Java nodes now have a method getText(), which provides access to the full text of the node. This is mostly intended as a debugging aid, and not as a way of writing rules (hence the @NoAttribute).

This implementation uses:

• a special token class, that is based on a start/end offset instead of lines and columns. This is more economic memory-wise, and allows us to cut a piece of the file text easily.
  • one downside of this representation is that getBeginLine/getEndline for nodes is less efficient. I don't think this will be a real problem though
• a special CharStream that records text offsets of tokens

Both of these classes are added to pmd-core instead of generated by Javacc then hacked through Ant search/replace. I think, most of our javacc parser implementations could use the same token implementations and get the feature for nearly free. It's also nice to remove the duplication of the Token class (currently one per javacc module).

Future work:

• Refactor and use the Document API that was introduced and abandoned in pmd-core, to represent text regions of a document.
• Make a separate AbstractNode class, that doesn't have those beginLine, endLine, etc. fields. These can be fetched from the tokens directly
• Share this feature with other language modules