Counts n top email domain names leaked from Gawker's database

gawker.rb

#!/usr/bin/env ruby
# -*-ruby-*-

data = 'gawker.data' if !File.file?(data = ARGV[1].to_s)
ntop = 50 if (ntop = ARGV[0].to_i) <= 5
raw = 0
mail = Hash.new 0

File.open(data) {|fp|
  while line = fp.gets
    if line.match(/[^[:space:]]+@([^[:space:]]+\.[^[:space:]]+)/)
      raw += 1
      mail[$1.downcase] += 1
    end
  end
}

printf "Mail total:              %10d\n", raw
printf "Mail with unique domain: %10d\n", mail.size

puts "\nTop #{ntop} domains:\n"

mail_sa = mail.sort_by {|k, v| v}
mail_sa.last(ntop).reverse.map {|k, v|
  printf "%30s%10d\n", k, v
}