Decompilation Jode Test

Program-Transformation.Org: The Program Transformation Wiki

JODE open source Java Decompiler Simple Tests

Fibo

For source, see DecompilerFiboTestSource. Decompiled source from JODE:
/* Fibo - Decompiled by JODE
 * Visit http://jode.sourceforge.net/
 */

class Fibo
{
    private static int fib(int i) {
        if (i > 1)
            return fib(i - 1) + fib(i - 2);
        return i;
    }
    
    public static void main(String[] strings) throws Exception {
        int i = 0;
        try {
            i = Integer.parseInt(strings[0]);
        } catch (Exception exception) {
            System.out.println("Input error");
            System.exit(1);
        }
        int i_0_ = fib(i);
        System.out.println("fibonacci(" + i + ") = " + i_0_);
    }
}

As you can see, the decompilation is almost identical to the source (save the strange local variable name). The output compiled and ran perfectly with no changes required.

Casting

For source, see DecompilerCastingTestSource. Here is the output from JODE:

/* Casting - Decompiled by JODE
 * Visit http://jode.sourceforge.net/
 */

public class Casting
{
    public static void main(String[] strings) {
        for (char c = '\0'; c < '\u0080'; c++)
            System.out.println("ascii " + (int) c + " character " + c);
    }
}
As you can see, it's quite readable, very similar to the original, and is correct. No modifications were needed to recompile it.

Inner classes

For source, see DecompilerInnerClassesTestSource. When decompiled with JODE, the result is

/* Usa - Decompiled by JODE
 * Visit http://jode.sourceforge.net/
 */

public class Usa
{
    public String name = "Detroit";
    
    public class England
    {
        public String name = "London";
        
        public class Ireland
        {
            public String name = "Dublin";
            
            public void print_names() {
                System.out.println(name);
            }
        }
    }
}

It reproduced the inner classes correctly.

Deuces Wild

This is a 38K applet with two dimensional arrays of integers, some floating point code, and so on. It decompiled in about 8 seconds, with no errors, and recompiled and ran perfectly with no modifications. The arrays of strings are initialised naturally, e.g.

String[] cardrank
        = { "Two", "Three", "Four", "Five", "Six", "Seven", "Eight", "Nine",
            "Ten", "Jack", "Queen", "King", "Ace" };
String[] cardsuit = { "clubs", "diamonds", "hearts", "spades" };

Sable Test Program

For source, see DecompilerSableTestSource. Here is the result for function f (as produced by JODE):

    public static void f(short i) {
        boolean bool;
        Drawable drawable;
        if (i > 10) {
            Rectangle rectangle = new Rectangle(i, i);
            bool = rectangle.isFat();
            drawable = rectangle;
        } else {
            Circle circle = new Circle(i);
            bool = circle.isFat();
            drawable = circle;
        }
        if (!bool)
            drawable.draw();
    }

As you can see, it had no trouble with the tricky d variable (here called drawable), and produced correct code which compiled without modification.

Optimised code.

For source, see DecompilerOptimisedTestSource. This was the result from JODE for the method f:

 
    public static void f(short i) {
        boolean bool;
        Drawable drawable;
        if (i > 10) {
            Rectangle rectangle = new Rectangle;
            ((UNCONSTRUCTED)rectangle).Rectangle(i, i);
            bool = rectangle.isFat();
            drawable = rectangle;
        } else {
            Circle circle = new Circle;
            ((UNCONSTRUCTED)circle).Circle(i);
            bool = circle.isFat();
            drawable = circle;
        }
        if (!bool)
            drawable.draw();
    }

As you can see, it is confused with the two constructors. I have no idea why this is any sort of problem for decompilers; they all see a call to new, then a call to the constructor. They must expect the constructor to immediately follow the call to new, or something.

When the above problems were corrected (e.g. Rectangle rectangle = new Rectangle(i, i), the code compiled and is correct.

Simple control flow

For source, see DecompilerControlFlowTestSource. JODE produces:

    public int foo(int i, int i_0_) {
        while (i < i_0_) {
            try {
                i = i_0_++ / i;
            } catch (RuntimeException runtimeexception) {
                i = 10;
            }
        }
        return i_0_;
    }

While not like the original source code, I think that this would be correct. After an exception, the continue (in the original source) should cause the program to escape to the outer while loop; the Jode version stays in the inner while loop, which is all that there really is in the outer loop.

Exceptions

For source, see DecompilerExceptionTestSource. Unfortunately, JODE bombs out with this assertion:

jode.AssertError: Exception handlers ranges are intersecting:
[16, 47] and [8, 24].

While this is true, and the exception handlers will never overlap in programs compiled from Java source code, this is not a good solution to the problem.

Life

This version of Conway's Game of Life was compiled from Ada 95 source code. The applet is at http://www.appletmagic.com/download/demo/LifeRect.html. The only way I could find to decompile the whole program at once was to jar up all the .class files. When this was done, the decompilation produced numerous type errors, and many of the decompiled results did not recompile. Some of the errors were due to one .class file being missing (interface/java/NonLimited_.class). The applet seems to run fine without this class file.

Conclusion

This is an excellent decompiler: fast, good output quality. It does have a problem with some optimised code, as shown above, but most of the time it's easy to correct. None of the Java decompilers I have tested so far (eight in total) has passed all tests. Of these, and with these limited tests, JODE has the fewer errors.

CategoryDecompilation