Terence Parr schrieb: > Ok,I'm trying to integrate a v3 ANTLR parser into the mainline. We > need this eventually as the new front end to ANTLR itself but also I > wanted to integrate Leon's cool Strip program that knows how to print > grammars back out with lots of different things stripped: > > usage: java org.antlr.morph.Strip [args] file.g [file2.g [file3.g ...]] > -strip-rewrite strip off rewrites and rule/token reference > labels > -strip-label strip off rule/token reference labels > -grammar-option keep grammar options > -grammar-action keep grammar actions, e.g. header, members. > -globalscope keep global scopes > -rule-argument keep rule arguments > -rule-returnvalue keep rule return values > -rule-throws keep rule throws specification > -rule-option keep rule options > -rule-scope keep rule scope > -rule-action keep rule actions, e.g. @init, @after. > -rule-exception keep rule exception group > -alt-action keep alternative actions > -alt-ruleargument keep alternative rule reference arguments > -alt-tokenargument keep alternative token reference arguments > -label keep rule/token reference labels > you need to provide a grammar file. > > I have make sure that the grammar in org/antlr/grammar/v3 is the > latest version we have. thanks to everyone including Leon gave me > fixes. The problem I have is that I will need only ASTs out of the > grammar but Leon's strip program needs parse trees, which requires - > debug option. It uses the debug events to build parsers. I have the > tree grammar that walks an ANTLR AST. I don't want to have two > versions of the grammar; well, actually a single version but with two > different builds: one with and one without debugging.
That would be a good place to be able to switch between parse trees and ASTs via a parameter (whichever way is used to pass it). It would be dumb if there is no general way to solve this problem. Johannes > > So, perhaps I adapt Leon's tool to use the rewrite stream and ASTs > with the tree grammar rather than the parse tree. I think I can get a > lot of his logic into the tree grammar from a parse tree. that is the > direction I'm going. > > With luck, all future stuff can see directly off of a generic AST and > the grammar can stay pristine. > > Ter > > _______________________________________________ > antlr-dev mailing list > [email protected] > http://www.antlr.org/mailman/listinfo/antlr-dev > _______________________________________________ antlr-dev mailing list [email protected] http://www.antlr.org/mailman/listinfo/antlr-dev
