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 ns_prefixes = {
  29         ["http://www.w3.org/XML/1998/namespace"] = "xml";
  30 };
  31
  32 local xmlns_streams = "http://etherx.jabber.org/streams";
  33
  34 local ns_separator = "\1";
  35 local ns_pattern = "^([^"..ns_separator.."]*)"..ns_separator.."?(.*)$";
  36
  37 _M.ns_separator = ns_separator;
  38 _M.ns_pattern = ns_pattern;
  39
  40 function new_sax_handlers(session, stream_callbacks)
  41         local xml_handlers = {};
  42
  43         local cb_streamopened = stream_callbacks.streamopened;
  44         local cb_streamclosed = stream_callbacks.streamclosed;
  45         local cb_error = stream_callbacks.error or function(session, e) error("XML stream error: "..tostring(e)); end;
  46         local cb_handlestanza = stream_callbacks.handlestanza;
  47
  48         local stream_ns = stream_callbacks.stream_ns or xmlns_streams;
  49         local stream_tag = stream_callbacks.stream_tag or "stream";
  50         if stream_ns ~= "" then
  51                 stream_tag = stream_ns..ns_separator..stream_tag;
  52         end
  53         local stream_error_tag = stream_ns..ns_separator..(stream_callbacks.error_tag or "error");
  54
  55         local stream_default_ns = stream_callbacks.default_ns;
  56
  57         local stack = {};
  58         local chardata, stanza = {};
  59         local non_streamns_depth = 0;
  60         function xml_handlers:StartElement(tagname, attr)
  61                 if stanza and #chardata > 0 then
  62                         -- We have some character data in the buffer
  63                         t_insert(stanza, t_concat(chardata));
  64                         chardata = {};
  65                 end
  66                 local curr_ns,name = tagname:match(ns_pattern);
  67                 if name == "" then
  68                         curr_ns, name = "", curr_ns;
  69                 end
  70
  71                 if curr_ns ~= stream_default_ns or non_streamns_depth > 0 then
  72                         attr.xmlns = curr_ns;
  73                         non_streamns_depth = non_streamns_depth + 1;
  74                 end
  75
  76                 -- FIXME !!!!!
  77                 for i=1,#attr do
  78                         local k = attr[i];
  79                         attr[i] = nil;
  80                         local ns, nm = k:match(ns_pattern);
  81                         if nm ~= "" then
  82                                 ns = ns_prefixes[ns];
  83                                 if ns then
  84                                         attr[ns..":"..nm] = attr[k];
  85                                         attr[k] = nil;
  86                                 end
  87                         end
  88                 end
  89
  90                 if not stanza then --if we are not currently inside a stanza
  91                         if session.notopen then
  92                                 if tagname == stream_tag then
  93                                         non_streamns_depth = 0;
  94                                         if cb_streamopened then
  95                                                 cb_streamopened(session, attr);
  96                                         end
  97                                 else
  98                                         -- Garbage before stream?
  99                                         cb_error(session, "no-stream");
 100                                 end
 101                                 return;
 102                         end
 103                         if curr_ns == "jabber:client" and name ~= "iq" and name ~= "presence" and name ~= "message" then
 104                                 cb_error(session, "invalid-top-level-element");
 105                         end
 106
 107                         stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
 108                 else -- we are inside a stanza, so add a tag
 109                         t_insert(stack, stanza);
 110                         local oldstanza = stanza;
 111                         stanza = setmetatable({ name = name, attr = attr, tags = {} }, stanza_mt);
 112                         t_insert(oldstanza, stanza);
 113                         t_insert(oldstanza.tags, stanza);
 114                 end
 115         end
 116         function xml_handlers:CharacterData(data)
 117                 if stanza then
 118                         t_insert(chardata, data);
 119                 end
 120         end
 121         function xml_handlers:EndElement(tagname)
 122                 if non_streamns_depth > 0 then
 123                         non_streamns_depth = non_streamns_depth - 1;
 124                 end
 125                 if stanza then
 126                         if #chardata > 0 then
 127                                 -- We have some character data in the buffer
 128                                 t_insert(stanza, t_concat(chardata));
 129                                 chardata = {};
 130                         end
 131                         -- Complete stanza
 132                         if #stack == 0 then
 133                                 if tagname ~= stream_error_tag then
 134                                         cb_handlestanza(session, stanza);
 135                                 else
 136                                         cb_error(session, "stream-error", stanza);
 137                                 end
 138                                 stanza = nil;
 139                         else
 140                                 stanza = t_remove(stack);
 141                         end
 142                 else
 143                         if tagname == stream_tag then
 144                                 if cb_streamclosed then
 145                                         cb_streamclosed(session);
 146                                 end
 147                         else
 148                                 local curr_ns,name = tagname:match(ns_pattern);
 149                                 if name == "" then
 150                                         curr_ns, name = "", curr_ns;
 151                                 end
 152                                 cb_error(session, "parse-error", "unexpected-element-close", name);
 153                         end
 154                         stanza, chardata = nil, {};
 155                         stack = {};
 156                 end
 157         end
 158
 159         local function restricted_handler(parser)
 160                 cb_error(session, "parse-error", "restricted-xml", "Restricted XML, see RFC 6120 section 11.1.");
 161                 if not parser.stop or not parser:stop() then
 162                         error("Failed to abort parsing");
 163                 end
 164         end
 165
 166         if lxp_supports_doctype then
 167                 xml_handlers.StartDoctypeDecl = restricted_handler;
 168         end
 169         xml_handlers.Comment = restricted_handler;
 170         xml_handlers.ProcessingInstruction = restricted_handler;
 171
 172         local function reset()
 173                 stanza, chardata = nil, {};
 174                 stack = {};
 175         end
 176
 177         local function set_session(stream, new_session)
 178                 session = new_session;
 179         end
 180
 181         return xml_handlers, { reset = reset, set_session = set_session };
 182 end
 183
 184 function new(session, stream_callbacks)
 185         local handlers, meta = new_sax_handlers(session, stream_callbacks);
 186         local parser = new_parser(handlers, ns_separator);
 187         local parse = parser.parse;
 188
 189         return {
 190                 reset = function ()
 191                         parser = new_parser(handlers, ns_separator);
 192                         parse = parser.parse;
 193                         meta.reset();
 194                 end,
 195                 feed = function (self, data)
 196                         return parse(parser, data);
 197                 end,
 198                 set_session = meta.set_session;
 199         };
 200 end
 201
 202 return _M;