Add built-in string support in dis/assembler

disasm.py

@@ -30,7 +30,7 @@ with open(args.file, 'rb') as data:
 	SYN_MEM = memory_from_file(data)
 	disassemble_end = len(SYN_MEM)
 
-labels, comments_before, comments_inline, replacements = {}, {}, {}, {}
+labels, comments_before, comments_inline, replacements, strings = {}, {}, {}, {}, []
 
 # Do smart things if requested
 if args.smart:
@@ -99,11 +99,18 @@ if args.hints:
 					code = line[line.index(' ', line.index(' ') + 1) + 1:].strip()
 					instruction = assemble_line(None, code)[0][0]
 					replacements[loc] = instruction
+				elif line.startswith('del '):
+					loc = int(line.split()[1], 16)
+					replacements[loc] = None
+				elif line.startswith('str '):
+					loc = int(line.split()[1], 16)
+					strings.append(loc)
 				else:
 					raise Exception('Invalid line in hint file: {}'.format(line))
 
 MODE_OUT = False
-MODE_DAT = False #False, 1 (data), 2 (text)
+MODE_DAT = False #False, 1 (data), 2 (text), 3 (unknown string), 4 (data string), 5 (text string)
+str_ctr = None
 
 SYN_PTR = 0
 
@@ -124,7 +131,7 @@ def set_mode_dat(mode):
 		pass
 	elif mode == False:
 		# Switching off
-		if MODE_DAT == 2:
+		if MODE_DAT == 2 or MODE_DAT == 5:
 			print('"', end='')
 		print()
 	elif MODE_DAT == 1:
@@ -135,11 +142,20 @@ def set_mode_dat(mode):
 		# Switching from text to data
 		print('"')
 		print('{:04x}: data'.format(SYN_PTR), end='')
-	else:
+	elif mode == 4:
+		# Detected data string
+		pass
+	elif mode == 5:
+		# Detected text string
+		print(' "', end='')
+	elif mode == 1 or mode == 2:
 		# Switching to a new mode
 		print('{:04x}: data'.format(SYN_PTR), end='')
 		if mode == 2:
 			print('"', end='')
+	elif mode == 3:
+		# Switching to a new string mode
+		print('{:04x}: str'.format(SYN_PTR), end='')
 	MODE_DAT = mode
 def clear_modes():
 	set_mode_out(False)
@@ -164,13 +180,41 @@ while SYN_PTR < len(SYN_MEM):
 	# Handle replacements
 	if SYN_PTR in replacements:
 		instruction = replacements[SYN_PTR]
+		if instruction is not None:
 			print('{:04x}: {}{}'.format(SYN_PTR, instruction.describe(), comment_inline))
 			SYN_PTR += len(instruction.assemble(None))
 		continue
 	
 	word = SYN_MEM[SYN_PTR]
 	
-	if SYN_PTR >= disassemble_end or word not in instructions_by_opcode:
+	if SYN_PTR in strings:
+		# String length
+		set_mode_dat(3)
+		str_ctr = word
+		SYN_PTR += 1
+	elif MODE_DAT == 3:
+		# Detect string type
+		if 32 <= word <= 126:
+			set_mode_dat(5)
+			print(escape_char(chr(word)), end='')
+		else:
+			set_mode_dat(4)
+			print(' {:04x}'.format(word), end='')
+		SYN_PTR += 1
+		str_ctr -= 1
+		if str_ctr <= 0:
+			set_mode_dat(False)
+	elif MODE_DAT == 4 or MODE_DAT == 5:
+		# String
+		if MODE_DAT == 4:
+			print(' {:04x}'.format(word), end='')
+		else:
+			print(escape_char(chr(word)), end='')
+		SYN_PTR += 1
+		str_ctr -= 1
+		if str_ctr <= 0:
+			set_mode_dat(False)
+	elif SYN_PTR >= disassemble_end or word not in instructions_by_opcode:
 		# Data
 		if 32 <= word <= 126:
 			# Looks like letters - unfortunately "\n" looks like MULT

disasm_hints/annotations.txt

@@ -73,6 +73,12 @@ ren label_059d self_test_mod_fail
 ren label_034d self_test_rwmem
 lbl 034b self_test_rwmem_data
 cmb 034b Test rmem/wmem
+rep 034d rmem R0 $self_test_rwmem_data
+rep 0357 add  R2 $self_test_rwmem_data 0001
+rep 0365 set  R0 $self_test_rwmem_data
+cmb 0377 Check the data has decrypted successfully
+rep 0377 rmem R0 $encrypted_data
+rep 0381 add  R2 $encrypted_data 0001
 ren label_04d7 self_test_rmem_fail
 ren label_04ee self_test_wmem_fail
 ren sub_06bb decrypt_data
@@ -80,14 +86,20 @@ cmb 0375 Sneaky!
 cmb 0375 call $decrypt_data
 rep 0375 nop
 rep 0376 nop
+cmb 038f Replace the "t" in "test string" with "T"
 lbl 17b4 encrypted_data
 ren label_03ad self_test_wmem_cmd_fail
+cmb 03a3 Try replacing the next instructions
+rep 03a3 wmem $self_test_dynamic 0015
+rep 03a6 wmem $($self_test_dynamic+1) 0007
 rep 03ab out $self_test_complete
 cmi 03a9 Becomes noop; jt 0013 $self_test_complete
+lbl 03a9 self_test_dynamic
 lbl 03d2 self_test_complete
 lbl 17c0 str_self_test_result
 cmi 03d6 The "F"
 lbl 17e4 str_complete
+str 17e4
 ren label_03e8 self_test_loop_copy_complete
 cmb 03e8 Copy the "complete" string over the "FAILED!!" substring of $str_self_test_result
 ren label_03ff self_test_concat_all_pass

libsynacor/assembly.py

@@ -103,7 +103,7 @@ def assemble_instruction(source, tokens):
 				instruction.args.append(OpLiteral(ord(char)))
 				instructions.append(instruction)
 			return instructions, []
-		elif isinstance(instruction, InstructionData):
+		elif isinstance(instruction, InstructionData) or isinstance(instruction, InstructionString):
 			if tokens[1].startswith('"'):
 				chars = unescape_char(tokens[1][1:-1])
 				instruction.args = [ord(char) for char in chars]
@@ -120,6 +120,10 @@ def assemble_instruction(source, tokens):
 					# Register
 					arg = OpRegister(int(argstr[1:]))
 				elif argstr.startswith('$'):
+					if argstr.startswith('$('):
+						# Expression
+						arg = OpExpression(argstr[2:-1])
+					else:
 						# Label
 						arg = OpLabel(argstr[1:])
 				else:

libsynacor/bytecode.py

@@ -64,6 +64,28 @@ class OpLabel(Operand):
 			raise Exception('Unknown label {}'.format(self.label))
 		return OpLiteral(labels[self.label]).assemble(labels)
 
+# Used only in assembling process
+class OpExpression(Operand):
+	def __init__(self, expr):
+		self.expr = expr
+	
+	def describe(self):
+		return '$({})'.format(self.expr)
+	def assemble(self, labels):
+		if labels is None:
+			# First pass
+			return 0xffff
+		
+		# Replace labels
+		expr = self.expr
+		for label in labels:
+			expr = expr.replace('$' + label, str(labels[label]))
+		
+		# Warning: Not safe for untrusted code!
+		value = eval(expr)
+		
+		return OpLiteral(value).assemble(labels)
+
 instructions_by_opcode = {}
 instructions_by_name = {}
 
@@ -246,3 +268,8 @@ class InstructionData(Instruction):
 	def assemble(self, labels):
 		return self.args
 instructions_by_name['data'] = InstructionData
+
+class InstructionString(Instruction):
+	def assemble(self, labels):
+		return [len(self.args)] + self.args
+instructions_by_name['str'] = InstructionString