Check the integrity of files in an AnalysisTools output directory

check_output_directory.py

import glob

import ROOT
import concurrent.futures


def check_file(path):
    f = ROOT.TFile.Open(path)
    # The file is somehow unreadable.
    if not f:
        return path, None
    # The file is incomplete or corrupt.
    if f.TestBit(ROOT.TFile.kRecovered) or f.IsZombie():
        return path, None
    try:
        events = f.Get('Events')
        n_entries = events.GetEntriesFast()
        return path, n_entries
    except Exception:
        return path, None
    finally:
        f.Close()
    
    
if __name__ == '__main__':
    
    # Output directory globbing pattern.
    SEARCH_PATTERN = '/path/to/the/output/dir/{0}/*.root'
    
    # The sample names.
    SAMPLES = ['ZH', 'DYToLL', 'QCDHT300']
    
    for sample in SAMPLES:
        print 'Checking {0}'.format(sample)

        with concurrent.futures.ThreadPoolExecutor(42) as executor:
            futures = [executor.submit(check_file, path) for path in glob.glob(SEARCH_PATTERN.format(sample))]

        bad_paths = []
        for future in concurrent.futures.as_completed(futures):
            path, n_entries = future.result()
            if n_entries is None:
                bad_paths.append(path)
            elif n_entries == 0:
                print '{0} has a tree with no events!'.format(path)
        
        print 'There are {0!s} good files out of a total of {1!s}'.format(len(futures) - len(bad_paths), len(futures))
        
        print 'The following files are likely corrupt: {0}'.format('\n'.join(bad_paths))