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
        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(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']
                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(("\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:	1.18
Committed:	Fri Nov 16 11:09:31 2007 UTC (17 years, 5 months ago) by slacapra
Content type:	text/x-python
Branch:	MAIN
CVS Tags:	CRAB_2_0_4, CRAB_2_0_4_pre2, CRAB_2_0_4_pre1, CRAB_2_0_3, CRAB_2_0_3_pre1, CRAB_2_0_2, CRAB_2_0_2_pre6
Changes since 1.17:	+95 -33 lines
Log Message:	remove support for DBS1, and remove DBS2 string from files/class
#	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			try:
92			dbs_url=self.cfg_params['CMSSW.dbs_url']
93			except KeyError:
94	slacapra	1.18	dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
95	afanfani	1.3
96	slacapra	1.18	common.logger.debug(3,"Accessing DBS at: "+dbs_url)
97
98			## check if runs are selected
99	slacapra	1.11	try:
100	slacapra	1.18	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
101			except:
102			runselection = []
103
104			## service API
105			args = {}
106			args['url'] = dbs_url
107			args['level'] = 'CRITICAL'
108
109			api = DBSAPI.dbsApi.DbsApi(args)
110	afanfani	1.5	try:
111	slacapra	1.18	if len(runselection) <= 0 :
112			files = api.listDatasetFiles(self.datasetPath)
113			else :
114			files = api.listFiles(path=self.datasetPath, details=True)
115			except DbsBadRequest, msg:
116	slacapra	1.11	raise DataDiscoveryError(msg)
117	slacapra	1.13	except DBSError, msg:
118			raise DataDiscoveryError(msg)
119	slacapra	1.11
120	slacapra	1.18	# parse files and fill arrays
121			for file in files :
122			filename = file['LogicalFileName']
123			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
153			# total number of events
154			self.maxEvents += events
155	slacapra	1.11
156	slacapra	1.18	for block in self.eventsPerBlock.keys() :
157			common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
158	slacapra	1.11
159			if len(self.eventsPerBlock) <= 0:
160	slacapra	1.18	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
161	slacapra	1.11	+ " dataset path variables in crab.cfg")
162	slacapra	1.18	% self.datasetPath)
163	afanfani	1.1
164
165			# #################################################
166			def getMaxEvents(self):
167			"""
168	slacapra	1.11	max events
169	afanfani	1.1	"""
170	slacapra	1.18	return self.maxEvents
171	afanfani	1.1
172			# #################################################
173	slacapra	1.11	def getEventsPerBlock(self):
174	afanfani	1.1	"""
175	slacapra	1.11	list the event collections structure by fileblock
176	afanfani	1.1	"""
177	slacapra	1.11	return self.eventsPerBlock
178	afanfani	1.1
179			# #################################################
180	slacapra	1.11	def getEventsPerFile(self):
181	afanfani	1.1	"""
182	slacapra	1.11	list the event collections structure by file
183	afanfani	1.1	"""
184	slacapra	1.11	return self.eventsPerFile
185	afanfani	1.1
186			# #################################################
187	slacapra	1.11	def getFiles(self):
188	afanfani	1.1	"""
189	slacapra	1.11	return files grouped by fileblock
190	afanfani	1.1	"""
191	slacapra	1.11	return self.blocksinfo
192	afanfani	1.1
193			########################################################################