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 1.4 by afanfani, Sun Mar 12 01:00:47 2006 UTC vs.
Revision 1.42 by ewv, Mon Mar 22 21:17:15 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	>	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)
51	<	pass
52	<
53	<	def getErrorMessage(self):
54	<	""" Return exception error """
55	<	return "%s" % (self.args)
48	>	def __init__(self, errorMessage):
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	>
58
17	–	# ####################################
59		class NotExistingDatasetError(exceptions.Exception):
60	<	def __init__(self, errorMessage):
61	<	args=errorMessage
62	<	exceptions.Exception.__init__(self, args)
63	<	pass
64	<
65	<	def getErrorMessage(self):
66	<	""" Return exception error """
67	<	return "%s" % (self.args)
60	>	def __init__(self, errorMessage):
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	>
70
28	–	# ####################################
71		class NoDataTierinProvenanceError(exceptions.Exception):
72	<	def __init__(self, errorMessage):
73	<	args=errorMessage
74	<	exceptions.Exception.__init__(self, args)
75	<	pass
76	<
77	<	def getErrorMessage(self):
78	<	""" Return exception error """
79	<	return "%s" % (self.args)
72	>	def __init__(self, errorMessage):
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
55	–	# ####################################
107		def fetchDBSInfo(self):
108		"""
109		Contact DBS
110		"""
111	<
112	<	## add the PU among the required data tiers if the Digi are requested
113	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
114	<	self.dataTiers.append('PU')
115	<
116	<	## get info about the requested dataset
117	<	dbs=DBSInfo()
118	<	self.datasets = dbs.getMatchingDatasets(self.owner, self.dataset)
119	<	if len(self.datasets) == 0:
120	<	raise DataDiscoveryError("Owner=%s, Dataset=%s unknown to DBS" % (self.owner, self.dataset))
121	<	if len(self.datasets) > 1:
122	<	raise DataDiscoveryError("Owner=%s, Dataset=%s is ambiguous" % (self.owner, self.dataset))
111	>	## get DBS URL
112	>	global_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
113	>	caf_url = "http://cmsdbsprod.cern.ch/cms_dbs_caf_analysis_01/servlet/DBSServlet"
114	>	dbs_url_map  =   {'glite':    global_url,
115	>	'glite_slc5':global_url,\
116	>	'glitecoll':global_url,\
117	>	'condor':   global_url,\
118	>	'condor_g': global_url,\
119	>	'glidein':  global_url,\
120	>	'lsf':      global_url,\
121	>	'caf':      caf_url,\
122	>	'sge':      global_url,\
123	>	'arc':      global_url,\
124	>	'pbs':      global_url
125	>	}
126	>
127	>	dbs_url_default = dbs_url_map[(common.scheduler.name()).lower()]
128	>	dbs_url=  self.cfg_params.get('CMSSW.dbs_url', dbs_url_default)
129	>	common.logger.info("Accessing DBS at: "+dbs_url)
130	>
131	>	## check if runs are selected
132	>	runselection = []
133	>	if (self.cfg_params.has_key('CMSSW.runselection')):
134	>	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
135	>
136	>	## check if various lumi parameters are set
137	>	self.lumiMask = self.cfg_params.get('CMSSW.lumi_mask',None)
138	>	self.lumiParams = self.cfg_params.get('CMSSW.total_number_of_lumis',None) or \
139	>	self.cfg_params.get('CMSSW.lumis_per_job',None)
140	>
141	>	lumiList = None
142	>	if self.lumiMask:
143	>	lumiList = LumiList(filename=self.lumiMask)
144	>
145	>	self.splitByRun = int(self.cfg_params.get('CMSSW.split_by_run', 0))
146	>
147	>	common.logger.log(10-1,"runselection is: %s"%runselection)
148	>	## service API
149	>	args = {}
150	>	args['url']     = dbs_url
151	>	args['level']   = 'CRITICAL'
152	>
153	>	## check if has been requested to use the parent info
154	>	useparent = int(self.cfg_params.get('CMSSW.use_parent',0))
155	>
156	>	## check if has been asked for a non default file to store/read analyzed fileBlocks
157	>	defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt'
158	>	fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName))
159	>
160	>	api = DBSAPI.dbsApi.DbsApi(args)
161	>	self.files = self.queryDbs(api,path=self.datasetPath,runselection=runselection,useParent=useparent)
162	>
163	>	anFileBlocks = []
164	>	if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName)
165	>
166	>	# parse files and fill arrays
167	>	for file in self.files :
168	>	parList  = []
169	>	fileLumis = [] # List of tuples
170	>	# skip already analyzed blocks
171	>	fileblock = file['Block']['Name']
172	>	if fileblock not in anFileBlocks :
173	>	filename = file['LogicalFileName']
174	>	# asked retry the list of parent for the given child
175	>	if useparent==1:
176	>	parList = [x['LogicalFileName'] for x in file['ParentList']]
177	>	if self.ads or self.lumiMask or self.lumiParams:
178	>	fileLumis = [ (x['RunNumber'], x['LumiSectionNumber'])
179	>	for x in file['LumiList'] ]
180	>	self.parent[filename] = parList
181	>	# For LumiMask, intersection of two lists.
182	>	if self.lumiMask:
183	>	self.lumis[filename] = lumiList.filterLumis(fileLumis)
184	>	else:
185	>	self.lumis[filename] = fileLumis
186	>	if filename.find('.dat') < 0 :
187	>	events    = file['NumberOfEvents']
188	>	# Count number of events and lumis per block
189	>	if fileblock in self.eventsPerBlock.keys() :
190	>	self.eventsPerBlock[fileblock] += events
191	>	else :
192	>	self.eventsPerBlock[fileblock] = events
193	>	# Number of events per file
194	>	self.eventsPerFile[filename] = events
195	>
196	>	# List of files per block
197	>	if fileblock in self.blocksinfo.keys() :
198	>	self.blocksinfo[fileblock].append(filename)
199	>	else :
200	>	self.blocksinfo[fileblock] = [filename]
201	>
202	>	# total number of events
203	>	self.maxEvents += events
204	>	self.maxLumis  += len(self.lumis[filename])
205	>
206	>	if  self.skipBlocks and len(self.eventsPerBlock.keys()) == 0:
207	>	msg = "No new fileblocks available for dataset: "+str(self.datasetPath)
208	>	raise  CrabException(msg)
209	>
210	>	saveFblocks=''
211	>	for block in self.eventsPerBlock.keys() :
212	>	saveFblocks += str(block)+'\n'
213	>	common.logger.log(10-1,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
214	>	writeTXTfile(self, fileBlocks_FileName , saveFblocks)
215	>
216	>	if len(self.eventsPerBlock) <= 0:
217	>	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
218	>	+ " dataset path variables in crab.cfg")
219	>	% self.datasetPath)
220	>
221	>
222	>	def queryDbs(self,api,path=None,runselection=None,useParent=None):
223	>
224	>	allowedRetriveValue = ['retrive_block', 'retrive_run']
225	>	if self.ads or self.lumiMask or self.lumiParams:
226	>	allowedRetriveValue.append('retrive_lumi')
227	>	if useParent == 1: allowedRetriveValue.append('retrive_parent')
228	>	common.logger.debug("Set of input parameters used for DBS query: %s" % allowedRetriveValue)
229		try:
230	<	self.dbsdataset = self.datasets[0].getDatasetPath()
231	<	self.blocksinfo = dbs.getDatasetContents(self.dbsdataset)
232	<	self.allblocks.append (self.blocksinfo.keys ()) # add also the current fileblocksinfo
233	<	self.dbspaths.append(self.dbsdataset)
234	<	except DBSError, ex:
235	<	raise DataDiscoveryError(ex.getErrorMessage())
236	<
237	<	if len(self.blocksinfo)<=0:
238	<	msg="\nERROR Data for %s do not exist in DBS! \n Check the dataset/owner variables in crab.cfg !"%self.dbsdataset
239	<	raise NotExistingDatasetError(msg)
230	>	if len(runselection) <=0 :
231	>	if useParent==1 or self.splitByRun==1 or self.ads or self.lumiMask or self.lumiParams:
232	>	if self.ads:
233	>	files = api.listFiles(analysisDataset=path, retriveList=allowedRetriveValue)
234	>	else :
235	>	files = api.listFiles(path=path, retriveList=allowedRetriveValue)
236	>	else:
237	>	files = api.listDatasetFiles(self.datasetPath)
238	>	else :
239	>	files=[]
240	>	for arun in runselection:
241	>	try:
242	>	if self.ads:
243	>	filesinrun = api.listFiles(analysisDataset=path,retriveList=allowedRetriveValue,runNumber=arun)
244	>	else:
245	>	filesinrun = api.listFiles(path=path,retriveList=allowedRetriveValue,runNumber=arun)
246	>	files.extend(filesinrun)
247	>	except:
248	>	msg="WARNING: problem extracting info from DBS for run %s "%arun
249	>	common.logger.info(msg)
250	>	pass
251	>
252	>	except DbsBadRequest, msg:
253	>	raise DataDiscoveryError(msg)
254	>	except DBSError, msg:
255	>	raise DataDiscoveryError(msg)
256
257	+	return files
258
85	–	## get info about the parents
86	–	try:
87	–	parents=dbs.getDatasetProvenance(self.dbsdataset, self.dataTiers)
88	–	except DBSInvalidDataTierError, ex:
89	–	msg=ex.getErrorMessage()+' \n Check the data_tier variable in crab.cfg !\n'
90	–	raise DataDiscoveryError(msg)
91	–	except DBSError, ex:
92	–	raise DataDiscoveryError(ex.getErrorMessage())
259
260	<	## check that the user asks for parent Data Tier really existing in the DBS provenance
261	<	self.checkParentDataTier(parents, self.dataTiers)
260	>	def getMaxEvents(self):
261	>	"""
262	>	max events
263	>	"""
264	>	return self.maxEvents
265
97	–	## for each parent get the corresponding fileblocks
98	–	try:
99	–	for p in parents:
100	–	## fill a list of dbspaths
101	–	parentPath = p.getDatasetPath()
102	–	self.dbspaths.append (parentPath)
103	–	parentBlocks = dbs.getDatasetContents (parentPath)
104	–	self.allblocks.append (parentBlocks.keys ())  # add parent fileblocksinfo
105	–	except DBSError, ex:
106	–	raise DataDiscoveryError(ex.getErrorMessage())
107	–
108	–	# #################################################
109	–	def checkParentDataTier(self, parents, dataTiers):
110	–	"""
111	–	check that the data tiers requested by the user really exists in the provenance of the given dataset
112	–	"""
113	–	startType = string.split(self.dbsdataset,'/')[2]
114	–	parentTypes = map(lambda p: p.getDataType(), parents)
115	–	for tier in dataTiers:
116	–	if parentTypes.count(tier) <= 0 and tier != startType:
117	–	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)
118	–	raise  NoDataTierinProvenanceError(msg)
266
267	+	def getMaxLumis(self):
268	+	"""
269	+	Return the number of lumis in the dataset
270	+	"""
271	+	return self.maxLumis
272
273	<	# #################################################
274	<	def getMaxEvents(self):
273	>
274	>	def getEventsPerBlock(self):
275		"""
276	<	max events of the primary dataset-owner
276	>	list the event collections structure by fileblock
277		"""
278	<	## loop over the fileblocks of the primary dataset-owner
127	<	nevts=0
128	<	for blockevts in self.blocksinfo.values():
129	<	nevts=nevts+blockevts
278	>	return self.eventsPerBlock
279
131	–	return nevts
280
281	<	# #################################################
134	<	def getDBSPaths(self):
281	>	def getEventsPerFile(self):
282		"""
283	<	list the DBSpaths for all required data
283	>	list the event collections structure by file
284		"""
285	<	return self.dbspaths
285	>	return self.eventsPerFile
286	>
287
288	<	# #################################################
141	<	def getEVC(self):
288	>	def getFiles(self):
289		"""
290	<	list the event collections structure by fileblock
290	>	return files grouped by fileblock
291		"""
292	<	print "To be used by a more complex job splitting... TODO later... "
146	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
292	>	return self.blocksinfo
293
294	<	# #################################################
295	<	def getFileBlocks(self):
294	>
295	>	def getParent(self):
296		"""
297	<	fileblocks for all required dataset-owners
297	>	return parent grouped by file
298		"""
299	<	return self.allblocks
299	>	return self.parent
300
301	<	########################################################################
301	>
302	>	def getLumis(self):
303	>	"""
304	>	return lumi sections grouped by file
305	>	"""
306	>	return self.lumis
307
308
309	+	def getListFiles(self):
310	+	"""
311	+	return parent grouped by file
312	+	"""
313	+	return self.files

Diff Legend

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