Changeset 58

Timestamp:

25.11.2006 00:37:28 (2 years ago)

Author:

jorge

Message:

+ small bug fixes,
+ added statistical information regarding sub-program execution times to help evaluate bottlenecks. (requires Time::HiRes? generally available);
+ connect to mysql db only if images are found.
+ when focr_verbose is greater than 1 you can send logging to stderr if focr_log_stderr is set to 1
+ reduced debug logging by requiring focr_verbose = 3 for to log most sql statements as well as other informational messages

Files:

• trunk/devel/FuzzyOcr.cf (modified) (1 diff)
• trunk/devel/FuzzyOcr.pm (modified) (7 diffs)
• trunk/devel/FuzzyOcr/Config.pm (modified) (6 diffs)
• trunk/devel/FuzzyOcr/Hashing.pm (modified) (2 diffs)
• trunk/devel/FuzzyOcr/Misc.pm (modified) (5 diffs)
• trunk/devel/INSTALL (modified) (1 diff)

trunk/devel/FuzzyOcr.cf

@@ -16 +16 @@
-# Verbosity level (see manual) Attention: Don't set to 0, but to 0.0 for quiet operation. (Default value: 1)
-#focr_verbose 2
+#
+# When focr_verbose>1 you can log to stderr by setting this variable to 1 (Default value: 0)
+#focr_log_stderr 1
-#
-# Logfile (make sure it is writable by the plugin) (Default value: /etc/mail/spamassassin/FuzzyOcr.log)

trunk/devel/FuzzyOcr.pm

@@ -19 +19 @@
-use Fcntl ':flock';
-
-
+
+
-use FuzzyOcr::Config qw(kill_pid get_tmpdir set_tmpdir get_pms save_pms get_timeout get_ddb get_thresholds get_scansets get_config get_wordlist set_config finish_parsing_end load_global_words load_personal_words debuglog logfile);
-use FuzzyOcr::Hashing qw(check_image_hash_db check_image_hash_db add_image_hash_db calc_image_hash);
@@ -93 +94 @@
-    }
-
+    my $ddb;
-    my $imgdir;
-    my %imgfiles = ();
@@ -113 +115 @@
-            debuglog("Variable \$ENV{HOME} not defined and getpwuid failed, personal wordlist function not available...");
-        }
-    }
-    if ($conf->{focr_enable_image_hashing} == 3) {
-        $conf->{focr_ddb} = get_ddb();
-    }
-
@@ -266 +265 @@
-    }
-    debuglog("Found: $cnt images"); $cnt = 0;
+    if ($conf->{focr_enable_image_hashing} == 3) {
+        $ddb = $conf->{focr_ddb} = get_ddb();
+    }
-
-    my $haserr = open RAWERR, ">$imgdir/raw.err";
@@ -602 +604 @@
-            $scan =~ s/\$efile/$efile/g;
-            #unlink $efile if (-e $efile);
-
+#
-            printf RAWERR qq(## $scan 2>>$efile\n) if ($haserr>0);
-
@@ -707 +709 @@
-        }
-        if ( $conf->{focr_verbose} > 0 and $conf->{focr_verbose} < 3 ) {
-
+ unless ($conf->{focr_enable_image_hashing} == 3)
-        }
-        for my $set ( 0 .. 3 ) {
@@ -719 +721 @@
-    }
-    if ($conf->{focr_enable_image_hashing} == 3) {
-conf->{focr_ddb}
+ddb
-    }
-    debuglog("FuzzyOcr ending successfully...");

trunk/devel/FuzzyOcr/Config.pm

@@ -134 +134 @@
-sub get_ddb {
-    return undef unless (HAS_DBI and HAS_DBD_MYSQL);
+    use DBI;
-    my $conf = get_config();
-    my %dopts = ( AutoCommit => 1 );
@@ -197 +198 @@
-        setting => 'focr_logfile',
-        type => $Mail::SpamAssassin::Conf::CONF_TYPE_STRING
+       });
+
+    push (@cmds, {
+        setting => 'focr_log_stderr',
+        default => 0,
+        type => $Mail::SpamAssassin::Conf::CONF_TYPE_BOOL
-       });
-
@@ -406 +413 @@
-    $conf = $opts->{conf};
-
+    #add_facilities('FuzzyOcr');
-    # find external binaries
-    @paths = split(/:/, $conf->{'focr_path_bin'});
@@ -521 +529 @@
-    }
-    if ($conf->{focr_enable_image_hashing} == 3) {
-$conf->{focr_ddb} = 
+
-        if (defined $ddb) {
-            my $db   = $conf->{focr_mysql_db};
@@ -594 +602 @@
-                debuglog("Stored [$err] Hashes in $db.$tab") if $err>0;
-            }
-
-
+
-            debuglog("done updating MySQL database");
-        }
@@ -685 +692 @@
-    my @lines = split('\n',$_[0]);
-    my $limit = defined $_[1] ? $_[1] : 1;
-    foreach (@lines) { dbg ("FuzzyOcr: $_"); }
-    if ( $conf->{focr_verbose} > $limit ) {
-        logfile($_[0]) if defined $conf->{focr_logfile};
+        if ($conf->{focr_log_stderr}) {
+            foreach (@lines) { info ("FuzzyOcr: $_"); }
+        }
+    } else {
+        foreach (@lines) { dbg  ("FuzzyOcr: $_"); }
-    }
-}

trunk/devel/FuzzyOcr/Hashing.pm

@@ -121 +121 @@
-            # Expire old records...
-            $sql = qq(delete from $db.$dbfile where $dbfile.check < $then);
-
+,2
-            $ddb->do($sql);
-        }
@@ -130 +130 @@
-            debuglog("Matched [$next] time(s). Prev match: ".fmt_time($now - $when));
-            $sql = qq(update $db.$dbfile set $dbfile.match='$next',$dbfile.check='$now' where $dbfile.key='$key');
-
+,2
-            $ddb->do($sql);
-        }

trunk/devel/FuzzyOcr/Misc.pm

@@ -43 +43 @@
-    my $t = get_timeout();
-    my ($cmd, $stdout, $stderr, $return_stdout) = @_;
+    my ($pgm,undef) = split(' ',$cmd,2);
-    $stdout = '>/dev/null' unless $stdout;
-    $stderr = '>/dev/null' unless $stderr;
-    my $retcode;
-    if ($conf->{'focr_global_timeout'}) {
+        my $begin = [gettimeofday];
-        my $pid = fork();
-        if (not defined $pid) {
@@ -52 +54 @@
-            return -1;
-        } elsif (not $pid) {
+            debuglog("Exec: $cmd $stdout $stderr",2);
-            open(STDOUT, $stdout);
-            open(STDERR, $stderr);
@@ -58 +61 @@
-        } else {
-            set_pid($pid);
-
-
+
+
+
+
-            unset_pid();
-            if ($return_stdout and $stdout !~ m,/dev/null,i) {
@@ -80 +85 @@
-                return -1;
-            } elsif (not $pid) {
-
+,2
-                open(STDOUT, $stdout);
-                open(STDERR, $stderr);
@@ -90 +95 @@
-                debuglog("Elapsed: ".
-                    sprintf ("%.6f",tv_interval($begin, [gettimeofday]))
-cmd
+pgm
-                unset_pid();
-                if ($return_stdout and $stdout !~ m,/dev/null,i) {

trunk/devel/INSTALL

@@ -57 +57 @@
-focr_max_size_<img_type> <max-size>
-
-
+
+
-
-- gif