X-Git-Url: http://dxcluster.net/gitweb/gitweb.cgi?a=blobdiff_plain;f=perl%2FDXDupe.pm;h=ebb2aac41a7c157f5be5986c9384cce2143a399a;hb=86d0e8cc034db0b0d23afc09b6f596c38a57885d;hp=cd4272c787530d24473dc821fcd7b724596762c9;hpb=334765d047bc6f15f8c15686afc07b0242c6919f;p=spider.git diff --git a/perl/DXDupe.pm b/perl/DXDupe.pm index cd4272c7..ebb2aac4 100644 --- a/perl/DXDupe.pm +++ b/perl/DXDupe.pm @@ -17,28 +17,44 @@ use vars qw{$lasttime $dbm %d $default $fn}; $default = 48*24*60*60; $lasttime = 0; -$fn = "$main::data/dupefile"; +localdata_mv("dupefile"); +$fn = localdata("dupefile"); sub init { - $dbm = tie (%d, 'DB_File', $fn) or confess "can't open dupe file: $fn ($!)"; + unlink $fn; + $dbm = tie (%d, 'DB_File', $fn); + confess "cannot open $fn $!" unless $dbm; } sub finish { undef $dbm; untie %d; + undef %d; + unlink $fn; } sub check { - my ($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 $s = shift; @@ -49,9 +65,11 @@ sub process { # once an hour if ($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; } } @@ -78,7 +96,7 @@ 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; }