main.rb

require 'progressbar'

class Main
  PARTIAL_VOLUME_FILE_NAME = "dataN.txt"

  def initialize(options: {})
    @source_file_name = options[:source_file_name] || 'data_large.txt'
    @count_lines = options[:count_lines]
    @with_gc = options[:with_gc] || true
    @process_file_progress_bar = options[:with_progress_bar] ? ProgressBar.create(title: "Process file") : nil
    @collect_report_progress_bar = options[:with_progress_bar] ? ProgressBar.create(title: "Collect report") : nil
    @write_to_result_file_progress_bar = options[:with_progress_bar] ? ProgressBar.create(title: "Write to result file") : nil
  end

  def call
    GC.disable unless with_gc
    work
    GC.start
  end

  private

  attr_reader :source_file_name, :count_lines, :with_gc, :process_file_progress_bar, :collect_report_progress_bar, :write_to_result_file_progress_bar

  def work
    `head -n #{count_lines} #{source_file_name} > #{PARTIAL_VOLUME_FILE_NAME}` if count_lines

    user_id_users = {}
    user_id_stats = {}
    total_sessions = 0
    uniq_browsers = Set.new

    with_progress_bar(process_file_progress_bar) do
      File.foreach(count_lines ? PARTIAL_VOLUME_FILE_NAME : source_file_name).each do |line|
        fields = line.split(',')
        case fields[0]
        when 'user'
          user = { 'id' => fields[1], 'first_name' => fields[2], 'last_name' => fields[3], 'age' => fields[4] }
          user_id_users[user['id']] = user
        when 'session'
          session = { 'user_id' => fields[1], 'session_id' => fields[2], 'browser' => fields[3], 'time' => fields[4], 'date' => fields[5] }
          user_stat = user_id_stats[session['user_id']]
          sessions_count = user_stat&.dig('sessions_count') || 0
          total_time = user_stat&.dig('total_time') || 0
          longest_session = user_stat&.dig('longest_session') || 0
          browsers = user_stat&.dig('browsers') || []
          dates = user_stat&.dig('dates') || []
          is_used_ie = user_stat&.dig('used_ie') || false
          is_always_used_chrome = user_stat&.dig('is_used_chrome') || false

          user_id_stats[session['user_id']] = (user_id_stats[session['user_id']] || {})
          user_id_stats[session['user_id']] = {
            'sessions_count' => sessions_count + 1,
            'total_time' => total_time + session['time'].to_i,
            'longest_session' => session['time'].to_i > longest_session ? session['time'].to_i : longest_session,
            'browsers' => browsers << session['browser'].upcase,
            'used_ie' => is_used_ie ? true : !!(session['browser'].upcase =~ /INTERNET EXPLORER/),
            'always_used_chrome' => is_always_used_chrome ? !!(session['browser'].upcase =~ /CHROME/) : false,
            'dates' => dates << Date.strptime(session['date'], '%Y-%m-%d').iso8601
          }
          total_sessions += 1
          uniq_browsers.add(session['browser'].upcase)
        end
      end
    end
  
    report = {}

    with_progress_bar(collect_report_progress_bar) do
      report['totalUsers'] = user_id_users.keys.count
      report['uniqueBrowsersCount'] = uniq_browsers.count
      report['totalSessions'] = total_sessions
      report['allBrowsers'] = uniq_browsers.sort.join(',')
      report['usersStats'] = {}
      
      user_id_stats.each do |user_id, stat|
        user = user_id_users[user_id]
        user_key = "#{user['first_name']} #{user['last_name']}"
        report['usersStats'][user_key] = {
          'sessionsCount' => stat['sessions_count'],
          'totalTime' => "#{stat['total_time']} min.",
          'longestSession' => "#{stat['longest_session']} min.",
          'browsers' => stat['browsers'].sort.join(', '),
          'usedIE' => stat['used_ie'],
          'alwaysUsedChrome' => stat['always_used_chrome'],
          'dates' => stat['dates'].sort.reverse
        }
      end
    end
  
    with_progress_bar(write_to_result_file_progress_bar) { File.write('result.json', "#{report.to_json}\n") }
  end

  def with_progress_bar(bar)
    bar&.progress = 0
    yield
    bar&.progress = 100
  end
end