parsing cloudfront logs

gistfile1.rb

require 'rubygems'
require 'zlib'
require 'aws/s3'

def parse_cloudfront_line(line)
  return if line.strip.match(/^#/)
  bytes = line.split("\t")[5]
  filename = line.split("\t")[13]
  info = {:file => filename, :bytes => bytes}
  info if filename.strip != '-'
rescue
  nil
end

def parse_cloudfront_logs
  object  = AWS::S3::Bucket.objects('logs').first

  usage_details = {}

  open(object.url) do |f|
    gz = Zlib::GzipReader.new(f)
    gz.each_line do |line|
      usage = parse_cloudfront_line(line)
      next if usage.nil?
      if usage_details[usage[:file]]
        usage_details[usage[:file]] += usage[:bytes].to_i
      else
        usage_details[usage[:file]] = usage[:bytes].to_i
      end
    end
  end

  puts usage_details.inspect
end

AWS::S3::Base.establish_connection!(
  :access_key_id => '', 
  :secret_access_key => ''
)

parse_cloudfront_logs