gistfile1.txt

ruby main.rb go

main.rb

require 'yaml'
require 'MeCab'
require 'pp'
require 'rubygems'
require 'twitter'

class MyTwitter

  def initialize
    config = "setting.yml"
    begin
      yaml = YAML.load_file(config)
    rescue Errno::ENOENT
      puts "#{config} is not found"
      exit
    end 
    @database = yaml['database'] || self.alert('database')
    @last_file = yaml['last'] || self.alert('last')
    @username = yaml['twitter']['username'] || self.alert('twitter username')
    @password = yaml['twitter']['password'] || self.alert('twitter password')
    @exclude_names = yaml['exclude_names'] || Array.new
    @hashtag = yaml['hashtag'] || self.alert('hashtag')
    @follow_reply = yaml['follow_reply'] || false
    @timeline_follow = yaml['timeline_follow'] || false
  end

  def stack_ids
    @stack_ids = @stack_ids || Array.new
  end

  def twitter
    @twitter = @twitter || Twitter::Base.new(Twitter::HTTPAuth.new(@username, @password))
  end

  def friend_ids
    @friend_ids = @friend_ids || self.twitter.friend_ids
  end

  def diff_follow
    diff_ids = self.friend_ids - self.twitter.follower_ids
    diff_ids.each do |id|
      self.stack_id(id)
    end
  end

  def stack_id(id)
     self.stack_ids.push(id) if !self.friend_ids.index(id) && !self.stack_ids.index(id)
  end

  def stack_follow
    self.stack_ids.each do |id|
      self.twitter.friendship_create(id, true)
      puts "follow: #{id}"
      sleep(1)
    end
  end

  def run str
    self.diff_follow if @follow_reply
    data = self.crawl
    if str == "go"
      word = self.markov data
      puts "markov word: #{word}"
      self.post word
    end
    self.stack_follow 
  end
  
  protected
  def alert str
    puts "#{str} is not found"
    puts "exit"
  end

  def save_last str
    file = File.open(@last_file, 'w')
    file.puts str
    file.close
  end

  def get_last
    begin
      f = open(@last_file) 
      last = f.read.chomp.to_i
      f.close
    rescue
      last = nil
    end
    return last
  end

  def save_data data
    f = open(@database, 'w')
    f.puts Marshal.dump(data)
    f.close
  end

  def get_data
    return (File.file? @database) ? Marshal.restore(open(@database, 'r')) : Array.new
  end

  def post str
    if str.split(//u).size <= 140
      self.twitter.update(str)
    else
      puts "over 140" 
    end
  end

  def crawl
    page = 1
    first = true
    modify = false
    last = self.get_last
    mecab = MeCab::Tagger.new("-Owakati")
    data = self.get_data

    catch :done do
      while(true)
        results = Twitter::Search.new(@hashtag).page(page).fetch().results
        break if results.empty?
        puts "Loading #{page} page"
        results.each do |result|
          next if @exclude_names.index result.from_user
          throw :done if result.id <= last
          self.stack_id Twitter.user(result.from_user).id if @timeline_follow
          puts "new twiit: #{result.text}"
          modify = true
          save_last result.id if first == true
          first = false
          mecab.parse('HEAD' + result.text.sub(/#{@hashtag}/, '') + "EOS").split(' ').each_cons(3) do |a|
            data.push h = {'head' => a[0], 'middle' => a[1], 'end' => a[2]}
          end
        end
        page += 1
      end
    end
    

    self.save_data data if modify
    return data
  end

  def markov data
    t1 = ''
    t2 = ''

    data.sort_by{rand}.each do |a|
      if a['head'] == 'HEAD'
        next if a['middle'] =‾ /^@.*$/
        t1 = a['middle']
        t2 = a['end']
        break
      end
    end

    new_text = t1 + t2 

    while true
      _a = Array.new
      data.each do |hash|
        _a.push hash if hash['head'] == t1 && hash['middle'] == t2
      end

      break if _a.size == 0
      num = rand(_a.size)
      new_text = new_text + _a[num]['end']
      break if _a[num]['end'] == "EOS"
      t1 = _a[num]['middle']
      t2 = _a[num]['end']
    end

    word = new_text.gsub(/EOS$/, " #{@hashtag}")
    return word
  end
end

main = MyTwitter.new
main.run ARGV.shift

setting.yml

database: 'data'
last: 'last'

exclude_names: []

twitter:
  username: "username"
  password: "password"

hashtag: "#hashtag"
follow_reply: false
timeline_follow: false