util/json.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 local type = type;
  10 local t_insert, t_concat, t_remove, t_sort = table.insert, table.concat, table.remove, table.sort;
  11 local s_char = string.char;
  12 local tostring, tonumber = tostring, tonumber;
  13 local pairs, ipairs = pairs, ipairs;
  14 local next = next;
  15 local error = error;
  16 local newproxy, getmetatable, setmetatable = newproxy, getmetatable, setmetatable;
  17 local print = print;
  18
  19 local has_array, array = pcall(require, "util.array");
  20 local array_mt = has_array and getmetatable(array()) or {};
  21
  22 --module("json")
  23 local json = {};
  24
  25 local null = newproxy and newproxy(true) or {};
  26 if getmetatable and getmetatable(null) then
  27         getmetatable(null).__tostring = function() return "null"; end;
  28 end
  29 json.null = null;
  30
  31 local escapes = {
  32         ["\""] = "\\\"", ["\\"] = "\\\\", ["\b"] = "\\b",
  33         ["\f"] = "\\f", ["\n"] = "\\n", ["\r"] = "\\r", ["\t"] = "\\t"};
  34 local unescapes = {
  35         ["\""] = "\"", ["\\"] = "\\", ["/"] = "/",
  36         b = "\b", f = "\f", n = "\n", r = "\r", t = "\t"};
  37 for i=0,31 do
  38         local ch = s_char(i);
  39         if not escapes[ch] then escapes[ch] = ("\\u%.4X"):format(i); end
  40 end
  41
  42 local function codepoint_to_utf8(code)
  43         if code < 0x80 then return s_char(code); end
  44         local bits0_6 = code % 64;
  45         if code < 0x800 then
  46                 local bits6_5 = (code - bits0_6) / 64;
  47                 return s_char(0x80 + 0x40 + bits6_5, 0x80 + bits0_6);
  48         end
  49         local bits0_12 = code % 4096;
  50         local bits6_6 = (bits0_12 - bits0_6) / 64;
  51         local bits12_4 = (code - bits0_12) / 4096;
  52         return s_char(0x80 + 0x40 + 0x20 + bits12_4, 0x80 + bits6_6, 0x80 + bits0_6);
  53 end
  54
  55 local valid_types = {
  56         number  = true,
  57         string  = true,
  58         table   = true,
  59         boolean = true
  60 };
  61 local special_keys = {
  62         __array = true;
  63         __hash  = true;
  64 };
  65
  66 local simplesave, tablesave, arraysave, stringsave;
  67
  68 function stringsave(o, buffer)
  69         -- FIXME do proper utf-8 and binary data detection
  70         t_insert(buffer, "\""..(o:gsub(".", escapes)).."\"");
  71 end
  72
  73 function arraysave(o, buffer)
  74         t_insert(buffer, "[");
  75         if next(o) then
  76                 for i,v in ipairs(o) do
  77                         simplesave(v, buffer);
  78                         t_insert(buffer, ",");
  79                 end
  80                 t_remove(buffer);
  81         end
  82         t_insert(buffer, "]");
  83 end
  84
  85 function tablesave(o, buffer)
  86         local __array = {};
  87         local __hash = {};
  88         local hash = {};
  89         for i,v in ipairs(o) do
  90                 __array[i] = v;
  91         end
  92         for k,v in pairs(o) do
  93                 local ktype, vtype = type(k), type(v);
  94                 if valid_types[vtype] or v == null then
  95                         if ktype == "string" and not special_keys[k] then
  96                                 hash[k] = v;
  97                         elseif (valid_types[ktype] or k == null) and __array[k] == nil then
  98                                 __hash[k] = v;
  99                         end
 100                 end
 101         end
 102         if next(__hash) ~= nil or next(hash) ~= nil or next(__array) == nil then
 103                 t_insert(buffer, "{");
 104                 local mark = #buffer;
 105                 if buffer.ordered then
 106                         local keys = {};
 107                         for k in pairs(hash) do
 108                                 t_insert(keys, k);
 109                         end
 110                         t_sort(keys);
 111                         for _,k in ipairs(keys) do
 112                                 stringsave(k, buffer);
 113                                 t_insert(buffer, ":");
 114                                 simplesave(hash[k], buffer);
 115                                 t_insert(buffer, ",");
 116                         end
 117                 else
 118                         for k,v in pairs(hash) do
 119                                 stringsave(k, buffer);
 120                                 t_insert(buffer, ":");
 121                                 simplesave(v, buffer);
 122                                 t_insert(buffer, ",");
 123                         end
 124                 end
 125                 if next(__hash) ~= nil then
 126                         t_insert(buffer, "\"__hash\":[");
 127                         for k,v in pairs(__hash) do
 128                                 simplesave(k, buffer);
 129                                 t_insert(buffer, ",");
 130                                 simplesave(v, buffer);
 131                                 t_insert(buffer, ",");
 132                         end
 133                         t_remove(buffer);
 134                         t_insert(buffer, "]");
 135                         t_insert(buffer, ",");
 136                 end
 137                 if next(__array) then
 138                         t_insert(buffer, "\"__array\":");
 139                         arraysave(__array, buffer);
 140                         t_insert(buffer, ",");
 141                 end
 142                 if mark ~= #buffer then t_remove(buffer); end
 143                 t_insert(buffer, "}");
 144         else
 145                 arraysave(__array, buffer);
 146         end
 147 end
 148
 149 function simplesave(o, buffer)
 150         local t = type(o);
 151         if t == "number" then
 152                 t_insert(buffer, tostring(o));
 153         elseif t == "string" then
 154                 stringsave(o, buffer);
 155         elseif t == "table" then
 156                 local mt = getmetatable(o);
 157                 if mt == array_mt then
 158                         arraysave(o, buffer);
 159                 else
 160                         tablesave(o, buffer);
 161                 end
 162         elseif t == "boolean" then
 163                 t_insert(buffer, (o and "true" or "false"));
 164         else
 165                 t_insert(buffer, "null");
 166         end
 167 end
 168
 169 function json.encode(obj)
 170         local t = {};
 171         simplesave(obj, t);
 172         return t_concat(t);
 173 end
 174 function json.encode_ordered(obj)
 175         local t = { ordered = true };
 176         simplesave(obj, t);
 177         return t_concat(t);
 178 end
 179 function json.encode_array(obj)
 180         local t = {};
 181         arraysave(obj, t);
 182         return t_concat(t);
 183 end
 184
 185 -----------------------------------
 186
 187
 188 function json.decode(json)
 189         json = json.." "; -- appending a space ensures valid json wouldn't touch EOF
 190         local pos = 1;
 191         local current = {};
 192         local stack = {};
 193         local ch, peek;
 194         local function next()
 195                 ch = json:sub(pos, pos);
 196                 if ch == "" then error("Unexpected EOF"); end
 197                 pos = pos+1;
 198                 peek = json:sub(pos, pos);
 199                 return ch;
 200         end
 201
 202         local function skipwhitespace()
 203                 while ch and (ch == "\r" or ch == "\n" or ch == "\t" or ch == " ") do
 204                         next();
 205                 end
 206         end
 207         local function skiplinecomment()
 208                 repeat next(); until not(ch) or ch == "\r" or ch == "\n";
 209                 skipwhitespace();
 210         end
 211         local function skipstarcomment()
 212                 next(); next(); -- skip '/', '*'
 213                 while peek and ch ~= "*" and peek ~= "/" do next(); end
 214                 if not peek then error("eof in star comment") end
 215                 next(); next(); -- skip '*', '/'
 216                 skipwhitespace();
 217         end
 218         local function skipstuff()
 219                 while true do
 220                         skipwhitespace();
 221                         if ch == "/" and peek == "*" then
 222                                 skipstarcomment();
 223                         elseif ch == "/" and peek == "/" then
 224                                 skiplinecomment();
 225                         else
 226                                 return;
 227                         end
 228                 end
 229         end
 230
 231         local readvalue;
 232         local function readarray()
 233                 local t = setmetatable({}, array_mt);
 234                 next(); -- skip '['
 235                 skipstuff();
 236                 if ch == "]" then next(); return t; end
 237                 t_insert(t, readvalue());
 238                 while true do
 239                         skipstuff();
 240                         if ch == "]" then next(); return t; end
 241                         if not ch then error("eof while reading array");
 242                         elseif ch == "," then next();
 243                         elseif ch then error("unexpected character in array, comma expected"); end
 244                         if not ch then error("eof while reading array"); end
 245                         t_insert(t, readvalue());
 246                 end
 247         end
 248
 249         local function checkandskip(c)
 250                 local x = ch or "eof";
 251                 if x ~= c then error("unexpected "..x..", '"..c.."' expected"); end
 252                 next();
 253         end
 254         local function readliteral(lit, val)
 255                 for c in lit:gmatch(".") do
 256                         checkandskip(c);
 257                 end
 258                 return val;
 259         end
 260         local function readstring()
 261                 local s = {};
 262                 checkandskip("\"");
 263                 while ch do
 264                         while ch and ch ~= "\\" and ch ~= "\"" do
 265                                 t_insert(s, ch); next();
 266                         end
 267                         if ch == "\\" then
 268                                 next();
 269                                 if unescapes[ch] then
 270                                         t_insert(s, unescapes[ch]);
 271                                         next();
 272                                 elseif ch == "u" then
 273                                         local seq = "";
 274                                         for i=1,4 do
 275                                                 next();
 276                                                 if not ch then error("unexpected eof in string"); end
 277                                                 if not ch:match("[0-9a-fA-F]") then error("invalid unicode escape sequence in string"); end
 278                                                 seq = seq..ch;
 279                                         end
 280                                         t_insert(s, codepoint_to_utf8(tonumber(seq, 16)));
 281                                         next();
 282                                 else error("invalid escape sequence in string"); end
 283                         end
 284                         if ch == "\"" then
 285                                 next();
 286                                 return t_concat(s);
 287                         end
 288                 end
 289                 error("eof while reading string");
 290         end
 291         local function readnumber()
 292                 local s = "";
 293                 if ch == "-" then
 294                         s = s..ch; next();
 295                         if not ch:match("[0-9]") then error("number format error"); end
 296                 end
 297                 if ch == "0" then
 298                         s = s..ch; next();
 299                         if ch:match("[0-9]") then error("number format error"); end
 300                 else
 301                         while ch and ch:match("[0-9]") do
 302                                 s = s..ch; next();
 303                         end
 304                 end
 305                 if ch == "." then
 306                         s = s..ch; next();
 307                         if not ch:match("[0-9]") then error("number format error"); end
 308                         while ch and ch:match("[0-9]") do
 309                                 s = s..ch; next();
 310                         end
 311                         if ch == "e" or ch == "E" then
 312                                 s = s..ch; next();
 313                                 if ch == "+" or ch == "-" then
 314                                         s = s..ch; next();
 315                                         if not ch:match("[0-9]") then error("number format error"); end
 316                                         while ch and ch:match("[0-9]") do
 317                                                 s = s..ch; next();
 318                                         end
 319                                 end
 320                         end
 321                 end
 322                 return tonumber(s);
 323         end
 324         local function readmember(t)
 325                 skipstuff();
 326                 local k = readstring();
 327                 skipstuff();
 328                 checkandskip(":");
 329                 t[k] = readvalue();
 330         end
 331         local function fixobject(obj)
 332                 local __array = obj.__array;
 333                 if __array then
 334                         obj.__array = nil;
 335                         for i,v in ipairs(__array) do
 336                                 t_insert(obj, v);
 337                         end
 338                 end
 339                 local __hash = obj.__hash;
 340                 if __hash then
 341                         obj.__hash = nil;
 342                         local k;
 343                         for i,v in ipairs(__hash) do
 344                                 if k ~= nil then
 345                                         obj[k] = v; k = nil;
 346                                 else
 347                                         k = v;
 348                                 end
 349                         end
 350                 end
 351                 return obj;
 352         end
 353         local function readobject()
 354                 local t = {};
 355                 next(); -- skip '{'
 356                 skipstuff();
 357                 if ch == "}" then next(); return t; end
 358                 if not ch then error("eof while reading object"); end
 359                 readmember(t);
 360                 while true do
 361                         skipstuff();
 362                         if ch == "}" then next(); return fixobject(t); end
 363                         if not ch then error("eof while reading object");
 364                         elseif ch == "," then next();
 365                         elseif ch then error("unexpected character in object, comma expected"); end
 366                         if not ch then error("eof while reading object"); end
 367                         readmember(t);
 368                 end
 369         end
 370
 371         function readvalue()
 372                 skipstuff();
 373                 while ch do
 374                         if ch == "{" then
 375                                 return readobject();
 376                         elseif ch == "[" then
 377                                 return readarray();
 378                         elseif ch == "\"" then
 379                                 return readstring();
 380                         elseif ch:match("[%-0-9%.]") then
 381                                 return readnumber();
 382                         elseif ch == "n" then
 383                                 return readliteral("null", null);
 384                         elseif ch == "t" then
 385                                 return readliteral("true", true);
 386                         elseif ch == "f" then
 387                                 return readliteral("false", false);
 388                         else
 389                                 error("invalid character at value start: "..ch);
 390                         end
 391                 end
 392                 error("eof while reading value");
 393         end
 394         next();
 395         return readvalue();
 396 end
 397
 398 function json.test(object)
 399         local encoded = json.encode(object);
 400         local decoded = json.decode(encoded);
 401         local recoded = json.encode(decoded);
 402         if encoded ~= recoded then
 403                 print("FAILED");
 404                 print("encoded:", encoded);
 405                 print("recoded:", recoded);
 406         else
 407                 print(encoded);
 408         end
 409         return encoded == recoded;
 410 end
 411
 412 return json;