util/serialization.lua

   1 -- Prosody IM
   2 -- Copyright (C) 2008-2010 Matthew Wild
   3 -- Copyright (C) 2008-2010 Waqas Hussain
   4 -- Copyright (C) 2018 Kim Alvefur
   5 --
   6 -- This project is MIT/X11 licensed. Please see the
   7 -- COPYING file in the source package for more information.
   8 --
   9
  10 local getmetatable = getmetatable;
  11 local next, type = next, type;
  12 local s_format = string.format;
  13 local s_gsub = string.gsub;
  14 local s_rep = string.rep;
  15 local s_char = string.char;
  16 local s_match = string.match;
  17 local t_concat = table.concat;
  18
  19 local to_hex = require "util.hex".to;
  20
  21 local pcall = pcall;
  22 local envload = require"util.envload".envload;
  23
  24 local pos_inf, neg_inf = math.huge, -math.huge;
  25 local m_type = math.type or function (n)
  26         return n % 1 == 0 and n <= 9007199254740992 and n >= -9007199254740992 and "integer" or "float";
  27 end;
  28
  29 local function rawpairs(t)
  30         return next, t, nil;
  31 end
  32
  33 local function fatal_error(obj, why)
  34         error("Can't serialize "..type(obj) .. (why and ": ".. why or ""));
  35 end
  36
  37 local function nonfatal_fallback(x, why)
  38         return s_format("{__type=%q,__error=%q}", type(x), why or "fail");
  39 end
  40
  41 local string_escapes = {
  42         ['\a'] = [[\a]]; ['\b'] = [[\b]];
  43         ['\f'] = [[\f]]; ['\n'] = [[\n]];
  44         ['\r'] = [[\r]]; ['\t'] = [[\t]];
  45         ['\v'] = [[\v]]; ['\\'] = [[\\]];
  46         ['\"'] = [[\"]]; ['\''] = [[\']];
  47 }
  48
  49 for i = 0, 255 do
  50         local c = s_char(i);
  51         if not string_escapes[c] then
  52                 string_escapes[c] = s_format("\\%03d", i);
  53         end
  54 end
  55
  56 local default_keywords = {
  57         ["do"] = true; ["and"] = true; ["else"] = true; ["break"] = true;
  58         ["if"] = true; ["end"] = true; ["goto"] = true; ["false"] = true;
  59         ["in"] = true; ["for"] = true; ["then"] = true; ["local"] = true;
  60         ["or"] = true; ["nil"] = true; ["true"] = true; ["until"] = true;
  61         ["elseif"] = true; ["function"] = true; ["not"] = true;
  62         ["repeat"] = true; ["return"] = true; ["while"] = true;
  63 };
  64
  65 local function new(opt)
  66         if type(opt) ~= "table" then
  67                 opt = { preset = opt };
  68         end
  69
  70         local types = {
  71                 table = true;
  72                 string = true;
  73                 number = true;
  74                 boolean = true;
  75                 ["nil"] = true;
  76         };
  77
  78         -- presets
  79         if opt.preset == "debug" then
  80                 opt.preset = "oneline";
  81                 opt.freeze = true;
  82                 opt.fatal = false;
  83                 opt.fallback = nonfatal_fallback;
  84                 opt.unquoted = true;
  85         end
  86         if opt.preset == "oneline" then
  87                 opt.indentwith = opt.indentwith or "";
  88                 opt.itemstart = opt.itemstart or " ";
  89                 opt.itemlast = opt.itemlast or "";
  90                 opt.tend = opt.tend or " }";
  91         elseif opt.preset == "compact" then
  92                 opt.indentwith = opt.indentwith or "";
  93                 opt.itemstart = opt.itemstart or "";
  94                 opt.itemlast = opt.itemlast or "";
  95                 opt.equals = opt.equals or "=";
  96                 opt.unquoted = true;
  97         end
  98
  99         local fallback = opt.fallback or opt.fatal == false and nonfatal_fallback or fatal_error;
 100
 101         local function ser(v)
 102                 return (types[type(v)] or fallback)(v);
 103         end
 104
 105         local keywords = opt.keywords or default_keywords;
 106
 107         -- indented
 108         local indentwith = opt.indentwith or "\t";
 109         local itemstart = opt.itemstart or "\n";
 110         local itemsep = opt.itemsep or ";";
 111         local itemlast = opt.itemlast or ";\n";
 112         local tstart = opt.tstart or "{";
 113         local tend = opt.tend or "}";
 114         local kstart = opt.kstart or "[";
 115         local kend = opt.kend or "]";
 116         local equals = opt.equals or " = ";
 117         local unquoted = opt.unquoted == true and "^[%a_][%w_]*$" or opt.unquoted;
 118         local hex = opt.hex;
 119         local freeze = opt.freeze;
 120         local maxdepth = opt.maxdepth or 127;
 121         local multirefs = opt.multiref;
 122         local table_pairs = opt.table_iterator or rawpairs;
 123
 124         -- serialize one table, recursively
 125         -- t - table being serialized
 126         -- o - array where tokens are added, concatenate to get final result
 127         --   - also used to detect cycles
 128         -- l - position in o of where to insert next token
 129         -- d - depth, used for indentation
 130         local function serialize_table(t, o, l, d)
 131                 if o[t] then
 132                         o[l], l = fallback(t, "table has multiple references"), l + 1;
 133                         return l;
 134                 elseif d > maxdepth then
 135                         o[l], l = fallback(t, "max table depth reached"), l + 1;
 136                         return l;
 137                 end
 138
 139                 -- Keep track of table loops
 140                 local ot = t; -- reference pre-freeze
 141                 o[t] = true;
 142                 o[ot] = true;
 143
 144                 if freeze == true then
 145                         -- opportunity to do pre-serialization
 146                         local mt = getmetatable(t);
 147                         if type(mt) == "table" then
 148                                 local tag = mt.__name;
 149                                 local fr = mt.__freeze;
 150
 151                                 if type(fr) == "function" then
 152                                         t = fr(t);
 153                                         if type(tag) == "string" then
 154                                                 o[l], l = tag, l + 1;
 155                                         end
 156                                 end
 157                         end
 158                 end
 159
 160                 o[l], l = tstart, l + 1;
 161                 local indent = s_rep(indentwith, d);
 162                 local numkey = 1;
 163                 local ktyp, vtyp;
 164                 local had_items = false;
 165                 for k,v in table_pairs(t) do
 166                         had_items = true;
 167                         o[l], l = itemstart, l + 1;
 168                         o[l], l = indent, l + 1;
 169                         ktyp, vtyp = type(k), type(v);
 170                         if k == numkey then
 171                                 -- next index in array part
 172                                 -- assuming that these are found in order
 173                                 numkey = numkey + 1;
 174                         elseif unquoted and ktyp == "string" and
 175                                 not keywords[k] and s_match(k, unquoted) then
 176                                 -- unquoted keys
 177                                 o[l], l = k, l + 1;
 178                                 o[l], l = equals, l + 1;
 179                         else
 180                                 -- quoted keys
 181                                 o[l], l = kstart, l + 1;
 182                                 if ktyp == "table" then
 183                                         l = serialize_table(k, o, l, d+1);
 184                                 else
 185                                         o[l], l = ser(k), l + 1;
 186                                 end
 187                                 -- =
 188                                 o[l], o[l+1], l = kend, equals, l + 2;
 189                         end
 190
 191                         -- the value
 192                         if vtyp == "table" then
 193                                 l = serialize_table(v, o, l, d+1);
 194                         else
 195                                 o[l], l = ser(v), l + 1;
 196                         end
 197                         o[l], l = itemsep, l + 1;
 198                 end
 199                 if had_items then
 200                         o[l - 1] = itemlast;
 201                         o[l], l = s_rep(indentwith, d-1), l + 1;
 202                 end
 203                 o[l], l = tend, l +1;
 204
 205                 if multirefs then
 206                         o[t] = nil;
 207                         o[ot] = nil;
 208                 end
 209
 210                 return l;
 211         end
 212
 213         function types.table(t)
 214                 local o = {};
 215                 serialize_table(t, o, 1, 1);
 216                 return t_concat(o);
 217         end
 218
 219         local function serialize_string(s)
 220                 return '"' .. s_gsub(s, "[%z\1-\31\"\'\\\127-\255]", string_escapes) .. '"';
 221         end
 222
 223         if type(hex) == "string" then
 224                 function types.string(s)
 225                         local esc = serialize_string(s);
 226                         if #esc > (#s*2+2+#hex) then
 227                                 return hex .. '"' .. to_hex(s) .. '"';
 228                         end
 229                         return esc;
 230                 end
 231         else
 232                 types.string = serialize_string;
 233         end
 234
 235         function types.number(t)
 236                 if m_type(t) == "integer" then
 237                         return s_format("%d", t);
 238                 elseif t == pos_inf then
 239                         return "(1/0)";
 240                 elseif t == neg_inf then
 241                         return "(-1/0)";
 242                 elseif t ~= t then
 243                         return "(0/0)";
 244                 end
 245                 return s_format("%.18g", t);
 246         end
 247
 248         -- Are these faster than tostring?
 249         types["nil"] = function()
 250                 return "nil";
 251         end
 252
 253         function types.boolean(t)
 254                 return t and "true" or "false";
 255         end
 256
 257         return ser;
 258 end
 259
 260 local function deserialize(str)
 261         if type(str) ~= "string" then return nil; end
 262         str = "return "..str;
 263         local f, err = envload(str, "=serialized data", {});
 264         if not f then return nil, err; end
 265         local success, ret = pcall(f);
 266         if not success then return nil, ret; end
 267         return ret;
 268 end
 269
 270 local default = new();
 271 return {
 272         new = new;
 273         serialize = function (x, opt)
 274                 if opt == nil then
 275                         return default(x);
 276                 else
 277                         return new(opt)(x);
 278                 end
 279         end;
 280         deserialize = deserialize;
 281 };