# <@LICENSE>
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to you under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License.  You may obtain a copy of the License at:
# 
#     http://www.apache.org/licenses/LICENSE-2.0
# 
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# </@LICENSE>

=head1 NAME

Mail::SpamAssassin::Plugin::AutoLearnThreshold - threshold-based discriminator for Bayes auto-learning

=head1 SYNOPSIS

  loadplugin     Mail::SpamAssassin::Plugin::AutoLearnThreshold

=head1 DESCRIPTION

This plugin implements the threshold-based auto-learning discriminator
for SpamAssassin's Bayes subsystem.  Auto-learning is a mechanism
whereby high-scoring mails (or low-scoring mails, for non-spam) are fed
into its learning systems without user intervention, during scanning.

Note that certain tests are ignored when determining whether a message
should be trained upon:

=over 4

=item * rules with tflags set to 'learn' (the Bayesian rules)

=item * rules with tflags set to 'userconf' (user configuration)

=item * rules with tflags set to 'noautolearn'

=back

Also note that auto-learning occurs using scores from either scoreset 0
or 1, depending on what scoreset is used during message check.  It is
likely that the message check and auto-learn scores will be different.

=cut

package Mail::SpamAssassin::Plugin::AutoLearnThreshold;

use Mail::SpamAssassin::Plugin;
use Mail::SpamAssassin::Logger;
use strict;
use warnings;
# use bytes;
use re 'taint';

our @ISA = qw(Mail::SpamAssassin::Plugin);

sub new {
  my $class = shift;
  my $mailsaobject = shift;

  $class = ref($class) || $class;
  my $self = $class->SUPER::new($mailsaobject);
  bless ($self, $class);

  $self->set_config($mailsaobject->{conf});

  return $self;
}

sub set_config {
  my($self, $conf) = @_;
  my @cmds;

=head1 USER OPTIONS

The following configuration settings are used to control auto-learning:

=over 4

=item bayes_auto_learn_threshold_nonspam n.nn   (default: 0.1)

The score threshold below which a mail has to score, to be fed into
SpamAssassin's learning systems automatically as a non-spam message.

=cut

  push (@cmds, {
    setting => 'bayes_auto_learn_threshold_nonspam',
    default => 0.1,
    type => $Mail::SpamAssassin::Conf::CONF_TYPE_NUMERIC
  });

=item bayes_auto_learn_threshold_spam n.nn      (default: 12.0)

The score threshold above which a mail has to score, to be fed into
SpamAssassin's learning systems automatically as a spam message.

Note: SpamAssassin requires at least 3 points from the header, and 3
points from the body to auto-learn as spam.  Therefore, the minimum
working value for this option is 6.

If test option C<autolearn_header> or C<autolearn_body> is set, points from
that rule are forced to count as coming from header or body accordingly. 
This can be useful for adjusting some meta rules.

If the test option C<autolearn_force> is set, the minimum value will 
remain at 6 points but there is no requirement that the points come
from body and header rules.  This option is useful for autolearning
with rules that are considered to be extremely safe indicators of 
the spaminess of a message.

=cut

  push (@cmds, {
    setting => 'bayes_auto_learn_threshold_spam',
    default => 12.0,
    type => $Mail::SpamAssassin::Conf::CONF_TYPE_NUMERIC
  });

=item bayes_auto_learn_on_error (0 | 1)        (default: 0)

With C<bayes_auto_learn_on_error> off, autolearning will be performed
even if bayes classifier already agrees with the new classification (i.e.
yielded BAYES_00 for what we are now trying to teach it as ham, or yielded
BAYES_99 for spam). This is a traditional setting, the default was chosen
to retain backward compatibility.

With C<bayes_auto_learn_on_error> turned on, autolearning will be performed
only when a bayes classifier had a different opinion from what the autolearner
is now trying to teach it (i.e. it made an error in judgement). This strategy
may or may not produce better future classifications, but usually works
very well, while also preventing unnecessary overlearning and slows down
database growth.

=cut

  push (@cmds, {
    setting => 'bayes_auto_learn_on_error',
    default => 0,
    type => $Mail::SpamAssassin::Conf::CONF_TYPE_BOOL
  });

  $conf->{parser}->register_commands(\@cmds);
}

sub autolearn_discriminator {
  my ($self, $params) = @_;

  my $scan = $params->{permsgstatus};
  my $conf = $scan->{conf};

  # Figure out min/max for autolearning.
  # Default to specified auto_learn_threshold settings
  my $min = $conf->{bayes_auto_learn_threshold_nonspam};
  my $max = $conf->{bayes_auto_learn_threshold_spam};

  # Find out what score we should consider this message to have ...
  my $score = $scan->get_autolearn_points();
  my $body_only_points = $scan->get_body_only_points();
  my $head_only_points = $scan->get_head_only_points();
  my $learned_points = $scan->get_learned_points();

  # find out if any of the tests added an autolearn_force status
  my $force_autolearn = $scan->get_autolearn_force_status();
  my $force_autolearn_names = $scan->get_autolearn_force_names();

  dbg("learn: auto-learn? ham=$min, spam=$max, ".
                "body-points=".$body_only_points.", ".
                "head-points=".$head_only_points.", ".
                "learned-points=".$learned_points);

  my $isspam;
  if ($score < $min) {
    $isspam = 0;
  } elsif ($score >= $max) {
    $isspam = 1;
  } else {
    dbg("learn: auto-learn? no: inside auto-learn thresholds, not considered ham or spam");
    return;
  }

  my $learner_said_ham_points = -1.0;
  my $learner_said_spam_points = 1.0;

  if ($isspam) {
    my $required_body_points = 3;
    my $required_head_points = 3;

    #Set a lower threshold of "just has to be spam" if autolearn_force was set on a rule
    if ($force_autolearn) {
      $required_body_points = -99;
      $required_head_points = -99;
      dbg("learn: auto-learn: autolearn_force flagged for a rule.  Removing separate body and head point threshold.  Body Only Points: $body_only_points ($required_body_points req'd) / Head Only Points: $head_only_points ($required_head_points req'd)");
      dbg("learn: auto-learn: autolearn_force flagged because of rule(s): $force_autolearn_names");
    } else {
      dbg("learn: auto-learn: autolearn_force not flagged for a rule. Body Only Points: $body_only_points ($required_body_points req'd) / Head Only Points: $head_only_points ($required_head_points req'd)");
    }

    if ($body_only_points < $required_body_points) {
      dbg("learn: auto-learn? no: scored as spam but too few body points (".
          $body_only_points." < ".$required_body_points.")");
      return;
    }
    if ($head_only_points < $required_head_points) {
      dbg("learn: auto-learn? no: scored as spam but too few head points (".
          $head_only_points." < ".$required_head_points.")");
      return;
    }
    if ($learned_points < $learner_said_ham_points) {
      dbg("learn: auto-learn? no: scored as spam but learner indicated ham (".
          $learned_points." < ".$learner_said_ham_points.")");
      return;
    }

    if (!$scan->is_spam()) {
      dbg("learn: auto-learn? no: scored as ham but autolearn wanted spam");
      return;
    }

  } else {
    if ($learned_points > $learner_said_spam_points) {
      dbg("learn: auto-learn? no: scored as ham but learner indicated spam (".
          $learned_points." > ".$learner_said_spam_points.")");
      return;
    }

    if ($scan->is_spam()) {
      dbg("learn: auto-learn? no: scored as spam but autolearn wanted ham");
      return;
    }
  }

  if ($conf->{bayes_auto_learn_on_error}) {
    # learn-on-error strategy chosen:
    # only allow learning if the autolearning classifier was unsure or
    # had a different opinion from what we are trying to make it learn
    #
    my $tests = $scan->get_tag('TESTS');
    if (defined $tests && $tests ne 'none') {
      my %t = map { ($_,1) } split(/,/, $tests);
      if ($isspam && $t{'BAYES_99'} || !$isspam && $t{'BAYES_00'}) {
        dbg("learn: auto-learn? no: learn-on-error, %s, already classified ".
            "as such",  $isspam ? 'spam' : 'ham');
        return;
      }
    }
  }

  dbg("learn: auto-learn? yes, ".($isspam?"spam ($score > $max)":"ham ($score < $min)")." autolearn_force=".($force_autolearn?"yes":"no"));
 
  #Return an array reference because call_plugins only carry's one return value 
  return [$isspam, $force_autolearn, $force_autolearn_names];
}

1;

=back

=cut