lp: bug fix (forgot to transfer registers)

- tables can now be (mostly) decompiled
- changed 'decompiled source' to 'pseudo-code' since the output doesn't typically match the compiled script source.
- misc. refactoring

README.md

@@ -12,51 +12,70 @@ Lua has a relatively small instruction set (only 38 different opcodes!). This ma
 
 ```sh
 > cat example.lua && luac5.1 -o example.luac example.lua
-local total = 0
+local tbl = {"He", "llo", " ", "Wo", "rld", "!"}
+local str = ""
 
-for i = 0, 9, 1 do
-    total = total + i
-    print(total)
+for i = 1, #tbl do
+    str = str .. tbl[i]
 end
+
+print(str)
 > python main.py example.luac
 example.luac
 
 ==== [[example.lua's constants]] ====
 
-0: [NUMBER] 0.0
-1: [NUMBER] 9.0
-2: [NUMBER] 1.0
-3: [STRING] print
+0: [STRING] He
+1: [STRING] llo
+2: [STRING]  
+3: [STRING] Wo
+4: [STRING] rld
+5: [STRING] !
+6: [STRING] 
+7: [NUMBER] 1.0
+8: [STRING] print
 
 ==== [[example.lua's locals]] ====
 
-R[0]: total
-R[1]: (for index)
-R[2]: (for limit)
-R[3]: (for step)
-R[4]: i
+R[0]: tbl
+R[1]: str
+R[2]: (for index)
+R[3]: (for limit)
+R[4]: (for step)
+R[5]: i
 
 ==== [[example.lua's dissassembly]] ====
 
-[  0]      LOADK :   R[0]   K[0]               ; load 0.0 into R[0]
-[  1]      LOADK :   R[1]   K[0]               ; load 0.0 into R[1]
-[  2]      LOADK :   R[2]   K[1]               ; load 9.0 into R[2]
-[  3]      LOADK :   R[3]   K[2]               ; load 1.0 into R[3]
-[  4]    FORPREP :   R[1]      4               ; 
-[  5]        ADD :   R[0]   R[0]   R[4]        ; add R[4] to R[0], place into R[0]
-[  6]  GETGLOBAL :   R[5]   K[3]               ; move _G["print"] into R[5]
-[  7]       MOVE :      6      0      0        ; move R[0] into R[6]
-[  8]       CALL :      5      2      1        ; 
-[  9]    FORLOOP :   R[1]     -5               ; 
-[ 10]     RETURN :      0      1      0        ; 
+[  0]   NEWTABLE :      0      6      0        ; 
+[  1]      LOADK :   R[1]   K[0]               ; load "He" into R[1]
+[  2]      LOADK :   R[2]   K[1]               ; load "llo" into R[2]
+[  3]      LOADK :   R[3]   K[2]               ; load " " into R[3]
+[  4]      LOADK :   R[4]   K[3]               ; load "Wo" into R[4]
+[  5]      LOADK :   R[5]   K[4]               ; load "rld" into R[5]
+[  6]      LOADK :   R[6]   K[5]               ; load "!" into R[6]
+[  7]    SETLIST :      0      6      1        ; 
+[  8]      LOADK :   R[1]   K[6]               ; load "" into R[1]
+[  9]      LOADK :   R[2]   K[7]               ; load 1 into R[2]
+[ 10]        LEN :      3      0      0        ; 
+[ 11]      LOADK :   R[4]   K[7]               ; load 1 into R[4]
+[ 12]    FORPREP :   R[2]      3               ; 
+[ 13]       MOVE :      6      1      0        ; move R[1] into R[6]
+[ 14]   GETTABLE :   R[7]      0   R[5]        ; 
+[ 15]     CONCAT :      1      6      7        ; concat 2 values from R[6] to R[7], store into R[1]
+[ 16]    FORLOOP :   R[2]     -4               ; 
+[ 17]  GETGLOBAL :   R[2]   K[8]               ; move _G["print"] into R[2]
+[ 18]       MOVE :      3      1      0        ; move R[1] into R[3]
+[ 19]       CALL :      2      2      1        ; 
+[ 20]     RETURN :      0      1      0        ; 
 
-==== [[example.lua's decompiled source]] ====
+==== [[example.lua's pseudo-code]] ====
 
-local total = 0.0
-for i = 0.0, 9.0, 1.0 do
-    total = (total + i)
-    print(total)
+local tbl = {"He", "llo", " ", "Wo", "rld", "!", }
+local str = ""
+for i = 1, #tbl, 1 do
+    str = str .. tbl[i]
 end
+print(str)
 
 
 ```

lparser.py

@@ -8,7 +8,6 @@
 
 from operator import concat
 from subprocess import call
-from xmlrpc.client import Boolean
 from lundump import Chunk, Constant, Instruction, Opcodes, whichRK, readRKasK
 
 class _Scope:
@@ -30,7 +29,12 @@ class _Line:
         self.scope = scope
 
 def isValidLocal(ident: str) -> bool:
-    for c in ident:
+    # has to start with an alpha or _
+    if ident[0] not in "abcdefghijklmnopqrstuvwxyz_":
+        return False
+
+    # then it can be alphanum or _
+    for c in ident[1:]:
         if c not in "abcdefghijklmnopqrstuvwxyz1234567890_":
             return False
 
@@ -63,7 +67,7 @@ class LuaDecomp:
             # end the scope (if we're supposed too)
             self.__checkScope()
 
-        print("\n==== [[" + str(self.chunk.name) + "'s decompiled source]] ====\n")
+        print("\n==== [[" + str(self.chunk.name) + "'s pseudo-code]] ====\n")
 
         for line in self.lines:
             if self.annotateLines:
@@ -121,7 +125,7 @@ class LuaDecomp:
         self.src = ""
 
     # walks traceback, if local wasn't set before, the local needs to be defined
-    def __needsDefined(self, reg) -> Boolean:
+    def __needsDefined(self, reg) -> bool:
         for _, trace in self.traceback.items():
             if reg in trace.sets:
                 return False
@@ -147,7 +151,7 @@ class LuaDecomp:
         # if the top indx is a local, get it
         return self.locals[indx] if indx in self.locals else self.top[indx]
 
-    def __setReg(self, indx: int, code: str) -> None:
+    def __setReg(self, indx: int, code: str, forceLocal: bool = False) -> None:
         # if the top indx is a local, set it
         if indx in self.locals:
             if self.__needsDefined(indx):
@@ -155,10 +159,9 @@ class LuaDecomp:
             else:
                 self.__addExpr(self.locals[indx] + " = " + code)
                 self.__endStatement()
-        elif self.aggressiveLocals: # 'every register is a local!!'
+        elif self.aggressiveLocals or forceLocal: # 'every register is a local!!'
             self.__newLocal(indx, code)
 
-
         self.__addSetTraceback(indx)
         self.top[indx] = code
 
@@ -202,6 +205,8 @@ class LuaDecomp:
         self.__addExpr("end")
         self.scope.pop()
 
+        self.__endStatement()
+
     # =====================================[[ Instructions ]]======================================
 
     def __emitOperand(self, a: int, b: str, c: str, op: str) -> None:
@@ -256,6 +261,38 @@ class LuaDecomp:
         else:
             return self.__getReg(rk)
 
+    # walk & peak ahead NEWTABLE
+    def __parseNewTable(self, indx: int):
+        # TODO: parse SETTABLE too?
+        tblOps = [Opcodes.LOADK, Opcodes.SETLIST]
+
+        instr = self.__getNextInstr()
+        cachedRegs = {}
+        tbl = "{"
+        while instr.opcode in tblOps:
+            if instr.opcode == Opcodes.LOADK: # operate on registers
+                cachedRegs[instr.A] = self.chunk.getConstant(instr.B).toCode()
+            elif instr.opcode == Opcodes.SETLIST:
+                numElems = instr.B
+
+                for i in range(numElems):
+                    tbl += "%s, " % cachedRegs[instr.A + i + 1]
+                    del cachedRegs[instr.A + i + 1]
+
+            self.pc += 1
+            instr = self.__getNextInstr()
+        tbl += "}"
+
+        # i use forceLocal here even though i don't know *for sure* that the register is a local.
+        # this does help later though if the table is reused (which is 99% of the time). the other 1%
+        # only affects syntax and may look a little weird but is fine and equivalent non-the-less
+        self.__setReg(indx, tbl, forceLocal=True)
+        self.__endStatement()
+
+        # if we have leftovers... oops, set those
+        for i, v in cachedRegs.items():
+            self.__setReg(i, v)
+
     def parseInstr(self):
         instr = self.__getCurrInstr()
 
@@ -280,6 +317,8 @@ class LuaDecomp:
         elif instr.opcode == Opcodes.SETTABLE:
             self.__addExpr(self.__getReg(instr.A) + "[" + self.__readRK(instr.B) + "] = " + self.__readRK(instr.C))
             self.__endStatement()
+        elif instr.opcode == Opcodes.NEWTABLE:
+            self.__parseNewTable(instr.A)
         elif instr.opcode == Opcodes.ADD:
             self.__emitOperand(instr.A, self.__readRK(instr.B), self.__readRK(instr.C), " + ")
         elif instr.opcode == Opcodes.SUB:
@@ -297,7 +336,7 @@ class LuaDecomp:
         elif instr.opcode == Opcodes.NOT:
             self.__setReg(instr.A, "not " + self.__getReg(instr.B))
         elif instr.opcode == Opcodes.LEN:
-            self.__setReg(instr.A, "#" + self.__getCurrInstr(instr.B))
+            self.__setReg(instr.A, "#" + self.__getReg(instr.B))
         elif instr.opcode == Opcodes.CONCAT:
             count = instr.C-instr.B+1
             concatStr = ""
@@ -353,5 +392,21 @@ class LuaDecomp:
         elif instr.opcode == Opcodes.FORPREP:
             self.__addExpr("for %s = %s, %s, %s " % (self.__getLocal(instr.A+3), self.__getReg(instr.A), self.__getReg(instr.A + 1), self.__getReg(instr.A + 2)))
             self.__startScope("do", self.pc, instr.B)
+        elif instr.opcode == Opcodes.SETLIST:
+            # LFIELDS_PER_FLUSH (50) is the number of elements that *should* have been set in the list in the *last* SETLIST
+            # eg.
+            # [ 49]      LOADK :  R[49]   K[1]               ; load 0.0 into R[49]
+            # [ 50]      LOADK :  R[50]   K[1]               ; load 0.0 into R[50]
+            # [ 51]    SETLIST :      0     50      1        ; sets list[1..50]
+            # [ 52]      LOADK :   R[1]   K[1]               ; load 0.0 into R[1]
+            # [ 53]    SETLIST :      0      1      2        ; sets list[51..51]
+            numElems = instr.B
+            startAt = ((instr.C - 1) * 50)
+            ident = self.__getLocal(instr.A)
+
+            # set each index (TODO: make tables less verbose)
+            for i in range(numElems):
+                self.__addExpr("%s[%d] = %s" % (ident, (startAt + i + 1), self.__getReg(instr.A + i + 1)))
+                self.__endStatement()
         else:
             raise Exception("unsupported instruction: %s" % instr.toString())

lundump.py

@@ -152,7 +152,7 @@ class Constant:
         self.data = data
 
     def toString(self):
-        return "[" + self.type.name + "] " + str(self.data)
+        return "[%s] %s" % (self.type.name, str(self.data))
 
     # format the constant so that it is parsable by lua
     def toCode(self):
@@ -164,7 +164,7 @@ class Constant:
             else:
                 return "false"
         elif self.type == ConstType.NUMBER:
-            return str(self.data)
+            return "%g" % self.data
         else:
             return "nil"