#!/usr/bin/perl -w

# Usage:  sa33badrules.pl [attachment_number [ham [diff]]]
# Where attachment_number is a bugzilla attachment with a proposed scoreset
# and ham is the minimum percentage of the ham corpus to always show up (def=1)
# and diff is a minimum difference between the ham% and the spam% (def=0.05).
# Results will include anything that hits the minimum ham OR the minimum diff.

my $rules = shift;
my $min_ham = shift;
my $min_diff = shift;
if (  not (defined($rules) )  ) 	{ $rules  = 4558; }
if (  not (defined($min_ham) )  ) 	{ $min_ham  = 1; }
if (  not (defined($min_diff) )  )	{ $min_diff = 0.05; }
my (@output, @scores);

my $attach = 'https://issues.apache.org/SpamAssassin/attachment.cgi?id';
open (SCORES, "elinks -source '$attach=$rules' |");
while (<SCORES>) { push(@scores, $_); }
close SCORES;

open (RULEQA, 'elinks -dump "http://ruleqa.spamassassin.org/" |'); # formatted
while (<RULEQA>) {

  my ($spam, $ham, $ratio, $rule)
    = /\s+[\d.]+\s+([\d.]+)\s+([\d.]+)(\s+[\d.]+){3}\s+(?!T_)(\w.*)/;
  next unless ( defined($spam) && ($ham > $min_ham || $ham > $min_diff+$spam) );

  my $score = "unknown";
  foreach (@scores) {
    if (/^score +$rule +([^#]+)/) {
      $score = $1;
      last;
    }
  }
  chomp ($score);

  my $fudge = 1;
  if ($ham > 1) { $fudge = $ham; } else { $fudge = 1; }
  my $extra = " ";
  if ($spam) { $ratio = $fudge*$ham/$spam; $extra = "*" if ($ham / $spam < 1); }
  else       { $ratio = $fudge*$ham*100; }
  if ($ratio !~ /\./) { $ratio .= ".0"; }
  if ( length($ratio) < 5 ) { $ratio .= "0000"; }
  $ratio =~ s/^(.....).*/$1$extra/;

  while ( length($score) < 23 ) { $score .= " "; }
  if ($ham  < 10) { $ham .= "     " if (!$ham);   $ham  = " $ham"; }
  if ($spam < 10) { $spam .= "     " if (!$spam); $spam = " $spam"; }

  push (@output, "$ratio  $ham $spam   $score    $rule\n");

}
close RULEQA;

print "H^2/S    HAM%    SPAM%    Score in attachment $rules   Rule\n";
print sort { substr($b, 0, 4) <=> substr($a, 0, 4) } @output;