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 1.5 by afanfani, Thu May 18 18:46:22 2006 UTC vs.
Revision 1.21 by spiga, Thu Jun 5 07:13:31 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.owner = owner
46	<	self.dataset = dataset
47	<	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	>	self.parent = {}       # DBS output: max events
84
85		# ####################################
86		def fetchDBSInfo(self):
#	Line 58 | Line 88 | class DataDiscovery:
88		Contact DBS
89		"""
90
91	<	## add the PU among the required data tiers if the Digi are requested
92	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
93	<	self.dataTiers.append('PU')
94	<
95	<	## get info about the requested dataset
96	<	dbs=DBSInfo()
91	>	## get DBS URL
92	>	dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
93	>	if (self.cfg_params.has_key('CMSSW.dbs_url')):
94	>	dbs_url=self.cfg_params['CMSSW.dbs_url']
95	>
96	>	common.logger.debug(3,"Accessing DBS at: "+dbs_url)
97	>
98	>	## check if runs are selected
99	>	runselection = []
100	>	if (self.cfg_params.has_key('CMSSW.runselection')):
101	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
102	>
103	>	common.logger.debug(6,"runselection is: %s"%runselection)
104	>	## service API
105	>	args = {}
106	>	args['url']     = dbs_url
107	>	args['level']   = 'CRITICAL'
108	>
109	>	## check if has been requested to use the parent info
110	>	if (self.cfg_params.has_key('CMSSW.runselection')):
111	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
112	>
113	>	useParent = self.cfg_params.get('CMSSW.use_parent',False)
114	>
115	>	allowedRetriveValue = [
116	>	'retrive_child',
117	>	'retrive_block',
118	>	'retrive_lumi',
119	>	'retrive_run'
120	>	]
121	>	if useParent:  allowedRetriveValue.append('retrive_parent')
122	>	common.logger.debug(5,"Set of input parameters used for DBS query : \n"+str(allowedRetriveValue))
123	>	common.logger.write("Set of input parameters used for DBS query : \n"+str(allowedRetriveValue))
124	>	api = DBSAPI.dbsApi.DbsApi(args)
125		try:
126	<	self.datasets = dbs.getMatchingDatasets(self.owner, self.dataset)
127	<	except DBSError, ex:
128	<	raise DataDiscoveryError(ex.getErrorMessage())
129	<	if len(self.datasets) == 0:
130	<	raise DataDiscoveryError("Owner=%s, Dataset=%s unknown to DBS" % (self.owner, self.dataset))
131	<	if len(self.datasets) > 1:
132	<	raise DataDiscoveryError("Owner=%s, Dataset=%s is ambiguous" % (self.owner, self.dataset))
133	<	try:
134	<	self.dbsdataset = self.datasets[0].get('datasetPathName')
135	<	self.blocksinfo = dbs.getDatasetContents(self.dbsdataset)
136	<	self.allblocks.append (self.blocksinfo.keys ()) # add also the current fileblocksinfo
137	<	self.dbspaths.append(self.dbsdataset)
138	<	except DBSError, ex:
139	<	raise DataDiscoveryError(ex.getErrorMessage())
140	<
141	<	if len(self.blocksinfo)<=0:
142	<	msg="\nERROR Data for %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset
143	<	raise NotExistingDatasetError(msg)
144	<
145	<
146	<	## get info about the parents
147	<	try:
148	<	parents=dbs.getDatasetProvenance(self.dbsdataset, self.dataTiers)
149	<	except DBSInvalidDataTierError, ex:
150	<	msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n'
151	<	raise DataDiscoveryError(msg)
152	<	except DBSError, ex:
153	<	raise DataDiscoveryError(ex.getErrorMessage())
154	<
155	<	## check that the user asks for parent Data Tier really existing in the DBS provenance
156	<	self.checkParentDataTier(parents, self.dataTiers)
126	>	if len(runselection) <= 0 :
127	>	files = api.listFiles(path=self.datasetPath,retriveList=allowedRetriveValue)
128	>	else :
129	>	files=[]
130	>	for arun in runselection:
131	>	try:
132	>	filesinrun = api.listFiles(path=self.datasetPath,retriveList=allowedRetriveValue,runNumber=arun)
133	>	files.extend(filesinrun)
134	>	except:
135	>	msg="WARNING: problem extracting info from DBS for run %s "%arun
136	>	common.logger.message(msg)
137	>	pass
138	>
139	>	except DbsBadRequest, msg:
140	>	raise DataDiscoveryError(msg)
141	>	except DBSError, msg:
142	>	raise DataDiscoveryError(msg)
143	>
144	>	# parse files and fill arrays
145	>	for file in files :
146	>	parList = []
147	>	filename = file['LogicalFileName']
148	>	# asked retry the list of parent for the given child
149	>	if useParent: parList = [x['LogicalFileName'] for x in file['ParentList']]
150	>	self.parent[filename] = parList
151	>	if filename.find('.dat') < 0 :
152	>	fileblock = file['Block']['Name']
153	>	events    = file['NumberOfEvents']
154	>	# number of events per block
155	>	if fileblock in self.eventsPerBlock.keys() :
156	>	self.eventsPerBlock[fileblock] += events
157	>	else :
158	>	self.eventsPerBlock[fileblock] = events
159	>	# number of events per file
160	>	self.eventsPerFile[filename] = events
161	>
162	>	# number of events per block
163	>	if fileblock in self.blocksinfo.keys() :
164	>	self.blocksinfo[fileblock].append(filename)
165	>	else :
166	>	self.blocksinfo[fileblock] = [filename]
167	>
168	>	# total number of events
169	>	self.maxEvents += events
170	>
171	>	for block in self.eventsPerBlock.keys() :
172	>	common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
173	>
174	>	if len(self.eventsPerBlock) <= 0:
175	>	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
176	>	+ " dataset path variables in crab.cfg")
177	>	% self.datasetPath)
178
100	–	## for each parent get the corresponding fileblocks
101	–	try:
102	–	for p in parents:
103	–	## fill a list of dbspaths
104	–	parentPath = p.get('parent').get('datasetPathName')
105	–	self.dbspaths.append (parentPath)
106	–	parentBlocks = dbs.getDatasetContents (parentPath)
107	–	self.allblocks.append (parentBlocks.keys ())  # add parent fileblocksinfo
108	–	except DBSError, ex:
109	–	raise DataDiscoveryError(ex.getErrorMessage())
179
180		# #################################################
181	<	def checkParentDataTier(self, parents, dataTiers):
181	>	def getMaxEvents(self):
182		"""
183	<	check that the data tiers requested by the user really exists in the provenance of the given dataset
183	>	max events
184		"""
185	<	startType = string.split(self.dbsdataset,'/')[2]
117	<	# for example 'type' is PU and 'dataTier' is Hit
118	<	parentTypes = map(lambda p: p.get('type'), parents)
119	<	for tier in dataTiers:
120	<	if parentTypes.count(tier) <= 0 and tier != startType:
121	<	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 !"%(self.dbsdataset,tier)
122	<	raise  NoDataTierinProvenanceError(msg)
123	<
185	>	return self.maxEvents
186
187		# #################################################
188	<	def getMaxEvents(self):
188	>	def getEventsPerBlock(self):
189		"""
190	<	max events of the primary dataset-owner
190	>	list the event collections structure by fileblock
191		"""
192	<	## loop over the fileblocks of the primary dataset-owner
131	<	nevts=0
132	<	for blockevts in self.blocksinfo.values():
133	<	nevts=nevts+blockevts
134	<
135	<	return nevts
192	>	return self.eventsPerBlock
193
194		# #################################################
195	<	def getDBSPaths(self):
195	>	def getEventsPerFile(self):
196		"""
197	<	list the DBSpaths for all required data
197	>	list the event collections structure by file
198		"""
199	<	return self.dbspaths
199	>	return self.eventsPerFile
200
201		# #################################################
202	<	def getEVC(self):
202	>	def getFiles(self):
203		"""
204	<	list the event collections structure by fileblock
204	>	return files grouped by fileblock
205		"""
206	<	print "To be used by a more complex job splitting... TODO later... "
150	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
206	>	return self.blocksinfo
207
208		# #################################################
209	<	def getFileBlocks(self):
209	>	def getParent(self):
210		"""
211	<	fileblocks for all required dataset-owners
211	>	return parent grouped by file
212		"""
213	<	return self.allblocks
213	>	return self.parent
214
215		########################################################################
160	–
161	–

Diff Legend

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