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
  10 local lxp = require "lxp";
  11 local st = require "util.stanza";
  12 local stanza_mt = st.stanza_mt;
  13
  14 local error = error;
  15 local tostring = tostring;
  16 local t_insert = table.insert;
  17 local t_concat = table.concat;
  18 local t_remove = table.remove;
  19 local setmetatable = setmetatable;
  20
  21 -- COMPAT: w/LuaExpat 1.1.0
  22 local lxp_supports_doctype = pcall(lxp.new, { StartDoctypeDecl = false });
  23
  24 module "xmppstream"
  25
  26 local new_parser = lxp.new;
  27
  28 local xml_namespace = {
  29         ["http://www.w3.org/XML/1998/namespace\1lang"] = "xml:lang";
  30         ["http://www.w3.org/XML/1998/namespace\1space"] = "xml:space";
  31         ["http://www.w3.org/XML/1998/namespace\1base"] = "xml:base";
  32         ["http://www.w3.org/XML/1998/namespace\1id"] = "xml:id";
  33 };
  34
  35 local xmlns_streams = "http://etherx.jabber.org/streams";
  36
  37 local ns_separator = "\1";
  38 local ns_pattern = "^([^"..ns_separator.."]*)"..ns_separator.."?(.*)$";
  39
  40 _M.ns_separator = ns_separator;
  41 _M.ns_pattern = ns_pattern;
  42
  43 function new_sax_handlers(session, stream_callbacks)
  44         local xml_handlers = {};
  45
  46         local cb_streamopened = stream_callbacks.streamopened;
  47         local cb_streamclosed = stream_callbacks.streamclosed;
  48         local cb_error = stream_callbacks.error or function(session, e, stanza) error("XML stream error: "..tostring(e)..(stanza and ": "..tostring(stanza) or ""),2); end;
  49         local cb_handlestanza = stream_callbacks.handlestanza;
  50
  51         local stream_ns = stream_callbacks.stream_ns or xmlns_streams;
  52         local stream_tag = stream_callbacks.stream_tag or "stream";
  53         if stream_ns ~= "" then
  54                 stream_tag = stream_ns..ns_separator..stream_tag;
  55         end
  56         local stream_error_tag = stream_ns..ns_separator..(stream_callbacks.error_tag or "error");
  57
  58         local stream_default_ns = stream_callbacks.default_ns;
  59
  60         local stack = {};
  61         local chardata, stanza = {};
  62         local non_streamns_depth = 0;
  63         function xml_handlers:StartElement(tagname, attr)
  64                 if stanza and #chardata > 0 then
  65                         -- We have some character data in the buffer
  66                         t_insert(stanza, t_concat(chardata));
  67                         chardata = {};
  68                 end
  69                 local curr_ns,name = tagname:match(ns_pattern);
  70                 if name == "" then
  71                         curr_ns, name = "", curr_ns;
  72                 end
  73
  74                 if curr_ns ~= stream_default_ns or non_streamns_depth > 0 then
  75                         attr.xmlns = curr_ns;
  76                         non_streamns_depth = non_streamns_depth + 1;
  77                 end
  78
  79                 for i=1,#attr do
  80                         local k = attr[i];
  81                         attr[i] = nil;
  82                         local xmlk = xml_namespace[k];
  83                         if xmlk then
  84                                 attr[xmlk] = attr[k];
  85                                 attr[k] = nil;
  86                         end
  87                 end
  88
  89                 if not stanza then --if we are not currently inside a stanza
  90                         if session.notopen then
  91                                 if tagname == stream_tag then
  92                                         non_streamns_depth = 0;
  93                                         if cb_streamopened then
  94                                                 cb_streamopened(session, attr);
  95                                         end
  96                                 else
  97                                         -- Garbage before stream?
  98                                         cb_error(session, "no-stream");
  99                                 end
 100                                 return;
 101                         end
 102                         if curr_ns == "jabber:client" and name ~= "iq" and name ~= "presence" and name ~= "message" then
 103                                 cb_error(session, "invalid-top-level-element");
 104                         end
 105
 106                         stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
 107                 else -- we are inside a stanza, so add a tag
 108                         t_insert(stack, stanza);
 109                         local oldstanza = stanza;
 110                         stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
 111                         t_insert(oldstanza, stanza);
 112                         t_insert(oldstanza.tags, stanza);
 113                 end
 114         end
 115         function xml_handlers:CharacterData(data)
 116                 if stanza then
 117                         t_insert(chardata, data);
 118                 end
 119         end
 120         function xml_handlers:EndElement(tagname)
 121                 if non_streamns_depth > 0 then
 122                         non_streamns_depth = non_streamns_depth - 1;
 123                 end
 124                 if stanza then
 125                         if #chardata > 0 then
 126                                 -- We have some character data in the buffer
 127                                 t_insert(stanza, t_concat(chardata));
 128                                 chardata = {};
 129                         end
 130                         -- Complete stanza
 131                         if #stack == 0 then
 132                                 if tagname ~= stream_error_tag then
 133                                         cb_handlestanza(session, stanza);
 134                                 else
 135                                         cb_error(session, "stream-error", stanza);
 136                                 end
 137                                 stanza = nil;
 138                         else
 139                                 stanza = t_remove(stack);
 140                         end
 141                 else
 142                         if cb_streamclosed then
 143                                 cb_streamclosed(session);
 144                         end
 145                 end
 146         end
 147
 148         local function restricted_handler(parser)
 149                 cb_error(session, "parse-error", "restricted-xml", "Restricted XML, see RFC 6120 section 11.1.");
 150                 if not parser.stop or not parser:stop() then
 151                         error("Failed to abort parsing");
 152                 end
 153         end
 154
 155         if lxp_supports_doctype then
 156                 xml_handlers.StartDoctypeDecl = restricted_handler;
 157         end
 158         xml_handlers.Comment = restricted_handler;
 159         xml_handlers.ProcessingInstruction = restricted_handler;
 160
 161         local function reset()
 162                 stanza, chardata = nil, {};
 163                 stack = {};
 164         end
 165
 166         local function set_session(stream, new_session)
 167                 session = new_session;
 168         end
 169
 170         return xml_handlers, { reset = reset, set_session = set_session };
 171 end
 172
 173 function new(session, stream_callbacks)
 174         local handlers, meta = new_sax_handlers(session, stream_callbacks);
 175         local parser = new_parser(handlers, ns_separator);
 176         local parse = parser.parse;
 177
 178         return {
 179                 reset = function ()
 180                         parser = new_parser(handlers, ns_separator);
 181                         parse = parser.parse;
 182                         meta.reset();
 183                 end,
 184                 feed = function (self, data)
 185                         return parse(parser, data);
 186                 end,
 187                 set_session = meta.set_session;
 188         };
 189 end
 190
 191 return _M;