The program is currently able to lex most useful tokens, and parse constructs associated with them on simple examples. Unit tests are still missing to formally ensure everything's right.