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("-q")
parser.remove_option("-n")
parser.remove_option("-u")
parser.remove_option("-e")
parser.remove_option("-R")
parser.remove_option("-b")
parser.remove_option("-t")
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 ()
    os.system("%s -l %g -m %s >> %s" % (cutFlowTable, intLumi,cutFlowArgs,texfile))
    #    print "Debug:  running command:  %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:	1.18
Committed:	Mon Jul 8 08:55:37 2013 UTC (11 years, 10 months ago) by wulsin
Content type:	text/x-python
Branch:	MAIN
Changes since 1.17:	+27 -2 lines
Log Message:	Add option to specify output file name; remove unused options
#	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("-q")
15			parser.remove_option("-n")
16			parser.remove_option("-u")
17			parser.remove_option("-e")
18			parser.remove_option("-R")
19			parser.remove_option("-b")
20			parser.remove_option("-t")
21			parser.remove_option("-y")
22			parser.remove_option("-p")
23	ahart	1.17	parser.remove_option("-r")
24	wulsin	1.18	parser.remove_option("-o")
25			parser.remove_option("--2D")
26	ahart	1.17
27			parser.add_option("-d", "--diff", action="store_true", dest="makeDiffPlots", default=False,
28			help="add a column for data-MC to the tables")
29			parser.add_option("-r", "--ratio", action="store_true", dest="makeRatioPlots", default=False,
30			help="add a colunn for (data-MC)/MC to the tables")
31			parser.add_option("-s", "--sToB", action="store_true", dest="signalToBackground", default=False,
32			help="add a column for S/sqrt(S+B) to the tables")
33	wulsin	1.18	parser.add_option("-o", "--output-file", dest="outputFileName",
34			help="specify an output file base name for the cutflow table (suffix will be appended), default is 'cutFlow'")
35
36	ahart	1.17
37	lantonel	1.13	(arguments, args) = parser.parse_args()
38	lantonel	1.1
39	lantonel	1.12	if arguments.localConfig:
40	lantonel	1.1	sys.path.append(os.getcwd())
41	lantonel	1.12	exec("from " + arguments.localConfig.rstrip('.py') + " import *")
42	lantonel	1.1
43	lantonel	1.12	condor_dir = set_condor_output_dir(arguments)
44	lantonel	1.1
45			from ROOT import TFile, gROOT, gStyle, gDirectory, TKey
46
47			channels = []
48			processed_datasets = []
49	wulsin	1.18
50			outputFileName = "cutFlow"
51			if arguments.outputFileName:
52			outputFileName = arguments.outputFileName
53			outputFileName = outputFileName.partition(".")[0] # If the input filename contains a period, take only the part before the period.
54
55			texfile = condor_dir + "/" + outputFileName + ".tex"
56	lantonel	1.1
57			replacements = {
58			">":"$>$",
59			"<":"$<$",
60	lantonel	1.11	" eta ":" $\\eta$ ",
61			" abs(eta) ":" $\|\\eta\|$ ",
62	lantonel	1.1	"#":"Num",
63
64			"\\rightarrow":"{\\rightarrow}",
65			"\\mu QCD":"\\mu$ $QCD",
66			"EM QCD":"EM$ $QCD",
67			"BCtoE QCD":"BCtoE$ $QCD",
68	ahart	1.3
69
70	lantonel	1.11	" pt ":" $p_{T}$ ",
71			" ht ":"$H_{T}$ ",
72	ahart	1.3
73	lantonel	1.1	"d0":"$d_{0}$",
74
75	lantonel	1.11	" MET ":" Met ",
76			" MEt ":" Met ",
77			" met ":" Met ",
78			" Met ":"$\\not\\!\\!{E}_{T}$ ",
79
80			" MHT ":" Mht ",
81			" MHt ":" Mht ",
82			" mht ":" Mht ",
83			" Mht ":"$\\not\\!\\!{H}_{T}$ ",
84	lantonel	1.1
85			"M_Z" : "$M_{Z}$",
86			"M_mumu" : "$M_{\\mu\\mu}$",
87			"M_ee" : "$M_{ee}$",
88			"M_ll" : "$M_{ll}$",
89
90			}
91
92			secondary_replacements = {
93			"$$<$":"$<"
94	lantonel	1.11
95	lantonel	1.1	}
96
97
98			#### check which input datasets have valid output files
99			for dataset in datasets:
100			fileName = condor_dir + "/" + dataset + ".root"
101			if not os.path.exists(fileName):
102	lantonel	1.5	#print "Couldn't find output file for",dataset,"dataset"
103	lantonel	1.1	continue
104			testFile = TFile(fileName)
105			if not (testFile.IsZombie()):
106			processed_datasets.append(dataset)
107
108			if len(processed_datasets) is 0:
109	lantonel	1.5	sys.exit("Can't find any output root files for the given list of datasets")
110	lantonel	1.1
111			#### open first input file and re-make its directory structure in the output file
112			testFile = TFile(condor_dir + "/" + processed_datasets[0] + ".root")
113			testFile.cd()
114			for key in testFile.GetListOfKeys():
115			if (key.GetClassName() != "TDirectoryFile"):
116			continue
117			rootDirectory = key.GetName()
118			testFile.cd(key.GetName())
119			for key2 in gDirectory.GetListOfKeys():
120			if (key2.GetClassName() != "TDirectoryFile"):
121			continue
122			channels.append(key2.GetName())
123
124	ahart	1.3	fout = open (texfile, "w")
125	ahart	1.9	fout.write ("\\documentclass[a2paper,8pt]{article}\n\n")
126	ahart	1.3	fout.write ("\\usepackage[landscape,margin=0.15cm]{geometry}\n\n")
127			fout.write ("\\usepackage{multirow}\n\n")
128			fout.write ("\\begin{document}\n\n")
129			fout.write ("\\pagestyle{empty}\n\n")
130			fout.close ()
131
132			firstChannel = True
133
134	ahart	1.15	#weight = intLumi / 10000.0
135			#for dataset in processed_datasets:
136			# dataset_file = "%s/%s.root" % (condor_dir,dataset)
137			# fin = TFile (dataset_file)
138			# flags = fin.Get ("flags")
139			# noWeights = flags and flags.GetBinContent (1)
140			# fin.Close ()
141			# if types[dataset] != "data" and not noWeights:
142			# os.system("mergeTFileServiceHistograms -i %s -o %s -w %g" % (dataset_file, dataset_file + "_tmp", weight))
143			# else:
144			# os.system("mergeTFileServiceHistograms -i %s -o %s -w %g" % (dataset_file, dataset_file + "_tmp", 1.0))
145	ahart	1.4
146	wulsin	1.18
147	lantonel	1.1	for channel in channels: # loop over final states, which each have their own directory
148	lantonel	1.11	formatted_channel = channel.replace("_"," ")
149	ahart	1.6	cutFlowArgs = ""
150			selectionArgs = ""
151			minusOneArgs = ""
152	lantonel	1.1	#print hist
153	wulsin	1.18
154	lantonel	1.1	for dataset in processed_datasets:
155	ahart	1.15	dataset_file = "%s/%s.root" % (condor_dir,dataset)
156	lantonel	1.1	#print dataset_file
157	ahart	1.17	if arguments.makeDiffPlots or arguments.makeRatioPlots:
158			if types[dataset] == "data":
159			dataset_file = "\"<" + dataset_file + "\""
160			elif types[dataset] == "bgMC":
161			dataset_file = "\">" + dataset_file + "\""
162			if arguments.signalToBackground:
163			if types[dataset] == "signalMC":
164			dataset_file = "\"<" + dataset_file + "\""
165			elif types[dataset] == "bgMC":
166			dataset_file = "\">" + dataset_file + "\""
167	ahart	1.6	cutFlowArgs = cutFlowArgs + " " + dataset_file
168			selectionArgs = selectionArgs + " " + dataset_file
169			minusOneArgs = minusOneArgs + " " + dataset_file
170	lantonel	1.16	cutFlowArgs = cutFlowArgs + " '" + channel + "CutFlow'"
171			selectionArgs = selectionArgs + " '" + channel + "Selection'"
172			minusOneArgs = minusOneArgs + " '" + channel + "MinusOne'"
173	ahart	1.3
174	lantonel	1.1	rawlabel = "$" + labels[dataset] + "$"
175			label = rawlabel.replace("#","\\")
176			label = "'" + label + "'"
177			#print label
178	ahart	1.6	cutFlowArgs = cutFlowArgs + " " + label
179			selectionArgs = selectionArgs + " " + label
180			minusOneArgs = minusOneArgs + " " + label
181	lantonel	1.1
182			#make cutFlowTable objects
183	ahart	1.6	fout = open (texfile, "a")
184	ahart	1.8	if not firstChannel:
185			fout.write ("\\pagebreak\n\n")
186	ahart	1.17	cutFlowTable = "cutFlowTable"
187			if arguments.makeDiffPlots:
188			cutFlowTable += " -d \"Data - MC\""
189			if arguments.makeRatioPlots:
190			cutFlowTable += " -r \"(Data - MC) / MC\""
191			if arguments.signalToBackground:
192			cutFlowTable += " -s \"$\\mathrm{S} / \\sqrt{\\mathrm{S} + \\mathrm{B}}$\""
193	ahart	1.8	firstChannel = False
194	lantonel	1.11	fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
195	ahart	1.8	fout.write ("\\subsection*{Cut flow}\n\n")
196	ahart	1.6	fout.close ()
197	ahart	1.17	os.system("%s -l %g -m %s >> %s" % (cutFlowTable, intLumi,cutFlowArgs,texfile))
198	wulsin	1.18	# print "Debug: running command: %s -l %g -m %s >> %s" % (cutFlowTable, intLumi,cutFlowArgs,texfile)
199	ahart	1.6	fout = open (texfile, "a")
200			fout.write ("\\pagebreak\n\n")
201	lantonel	1.11	fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
202	ahart	1.8	fout.write ("\\subsection*{Individual selection}\n\n")
203	ahart	1.6	fout.close ()
204	ahart	1.17	os.system("%s -l %g %s >> %s" % (cutFlowTable, intLumi,selectionArgs,texfile))
205	ahart	1.6	fout = open (texfile, "a")
206			fout.write ("\\pagebreak\n\n")
207	lantonel	1.11	fout.write ("\\section*{" + formatted_channel + " channel}\n\n")
208	ahart	1.8	fout.write ("\\subsection*{Minus one}\n\n")
209	ahart	1.6	fout.close ()
210	ahart	1.17	os.system("%s -l %g %s >> %s" % (cutFlowTable, intLumi,minusOneArgs,texfile))
211	ahart	1.4
212	ahart	1.15	#for dataset in processed_datasets:
213			# dataset_file = "%s/%s.root_tmp" % (condor_dir,dataset)
214			# os.remove(dataset_file)
215	ahart	1.3
216			#reformat tex files
217			for line in fileinput.FileInput(texfile,inplace=1):
218			for replacement in replacements.keys():
219			line = line.replace(replacement,replacements[replacement])
220			print line,
221
222			for line in fileinput.FileInput(texfile,inplace=1):
223			for replacement in secondary_replacements.keys():
224			line = line.replace(replacement,secondary_replacements[replacement])
225			print line,
226
227			fout = open (texfile, "a")
228			fout.write ("\\end{document}\n")
229			fout.close ()
230
231			#process tex files to make pdf files
232			command = "pdflatex -interaction=batchmode -output-directory=./%s %s > /dev/null" % (condor_dir,texfile)
233			os.system(command)
234			os.system(command)
235			#os.system("rm %s" % texfile)
236			os.system("rm %saux" % (texfile.rstrip("tex")))
237			os.system("rm %slog" % (texfile.rstrip("tex")))
238	wulsin	1.18
239			print "Finished writing cutFlow to " + condor_dir + "/" + outputFileName + "{.tex,.pdf}"
240