Egamma/test/egammaHltPrintRelvalEfficiencies.py

0001 #!/usr/bin/env python
0002
0003 from __future__ import print_function
0004 import sys, os
0005
0006 try:
0007     import ROOT
0008 except ImportError:
0009     print(file=sys.stderr)
0010     print("  Error importing the ROOT python module", file=sys.stderr)
0011     print("  Try e.g. initializing a CMSSW environment", file=sys.stderr)
0012     print("  prior to starting this script", file=sys.stderr)
0013     print(file=sys.stderr)
0014     sys.exit(1)
0015
0016 #----------------------------------------------------------------------
0017
0018 def splitAtCapitalization(text):
0019     """ splits a string before capital letters. Useful to make
0020     identifiers which consist of capitalized words easier to read
0021
0022     We should actually find a smarter algorithm in order to avoid
0023     splitting things like HLT or LW.
0024
0025     """
0026
0027     retval = ''
0028
0029     for ch in text:
0030         if ch.isupper() and len(retval) > 0:
0031             retval += ' '
0032
0033         retval += ch
0034
0035     return retval
0036
0037 #----------------------------------------------------------------------
0038 ## @param fin is the ROOT input file (the TFile, not the file name)
0039 #
0040 def findTopDir(fin):
0041     """tries to find a top directory for the DQM histograms. Note
0042     that the run number seems to be always 1 for MC but differs
0043     for data. If there is more than one top directory, this function
0044     prints an error message on stderr and exits (maybe this should
0045     be made more flexible in the future in order to allow DQM histogramming
0046     of data of multiple runs).
0047
0048     Returns None if no full path could be found.
0049
0050     """
0051
0052     import re
0053
0054     # an path looks like:
0055     # "DQMData/Run <run>/HLT/Run summary/HLTEgammaValidation"
0056
0057     theDir = fin.Get("DQMData")
0058
0059     if theDir == None:
0060         return None
0061
0062     # now look for directories of the form 'Run %d'
0063
0064     runSubdirName = None
0065
0066     for subdirName in [ x.GetName() for x in theDir.GetListOfKeys() ]:
0067
0068         if re.match("Run \d+$", subdirName):
0069             if runSubdirName != None:
0070                 # more than one run found
0071                 print("more than one run found in the DQM file, this is currently not supported", file=sys.stderr)
0072                 sys.exit(1)
0073
0074             runSubdirName = subdirName
0075
0076
0077     # check that we have at least (exactly) one directory
0078     if runSubdirName == None:
0079         return None
0080
0081     # get the rest
0082     return theDir.Get(runSubdirName + "/HLT/Run summary/HLTEgammaValidation")
0083
0084
0085 #----------------------------------------------------------------------
0086 # main
0087 #----------------------------------------------------------------------
0088 from optparse import OptionParser
0089
0090 parser = OptionParser("""
0091
0092   usage: %prog [options] root_file
0093
0094     given the output of the E/gamma HLT validation histogramming module,
0095     (DQM output) prints some information about path and module efficiencies.
0096
0097     Useful for determining which paths actually have some meaningful
0098     results in the file and which ones not.
0099 """)
0100
0101 parser.add_option("--summary",
0102                   dest="summary_mode",
0103                   default = False,
0104                   action="store_true",
0105                   help="print path efficiencies only, nothing about modules",
0106                   )
0107
0108 parser.add_option("--path",
0109                   dest="selected_paths",
0110                   default = [],
0111                   action="append",
0112                   help="restrict printout to specific path. "+
0113                        "This option can be given more than once to select several paths.",
0114                   )
0115
0116 parser.add_option("--ignore-empty",
0117                   dest="ignore_empty_paths",
0118                   action='store_true',
0119                   default = False,
0120                   help="Print only information about non-empty paths (i.e. those with at least one entry in the total_eff histogram).",
0121                   )
0122
0123 parser.add_option("--ignore-zero-eff",
0124                   dest="ignore_zero_efficiency",
0125                   action='store_true',
0126                   default = False,
0127                   help="Print only information about paths which have at least one entry in the bin of the last module in the overview histogram. Note that this also excludes those paths excluded by --ignore-empty .",
0128                   )
0129
0130 parser.add_option("--no-split-names",
0131                   dest="split_names",
0132                   action='store_false',
0133                   default = True,
0134                   help="Do not split module names.",
0135                   )
0136
0137
0138 (options, ARGV) = parser.parse_args()
0139
0140 if len(ARGV) != 1:
0141     parser.print_help()
0142     sys.exit(1)
0143
0144 #----------------------------------------
0145 # open the ROOT file
0146 #----------------------------------------
0147
0148 fin = ROOT.TFile.Open(ARGV[0])
0149
0150 top_dir = findTopDir(fin)
0151
0152 if top_dir == None:
0153     print("could not find a top directory inside root file", file=sys.stderr)
0154     print("A typical top directory for MC is 'DQMData/Run 1/HLT/Run summary/HLTEgammaValidation'", file=sys.stderr)
0155     print(file=sys.stderr)
0156     print("Exiting", file=sys.stderr)
0157     sys.exit(1)
0158
0159
0160 #--------------------
0161 # determine the length of the longest path name (for nice printout)
0162 #--------------------
0163
0164 maxPathNameLen = 100
0165 allPathNames = []
0166
0167 for path_key in top_dir.GetListOfKeys():
0168     pathName = path_key.GetName()
0169
0170     # just select directories (there are also other
0171     # objects in the top directory)
0172     path_dir = top_dir.Get(pathName)
0173     if not isinstance(path_dir,ROOT.TDirectoryFile):
0174         continue
0175
0176     if len(options.selected_paths) != 0 and not pathName in options.selected_paths:
0177         continue
0178
0179     # further checks which are done in the next
0180     # loop are not repeated here.
0181     # so we might get a maximum number of characters
0182     # which is slightly too high (but the code here
0183     # is more readable)
0184
0185     allPathNames.append(pathName)
0186
0187     maxPathNameLen = max(maxPathNameLen, len(pathName))
0188
0189 #--------------------
0190
0191 for path_name in allPathNames:
0192
0193     path_dir = top_dir.Get(path_name)
0194
0195     # just select directories (there are also other
0196     # objects in the top directory)
0197     if not isinstance(path_dir,ROOT.TDirectoryFile):
0198         continue
0199
0200     # find modules in order from total_eff_MC_matched histogram
0201     total_eff_histo = path_dir.Get("total_eff_MC_matched")
0202
0203     if total_eff_histo == None:
0204         # try with data:
0205         total_eff_histo = path_dir.Get("total_eff_RECO_matched")
0206
0207     # subtract 2 for 'Total' and 'Gen' bins
0208     num_modules = total_eff_histo.GetNbinsX() - 2
0209
0210     total = total_eff_histo.GetBinContent(num_modules)
0211     num_gen_events = total_eff_histo.GetBinContent(num_modules + 2)
0212
0213     if num_gen_events == 0 and options.ignore_empty_paths:
0214         continue
0215
0216     # check whether at least one event passed all modules
0217     if options.ignore_zero_efficiency:
0218         # get number of entries in last module
0219
0220         last_module_index = num_modules - 1
0221
0222         last_module_accepted_events = total_eff_histo.GetBinContent(last_module_index+1)
0223
0224         if last_module_accepted_events < 1:
0225             continue
0226
0227
0228     #--------------------
0229
0230     if not options.summary_mode:
0231         print("----------------------------------------")
0232
0233     print(("PATH: %-" + str(maxPathNameLen) + "s") % path_name, end=' ')
0234
0235     if num_gen_events > 0:
0236         print("(%5.1f%% eff.)" % (100 * total / float(num_gen_events)), end=' ')
0237
0238     elif options.summary_mode:
0239         print("(no entries)", end=' ')
0240
0241     print()
0242
0243     if not options.summary_mode:
0244         print("----------------------------------------")
0245
0246         print("  %-80s: %5d events" % ('generated', num_gen_events))
0247
0248     if options.summary_mode:
0249         continue
0250
0251     previous_module_output = num_gen_events
0252
0253     print()
0254
0255     for i in range(num_modules):
0256
0257         module_name = total_eff_histo.GetXaxis().GetBinLabel(i+1)
0258
0259         if options.split_names:
0260             module_name = splitAtCapitalization(module_name)
0261
0262         events = total_eff_histo.GetBinContent(i+1)
0263
0264
0265
0266
0267         print("  %-90s: %5d events" % (module_name, events), end=' ')
0268
0269         if previous_module_output > 0:
0270             eff = 100 * events / float(previous_module_output)
0271             print("(%5.1f%% eff.)" % (eff), end=' ')
0272             if eff > 100.:
0273                 if module_name.find("Unseeded") >= 0:
0274                     print(">100% Unseeded Filter", end=' ')
0275                 else:
0276                     print("ERROR", end=' ')
0277
0278         print()
0279
0280
0281         previous_module_output = events
0282
0283
0284     print()
0285
0286
0287
0288
0289