The ReplaceTags Plugin
This SpamAssassin plugin module allows you to create various character-classes and use them in your rules. The basic advantage is that you don't have to write huge rules, which are harder to debug, but can use more readable shortcuts. Anotherone is that it gets pretty easy to extend rules by only modifying the class. It is also pretty customizeable for individual start/end tags and which tests are parsed.
Code
Add the following to your local.cf file:
loadplugin ReplaceTags /path/to/plugin/ReplaceTags.pm
replace_rules VIAGRA_OBFU,CIALIS_OBFU # ... and so on
replace_start_sign <
replace_end_sign >
# Example classes
class A [\@\xc0\xc1\xc2\xc3\xc4\xc5\xc6\xe4\xe3\xe2\xe0\xe1\xe2\xe3\xe4\xe5\xe6]
class G [gk]
class I [il\|\!1y\?\xcc\xcd\xce\xcf\xec\xed\xee\xef]
class R [r3]
class V [v\\\/wu]
class SP [\s\d\_\-\*\$\%\(\)\,\.\:\;\?\!\}\{\[\]\|\/\?\^\#\~\xa1\\`\'\+]
# This is only an example for a pair of rules
body VIAGRA_OBFU /(?!viagra)<V>+<SP>*<I>+<SP>*<A>+<SP>*<G>+<SP>*<R>+<SP>*<A>+/i
describe VIAGRA_OBFU obfuscated match "viagra"
score VIAGRA_OBFU 3
body VIAGRA /viagra/i
describe VIAGRA match plain "viagra"
score VIAGRA 1.5
ReplaceTags.pm
=head1 NAME
ReplaceTags - Create character-classes for SpamAssassin-rules
A character-class may contain of any character which is valid in a regular expression.
The grouped characters are easy to maintain and the rules are more readable.
=head1 SYNOPSIS
loadplugin ReplaceTags /path/to/plugin/ReplaceTags.pm
replace_rules VIAGRA_OBFU,CIALIS_OBFU,...
replace_start_sign <
replace_end_sign >
class A [\@\xc0\xc1\xc2\xc3\xc4\xc5\xc6\xe4\xe3\xe2\xe0\xe1\xe2\xe3\xe4\xe5\xe6]
class G [gk]
class I [il\|\!1y\?\xcc\xcd\xce\xcf\xec\xed\xee\xef]
class R [r3]
class V [v\\\/wu]
class SP [\s\d\_\-\*\$\%\(\)\,\.\:\;\?\!\}\{\[\]\|\/\?\^\#\~\xa1\\`\'\+]
body VIAGRA_OBFU /(?!viagra)<V>+<SP>*<I>+<SP>*<A>+<SP>*<G>+<SP>*<R>+<SP>*<A>+/i
describe VIAGRA_OBFU obfuscated match "viagra"
score VIAGRA_OBFU 3
body VIAGRA /viagra/i
describe VIAGRA match plain "viagra"
score VIAGRA 1.5
This example displays how this plugin can be used to match obfuscated and "normal" phrases, which makes it easier
to increase the scores for rules matching only obfuscated patterns on a real world example.
=head1 CONFIGURATION
=over 4
=item replace_rules list_of_tests
Specify a commaspererated list of symbolic test names of tests which should be parsed. The test will only be
parsed if it is a body, header, uri, full or raw test.
=item replace_start_sign sign
=item replace_end_sign sign
Character(s) which indicate the start and end of a class inside a rule. If the class is not enclosed by this
signs it won't be found nor replaced. If you encounter problems run spamassassin from the commandline with
the -D flag and check the output. The default values are < (for replace_start_sign) and > (for replace_end_sign).
=item class classname characters
Define a character class. Valid characters are the same as in any usual regular expression. It is not a good idea to
put quantifiers inside the character class, better use them inside your rule is shown above in the example.
=cut
package ReplaceTags;
use strict;
use Mail::SpamAssassin;
use Mail::SpamAssassin::Plugin;
use Switch;
our @ISA = qw|Mail::SpamAssassin::Plugin|;
sub new {
my ($class, $mailsa) = @_;
$class = ref($class) || $class;
my $self = $class->SUPER::new($mailsa);
bless ($self, $class);
return $self;
}
sub check_start {
# This is the point where the rulesets get replaced with our stuff
my ($self,$pms) = @_;
my $start_tag = $self->{'replace_start_sign'};
my $end_tag = $self->{'replace_end_sign'};
# Put the names of the rules, which get replaced in a hash, for easier usage
my %Rules = ();
for my $rule_name (@{$self->{'rules_to_replace'}}) {
$Rules{$rule_name} = 1;
}
for my $rule_set (qw|body_tests rawbody_tests head_tests full_tests uri_tests|) {
# TODO check what that 0 is for (I guess 0 are the GLOBAL tests and user-tests will
# have a number (uid?)...guess is wrong :( )
for my $rule_name (keys %{$pms->{'permsgstatus'}->{'conf'}->{$rule_set}->{0}}) {
# If rulename doesn't match, skip
next unless ($Rules{$rule_name});
# Increase it here, so that we can do some small debugging at the end
$Rules{$rule_name}++;
my $rule_content = $pms->{'permsgstatus'}->{'conf'}->{$rule_set}->{0}->{$rule_name};
# Loop all available tags
for my $tag_name (keys %{$self->{'tags_to_replace'}}) {
# Check the rule for replacements and replace them
if ($rule_content =~ m|$start_tag$tag_name$end_tag|) {
my $replacement = get_replacement_value($self,$tag_name);
if ($replacement) {
dbg("ReplaceTags: modifying rule $rule_name, replacing $start_tag$tag_name$end_tag with $replacement");
$rule_content =~ s|$start_tag$tag_name$end_tag|$replacement|g;
$pms->{'permsgstatus'}->{'conf'}->{$rule_set}->{0}->{$rule_name} = $rule_content;
}
else {
dbg("ReplaceTags: No replacement found for rule $rule_name, ($start_tag$tag_name$end_tag)");
}
}
} # for my $tag_name (keys %{$self->{'tags_to_replace'}})
} # for my $rule_name (keys %{$pms->{'permsgstatus'}->{'conf'}->{$rule_set}->{0}})
} # for my $test (qw|body_tests rawbody_tests head_tests full_tests uri_tests|)
# See if we have rules, that we should parse and haven't been found
for my $rule (keys %Rules) {
dbg("ReplaceTags: Rule $rule not found.") if ($Rules{$rule} == 1);
}
}
sub get_replacement_value {
# Substitute replacement with the correct pattern
my ($self,$tag_name) = @_;
my $replacement = $self->{'tags_to_replace'}->{$tag_name};
if ($replacement) {
dbg("ReplaceTags: Replacement found $replacement");
}
else {
dbg("ReplaceTags: No Replacement for $tag_name");
}
return ($replacement);
}
sub parse_config {
# Used configuration pragmas are
# class
# replace_rules
# replace_start_sign
# replace_end_sign
my ($self, $opts) = @_;
switch ($opts->{'key'}) {
case 'class'
{
if ($opts->{'value'} =~ m|^(\S+)\s+(.*?)\s*$|) {
my $tag_name = $1;
my $tag_replacement = $2;
dbg("ReplaceTags: parse_config got $tag_name -> $tag_replacement");
$self->{'tags_to_replace'}->{$tag_name} = $tag_replacement;
#$opts->{'conf'}->{'tags_to_replace'}->{$TagName} = $TagReplacement;
}
}
case 'replace_rules'
{
# The replace_rules configuration-pragma contains a commasepareted
# list of all rules, which should get parsed by this module
$opts->{'value'} =~ s|\s*||g;
@{$self->{'rules_to_replace'}} = split(/\,/,$opts->{'value'});
}
case 'replace_start_sign'
{
# The replace_start_sign indicates the start of a replacement-tag. If this
# setting is omitted a < is used as default.
if ($opts->{'value'}) {
dbg("ReplaceTags: setting start sign to '".$opts->{'value'}."'");
$self->{'replace_start_sign'} = $opts->{'value'};
}
else {
dbg("ReplaceTags: no start sign specified. Fall back to default '<'");
$self->{'replace_start_sign'} = '<';
}
}
case 'replace_end_sign'
{
# The replace_end_sign indicates the end of a replacement-tag. If this
# setting is omitted a > is used as default.
if ($opts->{'value'}) {
dbg("ReplaceTags: setting end sign to '".$opts->{'value'}."'");
$self->{'replace_end_sign'} = $opts->{'value'};
}
else {
dbg("ReplaceTags: no end sign specified. Fall back to default '>'");
$self->{'replace_end_sign'} = '>';
}
}
}
}
sub dbg { Mail::SpamAssassin::dbg (@_); }
1;
How To Use It
See the pod or the example above.
Example Classes
class A [gra\@\xc0\xc1\xc2\xc3\xc4\xc5\xc6\xe4\xe3\xe2\xe0\xe1\xe2\xe3\xe4\xe5\xe60o]
class B [b8]
class C [kc\xc7\xe7@]
class D [d\xd0]
class E [e3\xc8\xc9\xca\xcb\xe8\xe9\xea\xeb\xa4]
class F [f]
class G [gk]
class H [h]
class I [il\|\!1y\?\xcc\xcd\xce\xcf\xec\xed\xee\xef]
class J [j]
class K [k]
class L [il\|\!1\xa3]
class M [m]
class N [n\xd1\xf1]
class O [go0\xd2\xd3\xd4\xd5\xd6\xd8\xf0\xf2\xf3\xf4\xf5\xf6\xf8]
class P [p\xfek]
class Q [q]
class R [r]
class S [s\xa6\xa7]
class T [t]
class U [uv\xb5\xd9\xda\xdb\xdc\xfc\xfb\xfa\xf9\xfd]
class V [v\\\/]
class W [wv]
class X [x\xd7]
class Y [y\xff\xfd\xa5j]
class Z [zs]
class PIC (jpe*g|gif|png)
class SP [\s\d\_\-\*\$\%\(\)\,\.\:\;\?\!\}\{\[\]\|\/\?\^\#\~\xa1\<B4>\`\'\+]
class CUR [\$\xa5\xa3\xa4\xa2]