フェイトちゃんちゅっちゅ

get_uri_title.rb

# -*- coding: utf-8 -*-

require "uri"
require "open-uri"
require "nokogiri"
require "kconv"

class Fatechan::Plugin::GetURITitle
  include Cinch::Plugin
  listen_to :channel

  def listen(m)
    return if not m.command == "PRIVMSG"
    URI.extract(m.message, %w{http}) do |uri|
      open(uri) do |f|
        title = nil
        if f.content_type == "text/html" then
          content = f.read
          doc = Nokogiri::HTML(content)

          # XXX: Should use lower-case function if XPath 2.0 is available.
          html_content_type = doc.xpath(%{
            //meta[
              translate(
                @http-equiv,
                'ABCDEFGHIJKLMNOPQRSTUVWXYZ',
                'abcdefghijklmnopqrstuvwxyz'
              ) = 'content-type'
            ][1]
          })

          if html_content_type =~ /charset=([\w_-]+)/ then
            encoding = $1
          else
            encoding = Kconv.guess(content)
          end

          content.force_encoding(encoding).encode
          doc = Nokogiri::HTML(content)
          title = doc.xpath("//title[1]/text()")
          title = nil if title =~ /^\s*$/
        elsif f.content_type == "text/plain" then
          title = f.gets.chomp
          title.force_encoding(Kconv.guess(title)).encode
        end

        title ||= f.content_type || "Untitled"
        m.channel.notice "#{m.user.nick}: #{title}"
      end
    end
  end
end