ParC is a recursive descent parser compiler, it's not finished but it's currently what I'm working on.
I was really exicted when Microsoft announced Oslo, I thought, finally! A generic tool I can use to play with text. But, they never got any further than CTP.
M was a nice language and the whole experience created with IntelliPad was actually nice to work with, you'd play interactivly with your DSL and grammar. This was great fun. Now, I find my self needing tools like this again becuase writing the lexer, parser and syntax tree your self just blows. It's just a lot of grunt work.
So, as I was writing that lexer, AGAIN, and I thought -- maybe I can solve this problem (at least to my own satisfaction) once and for all. Thus ParC was born.
What ParC will be able to do, once finished, is to take a grammar (similar to how you'd write grammar in M) and generate the lexer, parser and syntax tree (with a visitor template) in C++. You are then free to do what you will with the syntax tree.
Most tools I know of have taken the path of mixing the grammar and code and introduce what is sometimes referred to as semantic actions. ParC does not do this, it simply focuses on building that syntax tree. The benefit of going down this road is that ParC is really, language agnostic. While the run-time and compiler will be in C++ it's possible to target any language as long as you have a generator for that language (and there will be some plug-in architecture for this).
I personally like the idea of thinking of the parser compiler as a function that takes an input in the form of text and returns a structured data set. This is what I wish to do with ParC and every now and then I find myself wishing I had such a tool. To quickly be able to do something meaningful with just text.