def findlabels(code, opc):
"""Detect all offsets in a byte code which are jump targets.
Return the list of offsets.
"""
labels = []
# enumerate() is not an option, since we sometimes process
# multiple elements on a single pass through the loop
try:
n = len(code)
except:
code = code.co_code
n = len(code)
offset = 0
while offset < n:
op = code2num(code, offset)
offset += 1
if op >= opc.HAVE_ARGUMENT:
arg = code2num(code, offset) + code2num(code, offset+1)*256
offset += 2
label = -1
if op in opc.JREL_OPS:
label = offset + arg
elif op in opc.JABS_OPS:
label = arg
if label >= 0:
if label not in labels:
labels.append(label)
return labels
def unpack_opargs_bytecode(code, opc):
extended_arg = 0
try:
n = len(code)
except TypeError:
code = code.co_code
n = len(code)
offset = 0
while offset < n:
prev_offset = offset
op = code2num(code, offset)
offset += 1
if op_has_argument(op, opc):
arg = code2num(code, offset) | extended_arg
extended_arg = extended_arg_val(opc, arg) if op == opc.EXTENDED_ARG else 0
offset += 2
else:
arg = None
yield (prev_offset, op, arg)
def get_instructions_bytes(bytecode,
opc,
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 = opc.findlabels(bytecode, opc)
# label_maps = get_jump_target_maps(bytecode, opc)
extended_arg = 0
# FIXME: We really need to distinguish 3.6.0a1 from 3.6.a3.
# See below FIXME
python_36 = True if opc.python_version >= 3.6 else False
starts_line = None
# enumerate() is not an option, since we sometimes process
# multiple elements on a single pass through the loop
n = len(bytecode)
i = 0
extended_arg_count = 0
extended_arg = 0
extended_arg_size = op_size(opc.EXTENDED_ARG, opc)
while i < n:
op = code2num(bytecode, i)
offset = i
if linestarts is not None:
starts_line = linestarts.get(i, None)
if starts_line is not None:
starts_line += line_offset
if i in labels:
# come_from = label_maps[i]
if False: # come_from[0] > i:
is_jump_target = 'loop'
# print("XXX %s at %d" % (opc.opname[op], i))
# from trepan.api import debug; debug()
else:
is_jump_target = True
else:
is_jump_target = False
i += 1
arg = None
argval = None
argrepr = ''
has_arg = op_has_argument(op, opc)
optype = None
if has_arg:
if python_36:
arg = code2num(bytecode, i) | extended_arg
extended_arg = (arg << 8) if op == opc.EXTENDED_ARG else 0
# FIXME: Python 3.6.0a1 is 2, for 3.6.a3 we have 1
i += 1
else:
arg = code2num(
bytecode,
i) + code2num(bytecode, i + 1) * 256 + extended_arg
i += 2
extended_arg = arg * 65536 if op == opc.EXTENDED_ARG else 0
# 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 opc.CONST_OPS:
argval, argrepr = _get_const_info(arg, constants)
optype = 'const'
elif op in opc.NAME_OPS:
argval, argrepr = _get_name_info(arg, names)
optype = 'name'
elif op in opc.JREL_OPS:
argval = i + arg
argrepr = "to " + repr(argval)
optype = 'jrel'
elif op in opc.JABS_OPS:
argval = arg
argrepr = "to " + repr(argval)
optype = 'jabs'
elif op in opc.LOCAL_OPS:
argval, argrepr = _get_name_info(arg, varnames)
optype = 'local'
elif op in opc.COMPARE_OPS:
argval = opc.cmp_op[arg]
argrepr = argval
optype = 'compare'
elif op in opc.FREE_OPS:
argval, argrepr = _get_name_info(arg, cells)
optype = 'free'
elif op in opc.NARGS_OPS:
optype = 'nargs'
if not python_36:
argrepr = (
"%d positional, %d keyword pair" %
(code2num(bytecode, i - 2), code2num(bytecode, i - 1)))
# This has to come after hasnargs. Some are in both?
elif op in opc.VARGS_OPS:
optype = 'vargs'
if hasattr(
opc,
'opcode_arg_fmt') and opc.opname[op] in opc.opcode_arg_fmt:
argrepr = opc.opcode_arg_fmt[opc.opname[op]](arg)
elif python_36:
i += 1
opname = opc.opname[op]
inst_size = op_size(op, opc) + (extended_arg_count * extended_arg_size)
yield Instruction(opname, op, optype, inst_size, arg, argval, argrepr,
has_arg, offset, starts_line, is_jump_target,
extended_arg_count != 0)
extended_arg_count = extended_arg_count + 1 if op == opc.EXTENDED_ARG else 0
def get_instructions_bytes(bytecode, opc, 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 = opc.findlabels(bytecode, opc)
# label_maps = get_jump_target_maps(bytecode, opc)
extended_arg = 0
# FIXME: We really need to distinguish 3.6.0a1 from 3.6.a3.
# See below FIXME
python_36 = True if opc.python_version >= 3.6 else False
starts_line = None
# enumerate() is not an option, since we sometimes process
# multiple elements on a single pass through the loop
n = len(bytecode)
i = 0
extended_arg_count = 0
extended_arg = 0
extended_arg_size = op_size(opc.EXTENDED_ARG, opc)
while i < n:
op = code2num(bytecode, i)
offset = i
if linestarts is not None:
starts_line = linestarts.get(i, None)
if starts_line is not None:
starts_line += line_offset
if i in labels:
# come_from = label_maps[i]
if False: # come_from[0] > i:
is_jump_target = 'loop'
# print("XXX %s at %d" % (opc.opname[op], i))
# from trepan.api import debug; debug()
else:
is_jump_target = True
else:
is_jump_target = False
i += 1
arg = None
argval = None
argrepr = ''
has_arg = op_has_argument(op, opc)
optype = None
if has_arg:
if python_36:
arg = code2num(bytecode, i) | extended_arg
extended_arg = (arg << 8) if op == opc.EXTENDED_ARG else 0
# FIXME: Python 3.6.0a1 is 2, for 3.6.a3 we have 1
i += 1
else:
arg = code2num(bytecode, i) + code2num(bytecode, i+1)*256 + extended_arg
i += 2
extended_arg = arg*65536 if op == opc.EXTENDED_ARG else 0
# 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 opc.CONST_OPS:
argval, argrepr = _get_const_info(arg, constants)
optype = 'const'
elif op in opc.NAME_OPS:
argval, argrepr = _get_name_info(arg, names)
optype = 'name'
elif op in opc.JREL_OPS:
argval = i + arg
argrepr = "to " + repr(argval)
optype = 'jrel'
elif op in opc.JABS_OPS:
argval = arg
argrepr = "to " + repr(argval)
optype = 'jabs'
elif op in opc.LOCAL_OPS:
argval, argrepr = _get_name_info(arg, varnames)
optype = 'local'
elif op in opc.COMPARE_OPS:
argval = opc.cmp_op[arg]
argrepr = argval
optype = 'compare'
elif op in opc.FREE_OPS:
argval, argrepr = _get_name_info(arg, cells)
optype = 'free'
elif op in opc.NARGS_OPS:
optype = 'nargs'
if not python_36:
argrepr = ("%d positional, %d keyword pair" %
(code2num(bytecode, i-2), code2num(bytecode, i-1)))
# This has to come after hasnargs. Some are in both?
elif op in opc.VARGS_OPS:
optype = 'vargs'
if hasattr(opc, 'opcode_arg_fmt') and opc.opname[op] in opc.opcode_arg_fmt:
argrepr = opc.opcode_arg_fmt[opc.opname[op]](arg)
elif python_36:
i += 1
opname = opc.opname[op]
inst_size = op_size(op, opc) + (extended_arg_count * extended_arg_size)
yield Instruction(opname, op, optype, inst_size, arg, argval, argrepr,
has_arg, offset, starts_line, is_jump_target,
extended_arg_count != 0)
extended_arg_count = extended_arg_count + 1 if op == opc.EXTENDED_ARG else 0
