flexibility Flexibly navigate and filter the model on the RDF statements level Work with models that are valid RDF, but incomplete Aspect models, e.g. in Aspect model editors Use SPARQL [sparql] to execute complex queries on Aspect models	High level, focus on convenience and type-safety Use Aspect-model specific interfaces for type-safe code Use Java-native data types (e.g. `java.math.BigInteger` for `xsd:integer`) Easily traverse the model on the abstraction level of Aspect model elements

Low level, focus on power and flexibility
Flexibly navigate and filter the model on the RDF statements level
Work with models that are valid RDF, but incomplete Aspect models, e.g. in Aspect model editors
Use SPARQL [sparql] to execute complex queries on Aspect models

High level, focus on convenience and type-safety
Use Aspect-model specific interfaces for type-safe code
Use Java-native data types (e.g. java.math.BigInteger for xsd:integer)
Easily traverse the model on the abstraction level of Aspect model elements

As a rule of thumb, if your use case mostly consists of consuming Aspect models, you should prefer the Java Aspect model, if you create or edit Aspect models, this is better done using the RDF API.

Understanding Model Resolution

Loading an Aspect model in either the RDF-based or the Java Aspect model-based APIs does not only comprise parsing the Turtle file, but also the resolution of elements on the RDF level:

An Aspect model refers to meta model elements (everything that starts with samm:), and can refer to namespaces of shared elements (samm-c, samm-e and unit)
It can also refer to elements from other Aspect models or model elements defined in separate Turtle files

You use the model resolver to load an Aspect model, which takes care of the first point: The used meta model elements and elements from shared namespaces are automatically resolved and added to the loaded RDF model. For the second point, you need to provide the model resolver with the information on where to find other Aspect models: In memory, in a local file system, in a remote model repository etc. This is done using a resolution strategy, which is a function from some type T to a Try of a RDF model. Several commonly used resolution strategies are readily available:

The FileSystemStrategy resolves a java.nio.file.Path for a file relative to the models directory
The ClassPathStrategy resolves a model from resources in the Java class path
The EitherStrategy can be used to chain two different strategies of different types

To implement custom resolution strategies (e.g., to resolve models against a different blob storage API), you can base your implementation on the AbstractResolutionStrategy. Using the model resolver requires at least one resolution strategy that can resolve AspectModelUrns (because references in an Aspect model to external model elements use their respective URNs) and can use another resolution strategy, for example for file paths. The following sections show examples for the different variants.

Loading an Aspect Model to work on the RDF Level

The following example shows how to use the AspectModelResolver to load an Aspect Model

Show used imports

import java.io.File;
import java.nio.file.Path;
import java.nio.file.Paths;
import java.util.List;

import org.apache.jena.rdf.model.Model;
import org.apache.jena.rdf.model.Statement;
import org.apache.jena.vocabulary.RDF;
import org.eclipse.esmf.samm.KnownVersion;
import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.aspectmodel.resolver.FileSystemStrategy;
import org.eclipse.esmf.aspectmodel.resolver.ResolutionStrategy;
import org.eclipse.esmf.aspectmodel.resolver.services.VersionedModel;
import org.eclipse.esmf.aspectmodel.urn.AspectModelUrn;
import org.eclipse.esmf.aspectmodel.vocabulary.SAMM;
import io.vavr.control.Try;

final File modelFile = new File( "aspect-models/org.eclipse.esmf.examples.movement/1.0.0/Movement.ttl" ); (1)
final Try<VersionedModel> tryModel = AspectModelResolver.loadAndResolveModel( modelFile ); (2)

// Let's do something with the loaded model on RDF level
tryModel.forEach( versionedModel -> { (3)
   final SAMM samm = new SAMM( KnownVersion.fromVersionString( versionedModel.getMetaModelVersion().toString() ).get() );
   final Model rdfModel = versionedModel.getModel();
   final List<Statement> result = rdfModel.listStatements( null, RDF.type, samm.Aspect() ).toList();(4)
} );

1	The file must be located in a directory according to the models directory structure.
2	You can check if loading succeeded using `versionedModel.isSuccess()` and `versionedModel.isFailure()`. If loading fails, `versionedModel.getCause()` provides more information.
3	The code inside `forEach` is only executed if the model could be loaded successfully.
4	List all statements in the RDF model that declare a `samm:Aspect`.

If you want to choose which model resolution strategy is used, you use the resolveAspectModel() method of the AspectModelResolver. In this case, you need to provide the resolution strategy and the initial input (InputStream, Model or a String containing RDF/Turtle), or one or more URNs of model elements to resolve - have a look at the methods provided by AspectModelResolver.

Show used imports

import java.io.File;
import java.nio.file.Path;
import java.nio.file.Paths;
import java.util.List;

import org.apache.jena.rdf.model.Model;
import org.apache.jena.rdf.model.Statement;
import org.apache.jena.vocabulary.RDF;
import org.eclipse.esmf.samm.KnownVersion;
import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.aspectmodel.resolver.FileSystemStrategy;
import org.eclipse.esmf.aspectmodel.resolver.ResolutionStrategy;
import org.eclipse.esmf.aspectmodel.resolver.services.VersionedModel;
import org.eclipse.esmf.aspectmodel.urn.AspectModelUrn;
import org.eclipse.esmf.aspectmodel.vocabulary.SAMM;
import io.vavr.control.Try;

// The directory containing the models folder structure,
// see models directory structure
final Path modelsRoot = Paths.get( "aspect-models" );
final ResolutionStrategy fileSystemStrategy = new FileSystemStrategy( modelsRoot );
final Try<VersionedModel> tryModel = new AspectModelResolver().resolveAspectModel( fileSystemStrategy,
      AspectModelUrn.fromUrn( "urn:samm:org.eclipse.esmf.examples.movement:1.0.0#Movement" ) );

To include the model resolver artifact, use the following dependencies:

Maven

<dependency>
    <groupId>org.eclipse.esmf</groupId>
    <artifactId>esmf-aspect-model-resolver</artifactId>
    <version>2.7.0</version>
</dependency>

Gradle Groovy DSL

implementation 'org.eclipse.esmf:esmf-aspect-model-resolver:2.7.0'

Gradle Kotlin DSL

implementation("org.eclipse.esmf:esmf-aspect-model-resolver:2.7.0")

Loading an Aspect Model to work on the Java Aspect Model Level

If you have retrieved a Try<VersionedModel> as shown above, you can use the AspectModelLoader to turn the RDF graph representation of the Aspect Model into a type-safe Java object. You can use the getElements() method to get all model elements from the Aspect Model, or, if you expect the model to contain exactly one Aspect (for example in unit tests), you can use the getSingleAspect() method.

Show used imports

import java.io.File;
import java.util.List;
import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.metamodel.NamedElement;
import org.eclipse.esmf.metamodel.loader.AspectModelLoader;
import io.vavr.collection.Stream;

final File modelFile = new File( "aspect-models/org.eclipse.esmf.examples.movement/1.0.0/Movement.ttl" );
final List<String> result = AspectModelResolver.loadAndResolveModel( modelFile )
      .flatMap( AspectModelLoader::getElements )
      .toStream()
      .flatMap( Stream::ofAll )
      .filter( element -> element.is( NamedElement.class ) )
      .map( element -> element.as( NamedElement.class ) )
      .map( modelElement -> String.format( "Model element: %s has URN %s%n",
            modelElement.getName(), modelElement.getAspectModelUrn() ) )
      .toJavaList();

To include the Java Aspect model artifact, use the following dependencies:

Maven

<dependency>
    <groupId>org.eclipse.esmf</groupId>
    <artifactId>esmf-aspect-meta-model-java</artifactId>
    <version>2.7.0</version>
</dependency>

Gradle Groovy DSL

implementation 'org.eclipse.esmf:esmf-aspect-meta-model-java:2.7.0'

Gradle Kotlin DSL

implementation("org.eclipse.esmf:esmf-aspect-meta-model-java:2.7.0")

Serializing a Java Aspect Model into a RDF/Turtle Aspect Model

There are two serialization components available: One to turn a Java Aspect model into a Jena RDF model (the reverse operation of what the AspectModelLoader does) and one to serialize the resulting Jena RDF model in Turtle syntax, formatted according to the guidelines described in the SAMM specification.

Show used imports

import java.io.File;

import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.aspectmodel.serializer.AspectSerializer;
import org.eclipse.esmf.metamodel.Aspect;
import org.eclipse.esmf.metamodel.loader.AspectModelLoader;

// Aspect as created by the AspectModelLoader
final Aspect aspect = // ...

// A String that contains the pretty printed Aspect Model
String aspectString = AspectSerializer.INSTANCE.apply( aspect );

To include the serializer artifact, use the following dependencies:

Maven

<dependency>
    <groupId>org.eclipse.esmf</groupId>
    <artifactId>esmf-aspect-model-serializer</artifactId>
    <version>2.7.0</version>
</dependency>

Gradle Groovy DSL

implementation 'org.eclipse.esmf:esmf-aspect-model-serializer:2.7.0'

Gradle Kotlin DSL

implementation("org.eclipse.esmf:esmf-aspect-model-serializer:2.7.0")

Validating Aspect Models

Aspect Models are validated using the AspectModelValidator. Validation returns a list of Violations. A violation has a human-readable message and a unique error code and provides access to the EvaluationContext which contains references to the model element that caused the violation and the SHACL shape that triggered the violation.

Each possible type of violation is a subtype of the Violation interface and provides additional context information specific to this type, for example, the MinLengthViolation provides int min (the allowed length) and int actual (the length that was encountered in the model).

Consider the following example:

Show used imports

import java.util.List;

import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.aspectmodel.resolver.services.VersionedModel;
import org.eclipse.esmf.aspectmodel.shacl.violation.Violation;
import org.eclipse.esmf.aspectmodel.validation.services.AspectModelValidator;
import org.eclipse.esmf.aspectmodel.validation.services.DetailedViolationFormatter;
import org.eclipse.esmf.aspectmodel.validation.services.ViolationFormatter;
import io.vavr.control.Try;
import java.io.File;

// Try<VersionedModel> as returned by the AspectModelResolver
final Try<VersionedModel> tryModel = // ...

final List<Violation> violations = new AspectModelValidator().validateModel( tryModel );
if ( violations.isEmpty() ) {
   // Aspect Model is valid!
   return;
}

final String validationReport = new ViolationFormatter().apply( violations ); (1)
final String detailedReport = new DetailedViolationFormatter().apply( violations );

class MyViolationVisitor implements Violation.Visitor<String> { (2)
   // ...
}

// Turn the list of Violations into a list of custom descriptions
final Violation.Visitor<String> visitor = new MyViolationVisitor();
final List<String> result = violations.stream()
      .map( violation -> violation.accept( visitor ) )
      .toList();

1	To format the validation result into an human-readable form, use the `ViolationFormatter` or the `DetailedViolationFormatter`. Note that those are only intended for text-based interfaces such as CLIs.
2	Every application dealing with validation results that needs to transform the results into some different structure should instead implement the `Violation.Visitor` Interface with a suitable target type `(String` used as an example here) and use the visitor to handle the different types of Violations in a type-safe way.

To include the model validator, use the following dependencies:

Maven

<dependency>
    <groupId>org.eclipse.esmf</groupId>
    <artifactId>esmf-aspect-model-validator</artifactId>
    <version>2.7.0</version>
</dependency>

Gradle Groovy DSL

implementation 'org.eclipse.esmf:esmf-aspect-model-validator:2.7.0'

Gradle Kotlin DSL

implementation("org.eclipse.esmf:esmf-aspect-model-validator:2.7.0")

Generating Documentation for Aspect Models

Different types of artifacts can be generated from an Aspect model. All corresponding generators are included in the following dependency:

Maven

<dependency>
    <groupId>org.eclipse.esmf</groupId>
    <artifactId>esmf-aspect-model-document-generators</artifactId>
    <version>2.7.0</version>
</dependency>

Gradle Groovy DSL

implementation 'org.eclipse.esmf:esmf-aspect-model-document-generators:2.7.0'

Gradle Kotlin DSL

implementation("org.eclipse.esmf:esmf-aspect-model-document-generators:2.7.0")

The documentation generation APIs provide methods that take as an argument a Function<String, java.io.OutputStream>. This is a mapping function that takes a file name as an input (which is determined by the respective generator) and returns a corresponding OutputStream, for example (but not necessarily) a FileOutputStream. By providing this function when calling the generator method, you can control where the output is written to, even when the generator creates multiple files. For the code examples in the following subsections, we assume that the following method is defined for calling the generators:

OutputStream outputStreamForName( final String aspectFileName ) {
   // Create an OutputStream for the file name, e.g. a FileOutputStream
}

Generating SVG or PNG Diagrams

Using the AspectModelDiagramGenerator, automatically layouted diagrams can be created for Aspect models in the formats PNG, SVG and Graphviz/DOT.

Show used imports

import java.io.OutputStream;
import java.util.Locale;
import java.util.Set;
import org.eclipse.esmf.aspectmodel.generator.diagram.AspectModelDiagramGenerator;
import org.eclipse.esmf.aspectmodel.generator.diagram.AspectModelDiagramGenerator.Format;
import org.eclipse.esmf.metamodel.AspectContext;
import org.eclipse.esmf.metamodel.loader.AspectModelLoader;

import java.io.File;
import java.io.IOException;

// AspectContext as returned by the AspectModelLoader
final AspectContext model = // ...

final AspectModelDiagramGenerator generator = new AspectModelDiagramGenerator( model ); (1)

// Variant 1: Generate a diagram in SVG format using @en descriptions and preferredNames from the model
final OutputStream output = outputStreamForName( "diagram.svg" );
generator.generateDiagram( Format.SVG, Locale.ENGLISH, output ); (2)
output.close();

// Variant 2: Generate diagrams in multiple formats, for all languages that are present in the model.
generator.generateDiagrams( Set.of( Format.PNG, Format.SVG ), this::outputStreamForName ); (3)

1	The diagram generator is initialized with the loaded model.
2	The simple call for one output format and one language (i.e., descriptions and preferredNames of one locale) takes one output stream to write the image to.
3	It is also possible to generate multiple diagrams, one for each combination of output format and language. For that, the set of target formats is given as well as a mapping function.

Generating HTML Documentation

A HTML reference documentation for an Aspect model can be generated as shown in the following code sample. The documentation contains an overview diagram and describes the model elements as specified in the Aspect model. Preferred names and descriptions in the respective language from the Aspect model are shown in the resulting document as part of each model element.

Show used imports

import java.nio.file.Paths;
import java.util.Map;
import org.eclipse.esmf.aspectmodel.generator.docu.AspectModelDocumentationGenerator;
import org.eclipse.esmf.aspectmodel.generator.docu.AspectModelDocumentationGenerator.HtmlGenerationOption;
import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.aspectmodel.resolver.FileSystemStrategy;
import org.eclipse.esmf.aspectmodel.resolver.ResolutionStrategy;
import org.eclipse.esmf.aspectmodel.resolver.services.VersionedModel;
import org.eclipse.esmf.aspectmodel.urn.AspectModelUrn;
import org.eclipse.esmf.metamodel.Aspect;
import org.eclipse.esmf.metamodel.AspectContext;
import org.eclipse.esmf.metamodel.loader.AspectModelLoader;
import java.io.IOException;

final AspectModelUrn targetAspect
      = AspectModelUrn.fromUrn( "urn:samm:org.eclipse.esmf.examples.movement:1.0.0#Movement" );
// VersionedModel as returned by the AspectModelResolver
final VersionedModel model = // ...
final Aspect aspect = AspectModelLoader.getAspects( model ).toJavaStream() (1)
      .flatMap( aspects -> aspects.stream() )
      .filter( theAspect -> theAspect.getAspectModelUrn().map( urn -> urn.equals( targetAspect ) ).orElse( false ) )
      .findFirst().orElseThrow();
final AspectModelDocumentationGenerator generator = (2)
      new AspectModelDocumentationGenerator( new AspectContext( model, aspect ) );

final Map<HtmlGenerationOption, String> options = Map.of(); (3)
generator.generate( this::outputStreamForName, options );

1	Load and select the Aspect from the loaded model for which generation should be generated.
2	The HTML generator is initialized with the context consisting of the loaded RDF model and the selected Aspect.
3	HTML generation can be controlled via a map of options; using an empty map implies default values.

Generating Sample JSON Payload

The sample JSON payload generator is used to create a valid JSON payload for an Aspect model as it could be returned by an Aspect that implements that Aspect model. This follows the mapping rules as defined in the Meta Model specification. The generator uses samm:exampleValues of Properties if present, otherwise random values corresponding to the respective data types are generated.

Show used imports

import java.io.File;
import java.io.IOException;

import org.eclipse.esmf.aspectmodel.generator.json.AspectModelJsonPayloadGenerator;
import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.metamodel.Aspect;
import org.eclipse.esmf.metamodel.loader.AspectModelLoader;

// Aspect as created by the AspectModelLoader
final Aspect aspect = // ...

final AspectModelJsonPayloadGenerator generator = new AspectModelJsonPayloadGenerator( aspect );

// Variant 1: Direct generation into a String
final String jsonString = generator.generateJson();

// Variant 2: Generate using mapping function
generator.generateJson( this::outputStreamForName );

Generating JSON Schema

The JSON schema generator creates a JSON Schema that describes the payload for an Aspect model as it could be returned by an Aspect that implements that Aspect model.

Show used imports

import java.io.ByteArrayOutputStream;
import java.util.Locale;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import org.eclipse.esmf.aspectmodel.generator.jsonschema.AspectModelJsonSchemaGenerator;
import org.eclipse.esmf.aspectmodel.generator.jsonschema.JsonSchemaGenerationConfig;
import org.eclipse.esmf.aspectmodel.generator.jsonschema.JsonSchemaGenerationConfigBuilder;
import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.metamodel.Aspect;
import org.eclipse.esmf.metamodel.loader.AspectModelLoader;
import java.io.File;
import java.io.IOException;

// Aspect as created by the AspectModelLoader
final Aspect aspect = // ...

final JsonSchemaGenerationConfig config = JsonSchemaGenerationConfigBuilder.builder()
      .locale( Locale.ENGLISH )
      .build();
final JsonNode jsonSchema = AspectModelJsonSchemaGenerator.INSTANCE.apply( aspect, config ).getContent();

// If needed, print or pretty print it into a string
final ByteArrayOutputStream out = new ByteArrayOutputStream();
final ObjectMapper objectMapper = new ObjectMapper();

objectMapper.writerWithDefaultPrettyPrinter().writeValue( out, jsonSchema );
final String result = out.toString();

Generating OpenAPI Specification

The OpenAPI specification generator creates either a JSON Schema or a Yaml Spec that specifies an Aspect regarding to the OpenApi specification. The currently used versions corresponds Draft 4 of the JSON Schema specification, and 3.0.1.

Show used imports

import java.io.ByteArrayOutputStream;
import java.io.File;
import java.io.IOException;

import org.eclipse.esmf.aspectmodel.generator.openapi.AspectModelOpenApiGenerator;
import org.eclipse.esmf.aspectmodel.generator.openapi.OpenApiSchemaGenerationConfig;
import org.eclipse.esmf.aspectmodel.generator.openapi.OpenApiSchemaGenerationConfigBuilder;
import org.eclipse.esmf.aspectmodel.generator.openapi.PagingOption;
import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.metamodel.Aspect;
import org.eclipse.esmf.metamodel.loader.AspectModelLoader;
import org.junit.jupiter.api.Test;

import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.databind.node.ObjectNode;
import com.fasterxml.jackson.dataformat.yaml.YAMLMapper;

public class GenerateOpenApi {
   @Test
   public void generateYaml() throws JsonProcessingException {
      final File modelFile = new File( "aspect-models/org.eclipse.esmf.examples.movement/1.0.0/Movement.ttl" );

      // Aspect as created by the AspectModelLoader
      final Aspect aspect = // ...
            // exclude the actual loading from the example to reduce noise
            AspectModelResolver.loadAndResolveModel( modelFile ).flatMap( AspectModelLoader::getSingleAspect ).get();

      //language=yaml
      final OpenApiSchemaGenerationConfig config = OpenApiSchemaGenerationConfigBuilder.builder()
            // Server URL
            .baseUrl( "http://www.example.com" )
            // The resource path which shall be added
            .resourcePath( "/testPath/{parameter}" )
            // Determines whether semantic versioning should be used for the API
            // i.e., true = v1.2.3, false = v1
            .useSemanticVersion( false )
            // A String containing all the information for dynamic properties mentioned in the resource path.
            // The string must be syntactically valid YAML.
            .properties( readYaml( """
                  parameter:
                    name: parameter
                    in: path
                    description: "A parameter."
                    required: true
                    schema:
                      type: string
                  """ ) )
            // Should the query API be added to the generated specification?
            .includeQueryApi( true )
            // The paging Option to be chosen. In case paging is possible:
            // The default for a time related collection is time-based paging.
            // Otherwise the default is offset-based paging.
            .pagingOption( PagingOption.OFFSET_BASED_PAGING )
            .build();

      // Generate pretty-printed YAML
      final AspectModelOpenApiGenerator generator = new AspectModelOpenApiGenerator();
      final String yaml = generator.apply( aspect, config ).getContentAsYaml();
   }

   @Test
   public void generateJson() throws IOException {
      final File modelFile = new File( "aspect-models/org.eclipse.esmf.examples.movement/1.0.0/Movement.ttl" );
      // Aspect as created by the AspectModelLoader
      final Aspect aspect = // ...
            // exclude the actual loading from the example to reduce noise
            AspectModelResolver.loadAndResolveModel( modelFile ).flatMap( AspectModelLoader::getSingleAspect ).get();

      final ObjectMapper objectMapper = new ObjectMapper();
      final OpenApiSchemaGenerationConfig config = OpenApiSchemaGenerationConfigBuilder.builder()
            // Server URL
            .baseUrl( "http://www.example.com" )
            // The resource path which shall be added
            .resourcePath( "/testPath/{parameter}" )
            // Determines whether semantic versioning should be used for the API
            // i.e., true = v1.2.3, false = v1
            .useSemanticVersion( false )
            // A JsonNode containing all the information for variable properties mentioned
            // in the resource path .
            .properties( readYaml( """
                     {
                       "test-Id": {
                         "name": "test-Id",
                         "in": "path",
                         "description": "The ID of the unit.",
                         "required": true,
                         "schema": {
                           "type": "string"
                         }
                       }
                     }
                  """ ) )
            // Should the query API be added to the generated specification?
            .includeQueryApi( true )
            // The paging Option to be chosen. In case paging is possible:
            // The default for a time related collection is time-based paging.
            // Otherwise the default is offset-based paging.
            .pagingOption( PagingOption.OFFSET_BASED_PAGING )
            .build();

      // Generate the JSON
      final AspectModelOpenApiGenerator generator = new AspectModelOpenApiGenerator();
      final JsonNode json = generator.apply( aspect, config ).getContent();

      // If needed, print or pretty print it into a string
      final ByteArrayOutputStream out = new ByteArrayOutputStream();

      objectMapper.writerWithDefaultPrettyPrinter().writeValue( out, json );
      final String result = out.toString();
   }

   private static ObjectNode readYaml( String content ) throws JsonProcessingException {
      return (ObjectNode) new YAMLMapper().readTree( content );
   }
}

Generate OpenAPI YAML

// Aspect as created by the AspectModelLoader
final Aspect aspect = // ...

//language=yaml
final OpenApiSchemaGenerationConfig config = OpenApiSchemaGenerationConfigBuilder.builder()
      // Server URL
      .baseUrl( "http://www.example.com" )
      // The resource path which shall be added
      .resourcePath( "/testPath/{parameter}" )
      // Determines whether semantic versioning should be used for the API
      // i.e., true = v1.2.3, false = v1
      .useSemanticVersion( false )
      // A String containing all the information for dynamic properties mentioned in the resource path.
      // The string must be syntactically valid YAML.
      .properties( readYaml( """
            parameter:
              name: parameter
              in: path
              description: "A parameter."
              required: true
              schema:
                type: string
            """ ) )
      // Should the query API be added to the generated specification?
      .includeQueryApi( true )
      // The paging Option to be chosen. In case paging is possible:
      // The default for a time related collection is time-based paging.
      // Otherwise the default is offset-based paging.
      .pagingOption( PagingOption.OFFSET_BASED_PAGING )
      .build();

// Generate pretty-printed YAML
final AspectModelOpenApiGenerator generator = new AspectModelOpenApiGenerator();
final String yaml = generator.apply( aspect, config ).getContentAsYaml();

Generate OpenAPI JSON

// Aspect as created by the AspectModelLoader
final Aspect aspect = // ...

final ObjectMapper objectMapper = new ObjectMapper();
final OpenApiSchemaGenerationConfig config = OpenApiSchemaGenerationConfigBuilder.builder()
      // Server URL
      .baseUrl( "http://www.example.com" )
      // The resource path which shall be added
      .resourcePath( "/testPath/{parameter}" )
      // Determines whether semantic versioning should be used for the API
      // i.e., true = v1.2.3, false = v1
      .useSemanticVersion( false )
      // A JsonNode containing all the information for variable properties mentioned
      // in the resource path .
      .properties( readYaml( """
               {
                 "test-Id": {
                   "name": "test-Id",
                   "in": "path",
                   "description": "The ID of the unit.",
                   "required": true,
                   "schema": {
                     "type": "string"
                   }
                 }
               }
            """ ) )
      // Should the query API be added to the generated specification?
      .includeQueryApi( true )
      // The paging Option to be chosen. In case paging is possible:
      // The default for a time related collection is time-based paging.
      // Otherwise the default is offset-based paging.
      .pagingOption( PagingOption.OFFSET_BASED_PAGING )
      .build();

// Generate the JSON
final AspectModelOpenApiGenerator generator = new AspectModelOpenApiGenerator();
final JsonNode json = generator.apply( aspect, config ).getContent();

// If needed, print or pretty print it into a string
final ByteArrayOutputStream out = new ByteArrayOutputStream();

objectMapper.writerWithDefaultPrettyPrinter().writeValue( out, json );
final String result = out.toString();

For Enumerations with complex data types, the values are modelled as instances of the Entity defined as the Enumeration’s data type (see Declaring Enumerations for more information). In case the Entity instances contain Properties with a sorted collection as their data type, the order of the values of said collection in the Entity instances is not preserved in the generated OpenAPI specification. Preserving this order in OpenAPI is not possible at this point.

Generating AsyncAPI Specification

The AsyncAPI specification generator creates either a JSON Schema or a Yaml Spec that specifies an Aspect regarding to the AsyncApi specification. The currently used versions correspond to Draft 4 of the JSON Schema specification, and 3.0.0 of the AsyncAPI specification.

Show used imports

import java.io.ByteArrayOutputStream;
import java.io.File;
import java.io.IOException;
import java.nio.charset.StandardCharsets;

import org.eclipse.esmf.aspectmodel.generator.asyncapi.AspectModelAsyncApiGenerator;
import org.eclipse.esmf.aspectmodel.generator.asyncapi.AsyncApiSchemaGenerationConfig;
import org.eclipse.esmf.aspectmodel.generator.asyncapi.AsyncApiSchemaGenerationConfigBuilder;
import org.eclipse.esmf.aspectmodel.resolver.AspectModelResolver;
import org.eclipse.esmf.metamodel.Aspect;
import org.eclipse.esmf.metamodel.loader.AspectModelLoader;

import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.dataformat.yaml.YAMLGenerator;
import com.fasterxml.jackson.dataformat.yaml.YAMLMapper;

Generate AsyncAPI YAML

// Aspect as created by the AspectModelLoader
final Aspect aspect = // ...

final ObjectMapper yamlMapper = new YAMLMapper().enable( YAMLGenerator.Feature.MINIMIZE_QUOTES );

final AsyncApiSchemaGenerationConfig config = AsyncApiSchemaGenerationConfigBuilder.builder()
      // i.e., true = v1.2.3, false = v1
      .useSemanticVersion( false )
      // Set Application id
      .applicationId( "test:serve" )
      .channelAddress( "/123-456/789-012/movement/0.0.1/Movement" )
      .build();

// Generate pretty-printed YAML
final AspectModelAsyncApiGenerator generator = new AspectModelAsyncApiGenerator();
final JsonNode json = generator.apply( aspect, config ).getContent();
final String yaml = yamlMapper.writeValueAsString( json );

final ByteArrayOutputStream out = new ByteArrayOutputStream();

out.write( yaml.getBytes( StandardCharsets.UTF_8 ) );

final String result = out.toString();

Generate AsyncAPI JSON

// Aspect as created by the AspectModelLoader
final Aspect aspect = // ...

final ObjectMapper objectMapper = new ObjectMapper();
final AsyncApiSchemaGenerationConfig config = AsyncApiSchemaGenerationConfigBuilder.builder()
      // i.e., true = v1.2.3, false = v1
      .useSemanticVersion( false )
      // Set Application id
      .applicationId( "test:serve" )
      .channelAddress( "/123-456/789-012/movement/0.0.1/Movement" )
      .build();

// Generate the JSON
final AspectModelAsyncApiGenerator generator = new AspectModelAsyncApiGenerator();
final JsonNode json = generator.apply( aspect, config ).getContent();

// If needed, print or pretty print it into a string
final ByteArrayOutputStream out = new ByteArrayOutputStream();
objectMapper.writerWithDefaultPrettyPrinter().writeValue( out, json );
final String result = out.toString();

Generating Java Code for Aspect Models

Java code can be generated from an Aspect model in two ways:

The generated code represents the Aspect payload. Aspects and Entities become Java classes; their Properties become fields in the classes. Characteristics are not first-class elements, but are implicitly represented by the usage of corresponding data types (e.g. using java.util.Set as the type for the Set Characteristic of a Property) or javax.validation annotations. The generated classes can be used in a straightforward fashion, but they do not contain information about the underlying Aspect model such as its version number. Parts of the Aspect model that have no representation in its corresponding JSON payload are not part of those classes either, in particular descriptions and preferred names. These classes are called POJOs (Plain Old Java Objects), as they do not contain logic but serve mainly as data containers.
The generated code represents the Aspect model itself: It is a type-safe variant of the model and includes every information that is also present in the model such as Characteristics, descriptions including language tags and original XSD data types. It is however not intended to store payload corresponding to an Aspect. Theses classes are called static meta classes, because they are created at compile time (static) and talk about the structure of the information, not the information itself (meta).

Depending on the use case, you would either use one or both of the types simultaneously.

To include the Java generator, use the following dependencies:

Maven

<dependency>
    <groupId>org.eclipse.esmf</groupId>
    <artifactId>esmf-aspect-model-java-generator</artifactId>
    <version>2.7.0</version>
</dependency>

Gradle Groovy DSL

implementation 'org.eclipse.esmf:esmf-aspect-model-java-generator:2.7.0'

Gradle Kotlin DSL

implementation("org.eclipse.esmf:esmf-aspect-model-java-generator:2.7.0")

Type Mapping

In the Java code generated from an Aspect model, the scalar Aspect model data types are mapped to native Java types. The following table lists the correspondences.

Aspect model type Java native type

Aspect model type	Java native type
`xsd:string`	`java.lang.String`
`xsd:boolean`	`java.lang.Boolean`
`xsd:decimal`	`java.math.BigDecimal`
`xsd:integer`	`java.math.BigDecimal`
`xsd:double`	`java.lang.Double`
`xsd:float`	`java.lang.Float`
`xsd:date`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:time`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:dateTime`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:dateTimeStamp`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:gYear`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:gMonth`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:gDay`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:gYearMonth`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:gMonthDay`	`javax.xml.datatype.XMLGregorianCalendar`
`xsd:duration`	`javax.xml.datatype.Duration`
`xsd:yearMonthDuration`	`javax.xml.datatype.Duration`
`xsd:dayTimeDuration`	`javax.xml.datatype.Duration`
`xsd:byte`	`java.lang.Byte`
`xsd:short`	`java.lang.Short`
`xsd:int`	`java.lang.Integer`
`xsd:long`	`java.lang.Long`
`xsd:unsignedByte`	`java.lang.Short`
`xsd:unsignedShort`	`java.lang.Integer`
`xsd:unsignedInt`	`java.lang.Long`
`xsd:unsignedLong`	`java.math.BigInteger`
`xsd:positiveInteger`	`java.math.BigInteger`
`xsd:nonNegativeInteger`	`java.math.BigInteger`
`xsd:negativeInteger`	`java.math.BigInteger`
`xsd:nonPositiveInteger`	`java.math.BigInteger`
`xsd:hexBinary`	`byte[]`
`xsd:base64Binary`	`byte[]`
`xsd:anyURI`	`java.net.URI`
`samm:curie`	`org.eclipse.esmf.metamodel.datatypes.Curie`

xsd:string

java.lang.String

xsd:boolean

java.lang.Boolean

xsd:decimal

java.math.BigDecimal

xsd:integer

java.math.BigDecimal

xsd:double

java.lang.Double

xsd:float

java.lang.Float

xsd:date