projects
/
spider.git
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
truncate qrg in spot dupe check
[spider.git]
/
perl
/
DXDupe.pm
diff --git
a/perl/DXDupe.pm
b/perl/DXDupe.pm
index fb1210a2a15b7186c11980712b52ba6120d80a1d..1dd7c700ff9fdbc0fc40202e6c6a7df328a79d0a 100644
(file)
--- a/
perl/DXDupe.pm
+++ b/
perl/DXDupe.pm
@@
-27,7
+27,7
@@
$main::branch += $BRANCH;
sub init
{
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 $fn corrupted, removing...");
unless ($dbm) {
eval { untie %d };
dbg("Dupefile $fn corrupted, removing...");
@@
-46,21
+46,21
@@
sub finish
sub check
{
sub check
{
- my
($s, $t) = @_
;
+ my
$s = shift
;
return 1 if find($s);
return 1 if find($s);
- add($s,
$
t);
+ add($s,
shif
t);
return 0;
}
sub find
{
return 0;
}
sub find
{
- return
1 if exists
$d{$_[0]};
+ return $d{$_[0]};
}
sub add
{
}
sub add
{
- my
($s, $t) = @_
;
-
$t = $main::systime + $default unless $
t;
+ my
$s = shift
;
+
my $t = shift || $main::systime + $defaul
t;
$d{$s} = $t;
}
$d{$s} = $t;
}
@@
-74,9
+74,11
@@
sub process
{
# once an hour
if ($main::systime - $lasttime >= 3600) {
{
# once an hour
if ($main::systime - $lasttime >= 3600) {
+ my @del;
while (($k, $v) = each %d) {
while (($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;
}
}
$lasttime = $main::systime;
}
}
@@
-103,7
+105,7
@@
sub listdups
my @out;
for (sort { $d{$a} <=> $d{$b} } grep { m{$regex}i } keys %d) {
my ($dum, $key) = unpack "a1a*", $_;
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;
}
}
return @out;
}