class SamplesMerger :
def __init__(self) :
self.filenamesByReqid = collections.defaultdict(list)
self.masspointByReqid = collections.defaultdict(list)
self.reqidProvider = ModeAWhDbMergedFake2Lreqid()
self.parDb = ModeAWhDbPar()
self.reqDb = ModeAWhDbReqid()
self.overwrite = False
self.verbose = False
self.regexHist = '.*'
def idTagFromFilename(self, fname) :
'parse something like wA_noslep_WH_2Lep_9_May16_n0139.AnaHists'
match = re.search('wA_noslep_WH_2Lep_(?P<id>\d+?)_(?P<tag>.*?).AnaHist', fname)# nongreedy
return match.group('id'), match.group('tag')
def addFile(self, filename) :
sample = guessSampleFromFilename(filename)
mc1, mn1 = self.parDb.mc1Mn1ByReqid(self.reqDb.reqidBySample(sample))
fakeReqid = self.reqidProvider.reqidByMc1Mn1(mc1, mn1)
self.filenamesByReqid[fakeReqid].append(filename)
self.masspointByReqid[fakeReqid].append((mc1, mn1))
def mergeAndWrite(self, outdir) :
"""write merged histos, unless !overwrite, in which case just
compute merged coords and fnames"""
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()
self.mergedPointsByReqid = {}
self.outFnameByReqid = {}
reqids = self.filenamesByReqid.keys()
for rid in reqids :
fnames = self.filenamesByReqid[rid]
points = self.masspointByReqid[rid]
npoints = len(points)
assert npoints,"cannot merge 0 samples"
def avgPoint(points) :
np = float(len(points))
return (sum([x for x,y in points])/np,
sum([y for x,y in points])/np)
point = avgPoint(points)
ids, tags = zip(*[self.idTagFromFilename(f) for f in fnames]) # vertical slice
assert len(set(tags))==1,"cannot merge samples with different tags %s"%str(fnames)
outfname = outdir+'/'+os.path.basename(fnames[0]).replace(str(ids[0]), str(rid))
self.outFnameByReqid[rid], self.mergedPointsByReqid[rid] = outfname, point
if self.verbose : print str(point)+' : '+outfname
skipMerge = os.path.exists(outfname) and not self.overwrite
if skipMerge : continue
else : mergeFiles(outfname, fnames, self.regexHist)
def printMergedFiles(self) :
for rid in self.outFnameByReqid.keys() :
print self.mergedPointsByReqid[rid], ' : ', self.outFnameByReqid[rid]
countBkgTot[sel] += counts
print countBkgTot
reqDb = ModeAWhDbReqid()
parDb = ModeAWhDbPar()
def selIsRelevant(sel) : return sel.startswith('sr')
def selIsFinal(sel) : return sel in interestingSelections
mc1Range = {'min': min(parDb.allMc1()), 'max' : max(parDb.allMc1())}
mn1Range = {'min': min(parDb.allMn1()), 'max' : max(parDb.allMn1())}
histos = dict()
backgroundRelErr = 0.2
for sample, countsSel in countsSigSampleSel.iteritems() :
mc1, mn1 = parDb.mc1Mn1ByReqid(reqDb.reqidBySample(sample))
print "%s (%.1f, %.1f) " % (sample, mc1, mn1)
for sel, counts in sorted(countsSel.iteritems()) :
if not selIsFinal(sel) : continue
histo = histos[sel] if sel in histos else r.TH2F(sel+'_zn',
sel+" Z_{n} (#deltab=%.2f);mc_{1};mn_{1}"%backgroundRelErr,
50, float(mc1Range['min']), float(mc1Range['max']),
50, float(mn1Range['min']), float(mn1Range['max']))
if sel not in histos : histos[sel] = histo
sig, bkg, dBkg = counts, countBkgTot[sel], backgroundRelErr
zn = r.RooStats.NumberCountingUtils.BinomialExpZ(sigScale*sig, bkg, dBkg)
histo.Fill(mc1, mn1, zn)
print "%s : %.2f / %.2f -> %.2f"%(sel, sig, bkg, zn)
r.gStyle.SetPaintTextFormat('.2f')
for s, h in histos.iteritems() :
def selIsRelevant(sel):
return sel.startswith('sr')
def selIsFinal(sel):
return sel in interestingSelections
mc1Range = {'min': min(parDb.allMc1()), 'max': max(parDb.allMc1())}
mn1Range = {'min': min(parDb.allMn1()), 'max': max(parDb.allMn1())}
histos = dict()
backgroundRelErr = 0.2
for sample, countsSel in countsSigSampleSel.iteritems():
mc1, mn1 = parDb.mc1Mn1ByReqid(reqDb.reqidBySample(sample))
print "%s (%.1f, %.1f) " % (sample, mc1, mn1)
for sel, counts in sorted(countsSel.iteritems()):
if not selIsFinal(sel): continue
histo = histos[sel] if sel in histos else r.TH2F(
sel + '_zn', sel +
" Z_{n} (#deltab=%.2f);mc_{1};mn_{1}" % backgroundRelErr, 50,
float(mc1Range['min']), float(mc1Range['max']), 50,
float(mn1Range['min']), float(mn1Range['max']))
if sel not in histos: histos[sel] = histo
sig, bkg, dBkg = counts, countBkgTot[sel], backgroundRelErr
zn = r.RooStats.NumberCountingUtils.BinomialExpZ(
sigScale * sig, bkg, dBkg)
histo.Fill(mc1, mn1, zn)
print "%s : %.2f / %.2f -> %.2f" % (sel, sig, bkg, zn)
class SamplesMerger:
def __init__(self):
self.filenamesByReqid = collections.defaultdict(list)
self.masspointByReqid = collections.defaultdict(list)
self.reqidProvider = ModeAWhDbMergedFake2Lreqid()
self.parDb = ModeAWhDbPar()
self.reqDb = ModeAWhDbReqid()
self.overwrite = False
self.verbose = False
self.regexHist = '.*'
def idTagFromFilename(self, fname):
'parse something like wA_noslep_WH_2Lep_9_May16_n0139.AnaHists'
match = re.search(
'wA_noslep_WH_2Lep_(?P<id>\d+?)_(?P<tag>.*?).AnaHist',
fname) # nongreedy
return match.group('id'), match.group('tag')
def addFile(self, filename):
sample = guessSampleFromFilename(filename)
mc1, mn1 = self.parDb.mc1Mn1ByReqid(self.reqDb.reqidBySample(sample))
fakeReqid = self.reqidProvider.reqidByMc1Mn1(mc1, mn1)
self.filenamesByReqid[fakeReqid].append(filename)
self.masspointByReqid[fakeReqid].append((mc1, mn1))
def mergeAndWrite(self, outdir):
"""write merged histos, unless !overwrite, in which case just
compute merged coords and fnames"""
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()
self.mergedPointsByReqid = {}
self.outFnameByReqid = {}
reqids = self.filenamesByReqid.keys()
for rid in reqids:
fnames = self.filenamesByReqid[rid]
points = self.masspointByReqid[rid]
npoints = len(points)
assert npoints, "cannot merge 0 samples"
def avgPoint(points):
np = float(len(points))
return (sum([x for x, y in points]) / np,
sum([y for x, y in points]) / np)
point = avgPoint(points)
ids, tags = zip(*[self.idTagFromFilename(f)
for f in fnames]) # vertical slice
assert len(
set(tags)
) == 1, "cannot merge samples with different tags %s" % str(fnames)
outfname = outdir + '/' + os.path.basename(fnames[0]).replace(
str(ids[0]), str(rid))
self.outFnameByReqid[rid], self.mergedPointsByReqid[
rid] = outfname, point
if self.verbose: print str(point) + ' : ' + outfname
skipMerge = os.path.exists(outfname) and not self.overwrite
if skipMerge: continue
else: mergeFiles(outfname, fnames, self.regexHist)
def printMergedFiles(self):
for rid in self.outFnameByReqid.keys():
print self.mergedPointsByReqid[rid], ' : ', self.outFnameByReqid[
rid]
