RelMon/scripts/compare_using_files.py

0001 #! /usr/bin/env python3
0002 ################################################################################
0003 # https://twiki.cern.ch/twiki/bin/view/CMSPublic/RelMon
0004 #
0005 #
0006 #
0007 # Danilo Piparo CERN - danilo.piparo@cern.ch
0008 #
0009 ################################################################################
0010
0011
0012 from __future__ import print_function
0013 def getInfoFromFilename(filename):
0014   prefix,sample,cmssw_release,tier = filename[:-5].split("__")[:5]
0015   run=int(prefix.split("_")[-1][1:])
0016   return run,sample,cmssw_release,tier
0017
0018 from sys import argv,exit
0019 import os
0020
0021 # Default Configuration Parameters ---------------------------------------------
0022
0023 stat_test="Chi2"
0024 test_threshold=1e-5
0025
0026
0027 #run="1"
0028
0029 dir_name=""
0030 outdir_name=""
0031
0032 compare=False
0033 report=False
0034
0035 do_pngs=False
0036
0037 black_list_str=""
0038
0039 #-------------------------------------------------------------------------------
0040
0041 from optparse import OptionParser
0042
0043 parser = OptionParser(usage="usage: %prog file1 file2 [options]")
0044
0045
0046 #parser.add_option("-r","--run ",
0047                   #action="store",
0048                   #dest="run",
0049                   #default=run,
0050                   #help="The run to be checked \n(default is %s)" %run)
0051
0052 parser.add_option("-d","--dir_name",
0053                   action="store",
0054                   dest="dir_name",
0055                   default=dir_name,
0056                   help="The 'directory' to be checked in the DQM \n(default is %s)" %dir_name)
0057
0058 parser.add_option("-o","--outdir_name",
0059                   action="store",
0060                   dest="outdir_name",
0061                   default=outdir_name,
0062                   help="The directory where the output will be stored \n(default is %s)" %outdir_name)
0063
0064 parser.add_option("-p","--do_pngs",
0065                   action="store_true",
0066                   dest="do_pngs",
0067                   default=False,
0068                   help="Do the pngs of the comparison (takes 50%% of the total running time) \n(default is %s)" %False)
0069
0070 parser.add_option("--no_successes",
0071                   action="store_true",
0072                   dest="no_successes",
0073                   default=False,
0074                   help="Do not draw successes. Default is False.")
0075
0076 parser.add_option("-P","--pickle",
0077                   action="store",
0078                   dest="pklfile",
0079                   default="",
0080                   help="Pkl file of the dir structure ")
0081
0082 parser.add_option("--sample",
0083                   action="store",
0084                   dest="sample",
0085                   default="Sample",
0086                   help="The name of the sample to be displayed")
0087
0088 parser.add_option("--metas",
0089                   action="store",
0090                   dest="metas",
0091                   default="",
0092                   help="The Metas describing the two files (separated by @@@)")
0093
0094 parser.add_option("-t","--test_threshold",
0095                   action="store",
0096                   dest="test_threshold",
0097                   default=test_threshold,
0098                   help="Threshold for the statistical test \n(default is %s)" %test_threshold)
0099
0100 parser.add_option("-s","--stat_test",
0101                   action="store",
0102                   dest="stat_test",
0103                   default=stat_test,
0104                   help="Statistical test (KS or Chi2) \n(default is %s)" %stat_test)
0105
0106 parser.add_option("-C","--compare",
0107                   action="store_true",
0108                   dest="compare",
0109                   default=compare,
0110                   help="Make the comparison \n(default is %s)" %compare)
0111
0112 parser.add_option("-R","--Report",
0113                   action="store_true",
0114                   dest="report",
0115                   default=report,
0116                   help="Make the html report \n(default is %s)" %report)
0117
0118 parser.add_option("--specify_run",
0119                   action="store_true",
0120                   dest="specify_run",
0121                   default=False,
0122                   help="Append the run number to the output dir for data")
0123
0124
0125 parser.add_option("-B","--black_list",
0126                   action="store",
0127                   dest="black_list",
0128                   default=black_list_str,
0129                   help="Blacklist elements. form is name@hierarchy_level (i.e. HLT@1) \n(default is %s)" %black_list_str)
0130
0131 ##---HASHING---##
0132 parser.add_option("--hash_name",
0133                   action="store_true",
0134                   dest="hash_name",
0135                   default=False,
0136                   help="Set if you want to minimize & hash the output HTML files.")
0137 ##--Blacklist File --##
0138 parser.add_option("--use_black_file",
0139                   action="store_true",
0140                   dest="blacklist_file",
0141                   default=False,
0142                   help="Use a black list file of histograms located @ /RelMon/data")
0143 ##-- USE CSS files in web access, for stand-alone usage --##
0144 parser.add_option("--standalone",
0145                   action="store_true",
0146                   dest="standalone",
0147                   default=False,
0148                   help="Makes CSS files accessible over HTTP")
0149
0150 def blackListedHistos():
0151         ##GET a black-list file of histograms##
0152     if "RELMON_SA" in os.environ:
0153         black_list_file="../data/blacklist.txt"
0154     else:
0155         black_list_file="%s/src/Utilities/RelMon/data/blacklist.txt"%(os.environ["CMSSW_BASE"])
0156     bListFile = open(black_list_file,'r')
0157     black_listed_histograms = bListFile.read()
0158     bListFile.close()
0159
0160     histogramArray = black_listed_histograms.split("\n")
0161     histogramArray.remove("")  #remove the last element which is empty line
0162     newarray = []
0163     for elem in histogramArray:
0164         tmp = elem.split("/")  #screw windows as it is being run on lxbuild machines with Linux
0165         tmp.insert(1,"Run summary")  #insert "Run summary" dir in path as in ROOT files they exists but user haven't defined them
0166         newarray.append(("/").join(tmp))
0167     return newarray
0168     ##------##
0169
0170 (options, args) = parser.parse_args()
0171
0172 if len(args)!=2 and options.compare:
0173   print("Wrong number of RootFiles specified (%s)" %len(args))
0174   print(args)
0175
0176 #-------------------------------------------------------------------------------
0177 original_pickle_name=""
0178 if options.compare:
0179
0180   if "RELMON_SA" in os.environ:
0181     import definitions
0182     from dqm_interfaces import DirID,DirWalkerFile,string2blacklist
0183     from dirstructure import Directory
0184   else:
0185     import Utilities.RelMon.definitions as definitions
0186     from Utilities.RelMon.dqm_interfaces import DirID,DirWalkerFile,string2blacklist
0187     from Utilities.RelMon.dirstructure import Directory
0188
0189   import pickle
0190   from os import mkdir,chdir,getcwd
0191   from os.path import exists
0192
0193   #-------------------------------------------------------------------------------
0194   # Guess Releases and sample from filename
0195   rootfilename1,rootfilename2 = args
0196
0197   run1=-1
0198   sample1=''
0199   cmssw_release1=''
0200   tier1=''
0201   run2=-1
0202   sample2=''
0203   cmssw_release2=''
0204   tier2=''
0205
0206   if options.metas=='':
0207     run1,sample1,cmssw_release1,tier1= getInfoFromFilename(rootfilename1)
0208     run2,sample2,cmssw_release2,tier2= getInfoFromFilename(rootfilename2)
0209   else:
0210     print("Reading meta from commandline")
0211     sample1=sample2=options.sample
0212     cmssw_release1,cmssw_release2=options.metas.split('@@@')
0213     options.standalone = True
0214
0215   # check if the sample is the same
0216   if sample1!=sample2:
0217     print("I am puzzled. Did you choose two different samples?")
0218     #exit(1)
0219   sample = sample1
0220
0221   # check if the run is the same
0222   if run1!=run2:
0223     print("I am puzzled. Did you choose two different runs?")
0224 #    exit(1)
0225   run=run1
0226
0227   fulldirname=options.outdir_name
0228   if len(fulldirname)==0:
0229     fulldirname=options.dir_name
0230   if len(fulldirname)==0:
0231     fulldirname="%s_%s_%s" %(sample1,cmssw_release1,cmssw_release2)
0232
0233
0234   black_list=string2blacklist(options.black_list)
0235
0236   if options.blacklist_file:
0237     black_listed = blackListedHistos()
0238   else:
0239     black_listed = []
0240
0241 #-------------------------------------------------------------------------------
0242
0243   print("Analysing Histograms located in directory %s at: " %options.dir_name)
0244   for filename in rootfilename1,rootfilename2:
0245     print(" o %s" %filename)
0246
0247
0248   if len(black_list)>0:
0249     print("We have a Blacklist:")
0250     for dirid in black_list:
0251       print(" o %s" %dirid)
0252
0253   # Set up the fake directory structure
0254   directory=Directory(options.dir_name)
0255   dirwalker=DirWalkerFile(fulldirname,
0256                           options.dir_name,
0257                           rootfilename1,rootfilename2,
0258                           run,
0259                           black_list,
0260                           options.stat_test,
0261                           options.test_threshold,
0262                           not options.no_successes,
0263                           options.do_pngs,
0264                           set(black_listed)
0265                           )
0266
0267   # Start the walker
0268   outdir_name=options.outdir_name
0269   if run>1 and options.specify_run:
0270     outdir_name+="_%s" %run
0271     fulldirname+="_%s" %run
0272   print("+"*30)
0273   print("Output Directory will be ", outdir_name)
0274   options.outdir_name=outdir_name
0275   if not exists(outdir_name) and len(outdir_name )>0:
0276     mkdir(outdir_name)
0277   if len(outdir_name)>0:
0278     chdir(outdir_name)
0279   dirwalker.walk()
0280
0281   run =  dirwalker.run
0282
0283
0284   # Fetch the directory from the walker
0285   directory=dirwalker.directory
0286
0287   # Set some meta for the page generation
0288   directory.meta.sample1=sample1
0289   directory.meta.sample2=sample2
0290   directory.meta.run1=run1
0291   directory.meta.run2=run2
0292   directory.meta.release1=cmssw_release1
0293   directory.meta.release2=cmssw_release2
0294   directory.meta.tier1=tier1
0295   directory.meta.tier2=tier2
0296
0297   # Print a summary Report on screen
0298   directory.print_report(verbose=True)
0299
0300   # Remove this DQM FW reminescence.
0301   directory.prune("Run summary")
0302
0303   # Dump the directory structure on disk in a pickle
0304   original_pickle_name="%s.pkl" %fulldirname
0305   print("Pickleing the directory as %s in dir %s" %(original_pickle_name,getcwd()))
0306   output = open(original_pickle_name,"wb")
0307   pickle.dump(directory, output, -1)# use highest protocol available for the pickle
0308   output.close()
0309
0310 #-------------------------------------------------------------------------------
0311 if options.report:
0312
0313   if "RELMON_SA" in os.environ:
0314     from directories2html import directory2html
0315     from dirstructure import Directory
0316   else:
0317     from Utilities.RelMon.directories2html import directory2html
0318     from Utilities.RelMon.dirstructure import Directory
0319
0320   from os.path import exists
0321   from os import chdir,mkdir
0322   import os
0323   import pickle
0324
0325   pickle_name=options.pklfile
0326   if len(options.pklfile)==0:
0327     pickle_name=original_pickle_name
0328
0329   print("Reading directory from %s" %(pickle_name))
0330   ifile=open(pickle_name,"rb")
0331   directory=pickle.load(ifile)
0332   ifile.close()
0333
0334   if not options.compare:
0335     if not os.path.exists(options.outdir_name):
0336       mkdir(options.outdir_name)
0337
0338   if exists(options.outdir_name) and len(directory.name)==0:
0339     chdir(options.outdir_name)
0340
0341   # Calculate the results of the tests for each directory
0342   print("Calculating stats for the directory...")
0343   directory.calcStats()
0344
0345   print("Producing html...")
0346   directory2html(directory, options.hash_name, options.standalone)
0347
0348 if not (options.report or options.compare):
0349   print("Neither comparison nor report to be executed. A typo?")
0350
0351