Re: [abc-users] Obtaining a program AST using abc

Hi Luca.

As the Figure suggests there is not one single AST but there rather
exists one AST per compilation unit, i.e. per .java source file.
Retrieve the root node for each single AST is easy:

The method ExtensionInfo.passes(Job) is called by polyglot for any
single source file, passing in a Job object. On the Job you can call
ast(), which will return the root node for that AST.

Eric

2008/10/7 Luca Cavallaro <cavallaro_at_elet.polimi.it>:
> Dear Eric,
> sorry to bother you again with my rookie questions :-P
> I found very useful your suggestion to look at abc.aspectj.ExtensionInfo for
> an entry level tutorial. What I couldn't find is how to get a reference to
> the first node of an AspectJ program AST.
> I'll try to be as precise as possible. What I am trying to get access to is
> the "AspectJ AST" reported in figure 3 at page 9 of the report "**Building
> the abc AspectJ compiler with Polyglot and Soot**":
> http://abc.comlab.ox.ac.uk/documents/abc-2004-4.pdf
> Is it any way to get there?
> Thank you very much again.
> Luca
>
> Eric Bodden wrote:
>>
>> For polyglot-based ASTs the principal way of accessing these ASTs is
>> to use a visitor (any subtype of polyglot.visit.NodeVisitor). Examples
>> of those, and how those visitor passes are scheduled can be seen in
>> the different ExtensionInfo classes, e.g. abc.aspectj.ExtensionInfo
>> (for plain AspectJ) or abc.eaj.ExtensionInfo (for the EAJ language
>> extension).
>>
>> If by "store" you mean "store on disk" that may be tricky because you
>> would need to somehow serialize the entire AST. If you just mean
>> "store as a reference" that should be easily possible I suppose.
>>
>> Eric
>>
>> 2008/10/3 Luca Cavallaro <cavallaro_at_elet.polimi.it>:
>>
>>>
>>> Dear Eric,
>>> thank you for your reply. My question was quite generic, so I am trying
>>> to
>>> sharpen it.
>>> I am using polyglot and what I want to do is to compare the ASTs of two
>>> different versions of the same program. My idea is to store somewhere the
>>> first version AST and to compare the second one with the stored one.
>>> May I ask you some further hints on my problem?
>>> Thanks
>>> Luca
>>>
>>> Eric Bodden wrote:
>>>
>>>>
>>>> Hi Luca.
>>>>
>>>> The way in which to access and transform the AST depends on whether or
>>>> not you are using the JastAdd-based frontend for your abc extension.
>>>> So are you using polyglot or JastAdd?
>>>>
>>>> In cases where you are using polyglot, the extension abc.eaj is a good
>>>> entry-level tutorial (although it does not show anything in terms of
>>>> AST transformations IIRC), in cases where you are using JastAdd you
>>>> should rather have a look at abc.ja.eaj.
>>>>
>>>> Hope that helps.
>>>>
>>>> Eric
>>>>
>>>> 2008/10/3 Luca Cavallaro <cavallaro_at_elet.polimi.it>:
>>>>
>>>>
>>>>>
>>>>> Dear all,
>>>>> I am quite new to abc and I was trying to programmatically access and
>>>>> manipulate an aAspectJ program Abstract Syntax Tree using abc.
>>>>> Anybody can suggest me a solution please?
>>>>> Best regards
>>>>> Luca Cavallaro
>>>>>
>>>>> --
>>>>> Luca Cavallaro
>>>>> Ph.D. student
>>>>> DEI, Politecnico di Milano,
>>>>> Via Golgi 40
>>>>> 200133 Milan,
>>>>> Italy
>>>>> /------------------------------------------------------------/
>>>>> Aphasia is the loss of speech in computer scientists when asked:
>>>>> "But of what use is your research?"
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>
>>> --
>>> Luca Cavallaro
>>> Ph.D. student
>>> DEI, Politecnico di Milano,
>>> Via Golgi 40
>>> 200133 Milan,
>>> Italy
>>> /------------------------------------------------------------/
>>> Aphasia is the loss of speech in computer scientists when asked:
>>> "But of what use is your research?"
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>>
>
>
> --
> Luca Cavallaro
> Ph.D. student
> DEI, Politecnico di Milano,
> Via Golgi 40
> 200133 Milan,
> Italy
> /------------------------------------------------------------/
> Aphasia is the loss of speech in computer scientists when asked:
> "But of what use is your research?"
>
>
>
>

-- 
Eric Bodden
Sable Research Group
McGill University, Montréal, Canada