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 -- luacheck: std none
  29
  30 local new_parser = lxp.new;
  31
  32 local xml_namespace = {
  33         ["http://www.w3.org/XML/1998/namespace\1lang"] = "xml:lang";
  34         ["http://www.w3.org/XML/1998/namespace\1space"] = "xml:space";
  35         ["http://www.w3.org/XML/1998/namespace\1base"] = "xml:base";
  36         ["http://www.w3.org/XML/1998/namespace\1id"] = "xml:id";
  37 };
  38
  39 local xmlns_streams = "http://etherx.jabber.org/streams";
  40
  41 local ns_separator = "\1";
  42 local ns_pattern = "^([^"..ns_separator.."]*)"..ns_separator.."?(.*)$";
  43
  44 local function dummy_cb() end
  45
  46 local function new_sax_handlers(session, stream_callbacks, cb_handleprogress)
  47         local xml_handlers = {};
  48
  49         local cb_streamopened = stream_callbacks.streamopened;
  50         local cb_streamclosed = stream_callbacks.streamclosed;
  51         local cb_error = stream_callbacks.error or
  52                 function(_, e, stanza)
  53                         error("XML stream error: "..tostring(e)..(stanza and ": "..tostring(stanza) or ""),2);
  54                 end;
  55         local cb_handlestanza = stream_callbacks.handlestanza;
  56         cb_handleprogress = cb_handleprogress or dummy_cb;
  57
  58         local stream_ns = stream_callbacks.stream_ns or xmlns_streams;
  59         local stream_tag = stream_callbacks.stream_tag or "stream";
  60         if stream_ns ~= "" then
  61                 stream_tag = stream_ns..ns_separator..stream_tag;
  62         end
  63         local stream_error_tag = stream_ns..ns_separator..(stream_callbacks.error_tag or "error");
  64
  65         local stream_default_ns = stream_callbacks.default_ns;
  66
  67         local stream_lang = "en";
  68
  69         local stack = {};
  70         local chardata, stanza = {};
  71         local stanza_size = 0;
  72         local non_streamns_depth = 0;
  73         function xml_handlers:StartElement(tagname, attr)
  74                 if stanza and #chardata > 0 then
  75                         -- We have some character data in the buffer
  76                         t_insert(stanza, t_concat(chardata));
  77                         chardata = {};
  78                 end
  79                 local curr_ns,name = tagname:match(ns_pattern);
  80                 if name == "" then
  81                         curr_ns, name = "", curr_ns;
  82                 end
  83
  84                 if curr_ns ~= stream_default_ns or non_streamns_depth > 0 then
  85                         attr.xmlns = curr_ns;
  86                         non_streamns_depth = non_streamns_depth + 1;
  87                 end
  88
  89                 for i=1,#attr do
  90                         local k = attr[i];
  91                         attr[i] = nil;
  92                         local xmlk = xml_namespace[k];
  93                         if xmlk then
  94                                 attr[xmlk] = attr[k];
  95                                 attr[k] = nil;
  96                         end
  97                 end
  98
  99                 if not stanza then --if we are not currently inside a stanza
 100                         if lxp_supports_bytecount then
 101                                 stanza_size = self:getcurrentbytecount();
 102                         end
 103                         if session.notopen then
 104                                 if tagname == stream_tag then
 105                                         non_streamns_depth = 0;
 106                                         stream_lang = attr["xml:lang"] or stream_lang;
 107                                         if cb_streamopened then
 108                                                 if lxp_supports_bytecount then
 109                                                         cb_handleprogress(stanza_size);
 110                                                         stanza_size = 0;
 111                                                 end
 112                                                 cb_streamopened(session, attr);
 113                                         end
 114                                 else
 115                                         -- Garbage before stream?
 116                                         cb_error(session, "no-stream", tagname);
 117                                 end
 118                                 return;
 119                         end
 120                         if curr_ns == "jabber:client" and name ~= "iq" and name ~= "presence" and name ~= "message" then
 121                                 cb_error(session, "invalid-top-level-element");
 122                         end
 123
 124                         stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
 125                 else -- we are inside a stanza, so add a tag
 126                         if lxp_supports_bytecount then
 127                                 stanza_size = stanza_size + self:getcurrentbytecount();
 128                         end
 129                         t_insert(stack, stanza);
 130                         local oldstanza = stanza;
 131                         stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
 132                         t_insert(oldstanza, stanza);
 133                         t_insert(oldstanza.tags, stanza);
 134                 end
 135         end
 136
 137         function xml_handlers:StartCdataSection()
 138                 if lxp_supports_bytecount then
 139                         if stanza then
 140                                 stanza_size = stanza_size + self:getcurrentbytecount();
 141                         else
 142                                 cb_handleprogress(self:getcurrentbytecount());
 143                         end
 144                 end
 145         end
 146         function xml_handlers:EndCdataSection()
 147                 if lxp_supports_bytecount then
 148                         if stanza then
 149                                 stanza_size = stanza_size + self:getcurrentbytecount();
 150                         else
 151                                 cb_handleprogress(self:getcurrentbytecount());
 152                         end
 153                 end
 154         end
 155         function xml_handlers:CharacterData(data)
 156                 if stanza then
 157                         if lxp_supports_bytecount then
 158                                 stanza_size = stanza_size + self:getcurrentbytecount();
 159                         end
 160                         t_insert(chardata, data);
 161                 elseif lxp_supports_bytecount then
 162                         cb_handleprogress(self:getcurrentbytecount());
 163                 end
 164         end
 165         function xml_handlers:EndElement(tagname)
 166                 if lxp_supports_bytecount then
 167                         stanza_size = stanza_size + self:getcurrentbytecount()
 168                 end
 169                 if non_streamns_depth > 0 then
 170                         non_streamns_depth = non_streamns_depth - 1;
 171                 end
 172                 if stanza then
 173                         if #chardata > 0 then
 174                                 -- We have some character data in the buffer
 175                                 t_insert(stanza, t_concat(chardata));
 176                                 chardata = {};
 177                         end
 178                         -- Complete stanza
 179                         if #stack == 0 then
 180                                 if lxp_supports_bytecount then
 181                                         cb_handleprogress(stanza_size);
 182                                 end
 183                                 stanza_size = 0;
 184                                 if stanza.attr["xml:lang"] == nil then
 185                                         stanza.attr["xml:lang"] = stream_lang;
 186                                 end
 187                                 if tagname ~= stream_error_tag then
 188                                         cb_handlestanza(session, stanza);
 189                                 else
 190                                         cb_error(session, "stream-error", stanza);
 191                                 end
 192                                 stanza = nil;
 193                         else
 194                                 stanza = t_remove(stack);
 195                         end
 196                 else
 197                         if cb_streamclosed then
 198                                 cb_streamclosed(session);
 199                         end
 200                 end
 201         end
 202
 203         local function restricted_handler(parser)
 204                 cb_error(session, "parse-error", "restricted-xml", "Restricted XML, see RFC 6120 section 11.1.");
 205                 if not parser.stop or not parser:stop() then
 206                         error("Failed to abort parsing");
 207                 end
 208         end
 209
 210         if lxp_supports_xmldecl then
 211                 function xml_handlers:XmlDecl(version, encoding, standalone)
 212                         if lxp_supports_bytecount then
 213                                 cb_handleprogress(self:getcurrentbytecount());
 214                         end
 215                         if (encoding and encoding:lower() ~= "utf-8")
 216                         or (standalone == "no")
 217                         or (version and version ~= "1.0") then
 218                                 return restricted_handler(self);
 219                         end
 220                 end
 221         end
 222         if lxp_supports_doctype then
 223                 xml_handlers.StartDoctypeDecl = restricted_handler;
 224         end
 225         xml_handlers.Comment = restricted_handler;
 226         xml_handlers.ProcessingInstruction = restricted_handler;
 227
 228         local function reset()
 229                 stanza, chardata, stanza_size = nil, {}, 0;
 230                 stack = {};
 231         end
 232
 233         local function set_session(stream, new_session) -- luacheck: ignore 212/stream
 234                 session = new_session;
 235         end
 236
 237         return xml_handlers, { reset = reset, set_session = set_session };
 238 end
 239
 240 local function new(session, stream_callbacks, stanza_size_limit)
 241         -- Used to track parser progress (e.g. to enforce size limits)
 242         local n_outstanding_bytes = 0;
 243         local handle_progress;
 244         if lxp_supports_bytecount then
 245                 function handle_progress(n_parsed_bytes)
 246                         n_outstanding_bytes = n_outstanding_bytes - n_parsed_bytes;
 247                 end
 248                 stanza_size_limit = stanza_size_limit or default_stanza_size_limit;
 249         elseif stanza_size_limit then
 250                 error("Stanza size limits are not supported on this version of LuaExpat")
 251         end
 252
 253         local handlers, meta = new_sax_handlers(session, stream_callbacks, handle_progress);
 254         local parser = new_parser(handlers, ns_separator, false);
 255         local parse = parser.parse;
 256
 257         function session.open_stream(session, from, to) -- luacheck: ignore 432/session
 258                 local send = session.sends2s or session.send;
 259
 260                 local attr = {
 261                         ["xmlns:stream"] = "http://etherx.jabber.org/streams",
 262                         ["xml:lang"] = "en",
 263                         xmlns = stream_callbacks.default_ns,
 264                         version = session.version and (session.version > 0 and "1.0" or nil),
 265                         id = session.streamid,
 266                         from = from or session.host, to = to,
 267                 };
 268                 if session.stream_attrs then
 269                         session:stream_attrs(from, to, attr)
 270                 end
 271                 send("<?xml version='1.0'?>");
 272                 send(st.stanza("stream:stream", attr):top_tag());
 273                 return true;
 274         end
 275
 276         return {
 277                 reset = function ()
 278                         parser = new_parser(handlers, ns_separator, false);
 279                         parse = parser.parse;
 280                         n_outstanding_bytes = 0;
 281                         meta.reset();
 282                 end,
 283                 feed = function (self, data) -- luacheck: ignore 212/self
 284                         if lxp_supports_bytecount then
 285                                 n_outstanding_bytes = n_outstanding_bytes + #data;
 286                         end
 287                         local _parser = parser;
 288                         local ok, err = parse(_parser, data);
 289                         if lxp_supports_bytecount and n_outstanding_bytes > stanza_size_limit then
 290                                 return nil, "stanza-too-large";
 291                         end
 292                         if parser ~= _parser then
 293                                 _parser:parse();
 294                                 _parser:close();
 295                         end
 296                         return ok, err;
 297                 end,
 298                 set_session = meta.set_session;
 299         };
 300 end
 301
 302 return {
 303         ns_separator = ns_separator;
 304         ns_pattern = ns_pattern;
 305         new_sax_handlers = new_sax_handlers;
 306         new = new;
 307 };