Log replayer that flags URLs where the X-Check-Cacheable header returned by Akamai is NO

akamai_not_cacheable_replayer.rb

=begin
    This script will replay an Apache log file and verify whether each request
    is cacheable by Akamai.

    To run the script, provide the log file and your Akamai origin host name as
    arguments, for example:
        replayer.rb /var/log/apache/access.log http://dsd-www.example.com.edgesuite-staging.net
=end

require 'net/http'

class AkamaiCacheLogParser

    USAGE = "Usage: #{__FILE__} <INPUT_FILE> <HOST>"
    PRAGMA_HEADER = 'akamai-x-cache-on, akamai-x-check-cacheable'

    def initialize path = '', host = nil
        abort("Unable to read input file. #{USAGE}") unless File.exists?(path)
        abort("No host specified. #{USAGE}") unless !host.nil?

        @requests_processed = 0
        @not_cacheable = []

        @log = File.open(path).read
        @log_lines = @log.split("\n")

        uri = URI(host)
        @host = uri.host
    end

    def run
        line_parts = []

        # Maybe a big assumption, but we'll get the host and port from
        # the first line so that we can use only one connection
        first_line = @log_lines[0]
        host_header, port = first_line.split(' ')[0].split(':')

        Net::HTTP.start(@host, port) do |http|
            @log_lines.each do |line|

                line_parts = line.split(' ')
                path = line_parts[10]

                request = Net::HTTP::Get.new path
                request['Pragma'] = PRAGMA_HEADER
                request['Host'] = host_header

                begin
                    response = http.request request
                rescue Interrupt
                    puts "Aborted."
                    Process.exit
                rescue Exception => exc
                    next
                ensure
                    @requests_processed += 1

                    if @requests_processed % 1000 == 0
                        puts "Processed #{@requests_processed} requests."
                        $stdout.flush
                    end
                end

                @not_cacheable.push(path) if response['x-check-cacheable'] == 'NO' && response.code == '200'
            end
        end
    end

    def results
        output = "==========\n"
        output += "Results:\n"
        output += "Processed #{@requests_processed} requests and found #{@not_cacheable.length} uncacheable paths.\n\n"

        # Order the non-cacheable paths by the number of times they were requested
        b = Hash.new(0)

        @not_cacheable.each do |path|
            b[path] += 1
        end

        b = b.sort_by {|key, value| value}.reverse

        b.each do |path, count|
            output += "[#{count}] #{path}\n"
        end

        return output
    end

end

parser = AkamaiCacheLogParser.new(*ARGV)

begin
    puts 'Running...'
    parser.run
ensure
    puts parser.results
end