You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@spamassassin.apache.org by jm...@apache.org on 2005/11/04 00:25:08 UTC
svn commit: r330663 - in /spamassassin/trunk/masses: hit-frequencies
rule-qa/automc/ruleqa.cgi rule-qa/corpus-hourly
Author: jm
Date: Thu Nov 3 15:25:02 2005
New Revision: 330663
URL: http://svn.apache.org/viewcvs?rev=330663&view=rev
Log:
added hit-frequencies -S (score-map) and -d (XML output) switches; hook up -S support in the ruleqa.cgi view
Modified:
spamassassin/trunk/masses/hit-frequencies
spamassassin/trunk/masses/rule-qa/automc/ruleqa.cgi
spamassassin/trunk/masses/rule-qa/corpus-hourly
Modified: spamassassin/trunk/masses/hit-frequencies
URL: http://svn.apache.org/viewcvs/spamassassin/trunk/masses/hit-frequencies?rev=330663&r1=330662&r2=330663&view=diff
==============================================================================
--- spamassassin/trunk/masses/hit-frequencies (original)
+++ spamassassin/trunk/masses/hit-frequencies Thu Nov 3 15:25:02 2005
@@ -19,16 +19,17 @@
use strict;
use FindBin;
use Getopt::Std;
-getopts("fm:M:X:l:L:pxhc:at:s:ioT");
+getopts("fm:M:X:l:L:pxhc:at:s:ioTSd");
use vars qw {
$opt_f $opt_m $opt_M $opt_X $opt_p $opt_x $opt_h $opt_l $opt_L $opt_c
- $opt_a $opt_t $opt_s $opt_i $sorting $opt_o $opt_T
+ $opt_a $opt_t $opt_s $opt_i $sorting $opt_o $opt_T $opt_S $opt_X
+ $opt_d
};
sub usage {
die "hit-frequencies [-c rules dir] [-f] [-m RE] [-M RE] [-X RE] [-l LC]
- [-s SC] [-a] [-p] [-x] [-i] [-T] [-o] [spam log] [ham log]
+ [-s SC] [-a] [-p] [-x] [-i] [-T] [-S] [-o] [-d] [spam log] [ham log]
-c p use p as the rules directory
-f falses. count only false-negative or false-positive matches
@@ -45,6 +46,8 @@
-i use IG (information gain) for ranking
-T display rule times. implies -x, -p
-o display hit overlaps against all other rules
+ -S display score-map of hits
+ -d XML output. conflicts with -x, -p
options -l and -L are mutually exclusive.
@@ -57,11 +60,16 @@
}
usage() if($opt_h || ($opt_l && $opt_L));
+usage() if($opt_d && ($opt_x || $opt_p));
if ($opt_p) {
$opt_x = 1;
}
+if ($opt_d) {
+ $opt_x = $opt_p = 1;
+}
+
$opt_s = 0 if ( !defined $opt_s );
my $cffile = $opt_c || "$FindBin::Bin/../rules";
@@ -75,6 +83,8 @@
my %freq_ham = ();
my %hmap_spam = ();
my %hmap_ham = ();
+my %scoremap_spam = ();
+my %scoremap_ham = ();
my %freq = ();
my $num_spam = 0;
my $num_ham = 0;
@@ -113,7 +123,25 @@
my $sorting = $opt_i ? "IG" : "RANK";
-if ($opt_p) {
+if ($opt_d) {
+ $hdr_all ||= 0.00001; # avoid div by 0 in the next 2 statements
+ $hdr_spam = ($num_spam / $hdr_all) * 100.0;
+ $hdr_ham = ($num_ham / $hdr_all) * 100.0;
+
+ print qq{
+
+ <freqs scoreset='$opt_s'>
+ <allmessages>
+ <count class='spam'>$num_spam</count>
+ <count class='ham'>$num_spam</count>
+ <pc class='spam'>$hdr_spam</pc>
+ <pc class='ham'>$hdr_spam</pc>
+ </allmessages>
+
+ };
+
+}
+elsif ($opt_p) {
printf "%7s %7s %7s %6s %6s %6s %s\n",
"MSECS", $opt_f?"FNEG%":"SPAM%", $opt_f?"FPO%":"HAM%",
"S/O", $sorting, "SCORE", "NAME";
@@ -322,6 +350,9 @@
my $fs = $freq_spam{$test}; $fs ||= 0;
my $fn = $freq_ham{$test}; $fn ||= 0;
my $fa = $fs+$fn;
+ my $num_fs = $fs;
+ my $num_fn = $fn;
+ my $num_fa = $fa;
# match certain tests
next if ($opt_m && $test !~ m/$opt_m/);
@@ -357,7 +388,25 @@
$soratio{$test} = soratio ($fsadj, $fnadj);
}
- if ($opt_T) {
+ if ($opt_d) {
+ my $tflags = $rules{$test}->{tflags} || ''; # good to know
+ print qq{
+ <rule>
+ <time>}.($rule_times{$test}||0).qq{</time>
+ <count class='all'>$num_fa</count>
+ <count class='spam'>$num_fs</count>
+ <count class='ham'>$num_fn</count>
+ <pc class='all'>$fa</pc>
+ <pc class='spam'>$fs</pc>
+ <pc class='ham'>$fn</pc>
+ <so>$soratio</so>
+ <rank>$ranking{$test}</rank>
+ <score>}.($scores{$test}||0).qq{</score>
+ <test>$test</test>
+ <tflags>$tflags</tflags>
+ };
+
+ } elsif ($opt_T) {
printf "%7.5f %7.4f %7.4f %7.3f %6.2f %6.2f %s\n",
$rule_times{$test}||0, $fs, $fn, $soratio, $ranking{$test},
$scores{$test}||0,
@@ -375,13 +424,66 @@
printf "%10d %10d %10d %s\n", $fa, $fs, $fn, $test;
}
+ if ($opt_S) {
+ _print_scoremap("ham", $scoremap_ham{$test});
+ _print_scoremap("spam", $scoremap_spam{$test});
+ }
+
if ($opt_o) {
compute_overlaps_for_rule($test);
}
+
+ if ($opt_d) {
+ print qq{ </rule> };
+ }
}
exit;
+sub _print_scoremap {
+ my ($name, $smap) = @_;
+
+ if ($opt_d) {
+ print qq{ <scoremap class='$name'> };
+ }
+
+ $smap ||= { };
+ my @scores = (sort { $a <=> $b } keys %{$smap});
+
+ my $total = 0;
+ foreach my $score (@scores) {
+ $total += $smap->{$score};
+ }
+
+ foreach my $score (@scores) {
+ my $num = $smap->{$score};
+ my $pc = sprintf("%.4f", ($num / ($total||0.0001)) * 100);
+
+ if ($opt_d) {
+ print qq{
+ <si score='$score' pc='$pc' count='$num' /> };
+
+ }
+ else {
+ printf " scoremap %4s: %2d %6.2f%% %4d %s\n",
+ $name, $score, $pc, $num, _scoremap_graph($pc);
+
+ }
+ }
+
+ if ($opt_d) {
+ print qq{ </scoremap> };
+
+ } else {
+ print "\n";
+ }
+}
+
+sub _scoremap_graph {
+ my ($pc) = @_;
+ return '*' x ($pc * (40/100));
+}
+
sub readlogs {
my $spam = $ARGV[0] || "spam.log";
@@ -393,6 +495,7 @@
my $isspam = ($file eq $spam);
my $caught;
my $rules;
+ my $score;
# this is very speed-sensitive code. remove all possible
# conditionals using an eval('..').
@@ -414,7 +517,7 @@
# note: doing the match with a regexp shaves off no less than
# 7 opcodes. nice!
$evalstr .= '
- ($caught, undef, undef, $rules) = split(\' \', $_, 5);
+ ($caught, $score, undef, $rules) = split(\' \', $_, 5);
next unless ($caught =~ /^[Y\.]$/ && $rules);
';
@@ -424,7 +527,14 @@
';
}
+ if ($opt_S) {
+ $evalstr .= '
+ $score = int $score;
+ ';
+ }
+
my $hmapstr = '';
+ my $smapstr = '';
if ($isspam) {
if ($opt_o) {
$hmapstr = '
@@ -435,11 +545,15 @@
';
}
+ if ($opt_S) {
+ $smapstr = ' $scoremap_spam{$r}{$score}++; ';
+ }
+
$evalstr .= '
$num_spam++;
foreach my $r (split(/,/, $rules)) {
$freq_spam{$r}++;
- '.$hmapstr.'
+ '.$hmapstr.$smapstr.'
}
';
} else {
@@ -452,11 +566,15 @@
';
}
+ if ($opt_S) {
+ $smapstr = ' $scoremap_ham{$r}{$score}++; ';
+ }
+
$evalstr .= '
$num_ham++;
foreach my $r (split(/,/, $rules)) {
$freq_ham{$r}++;
- '.$hmapstr.'
+ '.$hmapstr.$smapstr.'
}
';
}
@@ -516,13 +634,29 @@
sub _print_overlap_ratios {
my ($r1, $hash, $type) = @_;
+ if ($opt_d) {
+ print qq{ <overlap class='$type'> };
+ }
+
foreach my $full (sort { $b <=> $a } keys %$hash) {
my $ratio = $full * 100;
last if ($ratio < 30); # 30% cutoff
my $rules = _prettify_overlap_rules($r1, $hash->{$full});
next if ($rules eq '');
- printf " overlap %s: %3d%% %s\n", $type, $ratio, $rules;
+
+ if ($opt_d) {
+ print qq{
+ <overlaprules ratio='$ratio'>$rules</overlaprules>
+ };
+
+ } else {
+ printf " overlap %s: %3d%% %s\n", $type, $ratio, $rules;
+ }
+ }
+
+ if ($opt_d) {
+ print qq{ </overlap };
}
}
Modified: spamassassin/trunk/masses/rule-qa/automc/ruleqa.cgi
URL: http://svn.apache.org/viewcvs/spamassassin/trunk/masses/rule-qa/automc/ruleqa.cgi?rev=330663&r1=330662&r2=330663&view=diff
==============================================================================
--- spamassassin/trunk/masses/rule-qa/automc/ruleqa.cgi (original)
+++ spamassassin/trunk/masses/rule-qa/automc/ruleqa.cgi Thu Nov 3 15:25:02 2005
@@ -29,6 +29,7 @@
'NET.age' => 'set 1 (network), by message age in weeks',
'NET.all' => 'set 1 (network), by contributor',
'NET.new' => 'set 1 (network), in aggregate',
+ 'SCOREMAP.new' => 'set 0, score-map',
'OVERLAP.new' => 'set 0, overlaps between rules',
);
@@ -58,6 +59,7 @@
$s{all} = 1;
$s{new} = 1;
$s{overlap} = 1;
+ $s{scoremap} = 1;
}
if (!grep { $_ } values %s) {
@@ -451,6 +453,7 @@
# special case: we only build this for one set, as it's quite slow
# to generate
+ $s{scoremap} and showfreqsubset("SCOREMAP.new", $strdate);
$s{overlap} and showfreqsubset("OVERLAP.new", $strdate);
}
@@ -543,6 +546,9 @@
elsif (/MSEC/) {
next; # just ignored for now
}
+ elsif (/\s+scoremap (.*)$/) {
+ $freqs_data{$key}{$lastrule}{scoremap} .= $_;
+ }
elsif (/\s+overlap (.*)$/) {
$freqs_data{$key}{$lastrule}{overlap} .= $_;
}
@@ -753,6 +759,16 @@
}, \$out) or die $ttk->error();
$line_counter++;
+ }
+
+ # add scoremap using the EXTRA_TEMPLATE if it's present
+ if ($obj->{scoremap}) {
+ my $ovl = $obj->{scoremap} || '';
+ # scoremap spam: 16 12.11% 777 ****
+
+ $ttk->process(\$EXTRA_TEMPLATE, {
+ EXTRA => $ovl,
+ }, \$out) or die $ttk->error();
}
# add overlap using the EXTRA_TEMPLATE if it's present
Modified: spamassassin/trunk/masses/rule-qa/corpus-hourly
URL: http://svn.apache.org/viewcvs/spamassassin/trunk/masses/rule-qa/corpus-hourly?rev=330663&r1=330662&r2=330663&view=diff
==============================================================================
--- spamassassin/trunk/masses/rule-qa/corpus-hourly (original)
+++ spamassassin/trunk/masses/rule-qa/corpus-hourly Thu Nov 3 15:25:02 2005
@@ -358,6 +358,7 @@
$flags = "-t net -s 1" if $class eq "NET";
$flags = "-M HTML_MESSAGE" if $class eq "HTML";
$flags = "-o" if $class eq "OVERLAP";
+ $flags = "-S" if $class eq "SCOREMAP";
if ($opt{rules_dir}) {
$flags .= " -c '$opt{rules_dir}'";
}