def extract_insn_bytes_features(f, bb, insn):
"""parse referenced byte sequences
args:
f (IDA func_t)
bb (IDA BasicBlock)
insn (IDA insn_t)
example:
push offset iid_004118d4_IShellLinkA ; riid
"""
for ref in insn.dr:
ref = str(ref)
found = None
# check string first (ghidra/ida put it there)
if ref in f.unit.obj.bin.strings:
found = f.unit.obj.bin.strings[ref]
yield Bytes(found.encode("utf-16le")), insn.ea
if found:
return
# then check referenced data
if ref in f.unit.obj.bin.data:
found = f.unit.obj.bin.data[ref]
# found = struct.pack('<Q', int(found, base=16))
# found = bytes.fromhex(found)
found = base64.b64decode(found)
yield Bytes(found), insn.ea
def extract_insn_bytes_features(f, bb, insn):
"""
parse byte sequence features from the given instruction.
example:
# push offset iid_004118d4_IShellLinkA ; riid
"""
for oper in insn.opers:
if insn.mnem == "call":
continue
if isinstance(oper, envi.archs.i386.disasm.i386ImmOper):
v = oper.getOperValue(oper)
elif isinstance(oper, envi.archs.i386.disasm.i386RegMemOper):
# handle case like:
# movzx ecx, ds:byte_423258[eax]
v = oper.disp
elif isinstance(oper, envi.archs.i386.disasm.i386SibOper):
# like 0x401000 in `mov eax, 0x401000[2 * ebx]`
v = oper.imm
elif isinstance(oper, envi.archs.amd64.disasm.Amd64RipRelOper):
# see: Lab21-01.exe_:0x1400010D3
v = oper.getOperAddr(insn)
else:
continue
for v in derefs(f.vw, v):
try:
buf = read_bytes(f.vw, v)
except envi.SegmentationViolation:
continue
if capa.features.extractors.helpers.all_zeros(buf):
continue
yield Bytes(buf), insn.va
def extract_insn_bytes_features(f, bb, insn):
"""
parse byte sequence features from the given instruction.
example:
# push offset iid_004118d4_IShellLinkA ; riid
"""
for data_ref in insn.getDataRefs():
for v in derefs(f.smda_report, data_ref):
bytes_read = read_bytes(f.smda_report, v)
if bytes_read is None:
continue
if capa.features.extractors.helpers.all_zeros(bytes_read):
continue
yield Bytes(bytes_read), insn.offset
def extract_insn_bytes_features(f, bb, insn):
""" parse referenced byte sequences
args:
f (IDA func_t)
bb (IDA BasicBlock)
insn (IDA insn_t)
example:
push offset iid_004118d4_IShellLinkA ; riid
"""
if idaapi.is_call_insn(insn):
# ignore call instructions
return
for ref in idautils.DataRefsFrom(insn.ea):
extracted_bytes = capa.features.extractors.ida.helpers.read_bytes_at(ref, MAX_BYTES_FEATURE_SIZE)
if extracted_bytes and not capa.features.extractors.helpers.all_zeros(extracted_bytes):
yield Bytes(extracted_bytes), insn.ea
def extract_insn_bytes_features(f, bb, insn):
"""parse referenced byte sequences
args:
f (IDA func_t)
bb (IDA BasicBlock)
insn (IDA insn_t)
example:
push offset iid_004118d4_IShellLinkA ; riid
"""
if idaapi.is_call_insn(insn):
return
ref = capa.features.extractors.ida.helpers.find_data_reference_from_insn(insn)
if ref != insn.ea:
extracted_bytes = capa.features.extractors.ida.helpers.read_bytes_at(ref, MAX_BYTES_FEATURE_SIZE)
if extracted_bytes and not capa.features.extractors.helpers.all_zeros(extracted_bytes):
yield Bytes(extracted_bytes), insn.ea
def extract_insn_bytes_features(f, bb, insn):
"""
parse byte sequence features from the given instruction.
example:
# push offset iid_004118d4_IShellLinkA ; riid
"""
for oper in insn.opers:
if insn.mnem == "call":
# ignore call instructions
continue
if isinstance(oper, envi.archs.i386.disasm.i386ImmOper):
v = oper.getOperValue(oper)
elif isinstance(oper, envi.archs.i386.disasm.i386RegMemOper):
# handle case like:
# movzx ecx, ds:byte_423258[eax]
v = oper.disp
elif isinstance(oper, envi.archs.amd64.disasm.Amd64RipRelOper):
# see: Lab21-01.exe_:0x1400010D3
v = oper.getOperAddr(insn)
else:
continue
segm = f.vw.getSegment(v)
if not segm:
continue
segm_end = segm[0] + segm[1]
try:
# Do not read beyond the end of a segment
if v + MAX_BYTES_FEATURE_SIZE > segm_end:
extracted_bytes = f.vw.readMemory(v, segm_end - v)
else:
extracted_bytes = f.vw.readMemory(v, MAX_BYTES_FEATURE_SIZE)
except envi.SegmentationViolation:
pass
else:
if not capa.features.extractors.helpers.all_zeros(extracted_bytes):
yield Bytes(extracted_bytes), insn.va
