def enter_code_object(self, co_obj):
labels = dis.findlabels(co_obj.co_code)
labels = opcode_util.extendlabels(co_obj.co_code, labels)
self.blocks = dict((index, []) for index in labels)
self.stack = []
self.loop_stack = []
self.blocks[0] = self.block = []
def __init__(self, func):
func = get_function_object(func)
code = get_code_object(func)
pysig = utils.pysignature(func)
if not code:
raise errors.ByteCodeSupportError(
"%s does not provide its bytecode" % func)
if code.co_cellvars:
raise NotImplementedError("cell vars are not supported")
table = utils.SortedMap(ByteCodeIter(code))
labels = set(dis.findlabels(code.co_code))
labels.add(0)
try:
func_qualname = func.__qualname__
except AttributeError:
func_qualname = func.__name__
self._mark_lineno(table, code)
super(ByteCode, self).__init__(func=func,
func_qualname=func_qualname,
is_generator=inspect.isgeneratorfunction(func),
pysig=pysig,
filename=code.co_filename,
co_names=code.co_names,
co_varnames=code.co_varnames,
co_consts=code.co_consts,
co_freevars=code.co_freevars,
table=table,
labels=list(sorted(labels)))
def __init__(self, func):
func = get_function_object(func)
code = get_code_object(func)
pysig = utils.pysignature(func)
if not code:
raise errors.ByteCodeSupportError(
"%s does not provide its bytecode" % func)
# A map of {offset: ByteCodeInst}
table = utils.SortedMap(ByteCodeIter(code))
labels = set(dis.findlabels(code.co_code))
labels.add(0)
try:
func_qualname = func.__qualname__
except AttributeError:
func_qualname = func.__name__
self._mark_lineno(table, code)
super(ByteCode, self).__init__(func=func,
func_qualname=func_qualname,
is_generator=inspect.isgeneratorfunction(func),
pysig=pysig,
filename=code.co_filename,
co_names=code.co_names,
co_varnames=code.co_varnames,
co_consts=code.co_consts,
co_freevars=code.co_freevars,
table=table,
labels=list(sorted(labels)))
def _get_instructions_bytes(code, varnames=None, names=None, constants=None,
cells=None, linestarts=None, line_offset=0):
"""Iterate over the instructions in a bytecode string.
Generates a sequence of Instruction namedtuples giving the details of each
opcode. Additional information about the code's runtime environment
(e.g. variable names, constants) can be specified using optional
arguments.
"""
labels = dis.findlabels(code)
extended_arg = 0
starts_line = None
free = None
# enumerate() is not an option, since we sometimes process
# multiple elements on a single pass through the loop
n = len(code)
i = 0
while i < n:
op = code[i]
offset = i
if linestarts is not None:
starts_line = linestarts.get(i, None)
if starts_line is not None:
starts_line += line_offset
is_jump_target = i in labels
i = i + 1
arg = None
argval = None
argrepr = ''
if op >= dis.HAVE_ARGUMENT:
arg = code[i] + code[i + 1] * 256 + extended_arg
extended_arg = 0
i = i + 2
if op == dis.EXTENDED_ARG:
extended_arg = arg * 65536
# Set argval to the dereferenced value of the argument when
# availabe, and argrepr to the string representation of argval.
# _disassemble_bytes needs the string repr of the
# raw name index for LOAD_GLOBAL, LOAD_CONST, etc.
argval = arg
if op in dis.hasconst:
argval, argrepr = dis._get_const_info(arg, constants)
elif op in dis.hasname:
argval, argrepr = dis._get_name_info(arg, names)
elif op in dis.hasjrel:
argval = i + arg
argrepr = "to " + repr(argval)
elif op in dis.haslocal:
argval, argrepr = dis._get_name_info(arg, varnames)
elif op in dis.hascompare:
argval = dis.cmp_op[arg]
argrepr = argval
elif op in dis.hasfree:
argval, argrepr = dis._get_name_info(arg, cells)
elif op in dis.hasnargs:
argrepr = "%d positional, %d keyword pair" % (code[i - 2], code[i - 1])
yield dis.Instruction(dis.opname[op], op,
arg, argval, argrepr,
offset, starts_line, is_jump_target)
def __init__(self, func):
func = get_function_object(func)
code = get_code_object(func)
if not code:
raise ByteCodeSupportError("%s does not provide its bytecode" %
func)
if code.co_cellvars:
raise ByteCodeSupportError("does not support cellvars")
table = utils.SortedMap(ByteCodeIter(code))
labels = set(dis.findlabels(code.co_code))
labels.add(0)
try:
func_qualname = func.__qualname__
except AttributeError:
func_qualname = func.__name__
self._mark_lineno(table, code)
super(ByteCode, self).__init__(func=func,
func_qualname=func_qualname,
argspec=inspect.getargspec(func),
filename=code.co_filename,
co_names=code.co_names,
co_varnames=code.co_varnames,
co_consts=code.co_consts,
co_freevars=code.co_freevars,
table=table,
labels=list(sorted(labels)))
def __init__(self, func):
func = get_function_object(func)
code = get_code_object(func)
if not code:
raise ByteCodeSupportError("%s does not provide its bytecode" %
func)
if code.co_cellvars:
raise ByteCodeSupportError("does not support cellvars")
table = utils.SortedMap(ByteCodeIter(code))
labels = set(dis.findlabels(code.co_code))
labels.add(0)
try:
func_qualname = func.__qualname__
except AttributeError:
func_qualname = func.__name__
self._mark_lineno(table, code)
super(ByteCode, self).__init__(func=func,
func_qualname=func_qualname,
argspec=inspect.getargspec(func),
filename=code.co_filename,
co_names=code.co_names,
co_varnames=code.co_varnames,
co_consts=code.co_consts,
co_freevars=code.co_freevars,
table=table,
labels=list(sorted(labels)))
def __init__(self, method_name, code_object): Declaration.__init__(self, Declaration.METHOD, code_object) self._method_name = method_name self._formal_parameters = [] self._body = None self._labels = dis.findlabels(code_object.co_code) self._nested_types = []
def disassemble(co, lasti=-1):
"""Disassemble a code object."""
# Taken from dis.disassemble, returns disassembled code instead of printing
# it (the f**k python ?).
# Also, unicodified.
# Also, use % operator instead of string operations.
# Also, one statement per line.
out = StringIO()
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(dis.findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = None
while i < n:
c = code[i]
op = ord(c)
if i in linestarts:
if i > 0:
print(end=u'\n', file=out)
print(u'%3d' % linestarts[i], end=u' ', file=out)
else:
print(u' ', end=u' ', file=out)
if i == lasti:
print(u'-->', end=u' ', file=out)
else:
print(u' ', end=u' ', file=out)
if i in labels:
print(u'>>', end=u' ', file=out)
else:
print(u' ', end=u' ', file=out)
print(u'%4i' % i, end=u' ', file=out)
print(u'%-20s' % dis.opname[op], end=u' ', file=out)
i = i + 1
if op >= dis.HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i = i + 2
if op == dis.EXTENDED_ARG:
extended_arg = oparg * 65536
print(u'%5i' % oparg, end=u' ', file=out)
if op in dis.hasconst:
print(u'(%r)' % co.co_consts[oparg], end=u' ', file=out)
elif op in dis.hasname:
print(u'(%s)' % co.co_names[oparg], end=u' ', file=out)
elif op in dis.hasjrel:
print(u'(to %r)' % (i + oparg), end=u' ', file=out)
elif op in dis.haslocal:
print(u'(%s)' % co.co_varnames[oparg], end=u' ', file=out)
elif op in dis.hascompare:
print(u'(%s)' % dis.cmp_op[oparg], end=u' ', file=out)
elif op in dis.hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
print(u'(%s)' % free[oparg], end=u' ', file=out)
print(end=u'\n', file=out)
return out.getvalue()
def build_basic_blocks(co_obj):
co_code = co_obj.co_code
labels = extendlabels(co_code, dis.findlabels(co_code))
labels.sort()
blocks = dict(
(index, list(itercode(co_code[index:next_index], index)))
for index, next_index in zip([0] + labels, labels + [len(co_code)]))
return blocks
def enter_code_object (self, co_obj):
labels = dis.findlabels(co_obj.co_code)
labels = opcode_util.extendlabels(co_obj.co_code, labels)
self.blocks = dict((index, [])
for index in labels)
self.stack = []
self.loop_stack = []
self.blocks[0] = self.block = []
def build_basic_blocks (co_obj):
co_code = co_obj.co_code
labels = extendlabels(co_code, dis.findlabels(co_code))
labels.sort()
blocks = dict((index, list(itercode(co_code[index:next_index], index)))
for index, next_index in zip([0] + labels,
labels + [len(co_code)]))
return blocks
def __init__(self, *args, **kwargs):
super(InstructionInterpreter, self).__init__(*args, **kwargs)
self._labels = dis.findlabels(self._co_code)
self.starts_line = None
self.is_jump_target = False
self.argval = None
self.argrepr = None
self.current_line = -1
def __init__(self, *args, **kwargs):
super(InstructionInterpreter, self).__init__(*args, **kwargs)
self._labels = dis.findlabels(self._co_code)
self.starts_line = None
self.is_jump_target = False
self.argval = None
self.argrepr = None
self.current_line = -1
def disassembler(co, lasti=-1):
"""Disassemble a code object.
:param co: code object
:param lasti: internal
:yields: Instructions.
"""
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(dis.findlinestarts(co))
i = 0
extended_arg = 0
lineno = 0
free = None
for i, op, oparg in _walk_ops(co):
if i in linestarts:
lineno = linestarts[i]
instr = Instruction(i=i, op=op, lineno=lineno)
instr.linestart = i in linestarts
if i == lasti:
instr.lasti = True
else:
instr.lasti = False
if i in labels:
instr.label = True
else:
instr.label = False
instr.oparg = oparg
extended_arg = 0
if op == dis.EXTENDED_ARG:
extended_arg = oparg * 65536
instr.extended_arg = extended_arg
if op >= dis.HAVE_ARGUMENT:
if op in dis.hasconst:
instr.arg = co.co_consts[oparg]
elif op in dis.hasname:
instr.arg = co.co_names[oparg]
elif op in dis.hasjrel:
instr.arg = i + oparg
elif op in dis.haslocal:
instr.arg = co.co_varnames[oparg]
elif op in dis.hascompare:
instr.arg = dis.cmp_op[oparg]
elif op in dis.hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
instr.arg = free[oparg]
yield instr
def __init__(self, func):
self.code = get_code_object(func)
#print dis.dis(self.code)
# assert not self.code.co_freevars, "does not support freevars"
# assert not self.code.co_cellvars, "does not support cellvars"
self.table = SortedMap(ByteCodeIter(self.code))
labels = set(dis.findlabels(self.code.co_code) + find_targets(self))
labels.add(0)
self.labels = sorted(labels)
self._mark_lineno()
def __init__(self, func):
self.code = get_code_object(func)
#print dis.dis(self.code)
# assert not self.code.co_freevars, "does not support freevars"
# assert not self.code.co_cellvars, "does not support cellvars"
self.table = SortedMap(ByteCodeIter(self.code))
labels = set(dis.findlabels(self.code.co_code) + find_targets(self))
labels.add(0)
self.labels = sorted(labels)
self._mark_lineno()
def read(func):
proc = Procedure(func)
tables = Tables(
jump = {},
labels = findlabels(func.func_code.co_code),
variables = [Local(n) for n in func.func_code.co_varnames])
entry = interpret(proc, 0, [], tables)
assert len(entry.prec) == 0 # If this assertion fails, then code in the
# interpret() for Arguments need to be fixed.
ssa_conversion(proc)
return proc
def dissco(co):
try:
code = co.co_code
except AttributeError:
raise TypeError(
'disassembly only works on objects with code components.')
ret = []
labels = findlabels(code)
linestarts = dict(findlinestarts(co))
print labels, linestarts
n = len(code)
i = 0
cur_line = 0
oparg = None
arg = None
extended_arg = 0
free = None
while i < n:
c = code[i]
op = ord(c)
if i in linestarts:
cur_line = linestarts[i]
op_name = opname[op]
i += 1
if op >= HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i += 2
if op == EXTENDED_ARG:
extended_arg = oparg * 65536L
if op in hasconst:
arg = co.co_consts[oparg]
elif op in hasname:
arg = co.co_names[oparg]
elif op in hasjrel:
arg = i + oparg
elif op in haslocal:
arg = co.co_varnames[oparg]
elif op in hascompare:
arg = cmp_op[oparg]
elif op in hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
arg = free[oparg]
ret.append(Instruction(op, oparg, arg, cur_line))
oparg = None
arg = None
return ret
def disassemble(co):
"""Disassemble a code object."""
lines = []
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(dis.findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = None
while i < n:
c = code[i]
op = ord(c)
data = {"raw_op": op}
if i in linestarts:
data["line_no"] = linestarts[i]
else:
data["line_no"] = None
if i in labels:
data["in_labels"] = True
else:
data["in_labels"] = False
data["i"] = i
data["opname"] = opname[op]
data["hex"] = "%02X" % op
i = i + 1
if op >= HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
data["hex"] += " %02X %02X" % (ord(code[i]), ord(code[i + 1]))
i = i + 2
if op == EXTENDED_ARG:
extended_arg = oparg * 65536L
data["raw_arg"] = oparg
if op in hasconst:
data['arg'] = {"value": repr(co.co_consts[oparg]), "type": "const"}
elif op in hasname:
data['arg'] = {"value": co.co_names[oparg], "type": "name"}
elif op in hasjrel:
data['arg'] = {"value": repr(i + oparg), "type": "jump_relative"}
elif op in haslocal:
data['arg'] = {"value": co.co_varnames[oparg], "type": "local_name"}
elif op in hascompare:
data['arg'] = {"value": cmp_op[oparg], "type": "compare"}
elif op in hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
data['arg'] = {"value": cmp_op[oparg], "type": "free_variable"}
lines.append(data)
return json.dumps(lines)
def dissco(co):
try:
code = co.co_code
except AttributeError:
raise TypeError("disassembly only works on objects with code components.")
ret = []
labels = findlabels(code)
linestarts = dict(findlinestarts(co))
print labels, linestarts
n = len(code)
i = 0
cur_line = 0
oparg = None
arg = None
extended_arg = 0
free = None
while i < n:
c = code[i]
op = ord(c)
if i in linestarts:
cur_line = linestarts[i]
op_name = opname[op]
i += 1
if op >= HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i += 2
if op == EXTENDED_ARG:
extended_arg = oparg * 65536L
if op in hasconst:
arg = co.co_consts[oparg]
elif op in hasname:
arg = co.co_names[oparg]
elif op in hasjrel:
arg = i + oparg
elif op in haslocal:
arg = co.co_varnames[oparg]
elif op in hascompare:
arg = cmp_op[oparg]
elif op in hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
arg = free[oparg]
ret.append(Instruction(op, oparg, arg, cur_line))
oparg = None
arg = None
return ret
def from_code(cls, co):
"""Disassemble a Python code object into a Code object"""
co_code = co.co_code
labels = dict((addr, Label()) for addr in findlabels(co_code))
linestarts = dict(cls._findlinestarts(co))
cellfree = co.co_cellvars + co.co_freevars
code = []
n = len(co_code)
i = extended_arg = 0
while i < n:
op = Opcode(co_code[i])
if i in labels:
code.append((labels[i], None))
if i in linestarts:
code.append((SetLineno, linestarts[i]))
i += 1
if op in hascode:
lastop, lastarg = code[-1]
if lastop != LOAD_CONST:
raise ValueError(
"%s should be preceded by LOAD_CONST code" % op)
code[-1] = (LOAD_CONST, Code.from_code(lastarg))
if op not in hasarg:
code.append((op, None))
else:
arg = co_code[i] | co_code[i + 1] << 8 | extended_arg
extended_arg = 0
i += 2
if op == opcode.EXTENDED_ARG: extended_arg = arg << 16
else:
code.append(
(op, co.co_consts[arg] if op in hasconst else
co.co_names[arg] if op in hasname else
labels[arg] if op in hasjabs else labels[i +
arg] if op in
hasjrel else co.co_varnames[arg] if op in
haslocal else cmp_op[arg] if op in hascompare else
cellfree[arg] if op in hasfree else arg))
varargs = not not co.co_flags & CO_VARARGS
varkwargs = not not co.co_flags & CO_VARKEYWORDS
return cls(code=code,
freevars=co.co_freevars,
args=co.co_varnames[:co.co_argcount + varargs + varkwargs],
varargs=varargs,
varkwargs=varkwargs,
newlocals=not not co.co_flags & CO_NEWLOCALS,
name=co.co_name,
filename=co.co_filename,
firstlineno=co.co_firstlineno,
docstring=co.co_consts[0] if co.co_consts
and isinstance(co.co_consts[0], str) else None)
def make_junk_code(code):
# code_blocks = {block_id:[block_code,block_range]}
code_blocks, begin, code_block_id_tab, code_block_id_xref_dict = make_code_block(
code.co_code)
add_junk_code_dict = {}
add_jmp_dict = {}
code_block_new_position_dict = {}
random.shuffle(code_block_id_tab)
junk_code_max_size = (0x10000 - len(code.co_code) - len(
dis.findlabels(code.co_code))) / len(code_block_id_tab) - 5
junk_code_max_size = 9 if junk_code_max_size > 9 else junk_code_max_size
junk_code = add_junk_code(
random.randint(junk_code_max_size / 2, junk_code_max_size))
begin_position = 3 + len(junk_code)
for code_block_id in code_block_id_tab:
add_junk_code_dict[code_block_id] = add_junk_code(
random.randint(junk_code_max_size / 2, junk_code_max_size))
add_jmp_dict[code_block_id] = find_jmp(code_blocks[code_block_id])
code_block_new_position_dict[code_block_id] = begin_position
begin_position += len(code_blocks[code_block_id][0]) + 3 + 3 * len(
add_jmp_dict[code_block_id]) + len(
add_junk_code_dict[code_block_id])
co_code = chr(dis.opmap['JUMP_ABSOLUTE']) + \
n2b(code_block_new_position_dict[begin]) + junk_code
for code_block_id in code_block_id_tab:
next_code_block_id = 0
if code_block_id in code_block_id_xref_dict:
next_code_block_id = code_block_id_xref_dict[code_block_id]
co_code += patch_code(
code_blocks, add_jmp_dict[code_block_id], code_block_id,
next_code_block_id,
code_block_new_position_dict) + add_junk_code_dict[code_block_id]
co_consts = []
for v in code.co_consts:
if isinstance(v, types.CodeType):
co_consts += [make_junk_code(v)]
else:
co_consts += [v]
co_consts = tuple(co_consts)
if len(co_code) > 0xffff:
print '[ERROR]code is too long.'
return types.CodeType(code.co_argcount, code.co_nlocals, code.co_stacksize,
code.co_flags, co_code, co_consts, code.co_names,
code.co_varnames, code.co_filename, code.co_name,
code.co_firstlineno, code.co_lnotab)
def disassemble(co, lasti=-1):
"""Disassemble a code object."""
code = co.co_code
labels = findlabels(code)
linestarts = dict(findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = None
while i < n:
c = code[i]
op = ord(c)
if i in linestarts:
if i > 0:
print
print "%03d" % linestarts[i],
else:
print ' ',
if i == lasti: print '-->',
else: print ' ',
if i in labels: print '>>',
else: print ' ',
print repr(i).rjust(4),
print opname[op].ljust(20),
i = i+1
if op >= HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i+1])*256 + extended_arg
extended_arg = 0
i = i+2
if op == EXTENDED_ARG:
extended_arg = oparg*65536L
print repr(oparg).rjust(5),
if op in hasconst:
print '(' + repr(co.co_consts[oparg]) + ')',
elif op in hasname:
print '(' + repr(co.co_names[oparg]) + ')',
elif op in hasjrel:
print '(to ' + repr(i + oparg) + ')',
elif op in haslocal:
print '(' + co.co_varnames[oparg] + ')',
elif op in hascompare:
print '(' + repr(cmp_op[oparg]) + ')',
elif op in hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
print '(' + free[oparg] + ')',
print
def _iter_as_bytecode_as_instructions_py2(co):
code = co.co_code
op_offset_to_line = dict(dis.findlinestarts(co))
labels = set(dis.findlabels(code))
bytecode_len = len(code)
i = 0
extended_arg = 0
free = None
op_to_name = opname
while i < bytecode_len:
c = code[i]
op = ord(c)
is_jump_target = i in labels
curr_op_name = op_to_name[op]
initial_bytecode_offset = i
i = i + 1
if op < HAVE_ARGUMENT:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), None, is_jump_target, initial_bytecode_offset)
else:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i = i + 2
if op == EXTENDED_ARG:
extended_arg = oparg * 65536
if op in hasconst:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), co.co_consts[oparg], is_jump_target, initial_bytecode_offset)
elif op in hasname:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), co.co_names[oparg], is_jump_target, initial_bytecode_offset)
elif op in hasjrel:
argval = i + oparg
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), argval, is_jump_target, initial_bytecode_offset)
elif op in haslocal:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), co.co_varnames[oparg], is_jump_target, initial_bytecode_offset)
elif op in hascompare:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), cmp_op[oparg], is_jump_target, initial_bytecode_offset)
elif op in hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), free[oparg], is_jump_target, initial_bytecode_offset)
else:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), oparg, is_jump_target, initial_bytecode_offset)
def _iter_as_bytecode_as_instructions_py2(co):
code = co.co_code
op_offset_to_line = dict(dis.findlinestarts(co))
labels = set(dis.findlabels(code))
bytecode_len = len(code)
i = 0
extended_arg = 0
free = None
op_to_name = opname
while i < bytecode_len:
c = code[i]
op = ord(c)
is_jump_target = i in labels
curr_op_name = op_to_name[op]
initial_bytecode_offset = i
i = i + 1
if op < HAVE_ARGUMENT:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), None, is_jump_target, initial_bytecode_offset)
else:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i = i + 2
if op == EXTENDED_ARG:
extended_arg = oparg * 65536
if op in hasconst:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), co.co_consts[oparg], is_jump_target, initial_bytecode_offset)
elif op in hasname:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), co.co_names[oparg], is_jump_target, initial_bytecode_offset)
elif op in hasjrel:
argval = i + oparg
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), argval, is_jump_target, initial_bytecode_offset)
elif op in haslocal:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), co.co_varnames[oparg], is_jump_target, initial_bytecode_offset)
elif op in hascompare:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), cmp_op[oparg], is_jump_target, initial_bytecode_offset)
elif op in hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), free[oparg], is_jump_target, initial_bytecode_offset)
else:
yield _Instruction(curr_op_name, op, _get_line(op_offset_to_line, initial_bytecode_offset, 0), oparg, is_jump_target, initial_bytecode_offset)
def _get_instructions_bytes(code,
varnames=None,
names=None,
constants=None,
cells=None,
linestarts=None,
line_offset=0):
"""Iterate over the instructions in a bytecode string.
Generates a sequence of Instruction namedtuples giving the details of each
opcode. Additional information about the code's runtime environment
(e.g. variable names, constants) can be specified using optional
arguments.
"""
labels = findlabels(code)
starts_line = None
free = None
for offset, op, arg in _unpack_opargs(code):
if linestarts is not None:
starts_line = linestarts.get(offset, None)
if starts_line is not None:
starts_line += line_offset
is_jump_target = offset in labels
argval = None
argrepr = ''
if arg is not None:
# Set argval to the dereferenced value of the argument when
# available, and argrepr to the string representation of argval.
# _disassemble_bytes needs the string repr of the
# raw name index for LOAD_GLOBAL, LOAD_CONST, etc.
argval = arg
if op in hasconst:
argval, argrepr = _get_const_info(arg, constants)
elif op in hasname:
argval, argrepr = _get_name_info(arg, names)
elif op in hasjrel:
argval = offset + 3 + arg
argrepr = "to " + repr(argval)
elif op in haslocal:
argval, argrepr = _get_name_info(arg, varnames)
elif op in hascompare:
argval = cmp_op[arg]
argrepr = argval
elif op in hasfree:
argval, argrepr = _get_name_info(arg, cells)
yield Instruction(opname[op], op, arg, argval, argrepr, offset,
starts_line, is_jump_target)
def __init__(self, func_id):
code = func_id.code
labels = set(dis.findlabels(code.co_code))
labels.add(0)
# A map of {offset: ByteCodeInst}
table = OrderedDict(ByteCodeIter(code))
self._compute_lineno(table, code)
self.func_id = func_id
self.co_names = code.co_names
self.co_varnames = code.co_varnames
self.co_consts = code.co_consts
self.co_freevars = code.co_freevars
self.table = table
self.labels = sorted(labels)
def __init__(self, method_name, code_object):
Declaration.__init__(self, Declaration.METHOD, code_object)
self._method_name = method_name
self._formal_parameters = []
self._body = None
self._labels = dis.findlabels(code_object.co_code)
self._nested_types = []
self.pretty_summary = []
self.one_param_summary = []
self.all_params_summary = {}
self.vuln_summary = []
self.inter_vuln_summary = []
self.all_params_vuln_summary = {}
self.returns_tainted = False
self.all_params_returns_tainted = {}
def __init__(self, func_id):
code = func_id.code
labels = set(dis.findlabels(code.co_code))
labels.add(0)
# A map of {offset: ByteCodeInst}
table = OrderedDict(ByteCodeIter(code))
self._compute_lineno(table, code)
self.func_id = func_id
self.co_names = code.co_names
self.co_varnames = code.co_varnames
self.co_consts = code.co_consts
self.co_freevars = code.co_freevars
self.table = table
self.labels = sorted(labels)
def __init__(self, func):
self.func = func
self.code = get_code_object(func)
self.filename = self.code.co_filename
# Do basic checking on support for the given bytecode
if not self.code:
raise ByteCodeSupportError("%s does not provide its bytecode" %
func)
if self.code.co_freevars:
raise ByteCodeSupportError("does not support freevars")
if self.code.co_cellvars:
raise ByteCodeSupportError("does not support cellvars")
self.table = utils.SortedMap(ByteCodeIter(self.code))
labels = set(dis.findlabels(self.code.co_code))
labels.add(0)
self.labels = list(sorted(labels))
self._mark_lineno()
def disassemble(co):
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(dis.findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = None
lineno = None
while i < n:
c = code[i]
op = ord(c)
lineno = linestarts.get(i, lineno)
is_label = i in labels
ist = i
i += 1
if op >= opcode.HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i += 2
if op == opcode.EXTENDED_ARG:
extended_arg = oparg * 65536L
if op in opcode.hasconst:
arg = co.co_consts[oparg]
elif op in opcode.hasname:
arg = co.co_names[oparg]
elif op in opcode.hasjrel:
arg = i + oparg
elif op in opcode.haslocal:
arg = co.co_varnames[oparg]
elif op in opcode.hascompare:
arg = opcode.cmp_op[oparg]
elif op in opcode.hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
arg = free[oparg]
else:
arg = NOVAL
else:
arg = NOVAL
yield ist, lineno, is_label, opcode.opname[op], arg
def disassemble(co):
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(dis.findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = None
lineno = None
while i < n:
c = code[i]
op = ord(c)
lineno = linestarts.get(i, lineno)
is_label = i in labels
ist = i
i += 1
if op >= opcode.HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i += 2
if op == opcode.EXTENDED_ARG:
extended_arg = oparg * 65536L
if op in opcode.hasconst:
arg = co.co_consts[oparg]
elif op in opcode.hasname:
arg = co.co_names[oparg]
elif op in opcode.hasjrel:
arg = i + oparg
elif op in opcode.haslocal:
arg = co.co_varnames[oparg]
elif op in opcode.hascompare:
arg = opcode.cmp_op[oparg]
elif op in opcode.hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
arg = free[oparg]
else:
arg = NOVAL
else:
arg = NOVAL
yield ist, lineno, is_label, opcode.opname[op], arg
def disassemble(co, lasti=-1):
"""Disassemble a code object."""
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(dis.findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = None
while i < n:
c = code[i]
op = ord(c)
i = i + 1
oparg = None
if op >= dis.HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i = i + 2
if op == dis.EXTENDED_ARG:
extended_arg = oparg * 65536L
yield (op, oparg)
def make_code_block(code, obfuscate=False):
i = 0
code_block_id_xref_dict = {}
code_labels = dis.findlabels(code) if obfuscate else []
code_blocks = {}
code_block_id_tab = []
while (i < len(code)):
code_size = random.randint(3, 5) if obfuscate else random.randint(
8, 10)
code_block = ''
tmp = i
while (code_size and i < len(code)):
c = ord(code[i])
if i in code_labels:
if not code_block == '':
break
if obfuscate and (c in dis.hasjabs or c in dis.hasjrel):
code_block += code[i:i + 3]
i += 3
break
elif c in has_args:
code_block += code[i:i + 3]
i += 3
else:
code_block += code[i]
i += 1
code_size -= 1
code_block_id = random.randint(1, 0xffffffff)
while code_block_id in code_block_id_tab:
code_block_id = random.randint(1, 0xffffffff)
code_block_id_tab += [code_block_id]
code_blocks[code_block_id] = [code_block, xrange(tmp, i)]
for i, v in enumerate(code_block_id_tab[:-1]):
code_block_id_xref_dict[v] = code_block_id_tab[i + 1]
begin = code_block_id_tab[0]
return code_blocks, begin, code_block_id_tab, code_block_id_xref_dict
def pydis(code):
n = len(code)
i = 0
extarg = 0
ops = []
# Find branch targets
lbls = dis.findlabels(code)
while i < n:
# Insert label pseudo-instruction
if i in lbls:
ops.append((LABEL, lbls.index(i)))
op = ord(code[i])
i += 1
if op >= dis.HAVE_ARGUMENT:
arg = dec_arg(code[i : i + 2]) | extarg << 16
extarg = 0
i += 2
if op == dis.EXTENDED_ARG:
extarg = arg
continue
targ = None
if op in dis.hasjrel:
targ = i + arg
elif op in dis.hasjabs:
targ = arg
if targ != None:
arg = lbls.index(targ)
else:
arg = None
ops.append((op, arg))
return ops
def get_call_function_name(frame):
"""If f_back is looking at a call function, return
the name for it. Otherwise return None"""
f_back = frame.f_back
if not f_back: return None
if 'CALL_FUNCTION' != op_at_frame(f_back): return None
co = f_back.f_code
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(dis.findlinestarts(co))
inst = f_back.f_lasti
while inst >= 0:
c = code[inst]
op = ord(c)
if inst in linestarts:
inst += 1
oparg = ord(code[inst]) + (ord(code[inst + 1]) << 8)
return co.co_names[oparg]
inst -= 1
pass
return None
def make_code_from_frame(frame):
"""Make a Code object from the given frame.
Returns a two-tuple giving the Code object, and the offset of the
last instruction executed by the frame.
"""
code = frame.f_code
offset = i = 0
while i < frame.f_lasti:
offset += 1
if ord(code.co_code[i]) < HAVE_ARGUMENT:
i += 1
else:
i += 3
assert i == frame.f_lasti
for j in findlabels(code.co_code):
if j <= frame.f_lasti:
offset += 1
c = Code.from_code(code)
c.code[:] = [op for op in c.code if op[0] != SetLineno]
assert c.code[offset][0] == ord(code.co_code[frame.f_lasti])
return (c,offset)
def get_call_function_name(frame):
"""If f_back is looking at a call function, return
the name for it. Otherwise return None"""
f_back = frame.f_back
if not f_back: return None
if 'CALL_FUNCTION' != op_at_frame(f_back): return None
co = f_back.f_code
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(dis.findlinestarts(co))
inst = f_back.f_lasti
while inst >= 0:
c = code[inst]
op = ord(c)
if inst in linestarts:
inst += 1
oparg = ord(code[inst]) + (ord(code[inst+1]) << 8)
return co.co_names[oparg]
inst -= 1
pass
return None
def listbytecode(co):
code=co.co_code
labels=dis.findlabels(code)
n=len(code)
i=0
extended_arg=0
free=None
codes=[]
cnt=0
while i<n:
c=code[i]
op=ord(c)
script={'pos':cnt,'line':i,'co':op,'code':dis.opname[op]}
cnt+=1
i=i+1
codes.append(script)
if op>dis.HAVE_ARGUMENT:
oparg=ord(code[i])+ord(code[i+1])*256+extended_arg
extended_arg=0
i=i+2
if op==dis.EXTENDED_ARG:
extended_arg=oparg*65536L
script['arg']=oparg
if op in dis.hasconst:
script['const']=co.co_consts[oparg]
elif op in dis.hasname:
script['name']=co.co_names[oparg]
elif op in dis.hasjrel:
script['target']=i+oparg
elif op in dis.haslocal:
script['local']=co.co_varnames[oparg]
elif op in dis.hascompare:
script['cmp']=dis.cmp_op[oparg]
elif op in dis.hasfree:
if free is None:
free=co.co_cellvars+co.co_frevars
script['free']=free[oparg]
print '\n'.join(map(repr,codes))
return codes
def from_code(cls, co):
"""Disassemble a Python code object into a Code object"""
co_code=co.co_code
labels={addr:Label() for addr in findlabels(co_code)}
linestarts=dict(cls._findlinestarts(co))
cellfree=co.co_cellvars+co.co_freevars
code = []
n = len(co_code)
i = extended_arg=0
while i<n:
op = Opcode(co_code[i])
if i in labels:code.append((labels[i], None))
if i in linestarts:code.append((SetLineno, linestarts[i]))
i += 1
if op in hascode:
lastop,lastarg=code[-2]
if lastop!=LOAD_CONST:raise ValueError("%s should be preceded by LOAD_CONST"%op)
code[-1]=(LOAD_CONST,cls.from_code(lastarg))
if op not in hasarg:code.append((op, None))
else:
arg=co_code[i]|co_code[i+1]<<8|extended_arg
extended_arg=0
i+=2
if op==opcode.EXTENDED_ARG:extended_arg=arg<<16
else:code.append((op,co.co_consts[arg] if op in hasconst else co.co_names[arg] if op in hasname else labels[arg] if op in hasjabs else labels[i+arg] if op in hasjrel else co.co_varnames[arg] if op in haslocal else cmp_op[arg] if op in hascompare else cellfree[arg] if op in hasfree else arg))
varargs = not not co.co_flags&CO_VARARGS
varkwargs = not not co.co_flags&CO_VARKEYWORDS
return cls(co=co,
code = code,
freevars = co.co_freevars,
args = co.co_varnames[:co.co_argcount+varargs+varkwargs],
kwonly = co.co_kwonlyargcount,
varargs = varargs,
varkwargs = varkwargs,
newlocals = not not co.co_flags&CO_NEWLOCALS,
name = co.co_name,
filename = co.co_filename,
firstlineno = co.co_firstlineno,
docstring = co.co_consts[0] if co.co_consts and isinstance(co.co_consts[0],str) else None)
def _disassemble(code_object):
code_string = code_object.co_code
linestarts = dict(dis.findlinestarts(code_object))
labels = list(dis.findlabels(code_string))
i = 0
while i < len(code_string):
opcode = ord(code_string[i])
opname = dis.opname[opcode]
i += 1
if opcode >= dis.HAVE_ARGUMENT:
oparg = (ord(code_string[i]) + ord(code_string[i+1])*256)
if opcode in dis.hasconst:
oparg = code_object.co_consts[oparg]
elif opcode in dis.hasname:
oparg = code_object.co_names[oparg]
yield opname, oparg
if hasattr(oparg, 'co_code'):
for oc, oa in _disassemble(oparg):
yield oc, oa
i += 2
else:
yield opname, None
def get_names_at(co, lasti): """Get the names that are referenced in the active part of code""" code = co.co_code labels = dis.findlabels(code) linestarts = dict(dis.findlinestarts(co)) n = len(code) i = 0 extended_arg = 0 free = None linenames = [] found = False while i < n: c = code[i] op = ord(c) if i in linestarts: if found: return linenames linenames = [] if i == lasti: found = True i = i+1 if op >= dis.HAVE_ARGUMENT: oparg = ord(code[i]) + ord(code[i+1])*256 + extended_arg extended_arg = 0 i = i+2 if op == dis.EXTENDED_ARG: extended_arg = oparg*65536L if op in dis.hasname: linenames.append(co.co_names[oparg]) elif op in dis.haslocal: linenames.append(co.co_varnames[oparg]) elif op in dis.hasfree: if free is None: free = co.co_cellvars + co.co_freevars linenames.append(free[oparg]) if found: return linenames
def disassemble(co, lasti=-1):
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = co.co_cellvars + co.co_freevars
while i < n:
op = ord(code[i])
o = Op(op, i)
i += 1
if i in linestarts and i > 0:
o.line = linestarts[i]
if i in labels:
o.target = True
if op > dis.HAVE_ARGUMENT:
arg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i += 2
if op == dis.EXTENDED_ARG:
extended_arg = arg << 16
o.argcode = arg
if op in dis.hasconst:
o.arg = co.co_consts[arg]
elif op in dis.hasname:
o.arg = co.co_names[arg]
elif op in dis.hasjrel:
o.arg = i + arg
elif op in dis.haslocal:
o.arg = co.co_varnames[arg]
elif op in dis.hascompare:
o.arg = dis.cmp_op[arg]
elif op in dis.hasfree:
o.arg = free[arg]
yield o
def disassemble(co, lasti=-1):
code = co.co_code
labels = dis.findlabels(code)
linestarts = dict(findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = co.co_cellvars + co.co_freevars
while i < n:
op = ord(code[i])
o = Op(op, i)
i += 1
if i in linestarts and i > 0:
o.line = linestarts[i]
if i in labels:
o.target = True
if op > dis.HAVE_ARGUMENT:
arg = ord(code[i]) + ord(code[i+1]) * 256 + extended_arg
extended_arg = 0
i += 2
if op == dis.EXTENDED_ARG:
extended_arg = arg << 16
o.argcode = arg
if op in dis.hasconst:
o.arg = co.co_consts[arg]
elif op in dis.hasname:
o.arg = co.co_names[arg]
elif op in dis.hasjrel:
o.arg = i + arg
elif op in dis.haslocal:
o.arg = co.co_varnames[arg]
elif op in dis.hascompare:
o.arg = dis.cmp_op[arg]
elif op in dis.hasfree:
o.arg = free[arg]
yield o
def from_code(cls, co):
"""Disassemble a Python code object into a Code object"""
free_cell_isection = set(co.co_cellvars) & set(co.co_freevars)
if free_cell_isection:
print(co.co_name + ': has non-empty co.co_cellvars & co.co_freevars', free_cell_isection)
return None
co_code = co.co_code
labels = {addr: Label() for addr in findlabels(co_code)}
linestarts = dict(cls._findlinestarts(co))
cellfree = co.co_cellvars + co.co_freevars
code = []
n = len(co_code)
i = extended_arg = 0
is_generator = False
if version_info >= (3, 5,):
is_coroutine = False
while i < n:
op = Opcode(co_code[i])
if i in labels:
code.append((labels[i], None))
if i in linestarts:
code.append((SetLineno, linestarts[i]))
i += 1
if op in hascode:
lastop, lastarg = code[-2]
if lastop != LOAD_CONST:
raise ValueError("%s should be preceded by LOAD_CONST" % op)
sub_code = Code.from_code(lastarg)
if sub_code is None:
print(co.co_name + ': has unexpected subcode block')
return None
code[-2] = (LOAD_CONST, sub_code)
if op not in hasarg:
code.append((op, None))
else:
arg = co_code[i] | co_code[i + 1] << 8 | extended_arg
extended_arg = 0
i += 2
if op == opcode.EXTENDED_ARG:
extended_arg = arg << 16
else:
byteplay_arg = co.co_consts[arg] if op in hasconst else \
co.co_names[arg] if op in hasname else \
labels[arg] if op in hasjabs else \
labels[i + arg] if op in hasjrel else \
co.co_varnames[arg] if op in haslocal else \
cmp_op[arg] if op in hascompare else \
cellfree[arg] if op in hasfree else \
arg
code.append((op, byteplay_arg))
if op == YIELD_VALUE or op == YIELD_FROM:
is_generator = True
if version_info >= (3, 5,) and op in coroutine_opcodes:
is_coroutine = True
varargs = not not co.co_flags & CO_VARARGS
varkwargs = not not co.co_flags & CO_VARKEYWORDS
force_generator = not is_generator and (co.co_flags & CO_GENERATOR)
if version_info >= (3, 5,):
force_coroutine = not is_coroutine and (co.co_flags & CO_COROUTINE)
force_iterable_coroutine = co.co_flags & CO_ITERABLE_COROUTINE
assert not (force_coroutine and force_iterable_coroutine)
future_generator_stop = co.co_flags & CO_FUTURE_GENERATOR_STOP
else:
force_coroutine = None
force_iterable_coroutine =None
future_generator_stop = None
return cls(code=code,
freevars=co.co_freevars,
args=co.co_varnames[:co.co_argcount + varargs + varkwargs + co.co_kwonlyargcount],
kwonly=co.co_kwonlyargcount,
varargs=varargs,
varkwargs=varkwargs,
newlocals=not not co.co_flags & CO_NEWLOCALS,
name=co.co_name,
filename=co.co_filename,
firstlineno=co.co_firstlineno,
docstring=co.co_consts[0] if co.co_consts and isinstance(co.co_consts[0], str) else None,
force_generator=force_generator,
force_coroutine=force_coroutine,
force_iterable_coroutine=force_iterable_coroutine,
future_generator_stop=future_generator_stop)
def getjoinpoints(self):
"""Compute the bytecode positions that are potential join points
(for FlowObjSpace)"""
# first approximation
return dis.findlabels(self.co_code)
def disassemble_bytes(orig_msg, orig_msg_nocr, code, lasti=-1, cur_line=0,
start_line=-1, end_line=None, relative_pos=False,
varnames=(), names=(), constants=(), cells=(),
freevars=(), linestarts={}, highlight='light',
start_offset=0, end_offset=None):
"""Disassemble byte string of code. If end_line is negative
it counts the number of statement linestarts to use."""
statement_count = 10000
if end_line is None:
end_line = 10000
elif relative_pos:
end_line += start_line -1
pass
labels = findlabels(code)
null_print = lambda x: None
if start_line > cur_line:
msg_nocr = null_print
msg = null_print
else:
msg_nocr = orig_msg_nocr
msg = orig_msg
for instr in get_instructions_bytes(code, opc, varnames, names,
constants, cells, linestarts):
offset = instr.offset
if end_offset and offset > end_offset:
break
if instr.starts_line:
if offset:
msg("")
cur_line = instr.starts_line
if (start_line and ((start_line > cur_line) or
start_offset and start_offset > offset)) :
msg_nocr = null_print
msg = null_print
else:
statement_count -= 1
msg_nocr = orig_msg_nocr
msg = orig_msg
pass
if ((cur_line > end_line) or
(end_offset and offset > end_offset)):
break
msg_nocr(format_token(Mformat.LineNumber,
"%3d" % cur_line,
highlight=highlight))
else:
if start_offset and offset and start_offset <= offset:
msg_nocr = orig_msg_nocr
msg = orig_msg
pass
msg_nocr(' ')
if offset == lasti: msg_nocr(format_token(Mformat.Arrow, '-->',
highlight=highlight))
else: msg_nocr(' ')
if offset in labels: msg_nocr(format_token(Mformat.Arrow, '>>',
highlight=highlight))
else: msg_nocr(' ')
msg_nocr(repr(offset).rjust(4))
msg_nocr(' ')
msg_nocr(format_token(Mformat.Opcode,
instr.opname.ljust(20),
highlight=highlight))
msg_nocr(repr(instr.arg).ljust(10))
msg_nocr(' ')
# Show argval?
msg(format_token(Mformat.Name,
instr.argrepr.ljust(20),
highlight=highlight))
pass
return code, offset
def from_code(cls, co):
"""Disassemble a Python code object into a Code object."""
co_code = co.co_code
labels = dict((addr, Label()) for addr in findlabels(co_code))
linestarts = dict(cls._findlinestarts(co))
cellfree = co.co_cellvars + co.co_freevars
code = CodeList()
n = len(co_code)
i = 0
extended_arg = 0
while i < n:
op = Opcode(ord(co_code[i]))
if i in labels:
code.append((labels[i], None))
if i in linestarts:
code.append((SetLineno, linestarts[i]))
i += 1
if op in hascode:
lastop, lastarg = code[-1]
if lastop != LOAD_CONST:
raise ValueError, \
"%s should be preceded by LOAD_CONST code" % op
code[-1] = (LOAD_CONST, Code.from_code(lastarg))
if op not in hasarg:
code.append((op, None))
else:
arg = ord(co_code[i]) + ord(co_code[i+1])*256 + extended_arg
extended_arg = 0
i += 2
if op == opcode.EXTENDED_ARG:
extended_arg = arg << 16
elif op in hasconst:
code.append((op, co.co_consts[arg]))
elif op in hasname:
code.append((op, co.co_names[arg]))
elif op in hasjabs:
code.append((op, labels[arg]))
elif op in hasjrel:
code.append((op, labels[i + arg]))
elif op in haslocal:
code.append((op, co.co_varnames[arg]))
elif op in hascompare:
code.append((op, cmp_op[arg]))
elif op in hasfree:
code.append((op, cellfree[arg]))
else:
code.append((op, arg))
varargs = bool(co.co_flags & CO_VARARGS)
varkwargs = bool(co.co_flags & CO_VARKEYWORDS)
newlocals = bool(co.co_flags & CO_NEWLOCALS)
args = co.co_varnames[:co.co_argcount + varargs + varkwargs]
if co.co_consts and isinstance(co.co_consts[0], basestring):
docstring = co.co_consts[0]
else:
docstring = None
return cls(code = code,
freevars = co.co_freevars,
args = args,
varargs = varargs,
varkwargs = varkwargs,
newlocals = newlocals,
name = co.co_name,
filename = co.co_filename,
firstlineno = co.co_firstlineno,
docstring = docstring,
)
def disassemble_bytes(orig_msg, orig_msg_nocr, code, lasti=-1, cur_line=0,
start_line=-1, end_line=None, relative_pos=False,
varnames=(), names=(), consts=(), cellvars=(),
freevars=(), linestarts={}, highlight='light',
start_offset=0, end_offset=None):
"""Disassemble byte string of code. If end_line is negative
it counts the number of statement linestarts to use."""
statement_count = 10000
if end_line is None:
end_line = 10000
elif relative_pos:
end_line += start_line -1
pass
labels = findlabels(code)
n = len(code)
i = 0
extended_arg = 0
free = None
null_print = lambda x: None
if start_line > cur_line:
msg_nocr = null_print
msg = null_print
else:
msg_nocr = orig_msg_nocr
msg = orig_msg
pass
while i < n and statement_count >= 0:
c = code[i]
op = ord(c)
if end_offset and i > end_offset:
break
if start_offset > i :
msg_nocr = null_print
msg = null_print
else:
msg_nocr = orig_msg_nocr
msg = orig_msg
if i in linestarts:
if i > 0:
msg("")
cur_line = linestarts[i]
if ((start_line and start_line > cur_line) or
start_offset > i) :
msg_nocr = null_print
msg = null_print
else:
statement_count -= 1
msg_nocr = orig_msg_nocr
msg = orig_msg
pass
if ((cur_line > end_line) or
(end_offset and i > end_offset)):
break
msg_nocr(format_token(Mformat.LineNumber,
"%3d" % cur_line,
highlight=highlight))
else:
msg_nocr(' ')
if i == lasti: msg_nocr(format_token(Mformat.Arrow, '-->',
highlight=highlight))
else: msg_nocr(' ')
if i in labels: msg_nocr(format_token(Mformat.Arrow, '>>',
highlight=highlight))
else: msg_nocr(' ')
msg_nocr(repr(i).rjust(4))
msg_nocr(' ')
msg_nocr(format_token(Mformat.Opcode,
opname[op].ljust(20),
highlight=highlight))
i += 1
if op >= HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i+1])*256 + extended_arg
extended_arg = 0
i += 2
if op == EXTENDED_ARG:
extended_arg = oparg*65536
msg_nocr(repr(oparg).rjust(5))
msg_nocr(' ')
if op in hasconst:
msg_nocr('(' +
format_token(Mformat.Const,
repr(consts[oparg]),
highlight=highlight)
+ ')')
pass
elif op in hasname:
msg_nocr('(' +
format_token(Mformat.Name,
names[oparg],
highlight=highlight)
+ ')')
elif op in hasjrel:
msg_nocr(format_token(Mformat.Label,
'(to ' + repr(i + oparg) + ')',
highlight=highlight))
elif op in haslocal:
msg_nocr('(' +
format_token(Mformat.Var,
varnames[oparg],
highlight=highlight) + ')')
elif op in hascompare:
msg_nocr('(' +
format_token(Mformat.Compare,
cmp_op[oparg],
highlight=highlight) + ')')
elif op in hasfree:
if free is None:
free = cellvars + freevars
msg_nocr('(' + free[oparg] + ')')
pass
pass
msg("")
return
def disassembler(co, lasti= -1):
"""Disassemble a code object.
:param co: code object
:param lasti: internal
:yields: Instructions.
"""
code = co.co_code
labels = findlabels(code)
linestarts = dict(findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
lineno = 0
free = None
while i < n:
c = code[i]
op = co_ord(c)
if i in linestarts:
lineno = linestarts[i]
instr = Instruction(i=i, op=op, lineno=lineno)
instr.linestart = i in linestarts
if i == lasti:
instr.lasti = True
else:
instr.lasti = False
if i in labels:
instr.label = True
else:
instr.label = False
i = i + 1
if op >= opcode.HAVE_ARGUMENT:
oparg = co_ord(code[i]) + co_ord(code[i + 1]) * 256 + extended_arg
instr.oparg = oparg
extended_arg = 0
i = i + 2
if op == opcode.EXTENDED_ARG:
extended_arg = oparg * 65536
instr.extended_arg = extended_arg
if op in opcode.hasconst:
instr.arg = co.co_consts[oparg]
elif op in opcode.hasname:
instr.arg = co.co_names[oparg]
elif op in opcode.hasjrel:
instr.arg = i + oparg
elif op in opcode.haslocal:
instr.arg = co.co_varnames[oparg]
elif op in opcode.hascompare:
instr.arg = opcode.cmp_op[oparg]
elif op in opcode.hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
instr.arg = free[oparg]
yield instr
def disassemble_bytes(orig_msg,
orig_msg_nocr,
code,
lasti=-1,
cur_line=0,
start_line=-1,
end_line=None,
relative_pos=False,
varnames=(),
names=(),
consts=(),
cellvars=(),
freevars=(),
linestarts={},
highlight='light',
start_offset=0,
end_offset=None):
"""Disassemble byte string of code. If end_line is negative
it counts the number of statement linestarts to use."""
statement_count = 10000
if end_line is None:
end_line = 10000
elif relative_pos:
end_line += start_line - 1
pass
labels = findlabels(code)
n = len(code)
i = 0
extended_arg = 0
free = None
null_print = lambda x: None
if start_line > cur_line:
msg_nocr = null_print
msg = null_print
else:
msg_nocr = orig_msg_nocr
msg = orig_msg
pass
while i < n and statement_count >= 0:
c = code[i]
op = ord(c)
if end_offset and i > end_offset:
break
if start_offset > i:
msg_nocr = null_print
msg = null_print
else:
msg_nocr = orig_msg_nocr
msg = orig_msg
if i in linestarts:
if i > 0:
msg("")
cur_line = linestarts[i]
if ((start_line and start_line > cur_line) or start_offset > i):
msg_nocr = null_print
msg = null_print
else:
statement_count -= 1
msg_nocr = orig_msg_nocr
msg = orig_msg
pass
if ((cur_line > end_line) or (end_offset and i > end_offset)):
break
msg_nocr(
format_token(Mformat.LineNumber,
"%3d" % cur_line,
highlight=highlight))
else:
msg_nocr(' ')
if i == lasti:
msg_nocr(format_token(Mformat.Arrow, '-->', highlight=highlight))
else:
msg_nocr(' ')
if i in labels:
msg_nocr(format_token(Mformat.Arrow, '>>', highlight=highlight))
else:
msg_nocr(' ')
msg_nocr(repr(i).rjust(4))
msg_nocr(' ')
msg_nocr(
format_token(Mformat.Opcode,
opname[op].ljust(20),
highlight=highlight))
i += 1
if op >= HAVE_ARGUMENT:
oparg = ord(code[i]) + ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i += 2
if op == EXTENDED_ARG:
extended_arg = oparg * 65536
msg_nocr(repr(oparg).rjust(5))
msg_nocr(' ')
if op in hasconst:
msg_nocr('(' + format_token(
Mformat.Const, repr(consts[oparg]), highlight=highlight) +
')')
pass
elif op in hasname:
msg_nocr('(' + format_token(
Mformat.Name, names[oparg], highlight=highlight) + ')')
elif op in hasjrel:
msg_nocr(
format_token(Mformat.Label,
'(to ' + repr(i + oparg) + ')',
highlight=highlight))
elif op in haslocal:
msg_nocr('(' + format_token(
Mformat.Var, varnames[oparg], highlight=highlight) + ')')
elif op in hascompare:
msg_nocr('(' + format_token(
Mformat.Compare, cmp_op[oparg], highlight=highlight) + ')')
elif op in hasfree:
if free is None:
free = cellvars + freevars
msg_nocr('(' + free[oparg] + ')')
pass
pass
msg("")
return
def from_code(cls, co):
"""Disassemble a Python code object into a Code object"""
free_cell_isection = set(co.co_cellvars) & set(co.co_freevars)
if free_cell_isection:
print(
co.co_name + ': has non-empty co.co_cellvars & co.co_freevars',
free_cell_isection)
return None
co_code = co.co_code
labels = {addr: Label() for addr in findlabels(co_code)}
linestarts = dict(cls._findlinestarts(co))
cellfree = co.co_cellvars + co.co_freevars
code = []
n = len(co_code)
i = extended_arg = 0
is_generator = False
if version_info >= (
3,
5,
):
is_coroutine = False
while i < n:
op = Opcode(co_code[i])
if i in labels:
code.append((labels[i], None))
if i in linestarts:
code.append((SetLineno, linestarts[i]))
i += 1
if op in hascode:
lastop, lastarg = code[-2]
if lastop != LOAD_CONST:
raise ValueError("%s should be preceded by LOAD_CONST" %
op)
sub_code = Code.from_code(lastarg)
if sub_code is None:
print(co.co_name + ': has unexpected subcode block')
return None
code[-2] = (LOAD_CONST, sub_code)
if op not in hasarg:
code.append((op, None))
else:
arg = co_code[i] | co_code[i + 1] << 8 | extended_arg
extended_arg = 0
i += 2
if op == opcode.EXTENDED_ARG:
extended_arg = arg << 16
else:
byteplay_arg = co.co_consts[arg] if op in hasconst else \
co.co_names[arg] if op in hasname else \
labels[arg] if op in hasjabs else \
labels[i + arg] if op in hasjrel else \
co.co_varnames[arg] if op in haslocal else \
cmp_op[arg] if op in hascompare else \
cellfree[arg] if op in hasfree else \
arg
code.append((op, byteplay_arg))
if op == YIELD_VALUE or op == YIELD_FROM:
is_generator = True
if version_info >= (
3,
5,
) and op in coroutine_opcodes:
is_coroutine = True
varargs = not not co.co_flags & CO_VARARGS
varkwargs = not not co.co_flags & CO_VARKEYWORDS
force_generator = not is_generator and (co.co_flags & CO_GENERATOR)
if version_info >= (
3,
5,
):
force_coroutine = not is_coroutine and (co.co_flags & CO_COROUTINE)
force_iterable_coroutine = co.co_flags & CO_ITERABLE_COROUTINE
assert not (force_coroutine and force_iterable_coroutine)
future_generator_stop = co.co_flags & CO_FUTURE_GENERATOR_STOP
else:
force_coroutine = None
force_iterable_coroutine = None
future_generator_stop = None
return cls(code=code,
freevars=co.co_freevars,
args=co.co_varnames[:co.co_argcount + varargs + varkwargs +
co.co_kwonlyargcount],
kwonly=co.co_kwonlyargcount,
varargs=varargs,
varkwargs=varkwargs,
newlocals=not not co.co_flags & CO_NEWLOCALS,
name=co.co_name,
filename=co.co_filename,
firstlineno=co.co_firstlineno,
docstring=co.co_consts[0] if co.co_consts
and isinstance(co.co_consts[0], str) else None,
force_generator=force_generator,
force_coroutine=force_coroutine,
force_iterable_coroutine=force_iterable_coroutine,
future_generator_stop=future_generator_stop)
def from_code(cls, co):
"""Disassemble a Python code object into a Code object."""
co_code = co.co_code
labels = dict((addr, Label()) for addr in findlabels(co_code))
linestarts = dict(cls._findlinestarts(co))
cellfree = co.co_cellvars + co.co_freevars
code = CodeList()
n = len(co_code)
i = 0
extended_arg = 0
while i < n:
op = Opcode(ord(co_code[i]))
if i in labels:
code.append((labels[i], None))
if i in linestarts:
code.append((SetLineno, linestarts[i]))
i += 1
if op in hascode:
lastop, lastarg = code[-1]
if lastop != LOAD_CONST:
raise ValueError, \
"%s should be preceded by LOAD_CONST code" % op
code[-1] = (LOAD_CONST, Code.from_code(lastarg))
if op not in hasarg:
code.append((op, None))
else:
arg = ord(
co_code[i]) + ord(co_code[i + 1]) * 256 + extended_arg
extended_arg = 0
i += 2
if op == opcode.EXTENDED_ARG:
extended_arg = arg << 16
elif op in hasconst:
code.append((op, co.co_consts[arg]))
elif op in hasname:
code.append((op, co.co_names[arg]))
elif op in hasjabs:
code.append((op, labels[arg]))
elif op in hasjrel:
code.append((op, labels[i + arg]))
elif op in haslocal:
code.append((op, co.co_varnames[arg]))
elif op in hascompare:
code.append((op, cmp_op[arg]))
elif op in hasfree:
code.append((op, cellfree[arg]))
else:
code.append((op, arg))
varargs = bool(co.co_flags & CO_VARARGS)
varkwargs = bool(co.co_flags & CO_VARKEYWORDS)
newlocals = bool(co.co_flags & CO_NEWLOCALS)
args = co.co_varnames[:co.co_argcount + varargs + varkwargs]
if co.co_consts and isinstance(co.co_consts[0], basestring):
docstring = co.co_consts[0]
else:
docstring = None
return cls(
code=code,
freevars=co.co_freevars,
args=args,
varargs=varargs,
varkwargs=varkwargs,
newlocals=newlocals,
name=co.co_name,
filename=co.co_filename,
firstlineno=co.co_firstlineno,
docstring=docstring,
)
def getjoinpoints(self):
"""Compute the bytecode positions that are potential join points
(for FlowObjSpace)"""
# first approximation
return dis.findlabels(self.co_code)
def cmp_code_objects(version, code_obj1, code_obj2, name=''):
"""
Compare two code-objects.
This is the main part of this module.
"""
#print code_obj1, type(code_obj2)
assert type(code_obj1) == types.CodeType
assert type(code_obj2) == types.CodeType
#print dir(code_obj1)
if isinstance(code_obj1, object):
# new style classes (Python 2.2)
# assume _both_ code objects to be new stle classes
assert dir(code_obj1) == dir(code_obj2)
else:
# old style classes
assert dir(code_obj1) == code_obj1.__members__
assert dir(code_obj2) == code_obj2.__members__
assert code_obj1.__members__ == code_obj2.__members__
if name == '__main__':
name = code_obj1.co_name
else:
name = '%s.%s' % (name, code_obj1.co_name)
if name == '.?': name = '__main__'
if isinstance(code_obj1, object) and cmp(code_obj1, code_obj2):
# use the new style code-classes' __cmp__ method, which
# should be faster and more sophisticated
# if this compare fails, we use the old routine to
# find out, what exactly is nor equal
# if this compare succeds, simply return
#return
pass
if isinstance(code_obj1, object):
members = filter(lambda x: x.startswith('co_'), dir(code_obj1))
else:
members = dir(code_obj1)
members.sort()
#members.reverse()
tokens1 = None
for member in members:
if member in __IGNORE_CODE_MEMBERS__:
pass
elif member == 'co_code':
if version == 2.7:
import scanner27 as scan
scanner = scan.Scanner27()
elif version == 2.6:
import scanner26 as scan
scanner = scan.Scanner26()
elif version == 2.5:
import scanner25 as scan
scanner = scan.Scanner25()
scanner.setShowAsm(showasm=0)
global JUMP_OPs
JUMP_OPs = scan.JUMP_OPs + ['JUMP_BACK']
# use changed Token class
# we (re)set this here to save exception handling,
# which would get 'unubersichtlich'
scanner.setTokenClass(Token)
try:
# disassemble both code-objects
tokens1, customize = scanner.disassemble(code_obj1)
del customize # save memory
tokens2, customize = scanner.disassemble(code_obj2)
del customize # save memory
finally:
scanner.resetTokenClass() # restore Token class
targets1 = dis.findlabels(code_obj1.co_code)
tokens1 = [t for t in tokens1 if t.type != 'COME_FROM']
tokens2 = [t for t in tokens2 if t.type != 'COME_FROM']
i1 = 0
i2 = 0
offset_map = {}
check_jumps = {}
while i1 < len(tokens1):
if i2 >= len(tokens2):
if len(tokens1) == len(tokens2) + 2 \
and tokens1[-1].type == 'RETURN_VALUE' \
and tokens1[-2].type == 'LOAD_CONST' \
and tokens1[-2].pattr == None \
and tokens1[-3].type == 'RETURN_VALUE':
break
else:
raise CmpErrorCodeLen(name, tokens1, tokens2)
offset_map[tokens1[i1].offset] = tokens2[i2].offset
for idx1, idx2, offset2 in check_jumps.get(
tokens1[i1].offset, []):
if offset2 != tokens2[i2].offset:
raise CmpErrorCode(name, tokens1[idx1].offset,
tokens1[idx1], tokens2[idx2],
tokens1, tokens2)
if tokens1[i1] != tokens2[i2]:
if tokens1[i1].type == 'LOAD_CONST' == tokens2[i2].type:
i = 1
while tokens1[i1 + i].type == 'LOAD_CONST':
i += 1
if tokens1[i1+i].type.startswith(('BUILD_TUPLE', 'BUILD_LIST')) \
and i == int(tokens1[i1+i].type.split('_')[-1]):
t = tuple(
[elem.pattr for elem in tokens1[i1:i1 + i]])
if t != tokens2[i2].pattr:
raise CmpErrorCode(name, tokens1[i1].offset,
tokens1[i1], tokens2[i2],
tokens1, tokens2)
i1 += i + 1
i2 += 1
continue
elif i == 2 and tokens1[
i1 + i].type == 'ROT_TWO' and tokens2[
i2 + 1].type == 'UNPACK_SEQUENCE_2':
i1 += 3
i2 += 2
continue
elif i == 2 and tokens1[i1 + i].type in BIN_OP_FUNCS:
f = BIN_OP_FUNCS[tokens1[i1 + i].type]
if f(tokens1[i1].pattr,
tokens1[i1 + 1].pattr) == tokens2[i2].pattr:
i1 += 3
i2 += 1
continue
elif tokens1[i1].type == 'UNARY_NOT':
if tokens2[i2].type == 'POP_JUMP_IF_TRUE':
if tokens1[i1 + 1].type == 'POP_JUMP_IF_FALSE':
i1 += 2
i2 += 1
continue
elif tokens2[i2].type == 'POP_JUMP_IF_FALSE':
if tokens1[i1 + 1].type == 'POP_JUMP_IF_TRUE':
i1 += 2
i2 += 1
continue
elif tokens1[i1].type in ('JUMP_FORWARD', 'JUMP_BACK') \
and tokens1[i1-1].type == 'RETURN_VALUE' \
and tokens2[i2-1].type in ('RETURN_VALUE', 'RETURN_END_IF') \
and int(tokens1[i1].offset) not in targets1:
i1 += 1
continue
elif tokens1[i1].type == 'JUMP_FORWARD' and tokens2[i2].type == 'JUMP_BACK' \
and tokens1[i1+1].type == 'JUMP_BACK' and tokens2[i2+1].type == 'JUMP_BACK' \
and int(tokens1[i1].pattr) == int(tokens1[i1].offset) + 3:
if int(tokens1[i1].pattr) == int(tokens1[i1 +
1].offset):
i1 += 2
i2 += 2
continue
raise CmpErrorCode(name, tokens1[i1].offset, tokens1[i1],
tokens2[i2], tokens1, tokens2)
elif tokens1[i1].type in JUMP_OPs and tokens1[
i1].pattr != tokens2[i2].pattr:
dest1 = int(tokens1[i1].pattr)
dest2 = int(tokens2[i2].pattr)
if tokens1[i1].type == 'JUMP_BACK':
if offset_map[dest1] != dest2:
raise CmpErrorCode(name, tokens1[i1].offset,
tokens1[i1], tokens2[i2],
tokens1, tokens2)
else:
#import pdb; pdb.set_trace()
if dest1 in check_jumps:
check_jumps[dest1].append((i1, i2, dest2))
else:
check_jumps[dest1] = [(i1, i2, dest2)]
i1 += 1
i2 += 1
del tokens1, tokens2 # save memory
elif member == 'co_consts':
# partial optimization can make the co_consts look different,
# so we'll just compare the code consts
codes1 = (c for c in code_obj1.co_consts
if type(c) == types.CodeType)
codes2 = (c for c in code_obj2.co_consts
if type(c) == types.CodeType)
for c1, c2 in zip(codes1, codes2):
cmp_code_objects(version, c1, c2, name=name)
else:
# all other members must be equal
if getattr(code_obj1, member) != getattr(code_obj2, member):
raise CmpErrorMember(name, member, getattr(code_obj1, member),
getattr(code_obj2, member))
def print_code(co, lasti=-1, level=0):
"""Disassemble a code object."""
code = co.co_code
for constant in co.co_consts:
print('| |' * level, end=' ')
print('constant:', constant)
labels = findlabels(code)
linestarts = dict(findlinestarts(co))
n = len(code)
i = 0
extended_arg = 0
free = None
while i < n:
have_inner = False
c = code[i]
op = co_ord(c)
if i in linestarts:
if i > 0:
print()
print('| |' * level, end=' ')
print("%3d" % linestarts[i], end=' ')
else:
print('| |' * level, end=' ')
print(' ', end=' ')
if i == lasti: print('-->', end=' ')
else: print(' ', end=' ')
if i in labels: print('>>', end=' ')
else: print(' ', end=' ')
print(repr(i).rjust(4), end=' ')
print(opcode.opname[op].ljust(20), end=' ')
i = i + 1
if op >= opcode.HAVE_ARGUMENT:
oparg = co_ord(code[i]) + co_ord(code[i + 1]) * 256 + extended_arg
extended_arg = 0
i = i + 2
if op == opcode.EXTENDED_ARG:
extended_arg = oparg * 65536
print(repr(oparg).rjust(5), end=' ')
if op in opcode.hasconst:
print('(' + repr(co.co_consts[oparg]) + ')', end=' ')
if type(co.co_consts[oparg]) == types.CodeType:
have_inner = co.co_consts[oparg]
elif op in opcode.hasname:
print('(' + co.co_names[oparg] + ')', end=' ')
elif op in opcode.hasjrel:
print('(to ' + repr(i + oparg) + ')', end=' ')
elif op in opcode.haslocal:
print('(' + co.co_varnames[oparg] + ')', end=' ')
elif op in opcode.hascompare:
print('(' + opcode.cmp_op[oparg] + ')', end=' ')
elif op in opcode.hasfree:
if free is None:
free = co.co_cellvars + co.co_freevars
print('(' + free[oparg] + ')', end=' ')
print()
if have_inner is not False:
print_code(have_inner, level=level + 1)
def cmp_code_objects(version,
is_pypy,
code_obj1,
code_obj2,
name='',
ignore_code=False):
"""
Compare two code-objects.
This is the main part of this module.
"""
# print code_obj1, type(code_obj2)
assert iscode(code_obj1), \
"cmp_code_object first object type is %s, not code" % type(code_obj1)
assert iscode(code_obj2), \
"cmp_code_object second object type is %s, not code" % type(code_obj2)
# print dir(code_obj1)
if isinstance(code_obj1, object):
# new style classes (Python 2.2)
# assume _both_ code objects to be new stle classes
assert dir(code_obj1) == dir(code_obj2)
else:
# old style classes
assert dir(code_obj1) == code_obj1.__members__
assert dir(code_obj2) == code_obj2.__members__
assert code_obj1.__members__ == code_obj2.__members__
if name == '__main__':
name = code_obj1.co_name
else:
name = '%s.%s' % (name, code_obj1.co_name)
if name == '.?': name = '__main__'
if isinstance(code_obj1, object) and code_equal(code_obj1, code_obj2):
# use the new style code-classes' __cmp__ method, which
# should be faster and more sophisticated
# if this compare fails, we use the old routine to
# find out, what exactly is nor equal
# if this compare succeds, simply return
# return
pass
if isinstance(code_obj1, object):
members = [x for x in dir(code_obj1) if x.startswith('co_')]
else:
members = dir(code_obj1)
members.sort() # ; members.reverse()
tokens1 = None
for member in members:
if member in __IGNORE_CODE_MEMBERS__ or ignore_code:
pass
elif member == 'co_code' and not ignore_code:
if version == 2.3:
import uncompyle6.scanners.scanner23 as scan
scanner = scan.Scanner23(show_asm=False)
elif version == 2.4:
import uncompyle6.scanners.scanner24 as scan
scanner = scan.Scanner24(show_asm=False)
elif version == 2.5:
import uncompyle6.scanners.scanner25 as scan
scanner = scan.Scanner25(show_asm=False)
elif version == 2.6:
import uncompyle6.scanners.scanner26 as scan
scanner = scan.Scanner26(show_asm=False)
elif version == 2.7:
if is_pypy:
import uncompyle6.scanners.pypy27 as scan
scanner = scan.ScannerPyPy27(show_asm=False)
else:
import uncompyle6.scanners.scanner27 as scan
scanner = scan.Scanner27()
elif version == 3.0:
import uncompyle6.scanners.scanner30 as scan
scanner = scan.Scanner30()
elif version == 3.1:
import uncompyle6.scanners.scanner32 as scan
scanner = scan.Scanner32()
elif version == 3.2:
if is_pypy:
import uncompyle6.scanners.pypy32 as scan
scanner = scan.ScannerPyPy32()
else:
import uncompyle6.scanners.scanner32 as scan
scanner = scan.Scanner32()
elif version == 3.3:
import uncompyle6.scanners.scanner33 as scan
scanner = scan.Scanner33()
elif version == 3.4:
import uncompyle6.scanners.scanner34 as scan
scanner = scan.Scanner34()
elif version == 3.5:
import uncompyle6.scanners.scanner35 as scan
scanner = scan.Scanner35()
elif version == 3.6:
import uncompyle6.scanners.scanner36 as scan
scanner = scan.Scanner36()
global JUMP_OPS
JUMP_OPS = list(scan.JUMP_OPS) + ['JUMP_BACK']
# use changed Token class
# We (re)set this here to save exception handling,
# which would get confusing.
scanner.setTokenClass(Token)
try:
# ingest both code-objects
tokens1, customize = scanner.ingest(code_obj1)
del customize # save memory
tokens2, customize = scanner.ingest(code_obj2)
del customize # save memory
finally:
scanner.resetTokenClass() # restore Token class
targets1 = dis.findlabels(code_obj1.co_code)
tokens1 = [t for t in tokens1 if t.kind != 'COME_FROM']
tokens2 = [t for t in tokens2 if t.kind != 'COME_FROM']
i1 = 0
i2 = 0
offset_map = {}
check_jumps = {}
while i1 < len(tokens1):
if i2 >= len(tokens2):
if len(tokens1) == len(tokens2) + 2 \
and tokens1[-1].kind == 'RETURN_VALUE' \
and tokens1[-2].kind == 'LOAD_CONST' \
and tokens1[-2].pattr is None \
and tokens1[-3].kind == 'RETURN_VALUE':
break
else:
raise CmpErrorCodeLen(name, tokens1, tokens2)
offset_map[tokens1[i1].offset] = tokens2[i2].offset
for idx1, idx2, offset2 in check_jumps.get(
tokens1[i1].offset, []):
if offset2 != tokens2[i2].offset:
raise CmpErrorCode(name, tokens1[idx1].offset,
tokens1[idx1], tokens2[idx2],
tokens1, tokens2)
if tokens1[i1].kind != tokens2[i2].kind:
if tokens1[i1].kind == 'LOAD_CONST' == tokens2[i2].kind:
i = 1
while tokens1[i1 + i].kind == 'LOAD_CONST':
i += 1
if tokens1[i1+i].kind.startswith(('BUILD_TUPLE', 'BUILD_LIST')) \
and i == int(tokens1[i1+i].kind.split('_')[-1]):
t = tuple(
[elem.pattr for elem in tokens1[i1:i1 + i]])
if t != tokens2[i2].pattr:
raise CmpErrorCode(name, tokens1[i1].offset,
tokens1[i1], tokens2[i2],
tokens1, tokens2)
i1 += i + 1
i2 += 1
continue
elif i == 2 and tokens1[
i1 + i].kind == 'ROT_TWO' and tokens2[
i2 + 1].kind == 'UNPACK_SEQUENCE_2':
i1 += 3
i2 += 2
continue
elif i == 2 and tokens1[i1 + i].kind in BIN_OP_FUNCS:
f = BIN_OP_FUNCS[tokens1[i1 + i].kind]
if f(tokens1[i1].pattr,
tokens1[i1 + 1].pattr) == tokens2[i2].pattr:
i1 += 3
i2 += 1
continue
elif tokens1[i1].kind == 'UNARY_NOT':
if tokens2[i2].kind == 'POP_JUMP_IF_TRUE':
if tokens1[i1 + 1].kind == 'POP_JUMP_IF_FALSE':
i1 += 2
i2 += 1
continue
elif tokens2[i2].kind == 'POP_JUMP_IF_FALSE':
if tokens1[i1 + 1].kind == 'POP_JUMP_IF_TRUE':
i1 += 2
i2 += 1
continue
elif tokens1[i1].kind in ('JUMP_FORWARD', 'JUMP_BACK') \
and tokens1[i1-1].kind == 'RETURN_VALUE' \
and tokens2[i2-1].kind in ('RETURN_VALUE', 'RETURN_END_IF') \
and int(tokens1[i1].offset) not in targets1:
i1 += 1
continue
elif tokens1[i1].kind == 'JUMP_BACK' and tokens2[
i2].kind == 'CONTINUE':
# FIXME: should make sure that offset is inside loop, not outside of it
i1 += 2
i2 += 2
continue
elif tokens1[i1].kind == 'JUMP_FORWARD' and tokens2[i2].kind == 'JUMP_BACK' \
and tokens1[i1+1].kind == 'JUMP_BACK' and tokens2[i2+1].kind == 'JUMP_BACK' \
and int(tokens1[i1].pattr) == int(tokens1[i1].offset) + 3:
if int(tokens1[i1].pattr) == int(tokens1[i1 +
1].offset):
i1 += 2
i2 += 2
continue
elif tokens1[i1].kind == 'LOAD_NAME' and tokens2[i2].kind == 'LOAD_CONST' \
and tokens1[i1].pattr == 'None' and tokens2[i2].pattr is None:
pass
elif tokens1[i1].kind == 'LOAD_GLOBAL' and tokens2[i2].kind == 'LOAD_NAME' \
and tokens1[i1].pattr == tokens2[i2].pattr:
pass
elif tokens1[i1].kind == 'LOAD_ASSERT' and tokens2[i2].kind == 'LOAD_NAME' \
and tokens1[i1].pattr == tokens2[i2].pattr:
pass
elif (tokens1[i1].kind == 'RETURN_VALUE'
and tokens2[i2].kind == 'RETURN_END_IF'):
pass
elif (tokens1[i1].kind == 'BUILD_TUPLE_0'
and tokens2[i2].pattr == ()):
pass
else:
raise CmpErrorCode(name, tokens1[i1].offset,
tokens1[i1], tokens2[i2], tokens1,
tokens2)
elif tokens1[i1].kind in JUMP_OPS and tokens1[
i1].pattr != tokens2[i2].pattr:
if tokens1[i1].kind == 'JUMP_BACK':
dest1 = int(tokens1[i1].pattr)
dest2 = int(tokens2[i2].pattr)
if offset_map[dest1] != dest2:
raise CmpErrorCode(name, tokens1[i1].offset,
tokens1[i1], tokens2[i2],
tokens1, tokens2)
else:
# import pdb; pdb.set_trace()
try:
dest1 = int(tokens1[i1].pattr)
if dest1 in check_jumps:
check_jumps[dest1].append((i1, i2, dest2))
else:
check_jumps[dest1] = [(i1, i2, dest2)]
except:
pass
i1 += 1
i2 += 1
del tokens1, tokens2 # save memory
elif member == 'co_consts':
# partial optimization can make the co_consts look different,
# so we'll just compare the code consts
codes1 = (c for c in code_obj1.co_consts
if hasattr(c, 'co_consts'))
codes2 = (c for c in code_obj2.co_consts
if hasattr(c, 'co_consts'))
for c1, c2 in zip(codes1, codes2):
cmp_code_objects(version, is_pypy, c1, c2, name=name)
elif member == 'co_flags':
flags1 = code_obj1.co_flags
flags2 = code_obj2.co_flags
if is_pypy:
# For PYPY for now we don't care about PYPY_SOURCE_IS_UTF8:
flags2 &= ~0x0100 # PYPY_SOURCE_IS_UTF8
# We also don't care about COROUTINE or GENERATOR for now
flags1 &= ~0x000000a0
flags2 &= ~0x000000a0
if flags1 != flags2:
raise CmpErrorMember(name, 'co_flags', pretty_flags(flags1),
pretty_flags(flags2))
else:
# all other members must be equal
if getattr(code_obj1, member) != getattr(code_obj2, member):
raise CmpErrorMember(name, member, getattr(code_obj1, member),
getattr(code_obj2, member))
