ApacheのアクセスログをJSONに変換する

access_log_to_json.rb

require "json"
require "time"
require "optparse"

# require "pry"
# require "awesome_print"

ApacheLog = Struct.new :line do
  IP_ADDR_PATTERN = %q((\d{1,3}\.){3}\d{1,3})
  FIELDS = %i(xforward_for ip remotelog remoteuser _time _request status bytes _referer _ua)

  def xforward_for
    @xforward_for ||= if line.match /\A[,\S]+,\s/
      line.match(/\A#{IP_ADDR_PATTERN},\s#{IP_ADDR_PATTERN}/)[0]
    else
      line.match(/\A#{IP_ADDR_PATTERN}/)[0]
    end
  end

  def time
    @time ||= Time.parse _time
  end

  def request
    @request ||= _request[1..-2]
  end

  def bytes
    @bytest ||= offseted_line_for(:bytes).match(/\A\S+/)[0].to_i
  end

  def to_h
    FIELDS.inject Hash.new do |hash, field|
      field = field[1..-1] if field.to_s.start_with? ?_
      hash.merge field => send(field)
    end
  end

  def to_json
    to_h.to_json
  end

  %i(ip remotelog remoteuser status).each do |field|
    define_method field do
      instance_variable_get "@#{field}".intern or instance_variable_set "@#{field}".intern, offseted_line_for(field).match(/\A\S+/)[0]
    end
  end

  %i(request referer ua).each do |field|
    alt = (?_ + field.to_s).intern

    define_method field do
      instance_variable_get "@#{field}".intern or instance_variable_set "@#{field}".intern, send(alt)[1..-2]
    end

    define_method alt do
      offseted_line_for(alt).match(/\A\"[^\"]+\"/)[0]
    end

    private alt
  end

  private

  def offseted_line_for field
    index = FIELDS.index field
    precedents = FIELDS[0..index - 1]
    offset = precedents.map {|field| send field}.join(?").size + 1
    offseted_line offset
  end

  def offseted_line offset
    line[offset..-1]
  end

  def _time
    offseted_line_for(:_time).match(/\A\[[^\]]+\]/)[0].sub ?:, ?\s
  end
end

options = OptionParser.new do |o|
  o.banner = "Usage: #{File.basename __FILE__} [options]"
  options = Hash.new

  o.on("-i", "--input FILE") {|v| options[:input] = v}
  o.on("-o", "--output FILE") {|v| options[:output] = v}

  o.parse! ARGV
  break options
end

File.open options[:input] do |logs|
  File.open options[:output], ?w do |json_logs|
    logs.each do |line|
      log = ApacheLog.new line
      puts log.to_h
      json_logs.puts ApacheLog.new(line).to_json
    end
  end
end