(************************************************************
   utf.ml		Created      : Fri Mar 14 23:15:14 2003
  			Last modified: Sat Mar 15 03:47:11 2003
  Compile: ocamlc -g utf.ml -o utf #
  FTP Directory: sources/ocaml #
************************************************************)
(**

  @author Takashi Masuyama <mamewo@dk9.so-net.ne.jp>

*)

let table_file_name = "JIS0208.TXT"
(*let make_table *)

let sjis_code_start = 0x8140 
let sjis_code_end = 0xEAA4 

let additional = 0x8080
let euc_code_start = 0x2121 + additional
let euc_code_end = 0x7426 + additional

let read_table_file s e input =
  let table = Array.make (e-s+1) 0 in
  let rec iter () =
    let line = input_line input in
    if line.[0] = '#' then
      iter ()
    else
      begin
	Scanf.sscanf line "%i\t%i\t%i%s"
	  (fun sjis code unicode hoge -> table.(code+additional-s) <- unicode);
	iter ()
      end in
  try
    iter ()
  with End_of_file ->
    table 

(*let read_euc_char input =*)
(*let data =*)
(*  let tmp = "Αύ" in*)
(*  ((Char.code tmp.[0]) lsl 8) lor (Char.code tmp.[1])*)

let pure_euc_to_utf8 code =
  let table = 
    read_table_file euc_code_start euc_code_end (open_in table_file_name) in
  let end_pos = String.length code in
  let rec iter res pos =
    if pos >= end_pos then
      res
    else
      let c1code = code.[pos] in
      let c1 = Char.code c1code in
      if c1 <= 0x7F then
	let result = String.make 1 '\000' in
	begin
	  result.[0] <- c1code;
	  iter (res^result) (pos+1)
	end
      else
      let code = 
	let c2 = Char.code code.[pos+1] in
	let index = (c1 lsl 8) lor c2 in
	table.(index - euc_code_start) in
      let new_result =
	res^(if code <= 0x7FF then
	  let result = String.make 2 '\000' in
	  let cl = (code land 0x3F) lor 0x80 in (* 6 bit *)
	  let ch = ((code lsr 6) land 0x1F) lor 0xC0 in
	  begin
	    result.[0] <- Char.chr ch;
	    result.[1] <- Char.chr cl;
	    result
	  end
	else if code <= 0xFFFF then
	  let result = String.make 3 '\000' in
	  begin
	    result.[2] <- Char.chr((code land 0x3F) lor 0x80); (* 6 bit *)
	    result.[1] <- Char.chr(((code lsr 6) land 0x3F) lor 0x80);
	    result.[0] <- Char.chr(((code lsr 12) land 0xF) lor 0xE0);
	    result;
	  end
	else 
	  raise Not_found) in
      iter new_result (pos+2) in
  iter "" 0

let rec encode_from_input input =
  let line = input_line input in
  try
    print_endline (pure_euc_to_utf8 line);
    encode_from_input input
  with End_of_file ->
    ()
    
let _ = 
  let input = open_in "../aki/aki.txt" in
  encode_from_input input;
  close_in input