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 1.23 by spiga, Mon Jun 9 17:50:45 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	>	import os
8	>
9	>
10	>	# #######################################
11	>	class DBSError(exceptions.Exception):
12	>	def __init__(self, errorName, errorMessage):
13	>	args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
14	>	exceptions.Exception.__init__(self, args)
15	>	pass
16	>
17	>	def getErrorMessage(self):
18	>	""" Return error message """
19	>	return "%s" % (self.args)
20	>
21	>	# #######################################
22	>	class DBSInvalidDataTierError(exceptions.Exception):
23	>	def __init__(self, errorName, errorMessage):
24	>	args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
25	>	exceptions.Exception.__init__(self, args)
26	>	pass
27	>
28	>	def getErrorMessage(self):
29	>	""" Return error message """
30	>	return "%s" % (self.args)
31	>
32	>	# #######################################
33	>	class DBSInfoError:
34	>	def __init__(self, url):
35	>	print '\nERROR accessing DBS url : '+url+'\n'
36	>	pass
37
38		# ####################################
39		class DataDiscoveryError(exceptions.Exception):
40	<	def __init__(self, errorMessage):
41	<	args=errorMessage
42	<	exceptions.Exception.__init__(self, args)
43	<	pass
44	<
45	<	def getErrorMessage(self):
46	<	""" Return exception error """
47	<	return "%s" % (self.args)
40	>	def __init__(self, errorMessage):
41	>	self.args=errorMessage
42	>	exceptions.Exception.__init__(self, self.args)
43	>	pass
44	>
45	>	def getErrorMessage(self):
46	>	""" Return exception error """
47	>	return "%s" % (self.args)
48
49		# ####################################
50		class NotExistingDatasetError(exceptions.Exception):
51	<	def __init__(self, errorMessage):
52	<	args=errorMessage
53	<	exceptions.Exception.__init__(self, args)
54	<	pass
55	<
56	<	def getErrorMessage(self):
57	<	""" Return exception error """
58	<	return "%s" % (self.args)
51	>	def __init__(self, errorMessage):
52	>	self.args=errorMessage
53	>	exceptions.Exception.__init__(self, self.args)
54	>	pass
55	>
56	>	def getErrorMessage(self):
57	>	""" Return exception error """
58	>	return "%s" % (self.args)
59
60		# ####################################
61		class NoDataTierinProvenanceError(exceptions.Exception):
62	<	def __init__(self, errorMessage):
63	<	args=errorMessage
64	<	exceptions.Exception.__init__(self, args)
65	<	pass
66	<
67	<	def getErrorMessage(self):
68	<	""" Return exception error """
69	<	return "%s" % (self.args)
62	>	def __init__(self, errorMessage):
63	>	self.args=errorMessage
64	>	exceptions.Exception.__init__(self, self.args)
65	>	pass
66	>
67	>	def getErrorMessage(self):
68	>	""" Return exception error """
69	>	return "%s" % (self.args)
70
71		# ####################################
72		# class to find and extact info from published data
73		class DataDiscovery:
74	<	def __init__(self, owner, dataset, dataTiers, cfg_params):
74	>	def __init__(self, datasetPath, cfg_params, skipAnBlocks):
75
76	<	#       Attributes
77	<	self.owner = owner
46	<	self.dataset = dataset
47	<	self.dataTiers = dataTiers
76	>	#       Attributes
77	>	self.datasetPath = datasetPath
78		self.cfg_params = cfg_params
79	+	self.skipBlocks = skipAnBlocks
80
81	<	self.dbspaths= []     # DBS output: list of dbspaths for all data
82	<	self.allblocks = []   # DBS output: list of map fileblocks-totevts for all dataset-owners
83	<	self.blocksinfo = {}  # DBS output: map fileblocks-totevts for the primary block, used internally to this class
84	<	#DBS output: max events computed by method getMaxEvents
81	>	self.eventsPerBlock = {}  # DBS output: map fileblocks-events for collection
82	>	self.eventsPerFile = {}   # DBS output: map files-events
83	>	self.blocksinfo = {}      # DBS output: map fileblocks-files
84	>	self.maxEvents = 0        # DBS output: max events
85	>	self.parent = {}       # DBS output: max events
86
87		# ####################################
88		def fetchDBSInfo(self):
#	Line 58 | Line 90 | class DataDiscovery:
90		Contact DBS
91		"""
92
93	<	## add the PU among the required data tiers if the Digi are requested
94	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
95	<	self.dataTiers.append('PU')
96	<
97	<	## get info about the requested dataset
98	<	dbs=DBSInfo()
93	>	## get DBS URL
94	>	dbs_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
95	>	if (self.cfg_params.has_key('CMSSW.dbs_url')):
96	>	dbs_url=self.cfg_params['CMSSW.dbs_url']
97	>
98	>	common.logger.debug(3,"Accessing DBS at: "+dbs_url)
99	>
100	>	## check if runs are selected
101	>	runselection = []
102	>	if (self.cfg_params.has_key('CMSSW.runselection')):
103	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
104	>
105	>	common.logger.debug(6,"runselection is: %s"%runselection)
106	>	## service API
107	>	args = {}
108	>	args['url']     = dbs_url
109	>	args['level']   = 'CRITICAL'
110	>
111	>	## check if has been requested to use the parent info
112	>	useParent = self.cfg_params.get('CMSSW.use_parent',False)
113	>
114	>	## check if has been asked for a non default file to store/read analyzed fileBlocks
115	>	defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt'
116	>	fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName))
117	>
118	>	api = DBSAPI.dbsApi.DbsApi(args)
119		try:
120	<	self.datasets = dbs.getMatchingDatasets(self.owner, self.dataset)
121	<	except DBSError, ex:
122	<	raise DataDiscoveryError(ex.getErrorMessage())
123	<	if len(self.datasets) == 0:
124	<	raise DataDiscoveryError("Owner=%s, Dataset=%s unknown to DBS" % (self.owner, self.dataset))
125	<	if len(self.datasets) > 1:
126	<	raise DataDiscoveryError("Owner=%s, Dataset=%s is ambiguous" % (self.owner, self.dataset))
127	<	try:
128	<	self.dbsdataset = self.datasets[0].get('datasetPathName')
129	<	self.blocksinfo = dbs.getDatasetContents(self.dbsdataset)
130	<	self.allblocks.append (self.blocksinfo.keys ()) # add also the current fileblocksinfo
131	<	self.dbspaths.append(self.dbsdataset)
132	<	except DBSError, ex:
133	<	raise DataDiscoveryError(ex.getErrorMessage())
134	<
135	<	if len(self.blocksinfo)<=0:
136	<	msg="\nERROR Data for %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset
137	<	raise NotExistingDatasetError(msg)
138	<
139	<
140	<	## get info about the parents
141	<	try:
142	<	parents=dbs.getDatasetProvenance(self.dbsdataset, self.dataTiers)
143	<	except DBSInvalidDataTierError, ex:
144	<	msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n'
145	<	raise DataDiscoveryError(msg)
146	<	except DBSError, ex:
147	<	raise DataDiscoveryError(ex.getErrorMessage())
148	<
149	<	## check that the user asks for parent Data Tier really existing in the DBS provenance
150	<	self.checkParentDataTier(parents, self.dataTiers)
120	>	if len(runselection) <= 0 :
121	>	if useParent:
122	>	allowedRetriveValue = ['retrive_parent',
123	>	'retrive_block',
124	>	'retrive_lumi',
125	>	'retrive_run'
126	>	]
127	>	files = api.listFiles(path=self.datasetPath, retriveList=allowedRetriveValue)
128	>	common.logger.debug(5,"Set of input parameters used for DBS query : \n"+str(allowedRetriveValue))
129	>	common.logger.write("Set of input parameters used for DBS query : \n"+str(allowedRetriveValue))
130	>	else:
131	>	files = api.listDatasetFiles(self.datasetPath)
132	>	else :
133	>	files=[]
134	>	for arun in runselection:
135	>	try:
136	>	filesinrun = api.listFiles(path=self.datasetPath,retriveList=allowedRetriveValue,runNumber=arun)
137	>	files.extend(filesinrun)
138	>	except:
139	>	msg="WARNING: problem extracting info from DBS for run %s "%arun
140	>	common.logger.message(msg)
141	>	pass
142	>
143	>	except DbsBadRequest, msg:
144	>	raise DataDiscoveryError(msg)
145	>	except DBSError, msg:
146	>	raise DataDiscoveryError(msg)
147	>
148	>	anFileBlocks = []
149	>	if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName)
150	>
151	>	# parse files and fill arrays
152	>	for file in files :
153	>	parList = []
154	>	# skip already analyzed blocks
155	>	fileblock = file['Block']['Name']
156	>	if fileblock not in anFileBlocks :
157	>	filename = file['LogicalFileName']
158	>	# asked retry the list of parent for the given child
159	>	if useParent: parList = [x['LogicalFileName'] for x in file['ParentList']]
160	>	self.parent[filename] = parList
161	>	if filename.find('.dat') < 0 :
162	>	events    = file['NumberOfEvents']
163	>	# number of events per block
164	>	if fileblock in self.eventsPerBlock.keys() :
165	>	self.eventsPerBlock[fileblock] += events
166	>	else :
167	>	self.eventsPerBlock[fileblock] = events
168	>	# number of events per file
169	>	self.eventsPerFile[filename] = events
170	>
171	>	# number of events per block
172	>	if fileblock in self.blocksinfo.keys() :
173	>	self.blocksinfo[fileblock].append(filename)
174	>	else :
175	>	self.blocksinfo[fileblock] = [filename]
176	>
177	>	# total number of events
178	>	self.maxEvents += events
179	>	if  self.skipBlocks and len(self.eventsPerBlock.keys()) == 0:
180	>	msg = "No new fileblocks available for dataset: "+str(self.datasetPath)
181	>	raise  CrabException(msg)
182	>
183	>	saveFblocks=''
184	>	for block in self.eventsPerBlock.keys() :
185	>	saveFblocks += str(block)+'\n'
186	>	common.logger.debug(6,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
187	>	writeTXTfile(self, fileBlocks_FileName , saveFblocks)
188	>
189	>	if len(self.eventsPerBlock) <= 0:
190	>	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
191	>	+ " dataset path variables in crab.cfg")
192	>	% self.datasetPath)
193
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())
194
195		# #################################################
196	<	def checkParentDataTier(self, parents, dataTiers):
196	>	def getMaxEvents(self):
197		"""
198	<	check that the data tiers requested by the user really exists in the provenance of the given dataset
198	>	max events
199		"""
200	<	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	<
200	>	return self.maxEvents
201
202		# #################################################
203	<	def getMaxEvents(self):
203	>	def getEventsPerBlock(self):
204		"""
205	<	max events of the primary dataset-owner
205	>	list the event collections structure by fileblock
206		"""
207	<	## 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
207	>	return self.eventsPerBlock
208
209		# #################################################
210	<	def getDBSPaths(self):
210	>	def getEventsPerFile(self):
211		"""
212	<	list the DBSpaths for all required data
212	>	list the event collections structure by file
213		"""
214	<	return self.dbspaths
214	>	return self.eventsPerFile
215
216		# #################################################
217	<	def getEVC(self):
217	>	def getFiles(self):
218		"""
219	<	list the event collections structure by fileblock
219	>	return files grouped by fileblock
220		"""
221	<	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"
221	>	return self.blocksinfo
222
223		# #################################################
224	<	def getFileBlocks(self):
224	>	def getParent(self):
225		"""
226	<	fileblocks for all required dataset-owners
226	>	return parent grouped by file
227		"""
228	<	return self.allblocks
228	>	return self.parent
229
230		########################################################################
160	–
161	–

Diff Legend

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