Page History

=head1 NAME

ReplaceTags class- Create character-classes for SpamAssassin-rules

A character-class may contain of Aany character which is valid in  [\@\xc0\xc1\xc2\xc3\xc4\xc5\xc6\xe4\xe3\xe2\xe0\xe1\xe2\xe3\xe4\xe5\xe6]
  class         G       [gk]
  classa regular expression.
The grouped characters are easy to maintain and the rules are more readable.

=head1 SYNOPSIS

  loadplugin            ReplaceTags  I/path/to/plugin/ReplaceTags.pm
  replace_rules     [il\|\!1y\?\xcc\xcd\xce\xcf\xec\xed\xee\xef]
  class  VIAGRA_OBFU CIALIS_OBFU
  replace_start    R     <
  [r3]
replace_end  class         V>

  replace_class     [v\\\/wu]
    classA         SP      [\s@\dxc0\_xc1\-xc2\*xc3\$xc4\%xc5\(xc6\)xe4\,xe3\.xe2\:xe0\;xe1\?xe2\!xe3\}xe4\{xe5\[\]\|\/\?\^\#\~\xa1\Ž\`\'\+]

  body   xe6]
  replace_class         G       VIAGRA  /<V>+<SP>*<I>+<SP>*<A>+<SP>*<G>+<SP>*<R>+<SP>*<A>+/i
  describe    [gk]
  VIAGRAreplace_class  Testrule for ReplaceTags Plugin
  score  I       VIAGRA  3

This VIAGRA rule would match many different variations of "viagra" (like "v*ì*@*g*3*Ä" for instance, which
is a real-world example).

=head1 CONFIGURATION

=over 4

=item replace_tests     list_of_tests

Specify a commaspererated list of names of tests which should be parsed. Valid values are C<body_tests>,
C<rawbody_tests>, C<head_tests>, C<full_tests> and C<uri_tests>.

=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'};

  for my $rule_set (@{$self->{'rules_to_replace'}}) {

    # 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' [il\|\!1y\?\xcc\xcd\xce\xcf\xec\xed\xee\xef]
  replace_class         R               [r3]
  replace_class         V               [v\\\/wu]
  replace_class         SP              [\s\d\_\-\*\$\%\(\)\,\.\:\;\?\!\}\{\[\]\|\/\?\^\#\~\xa1\<B4>\`\'\+]

  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 list of symbolic test names (seperated with whitespaces) of tests which should be parsed. The test will only be parsed if it is a body, heade
r, uri, full or raw test.

=item replace_start sign

=item replace_end   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 replace_class classname expression

Define a character class or a valid regular expression. 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};

  my $end_tag     = $self->{replace_end};

  # 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 :( )
    while (my ($rule_name, $rule_content) = each %{$pms->{permsgstatus}->{conf}->{$rule_set}->{0}}) {

      # If rulename doesn't match, skip
      next unless $Rules{$rule_name};

      dbg("ReplaceTags: parse rule $rule_name");

      my %SkippedTags;     # Check-Container to prevent endless loops
                           # If a tag isn't found for max 4 times the while-loops stops
                           # this seems to be a _really_ bad idea... This can also occur
                           # if somebody has content inside his classes which contains the
                           # start- and end-tag unquoted, but without a valid tag-name. In that
                           # case this loop, would run for a while...

      while ($pms->{permsgstatus}->{conf}->{$rule_set}->{0}->{$rule_name} =~ /$start_tag(.+?)$end_tag/) {

        my $tag_name = $1;

        dbg("ReplaceTags: Please quote your start and end-signs in rule $rule_name") and last if $SkippedTags{$tag_name} > 5;

        dbg("ReplaceTags: TagName = $tag_name");

        # If the tag exists, replace it with the corresponding phrase
        if ($tag_name) {
          dbg("Davor => $pms->{permsgstatus}->{conf}->{$rule_set}->{0}->{$rule_name}");
          $pms->{permsgstatus}->{conf}->{$rule_set}->{0}->{$rule_name}) {
      # skip __ rules
      next if ($rule_name =~ m|^__|);

= replace_rule($self,$tag_name,$rule_content,$rule_name);
         my $rule_content dbg("Danach => $pms->{'permsgstatus'}->{'conf'}->{$rule_set}->{0}->{$rule_name}");
        }
        else {
       for my $tag_name (keys %{$self->{'tags_to_replace'}}) {

dbg("ReplaceTags: unknown tag encountered ($start_tag$tag_name$end_tag) in rule $rule_name");
         if ($rule_content =~ m|$start_tag$tag_name$end_tag|) {$SkippedTags{$tag_name}++;
        }

      } #   my $replacement = get_replacement_value($self,$tag_name);while ($pms->{permsgstatus}->{conf}->{$rule_set}->{0}->{$rule_name} =~ /$start_tag(.+?)$end_tag/)

    }  while    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)");
          }

        }

      }

    }

  }

}

sub get_replacement_value {
  my ($self,$tag_name) = @_;

  my $replacement = $self->{'tags_to_replace'}->{$tag_name};

  if ($replacement)(my ($rule_name, $rule_content) = each %{$pms->{permsgstatus}->{conf}->{$rule_set}->{0}})

  } # for my $test (qw|body_tests rawbody_tests head_tests full_tests uri_tests|)

} # sub check_start

sub replace_rule {

  my ($self,$tag_name,$rule_content,$rule_name) = @_;
  my $start_tag   = $self->{replace_start};
  my $end_tag     = $self->{replace_end};

  my $replacement = $self->{replace_classes}->{$tag_name};

  if ($replacement) {
    # Do the modifications to the rule and return it
    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;

    return $rule_content;

  }
  else {
    dbg("ReplaceTags: ReplacementNo replacement found $replacement for rule $rule_name, ($start_tag$tag_name$end_tag)");
  }

  else {
    dbg# It should never ever get here
  #dbg("ReplaceTags: NoPlease Replacementreport for $tag_name");
  }
  return ($replacementthis to somebody including the full debug-log.");

}

sub parse_config {
  # Used configuration pragmas are
  #   tagsreplace_class
  #   replace_testsrules
  #   replace_start_sign
  #   replace_end_sign

  my ($self, $opts) = @_;

  switch ($opts->{'key'}) {

    case 'replace_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'_classes}->{$TagName$tag_name} = $TagReplacement$tag_replacement;
          }
        }

    case 'replace_testsrules'
        {
          # The replace_testsrules configuration-pragma iscontains useda tocomma specify theseparated
          # testslist whichof shouldall berules, checkedwhich for replacement-tags.
          $opts->{'value'} =~ s|\s*||g;should get parsed by this module

          @{$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;