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 1.32 by ewv, Wed Jul 29 21:41:50 2009 UTC

#	Line 1 | Line 1
1		#!/usr/bin/env python
2	<	import sys, os, string, re
3	<	from DBSInfo import *
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	>
23	>	class DBSInvalidDataTierError(exceptions.Exception):
24	>	def __init__(self, errorName, errorMessage):
25	>	args='\nERROR DBS %s : %s \n'%(errorName,errorMessage)
26	>	exceptions.Exception.__init__(self, args)
27	>	pass
28	>
29	>	def getErrorMessage(self):
30	>	""" Return error message """
31	>	return "%s" % (self.args)
32	>
33	>
34	>
35	>	class DBSInfoError:
36	>	def __init__(self, url):
37	>	print '\nERROR accessing DBS url : '+url+'\n'
38	>	pass
39	>
40
41
6	–	# ####################################
42		class DataDiscoveryError(exceptions.Exception):
43		def __init__(self, errorMessage):
44	<	args=errorMessage
45	<	exceptions.Exception.__init__(self, args)
44	>	self.args=errorMessage
45	>	exceptions.Exception.__init__(self, self.args)
46		pass
47
48		def getErrorMessage(self):
49		""" Return exception error """
50		return "%s" % (self.args)
51
52	<	# ####################################
52	>
53	>
54		class NotExistingDatasetError(exceptions.Exception):
55		def __init__(self, errorMessage):
56	<	args=errorMessage
57	<	exceptions.Exception.__init__(self, args)
56	>	self.args=errorMessage
57	>	exceptions.Exception.__init__(self, self.args)
58		pass
59
60		def getErrorMessage(self):
61		""" Return exception error """
62		return "%s" % (self.args)
63
64	<	# ####################################
64	>
65	>
66		class NoDataTierinProvenanceError(exceptions.Exception):
67		def __init__(self, errorMessage):
68	<	args=errorMessage
69	<	exceptions.Exception.__init__(self, args)
68	>	self.args=errorMessage
69	>	exceptions.Exception.__init__(self, self.args)
70		pass
71
72		def getErrorMessage(self):
73		""" Return exception error """
74		return "%s" % (self.args)
75
39	–	# ####################################
40	–	# class to find and extact info from published data
41	–	class DataDiscovery:
42	–	def __init__(self, owner, dataset, dataTiers, cfg_params):
76
77	<	#       Attributes
78	<	self.owner = owner
79	<	self.dataset = dataset
80	<	self.dataTiers = dataTiers
77	>
78	>	class DataDiscovery:
79	>	"""
80	>	Class to find and extact info from published data
81	>	"""
82	>	def __init__(self, datasetPath, cfg_params, skipAnBlocks):
83	>
84	>	#       Attributes
85	>	self.datasetPath = datasetPath
86	>	# Analysis dataset is primary/processed/tier/definition
87	>	self.ads = len(self.datasetPath.split("/")) > 3
88		self.cfg_params = cfg_params
89	+	self.skipBlocks = skipAnBlocks
90	+
91	+	self.eventsPerBlock = {}  # DBS output: map fileblocks-events for collection
92	+	self.eventsPerFile = {}   # DBS output: map files-events
93	+	#         self.lumisPerBlock = {}   # DBS output: number of lumis in each block
94	+	#         self.lumisPerFile = {}    # DBS output: number of lumis in each file
95	+	self.blocksinfo = {}      # DBS output: map fileblocks-files
96	+	self.maxEvents = 0        # DBS output: max events
97	+	self.maxLumis = 0         # DBS output: total number of lumis
98	+	self.parent = {}          # DBS output: parents of each file
99	+	self.lumis = {}           # DBS output: lumis in each file
100
50	–	self.dbspaths= []     # DBS output: list of dbspaths for all data
51	–	self.allblocks = []   # DBS output: list of map fileblocks-totevts for all dataset-owners
52	–	self.blocksinfo = {}  # DBS output: map fileblocks-totevts for the primary block, used internally to this class
53	–	#DBS output: max events computed by method getMaxEvents
101
55	–	# ####################################
102		def fetchDBSInfo(self):
103		"""
104		Contact DBS
105		"""
106	+	## get DBS URL
107	+	global_url="http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
108	+	caf_url = "http://cmsdbsprod.cern.ch/cms_dbs_caf_analysis_01/servlet/DBSServlet"
109	+	dbs_url_map  =   {'glite':    global_url,
110	+	'glitecoll':global_url,\
111	+	'condor':   global_url,\
112	+	'condor_g': global_url,\
113	+	'glidein':  global_url,\
114	+	'lsf':      global_url,\
115	+	'caf':      caf_url,\
116	+	'sge':      global_url,
117	+	'arc':      global_url
118	+	}
119	+
120	+	dbs_url_default = dbs_url_map[(common.scheduler.name()).lower()]
121	+	dbs_url=  self.cfg_params.get('CMSSW.dbs_url', dbs_url_default)
122	+	common.logger.debug("Accessing DBS at: "+dbs_url)
123	+
124	+	## check if runs are selected
125	+	runselection = []
126	+	if (self.cfg_params.has_key('CMSSW.runselection')):
127	+	runselection = parseRange2(self.cfg_params['CMSSW.runselection'])
128	+
129	+
130	+	self.splitByRun = int(self.cfg_params.get('CMSSW.split_by_run', 0))
131	+
132	+	common.logger.log(10-1,"runselection is: %s"%runselection)
133	+	## service API
134	+	args = {}
135	+	args['url']     = dbs_url
136	+	args['level']   = 'CRITICAL'
137	+
138	+	## check if has been requested to use the parent info
139	+	useparent = int(self.cfg_params.get('CMSSW.use_parent',0))
140	+
141	+	## check if has been asked for a non default file to store/read analyzed fileBlocks
142	+	defaultName = common.work_space.shareDir()+'AnalyzedBlocks.txt'
143	+	fileBlocks_FileName = os.path.abspath(self.cfg_params.get('CMSSW.fileblocks_file',defaultName))
144	+
145	+	api = DBSAPI.dbsApi.DbsApi(args)
146	+	self.files = self.queryDbs(api,path=self.datasetPath,runselection=runselection,useParent=useparent)
147	+
148	+	anFileBlocks = []
149	+	if self.skipBlocks: anFileBlocks = readTXTfile(self, fileBlocks_FileName)
150	+
151	+	# parse files and fill arrays
152	+	for file in self.files :
153	+	parList  = []
154	+	lumiList = [] # List of tuples
155	+	# skip already analyzed blocks
156	+	fileblock = file['Block']['Name']
157	+	if fileblock not in anFileBlocks :
158	+	filename = file['LogicalFileName']
159	+	# asked retry the list of parent for the given child
160	+	if useparent==1:
161	+	parList = [x['LogicalFileName'] for x in file['ParentList']]
162	+	if self.ads:
163	+	lumiList = [ (x['RunNumber'], x['LumiSectionNumber'])
164	+	for x in file['LumiList'] ]
165	+	self.parent[filename] = parList
166	+	self.lumis[filename] = lumiList
167	+	if filename.find('.dat') < 0 :
168	+	events    = file['NumberOfEvents']
169	+	# Count number of events and lumis per block
170	+	if fileblock in self.eventsPerBlock.keys() :
171	+	self.eventsPerBlock[fileblock] += events
172	+	else :
173	+	self.eventsPerBlock[fileblock] = events
174	+	# Number of events per file
175	+	self.eventsPerFile[filename] = events
176	+
177	+	# List of files per block
178	+	if fileblock in self.blocksinfo.keys() :
179	+	self.blocksinfo[fileblock].append(filename)
180	+	else :
181	+	self.blocksinfo[fileblock] = [filename]
182	+
183	+	# total number of events
184	+	self.maxEvents += events
185	+	self.maxLumis  += len(lumiList)
186	+
187	+	if  self.skipBlocks and len(self.eventsPerBlock.keys()) == 0:
188	+	msg = "No new fileblocks available for dataset: "+str(self.datasetPath)
189	+	raise  CrabException(msg)
190	+
191	+	saveFblocks=''
192	+	for block in self.eventsPerBlock.keys() :
193	+	saveFblocks += str(block)+'\n'
194	+	common.logger.log(10-1,"DBSInfo: total nevts %i in block %s "%(self.eventsPerBlock[block],block))
195	+	writeTXTfile(self, fileBlocks_FileName , saveFblocks)
196	+
197	+	if len(self.eventsPerBlock) <= 0:
198	+	raise NotExistingDatasetError(("\nNo data for %s in DBS\nPlease check"
199	+	+ " dataset path variables in crab.cfg")
200	+	% self.datasetPath)
201	+
202	+
203	+	def queryDbs(self,api,path=None,runselection=None,useParent=None):
204	+
205	+	allowedRetriveValue = ['retrive_block', 'retrive_run']
206	+	if self.ads: allowedRetriveValue.append('retrive_lumi')
207	+	if useParent == 1: allowedRetriveValue.append('retrive_parent')
208	+	common.logger.debug("Set of input parameters used for DBS query: %s" % allowedRetriveValue)
209	+	try:
210	+	if len(runselection) <=0 :
211	+	if useParent==1 or self.splitByRun==1 :
212	+	if self.ads:
213	+	files = api.listFiles(analysisDataset=path, retriveList=allowedRetriveValue)
214	+	else :
215	+	files = api.listFiles(path=path, retriveList=allowedRetriveValue)
216	+	else:
217	+	files = api.listDatasetFiles(self.datasetPath)
218	+	else :
219	+	files=[]
220	+	for arun in runselection:
221	+	try:
222	+	if self.ads:
223	+	filesinrun = api.listFiles(analysisDataset=path,retriveList=allowedRetriveValue,runNumber=arun)
224	+	else:
225	+	filesinrun = api.listFiles(path=path,retriveList=allowedRetriveValue,runNumber=arun)
226	+	files.extend(filesinrun)
227	+	except:
228	+	msg="WARNING: problem extracting info from DBS for run %s "%arun
229	+	common.logger.info(msg)
230	+	pass
231
232	<	## add the PU among the required data tiers if the Digi are requested
233	<	if (self.dataTiers.count('Digi')>0) & (self.dataTiers.count('PU')<=0) :
234	<	self.dataTiers.append('PU')
232	>	except DbsBadRequest, msg:
233	>	raise DataDiscoveryError(msg)
234	>	except DBSError, msg:
235	>	raise DataDiscoveryError(msg)
236
237	<	## 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)
237	>	return files
238
239
240	<	## get info about the parents
241	<	try:
242	<	parents=dbs.getDatasetProvenance(self.dbsdataset, self.dataTiers)
243	<	except DBSInvalidDataTierError, ex:
244	<	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())
240	>	def getMaxEvents(self):
241	>	"""
242	>	max events
243	>	"""
244	>	return self.maxEvents
245
97	–	## check that the user asks for parent Data Tier really existing in the DBS provenance
98	–	self.checkParentDataTier(parents, self.dataTiers)
246
247	<	## for each parent get the corresponding fileblocks
248	<	try:
249	<	for p in parents:
250	<	## fill a list of dbspaths
251	<	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)
247	>	def getEventsPerBlock(self):
248	>	"""
249	>	list the event collections structure by fileblock
250	>	"""
251	>	return self.eventsPerBlock
252
253
254	<	# #################################################
126	<	def getMaxEvents(self):
254	>	def getEventsPerFile(self):
255		"""
256	<	max events of the primary dataset-owner
256	>	list the event collections structure by file
257		"""
258	<	## loop over the fileblocks of the primary dataset-owner
131	<	nevts=0
132	<	for blockevts in self.blocksinfo.values():
133	<	nevts=nevts+blockevts
258	>	return self.eventsPerFile
259
135	–	return nevts
260
261	<	# #################################################
138	<	def getDBSPaths(self):
261	>	def getFiles(self):
262		"""
263	<	list the DBSpaths for all required data
263	>	return files grouped by fileblock
264		"""
265	<	return self.dbspaths
265	>	return self.blocksinfo
266	>
267
268	<	# #################################################
145	<	def getEVC(self):
268	>	def getParent(self):
269		"""
270	<	list the event collections structure by fileblock
270	>	return parent grouped by file
271		"""
272	<	print "To be used by a more complex job splitting... TODO later... "
273	<	print "it requires changes in what's returned by DBSInfo.getDatasetContents and then fetchDBSInfo"
272	>	return self.parent
273	>
274
275	<	# #################################################
153	<	def getFileBlocks(self):
275	>	def getLumis(self):
276		"""
277	<	fileblocks for all required dataset-owners
277	>	return lumi sections grouped by file
278		"""
279	<	return self.allblocks
279	>	return self.lumis
280
281	<	########################################################################
281	>
282	>	def getListFiles(self):
283	>	"""
284	>	return parent grouped by file
285	>	"""
286	>	return self.files

Diff Legend

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