Fix Pak2 and add position to Theirsce lib

pak2.py

@@ -15,7 +15,7 @@ class pak2_file:
     char_count: int = 0
     slot_count: int = 0
     image_count: int = 0
-    chunks: pak2_chunks = pak2_chunks()
+    chunks: pak2_chunks = field(default_factory=pak2_chunks)
 
 def get_file_name_noext(path):
     return os.path.splitext(os.path.basename(path))[0]

theirsce.py

@@ -1,3 +1,4 @@
+from itertools import tee, zip_longest
 from typing import Generator
 from FileIO import FileIO
 from dataclasses import dataclass, field
@@ -61,11 +62,13 @@ class Theirsce(FileIO):
         self.seek(start)
 
         while self.tell() < end:
+            opcode = self.read_opcode()
             pos = self.tell()
-            yield self.read_opcode()
+            yield opcode
             self.seek(pos)
 
     def read_opcode(self):
+        pos = self.tell()
         opcode = self.read_uint8()
 
         # Reference Block
@@ -103,11 +106,11 @@ class Theirsce(FileIO):
             else:
                 scope = ReferenceScope.LOCAL
 
-            return TheirsceReferenceInstruction(ref_type=VariableType(var_type), scope=scope, offset=value, shift=shift)
+            return TheirsceReferenceInstruction(ref_type=VariableType(var_type), scope=scope, offset=value, shift=shift, position=pos)
 
         # ALU operations
         elif opcode < 0xC0:
-            return TheirsceAluInstruction(operation = AluOperation(opcode & 0x3F))
+            return TheirsceAluInstruction(operation = AluOperation(opcode & 0x3F), position=pos)
 
         # PUSH block, the amount of bytes depend on encoding
         elif opcode < 0xE0:
@@ -129,29 +132,29 @@ class Theirsce(FileIO):
             # to signed
             value = value | (-(value & 0x80000000))
 
-            return TheirscePushInstruction(value=value)
+            return TheirscePushInstruction(value=value, position=pos)
 
         # CALL block, available commands are at 0x1e5300
         # first entry is number of parameters then function
         elif opcode < 0xF0:
             index = ((opcode & 0xF) << 8) | self.read_uint8()
-            return TheirsceSyscallInstruction(function_index=index, function_name=SYSCALL_NAMES[index])
+            return TheirsceSyscallInstruction(function_index=index, function_name=SYSCALL_NAMES[index], position=pos)
 
         # Flow related block
         elif opcode < 0xF8:
             if opcode == 0xF0:
-                return TheirsceReturnInstruction(is_void=True)
+                return TheirsceReturnInstruction(is_void=True, position=pos)
             elif opcode == 0xF1:
-                return TheirsceReturnInstruction(is_void=False)
+                return TheirsceReturnInstruction(is_void=False, position=pos)
             
             # Need to be offsetted to start of code
             elif opcode >= 0xF2 and opcode < 0xF5:
                 target = self.code_offset + self.read_uint16()
-                return TheirsceBranchInstruction(branch_type=BranchType(opcode - 0xF2), destination=target)
+                return TheirsceBranchInstruction(branch_type=BranchType(opcode - 0xF2), destination=target, position=pos)
             elif opcode == 0xF5:
                 target = self.code_offset + self.read_uint16()
                 reserve = self.read_uint16()
-                return TheirsceLocalCallInstruction(destination=target, reserve=reserve)
+                return TheirsceLocalCallInstruction(destination=target, reserve=reserve, position=pos)
             
             # ?
             elif opcode == 0xF6:
@@ -163,20 +166,20 @@ class Theirsce(FileIO):
                     else:
                         params.append(self.read_uint8())
             
-                return TheirsceAcquireInstruction(params=params,variables=variables)
+                return TheirsceAcquireInstruction(params=params,variables=variables, position=pos)
 
             elif opcode == 0xF7:
                 param = self.read_uint16() # Type?
-                return TheirsceBreakInstruction(param=param)
+                return TheirsceBreakInstruction(param=param, position=pos)
         
         # Get string
         elif opcode < 0xFC:
             value = ((opcode & 3) << 16) | self.read_uint16()
-            return TheirsceStringInstruction(offset=value,text="")
+            return TheirsceStringInstruction(offset=value,text="", position=pos)
 
         # ?
         elif opcode == 0xFE:
-            return TheirsceSpecialReferenceInstruction()
+            return TheirsceSpecialReferenceInstruction(position=pos)
         
         # Impossible
         else:

theirsce_instructions.py

@@ -22,6 +22,7 @@ class BranchType(Enum):
 
 
 class InstructionType(Enum):
+    INVALID    = -1
     ALU        = 0
     PUSH       = 1
     SYSCALL    = 2
@@ -90,6 +91,7 @@ class TheirsceBaseInstruction:
 @dataclass
 class TheirsceAluInstruction(TheirsceBaseInstruction):
     operation: AluOperation
+    position: int
     mnemonic = "ALU"
     type = InstructionType.ALU
 
@@ -100,6 +102,7 @@ class TheirsceAluInstruction(TheirsceBaseInstruction):
 @dataclass
 class TheirscePushInstruction(TheirsceBaseInstruction):
     value: int
+    position: int
     mnemonic = "PUSH"
     type = InstructionType.PUSH
 
@@ -108,6 +111,7 @@ class TheirscePushInstruction(TheirsceBaseInstruction):
 class TheirsceSyscallInstruction(TheirsceBaseInstruction):
     function_index: int
     function_name: str
+    position: int
     mnemonic = "SYSCALL"
     type = InstructionType.SYSCALL
 
@@ -116,6 +120,7 @@ class TheirsceSyscallInstruction(TheirsceBaseInstruction):
 class TheirsceLocalCallInstruction(TheirsceBaseInstruction):
     destination: int
     reserve: int
+    position: int
     mnemonic = "CALL"
     type = InstructionType.LOCAL_CALL
 
@@ -124,6 +129,7 @@ class TheirsceLocalCallInstruction(TheirsceBaseInstruction):
 class TheirsceAcquireInstruction(TheirsceBaseInstruction):
     params: list[int]
     variables: int
+    position: int
     mnemonic = "ACQUIRE"
     type = InstructionType.ACQUIRE
 
@@ -131,6 +137,7 @@ class TheirsceAcquireInstruction(TheirsceBaseInstruction):
 @dataclass
 class TheirsceBreakInstruction(TheirsceBaseInstruction):
     param: int
+    position: int
     mnemonic = "BREAK"
     type = InstructionType.BREAK
 
@@ -139,6 +146,7 @@ class TheirsceBreakInstruction(TheirsceBaseInstruction):
 class TheirsceBranchInstruction(TheirsceBaseInstruction):
     destination: int
     branch_type: BranchType
+    position: int
     mnemonic = ""
     type = InstructionType.BRANCH
 
@@ -149,6 +157,7 @@ class TheirsceBranchInstruction(TheirsceBaseInstruction):
 @dataclass
 class TheirsceReturnInstruction(TheirsceBaseInstruction):
     is_void: bool
+    position: int
     mnemonic = "RETURN"
     type = InstructionType.RETURN
 
@@ -161,6 +170,7 @@ class TheirsceReturnInstruction(TheirsceBaseInstruction):
 class TheirsceStringInstruction(TheirsceBaseInstruction):
     text: str
     offset: int
+    position: int
     mnemonic = "STRING"
     type = InstructionType.STRING
 
@@ -171,11 +181,13 @@ class TheirsceReferenceInstruction(TheirsceBaseInstruction):
     scope: ReferenceScope
     offset: int
     shift:  int
+    position: int
     mnemonic = "REF"
     type = InstructionType.REFERENCE
 
 
 @dataclass
 class TheirsceSpecialReferenceInstruction(TheirsceBaseInstruction):
+    position: int
     mnemonic = "SP_REF"
     type = InstructionType.SP_REF