Ticket #275: fuzzy-stats_tlyons.mysql.patch

fuzzy-stats

@@ -1 @@
-local/
+
-#-----
-
+use strict;
-use MLDBM qw(DB_File Storable);
+use DBI;
-use POSIX qw(strftime);
+use Getopt::Long;
-
-my %Files = (
-etc/mail/spamassassin
-etc/mail/spamassassin
+var/lib/fuzzyocr
+var/lib/fuzzyocr
-    );
-
+my %MySQL = (
+     db   => 'FuzzyOcr'
+    ,hash => 'Hash'
+    ,safe => 'Safe'
+    ,user => 'fuzzyocr'
+    ,pass => 'fuzzyocr'
+    ,host => 'localhost'
+    ,port => 3306
+);
+
+my $cfgfile = "/etc/mail/spamassassin/FuzzyOcr.cf";
+my %App;
+my $verbose = 0;
+my %opts;
+GetOptions( \%opts,
+        'all',
+        'colon|colons',
+        'config=s'      => \$cfgfile,
+        'dbtype=i',
+        'debug',
+        'help',
+        'verbose'       => \$verbose,
+);
+
+if ( $opts{'help'} ) { usage(); };
+if ( $opts{'debug'} ) { use Data::Dumper; }
+
-my %Stats = ();
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
-    }
+}
+
+close CONFIG;
+
+# Allow user to override location of data (2 -> MLDBM, 3 -> MySQL)
+$App{'hashing_type'} = $opts{'dbtype'} || $App{'hashing_type'};
+
+sub get_ddb {
+    my %dopts = ( AutoCommit => 1 );
+    my $dsn = "dbi:mysql:database=".$MySQL{db};
+    if (defined $MySQL{socket}) {
+        $dsn .= ";mysql_socket=$MySQL{socket}";
+    } else {
+        $dsn .= ";host=$MySQL{host}";
+        $dsn .= ";port=$MySQL{port}" unless $MySQL{port} == 3306;
+    }
+    printf "Connecting to: $dsn\n" if $verbose;
+    return DBI->connect($dsn,$MySQL{user},$MySQL{pass},\%dopts);
+}
+
+sub print_summary {
+    my ($f,$time,$gctr,@Top) = @_;
+    # %Stats is a global variable
-    my $s = $Stats{$f};
-    next unless $$s{'images'};
-
@@ -54 +98 @@
-    my $p2 = sprintf "%12.2f",$$s{'score'}/$$s{'images'} if ($$s{'images'});
-    my $p3 = sprintf "%12.2f",$$s{'score2'}/$$s{'today'} if ($$s{'today'});
-
-    my @stat = stat($Files{$f});
-    printf "\n<<%s>>\n",$f;
-
-
+
+
+
+
+
+
-    printf "Oldest Hash  : %s\n",scalar(localtime($$s{'oldest'}));
-    printf "Average Score: %s\n",$p2 if ($$s{'score'});
-    printf "Images in DB : %9d\n",$$s{'images'};
@@ -70 +117 @@
-        last if (--$count == 0);
-    }
-}
+
+sub process_items {
+    my ($Stats,$f,$db,$days,%opts) = @_;
+    my @Text = ();
+
+    $Stats->{$f}{'images'}++;
+    $Stats->{$f}{'score'} += $db->{score};
+    if (int($db->{check}/86400) == $days || $opts{'all'}) {
+        $Stats->{$f}{'today'}++;
+        $Stats->{$f}{'score2'} += $db->{score};
+        my @basic = split(':',$db->{basic});
+        my $line;
+        # Output is human readable by default, but if user wants to
+        # easily delete items, print in ::: format for easy cut/paste
+        my $format = $opts{'colon'} ? "%9d:%d:%d:%d" : "%9d %dx%dx%d";
+        if ($db->{score}) {
+            $line = sprintf "%5d Time(s) -> %8.3f $format"
+                ,$db->{match}+1,$db->{score},@basic;
+        } else {
+            $line = sprintf "%5d Time(s) -> $format"
+                ,$db->{match}+1,@basic;
+        }
+        foreach my $t (qw/fname ctype/) {
+            $line .= sprintf ' %s',$db->{$t} if defined $db->{$t};
+        }
+        push @Text, $line;
+    }
+    $Stats->{$f}{'oldest'} = $db->{input}
+        unless $Stats->{$f}{'oldest'};
+    $Stats->{$f}{'oldest'} = $db->{input}
+        if ($Stats->{$f}{'oldest'} > $db->{input});
+    #print Dumper @Text if $opts{debug};
+    return @Text;
+}
+
+#Days
+my $diff = shift @ARGV || 0;
+my $gctr = shift @ARGV || 5;
+my $time = time;
+my $days = int($time/86400) - $diff;
+
+if ($App{hashing_type} == 2) {
+    foreach my $f (keys %Files) {
+        my %DB; my $err = 0;
+        tie %DB, 'MLDBM', $Files{$f} or ++$err;
+        next if $err; my @Top = ();
+        foreach my $k (keys %DB) {
+            my $db = $DB{$k};
+            # Pass global %Stats reference and modify values in func
+            push @Top, &process_items(\%Stats,$f,$db,$days,%opts);
+        }
+        print_summary($f,$time,$gctr,@Top);
+    }
+} elsif ($App{hashing_type} == 3) {
+    my $key = '%';
+    my $dbb = get_ddb();
+    if ($dbb) {
+        my $now = time;
+        foreach my $f (keys %Files) {
+            my @Top = ();
+            my $db;
+            $f =~ s/db_//;
+            my $sql = "SELECT * FROM $MySQL{$f}";
+            my $data = $dbb->selectall_arrayref($sql,undef);
+            foreach my $dbref (@{$data}) {
+                # Shove into the generic hashref to pass to process func 
+                $db = { 'basic' => $dbref->[1],
+                        'fname' => $dbref->[2],
+                        'ctype' => $dbref->[3],
+                        'match' => $dbref->[5],
+                        'input' => $dbref->[6],
+                        'check' => $dbref->[7],
+                        'score' => $dbref->[8],
+                };
+                # Pass global %Stats reference and modify values in func
+                push @Top, &process_items(\%Stats,$f,$db,$days,%opts);
+            }
+            print Dumper @Top if $opts{debug};
+            print_summary($f,$time,$gctr,@Top);
+        }
+    }
+}