SoMoX Extractor / JDT-Based Java Extractor

Overview

The JDT-based Java extractor is an extractor plugin for the SoMoX Extractor able to parse java source code and build up according elements in a GAST model. The according workflow job reads the GAST Model from the SoMoX Extractor blackboard and uses this as an input for the Java extractor itself which is workflow independent by default. This GAST model handed over to the extractor is then filled with the information extracted from the source code.

Features

Auto Sourcepath Detection

The JDT-based extractor requires a GAST model as input which includes a representation of the file system (files and directories) of the software. It makes use of an auto detection, which receives all java files from the gast model and parse them to extract the java source code information and entities.

Reference Solving

While the code is parsed, a class may be processed which extends another class that has not been processed yet and no according element in the GAST model exists which could be linked. As a result, this reference could not be solved while the JDT AST ist traversed.

To handle those references a reference registry and solver are provided to take care for this. While the JDT AST is traversed, type specific references (e.g. "extends") are registered in the ReferenceRegistry. As a post processing, a ReferenceSolver checks this registry and tries to solve those references based on the GAST model that has been build up during the initial JDT AST processing.

JDT Elements to GAST Mapping

JAVA Model	GAST Model	Status	Used by SoMoX
Package	Package	implemented	yes
Class (Regular)	GASTClass	implemented	yes
Class (Inner)	GASTClass (inner: true)	open	unknown
Interface (Regular)	GASTClass (isInterface: true)	implemented	unknown
Class Extends Class Relationship	InheritanceTypeAccess	implemented	yes
Class Implements Interface Relationship	InheritanceTypeAccess	open	yes
	Root	open	yes
	BasePath	open	yes
	Directory	open	yes
	File	open	yes
	Method	open	yes
	Function	open	yes
	BlockStatement	open	yes
	Visibilities	open	yes
	FormalParameter	open	yes
	Access	open	yes
	FunctionAccess	open	yes
	CompositeAccess	open	yes
	ModelElement	open	yes
	GASTStruct	open	yes

SISSy SQL Statements to get SoMoX GAST usage

Java Language Enhancements: http://download.oracle.com/javase/7/docs/technotes/guides/language/enhancements.html

Extraction Method

The JDT provides two different extraction methods (Java Model and AST Parser) but SoMoX uses only the AST Parser based method.

This decision is done because of the pros and cons of those two. However, the main reason was the requirement of the Java Model alternative to have all the source code within a java project what might not be realistic in a general application of the parser. Further Details about the comparision of those two extraction alternatives:

AST – Abstract Syntax Tree

Advantages

• Complete representation of the code (including expressions)
• Allows code manipulation
• Is not limited to Eclipse JAVA projects

Disadvantages

• slower than java model because of more extracted information
• requires code without compilation errors (Not clear if this holds because it is also able access problems in the compilation units)
• not able to let the parser solve all references by himself when he is working with an incomplete context (i.e. parsing a single java file)

Java Model

Is used i.e. for the outline view because faster to be updated.

Advantages

• Faster because of less information extracted
• is able to handle code with compilation errors

Disadvantages

• Not all information (e.g., the body of a method is not available)
• Disadvantage using IJavaProject
  • The libraries available in the current eclipse environment become available, too
  • it is required to create a new project in the workspace. That needs to be cleaned up. The extractor cannot run within a simple unit test. A plugin unit test is required to ramp up an eclipse environment -> resource demanding
  • All sources must be located within the project or a subdirectory of the project. It is not possible to link external resources. -> it is required to copy all sources into the project

Resources

Good Tutorials about working with the JDT:

• http://www.eclipsecon.org/2005/presentations/EclipseCON2005_Tutorial29.pdf
• http://www.vogella.de/articles/EclipseJDT/article.html
• http://www.eclipse.org/articles/article.php?file=Article-JavaCodeManipulation_AST/index.html

GAST Elements used by SoMoX

Paket	Klasse	Methode	Attributbezug	Rückgabetyp
de.fzi.gast.accesses	Access	getAccessedClass		GASTClass
de.fzi.gast.accesses	Access	setAccessedTarget		void
de.fzi.gast.accesses	InheritanceTypeAccess	setImplementationInheritance	implementationInheritance	void
de.fzi.gast.core	BasePath	getDirectories		EList<Directory>
de.fzi.gast.core	BasePath	setPath	path	void
de.fzi.gast.core	BasePath	setRoot		void
de.fzi.gast.core	Directory	getFiles		EList<File>
de.fzi.gast.core	Directory	getFileSystemPath	fileSystemPath	String
de.fzi.gast.core	Directory	getParentDirectory		Directory
de.fzi.gast.core	Directory	getSubDirectory		EList<Directory>
de.fzi.gast.core	Directory	setBasePath		void
de.fzi.gast.core	Directory	setParentDirectory		void
de.fzi.gast.core	File	getDirectory		Directory
de.fzi.gast.core	File	getFileSystemPath	fileSystemPath	String
de.fzi.gast.core	File	getFullQualifiedPath	fullQualifiedPath	String
de.fzi.gast.core	File	getTypes		EList<GASTType>
de.fzi.gast.core	File	setDirectory		void
de.fzi.gast.core	File	setLinesOfCode	linesOfCode	void
de.fzi.gast.core	File	setSize	size	void
de.fzi.gast.core	File	setSourceFile	sourceFile	void
de.fzi.gast.core	ModelElement	getSissyId	sissyId	int
de.fzi.gast.core	ModelElement	setSissyId	sissyId	void
de.fzi.gast.core	ModelElement	setStatus		void
de.fzi.gast.core	Package	getClasses		EList<GASTClass>
de.fzi.gast.core	Package	getQualifiedName	qualifiedName	String
de.fzi.gast.core	Package	getSubPackages		EList<Package>
de.fzi.gast.core	Package	getSurroundingPackage		Package
de.fzi.gast.core	Package	setSurroundingPackage		void
de.fzi.gast.core	Root	getAllInterfaces	interface	EList<GASTClass>
de.fzi.gast.core	Root	getAllNormalClasses		EList<GASTClass>
de.fzi.gast.core	Root	getBasePaths		EList<BasePath>
de.fzi.gast.core	Root	getPackageByQualifiedName	qualifiedName (Package)	Package
de.fzi.gast.core	Root	getPackages		EList<Package>
de.fzi.gast.functions	Function	getBody		BlockStatement
de.fzi.gast.functions	Function	getFormalParameters		EList<FormalParameter>
de.fzi.gast.functions	Function	getReturnTypeDeclaration		DeclarationTypeAccess
de.fzi.gast.statements	BlockStatement	getStatements		EList<Statement>
de.fzi.gast.types	GASTClass	getAllAccessedClasses		EList<GASTClass>
de.fzi.gast.types	GASTClass	getAllAccesses		EList<Access>
de.fzi.gast.types	GASTClass	getInheritanceTypeAccesses		EList<InheritanceTypeAccess>
de.fzi.gast.types	GASTClass	getInnerClasses	inner	EList<GASTClass>
de.fzi.gast.types	GASTClass	getMethods		EList<Method>
de.fzi.gast.types	GASTClass	getSuperTypes		EList<GASTClass>
de.fzi.gast.types	GASTClass	getSurroundingPackage		Package
de.fzi.gast.types	GASTClass	isInner	inner	boolean
de.fzi.gast.types	GASTClass	isInterface	interface	boolean
de.fzi.gast.types	GASTClass	isPrimitive	primitive	boolean
de.fzi.gast.types	GASTClass	setSurroundingPackage		void
de.fzi.gast.accesses	DeclarationTypeAccess	(parameter from return type)
de.fzi.gast.variables	FormalParameter	(parameter from return type)
de.fzi.gast.statements	Statement	(parameter from return type)
de.fzi.gast.types	GASTType	(parameter from return type)
de.fzi.gast.functions	Method	(parameter from return type)