| 1 |
import os, sys, re
|
| 2 |
|
| 3 |
###############################################################
|
| 4 |
|
| 5 |
#Usage:
|
| 6 |
# python onlyLast.py <path to input directory on CASTOR>
|
| 7 |
|
| 8 |
#now updated to handle output from the latest CRAB with
|
| 9 |
#file names like this:
|
| 10 |
# myfilename_1_1_aBC.root
|
| 11 |
|
| 12 |
# 02/09/11 - Don
|
| 13 |
# updated to handle files on the T3
|
| 14 |
# this script now automatically checks if the input dir is pointing
|
| 15 |
# to the T3 and adapts accordingly
|
| 16 |
# e.g. python onlyLast.py "$CUSE/blah"
|
| 17 |
|
| 18 |
###############################################################
|
| 19 |
|
| 20 |
#handle input arguments
|
| 21 |
#this includes the name of the script
|
| 22 |
##for arg in sys.argv:
|
| 23 |
## print arg
|
| 24 |
|
| 25 |
#sys.argv[1] will have the first argument
|
| 26 |
tmpfile = '.onlyLast_py_'
|
| 27 |
mypid = os.getpid()
|
| 28 |
tmpfile += `mypid`
|
| 29 |
|
| 30 |
|
| 31 |
inputdir = sys.argv[1]
|
| 32 |
if inputdir[len(inputdir)-1] != '/':
|
| 33 |
inputdir+='/'
|
| 34 |
|
| 35 |
#check if the input directory is located on the T3
|
| 36 |
isT3 = False
|
| 37 |
checkT3 = 'echo '
|
| 38 |
checkT3 += inputdir
|
| 39 |
checkT3 += ' | awk -F "/" \'{print $1}\' '
|
| 40 |
p = os.popen(checkT3)
|
| 41 |
location = p.readline()
|
| 42 |
p.close()
|
| 43 |
#print location
|
| 44 |
if "srm" in location:
|
| 45 |
print "T3 directory specified. Using grid commands."
|
| 46 |
isT3 = True
|
| 47 |
|
| 48 |
|
| 49 |
outputdir = inputdir
|
| 50 |
outputdir += 'EXTRAS'
|
| 51 |
|
| 52 |
mkdircommand = "nsmkdir "
|
| 53 |
mkdircommand += outputdir
|
| 54 |
|
| 55 |
#if isT3 == 'T3':
|
| 56 |
# mkdircommand = "srmmkdir "
|
| 57 |
# mkdircommand += outputdir
|
| 58 |
|
| 59 |
|
| 60 |
#get a file listing into tmpfile
|
| 61 |
cmd = 'nsls -l '
|
| 62 |
cmd += inputdir
|
| 63 |
cmd += ' | awk \'// {print $5,$9;}\' > '
|
| 64 |
|
| 65 |
|
| 66 |
if isT3 == True:
|
| 67 |
cmd = 'srmls "'
|
| 68 |
cmd += inputdir
|
| 69 |
cmd += '" | awk -F "/" \'{print $1,$NF}\' | awk \'NF>0\' > '
|
| 70 |
|
| 71 |
|
| 72 |
cmd+= tmpfile
|
| 73 |
#print cmd
|
| 74 |
os.system(cmd)
|
| 75 |
|
| 76 |
#keep track of whether we made the directory or not
|
| 77 |
alreadymadedir = 0
|
| 78 |
#do not manually make the directory for the T3, lcg-cp will take care of it
|
| 79 |
if isT3 == True:
|
| 80 |
alreadymadedir = 1
|
| 81 |
|
| 82 |
|
| 83 |
|
| 84 |
|
| 85 |
f = open(tmpfile,'r')
|
| 86 |
|
| 87 |
#to deal with the arbitrary extension to the filename that crab
|
| 88 |
#now adds, I will add a second associative array to keep track of the
|
| 89 |
#whole file name. (completefilenames)
|
| 90 |
#The only subtle part is where the file names are sorted.
|
| 91 |
#Hopefully this does not create any bugs
|
| 92 |
indexdict = {}
|
| 93 |
completefilenames = {}
|
| 94 |
|
| 95 |
stub = 'stub'
|
| 96 |
|
| 97 |
#do some accounting
|
| 98 |
nmoved = 0
|
| 99 |
|
| 100 |
for line in f:
|
| 101 |
mypair = line.split()
|
| 102 |
#file size
|
| 103 |
size = int(mypair[0])
|
| 104 |
#parse filename
|
| 105 |
result = re.match('(.*)_([0-9]*)_([0-9]*)_(.*).root',mypair[1])
|
| 106 |
if size > 0:
|
| 107 |
if stub=='stub':
|
| 108 |
stub = result.group(1)
|
| 109 |
elif stub!=result.group(1):
|
| 110 |
print "WARNING -- there is more than one group of filenames here!"
|
| 111 |
|
| 112 |
if result.group(2) in indexdict:
|
| 113 |
#so the key already exists
|
| 114 |
indexdict[result.group(2)].append(result.group(3))
|
| 115 |
completefilenames[result.group(2)].append(mypair[1])
|
| 116 |
else:
|
| 117 |
indexdict[result.group(2)] = [result.group(3)]
|
| 118 |
completefilenames[result.group(2)] = [mypair[1]]
|
| 119 |
# print result.group(0)
|
| 120 |
# print result.group(1)
|
| 121 |
# print result.group(2)
|
| 122 |
# print result.group(3)
|
| 123 |
|
| 124 |
#print indexdict
|
| 125 |
|
| 126 |
for ii in indexdict:
|
| 127 |
if len(indexdict[ii])==1:
|
| 128 |
print "nothing to do for ",ii
|
| 129 |
else:
|
| 130 |
indexdict[ii].sort()
|
| 131 |
#will this work? in my one test case, yes
|
| 132 |
completefilenames[ii].sort()
|
| 133 |
goodindex = indexdict[ii].pop()
|
| 134 |
goodfilename = completefilenames[ii].pop()
|
| 135 |
print ii, ": keeping index ",goodindex
|
| 136 |
print ii, ": corresponds to file ",goodfilename
|
| 137 |
for jj in completefilenames[ii]:
|
| 138 |
if alreadymadedir == 0:
|
| 139 |
print mkdircommand
|
| 140 |
#this will give a harmless error if the dir already exists
|
| 141 |
os.system(mkdircommand)
|
| 142 |
|
| 143 |
s=inputdir
|
| 144 |
s+=jj
|
| 145 |
print "moving ", s
|
| 146 |
cpcmd = 'rfcp '
|
| 147 |
if isT3 == True:
|
| 148 |
cpcmd = 'lcg-cp --verbose -b -D srmv2 "'
|
| 149 |
cpcmd += s
|
| 150 |
|
| 151 |
if isT3 == True:
|
| 152 |
cpcmd += '" "'
|
| 153 |
else: cpcmd += ' '
|
| 154 |
|
| 155 |
cpcmd += outputdir
|
| 156 |
if isT3 == True:
|
| 157 |
cpcmd +='/'
|
| 158 |
cpcmd +=jj
|
| 159 |
cpcmd += '"'
|
| 160 |
print cpcmd
|
| 161 |
os.system(cpcmd)
|
| 162 |
rmcmd = 'rfrm '
|
| 163 |
if isT3 == True:
|
| 164 |
rmcmd = 'srmrm '
|
| 165 |
rmcmd += s
|
| 166 |
print rmcmd
|
| 167 |
os.system(rmcmd)
|
| 168 |
nmoved = nmoved+1
|
| 169 |
|
| 170 |
f.close()
|
| 171 |
os.remove(tmpfile)
|
| 172 |
|
| 173 |
print "----------------------------"
|
| 174 |
print "I moved this many files: ",nmoved
|
| 175 |
print "----------------------------"
|
| 176 |
|
| 177 |
|
| 178 |
|
