# safubot - an event-driven chatbot framework for Ruby

## <a name="overview">Overview</a>
Safubot is a chatbot framework for Twitter and XMPP which aims to abstract away the idiosyncracies of the underlying APIs, allowing you to focus on writing request-processing logic. Of course, if you want to use  service-specific features (such as responding to timeline tweets) it lets you do that too!

## <a name="installation">Installation</a>
	gem install safubot

## <a name="requirements">Requirements</a>
Safubot uses [MongoDB](http://www.mongodb.org) for storage. It's [easy to install](http://www.mongodb.org/display/DOCS/Quickstart+Unix) and pretty awesome!

## <a name="documentation">Documentation</a>
[http://rdoc.info/gems/safubot/frames](http://rdoc.info/gems/safubot/frames)

## <a name="example">Sample Usage</a>
	require 'safubot'

	class NiftyBot < Safubot::Bot
	  def initialize
		# Check the MongoMapper docs (http://www.mongomapper.com/documentation/) if you want
		# to do something more sophisticated than an authless localhost connection.
		super(:database => "niftybot")

		# This creates a Safubot::Twitter::Bot instance at @twitter
		enable_twitter({
			:username => "niftybot",
			:consumer_key => CONSUMER_KEY,
			:consumer_secret => CONSUMER_SECRET,
			:oauth_token => OAUTH_TOKEN,
			:oauth_token_secret => OAUTH_TOKEN_SECRET
		})

		# Similarly, this makes a Safubot::XMPP::Bot instance at @xmpp
		enable_xmpp({
			:jid => "niftybot@jabber.org/niftyhost",
			:password => JABBER_PASSWORD
		})

		# A Request can be sourced from a:
		#   * Twitter mention
		#   * Twitter DM
		#   * XMPP chat message
		# The "respond" method will reply using the appropriate medium.
		on(:request) do |req|
		  if req.text.match /nifty/i
			respond req, "Yep, I'm a nifty bot! :3"
		  else
			raise ArgumentError, "This isn't nifty at all! :("
		  end
		end

		# Any unhandled errors encountered during request processing
		# will come through here.
		on(:request_error) do |req, e|
		  respond req, "#{e}"
		end
	  end
	end

	# This will run the Twitter/XMPP streaming processes.
	# If both are needed, the first-enabled one will be forked
	# into its own process.
	NiftyBot.new.run

## <a name="logging">Logging</a>
Safubot::Log behaves as a Logger instance and will write to stdout by default. You can also tell it to write to a file:
	Safubot::Log.path = "/some/cool/logfile"

## <a name="example">Specific Examples</a>

### Find the last processed request from a user

	on(:request) do |req|
	  req.user.requests.where(:processed => true).sort(:created_at.desc).first
	end

### Update our twitter account

	@twitter.on(:ready) do
	  @twitter.client.update("Hello there, wonderful scary world of Twitter! I am so *not* a spambot.")
	end

### Reply to a non-request timeline tweet

	@twitter.on(:timeline) do |tweet|
	  if tweet.raw['user']['screen_name'] == "unnali"
		@twitter.reply tweet, "Is HaeSeun OSS yet? :O"
	  end
	end

### Start an unprompted conversation with an XMPP user

	@xmpp.on(:ready) do
	  @xmpp.tell('^_^@jabber.org', "You'd best be maintaining safubot!")
	end

## <a name="caveats">Caveats</a>

I've yet to determine how to make multiple libraries with their own EventMachine loops play nice together, so the TweetStream and Blather clients are currently run in separate processes. Keep in mind the effect this will have on the execution context of your event handlers. MongoDB is your friend!