in reply to Re: Re: file chunk algorithm
in thread file chunk algorithm
package Bin; use Carp; use strict; # Add an element to a bucket, returns the head of the list. sub add { my $self = shift; my $thing = shift; my $size = shift; if ($size <= ($self->{max_size} - $self->{size})) { $self->{size} += $size; push @{$self->{things}}, $thing; $self->{skipped} = 0; } elsif ($size > $self->{max_size}) { confess("'$thing' larger than a bucket ($size vs $self->{max_size} +)"); } else { $self->{next} ||= new Bin(@$self{'max_size', 'max_skips'}); if ($self->{max_skips} < ++$self->{skipped}) { return $self->{next}->add($thing, $size); } else { $self->{next} = $self->{next}->add($thing, $size); } } return $self; } sub DESTROY { my $self = shift; print "Bucket size $self->{size}: @{$self->{things}}\n"; <STDIN>; } sub new { my $self = bless {}, shift; $self->{max_size} = shift || 10_000_000; $self->{max_skips} = shift || 50; $self->{skipped} = 0; $self->{size} = 0; if (@_) { $self->add(@_); } $self; } package main; my $bin = new Bin; foreach my $cnt (9_800..10_001) { foreach (1, 10, 100) { $bin = $bin->add("$cnt\_$_", $cnt * $_); } } undef($bin);
|
|---|