util/xml.lua
author Jonas Schäfer <jonas@wielicki.name>
Mon, 10 Jan 2022 18:23:54 +0100
branch0.11
changeset 12185 783056b4e448
parent 8558 4f0f5b49bb03
child 12205 e5e0ab93d7f4
permissions -rw-r--r--
util.xml: Do not allow doctypes, comments or processing instructions Yes. This is as bad as it sounds. CVE pending. In Prosody itself, this only affects mod_websocket, which uses util.xml to parse the <open/> frame, thus allowing unauthenticated remote DoS using Billion Laughs. However, third-party modules using util.xml may also be affected by this. This commit installs handlers which disallow the use of doctype declarations and processing instructions without any escape hatch. It, by default, also introduces such a handler for comments, however, there is a way to enable comments nontheless. This is because util.xml is used to parse human-facing data, where comments are generally a desirable feature, and also because comments are generally harmless.


local st = require "util.stanza";
local lxp = require "lxp";
local t_insert = table.insert;
local t_remove = table.remove;
local error = error;

local _ENV = nil;
-- luacheck: std none

local parse_xml = (function()
	local ns_prefixes = {
		["http://www.w3.org/XML/1998/namespace"] = "xml";
	};
	local ns_separator = "\1";
	local ns_pattern = "^([^"..ns_separator.."]*)"..ns_separator.."?(.*)$";
	return function(xml, options)
		--luacheck: ignore 212/self
		local handler = {};
		local stanza = st.stanza("root");
		local namespaces = {};
		local prefixes = {};
		function handler:StartNamespaceDecl(prefix, url)
			if prefix ~= nil then
				t_insert(namespaces, url);
				t_insert(prefixes, prefix);
			end
		end
		function handler:EndNamespaceDecl(prefix)
			if prefix ~= nil then
				-- we depend on each StartNamespaceDecl having a paired EndNamespaceDecl
				t_remove(namespaces);
				t_remove(prefixes);
			end
		end
		function handler:StartElement(tagname, attr)
			local curr_ns,name = tagname:match(ns_pattern);
			if name == "" then
				curr_ns, name = "", curr_ns;
			end
			if curr_ns ~= "" then
				attr.xmlns = curr_ns;
			end
			for i=1,#attr do
				local k = attr[i];
				attr[i] = nil;
				local ns, nm = k:match(ns_pattern);
				if nm ~= "" then
					ns = ns_prefixes[ns];
					if ns then
						attr[ns..":"..nm] = attr[k];
						attr[k] = nil;
					end
				end
			end
			local n = {}
			for i=1,#namespaces do
				n[prefixes[i]] = namespaces[i];
			end
			stanza:tag(name, attr, n);
		end
		function handler:CharacterData(data)
			stanza:text(data);
		end
		function handler:EndElement()
			stanza:up();
		end
		local parser;
		-- SECURITY: These two handlers, especially the Doctype one, are required to prevent exploits such as Billion Laughs.
		function handler:StartDoctypeDecl()
			if not parser.stop or not parser:stop() then
				error("Failed to abort parsing");
			end
		end
		function handler:ProcessingInstruction()
			if not parser.stop or not parser:stop() then
				error("Failed to abort parsing");
			end
		end
		if not options or not options.allow_comments then
			-- NOTE: comments are generally harmless and can be useful when parsing configuration files or other data, even user-provided data
			function handler:Comment()
				if not parser.stop or not parser:stop() then
					error("Failed to abort parsing");
				end
			end
		end
		parser = lxp.new(handler, ns_separator);
		local ok, err, line, col = parser:parse(xml);
		if ok then ok, err, line, col = parser:parse(); end
		--parser:close();
		if ok then
			return stanza.tags[1];
		else
			return ok, err.." (line "..line..", col "..col..")";
		end
	end;
end)();

return {
	parse = parse_xml;
};