#!Lua-5.0.exe
-- Convert accented characters from a given string and return HTML entities in the output string.
-- by Philippe Lhoste <PhiLho(a)GMX.net> http://Phi.Lho.free.fr
-- v. 2.0 -- 2003/06/10 -- Better algorithm, using regular expression
-- v. 1.0 -- 2003/04/19 -- Naive implementation

local entities =
{
--	['&'] = "&amp;",
	['<'] = "&lt;",
	['>'] = "&gt;",
	-- French entities (the most common ones)
	['à'] = "&agrave;",
	['â'] = "&acirc;",
	['é'] = "&eacute;",
	['è'] = "&egrave;",
	['ê'] = "&ecirc;",
	['ë'] = "&euml;",
	['î'] = "&icirc;",
	['ï'] = "&iuml;",
	['ô'] = "&ocirc;",
	['ö'] = "&ouml;",
	['ù'] = "&ugrave;",
	['û'] = "&ucirc;",
	['ÿ'] = "&yuml;",
	['À'] = "&Agrave;",
	['Â'] = "&Acirc;",
	['É'] = "&Eacute;",
	['È'] = "&Egrave;",
	['Ê'] = "&Ecirc;",
	['Ë'] = "&Euml;",
	['Î'] = "&Icirc;",
	['Ï'] = "&Iuml;",
	['Ô'] = "&Ocirc;",
	['Ö'] = "&Ouml;",
	['Ù'] = "&Ugrave;",
	['Û'] = "&Ucirc;",
	['ç'] = "&ccedil;",
	['Ç'] = "&Ccedil;",
	['Ÿ'] = "&Yuml;",
	['«'] = "&laquo;",
	['»'] = "&raquo;",
	['©'] = "&copy;",
	['®'] = "&reg;",
	['æ'] = "&aelig;",
	['Æ'] = "&AElig;",
	['Œ'] = "&OElig;", -- Not understood by all browsers
	['œ'] = "&oelig;", -- Not understood by all browsers
}

function EncodeEntities1(toEncode)
	if toEncode == nil or type(toEncode) ~= "string" then
		return ''
	end

	local EncodeHighAscii = function (char)
		local code = string.byte(char)
		if code > 127 then
			return string.format("&#%d;", code)
		else
			return char
		end
	end

	local encodedString = toEncode
	-- First encode '&' char, to avoid re-encoding already encoded chars
	encodedString = string.gsub(encodedString, '&', "&amp;")
	-- Encode known entities
	for char, entity in entities do
		encodedString = string.gsub(encodedString, char, entity)
	end
	-- Encode unknown high Ascii characters to numerical entities
	encodedString = string.gsub(encodedString, '(.)', EncodeHighAscii)
	return encodedString
end

function EncodeEntities2(toEncode)
	if toEncode == nil or type(toEncode) ~= "string" then
		return ''
	end

	local EncodeToEntities = function (char)
		return entities[char] or char
	end

	local EncodeHighAscii = function (char)
		local code = string.byte(char)
		if code > 127 then
			return string.format("&#%d;", code)
		else
			return char
		end
	end

	local encodedString = toEncode
	local encodingString = "(["
	entities['&'] = "&amp;"
	-- Add all characters to encode to the encodingString
	for char, entity in entities do
		encodingString = encodingString .. char
	end
	encodingString = encodingString .. "])"
	-- Encode known characters to entities
	encodedString = string.gsub(encodedString, encodingString, EncodeToEntities)
	-- Encode unknown high Ascii characters to numerical entities
	encodedString = string.gsub(encodedString, '(.)', EncodeHighAscii)
	return encodedString
end

function EncodeEntities3(toEncode)
	if toEncode == nil or type(toEncode) ~= "string" then
		return ''
	end

	local EncodeToEntities = function (char)
		local entity = entities[char]
		if entity == nil then
			local code = string.byte(char)
			if code > 127 then
				entity = string.format("&#%d;", code)
			end
		end
		return entity or char
	end

	entities['&'] = "&amp;"
	-- I will replace '(.)' with '([^%c%s%w%p])'
	encodedString = string.gsub(toEncode, '(.)', EncodeToEntities)
	return encodedString
end