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 1.7 by slacapra, Thu Jul 6 10:22:47 2006 UTC vs.
Revision 1.46 by spiga, Wed Jun 2 13:55:14 2010 UTC

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

Diff Legend

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