CRAB/python/DataDiscovery_DBS2.py

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


# #######################################
class DBSError_DBS2(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_DBS2(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_DBS2:
    def __init__(self, url):
        print '\nERROR accessing DBS url : '+url+'\n'
        pass

# ####################################
class DataDiscoveryError_DBS2(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_DBS2(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_DBS2(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_DBS2:
    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
        try:
            dbs_url=self.cfg_params['CMSSW.dbs_url']
        except KeyError:
            dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"

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

        ## check if runs are selected
        try:
            runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
        except:
            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 = api.listFiles(path=self.datasetPath, details=True)
        except DbsBadRequest, msg:
            raise DataDiscoveryError_DBS2(msg)
        except DBSError_DBS2, msg:
            raise DataDiscoveryError_DBS2(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']
                continue_flag = 0
                if len(runselection) > 0 :
                    runslist = file['RunsList']
                    for run in runslist :
                        runnumber = run['RunNumber']
                        for selected_run in runselection :
                            if runnumber == selected_run :
                                continue_flag = 1
                else :
                    continue_flag = 1

                if continue_flag == 1 :
                    # 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_DBS2 (("\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:	1.9
Committed:	Wed Jul 4 00:50:09 2007 UTC (17 years, 10 months ago) by gutsche
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_0_2_pre5, CRAB_2_0_2_pre4, CRAB_2_0_2_pre3, CRAB_1_5_4_SLC3, CRAB_1_5_4_SLC3_pre4, CRAB_2_0_2_pre2, CRAB_2_0_2_pre1, CRAB_1_5_4_SLC3_pre3, CRAB_2_0_1, CRAB_1_5_4_SLC3_pre2, CRAB_2_0_1_pre1, CRAB_1_5_4_SLC3_pre1, CRAB_2_0_0, CRAB_2_0_0_pre10, CRAB_2_0_0_pre9, CRAB_1_5_4, CRAB_1_5_4_pre2, CRAB_1_5_4_pre1, CRAB_2_0_0_pre7, CRAB_2_0_0_pre6, CRAB_1_5_3, CRAB_1_5_3_pre5, CRAB_1_5_3_pre4, CRAB_2_0_0_pre5, CRAB_1_5_3_pre3, configure, CRAB_2_0_0_pre4
Branch point for:	CRAB_1_5_4_SLC3_pre4_br, CRAB_1_5_4_SLC3_start
Changes since 1.8:	+41 -20 lines
Log Message:	Introduced run selection within single PDS crab.cfg card: runselection = 1,2,3-4 Parsing is done on UI, DBS has no suited API to do parsing on DBS server (yet) api.listFiles is used with details=True which increases query time significantly. In addition, filtered for .dat files which can be present in PDS mixed with .root
#	User	Rev	Content
1	gutsche	1.1	#!/usr/bin/env python
2			import exceptions
3	gutsche	1.2	import DBSAPI.dbsApi
4			from DBSAPI.dbsApiException import *
5	gutsche	1.1	import common
6	gutsche	1.9	from crab_util import *
7	gutsche	1.1
8
9			# #######################################
10			class DBSError_DBS2(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_DBS2(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_DBS2:
33			def __init__(self, url):
34			print '\nERROR accessing DBS url : '+url+'\n'
35			pass
36
37			# ####################################
38			class DataDiscoveryError_DBS2(exceptions.Exception):
39			def __init__(self, errorMessage):
40			self.args=errorMessage
41			exceptions.Exception.__init__(self, self.args)
42			pass
43
44			def getErrorMessage(self):
45			""" Return exception error """
46			return "%s" % (self.args)
47
48			# ####################################
49			class NotExistingDatasetError_DBS2(exceptions.Exception):
50			def __init__(self, errorMessage):
51			self.args=errorMessage
52			exceptions.Exception.__init__(self, self.args)
53			pass
54
55			def getErrorMessage(self):
56			""" Return exception error """
57			return "%s" % (self.args)
58
59			# ####################################
60			class NoDataTierinProvenanceError_DBS2(exceptions.Exception):
61			def __init__(self, errorMessage):
62			self.args=errorMessage
63			exceptions.Exception.__init__(self, self.args)
64			pass
65
66			def getErrorMessage(self):
67			""" Return exception error """
68			return "%s" % (self.args)
69
70			# ####################################
71			# class to find and extact info from published data
72			class DataDiscovery_DBS2:
73			def __init__(self, datasetPath, cfg_params):
74
75			# Attributes
76			self.datasetPath = datasetPath
77			self.cfg_params = cfg_params
78
79			self.eventsPerBlock = {} # DBS output: map fileblocks-events for collection
80			self.eventsPerFile = {} # DBS output: map files-events
81			self.blocksinfo = {} # DBS output: map fileblocks-files
82			self.maxEvents = 0 # DBS output: max events
83
84			# ####################################
85			def fetchDBSInfo(self):
86			"""
87			Contact DBS
88			"""
89
90			## get DBS URL
91			try:
92			dbs_url=self.cfg_params['CMSSW.dbs_url']
93			except KeyError:
94	gutsche	1.6	dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
95	gutsche	1.1
96	gutsche	1.8	common.logger.debug(3,"Accessing DBS at: "+dbs_url)
97	gutsche	1.7
98	gutsche	1.9	## check if runs are selected
99			try:
100			runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
101			except:
102			runselection = []
103	gutsche	1.7
104	gutsche	1.1	## service API
105			args = {}
106	gutsche	1.3	args['url'] = dbs_url
107	gutsche	1.4	args['level'] = 'CRITICAL'
108	gutsche	1.1
109	gutsche	1.2	api = DBSAPI.dbsApi.DbsApi(args)
110	gutsche	1.1	try:
111	gutsche	1.9	if len(runselection) <= 0 :
112			files = api.listDatasetFiles(self.datasetPath)
113			else :
114			files = api.listFiles(path=self.datasetPath, details=True)
115	gutsche	1.1	except DbsBadRequest, msg:
116	gutsche	1.2	raise DataDiscoveryError_DBS2(msg)
117	gutsche	1.1	except DBSError_DBS2, msg:
118	gutsche	1.2	raise DataDiscoveryError_DBS2(msg)
119	gutsche	1.8
120	gutsche	1.1	# parse files and fill arrays
121			for file in files :
122			filename = file['LogicalFileName']
123	gutsche	1.9	if filename.find('.dat') < 0 :
124			fileblock = file['Block']['Name']
125			events = file['NumberOfEvents']
126			continue_flag = 0
127			if len(runselection) > 0 :
128			runslist = file['RunsList']
129			for run in runslist :
130			runnumber = run['RunNumber']
131			for selected_run in runselection :
132			if runnumber == selected_run :
133			continue_flag = 1
134			else :
135			continue_flag = 1
136
137			if continue_flag == 1 :
138			# number of events per block
139			if fileblock in self.eventsPerBlock.keys() :
140			self.eventsPerBlock[fileblock] += events
141			else :
142			self.eventsPerBlock[fileblock] = events
143
144			# number of events per file
145			self.eventsPerFile[filename] = events
146
147			# number of events per block
148			if fileblock in self.blocksinfo.keys() :
149			self.blocksinfo[fileblock].append(filename)
150			else :
151			self.blocksinfo[fileblock] = [filename]
152	gutsche	1.1
153	gutsche	1.9	# total number of events
154			self.maxEvents += events
155	gutsche	1.1
156	gutsche	1.7	for block in self.eventsPerBlock.keys() :
157			common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
158
159	gutsche	1.1	if len(self.eventsPerBlock) <= 0:
160	gutsche	1.2	raise NotExistingDatasetError_DBS2 (("\nNo data for %s in DBS\nPlease check"
161	gutsche	1.1	+ " dataset path variables in crab.cfg")
162			% self.datasetPath)
163
164
165			# #################################################
166			def getMaxEvents(self):
167			"""
168			max events
169			"""
170			return self.maxEvents
171
172			# #################################################
173			def getEventsPerBlock(self):
174			"""
175			list the event collections structure by fileblock
176			"""
177			return self.eventsPerBlock
178
179			# #################################################
180			def getEventsPerFile(self):
181			"""
182			list the event collections structure by file
183			"""
184			return self.eventsPerFile
185
186			# #################################################
187			def getFiles(self):
188			"""
189			return files grouped by fileblock
190			"""
191			return self.blocksinfo
192
193			########################################################################