Should now work properly

.gitignore

@@ -58,4 +58,7 @@ target/
 
 # Visual Studio
 *.pyproj
-*.sln
+*.sln
+
+# Custom
+1_M3_houseD_1F.txt

README.md

@@ -1,2 +1,3 @@
 # XDscriptTools
 XD script tools (disassembler, etc...)
+Requires Python 3

XDscriptDisassembler.py

@@ -2,14 +2,22 @@
 
 from XDscriptLib import *
 import argparse
+import sys
+import os
+
 
 if __name__ == '__main__':
+	if sys.version_info[0] < 3:
+		raise RuntimeError("Python 3 required")
+
 	parser = argparse.ArgumentParser()
 	parser.add_argument("files", help="XD script files to disassemble", nargs='+', type=str)
 	args = parser.parse_args()
+	fnames = args.files
 
-	for fname in args.files:
+	for fname in fnames:
 		with open(fname, "rb") as f:
-			out_fname = fname.split('.')[0]+".txt"
+			out_fname = os.path.splitext(fname)[0]+'.txt'
+			contents = f.read()
 			with open(out_fname, "w") as out_f:
-				out_f.write(str(ScriptCtx(f.read())))
+				out_f.write(str(ScriptCtx(contents)))

XDscriptLib/_Instruction.py

@@ -9,6 +9,7 @@ class Instruction(object):
 	Instruction -- represents a XD script instruction
 	
 	For the PAL version, the script interpreter is at 8023D83C
+	
 	ctx: ScriptCtx (attribute)
 	label: string
 	position: int, nextPosition
@@ -19,7 +20,7 @@ class Instruction(object):
 	parameter: 16-bit signed integer, not always used
 	
 	**Opcodes**: 
-		See below. int, float, str, and vector are the only 'copyable' types, ldcpvar is used for them.
+		See below. int, float, str, and vector are the only 'copyable' types, ldvar is used for them.
 		For all other existing types, ldncpvar is used (if ldncpvar is followed by setvar, the source variable is deleted)
 
 
@@ -29,6 +30,17 @@ class Instruction(object):
 	**Types**:
 		Refer to ScriptVar
 
+	**Stack**:
+		It is limited to 256 elements. Each task has its own stack. See FunctionInfo for further details.
+
+	**Tasks**:
+		The XD script engine is based on multitasking. 8 tasks max., either synchronous (blocking, and in the same thread),
+		or asynchronous (in a different thread). NB: GS uses its own threading system.
+
+		Task signature: either:
+			4 ints, 
+			or a character (for character interaction handlers)
+
 	Properties:
 	
 		- subSubOpcodes: 
@@ -54,7 +66,7 @@ class Instruction(object):
 	"""
 
 	instructionNames = ("nop", "operator", "ldimm",
-	"ldcpvar", "setvar", "setvector",
+	"ldvar", "setvar", "setvector",
 	"pop", "call", "return",
 	"callstd", "jmptrue", "jmpfalse",
 	"jmp", "reserve", "release",
@@ -198,7 +210,7 @@ def check(self):
 
 
 		elif self._opcode in (3, 4, 17):
-		    # ldcpvar, setvar, ldncpvar
+		    # ldvar, setvar, ldncpvar
 			self.checkVariable()
 			if self.subSubOpcodes[1] == 3 and self._opcode == 4:
 				warnings.warn("cannot change immutable reference (instruction #{0})".format(hex(self.position)))
@@ -223,6 +235,10 @@ def check(self):
 					self.name, hex(instrID)))
 
 				else:
+					if self._opcode == 7 and instrID not in self.ctx.sections["HEAD"].functionOffsets:
+						warnings.warn("call to unreferenced function ({0} \
+						at instruction #{1})".format(instrID, hex(self.position)))
+
 					lbl = ''.join(["sub_" if self._opcode == 7 else "loc_", hex(instrID)[2:]])
 					if not self.ctx.sections["CODE"].labels[instrID]: self.ctx.sections["CODE"].labels[instrID] = lbl
 					if self._opcode != 7 and (True if self.ctx is None 
@@ -366,7 +382,7 @@ def __str__(self):
 			else:
 				instrstr = "{0}, {1}".format(self._subOpcode, self._parameter & 0xffff)
 
-		elif self._opcode in (3, 4, 17):  # ldcpvar, setvar, ldncpvar
+		elif self._opcode in (3, 4, 17):  # ldvar, setvar, ldncpvar
 			instrstr = self.variableName
 
 		elif self._opcode == 5:  # setvector

XDscriptLib/_ScriptCtx.py

@@ -55,7 +55,8 @@ def parseFTBLSection(self):
 		for i in range(sec.nbElems):
 			code_off = struct.unpack_from(">I", sec.data, 8*i)[0]
 			nm_off = struct.unpack_from(">I", sec.data, 4 + 8*i)[0] - 0x20
-			if 0 > nm_off or nm_off + 0x20 >= len(sec.data): continue 
+			if 0 > nm_off or nm_off >= len(sec.data): 
+				continue 
 			nm = sec.data[nm_off:sec.data.find(b'\x00', nm_off)].decode('sjis')
 			sec.functionTable.append((code_off, nm))
 
@@ -114,6 +115,9 @@ def parseARRYSection(self):
 	def load(self, src):
 		if src[:4] != b'TCOD': warnings.warn("Apparently not a XD script file!")
 		self.totalSize = struct.unpack_from(">I", src, 4)[0]
+		if self.totalSize != len(src):
+			warnings.warn("len(src) is different from the script stored total size")
+
 		self.loadSections(src)
 		self.parseFTBLSection()
 		self.parseHEADSection()
@@ -126,11 +130,18 @@ def load(self, src):
 
 		ftbl = self.sections.get("FTBL")
 		code = self.sections["CODE"]
+		head = self.sections["HEAD"]
+
+
 		if ftbl is not None:
+			if not ftbl.nbElems == head.nbElems == code.nbElems:
+				warnings.warn("Inconsistent number of functions between FTBL, HEAD, and CODE")
 			for (off, nm) in ftbl.functionTable:
 				code.labels[off] = nm
+		elif head.nbElems != code.nbElems:
+			warnings.warn("Inconsistent number of functions between HEAD, and CODE")
 
-		entryPoint = self.sections["HEAD"].valueOffset
+		entryPoint = head.valueOffset
 		if not code.labels[entryPoint]: code.labels[entryPoint] = "__start" 
 
 	def __init__(self, src): self.load(src)
@@ -156,7 +167,9 @@ def __str__(self):
 
 		out.write('.section "HEAD":\n')
 		out.write('\t.set __ENTRY_POINT__, {0}\n'.format(code.labels[head.valueOffset]))
-		for off in head.functionOffsets: out.write('\t.function {0}\n'.format(code.labels[off]))
+		for off in head.functionOffsets:
+			if not code.labels[off]: code.labels[off] = 'sub_{0}'.format(hex(off))
+			out.write('\t.function {0}\n'.format(code.labels[off]))
 		out.write('\n')
 
 		out.write('.section "CODE":\n')
@@ -166,11 +179,11 @@ def __str__(self):
 			separator_printed = False
 			if (ftbl is not None and code.labels[instr.position] in [nm for (off, nm) in ftbl.functionTable])\
 			or code.labels[instr.position][:4] == 'sub_':
-				 out.write('\n\n;=================SUBROUTINE===================\n')
+				 out.write('\n\n;=============================SUBROUTINE==============================\n')
 				 separator_printed = True
 
 			elif code.labels[instr.position][:4] == 'loc_':
-				out.write(';----------------------------------------------\n')
+				out.write(';---------------------------------------------------------------------\n')
 				separator_printed = True
 
 			if code.labels[instr.position]:
@@ -186,7 +199,7 @@ def __str__(self):
 
 		if strg is not None:
 			out.write('.section "STRG":\n')
-			splt = strg.stringContents.split('\x00')
+			splt = strg.stringContents.rstrip('\x00').split('\x00')
 			if splt != ['']:
 				for s in splt:
 					out.write('\t"{0}",\n'.format(s))

XDscriptLib/_ScriptVar.py

@@ -20,7 +20,7 @@ class ScriptVar(object):
 	"""Script variable:
 	struct XDscriptVar {
 		s16 type; // 0x00 -- 0x01, s16
-		// 0x02--0x03: padding
+		0x02--0x03: unknown (unsused by the interpreter)
 		union{
 			s32 asInt;
 			float asFloat;