Configuration/scripts/makeCutFlows.py

#!/usr/bin/env python
import sys
import os
import fileinput
from array import *
from optparse import OptionParser
from OSUT3Analysis.Configuration.configurationOptions import *
from OSUT3Analysis.Configuration.processingUtilities import *

parser = OptionParser()
parser = set_commandline_arguments(parser)

parser.remove_option("-d")
parser.remove_option("-n")
parser.remove_option("-u")
parser.remove_option("-e")
parser.remove_option("-R")
parser.remove_option("-b")
parser.remove_option("-y")
parser.remove_option("-p")
parser.remove_option("-r")
parser.remove_option("-o")
parser.remove_option("--2D")

parser.add_option("-d", "--diff", action="store_true", dest="makeDiffPlots", default=False,
                  help="add a column for data-MC to the tables")
parser.add_option("-r", "--ratio", action="store_true", dest="makeRatioPlots", default=False,
                  help="add a colunn for (data-MC)/MC to the tables")
parser.add_option("-s", "--sToB", action="store_true", dest="signalToBackground", default=False,
                  help="add a column for S/sqrt(S+B) to the tables")
parser.add_option("-o", "--output-file", dest="outputFileName",
                  help="specify an output file base name for the cutflow table (suffix will be appended), default is 'cutFlow'")


(arguments, args) = parser.parse_args()

if arguments.localConfig:
    sys.path.append(os.getcwd())
    exec("from " + arguments.localConfig.rstrip('.py') + " import *")

condor_dir = set_condor_output_dir(arguments)

from ROOT import TFile, gROOT, gStyle, gDirectory, TKey

channels = []
processed_datasets = []

outputFileName = "cutFlow"
if arguments.outputFileName:
    outputFileName = arguments.outputFileName
outputFileName = outputFileName.partition(".")[0]  # If the input filename contains a period, take only the part before the period.  

texfile = condor_dir + "/" + outputFileName + ".tex"

replacements = {
    ">":"$>$",
    "<":"$<$",
    " eta ":" $\\eta$ ",
    " abs(eta) ":" $|\\eta|$ ",
    "#":"Num",

    "\\rightarrow":"{\\rightarrow}",
    "\\mu QCD":"\\mu$ $QCD",
    "EM QCD":"EM$ $QCD",
    "BCtoE QCD":"BCtoE$ $QCD",


    " pt ":" $p_{T}$ ",
    " ht ":"$H_{T}$ ",

    "d0":"$d_{0}$",

    " MET ":" Met ",
    " MEt ":" Met ",
    " met ":" Met ",
    " Met ":"$\\not\\!\\!{E}_{T}$ ",

    " MHT ":" Mht ",
    " MHt ":" Mht ",
    " mht ":" Mht ",
    " Mht ":"$\\not\\!\\!{H}_{T}$ ",

    "M_Z" : "$M_{Z}$",
    "M_mumu" : "$M_{\\mu\\mu}$",
    "M_ee" : "$M_{ee}$",
    "M_ll" : "$M_{ll}$",

}

secondary_replacements = {
    "$$<$":"$<"

}


#### check which input datasets have valid output files
for dataset in datasets:
    fileName = condor_dir + "/" + dataset + ".root"
    if not os.path.exists(fileName):
        #print "Couldn't find output file for",dataset,"dataset"
        continue
    testFile = TFile(fileName)
    if not (testFile.IsZombie()):
        processed_datasets.append(dataset)

if len(processed_datasets) is 0:
    sys.exit("Can't find any output root files for the given list of datasets")

#### open first input file and re-make its directory structure in the output file
testFile = TFile(condor_dir + "/" + processed_datasets[0] + ".root")
testFile.cd()
for key in testFile.GetListOfKeys():
    if (key.GetClassName() != "TDirectoryFile"):
        continue
    rootDirectory = key.GetName()
    testFile.cd(key.GetName())
    for key2 in gDirectory.GetListOfKeys():
        if (key2.GetClassName() != "TDirectoryFile"):
            continue
        channels.append(key2.GetName())

fout = open (texfile, "w")
fout.write ("\\documentclass[a2paper,8pt]{article}\n\n")
fout.write ("\\usepackage[landscape,margin=0.15cm]{geometry}\n\n")
fout.write ("\\usepackage{multirow}\n\n")
fout.write ("\\begin{document}\n\n")
fout.write ("\\pagestyle{empty}\n\n")
fout.close ()

firstChannel = True

#weight = intLumi / 10000.0
#for dataset in processed_datasets:
#    dataset_file = "%s/%s.root" % (condor_dir,dataset)
#    fin = TFile (dataset_file)
#    flags = fin.Get ("flags")
#    noWeights = flags and flags.GetBinContent (1)
#    fin.Close ()
#    if types[dataset] != "data" and not noWeights:
#        os.system("mergeTFileServiceHistograms -i %s -o %s -w %g" % (dataset_file, dataset_file + "_tmp", weight))
#    else:
#        os.system("mergeTFileServiceHistograms -i %s -o %s -w %g" % (dataset_file, dataset_file + "_tmp", 1.0))


for channel in channels: # loop over final states, which each have their own directory
    formatted_channel = channel.replace("_"," ")
    cutFlowArgs = ""
    selectionArgs = ""
    minusOneArgs = ""
    #print hist

    for dataset in processed_datasets:
            dataset_file = "%s/%s.root" % (condor_dir,dataset)
            #print dataset_file
            if arguments.makeDiffPlots or arguments.makeRatioPlots:
              if types[dataset] == "data":
                dataset_file = "\"<" + dataset_file + "\""
              elif types[dataset] == "bgMC":
                dataset_file = "\">" + dataset_file + "\""
            if arguments.signalToBackground:
              if types[dataset] == "signalMC":
                dataset_file = "\"<" + dataset_file + "\""
              elif types[dataset] == "bgMC":
                dataset_file = "\">" + dataset_file + "\""
            cutFlowArgs = cutFlowArgs + " " + dataset_file 
            selectionArgs = selectionArgs + " " + dataset_file 
            minusOneArgs = minusOneArgs + " " + dataset_file 
            cutFlowArgs = cutFlowArgs + " '" + channel + "CutFlow'"
            selectionArgs = selectionArgs + " '" + channel + "Selection'"
            minusOneArgs = minusOneArgs + " '" + channel + "MinusOne'"

            rawlabel = "$" + labels[dataset] + "$"
            label = rawlabel.replace("#","\\")
            label = "'" + label + "'"
            #print label
            cutFlowArgs = cutFlowArgs + " " + label
            selectionArgs = selectionArgs + " " + label
            minusOneArgs = minusOneArgs + " " + label

    #make cutFlowTable objects
    fout = open (texfile, "a")
    if not firstChannel:
        fout.write ("\\pagebreak\n\n")
    cutFlowTable = "cutFlowTable"
    if arguments.makeDiffPlots:
      cutFlowTable += " -d \"Data - MC\""
    if arguments.makeRatioPlots:
      cutFlowTable += " -r \"(Data - MC) / MC\""
    if arguments.signalToBackground:
      cutFlowTable += " -s \"$\\mathrm{S} / \\sqrt{\\mathrm{S} + \\mathrm{B}}$\""
    firstChannel = False
    fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
    fout.write ("\\subsection*{Cut flow}\n\n")
    fout.close ()
    #    print "Debug:  running command:  %s -l %g -m %s >> %s" % (cutFlowTable, intLumi,cutFlowArgs,texfile)
    os.system("%s -l %g -m %s >> %s" % (cutFlowTable, intLumi,cutFlowArgs,texfile))
    fout = open (texfile, "a")
    fout.write ("\\pagebreak\n\n")
    fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
    fout.write ("\\subsection*{Individual selection}\n\n")
    fout.close ()
    os.system("%s -l %g %s >> %s" % (cutFlowTable, intLumi,selectionArgs,texfile))
    fout = open (texfile, "a")
    fout.write ("\\pagebreak\n\n")
    fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
    fout.write ("\\subsection*{Minus one}\n\n")
    fout.close ()
    os.system("%s -l %g %s >> %s" % (cutFlowTable, intLumi,minusOneArgs,texfile))

#for dataset in processed_datasets:
#    dataset_file = "%s/%s.root_tmp" % (condor_dir,dataset)
#    os.remove(dataset_file)

#reformat tex files
for line in fileinput.FileInput(texfile,inplace=1):
    for replacement in replacements.keys():
        line = line.replace(replacement,replacements[replacement])
    print line,

for line in fileinput.FileInput(texfile,inplace=1):
    for replacement in secondary_replacements.keys():
        line = line.replace(replacement,secondary_replacements[replacement])
    print line,

fout = open (texfile, "a")
fout.write ("\\end{document}\n")
fout.close ()

#process tex files to make pdf files
command = "pdflatex -interaction=batchmode -output-directory=./%s %s > /dev/null" % (condor_dir,texfile)
os.system(command)
os.system(command)
#os.system("rm %s" % texfile)
os.system("rm %saux" % (texfile.rstrip("tex")))
os.system("rm %slog" % (texfile.rstrip("tex")))

print "Finished writing cutFlow to " + condor_dir + "/" + outputFileName + "{.tex,.pdf}"

Revision:	1.19
Committed:	Tue Jul 16 09:05:34 2013 UTC (11 years, 9 months ago) by wulsin
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	V02-03-02, HEAD
Changes since 1.18:	+1 -3 lines
Log Message:	Do not remove -t, -q options (they no longer exist previously)
#	User	Rev	Content
1	lantonel	1.1	#!/usr/bin/env python
2			import sys
3			import os
4			import fileinput
5			from array import *
6	lantonel	1.13	from optparse import OptionParser
7	lantonel	1.2	from OSUT3Analysis.Configuration.configurationOptions import *
8			from OSUT3Analysis.Configuration.processingUtilities import *
9	lantonel	1.1
10	lantonel	1.13	parser = OptionParser()
11	lantonel	1.1	parser = set_commandline_arguments(parser)
12	ahart	1.17
13			parser.remove_option("-d")
14	wulsin	1.18	parser.remove_option("-n")
15			parser.remove_option("-u")
16			parser.remove_option("-e")
17			parser.remove_option("-R")
18			parser.remove_option("-b")
19			parser.remove_option("-y")
20			parser.remove_option("-p")
21	ahart	1.17	parser.remove_option("-r")
22	wulsin	1.18	parser.remove_option("-o")
23			parser.remove_option("--2D")
24	ahart	1.17
25			parser.add_option("-d", "--diff", action="store_true", dest="makeDiffPlots", default=False,
26			help="add a column for data-MC to the tables")
27			parser.add_option("-r", "--ratio", action="store_true", dest="makeRatioPlots", default=False,
28			help="add a colunn for (data-MC)/MC to the tables")
29			parser.add_option("-s", "--sToB", action="store_true", dest="signalToBackground", default=False,
30			help="add a column for S/sqrt(S+B) to the tables")
31	wulsin	1.18	parser.add_option("-o", "--output-file", dest="outputFileName",
32			help="specify an output file base name for the cutflow table (suffix will be appended), default is 'cutFlow'")
33
34	ahart	1.17
35	lantonel	1.13	(arguments, args) = parser.parse_args()
36	lantonel	1.1
37	lantonel	1.12	if arguments.localConfig:
38	lantonel	1.1	sys.path.append(os.getcwd())
39	lantonel	1.12	exec("from " + arguments.localConfig.rstrip('.py') + " import *")
40	lantonel	1.1
41	lantonel	1.12	condor_dir = set_condor_output_dir(arguments)
42	lantonel	1.1
43			from ROOT import TFile, gROOT, gStyle, gDirectory, TKey
44
45			channels = []
46			processed_datasets = []
47	wulsin	1.18
48			outputFileName = "cutFlow"
49			if arguments.outputFileName:
50			outputFileName = arguments.outputFileName
51			outputFileName = outputFileName.partition(".")[0] # If the input filename contains a period, take only the part before the period.
52
53			texfile = condor_dir + "/" + outputFileName + ".tex"
54	lantonel	1.1
55			replacements = {
56			">":"$>$",
57			"<":"$<$",
58	lantonel	1.11	" eta ":" $\\eta$ ",
59			" abs(eta) ":" $\|\\eta\|$ ",
60	lantonel	1.1	"#":"Num",
61
62			"\\rightarrow":"{\\rightarrow}",
63			"\\mu QCD":"\\mu$ $QCD",
64			"EM QCD":"EM$ $QCD",
65			"BCtoE QCD":"BCtoE$ $QCD",
66	ahart	1.3
67
68	lantonel	1.11	" pt ":" $p_{T}$ ",
69			" ht ":"$H_{T}$ ",
70	ahart	1.3
71	lantonel	1.1	"d0":"$d_{0}$",
72
73	lantonel	1.11	" MET ":" Met ",
74			" MEt ":" Met ",
75			" met ":" Met ",
76			" Met ":"$\\not\\!\\!{E}_{T}$ ",
77
78			" MHT ":" Mht ",
79			" MHt ":" Mht ",
80			" mht ":" Mht ",
81			" Mht ":"$\\not\\!\\!{H}_{T}$ ",
82	lantonel	1.1
83			"M_Z" : "$M_{Z}$",
84			"M_mumu" : "$M_{\\mu\\mu}$",
85			"M_ee" : "$M_{ee}$",
86			"M_ll" : "$M_{ll}$",
87
88			}
89
90			secondary_replacements = {
91			"$$<$":"$<"
92	lantonel	1.11
93	lantonel	1.1	}
94
95
96			#### check which input datasets have valid output files
97			for dataset in datasets:
98			fileName = condor_dir + "/" + dataset + ".root"
99			if not os.path.exists(fileName):
100	lantonel	1.5	#print "Couldn't find output file for",dataset,"dataset"
101	lantonel	1.1	continue
102			testFile = TFile(fileName)
103			if not (testFile.IsZombie()):
104			processed_datasets.append(dataset)
105
106			if len(processed_datasets) is 0:
107	lantonel	1.5	sys.exit("Can't find any output root files for the given list of datasets")
108	lantonel	1.1
109			#### open first input file and re-make its directory structure in the output file
110			testFile = TFile(condor_dir + "/" + processed_datasets[0] + ".root")
111			testFile.cd()
112			for key in testFile.GetListOfKeys():
113			if (key.GetClassName() != "TDirectoryFile"):
114			continue
115			rootDirectory = key.GetName()
116			testFile.cd(key.GetName())
117			for key2 in gDirectory.GetListOfKeys():
118			if (key2.GetClassName() != "TDirectoryFile"):
119			continue
120			channels.append(key2.GetName())
121
122	ahart	1.3	fout = open (texfile, "w")
123	ahart	1.9	fout.write ("\\documentclass[a2paper,8pt]{article}\n\n")
124	ahart	1.3	fout.write ("\\usepackage[landscape,margin=0.15cm]{geometry}\n\n")
125			fout.write ("\\usepackage{multirow}\n\n")
126			fout.write ("\\begin{document}\n\n")
127			fout.write ("\\pagestyle{empty}\n\n")
128			fout.close ()
129
130			firstChannel = True
131
132	ahart	1.15	#weight = intLumi / 10000.0
133			#for dataset in processed_datasets:
134			# dataset_file = "%s/%s.root" % (condor_dir,dataset)
135			# fin = TFile (dataset_file)
136			# flags = fin.Get ("flags")
137			# noWeights = flags and flags.GetBinContent (1)
138			# fin.Close ()
139			# if types[dataset] != "data" and not noWeights:
140			# os.system("mergeTFileServiceHistograms -i %s -o %s -w %g" % (dataset_file, dataset_file + "_tmp", weight))
141			# else:
142			# os.system("mergeTFileServiceHistograms -i %s -o %s -w %g" % (dataset_file, dataset_file + "_tmp", 1.0))
143	ahart	1.4
144	wulsin	1.18
145	lantonel	1.1	for channel in channels: # loop over final states, which each have their own directory
146	lantonel	1.11	formatted_channel = channel.replace("_"," ")
147	ahart	1.6	cutFlowArgs = ""
148			selectionArgs = ""
149			minusOneArgs = ""
150	lantonel	1.1	#print hist
151	wulsin	1.18
152	lantonel	1.1	for dataset in processed_datasets:
153	ahart	1.15	dataset_file = "%s/%s.root" % (condor_dir,dataset)
154	lantonel	1.1	#print dataset_file
155	ahart	1.17	if arguments.makeDiffPlots or arguments.makeRatioPlots:
156			if types[dataset] == "data":
157			dataset_file = "\"<" + dataset_file + "\""
158			elif types[dataset] == "bgMC":
159			dataset_file = "\">" + dataset_file + "\""
160			if arguments.signalToBackground:
161			if types[dataset] == "signalMC":
162			dataset_file = "\"<" + dataset_file + "\""
163			elif types[dataset] == "bgMC":
164			dataset_file = "\">" + dataset_file + "\""
165	ahart	1.6	cutFlowArgs = cutFlowArgs + " " + dataset_file
166			selectionArgs = selectionArgs + " " + dataset_file
167			minusOneArgs = minusOneArgs + " " + dataset_file
168	lantonel	1.16	cutFlowArgs = cutFlowArgs + " '" + channel + "CutFlow'"
169			selectionArgs = selectionArgs + " '" + channel + "Selection'"
170			minusOneArgs = minusOneArgs + " '" + channel + "MinusOne'"
171	ahart	1.3
172	lantonel	1.1	rawlabel = "$" + labels[dataset] + "$"
173			label = rawlabel.replace("#","\\")
174			label = "'" + label + "'"
175			#print label
176	ahart	1.6	cutFlowArgs = cutFlowArgs + " " + label
177			selectionArgs = selectionArgs + " " + label
178			minusOneArgs = minusOneArgs + " " + label
179	lantonel	1.1
180			#make cutFlowTable objects
181	ahart	1.6	fout = open (texfile, "a")
182	ahart	1.8	if not firstChannel:
183			fout.write ("\\pagebreak\n\n")
184	ahart	1.17	cutFlowTable = "cutFlowTable"
185			if arguments.makeDiffPlots:
186			cutFlowTable += " -d \"Data - MC\""
187			if arguments.makeRatioPlots:
188			cutFlowTable += " -r \"(Data - MC) / MC\""
189			if arguments.signalToBackground:
190			cutFlowTable += " -s \"$\\mathrm{S} / \\sqrt{\\mathrm{S} + \\mathrm{B}}$\""
191	ahart	1.8	firstChannel = False
192	lantonel	1.11	fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
193	ahart	1.8	fout.write ("\\subsection*{Cut flow}\n\n")
194	ahart	1.6	fout.close ()
195	wulsin	1.19	# print "Debug: running command: %s -l %g -m %s >> %s" % (cutFlowTable, intLumi,cutFlowArgs,texfile)
196	ahart	1.17	os.system("%s -l %g -m %s >> %s" % (cutFlowTable, intLumi,cutFlowArgs,texfile))
197	ahart	1.6	fout = open (texfile, "a")
198			fout.write ("\\pagebreak\n\n")
199	lantonel	1.11	fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
200	ahart	1.8	fout.write ("\\subsection*{Individual selection}\n\n")
201	ahart	1.6	fout.close ()
202	ahart	1.17	os.system("%s -l %g %s >> %s" % (cutFlowTable, intLumi,selectionArgs,texfile))
203	ahart	1.6	fout = open (texfile, "a")
204			fout.write ("\\pagebreak\n\n")
205	lantonel	1.11	fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
206	ahart	1.8	fout.write ("\\subsection*{Minus one}\n\n")
207	ahart	1.6	fout.close ()
208	ahart	1.17	os.system("%s -l %g %s >> %s" % (cutFlowTable, intLumi,minusOneArgs,texfile))
209	ahart	1.4
210	ahart	1.15	#for dataset in processed_datasets:
211			# dataset_file = "%s/%s.root_tmp" % (condor_dir,dataset)
212			# os.remove(dataset_file)
213	ahart	1.3
214			#reformat tex files
215			for line in fileinput.FileInput(texfile,inplace=1):
216			for replacement in replacements.keys():
217			line = line.replace(replacement,replacements[replacement])
218			print line,
219
220			for line in fileinput.FileInput(texfile,inplace=1):
221			for replacement in secondary_replacements.keys():
222			line = line.replace(replacement,secondary_replacements[replacement])
223			print line,
224
225			fout = open (texfile, "a")
226			fout.write ("\\end{document}\n")
227			fout.close ()
228
229			#process tex files to make pdf files
230			command = "pdflatex -interaction=batchmode -output-directory=./%s %s > /dev/null" % (condor_dir,texfile)
231			os.system(command)
232			os.system(command)
233			#os.system("rm %s" % texfile)
234			os.system("rm %saux" % (texfile.rstrip("tex")))
235			os.system("rm %slog" % (texfile.rstrip("tex")))
236	wulsin	1.18
237			print "Finished writing cutFlow to " + condor_dir + "/" + outputFileName + "{.tex,.pdf}"
238