Bug 2207692

Samba has a python/samba/safe_tarfile.py [1] which is used by netcmd/domain_backup.py and others. However it doesn't overwrite extractall() only extract(). But looking at the test at [2] extratall seems to be covered as well.

So if we pass filter='data'


I think if we just pass filter='data' always to tarfile extract() we are fine? Or do we need to also add extractall() as it will print the warning otherwise, then we need to also implement extractall() in python/samba/safe_tarfile.py



 python/samba/safe_tarfile.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/python/samba/safe_tarfile.py b/python/samba/safe_tarfile.py
index cc19770d73f..489ccb1a65b 100644
--- a/python/samba/safe_tarfile.py
+++ b/python/samba/safe_tarfile.py
@@ -24,7 +24,8 @@ class TarFile(UnsafeTarFile):
     using '../../'.
     """
 
-    def extract(self, member, path="", set_attrs=True, *, numeric_owner=False):
+    def extract(self, member, path="", set_attrs=True, *, numeric_owner=False,
+                filter="data"):
         if isinstance(member, TarInfo):
             name = member.name
         else:
@@ -37,7 +38,7 @@ class TarFile(UnsafeTarFile):
             raise ExtractError(f"path '{name}' should not start with '/'")
 
         super().extract(member, path, set_attrs=set_attrs,
-                        numeric_owner=numeric_owner)
+                        numeric_owner=numeric_owner, filter=filter)
 
 
 open = TarFile.open


[1] https://gitlab.com/samba-team/samba/-/blob/master/python/samba/safe_tarfile.py
[2] https://gitlab.com/samba-team/samba/-/blob/master/python/samba/tests/safe_tarfile.py

For passing the filter, you'd also need to provide `extractall`. Beware that `extractall` does *not* call `extract`, they're speparate "frontends" for the internals.
Also, the filter argument isn't available on older/unpatched versions of Python, so you probably can't add it unconditionally.

But since you have a subclass, things are easier: you can just add an attribute to it:

    import tarfile

    class TarFile(UnsafeTarFile):
        try:
            extraction_filter = staticmethod(tarfile.data_filter)
        except AttributeError:
            ... # override extract/extractall for earlier/unpatched versions

See the docs: https://docs.python.org/3/library/tarfile.html#tarfile.TarFile.extraction_filter
(Please let me know if anything is unclear/missing, Samba would be the first project I know doing it this way.)


Note that the errors raised are subclasses of TarError, but *not* ExtractError like your existing safe TarFile.

Upstream Merge Request: https://gitlab.com/samba-team/samba/-/merge_requests/3114