Software / code / prosody
Annotate
util/json.lua @ 4284:20979f124ad9
util.json: Fixed handling of truncated JSON.
| author | Waqas Hussain <waqas20@gmail.com> |
|---|---|
| date | Thu, 02 Jun 2011 05:36:15 +0500 |
| parent | 4147:7f119ebcf55f |
| child | 4404:5356664ef9d4 |
| rev | line source |
|---|---|
| 4047 | 1 |
| 2 local type = type; | |
| 3 local t_insert, t_concat, t_remove = table.insert, table.concat, table.remove; | |
| 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 for k,v in pairs(hash) do | |
| 83 stringsave(k, buffer); | |
| 84 t_insert(buffer, ":"); | |
| 85 simplesave(v, buffer); | |
| 86 t_insert(buffer, ","); | |
| 87 end | |
| 88 if next(__hash) ~= nil then | |
| 89 t_insert(buffer, "\"__hash\":["); | |
| 90 for k,v in pairs(__hash) do | |
| 91 simplesave(k, buffer); | |
| 92 t_insert(buffer, ","); | |
| 93 simplesave(v, buffer); | |
| 94 t_insert(buffer, ","); | |
| 95 end | |
| 96 t_remove(buffer); | |
| 97 t_insert(buffer, "]"); | |
| 98 t_insert(buffer, ","); | |
| 99 end | |
| 100 if next(__array) then | |
| 101 t_insert(buffer, "\"__array\":"); | |
| 102 arraysave(__array, buffer); | |
| 103 t_insert(buffer, ","); | |
| 104 end | |
| 105 if mark ~= #buffer then t_remove(buffer); end | |
| 106 t_insert(buffer, "}"); | |
| 107 else | |
| 108 arraysave(__array, buffer); | |
| 109 end | |
| 110 end | |
| 111 | |
| 112 function simplesave(o, buffer) | |
| 113 local t = type(o); | |
| 114 if t == "number" then | |
| 115 t_insert(buffer, tostring(o)); | |
| 116 elseif t == "string" then | |
| 117 stringsave(o, buffer); | |
| 118 elseif t == "table" then | |
| 119 tablesave(o, buffer); | |
| 120 elseif t == "boolean" then | |
| 121 t_insert(buffer, (o and "true" or "false")); | |
| 122 else | |
| 123 t_insert(buffer, "null"); | |
| 124 end | |
| 125 end | |
| 126 | |
| 127 function json.encode(obj) | |
| 128 local t = {}; | |
| 129 simplesave(obj, t); | |
| 130 return t_concat(t); | |
| 131 end | |
| 132 | |
| 133 ----------------------------------- | |
| 134 | |
| 135 | |
| 136 function json.decode(json) | |
|
4284
20979f124ad9
util.json: Fixed handling of truncated JSON.
Waqas Hussain <waqas20@gmail.com>
parents:
4147
diff
changeset
|
137 json = json.." "; -- appending a space ensures valid json wouldn't touch EOF |
| 4047 | 138 local pos = 1; |
| 139 local current = {}; | |
| 140 local stack = {}; | |
| 141 local ch, peek; | |
| 142 local function next() | |
| 143 ch = json:sub(pos, pos); | |
|
4284
20979f124ad9
util.json: Fixed handling of truncated JSON.
Waqas Hussain <waqas20@gmail.com>
parents:
4147
diff
changeset
|
144 if ch == "" then error("Unexpected EOF"); end |
| 4047 | 145 pos = pos+1; |
| 146 peek = json:sub(pos, pos); | |
| 147 return ch; | |
| 148 end | |
| 149 | |
| 150 local function skipwhitespace() | |
| 151 while ch and (ch == "\r" or ch == "\n" or ch == "\t" or ch == " ") do | |
| 152 next(); | |
| 153 end | |
| 154 end | |
| 155 local function skiplinecomment() | |
| 156 repeat next(); until not(ch) or ch == "\r" or ch == "\n"; | |
| 157 skipwhitespace(); | |
| 158 end | |
| 159 local function skipstarcomment() | |
| 160 next(); next(); -- skip '/', '*' | |
| 161 while peek and ch ~= "*" and peek ~= "/" do next(); end | |
| 162 if not peek then error("eof in star comment") end | |
| 163 next(); next(); -- skip '*', '/' | |
| 164 skipwhitespace(); | |
| 165 end | |
| 166 local function skipstuff() | |
| 167 while true do | |
| 168 skipwhitespace(); | |
| 169 if ch == "/" and peek == "*" then | |
| 170 skipstarcomment(); | |
| 171 elseif ch == "/" and peek == "*" then | |
| 172 skiplinecomment(); | |
| 173 else | |
| 174 return; | |
| 175 end | |
| 176 end | |
| 177 end | |
| 178 | |
| 179 local readvalue; | |
| 180 local function readarray() | |
| 181 local t = {}; | |
| 182 next(); -- skip '[' | |
| 183 skipstuff(); | |
| 184 if ch == "]" then next(); return t; end | |
| 185 t_insert(t, readvalue()); | |
| 186 while true do | |
| 187 skipstuff(); | |
| 188 if ch == "]" then next(); return t; end | |
| 189 if not ch then error("eof while reading array"); | |
| 190 elseif ch == "," then next(); | |
| 191 elseif ch then error("unexpected character in array, comma expected"); end | |
| 192 if not ch then error("eof while reading array"); end | |
| 193 t_insert(t, readvalue()); | |
| 194 end | |
| 195 end | |
| 196 | |
| 197 local function checkandskip(c) | |
| 198 local x = ch or "eof"; | |
| 199 if x ~= c then error("unexpected "..x..", '"..c.."' expected"); end | |
| 200 next(); | |
| 201 end | |
| 202 local function readliteral(lit, val) | |
| 203 for c in lit:gmatch(".") do | |
| 204 checkandskip(c); | |
| 205 end | |
| 206 return val; | |
| 207 end | |
| 208 local function readstring() | |
| 209 local s = ""; | |
| 210 checkandskip("\""); | |
| 211 while ch do | |
| 212 while ch and ch ~= "\\" and ch ~= "\"" do | |
| 213 s = s..ch; next(); | |
| 214 end | |
| 215 if ch == "\\" then | |
| 216 next(); | |
| 217 if unescapes[ch] then | |
| 218 s = s..unescapes[ch]; | |
| 219 next(); | |
| 220 elseif ch == "u" then | |
| 221 local seq = ""; | |
| 222 for i=1,4 do | |
| 223 next(); | |
| 224 if not ch then error("unexpected eof in string"); end | |
| 225 if not ch:match("[0-9a-fA-F]") then error("invalid unicode escape sequence in string"); end | |
| 226 seq = seq..ch; | |
| 227 end | |
| 228 s = s..s.char(tonumber(seq, 16)); -- FIXME do proper utf-8 | |
| 229 next(); | |
| 230 else error("invalid escape sequence in string"); end | |
| 231 end | |
| 232 if ch == "\"" then | |
| 233 next(); | |
| 234 return s; | |
| 235 end | |
| 236 end | |
| 237 error("eof while reading string"); | |
| 238 end | |
| 239 local function readnumber() | |
| 240 local s = ""; | |
| 241 if ch == "-" then | |
| 242 s = s..ch; next(); | |
| 243 if not ch:match("[0-9]") then error("number format error"); end | |
| 244 end | |
| 245 if ch == "0" then | |
| 246 s = s..ch; next(); | |
| 247 if ch:match("[0-9]") then error("number format error"); end | |
| 248 else | |
| 249 while ch and ch:match("[0-9]") do | |
| 250 s = s..ch; next(); | |
| 251 end | |
| 252 end | |
| 253 if ch == "." then | |
| 254 s = s..ch; next(); | |
| 255 if not ch:match("[0-9]") then error("number format error"); end | |
| 256 while ch and ch:match("[0-9]") do | |
| 257 s = s..ch; next(); | |
| 258 end | |
| 259 if ch == "e" or ch == "E" then | |
| 260 s = s..ch; next(); | |
| 261 if ch == "+" or ch == "-" then | |
| 262 s = s..ch; next(); | |
| 263 if not ch:match("[0-9]") then error("number format error"); end | |
| 264 while ch and ch:match("[0-9]") do | |
| 265 s = s..ch; next(); | |
| 266 end | |
| 267 end | |
| 268 end | |
| 269 end | |
| 270 return tonumber(s); | |
| 271 end | |
| 272 local function readmember(t) | |
|
4147
7f119ebcf55f
util.json: Fixed a parse error caused by unexpected whitespace.
Waqas Hussain <waqas20@gmail.com>
parents:
4047
diff
changeset
|
273 skipstuff(); |
| 4047 | 274 local k = readstring(); |
|
4147
7f119ebcf55f
util.json: Fixed a parse error caused by unexpected whitespace.
Waqas Hussain <waqas20@gmail.com>
parents:
4047
diff
changeset
|
275 skipstuff(); |
| 4047 | 276 checkandskip(":"); |
| 277 t[k] = readvalue(); | |
| 278 end | |
| 279 local function fixobject(obj) | |
| 280 local __array = obj.__array; | |
| 281 if __array then | |
| 282 obj.__array = nil; | |
| 283 for i,v in ipairs(__array) do | |
| 284 t_insert(obj, v); | |
| 285 end | |
| 286 end | |
| 287 local __hash = obj.__hash; | |
| 288 if __hash then | |
| 289 obj.__hash = nil; | |
| 290 local k; | |
| 291 for i,v in ipairs(__hash) do | |
| 292 if k ~= nil then | |
| 293 obj[k] = v; k = nil; | |
| 294 else | |
| 295 k = v; | |
| 296 end | |
| 297 end | |
| 298 end | |
| 299 return obj; | |
| 300 end | |
| 301 local function readobject() | |
| 302 local t = {}; | |
| 303 next(); -- skip '{' | |
| 304 skipstuff(); | |
| 305 if ch == "}" then next(); return t; end | |
| 306 if not ch then error("eof while reading object"); end | |
| 307 readmember(t); | |
| 308 while true do | |
| 309 skipstuff(); | |
| 310 if ch == "}" then next(); return fixobject(t); end | |
| 311 if not ch then error("eof while reading object"); | |
| 312 elseif ch == "," then next(); | |
| 313 elseif ch then error("unexpected character in object, comma expected"); end | |
| 314 if not ch then error("eof while reading object"); end | |
| 315 readmember(t); | |
| 316 end | |
| 317 end | |
| 318 | |
| 319 function readvalue() | |
| 320 skipstuff(); | |
| 321 while ch do | |
| 322 if ch == "{" then | |
| 323 return readobject(); | |
| 324 elseif ch == "[" then | |
| 325 return readarray(); | |
| 326 elseif ch == "\"" then | |
| 327 return readstring(); | |
| 328 elseif ch:match("[%-0-9%.]") then | |
| 329 return readnumber(); | |
| 330 elseif ch == "n" then | |
| 331 return readliteral("null", null); | |
| 332 elseif ch == "t" then | |
| 333 return readliteral("true", true); | |
| 334 elseif ch == "f" then | |
| 335 return readliteral("false", false); | |
| 336 else | |
| 337 error("invalid character at value start: "..ch); | |
| 338 end | |
| 339 end | |
| 340 error("eof while reading value"); | |
| 341 end | |
| 342 next(); | |
| 343 return readvalue(); | |
| 344 end | |
| 345 | |
| 346 function json.test(object) | |
| 347 local encoded = json.encode(object); | |
| 348 local decoded = json.decode(encoded); | |
| 349 local recoded = json.encode(decoded); | |
| 350 if encoded ~= recoded then | |
| 351 print("FAILED"); | |
| 352 print("encoded:", encoded); | |
| 353 print("recoded:", recoded); | |
| 354 else | |
| 355 print(encoded); | |
| 356 end | |
| 357 return encoded == recoded; | |
| 358 end | |
| 359 | |
| 360 return json; |
