gig2.rb (Nokogiri+retry-handler+open-uri)

gig.rb

#!/usr/bin/env ruby
# vim:encoding=UTF-8:
=begin
# gig.rb

Github IRC Gateway

=end
require "logger"
require "ostruct"
require 'time'
require 'open-uri'
require 'net/irc/mala'
require 'nokogiri'
require 'retry-handler'

class Net::IRC::Github < Net::IRC::Server::Session
  EVENTS = {
    'DownloadEvent' => '6',
    'GistEvent'     => '10',
    'WatchEvent'    => '15',
    'FollowEvent'   => '15',
    'CreateEvent'   => '13',
    'ForkEvent'     => '3',
    'PushEvent'     => '14'
  }

  def server_name
    "github"
  end

  def server_version
    "0.0.0"
  end

  def main_channel
    @opts.main_channel || "#github"
  end

  def initialize(*args)
    super

    @last_retrieved = Time.now
    @wait_time = 30
    @retry_wait_time = 10
    @retry_max = 3
  end

  def on_disconnected
    @retrieve_thread.kill rescue nil
  end

  def on_user(m)
    super

    parse_real_name_field
    post @nick, JOIN, main_channel

    @retrieve_thread = Thread.start do
      loop do
        retrieve_from_github
      end
    end
  end

  private
  def parse_real_name_field
    @real, *@opts = real.split(/\s+/)
    @opts = OpenStruct.new @opts.inject({}) {|r, i|
      key, value = i.split("=", 2)
      value2 = case value
               when nil                      then true
               when /\A\d+\z/                then value.to_i
               when /\A(?:\d+\.\d*|\.\d+)\z/ then value.to_f
               else                               value
               end
      r.update key => value2
    }
  end

  def parse_github_rss(feed_data)
    ns = {"a" => "http://www.w3.org/2005/Atom"}
    Nokogiri::XML(feed_data).xpath("//a:feed/a:entry", ns).map do |entry|
      {
        :datetime => Time.parse(entry.xpath('string(a:published)', ns)),
        :id       => entry.xpath('string(a:id)', ns),
        :title    => entry.xpath('string(a:title)', ns),
        :author   => entry.xpath('string(a:author/a:name)', ns),
        :link     => entry.xpath('string(a:link/@href)', ns),
      }
    end
  end

  def retrieve_from_github
    begin
      @log.info 'retrieveing feed...'
      uri = "https://github.com/#{@real}.private.atom"

      feed_data = open(uri.to_s, :http_basic_authentication => [@real, @pass]).read
      entries = parse_github_rss(feed_data)

      entries.reverse_each do |entry|
        next if entry[:datetime] <= @last_retrieved
        type = entry[:id][%r|tag:github.com,2008:(.+?)/\d+|, 1]
        post entry[:author], PRIVMSG, main_channel,
              "\003#{EVENTS[type] || '5'}#{entry[:title]}\017 \00314#{entry[:link]}\017"
      end

      @last_retrieved = entries.first[:datetime]
      GC.start
      @log.info "sleep: #{@wait_time} seconds"

      sleep @wait_time
    end.retry(:max => @retry_max, :wait => @retry_wait_time, :accept_exception => Exception, :logger => @log)
  end
end

if __FILE__ == $0
  Net::IRC::CLI.run(Net::IRC::Github, ARGV)
end