Modul:UrlCheck

Die Dokumentation für dieses Modul kann unter Modul:UrlCheck/doc erstellt werden

local uc = {}

function uc.isUrl( url )
	-- return codes 0 and 1 reserved
	if ( url == nil ) or ( type( url ) ~= 'string' ) then return 2 end

	local s = mw.text.trim( url )
	if ( s == '' ) or ( #s > 2048 ) then return 3 end
		-- limitation because of search engines or IE
	if s:find( '%s' ) or s:find( '%c' ) or s:find( '%.%.' ) or s:find( '%.@' )
		or s:find( '@[%.%-]' ) or s:find( '%-%.' ) or s:find( '%.%-' )
		or s:find( '%./' ) or s:find( '/%.' ) or s:match( '^%.' )
		then return 4 end

	-- https://max:muster@www.example.com:8080/index.html?p1=A&p2=B#ressource

	-- protocol
	local count
	s,count = s:gsub( '^http://', '' )
	if count == 0 then
		s,count = s:gsub( '^https://', '' )
		if count == 0 then
			s,count = s:gsub( '^//', '' )
		end
	end
	if count == 0 then return 5 end -- missing or wrong protocol

	local user = ''
	local password = ''
	local host = ''
	local port = ''
	local path = ''
	local topLevel = ''
	local at

	-- user and password
	_,count = s:gsub( '@', '@' )
	if count > 1 then return 6 end
	if count == 1 then
		at = s:find( '@' )
		user = s:sub( 1, at - 1 )
		s = s:sub( at + 1, #s )
		if not user or not s then return 7 end

		_,count = user:gsub( ':', ':' )
		if count > 1 then return 8 end
		if count == 1 then
			at = user:find( ':' )
			password = user:sub( at + 1, #user )
			user = user:sub( 1, at - 1 )
			if not user or not password then return 9 end
			if #user > 64 then return 10 end
		end
	end

	-- host and path
	at = s:find( '/' )
	if at ~= nil then
		host = s:sub( 1, at - 1 )
		path = s:sub( at + 1, #s )
		if not host then return 11 end
	else
		host = s
		if host == '' then return 11 end
	end

	-- future: add path check
	-- split at '/', last part: t == mw.uri.encode( t ) ?

	-- host and port
	_,count = host:gsub( ':', ':' )
	if count > 1 then return 13 end
	if count == 1 then
		at = host:find( ':' )
		port = host:sub( at + 1, #host )
		host = host:sub( 1, at - 1 )
		if not host or not port then return 11 end
		if ( port:match( '^[1-9]%d*$' ) == nil ) or
			( tonumber( port ) > 65535 ) then return 14 end
	end

	-- handle host part
	if #host > 253 then return 15 end

	-- getting top-level domain
	at = host:match( '^.*()%.' ) -- find last dot
	if at == nil then return 16 end
	topLevel = host:sub( at + 1, #host )
	if not topLevel then return 17 end

	-- future: check of top-level domain

	if not mw.ustring.match( host, '^[%w%.%-]+%.%a%a+$' ) then return 20 end
	if not host:match( '^[%w%.%-]+%.%a%a+$' ) then
		return 1 -- matches domain only in UTF 8 mode
	end

	return 0
end

function uc.checkUrl( frame )
	local args = frame.args
	return uc.isUrl( args.url or '' )
end

return uc