* parse HTML input, instead of linebased processing * optionally use variables as input and output instead of files