util/json.lua

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