Decompilation Jrp Test
JReversePro (Java Decompiler) Simple Tests
I installed JReversePro version 1.4.1 (binary distribution; Sourceforge page is here).Fibo
For source, see DecompilerFiboTestSource. The output from JReversePro was this:
// JReversePro v 1.4.1 Fri Jan 10 11:31:40 EST 2003
import java.io.PrintStream;
class Fibo{
Fibo()
{
;
return;
}
private static int fib(int i)
{
if (i > 1)
return (Fibo_jrp.fib(i - 1) + Fibo_jrp.fib(i - 2));
return (i);
}
public static void main(String[] stringArr)
throws Exception
{
int i = 0;
int k;
try {
i = Integer.parseInt(stringArr[0]);
}
catch (Exception exception) {
System.out.println("Input error");
System.exit(1);
k = Fibo_jrp.fib(i);
System.out.println(new StringBuffer().append(
"fibonacci(").append(i).append(") = ").append(k).toString());
return;
}
}
}
When this is run, the result is blank (no output at all).
Obviously, the code to actually call the fib function is in the catch clause, instead of outsude the whole try/catch block. When this was corrected (moved a curly bracket), the result was correct.
Casting
For source, see DecompilerCastingTestSource. Here is the output from JReversePro for main:
public static void main(String[] stringArr)
{
int i = 0;
for (;j < 128;) {
System.out.println(new StringBuffer().append("ascii ")
.append(i).append(" character ").append(i).toString());
char j = (char)(i + 1);
}
return;
}
This doesn't compile because j is referenced before it is used. It needs extensive changes to make it work. Also, the nature of the code has changed; it's not working on integers whereas the original program worked on chars.
Inner classes
For source, see DecompilerInnerClassesTestSource. When decompiled with JReversePro, the result is
// Decompiled by JReversePro 1.4.1
public class Usa{
public String name;
public Usa()
{
;
name = "Detroit";
return;
}
}
It seems to sense that there is more there, but the inner classes are not reproduced.
Deuces Wild
As a much larger test (class file is about 30 times as large as Fibo's), I tried to decompile a file called deuceswild.class. It is an applet of some 27.5K bytes (.class file), which plays the casino game of Deuces Wild. It also calculates the optimal play, and prompts the user if an error (suboptimal play) is made. The decompilation was quite fast (for a decompiler written in Java); less than 5 seconds. (It takes javac a little longer to compile the source code than for JReversePro to produce it!) When recompiled, however, it had scores of errors. The first error was at this line of source code:
indeck = new int[][][ 14 ][ 5 ];It's pretty obvious that this should be
indeck = new int[ 14 ][ 5 ];With this correction, the next error was because of an unknown class "card". When this classfile, and a companion called cardeffects.class were downloaded, the errors included
deuceswild.java:650: cannot resolve symbol
symbol : class ActionEvent
^
deuceswild.java:85: incompatible types
found : card[]
required: card
card card = new card[6];
I fixed the first error with
import java.awt.AWTEvent; import java.awt.event.ActionEvent;The second was fixed by adding the missing brackets in the declaration for reference
card (should be type card[], as the compiler told me). There were several of these.
Next there were a lot of tedious problems like this one:
deuceswild.java:360: possible loss of precision
found : double
required: float
return (0.0);
Finally, this slightly interesting case:
do {
int j = 0;
...
} while (j == 1);
The variabe j is out of scope in the while part of the loop. Once all these things (perhaps a dozen edits in all) were fixed, the source code compiled, but did not run properly. For example, after advising me to hold all five cards, it told me "You have a null, You win $0".
As with most Java decompilations, the output looks very impressive, because the decompiler knows the names and types of all class members (methods and data members).
When I tried to decompile the latest version of deuceswild.class (38K verses 27.5K for the one used here), it bombed out with
jreversepro.parser.ClassParserException : Referring invalid constantpool index 658
The Jode, Jad, and Dava decompilers were able to successfully decompile this larger deuceswild.class file, except that Dava had one small error which was easily fixed.
Sable Test Program
For source, see DecompilerSableTestSource. Here is the result for function f (as produced by JReversePro):
public static void f(short i)
{
boolean j;
Drawable drawable;
Circle circle;
if (i > 10) {
Rectangle rectangle = new Rectangle(i , i);
j = rectangle.isFat();
Rectangle rectangle3 = rectangle;
}
else {
circle = new Circle(i);
j = circle.isFat();
drawable = circle;
}
if (!j)
drawable.draw();
return;
}
It types the local variable d correctly (here called drawable). The result is correct, except for the assignment to local variable rectangle3 (which immediately goes out of scope). (This corresponds to line 9 of the original method.)
Optimised bytecodes
For source, see DecompilerOptimisedTestSource. Output from JReversePro 1.4.1 for this code was (method f only):
public static void f(short i)
{
Drawable drawable;
if (i > 10) {
Rectangle rectangle = new Rectangle;
rectangle(i , i);
i = rectangle.isFat();
rectangle = rectangle;
}
else {
drawable = new Circle;
drawable(i);
i = drawable.isFat();
drawable = drawable;
}
if (i == 0)
drawable.draw();
return;
}
Now, in addition to the problems with the variable called d in the original source code, the decompiler does not separate local variable 0 into two distinct variables (initially it is a short integer parameter; later it is the boolean is_fat. Also, it didn't join the call to new Rectangle (and Circle) with the call to the constructor.
Simple control flow
For source, see DecompilerControlFlowTestSource. JReversePro produces:
public int foo(int i ,int j)
{
try {
for (;i < j;)
i = j++ / i;
}
catch (RuntimeException runtimeexception) {
i = 10;
return (j);
}
}
This is wrong. When the exception code is executed, the program should continue to loop. Also, if there is no exception, when i is no longer less than j, the program should return j (the code does not even compile).
Exceptions
For source, see DecompilerExceptionTestSource. Here is JReversePro's output:
public void foo()
{
System.out.println("a");
try {
System.out.println("b");
try {
System.out.println("c");
}
}
catch (**this_class** this) {
System.out.println("g");
System.out.println("d");
}
catch (**this_class** this) {
System.out.println("e");
System.out.println("f");
return;
}
}
The exception variable classes are not correct, the flow from block g should be to block f, d should be in a try clause, not a catch clause; f should not be in a catch clause.
Conclusion
This decompiler is still officially in "alpha" status, and the results confirm this. However, it is one of very few that attempts to type local references.
-- MikeVanEmmerik - 10 Jan 2003