CRAB/python/multicrab.py

#!/usr/bin/env python
import sys, os, time, string, shutil
from crab_util import *
from crab import *
import common

###########################################################################
class MultiCrab:
    def __init__(self, opts):
        self.prog_name='multicrab'
        # Configuration file
        self.cfg_fname = None
        # Continuation flag
        self.flag_continue = 0
        self.continue_dir = None
        self.processContinueOption_(opts)

        self.processIniFile_(opts)
        # configuration
        self.opts=opts

        if not self.flag_continue:
            self.createWorkSpace()

        print self.prog_name + ' running on ' +  time.ctime(time.time())
        print '  working directory   ' + self.continue_dir


    def processContinueOption_(self,opts):

        # Look for the '-continue' option.

        for opt in opts.keys():
            if ( opt in ('-continue','-c') ):
                self.flag_continue = 1
                val = opts[opt]
                if val:
                    if val[0] == '/': self.continue_dir = val     # abs path
                    else: self.continue_dir = os.getcwd() + '/' + val      # rel path
                    pass
                break
            pass

        # Look for actions which has sense only with '-continue'

        if "-create" not in opts.keys() :
            self.flag_continue = 1

        if not self.flag_continue: return

        if not self.continue_dir:
            prefix = self.prog_name + '_'
            self.continue_dir = findLastWorkDir(prefix)
            pass

        if not self.continue_dir:
            raise CrabException('Cannot find last working directory.')

        if not os.path.exists(self.continue_dir):
            msg = 'Cannot continue because the working directory <'
            msg += self.continue_dir
            msg += '> does not exist.'
            raise CrabException(msg)

        return

    def createWorkSpace(self):
        # create WorkingDir for Multicrab
        import os
        if not self.continue_dir:
            prefix = self.prog_name + '_'
            self.continue_dir = findLastWorkDir(prefix)
            pass
        # if 'MULTICRAB.working_dir' in self.opts.keys():    
        #     self.continue_dir = os.path.abspath(self.opts['MULTICRAB.working_dir'])
        if self.ui_working_dir:
            self.continue_dir = os.path.abspath(self.ui_working_dir)
        else:
            current_time = time.strftime('%y%m%d_%H%M%S', time.localtime(time.time()))
            self.continue_dir = os.getcwd() + '/' + self.prog_name + '_' + current_time

        if self.continue_dir and not os.path.exists(self.continue_dir):
            try:
                os.mkdir(self.continue_dir)
            except OSError:
                msg = 'Cannot create '+str(self.continue_dir) +' directory.\n'
                raise CrabException(msg)
            pass
        else:
            msg = 'Directory '+str(self.continue_dir) +' already exist.\n'
            raise CrabException(msg)

        os.putenv("MULTICRAB_WORKDIR",self.continue_dir)
        shutil.copyfile(self.cfg_fname,self.continue_dir+'/multicrab.cfg')
        
        return
        
    def processIniFile_(self, opts):
        """
        Processes a configuration INI-file.
        """

        # Extract cfg-file name from the cmd-line options.

        for opt in opts.keys():
            if ( opt == '-cfg' ):
                if self.flag_continue:
                    raise CrabException('-continue and -cfg cannot coexist.')
                if opts[opt] :
                    self.cfg_fname = opts[opt]
                    del opts[opt] # do not pass cfg further on
                else : processHelpOptions()
                pass
            pass

        # Set default cfg-fname

        if self.cfg_fname == None:
            if self.flag_continue:
                self.cfg_fname = self.continue_dir + '/multicrab.cfg'
            else:
                self.cfg_fname = 'multicrab.cfg'
                pass
            pass

        # Load cfg-file

        cfg_params = {}
        if self.cfg_fname != None:
            if os.path.exists(self.cfg_fname):
                cfg_params = self.loadMultiConfig(self.cfg_fname)
                pass
            else:
                msg = 'cfg-file '+self.cfg_fname+' not found.'
                raise CrabException(msg)
                pass
            pass

        # process the [CRAB] section

        lhp = len('MULTICRAB.')
        for k in cfg_params.keys():
            if len(k) >= lhp and k[:lhp] == 'MULTICRAB.':
                opt = '-'+k[lhp:]
                if len(opt) >= 3 and opt[:3] == '-__': continue
                if opt not in opts.keys():
                    opts[opt] = cfg_params[k]
                    pass
                pass
            pass

        self.cfg_params_dataset = {}
        common_opts = {}
        # first get common sections
        for sec in cfg_params:
            if sec in ['MULTICRAB']:
                if 'cfg' in cfg_params[sec]:
                    common_opts['cfg']=cfg_params[sec]['cfg']
                continue
            if sec in ['COMMON']:
                common_opts.update(cfg_params[sec])
                continue
            pass

        crab_cfg='crab.cfg'
        if common_opts.has_key('cfg') : crab_cfg=common_opts['cfg']

        # then Dataset's specific
        for sec in cfg_params:
            if sec in ['MULTICRAB', 'COMMON']: continue
            self.cfg_params_dataset[sec]=cfg_params[sec]
            # add common to all dataset
            for key in common_opts:
                self.cfg_params_dataset[sec][key]=common_opts[key]
            pass

        # read crab.cfg file and search for storage_path
        crab_cfg_params = loadConfig(crab_cfg,{})
        # also USER.ui_working_dir USER.outputdir and USER.logdir need special treatment
        if cfg_params.has_key("COMMON"):
            self.user_remote_dir = cfg_params["COMMON"].get("user.user_remote_dir", crab_cfg_params.get("USER.user_remote_dir",None))
            self.outputdir = cfg_params["COMMON"].get("user.outputdir", crab_cfg_params.get("USER.outputdir",None))
            self.logdir = cfg_params["COMMON"].get("user.logdir", crab_cfg_params.get("USER.logdir",None))
            self.ui_working_dir = cfg_params["COMMON"].get("user.ui_working_dir", crab_cfg_params.get("USER.ui_working_dir",None))
            self.publish_data_name = cfg_params["COMMON"].get("user.publish_data_name", crab_cfg_params.get("USER.publish_data_name",None))
        else:
            self.user_remote_dir = crab_cfg_params.get("USER.user_remote_dir",None)
            self.outputdir = crab_cfg_params.get("USER.outputdir",None)
            self.logdir = crab_cfg_params.get("USER.logdir",None)
            self.ui_working_dir = crab_cfg_params.get("USER.ui_working_dir",None)
            self.publish_data_name = crab_cfg_params.get("USER.publish_data_name",None)

        return

    def loadMultiConfig(self, file):
        """
        returns a dictionary with keys of the form
        <section>.<option> and the corresponding values
        """
        config={}
        cp = ConfigParser.ConfigParser()
        cp.read(file)
        for sec in cp.sections():
            # print 'Section',sec
            config[sec]={}
            for opt in cp.options(sec):
                #print 'config['+sec+'.'+opt+'] = '+string.strip(cp.get(sec,opt))
                config[sec][opt] = string.strip(cp.get(sec,opt))
        return config

    def run(self):
        #run crabs
        runFileName = self.continue_dir+'/multicrab.exe'
        runFile = open(runFileName,"w")
        for sec in self.cfg_params_dataset:
            options={}
            if self.flag_continue:
                options['-c']=sec
            # DatasetName to be used
            options['-USER.ui_working_dir']=sec
            # options from multicrab.cfg
            for opt in self.cfg_params_dataset[sec]:
                tmp = "-"+str(opt)
                if len(opt.split("."))==2:
                    tmp="-"+string.upper(opt.split(".")[0])+"."+opt.split(".")[1]
                
                options[tmp]=self.cfg_params_dataset[sec][opt]

            # check if user_remote_dir is set in multicrab.cfg
            # protect against no user_remote_dir
            self.user_remote_dir =self.cfg_params_dataset[sec].get("user.user_remote_dir",self.user_remote_dir)
            if not self.user_remote_dir:
                self.user_remote_dir = "./"
            # add section to storage_path if exist in crab.cfg
            if not self.cfg_params_dataset.has_key("USER.user_remote_dir") and self.user_remote_dir:
                options["-USER.user_remote_dir"]=self.user_remote_dir+"/"+sec
            # print options["-USER.user_remote_dir"]
            # also for ui_working_dir
            if not self.cfg_params_dataset.has_key("USER.ui_working_dir") and self.ui_working_dir:
                options["-USER.ui_working_dir"]=self.ui_working_dir+"/"+sec
            # if ui_working_dir is set, change -c dir accordnigly
                if self.flag_continue:
                    options['-c']=self.ui_working_dir+"/"+sec

            # also for logDir
            if not self.cfg_params_dataset.has_key("USER.logdir") and self.logdir:
                options["-USER.logdir"]=self.logdir+"/"+sec
            # also for outputdir
            if not self.cfg_params_dataset.has_key("USER.outputdir") and self.outputdir:
                options["-USER.outputdir"]=self.outputdir+"/"+sec
            # also for publish_data_name
            if not self.cfg_params_dataset.has_key("USER.publish_data_name") and self.publish_data_name:
                options["-USER.publish_data_name"]=self.publish_data_name+"_"+sec

            # Input options (command)
            for opt in self.opts:
                if opt != '-c':
                    options[opt]=self.opts[opt]
                # options[opt]=self.opts[opt]
                    if self.flag_continue and options.has_key("-cfg"):
                        del options['-cfg']
                    pass

            # write crab command to be executed later...
            cmd='crab '
            for o in options:
                if options[o]==None:
                    cmd+=str(o)+' '
                else:
                    options[o] = ''.join(options[o].split())
                    cmd+=str(o)+'='+str(options[o])+' '
                pass
            cmd+="\n"
            #print cmd

            runFile.write(cmd)

            # SL this does not work for complex, multi include pset.py 

            # crab = Crab()
            # try:
            #     crab.initialize_(options)
            #     crab.run()
            #     del crab
            #     print 'Log file is %s%s.log'%(common.work_space.logDir(),common.prog_name)  
            #     print '\n##############################  E N D  ####################################\n'
            # except CrabException, e:
            #     del crab
            #     print '\n' + common.prog_name + ': ' + str(e) + '\n'
            #     pass
            # pass
            # if (common.logger): common.logger.delete()
        pass
        return self.continue_dir
        

###########################################################################
if __name__ == '__main__':
    ## Get rid of some useless warning
    try:
        import warnings
        warnings.simplefilter("ignore", RuntimeWarning)
    except ImportError:
        pass # too bad, you'll get the warning

    # Parse command-line options and create a dictionary with
    # key-value pairs.
    options = parseOptions(sys.argv[1:])

    # Process "help" options, such as '-help', '-version'
    if processHelpOptions(options) : sys.exit(0)

    # Create, initialize, and run a Crab object
    try:
        multicrab = MultiCrab(options)
        continue_dir = multicrab.run()
        import os
        sys.exit(continue_dir)
    except CrabException, e:
        print '\n' + common.prog_name + ': ' + str(e) + '\n'

    pass
Revision:	1.29
Committed:	Wed Oct 26 11:43:40 2011 UTC (13 years, 6 months ago) by slacapra
Content type:	text/x-python
Branch:	MAIN
Changes since 1.28:	+27 -9 lines
Log Message:	Fix for publish and use of -cfg with multicrab
#	User	Rev	Content
1	slacapra	1.1	#!/usr/bin/env python
2			import sys, os, time, string, shutil
3			from crab_util import *
4			from crab import *
5			import common
6
7			###########################################################################
8			class MultiCrab:
9			def __init__(self, opts):
10			self.prog_name='multicrab'
11			# Configuration file
12			self.cfg_fname = None
13			# Continuation flag
14			self.flag_continue = 0
15			self.continue_dir = None
16			self.processContinueOption_(opts)
17
18			self.processIniFile_(opts)
19			# configuration
20			self.opts=opts
21
22			if not self.flag_continue:
23			self.createWorkSpace()
24
25			print self.prog_name + ' running on ' + time.ctime(time.time())
26			print ' working directory ' + self.continue_dir
27
28
29			def processContinueOption_(self,opts):
30
31			# Look for the '-continue' option.
32
33			for opt in opts.keys():
34			if ( opt in ('-continue','-c') ):
35			self.flag_continue = 1
36			val = opts[opt]
37			if val:
38			if val[0] == '/': self.continue_dir = val # abs path
39			else: self.continue_dir = os.getcwd() + '/' + val # rel path
40			pass
41			break
42			pass
43
44			# Look for actions which has sense only with '-continue'
45
46			if "-create" not in opts.keys() :
47			self.flag_continue = 1
48
49			if not self.flag_continue: return
50
51			if not self.continue_dir:
52			prefix = self.prog_name + '_'
53			self.continue_dir = findLastWorkDir(prefix)
54			pass
55
56			if not self.continue_dir:
57			raise CrabException('Cannot find last working directory.')
58
59			if not os.path.exists(self.continue_dir):
60			msg = 'Cannot continue because the working directory <'
61			msg += self.continue_dir
62			msg += '> does not exist.'
63			raise CrabException(msg)
64
65			return
66
67			def createWorkSpace(self):
68			# create WorkingDir for Multicrab
69	slacapra	1.29	import os
70			if not self.continue_dir:
71			prefix = self.prog_name + '_'
72			self.continue_dir = findLastWorkDir(prefix)
73			pass
74			# if 'MULTICRAB.working_dir' in self.opts.keys():
75			# self.continue_dir = os.path.abspath(self.opts['MULTICRAB.working_dir'])
76			if self.ui_working_dir:
77			self.continue_dir = os.path.abspath(self.ui_working_dir)
78	slacapra	1.1	else:
79			current_time = time.strftime('%y%m%d_%H%M%S', time.localtime(time.time()))
80			self.continue_dir = os.getcwd() + '/' + self.prog_name + '_' + current_time
81
82			if self.continue_dir and not os.path.exists(self.continue_dir):
83			try:
84			os.mkdir(self.continue_dir)
85			except OSError:
86			msg = 'Cannot create '+str(self.continue_dir) +' directory.\n'
87			raise CrabException(msg)
88			pass
89			else:
90			msg = 'Directory '+str(self.continue_dir) +' already exist.\n'
91			raise CrabException(msg)
92
93	slacapra	1.29	os.putenv("MULTICRAB_WORKDIR",self.continue_dir)
94			shutil.copyfile(self.cfg_fname,self.continue_dir+'/multicrab.cfg')
95	slacapra	1.1
96			return
97
98			def processIniFile_(self, opts):
99			"""
100			Processes a configuration INI-file.
101			"""
102
103			# Extract cfg-file name from the cmd-line options.
104
105			for opt in opts.keys():
106			if ( opt == '-cfg' ):
107			if self.flag_continue:
108			raise CrabException('-continue and -cfg cannot coexist.')
109	slacapra	1.8	if opts[opt] :
110			self.cfg_fname = opts[opt]
111	slacapra	1.11	del opts[opt] # do not pass cfg further on
112	slacapra	1.1	else : processHelpOptions()
113			pass
114			pass
115
116			# Set default cfg-fname
117
118			if self.cfg_fname == None:
119			if self.flag_continue:
120			self.cfg_fname = self.continue_dir + '/multicrab.cfg'
121			else:
122			self.cfg_fname = 'multicrab.cfg'
123			pass
124			pass
125
126			# Load cfg-file
127
128	slacapra	1.7	cfg_params = {}
129	slacapra	1.1	if self.cfg_fname != None:
130			if os.path.exists(self.cfg_fname):
131	slacapra	1.7	cfg_params = self.loadMultiConfig(self.cfg_fname)
132	slacapra	1.1	pass
133			else:
134			msg = 'cfg-file '+self.cfg_fname+' not found.'
135			raise CrabException(msg)
136			pass
137			pass
138
139			# process the [CRAB] section
140
141			lhp = len('MULTICRAB.')
142	slacapra	1.7	for k in cfg_params.keys():
143	slacapra	1.1	if len(k) >= lhp and k[:lhp] == 'MULTICRAB.':
144			opt = '-'+k[lhp:]
145			if len(opt) >= 3 and opt[:3] == '-__': continue
146			if opt not in opts.keys():
147	slacapra	1.7	opts[opt] = cfg_params[k]
148	slacapra	1.1	pass
149			pass
150			pass
151
152			self.cfg_params_dataset = {}
153	slacapra	1.3	common_opts = {}
154	slacapra	1.1	# first get common sections
155	slacapra	1.7	for sec in cfg_params:
156	slacapra	1.1	if sec in ['MULTICRAB']:
157	slacapra	1.8	if 'cfg' in cfg_params[sec]:
158			common_opts['cfg']=cfg_params[sec]['cfg']
159	slacapra	1.1	continue
160			if sec in ['COMMON']:
161	slacapra	1.8	common_opts.update(cfg_params[sec])
162	slacapra	1.1	continue
163			pass
164
165	slacapra	1.8	crab_cfg='crab.cfg'
166			if common_opts.has_key('cfg') : crab_cfg=common_opts['cfg']
167
168	slacapra	1.1	# then Dataset's specific
169	slacapra	1.7	for sec in cfg_params:
170	slacapra	1.1	if sec in ['MULTICRAB', 'COMMON']: continue
171	slacapra	1.7	self.cfg_params_dataset[sec]=cfg_params[sec]
172	slacapra	1.1	# add common to all dataset
173	slacapra	1.3	for key in common_opts:
174			self.cfg_params_dataset[sec][key]=common_opts[key]
175	slacapra	1.1	pass
176
177	slacapra	1.2	# read crab.cfg file and search for storage_path
178	slacapra	1.8	crab_cfg_params = loadConfig(crab_cfg,{})
179	slacapra	1.16	# also USER.ui_working_dir USER.outputdir and USER.logdir need special treatment
180	slacapra	1.7	if cfg_params.has_key("COMMON"):
181			self.user_remote_dir = cfg_params["COMMON"].get("user.user_remote_dir", crab_cfg_params.get("USER.user_remote_dir",None))
182	slacapra	1.16	self.outputdir = cfg_params["COMMON"].get("user.outputdir", crab_cfg_params.get("USER.outputdir",None))
183			self.logdir = cfg_params["COMMON"].get("user.logdir", crab_cfg_params.get("USER.logdir",None))
184			self.ui_working_dir = cfg_params["COMMON"].get("user.ui_working_dir", crab_cfg_params.get("USER.ui_working_dir",None))
185	slacapra	1.29	self.publish_data_name = cfg_params["COMMON"].get("user.publish_data_name", crab_cfg_params.get("USER.publish_data_name",None))
186	slacapra	1.6	else:
187	slacapra	1.7	self.user_remote_dir = crab_cfg_params.get("USER.user_remote_dir",None)
188	slacapra	1.16	self.outputdir = crab_cfg_params.get("USER.outputdir",None)
189			self.logdir = crab_cfg_params.get("USER.logdir",None)
190			self.ui_working_dir = crab_cfg_params.get("USER.ui_working_dir",None)
191	slacapra	1.29	self.publish_data_name = crab_cfg_params.get("USER.publish_data_name",None)
192	slacapra	1.16
193	slacapra	1.1	return
194
195			def loadMultiConfig(self, file):
196			"""
197			returns a dictionary with keys of the form
198			<section>.<option> and the corresponding values
199			"""
200			config={}
201			cp = ConfigParser.ConfigParser()
202			cp.read(file)
203			for sec in cp.sections():
204			# print 'Section',sec
205			config[sec]={}
206			for opt in cp.options(sec):
207			#print 'config['+sec+'.'+opt+'] = '+string.strip(cp.get(sec,opt))
208			config[sec][opt] = string.strip(cp.get(sec,opt))
209			return config
210
211			def run(self):
212			#run crabs
213	slacapra	1.29	runFileName = self.continue_dir+'/multicrab.exe'
214	slacapra	1.16	runFile = open(runFileName,"w")
215	slacapra	1.1	for sec in self.cfg_params_dataset:
216			options={}
217			if self.flag_continue:
218			options['-c']=sec
219			# DatasetName to be used
220			options['-USER.ui_working_dir']=sec
221			# options from multicrab.cfg
222			for opt in self.cfg_params_dataset[sec]:
223	slacapra	1.8	tmp = "-"+str(opt)
224			if len(opt.split("."))==2:
225			tmp="-"+string.upper(opt.split(".")[0])+"."+opt.split(".")[1]
226
227	slacapra	1.1	options[tmp]=self.cfg_params_dataset[sec][opt]
228	slacapra	1.16
229	slacapra	1.23	# check if user_remote_dir is set in multicrab.cfg
230	spiga	1.24	# protect against no user_remote_dir
231	slacapra	1.26	self.user_remote_dir =self.cfg_params_dataset[sec].get("user.user_remote_dir",self.user_remote_dir)
232			if not self.user_remote_dir:
233	slacapra	1.22	self.user_remote_dir = "./"
234	slacapra	1.2	# add section to storage_path if exist in crab.cfg
235	slacapra	1.6	if not self.cfg_params_dataset.has_key("USER.user_remote_dir") and self.user_remote_dir:
236	slacapra	1.5	options["-USER.user_remote_dir"]=self.user_remote_dir+"/"+sec
237	slacapra	1.27	# print options["-USER.user_remote_dir"]
238	slacapra	1.16	# also for ui_working_dir
239			if not self.cfg_params_dataset.has_key("USER.ui_working_dir") and self.ui_working_dir:
240			options["-USER.ui_working_dir"]=self.ui_working_dir+"/"+sec
241	slacapra	1.25	# if ui_working_dir is set, change -c dir accordnigly
242	slacapra	1.28	if self.flag_continue:
243			options['-c']=self.ui_working_dir+"/"+sec
244	slacapra	1.25
245	slacapra	1.16	# also for logDir
246			if not self.cfg_params_dataset.has_key("USER.logdir") and self.logdir:
247			options["-USER.logdir"]=self.logdir+"/"+sec
248			# also for outputdir
249			if not self.cfg_params_dataset.has_key("USER.outputdir") and self.outputdir:
250			options["-USER.outputdir"]=self.outputdir+"/"+sec
251	slacapra	1.29	# also for publish_data_name
252			if not self.cfg_params_dataset.has_key("USER.publish_data_name") and self.publish_data_name:
253			options["-USER.publish_data_name"]=self.publish_data_name+"_"+sec
254	slacapra	1.16
255	slacapra	1.1	# Input options (command)
256			for opt in self.opts:
257	slacapra	1.29	if opt != '-c':
258			options[opt]=self.opts[opt]
259			# options[opt]=self.opts[opt]
260			if self.flag_continue and options.has_key("-cfg"):
261			del options['-cfg']
262			pass
263	slacapra	1.20
264			# write crab command to be executed later...
265			cmd='crab '
266			for o in options:
267			if options[o]==None:
268			cmd+=str(o)+' '
269			else:
270			options[o] = ''.join(options[o].split())
271			cmd+=str(o)+'='+str(options[o])+' '
272			pass
273			cmd+="\n"
274			#print cmd
275
276			runFile.write(cmd)
277
278			# SL this does not work for complex, multi include pset.py
279
280			# crab = Crab()
281			# try:
282			# crab.initialize_(options)
283			# crab.run()
284			# del crab
285			# print 'Log file is %s%s.log'%(common.work_space.logDir(),common.prog_name)
286			# print '\n############################## E N D ####################################\n'
287			# except CrabException, e:
288			# del crab
289			# print '\n' + common.prog_name + ': ' + str(e) + '\n'
290			# pass
291			# pass
292			# if (common.logger): common.logger.delete()
293	slacapra	1.1	pass
294	slacapra	1.29	return self.continue_dir
295	slacapra	1.1
296
297			###########################################################################
298			if __name__ == '__main__':
299			## Get rid of some useless warning
300			try:
301			import warnings
302			warnings.simplefilter("ignore", RuntimeWarning)
303			except ImportError:
304			pass # too bad, you'll get the warning
305
306			# Parse command-line options and create a dictionary with
307			# key-value pairs.
308			options = parseOptions(sys.argv[1:])
309
310			# Process "help" options, such as '-help', '-version'
311			if processHelpOptions(options) : sys.exit(0)
312
313			# Create, initialize, and run a Crab object
314			try:
315			multicrab = MultiCrab(options)
316	slacapra	1.29	continue_dir = multicrab.run()
317			import os
318			sys.exit(continue_dir)
319	slacapra	1.1	except CrabException, e:
320			print '\n' + common.prog_name + ': ' + str(e) + '\n'
321
322			pass