X-Git-Url: http://dxcluster.net/gitweb/gitweb.cgi?a=blobdiff_plain;f=perl%2FDXDupe.pm;h=1dd7c700ff9fdbc0fc40202e6c6a7df328a79d0a;hb=f61fe3405f9cca29a7499709ed08aa66ee04be9a;hp=3dd245b2122497dd7c3986088933cd8f0ea41e41;hpb=70261345f844db52943884eb12c204c17105721b;p=spider.git diff --git a/perl/DXDupe.pm b/perl/DXDupe.pm index 3dd245b2..1dd7c700 100644 --- a/perl/DXDupe.pm +++ b/perl/DXDupe.pm @@ -19,9 +19,22 @@ $default = 48*24*60*60; $lasttime = 0; $fn = "$main::data/dupefile"; +use vars qw($VERSION $BRANCH); +$VERSION = sprintf( "%d.%03d", q$Revision$ =~ /(\d+)\.(\d+)/ ); +$BRANCH = sprintf( "%d.%03d", q$Revision$ =~ /\d+\.\d+\.(\d+)\.(\d+)/ || (0,0)); +$main::build += $VERSION; +$main::branch += $BRANCH; + sub init { - $dbm = tie (%d, 'DB_File', $fn) or confess "can't open dupe file: $fn ($!)"; + $dbm = tie (%d, 'DB_File', $fn); + unless ($dbm) { + eval { untie %d }; + dbg("Dupefile $fn corrupted, removing..."); + unlink $fn; + $dbm = tie (%d, 'DB_File', $fn) or confess "can't open dupe file: $fn ($!)"; + confess "cannot open $fn $!" unless $dbm; + } } sub finish @@ -33,13 +46,24 @@ sub finish sub check { - my ($s, $t) = @_; - return 1 if exists $d{$s}; - $t = $main::systime + $default unless $t; - $d{$s} = $t; + my $s = shift; + return 1 if find($s); + add($s, shift); return 0; } +sub find +{ + return $d{$_[0]}; +} + +sub add +{ + my $s = shift; + my $t = shift || $main::systime + $default; + $d{$s} = $t; +} + sub del { my $s = shift; @@ -50,9 +74,11 @@ sub process { # once an hour if ($main::systime - $lasttime >= 3600) { + my @del; while (($k, $v) = each %d) { - delete $d{$k} if $main::systime >= $v; + push @del, $k if $main::systime >= $v; } + delete $d{$_} for @del; $lasttime = $main::systime; } } @@ -79,7 +105,7 @@ sub listdups my @out; for (sort { $d{$a} <=> $d{$b} } grep { m{$regex}i } keys %d) { my ($dum, $key) = unpack "a1a*", $_; - push @out, "$key = " . cldatetime($d{$_} - $dupage); + push @out, "$key = " . cldatetime($d{$_} - $dupage) . " expires " . cldatetime($d{$_}); } return @out; }