Regular languages and their generating functions: The inverse problem