def add_filter(info: tarfile.TarInfo) -> Optional[tarfile.TarInfo]:
"""Filter files targeted to be added to tarfile.
Args:
info: Information on the file targeted to be added
Returns:
None: if file is not to be added
TarInfo: when file is to be added. Modified as needed.
Notes:
exclude is captured from parent
"""
if not (info.isfile() or info.isdir() or info.issym()):
return None
if _exclude_matcher(info.name, exclude):
return None
# Workaround https://bugs.python.org/issue32713. Fixed in Python 3.7
if info.mtime < 0 or info.mtime > 8 ** 11 - 1:
info.mtime = int(info.mtime)
# do not leak client information to service
info.uid = 0
info.uname = info.gname = "root"
if sys.platform == "win32":
info.mode = info.mode & 0o755 | 0o111
return info
def sanitize_tarinfo(tarinfo: tarfile.TarInfo):
path = Path(tarinfo.name)
# Ban absolute paths
if path.is_absolute():
return False
# Ban paths containing .. that would go outside
try:
base = Path("fake_path")
base.joinpath(path).relative_to(base)
except ValueError as e:
logger.warning(f"Unsafe path {path}")
return False
# Ban links that would point somewhere outside
if tarinfo.islnk() or tarinfo.issym():
try:
base = Path("fake_path")
link = path.parent / tarinfo.linkname
base.joinpath(link).relative_to(base)
except ValueError as e:
logger.warning(f"Unsafe path {path}")
return False
return True
def _is_file_type_forbidden(tarinfo: tarfile.TarInfo) -> bool:
return (tarinfo.islnk() or tarinfo.isblk() or tarinfo.ischr()
or tarinfo.isdev() or tarinfo.isfifo() or tarinfo.issym()
or tarinfo.islnk())
def _member_is_safe(member: tarfile.TarInfo) -> bool:
if member.isfile():
return getattr(member, 'sparse', None) is None
return member.isdir() or member.issym() or member.islnk()
def strip_prefix(self, prefix: str, member: tarfile.TarInfo):
member.name = self.strip_slash(prefix, member.name)
# Strip hardlinks
if member.islnk() and not member.issym():
member.linkname = self.strip_slash(prefix, member.linkname)
def _member_is_safe(member: tarfile.TarInfo) -> bool:
return (
member.isfile() or member.isdir() or member.issym() or
member.islnk()
)
