plot_discoveryyear.python

#!/usr/bin/python
import xml.etree.ElementTree as ET
import subprocess, glob, os, math, datetime

# Open pipe gnuplot. You may want to change the terminal from svg to pdf for publication quality plots.
gnuplot = subprocess.Popen(['gnuplot',"-persist"], stdin=subprocess.PIPE, stdout=subprocess.PIPE)
gnuplot.stdin.write("""
set terminal svg 
set output "plot_discoveryyear.svg" 
set xlabel "Discovery year"
set autoscale xfix 
set ylabel "Number of new planets"
set style fill solid 0.7
set key top left

set label "Data taken from the Open Exoplanet Catalogue. Last updated on """+datetime.date.today().isoformat()+"""." at screen 0.31,0.15 front font ",8"
set style data histograms 
set style histogram rowstacked


plot \
"-" using 2:xtic(1) title "Radial velocity planets" ,\
"" using 3 title "Transiting planets" ,\
"" using 4 title "Imaged planets" ,\
"" using 5 title "Planets detected by timing" ,\
"" using 6 lt 6 title "All others detection methods"
""")


def plotPlanetsDiscoveredBy(_discoverymethod=None):
	data = {}
	for filename in glob.glob("open_exoplanet_catalogue/systems/*.xml"):
		system = ET.parse(open(filename, 'r'))
		planets = system.findall(".//planet")
		for planet in planets:
			try:
				discoverymethod = planet.findtext("./discoverymethod")
				discoveryyear = int(planet.findtext("./discoveryyear"))
				if _discoverymethod==None: #plot all others
					if discoverymethod!="RV" and discoverymethod!="transit" and discoverymethod!="imaging" and discoverymethod!="timing":
						try:
							data[discoveryyear] += 1
						except:
							data[discoveryyear] = 1
				else:
					if discoverymethod==_discoverymethod:
						try:
							data[discoveryyear] += 1
						except:
							data[discoveryyear] = 1
			except:
				pass
	return data

methods = []
methods.append(plotPlanetsDiscoveredBy("RV"))
methods.append(plotPlanetsDiscoveredBy("transit"))
methods.append(plotPlanetsDiscoveredBy("imaging"))
methods.append(plotPlanetsDiscoveredBy("timing"))
methods.append(plotPlanetsDiscoveredBy())

def printToGnuplot():
	for year in range(1990,datetime.date.today().year+1):
		if year%5==0:  #only print some labels
			gnuplot.stdin.write("%d\t"%(year))
		else:
			gnuplot.stdin.write("\"\"\t")
		for method in methods:
			try:
				gnuplot.stdin.write("%d\t"% (int(method[year])))
			except:
				gnuplot.stdin.write("%d\t"% 0)
		gnuplot.stdin.write("\n")
	gnuplot.stdin.write("\ne\n")

# output needs to be repeated 5 times because we are using a pipe and not a file.
printToGnuplot()
printToGnuplot()
printToGnuplot()
printToGnuplot()
printToGnuplot()


gnuplot.stdin.close()