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