url.rb

# coding: utf-8
require 'nokogiri'
require 'kconv'

class URL
  include Donkun::Plugin
  set :help, <<-EOS
URL の情報を表示します
  EOS

  match /((https?):\/\/(([^[:space:].\/]+\.)+[a-z]+)(:\d+)?(\/[^[:space:]]*)?)/i, :use_prefix => false, :use_suffix => false
  def execute(m, url)
    text = fetch url
    m.target.notice text if text
  rescue RestClient::Exception => e
    m.target.notice "#{e.message} #{OWATA}"
  end

  private

  class Page
    attr_reader :plugin, :url

    def initialize(plugin, url, request_options = {}, options = {})
      @plugin = plugin
      @url = url
      @request_options = request_options
      @timeout = options[:timeput] || 5
    end

    def headers
      unless @headers
        timeout @timeout do
          begin
            request_head
          rescue RestClient::MethodNotAllowed
            request_get
          end
        end
      end
      @headers
    end

    def body
      unless @body
        timeout @timeout do
          request_get
        end
      end
      @body
    end

    def doc
      @doc ||= Nokogiri::HTML(body)
    end

    private

    def request_head
      res = @plugin.head @url, :options => @request_options
      @headers = res.headers
    end

    def request_get
      res = @plugin.get @url, :options => @request_options
      @headers = res.headers
      @body = res.body
      @body = @body.toutf8 unless @headers[:content_type] =~ /utf-?8/i
    end
  end

  def open(url)
    Page.new self, url, request_options(url)
  end

  def fetch(url)
    page = open url
    case page.headers[:content_type]
    when /html/
      on_html page
    else
      on_binary page
    end
  end

  def on_binary(page)
    content_length = page.headers[:content_length].to_i
    human_size = ['B', 'KB', 'MB', 'GB', 'TB'].inject(content_length) do |n, r|
      n < 1024 ? (break "#{n.round(2)} #{r}") : n / 1024.0
    end
    "#{page.headers[:content_type]} (#{human_size})"
  end

  def on_html(page)
    text = case page.url
           when %r(https?://(?:\w+\.)?twitter\.com/(\w+)/status(?:es)?/(\d+))
             on_twitter $1, $2
           else
             page.doc.title
           end

    sanitize text
  end

  def on_twitter(screen_name, id)
    page = open "https://twitter.com/#{screen_name}/status/#{id}"

    stream_protected = page.doc.at_css('.stream-protected h2')
    return stream_protected.text if stream_protected

    tweet = page.doc.at_css('.permalink-tweet')
    name = tweet['data-screen-name']
    text = tweet.at_css('.tweet-text').text
    "#{name}: #{text}"
  end

  def request_options(url)
    options = config[:request_options] || {}
    options.each do |k, v|
      return v if url =~ k
    end
    {}
  end
end