(* Example of writing a lexing function by hand.
 * 
 * The purpose of this file is mainly to illustrate that it's a bit
 * annoying and tedious to have to write lexers by hand. *)

(* Define the datatype of tokens *)
type token =
  | Int of int32
  | Ident of string
  | OP_LPAREN
  | KW_IF

(* This exception is raised when an unanticipated character is read *)
exception Lex_error of char



let is_character c = let i = Char.code c in 
  ((65 <= i)  && (i <= 90)) || ((97 <= i) || (i <= 122))

let is_digit c = let i = Char.code c in
  (48 <= i) && (i <= 57)

let is_underscore c = c = '_'

(* Reads an identifier from an input channel *)
(* An identifier matches the regular expression:
     character (digit | '_' | character)*  *)
(* Returns the token (Ident "foo") *)

(* Note: one problem with this approach is how to 
 * implement "lookahead" -- this example doesn't do a 
 * good job of that. *)
let rec lex_ident (input:in_channel) =
  let buf = Buffer.create 16 in
  let c = input_char input in
  let rec loop () =
    try 
      let c = input_char input in
	if (is_character c || is_digit c || is_underscore c) then
	  (Buffer.add_char buf c; loop () )
	else
	  Ident (Buffer.contents buf)
    with 
	End_of_file -> Ident(Buffer.contents buf)
  in if is_character c then (Buffer.add_char buf c; loop ()) 
    else raise (Lex_error c)
;;
try 
  (let x = lex_ident stdin in
    match x with
      | Ident s -> print_string s
      | _ -> ())
with
    Lex_error c -> print_char c