| 1 |
#!/usr/bin/python
|
| 2 |
|
| 3 |
#
|
| 4 |
# This is a wrapper script around the hadd tool.
|
| 5 |
# - hadd tool merges multiple root files with same tree
|
| 6 |
# but is limited to 1000 files
|
| 7 |
# - this script can merge up to 1000 * 1000 root files
|
| 8 |
#
|
| 9 |
# Todo:
|
| 10 |
# - extend to merge more files.
|
| 11 |
#
|
| 12 |
|
| 13 |
import subprocess
|
| 14 |
import sys
|
| 15 |
import os
|
| 16 |
|
| 17 |
def callHadd(outfile, filelist):
|
| 18 |
if os.path.exists(outfile):
|
| 19 |
print 'ERROR: file ', outfile, ' exists already.'
|
| 20 |
sys.exit()
|
| 21 |
|
| 22 |
cmd = 'hadd %s %s' % (outfile, ' '.join(filelist))
|
| 23 |
#print cmd
|
| 24 |
if not subprocess.call(cmd, shell=True) == 0:
|
| 25 |
print 'ERROR: while executing hadd.'
|
| 26 |
sys.exit()
|
| 27 |
|
| 28 |
|
| 29 |
def hadd(outfile, infiles):
|
| 30 |
nFiles = 0
|
| 31 |
files = []
|
| 32 |
nTempfiles = 0
|
| 33 |
tempfiles = []
|
| 34 |
|
| 35 |
for f in infiles:
|
| 36 |
if os.path.splitext(f)[1] == '.root':
|
| 37 |
files.append(f)
|
| 38 |
nFiles+=1
|
| 39 |
|
| 40 |
if nFiles > 0 and nFiles % 1000 == 0:
|
| 41 |
nTempfiles+=1
|
| 42 |
tmpfile = '%s_%03d.root' % (os.path.splitext(outfile)[0], nTempfiles)
|
| 43 |
callHadd(tmpfile, files)
|
| 44 |
files = []
|
| 45 |
tempfiles.append(tmpfile)
|
| 46 |
|
| 47 |
# join rest
|
| 48 |
if nFiles > 0 and not nFiles % 1000 == 0:
|
| 49 |
nTempfiles+=1
|
| 50 |
tmpfile = '%s_%03d.root' % (os.path.splitext(outfile)[0], nTempfiles)
|
| 51 |
callHadd(tmpfile, files)
|
| 52 |
tempfiles.append(tmpfile)
|
| 53 |
|
| 54 |
# join the tempfiles
|
| 55 |
callHadd(outfile, tempfiles)
|
| 56 |
|
| 57 |
# remove tempfiles
|
| 58 |
for f in tempfiles:
|
| 59 |
os.remove(f)
|
| 60 |
|
| 61 |
print '%d files joined in %s.' % (nFiles, outfile)
|
| 62 |
|
| 63 |
# ----------------------------------------------------------------------
|
| 64 |
|
| 65 |
if __name__ == '__main__':
|
| 66 |
|
| 67 |
if len(sys.argv) < 3:
|
| 68 |
print 'Usage: ', sys.argv[0], ' outfile infile1 [infile2] ...'
|
| 69 |
exit()
|
| 70 |
|
| 71 |
outfile = sys.argv[1]
|
| 72 |
|
| 73 |
if not os.path.splitext(outfile)[1] == '.root':
|
| 74 |
print 'ERROR: outfile ', outfile, 'must be a root file.'
|
| 75 |
exit()
|
| 76 |
|
| 77 |
if os.path.exists(outfile):
|
| 78 |
print 'ERROR: outfile ', outfile, ' must not exist.'
|
| 79 |
exit()
|
| 80 |
|
| 81 |
hadd(outfile, sys.argv[2:])
|
