be.gs irc bot

begs.rb

#!/usr/bin/env ruby

require 'rubygems'
require 'isaac'
require 'em-http-request'
require 'uri'

configure do |c|
  c.nick     = 'begs'
  c.server   = 'irc'
  c.realname = 'be.gs URL shortener (http://be.gs)'
  c.port     = 6667
  c.verbose  = true
end

# list of channels we should join upon connecting
CHANNELS = [
  { :channel => '#channel' },
  { :channel => '#anotherchannel', :key => 'sup3rs3kr1t' }
]

# options...
# * SHORTENER         url shortener we should use
# * LENGTH_THRESHOLD  minimum # of characters we should trigger shortening for
# * HTTP_OPTIONS      options for EM::HttpRequest
#
SHORTENER = URI.parse('http://be.gs/shorten')
LENGTH_THRESHOLD = 22 # characters
HTTP_OPTIONS = {
  :connect_timeout => 3,
  :inactivity_timeout => 5
}

helpers do
  # shorten a url for a requester...
  def shorten(target, url)
    # we don't want to clutter the channel with unnecessarily shortened links
    return nil if url.length < LENGTH_THRESHOLD
    
    # make sure this is a valid url, and not a loop back to us
    uri = URI.parse url rescue nil
    return nil if !uri or uri.host == SHORTENER.host

    # create our response hash
    response = {
      :url => nil,
      :title => nil
    }
    
    # spin up parallel requests for shortening and title fetching
    multi = EventMachine::MultiRequest.new
    multi.add(:shorten, EventMachine::HttpRequest.new(SHORTENER, HTTP_OPTIONS).post(:body => { :url => URI.encode(url) }))
    multi.add(:source, EventMachine::HttpRequest.new(uri.to_s, HTTP_OPTIONS).get(:redirects => 5, :head => { 'Accept-Encoding' => 'gzip, deflate', 'Range' => 'bytes=0-4095' }))

    # sort through our responses
    multi.callback do
      # if shortener returned 200, take text of response 
      if multi.responses[:callback][:shorten].response_header.status == 200
        response[:url] = multi.responses[:callback][:shorten].response
      end
      
      # if source returned 200 (aka not-redirect-hell), and contains a <title> tag in the first 512 bytes
      if multi.responses[:callback][:source] and multi.responses[:callback][:source].response_header.status == 200
        if multi.responses[:callback][:source].response[0..4095] =~ /<title>(.*)<\/title>/
          # we only want the first 64 characters of the title...
          response[:title] = $~[1].strip[0..63]
          response[:title] += '...' if $~[1].length > 64
        end
      end
      
      # send shortened url and title to requester
      display target, response
    end
  end
  
  # display shortened url and/or title
  def display(target, u)
    if u and (u[:url] or u[:title])
      msg target, "#{u[:url] ? u[:url].ljust(18) : nil} #{u[:title] ? ('## ' + u[:title]) : nil }"
    end
  end
end

# after connecting to the irc server, join our channels (w/keys if specified)
on :connect do
  CHANNELS.each do |channel|
    join channel[:channel] + " #{channel[:key]}"
  end
end

# look for urls in public channels, grab only the first one per line
on :channel, /(http[s]*:\/\/\S+)/ do
  url = match[0]
  shorten channel, url
end

# get urls from private messages, grab only the first one per line
on :private, /(http[s]*:\/\/\S+)/ do
  url = match[0]
  shorten nick, url
end