compareHistograms Namespace Reference

Functions
def	compare (shared_paths, pr_flat_dict, base_flat_dict, paths_to_save_in_pr, paths_to_save_in_base)
def	compare_TProfile (pr_item, base_item)
def	compareMP (shared_paths, pr_flat_dict, base_flat_dict, iProc, return_dict)
def	create_dif (base_file_path, pr_file_path, pr_number, test_number, cmssw_version, num_processes, output_dir_path)
def	create_dir (parent_dir, name)
def	flatten_file (file, run_nr)
def	get_node_name (node)
def	get_output_filename (input_file_path, pr_number, test_number, cmssw_version, isPr)
def	get_run_nr (file_path)
def	get_string_suffix ()
def	is_blacklisted (dirs_list, run_nr)
def	save_paths (flat_dict, paths, result_file_path)
def	save_to_file (flat_dict, path, output_file)
def	traverse_till_end (node, dirs_list, result, run_nr)

Variables
	args
	cmssw_version
	default
	description
	help
	IgnoreCommandLineOptions
	parser
	required
	type

Function Documentation

compare()

def compareHistograms.compare	(		shared_paths,
			pr_flat_dict,
			base_flat_dict,
			paths_to_save_in_pr,
			paths_to_save_in_base
	)

Definition at line 162 of file compareHistograms.py.

References compare_TProfile().

def compare(shared_paths, pr_flat_dict, base_flat_dict, paths_to_save_in_pr, paths_to_save_in_base):
   # Collect paths that have to be written to both output files
   for path in shared_paths:
      pr_item = pr_flat_dict[path]
      base_item = base_flat_dict[path]

      if pr_item == None or base_item == None:
         continue

      are_different=False

      if pr_item.InheritsFrom('TProfile2D') and base_item.InheritsFrom('TProfile2D'):
         # Compare TProfile (content, entries and errors)
         are_different = not compare_TProfile(pr_item, base_item)

      elif pr_item.InheritsFrom('TProfile') and base_item.InheritsFrom('TProfile'):
         # Compare TProfile (content, entries and errors)
         are_different = not compare_TProfile(pr_item, base_item)

      elif pr_item.InheritsFrom('TH1') and base_item.InheritsFrom('TH1'):
         # Compare bin by bin
         pr_array = np.array(pr_item)
         base_array = np.array(base_item)

         if pr_array.shape != base_array.shape or not np.allclose(pr_array, base_array, equal_nan=True):
            are_different = True
      else:
         # Compare non histograms
         if pr_item != base_item:
            are_different = True

      if are_different:
         paths_to_save_in_pr.append(path)
         paths_to_save_in_base.append(path)

# Returns False if different, True otherwise

compare_TProfile()

def compareHistograms.compare_TProfile	(		pr_item,
			base_item
	)

Definition at line 198 of file compareHistograms.py.

References isotrackApplyRegressor.range.

Referenced by compare(), and compareMP().

def compare_TProfile(pr_item, base_item):
   if pr_item.GetSize() != base_item.GetSize():
      return False

   for i in range(pr_item.GetSize()):
      pr_bin_content = pr_item.GetBinContent(i)
      base_bin_content = base_item.GetBinContent(i)

      pr_bin_entries = pr_item.GetBinEntries(i)
      base_bin_entries = base_item.GetBinEntries(i)

      pr_bin_error = pr_item.GetBinError(i)
      base_bin_error = base_item.GetBinError(i)

      if not np.isclose(pr_bin_content, base_bin_content, equal_nan=True):
         return False

      if not np.isclose(pr_bin_entries, base_bin_entries, equal_nan=True):
         return False

      if not np.isclose(pr_bin_error, base_bin_error, equal_nan=True):
         return False

   return True

compareMP()

def compareHistograms.compareMP	(		shared_paths,
			pr_flat_dict,
			base_flat_dict,
			iProc,
			return_dict
	)

Definition at line 123 of file compareHistograms.py.

References mps_setup.append, and compare_TProfile().

def compareMP(shared_paths, pr_flat_dict, base_flat_dict, iProc, return_dict):
   # Prepare output dictionary
   comparisons = {'pr': [], 'base': []}

   # Collect paths that have to be written to both output files
   for path in shared_paths:
      pr_item = pr_flat_dict[path]
      base_item = base_flat_dict[path]

      if pr_item == None or base_item == None:
         continue

      are_different=False

      if pr_item.InheritsFrom('TProfile2D') and base_item.InheritsFrom('TProfile2D'):
         # Compare TProfile (content, entries and errors)
         are_different = not compare_TProfile(pr_item, base_item)

      elif pr_item.InheritsFrom('TProfile') and base_item.InheritsFrom('TProfile'):
         # Compare TProfile (content, entries and errors)
         are_different = not compare_TProfile(pr_item, base_item)

      elif pr_item.InheritsFrom('TH1') and base_item.InheritsFrom('TH1'):
         # Compare bin by bin
         pr_array = np.array(pr_item)
         base_array = np.array(base_item)

         if pr_array.shape != base_array.shape or not np.allclose(pr_array, base_array, equal_nan=True):
            are_different = True
      else:
         # Compare non histograms
         if pr_item != base_item:
            are_different = True

      if are_different:
         comparisons['pr'].append(path)
         comparisons['base'].append(path)
   return_dict[iProc] = comparisons

create_dif()

def compareHistograms.create_dif	(		base_file_path,
			pr_file_path,
			pr_number,
			test_number,
			cmssw_version,
			num_processes,
			output_dir_path
	)

Definition at line 13 of file compareHistograms.py.

References flatten_file(), get_output_filename(), get_run_nr(), reco::helper::VirtualJetProducerHelper.intersection(), join(), print(), isotrackApplyRegressor.range, and save_paths().

def create_dif(base_file_path, pr_file_path, pr_number, test_number, cmssw_version, num_processes, output_dir_path):
   base_file = ROOT.TFile(base_file_path, 'read')
   ROOT.gROOT.GetListOfFiles().Remove(base_file)

   pr_file = ROOT.TFile(pr_file_path, 'read')
   ROOT.gROOT.GetListOfFiles().Remove(pr_file)

   if base_file.IsOpen():
      print('Baseline file successfully opened', file=sys.stderr)
   else:
      print('Unable to open base file', file=sys.stderr)
      return

   if pr_file.IsOpen():
      print('PR file successfully opened', file=sys.stderr)
   else:
      print('Unable to open PR file', file=sys.stderr)
      return

   run_nr = get_run_nr(pr_file_path)

   # Get list of paths (lists of directories)
   base_flat_dict = flatten_file(base_file, run_nr)
   pr_flat_dict = flatten_file(pr_file, run_nr)

   # Paths that appear in both baseline and PR data. (Intersection)
   shared_paths = list(set(pr_flat_dict).intersection(set(base_flat_dict)))

   # Paths that appear only in PR data. (Except)
   only_pr_paths = list(set(pr_flat_dict).difference(set(base_flat_dict)))

   # Paths that appear only in baseline data. (Except)
   only_base_paths = list(set(base_flat_dict).difference(set(pr_flat_dict)))

   # Histograms pointed to by these paths will be written to baseline output
   paths_to_save_in_base = []

   # Histograms pointed to by these paths will be written to pr output
   paths_to_save_in_pr = []

   # Make comparison
   if num_processes > 1:
       print("starting comparison using %d process(es)" % num_processes)
       manager = multiprocessing.Manager()
       return_dict = manager.dict()
       proc = []
       iProc = 0

       block = len(shared_paths)//num_processes
       for i in range(num_processes):
           p = multiprocessing.Process(target=compareMP, args=(shared_paths[i*block:(i+1)*block], pr_flat_dict, base_flat_dict, i, return_dict))
           proc.append(p)
           p.start()
           iProc += 1
       p = multiprocessing.Process(target=compareMP, args=(shared_paths[(i+1)*block:len(shared_paths)], pr_flat_dict, base_flat_dict, num_processes, return_dict))
       proc.append(p)
       p.start()
       iProc += 1

       for i in range(iProc):
           proc[i].join()
           paths_to_save_in_pr.extend(return_dict[i]['pr'])
           paths_to_save_in_base.extend(return_dict[i]['base'])

       paths_to_save_in_pr.sort()
       paths_to_save_in_base.sort()
       print("Done")
   else:
       compare(shared_paths, pr_flat_dict, base_flat_dict, paths_to_save_in_pr, paths_to_save_in_base)

   # Collect paths that have to be written to baseline output file
   for path in only_base_paths:
      item = base_flat_dict[path]

      if item == None:
         continue

      paths_to_save_in_base.append(path)

   # Collect paths that have to be written to PR output file
   for path in only_pr_paths:
      item = pr_flat_dict[path]

      if item == None:
         continue

      paths_to_save_in_pr.append(path)

   base_output_filename = get_output_filename(pr_file_path, pr_number, test_number, cmssw_version, False)
   pr_output_filename = get_output_filename(pr_file_path, pr_number, test_number, cmssw_version, True)

   # Write baseline output
   save_paths(base_flat_dict, paths_to_save_in_base, os.path.join(output_dir_path, 'base', base_output_filename))

   # Write PR output
   save_paths(pr_flat_dict, paths_to_save_in_pr, os.path.join(output_dir_path, 'pr', pr_output_filename))

   pr_file.Close()
   base_file.Close()

   # Info about changed, added and removed elements
   nr_of_changed_elements = len(set(paths_to_save_in_base).intersection(set(paths_to_save_in_pr)))
   nr_of_removed_elements = len(paths_to_save_in_base) - nr_of_changed_elements
   nr_of_added_elements = len(paths_to_save_in_pr) - nr_of_changed_elements

   print('Base output file. PR output file. Changed elements, removed elements, added elements:')
   print(base_output_filename)
   print(pr_output_filename)
   print('%s %s %s' % (nr_of_changed_elements, nr_of_removed_elements, nr_of_added_elements))

create_dir()

def compareHistograms.create_dir	(		parent_dir,
			name
	)

Definition at line 300 of file compareHistograms.py.

Referenced by save_to_file().

def create_dir(parent_dir, name):
   dir = parent_dir.Get(name)
   if not dir:
      dir = parent_dir.mkdir(name)
   return dir

flatten_file()

def compareHistograms.flatten_file	(		file,
			run_nr
	)

Definition at line 223 of file compareHistograms.py.

References traverse_till_end().

Referenced by create_dif().

def flatten_file(file, run_nr):
   result = {}
   for key in file.GetListOfKeys():
      try:
         traverse_till_end(key.ReadObj(), [], result, run_nr)
      except:
         pass

   return result

get_node_name()

def compareHistograms.get_node_name

(

node

)

Definition at line 243 of file compareHistograms.py.

References get_string_suffix(), and submitPVValidationJobs.split().

Referenced by traverse_till_end().

def get_node_name(node):
   if node.InheritsFrom('TObjString'):
      # Strip out just the name from a tag (<name>value</name>)
      name = node.GetName().split('>')[0][1:]
      return name + get_string_suffix()
   else:
      return node.GetName()

get_output_filename()

def compareHistograms.get_output_filename	(		input_file_path,
			pr_number,
			test_number,
			cmssw_version,
			isPr
	)

Definition at line 306 of file compareHistograms.py.

References python.rootplot.root2matplotlib.replace(), and submitPVValidationJobs.split().

Referenced by create_dif().

def get_output_filename(input_file_path, pr_number, test_number, cmssw_version, isPr):
   # Samples of correct output file format:
   # DQM_V0001_R000320822__wf136_892_pr__CMSSW_10_4_0_pre3-PR25518-1234__DQMIO.root
   # When run number is 1 we have to use RelVal naming pattern:
   # DQM_V0002_R000000001__RelVal_wf136_892_pr__CMSSW_10_4_0_pre3-PR25518-1234__DQMIO.root

   input_file_name = os.path.basename(input_file_path)

   run = input_file_name.split('_')[2]
   workflow = os.path.basename(os.path.dirname(input_file_path)).split('_')[0].replace('.', '_')
   if not workflow:
      workflow = 'Unknown'

   relval_prefix = ''
   if run == 'R000000001':
      relval_prefix = 'RelVal_'

   baseOrPr = 'base'
   if isPr:
      baseOrPr = 'pr'

   return 'DQM_V0001_%s__%swf%s_%s__%s-PR%s-%s__DQMIO.root' % (run, relval_prefix, workflow, baseOrPr, cmssw_version, pr_number, test_number)

get_run_nr()

def compareHistograms.get_run_nr

(

file_path

)

Definition at line 329 of file compareHistograms.py.

References submitPVValidationJobs.split().

Referenced by create_dif().

def get_run_nr(file_path):
   return os.path.basename(file_path).split('_')[2].lstrip('R').lstrip('0')

get_string_suffix()

def compareHistograms.get_string_suffix

(

)

Definition at line 251 of file compareHistograms.py.

Referenced by get_node_name(), and is_blacklisted().

def get_string_suffix():
   return '_string_monitor_element'

is_blacklisted()

def compareHistograms.is_blacklisted	(		dirs_list,
			run_nr
	)

Definition at line 254 of file compareHistograms.py.

References blacklist.get_blacklist(), get_string_suffix(), python.rootplot.root2matplotlib.replace(), and mkLumiAveragedPlots.tuple.

Referenced by traverse_till_end().

def is_blacklisted(dirs_list, run_nr):
   # Copy the list
   dirs_list = dirs_list[:]
   # Remove string suffix
   if dirs_list[-1].endswith(get_string_suffix()):
      dirs_list[-1] = dirs_list[-1].replace(get_string_suffix(), '')

   return tuple(dirs_list) in get_blacklist(run_nr)

save_paths()

def compareHistograms.save_paths	(		flat_dict,
			paths,
			result_file_path
	)

Definition at line 263 of file compareHistograms.py.

References print(), and save_to_file().

Referenced by create_dif().

def save_paths(flat_dict, paths, result_file_path):
   if len(paths) == 0:
      print('No differences were observed - output will not be written', file=sys.stderr)
      return

   # Make sure output dir exists
   result_dir = os.path.dirname(result_file_path)
   if not os.path.exists(result_dir):
      os.makedirs(result_dir)

   result_file = ROOT.TFile(result_file_path, 'recreate')
   ROOT.gROOT.GetListOfFiles().Remove(result_file)

   if not result_file.IsOpen():
      print('Unable to open %s output file' % result_file_path, file=sys.stderr)
      return

   for path in paths:
      save_to_file(flat_dict, path, result_file)

   result_file.Close()
   print('Output written to %s file' % result_file_path, file=sys.stderr)

# Saves file from flat_dict in the same dir of currently open file for writing

save_to_file()

def compareHistograms.save_to_file	(		flat_dict,
			path,
			output_file
	)

Definition at line 287 of file compareHistograms.py.

References create_dir().

Referenced by save_paths().

def save_to_file(flat_dict, path, output_file):
   histogram = flat_dict[path]

   current = output_file

   # Last item is filename. No need to create dir for it
   for directory in path[:-1]:
      current = create_dir(current, directory)
      current.cd()

   histogram.Write()

# Create dir in root file if it doesn't exist

traverse_till_end()

def compareHistograms.traverse_till_end	(		node,
			dirs_list,
			result,
			run_nr
	)

Definition at line 233 of file compareHistograms.py.

References get_node_name(), is_blacklisted(), and mkLumiAveragedPlots.tuple.

Referenced by flatten_file().

def traverse_till_end(node, dirs_list, result, run_nr):
   new_dir_list = dirs_list + [get_node_name(node)]
   if hasattr(node, 'GetListOfKeys'):
      for key in node.GetListOfKeys():
         traverse_till_end(key.ReadObj(), new_dir_list, result, run_nr)
   else:
      if not is_blacklisted(new_dir_list, run_nr):
         path = tuple(new_dir_list)
         result[path] = node

Variable Documentation

args

compareHistograms.args

Definition at line 342 of file compareHistograms.py.

cmssw_version

compareHistograms.cmssw_version

Definition at line 344 of file compareHistograms.py.

default

compareHistograms.default

Definition at line 337 of file compareHistograms.py.

description

compareHistograms.description

Definition at line 333 of file compareHistograms.py.

help

compareHistograms.help

Definition at line 335 of file compareHistograms.py.

IgnoreCommandLineOptions

compareHistograms.IgnoreCommandLineOptions

Definition at line 5 of file compareHistograms.py.

parser

compareHistograms.parser

Definition at line 333 of file compareHistograms.py.

required

compareHistograms.required

Definition at line 335 of file compareHistograms.py.

type

compareHistograms.type

Definition at line 340 of file compareHistograms.py.