#!/usr/bin/perl -w

use strict;
use File::Temp qw(tempfile);

sub get_available_num
{
  my $num = 1;

  opendir DIR, ".";
  while ($_ = readdir DIR)
    {
      next if !/^[0-9]{3}$/;
      s/^0+//;
      $num = $_ + 1 if ($_ >= $num);
    }
  closedir DIR;

  return $num;
}

sub split_uri
{
  my $uri = shift;
  my ($proto, $name, $dir);

  $uri =~ s/\?.*$//;

  if ($uri =~ /^([^:]+:\/\/)(.*)$/)
    {
      $proto = $1;
      $uri = $2;
    }
  else
    {
      $proto = undef;
    }

  if ($uri =~ /^([^\/]+)\/(.*)/)
    {
      $name = $1;
      $dir = $2;

      if ($dir =~ /^(.+)\/[^\/]*$/)
        {
	  $dir = $1;
	}
    }
  else
    {
      $name = $uri;
      $dir = undef;
    }

  return ($proto, $name, $dir)
}

sub get_file_from_uri
{
  my $uri = shift;

  my $old_umask = umask 077;

  my ($fh, $filename) = tempfile;
  close $fh;
  $uri =~ s/\\/\\\\/g;
  $uri =~ s/"/\\"/g;
  die unless (system("wget -U \"Mozilla/4.0 (compatible; MSIE 7.0b; Windows NT 6.0)\" -O \"".$filename."\" \"".$uri."\"") == 0);
  open F, "<$filename";
  my @lines = <F>;
  my $lines = join " ", @lines;
  close F;
  unlink $filename;

  umask $old_umask;

  return $lines;
}

sub extract_uris
{
  my $lines = shift;

  $lines =~ s/\s+/ /g;

  my %download_uris;

  my @uris = split /a href=/i, $lines;
  foreach my $uri (@uris)
    {
      next if (length($uri) == 0);
      my $t = substr $uri, 0, 1;
      if ($t =~ /['"]/)
	{
	  $uri =~ s/^$t([^$t]+)$t.*$/$1/;
	}
      else
        {
	  $uri =~ s/^([^ >]+)[ >].*$/$1/;
	}

      next if (!($uri =~ /\.(mpg|wmv|avi)$/));
      $download_uris{$uri} = 1;
    }

  @uris = sort keys %download_uris;

  return @uris;
}

sub escape
{
  shift;
  s/\\/\\\\/g;
  s/"/\\"/g;
  return $_;
}

while (@ARGV >= 1)
  {
    my $arg = shift @ARGV;

    my $lines = &get_file_from_uri($arg);
    my @uris = &extract_uris($lines);

    if (@uris == 0)
      {
	print "Nothing to download.\n";
	exit 0;
      }

    my $dirnum = &get_available_num;
    my $fnum = 1;

    my ($proto, $hostname, $path) = &split_uri($arg);

    mkdir sprintf("%03i", $dirnum);

    foreach (@uris)
      {
	my $extension = $_;
	$extension =~ s/^.*\.([^.]+)$/$1/;
	my $fname = sprintf("%03i/%03i.%s", $dirnum, $fnum, $extension);

	if ($_ =~ /^\//)
	  {
	    $_ = $proto.$hostname.$_;
	  }
	elsif (!($_ =~ /^[^:]+:\/\//))
	  {
	    $_ = $proto.$hostname."/".$path."/".$_;
	  }
	system "wget --referer=\"".&escape($arg)."\" -U \"Mozilla/4.0 (compatible; MSIE 7.0b; Windows NT 6.0)\" -O \"".$fname."\" \"".&escape($_)."\"\n";
	$fnum++;
      }
  }