Processing/bin/input.py

#!/usr/bin/env python
#---------------------------------------------------------------------------------------------------
# Simple interface to command line DBS to prepare my crabTask input files.
#---------------------------------------------------------------------------------------------------
import os,sys,types,string,getopt

# Define string to explain usage of the script
usage =  "Usage: input.py --dataset=<name>\n"
usage += "                --option=[ lfn, xml ]\n"
usage += "                [ --dbs= ]\n"
usage += "                --help\n"

def printLine(option,nEvents,block,lfn,iJob):
    if option == 'xml':
        print '  <Job MaxEvents="%d'%nEvents + '"  InputFiles="' + lfn \
              + '" SkipEvents="0" JobID="%d'%iJob + '" > </Job>'
    else:
        print "%s %s %d"%(block,lfn,nEvents)

    
# Define the valid options which can be specified and check out the command line
valid = ['db=','dbs=','dataset=','option=','help']
try:
    opts, args = getopt.getopt(sys.argv[1:], "", valid)
except getopt.GetoptError, ex:
    print usage
    print str(ex)
    sys.exit(1)

# --------------------------------------------------------------------------------------------------
# Get all parameters for the production
# --------------------------------------------------------------------------------------------------
# Set defaults for each option
db      = None
dbs     = None
dataset = None
option  = 'lfn'

# Read new values from the command line
for opt, arg in opts:
    if opt == "--help":
        print usage
        sys.exit(0)
    if opt == "--db":
        db      = arg
    if opt == "--dbs":
        dbs     = arg
    if opt == "--dataset":
        dataset = arg
    if opt == "--option":
        option  = arg

# Deal with obvious problems
if dataset == None:
    cmd = "--dataset=  required parameter not provided."
    raise RuntimeError, cmd

#---------------------------------------------------------------------------------------------------
# main
#---------------------------------------------------------------------------------------------------
if not db:
    # find relevant blocks
    if dbs != '':
        cmd = "dbs search --url=" + dbs + " --query=\"find block where dataset=" + dataset + "\""
    else:
        cmd = "dbs search --query=\"find block where dataset=" + dataset + "\""
    cmd += "| grep \# | sort"
    # never print #print "cmd: " + cmd

    blocks = []
    iJob = 1
    if option == 'xml':
        print '<arguments>'
    for line in os.popen(cmd).readlines():
        line = line[:-1]
        blocks.append(line)
    for block in blocks:
        if dbs != '':
            cmd = "dbs search --url=" + dbs + \
            " --query=\"find file,file.numevents where block=" + block + "\""
        else:
            cmd = "dbs search --query=\"find file,file.numevents where block=" + block + "\""
        cmd += "| grep store | sort"
        for line in os.popen(cmd).readlines():
            line = line[:-1]
            f       = line.split()
            lfn     = f[0]
            nEvents = int(f[1])
            f       = lfn.split("/")
            file    = f[-1]
            if nEvents != 0:
                printLine(option,nEvents,block,lfn,iJob)
                iJob = iJob + 1
    if option == 'xml':
        print '</arguments>'

if db:
    cmd = 'cat ' + db

    iJob = 1
    if option == 'xml':
        print '<arguments>'
    for line in os.popen(cmd).readlines():
        line    = line[:-1]
    
        f       = line.split()
        block   = f[0]
        lfn     = f[1]
        nEvents = int(f[2])
    
        f       = lfn.split("/")
        file    = f[-1]

        if nEvents != 0:
            printLine(option,nEvents,block,lfn,iJob)
            iJob = iJob + 1
    
    if option == 'xml':
        print '</arguments>'
Revision:	1.3
Committed:	Sat Mar 19 01:49:13 2011 UTC (14 years, 1 month ago) by paus
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	Mit_024b, Mit_025pre1, Mit_024a, Mit_024, Mit_023, Mit_022a, Mit_022, Mit_020d, TMit_020d, Mit_020c, Mit_021, Mit_021pre2, Mit_021pre1, Mit_020b, Mit_020a, Mit_020
Changes since 1.2:	+16 -3 lines
Log Message:	Small updates here and there.
#	User	Rev	Content
1	paus	1.2	#!/usr/bin/env python
2			#---------------------------------------------------------------------------------------------------
3			# Simple interface to command line DBS to prepare my crabTask input files.
4			#---------------------------------------------------------------------------------------------------
5			import os,sys,types,string,getopt
6
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	1.3	if dbs != '':
64			cmd = "dbs search --url=" + dbs + " --query=\"find block where dataset=" + dataset + "\""
65			else:
66			cmd = "dbs search --query=\"find block where dataset=" + dataset + "\""
67	paus	1.2	cmd += "\| grep \# \| sort"
68	paus	1.3	# never print #print "cmd: " + cmd
69
70	paus	1.2	blocks = []
71			iJob = 1
72			if option == 'xml':
73			print '<arguments>'
74			for line in os.popen(cmd).readlines():
75			line = line[:-1]
76			blocks.append(line)
77			for block in blocks:
78	paus	1.3	if dbs != '':
79			cmd = "dbs search --url=" + dbs + \
80			" --query=\"find file,file.numevents where block=" + block + "\""
81			else:
82			cmd = "dbs search --query=\"find file,file.numevents where block=" + block + "\""
83	paus	1.2	cmd += "\| grep store \| sort"
84			for line in os.popen(cmd).readlines():
85			line = line[:-1]
86			f = line.split()
87			lfn = f[0]
88			nEvents = int(f[1])
89			f = lfn.split("/")
90			file = f[-1]
91			if nEvents != 0:
92			printLine(option,nEvents,block,lfn,iJob)
93			iJob = iJob + 1
94			if option == 'xml':
95			print '</arguments>'
96
97			if db:
98			cmd = 'cat ' + db
99
100			iJob = 1
101			if option == 'xml':
102			print '<arguments>'
103			for line in os.popen(cmd).readlines():
104			line = line[:-1]
105
106			f = line.split()
107			block = f[0]
108			lfn = f[1]
109			nEvents = int(f[2])
110
111			f = lfn.split("/")
112			file = f[-1]
113
114			if nEvents != 0:
115			printLine(option,nEvents,block,lfn,iJob)
116			iJob = iJob + 1
117
118			if option == 'xml':
119			print '</arguments>'