a UNIXy program that can retrieve a schema for most file types ( some files like mp3 can't have a schema, there's some dynamic tricks in the format IIRC ) and displays it as an API. Then you query it using xpath/css:selector syntax.
That would solve the parser library problem of new languages*, and improve education.
*: well .. it would still be string all the way down, which I despise. 'the hell with strings, it should output json/yaml, it's a decent newer common pivot for atomic types and collections.
ps: there's nothing else but this rant embryo here. Sorry for the phising name
