Jeng/scripts/submitToCondor.py

#!/usr/bin/env python
############################################################################
#
# Author: Francisco Yumiceva
# yumiceva@fnal.gov
#
# Fermilab, 2007
#
# 2009/07/14: Modification to prevent empty string for input fileNames
# Geng-yuan Jeng
############################################################################

"""
   Submit jobs to condor. It will create directories in the output path
   to store configuration files and log files.
   
   e.g. submitToCondor.py -c ZprimeEventsReco_full.py -l datasets/zp1Tev.txt -n 10 -o /uscms_data/d1/yumiceva/CMSSW_1_3_6/TQAFAnalyzer/

   where ZprimeEventsReco_full.cfg is the configuration file. The configuration file
   needs to be modified a bit: (1) remove all input files in the PoolSource and add a
   keyword {FILENAME} you need also the brackets. The filename of the output file should
   be changed to another keyword called {OUTPUT_FILENAME}.
   
   To Generate MC: In your cfg file you need to add also the keywords {SEED1},...,{SEED6} for
   the random seeds which are going to be modified by the script using PYTHON random numbers,
   you also need to change the output filename to {OUTPUT_FILENAME} All keywords need to have
   brackets.
   
   usage: %prog
   -m, --mc   : to generate MC. it will handle random numbers.
   -e, --events = EVENTS: number of events to generate MC. If -m flag is turned on.
   -c, --cfg    = CFG: python configuration file
   -l, --list   = LIST: file with list of files (dataset)
   -n, --njobs  = NJOBS: number of jobs.
   -o, --output = OUTPUT: output path
   -i, --initial= INITIAL:  number of initial job
   -f, --final= FINAL: number of final job
   -s, --short : for short jobs that need high priority
   -t, --test : do not submit anything just show what I would do

   Author: Francisco Yumiceva (yumiceva@fnal.gov)
"""


import os,sys
try:
    import random
except:
    # now fix PYTHONPATH for the damn cmslpc python configuration
    atemp = os.getenv("PYTHONPATH")
    atempl = atemp.split(":")
    natemp = ""
    for iatemp in atempl:
        if iatemp.find("python2.3") == -1:
            natemp = natemp + iatemp
    print "\n FIX PYTHONPATH environment variable, try:\n"
    print "setenv PYTHONPATH "+ natemp
    sys.exit()
    #os.putenv("PYTHONPATH",natemp)
    # PYTHONPATH fixed

import string, re
from time import gmtime, localtime, strftime

#_______________OPTIONS__________________________________
import optparse
USAGE = re.compile(r'(?s)\s*usage: (.*?)(\n[ \t]*\n|$)')
def nonzero(self): # will become the nonzero method of optparse.Values
    "True if options were given"
    for v in self.__dict__.itervalues():
        if v is not None: return True
    return False

optparse.Values.__nonzero__ = nonzero # dynamically fix optparse.Values

class ParsingError(Exception): pass

optionstring=""

def exit(msg=""):
    raise SystemExit(msg or optionstring.replace("%prog",sys.argv[0]))

def parse(docstring, arglist=None):
    global optionstring
    optionstring = docstring
    match = USAGE.search(optionstring)
    if not match: raise ParsingError("Cannot find the option string")
    optlines = match.group(1).splitlines()
    try:
        p = optparse.OptionParser(optlines[0])
        for line in optlines[1:]:
            opt, help=line.split(':')[:2]
            short,long=opt.split(',')[:2]
            if '=' in opt:
                action='store'
                long=long.split('=')[0]
            else:
                action='store_true'
            p.add_option(short.strip(),long.strip(),
                         action = action, help = help.strip())
    except (IndexError,ValueError):
        raise ParsingError("Cannot parse the option string correctly")
    return p.parse_args(arglist)

#______________________________________________________________________

#
# path to this scripts and the used template file:
#
TemplatesPATH =  os.environ['CMSSW_BASE']+"/src/submit2uaf/"
#
# path to the output that is created by this script
#
#main_path   = "/uscmst1b_scratch/lpc1/cmsroc/yumiceva/CMSSW_1_2_0_pre3/src/RecoVertex/BeamSpotProducer/test/"
main_path = os.getcwd() + "/"

out_path      = main_path+"condorjobs/"

scripts_path  = main_path+"condorjobs/"
cfg_path      = scripts_path+"cfg/"
#condor_path   = scripts_path+"condor/"
csh_path      = scripts_path+"csh/"
logs_path     = scripts_path+"logs/"


istest = 0
isMC   = 0
Nevents = "-1"
isshort = 0

#
# Path to the input/output data:
#

cfg_tags        = ["{FILENAME}",
                   "{OUTPUT_FILENAME}"]

cfg_tags_mc     = ["{SEED1}",
                   "{SEED2}",
                   "{SEED3}",
                   "{SEED4}",
                   "{SEED5}",
                   "{SEED6}",
                   "{OUTPUT_FILENAME}"]

scripts_tags    = ["{PATHTOOUT}",
                   "{CONFIGFILE}",
                   "{OUTFILE}","{PATHTOBASE}","{OUTDIR}"]

condor_tags     = ["{EXECUTABLE}",
                   "{OUTPUT}",
                   "{ERROR}",
                   "{LOG}",
                   "{NOTIFY}"]


template_fnames = {}
template_fnames["cfg"]          =  TemplatesPATH+"/reco_template.cfg"
template_fnames["condor"]       =  TemplatesPATH+"/template.condor"
template_fnames["csh"]          =  TemplatesPATH+"/runcms_new2.csh"


def _mkdir(newdir):
    """works the way a good mkdir should :)
        - already exists, silently complete
        - regular file in the way, raise an exception
        - parent directory(ies) does not exist, make them as well
    """
    if os.path.isdir(newdir):
        pass
    elif os.path.isfile(newdir):
        raise OSError("a file with the same name as the desired " \
                      "dir, '%s', already exists." % newdir)
    else:
        head, tail = os.path.split(newdir)
        if head and not os.path.isdir(head):
            _mkdir(head)
        if tail:
            os.mkdir(newdir)

def change(infile,outfile,changearray,filearray):
    fin  = open(infile)
    fout = open(outfile,"w")
    
    #if line.find("untracked")!=-1 and line.find("PSet")!=-1 and line.find("maxEvents")!=-1:
    #        line = "untracked PSet maxEvents = {untracked int32 input = "+str(Nevents)+"}\n"

    for line in fin.readlines():
        for change in changearray:
            #if change[0] == "{FILENAME}" and line.find(change[0])!=-1:
            #    line=line.replace(change[0] ,"")
            #    itmp=0
             #   for ifile in filearray:
             #       ifile = ifile.strip("\n")
             #       itmp = itmp + 1
             #       suffix = "\","
             #       if itmp == len(filearray):
             #           suffix = "\""
             #       fout.write("\""+ifile+suffix+"\n")
                    
            #else:
                
            line=line.replace(change[0] ,change[1])
        
        fout.write(line)

    # 
    if outfile.find(".py")!=-1:
        fout.write('''
process.maxEvents = cms.untracked.PSet(input = cms.untracked.int32(-1) )

process.source = cms.Source("PoolSource",
fileNames = cms.untracked.vstring(
''')
        itmp=0
        for ifile in filearray:
            ifile = ifile.strip("\n")
            itmp = itmp + 1
            suffix = "\","
            if itmp == len(filearray):
                suffix = "\""
            fout.write("\""+ifile+suffix+"\n")

        fout.write(")\n")
        fout.write(", duplicateCheckMode = cms.untracked.string('noDuplicateCheck')\n")
        fout.write(")\n")

                                                                                                
def submit_jobs(njob,array,ini_cfgfile,output_path):
    
    cfgfile           = ini_cfgfile.replace("_template","")

    njobstr = str(njob)
    while len(njobstr)<4:
        njobstr = "0"+njobstr
    
    outfilename_cfg   = cfg_path  + cfgfile.replace(".py","_"+njobstr+".py")
    outfilename_root = output_path+ cfgfile.replace(".py","_"+njobstr+".root")
    outfilename_log   = logs_path + cfgfile.replace(".py","_"+njobstr+".log")
    outfilename_csh   = csh_path  + cfgfile.replace(".py","_"+njobstr+".csh")
    outfilename_condor= csh_path  + cfgfile.replace(".py","_"+njobstr+".condor")
    
    if  os.path.isfile(outfilename_root):
        print "Not submitting into condor batch farm since rootoutput already exists"
        print "delete or rename: "+outfilename_root+"\n to resubmit"
        sys.exit(main())
            
    #
    # First create config (.cfg) file:
    #
        
    changearray=[]

    if isMC:
        random.seed(njob)
        changearray.append((cfg_tags_mc[0],str(random.randint(1,987654321) ) ))
        changearray.append((cfg_tags_mc[1],str(random.randint(1,987654321) ) ))
        changearray.append((cfg_tags_mc[2],str(random.randint(1,987654321) ) ))
        changearray.append((cfg_tags_mc[3],str(random.randint(1,987654321) ) ))
        changearray.append((cfg_tags_mc[4],str(random.randint(1,987654321) ) ))
        changearray.append((cfg_tags_mc[5],str(random.randint(1,987654321) ) ))
        changearray.append((cfg_tags_mc[6],outfilename_root))
        change(template_fnames["cfg"],outfilename_cfg,changearray,array)
        #print outfilename_cfg + " has been written.\n"

    else:
        changearray.append((cfg_tags[0],""))
        changearray.append((cfg_tags[1],outfilename_root))
        change(template_fnames["cfg"],outfilename_cfg,changearray,array)
        #print outfilename_cfg + " has been written.\n"
    
    
    #
    # now create the script to process the file:
    #
    changearray=[]
    changearray.append((scripts_tags[0],output_path+"/"))
    changearray.append((scripts_tags[1],outfilename_cfg))
    changearray.append((scripts_tags[2],outfilename_root))
    changearray.append((scripts_tags[3],os.getenv("CMSSW_BASE")+"/src" ))
    changearray.append((scripts_tags[4],output_path))
    change(template_fnames["csh"],outfilename_csh,changearray,0)
    os.chmod(outfilename_csh,0775)
    #print outfilename_csh + " has been written.\n"
    
    #
    # finally create the condor job description file:
    #
    stdout         = outfilename_log.replace(".log",".stdout")
    stderr         = outfilename_log.replace(".log",".stderr")
    logfiles       = outfilename_log.replace(".log",".log")
    changearray    =[]
    changearray.append((condor_tags[0],outfilename_csh))
    changearray.append((condor_tags[1],stdout))
    changearray.append((condor_tags[2],stderr))
    changearray.append((condor_tags[3],logfiles))
    changearray.append((condor_tags[4],os.getenv("USER")+"@fnal.gov"))
    change(template_fnames["condor"],outfilename_condor,changearray,0)
    if isshort:
        fout = open(outfilename_condor,"a")
        fout.write("+LENGTH=\"SHORT\"\n")
        fout.close()
    
    #print outfilename_condor + " has been written.\n"

    submitcommand ="/opt/condor/bin/condor_submit  "+outfilename_condor
    print submitcommand+"\n"
    
    if not istest:
        child   = os.popen(submitcommand)
    else:
        print "Not submitting into condor batch farm. This is only a test"
        

#________________________________________________________________    
if __name__ =='__main__':


    #import optionparse
    option,args = parse(__doc__)
    if not args and not option: exit()

    if not option.mc and (not option.cfg or not option.njobs or not option.list):
        print " you need to provide configuration file, list of files, number of jobs"
        optionparse.exit()
        
    ini_cfgfile = option.cfg
    
    template_fnames["cfg"]          =  ini_cfgfile

    if not option.mc:
        list_of_files = option.list
    else:
        Nevents = str(option.events)
                      
    number_of_jobs = option.njobs
    
    ini_run = 0
    fin_run = 0

    if option.initial:
        ini_run = int(option.initial)
    if option.final:
        fin_run = int(option.final)
    
    istest = option.test
    isMC   = option.mc
    isshort = option.short
    
    output_path = ""
    
    if option.output:
        output_path = option.output+"/"
    
    cfg_path  = output_path+ "cfg/"
    csh_path  = output_path+ "csh/"
    logs_path = output_path+ "logs/"

    subset = []
    njob = 0
    notrootfiles = 0
    
    #make directories
    if not os.path.exists(cfg_path):
        _mkdir(cfg_path)
    if not os.path.exists(csh_path):
        _mkdir(csh_path)
    if not os.path.exists(logs_path):
        _mkdir(logs_path)
    
    
    #filesperjob = 0
    #inputfile = []
    if not option.mc:
        inputfile = open(list_of_files)
        totalfiles = len(inputfile.readlines())
        inputfile.seek(0,0)

        for iline in inputfile:
            #print iline
            if iline.find(".root")==-1:
                notrootfiles += 1

        #print notrootfiles
        totalfiles -= notrootfiles
        inputfile.seek(0,0)

        filesperjob = float(totalfiles)/float(number_of_jobs)
        filesperjob = int(filesperjob)
            
        #print "Number of total root files = "
        #print totalfiles    
        
        for ifile in inputfile:
            #print "ifile: "+ifile
            ignoreline = 0

            if ifile.find(".root")==-1:
                ignoreline = 1
            
            if ignoreline==0:
                #print "count line, where len subset is " +str(len(subset)) +"/"+str(filesperjob)
                ifile = ifile.strip("'")
                ifile = ifile.strip('\',\n')
                #ifile = ifile.strip("'")
                if len(subset) == filesperjob:
                    njob = njob + 1
                    if fin_run == 0 and njob >= ini_run:
                        submit_jobs(njob,subset,ini_cfgfile,output_path)
                    elif fin_run > 0 and njob <= fin_run and njob >= ini_run:
                        submit_jobs(njob,subset,ini_cfgfile,output_path)
            
                    subset = []
                #print "add line: " + ifile
                subset.append(ifile)

        #print " len = " + str(len(subset))+" subset = "
        #print subset
        
        if len(subset)>0:
            njob = njob + 1
            if fin_run == 0 and njob >= ini_run:
                submit_jobs(njob,subset,ini_cfgfile,output_path)
            elif fin_run > 0 and njob <= fin_run and njob >= ini_run:
                submit_jobs(njob,subset,ini_cfgfile,output_path)
        
        #submit_jobs(njob,subset,ini_cfgfile,output_path)

    else:
        ijob = 1
        while ijob <= int(number_of_jobs):
            submit_jobs(ijob,subset,ini_cfgfile,output_path)
            ijob = ijob + 1


Revision:	1.1
Committed:	Tue Jul 14 19:08:21 2009 UTC (15 years, 9 months ago) by jengbou
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	HEAD
Log Message:	add file
#	Content
1	#!/usr/bin/env python
2	############################################################################
3	#
4	# Author: Francisco Yumiceva
5	# yumiceva@fnal.gov
6	#
7	# Fermilab, 2007
8	#
9	# 2009/07/14: Modification to prevent empty string for input fileNames
10	# Geng-yuan Jeng
11	############################################################################
12
13	"""
14	Submit jobs to condor. It will create directories in the output path
15	to store configuration files and log files.
16
17	e.g. submitToCondor.py -c ZprimeEventsReco_full.py -l datasets/zp1Tev.txt -n 10 -o /uscms_data/d1/yumiceva/CMSSW_1_3_6/TQAFAnalyzer/
18
19	where ZprimeEventsReco_full.cfg is the configuration file. The configuration file
20	needs to be modified a bit: (1) remove all input files in the PoolSource and add a
21	keyword {FILENAME} you need also the brackets. The filename of the output file should
22	be changed to another keyword called {OUTPUT_FILENAME}.
23
24	To Generate MC: In your cfg file you need to add also the keywords {SEED1},...,{SEED6} for
25	the random seeds which are going to be modified by the script using PYTHON random numbers,
26	you also need to change the output filename to {OUTPUT_FILENAME} All keywords need to have
27	brackets.
28
29	usage: %prog
30	-m, --mc : to generate MC. it will handle random numbers.
31	-e, --events = EVENTS: number of events to generate MC. If -m flag is turned on.
32	-c, --cfg = CFG: python configuration file
33	-l, --list = LIST: file with list of files (dataset)
34	-n, --njobs = NJOBS: number of jobs.
35	-o, --output = OUTPUT: output path
36	-i, --initial= INITIAL: number of initial job
37	-f, --final= FINAL: number of final job
38	-s, --short : for short jobs that need high priority
39	-t, --test : do not submit anything just show what I would do
40
41	Author: Francisco Yumiceva (yumiceva@fnal.gov)
42	"""
43
44
45	import os,sys
46	try:
47	import random
48	except:
49	# now fix PYTHONPATH for the damn cmslpc python configuration
50	atemp = os.getenv("PYTHONPATH")
51	atempl = atemp.split(":")
52	natemp = ""
53	for iatemp in atempl:
54	if iatemp.find("python2.3") == -1:
55	natemp = natemp + iatemp
56	print "\n FIX PYTHONPATH environment variable, try:\n"
57	print "setenv PYTHONPATH "+ natemp
58	sys.exit()
59	#os.putenv("PYTHONPATH",natemp)
60	# PYTHONPATH fixed
61
62	import string, re
63	from time import gmtime, localtime, strftime
64
65	#_______________OPTIONS__________________________________
66	import optparse
67	USAGE = re.compile(r'(?s)\susage: (.?)(\n[ \t]*\n\|$)')
68	def nonzero(self): # will become the nonzero method of optparse.Values
69	"True if options were given"
70	for v in self.__dict__.itervalues():
71	if v is not None: return True
72	return False
73
74	optparse.Values.__nonzero__ = nonzero # dynamically fix optparse.Values
75
76	class ParsingError(Exception): pass
77
78	optionstring=""
79
80	def exit(msg=""):
81	raise SystemExit(msg or optionstring.replace("%prog",sys.argv[0]))
82
83	def parse(docstring, arglist=None):
84	global optionstring
85	optionstring = docstring
86	match = USAGE.search(optionstring)
87	if not match: raise ParsingError("Cannot find the option string")
88	optlines = match.group(1).splitlines()
89	try:
90	p = optparse.OptionParser(optlines[0])
91	for line in optlines[1:]:
92	opt, help=line.split(':')[:2]
93	short,long=opt.split(',')[:2]
94	if '=' in opt:
95	action='store'
96	long=long.split('=')[0]
97	else:
98	action='store_true'
99	p.add_option(short.strip(),long.strip(),
100	action = action, help = help.strip())
101	except (IndexError,ValueError):
102	raise ParsingError("Cannot parse the option string correctly")
103	return p.parse_args(arglist)
104
105	#______________________________________________________________________
106
107	#
108	# path to this scripts and the used template file:
109	#
110	TemplatesPATH = os.environ['CMSSW_BASE']+"/src/submit2uaf/"
111	#
112	# path to the output that is created by this script
113	#
114	#main_path = "/uscmst1b_scratch/lpc1/cmsroc/yumiceva/CMSSW_1_2_0_pre3/src/RecoVertex/BeamSpotProducer/test/"
115	main_path = os.getcwd() + "/"
116
117	out_path = main_path+"condorjobs/"
118
119	scripts_path = main_path+"condorjobs/"
120	cfg_path = scripts_path+"cfg/"
121	#condor_path = scripts_path+"condor/"
122	csh_path = scripts_path+"csh/"
123	logs_path = scripts_path+"logs/"
124
125
126	istest = 0
127	isMC = 0
128	Nevents = "-1"
129	isshort = 0
130
131	#
132	# Path to the input/output data:
133	#
134
135	cfg_tags = ["{FILENAME}",
136	"{OUTPUT_FILENAME}"]
137
138	cfg_tags_mc = ["{SEED1}",
139	"{SEED2}",
140	"{SEED3}",
141	"{SEED4}",
142	"{SEED5}",
143	"{SEED6}",
144	"{OUTPUT_FILENAME}"]
145
146	scripts_tags = ["{PATHTOOUT}",
147	"{CONFIGFILE}",
148	"{OUTFILE}","{PATHTOBASE}","{OUTDIR}"]
149
150	condor_tags = ["{EXECUTABLE}",
151	"{OUTPUT}",
152	"{ERROR}",
153	"{LOG}",
154	"{NOTIFY}"]
155
156
157	template_fnames = {}
158	template_fnames["cfg"] = TemplatesPATH+"/reco_template.cfg"
159	template_fnames["condor"] = TemplatesPATH+"/template.condor"
160	template_fnames["csh"] = TemplatesPATH+"/runcms_new2.csh"
161
162
163	def _mkdir(newdir):
164	"""works the way a good mkdir should :)
165	- already exists, silently complete
166	- regular file in the way, raise an exception
167	- parent directory(ies) does not exist, make them as well
168	"""
169	if os.path.isdir(newdir):
170	pass
171	elif os.path.isfile(newdir):
172	raise OSError("a file with the same name as the desired " \
173	"dir, '%s', already exists." % newdir)
174	else:
175	head, tail = os.path.split(newdir)
176	if head and not os.path.isdir(head):
177	_mkdir(head)
178	if tail:
179	os.mkdir(newdir)
180
181	def change(infile,outfile,changearray,filearray):
182	fin = open(infile)
183	fout = open(outfile,"w")
184
185	#if line.find("untracked")!=-1 and line.find("PSet")!=-1 and line.find("maxEvents")!=-1:
186	# line = "untracked PSet maxEvents = {untracked int32 input = "+str(Nevents)+"}\n"
187
188	for line in fin.readlines():
189	for change in changearray:
190	#if change[0] == "{FILENAME}" and line.find(change[0])!=-1:
191	# line=line.replace(change[0] ,"")
192	# itmp=0
193	# for ifile in filearray:
194	# ifile = ifile.strip("\n")
195	# itmp = itmp + 1
196	# suffix = "\","
197	# if itmp == len(filearray):
198	# suffix = "\""
199	# fout.write("\""+ifile+suffix+"\n")
200
201	#else:
202
203	line=line.replace(change[0] ,change[1])
204
205	fout.write(line)
206
207	#
208	if outfile.find(".py")!=-1:
209	fout.write('''
210	process.maxEvents = cms.untracked.PSet(input = cms.untracked.int32(-1) )
211
212	process.source = cms.Source("PoolSource",
213	fileNames = cms.untracked.vstring(
214	''')
215	itmp=0
216	for ifile in filearray:
217	ifile = ifile.strip("\n")
218	itmp = itmp + 1
219	suffix = "\","
220	if itmp == len(filearray):
221	suffix = "\""
222	fout.write("\""+ifile+suffix+"\n")
223
224	fout.write(")\n")
225	fout.write(", duplicateCheckMode = cms.untracked.string('noDuplicateCheck')\n")
226	fout.write(")\n")
227
228
229
230	def submit_jobs(njob,array,ini_cfgfile,output_path):
231
232	cfgfile = ini_cfgfile.replace("_template","")
233
234	njobstr = str(njob)
235	while len(njobstr)<4:
236	njobstr = "0"+njobstr
237
238	outfilename_cfg = cfg_path + cfgfile.replace(".py","_"+njobstr+".py")
239	outfilename_root = output_path+ cfgfile.replace(".py","_"+njobstr+".root")
240	outfilename_log = logs_path + cfgfile.replace(".py","_"+njobstr+".log")
241	outfilename_csh = csh_path + cfgfile.replace(".py","_"+njobstr+".csh")
242	outfilename_condor= csh_path + cfgfile.replace(".py","_"+njobstr+".condor")
243
244	if os.path.isfile(outfilename_root):
245	print "Not submitting into condor batch farm since rootoutput already exists"
246	print "delete or rename: "+outfilename_root+"\n to resubmit"
247	sys.exit(main())
248
249	#
250	# First create config (.cfg) file:
251	#
252
253	changearray=[]
254
255	if isMC:
256	random.seed(njob)
257	changearray.append((cfg_tags_mc[0],str(random.randint(1,987654321) ) ))
258	changearray.append((cfg_tags_mc[1],str(random.randint(1,987654321) ) ))
259	changearray.append((cfg_tags_mc[2],str(random.randint(1,987654321) ) ))
260	changearray.append((cfg_tags_mc[3],str(random.randint(1,987654321) ) ))
261	changearray.append((cfg_tags_mc[4],str(random.randint(1,987654321) ) ))
262	changearray.append((cfg_tags_mc[5],str(random.randint(1,987654321) ) ))
263	changearray.append((cfg_tags_mc[6],outfilename_root))
264	change(template_fnames["cfg"],outfilename_cfg,changearray,array)
265	#print outfilename_cfg + " has been written.\n"
266
267	else:
268	changearray.append((cfg_tags[0],""))
269	changearray.append((cfg_tags[1],outfilename_root))
270	change(template_fnames["cfg"],outfilename_cfg,changearray,array)
271	#print outfilename_cfg + " has been written.\n"
272
273
274	#
275	# now create the script to process the file:
276	#
277	changearray=[]
278	changearray.append((scripts_tags[0],output_path+"/"))
279	changearray.append((scripts_tags[1],outfilename_cfg))
280	changearray.append((scripts_tags[2],outfilename_root))
281	changearray.append((scripts_tags[3],os.getenv("CMSSW_BASE")+"/src" ))
282	changearray.append((scripts_tags[4],output_path))
283	change(template_fnames["csh"],outfilename_csh,changearray,0)
284	os.chmod(outfilename_csh,0775)
285	#print outfilename_csh + " has been written.\n"
286
287	#
288	# finally create the condor job description file:
289	#
290	stdout = outfilename_log.replace(".log",".stdout")
291	stderr = outfilename_log.replace(".log",".stderr")
292	logfiles = outfilename_log.replace(".log",".log")
293	changearray =[]
294	changearray.append((condor_tags[0],outfilename_csh))
295	changearray.append((condor_tags[1],stdout))
296	changearray.append((condor_tags[2],stderr))
297	changearray.append((condor_tags[3],logfiles))
298	changearray.append((condor_tags[4],os.getenv("USER")+"@fnal.gov"))
299	change(template_fnames["condor"],outfilename_condor,changearray,0)
300	if isshort:
301	fout = open(outfilename_condor,"a")
302	fout.write("+LENGTH=\"SHORT\"\n")
303	fout.close()
304
305	#print outfilename_condor + " has been written.\n"
306
307	submitcommand ="/opt/condor/bin/condor_submit "+outfilename_condor
308	print submitcommand+"\n"
309
310	if not istest:
311	child = os.popen(submitcommand)
312	else:
313	print "Not submitting into condor batch farm. This is only a test"
314
315
316
317	#________________________________________________________________
318	if __name__ =='__main__':
319
320
321	#import optionparse
322	option,args = parse(__doc__)
323	if not args and not option: exit()
324
325	if not option.mc and (not option.cfg or not option.njobs or not option.list):
326	print " you need to provide configuration file, list of files, number of jobs"
327	optionparse.exit()
328
329	ini_cfgfile = option.cfg
330
331	template_fnames["cfg"] = ini_cfgfile
332
333	if not option.mc:
334	list_of_files = option.list
335	else:
336	Nevents = str(option.events)
337
338	number_of_jobs = option.njobs
339
340	ini_run = 0
341	fin_run = 0
342
343	if option.initial:
344	ini_run = int(option.initial)
345	if option.final:
346	fin_run = int(option.final)
347
348	istest = option.test
349	isMC = option.mc
350	isshort = option.short
351
352	output_path = ""
353
354	if option.output:
355	output_path = option.output+"/"
356
357	cfg_path = output_path+ "cfg/"
358	csh_path = output_path+ "csh/"
359	logs_path = output_path+ "logs/"
360
361	subset = []
362	njob = 0
363	notrootfiles = 0
364
365	#make directories
366	if not os.path.exists(cfg_path):
367	_mkdir(cfg_path)
368	if not os.path.exists(csh_path):
369	_mkdir(csh_path)
370	if not os.path.exists(logs_path):
371	_mkdir(logs_path)
372
373
374	#filesperjob = 0
375	#inputfile = []
376	if not option.mc:
377	inputfile = open(list_of_files)
378	totalfiles = len(inputfile.readlines())
379	inputfile.seek(0,0)
380
381	for iline in inputfile:
382	#print iline
383	if iline.find(".root")==-1:
384	notrootfiles += 1
385
386	#print notrootfiles
387	totalfiles -= notrootfiles
388	inputfile.seek(0,0)
389
390	filesperjob = float(totalfiles)/float(number_of_jobs)
391	filesperjob = int(filesperjob)
392
393	#print "Number of total root files = "
394	#print totalfiles
395
396	for ifile in inputfile:
397	#print "ifile: "+ifile
398	ignoreline = 0
399
400	if ifile.find(".root")==-1:
401	ignoreline = 1
402
403	if ignoreline==0:
404	#print "count line, where len subset is " +str(len(subset)) +"/"+str(filesperjob)
405	ifile = ifile.strip("'")
406	ifile = ifile.strip('\',\n')
407	#ifile = ifile.strip("'")
408	if len(subset) == filesperjob:
409	njob = njob + 1
410	if fin_run == 0 and njob >= ini_run:
411	submit_jobs(njob,subset,ini_cfgfile,output_path)
412	elif fin_run > 0 and njob <= fin_run and njob >= ini_run:
413	submit_jobs(njob,subset,ini_cfgfile,output_path)
414
415	subset = []
416	#print "add line: " + ifile
417	subset.append(ifile)
418
419	#print " len = " + str(len(subset))+" subset = "
420	#print subset
421
422	if len(subset)>0:
423	njob = njob + 1
424	if fin_run == 0 and njob >= ini_run:
425	submit_jobs(njob,subset,ini_cfgfile,output_path)
426	elif fin_run > 0 and njob <= fin_run and njob >= ini_run:
427	submit_jobs(njob,subset,ini_cfgfile,output_path)
428
429	#submit_jobs(njob,subset,ini_cfgfile,output_path)
430
431	else:
432	ijob = 1
433	while ijob <= int(number_of_jobs):
434	submit_jobs(ijob,subset,ini_cfgfile,output_path)
435	ijob = ijob + 1
436
437
438