util/xmppstream.lua

   1 -- Prosody IM
   2 -- Copyright (C) 2008-2010 Matthew Wild
   3 -- Copyright (C) 2008-2010 Waqas Hussain
   4 --
   5 -- This project is MIT/X11 licensed. Please see the
   6 -- COPYING file in the source package for more information.
   7 --
   8
   9 local lxp = require "lxp";
  10 local st = require "util.stanza";
  11 local stanza_mt = st.stanza_mt;
  12
  13 local error = error;
  14 local tostring = tostring;
  15 local t_insert = table.insert;
  16 local t_concat = table.concat;
  17 local t_remove = table.remove;
  18 local setmetatable = setmetatable;
  19
  20 -- COMPAT: w/LuaExpat 1.1.0
  21 local lxp_supports_doctype = pcall(lxp.new, { StartDoctypeDecl = false });
  22 local lxp_supports_xmldecl = pcall(lxp.new, { XmlDecl = false });
  23 local lxp_supports_bytecount = not not lxp.new({}).getcurrentbytecount;
  24
  25 local default_stanza_size_limit = 1024*1024*10; -- 10MB
  26
  27 local _ENV = nil;
  28
  29 local new_parser = lxp.new;
  30
  31 local xml_namespace = {
  32         ["http://www.w3.org/XML/1998/namespace\1lang"] = "xml:lang";
  33         ["http://www.w3.org/XML/1998/namespace\1space"] = "xml:space";
  34         ["http://www.w3.org/XML/1998/namespace\1base"] = "xml:base";
  35         ["http://www.w3.org/XML/1998/namespace\1id"] = "xml:id";
  36 };
  37
  38 local xmlns_streams = "http://etherx.jabber.org/streams";
  39
  40 local ns_separator = "\1";
  41 local ns_pattern = "^([^"..ns_separator.."]*)"..ns_separator.."?(.*)$";
  42
  43 local function dummy_cb() end
  44
  45 local function new_sax_handlers(session, stream_callbacks, cb_handleprogress)
  46         local xml_handlers = {};
  47
  48         local cb_streamopened = stream_callbacks.streamopened;
  49         local cb_streamclosed = stream_callbacks.streamclosed;
  50         local cb_error = stream_callbacks.error or function(session, e, stanza) error("XML stream error: "..tostring(e)..(stanza and ": "..tostring(stanza) or ""),2); end;
  51         local cb_handlestanza = stream_callbacks.handlestanza;
  52         cb_handleprogress = cb_handleprogress or dummy_cb;
  53
  54         local stream_ns = stream_callbacks.stream_ns or xmlns_streams;
  55         local stream_tag = stream_callbacks.stream_tag or "stream";
  56         if stream_ns ~= "" then
  57                 stream_tag = stream_ns..ns_separator..stream_tag;
  58         end
  59         local stream_error_tag = stream_ns..ns_separator..(stream_callbacks.error_tag or "error");
  60
  61         local stream_default_ns = stream_callbacks.default_ns;
  62
  63         local stack = {};
  64         local chardata, stanza = {};
  65         local stanza_size = 0;
  66         local non_streamns_depth = 0;
  67         function xml_handlers:StartElement(tagname, attr)
  68                 if stanza and #chardata > 0 then
  69                         -- We have some character data in the buffer
  70                         t_insert(stanza, t_concat(chardata));
  71                         chardata = {};
  72                 end
  73                 local curr_ns,name = tagname:match(ns_pattern);
  74                 if name == "" then
  75                         curr_ns, name = "", curr_ns;
  76                 end
  77
  78                 if curr_ns ~= stream_default_ns or non_streamns_depth > 0 then
  79                         attr.xmlns = curr_ns;
  80                         non_streamns_depth = non_streamns_depth + 1;
  81                 end
  82
  83                 for i=1,#attr do
  84                         local k = attr[i];
  85                         attr[i] = nil;
  86                         local xmlk = xml_namespace[k];
  87                         if xmlk then
  88                                 attr[xmlk] = attr[k];
  89                                 attr[k] = nil;
  90                         end
  91                 end
  92
  93                 if not stanza then --if we are not currently inside a stanza
  94                         if lxp_supports_bytecount then
  95                                 stanza_size = self:getcurrentbytecount();
  96                         end
  97                         if session.notopen then
  98                                 if tagname == stream_tag then
  99                                         non_streamns_depth = 0;
 100                                         if cb_streamopened then
 101                                                 if lxp_supports_bytecount then
 102                                                         cb_handleprogress(stanza_size);
 103                                                         stanza_size = 0;
 104                                                 end
 105                                                 cb_streamopened(session, attr);
 106                                         end
 107                                 else
 108                                         -- Garbage before stream?
 109                                         cb_error(session, "no-stream", tagname);
 110                                 end
 111                                 return;
 112                         end
 113                         if curr_ns == "jabber:client" and name ~= "iq" and name ~= "presence" and name ~= "message" then
 114                                 cb_error(session, "invalid-top-level-element");
 115                         end
 116
 117                         stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
 118                 else -- we are inside a stanza, so add a tag
 119                         if lxp_supports_bytecount then
 120                                 stanza_size = stanza_size + self:getcurrentbytecount();
 121                         end
 122                         t_insert(stack, stanza);
 123                         local oldstanza = stanza;
 124                         stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
 125                         t_insert(oldstanza, stanza);
 126                         t_insert(oldstanza.tags, stanza);
 127                 end
 128         end
 129         if lxp_supports_xmldecl then
 130                 function xml_handlers:XmlDecl(version, encoding, standalone)
 131                         if lxp_supports_bytecount then
 132                                 cb_handleprogress(self:getcurrentbytecount());
 133                         end
 134                 end
 135         end
 136         function xml_handlers:StartCdataSection()
 137                 if lxp_supports_bytecount then
 138                         if stanza then
 139                                 stanza_size = stanza_size + self:getcurrentbytecount();
 140                         else
 141                                 cb_handleprogress(self:getcurrentbytecount());
 142                         end
 143                 end
 144         end
 145         function xml_handlers:EndCdataSection()
 146                 if lxp_supports_bytecount then
 147                         if stanza then
 148                                 stanza_size = stanza_size + self:getcurrentbytecount();
 149                         else
 150                                 cb_handleprogress(self:getcurrentbytecount());
 151                         end
 152                 end
 153         end
 154         function xml_handlers:CharacterData(data)
 155                 if stanza then
 156                         if lxp_supports_bytecount then
 157                                 stanza_size = stanza_size + self:getcurrentbytecount();
 158                         end
 159                         t_insert(chardata, data);
 160                 elseif lxp_supports_bytecount then
 161                         cb_handleprogress(self:getcurrentbytecount());
 162                 end
 163         end
 164         function xml_handlers:EndElement(tagname)
 165                 if lxp_supports_bytecount then
 166                         stanza_size = stanza_size + self:getcurrentbytecount()
 167                 end
 168                 if non_streamns_depth > 0 then
 169                         non_streamns_depth = non_streamns_depth - 1;
 170                 end
 171                 if stanza then
 172                         if #chardata > 0 then
 173                                 -- We have some character data in the buffer
 174                                 t_insert(stanza, t_concat(chardata));
 175                                 chardata = {};
 176                         end
 177                         -- Complete stanza
 178                         if #stack == 0 then
 179                                 if lxp_supports_bytecount then
 180                                         cb_handleprogress(stanza_size);
 181                                 end
 182                                 stanza_size = 0;
 183                                 if tagname ~= stream_error_tag then
 184                                         cb_handlestanza(session, stanza);
 185                                 else
 186                                         cb_error(session, "stream-error", stanza);
 187                                 end
 188                                 stanza = nil;
 189                         else
 190                                 stanza = t_remove(stack);
 191                         end
 192                 else
 193                         if cb_streamclosed then
 194                                 cb_streamclosed(session);
 195                         end
 196                 end
 197         end
 198
 199         local function restricted_handler(parser)
 200                 cb_error(session, "parse-error", "restricted-xml", "Restricted XML, see RFC 6120 section 11.1.");
 201                 if not parser.stop or not parser:stop() then
 202                         error("Failed to abort parsing");
 203                 end
 204         end
 205
 206         if lxp_supports_doctype then
 207                 xml_handlers.StartDoctypeDecl = restricted_handler;
 208         end
 209         xml_handlers.Comment = restricted_handler;
 210         xml_handlers.ProcessingInstruction = restricted_handler;
 211
 212         local function reset()
 213                 stanza, chardata, stanza_size = nil, {}, 0;
 214                 stack = {};
 215         end
 216
 217         local function set_session(stream, new_session)
 218                 session = new_session;
 219         end
 220
 221         return xml_handlers, { reset = reset, set_session = set_session };
 222 end
 223
 224 local function new(session, stream_callbacks, stanza_size_limit)
 225         -- Used to track parser progress (e.g. to enforce size limits)
 226         local n_outstanding_bytes = 0;
 227         local handle_progress;
 228         if lxp_supports_bytecount then
 229                 function handle_progress(n_parsed_bytes)
 230                         n_outstanding_bytes = n_outstanding_bytes - n_parsed_bytes;
 231                 end
 232                 stanza_size_limit = stanza_size_limit or default_stanza_size_limit;
 233         elseif stanza_size_limit then
 234                 error("Stanza size limits are not supported on this version of LuaExpat")
 235         end
 236
 237         local handlers, meta = new_sax_handlers(session, stream_callbacks, handle_progress);
 238         local parser = new_parser(handlers, ns_separator, false);
 239         local parse = parser.parse;
 240
 241         function session.open_stream(session, from, to)
 242                 local send = session.sends2s or session.send;
 243
 244                 local attr = {
 245                         ["xmlns:stream"] = "http://etherx.jabber.org/streams",
 246                         ["xml:lang"] = "en",
 247                         xmlns = stream_callbacks.default_ns,
 248                         version = session.version and (session.version > 0 and "1.0" or nil),
 249                         id = session.streamid,
 250                         from = from or session.host, to = to,
 251                 };
 252                 if session.stream_attrs then
 253                         session:stream_attrs(from, to, attr)
 254                 end
 255                 send("<?xml version='1.0'?>");
 256                 send(st.stanza("stream:stream", attr):top_tag());
 257                 return true;
 258         end
 259
 260         return {
 261                 reset = function ()
 262                         parser = new_parser(handlers, ns_separator, false);
 263                         parse = parser.parse;
 264                         n_outstanding_bytes = 0;
 265                         meta.reset();
 266                 end,
 267                 feed = function (self, data)
 268                         if lxp_supports_bytecount then
 269                                 n_outstanding_bytes = n_outstanding_bytes + #data;
 270                         end
 271                         local ok, err = parse(parser, data);
 272                         if lxp_supports_bytecount and n_outstanding_bytes > stanza_size_limit then
 273                                 return nil, "stanza-too-large";
 274                         end
 275                         return ok, err;
 276                 end,
 277                 set_session = meta.set_session;
 278         };
 279 end
 280
 281 return {
 282         ns_separator = ns_separator;
 283         ns_pattern = ns_pattern;
 284         new_sax_handlers = new_sax_handlers;
 285         new = new;
 286 };