KIT-CMS · conformist89 · May 13, 2022 · May 13, 2022 · May 13, 2022 · Jun 30, 2022
diff --git a/Dumbledraw/dumbledraw.py b/Dumbledraw/dumbledraw.py
diff --git a/Dumbledraw/labels.yaml b/Dumbledraw/labels.yaml
diff --git a/Dumbledraw/rootfile_parser.py b/Dumbledraw/rootfile_parser.py
@@ -4,23 +4,26 @@
 import logging
 import ROOT
 import copy
+
 logger = logging.getLogger(__name__)
 
 
 class Rootfile_parser(object):
-    def __init__(self, inputrootfilename, mode="CombineHarvester"):
+    def __init__(self, inputrootfilename, mode="CombineHarvester", prefit=False):
         self._rootfilename = inputrootfilename
         self._rootfile = ROOT.TFile(self._rootfilename, "READ")
         self._type = "control"
         content = [entry.GetName() for entry in self._rootfile.GetListOfKeys()]
         for entry in content:
             if entry.endswith("prefit"):
                 self._type = "prefit"
-        for entry in content:
-            if entry.endswith("postfit"):
-                self._type = "postfit"
-        logger.debug("Identified rootfile %s as %s shapes" %
-                     (inputrootfilename, self._type))
+        if not prefit:
+            for entry in content:
+                if entry.endswith("postfit"):
+                    self._type = "postfit"
+        logger.debug(
+            "Identified rootfile %s as %s shapes" % (inputrootfilename, self._type)
+        )
         if mode == "standard":
             self._hist_hash = "{channel}_{category}{plottype}/{process}{unc}"
         elif mode == "CombineHarvester":
@@ -36,41 +39,49 @@ def rootfile(self):
 
     def get(self, era, channel, category, process, syst=None):
         if syst != None and self._type != "control":
-            logger.fatal(
-                "Uncertainty shapes are only available in control plots!")
+            logger.fatal("Uncertainty shapes are only available in control plots!")
             raise Exception
         hist_hash = self._hist_hash.format(
             era=era,
             channel=channel,
             category=category,
             process=process,
             plottype="{plottype}",
-            unc="{unc}")
+            unc="{unc}",
+        )
         if self._type == "control":
             syst = "" if syst == None else "_" + syst
             hist_hash = hist_hash.format(plottype="", unc=syst)
         else:
             hist_hash = hist_hash.format(plottype="_" + self._type, unc="")
-        logger.debug(
-            "Try to access %s in %s" % (hist_hash, self._rootfilename))
+        logger.debug("Try to access %s in %s" % (hist_hash, self._rootfilename))
         # perform check if file is available and otherwise return some dummy TH1F
-        available_processes = [entry.GetName() for entry in self._rootfile.Get(hist_hash.split('/')[0]).GetListOfKeys()]
-        if hist_hash.split('/')[1] in available_processes:
+        available_processes = [
+            entry.GetName()
+            for entry in self._rootfile.Get(hist_hash.split("/")[0]).GetListOfKeys()
+        ]
+        if hist_hash.split("/")[1] in available_processes:
             return self._rootfile.Get(hist_hash)
         elif len(available_processes) != 0:
-            logger.warning("%s in %s does not exist !" % (hist_hash, self._rootfilename))
+            logger.warning(
+                "%s in %s does not exist !" % (hist_hash, self._rootfilename)
+            )
             logger.debug(" Available Histograms are: %s" % available_processes)
             logger.debug(" Returning a dummy histogram ")
-            dummy = self._rootfile.Get('{}/{}'.format(hist_hash.split('/')[0],available_processes[0]))
+            dummy = self._rootfile.Get(
+                "{}/{}".format(hist_hash.split("/")[0], available_processes[0])
+            )
             dummy.Reset()
             dummy.SetTitle(process)
             dummy.SetName(hist_hash)
             return dummy
         else:
-            logger.fatal(" None of the requested Histograms are available in %s. Aborting." % hist_hash.split('/')[0])
+            logger.fatal(
+                " None of the requested Histograms are available in %s. Aborting."
+                % hist_hash.split("/")[0]
+            )
             raise Exception
 
-
     def get_bins(self, era, channel, category, process, syst=None):
         hist = self.get(era, channel, category, process, syst)
         nbins = hist.GetNbinsX()

diff --git a/Dumbledraw/rootfile_parser_inputshapes.py b/Dumbledraw/rootfile_parser_inputshapes.py
@@ -4,6 +4,7 @@
 import logging
 import ROOT
 import copy
+
 logger = logging.getLogger(__name__)
 
 
@@ -29,10 +30,10 @@ def get(self, channel, category, process):
             analysis=self._analysis,
             epoch=self._epoch,
             variable=self._variable,
-            mass=self._mass)
-        logger.debug("Try to access %s in %s" % (hist_hash,
-                                                 self._rootfilename))
-        print "rootfile: " , self._rootfile.Get(hist_hash), " hash: ", hist_hash
+            mass=self._mass,
+        )
+        logger.debug("Try to access %s in %s" % (hist_hash, self._rootfilename))
+        print("rootfile: ", self._rootfile.Get(hist_hash), " hash: ", hist_hash)
 
         return self._rootfile.Get(hist_hash)
 

diff --git a/Dumbledraw/rootfile_parser_ntuple_processor_inputshapes.py b/Dumbledraw/rootfile_parser_ntuple_processor_inputshapes.py
@@ -0,0 +1,213 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+from __future__ import print_function
+
+import logging
+import ROOT
+import copy
+
+logger = logging.getLogger(__name__)
+
+
+class Rootfile_parser(object):
+
+    _dataset_map = {
+        "data": "data",
+        "ZTT": "DY",
+        "ZL": "DY",
+        "ZJ": "DY",
+        "TTT": "TT",
+        "TTL": "TT",
+        "TTJ": "TT",
+        "VVT": "VV",
+        "VVL": "VV",
+        "VVJ": "VV",
+        "W": "W",
+        "EMB": "EMB",
+        "QCDEMB": "QCD",
+        "QCD": "QCDMC",
+        "jetFakesEMB": "jetFakes",
+        "jetFakes": "jetFakesMC",
+        "ggH125": "ggH",
+        "qqH125": "qqH",
+        "wFakes": "wFakes",
+        "embminus2p5": "embminus2p5",
+        "embminus2p4": "embminus2p4",
+        "embminus2p3": "embminus2p3",
+        "embminus2p2": "embminus2p2",
+        "embminus2p1": "embminus2p1",
+        "embminus2p0": "embminus2p0",
+        "embminus1p9": "embminus1p9",
+        "embminus1p8": "embminus1p8",
+        "embminus1p7": "embminus1p7",
+        "embminus1p6": "embminus1p6",
+        "embminus1p5": "embminus1p5",
+        "embminus1p4": "embminus1p4",
+        "embminus1p3": "embminus1p3",
+        "embminus1p2": "embminus1p2",
+        "embminus1p1": "embminus1p1",
+        "embminus1p0": "embminus1p0",
+        "embminus0p9": "embminus0p9",
+        "embminus0p8": "embminus0p8",
+        "embminus0p7": "embminus0p7",
+        "embminus0p6": "embminus0p6",
+        "embminus0p5": "embminus0p5",
+        "embminus0p4": "embminus0p4",
+        "embminus0p3": "embminus0p3",
+        "embminus0p2": "embminus0p2",
+        "embminus0p1": "embminus0p1",
+        "emb0p0": "emb0p0",
+        "emb0p1": "emb0p1",
+        "emb0p2": "emb0p2",
+        "emb0p3": "emb0p3",
+        "emb0p4": "emb0p4",
+        "emb0p5": "emb0p5",
+        "emb0p6": "emb0p6",
+        "emb0p7": "emb0p7",
+        "emb0p8": "emb0p8",
+        "emb0p9": "emb0p9",
+        "emb1p0": "emb1p0",
+        "emb1p1": "emb1p1",
+        "emb1p2": "emb1p2",
+        "emb1p3": "emb1p3",
+        "emb1p4": "emb1p4",
+        "emb1p5": "emb1p5",
+        "emb1p6": "emb1p6",
+        "emb1p7": "emb1p7",
+        "emb1p8": "emb1p8",
+        "emb1p9": "emb1p9",
+        "emb2p0": "emb2p0",
+        "emb2p1": "emb2p1",
+        "emb2p2": "emb2p2",
+        "emb2p3": "emb2p3",
+        "emb2p4": "emb2p4",
+        "emb2p5": "emb2p5",
+    }
+
+    _process_map = {
+        "data": "data",
+        "ZTT": "DY-ZTT",
+        "ZL": "DY-ZL",
+        "ZJ": "DY-ZJ",
+        "TTT": "TT-TTT",
+        "TTL": "TT-TTL",
+        "TTJ": "TT-TTJ",
+        "VVT": "VV-VVT",
+        "VVL": "VV-VVL",
+        "VVJ": "VV-VVJ",
+        "W": "W",
+        "EMB": "Embedded",
+        "QCDEMB": "QCD",
+        "QCD": "QCDMC",
+        "jetFakesEMB": "jetFakes",
+        "jetFakes": "jetFakesMC",
+        "ggH125": "ggH125",
+        "qqH125": "qqH125",
+        "wFakes": "wFakes",
+        "embminus2p5": "Embedded",
+        "embminus2p4": "Embedded",
+        "embminus2p3": "Embedded",
+        "embminus2p2": "Embedded",
+        "embminus2p1": "Embedded",
+        "embminus2p0": "Embedded",
+        "embminus1p9": "Embedded",
+        "embminus1p8": "Embedded",
+        "embminus1p7": "Embedded",
+        "embminus1p6": "Embedded",
+        "embminus1p5": "Embedded",
+        "embminus1p4": "Embedded",
+        "embminus1p3": "Embedded",
+        "embminus1p2": "Embedded",
+        "embminus1p1": "Embedded",
+        "embminus1p0": "Embedded",
+        "embminus0p9": "Embedded",
+        "embminus0p8": "Embedded",
+        "embminus0p7": "Embedded",
+        "embminus0p6": "Embedded",
+        "embminus0p5": "Embedded",
+        "embminus0p4": "Embedded",
+        "embminus0p3": "Embedded",
+        "embminus0p2": "Embedded",
+        "embminus0p1": "Embedded",
+        "emb0p0": "Embedded",
+        "emb0p1": "Embedded",
+        "emb0p2": "Embedded",
+        "emb0p3": "Embedded",
+        "emb0p4": "Embedded",
+        "emb0p5": "Embedded",
+        "emb0p6": "Embedded",
+        "emb0p7": "Embedded",
+        "emb0p8": "Embedded",
+        "emb0p9": "Embedded",
+        "emb1p0": "Embedded",
+        "emb1p1": "Embedded",
+        "emb1p2": "Embedded",
+        "emb1p3": "Embedded",
+        "emb1p4": "Embedded",
+        "emb1p5": "Embedded",
+        "emb1p6": "Embedded",
+        "emb1p7": "Embedded",
+        "emb1p8": "Embedded",
+        "emb1p9": "Embedded",
+        "emb2p0": "Embedded",
+        "emb2p1": "Embedded",
+        "emb2p2": "Embedded",
+        "emb2p3": "Embedded",
+        "emb2p4": "Embedded",
+        "emb2p5": "Embedded",
+    }
+
+    def __init__(self, inputrootfilename, variable):
+        self._rootfilename = inputrootfilename
+        self._rootfile = ROOT.TFile(self._rootfilename, "READ")
+        self._variable = variable
+
+    @property
+    def rootfile(self):
+        return self._rootfile
+
+    def get(self, channel, process, category=None, shape_type="Nominal"):
+        dataset = self._dataset_map[process]
+        if category is None:
+            category = "" if "data" in process else "-" + self._process_map[process]
+        else:
+            category = (
+                "-" + category
+                if "data" in process
+                else "-" + "-".join([self._process_map[process], category])
+            )
+        hist_hash = "{dataset}#{channel}{category}#{shape_type}#{variable}".format(
+            dataset=dataset,
+            channel=channel,
+            category=category,
+            shape_type=shape_type,
+            variable=self._variable,
+        )
+        logger.debug("Try to access %s in %s" % (hist_hash, self._rootfilename))
+        print("rootfile: ", self._rootfile.Get(hist_hash), " hash: ", hist_hash)
+
+        return self._rootfile.Get(hist_hash)
+
+    def list_contents(self):
+        return [key.GetTitle() for key in self._rootfile.GetListOfKeys()]
+
+    def get_bins(self, channel, category):
+        hist = self.get(channel, category)
+        nbins = hist.GetNbinsX()
+        bins = []
+        for i in range(nbins):
+            bins.append(hist.GetBinLowEdge(i + 1))
+        bins.append(hist.GetBinLowEdge(i + 1) + hist.GetBinWidth(i + 1))
+        return bins
+
+    def get_values(self, channel, category):
+        hist = self.get(channel, category)
+        nbins = hist.GetNbinsX()
+        values = []
+        for i in range(nbins):
+            values.append(hist.GetBinContent(i + 1))
+        return values
+
+    def __del__(self):
+        logger.debug("Closing rootfile %s" % (self._rootfilename))
+        self._rootfile.Close()
diff --git a/Dumbledraw/sf_rootfile_parser.py b/Dumbledraw/sf_rootfile_parser.py
@@ -4,6 +4,7 @@
 import logging
 import ROOT
 import copy
+
 logger = logging.getLogger(__name__)
 
 
@@ -13,24 +14,22 @@ def __init__(self, inputrootfilename):
         self._rootfile = ROOT.TFile(self._rootfilename, "READ")
         content = [entry.GetName() for entry in self._rootfile.GetListOfKeys()]
         self.Nbins = len(content)
-        logger.debug("Identified {} histograms in rootfile {} ".format(len(content), inputrootfilename))
+        logger.debug(
+            "Identified {} histograms in rootfile {} ".format(
+                len(content), inputrootfilename
+            )
+        )
         self._hist_hash = "{variable}_projx_{etabin}"
 
     @property
     def rootfile(self):
         return self._rootfile
 
-
-
     def get(self, variable, etabin):
-        hist_hash = self._hist_hash.format(
-            variable=variable,
-            etabin=etabin)
-        logger.debug(
-            "Try to access %s in %s" % (hist_hash, self._rootfilename))
+        hist_hash = self._hist_hash.format(variable=variable, etabin=etabin)
+        logger.debug("Try to access %s in %s" % (hist_hash, self._rootfilename))
         return self._rootfile.Get(hist_hash)
 
-
     def get_bins(self, variable, etabin):
         hist = self.get(self, variable, etabin)
         nbins = hist.GetNbinsX()
@@ -50,8 +49,8 @@ def get_values(self, variable, etabin):
 
     def get_values_up(self, variable, etabin):
         hist = self.get(self, variable, etabin)
-        nbins=hist.GetNbinsX()
-        values=[]
+        nbins = hist.GetNbinsX()
+        values = []
         for i in range(nbins):
             values.append(hist.GetBinErrDown(i + 1))
         return values