tools/ejabberdsql2prosody.lua

   1 #!/usr/bin/env lua
   2 -- Prosody IM
   3 -- Copyright (C) 2008-2010 Matthew Wild
   4 -- Copyright (C) 2008-2010 Waqas Hussain
   5 --
   6 -- This project is MIT/X11 licensed. Please see the
   7 -- COPYING file in the source package for more information.
   8 --
   9
  10 prosody = {};
  11
  12 package.path = package.path ..";../?.lua";
  13 local serialize = require "util.serialization".serialize;
  14 local st = require "util.stanza";
  15 package.loaded["util.logger"] = {init = function() return function() end; end}
  16 local dm = require "util.datamanager"
  17 dm.set_data_path("data");
  18
  19 function parseFile(filename)
  20 ------
  21
  22 local file = nil;
  23 local last = nil;
  24 local function read(expected)
  25         local ch;
  26         if last then
  27                 ch = last; last = nil;
  28         else ch = file:read(1); end
  29         if expected and ch ~= expected then error("expected: "..expected.."; got: "..(ch or "nil")); end
  30         return ch;
  31 end
  32 local function pushback(ch)
  33         if last then error(); end
  34         last = ch;
  35 end
  36 local function peek()
  37         if not last then last = read(); end
  38         return last;
  39 end
  40
  41 local escapes = {
  42         ["\\0"] = "\0";
  43         ["\\'"] = "'";
  44         ["\\\""] = "\"";
  45         ["\\b"] = "\b";
  46         ["\\n"] = "\n";
  47         ["\\r"] = "\r";
  48         ["\\t"] = "\t";
  49         ["\\Z"] = "\26";
  50         ["\\\\"] = "\\";
  51         ["\\%"] = "%";
  52         ["\\_"] = "_";
  53 }
  54 local function unescape(s)
  55         return escapes[s] or error("Unknown escape sequence: "..s);
  56 end
  57 local function readString()
  58         read("'");
  59         local s = "";
  60         while true do
  61                 local ch = peek();
  62                 if ch == "\\" then
  63                         s = s..unescape(read()..read());
  64                 elseif ch == "'" then
  65                         break;
  66                 else
  67                         s = s..read();
  68                 end
  69         end
  70         read("'");
  71         return s;
  72 end
  73 local function readNonString()
  74         local s = "";
  75         while true do
  76                 if peek() == "," or peek() == ")" then
  77                         break;
  78                 else
  79                         s = s..read();
  80                 end
  81         end
  82         return tonumber(s);
  83 end
  84 local function readItem()
  85         if peek() == "'" then
  86                 return readString();
  87         else
  88                 return readNonString();
  89         end
  90 end
  91 local function readTuple()
  92         local items = {}
  93         read("(");
  94         while peek() ~= ")" do
  95                 table.insert(items, readItem());
  96                 if peek() == ")" then break; end
  97                 read(",");
  98         end
  99         read(")");
 100         return items;
 101 end
 102 local function readTuples()
 103         if peek() ~= "(" then read("("); end
 104         local tuples = {};
 105         while true do
 106                 table.insert(tuples, readTuple());
 107                 if peek() == "," then read() end
 108                 if peek() == ";" then break; end
 109         end
 110         return tuples;
 111 end
 112 local function readTableName()
 113         local tname = "";
 114         while peek() ~= "`" do tname = tname..read(); end
 115         return tname;
 116 end
 117 local function readInsert()
 118         if peek() == nil then return nil; end
 119         for ch in ("INSERT INTO `"):gmatch(".") do -- find line starting with this
 120                 if peek() == ch then
 121                         read(); -- found
 122                 else -- match failed, skip line
 123                         while peek() and read() ~= "\n" do end
 124                         return nil;
 125                 end
 126         end
 127         local tname = readTableName();
 128         for ch in ("` VALUES "):gmatch(".") do read(ch); end -- expect this
 129         local tuples = readTuples();
 130         read(";"); read("\n");
 131         return tname, tuples;
 132 end
 133
 134 local function readFile(filename)
 135         file = io.open(filename);
 136         if not file then error("File not found: "..filename); os.exit(0); end
 137         local t = {};
 138         while true do
 139                 local tname, tuples = readInsert();
 140                 if tname then
 141                         if t[tname] then
 142                                 local t_name = t[tname];
 143                                 for i=1,#tuples do
 144                                         table.insert(t_name, tuples[i]);
 145                                 end
 146                         else
 147                                 t[tname] = tuples;
 148                         end
 149                 elseif peek() == nil then
 150                         break;
 151                 end
 152         end
 153         return t;
 154 end
 155
 156 return readFile(filename);
 157
 158 ------
 159 end
 160
 161 -- XML parser
 162 local parse_xml = (function()
 163         local entity_map = setmetatable({
 164                 ["amp"] = "&";
 165                 ["gt"] = ">";
 166                 ["lt"] = "<";
 167                 ["apos"] = "'";
 168                 ["quot"] = "\"";
 169         }, {__index = function(_, s)
 170                         if s:sub(1,1) == "#" then
 171                                 if s:sub(2,2) == "x" then
 172                                         return string.char(tonumber(s:sub(3), 16));
 173                                 else
 174                                         return string.char(tonumber(s:sub(2)));
 175                                 end
 176                         end
 177                 end
 178         });
 179         local function xml_unescape(str)
 180                 return (str:gsub("&(.-);", entity_map));
 181         end
 182         local function parse_tag(s)
 183                 local name,sattr=(s):gmatch("([^%s]+)(.*)")();
 184                 local attr = {};
 185                 for a,b in (sattr):gmatch("([^=%s]+)=['\"]([^'\"]*)['\"]") do attr[a] = xml_unescape(b); end
 186                 return name, attr;
 187         end
 188         return function(xml)
 189                 local stanza = st.stanza("root");
 190                 local regexp = "<([^>]*)>([^<]*)";
 191                 for elem, text in xml:gmatch(regexp) do
 192                         if elem:sub(1,1) == "!" or elem:sub(1,1) == "?" then -- neglect comments and processing-instructions
 193                         elseif elem:sub(1,1) == "/" then -- end tag
 194                                 elem = elem:sub(2);
 195                                 stanza:up(); -- TODO check for start-end tag name match
 196                         elseif elem:sub(-1,-1) == "/" then -- empty tag
 197                                 elem = elem:sub(1,-2);
 198                                 local name,attr = parse_tag(elem);
 199                                 stanza:tag(name, attr):up();
 200                         else -- start tag
 201                                 local name,attr = parse_tag(elem);
 202                                 stanza:tag(name, attr);
 203                         end
 204                         if #text ~= 0 then -- text
 205                                 stanza:text(xml_unescape(text));
 206                         end
 207                 end
 208                 return stanza.tags[1];
 209         end
 210 end)();
 211 -- end of XML parser
 212
 213 local arg, host = ...;
 214 local help = "/? -? ? /h -h /help -help --help";
 215 if not(arg and host) or help:find(arg, 1, true) then
 216         print([[ejabberd SQL DB dump importer for Prosody
 217
 218   Usage: ejabberdsql2prosody.lua filename.txt hostname
 219
 220 The file can be generated using mysqldump:
 221   mysqldump db_name > filename.txt]]);
 222         os.exit(1);
 223 end
 224 local map = {
 225         ["last"] = {"username", "seconds", "state"};
 226         ["privacy_default_list"] = {"username", "name"};
 227         ["privacy_list"] = {"username", "name", "id"};
 228         ["privacy_list_data"] = {"id", "t", "value", "action", "ord", "match_all", "match_iq", "match_message", "match_presence_in", "match_presence_out"};
 229         ["private_storage"] = {"username", "namespace", "data"};
 230         ["rostergroups"] = {"username", "jid", "grp"};
 231         ["rosterusers"] = {"username", "jid", "nick", "subscription", "ask", "askmessage", "server", "subscribe", "type"};
 232         ["spool"] = {"username", "xml", "seq"};
 233         ["users"] = {"username", "password"};
 234         ["vcard"] = {"username", "vcard"};
 235         --["vcard_search"] = {};
 236 }
 237 local NULL = {};
 238 local t = parseFile(arg);
 239 for name, data in pairs(t) do
 240         local m = map[name];
 241         if m then
 242                 if #data > 0 and #data[1] ~= #m then
 243                         print("[warning] expected "..#m.." columns for table `"..name.."`, found "..#data[1]);
 244                 end
 245                 for i=1,#data do
 246                         local row = data[i];
 247                         for j=1,#m do
 248                                 row[m[j]] = row[j];
 249                                 row[j] = nil;
 250                         end
 251                 end
 252         end
 253 end
 254 --print(serialize(t));
 255
 256 for i, row in ipairs(t["users"] or NULL) do
 257         local node, password = row.username, row.password;
 258         local ret, err = dm.store(node, host, "accounts", {password = password});
 259         print("["..(err or "success").."] accounts: "..node.."@"..host);
 260 end
 261
 262 function roster(node, host, jid, item)
 263         local roster = dm.load(node, host, "roster") or {};
 264         roster[jid] = item;
 265         local ret, err = dm.store(node, host, "roster", roster);
 266         print("["..(err or "success").."] roster: " ..node.."@"..host.." - "..jid);
 267 end
 268 function roster_pending(node, host, jid)
 269         local roster = dm.load(node, host, "roster") or {};
 270         roster.pending = roster.pending or {};
 271         roster.pending[jid] = true;
 272         local ret, err = dm.store(node, host, "roster", roster);
 273         print("["..(err or "success").."] roster-pending: " ..node.."@"..host.." - "..jid);
 274 end
 275 function roster_group(node, host, jid, group)
 276         local roster = dm.load(node, host, "roster") or {};
 277         local item = roster[jid];
 278         if not item then print("Warning: No roster item "..jid.." for user "..node..", can't put in group "..group); return; end
 279         item.groups[group] = true;
 280         local ret, err = dm.store(node, host, "roster", roster);
 281         print("["..(err or "success").."] roster-group: " ..node.."@"..host.." - "..jid.." - "..group);
 282 end
 283 function private_storage(node, host, xmlns, stanza)
 284         local private = dm.load(node, host, "private") or {};
 285         private[stanza.name..":"..xmlns] = st.preserialize(stanza);
 286         local ret, err = dm.store(node, host, "private", private);
 287         print("["..(err or "success").."] private: " ..node.."@"..host.." - "..xmlns);
 288 end
 289 function offline_msg(node, host, t, stanza)
 290         stanza.attr.stamp = os.date("!%Y-%m-%dT%H:%M:%SZ", t);
 291         stanza.attr.stamp_legacy = os.date("!%Y%m%dT%H:%M:%S", t);
 292         local ret, err = dm.list_append(node, host, "offline", st.preserialize(stanza));
 293         print("["..(err or "success").."] offline: " ..node.."@"..host.." - "..os.date("!%Y-%m-%dT%H:%M:%SZ", t));
 294 end
 295 for i, row in ipairs(t["rosterusers"] or NULL) do
 296         local node, contact = row.username, row.jid;
 297         local name = row.nick;
 298         if name == "" then name = nil; end
 299         local subscription = row.subscription;
 300         if subscription == "N" then
 301                 subscription = "none"
 302         elseif subscription == "B" then
 303                 subscription = "both"
 304         elseif subscription == "F" then
 305                 subscription = "from"
 306         elseif subscription == "T" then
 307                 subscription = "to"
 308         else error("Unknown subscription type: "..subscription) end;
 309         local ask = row.ask;
 310         if ask == "N" then
 311                 ask = nil;
 312         elseif ask == "O" then
 313                 ask = "subscribe";
 314         elseif ask == "I" then
 315                 roster_pending(node, host, contact);
 316                 ask = nil;
 317         elseif ask == "B" then
 318                 roster_pending(node, host, contact);
 319                 ask = "subscribe";
 320         else error("Unknown ask type: "..ask); end
 321         local item = {name = name, ask = ask, subscription = subscription, groups = {}};
 322         roster(node, host, contact, item);
 323 end
 324 for i, row in ipairs(t["rostergroups"] or NULL) do
 325         roster_group(row.username, host, row.jid, row.grp);
 326 end
 327 for i, row in ipairs(t["vcard"] or NULL) do
 328         local ret, err = dm.store(row.username, host, "vcard", st.preserialize(parse_xml(row.vcard)));
 329         print("["..(err or "success").."] vCard: "..row.username.."@"..host);
 330 end
 331 for i, row in ipairs(t["private_storage"] or NULL) do
 332         private_storage(row.username, host, row.namespace, parse_xml(row.data));
 333 end
 334 table.sort(t["spool"] or NULL, function(a,b) return a.seq < b.seq; end); -- sort by sequence number, just in case
 335 local time_offset = os.difftime(os.time(os.date("!*t")), os.time(os.date("*t"))) -- to deal with timezones
 336 local date_parse = function(s)
 337         local year, month, day, hour, min, sec = s:match("(....)-?(..)-?(..)T(..):(..):(..)");
 338         return os.time({year=year, month=month, day=day, hour=hour, min=min, sec=sec-time_offset});
 339 end
 340 for i, row in ipairs(t["spool"] or NULL) do
 341         local stanza = parse_xml(row.xml);
 342         local last_child = stanza.tags[#stanza.tags];
 343         if not last_child or last_child ~= stanza[#stanza] then error("Last child of offline message is not a tag"); end
 344         if last_child.name ~= "x" and last_child.attr.xmlns ~= "jabber:x:delay" then error("Last child of offline message is not a timestamp"); end
 345         stanza[#stanza], stanza.tags[#stanza.tags] = nil, nil;
 346         local t = date_parse(last_child.attr.stamp);
 347         offline_msg(row.username, host, t, stanza);
 348 end