Re: [abc-users] Re: XML attributes in a reweaving analysis

Hi Eric!

I think you can have the weekend off - I appear to have fixed it :) I'm not
sure what I did; I think there were some superfluous tags from a previous stage
of the analysis that were causing the problems. Removing those seems to have
sorted it, and I can now see the LinkTags I added in the Jimple files. They
don't seem to be showing up in the .aj files for some reason, but I'll have a
fiddle around and see what the problem is there.

Many thanks!
Nathan

Eric Bodden wrote:
> That's good news. Could you email me a minimal example (maybe as an
> eclipse project) along with instructions to reproduce the exception?
> Then I may be able to look into it over the weekend.
>
> Eric
>
> 2009/5/29 Nathan Weston <npweston_at_gmail.com>:
>> Further discovery - if I manually insert LinkTags into an XML file after a
>> successful run of the Soot plugin, those changes are reflected in Eclipse.
>> So there should be no problem with integrating with the Soot plugin, as it
>> appears to just get its data from the XML files and doesn't keep a handle.
>>
>> Hope that helps!
>> Nathan
>>
>> Nathan Weston wrote:
>>> Hi Eric,
>>>
>>> Getting closer - having run Soot from the Eclipse plugin and choosing a
>>> similar analysis (Reaching Definitions), it adds the appropriate LinkTags to
>>> .java files but produces the same error as before with .aj files. Perhaps
>>> it isn't linking with the AJDT plugin properly?
>>>
>>> I'm running abc directly from the command-line; I wondered if that might
>>> be an issue but I've been able to add ColorTags in the past (this was a
>>> while ago, using previous versions of both Soot and abc, and Eclipse for
>>> that matter).
>>>
>>> Thanks,
>>> Nath
>>>
>>> Eric Bodden wrote:
>>>> Hi Nathan.
>>>>
>>>> Good to see that we are making process. The exception seems to be
>>>> caused by the fact that the DOM tree is not present when the Soot
>>>> Eclipse plugin tries to parse it. I am not sure why this happens. How
>>>> exactly are you executing abc?
>>>>
>>>> Usually, the way the Soot Eclipse plugin works is that the user starts
>>>> Soot *through the plugin*. Soot then produces the XML files, and -I
>>>> think- retains a handle to the files. That way when you open a Java or
>>>> Jimple file, Soot can access the attribute data from the XML files.
>>>>
>>>> Now if you run Soot (or abc for that matter) differently, i.e. not
>>>> through the UI provided by the Soot plugin I suspect that the plugin
>>>> has problems accessing the file. But that's just a "guesstimation". It
>>>> would really help to know what exactly you are doing to provoke the
>>>> exception.
>>>>
>>>> Eric
>>>>
>>>> P.S. The code for the eclipse plugin is available online in Soot's SVN
>>>> under soot/eclipse.
>>>>
>>>> 2009/5/28 Nathan Weston <npweston_at_gmail.com>:
>>>>> Hi Eric,
>>>>>
>>>>> Rather stupidly I'd instantiated a new Weaver rather than getting the
>>>>> appropriate instantiation from the AbcExtension. D'oh! That works now
>>>>> -
>>>>> the XML attributes are being written correctly. Sadly the color and
>>>>> link
>>>>> tags don't appear to be showing up in Eclipse (either in Jimple or in
>>>>> the
>>>>> original source files), and I get an error when I hover over the
>>>>> statements
>>>>> which have the tags:
>>>>>
>>>>> java.lang.NullPointerException
>>>>> at
>>>>>
>>>>> ca.mcgill.sable.soot.attributes.AttributeDomProcessor.processNode(AttributeDomProcessor.java:55)
>>>>> at
>>>>>
>>>>> ca.mcgill.sable.soot.attributes.AttributeDomProcessor.processAttributesDom(AttributeDomProcessor.java:48)
>>>>> at
>>>>>
>>>>> ca.mcgill.sable.soot.attributes.SootAttributeFilesReader.readFile(SootAttributeFilesReader.java:48)
>>>>> at
>>>>>
>>>>> ca.mcgill.sable.soot.attributes.AbstractAttributesComputer.computeAttributes(AbstractAttributesComputer.java:80)
>>>>> at
>>>>>
>>>>> ca.mcgill.sable.soot.attributes.AbstractAttributesComputer.getHandler(AbstractAttributesComputer.java:133)
>>>>> at
>>>>>
>>>>> ca.mcgill.sable.soot.attributes.AbstractAttributesComputer.getAttributesHandler(AbstractAttributesComputer.java:101)
>>>>> at
>>>>>
>>>>> ca.mcgill.sable.soot.attributes.SootAttributesJimpleHover.getAttributes(SootAttributesJimpleHover.java:58)
>>>>> at
>>>>>
>>>>> ca.mcgill.sable.soot.attributes.AbstractSootAttributesHover.getHoverInfo(AbstractSootAttributesHover.java:78)
>>>>> at
>>>>>
>>>>> org.eclipse.jface.text.TextViewerHoverManager$4.run(TextViewerHoverManager.java:168)
>>>>>
>>>>> If you have any suggestions I'd be glad to hear them, but you've been
>>>>> tremendously helpful already, thanks :)
>>>>>
>>>>> Nathan
>>>>>
>>>>>
>>>>> Eric Bodden wrote:
>>>>>> Hi Nathan.
>>>>>>
>>>>>> I just had a look at the Weaver class. The class initializes the
>>>>>> mapping in the method storeBindings, which is called *before* any
>>>>>> ReweavingAnalysis is executed. Therefore, when you call this method
>>>>>> from within a ReweavingAnalysis then this should actually work. Maybe
>>>>>> you are calling reverseRebind at the wrong point in time? I am using
>>>>>> reverseRebind in line 77 of the class
>>>>>> abc.eaj.weaving.weaver.maybeshared.TLOAnalysisManager if you want to
>>>>>> see an example.
>>>>>>
>>>>>> Hope that helps.
>>>>>>
>>>>>> Eric
>>>>>>
>>>>>> 2009/5/28 Nathan Weston <npweston_at_gmail.com>:
>>>>>>> Hi Eric,
>>>>>>>
>>>>>>> Thanks for your reply - I'd been using
>>>>>>> abc.weaving.weaver.Weaver.rebind(Object), so thanks for putting me
>>>>>>> right! However, now I've changed it to reverseRebind, I'm getting a
>>>>>>> NullPointerException at line 120 in the Weaver class:
>>>>>>>
>>>>>>> Object result= reverseUnitBindings.get(unitTrapOrLocal);
>>>>>>>
>>>>>>> unitTrapOrLocal is certainly not null, so I can only assume it's the
>>>>>>> reverseUnitBindings map which isn't being initialised. Is there
>>>>>>> another method I have to call on the weaver to get this to work?
>>>>>>>
>>>>>>> Many thanks for your help,
>>>>>>> Nathan
>>>>>>>
>>>>>>> 2009/5/21 Eric Bodden <eric.bodden_at_mail.mcgill.ca>:
>>>>>>>> Hi Nathan.
>>>>>>>>
>>>>>>>> Interesting question. Would be nice to see this working... I am not
>>>>>>>> quite sure about the exception that you are seeing but I think I have
>>>>>>>> an explanation for why your tags get lost on the way. The problem is
>>>>>>>> that after you added the tags in your reweaving analysis, abc will
>>>>>>>> (1)
>>>>>>>> replace all method bodies with the bodies that existed before weaving
>>>>>>>> and then (2) do the weaving over again. As you can imagine, (1) also
>>>>>>>> removes your tags.
>>>>>>>>
>>>>>>>> There is however a way to work around this. The method
>>>>>>>> abc.weaving.weaver.Weaver.reverseRebind(Object) gives you for every
>>>>>>>> unit, trap or local that you "see" in the reweaving analysis the
>>>>>>>> corresponding unit, trap or local that you had before weaving (except
>>>>>>>> for those units, traps or local that were introduced through the
>>>>>>>> weaving process, of course). When you attach your tags to those units
>>>>>>>> instead then I believe that abc should preserve them.
>>>>>>>>
>>>>>>>> Hope that helps,
>>>>>>>> Eric
>>>>>>>>
>>>>>>>> 2009/5/21 Nathan Weston <npweston_at_gmail.com>:
>>>>>>>>> Apologies, I should have said I'm using Eclipse 3.4.2, Soot Plugin
>>>>>>>>> 2.3.0 and
>>>>>>>>> abc 1.3.0.
>>>>>>>>>
>>>>>>>>> Thanks,
>>>>>>>>> Nathan
>>>>>>>>>
>>>>>>>>> Nathan Weston wrote:
>>>>>>>>>> Hello all,
>>>>>>>>>>
>>>>>>>>>> I've implemented a new reweaving analysis which I've hooked in to
>>>>>>>>>> abc
>>>>>>>>>> as a
>>>>>>>>>> new reweaving pass. All seems to work well except that I'd like it
>>>>>>>>>> to
>>>>>>>>>> integrate with the Soot Eclipse plugin by providing visualisations
>>>>>>>>>> in
>>>>>>>>>> terms
>>>>>>>>>> of Color and Link Tags. In my reweaving pass I add these tags to
>>>>>>>>>> Units in
>>>>>>>>>> the method bodies, and I've added the Soot arguments
>>>>>>>>>> "-xml-attributes"
>>>>>>>>>> and
>>>>>>>>>> "-f J".
>>>>>>>>>>
>>>>>>>>>> The XML files get generated and placed in an attributes folder
>>>>>>>>>> under
>>>>>>>>>> sootOutput in the appropriate project folder. However, these XML
>>>>>>>>>> files do
>>>>>>>>>> not contain the Link and Color tags I've added, though they do
>>>>>>>>>> contain
>>>>>>>>>> InPreinitializationTags. As well as this, trying to load the
>>>>>>>>>> Jimple
>>>>>>>>>> in
>>>>>>>>>> Eclipse throws the following exception:
>>>>>>>>>>
>>>>>>>>>> java.lang.NullPointerException
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ca.mcgill.sable.soot.attributes.AttributeDomProcessor.processNode(AttributeDomProcessor.java:55)
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ca.mcgill.sable.soot.attributes.AttributeDomProcessor.processAttributesDom(AttributeDomProcessor.java:48)
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ca.mcgill.sable.soot.attributes.SootAttributeFilesReader.readFile(SootAttributeFilesReader.java:48)
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ca.mcgill.sable.soot.attributes.AbstractAttributesComputer.computeAttributes(AbstractAttributesComputer.java:80)
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ca.mcgill.sable.soot.attributes.AbstractAttributesComputer.getHandler(AbstractAttributesComputer.java:129)
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ca.mcgill.sable.soot.attributes.AbstractAttributesComputer.getAttributesHandler(AbstractAttributesComputer.java:101)
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ca.mcgill.sable.soot.resources.SootPartManager.updatePart(SootPartManager.java:51)
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ca.mcgill.sable.soot.resources.EditorActivationListener.partActivated(EditorActivationListener.java:45)
>>>>>>>>>> at
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> org.eclipse.ui.internal.PartListenerList$1.run(PartListenerList.java:72)
>>>>>>>>>> at org.eclipse.core.runtime.SafeRunner.run(SafeRunner.java:37)
>>>>>>>>>> at org.eclipse.core.runtime.Platform.run(Platform.java:880)
>>>>>>>>>> ... lots more
>>>>>>>>>>
>>>>>>>>>> Any help as to where I'm going wrong would be greatly appreciated!
>>>>>>>>>> nathan
>>>>>>>> --
>>>>>>>> Eric Bodden
>>>>>>>> Sable Research Group, McGill University
>>>>>>>> Montréal, Québec, Canada
>>>>>>>>
>>>>>>>>
>>>>>>
>>>>
>>>>
>>
>
>
>
Received on Fri May 29 2009 - 14:47:09 BST