lua UTF16 BE lib

utf16.lua

local utf16 = {}

-- Big Endian

function utf16.toutf8(s)
	local surrogate
	return (s:gsub("..", function(utf16)
		local cp = string.unpack(">H", utf16)
		if (cp & 0xFC00) == 0xD800 then
			surrogate = cp
			return ""
		else
			if surrogate then
				cp = ((surrogate - 0xD800) << 10) + (cp - 0xDC00) + 0x10000
				surrogate = nil
			end
			return utf8.char(cp)
		end
	end))
end

function utf16.fromutf8(s)
	return (s:gsub(utf8.charpattern, function (u8)
		local cp = utf8.codepoint(u8)
		if cp <= 0xffff then
			return string.pack(">H", cp)
		else
			cp = cp - 0x10000
			local surrogate = (cp >> 10) + 0xD800
			local lo = (cp & 0x3ff) + 0xDC00
			return string.pack(">HH", surrogate, lo)
		end
	end))
end

--local s = utf16.fromutf8("😀")
--assert(utf16.toutf8(s) == "😀")

return utf16