amigaremix_dump

#!/usr/bin/perl
# abstract: rip all songs from amigaremix.com - in parallel

use strict;
use vars qw($VERSION);
$VERSION = '0.001';

use WWW::Mechanize;
use Parallel::ForkManager;

my $m = WWW::Mechanize->new(
  agent_alias => 'Linux Mozilla',
);

# nasty :-)                                                                  {{{
$m->get('http://www.amigaremix.com/remixes/9999');
$m->follow_link(url_regex => qr{remixes/\d+});
my($last_page) = ($m->uri =~ m{/(\d+)$});
$last_page++;
#}}}

my @files;
for my $u(1 .. $last_page) {
  $m->get("http://amigaremix.com/remixes/$u");
  push(@files,  grep{ m/[.]mp3$/ } map{ $_->url_abs } $m->find_all_links);
}

printf("Found %d files.\n", scalar @files);
my $PROCESS_MAX = scalar @files / 2;
printf("Will spawn %d processes, is that ok? [y/N] ", $PROCESS_MAX);
chomp(my $answer = <>);

die "Aborting...\n" unless lc($answer) eq 'y';


my $pm = Parallel::ForkManager->new( $PROCESS_MAX );

printf "\e[?7t";
for my $f(@files) {
  $pm->start and next;
  $m->get($f);

  my($dest) = $f =~ m{.+/(.+[.]mp3)$};
  $dest =~ s/%20/_/g, s/_+/_/g;
  print "$f => $dest\n" if $m->save_content($dest) == 0;
  $pm->finish;
}
printf "\e[?7t";