@@ -1,8 +1,18 @@
 -- [ʞ] cortav.lua
 --  ~ lexi hale <lexi@hale.su>
 --  © AGPLv3
 --  ? reference implementation of the cortav document language
+--
+--  ! TODO refactor encoding logic. it's a complete
+--         mess and i seem to have repeatedly gotten
+--         confused about how it's supposed to work.
+--         the whole shitshow needs to be replaced
+--         with a clean, simple paradigm: documents
+--         are translated to UTF8 on the way in, and
+--         translate back out on the way out. trying
+--         to cope with multiple simultaneous
+--         encodings in memory is a disaster zone.
 
 local ss = require 'sirsem'
 -- aliases for commonly used sirsem funcs
 local startswith = ss.str.begins
@@ -735,8 +745,15 @@
 				origin = o;
 			}};
 			origin = o;
 		}
+	end
+	local function unicodepoint(s,c)
+		local cp = tonumber(s, 16)
+		return {
+			kind = 'codepoint';
+			code = cp;
+		}
 	end
 	ct.spanctls = {
 		{seq = '!', parse = formatter 'emph'};
 		{seq = '*', parse = formatter 'strong'};
@@ -797,8 +814,14 @@
 		{seq = '→', parse = insert_link};
 		{seq = '🔗', parse = insert_link};
 		{seq = '##', parse = insert_var_ref(true)};
 		{seq = '#', parse = insert_var_ref(false)};
+
+		{seq = 'U+', parse = unicodepoint};
+		{seq = 'u+', parse = unicodepoint};
+		{seq = 'U',  parse = unicodepoint};
+		{seq = 'u',  parse = unicodepoint};
+
 		{seq = '%%', parse = function (s,c)
 			local com = s:match '^%%%%%s*(.*)$'
 			return {
 				kind = 'comment';