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