Parsing OCL Documents

Parsing OCL Documents

As we saw in the Parsing Constraints and Queries topic, the OCL Facade provides an API for parsing OCL expressions embedded in models as constraints.

The OCL specification defines a Complete OCL text document with which a UML (or Ecore) metamodel may be completed by providing many complementary constraints and expressions. In this case, the concrete syntax for context declarations indicates the context of constraints, equivalent to their placement in models.

As an example, consider the following Complete OCL document:

[Text for cut and paste]

The OCL Input

The Pivot binding provides a UML-aligned representation and so a Complete OCL document can be parsed to provide a similar Resource to that derived from a UML or Ecore metamodel. A Root contains a Model which contains Packages and Classes. The complementing Resource from the Complete OCL document is independent of the similarly structured complemented Resource of the completed metamodel.

The Pivot binding uses an Xtext parser with a UML-aligned output. The input text is therefore specified by a URI and loaded by the Xtext parser to create a Concrete Syntax Resource. This may then be converted to the Pivot Abstract Syntax Resource. The Abstract Syntax Resource has a conventional Model, Package, Class, Operation hierarchy in order to provide a coherent composition context for the Constraints.

The elements of the independent complementing and complemented Resources are merged within CompleteClasses and CompletePackages of the CompleteModel managed behind the OCL facade.

There are therefore two Class objects named Library, one for each Resource. The objects are distinct in so far as they belong to different resources, which can be separately serialized, and in so far as they may appear distinct to OCL expressions that use reflective access. However they are logically merged and the CompleteEnvironment provides utility methods that allow the multiple objects to be accessed as a merged object.

The Ecore binding provided an OCLInput class to supervise the OCL source text, and the result of parsing the document was a List<Constraint>. p.

The Complete OCL document is a textual Resource with an associated text tooling. The OCL facade provides an API to load a Resource from a given URI.

[Text for cut and paste]

Traversing the Constraints

The parsed resurce can be traversed in the same way as other EMF resources.

[Text for cut and paste]

Accessing the Constraints

The contents of the Complete OCL document contribute to a CompleteModel that merges all the contributions. The contributions can therefore be used as if defined in a primary metamodel.

[Text for cut and paste]

Using the Constraints to Validate a Model

The standard EMF validation makes use of an EValidatorRegistry that maps the URI of an EPackage to the derived EValidator that provides the constraints appilcable to the EPackage. If we want to exploit additional constraints defined in a Complete OCL document, we must extend the underlying EValidator. The ComposedValidator enables multiple EValidator to be composed and to behave as a single EValidator. ComposedEValidator.install() replaces the single EValidator by a composite initially containing just the replaced EValidator. A CompleteOCLEObjectValidator provides the additional validation of the given uri and EPackage.

[Text for cut and paste]

The standard Diagnostician does not directly support validation of a Resource. MyDiagnostician remedies this deficiency and provides a SubstitutionLabelProvider that provides slightly better labels within OCL diagnostics.

[Text for cut and paste]

The source for these examples may be found in the org.eclipse.ocl.examples.xtext.tests plugin in model/parsingDocumentsExample.ocl and in src/org/eclipse/ocl/examples/test/xtext/PivotDocumentationExamples.java.