Skip to content
/ NF-Save Public

Take in a data structure and output nf/iptables and ipset save data

License

Notifications You must be signed in to change notification settings

ag4ve/NF-Save

Repository files navigation

NAME

NF::Save - Module for storing, parsing, and restoring iptables and ipset data.

SYNOPSIS

use NF::Save;

my $oIPT = NF::Save->new(
  {
    'Policy'  =>
    {
      'filter'    =>
      {
        'INPUT'       => 'DROP',
        'OUTPUT'      => 'DROP',
        'FORWARD'     => 'DROP',
      },
    }
  }
);

my $paStruct = 
[
  {
    'in' => "eth0",
    'udp' => 
    {
      'sport' => "1024:65535",
      'dport' => "53",
    }, 
    'dst' => "192.168.15.1",
    'comment' => [qw/nameserver/],
    'jump' => "ACCEPT",
  },
  {
    # more rules
  },
];

# Add rules to the OUTPUT chain.
foreach my $phRule (@$paStruct)
{
  $oIPT->rule('OUTPUT', $phRule);
}

# Get a set of rules that could be used with: 
# $ ./firewall.pl | iptables-restore

# And to get the ipset set
print "$_\n" for ($oIPT->save());

DESCRIPTION

NF-Save is intended to let perl import/export firewall rules and ipsets in the exact syntax used by iptables-save / iptables-restore and ipset save / ipset restore: same order of arguments, same way of expanding tokens, same defaults when things weren't explicitly specified, etc.

That way, existing policies can be imported easily and completely, and policies composed in perl code can be written out in a way that is 100% diff-compatible with iptables-save output.

More less used methods are documented in NF::Save::Misc. If you wish to create new modules, see NF::Save::ModuleDoc.

Common methods

new(%Options)

%Options:

  • %UIDs contains a hash of {'username' => #id}
  • @IPTLookup contains replacement data to be used to handle the data structure (an index with an undefined value will not effect the original array).
  • @Flags contains a hash of flags to be used when --syn/mss/etc would have been used - (arbitrary names can be defined).
  • $UseIPSET boolean - whether or not to default lists as ipset.
  • %Policy default policy to use.
  • $Trace boolean - whether or not to print a stack trace with warnings.
  • $PreCheck boolean - whether or not to pre-check the structure passed to rule().
  • @Modules list of NF::Save modules to use. If this is a string, all modules in this namespace will be loaded.

get($sChain, $sTable)

Return the internal data structure used to store iptables information.

useipset($sBool)

Return whether ipset is used by default and optionally change whether or not ipset is used by default. So: $ipt->useipset(1); is the same as: $ipt->new({UseIPSET => 1});

rule($sChain, $sRule, $sTable, $sFunc)

An interface designed to look fairly similar to the iptables CLI.

The tcp '--syn' and '! --syn' options add masks from individual from the $rhFlags hashref.

The big difference is that the chain is seperate from the action.

  • This: iptables -I INPUT 5 -j ACCEPT
  • Turns into this: $ipt->rule('INPUT', {jump => 'ACCEPT'}, undef, 'I 5');
  • The default is to APPEND to the filter table, which means the pattern is: $ipt->rule('INPUT', {jump => 'ACCEPT'});
  • Delete and replace have been implemented for completeness - for replace: $ipt->rule('OUTPUT', {jump => 'ACCEPT'}, 'filter', 'R 5');

get_ipset_data($sName)

Return internal data for an ipset or all sets if no name was given.

save()

Return an array that can pe passed to iptables-restore. This data should duplicate iptables-save so that data generated with this and restored into iptables would show no differece when compared to the output of:

iptables-save

assemble(%$phParams, $sChain, $check)

Create an iptables rule for a data structure definition. The chain name and whether to check the ruleset are optional.

DOCUMENTATION SYNTAX

When defining the values in a hash, perl data types are used to describe the value type and variable names that are used to define the key name. Documentation may define an alternative data type not self documented in this syntax. A hash that looks like:

{ array => [qw/1 2 3/], hash => {"a" => "1", "b" => "2"}, string => "foo", }

Would be defined as:

  • @array description
  • %hash description
  • $string description

TODO

  • Need more tests (branch coverage is ~40%) and use cases
    • Need to handle more modules
  • Integration with libiptc using FFI or similar instead of using IPC
  • Make NF::RuleParser to make a NF::Save structure from iptables
  • IPT allows deletion on exact rule match - not supported here

AUTHOR

Shawn Wilson ag4ve.us@gmail.com

THANKS

A special thanks to Korelogic for giving inspiration and funding for most of this work and allowing it to be open sourced.

COPYRIGHT

Copyright 2017 - Shawn Wilson

LICENSE

The GNU Lesser General Public License, version 3.0 (LGPL-3.0) http://opensource.org/licenses/LGPL-3.0