Add label processing to disassembler

disasm.py

@@ -21,11 +21,61 @@ import argparse
 
 parser = argparse.ArgumentParser()
 parser.add_argument('file', help='.bin file containing the initial memory dump')
+parser.add_argument('--hints', help='File(s) outlining additional jmp/call targets, label names, comments, etc', action='append')
 args = parser.parse_args()
 
 with open(args.file, 'rb') as data:
 	SYN_MEM = memory_from_file(data)
 
+# Find things to label
+labels, comments_before, comments_inline = {}, {}, {}
+SYN_PTR = 0
+while SYN_PTR < len(SYN_MEM):
+	word = SYN_MEM[SYN_PTR]
+	if word in instructions_by_opcode:
+		instruction, SYN_PTR = Instruction.next_instruction(SYN_MEM, SYN_PTR)
+		if isinstance(instruction, InstructionJmp) or isinstance(instruction, InstructionJt) or isinstance(instruction, InstructionJf):
+			if isinstance(instruction, InstructionJmp):
+				op = instruction.args[0]
+			else:
+				op = instruction.args[1]
+			if isinstance(op, OpLiteral):
+				loc = op.get(None)
+				labels['label_{:04x}'.format(loc)] = loc
+		elif isinstance(instruction, InstructionCall):
+			if isinstance(instruction.args[0], OpLiteral):
+				loc = instruction.args[0].get(None)
+				labels['sub_{:04x}'.format(loc)] = loc
+	else:
+		SYN_PTR += 1
+# Read hints
+if args.hints:
+	for hintfile in args.hints:
+		with open(hintfile, 'r') as data:
+			while True:
+				line = data.readline()
+				if line == '':
+					break
+				if line.startswith('jmp '):
+					loc = int(line.split()[1], 16)
+					labels['label_{:04x}'.format(loc)] = loc
+				elif line.startswith('call '):
+					loc = int(line.split()[1], 16)
+					labels['sub_{:04x}'.format(loc)] = loc
+				elif line.startswith('ren '):
+					old_label = line.split()[1]
+					new_label = line.split()[2]
+					labels[new_label] = labels[old_label]
+					del labels[old_label]
+				elif line.startswith('cmb '):
+					loc = int(line.split()[1], 16)
+					comment = line[line.index(' ', line.index(' ') + 1) + 1:].strip()
+					if loc not in comments_before:
+						comments_before[loc] = []
+					comments_before[loc].append(comment)
+				else:
+					raise Exception('Invalid line in hint file: {}'.format(line))
+
 def escape_char(char):
 	return char.encode('unicode_escape').decode('utf-8').replace('"', '\\"')
 
@@ -35,6 +85,30 @@ MODE_DAT = False #False, 1 (data), 2 (text)
 SYN_PTR = 0
 
 while SYN_PTR < len(SYN_MEM):
+	if SYN_PTR in comments_before:
+		if MODE_OUT:
+			print('"')
+			MODE_OUT = False
+		if MODE_DAT == 1:
+			print()
+			MODE_DAT = False
+		if MODE_DAT == 2:
+			print('"')
+			MODE_DAT = False
+		for comment in comments_before[SYN_PTR]:
+			print('; {}'.format(comment))
+	if any(v == SYN_PTR for k, v in labels.items()):
+		if MODE_OUT:
+			print('"')
+			MODE_OUT = False
+		if MODE_DAT == 1:
+			print()
+			MODE_DAT = False
+		if MODE_DAT == 2:
+			print('"')
+			MODE_DAT = False
+		print('${}:'.format(next(k for k, v in labels.items() if v == SYN_PTR)))
+	
 	word = SYN_MEM[SYN_PTR]
 	
 	if MODE_OUT and word != 19:
@@ -72,7 +146,7 @@ while SYN_PTR < len(SYN_MEM):
 		SYN_PTR += 1
 	else:
 		# Instruction
-		instruction, SYN_PTR = Instruction.next_instruction(SYN_MEM, SYN_PTR)
+		instruction, next_SYN_PTR = Instruction.next_instruction(SYN_MEM, SYN_PTR)
 		# Special cases
 		if isinstance(instruction, InstructionOut):
 			if isinstance(instruction.args[0], OpLiteral):
@@ -89,5 +163,20 @@ while SYN_PTR < len(SYN_MEM):
 					print('"')
 					MODE_OUT = False
 				print('{:04x}: {}'.format(SYN_PTR, instruction.describe()))
+		elif isinstance(instruction, InstructionJmp) or isinstance(instruction, InstructionJt) or isinstance(instruction, InstructionJf) or isinstance(instruction, InstructionCall):
+			if isinstance(instruction, InstructionJmp) or isinstance(instruction, InstructionCall):
+				op = instruction.args[0]
+			else:
+				op = instruction.args[1]
+			if isinstance(op, OpLiteral):
+				loc = op.get(None)
+				if any(v == loc for k, v in labels.items()):
+					label = next(k for k, v in labels.items() if v == loc)
+					print('{:04x}: {: <4} ${}'.format(SYN_PTR, instruction.name, label))
 				else:
 					print('{:04x}: {}'.format(SYN_PTR, instruction.describe()))
+			else:
+				print('{:04x}: {}'.format(SYN_PTR, instruction.describe()))
+		else:
+			print('{:04x}: {}'.format(SYN_PTR, instruction.describe()))
+		SYN_PTR = next_SYN_PTR

disasm_hints/annotations.txt

@@ -0,0 +1,6 @@
+jmp 0000
+ren label_0000 start
+jmp 0110
+ren label_0110 self_test
+cmb 0140 Test jmp
+ren label_015b self_test_jmp1

disasm_hints/jmpcall.txt

@@ -0,0 +1,95 @@
+call 0505
+call 0507
+call 05b2
+call 05ee
+call 05f8
+call 05fb
+call 0607
+call 0623
+call 0634
+call 0645
+call 0653
+call 0670
+call 0683
+call 06bb
+call 06e7
+call 0731
+call 07d1
+call 084d
+call 0865
+call 086e
+call 0882
+call 08c8
+call 08e9
+call 0b94
+call 0cad
+call 0d48
+call 0df0
+call 0e48
+call 0e8f
+call 0e9e
+call 0ec0
+call 0eca
+call 0ede
+call 0f35
+call 0f98
+call 0fa9
+call 0fcd
+call 0fde
+call 1000
+call 1011
+call 1022
+call 1033
+call 1044
+call 107a
+call 10b7
+call 1135
+call 11a3
+call 11b5
+call 1203
+call 1252
+call 1270
+call 12bf
+call 1315
+call 1371
+call 14f0
+call 1501
+call 1512
+call 1523
+call 1534
+call 1545
+call 1659
+call 16b6
+call 16bf
+call 16d6
+call 16f4
+call 1705
+call 1721
+call 174c
+call 1766
+jmp 015b
+jmp 0166
+jmp 01e4
+jmp 02c4
+jmp 02db
+jmp 034d
+jmp 03e8
+jmp 061e
+jmp 06b2
+jmp 06fb
+jmp 07e3
+jmp 07ff
+jmp 0833
+jmp 08bc
+jmp 08cc
+jmp 08f8
+jmp 0aae
+jmp 0b86
+jmp 0c79
+jmp 0cfe
+jmp 0d99
+jmp 0e43
+jmp 112e
+jmp 1312
+jmp 1652
+jmp 1747