Hi Jean-Christophe, Yes, SmaCC has the infrastructure for doing that, in text form (match an ast subtree, produce a string) and linked to SmaCC produced ASTs (I believe by using a GLR parser to generate multiple matches on an input pattern). Using text is a deliberate design decision [1], as far as I know, and I can understand why ;) But under some conditions, you can easily move from text replacement to AST subtrees injection; ask me for more details.
For a system that transform ASTs, you want want to look at DMS [2]. Just significantly more expensive. Another idea would be a derivative of the RB engine you have in Pharo (after all, RB is from the same guys as SmaCC). At the moment, the rewritting part of SmaCC is not that tested (or not working, for the master/slave client/serveur rewritting engine part), so I'd be really interested to have some feedback if you decide to use it. Thierry [1]John Brant, Don Roberts, Bill Plendl, Jeff Prince, "Extreme maintenance: Transforming Delphi into C#," 2013 IEEE International Conference on Software Maintenance, pp. 1-8, 2010 IEEE International Conference on Software Maintenance, 2010 [2] http://www.semdesigns.com/Products/DMS/DMSToolkit.html 2015-02-04 15:57 GMT+01:00 Jean-Christophe Bach < jeanchristophe.b...@inria.fr>: > Hi, > > We are trying to transform programs (from Delphi to Java). We already > have source and target ASTs and we would like to write the > transformation itself. We are wondering if we could use SmaCC to do that > but we are not able to figure out how. The documentation is not clear > on the fact that it is possible or not to use any AST as input. Does > anyone know if it is possible? > > Then, in the tutorial, the example seems to be a string-based > transformation while we would like to transform ASTs. Are we wrong? What > have we missed? > In Python tests, there are ASTs but we did not find any transformation. > > Has someone a clear and concise example to help us to start? > > Thank you in advance, > > Jean-Christophe >
