Presumably some chap has written a module that'll do what you need, but I don't know of one. If you can do system("df"), though, why not? (After all, Perl is good at parsing this kind of stuff.)

By "the two line junk," do you mean the header line or the line wrapping/splitting when the device name is too long?

Filesystem  1k-blocks   Used Available Use% Mounted on
/dev/sdb2      614208 522868     59612  90% /
/dev/sdb3      614660 472372    110560  82% /usr
/devfs/scsi/host0/bus0/target8/lun0/part6
               711508 376520    298844  56% /build
/devfs/scsi/host0/bus0/target8/lun0/part7
               711508 574500    100864  86% /arc/1
[download]

My df (fileutils 4.1) takes a -P (a.k.a. POSIXLY_CORRECT) which forces each record to be on the same line. Uglier for humans; prettier for Perl. But in case yours doesn't:

# capacity($mountpoint)
#
# Returns the capacity (in 1024-byte units) of the
# filesystem at $mountpoint.
#
sub capacity
{
  my ($mp) = @_;
  my ($head, $line, $extra) = `df -kl $mp`;
  $line .= $extra if defined $extra;
  return (split ' ', $line)[1];
}
[download]

Before using this function, be sure to consider: security ($mp is passed to the shell), error handling, efficiency (you will have to spawn a subshell for every filesystem), and portability (not every OS has df).