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 1.48 by ewv, Tue Jul 6 16:31:55 2010 UTC

#	Line 1 | Line 1
1	<	#!/usr/bin/env python2
2	<	import sys, os, string, re
3	<	from DBSInfo import *
1	>	#!/usr/bin/env python
2	>
3	>	__revision__ = "$Id$"
4	>	__version__ = "$Revision$"
5	>
6	>	import exceptions
7	>	import DBSAPI.dbsApi
8	>	from DBSAPI.dbsApiException import *
9	>	import common
10	>	from crab_util import *
11	>	try: # Can remove when CMSSW 3.7 and earlier are dropped
12	>	from FWCore.PythonUtilities.LumiList import LumiList
13	>	except ImportError:
14	>	from LumiList import LumiList
15	>
16	>	import os
17	>
18	>
19	>
20	>	class DBSError(exceptions.Exception):
21	>	def __init__(self, errorName, errorMessage):
22	>	args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
23	>	exceptions.Exception.__init__(self, args)
24	>	pass
25	>
26	>	def getErrorMessage(self):
27	>	""" Return error message """
28	>	return "%s" % (self.args)
29	>
30	>
31	>
32	>	class DBSInvalidDataTierError(exceptions.Exception):
33	>	def __init__(self, errorName, errorMessage):
34	>	args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
35	>	exceptions.Exception.__init__(self, args)
36	>	pass
37	>
38	>	def getErrorMessage(self):
39	>	""" Return error message """
40	>	return "%s" % (self.args)
41	>
42	>
43	>
44	>	class DBSInfoError:
45	>	def __init__(self, url):
46	>	print '\nERROR accessing DBS url : '+url+'\n'
47	>	pass
48	>
49
50
6	–	# ####################################
51		class DataDiscoveryError(exceptions.Exception):
52	<	def __init__(self, errorMessage):
53	<	args=errorMessage
54	<	exceptions.Exception.__init__(self, args)
55	<	pass
56	<
57	<	def getErrorMessage(self):
58	<	""" Return exception error """
59	<	return "%s" % (self.args)
52	>	def __init__(self, errorMessage):
53	>	self.args=errorMessage
54	>	exceptions.Exception.__init__(self, self.args)
55	>	pass
56	>
57	>	def getErrorMessage(self):
58	>	""" Return exception error """
59	>	return "%s" % (self.args)
60	>
61	>
62
17	–	# ####################################
63		class NotExistingDatasetError(exceptions.Exception):
64	<	def __init__(self, errorMessage):
65	<	args=errorMessage
66	<	exceptions.Exception.__init__(self, args)
67	<	pass
68	<
69	<	def getErrorMessage(self):
70	<	""" Return exception error """
71	<	return "%s" % (self.args)
64	>	def __init__(self, errorMessage):
65	>	self.args=errorMessage
66	>	exceptions.Exception.__init__(self, self.args)
67	>	pass
68	>
69	>	def getErrorMessage(self):
70	>	""" Return exception error """
71	>	return "%s" % (self.args)
72	>
73	>
74
28	–	# ####################################
75		class NoDataTierinProvenanceError(exceptions.Exception):
76	<	def __init__(self, errorMessage):
77	<	args=errorMessage
78	<	exceptions.Exception.__init__(self, args)
79	<	pass
80	<
81	<	def getErrorMessage(self):
82	<	""" Return exception error """
83	<	return "%s" % (self.args)
76	>	def __init__(self, errorMessage):
77	>	self.args=errorMessage
78	>	exceptions.Exception.__init__(self, self.args)
79	>	pass
80	>
81	>	def getErrorMessage(self):
82	>	""" Return exception error """
83	>	return "%s" % (self.args)
84	>
85
39	–	# ####################################
40	–	# class to find and extact info from published data
41	–	class DataDiscovery:
42	–	def __init__(self, owner, dataset, dataTiers, cfg_params):
86
87	<	#       Attributes
88	<	self.owner = owner
89	<	self.dataset = dataset
90	<	self.dataTiers = dataTiers
87	>	class DataDiscovery:
88	>	"""
89	>	Class to find and extact info from published data
90	>	"""
91	>	def __init__(self, datasetPath, cfg_params, skipAnBlocks):
92	>
93	>	#       Attributes
94	>	self.datasetPath = datasetPath
95	>	# Analysis dataset is primary/processed/tier/definition
96	>	self.ads = len(self.datasetPath.split("/")) > 4
97		self.cfg_params = cfg_params
98	+	self.skipBlocks = skipAnBlocks
99
100	<	self.dbspaths= []     # DBS output: list of dbspaths for all data
101	<	self.allblocks = []   # DBS output: list of map fileblocks-totevts for all dataset-owners
102	<	self.blocksinfo = {}  # DBS output: map fileblocks-totevts for the primary block, used internally to this class
103	<	#DBS output: max events computed by method getMaxEvents
100	>	self.eventsPerBlock = {}  # DBS output: map fileblocks-events for collection
101	>	self.eventsPerFile = {}   # DBS output: map files-events
102	>	#         self.lumisPerBlock = {}   # DBS output: number of lumis in each block
103	>	#         self.lumisPerFile = {}    # DBS output: number of lumis in each file
104	>	self.blocksinfo = {}      # DBS output: map fileblocks-files
105	>	self.maxEvents = 0        # DBS output: max events
106	>	self.maxLumis = 0         # DBS output: total number of lumis
107	>	self.parent = {}          # DBS output: parents of each file
108	>	self.lumis = {}           # DBS output: lumis in each file
109	>	self.lumiMask = None
110	>	self.splitByLumi = False
111
55	–	# ####################################
112		def fetchDBSInfo(self):
113		"""
114		Contact DBS
115		"""
116	<
117	<	## add the PU among the required data tiers if the Digi are requested
118	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
119	<	self.dataTiers.append('PU')
120	<
121	<	## get info about the requested dataset
122	<	dbs=DBSInfo()
116	>	## get DBS URL
117	>	global_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
118	>	dbs_url=  self.cfg_params.get('CMSSW.dbs_url', global_url)
119	>	common.logger.info("Accessing DBS at: "+dbs_url)
120	>
121	>	## check if runs are selected
122	>	runselection = []
123	>	if (self.cfg_params.has_key('CMSSW.runselection')):
124	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
125	>
126	>	## check if various lumi parameters are set
127	>	self.lumiMask = self.cfg_params.get('CMSSW.lumi_mask',None)
128	>	self.lumiParams = self.cfg_params.get('CMSSW.total_number_of_lumis',None) or \
129	>	self.cfg_params.get('CMSSW.lumis_per_job',None)
130	>
131	>	lumiList = None
132	>	if self.lumiMask:
133	>	lumiList = LumiList(filename=self.lumiMask)
134	>	if runselection:
135	>	runList = LumiList(runs = runselection)
136	>
137	>	self.splitByRun = int(self.cfg_params.get('CMSSW.split_by_run', 0))
138	>	common.logger.log(10-1,"runselection is: %s"%runselection)
139	>
140	>	if not self.splitByRun:
141	>	self.splitByLumi = self.lumiMask or self.lumiParams or self.ads
142	>
143	>	if self.splitByRun and not runselection:
144	>	msg = "Error: split_by_run must be combined with a runselection"
145	>	raise CrabException(msg)
146	>
147	>	## service API
148	>	args = {}
149	>	args['url']     = dbs_url
150	>	args['level']   = 'CRITICAL'
151	>
152	>	## check if has been requested to use the parent info
153	>	useparent = int(self.cfg_params.get('CMSSW.use_parent',0))
154	>
155	>	## check if has been asked for a non default file to store/read analyzed fileBlocks
156	>	defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt'
157	>	fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName))
158	>
159	>	api = DBSAPI.dbsApi.DbsApi(args)
160	>	self.files = self.queryDbs(api,path=self.datasetPath,runselection=runselection,useParent=useparent)
161	>
162	>	# Check to see what the dataset is
163	>	pdsName = self.datasetPath.split("/")[1]
164	>	primDSs = api.listPrimaryDatasets(pdsName)
165	>	dataType = primDSs[0]['Type']
166	>	common.logger.debug("Datatype is %s" % dataType)
167	>	if dataType == 'data' and not (self.splitByRun or self.splitByLumi):
168	>	msg = 'Data must be split by lumi or by run. ' \
169	>	'Please see crab -help for the correct settings'
170	>	raise  CrabException(msg)
171	>
172	>
173	>
174	>	anFileBlocks = []
175	>	if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName)
176	>
177	>	# parse files and fill arrays
178	>	for file in self.files :
179	>	parList  = []
180	>	fileLumis = [] # List of tuples
181	>	# skip already analyzed blocks
182	>	fileblock = file['Block']['Name']
183	>	if fileblock not in anFileBlocks :
184	>	filename = file['LogicalFileName']
185	>	# asked retry the list of parent for the given child
186	>	if useparent==1:
187	>	parList = [x['LogicalFileName'] for x in file['ParentList']]
188	>	if self.splitByLumi:
189	>	fileLumis = [ (x['RunNumber'], x['LumiSectionNumber'])
190	>	for x in file['LumiList'] ]
191	>	self.parent[filename] = parList
192	>	# For LumiMask, intersection of two lists.
193	>	if self.lumiMask and runselection:
194	>	self.lumis[filename] = runList.filterLumis(lumiList.filterLumis(fileLumis))
195	>	elif runselection:
196	>	self.lumis[filename] = runList.filterLumis(fileLumis)
197	>	elif self.lumiMask:
198	>	self.lumis[filename] = lumiList.filterLumis(fileLumis)
199	>	else:
200	>	self.lumis[filename] = fileLumis
201	>	if filename.find('.dat') < 0 :
202	>	events    = file['NumberOfEvents']
203	>	# Count number of events and lumis per block
204	>	if fileblock in self.eventsPerBlock.keys() :
205	>	self.eventsPerBlock[fileblock] += events
206	>	else :
207	>	self.eventsPerBlock[fileblock] = events
208	>	# Number of events per file
209	>	self.eventsPerFile[filename] = events
210	>
211	>	# List of files per block
212	>	if fileblock in self.blocksinfo.keys() :
213	>	self.blocksinfo[fileblock].append(filename)
214	>	else :
215	>	self.blocksinfo[fileblock] = [filename]
216	>
217	>	# total number of events
218	>	self.maxEvents += events
219	>	self.maxLumis  += len(self.lumis[filename])
220	>
221	>	if  self.skipBlocks and len(self.eventsPerBlock.keys()) == 0:
222	>	msg = "No new fileblocks available for dataset: "+str(self.datasetPath)
223	>	raise  CrabException(msg)
224	>
225	>
226	>	if len(self.eventsPerBlock) <= 0:
227	>	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
228	>	+ " dataset path variables in crab.cfg")
229	>	% self.datasetPath)
230	>
231	>
232	>	def queryDbs(self,api,path=None,runselection=None,useParent=None):
233	>
234	>
235	>	allowedRetriveValue = []
236	>	if self.splitByLumi or self.splitByRun or useParent == 1:
237	>	allowedRetriveValue.extend(['retrive_block', 'retrive_run'])
238	>	if self.splitByLumi:
239	>	allowedRetriveValue.append('retrive_lumi')
240	>	if useParent == 1:
241	>	allowedRetriveValue.append('retrive_parent')
242	>	common.logger.debug("Set of input parameters used for DBS query: %s" % allowedRetriveValue)
243		try:
244	<	self.datasets = dbs.getMatchingDatasets(self.owner, self.dataset)
245	<	except DBSError, ex:
246	<	raise DataDiscoveryError(ex.getErrorMessage())
247	<	if len(self.datasets) == 0:
248	<	raise DataDiscoveryError("Owner=%s, Dataset=%s unknown to DBS" % (self.owner, self.dataset))
249	<	if len(self.datasets) > 1:
250	<	raise DataDiscoveryError("Owner=%s, Dataset=%s is ambiguous" % (self.owner, self.dataset))
251	<	try:
252	<	self.dbsdataset = self.datasets[0].get('datasetPathName')
253	<	self.blocksinfo = dbs.getDatasetContents(self.dbsdataset)
254	<	self.allblocks.append (self.blocksinfo.keys ()) # add also the current fileblocksinfo
255	<	self.dbspaths.append(self.dbsdataset)
256	<	except DBSError, ex:
257	<	raise DataDiscoveryError(ex.getErrorMessage())
258	<
259	<	if len(self.blocksinfo)<=0:
260	<	msg="\nERROR Data for %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset
261	<	raise NotExistingDatasetError(msg)
244	>	if self.splitByRun:
245	>	files = []
246	>	for arun in runselection:
247	>	try:
248	>	if self.ads:
249	>	filesinrun = api.listFiles(analysisDataset=path,retriveList=allowedRetriveValue,runNumber=arun)
250	>	else:
251	>	filesinrun = api.listFiles(path=path,retriveList=allowedRetriveValue,runNumber=arun)
252	>	files.extend(filesinrun)
253	>	except:
254	>	msg="WARNING: problem extracting info from DBS for run %s "%arun
255	>	common.logger.info(msg)
256	>	pass
257	>
258	>	else:
259	>	if allowedRetriveValue:
260	>	if self.ads:
261	>	files = api.listFiles(analysisDataset=path, retriveList=allowedRetriveValue)
262	>	else :
263	>	files = api.listFiles(path=path, retriveList=allowedRetriveValue)
264	>	else:
265	>	files = api.listDatasetFiles(self.datasetPath)
266	>
267	>	except DbsBadRequest, msg:
268	>	raise DataDiscoveryError(msg)
269	>	except DBSError, msg:
270	>	raise DataDiscoveryError(msg)
271
272	+	return files
273
88	–	## get info about the parents
89	–	try:
90	–	parents=dbs.getDatasetProvenance(self.dbsdataset, self.dataTiers)
91	–	except DBSInvalidDataTierError, ex:
92	–	msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n'
93	–	raise DataDiscoveryError(msg)
94	–	except DBSError, ex:
95	–	raise DataDiscoveryError(ex.getErrorMessage())
274
275	<	## check that the user asks for parent Data Tier really existing in the DBS provenance
276	<	self.checkParentDataTier(parents, self.dataTiers)
275	>	def getMaxEvents(self):
276	>	"""
277	>	max events
278	>	"""
279	>	return self.maxEvents
280
281	<	## for each parent get the corresponding fileblocks
282	<	try:
283	<	for p in parents:
284	<	## fill a list of dbspaths
285	<	parentPath = p.get('parent').get('datasetPathName')
286	<	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())
110	<
111	<	# #################################################
112	<	def checkParentDataTier(self, parents, dataTiers):
113	<	"""
114	<	check that the data tiers requested by the user really exists in the provenance of the given dataset
115	<	"""
116	<	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)
281	>
282	>	def getMaxLumis(self):
283	>	"""
284	>	Return the number of lumis in the dataset
285	>	"""
286	>	return self.maxLumis
287
288
289	<	# #################################################
126	<	def getMaxEvents(self):
289	>	def getEventsPerBlock(self):
290		"""
291	<	max events of the primary dataset-owner
291	>	list the event collections structure by fileblock
292		"""
293	<	## loop over the fileblocks of the primary dataset-owner
131	<	nevts=0
132	<	for blockevts in self.blocksinfo.values():
133	<	nevts=nevts+blockevts
293	>	return self.eventsPerBlock
294
135	–	return nevts
295
296	<	# #################################################
138	<	def getDBSPaths(self):
296	>	def getEventsPerFile(self):
297		"""
298	<	list the DBSpaths for all required data
298	>	list the event collections structure by file
299		"""
300	<	return self.dbspaths
300	>	return self.eventsPerFile
301
302	<	# #################################################
303	<	def getEVC(self):
302	>
303	>	def getFiles(self):
304		"""
305	<	list the event collections structure by fileblock
305	>	return files grouped by fileblock
306		"""
307	<	print "To be used by a more complex job splitting... TODO later... "
308	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
307	>	return self.blocksinfo
308	>
309
310	<	# #################################################
153	<	def getFileBlocks(self):
310	>	def getParent(self):
311		"""
312	<	fileblocks for all required dataset-owners
312	>	return parent grouped by file
313		"""
314	<	return self.allblocks
314	>	return self.parent
315
316	<	########################################################################
316	>
317	>	def getLumis(self):
318	>	"""
319	>	return lumi sections grouped by file
320	>	"""
321	>	return self.lumis
322
323
324	+	def getListFiles(self):
325	+	"""
326	+	return parent grouped by file
327	+	"""
328	+	return self.files

Diff Legend

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