def __init__(self, reader: StructReader):
reader.bigendian = True
entry_start_offset = reader.tell()
self.size_of_entry = reader.i32()
self.offset = reader.i32()
self.size_of_compressed_data = reader.i32()
self.size_od_uncompressed_data = reader.i32()
self.is_compressed = bool(reader.read_byte())
entry_type = bytes(reader.read(1))
name_length = self.size_of_entry - reader.tell() + entry_start_offset
if name_length > 0x1000:
raise RuntimeError(
F'Refusing to process TOC entry with name of size {name_length}.'
)
name, *_ = bytes(reader.read(name_length)).partition(B'\0')
try:
name = name.decode('utf8', 'backslashreplace')
except Exception:
name = None
if not all(part.isprintable() for part in re.split('\\s*', name)):
raise RuntimeError(
'Refusing to process TOC entry with non-printable name.')
name = name or str(uuid.uuid4())
if entry_type == B'Z':
entry_type = B'z'
try:
self.type = PiType(entry_type)
except ValueError:
xtpyi.logger.error(F'unknown type {entry_type!r} in field {name}')
self.type = PiType.UNKNOWN
self.name = name
def __init__(self, reader: StructReader, version: str):
reader.bigendian = True
self.base = reader.tell()
signature = reader.read(4)
if signature != self.MagicSignature:
raise ValueError('invalid magic')
magic = bytes(reader.read(4))
with contextlib.suppress(KeyError):
version = xtpyi._xdis.magics.versions[magic]
vtuple = version2tuple(version)
padding_size = 4
if vtuple >= (3, 3):
padding_size += 4
if vtuple >= (3, 7):
padding_size += 4
self.version = version
self.magic = magic + padding_size * b'\0'
self.toc_offset = reader.i32()
self.reader = reader
self.entries: List[PiMeta] = []
def test_bitreader_structured(self):
items = (
0b1100101, # noqa
-0x1337, # noqa
0xDEFACED, # noqa
0xC0CAC01A, # noqa
-0o1337, # noqa
2076.171875, # noqa
math.pi # noqa
)
data = struct.pack('<bhiLqfd', *items)
sr = StructReader(data)
self.assertEqual(sr.read_nibble(), 0b101)
self.assertRaises(sr.Unaligned, lambda: sr.read_exactly(2))
sr.seek(0)
self.assertEqual(sr.read_byte(), 0b1100101)
self.assertEqual(sr.i16(), -0x1337)
self.assertEqual(sr.i32(), 0xDEFACED)
self.assertEqual(sr.u32(), 0xC0CAC01A)
self.assertEqual(sr.i64(), -0o1337)
self.assertAlmostEqual(sr.read_struct('f', True), 2076.171875)
self.assertAlmostEqual(sr.read_struct('d', True), math.pi)
self.assertTrue(sr.eof)
def __init__(self,
reader: StructReader,
offset: int,
unmarshal: Unmarshal = Unmarshal.No):
reader.bigendian = True
reader.seekset(offset)
self.reader = reader
signature = reader.read_bytes(8)
if signature != self.MagicSignature:
raise ValueError(
F'offset 0x{offset:X} has invalid signature {signature.hex().upper()}; '
F'should be {self.MagicSignature.hex().upper()}')
self.size = reader.i32()
toc_offset = reader.i32()
toc_length = reader.i32()
self.py_version = '.'.join(str(reader.u32()))
self.py_libname = self._read_libname(reader)
self.offset = reader.tell() - self.size
self.toc: Dict[str, PiTOCEntry] = {}
toc_end = self.offset + toc_offset + toc_length
reader.seekset(self.offset + toc_offset)
while reader.tell() < toc_end:
try:
entry = PiTOCEntry(reader)
except EOF:
xtpyi.logger.warning('end of file while reading TOC')
break
except Exception as error:
xtpyi.logger.warning(
F'unexpected error while reading TOC: {error!s}')
break
if entry.name in self.toc:
raise KeyError(F'duplicate name {entry.name}')
self.toc[entry.name] = entry
self.files: Dict[str, PiMeta] = {}
no_pyz_found = True
pyz_entries: Dict[str, PYZ] = {}
for entry in list(self.toc.values()):
if entry.type is not PiType.PYZ:
continue
no_pyz_found = False
name, xt = os.path.splitext(entry.name)
name_pyz = F'{name}.pyz'
if name == entry.name:
del self.toc[name]
self.toc[name_pyz] = entry
entry.name = name_pyz
reader.seekset(self.offset + entry.offset)
if entry.is_compressed:
data = self.extract(entry.name).unpack()
else:
data = reader
pyz_entries[name] = PYZ(data, self.py_version)
magics = {pyz.magic for pyz in pyz_entries.values()}
if not magics:
if not no_pyz_found:
xtpyi.logger.warning(
'no magic signature could be recovered from embedded pyzip archives; this is '
'unsual and means that there is no way to guess the missing magic for source '
'file entries and it will likely not be possible to decompile them.'
)
return
elif len(magics) > 1:
xtpyi.logger.warning(
'more than one magic signature was recovered; this is unusual.'
)
magics = list(magics)
keys: Set[bytes] = set()
for entry in self.toc.values():
extracted = self.extract(entry.name)
if entry.type not in (PiType.SOURCE, PiType.MODULE):
self.files[entry.name] = extracted
continue
data = extracted.unpack()
name, _ = os.path.splitext(extracted.name)
del self.files[extracted.name]
extracted.name = F'{name}.pyc'
self.files[extracted.name] = extracted
if len(magics) == 1 and data[:4] != magics[0]:
extracted.data = magics[0] + data
decompiled = make_decompiled_item(name, data, *magics)
if entry.type is PiType.SOURCE:
decompiled.type = PiType.USERCODE
self.files[F'{name}.py'] = decompiled
if name.endswith('crypto_key'):
for key in decompiled.unpack() | carve('string', decode=True):
if len(key) != 0x10:
continue
xtpyi.logger.info(F'found key: {key.decode(xtpyi.codec)}')
keys.add(key)
if unmarshal is Unmarshal.No:
return
if not keys:
key = None
else:
key = next(iter(keys))
for name, pyz in pyz_entries.items():
pyz.unpack(unmarshal is Unmarshal.YesAndDecompile, key)
for unpacked in pyz.entries:
unpacked.name = path = F'{name}/{unpacked.name}'
if path in self.files:
raise ValueError(F'duplicate file name: {path}')
self.files[path] = unpacked
