def get_set_of_histogram_paths(input_file, seen_summed_hists = False):
global existing_bins, to_be_created, filter_on_folders, filter_on_histograms
histogram_list = []
add_histogram = histogram_list.append
checked_n_entries = 0
for histogram_path in input_file.readlines():
checked_n_entries += 1
if checked_n_entries % 10000 == 0:
logging.debug( 'Checked %d' %checked_n_entries)
if not filter_string(histogram_path, filter_on_folders):
continue
if not filter_string(histogram_path, filter_on_histograms):
continue
histogram_path = histogram_path.rstrip(' \n')
directory, histogram_name, b_tag_bin = get_histogram_info_tuple(histogram_path)
logging.debug('Searching %s' % histogram_path)
logging.debug('Found directory %s' % directory)
logging.debug('Found histogram_name %s' % histogram_name)
logging.debug('Found b_tag_bin %s' % b_tag_bin)
if b_tag_bin in existing_bins:
histogram_name = '_'.join(histogram_name.split('_')[:-1])
logging.debug('Adding histogram %s' % (directory + histogram_name))
add_histogram(directory + histogram_name)
if b_tag_bin in to_be_created:
seen_summed_hists = True
return set(histogram_list)#only unique ones
def sum_b_tag_bins_in_file(file_in_path):
global existing_bins, to_be_created, existing_histogram_file, input_folder, output_folder
logging.debug('Processing file %s' % file_in_path)
output_file_name = file_in_path.replace('.root', '_summed.root')
output_file_name = output_file_name.replace(input_folder, output_folder)
#run rootinfo on file
#or read the output (histogram list)
input_file = open(existing_histogram_file)
seen_summed_hists = False
histogram_set = get_set_of_histogram_paths(input_file, seen_summed_hists)
logging.debug('Found %d unique b-tag binned histograms' %len(histogram_set))
if seen_summed_hists:
logging.warn('Summed histograms have been detected. Will skip this part')
return
input_file.close()
directories = []
for path in histogram_set:
histogram_path, histogram_name, b_tag_bin = get_histogram_info_tuple(path)
directories.append(histogram_path)
logging.debug('opening file %s ' % output_file_name)
output_file = File(output_file_name, 'recreate')
cd = output_file.cd
logging.debug( 'creating folder structure')
create_folder_structure(output_file, directories)
logging.debug( 'created folder structure')
logging.debug('opening file %s ' % file_in_path)
input_file = File(file_in_path, 'read')
get_histogram = input_file.Get
logging.debug('opened file')
new_histograms = {}
for histogram in histogram_set:
cd()
logging.debug('Processing histogram: %s' % histogram)
histogram_path, histogram_name, b_tag_bin = get_histogram_info_tuple(histogram)
logging.debug('Found histogram_path %s' % histogram_path)
logging.debug('Found histogram_name %s' % histogram_name)
cd(histogram_path)
existing_histograms = [get_histogram(histogram + '_' + existing_bin).Clone() for existing_bin in existing_bins]
for bin_i, b_tag_bin in enumerate(existing_bins):#write existing b-tag bins
current_histogram_name = histogram_name + '_' + b_tag_bin
existing_histograms[bin_i].Write(current_histogram_name)
for bin_i, b_tag_bin in enumerate(to_be_created):#write new b-tag bins
current_histogram_name = histogram_name + '_' + b_tag_bin
new_histogram = existing_histograms[bin_i].Clone(current_histogram_name)
for existing_histogram in existing_histograms[bin_i + 1:]:
new_histogram.Add(existing_histogram)
new_histogram.Write(current_histogram_name)
input_file.Close()
output_file.Close()
logging.debug( 'Finished %s' % file_in_path)
logging.debug( 'Output: %s' % output_file_name)
del new_histograms, histogram_set, input_file, output_file
return
