util.ip: Fix protocol detection of IPv6 addresses beginning with : [Backported from...
[prosody.git] / util / xmppstream.lua
index 0f80742dc706d693dd4785e7738952a2b2a4dbd3..138c86b744832b4e45f98798e64caf30fac664b5 100644 (file)
@@ -6,7 +6,6 @@
 -- COPYING file in the source package for more information.
 --
 
-
 local lxp = require "lxp";
 local st = require "util.stanza";
 local stanza_mt = st.stanza_mt;
@@ -20,13 +19,20 @@ local setmetatable = setmetatable;
 
 -- COMPAT: w/LuaExpat 1.1.0
 local lxp_supports_doctype = pcall(lxp.new, { StartDoctypeDecl = false });
+local lxp_supports_xmldecl = pcall(lxp.new, { XmlDecl = false });
+local lxp_supports_bytecount = not not lxp.new({}).getcurrentbytecount;
+
+local default_stanza_size_limit = 1024*1024*10; -- 10MB
 
 module "xmppstream"
 
 local new_parser = lxp.new;
 
-local ns_prefixes = {
-       ["http://www.w3.org/XML/1998/namespace"] = "xml";
+local xml_namespace = {
+       ["http://www.w3.org/XML/1998/namespace\1lang"] = "xml:lang";
+       ["http://www.w3.org/XML/1998/namespace\1space"] = "xml:space";
+       ["http://www.w3.org/XML/1998/namespace\1base"] = "xml:base";
+       ["http://www.w3.org/XML/1998/namespace\1id"] = "xml:id";
 };
 
 local xmlns_streams = "http://etherx.jabber.org/streams";
@@ -37,13 +43,16 @@ local ns_pattern = "^([^"..ns_separator.."]*)"..ns_separator.."?(.*)$";
 _M.ns_separator = ns_separator;
 _M.ns_pattern = ns_pattern;
 
-function new_sax_handlers(session, stream_callbacks)
+local function dummy_cb() end
+
+function new_sax_handlers(session, stream_callbacks, cb_handleprogress)
        local xml_handlers = {};
        
        local cb_streamopened = stream_callbacks.streamopened;
        local cb_streamclosed = stream_callbacks.streamclosed;
-       local cb_error = stream_callbacks.error or function(session, e) error("XML stream error: "..tostring(e)); end;
+       local cb_error = stream_callbacks.error or function(session, e, stanza) error("XML stream error: "..tostring(e)..(stanza and ": "..tostring(stanza) or ""),2); end;
        local cb_handlestanza = stream_callbacks.handlestanza;
+       cb_handleprogress = cb_handleprogress or dummy_cb;
        
        local stream_ns = stream_callbacks.stream_ns or xmlns_streams;
        local stream_tag = stream_callbacks.stream_tag or "stream";
@@ -56,6 +65,7 @@ function new_sax_handlers(session, stream_callbacks)
        
        local stack = {};
        local chardata, stanza = {};
+       local stanza_size = 0;
        local non_streamns_depth = 0;
        function xml_handlers:StartElement(tagname, attr)
                if stanza and #chardata > 0 then
@@ -73,30 +83,33 @@ function new_sax_handlers(session, stream_callbacks)
                        non_streamns_depth = non_streamns_depth + 1;
                end
                
-               -- FIXME !!!!!
                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
+                       local xmlk = xml_namespace[k];
+                       if xmlk then
+                               attr[xmlk] = attr[k];
+                               attr[k] = nil;
                        end
                end
                
                if not stanza then --if we are not currently inside a stanza
+                       if lxp_supports_bytecount then
+                               stanza_size = self:getcurrentbytecount();
+                       end
                        if session.notopen then
                                if tagname == stream_tag then
                                        non_streamns_depth = 0;
                                        if cb_streamopened then
+                                               if lxp_supports_bytecount then
+                                                       cb_handleprogress(stanza_size);
+                                                       stanza_size = 0;
+                                               end
                                                cb_streamopened(session, attr);
                                        end
                                else
                                        -- Garbage before stream?
-                                       cb_error(session, "no-stream");
+                                       cb_error(session, "no-stream", tagname);
                                end
                                return;
                        end
@@ -106,6 +119,9 @@ function new_sax_handlers(session, stream_callbacks)
                        
                        stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
                else -- we are inside a stanza, so add a tag
+                       if lxp_supports_bytecount then
+                               stanza_size = stanza_size + self:getcurrentbytecount();
+                       end
                        t_insert(stack, stanza);
                        local oldstanza = stanza;
                        stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
@@ -113,12 +129,45 @@ function new_sax_handlers(session, stream_callbacks)
                        t_insert(oldstanza.tags, stanza);
                end
        end
+       if lxp_supports_xmldecl then
+               function xml_handlers:XmlDecl(version, encoding, standalone)
+                       if lxp_supports_bytecount then
+                               cb_handleprogress(self:getcurrentbytecount());
+                       end
+               end
+       end
+       function xml_handlers:StartCdataSection()
+               if lxp_supports_bytecount then
+                       if stanza then
+                               stanza_size = stanza_size + self:getcurrentbytecount();
+                       else
+                               cb_handleprogress(self:getcurrentbytecount());
+                       end
+               end
+       end
+       function xml_handlers:EndCdataSection()
+               if lxp_supports_bytecount then
+                       if stanza then
+                               stanza_size = stanza_size + self:getcurrentbytecount();
+                       else
+                               cb_handleprogress(self:getcurrentbytecount());
+                       end
+               end
+       end
        function xml_handlers:CharacterData(data)
                if stanza then
+                       if lxp_supports_bytecount then
+                               stanza_size = stanza_size + self:getcurrentbytecount();
+                       end
                        t_insert(chardata, data);
+               elseif lxp_supports_bytecount then
+                       cb_handleprogress(self:getcurrentbytecount());
                end
        end
        function xml_handlers:EndElement(tagname)
+               if lxp_supports_bytecount then
+                       stanza_size = stanza_size + self:getcurrentbytecount()
+               end
                if non_streamns_depth > 0 then
                        non_streamns_depth = non_streamns_depth - 1;
                end
@@ -130,6 +179,10 @@ function new_sax_handlers(session, stream_callbacks)
                        end
                        -- Complete stanza
                        if #stack == 0 then
+                               if lxp_supports_bytecount then
+                                       cb_handleprogress(stanza_size);
+                               end
+                               stanza_size = 0;
                                if tagname ~= stream_error_tag then
                                        cb_handlestanza(session, stanza);
                                else
@@ -140,19 +193,9 @@ function new_sax_handlers(session, stream_callbacks)
                                stanza = t_remove(stack);
                        end
                else
-                       if tagname == stream_tag then
-                               if cb_streamclosed then
-                                       cb_streamclosed(session);
-                               end
-                       else
-                               local curr_ns,name = tagname:match(ns_pattern);
-                               if name == "" then
-                                       curr_ns, name = "", curr_ns;
-                               end
-                               cb_error(session, "parse-error", "unexpected-element-close", name);
+                       if cb_streamclosed then
+                               cb_streamclosed(session);
                        end
-                       stanza, chardata = nil, {};
-                       stack = {};
                end
        end
 
@@ -170,7 +213,7 @@ function new_sax_handlers(session, stream_callbacks)
        xml_handlers.ProcessingInstruction = restricted_handler;
        
        local function reset()
-               stanza, chardata = nil, {};
+               stanza, chardata, stanza_size = nil, {}, 0;
                stack = {};
        end
        
@@ -181,19 +224,39 @@ function new_sax_handlers(session, stream_callbacks)
        return xml_handlers, { reset = reset, set_session = set_session };
 end
 
-function new(session, stream_callbacks)
-       local handlers, meta = new_sax_handlers(session, stream_callbacks);
-       local parser = new_parser(handlers, ns_separator);
+function new(session, stream_callbacks, stanza_size_limit)
+       -- Used to track parser progress (e.g. to enforce size limits)
+       local n_outstanding_bytes = 0;
+       local handle_progress;
+       if lxp_supports_bytecount then
+               function handle_progress(n_parsed_bytes)
+                       n_outstanding_bytes = n_outstanding_bytes - n_parsed_bytes;
+               end
+               stanza_size_limit = stanza_size_limit or default_stanza_size_limit;
+       elseif stanza_size_limit then
+               error("Stanza size limits are not supported on this version of LuaExpat")
+       end
+
+       local handlers, meta = new_sax_handlers(session, stream_callbacks, handle_progress);
+       local parser = new_parser(handlers, ns_separator, false);
        local parse = parser.parse;
 
        return {
                reset = function ()
-                       parser = new_parser(handlers, ns_separator);
+                       parser = new_parser(handlers, ns_separator, false);
                        parse = parser.parse;
+                       n_outstanding_bytes = 0;
                        meta.reset();
                end,
                feed = function (self, data)
-                       return parse(parser, data);
+                       if lxp_supports_bytecount then
+                               n_outstanding_bytes = n_outstanding_bytes + #data;
+                       end
+                       local ok, err = parse(parser, data);
+                       if lxp_supports_bytecount and n_outstanding_bytes > stanza_size_limit then
+                               return nil, "stanza-too-large";
+                       end
+                       return ok, err;
                end,
                set_session = meta.set_session;
        };