#!/usr/bin/python

# Minas Gjoka

import urllib,re, sys, time, random, datetime, time, signal, urllib2, cookielib, httplib, gzip, glob, os, csv
import Gnuplot, Gnuplot.funcutils

try:
	name_crawled = sys.argv[1]
	name_analytics = sys.argv[2]
except:
	print 'Usage:', sys.argv[0], 'rank_crawled rank_analytics'
	sys.exit(1)


crawled = {}
sum_crawled = 0

facebook_given = {}
sum_facebook_given = 0

f = open(name_crawled,"r")
comment_RegExp = re.compile('^#')
for line in f:
	if not comment_RegExp.search(line.strip()):
		fields=line.split()
		crawled[ int(fields[0]) ] = int(fields[1])
		sum_crawled = sum_crawled + int(fields[1])


f = open(name_analytics,"r")
comment_RegExp = re.compile('^#')
for line in f:
	if not comment_RegExp.search(line.strip()):
		fields=line.split()
		facebook_given[ int(fields[0]) ] = int(fields[1])
		sum_facebook_given = sum_facebook_given + int(fields[1])


fout = open("compare-sample-dataset.dat","w")
fout.write("#Rank Facebook Crawled RatioFacebook(%d) RatioCrawled(%d)\n" % (sum_facebook_given,sum_crawled) )
for i in range(1, min(len(crawled.keys()), len(facebook_given)) + 1 ):
	fout.write("%d %d %d %f %f\n" % (i, facebook_given[i], crawled[i],  \
				       facebook_given[i]/float(sum_facebook_given), crawled[i]/float(sum_crawled), ) ) 
		
fout.close()

g = Gnuplot.Gnuplot(debug=0)
g('set grid')
g('set pointsize 1.5')
g('set logscale x')	
g('set logscale y')	
g.xlabel('Ranked Applications\\n')
g.ylabel('Ratio of Installs over all application installs')
g('set terminal postscript monochrome enhanced "Arial" 20')
g('set output "compare-sample-dataset.eps"') 
g('plot "< cat compare-sample-dataset.dat | awk \'BEGIN{i=0; period=2} { if (NR>0) { i=i+1 } ; print  }\'"  using 1:4 with lines lt 4 lw 5  title "Facebook Statistics",\
        "< cat compare-sample-dataset.dat | awk \'BEGIN{i=0; period=2} { if (NR>0) { i=i+1 } ; print  }\'"  using 1:5 with lines lt 19  lw 5 title "Crawled Dataset"')