def _convert(self, value):
if isinstance(value, disasm.Module):
return self._covert(value.body)
elif isinstance(value, disasm.CodeObject):
return Block(*[self._convert(x) for x in value.opcodes])
elif isinstance(value, disasm.Opcode):
arg = value.argument
if isinstance(arg, list): arg = List(*[self._convert(x) for x in arg])
elif isinstance(arg, tuple): arg = Tuple(*[self._convert(x) for x in arg])
elif isinstance(arg, set): arg = Call(Ident('set'), Tuple(*[self._convert(x) for x in arg]))
elif isinstance(arg, frozenset): arg = Call(Ident('frozenset'), Tuple(*[self._convert(x) for x in arg]))
else: arg = self._convert(arg)
return Opcode(value.offset, value.size, value.opcode, arg if op.has_argument(value.opcode) else None)
else:
return Const(value)
def unmarshal_node(self):
type = self.read_int8()
# Global singletons
if type == _TYPE_NONE: return None
elif type == _TYPE_TRUE: return True
elif type == _TYPE_FALSE:
return False
# Collections
elif type == _TYPE_TUPLE:
return self.unmarshal_collection(tuple)
elif type == _TYPE_LIST:
return self.unmarshal_collection(list)
elif type == _TYPE_SET:
return self.unmarshal_collection(set)
elif type == _TYPE_FROZEN_SET:
return self.unmarshal_collection(frozenset)
# Numbers
elif type == _TYPE_INT:
return self.read_int32()
elif type == _TYPE_INT64:
return struct.unpack('=q', self.file.read(8))[0]
elif type == _TYPE_BINARY_FLOAT:
return struct.unpack('=d', self.file.read(8))[0]
elif type == _TYPE_BINARY_COMPLEX:
return complex(*struct.unpack('=dd', self.file.read(16)))
elif type == _TYPE_LONG:
nbits = self.read_int32()
if not nbits:
return 0
n = 0
for i in range(abs(nbits)):
digit = self.read_int16()
n |= digit << (i * 15)
return n if nbits > 0 else -n
# Strings
elif type == _TYPE_STRING:
return self.read_string_ascii()
elif type == _TYPE_UNICODE:
return self.read_string_utf8()
elif type == _TYPE_INTERNED:
data = self.read_string_ascii()
self.string_table.append(data)
return data
elif type == _TYPE_STRING_REF:
index = self.read_int32()
if index < 0 or index >= len(self.string_table):
raise DisassemblerException(
'String index %d is outside string table' % index)
return self.string_table[index]
# Code objects
elif type == _TYPE_CODE:
co = CodeObject()
co.co_argcount = self.read_int32()
co.co_kwonlyargcount = self.read_int32() if op.has_kwonlyargcount(
self.magic) else 0
co.co_nlocals = self.read_int32()
co.co_stacksize = self.read_int32()
co.co_flags = self.read_int32()
type = self.read_int8()
if type != _TYPE_STRING:
raise DisassemblerException(
'Bytecode was not marshalled as a string (type was 0x%02X instead of 0x%02X)'
% (type, _TYPE_STRING))
co.co_code = self.read_byte_array()
co.co_consts = self.unmarshal_node()
co.co_names = self.unmarshal_node()
co.co_varnames = self.unmarshal_node()
co.co_freevars = self.unmarshal_node()
co.co_cellvars = self.unmarshal_node()
co.co_filename = self.unmarshal_node()
co.co_name = self.unmarshal_node()
co.co_firstlineno = self.read_int32()
co.co_lnotab = self.unmarshal_node()
# Start disassembly
argument = 0
i = 0
while i < len(co.co_code):
offset = i
opcode = op.from_bytecode(co.co_code[i], self.magic)
if opcode is None:
raise DisassemblerException('Unknown bytecode 0x%02X' %
co.co_code[i])
i += 1
if op.has_argument(opcode):
lo, hi = co.co_code[i:i + 2]
argument |= (lo | (hi << 8))
i += 2
# The upper 16 bits of 32-bit arguments are stored in a fake
# EXTENDED_ARG opcode that precedes the actual opcode
if opcode == op.EXTENDED_ARG:
argument <<= 16
continue
# Decode the opcode argument if present
arg = None
if op.has_argument(opcode):
if opcode == op.LOAD_CONST:
if argument >= len(co.co_consts):
raise DisassemblerException(
'Invalid argument %d for opcode %s' %
(argument, opcode))
arg = co.co_consts[argument]
elif opcode in [
op.LOAD_NAME, op.STORE_NAME, op.DELETE_NAME,
op.LOAD_ATTR, op.STORE_ATTR, op.DELETE_ATTR,
op.LOAD_GLOBAL, op.STORE_GLOBAL, op.DELETE_GLOBAL,
op.IMPORT_NAME, op.IMPORT_FROM
]:
if argument >= len(co.co_names):
raise DisassemblerException(
'Invalid argument %d for opcode %s' %
(argument, opcode))
arg = co.co_names[argument]
elif opcode in [
op.LOAD_FAST, op.STORE_FAST, op.DELETE_FAST
]:
if argument >= len(co.co_varnames):
raise DisassemblerException(
'Invalid argument %d for opcode %s' %
(argument, opcode))
arg = co.co_varnames[argument]
else:
arg = argument
# Record disassembled opcode
co.opcodes.append(Opcode(offset, i - offset, opcode, arg))
argument = 0
return co
else:
raise DisassemblerException(
'Cannot unmarshal unknown type 0x%02X' % type)
def unmarshal_node(self):
type = self.read_int8()
# Global singletons
if type == _TYPE_NONE: return None
elif type == _TYPE_TRUE: return True
elif type == _TYPE_FALSE: return False
# Collections
elif type == _TYPE_TUPLE: return self.unmarshal_collection(tuple)
elif type == _TYPE_LIST: return self.unmarshal_collection(list)
elif type == _TYPE_SET: return self.unmarshal_collection(set)
elif type == _TYPE_FROZEN_SET: return self.unmarshal_collection(frozenset)
# Numbers
elif type == _TYPE_INT: return self.read_int32()
elif type == _TYPE_INT64: return struct.unpack('=q', self.file.read(8))[0]
elif type == _TYPE_BINARY_FLOAT: return struct.unpack('=d', self.file.read(8))[0]
elif type == _TYPE_BINARY_COMPLEX: return complex(*struct.unpack('=dd', self.file.read(16)))
elif type == _TYPE_LONG:
nbits = self.read_int32()
if not nbits:
return 0
n = 0
for i in range(abs(nbits)):
digit = self.read_int16()
n |= digit << (i * 15)
return n if nbits > 0 else -n
# Strings
elif type == _TYPE_STRING: return self.read_string_ascii()
elif type == _TYPE_UNICODE: return self.read_string_utf8()
elif type == _TYPE_INTERNED:
data = self.read_string_ascii()
self.string_table.append(data)
return data
elif type == _TYPE_STRING_REF:
index = self.read_int32()
if index < 0 or index >= len(self.string_table):
raise DisassemblerException('String index %d is outside string table' % index)
return self.string_table[index]
# Code objects
elif type == _TYPE_CODE:
co = CodeObject()
co.co_argcount = self.read_int32()
co.co_kwonlyargcount = self.read_int32() if op.has_kwonlyargcount(self.magic) else 0
co.co_nlocals = self.read_int32()
co.co_stacksize = self.read_int32()
co.co_flags = self.read_int32()
type = self.read_int8()
if type != _TYPE_STRING:
raise DisassemblerException('Bytecode was not marshalled as a string (type was 0x%02X instead of 0x%02X)' % (type, _TYPE_STRING))
co.co_code = self.read_byte_array()
co.co_consts = self.unmarshal_node()
co.co_names = self.unmarshal_node()
co.co_varnames = self.unmarshal_node()
co.co_freevars = self.unmarshal_node()
co.co_cellvars = self.unmarshal_node()
co.co_filename = self.unmarshal_node()
co.co_name = self.unmarshal_node()
co.co_firstlineno = self.read_int32()
co.co_lnotab = self.unmarshal_node()
# Start disassembly
argument = 0
i = 0
while i < len(co.co_code):
offset = i
opcode = op.from_bytecode(co.co_code[i], self.magic)
if opcode is None:
raise DisassemblerException('Unknown bytecode 0x%02X' % co.co_code[i])
i += 1
if op.has_argument(opcode):
lo, hi = co.co_code[i:i + 2]
argument |= (lo | (hi << 8))
i += 2
# The upper 16 bits of 32-bit arguments are stored in a fake
# EXTENDED_ARG opcode that precedes the actual opcode
if opcode == op.EXTENDED_ARG:
argument <<= 16
continue
# Decode the opcode argument if present
arg = None
if op.has_argument(opcode):
if opcode == op.LOAD_CONST:
if argument >= len(co.co_consts):
raise DisassemblerException('Invalid argument %d for opcode %s' % (argument, opcode))
arg = co.co_consts[argument]
elif opcode in [op.LOAD_NAME, op.STORE_NAME, op.DELETE_NAME,
op.LOAD_ATTR, op.STORE_ATTR, op.DELETE_ATTR,
op.LOAD_GLOBAL, op.STORE_GLOBAL, op.DELETE_GLOBAL,
op.IMPORT_NAME, op.IMPORT_FROM]:
if argument >= len(co.co_names):
raise DisassemblerException('Invalid argument %d for opcode %s' % (argument, opcode))
arg = co.co_names[argument]
elif opcode in [op.LOAD_FAST, op.STORE_FAST, op.DELETE_FAST]:
if argument >= len(co.co_varnames):
raise DisassemblerException('Invalid argument %d for opcode %s' % (argument, opcode))
arg = co.co_varnames[argument]
else:
arg = argument
# Record disassembled opcode
co.opcodes.append(Opcode(offset, i - offset, opcode, arg))
argument = 0
return co
else:
raise DisassemblerException('Cannot unmarshal unknown type 0x%02X' % type)