def main(filename1, filename2, outdir, regexp, verbose) :
# there are too many histos; by default compare only the ones vs. pt for few selections
relevantHistograms = ['l_pt_den','l_pt_num']
relevantSelections = ['CRPremT2','CR_SSInc','CR_WHSS','CRPremT2','CR_SSInc','CR_WHSS']
outdir = outdir if outdir else guessOutdirFromInputs(filename1, filename2)
if verbose : print "saving output plots to '%s'"%outdir
file1, file2 = r.TFile.Open(filename1), r.TFile.Open(filename2)
histonames1 = getAllHistoNames(file1, onlyTH1=True)
histonames2 = getAllHistoNames(file2, onlyTH1=True)
if verbose :
print '\n'.join(["%s: %d histograms"%(f, len(hs))
for f, hs in [(filename1, histonames1), (filename2, histonames2)]])
commonHistos = [h for h in histonames1 if h in histonames2]
def diff(a, b) : return list(set(a)-set(b))
if len(commonHistos)!=len(histonames2) :
missFrom1, missFrom2 = diff(histonames2, histonames1), diff(histonames1, histonames2)
print ('histograms not in common:\n'
+"%d missing from %s\n"%(len(missFrom1), filename1)
#+'\n\t'.join(missFrom1)
+"%d missing from %s\n"%(len(missFrom2), filename2)
#+'\n\t'.join(missFrom2)
)
label1, label2 = labelFromFilename(filename1), labelFromFilename(filename2)
commonHistos = (filterWithRegexp(commonHistos, regexp) if regexp is not None
else filter(lambda h:
any(s in h for s in relevantHistograms) and
any(s in h for s in relevantSelections),
commonHistos))
canvas = r.TCanvas('diff_fakeMatrix','diff_fakeMatrix')
for h in commonHistos :
outname = outdir+'/'+h
h1, h2 = file1.Get(h), file2.Get(h)
plotComparison(h1, h2, canvas, outname, label1, label2, verbose)
def processFile(filename, outdir, label=''):
file = r.TFile.Open(filename)
outfname = (outdir + '/%(histoname)s_' +
os.path.basename(filename).replace('.root', '.png'))
histonames = getAllHistoNames(file, onlyTH1=True)
probHistos = [
buildProbHisto(file, n, d)
for n, d in findHistonamePairs(histonames, '_num', '_den')
]
probHistos = filter(
None, probHistos) # remove empty graphs that make TAxis complain
c = r.TCanvas('p_tight', '')
for ph in probHistos:
c.cd()
c.Clear()
if label:
ph.SetTitle("%s %s" %
(ph.GetTitle(), ", %s" % label if label else ''))
ph.Draw('ap')
xAx, yAx = ph.GetXaxis(), ph.GetYaxis()
yAx.SetRangeUser(0.0, 1.1)
l = referenceLine(xAx.GetXmin(), xAx.GetXmax())
l.Draw()
c.Update()
c.SaveAs(outfname % {'histoname': ph.GetName()})
def main(filename1, filename2, outdir, regexp, verbose):
# there are too many histos; by default compare only the ones vs. pt for few selections
relevantHistograms = ['l_pt_den', 'l_pt_num']
relevantSelections = [
'CRPremT2', 'CR_SSInc', 'CR_WHSS', 'CRPremT2', 'CR_SSInc', 'CR_WHSS'
]
outdir = outdir if outdir else guessOutdirFromInputs(filename1, filename2)
if verbose: print "saving output plots to '%s'" % outdir
file1, file2 = r.TFile.Open(filename1), r.TFile.Open(filename2)
histonames1 = getAllHistoNames(file1, onlyTH1=True)
histonames2 = getAllHistoNames(file2, onlyTH1=True)
if verbose:
print '\n'.join([
"%s: %d histograms" % (f, len(hs))
for f, hs in [(filename1, histonames1), (filename2, histonames2)]
])
commonHistos = [h for h in histonames1 if h in histonames2]
def diff(a, b):
return list(set(a) - set(b))
if len(commonHistos) != len(histonames2):
missFrom1, missFrom2 = diff(histonames2,
histonames1), diff(histonames1,
histonames2)
print('histograms not in common:\n' + "%d missing from %s\n" %
(len(missFrom1), filename1)
#+'\n\t'.join(missFrom1)
+ "%d missing from %s\n" % (len(missFrom2), filename2)
#+'\n\t'.join(missFrom2)
)
label1, label2 = labelFromFilename(filename1), labelFromFilename(filename2)
commonHistos = (filterWithRegexp(commonHistos, regexp)
if regexp is not None else filter(
lambda h: any(s in h for s in relevantHistograms) and
any(s in h for s in relevantSelections), commonHistos))
canvas = r.TCanvas('diff_fakeMatrix', 'diff_fakeMatrix')
for h in commonHistos:
outname = outdir + '/' + h
h1, h2 = file1.Get(h), file2.Get(h)
plotComparison(h1, h2, canvas, outname, label1, label2, verbose)
def mergeFiles(targetFile, filenames, regex) :
infiles = [r.TFile.Open(f) for f in filenames]
histonames = [h for h in getAllHistoNames(infiles[0]) if re.search(regex, h)]
scale = 1.0/len(infiles)
outFile = r.TFile.Open(targetFile, 'recreate')
outFile.cd()
for hn in histonames :
h = infiles[0].Get(hn).Clone()
h.Scale(scale)
for inf in infiles[1:] : h.Add(inf.Get(hn), scale)
h.Write()
outFile.Close()
def main(filenameOrig, filenameDest, substr1, substr2) :
s1, s2 = substr1, substr2
input = r.TFile.Open(filenameOrig)
histonames = getAllHistoNames(input)
output = r.TFile.Open(filenameDest, 'recreate')
output.cd()
nhFixed = 0
for hn in histonames :
h = input.Get(hn)
hnn = hn.replace(s1,s2) if s1 in hn else hn.replace(s2,s1) # avoid swapping twice s1->s2->s1
h.Write(hnn)
if hn!=hnn : nhFixed += 1
print "renamed %d histograms"%nhFixed
output.Close()
input.Close()
def mergeFiles(targetFile, filenames, regex):
infiles = [r.TFile.Open(f) for f in filenames]
histonames = [
h for h in getAllHistoNames(infiles[0]) if re.search(regex, h)
]
scale = 1.0 / len(infiles)
outFile = r.TFile.Open(targetFile, 'recreate')
outFile.cd()
for hn in histonames:
h = infiles[0].Get(hn).Clone()
h.Scale(scale)
for inf in infiles[1:]:
h.Add(inf.Get(hn), scale)
h.Write()
outFile.Close()
def main(filenameOrig, filenameDest, substr1, substr2):
s1, s2 = substr1, substr2
input = r.TFile.Open(filenameOrig)
histonames = getAllHistoNames(input)
output = r.TFile.Open(filenameDest, 'recreate')
output.cd()
nhFixed = 0
for hn in histonames:
h = input.Get(hn)
hnn = hn.replace(s1, s2) if s1 in hn else hn.replace(
s2, s1) # avoid swapping twice s1->s2->s1
h.Write(hnn)
if hn != hnn: nhFixed += 1
print "renamed %d histograms" % nhFixed
output.Close()
input.Close()
def processFile(filename, outdir, label='') :
file = r.TFile.Open(filename)
outfname = (outdir+'/%(histoname)s_'
+os.path.basename(filename).replace('.root','.png'))
histonames = getAllHistoNames(file, onlyTH1=True)
probHistos = [buildProbHisto(file, n, d)
for n,d in findHistonamePairs(histonames, '_num', '_den')]
probHistos = filter(None, probHistos) # remove empty graphs that make TAxis complain
c = r.TCanvas('p_tight','')
for ph in probHistos :
c.cd()
c.Clear()
if label : ph.SetTitle("%s %s"%(ph.GetTitle(), ", %s"%label if label else ''))
ph.Draw('ap')
xAx, yAx = ph.GetXaxis(), ph.GetYaxis()
yAx.SetRangeUser(0.0, 1.1)
l = referenceLine(xAx.GetXmin(), xAx.GetXmax())
l.Draw()
c.Update()
c.SaveAs(outfname%{'histoname':ph.GetName()})
'pickleFile']])
print 'Input files:\n'+'\n'.join(inputFileNames)
# navigate the files and collect the histos
referenceType = HistoType(pr='', ch=channel, var=referenceHisto, syst=referenceSyst)
histosByType = collections.defaultdict(list)
classifier = HistoNameClassifier()
histoNames = []
for fname, infile in zip(inputFileNames, inputFiles) :
sample = guessGroupFromFilename(fname)
setType, setSample = setHistoType, setHistoSample
def getType(histoName) : return classifier.histoType(histoName)
def isRightType(histo) : return referenceType.matchAllAvailabeAttrs(histo.type)
histonamesCached = len(histoNames)>0
if not histonamesCached : histoNames = getAllHistoNames(infile, onlyTH1=True, nameStem=histoname)
histos = filter(isRightType, map(lambda hn :
setSample(setType(infile.Get(hn), getType(hn)), sample),
histoNames))
if not histonamesCached : histoNames = [h.GetName() for h in histos] # after filtering
organizeHistosByType(histosByType, histos)
refHistos = histosByType # already filtered histonames, all histosByType are refHistos
allSamples = sorted(list(set([h.sample for histos in refHistos.values() for h in histos])))
allSamples += ['totbkg'] if printTotBkg else []
allSelects = sorted(list(set([k.pr for k in histosByType.keys()])))
if verbose : print 'allSamples : ',allSamples
if verbose : print 'allSelects : ',allSelects
# get the counts (adding up what needs to be merged by samplename)
sampleCountsPerSel = dict() # counts[sample][sel]
countsSampleSel = dict([(s, collections.defaultdict(float)) for s in allSamples])
plotRegions = options.regions.split(',')
referenceSyst = options.syst
verbose = options.verbose
assert channel in validChannels,"Invalid channel %s (should be one of %s)" % (channel, str(validChannels))
inputFileNames = glob.glob(inputDir+'/'+'*'+prodTag+'*.root') + glob.glob(signalFname)
inputFiles = [r.TFile.Open(f) for f in inputFileNames]
assert len(inputFileNames)==len(inputFiles),"Cannot open some of the input files"
refHistoType = HistoType(pr='', ch=channel, var=referenceHisto, syst=referenceSyst)
histosByType = collections.defaultdict(list)
classifier = HistoNameClassifier()
for fname, infile in zip(inputFileNames, inputFiles) :
samplename = guessSampleFromFilename(fname)
histoNames = [n for n in getAllHistoNames(infile, onlyTH1=True)
if refHistoType.matchAllAvailabeAttrs( classifier.histoType( n ) )]
histos = [infile.Get(hn) for hn in histoNames]
for h in histos :
setHistoType(h, classifier.histoType(h.GetName()))
setHistoSample(h, samplename)
histos = [h for h in histos if h.type.pr in plotRegions]
organizeHistosByType(histosByType, histos)
refHistos = histosByType # already filtered histonames, all histosByType are refHistos
def isSignal(sampleName) : return 'WH_' in sampleName
allSamples = list(set([h.sample for histos in refHistos.values() for h in histos]))
allBkgNames = [s for s in allSamples if not isSignal(s)]
sigName = next(s for s in allSamples if isSignal(s))
if verbose : print '\n'.join("%s : %s" % (s,l) for s,l in zip(['bkg','sig'], [str(allBkgNames), sigName]))
'*.root') + glob.glob(signalFname)
inputFiles = [r.TFile.Open(f) for f in inputFileNames]
assert len(inputFileNames) == len(
inputFiles), "Cannot open some of the input files"
refHistoType = HistoType(pr='',
ch=channel,
var=referenceHisto,
syst=referenceSyst)
histosByType = collections.defaultdict(list)
classifier = HistoNameClassifier()
for fname, infile in zip(inputFileNames, inputFiles):
samplename = guessSampleFromFilename(fname)
histoNames = [
n for n in getAllHistoNames(infile, onlyTH1=True)
if refHistoType.matchAllAvailabeAttrs(classifier.histoType(n))
]
histos = [infile.Get(hn) for hn in histoNames]
for h in histos:
setHistoType(h, classifier.histoType(h.GetName()))
setHistoSample(h, samplename)
histos = [h for h in histos if h.type.pr in plotRegions]
organizeHistosByType(histosByType, histos)
refHistos = histosByType # already filtered histonames, all histosByType are refHistos
def isSignal(sampleName):
return 'WH_' in sampleName
signalScale = options.sigScale
justTest = options.test
verbose = options.verbose
inputFileNames = glob.glob(inputDir+'/'+'*'+prodTag+'*.root') + glob.glob(signalFname)
print 'input files:\n'+'\n'.join(inputFileNames)
inputFiles = [r.TFile.Open(f) for f in inputFileNames]
histosByType = collections.defaultdict(list)
classifier = HistoNameClassifier()
for fname, infile in zip(inputFileNames, inputFiles) :
print '-'*3 + fname + '-'*3
samplename = guessSampleFromFilename(fname)
histoNames = getAllHistoNames(inputFiles[0], onlyTH1=True)
histoNames = [h for h in histoNames if any([h.startswith(p) for p in ['sr6', 'sr7', 'sr8', 'sr9']])]
if justTest : histoNames = histoNames[:10] # just get 10 histos to run quick tests
histos = [infile.Get(hn) for hn in histoNames]
for h in histos :
setHistoType(h, classifier.histoType(h.GetName()))
setHistoSample(h, samplename)
organizeHistosByType(histosByType, histos)
def isSignal(sampleName) : return 'WH_' in sampleName
def cumsum(l, leftToRight=True) :
#return numpy.cumsum(l) # not available ?
return [sum(l[:i]) for i in range(1,len(l)+1)] if leftToRight \
else [sum(l[-i:]) for i in range(1,len(l)+1)][::-1]
def mergeOuter(bc, nOuter=2) : # add over/underflow in the first/last bin
def printHistoNames(inputFile) :
print 'histograms from ',inputFile.GetName()
print '\n'.join(getAllHistoNames(inputFile))
