util/json.lua

   1 -- Prosody IM
   2 -- Copyright (C) 2008-2010 Matthew Wild
   3 -- Copyright (C) 2008-2010 Waqas Hussain
   4 --
   5 -- utf8char copyright (C) 2007 Rici Lake
   6 --
   7 -- This project is MIT/X11 licensed. Please see the
   8 -- COPYING file in the source package for more information.
   9 --
  10
  11 local type = type;
  12 local t_insert, t_concat, t_remove, t_sort = table.insert, table.concat, table.remove, table.sort;
  13 local s_char = string.char;
  14 local tostring, tonumber = tostring, tonumber;
  15 local pairs, ipairs = pairs, ipairs;
  16 local next = next;
  17 local error = error;
  18 local newproxy, getmetatable = newproxy, getmetatable;
  19 local print = print;
  20
  21 --module("json")
  22 local json = {};
  23
  24 local null = newproxy and newproxy(true) or {};
  25 if getmetatable and getmetatable(null) then
  26         getmetatable(null).__tostring = function() return "null"; end;
  27 end
  28 json.null = null;
  29
  30 local escapes = {
  31         ["\""] = "\\\"", ["\\"] = "\\\\", ["\b"] = "\\b",
  32         ["\f"] = "\\f", ["\n"] = "\\n", ["\r"] = "\\r", ["\t"] = "\\t"};
  33 local unescapes = {
  34         ["\""] = "\"", ["\\"] = "\\", ["/"] = "/",
  35         b = "\b", f = "\f", n = "\n", r = "\r", t = "\t"};
  36 for i=0,31 do
  37         local ch = s_char(i);
  38         if not escapes[ch] then escapes[ch] = ("\\u%.4X"):format(i); end
  39 end
  40
  41 local function utf8char(i)
  42         if i >= 0 then
  43                 i = i - i%1
  44                 if i < 128 then
  45                         return s_char(i)
  46                 else
  47                         local c1 = i % 64
  48                         i = (i - c1) / 64
  49                         if i < 32 then
  50                                 return s_char(0xC0+i, 0x80+c1)
  51                         else
  52                                 local c2 = i % 64
  53                                 i = (i - c2) / 64
  54                                 if i < 16 and (i ~= 13 or c2 < 32) then
  55                                         return s_char(0xE0+i, 0x80+c2, 0x80+c1)
  56                                 elseif i >= 16 and i < 0x110 then
  57                                         local c3 = i % 64
  58                                         i = (i - c3) / 64
  59                                         return s_char(0xF0+i, 0x80+c3, 0x80+c2, 0x80+c1)
  60                                 end
  61                         end
  62                 end
  63         end
  64 end
  65
  66
  67 local valid_types = {
  68         number  = true,
  69         string  = true,
  70         table   = true,
  71         boolean = true
  72 };
  73 local special_keys = {
  74         __array = true;
  75         __hash  = true;
  76 };
  77
  78 local simplesave, tablesave, arraysave, stringsave;
  79
  80 function stringsave(o, buffer)
  81         -- FIXME do proper utf-8 and binary data detection
  82         t_insert(buffer, "\""..(o:gsub(".", escapes)).."\"");
  83 end
  84
  85 function arraysave(o, buffer)
  86         t_insert(buffer, "[");
  87         if next(o) then
  88                 for i,v in ipairs(o) do
  89                         simplesave(v, buffer);
  90                         t_insert(buffer, ",");
  91                 end
  92                 t_remove(buffer);
  93         end
  94         t_insert(buffer, "]");
  95 end
  96
  97 function tablesave(o, buffer)
  98         local __array = {};
  99         local __hash = {};
 100         local hash = {};
 101         for i,v in ipairs(o) do
 102                 __array[i] = v;
 103         end
 104         for k,v in pairs(o) do
 105                 local ktype, vtype = type(k), type(v);
 106                 if valid_types[vtype] or v == null then
 107                         if ktype == "string" and not special_keys[k] then
 108                                 hash[k] = v;
 109                         elseif (valid_types[ktype] or k == null) and __array[k] == nil then
 110                                 __hash[k] = v;
 111                         end
 112                 end
 113         end
 114         if next(__hash) ~= nil or next(hash) ~= nil or next(__array) == nil then
 115                 t_insert(buffer, "{");
 116                 local mark = #buffer;
 117                 if buffer.ordered then
 118                         local keys = {};
 119                         for k in pairs(hash) do
 120                                 t_insert(keys, k);
 121                         end
 122                         t_sort(keys);
 123                         for _,k in ipairs(keys) do
 124                                 stringsave(k, buffer);
 125                                 t_insert(buffer, ":");
 126                                 simplesave(hash[k], buffer);
 127                                 t_insert(buffer, ",");
 128                         end
 129                 else
 130                         for k,v in pairs(hash) do
 131                                 stringsave(k, buffer);
 132                                 t_insert(buffer, ":");
 133                                 simplesave(v, buffer);
 134                                 t_insert(buffer, ",");
 135                         end
 136                 end
 137                 if next(__hash) ~= nil then
 138                         t_insert(buffer, "\"__hash\":[");
 139                         for k,v in pairs(__hash) do
 140                                 simplesave(k, buffer);
 141                                 t_insert(buffer, ",");
 142                                 simplesave(v, buffer);
 143                                 t_insert(buffer, ",");
 144                         end
 145                         t_remove(buffer);
 146                         t_insert(buffer, "]");
 147                         t_insert(buffer, ",");
 148                 end
 149                 if next(__array) then
 150                         t_insert(buffer, "\"__array\":");
 151                         arraysave(__array, buffer);
 152                         t_insert(buffer, ",");
 153                 end
 154                 if mark ~= #buffer then t_remove(buffer); end
 155                 t_insert(buffer, "}");
 156         else
 157                 arraysave(__array, buffer);
 158         end
 159 end
 160
 161 function simplesave(o, buffer)
 162         local t = type(o);
 163         if t == "number" then
 164                 t_insert(buffer, tostring(o));
 165         elseif t == "string" then
 166                 stringsave(o, buffer);
 167         elseif t == "table" then
 168                 tablesave(o, buffer);
 169         elseif t == "boolean" then
 170                 t_insert(buffer, (o and "true" or "false"));
 171         else
 172                 t_insert(buffer, "null");
 173         end
 174 end
 175
 176 function json.encode(obj)
 177         local t = {};
 178         simplesave(obj, t);
 179         return t_concat(t);
 180 end
 181 function json.encode_ordered(obj)
 182         local t = { ordered = true };
 183         simplesave(obj, t);
 184         return t_concat(t);
 185 end
 186 function json.encode_array(obj)
 187         local t = {};
 188         arraysave(obj, t);
 189         return t_concat(t);
 190 end
 191
 192 -----------------------------------
 193
 194
 195 function json.decode(json)
 196         json = json.." "; -- appending a space ensures valid json wouldn't touch EOF
 197         local pos = 1;
 198         local current = {};
 199         local stack = {};
 200         local ch, peek;
 201         local function next()
 202                 ch = json:sub(pos, pos);
 203                 if ch == "" then error("Unexpected EOF"); end
 204                 pos = pos+1;
 205                 peek = json:sub(pos, pos);
 206                 return ch;
 207         end
 208
 209         local function skipwhitespace()
 210                 while ch and (ch == "\r" or ch == "\n" or ch == "\t" or ch == " ") do
 211                         next();
 212                 end
 213         end
 214         local function skiplinecomment()
 215                 repeat next(); until not(ch) or ch == "\r" or ch == "\n";
 216                 skipwhitespace();
 217         end
 218         local function skipstarcomment()
 219                 next(); next(); -- skip '/', '*'
 220                 while peek and ch ~= "*" and peek ~= "/" do next(); end
 221                 if not peek then error("eof in star comment") end
 222                 next(); next(); -- skip '*', '/'
 223                 skipwhitespace();
 224         end
 225         local function skipstuff()
 226                 while true do
 227                         skipwhitespace();
 228                         if ch == "/" and peek == "*" then
 229                                 skipstarcomment();
 230                         elseif ch == "/" and peek == "/" then
 231                                 skiplinecomment();
 232                         else
 233                                 return;
 234                         end
 235                 end
 236         end
 237
 238         local readvalue;
 239         local function readarray()
 240                 local t = {};
 241                 next(); -- skip '['
 242                 skipstuff();
 243                 if ch == "]" then next(); return t; end
 244                 t_insert(t, readvalue());
 245                 while true do
 246                         skipstuff();
 247                         if ch == "]" then next(); return t; end
 248                         if not ch then error("eof while reading array");
 249                         elseif ch == "," then next();
 250                         elseif ch then error("unexpected character in array, comma expected"); end
 251                         if not ch then error("eof while reading array"); end
 252                         t_insert(t, readvalue());
 253                 end
 254         end
 255
 256         local function checkandskip(c)
 257                 local x = ch or "eof";
 258                 if x ~= c then error("unexpected "..x..", '"..c.."' expected"); end
 259                 next();
 260         end
 261         local function readliteral(lit, val)
 262                 for c in lit:gmatch(".") do
 263                         checkandskip(c);
 264                 end
 265                 return val;
 266         end
 267         local function readstring()
 268                 local s = "";
 269                 checkandskip("\"");
 270                 while ch do
 271                         while ch and ch ~= "\\" and ch ~= "\"" do
 272                                 s = s..ch; next();
 273                         end
 274                         if ch == "\\" then
 275                                 next();
 276                                 if unescapes[ch] then
 277                                         s = s..unescapes[ch];
 278                                         next();
 279                                 elseif ch == "u" then
 280                                         local seq = "";
 281                                         for i=1,4 do
 282                                                 next();
 283                                                 if not ch then error("unexpected eof in string"); end
 284                                                 if not ch:match("[0-9a-fA-F]") then error("invalid unicode escape sequence in string"); end
 285                                                 seq = seq..ch;
 286                                         end
 287                                         s = s..utf8char(tonumber(seq, 16));
 288                                         next();
 289                                 else error("invalid escape sequence in string"); end
 290                         end
 291                         if ch == "\"" then
 292                                 next();
 293                                 return s;
 294                         end
 295                 end
 296                 error("eof while reading string");
 297         end
 298         local function readnumber()
 299                 local s = "";
 300                 if ch == "-" then
 301                         s = s..ch; next();
 302                         if not ch:match("[0-9]") then error("number format error"); end
 303                 end
 304                 if ch == "0" then
 305                         s = s..ch; next();
 306                         if ch:match("[0-9]") then error("number format error"); end
 307                 else
 308                         while ch and ch:match("[0-9]") do
 309                                 s = s..ch; next();
 310                         end
 311                 end
 312                 if ch == "." then
 313                         s = s..ch; next();
 314                         if not ch:match("[0-9]") then error("number format error"); end
 315                         while ch and ch:match("[0-9]") do
 316                                 s = s..ch; next();
 317                         end
 318                         if ch == "e" or ch == "E" then
 319                                 s = s..ch; next();
 320                                 if ch == "+" or ch == "-" then
 321                                         s = s..ch; next();
 322                                         if not ch:match("[0-9]") then error("number format error"); end
 323                                         while ch and ch:match("[0-9]") do
 324                                                 s = s..ch; next();
 325                                         end
 326                                 end
 327                         end
 328                 end
 329                 return tonumber(s);
 330         end
 331         local function readmember(t)
 332                 skipstuff();
 333                 local k = readstring();
 334                 skipstuff();
 335                 checkandskip(":");
 336                 t[k] = readvalue();
 337         end
 338         local function fixobject(obj)
 339                 local __array = obj.__array;
 340                 if __array then
 341                         obj.__array = nil;
 342                         for i,v in ipairs(__array) do
 343                                 t_insert(obj, v);
 344                         end
 345                 end
 346                 local __hash = obj.__hash;
 347                 if __hash then
 348                         obj.__hash = nil;
 349                         local k;
 350                         for i,v in ipairs(__hash) do
 351                                 if k ~= nil then
 352                                         obj[k] = v; k = nil;
 353                                 else
 354                                         k = v;
 355                                 end
 356                         end
 357                 end
 358                 return obj;
 359         end
 360         local function readobject()
 361                 local t = {};
 362                 next(); -- skip '{'
 363                 skipstuff();
 364                 if ch == "}" then next(); return t; end
 365                 if not ch then error("eof while reading object"); end
 366                 readmember(t);
 367                 while true do
 368                         skipstuff();
 369                         if ch == "}" then next(); return fixobject(t); end
 370                         if not ch then error("eof while reading object");
 371                         elseif ch == "," then next();
 372                         elseif ch then error("unexpected character in object, comma expected"); end
 373                         if not ch then error("eof while reading object"); end
 374                         readmember(t);
 375                 end
 376         end
 377
 378         function readvalue()
 379                 skipstuff();
 380                 while ch do
 381                         if ch == "{" then
 382                                 return readobject();
 383                         elseif ch == "[" then
 384                                 return readarray();
 385                         elseif ch == "\"" then
 386                                 return readstring();
 387                         elseif ch:match("[%-0-9%.]") then
 388                                 return readnumber();
 389                         elseif ch == "n" then
 390                                 return readliteral("null", null);
 391                         elseif ch == "t" then
 392                                 return readliteral("true", true);
 393                         elseif ch == "f" then
 394                                 return readliteral("false", false);
 395                         else
 396                                 error("invalid character at value start: "..ch);
 397                         end
 398                 end
 399                 error("eof while reading value");
 400         end
 401         next();
 402         return readvalue();
 403 end
 404
 405 function json.test(object)
 406         local encoded = json.encode(object);
 407         local decoded = json.decode(encoded);
 408         local recoded = json.encode(decoded);
 409         if encoded ~= recoded then
 410                 print("FAILED");
 411                 print("encoded:", encoded);
 412                 print("recoded:", recoded);
 413         else
 414                 print(encoded);
 415         end
 416         return encoded == recoded;
 417 end
 418
 419 return json;