make_lexer.cmlex

sml
name MakeLexer
alphabet 128 /* 1024 is maximum */

set digit = (range '0 '9)
set digit_start = (range '1 '9)

set lower = (range 'a 'z)
set upper = (range 'A 'Z)
set alpha = (or lower upper)
set alphanumeric = (or digit alpha)
set ident = (or '_ alphanumeric)

set whitechar = (or 32 9 10)  /* space, tab, lf */

function lex : t =
  '= => equal
  ': => colon
  '* => asterisk
  ', => comma
  '/ => slash

  '1 => one
  (+ digit) => number

  (seq '- '>) => rarrow
  (seq '< '-) => larrow
  '( => lparen
  ') => rparen
  '[ => lbrack
  '] => rbrack
  "(|" => lwhiteparen
  "|)" => rwhiteparen
  "<|" => lwhiteangle
  "|>" => rwhiteangle
  ":>" => seal
  (seq lower (* ident) (* '')) => lower_ident
  (seq upper (* ident) (* '')) => upper_ident
  (seq '' lower (* ident))     => quote_ident
  (seq '\ (* alphanumeric))    => backslash_ident
  '. => dot
  (+ whitechar) => whitespace

  epsilon => illegal
  eos => eof