ViewVC Help

View File | Revision Log | Show Annotations | Root Listing

root/cvsroot/COMP/CRAB/python/DataDiscovery.py

Comparing COMP/CRAB/python/DataDiscovery.py (file contents):
Revision 1.3 by afanfani, Sun Jan 29 01:46:08 2006 UTC vs.
Revision 1.19 by slacapra, Fri Jan 4 17:30:56 2008 UTC

#	Line 1 | Line 1
1	<	#!/usr/bin/env python2
2	<	import sys, os, string, re
3	<	from DBSInfo import *
4	<
1	>	#!/usr/bin/env python
2	>	import exceptions
3	>	import DBSAPI.dbsApi
4	>	from DBSAPI.dbsApiException import *
5	>	import common
6	>	from crab_util import *
7	>
8	>
9	>	# #######################################
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		# ####################################
38		class DataDiscoveryError(exceptions.Exception):
39	<	def __init__(self, errorMessage):
40	<	args=errorMessage
41	<	exceptions.Exception.__init__(self, args)
42	<	pass
43	<
44	<	def getErrorMessage(self):
45	<	""" Return exception error """
46	<	return "%s" % (self.args)
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(exceptions.Exception):
50	<	def __init__(self, errorMessage):
51	<	args=errorMessage
52	<	exceptions.Exception.__init__(self, args)
53	<	pass
54	<
55	<	def getErrorMessage(self):
56	<	""" Return exception error """
57	<	return "%s" % (self.args)
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(exceptions.Exception):
61	<	def __init__(self, errorMessage):
62	<	args=errorMessage
63	<	exceptions.Exception.__init__(self, args)
64	<	pass
65	<
66	<	def getErrorMessage(self):
67	<	""" Return exception error """
68	<	return "%s" % (self.args)
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:
73	<	def __init__(self, owner, dataset, dataTiers, cfg_params):
73	>	def __init__(self, datasetPath, cfg_params):
74
75	<	#       Attributes
76	<	self.dbsdataset='/'+dataset+'/datatier/'+owner
46	<	self.dataTiers = dataTiers
75	>	#       Attributes
76	>	self.datasetPath = datasetPath
77		self.cfg_params = cfg_params
78
79	<	self.dbspaths= []     # DBS output: list of dbspaths for all data
80	<	self.allblocks = []   # DBS output: list of map fileblocks-totevts for all dataset-owners
81	<	self.blocksinfo = {}  # DBS output: map fileblocks-totevts for the primary block, used internally to this class
82	<	#DBS output: max events computed by method getMaxEvents
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		"""
59	–	parents = []
60	–	parentsblocksinfo = {}
89
90	<	## add the PU among the required data tiers if the Digi are requested
91	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
92	<	self.dataTiers.append('PU')
90	>	## get DBS URL
91	>	dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
92	>	if (self.cfg_params.has_key('CMSSW.dbs_url')):
93	>	dbs_url=self.cfg_params['CMSSW.dbs_url']
94	>
95	>	common.logger.debug(3,"Accessing DBS at: "+dbs_url)
96	>
97	>	## check if runs are selected
98	>	runselection = []
99	>	if (self.cfg_params.has_key('CMSSW.runselection')):
100	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
101	>
102	>	## service API
103	>	args = {}
104	>	args['url']     = dbs_url
105	>	args['level']   = 'CRITICAL'
106
107	<	## get info about the requested dataset
67	<	dbs=DBSInfo(self.dbsdataset,self.dataTiers)
107	>	api = DBSAPI.dbsApi.DbsApi(args)
108		try:
109	<	self.blocksinfo=dbs.getDatasetContents()
110	<	except DBSError, ex:
111	<	raise DataDiscoveryError(ex.getErrorMessage())
112	<
113	<	if len(self.blocksinfo)<=0:
114	<	msg="\nERROR Data %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset
115	<	raise NotExistingDatasetError(msg)
116	<
117	<	currentdatatier=string.split(self.blocksinfo.keys()[0],'/')[2]
118	<	fakedatatier=string.split(self.dbsdataset,'/')[2]
119	<	currentdbsdataset=string.replace(self.dbsdataset, fakedatatier, currentdatatier)
120	<
121	<	self.dbspaths.append(currentdbsdataset)    # add the requested dbspath
122	<
123	<	## get info about the parents
124	<	try:
125	<	parents=dbs.getDatasetProvenance()
126	<	except DBSInvalidDataTierError, ex:
127	<	msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n'
128	<	raise DataDiscoveryError(msg)
129	<	except DBSError, ex:
130	<	raise DataDiscoveryError(ex.getErrorMessage())
131	<
132	<	## check that the user asks for parent Data Tier really existing in the DBS provenance
133	<	self.checkParentDataTier(parents, self.dataTiers, currentdbsdataset)
134	<
135	<	## for each parent get the corresponding fileblocks
136	<	for aparent in parents:
137	<	## fill a list of dbspaths
138	<	parentdbsdataset=aparent.getDatasetPath()
139	<	self.dbspaths.append(parentdbsdataset)
140	<	pdbs=DBSInfo(parentdbsdataset,[])
141	<	try:
142	<	parentsblocksinfo=pdbs.getDatasetContents()
143	<	except DBSError, ex:
144	<	raise DataDiscoveryError(ex.getErrorMessage())
145	<
146	<	self.allblocks.append(parentsblocksinfo.keys()) # add parent fileblocksinfo
147	<
148	<	## all the required blocks
149	<	self.allblocks.append(self.blocksinfo.keys()) # add also the current fileblocksinfo
150	<
151	<
152	<	# #################################################
153	<	def checkParentDataTier(self, parents, user_datatiers, currentdbsdataset ):
154	<	"""
155	<	check that the data tiers requested by the user really exists in the provenance of the given dataset
156	<	"""
157	<
158	<	current_datatier=string.split(currentdbsdataset,'/')[2]
159	<
160	<	parent_datatypes=[]
121	<	for aparent in parents:
122	<	parent_datatypes.append(aparent.getDataType())
123	<
124	<	for datatier in user_datatiers:
125	<	if parent_datatypes.count(datatier)<=0:
126	<	# the current datatier is not supposed to be in the provenance
127	<	if not (datatier == current_datatier):
128	<	msg="\nERROR Data %s not published in DBS with asked data tiers : the data tier not found is %s !\n  Check the data_tier variable in crab.cfg !"%(currentdbsdataset,datatier)
129	<	raise  NoDataTierinProvenanceError(msg)
109	>	if len(runselection) <= 0 :
110	>	files = api.listDatasetFiles(self.datasetPath)
111	>	else :
112	>	files = api.listFiles(path=self.datasetPath, details=True)
113	>	except DbsBadRequest, msg:
114	>	raise DataDiscoveryError(msg)
115	>	except DBSError, msg:
116	>	raise DataDiscoveryError(msg)
117	>
118	>	# parse files and fill arrays
119	>	for file in files :
120	>	filename = file['LogicalFileName']
121	>	if filename.find('.dat') < 0 :
122	>	fileblock = file['Block']['Name']
123	>	events    = file['NumberOfEvents']
124	>	continue_flag = 0
125	>	if len(runselection) > 0 :
126	>	runslist = file['RunsList']
127	>	for run in runslist :
128	>	runnumber = run['RunNumber']
129	>	for selected_run in runselection :
130	>	if runnumber == selected_run :
131	>	continue_flag = 1
132	>	else :
133	>	continue_flag = 1
134	>
135	>	if continue_flag == 1 :
136	>	# number of events per block
137	>	if fileblock in self.eventsPerBlock.keys() :
138	>	self.eventsPerBlock[fileblock] += events
139	>	else :
140	>	self.eventsPerBlock[fileblock] = events
141	>
142	>	# number of events per file
143	>	self.eventsPerFile[filename] = events
144	>
145	>	# number of events per block
146	>	if fileblock in self.blocksinfo.keys() :
147	>	self.blocksinfo[fileblock].append(filename)
148	>	else :
149	>	self.blocksinfo[fileblock] = [filename]
150	>
151	>	# total number of events
152	>	self.maxEvents += events
153	>
154	>	for block in self.eventsPerBlock.keys() :
155	>	common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
156	>
157	>	if len(self.eventsPerBlock) <= 0:
158	>	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
159	>	+ " dataset path variables in crab.cfg")
160	>	% self.datasetPath)
161
162
163		# #################################################
164		def getMaxEvents(self):
165		"""
166	<	max events of the primary dataset-owner
166	>	max events
167		"""
168	<	## loop over the fileblocks of the primary dataset-owner
138	<	nevts=0
139	<	for blockevts in self.blocksinfo.values():
140	<	nevts=nevts+blockevts
141	<
142	<	return nevts
168	>	return self.maxEvents
169
170		# #################################################
171	<	def getDBSPaths(self):
171	>	def getEventsPerBlock(self):
172		"""
173	<	list the DBSpaths for all required data
173	>	list the event collections structure by fileblock
174		"""
175	<	return self.dbspaths
175	>	return self.eventsPerBlock
176
177		# #################################################
178	<	def getEVC(self):
178	>	def getEventsPerFile(self):
179		"""
180	<	list the event collections structure by fileblock
180	>	list the event collections structure by file
181		"""
182	<	print "To be used by a more complex job splitting... TODO later... "
157	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
182	>	return self.eventsPerFile
183
184		# #################################################
185	<	def getFileBlocks(self):
185	>	def getFiles(self):
186		"""
187	<	fileblocks for all required dataset-owners
187	>	return files grouped by fileblock
188		"""
189	<	return self.allblocks
189	>	return self.blocksinfo
190
191		########################################################################
167	–
168	–

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines