| 1 |
paus |
1.2 |
#!/usr/bin/env python
|
| 2 |
|
|
#---------------------------------------------------------------------------------------------------
|
| 3 |
|
|
# Simple interface to command line DBS to prepare my crabTask input files.
|
| 4 |
|
|
#---------------------------------------------------------------------------------------------------
|
| 5 |
paus |
1.5 |
import os,sys,types,string,re,getopt
|
| 6 |
paus |
1.2 |
|
| 7 |
|
|
# Define string to explain usage of the script
|
| 8 |
|
|
usage = "Usage: input.py --dataset=<name>\n"
|
| 9 |
|
|
usage += " --option=[ lfn, xml ]\n"
|
| 10 |
paus |
1.3 |
usage += " [ --dbs= ]\n"
|
| 11 |
paus |
1.2 |
usage += " --help\n"
|
| 12 |
|
|
|
| 13 |
|
|
def printLine(option,nEvents,block,lfn,iJob):
|
| 14 |
|
|
if option == 'xml':
|
| 15 |
|
|
print ' <Job MaxEvents="%d'%nEvents + '" InputFiles="' + lfn \
|
| 16 |
|
|
+ '" SkipEvents="0" JobID="%d'%iJob + '" > </Job>'
|
| 17 |
|
|
else:
|
| 18 |
|
|
print "%s %s %d"%(block,lfn,nEvents)
|
| 19 |
|
|
|
| 20 |
|
|
|
| 21 |
|
|
# Define the valid options which can be specified and check out the command line
|
| 22 |
paus |
1.3 |
valid = ['db=','dbs=','dataset=','option=','help']
|
| 23 |
paus |
1.2 |
try:
|
| 24 |
|
|
opts, args = getopt.getopt(sys.argv[1:], "", valid)
|
| 25 |
|
|
except getopt.GetoptError, ex:
|
| 26 |
|
|
print usage
|
| 27 |
|
|
print str(ex)
|
| 28 |
|
|
sys.exit(1)
|
| 29 |
|
|
|
| 30 |
|
|
# --------------------------------------------------------------------------------------------------
|
| 31 |
|
|
# Get all parameters for the production
|
| 32 |
|
|
# --------------------------------------------------------------------------------------------------
|
| 33 |
|
|
# Set defaults for each option
|
| 34 |
|
|
db = None
|
| 35 |
paus |
1.3 |
dbs = None
|
| 36 |
paus |
1.2 |
dataset = None
|
| 37 |
|
|
option = 'lfn'
|
| 38 |
|
|
|
| 39 |
|
|
# Read new values from the command line
|
| 40 |
|
|
for opt, arg in opts:
|
| 41 |
|
|
if opt == "--help":
|
| 42 |
|
|
print usage
|
| 43 |
|
|
sys.exit(0)
|
| 44 |
|
|
if opt == "--db":
|
| 45 |
|
|
db = arg
|
| 46 |
paus |
1.3 |
if opt == "--dbs":
|
| 47 |
|
|
dbs = arg
|
| 48 |
paus |
1.2 |
if opt == "--dataset":
|
| 49 |
|
|
dataset = arg
|
| 50 |
|
|
if opt == "--option":
|
| 51 |
|
|
option = arg
|
| 52 |
|
|
|
| 53 |
|
|
# Deal with obvious problems
|
| 54 |
|
|
if dataset == None:
|
| 55 |
|
|
cmd = "--dataset= required parameter not provided."
|
| 56 |
|
|
raise RuntimeError, cmd
|
| 57 |
|
|
|
| 58 |
|
|
#---------------------------------------------------------------------------------------------------
|
| 59 |
|
|
# main
|
| 60 |
|
|
#---------------------------------------------------------------------------------------------------
|
| 61 |
|
|
if not db:
|
| 62 |
|
|
# find relevant blocks
|
| 63 |
paus |
1.4 |
if dbs == 'none':
|
| 64 |
|
|
cmd = 'dascli.py --query="block=' + dataset + '*" --limit=999999 --format=blocks'
|
| 65 |
|
|
elif dbs == '':
|
| 66 |
paus |
1.6 |
cmd = 'dbs search --query=\"find block where dataset=*' + dataset + '\"'
|
| 67 |
paus |
1.5 |
elif re.search('http://',dbs):
|
| 68 |
paus |
1.6 |
cmd = 'dbs search --url=' + dbs + ' --query="find block where dataset=*' + dataset + '"'
|
| 69 |
paus |
1.3 |
else:
|
| 70 |
paus |
1.5 |
cmd = 'echo ' + dataset + '#00000000-0000-0000-0000-000000000000'
|
| 71 |
|
|
|
| 72 |
paus |
1.4 |
#print "CMD " + cmd
|
| 73 |
paus |
1.2 |
cmd += "| grep \# | sort"
|
| 74 |
paus |
1.3 |
# never print #print "cmd: " + cmd
|
| 75 |
|
|
|
| 76 |
paus |
1.2 |
blocks = []
|
| 77 |
|
|
iJob = 1
|
| 78 |
|
|
if option == 'xml':
|
| 79 |
|
|
print '<arguments>'
|
| 80 |
|
|
for line in os.popen(cmd).readlines():
|
| 81 |
|
|
line = line[:-1]
|
| 82 |
|
|
blocks.append(line)
|
| 83 |
|
|
for block in blocks:
|
| 84 |
paus |
1.5 |
#print ' BLOCK: ' + block
|
| 85 |
|
|
|
| 86 |
paus |
1.4 |
if dbs == 'none':
|
| 87 |
|
|
cmd = 'dascli.py --query="file block=' + block + '" --limit=999999 --format=files'
|
| 88 |
|
|
elif dbs == '':
|
| 89 |
|
|
cmd = 'dbs search --query="find file,file.numevents where block=' + block + '"'
|
| 90 |
paus |
1.5 |
elif re.search('http://',dbs):
|
| 91 |
paus |
1.4 |
cmd = 'dbs search --url=' + dbs + \
|
| 92 |
|
|
' --query="find file,file.numevents where block=' + block + '"'
|
| 93 |
paus |
1.5 |
else:
|
| 94 |
|
|
cmd = 'cat /home/cmsprod/catalog/t2mit/private/' + dbs + dataset \
|
| 95 |
|
|
+ '/Files | sed \'s@XX-CATALOG-XX@@\' | sed \'s@root://xrootd1.cmsaf.mit.edu/@@\''
|
| 96 |
|
|
|
| 97 |
paus |
1.4 |
#print "CMD " + cmd
|
| 98 |
paus |
1.2 |
cmd += "| grep store | sort"
|
| 99 |
|
|
for line in os.popen(cmd).readlines():
|
| 100 |
paus |
1.5 |
#print "LINE >" + line
|
| 101 |
paus |
1.2 |
line = line[:-1]
|
| 102 |
|
|
f = line.split()
|
| 103 |
|
|
lfn = f[0]
|
| 104 |
|
|
nEvents = int(f[1])
|
| 105 |
|
|
f = lfn.split("/")
|
| 106 |
|
|
file = f[-1]
|
| 107 |
|
|
if nEvents != 0:
|
| 108 |
|
|
printLine(option,nEvents,block,lfn,iJob)
|
| 109 |
|
|
iJob = iJob + 1
|
| 110 |
|
|
if option == 'xml':
|
| 111 |
|
|
print '</arguments>'
|
| 112 |
|
|
|
| 113 |
|
|
if db:
|
| 114 |
|
|
cmd = 'cat ' + db
|
| 115 |
|
|
|
| 116 |
|
|
iJob = 1
|
| 117 |
|
|
if option == 'xml':
|
| 118 |
|
|
print '<arguments>'
|
| 119 |
|
|
for line in os.popen(cmd).readlines():
|
| 120 |
|
|
line = line[:-1]
|
| 121 |
|
|
|
| 122 |
|
|
f = line.split()
|
| 123 |
|
|
block = f[0]
|
| 124 |
|
|
lfn = f[1]
|
| 125 |
|
|
nEvents = int(f[2])
|
| 126 |
|
|
|
| 127 |
|
|
f = lfn.split("/")
|
| 128 |
|
|
file = f[-1]
|
| 129 |
|
|
|
| 130 |
|
|
if nEvents != 0:
|
| 131 |
|
|
printLine(option,nEvents,block,lfn,iJob)
|
| 132 |
|
|
iJob = iJob + 1
|
| 133 |
|
|
|
| 134 |
|
|
if option == 'xml':
|
| 135 |
|
|
print '</arguments>'
|
