Twitter の特定ユーザの 3200 発言を取得して YAML で保存する (適当)

twitter_get_all_statuses.rb

$KCODE = 'UTF-8'
require 'rubygems'
require 'json'
require 'ya2yaml'
require 'oauth/cli/twitter' # gem install oauth-cli-twitter

# usage ruby twitter_get_all_statuses.rb username path/to/file

# TODO: 今何発言とったか表示できるようにする
# TODO: 最古だけじゃなく最新も取得できるようにする
# BUG : 最後の発言を延々ととり続けてしまう

class AllStatuses
  include OAuth::CLI::Twitter

  API_URL = 'http://twitter.com/statuses/user_timeline.json'
  PIT_KEY = 'get-all-statuses'

  def access_token
    @access_token ||= get_access_token(:pit => PIT_KEY, :browser => true)
  end

  def url(max_id)
    url = API_URL + "?screen_name=#{@user}"
    url += "&max_id=#{max_id}" if max_id
    puts url
    url
  end

  def statuses(max_id)
    statuses = JSON.parse(access_token.get(url(max_id)).body)
    statuses.map {|s| s.delete('user'); s }
  end

  def last_id(path)
    return nil if !File.exist?(path)
    statuses = YAML.load(File.read(path))
    statuses.last['id']
  end

  def start(user, path)
    @user, @path = user, path
    max_id = last_id(path)
    file = File.open(path, 'a')
    loop do
      statuses = statuses(max_id)
      statuses.each {|s| puts s['text'] }
      file.write(statuses.to_yaml.sub(/---\s\n/, ''))
      max_id = (statuses.last['id'] + 1).to_s
      sleep 3
    end
  end
end

AllStatuses.new.start(ARGV[0], ARGV[1])