예제 #1
0
    def run(self):
        outpostfix = self.postfix if self.postfix != None else (
            "_Friend" if self.friend else "_Skim")
        if not self.noOut:

            if self.compression != "none":
                ROOT.gInterpreter.ProcessLine("#include <Compression.h>")
                (algo, level) = self.compression.split(":")
                compressionLevel = int(level)
                if algo == "LZMA":
                    compressionAlgo = ROOT.ROOT.kLZMA
                elif algo == "ZLIB":
                    compressionAlgo = ROOT.ROOT.kZLIB
                elif algo == "LZ4":
                    compressionAlgo = ROOT.ROOT.kLZ4
                else:
                    raise RuntimeError("Unsupported compression %s" % algo)
            else:
                compressionLevel = 0
            print("Will write selected trees to " + self.outputDir)
            if not self.justcount:
                if not os.path.exists(self.outputDir):
                    os.system("mkdir -p " + self.outputDir)
        else:
            compressionLevel = 0

        if self.noOut:
            if len(self.modules) == 0:
                raise RuntimeError(
                    "Running with --noout and no modules does nothing!")

        # Open histogram file, if desired
        if (self.histFileName != None and self.histDirName == None) or (
                self.histFileName == None and self.histDirName != None):
            raise RuntimeError(
                "Must specify both histogram file and histogram directory!")
        elif self.histFileName != None and self.histDirName != None:
            self.histFile = ROOT.TFile.Open(self.histFileName, "RECREATE")
        else:
            self.histFile = None

        for m in self.modules:
            if hasattr(m, 'writeHistFile') and m.writeHistFile:
                m.beginJob(histFile=self.histFile,
                           histDirName=self.histDirName)
            else:
                m.beginJob()

        fullClone = (len(self.modules) == 0)
        outFileNames = []
        t0 = time.time()
        totEntriesRead = 0
        for fname in self.inputFiles:
            ffnames = []
            if "," in fname:
                fnames = fname.split(',')
                fname, ffnames = fnames[0], fnames[1:]

            # open input file
            if self.prefetch:
                ftoread, toBeDeleted = self.prefetchFile(fname)
                inFile = ROOT.TFile.Open(ftoread)
            else:
                inFile = ROOT.TFile.Open(fname)

            # get input tree
            inTree = inFile.Get("Events")
            if inTree == None:
                inTree = inFile.Get("Friends")
            nEntries = min(inTree.GetEntries() - self.firstEntry,
                           self.maxEntries)
            totEntriesRead += nEntries
            # pre-skimming
            elist, jsonFilter = preSkim(inTree,
                                        self.json,
                                        self.cut,
                                        maxEntries=self.maxEntries,
                                        firstEntry=self.firstEntry)
            if self.justcount:
                print('Would select %d / %d entries from %s (%.2f%%)' %
                      (elist.GetN() if elist else nEntries, nEntries, fname,
                       (elist.GetN() if elist else nEntries) /
                       (0.01 * nEntries) if nEntries else 0))
                if self.prefetch:
                    if toBeDeleted:
                        os.unlink(ftoread)
                continue
            else:
                print('Pre-select %d entries out of %s (%.2f%%)' %
                      (elist.GetN() if elist else nEntries, nEntries,
                       (elist.GetN() if elist else nEntries) /
                       (0.01 * nEntries) if nEntries else 0))
                inAddFiles = []
                inAddTrees = []
            for ffname in ffnames:
                inAddFiles.append(ROOT.TFile.Open(ffname))
                inAddTree = inAddFiles[-1].Get("Events")
                if inAddTree == None:
                    inAddTree = inAddFiles[-1].Get("Friends")
                inAddTrees.append(inAddTree)
                inTree.AddFriend(inAddTree)

            if fullClone:
                # no need of a reader (no event loop), but set up the elist if available
                if elist:
                    inTree.SetEntryList(elist)
            else:
                # initialize reader
                inTree = InputTree(inTree, elist)

            # prepare output file
            if not self.noOut:
                outFileName = os.path.join(
                    self.outputDir,
                    os.path.basename(fname).replace(".root",
                                                    outpostfix + ".root"))
                outFile = ROOT.TFile.Open(outFileName, "RECREATE", "",
                                          compressionLevel)
                outFileNames.append(outFileName)
                if compressionLevel:
                    outFile.SetCompressionAlgorithm(compressionAlgo)
                # prepare output tree
                if self.friend:
                    outTree = FriendOutput(inFile, inTree, outFile)
                else:
                    outTree = FullOutput(
                        inFile,
                        inTree,
                        outFile,
                        branchSelection=self.branchsel,
                        outputbranchSelection=self.outputbranchsel,
                        fullClone=fullClone,
                        maxEntries=self.maxEntries,
                        firstEntry=self.firstEntry,
                        jsonFilter=jsonFilter,
                        provenance=self.provenance)
            else:
                outFile = None
                outTree = None
                if self.branchsel:
                    self.branchsel.selectBranches(inTree)

            # process events, if needed
            if not fullClone:
                eventRange = range(
                    self.firstEntry, self.firstEntry +
                    nEntries) if nEntries > 0 and not elist else None
                (nall, npass, timeLoop) = eventLoop(self.modules,
                                                    inFile,
                                                    outFile,
                                                    inTree,
                                                    outTree,
                                                    eventRange=eventRange,
                                                    maxEvents=self.maxEntries)
                print(
                    'Processed %d preselected entries from %s (%s entries). Finally selected %d entries'
                    % (nall, fname, nEntries, npass))
            else:
                nall = nEntries
                print('Selected %d / %d entries from %s (%.2f%%)' %
                      (outTree.tree().GetEntries(), nall, fname,
                       outTree.tree().GetEntries() /
                       (0.01 * nall) if nall else 0))

            # now write the output
            if not self.noOut:
                outTree.write()
                outFile.Close()
                print("Done %s" % outFileName)
            if self.jobReport:
                self.jobReport.addInputFile(fname, nall)
            if self.prefetch:
                if toBeDeleted:
                    os.unlink(ftoread)

        for m in self.modules:
            m.endJob()

        print("Total time %.1f sec. to process %i events. Rate = %.1f Hz." %
              ((time.time() - t0), totEntriesRead, totEntriesRead /
               (time.time() - t0)))

        if self.haddFileName:
            haddnano = "./haddnano.py" if os.path.isfile(
                "./haddnano.py") else "haddnano.py"
            os.system("%s %s %s" %
                      (haddnano, self.haddFileName, " ".join(outFileNames)))
        if self.jobReport:
            self.jobReport.addOutputFile(self.haddFileName)
            self.jobReport.save()
예제 #2
0
    def run(self):
        if not self.noOut:
            outpostfix = self.postfix if self.postfix != None else (
                "_Friend" if self.friend else "_Skim")
            if self.compression != "none":
                ROOT.gInterpreter.ProcessLine("#include <Compression.h>")
                (algo, level) = self.compression.split(":")
                compressionLevel = int(level)
                if algo == "LZMA": compressionAlgo = ROOT.ROOT.kLZMA
                elif algo == "ZLIB": compressionAlgo = ROOT.ROOT.kZLIB
                else: raise RuntimeError("Unsupported compression %s" % algo)
            else:
                compressionLevel = 0
            print "Will write selected trees to " + self.outputDir
            if not self.justcount:
                if not os.path.exists(self.outputDir):
                    os.system("mkdir -p " + self.outputDir)

        if self.noOut:
            if len(self.modules) == 0:
                raise RuntimeError(
                    "Running with --noout and no modules does nothing!")

        for m in self.modules:
            m.beginJob()

        fullClone = (len(self.modules) == 0)
        outFileNames = []
        t0 = time.clock()
        totEntriesRead = 0
        for fname in self.inputFiles:

            # open input file
            inFile = ROOT.TFile.Open(fname)

            #get input tree
            inTree = inFile.Get("Events")
            totEntriesRead += inTree.GetEntries()
            # pre-skimming
            elist, jsonFilter = preSkim(inTree, self.json, self.cut)
            if self.justcount:
                print 'Would select %d entries from %s' % (
                    elist.GetN() if elist else inTree.GetEntries(), fname)
                continue
            else:
                print 'Pre-select %d entries out of %s ' % (
                    elist.GetN() if elist else inTree.GetEntries(),
                    inTree.GetEntries())

            if fullClone:
                # no need of a reader (no event loop), but set up the elist if available
                if elist: inTree.SetEntryList(elist)
            else:
                # initialize reader
                inTree = InputTree(inTree, elist)

            # prepare output file
            outFileName = os.path.join(
                self.outputDir,
                os.path.basename(fname).replace(".root", outpostfix + ".root"))
            outFile = ROOT.TFile.Open(outFileName, "RECREATE", "",
                                      compressionLevel)
            outFileNames.append(outFileName)
            if compressionLevel:
                outFile.SetCompressionAlgorithm(compressionAlgo)

            # prepare output tree
            if self.friend:
                outTree = FriendOutput(inFile, inTree, outFile)
            else:
                outTree = FullOutput(inFile,
                                     inTree,
                                     outFile,
                                     branchSelection=self.branchsel,
                                     fullClone=fullClone,
                                     jsonFilter=jsonFilter,
                                     provenance=self.provenance)

            # process events, if needed
            if not fullClone:
                (nall, npass, timeLoop) = eventLoop(self.modules, inFile,
                                                    outFile, inTree, outTree)
                print 'Processed %d preselected entries from %s (%s entries). Finally selected %d entries' % (
                    nall, fname, inTree.GetEntries(), npass)
            else:
                print 'Selected %d entries from %s' % (
                    outTree.tree().GetEntries(), fname)

            # now write the output
            outTree.write()
            outFile.Close()
            print "Done %s" % outFileName
            if self.jobReport:
                self.jobReport.addInputFile(fname, nall)

        for m in self.modules:
            m.endJob()

        print totEntriesRead / (time.clock() - t0), "Hz"

        if self.haddFileName:
            os.system(
                "./haddnano.py %s %s" %
                (self.haddFileName, " ".join(outFileNames))
            )  #FIXME: remove "./" once haddnano.py is distributed with cms releases
        if self.jobReport:
            self.jobReport.addOutputFile(self.haddFileName)
            self.jobReport.save()
    def run(self):
        outpostfix = self.postfix if self.postfix != None else (
            "_Friend" if self.friend else "_Skim")
        fullClone = False
        outFileNames = []
        totEntriesRead = 0
        t0 = time.time()

        for fileName in self.inputFiles:
            # open file
            print("Opening file %s" % fileName)
            inFile = ROOT.TFile.Open(fileName)
            if (not inFile):  #check for null pointer
                print("Unable to open file %s, exting \n" % fileName)
                return 1

            # get input tree
            inTree = inFile.Get("Events")
            nEntries = min(inTree.GetEntries() - self.firstEntry,
                           self.maxEntries)
            totEntriesRead += nEntries

            # pre-skimming
            elist, jsonFilter = preSkim(inTree,
                                        self.json,
                                        self.cut,
                                        maxEntries=self.maxEntries,
                                        firstEntry=self.firstEntry)

            # number of events to be processed
            nTotal = elist.GetN() if elist else nEntries
            print('Pre-select %d entries out of %s ' % (nTotal, nEntries))

            inTree = InputTree(inTree, elist)

            # output
            outFileName = os.path.join(
                self.outputDir,
                os.path.basename(fileName).replace(".root",
                                                   outpostfix + ".root"))
            #compressionAlgo  = ROOT.ROOT.kLZMA
            #compressionLevel = int(9)
            compressionAlgo = ROOT.ROOT.kLZ4
            compressionLevel = int(4)
            outFile = ROOT.TFile.Open(outFileName, "RECREATE", "",
                                      compressionLevel)
            outFileNames.append(outFileName)
            outFile.SetCompressionAlgorithm(compressionAlgo)
            maxEntries = self.maxEntries
            if self.perJet:  #save two first jets
                maxEntries = self.maxEntries * 2
            outTree = FullOutput(inFile,
                                 inTree,
                                 outFile,
                                 branchSelection=self.branchsel,
                                 outputbranchSelection=self.outputbranchsel,
                                 fullClone=fullClone,
                                 maxEntries=maxEntries,
                                 firstEntry=self.firstEntry,
                                 jsonFilter=jsonFilter,
                                 provenance=self.provenance)

            t0 = time.time()
            tlast = t0
            doneEvents = 0
            acceptedEvents = 0
            if elist:
                eventRange = [(elist.GetEntry(0) if i == 0 else elist.Next())
                              for i in range(elist.GetN())]
            else:
                eventRange = range(self.firstEntry, self.firstEntry +
                                   nEntries) if nEntries > 0 else None

            entries = inTree.entries
            if eventRange:
                entries = len(eventRange)
            maxEvents = self.maxEntries
            if maxEvents > 0:
                entries = min(entries, self.maxEntries)
            entriesRange = range(entries) if eventRange == None else eventRange

            for m in self.modules:
                m.beginFile(inFile, outFile, inTree, outTree, entriesRange)

            for ie, i in enumerate(entriesRange):
                if maxEvents > 0 and ie >= maxEvents: break
                e = Event(inTree, ie)

                ret = True
                if self.perJet:
                    #print('ie ',ie)
                    for m in self.modules:
                        ret = m.analyze(e, ie)
                        if not ret: break
                        else:
                            clearExtraBranches(inTree)
                            m.fill(e, ie)
                else:
                    clearExtraBranches(inTree)
                    for m in self.modules:
                        ret = m.analyze(e, ie)
                        if not ret: break
                    if ret and outTree is not None:
                        outTree.fill()
                if ret:
                    acceptedEvents += 1
            for m in self.modules:
                m.endFile(inFile, outFile, inTree, outTree)

            outTree.write()
            outFile.Close()
            print("Done %s" % outFileName)

        for m in self.modules:
            m.endJob()
        print("Total time %.1f sec. to process %i events. Rate = %.1f Hz." %
              ((time.time() - t0), totEntriesRead, totEntriesRead /
               (time.time() - t0)))
예제 #4
0
    def run(self):
        outpostfix = self.postfix if self.postfix != None else (
            "_Friend" if self.friend else "_Skim")
        if not self.noOut:

            if self.compression != "none":
                ROOT.gInterpreter.ProcessLine("#include <Compression.h>")
                (algo, level) = self.compression.split(":")
                compressionLevel = int(level)
                if algo == "LZMA": compressionAlgo = ROOT.ROOT.kLZMA
                elif algo == "ZLIB": compressionAlgo = ROOT.ROOT.kZLIB
                else: raise RuntimeError("Unsupported compression %s" % algo)
            else:
                compressionLevel = 0
            print "Will write selected trees to " + self.outputDir
            if not self.justcount:
                if not os.path.exists(self.outputDir):
                    os.system("mkdir -p " + self.outputDir)
        else:
            compressionLevel = 0

        if self.noOut:
            if len(self.modules) == 0:
                raise RuntimeError(
                    "Running with --noout and no modules does nothing!")

        # Open histogram file, if desired
        if (self.histFileName != None and self.histDirName == None) or (
                self.histFileName == None and self.histDirName != None):
            raise RuntimeError(
                "Must specify both histogram file and histogram directory!")
        elif self.histFileName != None and self.histDirName != None:
            self.histFile = ROOT.TFile.Open(self.histFileName, "RECREATE")
        else:
            self.histFile = None

        for m in self.modules:
            if hasattr(m, 'writeHistFile') and m.writeHistFile:
                m.beginJob(histFile=self.histFile,
                           histDirName=self.histDirName)
            else:
                m.beginJob()

        fullClone = (len(self.modules) == 0)
        outFileNames = []
        t0 = time.clock()
        totEntriesRead = 0
        for fname in self.inputFiles:

            # open input file
            inFile = ROOT.TFile.Open(fname)

            #get input tree
            inTree = inFile.Get("Events")
            totEntriesRead += inTree.GetEntries()
            self.hcount.SetBinContent(1, inTree.GetEntries())
            ROOT.gROOT.SetBatch(True)
            if self.SMSMasses != None:
                inTree.Draw(
                    "MaxIf$(GenPart_mass, abs(GenPart_pdgId) == %i):MaxIf$(GenPart_mass, abs(GenPart_pdgId) == %i) >> hSMS(2000, -0.5, 1999.5, 2000, -0.5, 1999.5)"
                    % (self.SMSMasses[0], self.SMSMasses[1]))
                self.hsmscount = ROOT.gDirectory.Get('hSMS')
            if self.doISR != None:
                #Dirty ISR recipe for EWKinos
                #Need to correct for each mass point
                #Can't correct per sample (wrong normalization), need to save whole unskimmed histogram per point an then postprocess
                pt1 = "MaxIf$(GenPart_pt, abs(GenPart_pdgId) == %i && GenPart_status == 22)" % self.doISR[
                    0]
                pt2 = "MaxIf$(GenPart_pt, abs(GenPart_pdgId) == %i && GenPart_status == 22)" % self.doISR[
                    1]
                phi1 = "MaxIf$(GenPart_pt, abs(GenPart_pdgId) == %i && GenPart_status == 22)" % self.doISR[
                    0]
                phi2 = "MaxIf$(GenPart_pt, abs(GenPart_pdgId) == %i && GenPart_status == 22)" % self.doISR[
                    1]
                pt_ISR = "hypot(%s + %s * cos(%s-%s), %s*sin(%s - %s))" % (
                    pt1, pt2, phi2, phi1, pt2, phi2, phi1)
                inTree.Draw(
                    " %s : MaxIf$(GenPart_mass, abs(GenPart_pdgId) == %i) : MaxIf$(GenPart_mass, abs(GenPart_pdgId) == %i)  >> hISR(1000, -0.5, 1999.5, 1000, -0.5, 1999.5, 20, 0, 1000)"
                    % (pt_ISR, self.SMSMasses[0], self.SMSMasses[1]))
                self.hISR = ROOT.gDirectory.Get("hISR")
            if inTree.GetBranchStatus("genWeight"):
                inTree.Project("SumWeightsTemp", "1.0", "genWeight")
                sow = ROOT.gROOT.FindObject("SumWeightsTemp").Integral()
                self.hsumofweights.SetBinContent(1, sow)
            # pre-skimming
            elist, jsonFilter = preSkim(inTree, self.json, self.cut)
            if self.justcount:
                print 'Would select %d entries from %s' % (
                    elist.GetN() if elist else inTree.GetEntries(), fname)
                continue
            else:
                print 'Pre-select %d entries out of %s ' % (
                    elist.GetN() if elist else inTree.GetEntries(),
                    inTree.GetEntries())

            if fullClone:
                # no need of a reader (no event loop), but set up the elist if available
                if elist: inTree.SetEntryList(elist)
            else:
                # initialize reader
                inTree = InputTree(inTree, elist)

            # prepare output file
            if not self.noOut:
                outFileName = os.path.join(
                    self.outputDir,
                    os.path.basename(fname).replace(".root",
                                                    outpostfix + ".root"))
                outFile = ROOT.TFile.Open(outFileName, "RECREATE", "",
                                          compressionLevel)
                outFileNames.append(outFileName)
                if compressionLevel:
                    outFile.SetCompressionAlgorithm(compressionAlgo)
                # prepare output tree
                if self.friend:
                    outTree = FriendOutput(inFile, inTree, outFile)
                else:
                    outTree = FullOutput(
                        inFile,
                        inTree,
                        outFile,
                        branchSelection=self.branchsel,
                        outputbranchSelection=self.outputbranchsel,
                        fullClone=fullClone,
                        jsonFilter=jsonFilter,
                        provenance=self.provenance)
            else:
                outFile = None
                outTree = None

            # process events, if needed
            if not fullClone:
                (nall, npass, timeLoop) = eventLoop(self.modules, inFile,
                                                    outFile, inTree, outTree)
                print 'Processed %d preselected entries from %s (%s entries). Finally selected %d entries' % (
                    nall, fname, inTree.GetEntries(), npass)
            else:
                nall = inTree.GetEntries()
                print 'Selected %d entries from %s' % (
                    outTree.tree().GetEntries(), fname)

            # now write the output
            if not self.noOut:
                print "Start writing"
                self.hcount.Write()
                print "Start writing"
                if self.SMSMasses != None: self.hsmscount.Write()
                print "Start writing"
                if self.doISR != None: self.hISR.Write()
                print "Start writing"
                self.hsumofweights.Write()
                outTree.write()
                outFile.Close()
                print "Done %s" % outFileName
            if self.jobReport:
                self.jobReport.addInputFile(fname, nall)

        for m in self.modules:
            m.endJob()

        print totEntriesRead / (time.clock() - t0), "Hz"

        if self.haddFileName:
            os.system(
                "./haddnano.py %s %s" %
                (self.haddFileName, " ".join(outFileNames))
            )  #FIXME: remove "./" once haddnano.py is distributed with cms releases
        if self.jobReport:
            self.jobReport.addOutputFile(self.haddFileName)
            self.jobReport.save()
예제 #5
0
    def run(self):
        outpostfix = self.postfix if self.postfix is not None else (
            "_Friend" if self.friend else "_Skim")
        if self.allowNoPostfix and self.postfix is None:
            outpostfix = ""
        if not self.noOut:

            if self.compression != "none":
                ROOT.gInterpreter.ProcessLine("#include <Compression.h>")
                (algo, level) = self.compression.split(":")
                compressionLevel = int(level)
                if algo == "LZMA":
                    compressionAlgo = ROOT.ROOT.kLZMA
                elif algo == "ZLIB":
                    compressionAlgo = ROOT.ROOT.kZLIB
                elif algo == "LZ4":
                    compressionAlgo = ROOT.ROOT.kLZ4
                else:
                    raise RuntimeError("Unsupported compression %s" % algo)
            else:
                compressionLevel = 0
            print("Will write selected trees to " + self.outputDir)
            if not self.justcount:
                if not os.path.exists(self.outputDir):
                    os.system("mkdir -p " + self.outputDir)
        else:
            compressionLevel = 0

        if self.noOut:
            if len(self.modules) == 0:
                raise RuntimeError(
                    "Running with --noout and no modules does nothing!")

        # Open histogram file, if desired
        if (self.histFileName is not None and self.histDirName is None) or (
                self.histFileName is None and self.histDirName is not None):
            raise RuntimeError(
                "Must specify both histogram file and histogram directory!")
        elif self.histFileName is not None and self.histDirName is None:
            self.histFile = ROOT.TFile.Open(self.histFileName, "RECREATE")
        else:
            self.histFile = None

        for m in self.modules:
            if hasattr(m, 'writeHistFile') and m.writeHistFile:
                m.beginJob(histFile=self.histFile,
                           histDirName=self.histDirName)
            else:
                m.beginJob()

        fullClone = (len(self.modules) == 0)
        outFileNames = []
        t0 = time.time()
        totEntriesRead = 0
        for fname in self.inputFiles:
            ffnames = []
            if "," in fname:
                fnames = fname.split(',')
                fname, ffnames = fnames[0], fnames[1:]

            # open input file
            if self.prefetch:
                ftoread, toBeDeleted = self.prefetchFile(fname)
                inFile = ROOT.TFile.Open(ftoread)
            else:
                inFile = ROOT.TFile.Open(fname)

            if not inFile:
                print 'ERROR: file does not exist, check!'
                print '    filename:', fname
                exit(0)

            # get input tree
            inTree = inFile.Get("Events")
            if inTree is None:
                inTree = inFile.Get("Friends")
            nEntries = min(inTree.GetEntries() - self.firstEntry,
                           self.maxEntries)
            # first check that the histogram with weights is not already in the file
            hasWeightHistograms = False
            if inFile.GetListOfKeys().Contains(
                    "hGenWeights") and inFile.GetListOfKeys().Contains(
                        "hNumWeights"):
                hasWeightHistograms = True
                print "Histogram hGenWeights already exists, I will just copy it without recreating it"
            if self.saveHistoGenWeights and inTree.GetName(
            ) == "Events" and not hasWeightHistograms:
                print "Histogram hGenWeights does not exist yet, I will create it"
                # check that the tree contains all the original events, otherwise the sum of gen weights will miss some
                tmpTreeRuns = inFile.Get("Runs")
                for ievt, event in enumerate(tmpTreeRuns):
                    if ievt:
                        break  # only need first event (but there should be only 1 here)
                    nGenEvents = event.genEventCount
                if nGenEvents != inTree.GetEntries():
                    raise RuntimeError(
                        "I am creating the histogram with genWeight, but tree Events has less entries than genEventCount in tree Runs (%s instead of %s). The sum of weights will thus be wrong, please check"
                        % (str(inTree.GetEntries()), str(nGenEvents)))

                # saving distribution of genWeight for offline usage
                # idea is to fill the distribution of Log10(genWeight) with the sign, so to have a histogram from about -10 to 10
                # with about 10k bins (genWeights can take valus spanning several orders of magnitude, especially for fancy weights)
                # then one can compute the sum of genWeight in a given range using its integral (using Log10(threshold) ).
                # This somehow relies on having always |genWeight|>1, should it be < 1 the Log would change the sign.
                # So for the purpose of choosing the bin to be filled, we use |value| or 1.001, whatever is larger (this will not affect the integral)
                # then, need a second histogram to keep the integer number of events in each bin, so to allow for clipping of large weights
                hGenWeights = ROOT.TH1D("hGenWeights",
                                        "distribution of Log10(genWeight)",
                                        4800, -12.0, 12.0)
                hNumWeights = ROOT.TH1D(
                    "hNumWeights",
                    "distribution of Log10(genWeight) (unweighted)", 4800,
                    -12.0, 12.0)
                drawResult = inTree.Draw(
                    "TMath::Sign(1.0,genWeight)*TMath::Log10(max(1.001,abs(genWeight)))>>hGenWeights",
                    "genWeight", "goff", nEntries, self.firstEntry)
                drawResult = inTree.Draw(
                    "TMath::Sign(1.0,genWeight)*TMath::Log10(max(1.001,abs(genWeight)))>>hNumWeights",
                    "1", "goff", nEntries, self.firstEntry)
            totEntriesRead += nEntries
            # pre-skimming
            elist, jsonFilter = preSkim(inTree,
                                        self.json,
                                        self.cut,
                                        maxEntries=self.maxEntries,
                                        firstEntry=self.firstEntry)
            if self.justcount:
                print('Would select %d / %d entries from %s (%.2f%%)' %
                      (elist.GetN() if elist else nEntries, nEntries, fname,
                       (elist.GetN() if elist else nEntries) /
                       (0.01 * nEntries) if nEntries else 0))
                if self.prefetch:
                    if toBeDeleted:
                        os.unlink(ftoread)
                continue
            else:
                print('Pre-select %d entries out of %s (%.2f%%)' %
                      (elist.GetN() if elist else nEntries, nEntries,
                       (elist.GetN() if elist else nEntries) /
                       (0.01 * nEntries) if nEntries else 0))
                inAddFiles = []
                inAddTrees = []
            for ffname in ffnames:
                inAddFiles.append(ROOT.TFile.Open(ffname))
                inAddTree = inAddFiles[-1].Get("Events")
                if inAddTree is None:
                    inAddTree = inAddFiles[-1].Get("Friends")
                inAddTrees.append(inAddTree)
                inTree.AddFriend(inAddTree)

            if fullClone:
                # no need of a reader (no event loop), but set up the elist if available
                if elist:
                    inTree.SetEntryList(elist)
            else:
                # initialize reader
                inTree = InputTree(inTree, elist)

            # prepare output file
            if not self.noOut:
                outFileName = os.path.join(
                    self.outputDir,
                    os.path.basename(fname).replace(".root",
                                                    outpostfix + ".root"))
                outFile = ROOT.TFile.Open(outFileName, "RECREATE", "",
                                          compressionLevel)
                outFileNames.append(outFileName)
                if compressionLevel:
                    outFile.SetCompressionAlgorithm(compressionAlgo)
                # prepare output tree
                if self.friend:
                    outTree = FriendOutput(inFile, inTree, outFile)
                else:
                    outTree = FullOutput(
                        inFile,
                        inTree,
                        outFile,
                        branchSelection=self.branchsel,
                        outputbranchSelection=self.outputbranchsel,
                        fullClone=fullClone,
                        maxEntries=self.maxEntries,
                        firstEntry=self.firstEntry,
                        jsonFilter=jsonFilter,
                        provenance=self.provenance)
            else:
                outFile = None
                outTree = None
                if self.branchsel:
                    self.branchsel.selectBranches(inTree)

            # process events, if needed
            if not fullClone:
                eventRange = range(
                    self.firstEntry, self.firstEntry +
                    nEntries) if nEntries > 0 and not elist else None
                (nall, npass, timeLoop) = eventLoop(self.modules,
                                                    inFile,
                                                    outFile,
                                                    inTree,
                                                    outTree,
                                                    eventRange=eventRange,
                                                    maxEvents=self.maxEntries)
                print(
                    'Processed %d preselected entries from %s (%s entries). Finally selected %d entries'
                    % (nall, fname, nEntries, npass))
            else:
                nall = nEntries
                print('Selected %d / %d entries from %s (%.2f%%)' %
                      (outTree.tree().GetEntries(), nall, fname,
                       outTree.tree().GetEntries() /
                       (0.01 * nall) if nall else 0))

            # now write the output
            if not self.noOut:
                outTree.write()
                if not hasWeightHistograms:
                    if self.saveHistoGenWeights:
                        hGenWeights.Write(hGenWeights.GetName())
                        hNumWeights.Write(hNumWeights.GetName())
                outFile.Close()
                print("Done %s" % outFileName)
            if self.jobReport:
                self.jobReport.addInputFile(fname, nall)
            if self.prefetch:
                if toBeDeleted:
                    os.unlink(ftoread)

        for m in self.modules:
            m.endJob()

        print("Total time %.1f sec. to process %i events. Rate = %.1f Hz." %
              ((time.time() - t0), totEntriesRead, totEntriesRead /
               (time.time() - t0)))

        if self.haddFileName:
            haddnano = "./haddnano.py" if os.path.isfile(
                "./haddnano.py") else "haddnano.py"
            os.system("%s %s %s" %
                      (haddnano, self.haddFileName, " ".join(outFileNames)))
        if self.jobReport:
            self.jobReport.addOutputFile(self.haddFileName)
            self.jobReport.save()
예제 #6
0
        # prepare output file
        outFileName = os.path.join(
            outdir,
            os.path.basename(fname).replace(".root", outpostfix + ".root"))
        outFile = ROOT.TFile.Open(outFileName, "RECREATE", "",
                                  compressionLevel)
        if compressionLevel: outFile.SetCompressionAlgorithm(compressionAlgo)

        # prepare output tree
        if options.friend:
            outTree = FriendOutput(inFile, inTree, outFile)
        else:
            # FIXME process the other TTrees if there is a JSON
            outTree = FullOutput(inFile,
                                 inTree,
                                 outFile,
                                 branchSelection=branchsel,
                                 fullClone=fullClone)

        # process events, if needed
        if not fullClone:
            (nall, npass, time) = eventLoop(modules, inFile, outFile, inTree,
                                            outTree)
            print 'Processed %d entries from %s, selected %d entries' % (
                nall, fname, npass)
        else:
            print 'Selected %d entries from %s' % (outTree.tree().GetEntries(),
                                                   fname)

        # now write the output
        outTree.write()