8 files changed, 585 insertions, 11 deletions

diff --git a/tools/ejabberdsql2prosody.lua b/tools/ejabberdsql2prosody.lua
index ef4706ce..958cf0e2 100644
--- a/tools/ejabberdsql2prosody.lua
+++ b/tools/ejabberdsql2prosody.lua
@@ -7,6 +7,8 @@
 -- COPYING file in the source package for more information.
 --
 
+prosody = {};
+
 package.path = package.path ..";../?.lua";
 local serialize = require "util.serialization".serialize;
 local st = require "util.stanza";
diff --git a/tools/migration/Makefile b/tools/migration/Makefile
new file mode 100644
index 00000000..5998a5f7
--- /dev/null
+++ b/tools/migration/Makefile
@@ -0,0 +1,38 @@
+
+include ../../config.unix
+
+BIN = $(DESTDIR)$(PREFIX)/bin
+CONFIG = $(DESTDIR)$(SYSCONFDIR)
+SOURCE = $(DESTDIR)$(PREFIX)/lib/prosody
+DATA = $(DESTDIR)$(DATADIR)
+MAN = $(DESTDIR)$(PREFIX)/share/man
+
+INSTALLEDSOURCE = $(PREFIX)/lib/prosody
+INSTALLEDCONFIG = $(SYSCONFDIR)
+INSTALLEDMODULES = $(PREFIX)/lib/prosody/modules
+INSTALLEDDATA = $(DATADIR)
+
+SOURCE_FILES = migrator/*.lua
+
+all: prosody-migrator.install migrator.cfg.lua.install prosody-migrator.lua $(SOURCE_FILES)
+
+install: prosody-migrator.install migrator.cfg.lua.install
+	install -d $(BIN) $(CONFIG) $(SOURCE) $(SOURCE)/migrator
+	install -d $(MAN)/man1
+	install -d $(SOURCE)/migrator
+	install -m755 ./prosody-migrator.install $(BIN)/prosody-migrator
+	install -m644 $(SOURCE_FILES) $(SOURCE)/migrator
+	test -e $(CONFIG)/migrator.cfg.lua || install -m644 migrator.cfg.lua.install $(CONFIG)/migrator.cfg.lua
+
+clean:
+	rm -f prosody-migrator.install
+	rm -f migrator.cfg.lua.install
+
+prosody-migrator.install: prosody-migrator.lua
+	sed "s|^CFG_SOURCEDIR=.*;$$|CFG_SOURCEDIR='$(INSTALLEDSOURCE)';|; \
+		s|^CFG_CONFIGDIR=.*;$$|CFG_CONFIGDIR='$(INSTALLEDCONFIG)';|;" \
+			< prosody-migrator.lua > prosody-migrator.install
+
+migrator.cfg.lua.install: migrator.cfg.lua
+	sed "s|^local data_path = .*;$$|local data_path = '$(INSTALLEDDATA)';|;" \
+		< migrator.cfg.lua > migrator.cfg.lua.install
diff --git a/tools/migration/migrator.cfg.lua b/tools/migration/migrator.cfg.lua
new file mode 100644
index 00000000..fa37f2a3
--- /dev/null
+++ b/tools/migration/migrator.cfg.lua
@@ -0,0 +1,26 @@
+local data_path = "../../data";
+
+input {
+	type = "prosody_files";
+	path = data_path;
+}
+
+output {
+	type = "prosody_sql";
+	driver = "SQLite3";
+	database = data_path.."/prosody.sqlite";
+}
+
+--[[
+
+input {
+	type = "prosody_files";
+	path = data_path;
+}
+output {
+	type = "prosody_sql";
+	driver = "SQLite3";
+	database = data_path.."/prosody.sqlite";
+}
+
+]]
diff --git a/tools/migration/migrator/mtools.lua b/tools/migration/migrator/mtools.lua
new file mode 100644
index 00000000..e7b774bb
--- /dev/null
+++ b/tools/migration/migrator/mtools.lua
@@ -0,0 +1,56 @@
+
+
+local print = print;
+local t_insert = table.insert;
+local t_sort = table.sort;
+
+module "mtools"
+
+function sorted(params)
+
+	local reader = params.reader; -- iterator to get items from
+	local sorter = params.sorter; -- sorting function
+	local filter = params.filter; -- filter function
+
+	local cache = {};
+	for item in reader do
+		if filter then item = filter(item); end
+		if item then t_insert(cache, item); end
+	end
+	if sorter then
+		t_sort(cache, sorter);
+	end
+	local i = 0;
+	return function()
+		i = i + 1;
+		return cache[i];
+	end;
+
+end
+
+function merged(reader, merger)
+
+	local item1 = reader();
+	local merged = { item1 };
+	return function()
+		while true do
+			if not item1 then return nil; end
+			local item2 = reader();
+			if not item2 then item1 = nil; return merged; end
+			if merger(item1, item2) then
+			--print("merged")
+				item1 = item2;
+				t_insert(merged, item1);
+			else
+			--print("unmerged", merged)
+				item1 = item2;
+				local tmp = merged;
+				merged = { item1 };
+				return tmp;
+			end
+		end
+	end;
+
+end
+
+return _M;
diff --git a/tools/migration/migrator/prosody_files.lua b/tools/migration/migrator/prosody_files.lua
new file mode 100644
index 00000000..4e42f564
--- /dev/null
+++ b/tools/migration/migrator/prosody_files.lua
@@ -0,0 +1,134 @@
+
+local print = print;
+local assert = assert;
+local setmetatable = setmetatable;
+local tonumber = tonumber;
+local char = string.char;
+local coroutine = coroutine;
+local lfs = require "lfs";
+local loadfile = loadfile;
+local setfenv = setfenv;
+local pcall = pcall;
+local mtools = require "migrator.mtools";
+local next = next;
+local pairs = pairs;
+local json = require "util.json";
+local os_getenv = os.getenv;
+
+prosody = {};
+local dm = require "util.datamanager"
+
+module "prosody_files"
+
+local function is_dir(path) return lfs.attributes(path, "mode") == "directory"; end
+local function is_file(path) return lfs.attributes(path, "mode") == "file"; end
+local function clean_path(path)
+	return path:gsub("\\", "/"):gsub("//+", "/"):gsub("^~", os_getenv("HOME") or "~");
+end
+local encode, decode; do
+	local urlcodes = setmetatable({}, { __index = function (t, k) t[k] = char(tonumber("0x"..k)); return t[k]; end });
+	decode = function (s) return s and (s:gsub("+", " "):gsub("%%([a-fA-F0-9][a-fA-F0-9])", urlcodes)); end
+	encode = function (s) return s and (s:gsub("%W", function (c) return format("%%%02x", c:byte()); end)); end
+end
+local function decode_dir(x)
+	if x:gsub("%%%x%x", ""):gsub("[a-zA-Z0-9]", "") == "" then
+		return decode(x);
+	end
+end
+local function decode_file(x)
+	if x:match(".%.dat$") and x:gsub("%.dat$", ""):gsub("%%%x%x", ""):gsub("[a-zA-Z0-9]", "") == "" then
+		return decode(x:gsub("%.dat$", ""));
+	end
+end
+local function prosody_dir(path, ondir, onfile, ...)
+	for x in lfs.dir(path) do
+		local xpath = path.."/"..x;
+		if decode_dir(x) and is_dir(xpath) then
+			ondir(xpath, x, ...);
+		elseif decode_file(x) and is_file(xpath) then
+			onfile(xpath, x, ...);
+		end
+	end
+end
+
+local function handle_root_file(path, name)
+	--print("root file: ", decode_file(name))
+	coroutine.yield { user = nil, host = nil, store = decode_file(name) };
+end
+local function handle_host_file(path, name, host)
+	--print("host file: ", decode_dir(host).."/"..decode_file(name))
+	coroutine.yield { user = nil, host = decode_dir(host), store = decode_file(name) };
+end
+local function handle_store_file(path, name, store, host)
+	--print("store file: ", decode_file(name).."@"..decode_dir(host).."/"..decode_dir(store))
+	coroutine.yield { user = decode_file(name), host = decode_dir(host), store = decode_dir(store) };
+end
+local function handle_host_store(path, name, host)
+	prosody_dir(path, function() end, handle_store_file, name, host);
+end
+local function handle_host_dir(path, name)
+	prosody_dir(path, handle_host_store, handle_host_file, name);
+end
+local function handle_root_dir(path)
+	prosody_dir(path, handle_host_dir, handle_root_file);
+end
+
+local function decode_user(item)
+	local userdata = {
+		user = item[1].user;
+		host = item[1].host;
+		stores = {};
+	};
+	for i=1,#item do -- loop over stores
+		local result = {};
+		local store = item[i];
+		userdata.stores[store.store] = store.data;
+		store.user = nil; store.host = nil; store.store = nil;
+	end
+	return userdata;
+end
+
+function reader(input)
+	local path = clean_path(assert(input.path, "no input.path specified"));
+	assert(is_dir(path), "input.path is not a directory");
+	local iter = coroutine.wrap(function()handle_root_dir(path);end);
+	-- get per-user stores, sorted
+	local iter = mtools.sorted {
+		reader = function()
+			local x = iter();
+			if x then
+				dm.set_data_path(path);
+				x.data = assert(dm.load(x.user, x.host, x.store));
+				return x;
+			end
+		end;
+		sorter = function(a, b)
+			local a_host, a_user, a_store = a.host or "", a.user or "", a.store or "";
+			local b_host, b_user, b_store = b.host or "", b.user or "", b.store or "";
+			return a_host > b_host or (a_host==b_host and a_user > b_user) or (a_host==b_host and a_user==b_user and a_store > b_store);
+		end;
+	};
+	-- merge stores to get users
+	iter = mtools.merged(iter, function(a, b)
+		return (a.host == b.host and a.user == b.user);
+	end);
+
+	return function()
+		local x = iter();
+		return x and decode_user(x);
+	end
+end
+
+function writer(output)
+	local path = clean_path(assert(output.path, "no output.path specified"));
+	assert(is_dir(path), "output.path is not a directory");
+	return function(item)
+		if not item then return; end -- end of input
+		dm.set_data_path(path);
+		for store, data in pairs(item.stores) do
+			assert(dm.store(item.user, item.host, store, data));
+		end
+	end
+end
+
+return _M;
diff --git a/tools/migration/migrator/prosody_sql.lua b/tools/migration/migrator/prosody_sql.lua
new file mode 100644
index 00000000..50ae8c40
--- /dev/null
+++ b/tools/migration/migrator/prosody_sql.lua
@@ -0,0 +1,182 @@
+
+local assert = assert;
+local have_DBI, DBI = pcall(require,"DBI");
+local print = print;
+local type = type;
+local next = next;
+local pairs = pairs;
+local t_sort = table.sort;
+local json = require "util.json";
+local mtools = require "migrator.mtools";
+local tostring = tostring;
+local tonumber = tonumber;
+
+if not have_DBI then
+	error("LuaDBI (required for SQL support) was not found, please see http://prosody.im/doc/depends#luadbi", 0);
+end
+
+module "prosody_sql"
+
+local function create_table(connection, params)
+	local create_sql = "CREATE TABLE `prosody` (`host` TEXT, `user` TEXT, `store` TEXT, `key` TEXT, `type` TEXT, `value` TEXT);";
+	if params.driver == "PostgreSQL" then
+		create_sql = create_sql:gsub("`", "\"");
+	end
+	
+	local stmt = connection:prepare(create_sql);
+	if stmt then
+		local ok = stmt:execute();
+		local commit_ok = connection:commit();
+		if ok and commit_ok then
+			local index_sql = "CREATE INDEX `prosody_index` ON `prosody` (`host`, `user`, `store`, `key`)";
+			if params.driver == "PostgreSQL" then
+				index_sql = index_sql:gsub("`", "\"");
+			elseif params.driver == "MySQL" then
+				index_sql = index_sql:gsub("`([,)])", "`(20)%1");
+			end
+			local stmt, err = connection:prepare(index_sql);
+			local ok, commit_ok, commit_err;
+			if stmt then
+				ok, err = assert(stmt:execute());
+				commit_ok, commit_err = assert(connection:commit());
+			end
+		end
+	end
+end
+
+local function serialize(value)
+	local t = type(value);
+	if t == "string" or t == "boolean" or t == "number" then
+		return t, tostring(value);
+	elseif t == "table" then
+		local value,err = json.encode(value);
+		if value then return "json", value; end
+		return nil, err;
+	end
+	return nil, "Unhandled value type: "..t;
+end
+local function deserialize(t, value)
+	if t == "string" then return value;
+	elseif t == "boolean" then
+		if value == "true" then return true;
+		elseif value == "false" then return false; end
+	elseif t == "number" then return tonumber(value);
+	elseif t == "json" then
+		return json.decode(value);
+	end
+end
+
+local function decode_user(item)
+	local userdata = {
+		user = item[1][1].user;
+		host = item[1][1].host;
+		stores = {};
+	};
+	for i=1,#item do -- loop over stores
+		local result = {};
+		local store = item[i];
+		for i=1,#store do -- loop over store data
+			local row = store[i];
+			local k = row.key;
+			local v = deserialize(row.type, row.value);
+			if k and v then
+				if k ~= "" then result[k] = v; elseif type(v) == "table" then
+					for a,b in pairs(v) do
+						result[a] = b;
+					end
+				end
+			end
+			userdata.stores[store[1].store] = result;
+		end
+	end
+	return userdata;
+end
+
+function reader(input)
+	local dbh = assert(DBI.Connect(
+		assert(input.driver, "no input.driver specified"),
+		assert(input.database, "no input.database specified"),
+		input.username, input.password,
+		input.host, input.port
+	));
+	assert(dbh:ping());
+	local stmt = assert(dbh:prepare("SELECT * FROM prosody"));
+	assert(stmt:execute());
+	local keys = {"host", "user", "store", "key", "type", "value"};
+	local f,s,val = stmt:rows(true);
+	-- get SQL rows, sorted
+	local iter = mtools.sorted {
+		reader = function() val = f(s, val); return val; end;
+		filter = function(x)
+			for i=1,#keys do
+				if not x[keys[i]] then return false; end -- TODO log error, missing field
+			end
+			if x.host  == "" then x.host  = nil; end
+			if x.user  == "" then x.user  = nil; end
+			if x.store == "" then x.store = nil; end
+			return x;
+		end;
+		sorter = function(a, b)
+			local a_host, a_user, a_store = a.host or "", a.user or "", a.store or "";
+			local b_host, b_user, b_store = b.host or "", b.user or "", b.store or "";
+			return a_host > b_host or (a_host==b_host and a_user > b_user) or (a_host==b_host and a_user==b_user and a_store > b_store);
+		end;
+	};
+	-- merge rows to get stores
+	iter = mtools.merged(iter, function(a, b)
+		return (a.host == b.host and a.user == b.user and a.store == b.store);
+	end);
+	-- merge stores to get users
+	iter = mtools.merged(iter, function(a, b)
+		return (a[1].host == b[1].host and a[1].user == b[1].user);
+	end);
+	return function()
+		local x = iter();
+		return x and decode_user(x);
+	end;
+end
+
+function writer(output, iter)
+	local dbh = assert(DBI.Connect(
+		assert(output.driver, "no output.driver specified"),
+		assert(output.database, "no output.database specified"),
+		output.username, output.password,
+		output.host, output.port
+	));
+	assert(dbh:ping());
+	create_table(dbh, output);
+	local stmt = assert(dbh:prepare("SELECT * FROM prosody"));
+	assert(stmt:execute());
+	local stmt = assert(dbh:prepare("DELETE FROM prosody"));
+	assert(stmt:execute());
+	local insert_sql = "INSERT INTO `prosody` (`host`,`user`,`store`,`key`,`type`,`value`) VALUES (?,?,?,?,?,?)";
+	if output.driver == "PostgreSQL" then
+		insert_sql = insert_sql:gsub("`", "\"");
+	end
+	local insert = assert(dbh:prepare(insert_sql));
+
+	return function(item)
+		if not item then assert(dbh:commit()) return dbh:close(); end -- end of input
+		local host = item.host or "";
+		local user = item.user or "";
+		for store, data in pairs(item.stores) do
+			-- TODO transactions
+			local extradata = {};
+			for key, value in pairs(data) do
+				if type(key) == "string" and key ~= "" then
+					local t, value = assert(serialize(value));
+					local ok, err = assert(insert:execute(host, user, store, key, t, value));
+				else
+					extradata[key] = value;
+				end
+			end
+			if next(extradata) ~= nil then
+				local t, extradata = assert(serialize(extradata));
+				local ok, err = assert(insert:execute(host, user, store, "", t, extradata));
+			end
+		end
+	end;
+end
+
+
+return _M;
diff --git a/tools/migration/prosody-migrator.lua b/tools/migration/prosody-migrator.lua
new file mode 100644
index 00000000..2a8bf1c3
--- /dev/null
+++ b/tools/migration/prosody-migrator.lua
@@ -0,0 +1,134 @@
+#!/usr/bin/env lua
+
+CFG_SOURCEDIR=os.getenv("PROSODY_SRCDIR");
+CFG_CONFIGDIR=os.getenv("PROSODY_CFGDIR");
+
+-- Substitute ~ with path to home directory in paths
+if CFG_CONFIGDIR then
+        CFG_CONFIGDIR = CFG_CONFIGDIR:gsub("^~", os.getenv("HOME"));
+end
+
+if CFG_SOURCEDIR then
+        CFG_SOURCEDIR = CFG_SOURCEDIR:gsub("^~", os.getenv("HOME"));
+end
+
+local default_config = (CFG_CONFIGDIR or ".").."/migrator.cfg.lua";
+
+-- Command-line parsing
+local options = {};
+local handled_opts = 0;
+for i = 1, #arg do
+	if arg[i]:sub(1,2) == "--" then
+		local opt, val = arg[i]:match("([%w-]+)=?(.*)");
+		if opt then
+			options[(opt:sub(3):gsub("%-", "_"))] = #val > 0 and val or true;
+		end
+		handled_opts = i;
+	else
+		break;
+	end
+end
+table.remove(arg, handled_opts);
+
+-- Load config file
+local function loadfilein(file, env)
+	if loadin then
+		return loadin(env, io.open(file):read("*a"));
+	else
+		local chunk, err = loadfile(file);
+		if chunk then
+			setfenv(chunk, env);
+		end
+		return chunk, err;
+	end
+end
+
+local config_file = options.config or default_config;
+local from_store = arg[1] or "input";
+local to_store = arg[2] or "output";
+
+config = {};
+local config_env = setmetatable({}, { __index = function(t, k) return function(tbl) config[k] = tbl; end; end });
+local config_chunk, err = loadfilein(config_file, config_env);
+if not config_chunk then
+	print("There was an error loading the config file, check the file exists");
+	print("and that the syntax is correct:");
+	print("", err);
+	os.exit(1);
+end
+
+config_chunk();
+
+if CFG_SOURCEDIR then
+	package.path = CFG_SOURCEDIR.."/?.lua;"..package.path;
+	package.cpath = CFG_SOURCEDIR.."/?.so;"..package.cpath;
+elseif not package.loaded["util.json"] then
+	package.path = "../../?.lua;"..package.path
+	package.cpath = "../../?.so;"..package.cpath
+end
+
+local have_err;
+if #arg > 0 and #arg ~= 2 then
+	have_err = true;
+	print("Error: Incorrect number of parameters supplied.");
+end
+if not config[from_store] then
+	have_err = true;
+	print("Error: Input store '"..from_store.."' not found in the config file.");
+end
+if not config[to_store] then
+	have_err = true;
+	print("Error: Output store '"..to_store.."' not found in the config file.");
+end
+
+function load_store_handler(name)
+	local store_type = config[name].type;
+	if not store_type then
+		print("Error: "..name.." store type not specified in the config file");
+		return false;
+	else
+		local ok, err = pcall(require, "migrator."..store_type);
+		if not ok then
+			if package.loaded["migrator."..store_type] then
+				print(("Error: Failed to initialize '%s' store:\n\t%s")
+					:format(name, err));
+			else
+				print(("Error: Unrecognised store type for '%s': %s")
+					:format(from_store, store_type));
+			end
+			return false;
+		end
+	end
+	return true;
+end
+
+have_err = have_err or not(load_store_handler(from_store, "input") and load_store_handler(to_store, "output"));
+
+if have_err then
+	print("");
+	print("Usage: "..arg[0].." FROM_STORE TO_STORE");
+	print("If no stores are specified, 'input' and 'output' are used.");
+	print("");
+	print("The available stores in your migrator config are:");
+	print("");
+	for store in pairs(config) do
+		print("", store);
+	end
+	print("");
+	os.exit(1);
+end
+	
+local itype = config[from_store].type;
+local otype = config[to_store].type;
+local reader = require("migrator."..itype).reader(config[from_store]);
+local writer = require("migrator."..otype).writer(config[to_store]);
+
+local json = require "util.json";
+
+io.stderr:write("Migrating...\n");
+for x in reader do
+	--print(json.encode(x))
+	writer(x);
+end
+writer(nil); -- close
+io.stderr:write("Done!\n");
diff --git a/tools/xep227toprosody.lua b/tools/xep227toprosody.lua
index 313b2194..23e5948b 100755
--- a/tools/xep227toprosody.lua
+++ b/tools/xep227toprosody.lua
@@ -36,13 +36,15 @@ end
 
 local lxp = require "lxp";
 local st = require "util.stanza";
-local init_xmlhandlers = require "core.xmlhandlers";
+local xmppstream = require "util.xmppstream";
+local new_xmpp_handlers = xmppstream.new_sax_handlers;
 local dm = require "util.datamanager"
 dm.set_data_path("data");
 
-local ns_separator = "\1";
-local ns_pattern = "^([^"..ns_separator.."]*)"..ns_separator.."?(.*)$";
-local ns_xep227 = "http://www.xmpp.org/extensions/xep-0227.html#ns";
+local ns_separator = xmppstream.ns_separator;
+local ns_pattern = xmppstream.ns_pattern;
+
+local xmlns_xep227 = "http://www.xmpp.org/extensions/xep-0227.html#ns";
 
 -----------------------------------------------------------------------
 
@@ -114,7 +116,7 @@ function store_offline_messages(username, host, offline_messages)
 		--print("message :"..ch:pretty_print());
 		local ret, err = dm.list_append(username, host, "offline", st.preserialize(ch));
 		print("["..(err or "success").."] stored offline message: " ..username.."@"..host.." - "..ch.attr.from);
- 	end
+	end
 end
 
 
@@ -146,7 +148,7 @@ local user_name = "";
 
 local cb = {
 	stream_tag = "user",
-	stream_ns = ns_xep227,
+	stream_ns = xmlns_xep227,
 };
 function cb.streamopened(session, attr)
 	session.notopen = false;
@@ -176,7 +178,7 @@ function cb.handlestanza(session, stanza)
 	end
 end
 
-local user_handlers = init_xmlhandlers({ notopen = true, }, cb);
+local user_handlers = new_xmpp_handlers({ notopen = true }, cb);
 
 -----------------------------------------------------------------------
 
@@ -195,10 +197,10 @@ function lxp_handlers.StartElement(parser, elementname, attributes)
 	if curr_host ~= "" then
 		-- forward to xmlhandlers
 		user_handlers:StartElement(elementname, attributes);
-	elseif (curr_ns == ns_xep227) and (name == "host") then
+	elseif (curr_ns == xmlns_xep227) and (name == "host") then
 		curr_host = attributes["jid"]; -- start of host element
 		print("Begin parsing host "..curr_host);
-	elseif (curr_ns ~= ns_xep227) or (name ~= "server-data") then
+	elseif (curr_ns ~= xmlns_xep227) or (name ~= "server-data") then
 		io.stderr:write("Unhandled XML element: ", name, "\n");
 		os.exit(1);
 	end
@@ -213,14 +215,14 @@ function lxp_handlers.EndElement(parser, elementname)
 	--count = count - 1;
 	--io.write("- ", string.rep(" ", count), name, "  (", curr_ns, ")", "\n")
 	if curr_host ~= "" then
-		if (curr_ns == ns_xep227) and (name == "host") then
+		if (curr_ns == xmlns_xep227) and (name == "host") then
 			print("End parsing host "..curr_host);
 			curr_host = "" -- end of host element
 		else
 			-- forward to xmlhandlers
 			user_handlers:EndElement(elementname);
 		end
-	elseif (curr_ns ~= ns_xep227) or (name ~= "server-data") then
+	elseif (curr_ns ~= xmlns_xep227) or (name ~= "server-data") then
 		io.stderr:write("Unhandled XML element: ", name, "\n");
 		os.exit(1);
 	end