aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--stdlib/json.lua520
1 files changed, 249 insertions, 271 deletions
diff --git a/stdlib/json.lua b/stdlib/json.lua
index 711ef78..61b895e 100644
--- a/stdlib/json.lua
+++ b/stdlib/json.lua
@@ -31,111 +31,102 @@ local json = { _version = "0.1.2" }
31local encode 31local encode
32 32
33local escape_char_map = { 33local escape_char_map = {
34 [ "\\" ] = "\\", 34 [ "\\" ] = "\\",
35 [ "\"" ] = "\"", 35 [ "\"" ] = "\"",
36 [ "\b" ] = "b", 36 [ "\b" ] = "b",
37 [ "\f" ] = "f", 37 [ "\f" ] = "f",
38 [ "\n" ] = "n", 38 [ "\n" ] = "n",
39 [ "\r" ] = "r", 39 [ "\r" ] = "r",
40 [ "\t" ] = "t", 40 [ "\t" ] = "t",
41} 41}
42 42
43local escape_char_map_inv = { [ "/" ] = "/" } 43local escape_char_map_inv = { [ "/" ] = "/" }
44for k, v in pairs(escape_char_map) do 44for k, v in pairs(escape_char_map) do
45 escape_char_map_inv[v] = k 45 escape_char_map_inv[v] = k
46end 46end
47 47
48
49local function escape_char(c) 48local function escape_char(c)
50 return "\\" .. (escape_char_map[c] or string.format("u%04x", c:byte())) 49 return "\\" .. (escape_char_map[c] or string.format("u%04x", c:byte()))
51end 50end
52 51
53
54local function encode_nil(val) 52local function encode_nil(val)
55 return "null" 53 return "null"
56end 54end
57 55
58
59local function encode_table(val, stack) 56local function encode_table(val, stack)
60 local res = {} 57 local res = {}
61 stack = stack or {} 58 stack = stack or {}
62 59
63 -- Circular reference? 60 -- Circular reference?
64 if stack[val] then error("circular reference") end 61 if stack[val] then error("circular reference") end
65 62
66 stack[val] = true 63 stack[val] = true
67 64
68 if rawget(val, 1) ~= nil or next(val) == nil then 65 if rawget(val, 1) ~= nil or next(val) == nil then
69 -- Treat as array -- check keys are valid and it is not sparse 66 -- Treat as array -- check keys are valid and it is not sparse
70 local n = 0 67 local n = 0
71 for k in pairs(val) do 68 for k in pairs(val) do
72 if type(k) ~= "number" then 69 if type(k) ~= "number" then
73 error("invalid table: mixed or invalid key types") 70 error("invalid table: mixed or invalid key types")
74 end 71 end
75 n = n + 1 72 n = n + 1
76 end 73 end
77 if n ~= #val then 74 if n ~= #val then
78 error("invalid table: sparse array") 75 error("invalid table: sparse array")
79 end 76 end
80 -- Encode 77 -- Encode
81 for i, v in ipairs(val) do 78 for i, v in ipairs(val) do
82 table.insert(res, encode(v, stack)) 79 table.insert(res, encode(v, stack))
83 end 80 end
84 stack[val] = nil 81 stack[val] = nil
85 return "[" .. table.concat(res, ",") .. "]" 82 return "[" .. table.concat(res, ",") .. "]"
86 83
87 else 84 else
88 -- Treat as an object 85 -- Treat as an object
89 for k, v in pairs(val) do 86 for k, v in pairs(val) do
90 if type(k) ~= "string" then 87 if type(k) ~= "string" then
91 error("invalid table: mixed or invalid key types") 88 error("invalid table: mixed or invalid key types")
92 end 89 end
93 table.insert(res, encode(k, stack) .. ":" .. encode(v, stack)) 90 table.insert(res, encode(k, stack) .. ":" .. encode(v, stack))
94 end 91 end
95 stack[val] = nil 92 stack[val] = nil
96 return "{" .. table.concat(res, ",") .. "}" 93 return "{" .. table.concat(res, ",") .. "}"
97 end 94 end
98end 95end
99 96
100
101local function encode_string(val) 97local function encode_string(val)
102 return '"' .. val:gsub('[%z\1-\31\\"]', escape_char) .. '"' 98 return '"' .. val:gsub('[%z\1-\31\\"]', escape_char) .. '"'
103end 99end
104 100
105
106local function encode_number(val) 101local function encode_number(val)
107 -- Check for NaN, -inf and inf 102 -- Check for NaN, -inf and inf
108 if val ~= val or val <= -math.huge or val >= math.huge then 103 if val ~= val or val <= -math.huge or val >= math.huge then
109 error("unexpected number value '" .. tostring(val) .. "'") 104 error("unexpected number value '" .. tostring(val) .. "'")
110 end 105 end
111 return string.format("%.14g", val) 106 return string.format("%.14g", val)
112end 107end
113 108
114
115local type_func_map = { 109local type_func_map = {
116 [ "nil" ] = encode_nil, 110 [ "nil" ] = encode_nil,
117 [ "table" ] = encode_table, 111 [ "table" ] = encode_table,
118 [ "string" ] = encode_string, 112 [ "string" ] = encode_string,
119 [ "number" ] = encode_number, 113 [ "number" ] = encode_number,
120 [ "boolean" ] = tostring, 114 [ "boolean" ] = tostring,
121} 115}
122 116
123
124encode = function(val, stack) 117encode = function(val, stack)
125 local t = type(val) 118 local t = type(val)
126 local f = type_func_map[t] 119 local f = type_func_map[t]
127 if f then 120 if f then
128 return f(val, stack) 121 return f(val, stack)
129 end 122 end
130 error("unexpected type '" .. t .. "'") 123 error("unexpected type '" .. t .. "'")
131end 124end
132 125
133
134function json.encode(val) 126function json.encode(val)
135 return ( encode(val) ) 127 return ( encode(val) )
136end 128end
137 129
138
139------------------------------------------------------------------------------- 130-------------------------------------------------------------------------------
140-- Decode 131-- Decode
141------------------------------------------------------------------------------- 132-------------------------------------------------------------------------------
@@ -143,11 +134,11 @@ end
143local parse 134local parse
144 135
145local function create_set(...) 136local function create_set(...)
146 local res = {} 137 local res = {}
147 for i = 1, select("#", ...) do 138 for i = 1, select("#", ...) do
148 res[ select(i, ...) ] = true 139 res[ select(i, ...) ] = true
149 end 140 end
150 return res 141 return res
151end 142end
152 143
153local space_chars = create_set(" ", "\t", "\r", "\n") 144local space_chars = create_set(" ", "\t", "\r", "\n")
@@ -156,233 +147,220 @@ local escape_chars = create_set("\\", "/", '"', "b", "f", "n", "r", "t", "u")
156local literals = create_set("true", "false", "null") 147local literals = create_set("true", "false", "null")
157 148
158local literal_map = { 149local literal_map = {
159 [ "true" ] = true, 150 [ "true" ] = true,
160 [ "false" ] = false, 151 [ "false" ] = false,
161 [ "null" ] = nil, 152 [ "null" ] = nil,
162} 153}
163 154
164
165local function next_char(str, idx, set, negate) 155local function next_char(str, idx, set, negate)
166 for i = idx, #str do 156 for i = idx, #str do
167 if set[str:sub(i, i)] ~= negate then 157 if set[str:sub(i, i)] ~= negate then
168 return i 158 return i
169 end 159 end
170 end 160 end
171 return #str + 1 161 return #str + 1
172end 162end
173 163
174
175local function decode_error(str, idx, msg) 164local function decode_error(str, idx, msg)
176 local line_count = 1 165 local line_count = 1
177 local col_count = 1 166 local col_count = 1
178 for i = 1, idx - 1 do 167 for i = 1, idx - 1 do
179 col_count = col_count + 1 168 col_count = col_count + 1
180 if str:sub(i, i) == "\n" then 169 if str:sub(i, i) == "\n" then
181 line_count = line_count + 1 170 line_count = line_count + 1
182 col_count = 1 171 col_count = 1
183 end 172 end
184 end 173 end
185 error( string.format("%s at line %d col %d", msg, line_count, col_count) ) 174 error( string.format("%s at line %d col %d", msg, line_count, col_count) )
186end 175end
187 176
188
189local function codepoint_to_utf8(n) 177local function codepoint_to_utf8(n)
190 -- http://scripts.sil.org/cms/scripts/page.php?site_id=nrsi&id=iws-appendixa 178 -- http://scripts.sil.org/cms/scripts/page.php?site_id=nrsi&id=iws-appendixa
191 local f = math.floor 179 local f = math.floor
192 if n <= 0x7f then 180 if n <= 0x7f then
193 return string.char(n) 181 return string.char(n)
194 elseif n <= 0x7ff then 182 elseif n <= 0x7ff then
195 return string.char(f(n / 64) + 192, n % 64 + 128) 183 return string.char(f(n / 64) + 192, n % 64 + 128)
196 elseif n <= 0xffff then 184 elseif n <= 0xffff then
197 return string.char(f(n / 4096) + 224, f(n % 4096 / 64) + 128, n % 64 + 128) 185 return string.char(f(n / 4096) + 224, f(n % 4096 / 64) + 128, n % 64 + 128)
198 elseif n <= 0x10ffff then 186 elseif n <= 0x10ffff then
199 return string.char(f(n / 262144) + 240, f(n % 262144 / 4096) + 128, 187 return string.char(f(n / 262144) + 240, f(n % 262144 / 4096) + 128,
200 f(n % 4096 / 64) + 128, n % 64 + 128) 188 f(n % 4096 / 64) + 128, n % 64 + 128)
201 end 189 end
202 error( string.format("invalid unicode codepoint '%x'", n) ) 190 error( string.format("invalid unicode codepoint '%x'", n) )
203end 191end
204 192
205
206local function parse_unicode_escape(s) 193local function parse_unicode_escape(s)
207 local n1 = tonumber( s:sub(1, 4), 16 ) 194 local n1 = tonumber( s:sub(1, 4), 16 )
208 local n2 = tonumber( s:sub(7, 10), 16 ) 195 local n2 = tonumber( s:sub(7, 10), 16 )
209 -- Surrogate pair? 196 -- Surrogate pair?
210 if n2 then 197 if n2 then
211 return codepoint_to_utf8((n1 - 0xd800) * 0x400 + (n2 - 0xdc00) + 0x10000) 198 return codepoint_to_utf8((n1 - 0xd800) * 0x400 + (n2 - 0xdc00) + 0x10000)
212 else 199 else
213 return codepoint_to_utf8(n1) 200 return codepoint_to_utf8(n1)
214 end 201 end
215end 202end
216 203
217
218local function parse_string(str, i) 204local function parse_string(str, i)
219 local res = "" 205 local res = ""
220 local j = i + 1 206 local j = i + 1
221 local k = j 207 local k = j
222 208
223 while j <= #str do 209 while j <= #str do
224 local x = str:byte(j) 210 local x = str:byte(j)
225 211
226 if x < 32 then 212 if x < 32 then
227 decode_error(str, j, "control character in string") 213 decode_error(str, j, "control character in string")
228 214
229 elseif x == 92 then -- `\`: Escape 215 elseif x == 92 then -- `\`: Escape
230 res = res .. str:sub(k, j - 1) 216 res = res .. str:sub(k, j - 1)
231 j = j + 1 217 j = j + 1
232 local c = str:sub(j, j) 218 local c = str:sub(j, j)
233 if c == "u" then 219 if c == "u" then
234 local hex = str:match("^[dD][89aAbB]%x%x\\u%x%x%x%x", j + 1) 220 local hex = str:match("^[dD][89aAbB]%x%x\\u%x%x%x%x", j + 1)
235 or str:match("^%x%x%x%x", j + 1) 221 or str:match("^%x%x%x%x", j + 1)
236 or decode_error(str, j - 1, "invalid unicode escape in string") 222 or decode_error(str, j - 1, "invalid unicode escape in string")
237 res = res .. parse_unicode_escape(hex) 223 res = res .. parse_unicode_escape(hex)
238 j = j + #hex 224 j = j + #hex
239 else 225 else
240 if not escape_chars[c] then 226 if not escape_chars[c] then
241 decode_error(str, j - 1, "invalid escape char '" .. c .. "' in string") 227 decode_error(str, j - 1, "invalid escape char '" .. c .. "' in string")
242 end 228 end
243 res = res .. escape_char_map_inv[c] 229 res = res .. escape_char_map_inv[c]
244 end 230 end
245 k = j + 1 231 k = j + 1
246 232
247 elseif x == 34 then -- `"`: End of string 233 elseif x == 34 then -- `"`: End of string
248 res = res .. str:sub(k, j - 1) 234 res = res .. str:sub(k, j - 1)
249 return res, j + 1 235 return res, j + 1
250 end 236 end
251 237
252 j = j + 1 238 j = j + 1
253 end 239 end
254 240
255 decode_error(str, i, "expected closing quote for string") 241 decode_error(str, i, "expected closing quote for string")
256end 242end
257 243
258
259local function parse_number(str, i) 244local function parse_number(str, i)
260 local x = next_char(str, i, delim_chars) 245 local x = next_char(str, i, delim_chars)
261 local s = str:sub(i, x - 1) 246 local s = str:sub(i, x - 1)
262 local n = tonumber(s) 247 local n = tonumber(s)
263 if not n then 248 if not n then
264 decode_error(str, i, "invalid number '" .. s .. "'") 249 decode_error(str, i, "invalid number '" .. s .. "'")
265 end 250 end
266 return n, x 251 return n, x
267end 252end
268 253
269
270local function parse_literal(str, i) 254local function parse_literal(str, i)
271 local x = next_char(str, i, delim_chars) 255 local x = next_char(str, i, delim_chars)
272 local word = str:sub(i, x - 1) 256 local word = str:sub(i, x - 1)
273 if not literals[word] then 257 if not literals[word] then
274 decode_error(str, i, "invalid literal '" .. word .. "'") 258 decode_error(str, i, "invalid literal '" .. word .. "'")
275 end 259 end
276 return literal_map[word], x 260 return literal_map[word], x
277end 261end
278 262
279
280local function parse_array(str, i) 263local function parse_array(str, i)
281 local res = {} 264 local res = {}
282 local n = 1 265 local n = 1
283 i = i + 1 266 i = i + 1
284 while 1 do 267 while 1 do
285 local x 268 local x
286 i = next_char(str, i, space_chars, true) 269 i = next_char(str, i, space_chars, true)
287 -- Empty / end of array? 270 -- Empty / end of array?
288 if str:sub(i, i) == "]" then 271 if str:sub(i, i) == "]" then
289 i = i + 1 272 i = i + 1
290 break 273 break
291 end 274 end
292 -- Read token 275 -- Read token
293 x, i = parse(str, i) 276 x, i = parse(str, i)
294 res[n] = x 277 res[n] = x
295 n = n + 1 278 n = n + 1
296 -- Next token 279 -- Next token
297 i = next_char(str, i, space_chars, true) 280 i = next_char(str, i, space_chars, true)
298 local chr = str:sub(i, i) 281 local chr = str:sub(i, i)
299 i = i + 1 282 i = i + 1
300 if chr == "]" then break end 283 if chr == "]" then break end
301 if chr ~= "," then decode_error(str, i, "expected ']' or ','") end 284 if chr ~= "," then decode_error(str, i, "expected ']' or ','") end
302 end 285 end
303 return res, i 286 return res, i
304end 287end
305 288
306
307local function parse_object(str, i) 289local function parse_object(str, i)
308 local res = {} 290 local res = {}
309 i = i + 1 291 i = i + 1
310 while 1 do 292 while 1 do
311 local key, val 293 local key, val
312 i = next_char(str, i, space_chars, true) 294 i = next_char(str, i, space_chars, true)
313 -- Empty / end of object? 295 -- Empty / end of object?
314 if str:sub(i, i) == "}" then 296 if str:sub(i, i) == "}" then
315 i = i + 1 297 i = i + 1
316 break 298 break
317 end 299 end
318 -- Read key 300 -- Read key
319 if str:sub(i, i) ~= '"' then 301 if str:sub(i, i) ~= '"' then
320 decode_error(str, i, "expected string for key") 302 decode_error(str, i, "expected string for key")
321 end 303 end
322 key, i = parse(str, i) 304 key, i = parse(str, i)
323 -- Read ':' delimiter 305 -- Read ':' delimiter
324 i = next_char(str, i, space_chars, true) 306 i = next_char(str, i, space_chars, true)
325 if str:sub(i, i) ~= ":" then 307 if str:sub(i, i) ~= ":" then
326 decode_error(str, i, "expected ':' after key") 308 decode_error(str, i, "expected ':' after key")
327 end 309 end
328 i = next_char(str, i + 1, space_chars, true) 310 i = next_char(str, i + 1, space_chars, true)
329 -- Read value 311 -- Read value
330 val, i = parse(str, i) 312 val, i = parse(str, i)
331 -- Set 313 -- Set
332 res[key] = val 314 res[key] = val
333 -- Next token 315 -- Next token
334 i = next_char(str, i, space_chars, true) 316 i = next_char(str, i, space_chars, true)
335 local chr = str:sub(i, i) 317 local chr = str:sub(i, i)
336 i = i + 1 318 i = i + 1
337 if chr == "}" then break end 319 if chr == "}" then break end
338 if chr ~= "," then decode_error(str, i, "expected '}' or ','") end 320 if chr ~= "," then decode_error(str, i, "expected '}' or ','") end
339 end 321 end
340 return res, i 322 return res, i
341end 323end
342 324
343
344local char_func_map = { 325local char_func_map = {
345 [ '"' ] = parse_string, 326 [ '"' ] = parse_string,
346 [ "0" ] = parse_number, 327 [ "0" ] = parse_number,
347 [ "1" ] = parse_number, 328 [ "1" ] = parse_number,
348 [ "2" ] = parse_number, 329 [ "2" ] = parse_number,
349 [ "3" ] = parse_number, 330 [ "3" ] = parse_number,
350 [ "4" ] = parse_number, 331 [ "4" ] = parse_number,
351 [ "5" ] = parse_number, 332 [ "5" ] = parse_number,
352 [ "6" ] = parse_number, 333 [ "6" ] = parse_number,
353 [ "7" ] = parse_number, 334 [ "7" ] = parse_number,
354 [ "8" ] = parse_number, 335 [ "8" ] = parse_number,
355 [ "9" ] = parse_number, 336 [ "9" ] = parse_number,
356 [ "-" ] = parse_number, 337 [ "-" ] = parse_number,
357 [ "t" ] = parse_literal, 338 [ "t" ] = parse_literal,
358 [ "f" ] = parse_literal, 339 [ "f" ] = parse_literal,
359 [ "n" ] = parse_literal, 340 [ "n" ] = parse_literal,
360 [ "[" ] = parse_array, 341 [ "[" ] = parse_array,
361 [ "{" ] = parse_object, 342 [ "{" ] = parse_object,
362} 343}
363 344
364
365parse = function(str, idx) 345parse = function(str, idx)
366 local chr = str:sub(idx, idx) 346 local chr = str:sub(idx, idx)
367 local f = char_func_map[chr] 347 local f = char_func_map[chr]
368 if f then 348 if f then
369 return f(str, idx) 349 return f(str, idx)
370 end 350 end
371 decode_error(str, idx, "unexpected character '" .. chr .. "'") 351 decode_error(str, idx, "unexpected character '" .. chr .. "'")
372end 352end
373 353
374
375function json.decode(str) 354function json.decode(str)
376 if type(str) ~= "string" then 355 if type(str) ~= "string" then
377 error("expected argument of type string, got " .. type(str)) 356 error("expected argument of type string, got " .. type(str))
378 end 357 end
379 local res, idx = parse(str, next_char(str, 1, space_chars, true)) 358 local res, idx = parse(str, next_char(str, 1, space_chars, true))
380 idx = next_char(str, idx, space_chars, true) 359 idx = next_char(str, idx, space_chars, true)
381 if idx <= #str then 360 if idx <= #str then
382 decode_error(str, idx, "trailing garbage") 361 decode_error(str, idx, "trailing garbage")
383 end 362 end
384 return res 363 return res
385end 364end
386 365
387
388return json 366return json