#! /usr/local/bin/perl

#################################################################
### Cleanfeed tester, v0.1
###   Test your cleanfeed modifcations against a known corpus of
###     spam / ham.
###
### Put this module in the cleanfeed dir, create the corpus
### directory structure in the same dir :
###   corpus/
###   corpus/good/
###   corpus/bad/
###
### Populate with files, and run.
###
### This software is free to redistribute and modify, and comes
### without any guaarntee whatsoever.  Use at your own risk.
### You may eventually contact me at fred@lacave.net.
###                                                        -- Fred

# With this set, you can put exclusions in your cleanfeed.local
# based on this, for instance if you don't want to pollute your hash
# databases with the tests.
$MODE = 'test';

# If you use a server which sends all headers, turn this to off.
$restricthdr = 1;

# Turn to off if you wish to show only the final report.
$verbose = 1;

# This is taken directly from innd.c.
%khdr = ( "Approved" => 1, "Control" => 1, "Date" => 1, "Distribution" => 1, "Expires" => 1,
          "From" => 1, "Lines" => 1, "Message-ID" => 1, "Newsgroups" => 1, "Path" => 1,
          "Reply-To" => 1, "Sender" => 1, "Subject" => 1, "Supersedes" => 1, "Bytes" => 1,
          "Also-Control" => 1, "References" => 1, "Xref" => 1, "Keywords" => 1,
          "X-Trace" => 1, "Date-Received" => 1, "Posted" => 1, "Posting-Version" => 1,
          "Received" => 1, "Relay-Version" => 1, "NNTP-Posting-Host" => 1,
          "Followup-To" => 1, "Organization" => 1, "Content-Type" => 1, "Content-Base" => 1,
          "Content-Disposition" => 1, "X-Newsreader" => 1, "X-Mailer" => 1,
          "X-Newsposter" => 1, "X-Cancelled-By" => 1, "X-Canceled-By" => 1,
          "Cancel-Key" => 1, "User-Agent" => 1, "X-Original-Message-ID" => 1 );

# And rock'n'roll.  No user servicable parts below.
require "cleanfeed.local";
require "filter_innd.pl";

@r = (0, 0, 0,
      0, 0, 0);

foreach my $d (('corpus/good','corpus/bad')) {
  foreach my $f (<$d/*>) {
    %hdr = {};
    cft_parse_art("$f");
    my $r = filter_art();
    print "$f : " if($verbose);
    $r[2]++ if($d =~ /good$/);
    $r[5]++ if($d =~ /bad$/);
    if($r) {
      print $r."\n" if($verbose);
      $r[1]++ if($d =~ /good$/);
      $r[3]++ if($d =~ /bad$/);
    } else {
      print $r."accepted\n" if($verbose);
      $r[0]++ if($d =~ /good$/);
      $r[4]++ if($d =~ /bad$/);
    }
  }
}
print "\n" if($verbose);
print "Result :\n";
printf("\t".'Negatives : %4d ok, %4d bad,  %3d%%'."\n",$r[0], $r[1],
  ($r[0] * 100 / $r[2])) if($r[2]);
printf("\t".'Positives : %4d ok, %4d bad,  %3d%%'."\n",$r[3], $r[4],
  ($r[3] * 100 / $r[5])) if($r[5]);
printf("\t".'TOTAL     : %4d ok, %4d bad,  %3d%%'."\n",$r[0] + $r[3], $r[1] + $r[4],
  (($r[0] + $r[3]) * 100/ ($r[2] + $r[5]))) if($r[2] + $r[5]);

sub cft_parse_art {
  my $fn=shift;
  my @l;
  open(F, $fn) && do {
    @l = <F>;
    close(F);
  };
  $lk = '#';
  while (scalar(@l) > 0) {
    $_ = shift(@l);
    chomp;
    last if(/^$/);
    if(/^\s/) {
      $hdr{$lk} .= "\n".$_;
    } elsif(/([A-Za-z0-9-]+): (.*)$/) {
      if($restricthdr) {
        $k = ($khdr{$1} ? $1 : "#");
      } else {
        $k = $1;
      }
      $hdr{$k} = $2;
      $lk = $k;
    }
  }
  $hdr{__LINES__} = scalar(@l);
  $hdr{__BODY__} = join('', @l);
  delete($hdr{'#'});
}