It cause corrupted strings literal (spaces inserted after double-quote start) and can't detect token validation errors after code format. It should not format source code when specific parse error found.