nginx log parser, reports top 10 consumers ip by traffic + gives some additional info (pages, urls hits)

analyse_nginx_log.rb

bytes_per_ip = Hash.new(0)
urls_per_ip = Hash.new(0)
pages_per_ip = Hash.new(0)
files_extensions_per_ip = Hash.new()
# i = 0

File.foreach ARGV.shift do |line|
  # Line example: [23/Nov/2012:06:26:53 +0400] 123.123.123.123 "GET /images/socializ/google-buzz.png HTTP/1.1" 200 2283 "-" www.domain.ru "http://www.domain.ru/page.html" "Opera/9.80 (J2ME/MIDP; Opera Mini/5.1.24009/28.3126; U; ru) Presto/2.8.119 Version/11.10"
  /\A\[.+?\]\s(?<ip>.+?)\s"GET (?<url>.+?)\s.+?"\s[0-9]+\s(?<bytes>[0-9]+)\s/ =~ line
  bytes_per_ip[ip] += bytes.to_i
  urls_per_ip[ip] += 1

  /(?:\.(?<url_file_extension>[a-z]{3,4}))?\z/ =~ url
  pages_per_ip[ip] +=1 if url_file_extension == 'html'
  
  if url_file_extension and url_file_extension != 'html'
    files_extensions_per_ip[ip] = Hash.new(0) unless files_extensions_per_ip.has_key?(ip)
    files_extensions_per_ip[ip][url_file_extension] += 1
  end

  # i += 1
  # break if i > 1000
end

# After sort we got array of [[ip, bytes], [ip, bytes], ...]
bytes_per_ip.sort { |a, b| a[1] <=> b[1] }.reverse.take(10).each do |bpi|
  puts "%15s => %5.2f Gb [pages hits: %i; files hits: %i, extensions: %s]" %
       [ bpi[0],
         bpi[1].to_f/(1024**3), # converting bytes to Gb
         pages_per_ip[bpi[0]],
         urls_per_ip[bpi[0]],
         files_extensions_per_ip[bpi[0]] ]
end

Usage: ruby analyse_nginx_log.rb nginx-access.log

Output example:

 68.200.110.185 => 15.08 Gb [pages hits: 11358; files hits: 166721, extensions: {"jpg"=>145385, "jpeg"=>3139, "php"=>7, "net"=>1, "com"=>3, "xml"=>42, "swf"=>13, "flv"=>1, "gif"=>35, "css"=>1, "ico"=>1, "png"=>12}]
   62.212.76.84 =>  0.59 Gb [pages hits: 0; files hits: 41975, extensions: {"jpg"=>262}]
  217.69.133.69 =>  0.52 Gb [pages hits: 1; files hits: 4800, extensions: {"jpg"=>3590, "png"=>10, "jpeg"=>14, "gif"=>7}]
   89.250.14.40 =>  0.51 Gb [pages hits: 294; files hits: 18117, extensions: {"jpg"=>10359, "gif"=>4486, "png"=>2246, "php"=>530, "jpeg"=>6, "ico"=>11, "css"=>1}]
   66.249.76.10 =>  0.34 Gb [pages hits: 1417; files hits: 6187, extensions: {"jpg"=>3650, "jpeg"=>76, "swf"=>6, "xml"=>4, "txt"=>2, "gif"=>33, "png"=>13, "php"=>1}]
   194.8.130.79 =>  0.33 Gb [pages hits: 315; files hits: 6215, extensions: {"css"=>2, "ico"=>4, "gif"=>318, "jpg"=>5117, "php"=>273, "png"=>90, "jpeg"=>80}]
    2.94.140.87 =>  0.30 Gb [pages hits: 77; files hits: 1894, extensions: {"jpg"=>536, "jpeg"=>1155, "php"=>77, "gif"=>12, "png"=>10}]
 212.178.24.174 =>  0.28 Gb [pages hits: 123; files hits: 4908, extensions: {"css"=>1, "gif"=>104, "jpg"=>4226, "jpeg"=>138, "ico"=>1, "png"=>36, "php"=>100, "cur"=>1}]
  82.151.114.56 =>  0.28 Gb [pages hits: 19; files hits: 652, extensions: {"jpg"=>319, "gif"=>35, "css"=>1, "png"=>14, "ico"=>1, "php"=>18, "jpeg"=>241}]
   83.68.37.106 =>  0.28 Gb [pages hits: 237; files hits: 4784, extensions: {"php"=>224, "jpg"=>3863, "gif"=>24, "png"=>44, "css"=>1, "jpeg"=>345}]

3.2 millions of lines is parsed in 55 seconds. Any suggestions how to improve?