| 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 bit platforms. Modify |
| 13 | -- the Size_t type to be a 64 bit integer to make it work |
| 14 | -- for 64 bit systems, and set BIG_ENDIAN = true for |
| 15 | -- big-endian systems. |
| 16 | local string, table, math = string, table, math |
| 17 | local ipairs, setmetatable, type, assert = ipairs, setmetatable, type, assert |
| 18 | local _ = __END_OF_GLOBALS__ |
| 19 | local string_char, string_byte, string_sub = string.char, string.byte, string.sub |
| 20 | local table_concat = table.concat |
| 21 | local math_abs, math_ldexp, math_frexp = math.abs, math.ldexp, math.frexp |
| 22 | local Inf = math.huge |
| 23 | local Nan = Inf - Inf |
| 24 | |
| 25 | local BIG_ENDIAN = false --twiddle this for your platform. |
| 26 | |
| 27 | local function construct (class, ...) |
| 28 | return class.new(class, ...) |
| 29 | end |
| 30 | |
| 31 | local mt_memo = {} |
| 32 | |
| 33 | local Field = construct{ |
| 34 | new = |
| 35 | function (class, self) |
| 36 | local self = self or {} |
| 37 | local mt = mt_memo[class] or { |
| 38 | __index = class, |
| 39 | __call = construct |
| 40 | } |
| 41 | mt_memo[class] = mt |
| 42 | return setmetatable(self, mt) |
| 43 | end, |
| 44 | } |
| 45 | |
| 46 | local None = Field{ |
| 47 | unpack = function (self, bytes, ix) return nil, ix end, |
| 48 | pack = function (self, val) return "" end |
| 49 | } |
| 50 | |
| 51 | local char_memo = {} |
| 52 | |
| 53 | local function char(n) |
| 54 | local field = char_memo[n] or Field{ |
| 55 | unpack = function (self, bytes, ix) |
| 56 | return string_sub(bytes, ix, ix+n-1), ix+n |
| 57 | end, |
| 58 | pack = function (self, val) return string_sub(val, 1, n) end |
| 59 | } |
| 60 | char_memo[n] = field |
| 61 | return field |
| 62 | end |
| 63 | |
| 64 | local uint8 = Field{ |
| 65 | unpack = function (self, bytes, ix) |
| 66 | return string_byte(bytes, ix, ix), ix+1 |
| 67 | end, |
| 68 | pack = function (self, val) return string_char(val) end |
| 69 | } |
| 70 | |
| 71 | local uint32 = Field{ |
| 72 | unpack = |
| 73 | function (self, bytes, ix) |
| 74 | local a,b,c,d = string_byte(bytes, ix, ix+3) |
| 75 | if BIG_ENDIAN then a,b,c,d = d,c,b,a end |
| 76 | return a + b*256 + c*256^2 + d*256^3, ix+4 |
| 77 | end, |
| 78 | pack = |
| 79 | function (self, val) |
| 80 | assert(type(val) == "number", |
| 81 | "unexpected value type to pack as an uint32") |
| 82 | local a,b,c,d |
| 83 | d = val % 2^32 |
| 84 | a = d % 256; d = (d - a) / 256 |
| 85 | b = d % 256; d = (d - b) / 256 |
| 86 | c = d % 256; d = (d - c) / 256 |
| 87 | if BIG_ENDIAN then a,b,c,d = d,c,b,a end |
| 88 | return string_char(a,b,c,d) |
| 89 | end |
| 90 | } |
| 91 | |
| 92 | local int32 = uint32{ |
| 93 | unpack = function (self, bytes, ix) |
| 94 | local val, ix = uint32:unpack(bytes, ix) |
| 95 | return val < 2^32 and val or (val - 2^31), ix |
| 96 | end |
| 97 | } |
| 98 | |
| 99 | local Byte = uint8 |
| 100 | local Size_t = uint32 |
| 101 | local Integer = int32 |
| 102 | |
| 103 | -- Opaque types: |
| 104 | local Number = char(8) |
| 105 | local Insn = char(4) |
| 106 | |
| 107 | local Struct = Field{ |
| 108 | unpack = |
| 109 | function (self, bytes, ix) |
| 110 | local val = {} |
| 111 | local i,j = 1,1 |
| 112 | while self[i] do |
| 113 | local field = self[i] |
| 114 | local key = field.name |
| 115 | if not key then key, j = j, j+1 end |
| 116 | --print("unpacking struct field", key, " at index ", ix) |
| 117 | val[key], ix = field:unpack(bytes, ix) |
| 118 | i = i+1 |
| 119 | end |
| 120 | return val, ix |
| 121 | end, |
| 122 | pack = |
| 123 | function (self, val) |
| 124 | local data = {} |
| 125 | local i,j = 1,1 |
| 126 | while self[i] do |
| 127 | local field = self[i] |
| 128 | local key = field.name |
| 129 | if not key then key, j = j, j+1 end |
| 130 | data[i] = field:pack(val[key]) |
| 131 | i = i+1 |
| 132 | end |
| 133 | return table_concat(data) |
| 134 | end |
| 135 | } |
| 136 | |
| 137 | local List = Field{ |
| 138 | unpack = |
| 139 | function (self, bytes, ix) |
| 140 | local len, ix = Integer:unpack(bytes, ix) |
| 141 | local vals = {} |
| 142 | local field = self.type |
| 143 | for i=1,len do |
| 144 | --print("unpacking list field", i, " at index ", ix) |
| 145 | vals[i], ix = field:unpack(bytes, ix) |
| 146 | end |
| 147 | return vals, ix |
| 148 | end, |
| 149 | pack = |
| 150 | function (self, vals) |
| 151 | local len = #vals |
| 152 | local data = { Integer:pack(len) } |
| 153 | local field = self.type |
| 154 | for i=1,len do |
| 155 | data[#data+1] = field:pack(vals[i]) |
| 156 | end |
| 157 | return table_concat(data) |
| 158 | end |
| 159 | } |
| 160 | |
| 161 | local Boolean = Field{ |
| 162 | unpack = |
| 163 | function (self, bytes, ix) |
| 164 | local val, ix = Integer:unpack(bytes, ix) |
| 165 | assert(val == 0 or val == 1, |
| 166 | "unpacked an unexpected value "..val.." for a Boolean") |
| 167 | return val == 1, ix |
| 168 | end, |
| 169 | pack = |
| 170 | function (self, val) |
| 171 | assert(type(val) == "boolean", |
| 172 | "unexpected value type to pack as a Boolean") |
| 173 | return Integer:pack(val and 1 or 0) |
| 174 | end |
| 175 | } |
| 176 | |
| 177 | local String = Field{ |
| 178 | unpack = |
| 179 | function (self, bytes, ix) |
| 180 | local len, ix = Integer:unpack(bytes, ix) |
| 181 | local val = nil |
| 182 | if len > 0 then |
| 183 | -- len includes trailing nul byte; ignore it |
| 184 | local string_len = len - 1 |
| 185 | val = bytes:sub(ix, ix+string_len-1) |
| 186 | end |
| 187 | return val, ix + len |
| 188 | end, |
| 189 | pack = |
| 190 | function (self, val) |
| 191 | assert(type(val) == "nil" or type(val) == "string", |
| 192 | "unexpected value type to pack as a String") |
| 193 | if val == nil then |
| 194 | return Integer:pack(0) |
| 195 | end |
| 196 | return Integer:pack(#val+1) .. val .. "\000" |
| 197 | end |
| 198 | } |
| 199 | |
| 200 | local ChunkHeader = Struct{ |
| 201 | char(4){name = "signature"}, |
| 202 | Byte{name = "version"}, |
| 203 | Byte{name = "format"}, |
| 204 | Byte{name = "endianness"}, |
| 205 | Byte{name = "sizeof_int"}, |
| 206 | Byte{name = "sizeof_size_t"}, |
| 207 | Byte{name = "sizeof_insn"}, |
| 208 | Byte{name = "sizeof_Number"}, |
| 209 | Byte{name = "integral_flag"}, |
| 210 | } |
| 211 | |
| 212 | local ConstantTypes = { |
| 213 | [0] = None, |
| 214 | [1] = Boolean, |
| 215 | [3] = Number, |
| 216 | [4] = String, |
| 217 | } |
| 218 | local Constant = Field{ |
| 219 | unpack = |
| 220 | function (self, bytes, ix) |
| 221 | local t, ix = Byte:unpack(bytes, ix) |
| 222 | local field = ConstantTypes[t] |
| 223 | assert(field, "unknown constant type "..t.." to unpack") |
| 224 | local v, ix = field:unpack(bytes, ix) |
| 225 | return { |
| 226 | type = t, |
| 227 | value = v |
| 228 | }, ix |
| 229 | end, |
| 230 | pack = |
| 231 | function (self, val) |
| 232 | local t, v = val.type, val.value |
| 233 | return Byte:pack(t) .. ConstantTypes[t]:pack(v) |
| 234 | end |
| 235 | } |
| 236 | |
| 237 | local Local = Struct{ |
| 238 | String{name = "name"}, |
| 239 | Integer{name = "startpc"}, |
| 240 | Integer{name = "endpc"} |
| 241 | } |
| 242 | |
| 243 | local Function = Struct{ |
| 244 | String{name = "name"}, |
| 245 | Integer{name = "line"}, |
| 246 | Integer{name = "last_line"}, |
| 247 | Byte{name = "num_upvalues"}, |
| 248 | Byte{name = "num_parameters"}, |
| 249 | Byte{name = "is_vararg"}, |
| 250 | Byte{name = "max_stack_size"}, |
| 251 | List{name = "insns", type = Insn}, |
| 252 | List{name = "constants", type = Constant}, |
| 253 | List{name = "prototypes", type = nil}, --patch type below |
| 254 | List{name = "source_lines", type = Integer}, |
| 255 | List{name = "locals", type = Local}, |
| 256 | List{name = "upvalues", type = String}, |
| 257 | } |
| 258 | assert(Function[10].name == "prototypes", |
| 259 | "missed the function prototype list") |
| 260 | Function[10].type = Function |
| 261 | |
| 262 | local Chunk = Struct{ |
| 263 | ChunkHeader{name = "header"}, |
| 264 | Function{name = "body"} |
| 265 | } |
| 266 | |
| 267 | local function validate(chunk) |
| 268 | if type(chunk) == "function" then |
| 269 | return validate(string.dump(chunk)) |
| 270 | end |
| 271 | local f = Chunk:unpack(chunk, 1) |
| 272 | local chunk2 = Chunk:pack(f) |
| 273 | |
| 274 | if chunk == chunk2 then return true end |
| 275 | |
| 276 | local i |
| 277 | local len = math.min(#chunk, #chunk2) |
| 278 | for i=1,len do |
| 279 | local a = chunk:sub(i,i) |
| 280 | local b = chunk:sub(i,i) |
| 281 | if a ~= b then |
| 282 | return false, ("chunk roundtripping failed: ".. |
| 283 | "first byte difference at index %d"):format(i) |
| 284 | end |
| 285 | end |
| 286 | return false, ("chunk round tripping failed: ".. |
| 287 | "original length %d vs. %d"):format(#chunk, #chunk2) |
| 288 | end |
| 289 | |
| 290 | return { |
| 291 | disassemble = function (chunk) return Chunk:unpack(chunk, 1) end, |
| 292 | assemble = function (disassembled) return Chunk:pack(disassembled) end, |
| 293 | validate = validate |
| 294 | } |