#!/usr/bin/perl

use Algorithm::Bucketizer;
use Number::Format qw(:subs);

@patterns = (
        qr/(.*)\(*dis[ck]\s\d+\)*$/i,
        qr/(.*)cd\s*\d+$/,
        qr/^([^-]+\s\-\s).*$/,
);

$mediasize = 4.4 * 1000 * 1024 * 1024;

print "Buckets are ", format_bytes($mediasize), " each.\n";

# Create a bucketizer
my $b = Algorithm::Bucketizer->new( bucketsize => $mediasize, algorithm => "retry" );

while (<DATA>) {
    chomp; chomp;
    ( $name, $asize, $path ) = split(/\t/);
        print $path,"\n";

        if($path eq "") { $skipped_size += $asize; } else { $sizes{$path} = $asize; }
}

print "\n";
$groupsize=0; $whole_line="";

foreach $line (sort keys %sizes)
{
        $group = 0;
        $mysize = $sizes{$line};

        foreach $re (@patterns)
        {
                ($result) = ($line =~ m/$re/);
                if( ($result ne "") && ($result eq $prevresults{$re}))
                {
                        $group=1;
                }
                $prevresults{$re} = $result;
        }

        # override the pattern matching, if we're about to create an unfileable lump
        if( ($groupsize + $mysize) > $mediasize) { $group = 0; }

        if( (!$group) && ($groupsize>0))
        {
                print "COMMITING $groupcount ($groupsize): $whole_line\n";
                $b->add_item($whole_line,$groupsize) || die($!);

                $whole_line = "";
                $groupsize = 0;
                $groupcount = 0;
        }

        $whole_line .= $line."\t";
        $groupcount++;
        $groupsize += $mysize;
}
print "COMMITING $groupcount ($groupsize): $whole_line\n";
$b->add_item($whole_line,$groupsize) || die($!);

$b->optimize( algorithm => "random", maxtime => 60 );

for my $bucket ( $b->buckets() ) {

    print "\n\n# Bucket ",$bucket->serial(), " has ", format_bytes($bucket->level()), " in it.\n";
    for my $item ( $bucket->items() ) {
        $item =~ s/\t$//;
        print "    ",join("\n    ",split(/\t/,$item)),"\n";
    }
    print "\n";
}

print "\n\nWe skipped ",format_bytes($skipped_size)," of Misc.\n\n";

__DATA__
The Creatures - A bestiary of   85907653        /var/music/WholeAlbums2/The Creatures - A bestiary of
The Cure - The Head On The Door 54978322        /var/music/WholeAlbums2/The Cure - The Head On The Door
The Disposable Heroes Of Hiphoprisy - Hypocracy Is The Greatest Luxury  99314506        /var/music/WholeAlbums2/The Disposable Heroes Of Hiphoprisy - Hypocracy Is The Greatest Luxury
The Hives - Your New Favourite Band     33943504        /var/music/WholeAlbums2/The Hives - Your New Favourite Band
The Jimi Hendrix Experience - Electric Ladyland 110201733       /var/music/WholeAlbums2/The Jimi Hendrix Experience - Electric Ladyland
Prince - Emancipation - Disc 1  78945501        /var/music/WholeAlbums2/Prince - Emancipation - Disc 1
Prince - Emancipation - Disc 2  80622538        /var/music/WholeAlbums2/Prince - Emancipation - Disc 2
Prince - Emancipation - Disc 3  77673429        /var/music/WholeAlbums2/Prince - Emancipation - Disc 3
Talking Heads - Stop Making Sense       1992195 /var/music/WholeAlbums2/Talking Heads - Stop Making Sense
Talking Heads - The Name Of This Band Is Talking Heads  125134992       /var/music/WholeAlbums2/Talking Heads - The Name Of This Band Is Talking Heads
Tracy Chapman - Matters of the Heart    63415367        /var/music/WholeAlbums2/Tracy Chapman - Matters of the Heart
Turin Brakes - the Optimist LP  52981417        /var/music/WholeAlbums2/Turin Brakes - the Optimist LP
The Vines - Highly Evolved      127977590       /var/music/WholeAlbums2/The Vines - Highly Evolved
The Wannadies - Bagsy Me        47384936        /var/music/WholeAlbums2/The Wannadies - Bagsy Me