Using Spoon as a Java library

The Launcher class

The Spoon Launcher (JavaDoc) is used to create the AST model of a project. It can be as short as:

CtClass l = Launcher.parseClass("class A { void m() { System.out.println(\"yeah\");} }");

Or with a plain object:

Launcher launcher = new Launcher();

// path can be a folder or a file
// addInputResource can be called several times
launcher.addInputResource("<path_to_source>"); 
// the compliance level should be set to the java version targeted by the input resources, e.g. Java 17
launcher.getEnvironment().setComplianceLevel(17);

launcher.buildModel();

CtModel model = launcher.getModel();

The current default value for the compliance level is 8. It might cause unexpected issues to have a compliance level lower than the Java version of the input resources.

Pretty-printing modes

See http://spoon.gforge.inria.fr/custom-pretty-printing.html.

The MavenLauncher class

The Spoon MavenLauncher (JavaDoc) is used to create the AST model of a Maven project. It automatically infers the list of source folders and the dependencies from the pom.xml file. This launcher handles multi-module Maven projects.

// the second parameter can be APP_SOURCE / TEST_SOURCE / ALL_SOURCE
MavenLauncher launcher = new MavenLauncher("<path_to_maven_project>", MavenLauncher.SOURCE_TYPE.APP_SOURCE);
launcher.buildModel();
CtModel model = launcher.getModel();

// list all packages of the model
for(CtPackage p : model.getAllPackages()) {
  System.out.println("package: " + p.getQualifiedName());
}
// list all classes of the model
for(CtType<?> s : model.getAllTypes()) {
  System.out.println("class: " + s.getQualifiedName());
}

Note that by default, MavenLauncher relies on an existing local Maven binary to build the project’s classpath. But a constructor allowing the user to skip this step and to provide a custom classpath is available.

MavenLauncher launcher = new MavenLauncher("<path_to_maven_project>",
        MavenLauncher.SOURCE_TYPE.APP_SOURCE,
        new String[] {
            "/home/user/.m2/repository/org/my/jar/1.0/org-my-jar-1.0.jar"
        }
    );
launcher.buildModel();
CtModel model = launcher.getModel();

To avoid invoking Maven over and over to build a classpath that has not changed, it is stored in the file spoon.classpath.tmp (or depending on the scope spoon.classpath-app.tmp or spoon.classpath-test.tmp) in the same folder as the pom.xml. This classpath will be refreshed if the file is deleted or if it has not been modified for 1h.

Analyzing bytecode with JarLauncher

There are two ways to analyze bytecode with Spoon:

• Bytecode resources can be added to the classpath (some information will be extracted through reflection).
• A decompiler may be used, and then, the analyzes will be performed on the decompiled sources.

The Spoon JarLauncher (JavaDoc) is used to create the AST model from a jar. It automatically decompiles class files contained in the jar and analyzes them. If a pom file corresponding to the jar is provided, it will be used to build the classpath containing all dependencies.

//More constructors are available, check the JavaDOc for more information.
JarLauncher launcher = JarLauncher("<path_to_jar>", "<path_to_output_src_dir>", "<path_to_pom>");
launcher.buildModel();
CtModel model = launcher.getModel();

Note that the default decompiler CFR can be changed by providing an instance implementing spoon.decompiler.Decompiler as a parameter.

JarLauncher launcher = new JarLauncher("<path_to_jar>", "<path_to_output_src_dir>", "<path_to_pom>",
    new Decompiler() {
        @Override
        public void decompile(String inputPath, String outputPath, String[] classpath) {
            //Custom decompiler call
        }
    }
);

Spoon provides two out of the shelf decompilers, CFR by default, and Fernflower. You can use the latter like this:

JarLauncher launcher = new JarLauncher(
        "<path_to_jar>",
        "<path_to_output_src_dir>",
        "<path_to_pom>",
        new FernflowerDecompiler(new File("<path_to_output_src_dir>/src/main/java"))
);

Optionally, the classic launcher can be used with DecompiledResource like this:

Launcher launcher = new Launcher();
launcher.addInputResource(
    new DecompiledResource(baseDir.getAbsolutePath(), new String[]{}, new CFRDecompiler(), pathToDecompiledRoot.getPath())
);

Warning The JarLauncher feature (and all features relying on decompilation) are not included in spoon-core but in spoon-decompiler. If you want to use them you should declare a dependency to spoon-decompiler.

Resolution of elements and classpath

Spoon analyzes source code. However, this source code may refer to libraries (as a field, parameter, or method return type). There are two cases:

• Full classpath: all dependencies are in the JVM classpath or are given to the launcher with launcher.getEnvironment().setSourceClasspath("<classpath_project>"); (optional).
• No classpath: some dependencies are unknown and launcher.getEnvironment().setNoClasspath(true) is set.

This has a direct impact on Spoon references. When you consider a reference object (say, a CtTypeReference), there are three cases:

• Case 1 (code available as source code): the reference points to a code element for which the source code is present. In this case, reference.getDeclaration() returns this code element (e.g. TypeReference.getDeclaration returns the CtType representing the given Java file). reference.getTypeDeclaration() is identical to reference.getDeclaration().
• Case 2 (code available as binary in the classpath): the reference points to a code element for which the source code is NOT present, but for which the binary class is in the classpath (either the JVM classpath or the --source-classpath argument). In this case, reference.getDeclaration() returns null and reference.getTypeDeclaration returns a partial CtType built using runtime reflection. Those objects built using runtime reflection are called shadow objects, and you can identify them with method isShadow. (This also holds for getFieldDeclaration and getExecutableDeclaration).
• Case 3 (code not available, aka noclasspath): the reference points to a code element for which the source code is NOT present, but for which the binary class is NOT in the classpath. This is called in Spoon the noclasspath mode. In this case, both reference.getDeclaration() and reference.getTypeDeclaration() return null. (This also holds for getFieldDeclaration and getExecutableDeclaration).

Default and custom classloaders

Spoon uses a URLClassLoader by default to load binaries before parsing them. A URLClassLoader will load a class with the parent classloader first. This means that if there is a name conflict between the JVM classpath and the manual classpath then the former will be chosen. To give priority to the manual classpath, a child-first classloader can be used. Classloaders are set with launcher.getEnvironment().setInputClassLoader(customClassloader).

Declaring the dependency to Spoon

Maven

<dependency>
    <groupId>fr.inria.gforge.spoon</groupId>
    <artifactId>spoon-core</artifactId>
    <version></version>
</dependency>

Gradle

compile 'fr.inria.gforge.spoon:spoon-core:'

Advanced launchers

Incremental Launcher

IncrementalLauncher (JavaDoc) allows cached ASTs and compiled classes. Any Spoon analysis can then be restarted from where it stopped instead of restarting from scratch.

final File cache = new File("<path_to_cache>");
Set<File> inputResources = Collections.singleton(new File("<path_to_sources>"));
Set<String> sourceClasspath = Collections.emptySet(); // Empty classpath

//Start build from cache
IncrementalLauncher launcher = new IncrementalLauncher(inputResources, sourceClasspath, cache);

if (launcher.changesPresent()) {
    System.out.println("There are changes since last save to cache.");
}

CtModel newModel = launcher.buildModel();
//Model is now up to date

launcher.saveCache();
//Cache is now up to date

Fluent LauncherAPI

FluentLauncher (JavaDoc) allows setting most options directly for simple and fluent launcher usage.

For the classic launcher it’s simply:

CtModel model = new FluentLauncher()
                .inputResource("<path_to_sources>")
                .noClasspath(true)
                .outputDirectory("<path_to_outputdir>")
                .processor(....)
                .buildModel();

If you want to use other launchers like the MavenLauncher:

MavenLauncher launcher = new MavenLauncher(....);
CtModel model = new FluentLauncher(launcher)
                .processor(....)
                .encoding(...)
                .buildModel();