module Unicode where

-- | Like HBC's Char.decodeUTF8, but with more graceful failure.
decodeUTF8 :: String -> String
decodeUTF8 "" = ""
decodeUTF8 (c:cs) | c < '\x80' = c : decodeUTF8 cs
decodeUTF8 (c:c':cs) | '\xc0' <= c  && c  <= '\xdf' && 
		      '\x80' <= c' && c' <= '\xbf' =
	toEnum ((fromEnum c `mod` 0x20) * 0x40 + fromEnum c' `mod` 0x40) : decodeUTF8 cs
decodeUTF8 (c:c':c'':cs) | '\xe0' <= c   && c   <= '\xef' && 
		          '\x80' <= c'  && c'  <= '\xbf' &&
		          '\x80' <= c'' && c'' <= '\xbf' =
	toEnum ((fromEnum c `mod` 0x10 * 0x1000) + (fromEnum c' `mod` 0x40) * 0x40 + fromEnum c'' `mod` 0x40) : decodeUTF8 cs
--decodeUTF8 _ = error "UniChar.decodeUTF8: bad data"
decodeUTF8 (_:cs) = decodeUTF8 cs

encodeUTF8 :: String -> String
encodeUTF8 "" = ""
encodeUTF8 (c:cs) =
	if c > '\x0000' && c < '\x0080' then
	    c : encodeUTF8 cs
	else if c < toEnum 0x0800 then
	    let i = fromEnum c
	    in  toEnum (0xc0 + i `div` 0x40) : 
	        toEnum (0x80 + i `mod` 0x40) : 
		encodeUTF8 cs
	else
	    let i = fromEnum c
	    in  toEnum (0xe0 + i `div` 0x1000) : 
	        toEnum (0x80 + (i `mod` 0x1000) `div` 0x40) : 
		toEnum (0x80 + i `mod` 0x40) : 
		encodeUTF8 cs

-- | Compose selected decomposed Unicode characters
compose s =
  case s of
    []        -> []
    'A':'\778':s -> 'Å':compose s
    'A':'\776':s -> 'Ä':compose s
    'O':'\776':s -> 'Ö':compose s
    'E':'\769':s -> 'É':compose s
    'E':'\768':s -> 'È':compose s
    'U':'\776':s -> 'Ü':compose s
    'a':'\778':s -> 'å':compose s
    'a':'\776':s -> 'ä':compose s
    'o':'\776':s -> 'ö':compose s
    'e':'\769':s -> 'é':compose s
    'e':'\768':s -> 'è':compose s
    'u':'\776':s -> 'ü':compose s
    c:s       -> c  :compose s