[Soot-list] How transform local variables into class' s fields.

[Eric Bodden]

Adriano,

the best idea would be to not do these transformations manually on the
.jimple files but instead to implement an actual transformation that
does this automatically for you, e.g. using a BodyTransformer in Soot.
The Soot website has many tutorials that show how to do this kind of
thing. Using a BodyTransformer, you can easily modify the Jimple
representation in memory.

Eric
--
Dr. Eric Bodden
Software Technology Group, Technische Universität Darmstadt, Germany
Tel: +49 6151 16-5478    Fax: +49 6151 16-5410
Mailing Address: S2|02 A209, Hochschulstraße 10, 64289 Darmstadt



On 30 December 2010 13:57, adriano cosimati <agenteborghese at gmail.com> wrote:
> In the code .jimple code I changed some variable was missing,
> but the .jimple code format is actually very difficult to modify and
> maintain,
> especially because the same variables are repeated several times, and there
> is a risk of confusion.
> So I chose the output format .grimp that is lexically much closer to the
> Java source than to Java bytecode.
> I also tried the format .dava ; it would be the ideal solution, but because
> the input code  tested had  control flow obfuscated, soot program gave me a
> NullPointerException at time of the rearrangement of  labels in the code.
>
> Any more elegant and practical solution to my problem to know the status of
> a local variables of an obfuscated method , is welcome.
>
> Thanks for your quick response, and many compliments to the creator of
> 'soot' program .
>
>
> 2010/12/30 Eric Bodden <bodden at st.informatik.tu-darmstadt.de>
>>
>> Hi Adriano.
>>
>> > I thought to turn the .class file into a .jimple file and then change
>> > the
>> > code, and turn local variables into instance fields, so I can debug
>> > them.
>>
>> Note that this will only work if your program does not use any form of
>> recursion. In case of recursion you have multiple copies of the same
>> locals on the call stack and a single field is not enough to simulate
>> all those copies.
>>
>> But about your problem...
>>
>> > When i compile into the .class file, the corresponding .java source is
>> > the
>> > same  i started from
>> > (with a little optimization), but no trace of the two added instance's
>> > fileds.
>> > Where am I wrong ?
>>
>> I think I do not understand your question. The transformation
>> apparently shows up in the jimple code, and hence it should also show
>> up in the .class files that Soot outputs.
>>
>> Eric
>
>