apache_log_analysis らしい。何かあった。

access_log.rb

# coding: utf-8
module AccessLog
  CLF_REGEXP = /
    \A                       (?# 行頭)
    (\S+)\s                  (?# 1 address)
    (\S+)\s                  (?# 2 ident)
    (\S+)\s                  (?# 3 authuser)
    \[([^\]]+)\]\s           (?# 4 time)
    "(\S+)\s(\S+)\s(\S+)"\s  (?# 5 6 7 method url version)
    (\d+)\s                  (?# 8 status)
    (\d+|-)\s                (?# 9 bytes)
    "([^"]*)"\s              (?# 10 referer)
    "([^"]*)"                (?# 11 user-agent)
    \Z                       (?# 行末)
  /x

  Entry = Struct.new(
    :address, :ident, :user, :time,
    :method, :url, :version, :status, :byte,
    :referer, :user_agent
  )

  module_function
  def each_entry(file)
    file.each_line do |line|
      if entry = parse(line)
        yield(entry)
      end
    end
  end

  def parse(line)
    if m = CLF_REGEXP.match(line)
      return Entry.new(*m.captures)
    end
    $stderr.puts("parse failure: #{line.dump}")
    return nil
  end
end

Rakefile

# coding: utf-8
require './access_log'

entries = []
task :load do
  logfile = ENV["LOGFILE"] || "access_log"
  puts "loading #{logfile}."
  File.open(logfile) do |log|
    AccessLog.each_entry(log) do |entry|
      entries << entry
    end
  end
end

desc "Bus Timerのアクセス数を集計する"
task bus_timer: [:load] do
  bt_count = Hash.new(0)
  entries.each do |entry|
    if /"common\/bt\/"/ =~ entry.url
      bt_count[entry.url] += 1
    end
  end
  ranking = bt_count.sort_by {|url, count| -count }
  ranking.each do |url, count|
    printf("%d: %p\n", count, url)
  end
end

desc "時間帯別のアクセス数を集計する"
task time: [:load] do
  hour_count = Hash.new(0)
  entries.each do |entry|
    times = entry.time.split(/[:\/ ]/) # 時刻を分割する
    hour_count[times[3]] += 1 # 「時」のカウントを追加する
  end
  hours = hour_count.keys.sort
  hours.each do |h|
    printf("%s: %s\n", h, "#" * (hour_count[h] / 3))
  end
end

desc "URL別にアクセス数を集計する"
task url: [:load] do
  url_count = Hash.new(0)
  entries.each do |entry|
    url_count[entry.url] += 1
  end
  ranking = url_count.sort_by {|url, count| -count }
  ranking.each do |url, count|
    printf("%d: %p\n", count, url)
  end
end

desc "エラーになったアクセスを表示する"
task error: [:load] do
  entries.each do |entry|
    if /^[45]/ =~ entry.status # ステータスが4xxか5xxなら表示する
      printf("%p %p %p\n", entry.time, entry.status, entry.url)
    end
  end
end

task default: [:time, :url, :error]