I've been pretty sick this last week so I decided to work on something that was fun. After talking with Don Bradshaw at http://www.quipoz.com in Sydney and after trying to convert ANTLR v3's grammar (written in v2) into v3, I decided to build TreeWizard. ANTLR has a nice facility for creating trees from input symbols like:
decl : 'int' ID ';' ; -> ^(DECL 'int' ID)
But, sometimes outside of the grammar file (or in severe cases in a grammar action) you need to build a tree in raw code. Further, there are many cases, as Andy Tripp will tell you, that building a tree grammar is a overkill just to examine a few subtrees. You might want to simply print out all of the declarations defined in the tree. You don't care about the tree structure except for the declaration itself. A grammar for the full tree is a lot of unnecessary work.
TreeWizard is my first attempt at a utility class that lets you create and navigate trees without relying on an ANTLR grammar. You will find it attached with about 30 unit tests. This is only my development branch so we are free to update, change, and otherwise mangle the definition. I just want to make sure that we have something out there that people can try out to make sure I'm going in the right direction.
Creating trees
Creating a tree is a simple matter of supplying a pattern for the tree using the token names from an ANTLR grammar:
TreeAdaptor adaptor = new CommonTreeAdaptor();
String[] tokens = myparser.getTokenNames();
TreeWizard wiz = new TreeWizard(adaptor, tokens);
CommonTree t = (CommonTree)wiz.create("(A B C)");
where the tokens from your parser look something like:
public static final String[] tokenNames = new String[] {
"<invalid>", "<EOR>", "<DOWN>", "<UP>", "A", "B", "C"
};