You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
ipaddress.reverse_pointer returns nonsense if prefixed input is given #91601
ipaddress.reverse_pointer does not seem to validate its input at all. For example:
$ python3
Python 3.10.4 (main, Apr 2 2022, 09:04:19) [GCC 11.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import ipaddress
>>> ipaddress.ip_network('2001:db8:1:2:3:4:5::/64',False).reverse_pointer
'4.6./.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0.1.0.0.0.8.b.d.0.1.0.0.2.ip6.arpa'
>>>
This is very obviously wrong.
In this case of prefixed input, the output should be cut to the appropriate prefix length, yielding
2.0.0.0.1.0.0.0.8.b.d.0.1.0.0.2.ip6.arpa as output. It is ok to assume /128 as the default, converting 2001:db8:1:2:3:4:5:: to 0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0.1.0.0.0.8.b.d.0.1.0.0.2.ip6.arpa, but input longer than that is an error.
Prefixed address strings are for example important when generating zone names for the server side of IPv6 reverse DNS. Run-of-the-mill network client software is unlikely to use this for addresses shorter than /128, but for the server side it is an important use case.
Still reproducible on released versions, and the snippet in the report no longer runs as written.
On 3.14.8 and 3.15.2, ipaddress.ip_network('2001:db8:1:2:3:4:5::/64') raises ValueError: 2001:db8:1:2:3:4:5:0/64 has host bits set, because strict defaults to true — so the exact example in the report errors out before reaching reverse_pointer.
Going through ip_interface, the nonsense is unchanged on both versions:
The 4.6./. prefix is the artifact the report describes, and the interface still knows better: its .network is 2001:db8:1:2::/64, so the host bits are only kept in the address that reverse_pointer renders.
Activity
iacore commentedon Nov 3, 2026
Still reproducible on released versions, and the snippet in the report no longer runs as written.
On 3.14.8 and 3.15.2,
ipaddress.ip_network('2001:db8:1:2:3:4:5::/64')raisesValueError: 2001:db8:1:2:3:4:5:0/64 has host bits set, becausestrictdefaults to true — so the exact example in the report errors out before reachingreverse_pointer.Going through
ip_interface, the nonsense is unchanged on both versions:The
4.6./.prefix is the artifact the report describes, and the interface still knows better: its.networkis2001:db8:1:2::/64, so the host bits are only kept in the address thatreverse_pointerrenders.