geo_tagged_counts.rb

require 'json'
require 'csv'
require 'time'

file_in = '/Users/jenningsanderson/Documents/Boulder_Floods/2013ColoradoFloods.json'
file_out = "/Users/jenningsanderson/Dropbox/SocialComputing/GroupMiniProject-GeoCoded/geo_tagged_counts.csv"

def get_geo_users(in_file)
	screen_names = []
	ticker=0
	File.open( in_file ).each do |line|
		#Load the tweet object -- this is computationally expensive, do it once
		tweet = JSON.load(line)	
		if tweet['coordinates']
			unless screen_names.include? tweet['user']['screen_name']
				screen_names << tweet['user']['screen_name']
			end
		end
		if ticker %10000 == 0 #Logging
			puts "Processed #{ticker} tweets, found #{screen_names.size} users"
		end 
		ticker+=1
	end
	screen_names
end

def write_csv(geo, in_file, write_file)
	ticker = 0
	#Open the CSV
	val = []
	mobiles = ['phone','instagram','ipad','ios','android','blackberry','mobile','vine', 'foursquare']

	CSV.open(write_file, "w") do |csv|
		csv << ['Handle' , 'Total' , 'Geo' , 'Ratio' , 'ChangeTimes', 'MobileCount', 'Mobile/Geo', 'RT', 'InterestLevel']
		#Open the file
		File.open( in_file ).each do |line|
			#Load the tweet object -- this is computationally expensive, do it once
			tweet = JSON.load(line)	
			handle = tweet['user']['screen_name']
			
			#Processing
			if geo.keys.include? handle
				geo[handle][:total]+=1

				#Get the *initial* status for our dataset:
				if geo[handle][:total]<2 
					geo[handle][:changed]=0
					if tweet['coordinates']
						geo[handle][:on] = 1
					else
						geo[handle][:on] = 0
					end
				end

				retweet = not((tweet['text'].downcase =~ /^(MT|RT)/i).nil?)

				if tweet['coordinates']
					geo[handle][:geo]+=1

					if retweet:
						geo[handle][:TaggedRT]+=1
					end
					
					#If here, then we know this particular tweet is geo-tagged.
					unless geo[handle][:on] > 0 #Meaning it was already tagged,
						geo[handle][:changed]+=1
					end
					geo[handle][:on] =1
				else #Not Geo-tagged

					if retweet:
						geo[handle][:nonTaggedRT]+=1
					end

					unless geo[handle][:on] < 1
						geo[handle][:changed]+=1
					end
					geo[handle][:on] =0
				end

				source = tweet['source'].downcase


				mobiles.each do |mobile|
					if source.include? mobile
						geo[handle][:mobile]+=1
					end
				end
			end

			#Logging
			ticker+=1
			if ticker %10000 == 0 #Logging
				puts "Processed #{ticker} tweets"
			end
		end #close file
		#Writing
		geo.each { |k,v|
			csv << [k, v[:total], v[:geo], v[:geo]/v[:total], v[:changed], v[:mobile], v[:mobile] / v[:geo], v[:RT], v[:geo]/v[:RT]]
		}
	end #close the csv
end #end function

users = get_geo_users(file_in)

puts "Total number of unique screen_names: #{users.size}, now calculating geo"

#Make the users hash
geo = {}
users.each { |user_name| geo[user_name] ||= Hash.new(0.0)}

write_csv(geo, file_in, file_out)