CRAB/python/DataDiscovery.py

#!/usr/bin/env python
import exceptions
import DBSAPI.dbsApi
from DBSAPI.dbsApiException import * 
import common
from crab_util import *


# #######################################
class DBSError(exceptions.Exception):
    def __init__(self, errorName, errorMessage):
        args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
        exceptions.Exception.__init__(self, args)
        pass
    
    def getErrorMessage(self):
        """ Return error message """
        return "%s" % (self.args)

# #######################################
class DBSInvalidDataTierError(exceptions.Exception):
    def __init__(self, errorName, errorMessage):
        args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
        exceptions.Exception.__init__(self, args)
        pass
    
    def getErrorMessage(self):
        """ Return error message """
        return "%s" % (self.args)

# #######################################
class DBSInfoError:
    def __init__(self, url):
        print '\nERROR accessing DBS url : '+url+'\n'
        pass

# ####################################
class DataDiscoveryError(exceptions.Exception):
    def __init__(self, errorMessage):
        self.args=errorMessage
        exceptions.Exception.__init__(self, self.args)
        pass

    def getErrorMessage(self):
        """ Return exception error """
        return "%s" % (self.args)

# ####################################
class NotExistingDatasetError(exceptions.Exception):
    def __init__(self, errorMessage):
        self.args=errorMessage
        exceptions.Exception.__init__(self, self.args)
        pass

    def getErrorMessage(self):
        """ Return exception error """
        return "%s" % (self.args)

# ####################################
class NoDataTierinProvenanceError(exceptions.Exception):
    def __init__(self, errorMessage):
        self.args=errorMessage
        exceptions.Exception.__init__(self, self.args)
        pass

    def getErrorMessage(self):
        """ Return exception error """
        return "%s" % (self.args)

# ####################################
# class to find and extact info from published data
class DataDiscovery:
    def __init__(self, datasetPath, cfg_params):

        #       Attributes
        self.datasetPath = datasetPath
        self.cfg_params = cfg_params

        self.eventsPerBlock = {}  # DBS output: map fileblocks-events for collection
        self.eventsPerFile = {}   # DBS output: map files-events
        self.blocksinfo = {}      # DBS output: map fileblocks-files 
        self.maxEvents = 0        # DBS output: max events

# ####################################
    def fetchDBSInfo(self):
        """
        Contact DBS
        """

        ## get DBS URL
        dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
        if (self.cfg_params.has_key('CMSSW.dbs_url')):
            dbs_url=self.cfg_params['CMSSW.dbs_url']

        common.logger.debug(3,"Accessing DBS at: "+dbs_url)

        ## check if runs are selected
        runselection = []
        if (self.cfg_params.has_key('CMSSW.runselection')):
            runselection = parseRange2(self.cfg_params['CMSSW.runselection'])

        common.logger.debug(6,"runselection is: %s"%runselection)
        ## service API
        args = {}
        args['url']     = dbs_url
        args['level']   = 'CRITICAL'

        api = DBSAPI.dbsApi.DbsApi(args)
        try:
            if len(runselection) <= 0 :
                files = api.listDatasetFiles(self.datasetPath)
            else :
                files=[]
                for arun in runselection:
                    try:
                        filesinrun = api.listFiles(path=self.datasetPath, details=True,runNumber=arun)
                        files.extend(filesinrun)
                    except:
                        msg="WARNING: problem extracting info from DBS for run %s "%arun
                        common.logger.message(msg)
                        pass

        except DbsBadRequest, msg:
            raise DataDiscoveryError(msg)
        except DBSError, msg:
            raise DataDiscoveryError(msg)

        # parse files and fill arrays
        for file in files :
            filename = file['LogicalFileName']
            if filename.find('.dat') < 0 :
                fileblock = file['Block']['Name']
                events    = file['NumberOfEvents']
                # number of events per block
                if fileblock in self.eventsPerBlock.keys() :
                    self.eventsPerBlock[fileblock] += events
                else :
                    self.eventsPerBlock[fileblock] = events
                # number of events per file
                self.eventsPerFile[filename] = events

                # number of events per block
                if fileblock in self.blocksinfo.keys() :
                    self.blocksinfo[fileblock].append(filename)
                else :
                    self.blocksinfo[fileblock] = [filename]

                # total number of events
                self.maxEvents += events

        for block in self.eventsPerBlock.keys() :
            common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))

        if len(self.eventsPerBlock) <= 0:
            raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
                                            + " dataset path variables in crab.cfg")
                                            % self.datasetPath)


# #################################################
    def getMaxEvents(self):
        """
        max events 
        """
        return self.maxEvents

# #################################################
    def getEventsPerBlock(self):
        """
        list the event collections structure by fileblock 
        """
        return self.eventsPerBlock

# #################################################
    def getEventsPerFile(self):
        """
        list the event collections structure by file 
        """
        return self.eventsPerFile

# #################################################
    def getFiles(self):
        """
        return files grouped by fileblock 
        """
        return self.blocksinfo        

########################################################################
Revision:	1.20
Committed:	Fri Jan 11 22:11:55 2008 UTC (17 years, 3 months ago) by afanfani
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_2_2_pre2, CRAB_2_2_2_pre1, CRAB_2_2_1, CRAB_2_2_1_pre6, CRAB_2_2_1_pre5, CRAB_2_2_1_pre4, PRODCOMMON_0_10_7_testCS2, CRAB_2_2_1_pre3, CRAB_2_2_1_pre2, CRAB_2_2_1_pre1, CRAB_2_2_0, CRAB_2_2_0_pre21, CRAB_2_2_0_pre19, CRAB_2_2_0_pre18, CRAB_2_2_0_pre17, CRAB_2_2_0_pre16, CRAB_2_2_0_pre15, CRAB_2_2_0_pre13, CRAB_2_2_0_pre12, CRAB_2_2_0_pre11, CRAB_2_2_0_pre10, bp_osg_bdii, CRAB_2_2_0_pre9, CRAB_2_2_0_pre8, CRAB_2_2_0_pre7, CRAB_2_1_2, CRAB_2_2_0_pre5, CRAB_2_1_2_pre2, CRAB_2_1_2_pre1, CRAB_2_2_0_pre4, CRAB_2_2_0_pre2, CRAB_2_1_1, CRAB_2_1_1_pre3, CRAB_2_2_0_pre1, CRAB_2_1_1_pre1, CRAB_2_1_0, CRAB_2_1_0_pre6, CRAB_2_1_0_pre5, CRAB_2_1_0_pre4, CRAB_2_1_0_pre3
Branch point for:	osg_bdii, CRAB_2_1_2_br, CRAB_2_1_1_pre2
Changes since 1.19:	+25 -28 lines
Log Message:	if runselection is provided switch to use listFiles DBS API with run option since it's much faster.
#	User	Rev	Content
1	gutsche	1.6	#!/usr/bin/env python
2	slacapra	1.18	import exceptions
3			import DBSAPI.dbsApi
4			from DBSAPI.dbsApiException import *
5			import common
6			from crab_util import *
7	afanfani	1.1
8	afanfani	1.3
9	slacapra	1.18	# #######################################
10			class DBSError(exceptions.Exception):
11			def __init__(self, errorName, errorMessage):
12			args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
13			exceptions.Exception.__init__(self, args)
14			pass
15
16			def getErrorMessage(self):
17			""" Return error message """
18			return "%s" % (self.args)
19
20			# #######################################
21			class DBSInvalidDataTierError(exceptions.Exception):
22			def __init__(self, errorName, errorMessage):
23			args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
24			exceptions.Exception.__init__(self, args)
25			pass
26
27			def getErrorMessage(self):
28			""" Return error message """
29			return "%s" % (self.args)
30
31			# #######################################
32			class DBSInfoError:
33			def __init__(self, url):
34			print '\nERROR accessing DBS url : '+url+'\n'
35			pass
36
37	afanfani	1.1	# ####################################
38	afanfani	1.3	class DataDiscoveryError(exceptions.Exception):
39	slacapra	1.7	def __init__(self, errorMessage):
40	gutsche	1.15	self.args=errorMessage
41	slacapra	1.14	exceptions.Exception.__init__(self, self.args)
42	slacapra	1.7	pass
43
44			def getErrorMessage(self):
45			""" Return exception error """
46			return "%s" % (self.args)
47	afanfani	1.3
48	afanfani	1.1	# ####################################
49	afanfani	1.3	class NotExistingDatasetError(exceptions.Exception):
50	slacapra	1.7	def __init__(self, errorMessage):
51	gutsche	1.15	self.args=errorMessage
52	slacapra	1.14	exceptions.Exception.__init__(self, self.args)
53	slacapra	1.7	pass
54
55			def getErrorMessage(self):
56			""" Return exception error """
57			return "%s" % (self.args)
58	afanfani	1.1
59			# ####################################
60	afanfani	1.3	class NoDataTierinProvenanceError(exceptions.Exception):
61	slacapra	1.7	def __init__(self, errorMessage):
62	gutsche	1.15	self.args=errorMessage
63	slacapra	1.14	exceptions.Exception.__init__(self, self.args)
64	slacapra	1.7	pass
65
66			def getErrorMessage(self):
67			""" Return exception error """
68			return "%s" % (self.args)
69	afanfani	1.1
70			# ####################################
71			# class to find and extact info from published data
72			class DataDiscovery:
73	gutsche	1.15	def __init__(self, datasetPath, cfg_params):
74	afanfani	1.1
75	slacapra	1.18	# Attributes
76	slacapra	1.11	self.datasetPath = datasetPath
77	afanfani	1.1	self.cfg_params = cfg_params
78
79	slacapra	1.11	self.eventsPerBlock = {} # DBS output: map fileblocks-events for collection
80			self.eventsPerFile = {} # DBS output: map files-events
81	slacapra	1.18	self.blocksinfo = {} # DBS output: map fileblocks-files
82			self.maxEvents = 0 # DBS output: max events
83	afanfani	1.1
84			# ####################################
85			def fetchDBSInfo(self):
86			"""
87			Contact DBS
88			"""
89
90	slacapra	1.11	## get DBS URL
91	slacapra	1.19	dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
92			if (self.cfg_params.has_key('CMSSW.dbs_url')):
93	slacapra	1.11	dbs_url=self.cfg_params['CMSSW.dbs_url']
94	afanfani	1.3
95	slacapra	1.18	common.logger.debug(3,"Accessing DBS at: "+dbs_url)
96
97			## check if runs are selected
98	slacapra	1.19	runselection = []
99			if (self.cfg_params.has_key('CMSSW.runselection')):
100	slacapra	1.18	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
101
102	afanfani	1.20	common.logger.debug(6,"runselection is: %s"%runselection)
103	slacapra	1.18	## service API
104			args = {}
105			args['url'] = dbs_url
106			args['level'] = 'CRITICAL'
107
108			api = DBSAPI.dbsApi.DbsApi(args)
109	afanfani	1.5	try:
110	slacapra	1.18	if len(runselection) <= 0 :
111			files = api.listDatasetFiles(self.datasetPath)
112			else :
113	afanfani	1.20	files=[]
114			for arun in runselection:
115			try:
116			filesinrun = api.listFiles(path=self.datasetPath, details=True,runNumber=arun)
117			files.extend(filesinrun)
118			except:
119			msg="WARNING: problem extracting info from DBS for run %s "%arun
120			common.logger.message(msg)
121			pass
122
123	slacapra	1.18	except DbsBadRequest, msg:
124	slacapra	1.11	raise DataDiscoveryError(msg)
125	slacapra	1.13	except DBSError, msg:
126			raise DataDiscoveryError(msg)
127	slacapra	1.11
128	slacapra	1.18	# parse files and fill arrays
129			for file in files :
130			filename = file['LogicalFileName']
131			if filename.find('.dat') < 0 :
132			fileblock = file['Block']['Name']
133			events = file['NumberOfEvents']
134	afanfani	1.20	# number of events per block
135			if fileblock in self.eventsPerBlock.keys() :
136			self.eventsPerBlock[fileblock] += events
137			else :
138			self.eventsPerBlock[fileblock] = events
139			# number of events per file
140			self.eventsPerFile[filename] = events
141
142			# number of events per block
143			if fileblock in self.blocksinfo.keys() :
144			self.blocksinfo[fileblock].append(filename)
145	slacapra	1.18	else :
146	afanfani	1.20	self.blocksinfo[fileblock] = [filename]
147	slacapra	1.18
148	afanfani	1.20	# total number of events
149			self.maxEvents += events
150	slacapra	1.11
151	slacapra	1.18	for block in self.eventsPerBlock.keys() :
152			common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
153	slacapra	1.11
154			if len(self.eventsPerBlock) <= 0:
155	slacapra	1.18	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
156	slacapra	1.11	+ " dataset path variables in crab.cfg")
157	slacapra	1.18	% self.datasetPath)
158	afanfani	1.1
159
160			# #################################################
161			def getMaxEvents(self):
162			"""
163	slacapra	1.11	max events
164	afanfani	1.1	"""
165	slacapra	1.18	return self.maxEvents
166	afanfani	1.1
167			# #################################################
168	slacapra	1.11	def getEventsPerBlock(self):
169	afanfani	1.1	"""
170	slacapra	1.11	list the event collections structure by fileblock
171	afanfani	1.1	"""
172	slacapra	1.11	return self.eventsPerBlock
173	afanfani	1.1
174			# #################################################
175	slacapra	1.11	def getEventsPerFile(self):
176	afanfani	1.1	"""
177	slacapra	1.11	list the event collections structure by file
178	afanfani	1.1	"""
179	slacapra	1.11	return self.eventsPerFile
180	afanfani	1.1
181			# #################################################
182	slacapra	1.11	def getFiles(self):
183	afanfani	1.1	"""
184	slacapra	1.11	return files grouped by fileblock
185	afanfani	1.1	"""
186	slacapra	1.11	return self.blocksinfo
187	afanfani	1.1
188			########################################################################