parser.rb

#!/usr/bin/env ruby

require 'http_log_parser'
require 'timeout'

parser = HttpLogParser.new

UPDATE_INTERVAL = 10 #Sec
ALERT_INTERVAL = 120
ALERT_THRESHOLD = 10 #RPS
TOPX = 5

def print_stats ( time, sections, hps, total_hits, alerts )
  system "clear"
  p '### HTTP Stats ###'
  p "Current Time: #{time}"
  p 'Top Page Hits:'
  sections.each do |x|
    p "Page: #{x[0]}: #{x[1]}"
  end
  p "Hits Per Second: #{hps}"
  p "Total Hits: #{total_hits}"
  p 'Alerts:'
  alerts.each do |alert|
    p alert
  end

  sleep 1
end

hits = 0
alert_hits = 0
total_hits = 0
sections = {}
alerts = []
t1 = Time.now
sleep 1
loop do
  begin
    Timeout::timeout(0.2) {
      parsed_data = parser.parse_line($stdin.gets)
      /(?<section>(http:\/\/\w+\.?\w*)?(\/[\w0-9]*))/ =~ parsed_data[:request]
      sections["#{section}"] ||= 0
      sections["#{section}"] += 1
      hits += 1
    }
  rescue Timeout::Error
  end

  t2 = Time.now
  if ( ( t2 - t1 ).to_i % UPDATE_INTERVAL == 0 )
    sections = sections.sort_by{|k,v| v}[0..TOPX]

    alert_hits += hits
    total_hits += hits
    hps = hits / 10

    if ( ( t2 - t1 ).to_i % ALERT_INTERVAL == 0) &&
      alert_hits / ALERT_INTERVAL >= ALERT_THRESHOLD
      alerts << "High traffic generated an alert - hits = #{alert_hps}, triggered at #{t2}"
    end
    print_stats( t2, sections, hps, total_hits, alerts )

    hits = 0
  end

end