From 0cba3cd26003a0b17465353ebb7a159d02ad7056 Mon Sep 17 00:00:00 2001
From: Waqas Hussain <waqas20@gmail.com>
Date: Tue, 23 Apr 2013 15:55:49 -0400
Subject: util.json: New, improved, fixed codepoint to UTF-8 conversion.

---
 util/json.lua | 37 +++++++++++--------------------------
 1 file changed, 11 insertions(+), 26 deletions(-)

(limited to 'util/json.lua')

diff --git a/util/json.lua b/util/json.lua
index e8de4d2d..9c2dd2c6 100644
--- a/util/json.lua
+++ b/util/json.lua
@@ -2,8 +2,6 @@
 -- Copyright (C) 2008-2010 Matthew Wild
 -- Copyright (C) 2008-2010 Waqas Hussain
 --
--- utf8char copyright (C) 2007 Rici Lake
---
 -- This project is MIT/X11 licensed. Please see the
 -- COPYING file in the source package for more information.
 --
@@ -41,32 +39,19 @@ for i=0,31 do
 	if not escapes[ch] then escapes[ch] = ("\\u%.4X"):format(i); end
 end
 
-local function utf8char(i)
-	if i >= 0 then
-		i = i - i%1
-		if i < 128 then
-			return s_char(i)
-		else
-			local c1 = i % 64
-			i = (i - c1) / 64
-			if i < 32 then
-				return s_char(0xC0+i, 0x80+c1)
-			else
-        			local c2 = i % 64
-        			i = (i - c2) / 64
-        			if i < 16 and (i ~= 13 or c2 < 32) then
-        				return s_char(0xE0+i, 0x80+c2, 0x80+c1)
-        			elseif i >= 16 and i < 0x110 then
-        				local c3 = i % 64
-        				i = (i - c3) / 64
-        				return s_char(0xF0+i, 0x80+c3, 0x80+c2, 0x80+c1)
-        			end
-			end
-		end
+local function codepoint_to_utf8(code)
+	if code < 0x80 then return s_char(code); end
+	local bits0_6 = code % 64;
+	if code < 0x800 then
+		local bits6_5 = (code - bits0_6) / 64;
+		return s_char(0x80 + 0x40 + bits6_5, 0x80 + bits0_6);
 	end
+	local bits0_12 = code % 4096;
+	local bits6_6 = (bits0_12 - bits0_6) / 64;
+	local bits12_4 = (code - bits0_12) / 4096;
+	return s_char(0x80 + 0x40 + 0x20 + bits12_4, 0x80 + bits6_6, 0x80 + bits0_6);
 end
 
-
 local valid_types = {
 	number  = true,
 	string  = true,
@@ -292,7 +277,7 @@ function json.decode(json)
 						if not ch:match("[0-9a-fA-F]") then error("invalid unicode escape sequence in string"); end
 						seq = seq..ch;
 					end
-					s = s..utf8char(tonumber(seq, 16));
+					s = s..codepoint_to_utf8(tonumber(seq, 16));
 					next();
 				else error("invalid escape sequence in string"); end
 			end
-- 
cgit v1.2.3