Twitter stream stats

tweetstream.rb

require 'twitter'
require 'uri'
require 'emoji_data'

smile_emoji = "😄"

def nilstuff(a,stuff="")
	if(a.nil?)
		return stuff
	else
		return a
	end
end

# Go to https://apps.twitter.com and register yourself to get your API keys.
client = Twitter::Streaming::Client.new do |config|
	config.consumer_key        = ""
	config.consumer_secret     = ""
	config.access_token        = ""
	config.access_token_secret = ""
end

tweet_counter = 0
emoji_histo = {}
hashtag_histo ={}
uri_histo ={}
uri_tweet_counter = 0
emoji_tweet_counter = 0
start_time = Time.new  

client.sample do |object|
	if object.is_a?(Twitter::Tweet)
   		tweet_counter = tweet_counter+1
   		str = object.text
   		add_emoji = 0
   		matches = EmojiData::EMOJI_CHARS.select{ |ec| str.include? ec.char }
   		matches.each{ |matched_char|
   			add_emoji = 1
   			emoji_histo[matched_char.char] = 1 + nilstuff(emoji_histo[matched_char.char], 0)
   		}
   		emoji_tweet_counter = emoji_tweet_counter + add_emoji
   		mat = /#(\w+)/.match(str)
   		if(mat != nil)
   			mat.captures.each{ |cap|
   				hashtag_histo[cap] = 1 + nilstuff(hashtag_histo[cap],0)
   			}
   		end
 		#Also builtin to Twitter gem: object.hashtags.each{|h| }
 		add_uri = 0
 		object.uris.each{|h| 
 			add_uri = 1
 			begin
 				uri = URI(h.expanded_url).host
 				uri_histo[uri] = 1 +  nilstuff(uri_histo[uri],0)
 			rescue
				#ignore malformed URIs
			end
		}
		uri_tweet_counter = uri_tweet_counter + add_uri	
	end
	if(tweet_counter % 100 == 0)
		puts "========================================"
		puts "Total number of tweets received: #{tweet_counter.inspect} tweets"
		puts "Average tweets per hour/minute/second: " + (tweet_counter/(Time.now - start_time)).inspect + " tweets/sec" 
		puts "Top emojis in tweets: " + emoji_histo.sort_by{|k,v| v}.last(5).reverse.inspect
		puts "Percent of tweets that contains emojis: " + ((100.0 *emoji_tweet_counter)/tweet_counter).to_s + "\%"
		puts "Top hashtags: " + hashtag_histo.sort_by{|k,v| v}.last(5).reverse.inspect
		url_pct = ((100.0 *uri_tweet_counter)/tweet_counter).to_s 
		puts "Percent of tweets that contain a url: " + url_pct + "\%"
		pic_pct = ((100.0 *(nilstuff(uri_histo["instagram.com"],0) + nilstuff(uri_histo["pic.twitter.com"],0) ) )/tweet_counter).to_s
		puts "Percent of tweets that contain a photo url (pic.twitter.com or instagram): " + pic_pct + "\%" 
		puts "Top domains of urls in tweets: " + uri_histo.sort_by{|k,v| v}.last(5).reverse.inspect
	end
end