| 1 | -- Minichunkspy: Disassemble and reassemble chunks. |
| 2 | -- Copyright M Joonas Pihlaja 2009 |
| 3 | -- MIT license |
| 4 | -- |
| 5 | -- minichunkspy = require"minichunkspy" |
| 6 | -- |
| 7 | -- chunk = string.dump(loadfile"blabla.lua") |
| 8 | -- disassembled_chunk = minichunkspy.disassemble(chunk) |
| 9 | -- chunk = minichunkspy.assemble(disassembled_chunk) |
| 10 | -- assert(minichunkspy.validate(<function or chunk>)) |
| 11 | -- |
| 12 | -- Tested on little-endian 32 and 64 bit platforms. |
| 13 | local string, table, math = string, table, math |
| 14 | local ipairs, setmetatable, type, assert = ipairs, setmetatable, type, assert |
| 15 | local _ = __END_OF_GLOBALS__ |
| 16 | local string_char, string_byte, string_sub = string.char, string.byte, string.sub |
| 17 | local math_frexp, math_ldexp, math_abs = math.frexp, math.ldexp, math.abs |
| 18 | local table_concat = table.concat |
| 19 | local Inf = math.huge |
| 20 | local NaN = Inf - Inf |
| 21 | |
| 22 | local BIG_ENDIAN = false |
| 23 | local SIZEOF_SIZE_T = 4 |
| 24 | local SIZEOF_INT = 4 |
| 25 | local SIZEOF_NUMBER = 8 |
| 26 | |
| 27 | local save_stack = {} |
| 28 | |
| 29 | local function save() |
| 30 | save_stack[#save_stack+1] |
| 31 | = {BIG_ENDIAN, SIZEOF_SIZE_T, SIZEOF_INT, SIZEOF_NUMBER} |
| 32 | end |
| 33 | local function restore () |
| 34 | BIG_ENDIAN, SIZEOF_SIZE_T, SIZEOF_INT, SIZEOF_NUMBER |
| 35 | = unpack(save_stack[#save_stack]) |
| 36 | save_stack[#save_stack] = nil |
| 37 | end |
| 38 | |
| 39 | local function construct (class, self) |
| 40 | return class.new(class, self) |
| 41 | end |
| 42 | |
| 43 | local mt_memo = {} |
| 44 | |
| 45 | local Field = construct{ |
| 46 | new = |
| 47 | function (class, self) |
| 48 | local self = self or {} |
| 49 | local mt = mt_memo[class] or { |
| 50 | __index = class, |
| 51 | __call = construct |
| 52 | } |
| 53 | mt_memo[class] = mt |
| 54 | return setmetatable(self, mt) |
| 55 | end, |
| 56 | } |
| 57 | |
| 58 | local None = Field{ |
| 59 | unpack = function (self, bytes, ix) return nil, ix end, |
| 60 | pack = function (self, val) return "" end |
| 61 | } |
| 62 | |
| 63 | local char_memo = {} |
| 64 | |
| 65 | local function char(n) |
| 66 | local field = char_memo[n] or Field{ |
| 67 | unpack = function (self, bytes, ix) |
| 68 | return string_sub(bytes, ix, ix+n-1), ix+n |
| 69 | end, |
| 70 | pack = function (self, val) return string_sub(val, 1, n) end |
| 71 | } |
| 72 | char_memo[n] = field |
| 73 | return field |
| 74 | end |
| 75 | |
| 76 | local uint8 = Field{ |
| 77 | unpack = function (self, bytes, ix) |
| 78 | return string_byte(bytes, ix, ix), ix+1 |
| 79 | end, |
| 80 | pack = function (self, val) return string_char(val) end |
| 81 | } |
| 82 | |
| 83 | local uint32 = Field{ |
| 84 | unpack = |
| 85 | function (self, bytes, ix) |
| 86 | local a,b,c,d = string_byte(bytes, ix, ix+3) |
| 87 | if BIG_ENDIAN then a,b,c,d = d,c,b,a end |
| 88 | return a + b*256 + c*256^2 + d*256^3, ix+4 |
| 89 | end, |
| 90 | pack = |
| 91 | function (self, val) |
| 92 | assert(type(val) == "number", |
| 93 | "unexpected value type to pack as an uint32") |
| 94 | local a,b,c,d |
| 95 | d = val % 2^32 |
| 96 | a = d % 256; d = (d - a) / 256 |
| 97 | b = d % 256; d = (d - b) / 256 |
| 98 | c = d % 256; d = (d - c) / 256 |
| 99 | if BIG_ENDIAN then a,b,c,d = d,c,b,a end |
| 100 | return string_char(a,b,c,d) |
| 101 | end |
| 102 | } |
| 103 | |
| 104 | local uint64 = Field{ |
| 105 | unpack = |
| 106 | function (self, bytes, ix) |
| 107 | local a = uint32:unpack(bytes, ix) |
| 108 | local b = uint32:unpack(bytes, ix+4) |
| 109 | if BIG_ENDIAN then a,b = b,a end |
| 110 | return a + b*2^32, ix+8 |
| 111 | end, |
| 112 | pack = |
| 113 | function (self, val) |
| 114 | assert(type(val) == "number", |
| 115 | "unexpected value type to pack as an uint64") |
| 116 | local a = val % 2^32 |
| 117 | local b = (val - a) / 2^32 |
| 118 | if BIG_ENDIAN then a,b = b,a end |
| 119 | return uint32:pack(a) .. uint32:pack(b) |
| 120 | end |
| 121 | } |
| 122 | |
| 123 | local function explode_double(bytes, ix) |
| 124 | local a = uint32:unpack(bytes, ix) |
| 125 | local b = uint32:unpack(bytes, ix+4) |
| 126 | if BIG_ENDIAN then a,b = b,a end --XXX: ARM mixed-endian |
| 127 | |
| 128 | local sig_hi = b % 2^20 |
| 129 | local sig_lo = a |
| 130 | local significand = sig_lo + sig_hi*2^32 |
| 131 | |
| 132 | b = (b - sig_hi) / 2^20 |
| 133 | |
| 134 | local biased_exp = b % 2^11 |
| 135 | local sign = b <= biased_exp and 1 or -1 |
| 136 | |
| 137 | --print(sign, significand, biased_exp, "explode") |
| 138 | return sign, biased_exp, significand |
| 139 | end |
| 140 | |
| 141 | local function implode_double(sign, biased_exp, significand) |
| 142 | --print(sign, significand, biased_exp, "implode") |
| 143 | local sig_lo = significand % 2^32 |
| 144 | local sig_hi = (significand - sig_lo) / 2^32 |
| 145 | |
| 146 | local a = sig_lo |
| 147 | local b = ((sign < 0 and 2^11 or 0) + biased_exp)*2^20 + sig_hi |
| 148 | |
| 149 | if BIG_ENDIAN then a,b = b,a end --XXX: ARM mixed-endian |
| 150 | return uint32.pack(nil, a) .. uint32.pack(nil, b) |
| 151 | end |
| 152 | |
| 153 | local function math_sign(x) |
| 154 | if x ~= x then return x end --sign of NaN is NaN |
| 155 | if x == 0 then x = 1/x end --extract sign of zero |
| 156 | return x > 0 and 1 or -1 |
| 157 | end |
| 158 | |
| 159 | local SMALLEST_SUBNORMAL = math_ldexp(1, -1022 - 52) |
| 160 | local SMALLEST_NORMAL = SMALLEST_SUBNORMAL * 2^52 |
| 161 | local LARGEST_SUBNORMAL = math_ldexp(2^52 - 1, -1022 - 52) |
| 162 | local LARGEST_NORMAL = math_ldexp(2^53 - 1, 1023 - 52) |
| 163 | assert(SMALLEST_SUBNORMAL ~= 0.0 and SMALLEST_SUBNORMAL / 2 == 0.0) |
| 164 | assert(LARGEST_NORMAL ~= Inf) |
| 165 | assert(LARGEST_NORMAL * 2 == Inf) |
| 166 | |
| 167 | local double = Field{ |
| 168 | unpack = |
| 169 | function (self, bytes, ix) |
| 170 | local sign, biased_exp, significand = explode_double(bytes, ix) |
| 171 | |
| 172 | local val |
| 173 | if biased_exp == 0 then --subnormal |
| 174 | val = math_ldexp(significand, -1022 - 52) |
| 175 | elseif biased_exp == 2047 then |
| 176 | val = significand == 0 and Inf or NaN --XXX: loses NaN mantissa |
| 177 | else --normal |
| 178 | val = math_ldexp(2^52 + significand, biased_exp - 1023 - 52) |
| 179 | end |
| 180 | val = sign*val |
| 181 | return val, ix+8 |
| 182 | end, |
| 183 | |
| 184 | pack = |
| 185 | function (self, val) |
| 186 | if val ~= val then |
| 187 | return implode_double(1,2047,2^52-1) --XXX: loses NaN mantissa |
| 188 | end |
| 189 | |
| 190 | local sign = math_sign(val) |
| 191 | val = math_abs(val) |
| 192 | |
| 193 | if val == Inf then return implode_double(sign, 2047, 0) end |
| 194 | if val == 0 then return implode_double(sign, 0, 0) end |
| 195 | |
| 196 | local biased_exp, significand |
| 197 | |
| 198 | if val <= LARGEST_SUBNORMAL then |
| 199 | biased_exp = 0 |
| 200 | significand = val / SMALLEST_SUBNORMAL |
| 201 | else |
| 202 | local frac, exp = math_frexp(val) |
| 203 | significand = (2*frac - 1)*2^52 |
| 204 | biased_exp = exp + 1022 |
| 205 | end |
| 206 | return implode_double(sign, biased_exp, significand) |
| 207 | end |
| 208 | } |
| 209 | |
| 210 | local Byte = uint8 |
| 211 | |
| 212 | local IntegralTypes = { |
| 213 | [4] = uint32, |
| 214 | [8] = uint64 |
| 215 | } |
| 216 | |
| 217 | local FloatTypes = { |
| 218 | [4] = float, |
| 219 | [8] = double |
| 220 | } |
| 221 | |
| 222 | local Size_t = Field{ |
| 223 | unpack = function (self, bytes, ix) |
| 224 | return IntegralTypes[SIZEOF_SIZE_T]:unpack(bytes, ix) |
| 225 | end, |
| 226 | pack = function (self, val) |
| 227 | return IntegralTypes[SIZEOF_SIZE_T]:pack(val) |
| 228 | end, |
| 229 | } |
| 230 | |
| 231 | local Integer = Field{ |
| 232 | unpack = function (self, bytes, ix) |
| 233 | return IntegralTypes[SIZEOF_INT]:unpack(bytes, ix) |
| 234 | end, |
| 235 | pack = function (self, val) |
| 236 | return IntegralTypes[SIZEOF_INT]:pack(val) |
| 237 | end, |
| 238 | } |
| 239 | |
| 240 | local Number = Field{ |
| 241 | unpack = function (self, bytes, ix) |
| 242 | return FloatTypes[SIZEOF_NUMBER]:unpack(bytes, ix) |
| 243 | end, |
| 244 | pack = function (self, val) |
| 245 | return FloatTypes[SIZEOF_NUMBER]:pack(val) |
| 246 | end, |
| 247 | } |
| 248 | |
| 249 | -- Opaque types: |
| 250 | local Insn = char(4) |
| 251 | |
| 252 | local Struct = Field{ |
| 253 | unpack = |
| 254 | function (self, bytes, ix) |
| 255 | local val = {} |
| 256 | local i,j = 1,1 |
| 257 | while self[i] do |
| 258 | local field = self[i] |
| 259 | local key = field.name |
| 260 | if not key then key, j = j, j+1 end |
| 261 | --print("unpacking struct field", key, " at index ", ix) |
| 262 | val[key], ix = field:unpack(bytes, ix) |
| 263 | i = i+1 |
| 264 | end |
| 265 | return val, ix |
| 266 | end, |
| 267 | pack = |
| 268 | function (self, val) |
| 269 | local data = {} |
| 270 | local i,j = 1,1 |
| 271 | while self[i] do |
| 272 | local field = self[i] |
| 273 | local key = field.name |
| 274 | if not key then key, j = j, j+1 end |
| 275 | data[i] = field:pack(val[key]) |
| 276 | i = i+1 |
| 277 | end |
| 278 | return table_concat(data) |
| 279 | end |
| 280 | } |
| 281 | |
| 282 | local List = Field{ |
| 283 | unpack = |
| 284 | function (self, bytes, ix) |
| 285 | local len, ix = Integer:unpack(bytes, ix) |
| 286 | local vals = {} |
| 287 | local field = self.type |
| 288 | for i=1,len do |
| 289 | --print("unpacking list field", i, " at index ", ix) |
| 290 | vals[i], ix = field:unpack(bytes, ix) |
| 291 | end |
| 292 | return vals, ix |
| 293 | end, |
| 294 | pack = |
| 295 | function (self, vals) |
| 296 | local len = #vals |
| 297 | local data = { Integer:pack(len) } |
| 298 | local field = self.type |
| 299 | for i=1,len do |
| 300 | data[#data+1] = field:pack(vals[i]) |
| 301 | end |
| 302 | return table_concat(data) |
| 303 | end |
| 304 | } |
| 305 | |
| 306 | local Boolean = Field{ |
| 307 | unpack = |
| 308 | function (self, bytes, ix) |
| 309 | local val, ix = Integer:unpack(bytes, ix) |
| 310 | assert(val == 0 or val == 1, |
| 311 | "unpacked an unexpected value "..val.." for a Boolean") |
| 312 | return val == 1, ix |
| 313 | end, |
| 314 | pack = |
| 315 | function (self, val) |
| 316 | assert(type(val) == "boolean", |
| 317 | "unexpected value type to pack as a Boolean") |
| 318 | return Integer:pack(val and 1 or 0) |
| 319 | end |
| 320 | } |
| 321 | |
| 322 | local String = Field{ |
| 323 | unpack = |
| 324 | function (self, bytes, ix) |
| 325 | local len, ix = Size_t:unpack(bytes, ix) |
| 326 | local val = nil |
| 327 | if len > 0 then |
| 328 | -- len includes trailing nul byte; ignore it |
| 329 | local string_len = len - 1 |
| 330 | val = bytes:sub(ix, ix+string_len-1) |
| 331 | end |
| 332 | return val, ix + len |
| 333 | end, |
| 334 | pack = |
| 335 | function (self, val) |
| 336 | assert(type(val) == "nil" or type(val) == "string", |
| 337 | "unexpected value type to pack as a String") |
| 338 | if val == nil then |
| 339 | return Size_t:pack(0) |
| 340 | end |
| 341 | return Size_t:pack(#val+1) .. val .. "\000" |
| 342 | end |
| 343 | } |
| 344 | |
| 345 | local ChunkHeader = Struct{ |
| 346 | char(4){name = "signature"}, |
| 347 | Byte{name = "version"}, |
| 348 | Byte{name = "format"}, |
| 349 | Byte{name = "endianness"}, |
| 350 | Byte{name = "sizeof_int"}, |
| 351 | Byte{name = "sizeof_size_t"}, |
| 352 | Byte{name = "sizeof_insn"}, |
| 353 | Byte{name = "sizeof_Number"}, |
| 354 | Byte{name = "integral_flag"}, |
| 355 | } |
| 356 | |
| 357 | local ConstantTypes = { |
| 358 | [0] = None, |
| 359 | [1] = Boolean, |
| 360 | [3] = Number, |
| 361 | [4] = String, |
| 362 | } |
| 363 | local Constant = Field{ |
| 364 | unpack = |
| 365 | function (self, bytes, ix) |
| 366 | local t, ix = Byte:unpack(bytes, ix) |
| 367 | local field = ConstantTypes[t] |
| 368 | assert(field, "unknown constant type "..t.." to unpack") |
| 369 | local v, ix = field:unpack(bytes, ix) |
| 370 | if t == 3 then |
| 371 | assert(type(v) == "number") |
| 372 | end |
| 373 | return { |
| 374 | type = t, |
| 375 | value = v |
| 376 | }, ix |
| 377 | end, |
| 378 | pack = |
| 379 | function (self, val) |
| 380 | local t, v = val.type, val.value |
| 381 | return Byte:pack(t) .. ConstantTypes[t]:pack(v) |
| 382 | end |
| 383 | } |
| 384 | |
| 385 | local Local = Struct{ |
| 386 | String{name = "name"}, |
| 387 | Integer{name = "startpc"}, |
| 388 | Integer{name = "endpc"} |
| 389 | } |
| 390 | |
| 391 | local Function = Struct{ |
| 392 | String{name = "name"}, |
| 393 | Integer{name = "line"}, |
| 394 | Integer{name = "last_line"}, |
| 395 | Byte{name = "num_upvalues"}, |
| 396 | Byte{name = "num_parameters"}, |
| 397 | Byte{name = "is_vararg"}, |
| 398 | Byte{name = "max_stack_size"}, |
| 399 | List{name = "insns", type = Insn}, |
| 400 | List{name = "constants", type = Constant}, |
| 401 | List{name = "prototypes", type = nil}, --patch type below |
| 402 | List{name = "source_lines", type = Integer}, |
| 403 | List{name = "locals", type = Local}, |
| 404 | List{name = "upvalues", type = String}, |
| 405 | } |
| 406 | assert(Function[10].name == "prototypes", |
| 407 | "missed the function prototype list") |
| 408 | Function[10].type = Function |
| 409 | |
| 410 | local Chunk = Field{ |
| 411 | unpack = |
| 412 | function (self, bytes, ix) |
| 413 | local chunk = {} |
| 414 | local header, ix = ChunkHeader:unpack(bytes, ix) |
| 415 | assert(header.signature == "\027Lua", "signature check failed") |
| 416 | assert(header.version == 81, "version mismatch") |
| 417 | assert(header.format == 0, "format mismatch") |
| 418 | assert(header.endianness == 0 or |
| 419 | header.endianness == 1, "endianness mismatch") |
| 420 | assert(IntegralTypes[header.sizeof_int], "int size unsupported") |
| 421 | assert(IntegralTypes[header.sizeof_size_t], "size_t size unsupported") |
| 422 | assert(header.sizeof_insn == 4, "insn size unsupported") |
| 423 | assert(FloatTypes[header.sizeof_Number], "number size unsupported") |
| 424 | assert(header.integral_flag == 0, "integral flag mismatch; only floats supported") |
| 425 | |
| 426 | save() |
| 427 | BIG_ENDIAN = header.endianness == 0 |
| 428 | SIZEOF_SIZE_T = header.sizeof_size_t |
| 429 | SIZEOF_INT = header.sizeof_int |
| 430 | SIZEOF_NUMBER = header.sizeof_Number |
| 431 | chunk.header = header |
| 432 | chunk.body, ix = Function:unpack(bytes, ix) |
| 433 | restore() |
| 434 | return chunk, ix |
| 435 | end, |
| 436 | |
| 437 | pack = |
| 438 | function (self, val) |
| 439 | local data |
| 440 | save() |
| 441 | local header = val.header |
| 442 | BIG_ENDIAN = header.endianness == 0 |
| 443 | SIZEOF_SIZE_T = header.sizeof_size_t |
| 444 | SIZEOF_INT = header.sizeof_int |
| 445 | SIZEOF_NUMBER = header.sizeof_Number |
| 446 | data = ChunkHeader:pack(val.header) .. Function:pack(val.body) |
| 447 | restore() |
| 448 | return data |
| 449 | end |
| 450 | } |
| 451 | |
| 452 | local function validate(chunk) |
| 453 | if type(chunk) == "function" then |
| 454 | return validate(string.dump(chunk)) |
| 455 | end |
| 456 | local f = Chunk:unpack(chunk, 1) |
| 457 | local chunk2 = Chunk:pack(f) |
| 458 | |
| 459 | if chunk == chunk2 then return true end |
| 460 | |
| 461 | local i |
| 462 | local len = math.min(#chunk, #chunk2) |
| 463 | for i=1,len do |
| 464 | local a = chunk:sub(i,i) |
| 465 | local b = chunk:sub(i,i) |
| 466 | if a ~= b then |
| 467 | return false, ("chunk roundtripping failed: ".. |
| 468 | "first byte difference at index %d"):format(i) |
| 469 | end |
| 470 | end |
| 471 | return false, ("chunk round tripping failed: ".. |
| 472 | "original length %d vs. %d"):format(#chunk, #chunk2) |
| 473 | end |
| 474 | |
| 475 | return { |
| 476 | disassemble = function (chunk) return Chunk:unpack(chunk, 1) end, |
| 477 | assemble = function (disassembled) return Chunk:pack(disassembled) end, |
| 478 | validate = validate |
| 479 | } |